cs.RO / 1 / 2606.20641
MAGNIFIED: RL Fine-tuning of Multimodal Large Language Models for Motion Planning
MAGNIFIED：针对运动规划的多模态大型语言模型的强化学习微调
Abstract
Multi-modal Large Language Models (MLLMs) have demonstrated remarkable capabilities in semantic understanding and common sense reasoning, making them promising candidates for solving planning problems in autonomous driving. However, the next-token text prediction objectives traditionally used in pre-training and supervised fine-tuning (SFT) of MLLMs may fall short of fulfilling the planning objectives for autonomous vehicles. The next-token prediction objective merely encourages per-token imitation in text, often irrespective of multi-step consequences and the alignment with crucial planning considerations such as giving space to other road actors. To overcome these limitations, we propose a reinforcement learning fine-tuning (RLFT) approach, MAGNIFIED, that aligns the MLLM-based driving agent with planning objectives by learning from token-level rewards. By mapping a sequence of predicted tokens to corresponding vehicle trajectories and learning from planning rewards, MAGNIFIED optimizes for the true planning objectives rather than focusing solely on token prediction accuracy, enabling the model to refine its understanding of the planning task beyond simple imitation. We validate our approach on the Waymo Open Motion Dataset with a novel setup incorporating rasterized birds-eye views and tokenized trajectories as inputs and planning-oriented outputs. An initial SFT phase establishes a strong baseline in outputting plan trajectories as sequences of X-Y coordinates in text, while subsequent RL fine-tuning substantially enhances planning performance relative to the SFT baseline (demonstrating over a 10.5% reduction in overlap rate and a 38.9% reduction in off-road rate), underscoring the potential of RLFT on MLLMs to achieve vehicle planning that is better aligned with compliant, comfortable, and efficient driving.
Chinese Translation
多模态大型语言模型（MLLMs）在语义理解和常识推理方面展现出了显著的能力，使其成为解决自动驾驶规划问题的有希望的候选者。然而，传统上用于MLLMs的预训练和监督微调（SFT）的下一个标记文本预测目标，可能无法满足自动驾驶车辆的规划目标。下一个标记预测目标仅鼓励文本中的逐标记模仿，往往忽视多步骤后果以及与关键规划考虑（如为其他道路参与者留出空间）的对齐。为克服这些局限性，我们提出了一种强化学习微调（RLFT）方法MAGNIFIED，通过从标记级奖励中学习，将基于MLLM的驾驶代理与规划目标对齐。通过将一系列预测的标记映射到相应的车辆轨迹并从规划奖励中学习，MAGNIFIED优化真实的规划目标，而不仅仅关注标记预测的准确性，使模型能够超越简单模仿，提升对规划任务的理解。我们在Waymo开放运动数据集上验证了我们的方法，采用了新的设置，将栅格化的鸟瞰图和标记化的轨迹作为输入，并生成以规划为导向的输出。初步的SFT阶段建立了输出计划轨迹作为文本中X-Y坐标序列的强基线，而随后的RL微调相较于SFT基线显著提升了规划性能（重叠率减少超过10.5%，越界率减少38.9%），突显了RLFT在MLLMs上实现更符合合规、舒适和高效驾驶的车辆规划的潜力。
cs.RO / 2 / 2606.20645
TACT-ful: Multi-Channel Terrain Affordance and Compliance Training for Payload-Robust Perceptive Humanoid Locomotion
TACT-ful：多通道地形可供性与顺应性训练以实现负载稳健的感知类人步态
Abstract
Foothold selection on structured terrain requires explicit reasoning about contact planarity, surface steepness, and kinematic reachability, properties not captured by a single height-based terrain signal. We propose a multi-channel terrain cost combining flatness, steepness, and velocity-aware height feasibility, plus a forward climb reward, that simultaneously drives a GPU-parallel divergent component of motion (DCM) foothold planner and shapes a dense per-step affordance reward for an asymmetric actor-critic policy trained with proximal policy optimization (PPO) from depth images. A B\'ezier swing trajectory with adaptive apex bias extends foothold tracking to joint position-and-orientation, using the arc tangent to guide sole orientation through riser crossings and tread landings. To support payload tasks, we introduce a lower-body compliance training procedure in which a virtual wrench is injected at a sampled load attachment point, generating physically consistent force and moment; wrench-aware compliance targets replace rigid pose penalties, and the policy learns to yield to load-induced perturbations without force sensing. The full system trains end-to-end with standard PPO, no distillation, and no teacher-student staging, and is deployed on a humanoid directly from simulation with configuration changes only. In simulation, the policy reaches $1.0~\mathrm{m/s}$ on stairs with risers up to $0.20~\mathrm{m}$ and improves payload robustness up to ${\sim}15~\mathrm{kg}$ centered load and for moment-dominated wrist loads without fine-tuning. We also provide a qualitative hardware demonstration on structured terrain. Project website: https://fai-rl-tech.github.io/tact-locomotion.github.io/
Chinese Translation
在结构化地形上选择支撑点需要对接触平面性、表面陡峭度和运动学可达性进行明确推理，这些特性无法通过单一的基于高度的地形信号捕捉。我们提出了一种多通道地形成本，结合了平坦度、陡峭度和考虑速度的高度可行性，以及前进攀爬奖励，同时驱动GPU并行的发散运动组件（DCM）支撑点规划器，并为使用近端策略优化（PPO）从深度图像训练的非对称演员-评论者策略塑造了密集的每步可供性奖励。自适应顶点偏差的Bézier摆动轨迹将支撑点跟踪扩展到关节位置和方向，利用反正切引导鞋底方向通过踏步交叉和踏面着陆。为了支持负载任务，我们引入了一种下肢顺应性训练程序，在采样的负载附着点注入虚拟扭矩，生成物理一致的力和力矩；考虑扭矩的顺应性目标替代了刚性姿态惩罚，策略学习在没有力传感的情况下对负载引起的扰动做出反应。整个系统通过标准PPO进行端到端训练，无需蒸馏和教师-学生阶段，并且仅通过配置更改直接从仿真部署到类人机器人。在仿真中，策略在高达$0.20~ ext{m}$的台阶上达到$1.0~ ext{m/s}$，并在不进行微调的情况下提高了负载稳健性，适用于约${ ilde{15}~ ext{kg}}$的集中负载和以力矩为主的手腕负载。我们还提供了在结构化地形上的定性硬件演示。项目网站：https://fai-rl-tech.github.io/tact-locomotion.github.io/
cs.RO / 3 / 2606.20647
Tessellated Biomes: Distributed Robotic Assemblies for Architectural Resilience
镶嵌生物群落：用于建筑韧性的分布式机器人组装
Abstract
This paper presents Tessellated Biomes, a cyber-physical framework for the adaptive robotic construction and reconfiguration of modular multi-material assemblies. It challenges the linear lifecycle of standard construction by fusing (1) local microfactory fabrication, (2) discrete multi-material optimization, and (3) distributed robotic assembly into a unified circular process of spatial adaptation. The research details methods for the digital fabrication of self-aligning modular primitives in multiple materials (PLA, timber, and concrete) produced in local microfactories; the aggregation and optimization of these primitives into compression-based discrete structures; and the deployment of custom quadrupedal robots that collaboratively relocate material into realized physical aggregations. The framework is validated through the fabrication, optimization, and robotic assembly of discrete structures. Together, these results position Tessellated Biomes as a model for resilient, reconfigurable architecture.
Chinese Translation
本文提出了镶嵌生物群落（Tessellated Biomes），这是一个用于模块化多材料组装的自适应机器人建造与重构的网络物理框架。它通过融合（1）地方微工厂制造，（2）离散多材料优化，以及（3）分布式机器人组装，挑战了标准建筑的线性生命周期，形成一个统一的空间适应循环过程。研究详细介绍了在地方微工厂中生产的多种材料（PLA、木材和混凝土）自对准模块原件的数字化制造方法；将这些原件聚合并优化为基于压缩的离散结构；以及部署定制的四足机器人，协作将材料重新定位到已实现的物理聚合体中。通过离散结构的制造、优化和机器人组装验证了该框架。这些结果共同将镶嵌生物群落定位为一种韧性和可重构建筑的模型。
cs.RO / 4 / 2606.20660
Learning Control as Enabling Layer for Embodied Intelligence Research explored with Soft Robotic Swimming in diverse Flow Speeds
学习控制作为赋能层：通过软体机器人游泳研究不同流速下的具身智能
Abstract
Soft robots are valuable robophysical platforms for studying body-caudal undulatory locomotion, but their compliant bodies are difficult to control precisely under changing hydrodynamic loading. Conventional proportional-integral-derivative (PID) feedback stabilizes periodic undulation in static water, but can accumulate flow-dependent tracking delay and increasing inter-trial variability when environmental flow becomes non-trivial. Here, we evaluate whether augmenting PID control with a Linear Repetitive Learning Estimation Scheme (PID-LRLES) recovers tracking accuracy and repeatability under dynamic flow. The LRLES generalizes classical integral action from constant to periodic, non-constant references, while using a stable transfer-function realization whose poles have negative real parts to avoid the long-term instability issues of classical repetitive control. Closed-loop experiments were carried out in a recirculating flow tank at five bulk flow speeds spanning 0 to 32.6 cm s^-1, using an embedded soft capacitive bending sensor at a 1 kHz control-loop rate. With controller gains tuned once in static water and then held fixed across all conditions, PID-LRLES tracked the periodic bending-envelope reference more closely than the PID baseline and significantly reduced the inter-trial spread of the per-trial RMSE (paired Wilcoxon signed-rank test, p = 1.8 x 10^-4, n = 25). Embedded soft proprioception and cycle-to-cycle learning act as complementary contributors to robustness: the sensor exposes the periodic hydrodynamic bias in body deformation, while the learning term absorbs it over recent oscillation cycles. By reducing flow-dependent control-induced variability, the approach provides an enabling layer for future robophysical studies seeking to isolate the effects of morphology, sensing, and environmental flow on aquatic locomotion.
Chinese Translation
软体机器人是研究身体-尾部波动运动的宝贵机器人物理平台，但其柔性身体在变化的水动力负载下难以精确控制。传统的比例-积分-微分（PID）反馈能够在静水中稳定周期性波动，但在环境流动变得复杂时，可能会积累流动依赖的跟踪延迟，并增加试验间的变异性。在此，我们评估通过线性重复学习估计方案（PID-LRLES）增强PID控制是否能够在动态流动下恢复跟踪精度和重复性。LRLES将经典积分作用从恒定参考推广到周期性、非恒定参考，同时使用稳定的传递函数实现，其极点具有负实部，以避免经典重复控制的长期不稳定性问题。在一个循环流动水槽中进行了闭环实验，涵盖了0至32.6 cm s^-1的五种体积流速，使用嵌入式软电容弯曲传感器，以1 kHz的控制循环频率进行测量。在静水中调节一次控制器增益后，在所有条件下保持不变，PID-LRLES比PID基线更紧密地跟踪周期性弯曲包络参考，并显著减少了每次试验均方根误差（RMSE）的试验间差异（配对Wilcoxon符号秩检验，p = 1.8 x 10^-4，n = 25）。嵌入式软本体感知和循环间学习作为增强鲁棒性的互补因素：传感器揭示了身体变形中的周期性水动力偏差，而学习项则在最近的振荡周期中吸收了这种偏差。通过减少流动依赖的控制引起的变异性，该方法为未来的机器人物理研究提供了一个赋能层，旨在隔离形态、感知和环境流动对水生运动的影响。
cs.RO / 5 / 2606.20679
MemoryVAM: Integrating Memory into Video Action Model for Robot Manipulation
MemoryVAM：将记忆集成到视频动作模型中以实现机器人操作
Abstract
Video-world-model policies learn action-relevant representations by predicting future observations. However, they condition on only a short observation window, which renders long-horizon manipulation non-Markovian when the correct action depends on earlier events that are no longer visible. We present MemoryVAM, an episodic memory mechanism for video-world-model policies. We employ a Recap-Cue (RC) module, in which a Perceiver-based Recap Compressor maps per-frame CLIP embeddings into compact memory tokens, and a lightweight Cue Gate estimates task completion from memory and language. These tokens are injected into both the video backbone and the action decoder, aligning policy imagination with episode progress and conditioning actions on history. Our model trains the memory module with video prediction, a delta-reconstruction auxiliary loss, and episode-boundary supervision, requiring no per-frame progress labels. The same mechanism applies to UNet and Diffusion Transformer (DiT) backbones by changing only the cross-attention injection interface. On LIBERO-Mem, our model improves average success from 5% to 42.5%. On real robots, it achieves 78.3% success on counting tasks, 80.0% on spatial recall, and 75.0% on sequential tracking. Project page: https://MemoryVAM.github.io/
Chinese Translation
视频世界模型策略通过预测未来观察来学习与动作相关的表征。然而，它们仅依赖于短期观察窗口，这使得当正确的动作依赖于不再可见的早期事件时，长时间跨度的操作变得非马尔可夫。我们提出了MemoryVAM，一种用于视频世界模型策略的情节记忆机制。我们采用了Recap-Cue (RC) 模块，其中基于Perceiver的Recap Compressor将每帧的CLIP嵌入映射为紧凑的记忆标记，而轻量级的Cue Gate则从记忆和语言中估计任务完成情况。这些标记被注入到视频主干和动作解码器中，使策略的想象与情节进展对齐，并根据历史条件化动作。我们的模型通过视频预测、增量重建辅助损失和情节边界监督来训练记忆模块，无需每帧的进展标签。相同的机制适用于UNet和Diffusion Transformer (DiT) 主干，仅需更改交叉注意力注入接口。在LIBERO-Mem上，我们的模型将平均成功率从5%提高到42.5%。在真实机器人上，它在计数任务上实现了78.3%的成功率，在空间回忆上为80.0%，在顺序跟踪上为75.0%。项目页面：https://MemoryVAM.github.io/
cs.RO / 6 / 2606.20686
JPPD: Joint Prediction_Planning Diffusion with Differentiable Safety Guidance for Dynamic Obstacle Avoidance in Intelligent Transportation Systems
JPPD：具有可微分安全引导的联合预测-规划扩散框架，用于智能交通系统中的动态障碍物避让
Abstract
Shared-space transportation operation requires low-speed autonomous platforms to navigate safely and efficiently among pedestrians, service robots, micromobility users, carts, and other road users. Most existing systems decompose this problem into trajectory prediction followed by motion planning, which creates one-way information flow: predicted participant futures influence the robot plan, but the selected robot plan cannot influence the predicted multi-agent evolution. This paper presents a joint prediction-planning diffusion framework that treats participant prediction and robot planning as a single conditional trajectory generation problem, where the model samples the future robot trajectory and all participant trajectories from one coupled distribution using a causal Transformer with cross-trajectory attention. To replace heuristic repulsive post-processing, the framework introduces differentiable safety potential guidance, a time-varying occupancy-probability potential whose gradient directly guides the joint sampler, and conditional flow matching is used to reduce inference steps while preserving multimodal trajectory diversity. The evaluation emphasizes shared-space operational effects, including near misses, blockage time, induced participant deviation, hard-braking events, and embedded latency, rather than treating average displacement error and final displacement error as the main result. Experiments in scenario-grounded simulation, naturalistic pedestrian replay, Isaac Sim validation, and ROS/Orin deployment show that joint sampling improves tail safety and runtime efficiency over a separated prediction-then-planning baseline.
Chinese Translation
共享空间的交通运营要求低速自主平台在行人、服务机器人、微型出行用户、手推车及其他道路使用者之间安全高效地导航。大多数现有系统将此问题分解为轨迹预测和运动规划，这造成了单向信息流：预测的参与者未来影响机器人规划，但选定的机器人规划无法影响预测的多智能体演化。本文提出了一种联合预测-规划扩散框架，将参与者预测和机器人规划视为一个单一的条件轨迹生成问题，其中模型使用具有跨轨迹注意力的因果Transformer从一个耦合分布中采样未来机器人轨迹和所有参与者轨迹。为了替代启发式排斥后处理，该框架引入了可微分安全势引导，这是一种时变的占用概率势，其梯度直接引导联合采样器，并使用条件流匹配来减少推理步骤，同时保持多模态轨迹的多样性。评估强调共享空间的操作效果，包括近失误、阻塞时间、引发的参与者偏差、急刹车事件和嵌入延迟，而不是将平均位移误差和最终位移误差视为主要结果。在情景基础模拟、自然行人重放、Isaac Sim验证和ROS/Orin部署的实验中，联合采样在尾部安全性和运行时效率上优于分离的预测-再规划基线。
cs.RO / 7 / 2606.20698
SafeDojo: Safe Reinforcement Learning for VLA via Interactive World Model
SafeDojo：通过交互式世界模型实现的安全强化学习用于视觉-语言-动作
Abstract
Safe control is a prerequisite for real-world embodied intelligence, for which safe reinforcement learning has emerged as a promising paradigm. However, existing safe reinforcement learning methods either require costly real-world exploration or depend on hand-crafted safety functions. Neither scales to vision-language-action models deployed in open-world physical environments. We propose SafeDojo, the first model-based safe reinforcement learning framework for vision-language-action policies designed to learn safe actions through world model-based imagination. Specifically, SafeDojo performs online reinforcement learning on top of an interactive video world model. The world model generates action-conditioned future predictions, from which a tailored ResNet success classifier estimates per-step task progress from imagined frames and a lightweight safety head predicts per-step safety costs from latent context together with the proposed action chunk, enabling simultaneous assessment of task execution and trajectory safety. The decoupled task-reward and safety-cost signals are balanced through a Lagrangian-based constrained GRPO objective, enabling coordinated improvement of task success and safety under explicit constraints. On SafeLIBERO, SafeDojo achieves the best aggregate task success, safe success, and execution efficiency among inference-time safety, model-free RL, and model-based RL baselines, with the best average safe-success rate on both levels and an 8.25 percentage-point improvement over the strongest baseline on Level I. Real-world Franka deployment further shows the best average task and safe-success rates across five tasks. Our results position world model-based safe reinforcement learning as a scalable and generalizable path toward safe embodied intelligence.
Chinese Translation
安全控制是实现现实世界具身智能的前提，而安全强化学习已成为一种有前景的范式。然而，现有的安全强化学习方法要么需要昂贵的现实世界探索，要么依赖于手工设计的安全函数，这两者都无法扩展到在开放世界物理环境中部署的视觉-语言-动作模型。我们提出了SafeDojo，这是第一个基于模型的安全强化学习框架，旨在通过基于世界模型的想象学习安全动作。具体而言，SafeDojo在交互式视频世界模型之上进行在线强化学习。该世界模型生成基于动作的未来预测，定制的ResNet成功分类器从想象的帧中估计每一步的任务进展，而轻量级安全头则从潜在上下文和所提议的动作块中预测每一步的安全成本，从而实现任务执行和轨迹安全的同时评估。解耦的任务奖励和安全成本信号通过基于拉格朗日的约束GRPO目标进行平衡，从而在明确约束下协调任务成功和安全的提升。在SafeLIBERO上，SafeDojo在推理时安全、无模型强化学习和基于模型的强化学习基准中实现了最佳的综合任务成功率、安全成功率和执行效率，在两个层次上都达到了最佳的平均安全成功率，并在Level I上比最强基线提高了8.25个百分点。现实世界的Franka部署进一步显示出在五个任务中最佳的平均任务和安全成功率。我们的结果将基于世界模型的安全强化学习定位为通向安全具身智能的可扩展和可推广的路径。
cs.RO / 8 / 2606.20712
Real-World Deployment of Massively Parallel Sampling-Based MPC for Contact-Rich Manipulation
大规模并行采样基础模型预测控制在接触丰富的操作中的实际部署
Abstract
Sampling-based Model Predictive Control (SMPC) is a promising strategy for contact-rich robotic manipulation, combining gradient-free optimization with massively parallel GPU simulation. Yet, most prior work relies on simplified dynamics or remains confined to simulation. We present an MPC framework that leverages JAX for large-scale parallelization and efficient computation, coupled with the high-fidelity MuJoCo MJX simulator, and deploy it on a Franka Research 3 executing the Push-T manipulation task through a complete real-to-sim-to-real pipeline. The MTP variant with structured global sampling outperforms unimodal baselines such as CEM, MPPI, and PS across tasks that require mode switching, both in simulation and on hardware. Furthermore, we evaluate online domain randomization within the MPC sample budget, showing that contact-initiation parameters yield interpretable adaptation signals, whereas global physics parameters provide feedback that is too weak for reliable exploitation at typical replanning frequencies. These findings highlight key challenges for sampling-based MPC in contact-rich manipulation-contact sensitivity, tight compute budgets, and the difficulty of obtaining informative domain-randomization signals in real time.
Chinese Translation
基于采样的模型预测控制（SMPC）是一种有前景的策略，适用于接触丰富的机器人操作，结合了无梯度优化和大规模并行GPU仿真。然而，大多数先前的工作依赖于简化的动力学模型或局限于仿真。我们提出了一种MPC框架，利用JAX实现大规模并行化和高效计算，并结合高保真度的MuJoCo MJX仿真器，将其部署在Franka Research 3上，执行Push-T操作任务，采用完整的真实-仿真-真实管道。具有结构化全局采样的MTP变体在需要模式切换的任务中，在仿真和硬件上均优于单峰基线算法，如CEM、MPPI和PS。此外，我们在MPC样本预算内评估了在线领域随机化，结果表明，接触启动参数产生可解释的适应信号，而全局物理参数提供的反馈在典型的重新规划频率下过于微弱，难以可靠利用。这些发现突显了基于采样的MPC在接触丰富操作中的关键挑战，包括接触敏感性、紧凑的计算预算，以及实时获取信息丰富的领域随机化信号的困难。
cs.RO / 9 / 2606.20739
Coupled Routing and Configuration Optimization for Multi-Viewpoint Robotic Inspection
多视角机器人检测的耦合路径与配置优化
Abstract
We present a unified framework that turns a set of 6-DoF inspection viewpoints into a time-optimal, collision-free route for a 9-DoF robotic system. Unlike modular pipelines that fix a single inverse-kinematics (IK) configuration per viewpoint, build an all-pairs travel-time map, and then route, our method jointly optimizes the visiting order and the per-viewpoint configuration in a single global search. The three-dimensional self-motion manifold of each viewpoint is parameterized in closed form so that the pose constraint holds by construction, the rest-to-rest travel time is approximated by a closed-form admissible double-integrator surrogate, and the tour is encoded by random keys. A derivative-free optimizer (CMA-ES) minimizes a cheap penalized objective over order and configuration, after which direct-collocation trajectory optimization is applied only to the edges of the selected route to certify dynamic feasibility and torque limits, and to return exact timings. This reduces the trajectory solves from quadratic to linear in the number of viewpoints and removes the decoupling that prevents modular pipelines from being globally time-optimal. Simulations and real-robot experiments on a KUKA LBR iiwa with a 2-DoF linear stage validate feasibility, smooth execution, and reduced end-to-end inspection time relative to modular and naive distance-based baselines.
Chinese Translation
我们提出了一个统一框架，将一组6自由度（6-DoF）检测视角转化为9自由度（9-DoF）机器人系统的时间最优、无碰撞路径。与固定每个视角单一逆运动学（IK）配置的模块化管道不同，我们的方法在单一全局搜索中联合优化访问顺序和每个视角的配置。每个视角的三维自运动流形以封闭形式参数化，从而确保姿态约束在构造时成立，静止到静止的旅行时间通过封闭形式的可接受双积分器替代物进行近似，巡回路径通过随机键进行编码。无导数优化器（CMA-ES）在顺序和配置上最小化一个廉价的惩罚目标，之后仅对所选路径的边缘应用直接配合轨迹优化，以验证动态可行性和扭矩限制，并返回精确的时间。这将轨迹求解的复杂度从与视角数量的平方成正比降低到线性，并消除了阻止模块化管道实现全局时间最优的解耦。对KUKA LBR iiwa机器人及其2自由度线性平台的仿真和实机器人实验验证了相对于模块化和简单距离基础线的可行性、平滑执行和减少的端到端检测时间。
cs.RO / 10 / 2606.20742
A Digital Twin Framework for Traffic-Aware UAV Pavement Monitoring without Lane Closure
一种无需封闭车道的交通感知无人机路面监测数字双胞胎框架
Abstract
UAV-based pavement inspection can reduce the cost and risk of road-surface monitoring, but real-world deployment remains difficult when traffic, pedestrians, and temporary occlusions affect the visibility of defects. This paper presents a Unity-based digital twin framework for traffic-aware UAV pavement monitoring without lane closure. The proposed environment integrates procedurally generated road defects, dynamic vehicles and pedestrians, autonomous UAV navigation, and an embedded road-damage perception pipeline. The perception module uses a two-stage approach: a lightweight YOLOv8n detector first localises road defects, pedestrians, and vehicles, while a second classifier distinguishes among potholes, single cracks, and crocodile cracks. On the simulator test set, the full pipeline achieved 99.26% overall accuracy across five classes. The digital twin was then used to evaluate three recovery strategies for occluded road segments: hover-and-recheck, micro-repositioning, and skip-and-revisit. Experiments were conducted across different traffic densities and flight altitudes using coverage, mission time, energy consumption, and revisit ratio as operational metrics. Results show that flight altitude has a strong influence on inspection coverage and that adaptive recovery improves performance under occlusion. In particular, hover-and-recheck achieved the most consistent coverage under medium and high traffic conditions, reaching up to 97.03% coverage, while skip-and-revisit was most effective in low-traffic scenarios, reaching 97.95\% coverage at medium altitude. These results demonstrate that digital twins can support the development and evaluation of traffic-aware UAV inspection strategies before real-world deployment.
Chinese Translation
基于无人机的路面检查可以降低道路表面监测的成本和风险，但在交通、行人和临时遮挡影响缺陷可见性的情况下，实际部署仍然困难。本文提出了一种基于Unity的数字双胞胎框架，用于无需封闭车道的交通感知无人机路面监测。所提出的环境集成了程序生成的道路缺陷、动态车辆和行人、自动无人机导航以及嵌入式道路损伤感知管道。感知模块采用两阶段方法：轻量级YOLOv8n检测器首先定位道路缺陷、行人和车辆，而第二个分类器则区分坑洞、单一裂缝和鳄鱼裂缝。在模拟器测试集上，完整管道在五个类别中实现了99.26%的总体准确率。随后，数字双胞胎用于评估三种遮挡道路段的恢复策略：悬停重检、微调位置和跳过重访。实验在不同交通密度和飞行高度下进行，使用覆盖率、任务时间、能耗和重访比例作为操作指标。结果表明，飞行高度对检查覆盖率有显著影响，且自适应恢复在遮挡情况下提高了性能。特别是，在中高交通条件下，悬停重检实现了最一致的覆盖率，达到97.03%的覆盖率，而在低交通场景中，跳过重访在中等高度下最为有效，达到97.95%的覆盖率。这些结果表明，数字双胞胎可以支持交通感知无人机检查策略的开发和评估，为实际部署做好准备。
cs.RO / 11 / 2606.20748
Toward Machine Risk Perception: Integrating Trust Calibration and Precursor-Based Risk Estimation for Humanoid
迈向机器风险感知：整合信任校准与基于前兆的风险评估用于类人机器人
Abstract
Humanoid robots are emerging as co-workers in smart manufacturing, yet their dynamic, human-like movements introduce safety risks that differ fundamentally from those of fixed or wheeled robots. Conventional safety paradigms based on reactive force or distance limits fail to capture the sequential, uncertain nature of humanoid failures. This study proposes a precursor-driven, trust-calibrated framework to enable proactive humanoid risk perception. Accident evolution is modeled through sequential precursor cues using a Logistic-Exponential (LE) formulation that couples logistic escalation from diverse precursors with exponential decay for temporal dissipation. Trust is defined as the inverse of the estimated accident probability, allowing humanoids to adapt behavior in real time, reducing aggressiveness when risk intensifies, and restoring confidence as stability returns. A multi-source dataset of 126 documented events and 241 precursors revealed twelve dominant accident modes, most evolving through overlapping cues within one second. A simulated case study ("fall-onto-human") demonstrated how the LE-Trust coupling can trigger early intervention and prevent collapse. The results advance humanoid safety from static thresholds toward dynamic, evidence-based inference, establishing a foundation for risk-aware and trustworthy human-robot collaboration in Industry 5.0 environments.
Chinese Translation
类人机器人作为智能制造中的协作伙伴正在崭露头角，但其动态的人类般运动带来了与固定或轮式机器人根本不同的安全风险。基于反应力或距离限制的传统安全范式未能捕捉类人机器人故障的序列性和不确定性。本研究提出了一种基于前兆驱动和信任校准的框架，以实现主动的类人风险感知。事故演变通过序列前兆线索建模，采用Logistic-Exponential (LE) 公式，将来自不同前兆的逻辑升级与时间衰减的指数衰减相结合。信任被定义为估计事故概率的倒数，使类人机器人能够实时调整行为，在风险加剧时降低攻击性，并在稳定性恢复时恢复信心。126个记录事件和241个前兆的多源数据集揭示了十二种主要的事故模式，其中大多数在一秒内通过重叠线索演变。一个模拟案例研究（“跌倒-碰撞人类”）展示了LE-Trust耦合如何触发早期干预并防止崩溃。研究结果将类人机器人的安全性从静态阈值推进到动态、基于证据的推理，为在工业5.0环境中实现风险意识和可信赖的人机协作奠定了基础。
cs.RO / 12 / 2606.20754
Perturbation-Based Uncertainty for Failure Detection in Vision-Language-Action Models
基于扰动的不确定性在视觉-语言-动作模型中的故障检测
Abstract
Vision-Language-Action (VLA) models have shown strong performance in robotic manipulation, but reliable uncertainty quantification remains challenging, particularly under distribution shift. Unlike autoregressive policies, many modern VLA models generate continuous actions through regression or flow-based generation, where explicit predictive probabilities are unavailable. Moreover, existing approaches often rely on stochastic action sampling or supervised failure labels, limiting their applicability across diverse pretrained VLA models. In this work, we propose a label-free and model-agnostic framework for inference-time uncertainty estimation through hidden activation perturbations, motivated by Bayesian perspectives on local model variations. Specifically, we inject Gaussian perturbations into transformer hidden activations and estimate epistemic signals from disagreement across perturbed action predictions. Experiments on LIBERO and LIBERO-PRO show that perturbation-based uncertainty consistently improves failure detection under distribution shift compared to sampling-based uncertainty, providing a practical uncertainty signal for VLA models.
Chinese Translation
视觉-语言-动作（VLA）模型在机器人操作中表现出色，但在分布转移下可靠的不确定性量化仍然具有挑战性。与自回归策略不同，许多现代VLA模型通过回归或基于流的生成生成连续动作，而缺乏明确的预测概率。此外，现有方法通常依赖于随机动作采样或监督故障标签，这限制了它们在多样化的预训练VLA模型中的适用性。在本研究中，我们提出了一种无标签且模型无关的推理时不确定性估计框架，通过隐藏激活扰动来实现，这一思路受到对局部模型变异的贝叶斯视角的启发。具体而言，我们将高斯扰动注入到变换器的隐藏激活中，并通过对扰动后动作预测的不一致性来估计认知信号。在LIBERO和LIBERO-PRO上的实验表明，与基于采样的不确定性相比，基于扰动的不确定性在分布转移下持续改善了故障检测，为VLA模型提供了实用的不确定性信号。
cs.RO / 13 / 2606.20755
UNSEEN: Uncertainty-aware Navigation via Sparse Estimation in Unknown Environments
UNSEEN：基于稀疏估计的不确定性感知导航在未知环境中的应用
Abstract
Visual navigation in unknown environments remains a core challenge in mobile robotics, especially for resource-constrained platforms. Most existing approaches rely on loosely coupled modular pipelines and strong assumptions on perception quality or environmental structure, often resorting to multi-modal sensor suites that increase system complexity and deployment cost. Vision-only navigation offers a lightweight alternative, but its performance degrades severely under motion blur, low texture, and illumination changes, largely because they neglect the tight coupling between commanded motion and perception. While perception-aware methods partially address this issue, they typically optimize individual modules and fail to propagate uncertainty consistently across the navigation stack. In this paper, we present UNSEEN, a unified uncertainty- and perception-aware navigation framework that explicitly couples localization, mapping, and planning using only a front-mounted camera. UNSEEN estimates sparse maps and robot poses with associated uncertainties at 6Hz, and leverages them to plan trajectories that jointly optimize task progress and estimation accuracy in receding-horizon. Simulations and extensive real-world experiments in unknown environments demonstrate the robustness of the proposed approach, with UNSEEN-SLAM reducing absolute translational error by 9.8% and UNSEEN-Plan improving estimation accuracy by up to 45% compared to state-of-the-art methods, while achieving a 100% task success rate.
Chinese Translation
在未知环境中进行视觉导航仍然是移动机器人领域的一项核心挑战，特别是对于资源受限的平台。现有的大多数方法依赖于松散耦合的模块化管道，并对感知质量或环境结构做出强假设，通常需要多模态传感器组合，这增加了系统的复杂性和部署成本。仅使用视觉的导航提供了一种轻量级的替代方案，但在运动模糊、低纹理和光照变化下，其性能严重下降，这主要是因为它们忽视了指令运动与感知之间的紧密耦合。尽管感知感知方法部分解决了这一问题，但它们通常优化单个模块，未能在导航堆栈中一致地传播不确定性。在本文中，我们提出了UNSEEN，一个统一的不确定性和感知感知导航框架，明确耦合定位、地图构建和规划，仅使用前置摄像头。UNSEEN以6Hz的频率估计稀疏地图和机器人姿态及其相关不确定性，并利用这些信息规划轨迹，联合优化任务进展和估计精度。模拟和在未知环境中的大量实地实验展示了所提方法的鲁棒性，UNSEEN-SLAM将绝对平移误差降低了9.8%，UNSEEN-Plan的估计精度提高了最高45%，同时实现了100%的任务成功率。
cs.RO / 14 / 2606.20772
Mind the Privileged-to-Camera Gap: Actor-Centric Sidecar Supervision for Camera-First Open-Loop Waypoint Prediction
注意特权到相机的差距：以演员为中心的副车监督用于相机优先的开放环路航点预测
Abstract
Camera-first autonomous-driving models predict future ego waypoints from images, ego-state features, and route commands, but waypoint supervision alone does not explicitly supervise actor-level representations of nearby road users. We study this as supervised representation learning for open-loop waypoint prediction. The deployable model uses multi-view RGB, ego state, and route command at inference. During training, simulator-derived sidecar labels supervise actor grounding, privileged hindsight actor relevance relative to the logged ego trajectory, and selected-actor short-horizon motion; these labels are never inference inputs. We evaluate route-disjoint splits with matched architecture, optimizer, validation criterion, checkpoint selection, and three seeds. A plain waypoint-only RGB baseline obtains 1.815$\pm$0.02 m final displacement error (FDE), and the matched no-teacher non-sidecar RGB control obtains 1.716$\pm$0.02 m. Road-user sidecar supervision (RU-sidecar) reduces FDE to 1.223$\pm$0.01 m, a 32.6% reduction over the plain baseline and 28.7% over the matched no-teacher non-sidecar RGB control. It improves over the plain baseline on 1445/1494 routes and over the matched no-teacher non-sidecar RGB control on 1417/1494 routes. Actor-conditioned slices show gains in all nonempty subsets, including 29.1% reduction for samples with at least four valid sidecar actors and 30.0% when a vulnerable road user is present. Optional simulator-state teacher alignment reaches 1.186$\pm$0.15 m FDE, but higher seed variability makes it secondary. Non-deployable simulator-state diagnostics remain stronger, indicating a privileged-to-camera gap. The evidence is limited to open-loop simulation diagnostics.
Chinese Translation
相机优先的自动驾驶模型通过图像、自动驾驶状态特征和路线指令预测未来的自我航点，但仅依赖航点监督并未明确监督附近道路用户的演员级表示。我们将其研究为开放环路航点预测的监督表示学习。可部署模型在推理时使用多视角RGB、自动驾驶状态和路线指令。在训练过程中，模拟器生成的副车标签监督演员定位、相对于记录的自我轨迹的特权回顾演员相关性以及选定演员的短期运动；这些标签从未作为推理输入。我们评估了具有匹配架构、优化器、验证标准、检查点选择和三个随机种子的路线不重叠拆分。一个仅使用航点的RGB基线模型获得了1.815±0.02米的最终位移误差（FDE），而匹配的无教师非副车RGB控制模型获得了1.716±0.02米。道路用户副车监督（RU-sidecar）将FDE降低至1.223±0.01米，相较于普通基线减少了32.6%，相较于匹配的无教师非副车RGB控制减少了28.7%。在1494条路线中，该方法在1445条上优于普通基线，在1417条上优于匹配的无教师非副车RGB控制。针对演员条件切片的结果显示，在所有非空子集中均有提升，其中对于至少有四个有效副车演员的样本减少了29.1%，而当存在脆弱道路用户时减少了30.0%。可选的模拟器状态教师对齐达到了1.186±0.15米的FDE，但更高的随机种子变异性使其成为次要因素。不可部署的模拟器状态诊断结果仍然更强，表明存在特权到相机的差距。证据仅限于开放环路模拟诊断。
cs.RO / 15 / 2606.20781
World Action Models: A Survey
世界行动模型：综述
Abstract
World Action Models (WAMs) are embodied predictive-action models that make a forecast of the future available to action. Recent WAMs repurpose large video generation models, and a parallel line relies on language or vision-language backbones without a video-generation core. This rapid expansion has blurred the boundary among broad world models, video generation models, action-grounded video world models, Vision-Language-Action policies, and WAMs. This survey gives the field a common account. It first clarifies these boundaries, then organizes existing works through two complementary views. The first view asks what each method is required to generate, spanning rendered futures, latent futures, and video-generation-free action reasoning. The second view decomposes each method by predictive substrate, backbone, action coupling, and deployment regime. This anatomy supports a unified discussion of interactability, causality, persistence, physical plausibility, and generalization, followed by data, evaluation, and open challenges. Across these axes, a consistent design pattern emerges: WAMs are not simply video generators with action heads, but predictive-action methods whose design choices trade representational richness against compute, memory, latency, and action-label cost. The field is moving toward methods that generate less of the future while preserving what control requires. The survey homepage is available at https://world-action-models.github.io/.
Chinese Translation
世界行动模型（World Action Models, WAMs）是具身的预测行动模型，能够为行动提供未来的预测。近期的WAMs重新利用了大型视频生成模型，而另一条平行的研究路线则依赖于语言或视觉-语言骨干网络，而不依赖视频生成核心。这一快速扩展模糊了广义世界模型、视频生成模型、基于行动的视频世界模型、视觉-语言-行动策略以及WAMs之间的界限。本综述为该领域提供了一个共同的框架。首先，它澄清了这些边界，然后通过两种互补的视角组织现有的研究工作。第一种视角关注每种方法需要生成的内容，包括渲染的未来、潜在的未来和无视频生成的行动推理。第二种视角则通过预测基础、骨干网络、行动耦合和部署机制对每种方法进行解构。这种解剖支持对可交互性、因果性、持久性、物理合理性和泛化能力的统一讨论，随后讨论数据、评估和开放挑战。在这些维度中，出现了一种一致的设计模式：WAMs并不仅仅是带有行动头的视频生成器，而是预测行动方法，其设计选择在表示丰富性与计算、内存、延迟和行动标签成本之间进行权衡。该领域正朝着生成较少未来的方向发展，同时保留控制所需的内容。综述主页可访问 https://world-action-models.github.io/。
cs.RO / 16 / 2606.20871
Geometric Entropy: When Trajectory Diversity Helps and Hurts in Imitation Learning
几何熵：轨迹多样性在模仿学习中的利与弊
Abstract
We study how trajectory-shape diversity in demonstrations affects imitation learning (IL) performance across models, tasks, and data scales. We introduce Geometric Entropy (H_G), a task-agnostic metric that quantifies the intrinsic diversity of transit trajectories after normalizing away extrinsic variation, such as goal pose and workspace scale, via target-frame alignment. Across multiple IL architectures and both simulated and real-robot contact-rich manipulation tasks, we observe a consistent inverted-U relationship between success and H_G: increasing geometric diversity improves robustness in low-diversity regimes but degrades performance once diversity induces strategy ambiguity. Moreover, the optimal entropy shifts toward lower values as task mastery increases through more data, easier tasks, or stronger priors, and for a pretrained vision-language-action model the trend becomes effectively monotonic decreasing. Practically, H_G enables fast pre-training auditing of demonstration datasets and offers a simple guideline for calibrating demonstrations toward the learnable regime.
Chinese Translation
我们研究了演示中的轨迹形状多样性如何影响不同模型、任务和数据规模下的模仿学习（IL）性能。我们引入了几何熵（H_G），这是一种与任务无关的度量，量化了在通过目标框架对齐消除外部变化（如目标姿态和工作空间规模）后，过渡轨迹的内在多样性。在多种IL架构以及模拟和真实机器人接触丰富的操作任务中，我们观察到成功与H_G之间存在一致的倒U型关系：在低多样性状态下，增加几何多样性提高了鲁棒性，但一旦多样性导致策略模糊，性能就会下降。此外，随着通过更多数据、更简单的任务或更强的先验知识提高任务掌握度，最佳熵值向较低值移动，而对于一个预训练的视觉-语言-动作模型，这一趋势变得有效单调递减。从实践角度来看，H_G能够快速审计演示数据集的预训练，并为调整演示以适应可学习的范围提供简单的指导。
cs.RO / 17 / 2606.20905
Vesta: A Generalist Embodied Reasoning Model
Vesta：一个通用的具身推理模型
Bjorck, Johan, Li, Zhiqi, Man, Yunze, Wang, Jing, Cheng, An-Chieh, Liu, Sifei, Wang, Shihao, Yu, Zhiding, Badki, Abhishek, Birchfield, Stan, Blukis, Valts, Chebotar, Yevgen, Chen, Siyi, Leng, Sicong, Chou, Yu-Cheng, Ding, Tianli, Li, Boyi, Luo, Zhengyi, Su, Hang, Tremblay, Jonathan, Wang, Tingwu, Wen, Bowen, Wu, Jimmy, Xie, Xianghui, Ye, Hanrong, Yin, Hongxu, Zentner, K. R., Gui, Liangyan, Wang, Yu-Xiong, Zhu, Yuke, Fan, Linxi "Jim", Kautz, Jan
Abstract
Robots operating in open-world environments must seamlessly integrate localization, spatial reasoning, navigation, and long-horizon planning. While specialist models excel at individual tasks, deploying a multi-model stack is computationally expensive and prone to cascading errors. We present Vesta, a unified embodied generalist that consolidates these capabilities into a single foundation model. Our approach combines a diverse and massive curated corpus designed to induce spatial grounding and a simple multimodal memory harness that enables reasoning over extended time horizons. Across diverse benchmarks, Vesta on average beats individual SOTA baselines by >$20\%$ and beats an ensemble of per-category-best baselines by $>10\%$ -- thus demonstrating that a generalist model can match or exceed specialists. On real-world robotic tasks requiring memory and reasoning, Vesta improves task success by >35\%. Our work thus demonstrates that a single generalist is a feasible, scalable, and arguably preferable alternative to combining specialists.
Chinese Translation
在开放世界环境中操作的机器人必须无缝整合定位、空间推理、导航和长时间规划。尽管专门模型在各自的任务中表现出色，但部署多模型堆栈在计算上成本高昂且容易导致级联错误。我们提出了Vesta，一个统一的具身通用模型，将这些能力整合为一个单一的基础模型。我们的方法结合了一个多样化且庞大的策划语料库，旨在引导空间基础，并利用一个简单的多模态记忆系统，使得能够在较长时间范围内进行推理。在各种基准测试中，Vesta的平均表现超过了单个最先进(SOTA)基线超过20%，并且超过了每个类别最佳基线的集成模型超过10%——这表明通用模型可以与专门模型相匹配或超越。在需要记忆和推理的真实世界机器人任务中，Vesta的任务成功率提高了超过35%。因此，我们的工作表明，单一的通用模型是一个可行的、可扩展的，并且可以说是优于结合专门模型的替代方案。
cs.RO / 18 / 2606.20958
Learning-Based Modeling of Soft Robots via Cosserat Rod Theory
基于学习的软机器人建模：科塞拉特杆理论
Abstract
Modeling soft robot dynamics is challenging due to their continuum structure and typically nonlinear dynamics. Creating models based on first-order principles is typically time-demanding, and their expressiveness is limited, whereas data-driven models lack interpretability and physical consistency. This work aims to overcome these challenges by introducing a port-Hamiltonian Gaussian Process Regression framework for learning and simulating the dynamics of planar, rod-like soft robots. In detail, the proposed model integrates Cosserat rod theory and Hamiltonian physics with data-driven inference to preserve the system's energy structure while accurately learning the rod dynamics. Numerical simulations show that we can achieve accurate and energy-consistent representations of a rod-like soft robot, showing the potential for a robust and interpretable pathway for modeling complex continuum mechanics.
Chinese Translation
软机器人动力学建模具有挑战性，因为它们的连续结构和通常的非线性动力学。基于第一原理创建模型通常需要耗费大量时间，并且其表达能力有限，而数据驱动模型则缺乏可解释性和物理一致性。本研究旨在通过引入一个端口哈密顿高斯过程回归框架，来学习和模拟平面杆状软机器人的动力学，从而克服这些挑战。具体而言，所提出的模型将科塞拉特杆理论和哈密顿物理与数据驱动推断相结合，以保持系统的能量结构，同时准确学习杆的动力学。数值模拟表明，我们能够实现对杆状软机器人的准确且能量一致的表征，展示了建模复杂连续介质力学的稳健且可解释的潜力。
cs.RO / 19 / 2606.20962
Heterogeneous Policy Networks for Composite Robot Team Communication and Coordination
复合机器人团队通信与协调的异构策略网络
Abstract
High-performing human-human teams learn intelligent and efficient communication and coordination strategies to maximize their joint utility. These teams implicitly understand the different roles of heterogeneous team members and adapt their communication protocols accordingly. Multi-Agent Reinforcement Learning (MARL) has attempted to develop computational methods for synthesizing such joint coordination-communication strategies, but emulating heterogeneous communication patterns across agents with different state, action, and observation spaces has remained a challenge. Without properly modeling agent heterogeneity, as in prior MARL work that leverages homogeneous graph networks, communication becomes less helpful and can even deteriorate the team's performance. In the past, we proposed Heterogeneous Policy Networks (HetNet) to learn efficient and diverse communication models for coordinating cooperative heterogeneous teams. In this extended work, we extend Heterogeneous Policy Networks (HetNet) to support scaling heterogeneous robot teams. Building on heterogeneous graph-attention networks, we show that HetNet not only facilitates learning heterogeneous collaborative policies but also enables end-to-end training for learning highly efficient binarized messaging. Our empirical evaluation shows that HetNet sets a new state of the art in learning coordination and communication strategies for heterogeneous multi-agent teams by achieving an 5.84% to 707.65% performance improvement over the next-best baseline across multiple domains while simultaneously achieving a 200x reduction in the required communication bandwidth.
Chinese Translation
高效的人际团队学习智能且高效的通信与协调策略，以最大化其联合效用。这些团队隐含地理解异构团队成员的不同角色，并相应地调整其通信协议。多智能体强化学习（Multi-Agent Reinforcement Learning, MARL）试图开发计算方法，以合成这种联合协调-通信策略，但在不同状态、动作和观察空间的智能体之间模拟异构通信模式仍然是一项挑战。如果没有正确建模智能体的异构性，如先前利用同质图网络的MARL工作，通信将变得不那么有帮助，甚至可能恶化团队的表现。在过去，我们提出了异构策略网络（Heterogeneous Policy Networks, HetNet），以学习高效且多样的通信模型来协调合作的异构团队。在这项扩展工作中，我们扩展了异构策略网络（HetNet），以支持异构机器人团队的规模化。在异构图注意力网络的基础上，我们展示了HetNet不仅促进了异构协作策略的学习，还实现了端到端训练，以学习高效的二进制消息传递。我们的实证评估表明，HetNet在学习异构多智能体团队的协调与通信策略方面设定了新的最优状态，在多个领域中实现了5.84%至707.65%的性能提升，同时通信带宽需求减少了200倍。
cs.RO / 20 / 2606.20990
Duet: Dual-Robot Understanding via Efficient Teaching
二重奏：通过高效教学实现双机器人理解
Abstract
Dual-robot collaboration enables tasks that exceed the reach and payload of a single robot, such as collaboratively transporting objects across environments and executing coordinated handovers. Data acquisition is the primary bottleneck for training these systems. To this end, we introduce DUET, a dual-robot learning framework for mobile manipulation. For efficient data collection, we create a unified dual-embodiment synchronized VR-based teleoperation system for in-domain heterogeneous robot data collection. We further develop a complementary tracking pipeline that records human-human coordination and collaborative mobile manipulation priors. To allow efficient learning, we introduce an Action Chunking Transformer based architecture that first pretrains collaborative policies on efficient human-human demonstrations, before finetuning them on a minimal set of real-robot teleoperation trajectories. We develop a benchmark of four collaborative tasks to evaluate our framework using a Unitree G1 humanoid and a Dexmate Vega1 mobile manipulator. The results demonstrate that harnessing human priors not only yields superior task performance compared to baselines trained only on robot data, but also reduces the total human effort required for data collection. Our human data collection pipeline achieves 5.4x acceleration on average from teleoperation, but we perform equally or better than robot-only data trained policies across all tasks. Our project page is available at https://zhaoy37.github.io/Duet/.
Chinese Translation
双机器人协作能够完成超出单个机器人能力范围的任务，例如在不同环境中协同运输物体和执行协调交接。数据采集是训练这些系统的主要瓶颈。为此，我们提出了DUET，一个用于移动操控的双机器人学习框架。为了高效的数据收集，我们创建了一个统一的双重体现同步虚拟现实（VR）远程操作系统，以便进行领域内异构机器人数据的收集。我们进一步开发了一个互补的跟踪管道，用于记录人际协调和协同移动操控的先验知识。为了实现高效学习，我们引入了一种基于动作分块变换器（Action Chunking Transformer）的架构，该架构首先在高效的人际示范上预训练协作策略，然后在一组最小的真实机器人远程操作轨迹上进行微调。我们开发了四个协作任务的基准，以评估我们的框架，使用Unitree G1人形机器人和Dexmate Vega1移动操控器。结果表明，利用人类先验不仅在任务表现上优于仅基于机器人数据训练的基线，而且减少了数据收集所需的人力总投入。我们的人类数据收集管道在远程操作中平均实现了5.4倍的加速，但在所有任务中，我们的表现与仅基于机器人数据训练的策略相当或更好。我们的项目页面可访问：https://zhaoy37.github.io/Duet/。
cs.RO / 21 / 2606.20999
Inductive Generalization for Robotic Manipulation
机器人操作的归纳泛化
Abstract
Understanding the generalization capabilities of visuomotor policies is essential in the development of capable robotic agents. Generalizable models learn structures that transfer across domains. However, in practice, visuomotor policies test performance by interpolation on known distributions using unstructured domain shifts (e.g. lighting, clutter, diverse objects). We argue that to measure generalization capabilities we must instead test the inductive capacity of policies on progressively harder, out-of-distribution task variants. We call this inductive generalization, drawing directly on how axis-based evaluation has revealed inherent generalization limitations in language models (e.g. sequence length, counting) arXiv:2502.00197 . We provide a reusable and formal evaluation protocol for measuring inductive generalization in any manipulation policy, and establish baselines showing that existing paradigms fail this test; e.g. SoTA Vision-Language-Action models and find that policies that appear to generalize to prior domain shifts (distractors, etc) fail inductive generalization tests. These results expose a class of learning challenges orthogonal to those addressed by data and model scaling in robot learning, yet are imperative to solve in order to realize general purpose robots.
Chinese Translation
理解视觉运动策略的泛化能力对于开发高效的机器人代理至关重要。可泛化模型学习能够跨领域转移的结构。然而，在实践中，视觉运动策略通过在已知分布上进行插值来测试性能，使用非结构化的领域转移（例如，光照、杂乱、不同对象）。我们认为，为了衡量泛化能力，我们必须在逐渐更难的、超出分布的任务变体上测试策略的归纳能力。我们称之为归纳泛化，这直接借鉴了基于轴的评估如何揭示语言模型（例如，序列长度、计数）固有的泛化限制。我们提供了一种可重用的正式评估协议，用于测量任何操作策略的归纳泛化，并建立基准，表明现有范式未能通过此测试；例如，最先进的视觉-语言-动作模型，并发现那些看似能够泛化到先前领域转移（干扰物等）的策略未能通过归纳泛化测试。这些结果揭示了一类与机器人学习中数据和模型扩展所解决的问题正交的学习挑战，但解决这些问题对于实现通用机器人至关重要。
cs.RO / 22 / 2606.21011
R2HandoverSim: A Simulation Framework and Benchmark for Robot-to-Human Object Handovers
R2HandoverSim：一个机器人与人类物体交接的仿真框架与基准测试
Abstract
We present R2HandoverSim, a simulation benchmark for robot-to-human (R2H) object handovers. Although R2H handover methods have advanced rapidly, the lack of standardized evaluation protocols impedes objective comparison. Our benchmark enables reproducible evaluation by systematically comparing four baselines on their predicted shared grasp poses. We conduct a user study with 30 participants, analyze baseline performance, and show that simulation results correlate with real-world evaluation outcomes. Crucially, five complementary metrics (planning feasibility, reachability, grasp stability, grasp affordance, and safety) better reflect user-perceived handover quality than overall success rate alone. Website and code: https://robot-future.github.io/r2handoversim/.
Chinese Translation
我们提出了R2HandoverSim，这是一个用于机器人与人类（R2H）物体交接的仿真基准测试。尽管R2H交接方法发展迅速，但缺乏标准化的评估协议妨碍了客观比较。我们的基准测试通过系统地比较四个基线在其预测的共享抓取姿态上的表现，实现了可重复的评估。我们进行了一个包含30名参与者的用户研究，分析了基线性能，并展示了仿真结果与现实世界评估结果之间的相关性。重要的是，五个互补指标（规划可行性、可达性、抓取稳定性、抓取适应性和安全性）比单一的整体成功率更能反映用户感知的交接质量。网站和代码： https://robot-future.github.io/r2handoversim/。
cs.RO / 23 / 2606.21014
BayesFP: Posterior Estimation for Flow-Based Policies via Feynman-Kac Sampling
BayesFP：通过费曼-卡克采样进行基于流的策略的后验估计
Abstract
Robots must generate trajectories that remain faithful to learned expert behavior while satisfying safety constraints and task-specific objectives specified only at inference time. We formulate constrained trajectory generation for pretrained diffusion and flow-matching policies as Bayesian posterior sampling, with the learned demonstration distribution as a prior and an inference-time, cost-derived likelihood tilting it toward feasible, optimal trajectories. To sample from this posterior without any retraining of the base policy, we leverage the Feynman--Kac corrector framework, originally formulated for diffusion models, and extend it to deterministic flow-matching policies. The result is a unified, inference-time, retraining-free sampler for diffusion and flow policies. We validate the approach on pretrained Diffusion Policy, GR00T-N1.6, and $\pi_{0.5}$ checkpoints across simulated and real-world manipulation tasks, including planning around non-convex obstacles introduced at inference time, and show improvements over the base $\pi_{0.5}$ on zero-shot tasks.
Chinese Translation
机器人必须生成忠实于学习到的专家行为的轨迹，同时满足仅在推理时指定的安全约束和任务特定目标。我们将预训练的扩散和流匹配策略的受限轨迹生成形式化为贝叶斯后验采样，以学习到的示范分布作为先验，并在推理时通过成本导出的似然倾斜到可行的最优轨迹。为了在不对基础策略进行任何再训练的情况下从该后验中采样，我们利用了最初为扩散模型制定的费曼-卡克修正器框架，并将其扩展到确定性流匹配策略。结果是一个统一的、推理时的、无再训练的扩散和流策略采样器。我们在预训练的扩散策略GR00T-N1.6和$ ext{π}_{0.5}$检查点上验证了该方法，涵盖了模拟和真实世界的操作任务，包括在推理时围绕非凸障碍物进行规划，并在零-shot任务上显示出相较于基础$ ext{π}_{0.5}$的改进。
cs.RO / 24 / 2606.21047
Membrane-based Acoustic Microrobots
基于膜的声学微型机器人
Abstract
Acoustic microrobots have emerged as a promising frontier for targeted drug delivery and minimally invasive medicine due to their high-power density and biocompatibility. Despite wide-ranging designs, conventional acoustic microrobots mostly rely on air microbubbles trapped within confined microcavities within the robot body, which suffer from limited operational longevity due to rapid gas dissolution and resultant shifts in resonance frequency. In this paper, we propose a robust, membrane-based acoustic microrobot that overcomes these limitations by employing a thin flexible Polydimethylsiloxane (PDMS) membrane bonded over confined microcavities for microstreaming. The introduced design physically prevents gas diffusion, ensuring stable performance over extended periods at high actuation voltages. We systematically characterized the membrane-based acoustic actuator longevity, demonstrating consistent streaming and propulsion for over 24 hours of continuous operation. In addition, by embedding magnetic microparticles into the structural body, these actuators were successfully employed as microswimmers with directional control using low-intensity (2 mT) external magnetic fields. Finally, we demonstrate the scalability of the proposed design architecture down to ~100 um. This membrane-based approach establishes a reliable framework for the development of high-endurance acoustic microactuators and microrobots capable of performing long-term tasks.
Chinese Translation
声学微型机器人因其高功率密度和生物相容性而成为靶向药物输送和微创医学的有前景的前沿技术。尽管设计多样，传统的声学微型机器人主要依赖于困在机器人主体内的微腔中的空气微气泡，这导致其操作寿命有限，因为气体迅速溶解并导致共振频率的变化。本文提出了一种稳健的基于膜的声学微型机器人，通过采用粘合在微腔上的薄型柔性聚二甲基硅氧烷（PDMS）膜来克服这些限制，以实现微流动。所提出的设计在物理上防止气体扩散，确保在高驱动电压下长时间稳定运行。我们系统地表征了基于膜的声学驱动器的寿命，展示了在超过24小时的连续操作中保持一致的流动和推进。此外，通过将磁性微粒嵌入结构体，这些驱动器成功地作为微游动器在低强度（2 mT）外部磁场下实现了方向控制。最后，我们展示了所提设计架构的可扩展性，缩小至约100微米。该基于膜的方法为开发能够执行长期任务的高耐久性声学微驱动器和微型机器人建立了可靠的框架。
cs.RO / 25 / 2606.21086
ReFPO: Reflow Regularization for Flow Matching Policy Gradients
ReFPO：用于流匹配策略梯度的重流正则化
Abstract
We present Reflow-regularized Flow Matching Policy Gradients (ReFPO), a simple online RL method that adds explicit Reflow regularization to FPO for efficient flow-based control. We uncover a key structural property: the gradient updates in Flow Matching Policy Gradients (FPO) can be interpreted as an implicit advantage-weighted Reflow process, providing a new geometric perspective on flow-based policy gradients. Building on this insight, ReFPO introduces an explicit geometric regularizer that can be implemented with a single line of code change without incurring additional computational overhead or auxiliary distillation stages. By synergizing advantage-guided updates with path rectification, our method reduces CFM proxy-ratio spikes, stabilizes PPO-style training, and enables high-fidelity one-step inference that often matches or exceeds multi-step performance. We experimentally demonstrate that ReFPO improves average performance and discretization robustness across GridWorld, MuJoCo Playground, and high-dimensional Humanoid Control tasks, providing a scalable and stable approach for generative policies in complex physical simulations.
Chinese Translation
我们提出了重流正则化的流匹配策略梯度（ReFPO），这是一种简单的在线强化学习方法，通过向FPO添加显式的重流正则化，实现高效的基于流的控制。我们发现了一个关键的结构特性：流匹配策略梯度（FPO）中的梯度更新可以被解释为一种隐式的优势加权重流过程，为基于流的策略梯度提供了新的几何视角。在此基础上，ReFPO引入了一种显式的几何正则化器，可以通过一行代码的修改实现，而不会增加额外的计算开销或辅助蒸馏阶段。通过将优势引导的更新与路径修正相结合，我们的方法减少了CFM代理比率的峰值，稳定了PPO风格的训练，并实现了高保真的一步推断，通常能够匹配或超过多步性能。我们通过实验展示了ReFPO在GridWorld、MuJoCo Playground和高维人形控制任务中的平均性能和离散化鲁棒性得到了改善，为复杂物理仿真中的生成策略提供了一种可扩展且稳定的方法。
cs.RO / 26 / 2606.21088
MV-WAM: Manifold-Aware World Action Model with Value Augmentation
MV-WAM：具有价值增强的流形感知世界行动模型
Abstract
Achieving robust and generalizable manipulation across diverse environments remains a fundamental challenge in embodied robotics. Recent world action models achieve strong in-domain performance, yet their gains do not extend proportionally to out-of-distribution scenarios. We attribute this to a structural mismatch between visual and action modalities, whose intrinsically heterogeneous manifolds cause joint optimization to disproportionately degrade action robustness under distribution shift. To address this, we propose MV-WAM, a novel end-to-end framework that jointly models visual prediction, action generation, and value estimation designed to effectively leverage video priors during both training and inference for enhanced action generalization. Key to this unification is a cross-modality causal mask that hierarchically grounds actions in predicted video frames and value function tokens in both modalities. To further narrow the generalization gap, MV-WAM adopts a manifold-aware optimization scheme that explicitly accounts for the structural heterogeneity across modalities. Finally, MV-WAM introduces a progress-value regulation mechanism that estimates task completion and detects misalignment between predicted frames and generated actions, enabling the policy to autonomously identify execution deviations and recover through value-guided rollback. On the RoboTwin simulation, MV-WAM achieves a 55.7% mean success rate on random scenarios without any randomized action supervision, outperforming the strongest baseline by 29.3%. MV-WAM achieves a 77.5% mean success rate across four real-world tasks of varying difficulty on a dual-arm robot. Our results demonstrate that manifold-aware cross-modal alignment is essential for robust policy generalization, offering a path toward deployable robotic manipulation.
Chinese Translation
在多样化环境中实现稳健且可泛化的操作仍然是具身机器人领域的一项基本挑战。近期的世界行动模型在领域内表现出色，但其性能提升并未在分布外场景中成比例扩展。我们将这一现象归因于视觉和行动模态之间的结构不匹配，这种内在异质的流形导致联合优化在分布转移下不成比例地降低了行动的稳健性。为了解决这一问题，我们提出了MV-WAM，一种新颖的端到端框架，联合建模视觉预测、行动生成和价值估计，旨在有效利用视频先验以增强行动的泛化能力，适用于训练和推理阶段。该统一的关键在于跨模态因果掩码，它将行动在预测的视频帧和两种模态中的价值函数标记进行分层关联。为了进一步缩小泛化差距，MV-WAM采用了一种流形感知优化方案，明确考虑了模态之间的结构异质性。最后，MV-WAM引入了一种进度-价值调节机制，估计任务完成情况并检测预测帧与生成行动之间的错位，使得策略能够自主识别执行偏差并通过价值引导的回滚进行恢复。在RoboTwin仿真中，MV-WAM在随机场景下实现了55.7%的平均成功率，且未使用任何随机化的行动监督，超越了最强基线29.3%。在双臂机器人上，MV-WAM在四个难度各异的真实任务中实现了77.5%的平均成功率。我们的结果表明，流形感知的跨模态对齐对于稳健的策略泛化至关重要，为可部署的机器人操作提供了一条路径。
cs.RO / 27 / 2606.21093
How Should a Robot Configure Its Laser Scanner for Inspection?
机器人应如何配置其激光扫描仪以进行检测？
Abstract
Robotic inspection relies on accurate sensing to acquire high-fidelity geometric measurements for defect detection and metrology. While prior work has focused on robot motion and viewpoint planning, how to configure sensing parameters remains largely underexplored, despite their decisive impact on measurement quality. We propose SenseHD, a robotic sensing system that formulates scanner configuration as an instruction-conditioned sensing decision. Instead of predicting precise parameter values, SenseHD treats sensing parameters as discrete sensing actions and selects stable sensing regimes through hyperdimensional associative memory. Experiments on a real robotic inspection platform demonstrate that SenseHD robustly selects appropriate sensing configurations and significantly improves inspection reliability, while remaining lightweight and efficient compared to baseline methods.
Chinese Translation
机器人检测依赖于准确的传感以获取高保真几何测量，以便进行缺陷检测和计量。尽管之前的研究集中在机器人运动和视点规划上，但如何配置传感参数仍然在很大程度上未被探讨，尽管这对测量质量有决定性的影响。我们提出了SenseHD，一种将扫描仪配置形式化为基于指令的传感决策的机器人传感系统。SenseHD并不预测精确的参数值，而是将传感参数视为离散的传感动作，并通过超维关联记忆选择稳定的传感模式。在真实的机器人检测平台上的实验表明，SenseHD能够稳健地选择合适的传感配置，并显著提高检测的可靠性，同时与基线方法相比，保持轻量和高效。
cs.RO / 28 / 2606.21100
Factor-Aware Mixture-of-Experts with Pretrained Encoder for Combinatorial Generalization
具有预训练编码器的因子感知专家混合模型用于组合泛化
Abstract
The integration of pretrained encoders with diffusion policies has become a dominant paradigm for visual robotic manipulation. However, it still struggles to generalize across complex environments with varying factors such as lighting and surface textures. To address this, we propose FAME, a framework that integrates a factor-aware mixture-of-experts (MoE) with a pretrained encoder to enhance generalization to environmental variations. FAME follows a three-stage training process: (1) policy warmup, where a diffusion policy is trained on standard-environment data with a frozen encoder; (2) factor-specific adapter training, where lightweight adapters inserted between the frozen encoder and the temporarily frozen policy are trained on customized datasets, each targeting a distinct environmental variation; and (3) joint fine-tuning, where a central router and the warmed policy are trained on mixed data to handle multiple factors jointly. FAME is ``factor-aware'' because the central router softly weights frozen factor-specific adapters as a dense MoE, enabling combinatorial generalization across multiple factors. Evaluations on the Meta-World benchmark show that FAME outperforms diffusion policy baselines by 34%. We further validate FAME in a real-world pick-and-place task using a compact model trained on newly collected data, where FAME achieves a 35% improvement in generalization under real-world variations.
Chinese Translation
将预训练编码器与扩散策略相结合已成为视觉机器人操作的主流范式。然而，它在复杂环境中仍然难以实现泛化，尤其是在光照和表面纹理等变化因素的影响下。为了解决这个问题，我们提出了 FAME，一个将因子感知专家混合模型（MoE）与预训练编码器相结合的框架，以增强对环境变化的泛化能力。FAME 遵循三阶段的训练过程：（1）策略预热，在此阶段，扩散策略在标准环境数据上进行训练，编码器保持冻结；（2）因子特定适配器训练，在此阶段，插入在冻结编码器与暂时冻结的策略之间的轻量级适配器在定制数据集上进行训练，每个数据集针对特定的环境变化；（3）联合微调，在此阶段，中央路由器和预热策略在混合数据上进行训练，以共同处理多个因素。FAME 被称为“因子感知”，因为中央路由器以稠密的 MoE 方式对冻结的因子特定适配器进行软加权，从而实现跨多个因素的组合泛化。在 Meta-World 基准测试中的评估显示，FAME 的表现比扩散策略基线提高了 34%。我们进一步在真实世界的抓取与放置任务中验证了 FAME，使用在新收集数据上训练的紧凑模型，FAME 在真实世界变化下的泛化能力提高了 35%。
cs.RO / 29 / 2606.21139
PoLAR: Factorizing Extent and Mode in Latent Actions for Robot Policy Learning
PoLAR：在潜在动作中因式分解范围和模式以实现机器人策略学习
Abstract
Latent action pretraining learns representations of visual change from pairs of observations, but existing methods typically encode each transition as a single unstructured representation that entangles transition extent and transition mode. We introduce Polar Latent Actions with Radial structure (PoLAR), which imposes a radial-direction structure on latent actions, encouraging radius to encode transition extent and direction to retain transition mode. PoLAR uses temporal offset between two observations as a weak proxy for transition extent, encouraging latent action from observation pairs separated by larger temporal gaps to occupy larger radii. We instantiate this structure in hyperbolic space, whose expanding volume with radius offers a natural fit for more diverse transition modes at larger extents. Across in-task and large-scale pretraining settings, PoLAR improves downstream policy performance in simulation and real-world robot experiments, outperforming latent action baselines and strong pretrained VLAs. These results suggest that the geometry of the latent action space is an important design choice for transferring visual pretraining to downstream robot policy learning.
Chinese Translation
潜在动作预训练通过观察对学习视觉变化的表示，但现有方法通常将每个转变编码为单一的非结构化表示，这使得转变范围和转变模式相互纠缠。我们引入了具有径向结构的极性潜在动作（PoLAR），该方法在潜在动作上施加了径向方向结构，鼓励半径编码转变范围，而方向则保留转变模式。PoLAR使用两个观察之间的时间偏移作为转变范围的弱代理，鼓励来自时间间隔较大的观察对的潜在动作占据更大的半径。我们在双曲空间中实例化这一结构，其随半径扩展的体积为更大范围内的多样化转变模式提供了自然的适配。在任务内和大规模预训练设置中，PoLAR在模拟和真实世界的机器人实验中提高了下游策略的性能，超越了潜在动作基线和强大的预训练视觉语言代理（VLA）。这些结果表明，潜在动作空间的几何结构是将视觉预训练转移到下游机器人策略学习的重要设计选择。
cs.RO / 30 / 2606.21148
Pose-Agnostic Robotic Functional Grasping via Observation-Action Canonicalization
无姿态依赖的机器人功能性抓取通过观察-动作规范化
Abstract
Functional robotic grasping requires a policy that generalizes across diverse object geometries and poses while maintaining task-specific contact precision. We study this challenge through mug-handle grasping, where thin handles, instance variation, and upright or inverted placements make both perception and control sensitive to object configuration. Grasp pose detection methods operate open-loop and are sensitive to estimation errors on thin handle structures. Learned visuomotor policies must implicitly learn to handle the coupled variation in visual appearance and action direction induced by different object placements, limiting generalization. We propose AnyMug, a canonicalized visuomotor reinforcement learning framework for functional grasping that trains a single closed-loop policy entirely in simulation and deploys it zero-shot on a real robot. AnyMug introduces observation-action canonicalization, which transforms both the depth observation and the predicted end-effector action into a shared object-centric frame. The policy therefore sees a consistent mug-centered view and emits actions in a canonical direction regardless of mug placement, allowing the same grasping behavior to be reused across configurations. A handle-aware reward further encourages precise approach, gripper alignment, and opposing-finger placement, while a pose curriculum and domain randomization improve training stability and sim-to-real transfer. In simulation, AnyMug achieves over 93% success rate on both unseen upright and inverted mugs and transfers zero-shot to a real Franka Panda, reaching 80% success rate on 5 held-out physical mugs across both pose categories.
Chinese Translation
功能性机器人抓取需要一种能够在多样的物体几何形状和姿态之间进行泛化的策略，同时保持任务特定的接触精度。我们通过杯柄抓取研究这一挑战，在这种情况下，细长的把手、实例变异以及直立或倒置的放置方式使得感知和控制对物体配置非常敏感。抓取姿态检测方法采用开环操作，对细把手结构的估计误差非常敏感。学习的视觉运动策略必须隐式地学习处理由不同物体放置引起的视觉外观和动作方向的耦合变异，这限制了泛化能力。我们提出了AnyMug，一个用于功能性抓取的规范化视觉运动强化学习框架，该框架在模拟环境中训练一个完全闭环的策略，并在真实机器人上进行零-shot部署。AnyMug引入了观察-动作规范化，将深度观察和预测的末端执行器动作转换为共享的物体中心框架。因此，策略能够以一致的杯子中心视图进行观察，并在规范化方向上发出动作，无论杯子的放置如何，从而允许在不同配置中重用相同的抓取行为。一个关注把手的奖励进一步鼓励精确的接近、夹具对齐和对指放置，同时姿态课程和领域随机化提高了训练的稳定性和模拟到现实的转移。在模拟中，AnyMug在未见过的直立和倒置杯子上实现了超过93%的成功率，并在真实的Franka Panda机器人上零-shot转移，达到在两个姿态类别下对5个保留物理杯子的80%成功率。
cs.RO / 31 / 2606.21165
OmniV2X: A Generative Foundation Planner for Efficient End-to-End Cooperative Driving
OmniV2X：高效端到端协同驾驶的生成基础规划模型
Abstract
We present OmniV2X, a generative foundation model for vehicle-to-everything (V2X) cooperative driving. The model directly interprets independent context sequences comprising multi-modal and multi-agent observations. The new design mitigates the computational cost of dense 3D perception, the vulnerability to data scarcity in cooperative scenarios, and the poor compliance with standardized messaging in existing methods that fuse multi-modal inputs into a shared representation. For training, we present an end-to-end supervised pipeline using a downstream trajectory generation loss, in which a high-capacity generative sequence planner implicitly learns to steer the model and leverage multi-modal inputs via cross-attention injection. As a foundation model, we demonstrate that OmniV2X pre-trained on large-scale single-agent planning datasets can efficiently adapt to cooperative environments by integrating the conditioning context with lightweight, standard-compliant V2X tokens. Evaluated on the DAIR-V2X-Seq dataset, OmniV2X outperforms existing end-to-end cooperative driving baselines, achieving state-of-the-art performance with less than 10% of the fine-tune V2X dataset and less than 1% of the communication bandwidth. We conduct comprehensive evaluations to demonstrate its computational efficiency and robustness under real-world constraints.
Chinese Translation
我们提出了OmniV2X，一种用于车与万物（V2X）协同驾驶的生成基础模型。该模型直接解释由多模态和多智能体观察组成的独立上下文序列。新设计减轻了密集3D感知的计算成本、在协同场景中对数据稀缺的脆弱性，以及现有方法在将多模态输入融合为共享表示时对标准化消息的低遵循性。为了进行训练，我们提出了一种端到端的监督管道，使用下游轨迹生成损失，其中高容量的生成序列规划器隐式学习如何引导模型并通过交叉注意力注入利用多模态输入。作为基础模型，我们展示了在大规模单智能体规划数据集上预训练的OmniV2X如何通过将条件上下文与轻量级、符合标准的V2X令牌相结合，能够高效适应协同环境。在DAIR-V2X-Seq数据集上的评估表明，OmniV2X超越了现有的端到端协同驾驶基线，凭借不到10%的微调V2X数据集和不到1%的通信带宽，实现了最先进的性能。我们进行了全面评估，以展示其在现实世界约束下的计算效率和鲁棒性。
cs.RO / 32 / 2606.21188
Remember what you did?: Learning Behavioral Memories for Partially Observable Object Manipulation
记住你做过的事情吗？：学习部分可观察物体操控的行为记忆
Abstract
Long horizon, contact-rich manipulation is inherently partially observable. This is as a single visual observation rarely captures a robot's full action context, including prior attempts, interactions, or progress. Consequently, standard visuomotor policies or vision-language-action models are prone to struggle in such tasks due to a lack of memory. To address this, we introduce Compressed Action Memory Policy (CAMP) based on the insight that a robot's own action history serves as a highly informative, self-supervised signal, enabling the policy to learn a robust, compact history representation. In our approach, we train a memory module to maintain a compressed representation of past actions, forcing it to encode a latent behavioral memory of all the robot's past interactions that can then be used to better contextualize future actions. This allows our approach to implicitly track generalized task progress and learn from failed attempts without any additional supervision, or external oversight. We evaluate CAMP across four real-robot setups and two novel simulation benchmarks: Memory-T-Bench and Memory-Manip-Bench. By demonstrating substantial gains over state-of-the-art baselines, CAMP is, to our knowledge, the first policy to demonstrate substantial success on contact-rich partially observable manipulation tasks purely through learned memory.
Chinese Translation
长时间跨度、接触丰富的操控本质上是部分可观察的。因为单一的视觉观察很少能捕捉到机器人完整的动作上下文，包括之前的尝试、交互或进展。因此，标准的视觉运动策略或视觉-语言-动作模型在此类任务中往往会因缺乏记忆而面临困难。为了解决这个问题，我们提出了压缩动作记忆策略（Compressed Action Memory Policy，CAMP），基于这样的洞察：机器人的自身动作历史作为一种高度信息化的自我监督信号，使得策略能够学习到一个稳健、紧凑的历史表示。在我们的方法中，我们训练一个记忆模块来维护过去动作的压缩表示，迫使其编码所有机器人过去交互的潜在行为记忆，从而更好地为未来的动作提供上下文。这使得我们的方法能够隐式跟踪通用任务进展，并在没有任何额外监督或外部监督的情况下从失败的尝试中学习。我们在四个真实机器人设置和两个新颖的仿真基准（Memory-T-Bench 和 Memory-Manip-Bench）上评估CAMP。通过展示相较于最先进基线的显著提升，CAMP是我们所知的第一个在接触丰富的部分可观察操控任务中通过学习记忆取得显著成功的策略。
cs.RO / 33 / 2606.21216
A scalar per patch from pre-trained ViTs enables fast moving navigation in the real world
来自预训练视觉变换器的每个补丁标量实现快速的现实世界导航
Abstract
Trained policies for real-world robotics rely on computer vision components, typically in the form of pre-trained visual encoders. These encoders are an essential component and it has been shown that their power does not emerge from training on robotics downstream losses alone. Pre-training with auxiliary losses in the form of computer-vision pre-text tasks is a defining factor and heavily conditions agent performance in robotics tasks. In this unprecedented large-scale study, we ran 966 navigation episodes of static point goal navigation in a real-world building for 24km and asked which components really matter for the computer vision aspects of robotics: we evaluate state-of-the art visual encoders in realistic conditions. We explore the usefulness of heterogeneous multi-teacher distillation leading to encoders with multiple different and complementary skills. We investigate how much information from these encoders is necessary for robotics by bottlenecking them in a principled and spatially useful way and we show that this leads to the emergence of interpretable features linked to affordances. We also argue that training policies on RGB data alone does not lead to an optimal usage of visual features and show this by finetuning policies pre-trained on privileged information. All in all, we paint a more complete picture of what aspects of computer vision are relevant for real-world navigation.
Chinese Translation
针对现实世界机器人技术的训练策略依赖于计算机视觉组件，通常以预训练视觉编码器的形式存在。这些编码器是一个重要组成部分，已有研究表明，它们的能力并非仅仅来源于对机器人下游损失的训练。以计算机视觉预文本任务形式的辅助损失进行预训练是一个决定性因素，并且在机器人任务中严重影响代理的表现。在这项前所未有的大规模研究中，我们在一个真实建筑中进行了966次静态目标导航的导航实验，总行程达到24公里，并探讨了哪些组件对机器人计算机视觉方面真正重要：我们在现实条件下评估了最先进的视觉编码器。我们探索了异构多教师蒸馏的有效性，从而获得具有多种不同且互补技能的编码器。我们研究了这些编码器中有多少信息对于机器人是必要的，通过以原则性和空间上有用的方式对其进行瓶颈化，我们展示了这导致了与可供性相关的可解释特征的出现。我们还认为，仅在RGB数据上训练策略并不能实现视觉特征的最佳使用，并通过对在特权信息上预训练的策略进行微调来证明这一点。总的来说，我们描绘了计算机视觉在现实世界导航中相关的各个方面的更完整的图景。
cs.RO / 34 / 2606.21220
A UAV-Mounted Sensor Network for Close-Range Inspection of Wind Turbine Rotor Blades
用于风力涡轮机叶片近距离检查的无人机搭载传感器网络
Abstract
Inspection of offshore wind turbine rotor blades is critical for predictive maintenance to maximise efficiency and extend operational lifetime. However, it remains a challenging task due to remote locations, large structural dimensions, and the limitations of current UAV-compatible sensor systems. While existing approaches can detect certain types of surface anomalies, reliable classification of defect types often remains a manual and error-prone process. This paper presents the design of a UAV-mounted multimodal sensor network combining an industrial RGB camera, a passive thermal infrared camera, and an in-house developed 3D scanner. All sensors are co-calibrated into a common coordinate frame, enabling spatial superimposition of geometric, colour, and thermal data. The system is designed to operate at close range, addressing three fundamental sensing challenges: platform motion, large field of view, and millimetre-level measurement accuracy. Preliminary laboratory results demonstrate synchronised multi-sensor acquisition and initial point cloud reconstructions, forming the basis for future airborne inspection trials.
Chinese Translation
离岸风力涡轮机叶片的检查对于预测性维护至关重要，以最大化效率并延长操作寿命。然而，由于远程位置、大型结构尺寸以及当前无人机兼容传感器系统的局限性，这仍然是一项具有挑战性的任务。虽然现有方法能够检测某些类型的表面异常，但缺陷类型的可靠分类通常仍然是一个手动且容易出错的过程。本文提出了一种无人机搭载的多模态传感器网络的设计，该网络结合了工业RGB相机、被动热红外相机和自主开发的3D扫描仪。所有传感器都被共同标定到一个公共坐标系中，从而实现几何、颜色和热数据的空间叠加。该系统设计用于近距离操作，解决了三个基本的传感挑战：平台运动、大视场和毫米级测量精度。初步的实验室结果展示了同步的多传感器采集和初步的点云重建，为未来的空中检查试验奠定了基础。
cs.RO / 35 / 2606.21222
FleetAgent: Teleoperation Assistant for Autonomous Fleets via Vectorized V2N Messages
FleetAgent：通过向量化的车辆到网络（V2N）消息为自主车队提供远程操作助手
Abstract
Large-scale autonomous fleets rely on teleoperation to resolve rare failures, yet streaming raw sensor data from many vehicles is costly, and remote operators can only monitor a limited number of vehicles at a time. We introduce FleetAgent, a cloud-hosted multimodal large language model (MLLM) assistant that consumes compact vectorized vehicle-to-network (V2N) messages, such as map elements, detected objects, and the ego planned path. It provides a structured natural-language response (including narration, explanation, and evaluation of the plan and scene), along with an intervention urgency score for operator prioritization. To make structured messages compatible with token-based MLLMs, we propose VecFormer, a vector-to-embedding interface with differentiable top-K context selection that bounds context length and GPU KV-cache growth, enabling more efficient batch processing, which is important under the context of cloud-hosted large-scale fleet management. We also construct VecEval, a nuScenes-derived dataset with paired human and synthetic imperfect plans and human-verified language labels, to facilitate the training and evaluation of our proposed system. Our proposed system can reduce uplink payload by up to 625 times compared with raw images and reduce KV-cache memory by 16.54 times compared with original text descriptions. On VecEval, FleetAgent improves Lingo-Judge score by 16.8% and reduces intervention failure rate by 19.9%, compared with Qwen2.5-VL-7B using language descriptions. These results demonstrate that FleetAgent can utilize compact structured V2N messaging to enable efficient, explainable teleoperation monitoring for autonomous fleets.
Chinese Translation
大规模自主车队依赖远程操作来解决罕见故障，但从多辆车辆流式传输原始传感器数据的成本较高，且远程操作员一次只能监控有限数量的车辆。我们提出了FleetAgent，这是一种云托管的多模态大型语言模型（MLLM）助手，能够处理紧凑的向量化车辆到网络（V2N）消息，如地图元素、检测到的物体和自我规划路径。它提供结构化的自然语言响应（包括叙述、解释以及对计划和场景的评估），并附带干预紧急程度评分以便操作员优先处理。为了使结构化消息与基于令牌的MLLM兼容，我们提出了VecFormer，这是一种具有可微分的前K上下文选择的向量到嵌入接口，能够限制上下文长度和GPU KV缓存的增长，从而实现更高效的批处理，这在云托管的大规模车队管理背景下尤为重要。我们还构建了VecEval，这是一个基于nuScenes的数据集，包含配对的人类和合成不完美计划以及经过人工验证的语言标签，以促进我们提出的系统的训练和评估。与原始图像相比，我们提出的系统可以将上行负载减少多达625倍，并且与原始文本描述相比，KV缓存内存减少16.54倍。在VecEval上，FleetAgent的Lingo-Judge得分提高了16.8%，干预失败率降低了19.9%，与使用语言描述的Qwen2.5-VL-7B相比。这些结果表明，FleetAgent能够利用紧凑的结构化V2N消息，实现对自主车队的高效、可解释的远程操作监控。
cs.RO / 36 / 2606.21223
Ultra-Fusion: A Resilient Tightly-Coupled Multi-Sensor Fusion SLAM Framework under Sensor Degradation and Spatiotemporal Perturbation for Intelligent Transportation Systems
超融合：一种在传感器退化和时空扰动下的抗干扰紧耦合多传感器融合SLAM框架，用于智能交通系统
Abstract
Reliable localization is essential for intelligent transportation systems (ITS), including autonomous vehicles, quadruped last-mile carriers, and infrastructure-inspection unmanned aerial vehicles (UAVs). Although tightly-coupled multi-sensor fusion improves accuracy in favorable conditions, deployed systems remain vulnerable to sensor degradation -- poor illumination, LiDAR degeneracy, wheel slippage, and GNSS outage -- and to spatiotemporal calibration errors. These failures are common in urban canyons, tunnels, and high-speed corridors, where localization drift can degrade route tracking, tunnel passage continuity, and local map alignment. This paper presents Ultra-Fusion, a tightly-coupled multi-sensor localization framework based on a unified sliding-window estimator. Asynchronous measurements are timestamp-ordered and converted into optional factors within one optimization window, supporting WIO, VIO, LIO, and LVIO with optional wheel and GNSS augmentation. Observability-aware initialization selects the bootstrap mode, factor-wise reliability scheduling gates degraded measurements, and online LiDAR--IMU spatiotemporal calibration refines temporal offsets and rotational extrinsics during operation. We extend the M3DGR benchmark with simulation trajectories and evaluate more than 60 open-source SLAM systems on M3DGR, M2DGR-Plus, KAIST, GrandTour, and MARS-LVIG. The results show competitive accuracy across wheeled, legged, and aerial platforms under long-duration and high-speed operation, degradation, and calibration perturbation, improving localization availability for road-level autonomy, campus and warehouse mobility, and low-altitude aerial inspection. To benefit the industrial and academic community, we will release source code and datasets upon paper acceptance.
Chinese Translation
可靠的定位对于智能交通系统（ITS）至关重要，包括自动驾驶汽车、四足最后一公里运输工具和基础设施检查无人机（UAV）。尽管紧耦合多传感器融合在有利条件下提高了精度，但已部署的系统仍然容易受到传感器退化（如光照不足、激光雷达退化、轮滑和全球导航卫星系统（GNSS）失效）以及时空校准误差的影响。这些故障在城市峡谷、隧道和高速走廊中很常见，定位漂移可能会降低路线跟踪、隧道通行连续性和局部地图对齐的效果。本文提出了超融合（Ultra-Fusion），一种基于统一滑动窗口估计器的紧耦合多传感器定位框架。异步测量按时间戳排序，并在一个优化窗口内转换为可选因子，支持WIO、VIO、LIO和LVIO，并可选择轮子和GNSS增强。可观测性感知初始化选择引导模式，因子级可靠性调度门控退化测量，在线激光雷达-惯性测量单元（LiDAR-IMU）时空校准在操作过程中细化时间偏移和旋转外参。我们扩展了M3DGR基准，增加了模拟轨迹，并在M3DGR、M2DGR-Plus、KAIST、GrandTour和MARS-LVIG上评估了60多个开源SLAM系统。结果显示，在长时间和高速操作、退化和校准扰动下，轮式、腿式和空中平台的精度具有竞争力，提高了道路级自主、校园和仓库移动性以及低空空中检查的定位可用性。为了惠及工业和学术界，我们将在论文接受后发布源代码和数据集。
cs.RO / 37 / 2606.21245
A Human-Inspired Thumb-Index Robotic Hand with Strain Gauges Embedded in Soft Joints
一种人类启发的拇指-食指机器人手，软关节中嵌入应变计
Abstract
Human hand grasp adaptation depends mainly on the synergy between physical structure and biological feedback. Inspired by this biomechanical principle, the Safe Thumb-Index Robotic (STIR) Hand was developed as a minimal, lightweight, and low-cost two-digit prototype featuring an asymmetric thumb-index configuration. By pairing an underactuated, tendon-driven mechanical design with flexible strain gauges embedded into silicone-encapsulated soft joints, the system achieves passive grasp adaptation while establishing both internal proprioception and external perception. Unsupervised analysis was carried out on a dataset of the STIR hand grasping 20 different objects, along with an object classification task and an ablation study to highlight the contribution of the soft joint sensors. The object classification task discriminated object size, shape, and material stiffness with a high classification accuracy. In contrast to traditional industrial grippers and robotic hands, the STIR Hand demonstrates that sensorized compliant joints significantly improve overall sensitivity and ensure safe grasping, while remaining independent of additional fingertip tactile elements or external vision systems. Finally, a comparison to similar devices grasping identical objects validates the utility of the STIR Hand.
Chinese Translation
人手抓握适应性主要依赖于物理结构与生物反馈之间的协同作用。受这一生物力学原理的启发，开发了安全拇指-食指机器人手（Safe Thumb-Index Robotic Hand，STIR），该原型具有最小化、轻量化和低成本的特点，采用不对称的拇指-食指配置。通过将一种欠驱动的、腱驱动的机械设计与嵌入硅胶包裹软关节的柔性应变计相结合，该系统实现了被动抓握适应，同时建立了内部本体感知和外部感知。对STIR手抓握20种不同物体的数据集进行了无监督分析，并进行了物体分类任务和消融研究，以突出软关节传感器的贡献。物体分类任务能够高效区分物体的大小、形状和材料刚度，且分类准确率高。与传统工业夹具和机器人手相比，STIR手表明，传感器化的柔性关节显著提高了整体灵敏度，并确保安全抓握，同时不依赖于额外的指尖触觉元素或外部视觉系统。最后，与抓握相同物体的类似设备的比较验证了STIR手的实用性。
cs.RO / 38 / 2606.21258
Spectral GS-SLAM: Observability-Aware, Degeneracy-Robust Tracking for Real-Time 3D Gaussian Splatting SLAM
谱系GS-SLAM：具可观测性意识的、抗退化的实时3D高斯点云SLAM跟踪
Abstract
Recent 3DGS-SLAM systems enable real-time operation by leveraging conventional feature matching or ICP-based tracking, thereby avoiding the heavy dense photometric optimization used in earlier approaches. However, feature matching remains prone to failure in textureless environments, while ICP-based tracking struggles in structureless or geometrically degenerate scenes due to ill-conditioned optimization. To address this issue, we propose Spectral GS-SLAM, an efficient yet robust tracking framework that integrates ICP with complementary feature-based constraints. Our method mitigates numerical instability by adaptively compensating under-constrained directions in degenerate scenarios, without interfering with the shared Gaussian representation used for mapping. We further introduce a Gaussian-aware planarity weighting mechanism that exploits the intrinsic covariance structure of 3D Gaussians to characterize scene geometry and guide information fusion. Extensive evaluations on challenging TUM RGB-D sequences demonstrate that Spectral GS-SLAM achieves real-time performance (40.14 FPS) while maintaining consistent tracking in both structureless and featureless environments. The proposed method preserves trajectory integrity in degenerate scenes while maintaining competitive performance in non-adverse conditions.
Chinese Translation
近期的3DGS-SLAM系统通过利用传统特征匹配或基于ICP的跟踪实现了实时操作，从而避免了早期方法中使用的繁重的密集光度优化。然而，特征匹配在无纹理环境中仍然容易失败，而基于ICP的跟踪在无结构或几何退化场景中由于优化条件不良而面临困难。为了解决这一问题，我们提出了谱系GS-SLAM，这是一种高效且稳健的跟踪框架，集成了ICP与互补的基于特征的约束。我们的方法通过自适应补偿退化场景中的欠约束方向来减轻数值不稳定性，而不干扰用于映射的共享高斯表示。我们进一步引入了一种高斯感知的平面加权机制，利用3D高斯的内在协方差结构来表征场景几何并指导信息融合。在具有挑战性的TUM RGB-D序列上的广泛评估表明，谱系GS-SLAM在结构无关和特征缺失环境中实现了实时性能（40.14 FPS），同时保持了一致的跟踪。所提出的方法在退化场景中保持轨迹完整性，同时在非不利条件下保持竞争力的性能。
cs.RO / 39 / 2606.21264
RogueRover: Autonomous Rogue Device Localization for Incident Response
RogueRover：用于事件响应的自主恶意设备定位
Abstract
Physically localizing unauthorized wireless devices remains a critical bottleneck in cyber-physical security operations, where rogue access points can provide entry points for lateral movement and persistent compromise. While such devices can often be detected through network-side mechanisms, determining their physical location typically requires dense sensing infrastructure, site-specific RF fingerprinting, or manual inspection, limiting timely incident response. We investigate whether a single commodity robot can autonomously detect and localize rogue wireless devices under zero-configuration constraints, without RF fingerprinting, pre-installed sensors, or site calibration. We present RogueRover, an end-to-end system in which a quadruped robot autonomously patrols, collects spatially labeled RSSI measurements via a standard 802.11 interface, and estimates device locations offline. We evaluate the system across 11 patrol runs in a real indoor environment, with 6 rogue devices deployed under heterogeneous propagation conditions. Across 62 AP-patrol sessions, RogueRover achieves a median single-patrol localization error of 1.62 m without prior RF knowledge. Under multi-run aggregation, five of six devices are localized within 1 m. A blind trial validates the full pipeline, correctly identifying rogue devices among 73 observed BSSIDs and localizing them with errors of 0.34 m and 1.84 m. Across environments, simple weighted-centroid estimators perform comparably to, or better than, parametric path-loss models, indicating that measurement coverage from autonomous patrols is the primary determinant of localization accuracy under zero-prior constraints. Our results demonstrate that infrastructure-free, autonomous localization is feasible in practice, enabling rapid physical incident response in cyber-physical environments without additional sensing infrastructure.
Chinese Translation
在网络物理安全操作中，物理定位未经授权的无线设备仍然是一个关键瓶颈，因为恶意接入点可能为横向移动和持续妥协提供入口。虽然这些设备通常可以通过网络侧机制进行检测，但确定其物理位置通常需要密集的传感基础设施、特定场地的射频指纹识别或人工检查，这限制了及时的事件响应。我们研究了在零配置约束下，单个商品机器人是否能够自主检测和定位恶意无线设备，而无需射频指纹识别、预装传感器或场地校准。我们提出了RogueRover，这是一个端到端系统，其中四足机器人自主巡逻，通过标准802.11接口收集空间标记的RSSI测量，并离线估计设备位置。我们在一个真实的室内环境中进行了11次巡逻测试，部署了6个在异构传播条件下的恶意设备。在62次接入点巡逻会话中，RogueRover在没有先前射频知识的情况下，实现了1.62米的中位单次巡逻定位误差。在多次运行聚合下，六个设备中有五个被定位在1米以内。一项盲测验证了整个流程，在观察到的73个基本服务集标识符（BSSID）中正确识别恶意设备，并以0.34米和1.84米的误差进行定位。在不同环境中，简单的加权质心估计器的表现与参数路径损耗模型相当或更好，表明自主巡逻的测量覆盖是零先验约束下定位精度的主要决定因素。我们的结果表明，基础设施无关的自主定位在实践中是可行的，使得在网络物理环境中能够快速进行物理事件响应，而无需额外的传感基础设施。
cs.RO / 40 / 2606.21372
NAC: Neural Action Codec for Vision-Language-Action Models
NAC：用于视觉-语言-动作模型的神经动作编码器
Abstract
Vision-language-action (VLA) models rely on discrete action tokenizers to bridge continuous robot control and autoregressive sequence modeling, yet existing tokenizers often trade off between compression, latency, and downstream performance. We revisit this design through the lens of neural audio codecs-convolutional encoder-decoder architectures with residual vector quantization that serve as the standard front end for audio foundation models. Motivated by their success, we introduce the Neural Action Codec (NAC), which treats short robot action trajectories as multi-channel 1D signals and compresses them using a multi-scale RVQGAN architecture. We observe that audio-specific mel-spectrogram objectives are ill-suited for kinematic signals; however, by replacing them with simple time-domain and non-mel spectral reconstruction losses, audio-codec-style models can autoencode actions with high fidelity without substantial architectural changes. NAC provides a compact, ordered token space via offset codebooks, enabling standard autoregressive policies to operate over short, structured sequences. Meanwhile, a Vocos-style decoder with an ISTFT head and adversarial discriminators recovers smooth, detailed trajectories. Across LIBERO-10, RoboMimic, and a suite of real-world manipulation tasks, NAC achieves lower reconstruction error and higher success rates than binning, FAST, and prior VQ-based tokenizers at comparable or better compression rates. These results demonstrate that repurposed neural audio codecs offer a strong, practical backbone for learned action tokenization in modern VLAs.
Chinese Translation
视觉-语言-动作（VLA）模型依赖于离散动作标记器来连接连续的机器人控制与自回归序列建模，但现有的标记器往往在压缩、延迟和下游性能之间存在权衡。我们通过神经音频编码器的视角重新审视这一设计——卷积编码器-解码器架构与残差向量量化，作为音频基础模型的标准前端。受到其成功的启发，我们引入了神经动作编码器（NAC），将短机器人动作轨迹视为多通道一维信号，并使用多尺度RVQGAN架构对其进行压缩。我们观察到，音频特定的梅尔谱目标不适合运动学信号；然而，通过用简单的时域和非梅尔谱重建损失替换它们，音频编码器风格的模型可以在不进行实质性架构更改的情况下高保真地自编码动作。NAC通过偏移代码本提供了紧凑、有序的标记空间，使标准自回归策略能够在短的、结构化的序列上运行。同时，带有ISTFT头和对抗判别器的Vocos风格解码器能够恢复平滑、详细的轨迹。在LIBERO-10、RoboMimic以及一系列现实世界操作任务中，NAC在可比或更好的压缩率下实现了低重建误差和更高的成功率，优于分箱、FAST和先前的基于VQ的标记器。这些结果表明，重新利用的神经音频编码器为现代VLA中的学习动作标记化提供了强大且实用的基础。
cs.RO / 41 / 2606.21387
Long-Distance Real-World Navigation of the Legged-Wheeled Robot Go2-W Using Deep Reinforcement Learning
基于深度强化学习的腿轮机器人Go2-W的远程真实世界导航
Abstract
Legged-wheeled robots have long been studied for their potential to combine the efficient flat-ground mobility of wheels with the rough-terrain capability of legs. However, examples of their application to long-range autonomous navigation in real environments remain limited. This paper reports our effort to build a deep reinforcement learning (DRL) based locomotion controller and an autonomous navigation system for the commercially available legged-wheeled robot Go2-W, and to apply them to long-range autonomous navigation in a real environment. For locomotion control, we extended a proprioception-only policy, which we had previously developed for quadruped robots, to the 16-DoF legged-wheeled robot. We also found that wheeled locomotion concentrates the load on the hip joints and causes heat concentration that hinders sustained travel, and obtained a policy that suppresses it by distributing the load. We evaluated the system at the Tsukuba Challenge 2025, demonstrating that it can autonomously traverse an approximately 2.8 km route including sidewalks, a park, and stairs without stopping due to overheating.
Chinese Translation
腿轮机器人因其将轮子的高效平地移动能力与腿部的崎岖地形适应能力相结合的潜力而受到广泛研究。然而，应用于真实环境中的远程自主导航的实例仍然有限。本文报告了我们为市售腿轮机器人Go2-W构建基于深度强化学习（DRL）的运动控制器和自主导航系统的努力，并将其应用于真实环境中的远程自主导航。在运动控制方面，我们将之前为四足机器人开发的仅基于本体感觉的策略扩展到16自由度的腿轮机器人。我们还发现，轮式运动会将负载集中在髋关节上，导致热量集中，从而妨碍持续行驶，并获得了一种通过分散负载来抑制这种现象的策略。我们在筑波挑战赛2025上评估了该系统，证明其能够自主穿越约2.8公里的路线，包括人行道、公园和楼梯，而不会因过热而停止。
cs.RO / 42 / 2606.21396
Overcoming Imperfect Kinematics in Surgical Robotics Through Sim-to-Real Visuomotor Learning
通过模拟到现实的视觉运动学习克服外科机器人中的不完美运动学
Abstract
Robot-Assisted Surgery is integral to modern minimally invasive procedures, with automation emerging as the next frontier to enhance precision and reduce surgeon fatigue. This evolution is largely impeded by the inherent kinematic inaccuracies of surgical robots, where unreliable internal sensors lead to significant control errors. While previous methods attempted to mitigate these issues through complex model-based calibration, they often suffer from high cost and limited effectiveness. This work utilises a learning-policy to actively compensate for hardware inaccuracies using closed-loop visual feedback that was trained from a teacher-student learning framework. The policy can fuse unreliable internal readings with precise external visual data, allowing it to correct for kinematic errors in real time without needing a perfect physical model. The learned policy was successfully deployed on the da Vinci Research Kit, where experiments validated the fundamental feasibility of using external vision to overcome internal sensor deficits. This research provides a foundational and reliable control methodology, paving the way for more advanced and robust surgical automation.
Chinese Translation
机器人辅助手术是现代微创手术的重要组成部分，自动化正成为提升精确度和减少外科医生疲劳的下一个前沿。这一进展在很大程度上受到外科机器人固有运动学不准确性的阻碍，其中不可靠的内部传感器导致显著的控制误差。尽管之前的方法试图通过复杂的基于模型的校准来缓解这些问题，但它们往往面临高成本和有限的有效性。本研究利用学习策略，通过从师生学习框架中训练的闭环视觉反馈，主动补偿硬件不准确性。该策略能够将不可靠的内部读数与精确的外部视觉数据融合，从而在不需要完美物理模型的情况下实时纠正运动学误差。学习到的策略成功应用于达芬奇研究套件（da Vinci Research Kit），实验验证了使用外部视觉克服内部传感器缺陷的基本可行性。本研究提供了一种基础且可靠的控制方法，为更先进和稳健的外科自动化铺平了道路。
cs.RO / 43 / 2606.21398
BIT-Nav: Brain-Inspired Trajectory Memory for Embodied Navigation
BIT-Nav：基于大脑启发的轨迹记忆用于具身导航
Abstract
Vision-Language Models (VLMs) for embodied navigation rely on selecting a fixed number of frames from a growing trajectory history. As episodes extend, this selection grows increasingly sparse, yet prior work shows no accuracy gain when scaling from 8 to 64 frames, suggesting the bottleneck is not frame quantity but the representation itself. Sparse frame selection cannot capture the structured behavioral signal that long-horizon reasoning requires: turning patterns, cumulative displacement, and path topology. We introduce BIT-Nav (Brain-Inspired Trajectory Memory for Navigation), a framework that augments frozen VLM navigation pipelines with a compact learned trajectory memory. Motivated by hippocampal path integration, where spatial experience is compressed into structured episodic traces rather than stored as raw sensory replay, BIT-Nav trains a Bi-GRU encoder over action and relative pose sequences via a multi-positive InfoNCE contrastive objective on trajectory prefixes sharing the same behavioral intent. The resulting embedding is projected into the VLM token space via a lightweight MLP and injected as a single memory token at each decision step, conditioning the model on structured motion history at constant token cost regardless of episode length
Chinese Translation
用于具身导航的视觉-语言模型（VLMs）依赖于从不断增长的轨迹历史中选择固定数量的帧。随着情节的延续，这种选择变得越来越稀疏，但先前的研究表明，从8帧扩展到64帧并未带来准确性的提升，这表明瓶颈不在于帧的数量，而在于表示本身。稀疏的帧选择无法捕捉长时间推理所需的结构化行为信号：转向模式、累积位移和路径拓扑。我们提出了BIT-Nav（基于大脑启发的导航轨迹记忆），这是一个增强冻结VLM导航管道的框架，配备了紧凑的学习轨迹记忆。BIT-Nav的灵感来源于海马体路径整合，其中空间经验被压缩为结构化的情节痕迹，而不是作为原始感官重放存储。BIT-Nav通过对共享相同行为意图的轨迹前缀进行多正样本信息对比（InfoNCE）对比目标，训练一个双向门控循环单元（Bi-GRU）编码器，处理动作和相对姿态序列。生成的嵌入通过轻量级多层感知器（MLP）投影到VLM标记空间，并在每个决策步骤注入为单个记忆标记，使模型在每个情节长度不变的情况下，能够以恒定的标记成本对结构化运动历史进行条件化。
cs.RO / 44 / 2606.21406
Robot Self-Improvement via Human-Video Dynamics Models
通过人类视频动态模型实现机器人自我改进
Abstract
A central question in robot learning is how to acquire skills from the kinds of data that humans learn from: passive observation, embodied practice, and the experience of failure. Human videos provide the first of these in abundance, and prior work has shown they can initialize useful policies. Far less clear is whether they can support the second and third: whether priors extracted from human videos can ground a robot's own attempts well enough to evaluate them, correct them, and improve from them. In this work, we show that human videos can be used to learn embodiment-agnostic action, dynamics, and value representations that transfer across robot embodiments, providing the predictive foundation required for robots to autonomously improve from their own rollouts and failures. We introduce Dynamics-Guided Action Correction (DGAC), a training-free approach that leverages these adapted models to repair failed states: each failure becomes a query for which the learned models propose and rank corrective actions, turning failures into supervision for the next policy update. Across seven real-world manipulation tasks spanning both a mobile manipulator and a static manipulator arm, our approach improves success rates from 40% to 81% across multiple policy backbones, demonstrating cross-embodiment robot self-improvement from human-video priors. These results show that human priors and robot failures can be combined to enable scalable autonomous policy improvement. Project page: https://ethz-mrl.github.io/robot-self-improvement-website/.
Chinese Translation
机器人学习中的一个核心问题是如何从人类学习的数据中获取技能：被动观察、具身实践和失败经验。人类视频在这些数据中提供了丰富的第一类信息，先前的研究表明它们可以初始化有用的策略。然而，是否能够支持第二类和第三类问题仍然不够明确：从人类视频中提取的先验知识是否能够为机器人自身的尝试提供足够的基础，以便评估、纠正并从中改进。在本研究中，我们展示了人类视频可以用于学习与具身性无关的动作、动态和价值表示，这些表示可以在不同的机器人具身之间迁移，为机器人从自身的执行和失败中自主改进提供所需的预测基础。我们提出了动态引导的动作修正（Dynamics-Guided Action Correction, DGAC），这是一种无训练的方法，利用这些适应后的模型来修复失败状态：每次失败都成为一个查询，学习模型提出并排序纠正动作，将失败转化为下一个策略更新的监督。在七个真实世界的操作任务中，包括移动操纵器和静态操纵臂，我们的方法将成功率从40%提高到81%，展示了基于人类视频先验的跨具身机器人自我改进。这些结果表明，人类先验和机器人失败可以结合起来，实现可扩展的自主策略改进。项目页面：https://ethz-mrl.github.io/robot-self-improvement-website/
cs.RO / 45 / 2606.21438
Temporal logics and formal synthesis for robot planning and control
机器人规划与控制的时序逻辑与形式合成
Abstract
As robots move from controlled environments into real-world settings, it becomes increasingly crucial to ensure that they perform as expected. A key step toward that goal is a rigorous specification of the desired robot behavior, capturing intricate temporal, spatial, and logical requirements. Complementing this, plan and control synthesis methods are needed to fulfill these specifications with provable guarantees. This manuscript presents temporal logics - particularly linear and signal temporal logic - as expressive specification languages for robot behavior over time. We then discuss principles of formal synthesis, from discrete graph- and game-based approaches to sampling-based motion planning, trajectory optimization, and control-certificate-based synthesis. Finally, we outline challenges in deploying formal synthesis in real-world robotics, emphasizing the interplay between modeling fidelity, computational tractability, and the types of rigorous guarantees that can be achieved.
Chinese Translation
随着机器人从受控环境转向现实世界，确保它们按预期执行变得愈发重要。实现这一目标的关键步骤是对期望的机器人行为进行严格的规范，捕捉复杂的时序、空间和逻辑要求。为此，需要规划和控制合成方法，以可证明的保证来满足这些规范。本文提出了时序逻辑——特别是线性时序逻辑（linear temporal logic）和信号时序逻辑（signal temporal logic）——作为描述机器人行为的表达性规范语言。接着，我们讨论了形式合成的原则，包括离散图和基于博弈的方法、基于采样的运动规划、轨迹优化以及基于控制证书的合成。最后，我们概述了在现实世界机器人中部署形式合成所面临的挑战，强调建模精度、计算可行性与可实现的严格保证之间的相互作用。
cs.RO / 46 / 2606.21461
Manipulider: A Multi-Engine Buoyancy-Controlled Robot for Thrusterless Underwater Gliding and Manipulation
Manipulider：一种多引擎浮力控制机器人，用于无推进器的水下滑行和操作
Abstract
The Manipulider is a buoyancy-actuated underwater robot that enables thrusterless, glide-like locomotion and attitude-based manipulation, while providing a magnetic modular interface for rapid payload swapping (e.g., a gripper or sensors). Four syringe-based buoyancy engines distributed around the body jointly regulate net buoyancy and the center of buoyancy, allowing the vehicle to maintain large tilt angles through static force balance without continuous thrust and to avoid propeller entanglement risks. We present the mechanical and electrical design, calibration procedure, and control architecture. Experiments with a gripper attached (no external payload) show a controllable buoyancy-displacement range of 40 mL per engine ({\approx}160 g total buoyancy authority), maximum statically stable tilts of 64.6{\deg} (single-engine) and 61.8{\deg} (dual-engine), and representative vertical and tilt-transition dynamics. We further demonstrate tilt regulation, controlled ascent/descent primitives, and a proof-of-concept gripper-based payload-transport sequence without thrusters.
Chinese Translation
Manipulider是一种浮力驱动的水下机器人，能够实现无推进器的滑行式运动和基于姿态的操作，同时提供一个磁性模块化接口，以便快速更换载荷（例如，抓手或传感器）。四个分布在机身周围的注射器式浮力引擎共同调节净浮力和浮心，使得该机器人能够通过静态力平衡维持较大的倾斜角度，而无需持续推进，从而避免螺旋桨缠绕的风险。我们展示了机械和电气设计、校准程序和控制架构。附加抓手（无外部载荷）的实验表明，每个引擎可控的浮力位移范围为40 mL（约160 g的总浮力权威），最大静态稳定倾斜角为64.6°（单引擎）和61.8°（双引擎），以及代表性的垂直和倾斜过渡动态。我们进一步展示了倾斜调节、受控上升/下降原语，以及一个无推进器的基于抓手的载荷运输序列的概念验证。
cs.RO / 47 / 2606.21470
ASCII Art Turns LLMs into VLA Controllers
ASCII艺术将大型语言模型转变为视觉-语言-动作控制器
Abstract
Vision--Language--Action (VLA) controllers are often built by extending vision--language models (VLMs) with action supervision, relying on multimodal backbones with large data and compute requirements. We demonstrate that a text-only large language model (LLM) can be adapted into a VLA-style controller when visual observations are rendered into a text input using an ASCII representation. This ASCII-as-vision interface enables existing training and deployment stacks for LLMs to efficiently condition on visual state, follow natural-language instructions, and produce constrained, executable actions. We fine-tune and compare multiple LLMs and VLMs across model families and scales, using both expert demonstrations from a planning-based teacher, as well as DAgger for iterative improvement. In a 2D manipulation benchmark, in both simulation and on a physical manipulator, the resulting controllers can identify task-relevant entities and plan feasible action sequences. Our results suggest that ASCII rendering can serve as a lightweight, interpretable modality bridge from images to text, complementing conventional VLA pipelines, and opening directions for VLA research with text-only backbones.
Chinese Translation
视觉-语言-动作（VLA）控制器通常通过在视觉-语言模型（VLMs）上扩展动作监督来构建，这依赖于具有大量数据和计算需求的多模态骨干网络。我们证明了一个仅基于文本的大型语言模型（LLM）可以通过将视觉观察渲染为ASCII表示的文本输入，适应为VLA风格的控制器。这种ASCII作为视觉接口使现有的LLM训练和部署框架能够高效地对视觉状态进行条件处理，遵循自然语言指令，并生成受限的可执行动作。我们对多个LLM和VLM进行了微调和比较，涵盖了不同的模型家族和规模，使用了来自基于规划的教师的专家演示，以及用于迭代改进的DAgger。在一个二维操作基准测试中，无论是在模拟环境中还是在物理操控器上，所得到的控制器都能够识别与任务相关的实体并规划可行的动作序列。我们的结果表明，ASCII渲染可以作为从图像到文本的轻量级、可解释的模态桥梁，补充传统的VLA流程，并为基于文本的VLA研究开辟新的方向。
cs.RO / 48 / 2606.21496
Decoupling the Declarative from the Procedural in Vision-Language-Action Models
在视觉-语言-动作模型中解耦声明性与程序性
Abstract
Deploying generalist robotic agents in the real world requires transferable skills. Specifically, a policy trained to clone a behavior from object-specific demonstrations must generalize beyond that object, otherwise data collection requirements become intractable. Recently, fine-tuning of pre-trained billion-parameter Vision-Language Models (VLMs), initially on large-scale robot datasets and then on fewer scenario-specific demonstrations, has emerged as the predominant paradigm for designing Vision-Language-Action (VLA) models. While these policies achieve state-of-the-art manipulation performance in-distribution, they remain brittle to minor spatial, semantic, and task variations. In this work, we address the inability of current models to decouple the declarative (i.e., concepts and entity semantics) from the procedural knowledge (i.e., how to do something) encoded in their parameters, which is a fundamental bottleneck for zero-shot skill transfer to novel objects. To address this, we propose w$^{2}$VLA, a new VLA model with restructured information flow. Rather than feeding all multimodal tokens from the VLM encoder into a large, opaque transformer-based action expert, our approach modulates the robot state sequence with visual, spatial, and skill information in a compositional and interpretable manner. Unlike popular, state-of-the-art VLAs, we show that our modular approach successfully decouples knowledge representations, enabling robust behavior cloning and unprecedented zero-shot skill transfer capabilities across dissimilar, unseen objects.
Chinese Translation
在现实世界中部署通用机器人代理需要可转移的技能。具体而言，训练一个政策以克隆来自特定对象演示的行为必须超越该对象进行泛化，否则数据收集的要求将变得不可处理。最近，在大规模机器人数据集上进行预训练的亿参数视觉-语言模型（Vision-Language Models, VLMs）的微调，然后在较少的场景特定演示上进行微调，已成为设计视觉-语言-动作（Vision-Language-Action, VLA）模型的主要范式。虽然这些政策在分布内达到了最先进的操作性能，但它们对微小的空间、语义和任务变化仍然脆弱。在本研究中，我们解决了当前模型无法将声明性（即概念和实体语义）与其参数中编码的程序性知识（即如何做某事）解耦的问题，这对新对象的零样本技能转移构成了根本性瓶颈。为了解决这个问题，我们提出了w$^{2}$VLA，这是一种具有重构信息流的新型VLA模型。我们的方法不是将来自VLM编码器的所有多模态标记输入到一个大型、不透明的基于变换器的动作专家中，而是以组合和可解释的方式调节机器人状态序列，结合视觉、空间和技能信息。与流行的最先进VLA不同，我们展示了我们的模块化方法成功地解耦了知识表示，使得在不同的、未见过的对象之间实现稳健的行为克隆和前所未有的零样本技能转移能力成为可能。
cs.RO / 49 / 2606.21501
UniviewVLA: A Unified Multiview Vision-Language-Action Model with World Modeling
UniviewVLA：一种具有世界建模的统一多视角视觉-语言-动作模型
Abstract
Occluded tasks remain a bottleneck in robot manipulation. Existing solutions either deploy additional physical cameras requiring training-inference camera parity, or rely on explicit 3D reconstruction with high computational cost. Moreover, both approaches rely on standard agent-view and wrist-view observations, while failing to capture occlusion information and future scene evolution. To this end, we propose UniviewVLA, a unified multiview Vision-Language-Action model with world modeling, which infers multiview scene evolution for action prediction from only standard two-camera observations. We demonstrate that by leveraging generated multiview future views from the world model, UniviewVLA reveals occluded cues and models future scene evolution, improving action prediction and removing the need for extra hardware or explicit reconstruction. Besides, to accelerate inference while preserving prediction accuracy, UniviewVLA develops Motion-Informative Token Compression, which compresses each generated view from 625 to 16 tokens and reduces per-view latency from 6-7s to 0.2-0.3s. UniviewVLA also proposes training-free Action-Entropy View Selection, which dynamically identifies the most action-informative view at different inference stages. Extensive experiments show that UniviewVLA achieves 95.8% on LIBERO and 4.60 on CALVIN ABCD to D, both standard occlusion-free benchmarks. On customized occlusion-focused tasks, it improves success rate from 40.0% to 73.3%, and average real-robot success rate by 33.4 points, demonstrating stronger occlusion-focused performance without sacrificing standard occlusion-free benchmarks.
Chinese Translation
遮挡任务仍然是机器人操作中的瓶颈。现有解决方案要么部署额外的物理摄像头，需要进行训练-推理摄像头一致性，要么依赖于显式的3D重建，计算成本高。此外，这两种方法都依赖于标准的代理视角和腕部视角观测，未能捕捉遮挡信息和未来场景演变。为此，我们提出了UniviewVLA，一种具有世界建模的统一多视角视觉-语言-动作模型，它仅通过标准的双摄像头观测推断多视角场景演变以进行动作预测。我们证明，通过利用来自世界模型生成的多视角未来视图，UniviewVLA揭示了遮挡线索并建模未来场景演变，从而提高了动作预测能力，并消除了对额外硬件或显式重建的需求。此外，为了加速推理同时保持预测准确性，UniviewVLA开发了运动信息化令牌压缩，将每个生成视图的令牌数从625压缩到16，并将每视图的延迟从6-7秒减少到0.2-0.3秒。UniviewVLA还提出了无训练的动作熵视图选择，动态识别在不同推理阶段最具动作信息的视图。大量实验表明，UniviewVLA在LIBERO上达到了95.8%的准确率，在CALVIN ABCD到D上达到了4.60，均为标准的无遮挡基准。在定制的遮挡聚焦任务中，它将成功率从40.0%提高到73.3%，并将真实机器人平均成功率提高了33.4个百分点，展示了在不牺牲标准无遮挡基准的情况下，具有更强的遮挡聚焦性能。
cs.RO / 50 / 2606.21509
A Stitch in Time Saves Nine: Preserving Policy Compatibility Under Perception Updates in End-to-End Autonomous Driving
及时缝合可省九针：在端到端自动驾驶中保持政策兼容性以应对感知更新
Abstract
End-to-end autonomous driving systems tightly couple perception and decision-making through latent representations. Consequently, updates to perception models can alter these representations and degrade the performance of downstream policies that remain fixed. Existing solutions typically rely on policy retraining or architectural decoupling, both of which incur substantial computation and validation costs. In this paper, we formulate the model stitching problem for end-to-end autonomous driving and test the hypothesis that policy compatibility can be preserved through lightweight latent-space alignment. We study low-complexity model stitching methods, including linear and convolutional stitchers, for restoring compatibility between updated perception modules and frozen downstream policy modules. Experiments demonstrate that stitching effectively preserves downstream driving behavior under diverse perception updates, including changes in random initialization, sensor configuration, and training domain. In the most challenging cross-domain setting from nuScenes to CARLA, convolutional stitching retains over 91\% of the no-shift driving score while reducing adaptation time from \SI{22.18}{h} to \SI{0.91}{h}. These results suggest that model stitching provides an effective and computationally efficient alternative to retraining or fine-tuning for maintaining end-to-end autonomous driving systems. The model will be open-sourced upon paper acceptance at https://github.com/SCP-CN-001/model-stitching to support further research and development in autonomous driving.
Chinese Translation
端到端自动驾驶系统通过潜在表示紧密耦合感知与决策。因此，对感知模型的更新可能会改变这些表示，并降低固定的下游政策的性能。现有解决方案通常依赖于政策重训练或架构解耦，这两者都需要大量的计算和验证成本。本文提出了端到端自动驾驶的模型缝合问题，并检验了通过轻量级潜在空间对齐可以保持政策兼容性的假设。我们研究了低复杂度的模型缝合方法，包括线性和卷积缝合器，以恢复更新的感知模块与冻结的下游政策模块之间的兼容性。实验表明，缝合在多种感知更新下有效保持了下游驾驶行为，包括随机初始化、传感器配置和训练领域的变化。在从 nuScenes 到 CARLA 的最具挑战性的跨领域设置中，卷积缝合保持了超过 91\% 的无偏移驾驶得分，同时将适应时间从 22.18 小时减少到 0.91 小时。这些结果表明，模型缝合为保持端到端自动驾驶系统提供了一种有效且计算上高效的替代方案，而不是重训练或微调。该模型将在论文接受后开源，网址为 https://github.com/SCP-CN-001/model-stitching，以支持自动驾驶领域的进一步研究和开发。
cs.RO / 51 / 2606.21527
LOGOS: LiDAR-Only Gaussian Elevation Splatting for Unified Tiny Obstacle Segmentation
LOGOS：仅基于激光雷达的高斯高程点云融合用于统一的小障碍物分割
Abstract
Robust obstacle segmentation is essential for the safety of intelligent robots, where LiDAR-based perception systems play a fundamental role in the robot-environment interaction. While extensive LiDAR-based approaches have demonstrated high performance on common obstacles in urban scenarios, their results on tiny obstacles such as curbs, gravel, and potholes remain unsatisfactory due to the significant similarity between tiny obstacles and inherent road undulations. Moreover, their segmentation accuracy even deteriorates sharply when the LiDAR scans suffer from degradation in challenging off-road scenes. To overcome these bottlenecks, we propose LOGOS, a LiDAR-only unified tiny obstacle segmentation system, which models the road surface as a continuous mixture of 2D Gaussian primitives and distinguishes tiny obstacles via high-presicion elevation estimation. Unlike existing Gaussian splatting methods that rely on iterative RGB training, LOGOS is a backpropagation-free LiDAR-only approach. It directly estimates Gaussian parameters via a freespace-aware initialization by incrementally pruning non-road primitives using smoothness constraints. Subsequently, pointwise signed distances are computed via a novel normal-aware elevation splatting function, ensuring robustness to both flat and sloped terrains. We evaluate LOGOS on a highly heterogeneous benchmark of point cloud frames collected from urban mobility scenarios and mining haulage off-road environments. These data are practically acquired using different LiDAR sensors and exhibit large variations in point density, terrain roughness, and obstacle types. Experiments on the road and off-road scenes demonstrate that LOGOS significantly outperforms other state-of-the-art methods, particularly in degraded point cloud regions and challenging off-road scenarios, while maintaining real-time efficiency.
Chinese Translation
稳健的障碍物分割对于智能机器人的安全至关重要，其中基于激光雷达的感知系统在机器人与环境的交互中发挥着基础性作用。尽管大量基于激光雷达的方法在城市场景中的常见障碍物上表现出高性能，但对于小障碍物（如路缘石、碎石和坑洼）的结果仍然不尽如人意，因为小障碍物与固有的道路起伏之间存在显著的相似性。此外，当激光雷达扫描在具有挑战性的越野场景中遭受退化时，其分割精度甚至会急剧下降。为了解决这些瓶颈，我们提出了LOGOS，一个仅基于激光雷达的统一小障碍物分割系统，该系统将路面建模为2D高斯基元的连续混合，并通过高精度的高程估计来区分小障碍物。与现有依赖于迭代RGB训练的高斯点云融合方法不同，LOGOS是一种无反向传播的仅基于激光雷达的方法。它通过使用平滑性约束逐步修剪非道路基元，直接通过自由空间感知初始化来估计高斯参数。随后，通过一种新颖的法向感知高程点云融合函数计算逐点有符号距离，确保在平坦和倾斜地形上的稳健性。我们在从城市移动场景和矿山运输越野环境中收集的高度异质的点云帧基准上评估LOGOS。这些数据是通过不同的激光雷达传感器实际获取的，表现出点密度、地形粗糙度和障碍物类型的巨大变化。在道路和越野场景中的实验表明，LOGOS显著优于其他最先进的方法，特别是在退化的点云区域和具有挑战性的越野场景中，同时保持实时效率。
cs.RO / 52 / 2606.21572
Robot Critics that Sweat the Small Stuff
关注细节的机器人评估者
Abstract
Large vision-language models contain several priors about the world and object interactions, making them useful critics during inference to steer robot policies towards success. However, closed-loop robot manipulation requires judging small visual differences between success and failure, which remains a challenge for current VLMs. We introduce a method to fine-tune critics by constructing pairwise progress supervision using success and failure rollouts obtained from a policy. Our fine-tuned critic excels at fine-grained progress reasoning and subtle failure detection, outperforming prior progress reasoning baselines. Additionally, we use an action-conditioned video model to predict the visual effect of several candidate actions sampled from a policy, and show that our critic can correctly identify successful candidates to execute, improving the average policy success rate by 11% across real-world tasks and 5.9% across simulation tasks.
Chinese Translation
大型视觉-语言模型包含关于世界和物体交互的多个先验知识，使其在推理过程中能够作为有效的评估者，引导机器人策略走向成功。然而，闭环机器人操作需要判断成功与失败之间细微的视觉差异，这对当前的视觉-语言模型（VLMs）仍然是一个挑战。我们提出了一种通过构建基于成功和失败回放的成对进展监督来微调评估者的方法。我们的微调评估者在细粒度进展推理和细微失败检测方面表现优异，超越了之前的进展推理基准。此外，我们使用一个基于动作条件的视频模型来预测从策略中采样的多个候选动作的视觉效果，并展示我们的评估者能够正确识别出成功的候选动作，从而在现实世界任务中将平均策略成功率提高了11%，在模拟任务中提高了5.9%。
cs.RO / 53 / 2606.21577
Conflict-Aware Switching for CBF-CLF-Based Multi-Goal Navigation
基于控制障碍函数和控制李雅普诺夫函数的多目标导航中的冲突感知切换
Abstract
Quadratic programs (QPs) using Control Barrier Functions (CBFs) and Control Lyapunov Functions (CLFs) are widely used for safe control in reach-and-avoid navigation. However, the inherently conflicting nature of CBF and CLF constraints can lead to performance degradation, including slowdowns and deadlocks. This issue is exacerbated in multi-goal scenarios, where multiple nominal control objectives must be satisfied under shared safety constraints. Existing approaches for preemptive safety are often computationally expensive or overly conservative, while methods that relax or switch between nominal objectives are not well-suited for sequential goal-to-goal navigation. To address these limitations, we propose a conflict-aware switching strategy that detects high-conflict conditions and switches between available nominal control objectives to reduce constraint conflict. We apply this approach to multi-agent, multi-goal reach-and-avoid scenarios under CBF-CLF-QP control. Compared to a baseline sequential goal traversal strategy, our method reduces both completion time and timeout rates, demonstrating improved performance in satisfying all nominal control objectives while respecting safety constraints.
Chinese Translation
使用控制障碍函数（CBFs）和控制李雅普诺夫函数（CLFs）的二次规划（QPs）广泛应用于安全控制的到达与避让导航。然而，CBF和CLF约束之间固有的冲突特性可能导致性能下降，包括减速和死锁。在多目标场景中，这一问题尤为严重，因为必须在共享安全约束下满足多个名义控制目标。现有的预防性安全方法通常计算开销大或过于保守，而放宽或在名义目标之间切换的方法并不适合顺序目标导航。为了解决这些局限性，我们提出了一种冲突感知切换策略，该策略能够检测高冲突条件并在可用的名义控制目标之间切换，以减少约束冲突。我们将该方法应用于基于CBF-CLF-QP控制的多智能体、多目标的到达与避让场景。与基线顺序目标遍历策略相比，我们的方法在满足所有名义控制目标的同时，减少了完成时间和超时率，展示了在遵守安全约束的情况下的性能提升。
cs.RO / 54 / 2606.21600
VQActFlow: Vector-Quantized Action Mode Steering for Multi-Task Robot Manipulation
VQActFlow：用于多任务机器人操作的向量量化动作模式引导
Abstract
Multi-task robot manipulation policies are challenging to learn from demonstration because traditionally a single network must select among qualitatively different action modes from a multimodal demonstration distribution, conditioned on language and visual context. A wrong mode selection means executing the wrong task or an action infeasible in the scene. Tokenizing continuous actions into a learned discrete codebook separates these modes at the representation level, offering structural advantages for multi-task learning. We propose VQActFlow, a multi-task manipulation policy that tokenizes action chunks and generates code sequences via Variational Flow Matching. VQActFlow maintains an explicit preference over action modes throughout generation. Inference-time guidance acts on this preference to steer mode commitment. We instantiate this with classifier-free guidance over language conditioning, which steers the policy toward the instructed action mode, and a learned codebook critic that supplies a complementary feasibility signal. We evaluate VQActFlow on three platforms: the LIBERO simulation benchmarks, a Unitree G1 humanoid performing whole-body pick-and-place, and an ALOHA-style bimanual platform performing contact-rich tasks. Across these benchmarks, VQActFlow outperforms both continuous and discrete baselines.
Chinese Translation
从演示中学习多任务机器人操作策略具有挑战性，因为传统上单一网络必须在多模态演示分布中选择不同的动作模式，这些选择依赖于语言和视觉上下文。错误的模式选择意味着执行错误的任务或在场景中不可行的动作。将连续动作标记为学习到的离散代码本在表征层面上分离了这些模式，为多任务学习提供了结构性优势。我们提出了VQActFlow，这是一种多任务操作策略，通过变分流匹配（Variational Flow Matching）对动作块进行标记并生成代码序列。VQActFlow在生成过程中保持对动作模式的明确偏好。推理时的引导作用于这一偏好，以引导模式承诺。我们通过无分类器引导（classifier-free guidance）与语言条件相结合来实现这一点，从而引导策略朝向指令的动作模式，并使用学习到的代码本评估器提供补充的可行性信号。我们在三个平台上评估了VQActFlow：LIBERO仿真基准、执行全身抓取和放置的Unitree G1人形机器人，以及执行接触丰富任务的ALOHA风格双手平台。在这些基准测试中，VQActFlow的表现优于连续和离散基线。
cs.RO / 55 / 2606.21646
Energy-based Compositional Diffusion Planning
基于能量的组合扩散规划
Abstract
Compositional diffusion planners aim to solve long-horizon robotic tasks using short training trajectories. Yet, current approaches often rely on the heuristic stitching of local predictions. We show that the resulting stitched update is generally a non-conservative field} that does not mathematically correspond to any valid global trajectory log-density function. We propose Energy-based Compositional Diffuser (ECD), a framework that formulates the global trajectory as the minimizer of the sum of local bridge potentials. This energy-based perspective defines a conservative correction field and contains a boundary reaction term that heuristic stitching omits. To enable efficient inference, we further introduce a Markov-based score approximation that computes the reaction term via a single block-tridiagonal solve, maintaining time complexity linear in the planning horizon. Empirically, ECD achieves state-of-the-art success rates on a range of OGBench stitching tasks, while nearly matching the inference speed of heuristic stitching methods. Code is available at https://github.com/GradientSpaces/ECD.
Chinese Translation
组合扩散规划器旨在利用短期训练轨迹解决长时间范围的机器人任务。然而，目前的方法往往依赖于局部预测的启发式拼接。我们表明，所得到的拼接更新通常是一个非保守场，并且在数学上与任何有效的全局轨迹对数密度函数不对应。我们提出了基于能量的组合扩散器（Energy-based Compositional Diffuser, ECD），该框架将全局轨迹公式化为局部桥接势能之和的最小化者。这种基于能量的视角定义了一个保守的修正场，并包含一个启发式拼接所忽略的边界反应项。为了实现高效推理，我们进一步引入了一种基于马尔可夫的得分近似，通过单个块三对角求解计算反应项，保持时间复杂度与规划时间范围线性相关。在实证上，ECD在一系列OGBench拼接任务中实现了最先进的成功率，同时几乎匹配了启发式拼接方法的推理速度。代码可在 https://github.com/GradientSpaces/ECD 获取。
cs.RO / 56 / 2606.21669
Online Learning of Robust Legged Odometry with Minimal Exteroceptive Supervision
最小外部监督下的鲁棒腿部里程计在线学习
Abstract
Robust locomotion and navigation for legged robots relies heavily on dependable odometry. Traditional multi-sensor fusion for such state estimation requires meticulous sensor calibration and platform-specific kinematic modeling, which complicates deployment. Industrially packaged exteroceptive sensors can provide accurate motion tracking but remain vulnerable to perceptually degraded conditions. We thus develop a plug-and-play, robust legged odometry system that eliminates the need for explicit exteroceptive-to-proprioceptive calibration or system kinematic modeling. Our approach leverages established exteroceptive motion pipelines as a continuous supervisory signal to train an online learned velocity neural network directly from proprioceptive data. An Invariant EKF (InEKF) is then used to fuse the learned proprioceptive or exteroceptive velocity (if any) and IMU data. When exteroception fails due to environmental degradation, the system seamlessly falls back to using the learned proprioceptive model, yielding a resilient legged odometry that readily adapts to new hardware. We demonstrate the platform-agnostic, easily deployable nature of our approach on different quadruped platforms, showcasing promising results in maintaining robust motion estimation across challenging scenarios.
Chinese Translation
腿部机器人稳健的运动和导航在很大程度上依赖于可靠的里程计。传统的多传感器融合用于这种状态估计需要精细的传感器校准和特定平台的运动学建模，这使得部署变得复杂。工业化包装的外部传感器可以提供准确的运动跟踪，但在感知受损的条件下仍然脆弱。因此，我们开发了一种即插即用的鲁棒腿部里程计系统，消除了对显式外部到本体感知校准或系统运动学建模的需求。我们的方法利用已建立的外部运动管道作为连续的监督信号，直接从本体感知数据训练在线学习的速度神经网络。然后使用不变扩展卡尔曼滤波器（Invariant EKF, InEKF）融合学习到的本体感知或外部速度（如果有的话）和IMU数据。当外部感知因环境退化而失效时，系统无缝地回退到使用学习到的本体感知模型，从而产生一种鲁棒的腿部里程计，能够轻松适应新硬件。我们在不同的四足平台上展示了我们方法的平台无关性和易于部署的特性，展示了在具有挑战性的场景中维持鲁棒运动估计的良好结果。
cs.RO / 57 / 2606.21672
Imitation from Heterogeneous Demonstrations using Grounded Latent-Action World Models
利用基于预测的潜在动作世界模型进行异质示范的模仿学习
Abstract
Imitation learning has emerged as a powerful paradigm for learning visuomotor policies, but its generalisation and stability are limited by the scale and quality of demonstration data needed. A promising direction is to leverage more abundant but heterogeneous data sources, which differ in action space and often lack action labels altogether. Existing co-training approaches that combine heterogeneous data sources rely on heuristic and hand-engineered alignment techniques. In contrast, we argue that action representations should be grounded in prediction: actions that produce the same effect on the environment should share the same representation, regardless of their sources. To this end, we instantiate this principle by using a grounded latent-action world model (GLAM), a pair of generative models with a shared latent action space across data sources that is grounded by predicting future observations consistently across sources. This latent action space is used to train downstream behavioural cloning (BC) policies which map observations to latent actions and decode them back to robot actions, providing a paradigm for learning from heterogeneous data. Empirically, we demonstrate that GLAM successfully learns an aligned latent action space that facilitates action transfer across data sources with and without action labels. Across five manipulation tasks in simulation and in the real world, GLAM-aligned policies significantly outperform BC baselines and prior latent-action methods, achieving an average of +48% improvement in task success rate with the same data-scarce setting. Videos and code are available at https://viccccciv.github.io/glam/.
Chinese Translation
模仿学习已成为学习视觉运动策略的强大范式，但其泛化能力和稳定性受到所需示范数据的规模和质量的限制。一个有前景的方向是利用更丰富但异质的数据源，这些数据源在动作空间上存在差异，且通常完全缺乏动作标签。现有的联合训练方法依赖于启发式和手工设计的对齐技术，以结合异质数据源。相反，我们认为动作表示应基于预测：对环境产生相同效果的动作应共享相同的表示，无论其来源如何。为此，我们通过使用基于预测的潜在动作世界模型（GLAM）来实现这一原则，该模型是一对生成模型，具有跨数据源共享的潜在动作空间，并通过一致地预测未来观察结果来实现对齐。这个潜在动作空间用于训练下游行为克隆（BC）策略，将观察映射到潜在动作，并将其解码回机器人动作，从而提供了一种从异质数据中学习的范式。在实证研究中，我们证明了GLAM成功学习了一个对齐的潜在动作空间，促进了跨数据源的动作转移，无论是否有动作标签。在五个仿真和现实世界的操作任务中，GLAM对齐的策略显著优于BC基线和先前的潜在动作方法，在相同的数据稀缺设置下，任务成功率平均提高了48%。视频和代码可在 https://viccccciv.github.io/glam/ 获取。
cs.RO / 58 / 2606.21716
Shape-programmable Magnetic Soft Membranes for Mechanically Active Microchannels
形状可编程磁性柔性膜用于机械活性微通道
Abstract
The capability to encode spatially distinct magnetization patterns within soft materials enables remote control over complex deformations. This characteristic is especially important for microfluidic platforms, where limited dynamic control of channel boundaries and laminar flow conditions usually restrict fluid transport and interactions. The present study introduces a shape-programmable magnetic soft membrane actuator as an active microchannel component that can dynamically modulate its shape under magnetic fields and therefore the microfluidic environment. The membrane is magnetically programmed using a template-based approach, which allows it to be controllably deformed in the form of a sinusoid under the influence of an external magnetic field. The membrane's integration into a microchannel converts a passive channel wall into a dynamically changeable interface, allowing active fluid manipulation and enhancing micromixing in laminar flow conditions. The proposed approach establishes a versatile platform for wirelessly controlled deformable interfaces in next-generation microfluidic and lab-on-chip systems.
Chinese Translation
在柔性材料中编码空间上不同的磁化模式的能力使得对复杂变形的远程控制成为可能。这一特性对于微流体平台尤为重要，因为通道边界和层流条件的动态控制有限，通常会限制流体的输送和相互作用。本研究提出了一种形状可编程的磁性柔性膜驱动器，作为一种能够在磁场下动态调节形状的活性微通道组件，从而改变微流体环境。该膜采用基于模板的方法进行磁性编程，使其在外部磁场的影响下可控地变形为正弦波形。将膜集成到微通道中，将被动的通道壁转变为动态可变的界面，允许主动流体操控，并在层流条件下增强微混合。所提出的方法为下一代微流体和芯片实验系统中的无线控制可变界面建立了一个多功能平台。
cs.RO / 59 / 2606.21737
Programmable magnetic soft robots with controlled locomotion and directional liquid cargo release
可编程磁性软机器人具有可控的运动和定向液体货物释放
Abstract
Magnetically programmable soft elastomers enable complex shape morphing and locomotion dynamics in small scale soft robots under external magnetic fields. Benefiting from their programmed deformation and wireless actuation capabilities, magnetic soft robots have emerged as promising platforms for targeted drug delivery, especially in human gastrointestinal tract. However, achieving controlled directional liquid cargo release toward desired tissue interface while preserving the encoded shape morphing and locomotion capabilities remain a significant challenge. Here, we report a new design strategy that employs an optimized magnetization profile to enable controlled directional release of aqueous cargo without compromising shape morphing and locomotion capabilities. Magnetic soft robots with a specific spatially distributed magnetization profile allow directional alignment of the release interface with the orientation of the external magnetic field. This orientation control ensures active alignment of the release interface toward the intestinal wall prior to drug release. An interconnected microporous elastomer is embedded within the robot for aqueous cargo storage, while a thin microcrystalline wax layer seals the release opening hole to isolate the stored liquid cargo from external environment during transport. Triggered release is achieved by mechanically rupturing the wax sealing layer under a higher magnitude external magnetic field. Controlled directional flipping, locomotion, and triggered release are decoupled through external magnetic field's direction and strength. The controlled directional release strategy reported here integrates directional targeted liquid cargo release, shape morphing, and locomotion, which establishes the groundwork for target drug delivery in gastrointestinal tract applications.
Chinese Translation
磁性可编程软弹性体使得在外部磁场下的小型软机器人能够实现复杂的形状变形和运动动态。得益于其编程变形和无线驱动能力，磁性软机器人已成为靶向药物输送的有前景平台，特别是在人体胃肠道中。然而，实现朝向所需组织界面的可控定向液体货物释放，同时保持编码的形状变形和运动能力，仍然是一个重大挑战。在此，我们报告了一种新的设计策略，采用优化的磁化分布，以实现水性货物的可控定向释放，而不影响形状变形和运动能力。具有特定空间分布的磁化特征的磁性软机器人允许释放界面与外部磁场的方向进行定向对齐。这种方向控制确保在药物释放之前，释放界面能够主动对准肠壁。一个互连的微孔弹性体嵌入机器人内部，用于储存水性货物，同时一层薄薄的微晶蜡层封闭释放开口，以在运输过程中将储存的液体货物与外部环境隔离。通过在更高强度的外部磁场下机械性破坏蜡封层，实现触发释放。可控的定向翻转、运动和触发释放通过外部磁场的方向和强度进行解耦。这里报告的可控定向释放策略整合了定向靶向液体货物释放、形状变形和运动，为胃肠道应用中的靶向药物输送奠定了基础。
cs.RO / 60 / 2606.21755
RoverDevKit: An open, physics-grounded tradespace toolkit for conceptual design of lunar micro-rovers
RoverDevKit：一个开放的、基于物理的交易空间工具包，用于月球微型探测器的概念设计
Abstract
Pre-Phase-A design of lunar micro-rovers is dominated by tightly coupled mobility, power, thermal, and mass trades, yet conceptual-design tooling for the rapidly growing sub-50 kg class is typically proprietary, weakly benchmarked, or too slow to drive optimization. We contribute RoverDevKit, an open analytical evaluator coupling terramechanics, mass, power, thermal survival, and traverse that runs in 30ms per mission, fast enough to serve directly as a multi-objective optimizer's fitness function. Across mare, polar, highland, and crater-rim scenarios, NSGA-II Pareto fronts show that the binding design trade changes with mission profile within a single mass class: energy storage dominates at high latitude, slope traction on loose highland regolith, and traverse range on mare and crater-rim missions. Notably, rigid four-wheel layouts Pareto-dominate the full modeled mass range under smooth-regolith range-mass-slope objectives, contrary to the expectation that six-wheel architectures become optimal at heavier masses; six-wheel rocker-bogie layouts enter the Pareto set only once missions impose an obstacle-navigation requirement. The evaluator performance is benchmarked using both component and system checks: the terramechanics kernel matches measured single-wheel drawbar pull within the literature model-form band on two independent datasets, the bottom-up mass model predicts published in-class (5-50 kg) rover masses to 13.3% median absolute error, and a rediscovery check places real micro-rovers near the optimizer's fronts. Propagating the measured terramechanics error through the optimizer leaves the qualitative design rules unchanged. The tool, data, validation artifacts, and figure-generation scripts are released openly.
Chinese Translation
月球微型探测器的前相位A设计受限于紧密耦合的机动性、功率、热量和质量之间的权衡，然而，针对快速增长的50公斤以下类别的概念设计工具通常是专有的、基准测试不足或优化速度过慢。我们贡献了RoverDevKit，一个开放的分析评估器，结合了土力学、质量、功率、热生存和行进能力，能够在每次任务中以30毫秒的速度运行，足够快以直接作为多目标优化器的适应度函数。在月海、极地、高地和陨石坑边缘场景中，NSGA-II帕累托前沿显示，在单一质量类别内，绑定设计权衡随着任务配置的变化而变化：在高纬度地区，能量存储占主导地位；在松散的高地表面，坡度牵引力占优；而在月海和陨石坑任务中，行进范围则是关键。值得注意的是，在平滑的土壤范围-质量-坡度目标下，刚性四轮布局在整个建模质量范围内占据帕累托优势，这与六轮结构在较重质量下变得最优的预期相悖；六轮摇臂布局仅在任务要求障碍物导航时才进入帕累托集合。评估器的性能通过组件和系统检查进行了基准测试：土力学内核在两个独立数据集上与文献模型范围内的单轮牵引力测量相匹配，底层质量模型对已发布的同类（5-50公斤）探测器质量的预测中位绝对误差为13.3%，而重新发现检查将真实微型探测器置于优化器的前沿附近。通过优化器传播测得的土力学误差不会改变定性设计规则。该工具、数据、验证文档和图形生成脚本均已公开发布。
cs.RO / 61 / 2606.21771
A Novel Bio-Inspired Fish Robot with Tunable Stiffness via Particle Jamming
一种新型的生物启发鱼类机器人，通过颗粒堵塞实现可调刚度
Abstract
Fish achieve efficient swimming across varied speeds through active modulation of their body flexibility. To explore the effects of tunable stiffness on swimming performance, we present a bio-inspired freely swimming fish robot with a rapidly tunable particle-jamming body. This design enables rapid stiffness adjustments with negligible changes in shape or volume, achieving a 54% variation in flexural rigidity across vacuum pressures of 0 to -40 kPa. We visualize the midline of the oscillating body under both low- and high-stiffness conditions, and the comparison confirms that the body curvature varies with stiffness. We further experimentally evaluate the tunable stiffness body's effects on swimming performance using velocity and cost of transport (CoT) measurements obtained via a motion tracking system. Results show that active stiffness tuning is essential for sustaining efficient and high-speed swimming across beating frequencies of 1-3 Hz. At low frequencies (1-1.5 Hz), a softer body (0 kPa) maximizes velocity and minimizes CoT, whereas at high frequencies (2.5-3 Hz), a stiffer body (-40 kPa) delivers superior velocity and reduced transport cost. These findings highlight stiffness modulation as a key strategy for adaptive and efficient propulsion in bio-inspired robotic swimmers.
Chinese Translation
鱼类通过主动调节身体柔韧性在不同速度下实现高效游泳。为了探索可调刚度对游泳性能的影响，我们提出了一种具有快速可调颗粒堵塞体的生物启发自由游泳鱼类机器人。该设计能够在形状或体积变化极小的情况下快速调整刚度，实现了在0到-40 kPa的真空压力下弯曲刚度的54%变化。我们可视化了在低刚度和高刚度条件下振荡体的中线，并通过比较确认了身体曲率随刚度变化。我们进一步通过运动跟踪系统对可调刚度体对游泳性能的影响进行了实验评估，使用速度和运输成本（CoT）测量。结果表明，主动刚度调节对于在1-3 Hz的拍打频率下维持高效和高速游泳至关重要。在低频（1-1.5 Hz）下，较软的身体（0 kPa）最大化速度并最小化CoT，而在高频（2.5-3 Hz）下，较硬的身体（-40 kPa）则提供了更高的速度和更低的运输成本。这些发现突显了刚度调节作为生物启发机器人游泳者适应性和高效推进的关键策略。
cs.RO / 62 / 2606.21788
Rotation-Aware Point-Cloud Embeddings for Vision-Based In-Hand Reorientation
基于视觉的手内重定向的旋转感知点云嵌入
Abstract
Point-cloud goals provide a direct way to specify dexterous in-hand reorientation: instead of defining an object-specific pose frame or estimating 6D pose at test time, the policy is given the desired 3D geometry of the object. Yet raw point-cloud goal conditioning is poorly conditioned for policy learning. Current and goal clouds are unordered, independently sampled, and often visibility-dependent, so their discrepancy entangles object rotation with permutation, resampling, and unstable correspondence structure. For this reason, prior point-cloud manipulation methods typically add structure outside the representation itself, such as explicit pose or relative-pose inputs, dense flow features, or distillation from privileged teachers. We close this gap by learning a rotation-aware point-cloud embedding whose Euclidean latent distance is calibrated to the SO(3) geodesic error between object orientations. The resulting representation turns current-goal comparison into a smooth control signal, allowing a model-free RL policy to act from current and goal point-cloud embeddings, proprioception, and centroid metadata, without object pose, relative pose, dense flow, or teacher-action supervision. In in-hand reorientation experiments, this interface matches privileged-state and distillation-based baselines while avoiding brittle test-time computation of structured pose or flow inputs. These results suggest that point-cloud goals become practical for this task when the representation, rather than an external module, encodes the task-relevant geometry of rotation. We also show evidence that generic visual point-cloud pretraining is insufficient for such a current-goal comparison because it discards the task-relevant state and preserves only shape features.
Chinese Translation
点云目标提供了一种直接指定灵巧手内重定向的方法：政策不再需要定义特定对象的姿态框架或在测试时估计6D姿态，而是直接给出对象的期望3D几何形状。然而，原始点云目标条件对于策略学习来说条件较差。当前和目标点云是无序的、独立采样的，并且通常依赖于可见性，因此它们之间的差异将对象旋转与排列、重采样和不稳定的对应结构纠缠在一起。因此，之前的点云操作方法通常在表示本身之外添加结构，例如显式姿态或相对姿态输入、密集流特征或来自特权教师的蒸馏。我们通过学习一种旋转感知的点云嵌入来填补这一空白，其欧几里得潜在距离经过校准，以匹配对象方向之间的SO(3)测地误差。由此产生的表示将当前与目标的比较转化为平滑的控制信号，使得无模型强化学习政策能够基于当前和目标点云嵌入、运动觉和质心元数据进行操作，而无需对象姿态、相对姿态、密集流或教师动作监督。在手内重定向实验中，该接口与特权状态和基于蒸馏的基线相匹配，同时避免了在测试时计算结构化姿态或流输入的脆弱性。这些结果表明，当表示而非外部模块编码与旋转相关的任务几何时，点云目标在此任务中变得实用。我们还展示了通用视觉点云预训练对于这样的当前-目标比较是不够的，因为它丢弃了与任务相关的状态，仅保留形状特征。
cs.RO / 63 / 2606.21792
THREAD: Trajectory Planning for Hybrid Rigid-Soft Manipulators with Environment-Aware Diffusion
THREAD：具有环境感知扩散的混合刚性-柔性操纵器轨迹规划
Abstract
Manipulation in confined environments, such as threading a manipulator through narrow apertures, remains a fundamental challenge, especially for conventional rigid robots. Hybrid rigid-soft manipulators offer promise but face two compounding planning challenges: backbone shapes feasible in free space become infeasible under environmental contact, and planning rigid and soft segments independently ignores their kinematic coupling. We present THREAD, the first diffusion-based trajectory planner for hybrid manipulation, learning a generative prior over physically realizable backbone trajectories conditioned on local environment geometry, with physics-inspired losses encoding curvature, smoothness, and collision constraints jointly across both segments. Trained in simulation, THREAD achieves 92.4% task success with 5x fewer collisions than the strongest baseline. We show cross-embodiment real-world transfer with minimal online updates, successfully threading through apertures as small as 1.3x the soft segment diameter.
Chinese Translation
在狭窄环境中进行操控，例如将操纵器穿过狭小的开口，仍然是一个基本挑战，尤其对于传统的刚性机器人而言。混合刚性-柔性操纵器展现出潜力，但面临两个复合的规划挑战：在自由空间中可行的主干形状在与环境接触时变得不可行，而独立规划刚性和柔性部分则忽视了它们的运动学耦合。我们提出了THREAD，这是首个基于扩散的混合操控轨迹规划器，学习在局部环境几何条件下的物理可实现主干轨迹的生成先验，采用受物理启发的损失函数共同编码两个部分的曲率、平滑性和碰撞约束。在仿真中训练后，THREAD实现了92.4%的任务成功率，碰撞次数比最强基线少5倍。我们展示了跨体现的真实世界迁移，经过最小的在线更新，成功穿过直径仅为柔性部分1.3倍的小开口。
cs.RO / 64 / 2606.21866
SurGE: Surrogate Gradient-guided Evolution for Co-design of Legged Robots with Parallel Elasticity
SurGE：基于代理梯度引导的腿式机器人并行弹性协同设计
Abstract
Co-design of legged robots with elastic elements is challenging due to the non-differentiability of contact dynamics and mechanism engagement. This paper presents SurGE, a framework that computes surrogate gradients of the design objective through a differentiable pipeline consisting of a kinodynamic single-rigid-body (Kino-SRB) model and a design-aware control policy, and injects them into CMA-ES via mean shift with cosine-annealed step decay. On a 4-DOF design space of a hopping robot with unidirectional parallel spring, SurGE achieves 6 times lower cross-seed standard deviation and 18% tighter population concentration compared to vanilla CMA-ES, while matching or improving the best objective. Hardware experiments on a 2D design subspace show that, starting from a hand-tuned initial design, SurGE reduces the design objective by 37.65% on hardware, with the improvement trend identified in simulation transferring consistently to the physical system. SurGE provides the potential to accelerate non-differentiable co-design problems in legged robots via surrogate model gradients.
Chinese Translation
腿式机器人与弹性元件的协同设计面临挑战，因为接触动力学和机制耦合的非可微性。本文提出了SurGE，一个通过包含运动动力学单刚体（Kino-SRB）模型和设计感知控制策略的可微管道计算设计目标的代理梯度的框架，并通过均值偏移与余弦退火步长衰减将其注入CMA-ES。在具有单向并行弹簧的跳跃机器人4自由度设计空间中，SurGE实现了比传统CMA-ES低6倍的交叉种子标准差和18%的种群集中度提升，同时匹配或改善了最佳目标。在二维设计子空间的硬件实验表明，从手动调优的初始设计开始，SurGE在硬件上将设计目标降低了37.65%，且在仿真中识别的改进趋势在物理系统中一致转移。SurGE为通过代理模型梯度加速腿式机器人中的非可微协同设计问题提供了潜力。
cs.RO / 65 / 2606.21920
Predictive Gaze Is Preserved but Reorganized toward Monitoring during Robot-Mediated Manipulation
预测性注视在机器人介导的操作中得以保留但重新组织以适应监控
Abstract
Goal-directed eye movements are a fundamental component of visuomotor control, enabling humans to anticipate and guide their actions. Whether this anticipatory and task-driven behavior is preserved when actions are executed through a robot rather than through one's own body remains unclear. Here we address this question by investigating gaze behavior during goal-directed telemanipulation to determine how visuomotor control adapts to altered embodiment. Our findings show that gaze remains strongly aligned with task goals, preserving its predictive role even during robot-mediated manipulation. At the same time, teleoperation systematically redistributes visual attention toward the robotic end-effector and manipulated objects, increasing online monitoring. These findings show that predictive gaze is not lost under altered embodiment, but reorganized in response to changes in sensory feedback and control demands. More broadly, they reveal the flexibility of the human visuomotor system when the natural sensorimotor coupling is disrupted and identify gaze as an informative signal for inferring action intentions in human-robot interaction.
Chinese Translation
目标导向的眼动是视觉运动控制的基本组成部分，使人类能够预测和引导自己的动作。当通过机器人而非自身身体执行动作时，这种预测性和任务驱动的行为是否得以保留仍不清楚。本文通过研究目标导向的远程操作中的注视行为，探讨视觉运动控制如何适应改变的身体感知。我们的研究发现，注视仍然与任务目标紧密对齐，即使在机器人介导的操作中也保留了其预测性角色。同时，远程操作系统性地将视觉注意力重新分配到机器人末端执行器和被操作物体上，增加了在线监控。这些发现表明，在改变的身体感知下，预测性注视并未丧失，而是根据感官反馈和控制需求的变化进行了重新组织。更广泛地说，这些结果揭示了当自然的感觉运动耦合被打破时，人类视觉运动系统的灵活性，并将注视识别为推断人机交互中行动意图的重要信号。
cs.RO / 66 / 2606.21935
CoRDE: Concept-Prior Routed Diffusion Experts for Structural Generalization in Robot Manipulation
CoRDE：用于机器人操作中结构泛化的概念优先路由扩散专家
Abstract
Diffusion models excel at capturing multi-modal action distributions in robot imitation learning. However, in multi-task and long-horizon scenarios, monolithic architectures lack structural generalization capabilities, suffering from gradient conflicts between distinct semantic sub-stages. While pure data-driven Mixture-of-Experts (MoE) methods introduce labor division, they frequently trigger routing collapse, and instantiating full-scale experts causes parameter explosion and high expansion costs. To address these issues, we propose Concept-prior Routed Diffusion Experts (CoRDE), a structure-guided variational distillation framework. CoRDE extracts semantic distributions from a frozen concept encoder to guide the variational posterior responsibility via a learnable soft mapping matrix. This mechanism introduces an entropy-controlled responsibility inference process that encourages confident routing under reliable semantic predictions while preserving the stochastic diffusion term for behavioral diversity. To overcome parameter inflation, CoRDE employs a parameter-efficient expert pool using Low-Rank Adaptation (LoRA) on a shared frozen backbone. Theoretical analysis shows that the mixture score discrepancy is bounded by responsibility-weighted local expert errors, supporting high-fidelity generation under low-rank expert adaptation. Empirical evaluations confirm that, compared to existing baselines, CoRDE systematically reduces routing collapse, forming robust, semantically aligned expert allocations while achieving superior action quality and incremental learning efficiency.
Chinese Translation
扩散模型在机器人模仿学习中擅长捕捉多模态动作分布。然而，在多任务和长时间跨度的场景中，单一架构缺乏结构泛化能力，面临不同语义子阶段之间的梯度冲突。虽然纯数据驱动的专家混合（Mixture-of-Experts, MoE）方法引入了劳动分工，但它们经常引发路由崩溃，并且实例化全规模专家会导致参数膨胀和高昂的扩展成本。为了解决这些问题，我们提出了概念优先路由扩散专家（Concept-prior Routed Diffusion Experts, CoRDE），这是一种结构引导的变分蒸馏框架。CoRDE 从冻结的概念编码器中提取语义分布，通过可学习的软映射矩阵引导变分后验责任。该机制引入了一种熵控制的责任推断过程，鼓励在可靠的语义预测下进行自信路由，同时保留行为多样性的随机扩散项。为了克服参数膨胀，CoRDE 使用低秩适应（Low-Rank Adaptation, LoRA）在共享的冻结主干上采用参数高效的专家池。理论分析表明，混合得分差异受到责任加权的局部专家误差的限制，支持在低秩专家适应下的高保真生成。实证评估证实，与现有基线相比，CoRDE 系统性地减少了路由崩溃，形成了稳健且语义对齐的专家分配，同时实现了更优的动作质量和增量学习效率。
cs.RO / 67 / 2606.22027
RARM: Confidence-Gated Progress Reward Modeling for RL in Manipulation
RARM：用于操作中的强化学习的信心门控进展奖励建模
Abstract
Reinforcement learning for robot manipulation is often bottlenecked by reward design, especially in long-horizon tasks: sparse success rewards provide weak supervision, while hand-crafted dense rewards are tedious to design and generalize poorly across tasks. Progress-based reward models offer a promising alternative by estimating how far an observation has advanced toward task completion, but existing approaches often require task-specific demonstrations or progress labels, and can assign high rewards to visually plausible but physically incorrect states. We introduce the Reference-Anchored Reward Model (RARM), a lightweight visual comparator that converts a single successful demonstration into a dense, progress-aware reward. RARM is trained once on general-purpose videos with a contrastive temporal objective, requiring no robot-specific data, task-specific reward labels, or per-task reward engineering. At deployment, RARM matches rollout clips to reference clips and rewards only confident forward progress, suppressing uncertain matches that may otherwise produce false-positive rewards. Across 9 simulated manipulation tasks from LIBERO and MetaWorld and 4 real-world tasks, RARM achieves the best overall success rates in subsequent RL training, with particularly large gains on long-horizon tasks such as cloth folding, where unreliable progress estimates are especially harmful.
Chinese Translation
机器人操作中的强化学习常常受到奖励设计的瓶颈，尤其是在长时间任务中：稀疏的成功奖励提供了弱监督，而手工设计的密集奖励则设计繁琐且在任务间泛化效果较差。基于进展的奖励模型通过估计观察在任务完成方面的进展程度提供了一个有前景的替代方案，但现有方法通常需要特定任务的示范或进展标签，并且可能会对视觉上合理但物理上不正确的状态分配高奖励。我们引入了参考锚定奖励模型（Reference-Anchored Reward Model, RARM），这是一种轻量级的视觉比较器，将单个成功示范转换为密集的、关注进展的奖励。RARM 在通用视频上进行了一次训练，采用对比时间目标，不需要机器人特定数据、任务特定奖励标签或每个任务的奖励工程。在部署时，RARM 将回放片段与参考片段匹配，仅对自信的前进进展给予奖励，抑制可能产生假阳性奖励的不确定匹配。在来自 LIBERO 和 MetaWorld 的 9 个模拟操作任务以及 4 个现实世界任务中，RARM 在后续的强化学习训练中实现了最佳的整体成功率，尤其在如折叠布料等长时间任务上取得了显著的提升，因为在这些任务中，不可靠的进展估计尤其有害。
cs.RO / 68 / 2606.22040
Deep RL- Tuned Mo del-Free Adaptive Control for Lower-Limb Exoskeletons During Sit-to-Stand Transitions
深度强化学习调优的无模型自适应控制用于下肢外骨骼在坐立转换过程中的应用
Abstract
Sit-to-stand (STS) transitions impose significant joint-loading demands on elderly individuals, making them a primary target for lower-limb exoskeleton assistance. However, accurate trajectory tracking during STS is challenging due to complex, time-varying human exoskeleton interaction dynamics and inter-subject variability that render model-based control approaches difficult to apply in practice. This paper presents an intelligent model free adaptive backstepping control strategy for a bilateral lower-limb exoskeleton during STS motion. The proposed controller design uses an ultra-local second-order model to avoid explicit system identification, while a Gaussian radial basis function (RBF) neural network estimates the unknown lumped dynamics online. To further improve phase-aware tracking performance, a Twin Delayed Deep Deterministic Policy Gradient (TD3) reinforcement learning agent is integrated as a supervisory gain scheduler that adaptively adjusts controller gains across the distinct phases of STS motion. The proposed controller is evaluated through co-simulation in MATLAB/Simulink and Simscape Multibody using OpenSim-derived reference trajectories and benchmarked against state-of-the-art controllers. Results demonstrate that the proposed controller achieves the lowest average RMSE of 0.078 degree across all joints, representing improvements of 60.2%, 54.4%, 48.7%, and 42.6% over proportional integral derivative (PID), model-free adaptive control (MFAC), linear quadratic regulator (LQR), and sliding-mode control (SMC), respectively. TD3 integration further reduces tracking error by 35%, 33%, and 79% at the hip, knee, and ankle joints compared to the standalone RBF-MFAC baseline. These results demonstrate the effectiveness and robustness of the proposed controller design for assistive exoskeleton control during STS transitions.
Chinese Translation
坐立转换（STS）对老年人的关节负荷要求极高，使其成为下肢外骨骼辅助的主要目标。然而，由于复杂的、时变的人体与外骨骼的交互动态以及个体间的差异，使得在实践中应用基于模型的控制方法变得困难。本文提出了一种智能的无模型自适应反向步进控制策略，用于双侧下肢外骨骼在STS运动中的控制。所提出的控制器设计采用超局部二阶模型，以避免显式的系统识别，同时高斯径向基函数（RBF）神经网络在线估计未知的集中动态。为了进一步提高对阶段的感知跟踪性能，集成了双延迟深度确定性策略梯度（TD3）强化学习代理，作为自适应增益调度器，在STS运动的不同阶段动态调整控制器增益。通过在MATLAB/Simulink和Simscape Multibody中进行共同仿真，使用OpenSim衍生的参考轨迹对所提出的控制器进行评估，并与最先进的控制器进行基准测试。结果表明，所提出的控制器在所有关节上实现了最低的平均均方根误差（RMSE）为0.078度，分别比比例-积分-微分（PID）、无模型自适应控制（MFAC）、线性二次调节器（LQR）和滑模控制（SMC）提高了60.2%、54.4%、48.7%和42.6%。与独立的RBF-MFAC基线相比，TD3集成进一步将髋关节、膝关节和踝关节的跟踪误差分别降低了35%、33%和79%。这些结果证明了所提出的控制器设计在STS转换过程中对辅助外骨骼控制的有效性和鲁棒性。
cs.RO / 69 / 2606.22046
A Multimodal Tiltwing Framework for Bioinspired Aerial Robots
一种多模态倾翼框架用于仿生空中机器人
Abstract
Tailless flapping-wing micro-aerial vehicles (FWMAVs) mimic the impressive flight performance of hummingbirds, utilising unsteady aerodynamic effects. However, existing designs are still limited and purpose-built with a restricted flight envelope and poor endurance. We therefore propose an adaptable tiltwing framework enabling bioinspired aerial robots to switch between hovering flight, high-speed directional flight, and energy-efficient gliding flight. The proposed framework utilises thrust vectoring with a wide actuation range via two fully independent propulsion units, each flapping a single wing, for effective control and enhanced manoeuvrability. For this, we developed a hybrid Scotch-yoke-based flapping mechanism that ensures a symmetric motion profile with a modular design guaranteeing an arbitrarily wide flapping angle to exploit the lift-enhancing clap-and-fling effect. Additionally, we implemented a passive wing-rotation mechanism, which, in combination with our dual-wing thrust-vectoring approach, allows unprecedented wing-design freedom, unlocking potential for precise optimisation. A contactless leading-edge tracking sensor provides accurate feedback on the wing's orientation and, in the gliding mode, enables dihedral-angle control, augmenting the active wing-pitch control. Extensive testing of a propulsion unit was conducted with a six-axis force/torque sensor, demonstrating the flapping mechanism's performance while optimising transmission efficiency and the passive wing-pitch mechanism. At full throttle, the average lift force generated by a single wing, flapping with a 188{\deg} amplitude, was 21.1 gf for a small 3.1 g 1S BLDC motor. Additional tests covering the full range of the wide-angle tilting capability showed an effective thrust-vectoring control architecture with a linear and symmetric response curve of the moments generated.
Chinese Translation
无尾拍翼微型空中飞行器（FWMAVs）模仿蜂鸟的卓越飞行性能，利用不稳定的空气动力学效应。然而，现有设计仍然有限，专门构建的飞行包络范围受限且耐久性差。因此，我们提出了一种可适应的倾翼框架，使仿生空中机器人能够在悬停飞行、高速定向飞行和节能滑翔飞行之间切换。所提出的框架利用了通过两个完全独立的推进单元进行的推力矢量控制，每个单元拍动一只翅膀，以实现有效控制和增强机动性。为此，我们开发了一种基于混合苏格兰摇杆的拍动机制，确保对称的运动特征，并采用模块化设计，保证了任意宽的拍动角度，以利用提升增强的拍击和抛掷效应。此外，我们实现了一种被动翼旋转机制，结合我们的双翼推力矢量控制方法，允许前所未有的翼设计自由度，开启精确优化的潜力。无接触的前缘跟踪传感器提供了翅膀方向的准确反馈，并在滑翔模式下实现了二面角控制，增强了主动翼俯仰控制。我们对推进单元进行了广泛测试，使用六轴力/扭矩传感器，展示了拍动机制的性能，同时优化了传动效率和被动翼俯仰机制。在全油门下，单翼以188°的振幅拍动所产生的平均升力为21.1 gf，适用于小型3.1 g的1S无刷直流电机。覆盖宽角倾斜能力全范围的额外测试显示了有效的推力矢量控制架构，生成的力矩具有线性和对称的响应曲线。
cs.RO / 70 / 2606.22051
GeoFlow-SLAM++: A Robust Multi-Camera Visual-Inertial SLAM System with Relocalization
GeoFlow-SLAM++：一种具有重定位能力的鲁棒多摄像头视觉惯性SLAM系统
Abstract
Monocular and RGB-D visual-inertial SLAM systems remain susceptible to limited field of view, sensor-specific failure modes, and unreliable cross-session relocalization. To address these issues, we present GeoFlow-SLAM++, a tightly coupled multi-camera visual-inertial SLAM system that extends GeoFlow-SLAM from a single RGB-D sensor to a calibrated multi-camera rig with a unified body-centric formulation. Within this multi-camera framework, GeoFlow-SLAM++ supports two interchangeable visual front-ends: a conventional ORB front-end and a neural network feature (NN-Feature) front-end built on SuperPoint and LightGlue. The system unifies tracking, mapping, and relocalization on a shared body state, and combines multi-camera reprojection constraints, IMU pre-integration, cross-view place recognition, and dual-stream optical flow/NN-Feature tracking for robust localization. As an optional extension, the system can further incorporate cross-view-consistent pseudo-depth predictions from RGB images as auxiliary geometric constraints. We evaluate GeoFlow-SLAM++ on EuRoC, OpenLORIS, TUM, Hilti, and a self-collected handheld multi-camera dataset. Results show that the NN-Feature front-end improves robustness in appearance-challenging scenarios, the multi-camera formulation achieves competitive localization accuracy on Hilti, and the unified cross-view relocalization design reaches LiDAR-comparable performance on the handheld dataset.
Chinese Translation
单目和RGB-D视觉惯性SLAM系统仍然容易受到视场限制、传感器特定故障模式和不可靠的跨会话重定位的影响。为了解决这些问题，我们提出了GeoFlow-SLAM++，这是一种紧密耦合的多摄像头视觉惯性SLAM系统，它将GeoFlow-SLAM从单一的RGB-D传感器扩展到一个经过校准的多摄像头装置，并采用统一的以主体为中心的公式。在这个多摄像头框架中，GeoFlow-SLAM++支持两种可互换的视觉前端：传统的ORB前端和基于SuperPoint和LightGlue构建的神经网络特征（NN-Feature）前端。该系统在共享的主体状态上统一了跟踪、映射和重定位，并结合了多摄像头重投影约束、IMU预积分、跨视图位置识别以及双流光流/NN-Feature跟踪，以实现鲁棒的定位。作为一个可选扩展，该系统还可以进一步结合来自RGB图像的跨视图一致的伪深度预测作为辅助几何约束。我们在EuRoC、OpenLORIS、TUM、Hilti以及自收集的手持多摄像头数据集上评估了GeoFlow-SLAM++。结果表明，NN-Feature前端在外观挑战场景中提高了鲁棒性，多摄像头公式在Hilti上实现了竞争性的定位精度，而统一的跨视图重定位设计在手持数据集上达到了与LiDAR相当的性能。
cs.RO / 71 / 2606.22062
How Should a Simulation-to-Reality Transfer Budget Be Spent?
模拟到现实转移预算应如何支出？
Abstract
Simulation-to-reality transfer, often called sim-to-real transfer, is a central challenge in robot learning. Yet, the tradeoff between measuring a system more accurately and training over a broader range of simulated dynamics is still poorly understood. In this work, we focused on the allocation of real-robot measurement time between system identification and domain randomization. We studied this tradeoff in a controlled sim-to-sim pendulum setting, where a hidden-parameter model stands in for the physical robot, and the experiment sweeps identification rollouts against the width of the randomization distribution. Across the reality gaps and noise levels we tested, the measurement budget did most of the work. A small number of identification rollouts closed most of the transfer gap, and once any real data was available, policies performed best when trained at the estimated parameters rather than over a widened randomization band. Broad randomization that contained the true system still did not substitute for measurement. These results hold in a benign regime where the dynamics are identifiable and only two parameters are unknown, so structural model mismatch remains the setting where randomization breadth may become more valuable. Overall, our results suggest that sim-to-real pipelines should first measure the parameters they can and reserve randomization for the uncertainty that remains.
Chinese Translation
模拟到现实转移，通常称为 sim-to-real 转移，是机器人学习中的一个核心挑战。然而，在更准确地测量系统与在更广泛的模拟动态范围内进行训练之间的权衡仍然不够清晰。在本研究中，我们重点关注了真实机器人测量时间在系统识别和领域随机化之间的分配。我们在一个受控的 sim-to-sim 摆动设置中研究了这一权衡，其中一个隐藏参数模型代表物理机器人，实验通过识别回合与随机化分布的宽度进行比较。在我们测试的现实差距和噪声水平中，测量预算发挥了主要作用。少量的识别回合就缩小了大部分转移差距，一旦有任何真实数据可用，策略在估计参数下训练时表现最佳，而不是在扩展的随机化范围内进行训练。尽管广泛的随机化包含了真实系统，但仍无法替代测量。这些结果适用于一个良好的环境，其中动态是可识别的，且仅有两个参数未知，因此结构模型不匹配仍然是随机化宽度可能更有价值的情境。总体而言，我们的结果表明，sim-to-real 流程应首先测量它们能够测量的参数，并将随机化保留用于剩余的不确定性。
cs.RO / 72 / 2606.22073
Dynamics, stability, and energy efficiency of an energy-recycling rimless wheel with spring-clutch legs
具有弹簧离合腿的能量回收无轮辋轮的动力学、稳定性和能量效率
Abstract
This paper proposes an energy-recycling rimless wheel with spring-clutch legs. The proposed mechanism uses a lockable clutch to store part of the impact-induced elastic energy after foot contact and reinject it in the next gait cycle. First, we develop a hybrid dynamic model of the energy-recycling rimless wheel. Second, numerical simulations are used to examine the dynamics, local stability of periodic gaits, and the Cost of Transport (CoT) of the proposed mechanism. The simulation results show that the proposed mechanism reduces the CoT by up to 16.13% compared with a benchmark viscoelastic-legged rimless wheel with telescopic spring-damper legs. Compared with the rigid rimless wheel, the viscoelastic-legged and energy-recycling models reduce the CoT by more than 50%. The energy-recycling model also maintains locally stable periodic gaits over the tested slope and stiffness ranges. Finally, prototype experiments on an inclined plane are conducted to examine the feasibility of the proposed mechanism. The experimental results show that the proposed rimless wheel achieves passive walking on a shallow 1{\deg} slope, corresponding to a CoT of approximately 0.02. These results suggest that the proposed spring-clutch mechanism can improve the simulated walking efficiency of the energy-recycling rimless wheel, while the prototype experiments support the feasibility of passive walking with the mechanism.
Chinese Translation
本文提出了一种具有弹簧离合腿的能量回收无轮辋轮。该机制使用可锁定的离合器在脚接触后储存部分冲击引起的弹性能量，并在下一个步态周期中重新注入。首先，我们开发了能量回收无轮辋轮的混合动态模型。其次，使用数值模拟来研究所提机制的动力学、周期步态的局部稳定性以及运输成本（Cost of Transport, CoT）。模拟结果表明，与基准的具有伸缩弹簧阻尼腿的粘弹性腿无轮辋轮相比，所提机制将CoT降低了最高16.13%。与刚性无轮辋轮相比，粘弹性腿和能量回收模型将CoT降低了超过50%。能量回收模型在测试的坡度和刚度范围内也保持局部稳定的周期步态。最后，在倾斜平面上进行原型实验，以检验所提机制的可行性。实验结果表明，所提无轮辋轮在1°的浅坡上实现了被动行走，对应的CoT约为0.02。这些结果表明，所提的弹簧离合机制可以提高能量回收无轮辋轮的模拟行走效率，同时原型实验支持该机制的被动行走的可行性。
cs.RO / 73 / 2606.22091
ACEsplat: Accelerated 3D Gaussian Scene Regression via RGB and Poses Only
ACEsplat：仅通过RGB和姿态加速的3D高斯场景回归
Abstract
Per-scene 3D Gaussian Splatting (3DGS) enables high-fidelity rendering, but practical robotic and AR scene capture pipelines often depend on external geometric initialization (e.g., SfM point clouds or depth estimates), which can be slow and brittle in on-site deployment. We present ACEsplat, a fast per-scene optimization framework that reconstructs 3D Gaussian representations from RGB images and camera poses only, without requiring external 3D priors (e.g., precomputed SfM models or supervised depth maps). ACEsplat uses a two-stage pipeline: (1) a self-supervised scene coordinate regression (SCR) module builds an internal geometry prior within 4--5 minutes; (2) SCR features and coordinate priors are fused by a lightweight Gaussian initialization head, followed by per-scene 3DGS optimization. On static-view rendering, ACEsplat achieves 29.11 dB PSNR on Wayspots with real-time SLAM poses and 33.20 dB on Cambridge Landmarks with SfM-refined poses. On RealEstate10K sparse-view novel view synthesis, it achieves competitive image fidelity under a challenging 2-view setting. ACEsplat completes scene-specific SCR mapping and 3DGS reconstruction within 15--25 minutes on a single GPU, making it a practical RGB+pose-only solution for rapid scene setup in robotics and mixed-reality applications.
Chinese Translation
每场景的3D高斯点云（3DGS）能够实现高保真渲染，但实际的机器人和增强现实场景捕捉流程通常依赖于外部几何初始化（例如，结构从运动（SfM）点云或深度估计），这在现场部署时可能会缓慢且不稳定。我们提出了ACEsplat，这是一种快速的每场景优化框架，仅通过RGB图像和相机姿态重建3D高斯表示，而无需外部3D先验（例如，预计算的SfM模型或监督深度图）。ACEsplat采用两阶段流程：（1）自监督场景坐标回归（SCR）模块在4到5分钟内构建内部几何先验；（2）SCR特征和坐标先验通过轻量级高斯初始化头进行融合，随后进行每场景的3DGS优化。在静态视图渲染中，ACEsplat在使用实时SLAM姿态的Wayspots上达到了29.11 dB的峰值信噪比（PSNR），在使用SfM精炼姿态的剑桥地标上达到了33.20 dB。在RealEstate10K稀疏视图的新视图合成中，它在具有挑战性的2视图设置下实现了竞争性的图像保真度。ACEsplat在单个GPU上完成场景特定的SCR映射和3DGS重建仅需15到25分钟，使其成为机器人和混合现实应用中快速场景设置的实用RGB+姿态解决方案。
cs.RO / 74 / 2606.22113
KITE: Decoupling Kinematics and Interaction for Zero-Shot Cross-Embodiment Manipulation
KITE：解耦运动学与交互以实现零-shot跨体现操控
Abstract
Generalizing manipulation policies across robot embodiments remains difficult because standard policies entangle task reasoning with embodiment-specific motor control. We study zero-shot cross-embodiment manipulation, where a policy trained on source embodiments must be deployed on a structurally different target embodiment without additional task demonstrations. We introduce Kinematic Interaction Transfer across Embodiments (KITE), which decouples manipulation into embodiment-agnostic task reasoning and embodiment-specific motor control, connected through a learned latent representation of interaction intent based on contact patterns. Task reasoning is performed by a shared policy that predicts latent intents from source demonstrations, while motor control is performed by an intent-conditioned action decoder learned from each embodiment's kinematic model. With KITE, adaptation to a new embodiment requires only training a new action decoder using its kinematic model, without recollecting demonstration data. We evaluate KITE on three manipulation tasks spanning transfer between parallel grippers, dexterous hands, and composite embodiments. KITE consistently achieves zero-shot transfer to structurally different target embodiments, outperforming state-of-the-art baselines in transfer success and task-embodiment scope.
Chinese Translation
跨机器人体现推广操控策略仍然困难，因为标准策略将任务推理与特定体现的运动控制纠缠在一起。我们研究零-shot跨体现操控，其中在源体现上训练的策略必须在结构上不同的目标体现上部署，而无需额外的任务演示。我们提出了跨体现的运动交互转移（Kinematic Interaction Transfer across Embodiments，KITE），它将操控解耦为与体现无关的任务推理和特定体现的运动控制，通过基于接触模式的交互意图的学习潜在表示相连接。任务推理由一个共享策略执行，该策略从源演示中预测潜在意图，而运动控制则由从每个体现的运动学模型学习的意图条件下的动作解码器执行。使用KITE，适应新的体现只需使用其运动学模型训练一个新的动作解码器，而无需重新收集演示数据。我们在三个操控任务上评估KITE，这些任务涵盖了在平行夹持器、灵巧手和复合体现之间的转移。KITE在结构上不同的目标体现上始终实现零-shot转移，在转移成功率和任务-体现范围方面超越了最先进的基线。
cs.RO / 75 / 2606.22116
DeformX: A Versatile Co-Simulation Framework for Deformable Linear Objects
DeformX：一种多功能的可变形线性物体联合仿真框架
Abstract
Deformable linear objects (DLOs) such as wires, cables, and ropes are common in robotic manipulation tasks, yet simulating them with both visual realism and physical accuracy remains challenging. Existing visual simulation methods typically rely on procedural geometric primitives that lack physically grounded deformation behavior, while physics-based approaches with robot learning support often approximate DLOs as rigid-link chains or generic soft bodies, failing to accurately capture the bending, twisting, and shear mechanics of slender elastic structures. In this work, we introduce DeformX, a co-simulation framework that integrates a dedicated Cosserat rod physics engine with NVIDIA Isaac Sim, enabling DLO simulations that are both physically faithful and visually realistic. Our Cosserat rod engine simulates the dynamics and self-collisions of DLOs, and contact interactions with arbitrary free-form meshes. To achieve high-fidelity visualization, we employ mesh skinning to map discrete rod deformations onto imported CAD models. To the best of our knowledge, DeformX is the one of the first frameworks for DLO simulation that unifies realistic visualization, principled physics, and compatibility with robot learning pipelines. We demonstrate its versatility across synthetic data generation and policy learning for DLO manipulation, and validate visual and physical fidelity through comparisons against real-world experiments. Notably, fine-tuning Segment Anything Model 3 (SAM3) on DeformX-generated data yields a 10.2% mAP@75 improvement in real-image wire segmentation, and a rope-swinging policy trained entirely in DeformX achieves a mean target-hitting error of 6.6 cm on a UR5e manipulator in real-world trials, highlighting its strong sim-to-real transfer capability.
Chinese Translation
可变形线性物体（DLOs），如电线、缆绳和绳索，在机器人操作任务中十分常见，但在视觉真实感和物理准确性方面的仿真仍然具有挑战性。现有的视觉仿真方法通常依赖于缺乏物理基础变形行为的程序几何原语，而基于物理的机器人学习支持方法则常常将DLOs近似为刚性链或通用软体，未能准确捕捉细长弹性结构的弯曲、扭转和剪切力学。在本研究中，我们介绍了DeformX，一个将专用的Cosserat杆物理引擎与NVIDIA Isaac Sim集成的联合仿真框架，使得DLO仿真既具物理真实性又具视觉真实感。我们的Cosserat杆引擎模拟DLO的动力学和自碰撞，以及与任意自由形状网格的接触交互。为了实现高保真可视化，我们采用网格蒙皮技术将离散杆变形映射到导入的CAD模型上。据我们所知，DeformX是首个统一真实可视化、原则性物理和与机器人学习管道兼容的DLO仿真框架。我们展示了其在合成数据生成和DLO操作策略学习中的多功能性，并通过与真实世界实验的比较验证了视觉和物理的保真度。值得注意的是，在DeformX生成的数据上微调Segment Anything Model 3（SAM3）在真实图像电线分割中实现了10.2%的mAP@75提升，而在DeformX中完全训练的绳摆策略在真实世界试验中在UR5e操纵器上的平均目标击中误差为6.6厘米，突显了其强大的仿真到现实转移能力。
cs.RO / 76 / 2606.22129
Durability-Aware Multi-Objective Optimization of the Jansen Linkage: Trading Gait Quality Against Joint Wear
耐久性意识的Jansen连杆多目标优化：在步态质量与关节磨损之间的权衡
Abstract
The Jansen linkage is a single-degree-of-freedom planar leg mechanism whose eleven "holy numbers" were evolved by Theo Jansen to optimize the foot-path gait alone, with no regard for the wear of its revolute joints. This paper introduces a durability objective into the design of the Jansen leg. A parametric forward-kinematic model (two-circle-intersection solver), an inverse-dynamic model (constraint-Jacobian / Lagrange-multiplier formulation of a seven-body, ten-joint system, independently cross-verified by a reduced-DOF energy method), and an Archard wear model are coupled to evaluate, for any set of link lengths, both gait quality and the per-cycle sliding wear at every pin. Because the wear is computed on ideal, clearance-free revolute joints, the resulting wear figures are a relative comparative ranking rather than an absolute life prediction. A bi-objective problem -- composite gait error versus total joint wear, subject to step-length, ground-clearance, duty-factor and assembly constraints -- is solved with NSGA-II. Under the adopted gait metric the classical Jansen design is Pareto-dominated: for a representative design, link-length adjustments within +/-29% simultaneously flatten the stance (-28%), smooth the stance velocity (-58%) and reduce total joint wear by ~56%. A sensitivity study shows the wear advantage is robust across a crank-speed x payload envelope (48%-56%) and identifies the link lengths that most strongly govern wear. A variance-based global (Sobol) analysis confirms that two link lengths dominate the wear variance, and a Monte-Carlo manufacturing-tolerance study shows the wear advantage degrades gracefully under realistic fabrication error. The framework provides a practical route to longer-lived walking linkages and a baseline for future wear-clearance-impact coupled studies.
Chinese Translation
Jansen连杆是一种单自由度的平面腿部机制，其十一组“神圣数字”由Theo Jansen演化而来，旨在单独优化足迹步态，而不考虑其旋转关节的磨损。本文在Jansen腿部设计中引入了耐久性目标。通过一个参数化的正向运动学模型（双圆交点求解器）、一个逆向动力学模型（约束雅可比矩阵/拉格朗日乘子形式的七体十关节系统，独立通过降低自由度的能量方法进行交叉验证）以及一个Archard磨损模型的结合，评估任何一组连杆长度下的步态质量和每个销钉的每周期滑动磨损。由于磨损是在理想的无间隙旋转关节上计算的，因此得到的磨损数据是相对比较排名，而非绝对寿命预测。通过NSGA-II解决一个双目标问题——复合步态误差与总关节磨损，受到步长、地面间隙、占空比和装配约束的限制。在采用的步态指标下，经典的Jansen设计被Pareto主导：对于一个代表性设计，连杆长度在±29%的调整同时使支撑阶段变平坦（-28%）、支撑速度变平滑（-58%）并减少总关节磨损约56%。灵敏度研究表明，磨损优势在曲柄速度与载荷范围（48%-56%）内是稳健的，并识别出对磨损影响最大的连杆长度。基于方差的全局（Sobol）分析确认了两个连杆长度主导磨损方差，而蒙特卡洛制造公差研究显示，在现实的制造误差下，磨损优势逐渐降低。该框架为更长寿命的步行连杆提供了一条实用的途径，并为未来的磨损-间隙-影响耦合研究提供了基准。
cs.RO / 77 / 2606.22136
Wh0: Generative World Models as Scalable Sources of Egocentric Human Hand Manipulation Data
Wh0：可扩展的人类手部操控数据生成世界模型
Abstract
Scaling dexterous manipulation requires generalization across objects, scenes, and tasks, yet existing data sources face a trade-off between scale and scene/embodiment alignment: teleoperation data is well aligned with robot deployment but expensive to collect; simulation is scalable but limited by the sim-to-real gap; and real egocentric videos scale effectively but remain misaligned with robot deployment. We propose Wh0, a framework that uses generative video world models as scalable and controllable sources of egocentric human-hand manipulation data to unlock the manipulation capabilities of pretrained dexterous VLA models. Conditioned on language, objects, and scenes, Wh0 uses a generative world model to produce WM-H, a 50k-episode dataset of egocentric human-object interaction videos. Wh0 then converts the generated videos into robot-trainable supervision through hand motion reconstruction and visual editing. Co-trained with a limited amount of real robot data, WM-H adapts pretrained VLA models to dexterous manipulation deployment. Across 18 real-world dexterous manipulation tasks, compared with a model post-trained only on robot data, Wh0 improves zero-shot success on unseen tasks from 8.3% to 38.9%. Ablation studies further show that scalable generation and scene/embodiment alignment are key drivers of performance gains. Videos and open-source code can be found on our project website: https://chenyt31.github.io/wh0.github.io/.
Chinese Translation
扩展灵巧操控需要在物体、场景和任务之间进行泛化，然而现有的数据源在规模与场景/体现对齐之间存在权衡：遥操作数据与机器人部署高度对齐，但收集成本高；仿真数据具有可扩展性，但受到仿真与现实之间差距的限制；而真实的自我中心视频在扩展性上表现良好，但与机器人部署仍存在不对齐。我们提出了Wh0，一个利用生成视频世界模型作为可扩展和可控的自我中心人手操控数据源的框架，以解锁预训练灵巧VLA模型的操控能力。Wh0基于语言、物体和场景进行条件生成，使用生成世界模型生成WM-H，一个包含50,000集自我中心人-物体交互视频的数据集。随后，Wh0通过手部运动重建和视觉编辑将生成的视频转换为可用于机器人训练的监督信号。与有限的真实机器人数据共同训练后，WM-H使预训练的VLA模型适应灵巧操控的部署。在18个真实世界的灵巧操控任务中，与仅在机器人数据上进行后训练的模型相比，Wh0在未见任务上的零-shot成功率从8.3%提升至38.9%。消融研究进一步表明，可扩展生成和场景/体现对齐是性能提升的关键驱动因素。视频和开源代码可在我们的项目网站找到：https://chenyt31.github.io/wh0.github.io/
cs.RO / 78 / 2606.22142
RoboLineage: Agent-Native Data Lifecycle Governance Across Robot Policy Iterations
RoboLineage：机器人政策迭代中的代理原生数据生命周期治理
Abstract
We present RoboLineage, an agent-native data lifecycle governance system for robot policy iteration. Modern robot policies improve through repeated data collection, review, retraining, evaluation, and release decisions, but the evidence connecting these steps is often scattered across local tools, scripts, and expert memory. RoboLineage makes this lifecycle explicit by representing rollouts, reviews, dataset decisions, training runs, policy metadata, evaluations, deployment recommendations, and next-collection plans as typed lineage artifacts. Agents interpret embodied rollout evidence, adapt accepted data to existing training stacks, maintain data health, and summarize cross-iteration state under explicit artifact boundaries. In real-robot manipulation workflows, RoboLineage makes routine policy iteration faster and more auditable while maintaining downstream policy performance. We open source RoboLineage as a lightweight lifecycle layer for different robot embodiments and training families. Project page: https://robolineage.github.io/
Chinese Translation
我们提出了RoboLineage，一个用于机器人政策迭代的代理原生数据生命周期治理系统。现代机器人政策通过重复的数据收集、审查、再训练、评估和发布决策不断改进，但连接这些步骤的证据往往分散在本地工具、脚本和专家记忆中。RoboLineage通过将回滚、审查、数据集决策、训练运行、政策元数据、评估、部署建议和下一次收集计划表示为类型化的血统工件，使这一生命周期变得明确。代理解释具体的回滚证据，将接受的数据适应现有的训练堆栈，维护数据健康，并在明确的工件边界下总结跨迭代状态。在真实机器人操作工作流程中，RoboLineage使常规政策迭代更快且更具可审计性，同时保持下游政策性能。我们将RoboLineage开源，作为不同机器人体现和训练系列的轻量级生命周期层。项目页面：https://robolineage.github.io/
cs.RO / 79 / 2606.22143
Physics-Informed Eikonal Caging for Whole-Arm Manipulation Planning
基于物理信息的全臂操作规划中的包围策略
Abstract
Planning contact-rich whole-arm manipulation is challenging because interactions that involve extended robot geometry give rise to complex contact dynamics that are difficult to model accurately. This creates a need for planning principles that do not rely heavily on precise contact models. Caging offers one such geometric notion of robustness to modeling inaccuracy by restricting object escape through geometrically enclosing the object. However, existing caging formulations are difficult to incorporate into continuous optimization-based manipulation planning. We reformulate caging as a minimum-time escape problem in which the object seeks to leave an enclosing robot geometry in the shortest time. This yields a continuous escape-time field that measures the robot's enclosure quality and we show it satisfies an eikonal equation. We therefore can approximate this field using a physics-informed neural network, producing a smooth differentiable representation that can be embedded directly into manipulation planning. The resulting objective supports whole-arm manipulation planning to favor robot configurations resisting object escape. This improves the manipulation robustness to contact model mismatch, thus enabling planning with simplified contact models, including quasi-dynamic approximations and simplified object geometry. Across simulation and real-world experiments, we show improved robustness to disturbances and contact-model mismatch relative to baselines. These results suggest that geometric enclosure can serve as a practical robustness primitive for whole-arm manipulation. A supplementary video, which includes an intuitive overview of our method and experiment video results, is available on our project webpage.
Chinese Translation
规划接触丰富的全臂操作具有挑战性，因为涉及扩展机器人几何形状的交互会产生复杂的接触动态，这些动态难以准确建模。这就需要一些不依赖于精确接触模型的规划原则。包围（Caging）提供了一种几何上的稳健性概念，通过几何上封闭物体来限制其逃逸。然而，现有的包围公式难以融入基于连续优化的操作规划中。我们将包围重新表述为一个最小时间逃逸问题，其中物体寻求在最短时间内离开封闭的机器人几何形状。这产生了一个连续的逃逸时间场，用于测量机器人的包围质量，我们证明它满足一个艾可诺方程（eikonal equation）。因此，我们可以使用基于物理信息的神经网络来近似这个场，生成一个平滑的可微分表示，可以直接嵌入到操作规划中。最终的目标支持全臂操作规划，倾向于抵抗物体逃逸的机器人配置。这提高了对接触模型不匹配的操作稳健性，从而使得可以使用简化的接触模型进行规划，包括准动态近似和简化的物体几何形状。在模拟和真实世界实验中，我们展示了相对于基线的干扰和接触模型不匹配的稳健性得到了改善。这些结果表明，几何包围可以作为全臂操作的实用稳健性原语。我们项目网页上提供了一段补充视频，其中包括我们方法的直观概述和实验视频结果。
cs.RO / 80 / 2606.22145
Zero-shot Transfer of Reinforcement Learning Control Policies for the Swing-Up and Stabilization of a Cart-Pole System
强化学习控制策略在摆杆系统的摆起与稳定中的零-shot迁移
Abstract
Reinforcement learning (RL) is a powerful and convenient tool to modernize controller design. In this work, we study the zero-shot transfer of RL-based control policies from simulation to hardware for cart-pole swing-up and stabilization. The two policies are trained independently, and the handoff is implemented in Simulink via switching logic. We apply a first-order action smoothing filter to prevent hardware damage from high-frequency oscillatory actuation. Pairing this bandwidth-aware filtering with sensitivity-guided domain randomization (DR) and a simple linear curriculum learning (CL) schedule, we obtain a swing-up policy that in all of our experiments injects sufficient energy for handoff into the stabilizer's region of attraction. The stabilization policy rejects disturbances within the tested range, and the swing-up policy can re-engage after larger perturbations and restores the pendulum to the inverted position.
Chinese Translation
强化学习（RL）是现代控制器设计的一种强大且便捷的工具。在本研究中，我们探讨了基于RL的控制策略从仿真到硬件的零-shot迁移，以实现摆杆的摆起与稳定。这两种策略是独立训练的，交接过程通过Simulink中的切换逻辑实现。我们应用了一阶动作平滑滤波器，以防止高频振荡驱动对硬件造成损害。将这种带宽感知的滤波与敏感性引导的领域随机化（DR）和简单的线性课程学习（CL）计划相结合，我们获得了一种摆起策略，在所有实验中都能为交接注入足够的能量，进入稳定器的吸引区。稳定策略能够抵抗测试范围内的干扰，而摆起策略在经历较大扰动后能够重新启动，并将摆锤恢复到倒立位置。
cs.RO / 81 / 2606.22169
Full Nonlinear Nonholonomic Dynamics and Motion Analysis of a 3-DoF Underactuated Spherical Rolling Robot
三自由度欠驱动球形滚动机器人的全非线性非完整动力学与运动分析
Abstract
This paper presents a full nonlinear constrained dynamic model of MonoRollBot, a novel 3-DoF spherical rolling robot driven by a single motor, a lead-screw transmission, and a spring-coupled internal moving mass, together with motion analysis of its behavior. To the best of our knowledge, this is one of the first full nonlinear nonholonomic models reported for a mono-actuated, super-underactuated spherical rolling robot of this kind. Because rolling without slipping is nonholonomic, the dynamics are derived using the Lagrange--d'Alembert formulation, with the lead-screw relation imposed as a holonomic constraint and the rolling condition imposed in Pfaffian form. The formulation retains the complete generalized coordinates of shell translation, shell attitude, screw travel, nut rotation, and radial mass motion. Simulations and representative motion studies show qualitative agreement with prototype behavior and reveal how gravity, compliance, and inertia jointly shape the locomotion and motion capabilities of this strongly underactuated robot. The resulting model also provides a mechanically consistent basis for future state estimation and hybrid controller design for this nonholonomic mono-actuated rolling robot.
Chinese Translation
本文提出了一种MonoRollBot的全非线性约束动力学模型，这是一种由单个电机、丝杠传动和弹簧耦合内部移动质量驱动的新型三自由度球形滚动机器人，并对其行为进行了运动分析。据我们所知，这是首次针对这种单驱动、超欠驱动球形滚动机器人报告的全非线性非完整模型。由于无滑滚动是非完整的，动力学是通过拉格朗日-达朗贝尔公式推导的，丝杠关系作为完整约束施加，而滚动条件则以Pfaffian形式施加。该公式保留了壳体平移、壳体姿态、丝杠行程、螺母旋转和径向质量运动的完整广义坐标。仿真和代表性运动研究显示出与原型行为的定性一致性，并揭示了重力、顺应性和惯性如何共同影响这一强欠驱动机器人的运动和运动能力。所得到的模型还为未来的状态估计和混合控制器设计提供了机械一致的基础，适用于这种非完整单驱动滚动机器人。
cs.RO / 82 / 2606.22174
OpenHLM: An Empirical Recipe for Whole-Body Humanoid Loco-Manipulation
OpenHLM：全身类人机器人运动操控的实证方法
Abstract
Whole-body humanoid loco-manipulation requires coordinating the robot's entire kinematic chain. However, most existing systems typically decouple the upper and lower bodies into separate controllers, limiting such coordination and yielding behaviors similar to those of a wheeled dual-arm platform. In this paper, we ask what it takes to build a whole-body native vision-language-action (VLA) model that maps language and pixels directly to all of the humanoid's degrees of freedom. We conduct a systematic empirical study organized as a roadmap of one-variable-at-a-time experiments across three phases: whole-body teleoperation, VLA model design, and heterogeneous co-training. Our study yields several intriguing findings: a joint-based whole-body teleoperation interface outperforms alternatives that only partially expose the humanoid's degrees of freedom; a VLA pretrained on static and wheeled dual-arm platforms transfers surprisingly well to a humanoid's full action space; and co-training with HuMI, the humanoid analog of UMI, extends the policy to new objects and instructions without additional whole-body teleoperation on those targets. Following this roadmap yields OpenHLM, an open-source recipe for whole-body humanoid loco-manipulation. In a challenging long-horizon task that spans a wide vertical range of the humanoid, OpenHLM outperforms two state-of-the-art humanoid VLA baselines (GR00T N1.6 and $\Psi_0$) using less than half the total demonstration time. Our code, training data, and model checkpoints are available at [https://openhlm-project.github.io/].
Chinese Translation
全身类人机器人运动操控需要协调机器人的整个运动链。然而，大多数现有系统通常将上下半身解耦为独立的控制器，这限制了这种协调性，并导致行为类似于轮式双臂平台。在本文中，我们探讨构建一个全身原生视觉-语言-动作（VLA）模型所需的条件，该模型将语言和像素直接映射到类人的所有自由度。我们进行了一项系统的实证研究，组织为一个逐变量实验的路线图，分为三个阶段：全身遥操作、VLA模型设计和异构共同训练。我们的研究得出了一些引人注目的发现：基于关节的全身遥操作界面优于仅部分暴露类人自由度的替代方案；在静态和轮式双臂平台上预训练的VLA模型意外地很好地迁移到类人的完整动作空间；与HuMI（类人机器人对应于UMI）的共同训练扩展了策略到新对象和指令，而无需对这些目标进行额外的全身遥操作。遵循这一路线图产生了OpenHLM，一个用于全身类人机器人运动操控的开源方法。在一个跨越类人机器人广泛垂直范围的具有挑战性的长时间任务中，OpenHLM在总演示时间不到一半的情况下，超越了两个最先进的类人VLA基线（GR00T N1.6和$ ext{Ψ}_0$）。我们的代码、训练数据和模型检查点可在[https://openhlm-project.github.io/]获取。
cs.RO / 83 / 2606.22251
Geometric Reconstruction of Extrinsic Contact Trajectories using Tactile Sensing and Proprioception for Tool Manipulation
利用触觉感知和本体感知进行工具操作的外部接触轨迹几何重建
Abstract
Tactile sensing enables robots to perceive rich contact information at the grasp, supporting tasks such as object recognition, in-hand pose estimation, and slip detection. However, in many tool-mediated manipulation tasks, the interaction that determines task success occurs at the tool tip, away from the tactile sensor, making direct sensing of tool-environment contact difficult, particularly when the contact moves during interaction. In this work, we reconstruct the trajectory of extrinsic tool-tip contact using tactile sensing and robot proprioception. We formulate tool-tip trajectory reconstruction as a geometric inference problem under a single-point contact assumption. Our method first estimates the global tool-tip contact location from a calibration segment designed to approximate fixed-point behavior, and then reconstructs the full trajectory by composing relative tool motion estimated from tactile marker observations under continuous contact. Across n=51 trials with multiple trajectories, tools, wrist poses, and grasp configurations, the proposed pipeline achieves a trajectory RMSE of 8.59 +/- 2.41 mm in the world frame and a shape RMSE of 5.96 +/- 1.16 mm, while operating online at 14.00 +/- 4.11 Hz. Overall, the results show that extrinsic tool-tip trajectory geometry can be recovered consistently from grasp-level tactile sensing, with trajectory shape remaining stable across variations in tools, wrist poses, and grasp configurations.
Chinese Translation
触觉感知使机器人能够感知抓取时丰富的接触信息，支持物体识别、手内姿态估计和滑动检测等任务。然而，在许多工具介导的操作任务中，决定任务成功的交互发生在工具尖端，远离触觉传感器，这使得直接感知工具与环境的接触变得困难，特别是在交互过程中接触发生移动的情况下。在本研究中，我们利用触觉感知和机器人本体感知重建外部工具尖端接触的轨迹。我们将工具尖端轨迹重建形式化为在单点接触假设下的几何推断问题。我们的方法首先从设计用于近似固定点行为的标定段中估计全局工具尖端接触位置，然后通过组合在连续接触下从触觉标记观察中估计的相对工具运动来重建完整轨迹。在n=51次试验中，涉及多条轨迹、工具、手腕姿态和抓取配置，所提出的流程在世界坐标系中实现了8.59 +/- 2.41 mm的轨迹均方根误差（RMSE）和5.96 +/- 1.16 mm的形状均方根误差，同时以14.00 +/- 4.11 Hz的频率在线运行。总体而言，结果表明可以从抓取级别的触觉感知中一致地恢复外部工具尖端轨迹的几何形状，且轨迹形状在工具、手腕姿态和抓取配置的变化中保持稳定。
cs.RO / 84 / 2606.22278
Any-Body Guard: Universal Safeguarding for Manipulation Policies via Action Masking
任何身体守卫：通过动作屏蔽实现操控策略的通用安全保障
Abstract
Ensuring safety of learning-enabled robotic manipulation across diverse embodiments and tasks still requires significant manual engineering. Existing approaches typically rely on heuristically designed fallback controllers or complex forward invariance assessments. These methods are often too conservative for task success, too computationally expensive for real-time execution, too heuristic to provide useful safety guarantees, or too engineering-heavy to transfer between setups. In this paper, we propose a universal safeguarding approach, X-Safe, which reasons directly in the robot's configuration space to provide formal probabilistic guarantees for collision avoidance. By operating in the configuration space, our method transfers across embodiments while relying solely on an object-based, quasi-static scene representation and a forward kinematics model of the robotic manipulator. Thus, X-Safe provides useful formal safety guarantees without requiring additional data, or engineering effort for different embodiments or scenes. We demonstrate X-Safe for diverse embodiments and policies, both in simulation and on hardware. We observe less degradation in task performance compared to state-of-the-art safeguarding, no collisions on hardware experiments, and empirically corroborate our formal guarantees.
Chinese Translation
确保学习驱动的机器人操控在多样化的形态和任务中安全性仍然需要大量的手动工程。现有的方法通常依赖于启发式设计的后备控制器或复杂的前向不变性评估。这些方法往往对任务成功过于保守，对实时执行计算开销过大，过于启发式以提供有用的安全保障，或者在不同设置之间转移时过于依赖工程。本文提出了一种通用的安全保障方法，X-Safe，它直接在机器人的配置空间中推理，以提供正式的概率性碰撞避免保障。通过在配置空间中操作，我们的方法在不同形态间转移，同时仅依赖于基于对象的准静态场景表示和机器人操控器的前向运动学模型。因此，X-Safe在不需要额外数据或工程努力的情况下，为不同形态或场景提供有用的正式安全保障。我们在多样化的形态和策略中演示了X-Safe，包括在仿真和硬件上。我们观察到与最先进的安全保障相比，任务性能的下降较小，硬件实验中没有发生碰撞，并且经验上证实了我们的正式保障。
cs.RO / 85 / 2606.22303
FlowDPG: Deterministic Policy Gradient on Flow Matching Policies for Real-World Manipulation
FlowDPG：针对流匹配策略的确定性策略梯度方法在现实世界操控中的应用
Abstract
Real-world reinforcement learning for robotic manipulation remains challenging, and this difficulty is amplified for flow matching policies: applying policy gradient methods to these policies is fundamentally limited by the need to backpropagate through time(BPTT) along the multi-step ODE that maps noise to actions, which is computationally prohibitive and numerically fragile. We propose FlowDPG, a DDPG-style method specifically designed for flow matching policies that distills the critic gradient into the velocity field at training time, bypassing BPTT entirely. Intuitively, FlowDPG combines two complementary vectors: the demonstration-driven velocity that keeps the action feasible, and the critic-driven correction that steers it toward higher value. Our contributions are threefold: (1) a BPTT-free distillation framework that enables stable DDPG-style policy improvement on flow matching policies, (2) a formal connection between the FlowDPG update direction and vanilla Deterministic Policy Gradient via three explicit approximations, and (3) real-world validation on a long-horizon, multi-stage, dual-arm AirPods assembly task, where FlowDPG attains a 92% end-to-end success rate, substantially outperforming recent RL methods spanning value-conditioning, auxiliary-module adaptation, and adjoint-based critic-gradient approaches. Videos and more results are provided on the project page https://flowdpg.github.io.
Chinese Translation
现实世界中的机器人操控强化学习仍然面临挑战，而对于流匹配策略，这种困难更为明显：将策略梯度方法应用于这些策略的根本限制在于需要沿着将噪声映射到动作的多步常微分方程（ODE）进行时间反向传播（BPTT），这在计算上是不可行的且数值上不稳定。我们提出了FlowDPG，这是一种专门为流匹配策略设计的DDPG风格方法，它在训练时将评论家梯度提炼为速度场，完全绕过了BPTT。直观上，FlowDPG结合了两个互补的向量：保持动作可行的演示驱动速度和引导其朝向更高价值的评论家驱动修正。我们的贡献有三方面：（1）一个无BPTT的提炼框架，使得在流匹配策略上实现稳定的DDPG风格策略改进，（2）通过三个明确的近似建立FlowDPG更新方向与普通确定性策略梯度之间的正式联系，以及（3）在一个长时间跨度、多阶段、双臂AirPods组装任务上的现实世界验证，其中FlowDPG达到了92%的端到端成功率，显著优于最近的强化学习方法，包括价值条件、辅助模块适应和基于伴随的评论家梯度方法。视频和更多结果请访问项目页面 https://flowdpg.github.io。
cs.RO / 86 / 2606.22319
EmbodiedUS-FS: Fast Slow Intelligence for Ultrasound Robotics
EmbodiedUS-FS：超声机器人快速慢速智能
Abstract
Robotic ultrasound scanning in real clinical environments requires both high-level clinical workflow reasoning and low-level closed-loop execution. Physicians natural-language instructions often contain implicit anatomical targets, procedural logic, image-quality requirements, and safety constraints, while execution is affected by patient motion, contact variations, and target drift. We propose a fast and slow hierarchical embodied ultrasound system for safe and interpretable robotic ultrasound assistance. The Slow Brain performs intent parsing and stage-wise task planning with knowledge augmentation from an API and handbook corpus, and generates executable plans through task-graph construction and structured plan verification. The Fast Brain fuses multimodal feedback, including ultrasound images, robot pose and force states, and patient-motion information, to refine local actions and perform image-quality-guided recovery behaviors. The system further integrates a Safety Shield and a hierarchical escalation policy to constrain risky actions and trigger replanning or human confirmation under persistent failures or safety-bound violations. Experiments on planning evaluation, closed-loop execution under dynamic perturbations, and safety-mechanism validation demonstrate that the proposed hierarchical design improves task success rates while reducing safety violations.
Chinese Translation
在真实临床环境中，机器人超声扫描需要高层次的临床工作流程推理和低层次的闭环执行。医生的自然语言指令通常包含隐含的解剖目标、程序逻辑、图像质量要求和安全约束，而执行过程受到患者运动、接触变化和目标漂移的影响。我们提出了一种快速和慢速层次化的具身超声系统，用于安全且可解释的机器人超声辅助。慢速大脑通过API和手册语料库进行意图解析和阶段性任务规划，并通过任务图构建和结构化计划验证生成可执行计划。快速大脑融合多模态反馈，包括超声图像、机器人姿态和力状态以及患者运动信息，以优化局部动作并执行基于图像质量的恢复行为。该系统进一步集成了安全保护机制和层次化升级策略，以限制风险行为，并在持续失败或安全边界违规时触发重新规划或人工确认。对规划评估、动态扰动下的闭环执行和安全机制验证的实验表明，所提出的层次化设计提高了任务成功率，同时减少了安全违规。
cs.RO / 87 / 2606.22332
Tactile Genesis: Exploring Tactile Sensors at Scale for Learning Dexterous Tasks
触觉生成：探索大规模触觉传感器在灵巧任务学习中的应用
Abstract
Tactile sensing is critical for contact-rich dexterous manipulation, yet it remains unclear which tactile abstractions a policy needs and when richer tactile fields justify their hardware cost. This is hard to study empirically: each sensor effectively defines a new robot, and no lab can replicate the same learning experiment across all of them. We present Tactile Genesis, a GPU-parallel tactile sensor simulation platform that exposes binary contact, contact depth, per-taxel kinematic force/torque, elastomer marker displacement, geometry-aware proximity, contact audio, and a voxelized temperature field (the first of its kind in robot learning physics simulation platforms) under a common interface, with configurable placement, resolution, and a realistic noise model (drift, hysteresis, dead taxels, crosstalk). It scales past 20,000 parallel environments and 1,000 taxels on a single GPU, improving throughput by 3 to 20 times over previous tactile simulators. We train teacher-student policies on three dexterous tasks, ablating sensor type, placement, resolution, and noise, and verify transfer to the real XHand1. Proprioception alone is insufficient on every task. Sensor placement dominates sensor type: fingertip-only coverage trails whole-hand coverage by a wide margin, while adding the palm and proximal phalanges closes most of the gap to the privileged teacher. Resolution matters far less than coverage: placing 200 taxels across the whole hand suffices across tasks. We find that force/torque per taxel is consistently the most useful sensor type. These results give concrete guidance for both future tactile hardware design for improving robot hands and policy-side observation choice in dexterous manipulation. https://neuroagents-lab.github.io/2026-tactile-genesis/
Chinese Translation
触觉感知对于接触丰富的灵巧操作至关重要，但仍不清楚政策需要哪些触觉抽象，以及何时更丰富的触觉场能够证明其硬件成本的合理性。这一问题难以通过实证研究来探讨：每个传感器实际上定义了一个新的机器人，没有实验室能够在所有传感器上重复相同的学习实验。我们提出了触觉生成（Tactile Genesis），这是一个GPU并行触觉传感器仿真平台，提供二进制接触、接触深度、每个传感器单元的运动力/扭矩、弹性体标记位移、几何感知的接近度、接触音频以及体素化温度场（在机器人学习物理仿真平台中首创）等功能，所有这些都在一个统一接口下实现，支持可配置的放置、分辨率和真实的噪声模型（漂移、滞后、死传感器单元、串扰）。该平台可扩展至超过20,000个并行环境和1,000个传感器单元，在单个GPU上提高了3到20倍的吞吐量，超越了之前的触觉仿真器。我们在三个灵巧任务上训练了教师-学生政策，消融了传感器类型、放置、分辨率和噪声，并验证了对真实XHand1的迁移。仅靠本体感知在每个任务中都不足够。传感器的放置方式主导了传感器类型：仅指尖覆盖的效果远不及整个手的覆盖，而添加手掌和近节指骨则大大缩小了与特权教师之间的差距。分辨率的重要性远不及覆盖：在整个手上放置200个传感器单元在各个任务中都足够。我们发现每个传感器单元的力/扭矩始终是最有用的传感器类型。这些结果为未来改进机器人手的触觉硬件设计和灵巧操作中的政策观察选择提供了具体指导。
cs.RO / 88 / 2606.22335
Multi-AUV Marine Life Tracking with Single Hydrophone Payloads via a Hidden Markov Model Equipped Particle Filter
通过配备隐马尔可夫模型的粒子滤波器实现多无人水下航行器的海洋生物追踪
Abstract
Researchers tag and track marine animals to study migration patterns, human impacts on behavior, and behavioral shifts due to climate change. Accurate data collection often requires tagging individual animals to collect spatio-temporal state estimates of the animal's geo-position and depth. Acoustic transmitters are prominent due to their continuous communication without requiring retrieval or surfacing to collect data. These transmitters emit underwater acoustic pulses that can be detected by hydrophones. However, the frequent movement of aquatic animals results in high data loss when the animal moves out of the detection range of a stationary hydrophone. Autonomous underwater vehicle (AUV) systems offer a solution for localizing transmitters with higher resolution over longer periods of time. Such systems previously deployed have often required multiple hydrophones mounted on a large frame carried by the AUV. This increases drag, limiting the speed at which the AUV can track highly mobile animals. This work provides an alternative by equipping multiple AUVs with a single compact hydrophone payload. A particle filter algorithm equipped with a hidden Markov model (HMM) behavioral motion model fuses measurements from multiple AUVs to estimate the transmitter's position. Real-world data shows a root mean square error (RMSE) of approximately 10 meters for short-term deployments, and a larger simulated dataset shows an RMSE of approximately 15 meters for longer deployments over a larger area. The HMM fit to historical animal movement data outperforms a generic velocity motion model, and both outperform a baseline random walk motion model.
Chinese Translation
研究人员对海洋动物进行标记和追踪，以研究迁徙模式、人类对行为的影响以及气候变化导致的行为变化。准确的数据收集通常需要对个体动物进行标记，以获取动物的地理位置和深度的时空状态估计。声学发射器因其无需回收或浮出水面即可进行持续通信而备受关注。这些发射器发出水下声脉冲，可以被水听器检测到。然而，水生动物的频繁移动导致当动物移出固定水听器的检测范围时，数据损失严重。自主水下航行器（AUV）系统提供了一种解决方案，可以在更长时间内以更高分辨率定位发射器。以往部署的此类系统通常需要多个水听器安装在AUV携带的大框架上。这增加了阻力，限制了AUV追踪高度移动动物的速度。本研究提供了一种替代方案，通过为多个AUV配备单个紧凑型水听器负载来实现。配备隐马尔可夫模型（HMM）行为运动模型的粒子滤波算法融合多个AUV的测量数据，以估计发射器的位置。实际数据表明，短期部署的均方根误差（RMSE）约为10米，而在更大区域的较大模拟数据集中，RMSE约为15米。HMM模型对历史动物运动数据的拟合优于通用速度运动模型，且两者均优于基线随机游走运动模型。
cs.RO / 89 / 2606.22338
Benchmarking Robot Memory Under Interference
干扰下的机器人记忆基准测试
Abstract
Robots deployed in realistic settings will accumulate experience across many sessions and tasks over their deployment. The robot's tasks may often require it to remember information from multiple sessions ago, making long-context robot memory important for real-world deployments. However, most robot-memory benchmarks today are based on single episodes or a short context. To measure how current robot memory systems perform on longer sessions with more distractions, we introduce RoboMME-Interference, a cross-session benchmark built on RoboMME. For each query episode, we construct a session history using the query's relevant prior demonstration followed by a controlled number of unrelated sessions, which we provide to the VLA as memory and measure accuracy. Running RoboMME's released memory-augmented $\pi_{0.5}$ variants unmodified through this benchmark, we find that while perceptual memory variants improve success when given the history without any distractors, they decay strongly and steadily as unrelated sessions accumulate. With this release, we emphasize the importance of long-context memory and robustness to interference and show that current systems largely fail on such capabilities. The project page, videos, code, and data are at https://robotmemorybench.com.
Chinese Translation
在现实环境中部署的机器人将在其使用过程中积累跨多个会话和任务的经验。机器人的任务往往需要其记住多次会话之前的信息，因此长时记忆对于实际部署的机器人至关重要。然而，目前大多数机器人记忆基准测试都是基于单一会话或短期上下文。为了评估当前机器人记忆系统在更长会话和更多干扰下的表现，我们引入了RoboMME-Interference，这是一个基于RoboMME构建的跨会话基准测试。对于每个查询会话，我们使用查询相关的先前演示构建会话历史，随后添加一定数量的无关会话，并将其提供给VLA作为记忆以测量准确性。在此基准测试中，运行RoboMME发布的记忆增强型$ ext{π}_{0.5}$变体，我们发现，尽管感知记忆变体在没有任何干扰的历史信息下提高了成功率，但随着无关会话的增加，其性能却显著且稳定地下降。通过此次发布，我们强调了长时记忆和对干扰的鲁棒性的重要性，并展示了当前系统在这些能力上大多存在不足。项目页面、视频、代码和数据可在https://robotmemorybench.com找到。
cs.RO / 90 / 2606.22360
A Taxonomy of Conceptual Alignment in Human-Robot Dialogue
人机对话中概念对齐的分类
Abstract
Successful conversations require speakers to align on the meaning of concepts, a challenging but crucial task for human-robot interaction. Understanding the process of establishing such alignment is hindered by competing interpretations of the term and isolated, unidirectional investigations of its design space. This paper argues for a design-centric understanding of conceptual alignment as a bidirectional and co-constructive process. We introduce a taxonomy that characterizes conceptual alignment dialogues along what triggers its initiation and what level(s) of conceptual understanding it concerns. We further present a dialogue act schema as an operational tool that captures the interactional moves through which alignment is achieved. Together, these contributions provide a structured foundation for analyzing, comparing, and designing conceptual alignment in human-robot interaction.
Chinese Translation
成功的对话要求发言者在概念的意义上达成一致，这是人机交互中一项具有挑战性但至关重要的任务。理解建立这种对齐的过程受到对该术语的不同解释以及对其设计空间的孤立、单向研究的阻碍。本文主张将概念对齐理解为一种双向和共同构建的过程，强调设计中心的视角。我们提出了一种分类法，描述了概念对齐对话的特征，包括其启动的触发因素以及涉及的概念理解层次。我们进一步提出了一种对话行为框架，作为一种操作工具，捕捉实现对齐的互动行为。综合这些贡献，为分析、比较和设计人机交互中的概念对齐提供了结构化的基础。
cs.RO / 91 / 2606.22397
Do Rigid-Body Simulators Dream of Soft Robots? Learning Contact-Rich Manipulation for Tendon-Driven Continuum Robots
刚体模拟器是否梦想着软机器人？为腱驱动连续机器人学习接触丰富的操控
Abstract
Learning contact-rich, whole-body manipulation for soft continuum robots is held back by the lack of simulation infrastructure that has accelerated rigid-robot manipulation. Existing soft robot simulators are physically grounded but lack the contact handling, actuation support, or learning integration needed for contact-rich manipulation; rigid-body approximations offer these capabilities but sacrifice physical grounding. We bridge this gap for tendon-driven continuum robots (TDCRs) by deriving a continuum-mechanics-informed discretization that places the soft robot natively inside MuJoCo, unifying tendon forces, body contact, and dynamics in a single physics pipeline. We validate the simulator against a Cosserat rod reference (static and dynamic) and real TDCR hardware. We then train state-based imitation learning policies via teleoperation in simulation and deploy them zero-shot to a physical 3-segment TDCR on a 7-DoF Franka arm across two contact-rich manipulation tasks. To our knowledge, this is the first demonstration of sim-to-real transfer for contact-rich manipulation with continuum robots.
Chinese Translation
由于缺乏能够加速刚体机器人操控的模拟基础设施，学习软连续机器人的接触丰富的全身操控受到限制。现有的软机器人模拟器虽然在物理上是有依据的，但缺乏进行接触丰富操控所需的接触处理、驱动支持或学习整合；而刚体近似虽然提供了这些能力，但牺牲了物理基础。我们通过推导一种基于连续介质力学的离散化方法，填补了这一空白，使腱驱动连续机器人（TDCR）能够原生地嵌入MuJoCo中，将腱力、身体接触和动力学统一在一个物理管道中。我们通过与Cosserat杆参考模型（静态和动态）以及真实的TDCR硬件进行验证，确保模拟器的准确性。随后，我们通过在模拟环境中的遥操作训练基于状态的模仿学习策略，并将其零样本部署到一个物理3段TDCR上，利用7自由度的Franka手臂完成两个接触丰富的操控任务。据我们所知，这是首次展示连续机器人在接触丰富操控中的模拟到现实转移。
cs.RO / 92 / 2606.22443
SPiralRoll: A Novel Adjustable-Stiffness Underactuated 3-DoF Joint with Torsion Springs for Rolling Robots
SPiralRoll：一种新型可调刚度的欠驱动三自由度关节，采用扭转弹簧用于滚动机器人
Abstract
Compliant mechanisms are important in robotics because they can improve adaptability, safety, and energy efficiency while reducing hardware complexity. This paper presents SPiralRoll, a novel torsion-spring-based underactuated compliant mechanism for rolling robots and compliant robotic actuation. The mechanism uses arc-distributed elastic members and two motor inputs to realize three physically observable output motions: rotational motion, radial expansion/contraction, and axial spin induced by nonlinear compliant deformation. Two configurations, namely full-arc and single-arc designs, are developed and experimentally evaluated. Beyond benchtop validation, the mechanism is integrated into a spherical rolling robot, where proof-of-concept experiments demonstrate forward rolling and turning. The results show that the full-arc design provides better structural support and smoother deformation, whereas the single-arc design yields larger deformation and stronger inertial excitation, making it more suitable for pendulum-driven rolling locomotion. Overall, SPiralRoll provides a low-cost, compact, and fully 3D-printable solution for underactuated compliant rolling robots and adaptive robotic joints.
Chinese Translation
顺应机制在机器人技术中至关重要，因为它们可以提高适应性、安全性和能效，同时减少硬件复杂性。本文提出了SPiralRoll，一种基于扭转弹簧的欠驱动顺应机制，适用于滚动机器人和顺应机器人驱动。该机制利用弧形分布的弹性元件和两个电机输入，实现三种物理可观察的输出运动：旋转运动、径向膨胀/收缩，以及由非线性顺应变形引起的轴向旋转。开发并实验评估了两种配置，即全弧设计和单弧设计。除了台面验证外，该机制还集成到一个球形滚动机器人中，概念验证实验展示了向前滚动和转向。结果表明，全弧设计提供了更好的结构支撑和更平滑的变形，而单弧设计则产生了更大的变形和更强的惯性激励，使其更适合于摆动驱动的滚动运动。总体而言，SPiralRoll为欠驱动顺应滚动机器人和自适应机器人关节提供了一种低成本、紧凑且完全可3D打印的解决方案。
cs.RO / 93 / 2606.22456
BLENDS: Bayesian Learning-Enhanced Deep Smoothing for GNSS-Denied Environments
BLENDS：基于贝叶斯学习增强的深度平滑算法用于GNSS缺失环境
Abstract
Maintaining accurate navigation during GNSS outages remains a significant challenge for autonomous systems relying on low-cost inertial sensors. While classical smoothing methods, such as the two-filter smoother and Rauch-Tung-Striebel smoother, exploit measurements collected before and after an outage, their performance remains limited by the accuracy of conventional GNSS measurements. This paper presents Bayesian learning-enhanced navigation with deep smoothing (BLENDS), a transformer-based framework that augments Bayesian smoothing with learned covariance adaptation and state correction. The proposed method preserves the statistical foundations of Bayesian estimation while leveraging data-driven learning to improve navigation accuracy. Evaluations on the quadrotor dataset with GNSS outages demonstrate that BLENDS consistently outperforms both model-based smoothers, achieving up to 25.6% improvement in the position root mean square error while also reducing estimation uncertainty. Furthermore, BLENDS learns to compensate for the systematic bias between conventional GNSS positioning and RTK ground truth, enabling accuracy beyond that achievable with conventional GNSS measurements alone. The results demonstrate the potential of learning-enhanced Bayesian smoothing for resilient and high-accuracy navigation in GNSS-challenged environments.
Chinese Translation
在GNSS失效期间保持准确导航仍然是依赖低成本惯性传感器的自主系统面临的重大挑战。虽然经典的平滑方法，如双滤波器平滑器和Rauch-Tung-Striebel平滑器，利用失效前后收集的测量数据，但其性能仍受限于传统GNSS测量的准确性。本文提出了一种基于贝叶斯学习增强的导航深度平滑框架（BLENDS），该框架通过学习的协方差自适应和状态修正来增强贝叶斯平滑。所提出的方法在保留贝叶斯估计的统计基础的同时，利用数据驱动学习来提高导航准确性。在带有GNSS失效的四旋翼数据集上的评估表明，BLENDS始终优于基于模型的平滑器，在位置均方根误差上提高了多达25.6%，同时减少了估计的不确定性。此外，BLENDS学习补偿传统GNSS定位与RTK真实值之间的系统偏差，使得准确性超越仅依赖传统GNSS测量所能达到的水平。结果表明，学习增强的贝叶斯平滑在GNSS挑战环境中具有实现韧性和高准确度导航的潜力。
cs.RO / 94 / 2606.22471
Scalable Multi-Task Data Generation via Reinforcement Learning for Language-Conditioned Bimanual Dexterous Manipulation
基于强化学习的可扩展多任务数据生成用于语言条件下的双手灵巧操作
Abstract
A key bottleneck in training generalist policies for bimanual dexterous manipulation is the lack of large-scale, high-quality datasets. Synthetic data generation in simulation provides a scalable alternative to human video demonstrations by overcoming challenges such as morphology mismatch, missing physical interactions, and the generation of robot actions. However, existing approaches based on human teleoperation offer limited task diversity, as object-centric trajectory matching often neglects the feasibility of robot execution. Reinforcement learning (RL) enables broader scalability but is often constrained by handcrafted, task-specific rewards. In this work, we propose a systematic RL-based data generation pipeline that integrates generalizable reward design, effective domain randomization, and language-conditioned task annotations. This pipeline synthesizes diverse, high-quality datasets for dexterous bimanual manipulation and enables training of language-conditioned multi-task policies. Our experiments show that the generated data significantly improves generalization across three representative manipulation tasks.
Chinese Translation
训练通用策略以进行双手灵巧操作的一个关键瓶颈是缺乏大规模、高质量的数据集。模拟中的合成数据生成提供了一种可扩展的替代方案，克服了形态不匹配、缺失物理交互和机器人动作生成等挑战，优于人类视频演示。然而，基于人类遥操作的现有方法提供的任务多样性有限，因为以物体为中心的轨迹匹配往往忽视了机器人执行的可行性。强化学习（RL）能够实现更广泛的可扩展性，但通常受到手工设计的特定任务奖励的限制。在本研究中，我们提出了一种系统的基于RL的数据生成管道，该管道集成了可推广的奖励设计、有效的领域随机化和语言条件下的任务注释。该管道合成了多样化的高质量数据集，用于灵巧的双手操作，并能够训练语言条件下的多任务策略。我们的实验表明，生成的数据显著提高了在三个代表性操作任务中的泛化能力。
cs.RO / 95 / 2606.22480
ARP: Enhancing Quantized Skill Abstractions via Visual Alignment and Iterative Refinement for Robotic Manipulation
ARP：通过视觉对齐和迭代精炼增强量化技能抽象以实现机器人操作
Abstract
Learning visuomotor policies for long-horizon manipulation remains a fundamental challenge. Recent skill-based imitation learning methods based on discrete quantization have shown promising results by representing complex behaviors as temporally extended skills. However, most existing approaches primarily encode action trajectories into latent skills, yielding weak visual-semantic grounding and limiting the ability to leverage visual observations for skill selection. Moreover, discrete tokenization inevitably incurs precision errors during continuous action generation. To alleviate these issues, we propose Aligned Refinement Policy (ARP), a discrete-skill framework that couples semantic grounding with execution-level refinement. Specifically, ARP introduces (i) a visual--action alignment objective that contrastively aligns visual embeddings with pre-quantized action representations in a shared latent space while preserving a state-independent skill decoder, and (ii) a lightweight Iterative Residual Head (IRH) that performs a two-step refinement to recover fine-grained control for precise execution. Extensive experiments show that ARP achieves state-of-the-art performance on the LIBERO and Meta-World benchmarks. Moreover, real-robot experiments on the Kuavo 4 Pro humanoid platform further validate its effectiveness, yielding consistent performance gains over several baselines on two challenging manipulation tasks.
Chinese Translation
学习长时间跨度的视觉运动策略仍然是一个基本挑战。基于离散量化的近期技能模仿学习方法通过将复杂行为表示为时间扩展的技能，显示出良好的效果。然而，大多数现有方法主要将动作轨迹编码为潜在技能，导致视觉-语义基础较弱，限制了利用视觉观察进行技能选择的能力。此外，离散标记化在连续动作生成过程中不可避免地会产生精度误差。为了解决这些问题，我们提出了对齐精炼策略（Aligned Refinement Policy，ARP），这是一种将语义基础与执行级精炼相结合的离散技能框架。具体而言，ARP引入了（i）一种视觉-动作对齐目标，该目标在共享潜在空间中对比性地将视觉嵌入与预量化的动作表示对齐，同时保持状态无关的技能解码器，以及（ii）一种轻量级的迭代残差头（Iterative Residual Head，IRH），它执行两步精炼以恢复精细控制以实现精确执行。大量实验表明，ARP在LIBERO和Meta-World基准测试中达到了最先进的性能。此外，在Kuavo 4 Pro人形平台上的真实机器人实验进一步验证了其有效性，在两个具有挑战性的操作任务中相较于多个基线实现了一致的性能提升。
cs.RO / 96 / 2606.22491
PenduMorph: Development and Motion Analysis of Pendulum-Actuated Rolling Reconfigurable Spherical Robot with Magnetic-Coupling
PenduMorph：具有磁耦合的摆动驱动可重构滚动球形机器人的开发与运动分析
Abstract
This paper presents "PenduMorph", a wireless reconfigurable rolling spherical robot designed as a modular platform for enclosed locomotion and inter-module interaction in challenging environments. The proposed robot extends our previous pendulum-actuated rolling disk concept to a fully enclosed spherical architecture integrating a 2-DoF internal pendulum, onboard control, battery-powered operation, and magnetic docking. The design aims to combine independent rolling mobility with protected hardware and reliable reconfigurability. We first present the robot design and an analytical study of the magnetic coupling mechanism to evaluate retention and interaction between coupled modules. We then experimentally investigate key motion behaviors at both the single-module and dual-module levels, including independent rolling, magnetic coupling, and coordinated coupled motion. The results show that the proposed platform enables stable wireless operation and a set of distinctive reconfigurable rolling behaviors, providing a useful foundation for future modular spherical robots operating in contact-rich and demanding environments.
Chinese Translation
本文介绍了 "PenduMorph"，一种无线可重构滚动球形机器人，旨在作为一个模块化平台，用于在复杂环境中实现封闭的运动和模块间的交互。所提出的机器人将我们之前的摆动驱动滚动盘概念扩展到一个完全封闭的球形结构，集成了2自由度（2-DoF）内部摆锤、车载控制、蓄电池供电操作和磁性对接。该设计旨在将独立的滚动移动性与受保护的硬件和可靠的可重构性相结合。我们首先介绍了机器人的设计以及对磁耦合机制的分析研究，以评估耦合模块之间的保持和交互。然后，我们在单模块和双模块层面上实验性地研究了关键运动行为，包括独立滚动、磁耦合和协调耦合运动。结果表明，所提出的平台能够实现稳定的无线操作和一系列独特的可重构滚动行为，为未来在接触丰富和要求苛刻的环境中运行的模块化球形机器人提供了有用的基础。
cs.RO / 97 / 2606.22670
Semantic-Aware Autonomous Exploration for UAVs in Unknown Indoor Environments
面向无人机在未知室内环境中的语义感知自主探索
Abstract
Autonomous exploration in unknown environments requires unmanned aerial vehicles (UAVs) to efficiently generate informative trajectories while simultaneously constructing accurate maps. Although many existing exploration methods rely on geometric information, they often lack semantic awareness, resulting in suboptimal exploration efficiency and limited environmental understanding. To address this limitation, this paper proposes a semantic-aware exploration framework that adds semantic information to a roadmap-based exploration strategy. The proposed method builds on the Dynamic Exploration Planner (DEP), which incrementally constructs a Probabilistic Roadmap (PRM), and augments this roadmap with a semantic layer. A semantic reward function is introduced to prioritize regions containing meaningful objects and structures, enabling the UAV to focus on areas with higher information value. Furthermore, the roadmap is continuously updated to support efficient frontier selection and path planning during exploration. The proposed framework is implemented in ROS Noetic and Gazebo using an RGB-D sensor for simultaneous acquisition of geometric and semantic information. Experimental results in multiple simulated environments demonstrate that the proposed approach achieves exploration coverage rates between 90% and 94% while reducing exploration time and travel distance compared with conventional geometry-based exploration methods.
Chinese Translation
在未知环境中的自主探索要求无人机（UAV）有效生成信息丰富的轨迹，同时构建准确的地图。尽管许多现有的探索方法依赖于几何信息，但它们往往缺乏语义感知，导致探索效率低下和环境理解有限。为了解决这一局限性，本文提出了一种语义感知探索框架，该框架将语义信息添加到基于路线图的探索策略中。所提出的方法基于动态探索规划器（Dynamic Exploration Planner, DEP），该规划器逐步构建概率路线图（Probabilistic Roadmap, PRM），并在此路线图上增加一个语义层。引入了一种语义奖励函数，以优先考虑包含有意义物体和结构的区域，使无人机能够专注于信息价值更高的区域。此外，路线图会不断更新，以支持在探索过程中的高效前沿选择和路径规划。所提出的框架在ROS Noetic和Gazebo中实现，使用RGB-D传感器同时获取几何和语义信息。在多个模拟环境中的实验结果表明，所提方法实现了90%到94%之间的探索覆盖率，同时与传统的基于几何的探索方法相比，减少了探索时间和行驶距离。
cs.RO / 98 / 2606.22672
Efficient Continuous Semantic Mapping based on Spatio-Temporal Awareness
基于时空意识的高效连续语义映射
Abstract
Continuous semantic mapping allows autonomous robots to understand both the spatial structure and the semantic content of complex environments. However, most existing methods process the entire space, treat voxels as independent units, and do not keep the semantic labels consistent over time. This leads to high computational cost and reduced robustness in dynamic scenes. This paper proposes a semantic mapping method that brings spatial and temporal relationships into the semantic inference process. The method adjusts the inference range according to the local semantic uncertainty and fuses labels over time to improve map stability and computational efficiency. Experiments on the SemanticKITTI dataset show that the proposed method improves mapping accuracy by about 12% and reaches an mIoU of 54.92%, which is 13.18 percentage points higher than spatial-only mapping. These results show that spatiotemporal reasoning is effective for continuous semantic mapping in autonomous robotic systems.
Chinese Translation
连续语义映射使自主机器人能够理解复杂环境的空间结构和语义内容。然而，大多数现有方法处理整个空间，将体素视为独立单元，并且未能在时间上保持语义标签的一致性。这导致了高计算成本和在动态场景中降低的鲁棒性。本文提出了一种语义映射方法，将空间和时间关系引入语义推理过程。该方法根据局部语义不确定性调整推理范围，并随着时间的推移融合标签，以提高地图的稳定性和计算效率。在SemanticKITTI数据集上的实验表明，所提出的方法将映射精度提高了约12%，并达到了54.92%的mIoU，比仅基于空间的映射高出13.18个百分点。这些结果表明，时空推理在自主机器人系统的连续语义映射中是有效的。
cs.RO / 99 / 2606.22682
Integrated cloud-based architecture for robot-robot and human-robot collaboration using ROS 2--MQTT in Mediterranean Greenhouses
基于云的集成架构用于地中海温室中的机器人-机器人和人-机器人协作，采用ROS 2--MQTT
Abstract
The imperative to develop more sustainable agriculture demands a transition from isolated automation toward the deployment of multi-robot systems (MRS) in agrifood environments. However, Mediterranean greenhouse settings-characterized by narrow corridors, dense biomass, and structural metallic interference-pose significant challenges for robust and scalable communication between agents. Traditional robotic frameworks, such as ROS 2, frequently encounter node discovery issues and latency spikes due to dynamic obstacles, dense foliage, and other characteristic greenhouse elements, creating a critical bottleneck for real-time coordination. This paper proposes an innovative cloud-based hybrid architecture that establishes a two-way communication bridge between ROS 2, acting as an edge computing platform, and iVeg as a Decision Support System (DSS), using MQTT and the European FIWARE platform. The proposed framework enables seamless interoperability between fleets of multiple robots in environments with communication constraints, facilitating the synchronised exchange of high-level telemetry, point cloud data and farmer identification for collaboration, amongst other critical data. The architecture was validated in a high-fidelity simulation environment and subsequently tested in a real-world greenhouse scenario, demonstrating its ability to maintain persistent connectivity and data integrity under adverse network conditions. The results indicate that the integration of MQTT effectively eliminates information silos, providing a scalable and decentralised solution for managing complex robotic missions, which are executed locally via Edge Computing. This work sets a new methodological precedent for the concept of Greenhouse Models as a Service (GMaaS), bridging the gap between low-level robotic control and high-level, cloud-based IoT decision-making.
Chinese Translation
发展更可持续的农业的迫切需求要求从孤立的自动化转向在农业食品环境中部署多机器人系统（MRS）。然而，地中海温室环境——以狭窄的走廊、密集的生物量和结构性金属干扰为特征——对代理之间的稳健和可扩展通信提出了重大挑战。传统的机器人框架，如ROS 2，常常因动态障碍物、密集的植被和其他温室特征而遇到节点发现问题和延迟峰值，这为实时协调创造了关键瓶颈。本文提出了一种创新的基于云的混合架构，建立了ROS 2（作为边缘计算平台）与iVeg（作为决策支持系统，DSS）之间的双向通信桥梁，使用MQTT和欧洲FIWARE平台。所提出的框架使得在通信受限的环境中，多个机器人队伍之间实现无缝互操作成为可能，促进了高层遥测、点云数据和农民身份等关键数据的同步交换，以便进行协作。该架构在高保真模拟环境中进行了验证，并随后在真实的温室场景中进行了测试，证明其在不利网络条件下保持持续连接和数据完整性的能力。结果表明，MQTT的集成有效消除了信息孤岛，为管理复杂的机器人任务提供了可扩展和去中心化的解决方案，这些任务通过边缘计算在本地执行。本研究为温室模型即服务（GMaaS）的概念设定了新的方法论先例，弥合了低级机器人控制与基于云的高层物联网决策之间的差距。
cs.RO / 100 / 2606.22729
Temporal Logic Guidance for Action-Only Diffusion Policies with World Models
基于时间逻辑的仅动作扩散策略引导与世界模型
Abstract
Diffusion policies enable multimodal robot behavior but offer limited ability to choose among behavior modes at inference time, even though such control is desirable in human-robot settings. Prior solutions to this lack of control have utilized Signal Temporal Logic (STL) to express human intentions and provide corresponding guidance for diffusion policy inference. However, these approaches can only guide diffusion policies that jointly generate future actions and states, increasing both complexity and runtime. We propose a novel guidance method for action-only diffusion policies that uses a separate learned world model to enable differentiable evaluation of STL robustness, with its gradient then injected into the diffusion process. This steers behavior toward constraint satisfaction without retraining, improving constraint adherence while preserving task performance. On the Can Transport task from Robomimic, our method maintains 100% task success while reducing constraint violations from over 80% for baseline methods to 4%. We also discuss extensions toward improved robustness and more complex constraints.
Chinese Translation
扩散策略能够实现多模态机器人行为，但在推理时选择行为模式的能力有限，尽管这种控制在人与机器人交互中是可取的。之前针对这一控制缺失的解决方案利用信号时间逻辑（Signal Temporal Logic, STL）来表达人类意图，并为扩散策略推理提供相应的引导。然而，这些方法只能引导同时生成未来动作和状态的扩散策略，从而增加了复杂性和运行时间。我们提出了一种新颖的仅动作扩散策略引导方法，该方法使用单独学习的世界模型来实现STL鲁棒性的可微评估，并将其梯度注入到扩散过程中。这使得行为朝向约束满足的方向引导，而无需重新训练，改善了约束遵循性，同时保持任务性能。在Robomimic的Can Transport任务中，我们的方法在保持100%任务成功率的同时，将基线方法的约束违反率从80%以上降低至4%。我们还讨论了向改进鲁棒性和更复杂约束的扩展。
cs.RO / 101 / 2606.22756
HERCULES: An Open-Source Simulation Framework for Heterogeneous Multi-Robot SLAM, Collaborative Perception, and Exploration
HERCULES：一个用于异构多机器人SLAM、协作感知和探索的开源仿真框架
Abstract
We present HERCULES, an open-source simulator and data-collection pipeline for heterogeneous multi-robot autonomy. Built upon the Unreal Engine 5 (UE5)-based simulators AirSim and Cosys-AirSim, HERCULES resolves key architectural limitations of prior frameworks to enable concurrent unmanned aerial and ground vehicle (UAV-UGV) operation in large-scale, photorealistic, dynamic environments. It introduces a new waypoint-tracking UGV controller that mirrors existing UAV control interfaces, and provides a shared navigation stack for mapping, traversability analysis, planning, and control across heterogeneous platforms. Expanding inherited sensor suites, it adds physics-based long-wave infrared (LWIR) cameras and configurable night-vision modes for degraded visual environments. HERCULES provides lightweight APIs, ROS 2 wrappers, and rigorous time synchronization across sensors and platforms, and brings state-of-the-art game-engine capabilities into robotics simulation, integrating intelligent agents such as pedestrians, traffic, and wildlife with high-fidelity dynamic phenomena, including fire, flooding, and crop disease spread. HERCULES runs in two modes: passively, replaying offline-designed trajectories to generate reproducible multi-modal datasets, and actively, running an online planner in closed loop from live observations. Our experiments in heterogeneous multi-robot SLAM, collaborative perception, and exploration, using both HERCULES-generated data and active closed-loop execution, demonstrate its utility for advancing heterogeneous multi-robot autonomy. We publicly release our source code, experiment code, documentation, and datasets, including a heterogeneous multi-robot SLAM benchmark collected with two UAVs and two UGVs across kilometer-scale desert, forest, and city environments, at https://lunarlab-gatech.github.io/HERCULES-website.
Chinese Translation
我们提出了HERCULES，一个用于异构多机器人自主的开源仿真器和数据收集管道。HERCULES基于Unreal Engine 5 (UE5)的仿真器AirSim和Cosys-AirSim，解决了先前框架的关键架构限制，使无人机（UAV）和地面车辆（UGV）能够在大规模、逼真的动态环境中并行操作。它引入了一种新的航点跟踪UGV控制器，模仿现有的UAV控制接口，并提供一个共享的导航栈，用于映射、可通行性分析、规划和跨异构平台的控制。HERCULES扩展了继承的传感器套件，增加了基于物理的长波红外（LWIR）摄像头和可配置的夜视模式，以应对视觉环境的退化。HERCULES提供轻量级API、ROS 2包装器，以及跨传感器和平台的严格时间同步，并将最先进的游戏引擎能力引入机器人仿真，集成了行人、交通和野生动物等智能代理，以及火灾、洪水和作物病害传播等高保真动态现象。HERCULES运行在两种模式下：被动模式，通过重放离线设计的轨迹生成可重复的多模态数据集；主动模式，从实时观测中运行闭环在线规划器。我们在异构多机器人SLAM、协作感知和探索方面的实验，使用HERCULES生成的数据和主动闭环执行，展示了其在推动异构多机器人自主方面的实用性。我们公开发布了源代码、实验代码、文档和数据集，包括在公里级沙漠、森林和城市环境中使用两架UAV和两辆UGV收集的异构多机器人SLAM基准，网址为https://lunarlab-gatech.github.io/HERCULES-website。
cs.RO / 102 / 2606.22757
Cooperative-ORCA*: Real-Time Proactive Deadlock Avoidance for Continuous-Space Multi-Agent Navigation
合作-ORCA*: 连续空间多智能体导航的实时主动死锁避免
Abstract
Multi-Agent Path Finding (MAPF) is a problem that requires computing collision-free paths for a set of agents from their start locations to designated goal locations. The problem has broad applications in domains where teams of robots must operate in a coordinated manner. ORCA* is a real time MAPF solver that assigns for each timestep a velocity for each agent. Due to its real time nature, it is myopic to future deadlocks that result from current decisions. ORCA*-MAPF attempts to remedy this limitation by introducing fallback mechanisms when deadlocks are detected. However, post hoc interventions often introduce significant flowtime overhead. In this paper, we introduce C-ORCA* and C-ORCA*-MAPF, continuous space MAPF algorithms that incorporate agents' entire spatial trajectory and their spatial dependencies to proactively prevent deadlocks from occurring, thus avoiding the high flowtime overhead associated with post hoc corrections in ORCA*-MAPF. The C-ORCA* family of algorithms significantly outperform previous state-of-the-art in terms of solve rate, runtime, and flowtime.
Chinese Translation
多智能体路径规划（MAPF）是一个需要为一组智能体从其起始位置计算无碰撞路径到指定目标位置的问题。该问题在机器人团队需要协调操作的领域具有广泛的应用。ORCA* 是一个实时的 MAPF 求解器，为每个时间步分配每个智能体的速度。由于其实时特性，它对当前决策导致的未来死锁具有短视性。ORCA*-MAPF 试图通过在检测到死锁时引入后备机制来弥补这一限制。然而，事后干预往往会引入显著的流动时间开销。在本文中，我们提出了 C-ORCA* 和 C-ORCA*-MAPF，这两种连续空间 MAPF 算法整合了智能体的整个空间轨迹及其空间依赖关系，以主动防止死锁的发生，从而避免与 ORCA*-MAPF 中事后修正相关的高流动时间开销。C-ORCA* 系列算法在求解率、运行时间和流动时间方面显著超越了之前的最先进技术。
cs.RO / 103 / 2606.22794
UniFS: Unified Fast-to-Slow Hierarchical Architecture for Vision-Language-Action Models
UniFS：用于视觉-语言-动作模型的统一快慢层次架构
Abstract
Mainstream Fast-Slow dual system vision-language-action models decouple a high-frequency action expert from a low-frequency vision-language model for efficiency, yet they face a fundamental frequency dilemma: large update gaps cause semantic drift from stale context, while small gaps erode the intended computational savings. Moreover, because the action expert receives only the VLM's final-layer representation at a single fixed frequency, rich intermediate features are discarded, limiting both information coupling and manipulation precision. Inspired by multi-timescale neural processing in the human brain, we introduce UniFS, a unified fast-to-slow architecture that resolves these challenges through three key designs. First, we stratify the VLM layers into groups with progressively decreasing update frequencies, enabling shallow layers to capture fast-changing dynamics while deeper layers cache stable semantic context. Second, a latent vector inversion mechanism re-routes the interaction order between multi-scale VLM features and the action expert, aligning fast-varying representations with fine-grained action decoding and slow-varying ones with coarse planning. Third, a multi-level supervision strategy enforces a coarse-to-fine learning hierarchy across temporal scales. Together, these designs enable richer cross-frequency information transfer within a single backbone, while the low-frequency pathways additionally preserve temporal context across steps. Experiments on LIBERO show that UniFS achieves state-of-the-art performance (98.3\% average success rate, a 2.5\% gain over VLA-Adapter baseline) while reducing average inference latency from 36.5~ms to 17.8~ms (2.1$\times$ speedup). Real-robot experiments on a Franka platform further validate its practical applicability. Code is opensourced at https://github.com/linsun449/UniFS.
Chinese Translation
主流的快慢双系统视觉-语言-动作模型将高频动作专家与低频视觉-语言模型解耦以提高效率，但它们面临一个基本的频率困境：大的更新间隔会导致语义漂移，因而失去过时的上下文，而小的间隔则削弱了预期的计算节省。此外，由于动作专家仅在单一固定频率下接收视觉-语言模型（VLM）最后一层的表示，丰富的中间特征被丢弃，从而限制了信息耦合和操作精度。受到人脑多时间尺度神经处理的启发，我们提出了UniFS，这是一种统一的快慢架构，通过三个关键设计来解决这些挑战。首先，我们将VLM层分层为更新频率逐渐降低的组，使得浅层能够捕捉快速变化的动态，而深层则缓存稳定的语义上下文。其次，潜在向量反转机制重新调整多尺度VLM特征与动作专家之间的交互顺序，将快速变化的表示与精细的动作解码对齐，而将慢变化的表示与粗略规划对齐。第三，多层次监督策略在时间尺度上强制实施粗到细的学习层次。综合这些设计，使得在单一主干内实现更丰富的跨频率信息传递，同时低频路径还保留了跨步骤的时间上下文。在LIBERO上的实验表明，UniFS实现了最先进的性能（98.3\%的平均成功率，比VLA-Adapter基线提高了2.5\%），同时将平均推理延迟从36.5毫秒减少到17.8毫秒（加速2.1倍）。在Franka平台上的真实机器人实验进一步验证了其实际应用性。代码已开源，地址为https://github.com/linsun449/UniFS。
cs.RO / 104 / 2606.22836
Cloak: Zero-Shot Cross-Embodiment Manipulation by Masking the End-Effector from the VLA
Cloak：通过遮蔽末端执行器实现零-shot跨体现操控
Abstract
We present Cloak, a training recipe that endows a Vision-Language-Action (VLA) model with zero-shot cross-embodiment transfer by cloaking the end-effector from its own wrist camera. The end-effector occupies a large and consistent region of the wrist view and masking it allows for embodiment-agnostic visual reasoning. Cloak renders a mask in simulation from the robot's known geometry, accurately and in real time, with no segmentation or generative models. During training, we augment the mask so the model generalizes to embodiments unseen at training time. We demonstrate the recipe with Cloak-VLA, a VLA trained with Cloak on a single parallel-jaw gripper dataset. No data of new embodiments is ever collected. Cloak-VLA transfers zero-shot to various unseen embodiments, including another gripper, another arm, and a five-fingered hand, while preserving the source embodiment's performance. By decoupling the wrist view from its own embodiment, Cloak allows data to outlive the hardware it was collected on.
Chinese Translation
我们提出了Cloak，这是一种训练方案，使得视觉-语言-动作（VLA）模型能够通过遮蔽末端执行器与其自身腕部摄像头之间的视角，实现零-shot跨体现转移。末端执行器在腕部视角中占据了一个大而一致的区域，遮蔽它可以实现与体现无关的视觉推理。Cloak在仿真中根据机器人的已知几何形状实时准确地渲染出一个遮蔽，而无需分割或生成模型。在训练过程中，我们增强了遮蔽，以便模型能够推广到训练时未见过的体现。我们用Cloak-VLA演示了这一方案，Cloak-VLA是一个在单一平行夹爪数据集上使用Cloak训练的VLA模型。未曾收集新的体现数据。Cloak-VLA能够零-shot转移到各种未见的体现，包括另一个夹爪、另一只手臂以及一个五指手，同时保持源体现的性能。通过将腕部视角与其自身体现解耦，Cloak使得数据能够超越其收集时所用的硬件。
cs.RO / 105 / 2606.22838
FPAS: Frontier-Based Path Planning with Adaptive Sampling for Large-Scale Unknown Environments
FPAS：基于边界的路径规划与自适应采样在大规模未知环境中的应用
Abstract
In this work, we propose Frontier-based Path Planning with Adaptive Sampling (FPAS), a novel framework designed for efficient goal-reaching in large-scale, unknown environments. While existing planners often struggle with computational bottlenecks or inefficient paths during long-range navigation, FPAS overcomes these challenges by reinterpreting the frontier concept for goal-directed tasks. Specifically, our method leverages frontiers to effectively guide forward progression into unobserved regions and to select promising subgoals for backtracking from dead-ends or inefficient paths. Furthermore, FPAS introduces an adaptive sampling mechanism based on a frontier-derived openness metric. This mechanism dynamically adjusts the global graph's density by employing sparse nodes in open areas to alleviate computational burdens, while preserving denser sampling in narrow passages to ensure connectivity. Extensive evaluations demonstrate that FPAS substantially improves computational efficiency over baseline methods while maintaining highly competitive goal-reaching performance.
Chinese Translation
在本研究中，我们提出了一种基于边界的路径规划与自适应采样（FPAS）框架，旨在实现大规模未知环境中的高效目标到达。现有的规划方法在长距离导航中常常面临计算瓶颈或路径效率低下的问题，而FPAS通过重新诠释边界概念来应对这些挑战，特别是在目标导向任务中。具体而言，我们的方法利用边界有效地引导向未观察区域的前进，并选择有前景的子目标，以便从死胡同或低效路径中回溯。此外，FPAS引入了一种基于边界派生开放度指标的自适应采样机制。该机制通过在开放区域采用稀疏节点来动态调整全局图的密度，以减轻计算负担，同时在狭窄通道中保持更密集的采样以确保连通性。广泛的评估表明，FPAS在保持高度竞争的目标到达性能的同时，显著提高了计算效率。
cs.RO / 106 / 2606.22860
HiL-ResRL: A Model-Agnostic Finetuning Adapter via Human-in-the-loop Residual Reinforcement Learning
HiL-ResRL：一种通过人机协作残差强化学习的模型无关微调适配器
Abstract
Recent advancements in generative imitation learning have significantly propelled the field of robotic manipulation. However, the majority of existing models rely heavily on Behavior Cloning (BC), a paradigm that suffers from compounding errors and distributional shift. Consequently, the efficacy of these models in practical industrial deployments remains limited. To address these challenges, we introduce a novel, plug-and-play fine-tuning pipeline designed to facilitate the robust deployment of Vision-Language-Action (VLA) models in real-world environments. In contrast to contemporary reinforcement learning (RL) fine-tuning strategies, which are often constrained by specific model architectures, our proposed framework is model-agnostic and adaptable to a diverse range of VLA models. We conceptualize VLA-generated actions as a unified interface, upon which we train a residual policy. This policy is designed to rectify suboptimal actions and address the distributional shift inherent in imitation learning. Additionally, we incorporate human-in-the-loop guidance to ensure safe exploration and maximize training efficiency. We conduct experiments directly in real-world robotic settings. The results demonstrate that within only 1.5 hour of real-world online RL training, the average success rate exceeds 95% on real robots. Our work presents a practical solution for deploying behavior cloning models in industrial scenarios.
Chinese Translation
最近，生成模仿学习的进展显著推动了机器人操作领域的发展。然而，现有大多数模型在很大程度上依赖于行为克隆（Behavior Cloning, BC），这一范式存在累积误差和分布偏移的问题。因此，这些模型在实际工业部署中的有效性仍然有限。为了解决这些挑战，我们提出了一种新颖的即插即用微调管道，旨在促进视觉-语言-动作（Vision-Language-Action, VLA）模型在现实环境中的稳健部署。与当代强化学习（Reinforcement Learning, RL）微调策略相比，后者通常受到特定模型架构的限制，我们提出的框架是模型无关的，能够适应多种VLA模型。我们将VLA生成的动作概念化为一个统一接口，并在此基础上训练一个残差策略。该策略旨在纠正次优动作并解决模仿学习中固有的分布偏移。此外，我们引入了人机协作指导，以确保安全探索并最大化训练效率。我们在真实的机器人环境中直接进行实验。结果表明，在仅1.5小时的现实在线RL训练中，真实机器人上的平均成功率超过95%。我们的工作为在工业场景中部署行为克隆模型提供了一个实用的解决方案。
cs.RO / 107 / 2606.22881
A Vendor-Agnostic LiDAR Data Conversion System with Multi-Signal Detection and Multi-Format Output
一种与供应商无关的激光雷达数据转换系统，具备多信号检测和多格式输出
Abstract
LiDAR (Light Detection and Ranging) sensors capture the surrounding environment as dense 3D point clouds by measuring the time-of-flight of emitted laser pulses, making them foundational across autonomous vehicles, robotics, and large-scale mapping. PCAP (Packet Capture) files from these sensors are the starting point of most 3D perception pipelines, yet internal packet structures, UDP (User Datagram Protocol) port conventions and encoding schemes differ enough across manufacturers that no single tool reads them all. Ouster, Velodyne, Hesai, and Livox each require their own SDK (Software Development Kit), their own environment setup, and their own conversion workflow. Supporting all four means maintaining four disconnected pipelines with no shared infrastructure. The pipeline described here takes a raw PCAP as input and handles vendor identification automatically, scoring six independent file characteristics through a weighted multi-signal approach to determine the source sensor. C++ SDKs handle Ouster and Velodyne, while Hesai and Livox rely on Python-based dpkt parsing where no open source SDK exists. From there, a single command writes output to any of five industry-standard formats. We tested on real outdoor captures. Ouster peaks at 2.08M points per second, Velodyne at 1.47M, both running through native C++ packet decoding. Hesai and Livox land at 110K and 150K respectively, where Python-layer parsing introduces overhead that compounds under sustained load. The 8-10x gap held consistently across runs. Tested on a consumer-grade i3 with 8GB RAM, no vendor configuration required
Chinese Translation
激光雷达（LiDAR，光学探测与测距）传感器通过测量发射激光脉冲的飞行时间，将周围环境捕捉为密集的三维点云，使其在自动驾驶车辆、机器人和大规模制图中具有基础性作用。这些传感器生成的PCAP（数据包捕获）文件是大多数三维感知管道的起点，然而，不同制造商之间的内部数据包结构、UDP（用户数据报协议）端口约定和编码方案差异足够大，以至于没有单一工具可以读取所有数据。Ouster、Velodyne、Hesai和Livox各自需要自己的SDK（软件开发工具包）、环境设置和转换工作流程。支持这四种设备意味着需要维护四个不相连的管道，且没有共享基础设施。本文描述的管道以原始PCAP作为输入，自动处理供应商识别，通过加权多信号方法对六个独立文件特征进行评分，以确定源传感器。C++ SDK处理Ouster和Velodyne，而Hesai和Livox则依赖于Python基础的dpkt解析，因为没有开放源代码的SDK可用。在此基础上，单个命令可以将输出写入五种行业标准格式中的任何一种。我们在真实的户外捕获数据上进行了测试。Ouster的峰值为每秒2.08百万点，Velodyne为每秒1.47百万点，均通过原生C++数据包解码运行。Hesai和Livox分别为每秒110K和150K，其中Python层解析引入的开销在持续负载下会加重。8-10倍的差距在多次运行中保持一致。在一台配备8GB RAM的消费级i3上进行测试，无需供应商配置。
cs.RO / 108 / 2606.22907
Improving Robotic Imitation Learning via Trajectory Standardization
通过轨迹标准化提升机器人模仿学习
Abstract
Imitation learning for robotic manipulation relies on large sets of human demonstration trajectories, which are often noisy and temporally irregular due to variable operator speed, intermittent pauses, and inconsistent action density. A common preprocessing strategy is time-uniform downsampling to shorten sequences, but it cannot effectively remove speed-induced non-uniformity or redundant pauses. This mismatch degrades data quality and hinders policy learning. To address this issue, we propose Information-Standardized Trajectory Resampling (ISR), an offline preprocessing method for effective imitation learning. ISR resamples each trajectory by enforcing approximately equal information distance between adjacent points. Specifically, we map trajectories onto an information-modulated Riemannian manifold and perform geodesic-equidistant parameterization. We construct an information-intensity field from velocity and acceleration norms: the velocity term removes small-motion redundancy, while the acceleration term preserves high-curvature and fine-manipulation phases. We evaluate ISR on three real-world manipulation tasks with mainstream imitation learning policies. Compared with the baseline time-uniform 3x downsampling, ISR improves task success rates by about 25%, remains robust across datasets collected from different operators, and reduces both dataset size and training cost. The code and videos are publicly available at https://d-robotics-ai-lab.github.io/isr.page.
Chinese Translation
机器人操作的模仿学习依赖于大量人类示范轨迹，这些轨迹由于操作员速度的变化、间歇性暂停和不一致的动作密度，往往存在噪声和时间不规则性。一种常见的预处理策略是时间均匀下采样，以缩短序列，但它无法有效去除速度引起的非均匀性或冗余暂停。这种不匹配降低了数据质量，阻碍了策略学习。为了解决这个问题，我们提出了信息标准化轨迹重采样（Information-Standardized Trajectory Resampling, ISR），这是一种用于有效模仿学习的离线预处理方法。ISR通过强制相邻点之间的近似等信息距离来重采样每个轨迹。具体而言，我们将轨迹映射到信息调制的黎曼流形上，并执行测地线等距参数化。我们从速度和加速度范数构建信息强度场：速度项去除小运动冗余，而加速度项保留高曲率和精细操作阶段。我们在三个真实世界的操作任务上评估ISR，并与主流模仿学习策略进行比较。与基线的时间均匀3倍下采样相比，ISR将任务成功率提高了约25%，在不同操作员收集的数据集上表现出稳健性，并减少了数据集大小和训练成本。代码和视频可在 https://d-robotics-ai-lab.github.io/isr.page 上公开获取。
cs.RO / 109 / 2606.22923
PanoVine: Whole-Body Visuomotor Control for Soft Growing Vine Robot
PanoVine：软体生长藤蔓机器人全身视觉运动控制
Abstract
Vine robots, a class of soft, growing robots, are suitable for navigating complex and confined environments due to their compliant bodies and self-supporting growth mechanism. However, hysteresis, tether interactions, and deformations make them difficult to predict and model, which in turn limits the effectiveness of conventional planning and control approaches. In this work, we present a data-driven, vision-based control framework for the first autonomous vine robot system. Our system integrates 19 cameras distributed along the robot's body to provide comprehensive feedback of both the robot state and the surrounding environment. Using this rich whole-body vision feedback, we train an end-to-end visuomotor policy from demonstrations for closed-loop autonomous control in complex environments. The policy efficiently aggregates information from distributed sensing while maintaining robustness to inaccurate robot states and actuation. Experimental results demonstrate that the learned policy enables robust navigation and manipulation in challenging scenarios, including steering through branched structures, climbing up slopes, traversing unsupported terrain, reaching objects precisely, and maneuvering through confined spaces and obstacles. Project website https://panovine-bot.github.io
Chinese Translation
藤蔓机器人是一类软体生长机器人，因其柔韧的身体和自支撑的生长机制，适合在复杂和狭窄的环境中导航。然而，滞后效应、缆绳交互和变形使得它们难以预测和建模，从而限制了传统规划和控制方法的有效性。在本研究中，我们提出了一种数据驱动的基于视觉的控制框架，适用于首个自主藤蔓机器人系统。我们的系统集成了分布在机器人身体上的19个摄像头，以提供对机器人状态和周围环境的全面反馈。利用这一丰富的全身视觉反馈，我们从演示中训练了一个端到端的视觉运动策略，以实现复杂环境中的闭环自主控制。该策略有效地整合了来自分布式传感器的信息，同时保持对不准确的机器人状态和驱动的鲁棒性。实验结果表明，学习到的策略能够在具有挑战性的场景中实现稳健的导航和操作，包括在分支结构中转向、爬坡、穿越不支持的地形、精确到达物体以及在狭小空间和障碍物中灵活移动。项目网站：https://panovine-bot.github.io
cs.RO / 110 / 2606.22971
Humanoid-OmniOcc: Stereo-Based Full-View Occupancy Dataset for Embodied AI
类人机器人全景占用数据集 Humanoid-OmniOcc：基于立体视觉的全视角占用数据集
Abstract
Occupancy prediction at voxel-level granularity is essential for safe robotic navigation and interaction in complex environments. Existing occupancy datasets, however, are predominantly designed for autonomous driving with vehicle-centric biases -- forward-facing cameras, far-field geometry, and static road priors -- limiting their applicability to embodied humanoid perception. We present Humanoid-OmniOcc, a large-scale panoramic stereo-based occupancy dataset tailored for humanoid robots. The dataset encompasses 15 diverse simulated indoor scenes and 5 real-world environments, yielding over 155K samples with broad scene and style diversity. Importantly, the dataset is designed around a Real2Sim2Real closed-loop paradigm: real sensor specifications drive physically accurate simulation, simulation produces large-scale annotated training data, and models trained in simulation are directly evaluated on real-world captures -- enabling iterative refinement of the sim-to-real pipeline. We further propose \textbf{H}umanoid \textbf{S}urround \textbf{S}tereo-guided \textbf{Occ}upancy model (Humanoid-OmniOcc) that exploits robust depth priors for accurate 2D-to-3D lifting. Extensive experiments show that Humanoid-OmniOcc consistently outperforms monocular baselines and generalizes well to both unseen simulated test scenes and real-world environments, validating the effectiveness of the Real2Sim2Real design. Code and data will be available upon acceptance at https://d-robotics-ai-lab.github.io/humanoid-omniocc.
Chinese Translation
在体素级别进行占用预测对于在复杂环境中安全的机器人导航和交互至关重要。然而，现有的占用数据集主要是为自动驾驶设计的，存在以车辆为中心的偏见——前向摄像头、远场几何和静态道路先验——限制了它们在类人机器人感知中的适用性。我们提出了 Humanoid-OmniOcc，这是一个大规模的基于全景立体视觉的占用数据集，专为类人机器人量身定制。该数据集涵盖了15个多样化的模拟室内场景和5个真实世界环境，产生了超过155K个样本，具有广泛的场景和风格多样性。重要的是，该数据集围绕 Real2Sim2Real 闭环范式设计：真实传感器规格驱动物理准确的模拟，模拟生成大规模标注训练数据，在模拟中训练的模型直接在真实世界捕获中进行评估——使得模拟到真实的流程能够进行迭代优化。我们进一步提出了类人周围立体引导占用模型（Humanoid-OmniOcc），该模型利用稳健的深度先验进行准确的2D到3D提升。大量实验表明，Humanoid-OmniOcc 在性能上始终优于单目基线，并且在未见过的模拟测试场景和真实世界环境中具有良好的泛化能力，验证了 Real2Sim2Real 设计的有效性。代码和数据将在接受后提供，网址为 https://d-robotics-ai-lab.github.io/humanoid-omniocc。
cs.RO / 111 / 2606.22982
Distilling Collaborative Dynamics into Latent Space for Implicit Coordination in Decentralized Multi-Agent Manipulation
将协作动态提炼至潜在空间以实现去中心化多智能体操控中的隐式协调
Abstract
Multi-arm manipulation demands precise spatiotemporal coordination, yet many centralized approaches scale poorly as team size increases. To address this, we propose CLS-DP, a decentralized multi-agent framework that enables implicit coordination under partial observability without shared global views, explicit state information, or inter-agent communication. Under the centralized training and decentralized execution (CTDE) paradigm, CLS-DP distills privileged multi-agent dynamics into a latent space. At deployment, each agent infers a collaborative latent from its local RGB observation and a shared task instruction; it then conditions the diffusion denoising process on this latent. This design enables implicit coordination with a per-agent cost independent of team size. Across six RoboFactory benchmark tasks spanning two to four agents, CLS-DP achieves a 38% mean success rate, outperforming the best centralized baseline (20%) and a decentralized ablation without the collaborative latent (9%). It also maintains superior parameter efficiency across all agent configurations. Attribution maps show that an agent conditioned on the collaborative latent places high attribution on the joints and grippers of both itself and its teammates throughout execution. This suggests that the learned latent efficiently encodes collaborative dynamics from local observation, which facilitates implicit coordination in realistic settings characterized by partial observability.
Chinese Translation
多臂操控要求精确的时空协调，然而许多集中式方法在团队规模增加时表现不佳。为了解决这个问题，我们提出了CLS-DP，一个去中心化的多智能体框架，能够在部分可观测性下实现隐式协调，而无需共享全局视图、显式状态信息或智能体间通信。在集中训练与去中心化执行（CTDE）范式下，CLS-DP将特权多智能体动态提炼到潜在空间。在部署时，每个智能体根据其本地RGB观测和共享任务指令推断出一个协作潜在；然后，它在此潜在上条件化扩散去噪过程。该设计使得每个智能体的成本与团队规模无关，从而实现隐式协调。在涵盖两个至四个智能体的六个RoboFactory基准任务中，CLS-DP实现了38%的平均成功率，超越了最佳集中式基线（20%）和没有协作潜在的去中心化消融实验（9%）。它在所有智能体配置中也保持了优越的参数效率。归因图显示，基于协作潜在条件化的智能体在执行过程中对自身及其队友的关节和夹具给予了高度的归因。这表明，学习到的潜在有效地编码了来自本地观测的协作动态，从而促进了在部分可观测性特征的现实环境中的隐式协调。
cs.RO / 112 / 2606.22998
TEXEDO : Test Time Scaling for Controller-aware Language-conditioned Humanoid Motion Generation
TEXEDO：面向控制器感知的语言条件类人运动生成的测试时间缩放
Abstract
Text-conditioned motion generation is a promising interface for programming humanoid robots, yet current generators are often trained on human motion datasets retargeted to robot morphologies. Although such data provides rich semantic and kinematic priors, it fails to capture the nuances of whole-body tracking controllers, including balance, contact dynamics, actuation limits, and controller-specific failure modes. As a result, generated motions can be semantically plausible but difficult or impossible for the robot to execute. We introduce TEXEDO, a test-time scaling framework for humanoid motion generation that improves motion quality without requiring a stronger underlying generator. Given a text prompt, TEXEDO samples multiple candidate motions from a pretrained text-conditioned generator and selects the best motion that is both executable and task-aligned. The reward model combines a dynamic feasibility verifier, distilled from whole-body tracking rollouts to predict physical executability, with a semantic alignment verifier that measures text-motion alignment in a learned co-embedding space. Our pipeline treats dynamic feasibility as a hard constraint and semantic alignment as the selection objective within the feasible set. Through large-scale simulation studies and real-world deployment on a Unitree G1 humanoid robot, we show that TEXEDO consistently improves both tracking fidelity and text alignment. These results demonstrate that grounded verification is an effective path toward deployable language-guided humanoid motion generation. Project website: https://jianuocao.github.io/TEXEDO/
Chinese Translation
基于文本的运动生成是编程类人机器人一种有前景的接口，但目前的生成器通常是在针对机器人形态进行重定向的人类运动数据集上训练的。尽管这些数据提供了丰富的语义和运动学先验，但未能捕捉到全身跟踪控制器的细微差别，包括平衡、接触动态、驱动限制和特定于控制器的故障模式。因此，生成的运动可能在语义上是合理的，但对于机器人来说执行起来可能困难或不可能。我们提出了TEXEDO，一种用于类人运动生成的测试时间缩放框架，它在不需要更强大的基础生成器的情况下提高运动质量。给定一个文本提示，TEXEDO从一个预训练的基于文本的生成器中采样多个候选运动，并选择最佳的既可执行又符合任务要求的运动。奖励模型结合了一个动态可行性验证器，该验证器从全身跟踪回放中提炼，以预测物理可执行性，以及一个语义对齐验证器，该验证器在学习的共同嵌入空间中测量文本与运动的对齐。我们的流程将动态可行性视为硬约束，而将语义对齐视为在可行集合内的选择目标。通过大规模的模拟研究和在Unitree G1类人机器人上的实际部署，我们展示了TEXEDO在跟踪保真度和文本对齐方面的一致改善。这些结果表明，基于验证的生成是实现可部署的语言引导类人运动生成的有效途径。项目网站：https://jianuocao.github.io/TEXEDO/
cs.RO / 113 / 2606.23006
ISOPoT: Imaging Sonar Odometry by Point Tracking
ISOPoT：基于点跟踪的成像声纳里程计
Abstract
Reliable navigation in underwater environments remains a key challenge in marine robotics. In such scenarios, forward-looking sonars are a natural choice for long-range perception, offering wide coverage even in turbid, low-visibility conditions. However, sonar images are inherently noisy, contain artifacts, and lack rich semantic structure, causing standard computer vision methods for keypoint detection and matching to perform poorly. In this paper, we introduce ISOPoT, an imaging sonar odometry method based on modern point tracking techniques. We propose a sonar odometry pipeline that uses multi-frame point tracks as its primary correspondence representation, augmented with lightweight optimizations to improve robustness. We evaluated the proposed method on the Aracati 2017 dataset, as well as on an internal sonar dataset collected in real-world underwater environments. Our results show that ISOPoT outperforms previous state-of-the-art methods consistently in both sonar-only scenarios and in multi-sensor settings.
Chinese Translation
在水下环境中可靠导航仍然是海洋机器人技术面临的关键挑战。在这种情况下，前视声纳是进行远程感知的自然选择，即使在浑浊、低能见度的条件下也能提供广泛的覆盖。然而，声纳图像本质上是嘈杂的，包含伪影，并且缺乏丰富的语义结构，导致标准计算机视觉方法在关键点检测和匹配方面表现不佳。在本文中，我们介绍了ISOPoT，一种基于现代点跟踪技术的成像声纳里程计方法。我们提出了一种声纳里程计管道，使用多帧点轨迹作为主要对应表示，并通过轻量级优化来增强鲁棒性。我们在Aracati 2017数据集以及在真实水下环境中收集的内部声纳数据集上评估了所提出的方法。我们的结果表明，ISOPoT在声纳单独场景和多传感器设置中始终优于之前的最先进方法。
cs.RO / 114 / 2606.23079
AdaReP:Adaptive Re-Planning under Model Mismatch for Neural World-Model Predictive Control
AdaReP：在模型不匹配下的自适应重新规划用于神经世界模型预测控制
Abstract
Neural world models coupled with model predictive control (MPC) replan at every environment step to bound accumulated prediction error, but this incurs substantial computational overhead. Reusing a cached plan reduces this overhead, yet its effectiveness depends on how prediction mismatch propagates through the local dynamics. We analyze this trade-off with a perturbation-based dynamic-regret framework and show that stale-plan penalties scale with the reuse tolerance, the accumulated mismatch since the last replanning step, and the local dynamics sensitivity. Based on this structure, we propose AdaReP, a training-free wrapper that adapts the replanning tolerance online using the current deviation from the cached rollout and a local sensitivity estimate, without modifying the learned world model or planner. Across image-space planning, latent-space control, and real-world robotic manipulation, AdaReP substantially reduces planner-side computation while maintaining comparable task performance, including over 80% fewer queries on a 50-trial physical robot study.
Chinese Translation
神经世界模型结合模型预测控制（MPC）在每个环境步骤重新规划，以限制累积的预测误差，但这会带来相当大的计算开销。重用缓存计划可以减少这种开销，但其有效性依赖于预测不匹配如何在局部动态中传播。我们通过基于扰动的动态后悔框架分析了这种权衡，并表明过时计划的惩罚与重用容忍度、上次重新规划步骤以来的累积不匹配以及局部动态的敏感性成比例。基于这一结构，我们提出了AdaReP，一个无训练的封装器，能够在线适应重新规划容忍度，利用当前与缓存展开的偏差和局部敏感性估计，而无需修改学习到的世界模型或规划器。在图像空间规划、潜在空间控制和真实世界机器人操作中，AdaReP显著减少了规划器端的计算，同时保持了可比的任务性能，包括在50次试验的物理机器人研究中减少超过80%的查询。
cs.RO / 115 / 2606.23085
Foresight: Failure Detection for Long-Horizon Robotic Manipulation with Action-Conditioned World Model Latents
Foresight：基于动作条件世界模型潜变量的长时间跨度机器人操作失败检测
Abstract
Long-horizon tasks are common in real-world robotic deployments, yet failure detection for such tasks remains underexplored. Detecting failures in long-horizon robotic tasks is particularly challenging because failure onset is often ambiguous and dense temporal annotations are typically unavailable. We present Foresight, a failure detection framework that monitors manipulation trajectories using latent representations from an action-conditioned world model. Foresight is trained using only final task-level success or failure labels. By leveraging predictive world-model embeddings, our method provides a unified framework for failure detection across different policies. We further use functional conformal prediction (FCP) to calibrate detection thresholds adaptively. We evaluate Foresight with state-of-the-art vision-language-action policies in simulation on LIBERO-Long, ManiSkill-Long, and BEHAVIOR-1K, compare it against state-of-the-artfailure detection methods, and validate it on real robots with three long-horizon tasks on a ReactorX-200 arm and one task on a Franka arm. Our results suggest that action-conditioned world-model embeddings provide a scalable representation for reliable failure monitoring in long-horizon manipulation.
Chinese Translation
长时间跨度的任务在现实世界的机器人应用中很常见，但对这类任务的失败检测仍然未得到充分探索。检测长时间跨度机器人任务中的失败尤其具有挑战性，因为失败的发生往往模糊不清，并且通常缺乏密集的时间注释。我们提出了Foresight，一个利用动作条件世界模型的潜在表示监控操作轨迹的失败检测框架。Foresight仅使用最终的任务级成功或失败标签进行训练。通过利用预测性世界模型嵌入，我们的方法为不同策略的失败检测提供了统一的框架。我们进一步使用功能性符合预测（Functional Conformal Prediction, FCP）来自适应地校准检测阈值。我们在LIBERO-Long、ManiSkill-Long和BEHAVIOR-1K的仿真环境中，使用最先进的视觉-语言-动作策略评估Foresight，并将其与最先进的失败检测方法进行比较，同时在ReactorX-200臂上进行三个长时间跨度任务和在Franka臂上进行一个任务的真实机器人上进行验证。我们的结果表明，基于动作条件的世界模型嵌入为长时间跨度操作中的可靠失败监控提供了可扩展的表示。
cs.RO / 116 / 2606.23090
Flow as Flow: Modeling Robot Velocity Fields as Probability Velocity Fields for Flow-Based Object Manipulation
流即流：将机器人速度场建模为基于流的物体操作的概率速度场
Abstract
Cross-embodiment data have become central to training robotic foundation models. To leverage such heterogeneous data, we focus on flow-based object manipulation, where robot flows (robot velocity fields) serve as embodiment-agnostic motion representations. Previous studies do not formulate robot flows as dense velocity fields, but as displacements of sparse keypoints, while such velocity fields better match the continuous-time nature of motions. We propose Flow as Flow, a framework that models robot flows as probability flows based on a flow matching formulation. By naturally modeling such velocity fields within this formulation, our method achieves efficient and high-quality robot flow generation. Across standard benchmarks, our method outperforms representative baseline methods on standard metrics, while achieving approximately 33$\times$ faster generation. Furthermore, through real-world experiments evaluating 9 methods with 260 trials per method across 13 manipulation tasks, we show that our method achieves a higher average success rate than the baseline methods. Our project page is available at https://flow-as-flow-u0n5y.kinsta.page.
Chinese Translation
跨体现数据已成为训练机器人基础模型的核心。为了利用这种异构数据，我们专注于基于流的物体操作，其中机器人流（机器人速度场）作为与体现无关的运动表示。以往研究并未将机器人流表述为密集速度场，而是作为稀疏关键点的位移，而这种速度场更好地匹配运动的连续时间特性。我们提出了“流即流”（Flow as Flow）框架，该框架基于流匹配公式将机器人流建模为概率流。通过在该公式内自然建模这些速度场，我们的方法实现了高效且高质量的机器人流生成。在标准基准测试中，我们的方法在标准指标上超越了代表性基线方法，同时实现了约33倍的生成速度提升。此外，通过对9种方法进行的现实世界实验，每种方法进行了260次试验，涵盖13个操作任务，我们展示了我们的方法在平均成功率上高于基线方法。我们的项目页面可访问 https://flow-as-flow-u0n5y.kinsta.page。
cs.RO / 117 / 2606.23147
Assistron: Bayesian Shared Autonomy with Off-the-shelf Vision-Language-Action Models
Assistron：基于贝叶斯的共享自主性与现成视觉-语言-动作模型
Abstract
We propose Assistron, a shared autonomy model that leverages Vision-Language-Action (VLA) models to assist the user in daily activities. Our approach is grounded in two core principles: (1)~minimizing human cognitive and physical effort by leveraging VLA-driven autonomy for macro-movements, and (2)~prioritizing human intervention specifically at critical failure points. Driven by the user's verbal language commands, Assistron utilizes the VLA to autonomously execute macro-reaching trajectories, saving users' effort. In contact-rich interactions where VLAs tend to fail, Assistron employs a phase-aware interaction detection mechanism and solicits the user to intervene, in turn adjusting the VLA's action generation via flow matching guidance. Critically, our formulation eliminates the need for VLA fine-tuning, protecting its broad behavioral priors from catastrophic forgetting and ensuring the model does not become a narrow specialist. We validate our approach on a comprehensive multi-task scene recovery benchmark encompassing diverse daily manipulation skills. Empirical results demonstrate that Assistron significantly improves task success rates over pure autonomous baselines while significantly reducing human cognitive and physical workload compared to traditional teleoperation, offering a scalable, smooth, and effortless paradigm for assistive manipulation. The code is available in https://github.com/mousecpn/Assistron.git.
Chinese Translation
我们提出了Assistron，这是一种共享自主模型，利用视觉-语言-动作（VLA）模型来协助用户进行日常活动。我们的方法基于两个核心原则：（1）通过利用VLA驱动的自主性来最小化人类的认知和身体努力，特别是在宏观运动中；（2）在关键故障点优先考虑人类干预。Assistron根据用户的语言指令，利用VLA自主执行宏观到达轨迹，从而节省用户的精力。在接触丰富的交互中，VLA往往会失败，Assistron采用了一种阶段感知的交互检测机制，并请求用户进行干预，从而通过流匹配指导调整VLA的动作生成。重要的是，我们的公式消除了对VLA微调的需求，保护了其广泛的行为先验，避免了灾难性遗忘，确保模型不会成为狭窄的专家。我们在一个全面的多任务场景恢复基准上验证了我们的方法，该基准涵盖了多种日常操作技能。实证结果表明，Assistron显著提高了任务成功率，相较于纯自主基线，同时在与传统遥操作相比时显著降低了人类的认知和身体负担，为辅助操作提供了一种可扩展、顺畅且轻松的范式。代码可在 https://github.com/mousecpn/Assistron.git 获取。
cs.RO / 118 / 2606.23152
ShotcreteDepth: A Bi-modal Dataset for Robust Robotic Depth Perception in Shotcrete Construction Environments
ShotcreteDepth：用于喷射混凝土施工环境中稳健机器人深度感知的双模态数据集
Abstract
We introduce ShotcreteDepth, a bi-modal dataset from the construction domain that captures both an active shotcreting process and general construction environments. The dataset comprises stereo RGB imagery and LiDAR point clouds acquired under harsh real-world conditions, including high turbidity and poor illumination. Such conditions adversely affect sensor measurements, leading to incomplete and noisy observations that pose significant challenges for perception systems in autonomous applications. Alongside the dataset, we release a lightweight annotation tool designed for time-efficient labeling of LiDAR point clouds. ShotcreteDepth consists of 11,252 temporally synchronized data samples, of which 220 are annotated for evaluation purposes. The dataset supports research in stereo matching, depth completion, and depth estimation under conditions that closely reflect the operational complexities found in industrial settings. Project repository: https://github.com/dtu-pas/shotcrete-depth
Chinese Translation
我们介绍了ShotcreteDepth，这是一个来自施工领域的双模态数据集，捕捉了活跃的喷射混凝土过程和一般施工环境。该数据集包括在恶劣的现实条件下获取的立体RGB图像和LiDAR点云，这些条件包括高浑浊度和较差的照明。这些条件对传感器测量产生不利影响，导致不完整和噪声较大的观测结果，从而对自主应用中的感知系统构成重大挑战。除了数据集，我们还发布了一个轻量级标注工具，旨在高效地对LiDAR点云进行标注。ShotcreteDepth包含11,252个时间同步的数据样本，其中220个样本已标注用于评估目的。该数据集支持在与工业环境中的操作复杂性密切相关的条件下进行立体匹配、深度补全和深度估计的研究。项目仓库： https://github.com/dtu-pas/shotcrete-depth
cs.RO / 119 / 2606.23153
Asymmetric physics enables efficient learning in quadrupedal robot swarms
非对称物理促进四足机器人群体的高效学习
Abstract
Animal collectives navigate cluttered environments through local coordination, yet robot swarms still struggle to reproduce this capability in the physical world. End-to-end learning offers a route to such coordination, but scaling it to embodied swarms remains difficult: standard sampling-based reinforcement learning becomes inefficient when visual perception, dense robot-robot interaction, and contact-rich locomotion must be learned together. Here we show that asymmetric physics enables efficient end-to-end learning of vision-based, decentralized control in large swarms of quadrupedal robots. During training, quadrupeds interact in shared environments, where a high-fidelity, non-differentiable simulator generates realistic motion and contact dynamics, and differentiable surrogate models provide gradients for navigation and locomotion policies. This separation enables up to 512 quadrupeds to learn coordinated navigation policies in obstacle-rich environments. At deployment, each robot acts from a single forward-facing depth camera, without explicit communication, centralized planning, or global maps. The policies generalize across forests, bridges, enclosures, narrow passages, and mazes, and zero-shot transfer to six physical quadrupeds across five real-world scenarios. The resulting swarms exhibit predictive avoidance, right-side yielding, pausing before bottlenecks, and wall following, showing that asymmetric physics enables efficient training of scalable decentralized control policies for quadrupedal robot swarms.
Chinese Translation
动物群体通过局部协调在复杂环境中导航，但机器人群体在物理世界中仍然难以重现这一能力。端到端学习提供了实现这种协调的途径，但将其扩展到具身群体仍然困难：当视觉感知、密集的机器人间交互和丰富接触的运动必须一起学习时，基于标准采样的强化学习变得低效。在此，我们展示了非对称物理使得在大规模四足机器人群体中高效实现基于视觉的去中心化控制的端到端学习成为可能。在训练过程中，四足机器人在共享环境中进行交互，其中高保真、不可微分的模拟器生成逼真的运动和接触动态，而可微分的替代模型为导航和运动策略提供梯度。这种分离使得多达512只四足机器人能够在障碍物丰富的环境中学习协调导航策略。在部署时，每个机器人仅依靠单个朝前的深度相机进行操作，无需显式通信、集中规划或全局地图。这些策略在森林、桥梁、围栏、狭窄通道和迷宫中具有良好的泛化能力，并且在五个真实场景中实现了对六只物理四足机器人的零样本迁移。最终形成的群体展现了预测性避让、右侧让行、在瓶颈处暂停和沿墙行驶的能力，表明非对称物理能够高效训练可扩展的去中心化控制策略，以适应四足机器人群体。
cs.RO / 120 / 2606.23157
Bridging Semantics and Kinematics: A Modular Framework for Zero-Shot Robotic Manipulation
弥合语义与运动学的鸿沟：一种用于零-shot机器人操作的模块化框架
Abstract
This paper presents a modular training-free framework for zero-shot, language-guided robotic manipulation in semi-structured environments. The architecture bridges the gap between high-level reasoning and low-level kinematics by decomposing the vision-action pipeline into three stages: visual perception, semantic interpretation, and task execution. To overcome the spatial ambiguity and semantic hallucinations inherent in standard Vision-Language Models (VLMs), the perception module employs FastSAM and Set-of-Mark (SoM) prompting to dynamically generate grounded, alphanumeric visual anchors. The same foundation model then operates purely as a Large Language Model (LLM) to act as a semantic router, translating unconstrained human directives into verifiable, reconfigurable configurations. Finally, these configurations are dynamically parsed by a Task Orchestrator into MoveIt Task Constructor (MTC) to generate collision-free trajectories. The framework is evaluated across two zero-shot experimental setups: unconstrained open-world sequential manipulation and dense relational spatial reasoning, achieving a 62% end-to-end task success rate across both scenarios, demonstrating its capacity to reliably execute complex physical actions without domain-specific training or manual coordinate programming.
Chinese Translation
本文提出了一种模块化的无训练框架，用于在半结构化环境中进行零-shot、语言引导的机器人操作。该架构通过将视觉-动作管道分解为三个阶段：视觉感知、语义解释和任务执行，弥合了高层推理与低层运动学之间的差距。为了克服标准视觉-语言模型（Vision-Language Models, VLMs）固有的空间歧义和语义幻觉，感知模块采用了FastSAM和Set-of-Mark (SoM) 提示，动态生成有根基的字母数字视觉锚点。相同的基础模型随后纯粹作为大型语言模型（Large Language Model, LLM）运作，充当语义路由器，将不受限制的人类指令翻译为可验证的、可重构的配置。最后，这些配置由任务协调器动态解析为MoveIt任务构造器（MoveIt Task Constructor, MTC），以生成无碰撞的轨迹。该框架在两个零-shot实验设置中进行了评估：不受限制的开放世界顺序操作和密集关系空间推理，在这两种场景中实现了62%的端到端任务成功率，展示了其在没有领域特定训练或手动坐标编程的情况下可靠执行复杂物理动作的能力。
cs.RO / 121 / 2606.23246
Lessons from the Field: A Case Study of Robotic Intervention in an Industrial Emergency
来自现场的经验教训：工业紧急情况下的机器人干预案例研究
Abstract
Incidents in chemical plants can pose a high level of risk and harsh environments for first responders. Contamination and explosion hazards can deny human access to the affected infrastructure, underscoring the need for capable robot systems. This field report documents the successful deployment of a robotic task force to neutralize an explosive gas hazard at a chemical plant after a fire incident. An Unmanned Ground Vehicle (UGV) with a custom manipulation tool opened a critical valve under hazardous conditions, averting the threat of a large-scale explosion. We provide insights into robot deployment and use the mission results to highlight both the importance of rescue robotics and limitations of using research platforms in real emergency deployments, such as communication constraints and the need for enhanced operator-assistance functions.
Chinese Translation
化工厂的事故可能对第一响应者构成高风险和恶劣环境。污染和爆炸危险可能使人类无法接触受影响的基础设施，这突显了能够应对这一挑战的机器人系统的必要性。本现场报告记录了一支机器人特遣队在火灾事故后成功部署，以中和化工厂的爆炸气体危险。一辆配备定制操作工具的无人地面车辆（UGV）在危险条件下打开了一个关键阀门，避免了大规模爆炸的威胁。我们提供了关于机器人部署的见解，并利用任务结果强调了救援机器人技术的重要性以及在实际紧急部署中使用研究平台的局限性，例如通信限制和对增强操作员辅助功能的需求。
cs.RO / 122 / 2606.23249
LP-NavOA: Integrated Local Navigation and Obstacle Avoidance for Humanoid Robots under Limited Perception
LP-NavOA：在有限感知下的人形机器人综合本地导航与障碍物避让
Abstract
Humanoid local navigation in cluttered environments must jointly resolve obstacle avoidance, sparse-goal recovery, and stable whole-body locomotion under short-range and partially observable sensing. Explicit planner-control decompositions introduce latency and can mismatch agile humanoid command-tracking limits, while purely reactive controllers may lose the goal after obstacle occlusion. We present LP-NavOA, a limited-perception navigation and obstacle-avoidance framework for humanoid robots. A raycast-conditioned perception-action proximal policy optimization (PPO) locomotion backbone is first trained with a robot-centered circular heading-speed command and a shared command-side safety filter. With this backbone frozen, A-star and waypoint teachers generate rollouts for distilling a recurrent local planner that overwrites only the heading command at deployment, leaving the whole-body policy intact. At runtime, LP-NavOA uses proprioception, short-range local range sensing, and a body-frame goal direction, requiring no global map, waypoint stream, or external planner. In MuJoCo open-wall and indoor layouts, the distilled planner produces obstacle bypassing and post-avoidance goal recovery, raising teacher-calibrated on-time arrival from 38--40\% to 85--97\% and reducing brush/contact-heavy progress relative to a backbone-only controller. Ablations show that dynamic route shaping, teacher-active data collection, and the circular command interface are important for navigation efficiency and for training the 3.0\,m/s backbone. A Unitree G1 deployment analysis demonstrates hardware executability without continuous joystick steering.
Chinese Translation
在人形机器人在复杂环境中的本地导航中，必须共同解决障碍物避让、稀疏目标恢复和在短距离及部分可观测感知下的稳定全身运动。显式的规划-控制分解引入了延迟，并可能与灵活的人形机器人指令跟踪限制不匹配，而纯粹的反应式控制器在障碍物遮挡后可能会失去目标。我们提出了LP-NavOA，一个针对人形机器人的有限感知导航与障碍物避让框架。首先，使用机器人中心的圆形航向-速度指令和共享的指令侧安全过滤器训练一个基于射线投射条件的感知-动作近端策略优化（PPO）运动骨干。冻结该骨干后，A*算法和航点教师生成回放，以提炼一个递归本地规划器，该规划器在部署时仅覆盖航向指令，保持全身策略不变。在运行时，LP-NavOA利用本体感知、短距离本地范围感知和身体框架目标方向，不需要全局地图、航点流或外部规划器。在MuJoCo开放墙体和室内布局中，提炼的规划器实现了障碍物绕行和避让后的目标恢复，使教师校准的准时到达率从38-40%提高到85-97%，并减少了相对于仅使用骨干控制器的接触/摩擦重进展。消融实验表明，动态路线塑造、教师主动数据收集和圆形指令接口对导航效率和训练3.0 m/s骨干至关重要。Unitree G1的部署分析展示了在没有持续操纵杆操控的情况下的硬件可执行性。
cs.RO / 123 / 2606.23258
Conceptual Design of an Ecosystem for Real Farm Data Collection toward Agricultural AI Foundation Models
面向农业人工智能基础模型的真实农场数据收集生态系统的概念设计
Abstract
Data scarcity is a fundamental challenge in developing AI and foundation models for agricultural robots. Existing open-source data platforms do not provide sufficient incentives for data providers so long-term data collection remains difficult. Furthermore, advances in generative AI have introduced a new challenge of verifying that collected data genuinely originates from real farm environments. We propose an ecosystem for the sustainable collection and distribution of real farm data, integrating automatic pricing driven by demand and rarity, revenue sharing that distributes earnings to farmers as an incentive to keep providing data, and data authenticity guarantees through authenticated device uploads. To demonstrate the economic sustainability for all three parties among farmers, AI companies, and the platform, we estimate the economic value that agricultural robots stand to generate.
Chinese Translation
数据稀缺是开发农业机器人所需的人工智能和基础模型的一个根本挑战。现有的开源数据平台未能为数据提供者提供足够的激励，因此长期数据收集仍然困难。此外，生成性人工智能的进展带来了一个新的挑战，即验证收集的数据确实源自真实的农场环境。我们提出了一个可持续的真实农场数据收集和分发生态系统，整合了基于需求和稀缺性的自动定价、将收益分配给农民以激励其持续提供数据的收入分享机制，以及通过认证设备上传确保数据真实性的保障。为了展示农民、人工智能公司和平台三方之间的经济可持续性，我们估算了农业机器人可能产生的经济价值。
cs.RO / 124 / 2606.23280
Causal Reward World Models: Zero-shot Reward Design for Automated Skill Generation
因果奖励世界模型：用于自动化技能生成的零-shot 奖励设计
Abstract
Automated Reward Design (ARD) aims to replace manual reward engineering in reinforcement learning with language-driven reward function synthesis. However, existing approaches based on large language models (LLMs) remain inherently correlation-driven, relying on iterative environmental feedback to refine reward hypotheses for each specific task. This paradigm not only results in inefficient reasoning but also makes LLMs susceptible to semantically plausible yet causally spurious reward components, leading to ineffective optimization. To address these limitations, we propose the Causal Reward World Model (CRWM), which explicitly models the causal topological relationships between candidate reward components and task-targeted physical variables through offline pre-training on multi-task interaction data. Based on a coarse-to-fine pre-training strategy, we introduce a joint optimization module that integrates Explicit Mechanism Decoupling with Confidence-Aware Soft Fusion to refine coarse structural priors using micro-level trajectories, thereby constructing a robust and interpretable causal skeleton. During inference, LLMs leverage CRWM as a task-irrelevant causal prior to constrain the reward generation, enabling zero-shot reward function design. Our work opens up a new white-box paradigm for the ARD problem. Extensive experiments on complex continuous control benchmarks demonstrate that CRWM generates executable reward functions without feedback-driven reward refinement, significantly reducing the design latency for acquiring new robotic skills while matching or surpassing state-of-the-art performance, and further exhibits strong generalization capabilities across unseen tasks and diverse robotic embodiments.
Chinese Translation
自动化奖励设计（Automated Reward Design, ARD）旨在通过基于语言的奖励函数合成替代强化学习中的手动奖励工程。然而，现有基于大型语言模型（Large Language Models, LLMs）的方法本质上仍然是基于相关性的，依赖于迭代的环境反馈来细化每个特定任务的奖励假设。这一范式不仅导致推理效率低下，还使得 LLMs 易受语义上合理但因果上虚假的奖励成分的影响，从而导致优化效果不佳。为了解决这些局限性，我们提出了因果奖励世界模型（Causal Reward World Model, CRWM），该模型通过对多任务交互数据进行离线预训练，明确建模候选奖励成分与任务目标物理变量之间的因果拓扑关系。基于粗到细的预训练策略，我们引入了一个联合优化模块，该模块结合了显式机制解耦（Explicit Mechanism Decoupling）与置信感知软融合（Confidence-Aware Soft Fusion），利用微观级别的轨迹来细化粗略的结构先验，从而构建一个稳健且可解释的因果骨架。在推理过程中，LLMs 利用 CRWM 作为与任务无关的因果先验来约束奖励生成，实现零-shot 奖励函数设计。我们的工作为 ARD 问题开辟了一个新的白盒范式。在复杂的连续控制基准上进行的大量实验表明，CRWM 能够生成可执行的奖励函数，而无需基于反馈的奖励细化，显著减少了获取新机器人技能的设计延迟，同时匹配或超越了最先进的性能，并进一步展现出在未见任务和多样化机器人形态上的强泛化能力。
cs.RO / 125 / 2606.23296
IOI: Decoupling Kinematics and Physics for Interactive World Models
IOI：为互动世界模型解耦运动学与物理学
Abstract
Developing generalist embodied agents requires interactive environments providing visually realistic feedback and accurate action-conditioned dynamics. Interactive world models address this by simulating such complex dynamics. However, purely data-driven methods struggle to ensure precise control alignment and physically plausible visual feedback due to a lack of explicit structural constraints. To address this, we propose IOI, a hybrid interactive world model integrating analytical kinematic priors with learned physical dynamics. Unlike data-driven approaches prone to spatiotemporal drift, IOI introduces explicit kinematic guidance, computing forward kinematics from action sequences for accurate motion trajectories. These trajectories are rendered into synchronized front, side, and top orthographic projections, eliminating the need for extrinsic camera calibration. A Multi-view Kinematic Aggregation and Injection module fuses these geometric cues and injects them into the video generator, providing geometry-consistent guidance. Conditioning video generation on these deterministic trajectories establishes a synergy between the analytical simulator and the world model. Decoupling deterministic motion into the kinematic prior frees the generator to model stochastic physical interactions. Experiments on the RoboTwin benchmark validate IOI across kinematic fidelity, out-of-distribution (OOD) generalization, and policy evaluation. IOI achieves state-of-the-art simulation performance and robust zero-shot generalization to unseen OOD tasks. Furthermore, IOI serves as a reliable policy evaluator, yielding success rates closely aligning with ground-truth physics simulators. On real-world platforms, policies trained on IOI-synthesized data match those trained on teleoperation demonstrations, solidifying its practical value for embodied policy learning.
Chinese Translation
开发通用的具身智能体需要提供视觉上逼真的反馈和准确的动作条件动态的互动环境。互动世界模型通过模拟这些复杂动态来解决这一问题。然而，纯数据驱动的方法由于缺乏明确的结构约束，难以确保精确的控制对齐和物理上合理的视觉反馈。为了解决这个问题，我们提出了IOI，一种将分析运动学先验与学习的物理动态相结合的混合互动世界模型。与容易受到时空漂移影响的数据驱动方法不同，IOI引入了明确的运动学指导，从动作序列计算前向运动学以获得准确的运动轨迹。这些轨迹被渲染为同步的正面、侧面和顶部正交投影，消除了对外部相机标定的需求。一个多视角运动学聚合与注入模块融合了这些几何线索，并将其注入视频生成器，提供几何一致的指导。基于这些确定性轨迹进行视频生成建立了分析模拟器与世界模型之间的协同作用。将确定性运动解耦为运动学先验使生成器能够建模随机物理交互。在RoboTwin基准上的实验验证了IOI在运动学保真度、分布外（OOD）泛化和策略评估方面的表现。IOI实现了最先进的仿真性能，并在未见的OOD任务上展现出强大的零样本泛化能力。此外，IOI作为一个可靠的策略评估器，其成功率与真实物理模拟器的结果紧密对齐。在真实世界平台上，基于IOI合成数据训练的策略与基于遥操作演示训练的策略相匹配，巩固了其在具身策略学习中的实际价值。
cs.RO / 126 / 2606.23312
From Pixels to Concepts: Growing Rich 3D Semantic Scene Graph Forests utilizing Foundation Models
从像素到概念：利用基础模型构建丰富的三维语义场景图森林
Abstract
Operating in complex real-world environments requires robots to understand their surroundings on a functional semantic level. This demands a detailed multi-layer world model capturing the complex relations of its surroundings. Hierarchical 3D scene graphs address this challenge by integrating geometric, semantic, and relational data within a unified spatial framework. However, current 3D scene graph approaches often restrict themselves to rigid structures of pre-determined relationship classes, mostly neglecting important semantic connections, like causal connections or environmental contexts. This paper explores the potential of foundation models to build forests of 3D scene graphs with open semantic relationships to improve scene understanding and robotic task execution. We propose a method where instance-specific concept-nodes and relationships are first identified by a VLM and extended upon by a LLM, inferring broader, more abstract concept-nodes and relationships through reasoning. These object-nodes, concept-nodes, and relationships are then assembled into a forest of hierarchical 3D scene graphs, enhanced with concept-nodes to represent abstract concepts. Evaluations were conducted on the uHumans2 and ScanNet indoor dataset, validating the accuracy and relevance of the generated relationships. Downstream suitability of scene-graph forests for robotics applications is demonstrated in an open-vocabulary object-retrieval task utilizing both ScanNet data and a real-world indoor deployment using a Boston Dynamics Spot. This paper leverages foundation models to create more expressive, semantically deep 3D hierarchical scene graphs and demonstrates their potential to advance semantic and environmental understanding in robotics.
Chinese Translation
在复杂的现实环境中操作要求机器人在功能语义层面理解其周围环境。这需要一个详细的多层次世界模型，以捕捉其周围环境的复杂关系。层次化的三维场景图通过在统一的空间框架内整合几何、语义和关系数据来应对这一挑战。然而，当前的三维场景图方法往往限制于预先确定的关系类别的刚性结构，通常忽视重要的语义连接，如因果关系或环境上下文。本文探讨了基础模型在构建具有开放语义关系的三维场景图森林方面的潜力，以改善场景理解和机器人任务执行。我们提出了一种方法，其中实例特定的概念节点和关系首先由视觉语言模型（VLM）识别，并通过大型语言模型（LLM）进行扩展，通过推理推断出更广泛、更抽象的概念节点和关系。这些对象节点、概念节点和关系随后被组装成一个层次化的三维场景图森林，并通过概念节点增强以表示抽象概念。在uHumans2和ScanNet室内数据集上进行了评估，验证了生成关系的准确性和相关性。通过在开放词汇对象检索任务中展示场景图森林在机器人应用中的下游适用性，利用ScanNet数据和使用波士顿动力公司Spot的现实室内部署。本文利用基础模型创建了更具表现力、语义深度的三维层次场景图，并展示了它们在推进机器人语义和环境理解方面的潜力。
cs.RO / 127 / 2606.23339
When Robots Rate Their Own Interactions: Engagement Validity and the Strangeness Failure
当机器人评估自己的互动：参与有效性与陌生性失效
Abstract
Human-robot interaction (HRI) evaluation relies almost exclusively on human-completed questionnaires, leaving the robot's perspective unexamined. We propose an \textit{inverted evaluation}, in which LLM-powered robots complete the same standardized instruments from their own perspective, and test whether these ratings agree with human ground truth. In Study~1, five LLMs completed HRI-CUES, Godspeed, and RoSAS questionnaires for 25~interactions ($N = 1{,}522$ evaluations) from the HRI-CUES dataset. LLMs achieved moderate-to-strong agreement on engagement dimensions (satisfaction $r$ up to $.65$ and enjoyment $r$ up to $.72$) with excellent test-retest reliability (ICC $\geq .82$), but \textit{systematically inverted} the comfort/strangeness dimension ($r = -.44$ to $-.67$, all $p < .05$), conflating engagement with comfort. In Study~2, a Nao robot running Claude~Sonnet~4.5 replicated these patterns in live interactions ($N = 4$), including real-time turn-by-turn assessment. The strangeness failure persisted across five models, synthetic controls, and embodied deployment for two participants. We argue that current LLM-based robots lack access to the internal affective states needed to assess constructs like strangeness, and that inverted evaluation requires supplementary modalities (e.g., physiological signals, gaze, proxemics) to move beyond behavioral proxies. These findings establish boundary conditions for using LLMs as interaction evaluators in HRI.
Chinese Translation
人机交互（HRI）评估几乎完全依赖于人类填写的问卷，未能考察机器人的视角。我们提出了一种 extit{反向评估}，即由大型语言模型（LLM）驱动的机器人从自身视角完成相同的标准化工具，并测试这些评分是否与人类的真实情况一致。在研究1中，五个LLM为来自HRI-CUES数据集的25次互动（$N = 1{,}522$评估）完成了HRI-CUES、Godspeed和RoSAS问卷。LLM在参与维度上达成了中等到强的一致性（满意度$r$高达$.65$，享受度$r$高达$.72$），并且具有优秀的重测信度（ICC $
geq .82$），但 extit{系统性地反转}了舒适/陌生维度（$r = -.44$到$-.67$，所有$p < .05$），将参与与舒适混淆。在研究2中，一台运行Claude~Sonnet~4.5的Nao机器人在实时互动中复制了这些模式（$N = 4$），包括逐轮的实时评估。陌生性失效在五个模型、合成控制和两个参与者的具身部署中持续存在。我们认为，目前基于LLM的机器人缺乏评估陌生性等构念所需的内部情感状态，反向评估需要补充的模式（例如，生理信号、注视、亲密距离）以超越行为代理。这些发现为在HRI中使用LLM作为互动评估者建立了边界条件。
cs.RO / 128 / 2606.23355
A Relaxed Quadratic-Program-based Framework for Trajectory Tracking of Unicycle Robots with Singularity Avoidance
基于放松二次规划的单轮机器人轨迹跟踪框架及奇异性避免
Abstract
Dynamic feedback linearization (DFL) is a classical technique for trajectory tracking of unicycle-type mobile robots, but the resulting DFL-based controller becomes singular when the linear velocity vanishes, rendering standard DFL-based controllers unsuitable for stop-and-reverse maneuvers. This paper proposes a quadratic-program (QP)-based optimal control framework that avoids this singularity, while establishing local Lipschitz continuity of the resulting feedback law. Our approach reformulates the DFL constraints as an equality-constrained QP with a slack variable, ensuring feasibility for all states and reference signals, including at points where the robot's velocity vanishes. By introducing slack variables and tunable parameters, we demonstrate that the singular configuration can be avoided for a large class of reference trajectories. The effectiveness of the proposed approach for trajectory tracking is demonstrated through ROS 2-Gazebo simulations on a TurtleBot3 Waffle robot. The code is available at https://gradslab.github.io/DFL_QP_Unicycle/
Chinese Translation
动态反馈线性化（DFL）是一种经典的单轮移动机器人轨迹跟踪技术，但当线性速度消失时，基于DFL的控制器会变得奇异，从而使得标准的基于DFL的控制器不适用于停止和反向操作。本文提出了一种基于二次规划（QP）的最优控制框架，避免了这一奇异性，同时建立了所得到的反馈法则的局部利普希茨连续性。我们的方法将DFL约束重新表述为带有松弛变量的等式约束QP，确保在所有状态和参考信号下的可行性，包括在机器人速度消失的点。通过引入松弛变量和可调参数，我们证明可以避免奇异配置，适用于大类参考轨迹。通过在TurtleBot3 Waffle机器人上的ROS 2-Gazebo仿真，验证了所提方法在轨迹跟踪中的有效性。代码可在https://gradslab.github.io/DFL_QP_Unicycle/获取。
cs.RO / 129 / 2606.23371
TSD: A Physics-Inspired Trajectory Saliency Detector for Efficient Imitation Learning
TSD：一种受物理启发的轨迹显著性检测器，用于高效模仿学习
Abstract
For imitation learning in robotic manipulation, high data collection costs result in the scarcity of high quality data. In this paper, we leverage the inherent heterogeneity of trajectories to address this challenge. Based on our observations of manipulation tasks, we categorize motions into transitional, precise, and agile types, defining the latter two as trajectory saliency due to their criticality to task success in contrast to the prevalent but less relevant transitional motions. Therefore, we propose the Trajectory Saliency Detector (TSD), a training-free and plug-and-play framework to identify trajectory saliency. TSD employs two physically-grounded metrics: spatial entropy to capture fine-grained manipulation and centripetal acceleration to detect agile maneuvering. We further leverage TSD to develop a dataset compression method that reduces training costs and a dataset expansion strategy that improves data collection efficiency. Extensive experiments in both simulation and real-world settings demonstrate that models trained on TSD-condensed datasets achieve comparable or even superior performance with 25% less data on average. These results validate the effectiveness of our dataset compression and expansion strategies, thereby confirming the utility of TSD. Consequently, TSD offers a scalable and cost-effective pathway to synthesize information-dense datasets for efficient robot learning. Project page: https://trajectory-saliency-detector.github.io/trajectory-saliency-detector/
Chinese Translation
在机器人操作的模仿学习中，高昂的数据收集成本导致高质量数据的稀缺。本文利用轨迹的内在异质性来应对这一挑战。基于我们对操作任务的观察，我们将运动分为过渡型、精确型和灵活型，并将后两者定义为轨迹显著性，因为它们对任务成功至关重要，相较于普遍存在但相关性较低的过渡型运动。因此，我们提出了轨迹显著性检测器（Trajectory Saliency Detector, TSD），这是一种无训练且即插即用的框架，用于识别轨迹显著性。TSD采用了两个基于物理的度量标准：空间熵用于捕捉细粒度的操作，向心加速度用于检测灵活的机动。我们进一步利用TSD开发了一种数据集压缩方法，以降低训练成本，以及一种数据集扩展策略，以提高数据收集效率。在模拟和现实环境中进行的大量实验表明，基于TSD压缩数据集训练的模型在数据量平均减少25%的情况下，能够实现可比甚至更优的性能。这些结果验证了我们数据集压缩和扩展策略的有效性，从而确认了TSD的实用性。因此，TSD为合成信息密集型数据集提供了一条可扩展且具有成本效益的途径，以实现高效的机器人学习。项目页面：https://trajectory-saliency-detector.github.io/trajectory-saliency-detector/
cs.RO / 130 / 2606.23420
Flowing With Purpose: Latent Action Guided Flow Matching Policies For Robotic Manipulation
有目的地流动：潜在动作引导的流匹配策略用于机器人操作
Abstract
Flow matching has recently become a new standard for behavior cloning in robotic manipulation. However, state-of-the-art flow matching policies suffer from a systematic structural mismatch: they rely on a globally fixed isotropic source distribution despite the strongly fragmented and heteroscedastic structure of robotic action spaces. This agnostic initialization forces the model to learn highly entangled vector fields, bottlenecking training efficiency and limiting overall policy performance. To address this limitation, we introduce Latent Action Guided Flow Matching (LAFM), a novel framework that replaces the monolithic Gaussian with an adaptive library of learned prior distributions. By grounding these distributions using a latent action model, LAFM maps current observations to discrete motion primitives, selecting a specialized base distribution that provides an informed, structurally aligned initialization for the denoising process. This dynamic adaptivity naturally accommodates heteroscedasticity in human demonstrations and makes transport trajectories shorter and less entangled. Empirically, LAFM substantially outperforms standard flow matching formulations, increasing task success rates by 23.4% in real-world robotic deployments and by 10.4% on the LIBERO-90 benchmark. Furthermore, we demonstrate that LAFM achieves state-of-the-art results, surpassing massively pre-trained vision-language-action models while utilizing significantly smaller architectures.
Chinese Translation
流匹配最近成为机器人操作中行为克隆的新标准。然而，最先进的流匹配策略存在系统性的结构不匹配：它们依赖于全球固定的各向同性源分布，尽管机器人动作空间的结构高度碎片化且异方差。这种无知的初始化迫使模型学习高度纠缠的向量场，瓶颈化了训练效率并限制了整体策略性能。为了解决这一限制，我们引入了潜在动作引导的流匹配（Latent Action Guided Flow Matching, LAFM），这是一个新颖的框架，它用自适应的学习先验分布库替代了单一的高斯分布。通过使用潜在动作模型来基于这些分布进行定位，LAFM将当前观察映射到离散的运动原语，选择一个提供信息丰富、结构对齐初始化的专门基础分布，以用于去噪过程。这种动态适应性自然适应于人类示范中的异方差，使得传输轨迹更短且不那么纠缠。实证结果表明，LAFM在标准流匹配公式中表现出显著优势，在真实世界的机器人部署中任务成功率提高了23.4%，在LIBERO-90基准上提高了10.4%。此外，我们证明LAFM达到了最先进的结果，超越了大规模预训练的视觉-语言-动作模型，同时使用了显著更小的架构。
cs.RO / 131 / 2606.23431
DexTeleop-0: Force-Aware Bimanual Dexterous Teleoperation with Ego-Centric Perception towards Shared Autonomy
DexTeleop-0：基于自我中心感知的力感知双手灵巧遥操作以实现共享自主性
Abstract
Fine-grained, bimanual dexterous manipulation remains a foundational challenge in robotics. Traditional teleoperation systems often fail in contact-rich tasks because embodiment gaps hinder accurate kinematic mapping, while tactile and force feedback remain absent. Consequently, data collection efficiency for high-precision tasks remains prohibitively low. To address these limitations, we propose a tactile-driven adaptation strategy designed to enable fine-grained manipulation on top of teleoperation pipelines. Instantiated within our bimanual dexterous framework, DexTeleop-0, this strategy introduces a real-time optimization loop that bridges the embodiment gap by translating coarse human tracking intents into precise, force-compliant robotic commands with tactile sensing. By estimating accurate contact points and leveraging a tactile-enabled fingertip force-sensing profile, the system dynamically computes localized corrections using the operational space Jacobian with respect to joint angle updates. We rigorously evaluate this tactile-driven adaptation strategy across both simulated environments and real-world hardware. Compared with representative baselines, the proposed method consistently achieves higher task success rates and improved execution efficiency in robust grasping, disturbance-resilient manipulation, and complex dexterous tasks.
Chinese Translation
细粒度的双手灵巧操作仍然是机器人技术中的一个基础性挑战。传统的遥操作系统在接触丰富的任务中往往失败，因为身体化差距阻碍了准确的运动学映射，同时缺乏触觉和力反馈。因此，高精度任务的数据收集效率仍然极低。为了解决这些限制，我们提出了一种基于触觉驱动的适应策略，旨在支持在遥操作管道上进行细粒度操作。该策略在我们的双手灵巧框架DexTeleop-0中实现，引入了一个实时优化循环，通过将粗略的人类跟踪意图转化为精确的、符合力的机器人指令，来弥补身体化差距。通过估计准确的接触点并利用触觉启用的指尖力传感配置，系统使用操作空间雅可比矩阵动态计算相对于关节角度更新的局部修正。我们在模拟环境和真实硬件上严格评估了这种基于触觉驱动的适应策略。与代表性基线相比，所提出的方法在稳健抓取、抗干扰操作和复杂灵巧任务中始终实现了更高的任务成功率和更好的执行效率。
cs.RO / 132 / 2606.23444
SkyJEPA: Learning Long-Horizon World Models for Zero-Shot Sim-to-Real Control of Quadrotors
SkyJEPA：用于四旋翼零-shot模拟到真实控制的长时间世界模型学习
Abstract
Accurate dynamics models are critical for informed decision-making in robotic systems, particularly for agile aerial vehicles operating under uncertainty. Neural network dynamics models are attractive for capturing complex nonlinear effects, but existing predictive approaches struggle with long-horizon forecasting because their autoregressive rollout mechanism amplifies errors over time. Joint Embedding Predictive Architectures (JEPAs) offer a compelling alternative by modeling dynamics in latent space, yet prior JEPA-style methods for robot navigation have been studied primarily for kinematic-level planning, with limited investigation in high-frequency control. In this work, we introduce the JEPA-style model for real-time quadrotor control. The proposed approach combines a latent dynamics model with a novel physics-inspired prober that maps frozen latents to interpretable state, enabling physically grounded long-horizon prediction. Additionally, we combine the learned model with a sampling-based optimal control solution to take advantage of its predictive capabilities for real-time control on embedded hardware. Finally, to reduce the dependence on expensive and unsafe real-world data collection, we develop a structured pipeline for automated dataset generation. Extensive open-loop and outdoor closed-loop experiments demonstrate accurate prediction, robust zero-shot sim-to-real transfer, and strong generalization across diverse operating conditions.
Chinese Translation
准确的动力学模型对于机器人系统中的知情决策至关重要，特别是在不确定性下操作的灵活空中飞行器。神经网络动力学模型因其能够捕捉复杂的非线性效应而受到青睐，但现有的预测方法在长时间预测方面面临挑战，因为其自回归展开机制会随着时间的推移放大误差。联合嵌入预测架构（Joint Embedding Predictive Architectures, JEPA）通过在潜在空间中建模动力学提供了一种引人注目的替代方案，但之前的JEPA风格方法主要集中在运动学层面的规划，对高频控制的研究相对有限。在本研究中，我们引入了用于实时四旋翼控制的JEPA风格模型。所提出的方法结合了潜在动力学模型和一种新颖的物理启发式探测器，该探测器将冻结的潜在变量映射到可解释的状态，从而实现基于物理的长时间预测。此外，我们将学习到的模型与基于采样的最优控制解决方案相结合，以利用其预测能力进行嵌入式硬件上的实时控制。最后，为了减少对昂贵且不安全的真实世界数据收集的依赖，我们开发了一个结构化的自动数据集生成管道。广泛的开环和户外闭环实验表明，准确的预测、强健的零-shot模拟到真实转移以及在多样化操作条件下的强泛化能力。
cs.RO / 133 / 2606.23502
DVL-DeepONet: A Physics-Guided Operator Learning for Resilient Underwater Navigation
DVL-DeepONet：一种物理引导的算子学习方法用于抗干扰水下导航
Abstract
Autonomous Underwater Vehicles (AUVs) rely heavily on the fusion of inertial sensors and Doppler velocity logs (DVLs) for navigation. In standard autonomous navigation systems, the DVL measures four beam velocities, thereby enabling the estimation of the AUV velocity vector. However, during real-world missions, the DVL may receive noisy or incomplete beam measurements due to marine obstacles, seabed reflections, or environmental disturbances. Furthermore, some low-cost underwater platforms operate without inertial sensors to reduce system complexity and cost. In such cases, reliable estimation of the AUV velocity vector in real-world missing beam scenarios becomes challenging, leading to degraded navigation solutions. To circumvent these challenges and enable resilient underwater navigation, we propose DVL-DeepONet, a physics-guided deep neural operator framework along with three variants. The proposed models are designed to estimate DVL-based velocity information under multiple operational scenarios, including (i) noise-resilient estimation in coupled inertial/DVL measurements, (ii) DVL-only learning, and (iii) beam measurement recovery. By learning a nonlinear operator that maps temporal inertial/DVL observations directly to vehicle velocity while enforcing DVL measurement physics through a consistency constraint, the proposed approach enables robust velocity estimation even under degraded sensing conditions. The proposed framework is validated using real-world AUV experiments, comprising a cumulative path length of approximately 10,000 m. Experimental results demonstrate that the proposed DVL-DeepONet architectures outperform baseline model-based approaches and learning-based algorithms by 40%.
Chinese Translation
自主水下航行器（AUV）在导航中高度依赖惯性传感器与多普勒速度日志（DVL）的融合。在标准的自主导航系统中，DVL测量四束速度，从而能够估计AUV的速度向量。然而，在实际任务中，由于海洋障碍物、海床反射或环境干扰，DVL可能会接收到噪声或不完整的束测量。此外，一些低成本水下平台为了降低系统复杂性和成本，通常不配备惯性传感器。在这种情况下，在实际缺失束测量的场景中可靠地估计AUV的速度向量变得具有挑战性，导致导航解决方案性能下降。为了解决这些挑战并实现抗干扰的水下导航，我们提出了DVL-DeepONet，一种物理引导的深度神经算子框架及其三个变体。所提模型旨在在多种操作场景下估计基于DVL的速度信息，包括（i）在耦合的惯性/DVL测量中进行抗噪声估计，（ii）仅基于DVL的学习，以及（iii）束测量恢复。通过学习一个非线性算子，将时间上的惯性/DVL观测直接映射到车辆速度，同时通过一致性约束强制执行DVL测量物理，所提方法能够在恶劣的传感条件下实现稳健的速度估计。该框架通过真实世界的AUV实验进行了验证，累计路径长度约为10,000米。实验结果表明，所提的DVL-DeepONet架构比基线模型方法和基于学习的算法提高了40%的性能。
cs.RO / 134 / 2606.23531
BiliVLA: Scene-Aware Vision-Language-Action Model with Reinforcement Learning for Autonomous Biliary Endoscopic Navigation
BiliVLA：基于强化学习的场景感知视觉-语言-动作模型用于自主胆道内窥镜导航
Abstract
Endoscopic retrograde cholangiopancreatography (ERCP) demands precise endoscopic navigation and stable biliary cannulation within a narrow monocular field characterized by specular reflections, partial occlusions, and frequent tissue contact. Although recent robotic systems and vision-based assistance techniques improve operator ergonomics and provide perceptual cues, their performance degrades under pronounced anatomical variability and safety-critical visual artifacts, which hinders reliable autonomy in cannulation-grade procedures. Here, we present BiliVLA, a scene-aware Vision-Language-Action (VLA) framework that formulates biliary endoscopic navigation as an instruction-conditioned visuomotor learning problem. Given an endoscopic observation and a stage-specific language instruction, BiliVLA jointly predicts the target category, a grounded bounding box, and a discrete three degrees of freedom (DoF) motor command for a continuum endoscope. The proposed framework incorporates scene-aware supervision to enhance semantic target consistency and safety-aware recovery supervision to induce conservative retreat behaviors under luminal wall contact. A key component of BiliVLA is a two-stage training paradigm that combines grounding-enhanced supervised fine-tuning (SFT) with Group Relative Policy Optimization (GRPO), which significantly improves action reliability and decision consistency during closed-loop navigation. Across three ERCP subtasks, BiliVLA achieves an average action precision of 91.96\% and an overall success rate (SR) of 84.85\% in real-world phantom experiments. These results indicate that integrating semantic grounding, scene-aware learning, and reward-guided optimization improves perception-action alignment and enables robust autonomous endoscopic navigation.
Chinese Translation
内窥镜逆行胰胆管造影（ERCP）要求在狭窄的单目视野中进行精确的内窥镜导航和稳定的胆道插管，该视野特征包括镜面反射、部分遮挡和频繁的组织接触。尽管近期的机器人系统和基于视觉的辅助技术改善了操作员的人体工学并提供了感知线索，但在明显的解剖变异和安全关键的视觉伪影下，它们的性能会下降，这阻碍了在插管级程序中实现可靠的自主性。在此，我们提出了BiliVLA，一个场景感知的视觉-语言-动作（VLA）框架，将胆道内窥镜导航表述为一个基于指令的视觉运动学习问题。在给定内窥镜观察和特定阶段的语言指令的情况下，BiliVLA共同预测目标类别、一个有根据的边界框以及一个用于连续内窥镜的离散三自由度（DoF）运动指令。所提出的框架结合了场景感知的监督，以增强语义目标的一致性，并采用安全感知的恢复监督，以在腔道壁接触下诱导保守的撤退行为。BiliVLA的一个关键组成部分是一个两阶段训练范式，该范式结合了增强基础的监督微调（SFT）与组相对策略优化（GRPO），显著提高了闭环导航中的动作可靠性和决策一致性。在三个ERCP子任务中，BiliVLA在真实世界的幻影实验中实现了91.96%的平均动作精度和84.85%的整体成功率（SR）。这些结果表明，整合语义基础、场景感知学习和奖励引导优化能够改善感知-动作对齐，并实现稳健的自主内窥镜导航。
cs.RO / 135 / 2606.23565
HoloAgent-0: A Unified Embodied Agent Framework with 3D Spatial Memory
HoloAgent-0：一个具有三维空间记忆的统一具身代理框架
Abstract
LLM agents follow a practical execution loop in digital environments: they reason over structured states, invoke tools, inspect feedback, and revise actions. Extending this loop to physical robots is difficult because physical execution is continuous, embodiment-dependent, uncertain, and constrained by safety. Existing embodied-AI systems have advanced manipulation, spatial understanding, navigation, and humanoid control, but these capabilities often remain specialized modules or loosely coupled decision loops. In this work, we introduce HoloAgent-0, a unified embodied agent framework for real-world robot deployment. Embodied AgentOS converts language instructions into executable skill graphs, schedules robot resources, monitors execution, and triggers clarification or re-planning from runtime feedback. HoloAgent-0 organizes heterogeneous robot models and controllers through three coupled layers: Embodied AgentOS for closed-loop execution, 3D spatial memory for physical world grounding, and embodied skills for robot action. We deploy HoloAgent-0 on real hardware and evaluate its spatial memory, long-horizon navigation, and closed-loop execution across motion generation, object search, cross-robot coordination, and mobile manipulation.
Chinese Translation
大型语言模型（LLM）代理在数字环境中遵循一个实用的执行循环：它们对结构化状态进行推理，调用工具，检查反馈，并修正行动。将这一循环扩展到物理机器人中是困难的，因为物理执行是连续的、依赖于具身性、不确定的，并受到安全性的限制。现有的具身人工智能系统在操作、空间理解、导航和类人控制方面取得了进展，但这些能力往往仍然是专门的模块或松散耦合的决策循环。在本研究中，我们介绍了HoloAgent-0，一个用于现实世界机器人部署的统一具身代理框架。具身AgentOS将语言指令转换为可执行的技能图，调度机器人资源，监控执行，并根据运行时反馈触发澄清或重新规划。HoloAgent-0通过三个耦合层组织异构的机器人模型和控制器：用于闭环执行的具身AgentOS、用于物理世界基础的三维空间记忆，以及用于机器人行动的具身技能。我们在真实硬件上部署HoloAgent-0，并评估其空间记忆、长时间导航和在运动生成、物体搜索、跨机器人协调和移动操作中的闭环执行能力。
cs.RO / 136 / 2606.23588
A Generative Model for Closed-Loop Microsimulation of Signalized Intersections
用于信号交叉口闭环微观仿真的生成模型
Abstract
Traffic microsimulators rely on hand-crafted behavior models that reproduce aggregate flow but miss the heterogeneous interactions between vehicles at signalized intersections. Learned trajectory predictors capture richer interactions but are short-horizon and tend to be unstable when run in closed loop. We present Enactor, an actor-centric generative model for closed-loop intersection microsimulation. The model focuses on vehicles; pedestrians are included as context that can influence vehicle decisions but not predicted. Dynamic actors and lane polylines are encoded in polar coordinates referenced to the intersection center. A transformer with separate spatial and temporal attention blocks predicts a distribution over each actor's next-step motion ($s$, $\alpha$). Training uses a closed-loop curriculum so the model is exposed to its own predictions. We evaluate Enactor in two regimes. In a 4000-second simulation-in-the-loop test at two intersection geometries, Enactor controls every dynamic vehicle against a continuously refreshing actor set rather than the fixed cohort that learned trajectory predictors are usually evaluated against. It recovers the SUMO data generator's speed and travel-time distributions with KL divergence over an order of magnitude lower than a recent transformer baseline on travel time, and substantially lower on speed (roughly $5\times$ lower at Site 1), and reduces red-light violations relative to the same baseline by more than an order of magnitude. An ablation isolates the leader rear-bumper feature as the change with the largest effect on intersection-aware safety metrics. We also evaluate on real-world field data and apply the same architecture to naturalistic vehicle trajectories from a fish-eye camera at a signalized intersection and evaluate it on multi-horizon predictive tasks. Enactor outperforms a constant-velocity baseline at every horizon evaluated.
Chinese Translation
交通微观仿真器依赖于手工制作的行为模型，这些模型能够再现总体流量，但忽略了信号交叉口车辆之间的异质交互。学习的轨迹预测器捕捉了更丰富的交互，但通常是短期的，并且在闭环运行时往往不稳定。我们提出了Enactor，一个以行为者为中心的生成模型，用于闭环交叉口微观仿真。该模型专注于车辆；行人作为上下文被纳入其中，能够影响车辆决策，但不进行预测。动态行为者和车道多边形以极坐标形式编码，参考交叉口中心。一个具有独立空间和时间注意力模块的变换器预测每个行为者下一步运动的分布（$s$，$eta$）。训练采用闭环课程，使模型能够接触到自身的预测。我们在两个领域评估Enactor。在一个4000秒的循环仿真测试中，针对两种交叉口几何形状，Enactor控制每个动态车辆，而不是通常用于评估学习轨迹预测器的固定队列，而是一个不断更新的行为者集合。它恢复了SUMO数据生成器的速度和旅行时间分布，KL散度比最近的变换器基线在旅行时间上低了一个数量级，在速度上显著更低（在站点1大约低$5 imes$），并且相对于同一基线减少了红灯违规行为超过一个数量级。一项消融实验将领导者后保险杠特征隔离为对交叉口安全指标影响最大的变化。我们还在真实世界的现场数据上进行评估，并将相同的架构应用于信号交叉口鱼眼摄像头捕捉的自然车辆轨迹，并在多时间范围预测任务上进行评估。Enactor在每个评估的时间范围内均优于恒速基线。
cs.RO / 137 / 2606.23589
KEMO: Event-Driven Keyframe Memory for Long-Horizon Robot Manipulation with VLA Policies
KEMO：基于事件驱动的关键帧记忆框架用于长时间范围机器人操作与VLA策略
Abstract
Long-horizon robot manipulation remains challenging because similar observations may occur at different execution stages, while the appropriate action depends on previously completed operations. Memory can address this ambiguity by enabling policies to infer task progress from execution history. However, existing memory-augmented approaches often either retain dense histories that require compression or rely primarily on recent context that may discard earlier task-relevant events. In this work, we propose propose KEMO, a lightweight plug-in memory framework that automatically selectively preserves keyframes associated with task-relevant state changes for VLA policies. KEMO combines robot kinematics with visual filtering to detect events, encodes the selected keyframes as compact temporally ordered memory tokens, and integrates them with current visual features through cross-attention and gated residual fusion for VLA training. The detected events also define higher-weight training samples near critical transitions. We evaluate KEMO on various real-world dual-arm manipulation tasks spanning 2 to 6 scored subtasks, and trajectory length ranging from 830 steps to 2846 execution steps (durations from 28 to 95 seconds). Compared with the memory-free baseline (e.g., $\pi_{0.5}$), KEMO improves aggregate Task Success Rate by 23.6\% and Stage Completion Rate by 34.1\%. Ablations show that event-driven keyframe selection outperforms uniform sampling and recent-frame retention, while the proposed gated fusion and keyframe-aligned loss weighting provide complementary gains.
Chinese Translation
长时间范围的机器人操作仍然面临挑战，因为在不同的执行阶段可能会出现相似的观察，而适当的动作依赖于之前完成的操作。记忆可以通过使策略能够从执行历史中推断任务进展来解决这种模糊性。然而，现有的增强记忆的方法往往要么保留需要压缩的密集历史，要么主要依赖于可能丢弃早期与任务相关事件的近期上下文。在本研究中，我们提出了KEMO，一个轻量级的插件记忆框架，能够自动选择性地保留与任务相关状态变化相关的关键帧，以用于VLA策略。KEMO结合了机器人运动学和视觉过滤来检测事件，将选定的关键帧编码为紧凑的时间顺序记忆标记，并通过交叉注意力和门控残差融合将其与当前视觉特征集成，以进行VLA训练。检测到的事件还定义了在关键过渡附近的高权重训练样本。我们在各种现实世界的双臂操作任务上评估了KEMO，这些任务涵盖了2到6个评分子任务，轨迹长度从830步到2846步（持续时间从28秒到95秒）。与无记忆基线（例如，$ ext{π}_{0.5}$）相比，KEMO将整体任务成功率提高了23.6\%，阶段完成率提高了34.1\\%。消融实验表明，基于事件的关键帧选择优于均匀采样和近期帧保留，而所提出的门控融合和关键帧对齐损失加权提供了互补的增益。
cs.RO / 138 / 2606.23593
Real-Time Multimodal Activity-Aware Error Detection in Robot-Assisted Surgery
实时多模态活动感知的机器人辅助手术错误检测
Abstract
Robot-assisted minimally invasive surgery improves surgical precision but introduces complexity, making technical error detection essential for ensuring patient safety. Current executional error detection methods using video data often overlook fine-grained contextual descriptions of activities and error types within the hierarchical structure of surgical procedures. They also under-utilize complementary multimodal information. We propose a unified framework for executional error detection that leverages multimodal input, including video, kinematics, and descriptive textual prompts. Through activity prompting, we integrate descriptive language in gesture-level activities, instrument-object interactions, and error definitions. We also introduce activity-aware visual embeddings derived from vision encoders pretrained on surgical activity labels to compare the effectiveness of contrastive language-image embeddings with traditional image-based embeddings for error detection. By seamlessly integrating kinematic data with video and textual modalities, our framework significantly improves error detection performance. Achieving up to 5\% and 16.6\% F1 score improvements over state-of-the-art baselines on the JIGSAWS and SAR-RARP50 datasets, respectively, we demonstrate the value of combining curated textual prompts with multimodal data for accurate error detection.
Chinese Translation
机器人辅助的微创手术提高了外科手术的精确性，但也带来了复杂性，使得技术错误检测对于确保患者安全至关重要。目前基于视频数据的执行错误检测方法往往忽视了手术过程中的活动和错误类型的细粒度上下文描述及其层次结构，同时也未充分利用互补的多模态信息。我们提出了一种统一的执行错误检测框架，利用包括视频、运动学和描述性文本提示在内的多模态输入。通过活动提示，我们将描述性语言整合到手势级活动、器械-物体交互和错误定义中。我们还引入了基于视觉编码器预训练的手术活动标签生成的活动感知视觉嵌入，以比较对比语言-图像嵌入与传统图像嵌入在错误检测中的有效性。通过将运动学数据与视频和文本模态无缝集成，我们的框架显著提高了错误检测性能。在JIGSAWS和SAR-RARP50数据集上，F1分数分别比最先进的基线提高了5 ext{%}和16.6 ext{%}，证明了将精心策划的文本提示与多模态数据结合用于准确错误检测的价值。
cs.RO / 139 / 2606.23606
Autonomous Subsea Cable Search and Tracking with Graph-Optimised Priors and Visual Tracking
基于图优化先验和视觉跟踪的自主海底电缆搜索与追踪
Abstract
Global communications rely on subsea cable infrastructure that remains vulnerable to damage from natural hazards and human activity. Autonomous underwater vehicles (AUVs) offer an efficient means to inspect long sections of exposed cable, but uncertainty in cable route maps, small cable diameters and partial burial makes continuous tracking a challenge. This paper presents a novel cable search and tracking method that leverages uncertain prior cable route maps. Graph-based optimisation continuously update the cable route to remain consistent with visual observations. Route uncertainty is constrained as a function of distance from observations using physics-based catenary models that account for cable parameters (i.e., lay depth, diameter, and density), bounding the search space to physically feasible regions and improving search efficiency. Cable detection is performed using a semi-supervised classifier running in real-time on-board a camera-equipped AUV. These detections both update the graph-based optimisation and enable visual cable tracking. When tracking is lost due to misclassification, burial or imperfect control, the bounded search space enables efficient recovery. The approach was demonstrated in field trials using the University of Southampton's Smarty200 AUV. The system successfully located the cable despite deliberate errors in it initial cable route map, updating this to be consistent with observations and using visual tracking to inspect up to 59% of a 120m test cable, with successful recovered after tracking loss.
Chinese Translation
全球通信依赖于海底电缆基础设施，而这些基础设施易受到自然灾害和人类活动的损害。自主水下航行器（AUV）提供了一种高效的手段来检查长段暴露的电缆，但电缆路线图的不确定性、小电缆直径和部分埋藏使得连续追踪面临挑战。本文提出了一种新颖的电缆搜索与追踪方法，该方法利用不确定的先验电缆路线图。基于图的优化方法持续更新电缆路线，以保持与视觉观测的一致性。通过基于物理的悬链线模型约束路线的不确定性，该模型考虑了电缆参数（即铺设深度、直径和密度），将搜索空间限制在物理可行区域，从而提高搜索效率。电缆检测使用在配备摄像头的AUV上实时运行的半监督分类器进行。这些检测不仅更新了基于图的优化，还实现了视觉电缆追踪。当由于误分类、埋藏或控制不完善而导致追踪丢失时，有限的搜索空间使得高效恢复成为可能。该方法在南安普敦大学的Smarty200 AUV上进行了现场试验，系统成功定位了电缆，尽管初始电缆路线图中存在故意错误，更新后的路线与观测结果一致，并使用视觉跟踪检查了多达59%的120米测试电缆，在追踪丢失后成功恢复。
cs.RO / 140 / 2606.23617
RECALL: Recovery Experience Collection for Active Lifelong Learning in Vision-Language-Action Models
RECALL：用于视觉-语言-行动模型的主动终身学习恢复经验收集
Abstract
Vision-Language-Action (VLA) models are commonly fine-tuned through passive imitation learning, where additional demonstrations are collected for tasks where the policy performs poorly. This approach incurs several downsides: it requires the robot to fail before data collection is triggered, provides little guidance about which states require supervision, and wastes demonstrator effort on redundant parts of the task where the policy already performs well. In this paper, we propose an active, continual learning paradigm for VLAs. We demonstrate that active, uncertainty-guided data collection leads to more efficient fine-tuning than when using passively-collected demonstrations. However, we also find that fine-tuning only on actively-collected recovery data leads to catastrophic forgetting. We evaluate techniques for continual learning, including replay-based data mixing and elastic weight consolidation, and identify tradeoffs between plasticity to uncertainty-guided recovery data and retention of previously learned behaviors. Overall, our work contributes an empirical study of active continual learning for autoregressive VLAs, establishing that uncertainty-guided recovery demonstrations can improve adaptation efficiency while also revealing open challenges when targeted new data is incorporated into large robot policies.
Chinese Translation
视觉-语言-行动（VLA）模型通常通过被动模仿学习进行微调，在这种情况下，会收集额外的演示数据以应对策略表现不佳的任务。这种方法存在几个缺点：它要求机器人在数据收集被触发之前先失败，提供的指导信息有限，无法明确指出哪些状态需要监督，并且浪费了演示者在策略已经表现良好的任务冗余部分上的努力。在本文中，我们提出了一种针对VLA的主动持续学习范式。我们证明了主动的不确定性引导数据收集比使用被动收集的演示数据更有效地进行微调。然而，我们也发现仅在主动收集的恢复数据上进行微调会导致灾难性遗忘。我们评估了持续学习的技术，包括基于重放的数据混合和弹性权重整合，并识别出对不确定性引导的恢复数据的可塑性与保留先前学习行为之间的权衡。总体而言，我们的工作为自回归VLA的主动持续学习提供了实证研究，确立了不确定性引导的恢复演示可以提高适应效率，同时也揭示了在将目标新数据纳入大型机器人策略时所面临的开放挑战。
cs.RO / 141 / 2606.23623
dVLA-RL: Reinforcement Learning over Denoising Trajectories for Discrete Diffusion Vision-Language-Action Models
dVLA-RL：基于去噪轨迹的强化学习在离散扩散视觉-语言-动作模型中的应用
Abstract
Vision-Language-Action (VLA) models have established a powerful paradigm for generalist robotic manipulation by grounding control into the semantic reasoning of VLMs. Prevailing architectures typically model actions continuously via diffusion or flow processes, or discretely through either autoregressive generation or parallel decoding. Recently, Discrete Diffusion VLAs (dVLAs) have emerged as a distinct alternative, unifying vision, language, and action into a single discrete token space via masked generative modeling. While combining iterative refinement with unified representations, its training has thus far been restricted to Supervised Fine-Tuning (SFT), leaving the potential of Reinforcement Learning (RL) for further policy refinement largely unexplored. A fundamental challenge in RL for dVLAs is that the marginal probability of the final action generated by dVLAs remains intractable. To solve this problem, we propose \textbf{dVLA-RL}, shifting the learning objective from the marginal action probability to the joint probability of the sampled generation path. Specifically, by modeling the denoising process as a Markov Decision Process (MDP), we mathematically formulate this path probability as a product of step-wise transitions. This trajectory-level objective provides a unified formulation that natively accommodates variable denoising steps. Leveraging this intrinsic fexibility, we introduce a unified step scheduling approach for complex multi-task learning, tailoring denoising steps to specific task complexities to maximize both success rates and computational effciency. Extensive evaluations demonstrate that our approach achieves a success rate of \textbf{99.7\%} on LIBERO. Furthermore, it establishes strong VLA-based results on RoboTwin 2.0 by delivering a \textbf{30.6\%} improvement over the SFT baseline, remaining competitive with strong World-Action Model baselines.
Chinese Translation
视觉-语言-动作（VLA）模型通过将控制与视觉语言模型（VLMs）的语义推理相结合，建立了一个强大的通用机器人操作范式。现有架构通常通过扩散或流动过程连续建模动作，或通过自回归生成或并行解码离散建模动作。最近，离散扩散VLA（dVLA）作为一种独特的替代方案出现，通过掩蔽生成建模将视觉、语言和动作统一到一个离散的标记空间中。尽管将迭代精炼与统一表示相结合，但其训练迄今为止仅限于监督微调（SFT），使得强化学习（RL）在进一步策略精炼中的潜力尚未得到充分探索。dVLA在RL中的一个基本挑战是，dVLA生成的最终动作的边际概率仍然是不可处理的。为了解决这个问题，我们提出了 extbf{dVLA-RL}，将学习目标从边际动作概率转移到采样生成路径的联合概率。具体而言，通过将去噪过程建模为马尔可夫决策过程（MDP），我们将该路径概率数学上表述为逐步转移的乘积。这个轨迹级别的目标提供了一个统一的公式，能够自然地适应可变的去噪步骤。利用这种内在的灵活性，我们引入了一种统一的步骤调度方法，用于复杂的多任务学习，根据特定任务的复杂性调整去噪步骤，以最大化成功率和计算效率。广泛的评估表明，我们的方法在LIBERO上达到了 extbf{99.7 ext{%}}的成功率。此外，它在RoboTwin 2.0上建立了强大的基于VLA的结果，相较于SFT基线提高了 extbf{30.6 ext{%}}，并与强大的世界动作模型基线保持竞争力。
cs.RO / 142 / 2606.23625
Learning to See While Learning to Act: Diffusion Models for Active Perception in Robot Imitation
在学习行动的同时学习观察：用于机器人模仿的扩散模型的主动感知
Abstract
Most imitation learning methods assume full observability in table-top settings. In practice, objects are often occluded, requiring robots to both search and act, and learning this coupled behavior from limited demonstrations remains challenging. We propose See2Act, an imitation learning approach that conditions action prediction on a sequence of actively-inferred viewpoints at test time, by coupling action denoising with viewpoint refinement. The policy is trained using camera poses anchored to keyframe actions from offline demonstrations, enabling implicit learning of where to see, while learning how to act. We empirically demonstrate that in Ravens the policy recovers informative viewpoints under severe occlusions, and on RLBench tasks it improves performance by up to 34% over prior methods. In the real world, we collect 50 demonstrations in a digital twin and achieve zero-shot sim-to-real transfer on pick-and-place tasks using depth observations. The policy handles significant occlusions, showing that learned viewpoint reasoning enables robust manipulation under partial observability.
Chinese Translation
大多数模仿学习方法假设在桌面设置中具有完全可观察性。实际上，物体常常被遮挡，这要求机器人既要搜索又要行动，而从有限的演示中学习这种耦合行为仍然具有挑战性。我们提出了See2Act，这是一种模仿学习方法，在测试时通过将动作去噪与视点优化相结合，基于一系列主动推断的视点来进行动作预测。该策略使用锚定于离线演示中的关键帧动作的相机姿态进行训练，从而实现了在学习如何行动的同时隐式学习观察的地点。我们通过实验证明，在Ravens环境中，该策略能够在严重遮挡下恢复有信息的视点，并且在RLBench任务中，其性能比之前的方法提高了多达34%。在现实世界中，我们在数字双胞胎中收集了50个演示，并在抓取和放置任务中实现了零样本的模拟到现实转移，使用深度观测。该策略能够处理显著的遮挡，显示出学习到的视点推理能够在部分可观察性下实现稳健的操作。
cs.RO / 143 / 2606.23641
Flatness Preserves Instruction Following in Vision-Language-Action Models
平坦性保持了视觉-语言-动作模型中的指令遵循
Abstract
Vision-language-action (VLA) models have the potential for open-world generalization by leveraging pretrained vision-language representations, yet downstream finetuning on limited robot data often degrades these representations, leading to brittle policies that ignore language instructions in favor of visual shortcuts, a failure mode we term instruction blindness. We hypothesize that standard finetuning with limited data applies gradients to a sparse set of points, which manifests as a sharp loss landscape with high-curvature minima. We propose to address this directly through flatness-preserving optimization while finetuning on the exact same data, where learning a flatter landscape results in a model more robust to perturbations in the weight space. Specifically, we demonstrate that simply applying sharpness-aware minimization during VLA finetuning significantly improves instruction following by over 60% across multiple simulation and real-world benchmarks without additional data, architectural modification, or retraining. We further analyze the effect of selective sharpness, quantify its effects, and show that our approach is complementary to existing guidance techniques. Project page can be found at https://haochenz11.github.io/papers/flatness-vla/.
Chinese Translation
视觉-语言-动作（VLA）模型通过利用预训练的视觉-语言表示，具有开放世界泛化的潜力。然而，在有限的机器人数据上进行下游微调往往会降低这些表示的效果，导致脆弱的策略忽视语言指令而偏向视觉捷径，这种失败模式我们称之为指令盲目性。我们假设，标准的有限数据微调在稀疏的点集上应用梯度，这表现为具有高曲率极小值的尖锐损失景观。我们提出通过在完全相同的数据上进行平坦性保持优化来直接解决这个问题，学习一个更平坦的景观使模型在权重空间中的扰动下更加稳健。具体而言，我们展示了在VLA微调过程中简单地应用尖锐性感知最小化显著提高了指令遵循的效果，提升幅度超过60%，这一改善是在多个模拟和现实世界基准测试中实现的，且无需额外的数据、架构修改或重新训练。我们进一步分析了选择性尖锐性的影响，量化其效果，并表明我们的方法与现有的指导技术是互补的。项目页面可在 https://haochenz11.github.io/papers/flatness-vla/ 找到。
cs.RO / 144 / 2606.23658
A Reduced Order Model for Emergent Mechanics in Woven Systems
一种用于编织系统中涌现力学的降阶模型
Abstract
Woven structures exhibit rich mechanical behaviors including anisotropic stiffness, shear-induced locking, and crimp interchange that emerge purely from the geometric arrangement of individual weavers rather than from constituent material properties. Existing models either homogenize these interactions or resolve them at prohibitive computational cost. We introduce a reduced-order model that bridges this gap by representing individual weaver interactions through a system of nodes and four physically interpretable stiffness elements capturing axial deformation, in-plane uncrimping, inter-weaver shear, and frictional slip. Eigenvalue analysis of the unit cell confirms that the lowest-energy deformation modes correspond directly to known weave-specific phenomena, and that each element is necessary for a complete kinematic and mechanistic description. Element stiffness parameters are calibrated against empirical three-point bending and shear data, achieving agreement within 5% across varied weaver widths and spacings. The validated model is then applied to demonstrate capabilities beyond the reach of continuum approaches including: the emergent Poisson's response arising from crimp interchange, stepwise force reduction during progressive weaver pullout, stress localization under three distinct tearing configurations, and programmable mechanical anisotropy through spatially graded weaver stiffness. The physical transparency and computational efficiency of the framework position it as a practical tool for the analysis and design of woven architected materials with programmable mechanical response.
Chinese Translation
编织结构展现出丰富的机械行为，包括各向异性刚度、剪切引起的锁定以及由于单个编织者的几何排列而产生的褶皱互换，这些行为并非源于材料的组成特性。现有模型要么对这些相互作用进行均匀化，要么以高昂的计算成本进行解析。我们提出了一种降阶模型，通过一组节点及四个具有物理解释的刚度元素来表示单个编织者的相互作用，这些元素捕捉了轴向变形、平面去褶皱、编织者间剪切和摩擦滑移。单位胞的特征值分析确认，最低能量的变形模式直接对应于已知的编织特定现象，并且每个元素对于完整的运动学和力学描述都是必要的。元素刚度参数经过与经验三点弯曲和剪切数据的校准，在不同编织者宽度和间距下实现了5%以内的一致性。经过验证的模型随后被应用于展示超越连续方法的能力，包括：由于褶皱互换而产生的涌现泊松响应、逐步减小的编织者拔出过程中的力、在三种不同撕裂配置下的应力局部化，以及通过空间分级编织者刚度实现的可编程机械各向异性。该框架的物理透明性和计算效率使其成为分析和设计具有可编程机械响应的编织结构材料的实用工具。
cs.RO / 145 / 2606.23680
CoorDex: Coordinating Body and Hand Priors for Continuous Dexterous Humanoid Loco-Manipulation
CoorDex：协调身体与手部先验以实现连续灵巧的人形机器人运动操作
Abstract
Humanoid loco-manipulation is often simplified into a stop-and-go process: walking to an object, stopping to manipulate it, and then resuming locomotion. It also commonly relies on low degree-of-freedom (DoF) end effectors that behave like an open-close grasp primitive. We introduce CoorDex, a learning pipeline that converts high-dimensional body and dexterous hand control into coordinated latent residual control, enabling high-DoF dexterous loco-manipulation on the move. Starting from simulated whole-body and hand demonstrations, CoorDex trains privileged motion tracking teachers for the humanoid body and dexterous hand, distills them into proprioception-conditioned latent priors, and uses the frozen priors as the action space for downstream residual reinforcement learning. A coordinated latent residual policy composes these priors through shared task context and separate body-hand residual heads, preserving natural whole-body motion while improving finger-level contact reliability. CoorDex enables a Unitree G1 humanoid with a 20-DoF WUJI hand to execute dexterous manipulation while in motion, including non-stop bottle grasping and carrying, fridge door opening on the move, and cube pick-and-turn. Ablations on the walk-grasp-carry task show that joint-space PPO, joint-space hand control, and monolithic latent prediction all fail under the same reward budget, while the latent-prior interface and coordinated residual structure make high-dimensional contact-rich loco-manipulation trainable. Project Page: https://skevinci.github.io/coordex/
Chinese Translation
人形机器人的运动操作通常被简化为一个停顿与移动的过程：走向物体，停下来进行操作，然后恢复运动。它通常依赖于低自由度（DoF）的末端执行器，这些执行器的行为类似于开合抓取原语。我们提出了CoorDex，一个学习管道，将高维的身体和灵巧手部控制转换为协调的潜在残差控制，从而实现高自由度的灵巧运动操作。CoorDex从模拟的全身和手部演示开始，为人形机器人身体和灵巧手部训练特权运动跟踪教师，将其提炼为本体感知条件的潜在先验，并将冻结的先验作为下游残差强化学习的动作空间。协调的潜在残差策略通过共享任务上下文和独立的身体-手部残差头组合这些先验，保持自然的全身运动，同时提高指尖级接触的可靠性。CoorDex使得一台配备20自由度WUJI手的Unitree G1人形机器人能够在运动中执行灵巧操作，包括不停地抓取和携带瓶子、在移动中打开冰箱门以及立方体的拾取与旋转。在行走-抓取-携带任务上的消融实验表明，关节空间PPO、关节空间手部控制和单一潜在预测在相同的奖励预算下均未能成功，而潜在先验接口和协调残差结构使得高维接触丰富的运动操作可训练。项目页面：https://skevinci.github.io/coordex/
cs.RO / 146 / 2606.23685
LaST-HD: Learning Latent Physical Reasoning from Scalable Human Data for Robot Manipulation
LaST-HD：从可扩展人类数据中学习潜在物理推理以进行机器人操作
Abstract
Human-hand demonstrations provide a direct and scalable source of physical interaction data for robot learning. While manual retargeting is indispensable for establishing kinematic action correspondence across different morphologies, robust transfer requires going beyond geometry to address the underlying alignment of physical dynamics between human and robot manipulation. To address this, we introduce LaST-HD, a novel human-to-robot action learning paradigm that extends reasoning-before-acting VLA by aligning human-hand and robot demonstrations in a shared latent reasoning space. Rather than mimicking human kinematics, LaST-HD trains an auxiliary action-conditioned world model on unpaired human-hand and robot trajectories to synthesize unified latent targets. After aligning cross-embodiment representations in this shared forward-dynamics space, these targets supervise LaST-HD's latent reasoning process, enabling it to internalize shared physical dynamics and drive efficient human-hand action learning. Moreover, we develop Out-of-Lab (OOL) Glove, a low-cost motion-capture glove tailored to LaST-HD for human-hand data collection. The captured human data provide precise keypoints and serve as universal action supervision across grippers and dexterous hands. Armed with the aligned latent space and high-fidelity human-hand data, we develop a progressive mixed-to-human training recipe comprising mixed human-robot co-training and human-hand online correction post-training. Through mixed co-training, LaST-HD improves generalization to novel objects, scenes, and positions using only human-hand demonstrations. With online correction, LaST-HD further adapts to novel environments and achieves over 90\% accuracy using only 20 minutes of OOL glove data.
Chinese Translation
人手示范为机器人学习提供了直接且可扩展的物理交互数据来源。虽然手动重定向对于在不同形态之间建立运动学动作对应关系是不可或缺的，但稳健的迁移需要超越几何学，解决人类与机器人操作之间物理动态的基本对齐。为此，我们提出了LaST-HD，一种新颖的人机动作学习范式，通过在共享的潜在推理空间中对齐人手和机器人示范，扩展了行动前推理的变分推断学习（VLA）。LaST-HD并不是简单模仿人类的运动学，而是在未配对的人手和机器人轨迹上训练一个辅助的动作条件世界模型，以合成统一的潜在目标。在这个共享的前向动态空间中对齐跨体现表示后，这些目标监督LaST-HD的潜在推理过程，使其能够内化共享的物理动态并推动高效的人手动作学习。此外，我们开发了Out-of-Lab (OOL) 手套，这是一种低成本的动作捕捉手套，专为LaST-HD设计以进行人手数据收集。捕获的人类数据提供了精确的关键点，并作为夹具和灵巧手之间的通用动作监督。借助对齐的潜在空间和高保真的人手数据，我们开发了一种渐进式混合到人类的训练方案，包括混合的人机共同训练和训练后的人手在线校正。通过混合共同训练，LaST-HD在仅使用人手示范的情况下提高了对新物体、新场景和新位置的泛化能力。通过在线校正，LaST-HD进一步适应新环境，并在仅使用20分钟的OOL手套数据的情况下实现了超过90%的准确率。
cs.RO / 147 / 2606.23686
LIBERO-Safety: A Comprehensive Benchmark for Physical and Semantic Safety in Vision-Language-Action Models
LIBERO-Safety：视觉-语言-动作模型中物理与语义安全的综合基准
Abstract
Despite the impressive manipulation capabilities of Vision-Language-Action (VLA) models, their operational safety under strict constraints remains largely unverified. To address this, we introduce a parametric safety benchmark to procedurally generate safety-critical scenarios with comprehensive stochasticity. To overcome the scalability bottlenecks of human teleoperation, we develop a novel keypose-driven data generation pipeline. Leveraging this infrastructure, we curate a large-scale dataset of 19,664 strictly collision-free demonstrations with extensive domain randomization. We then conduct a systematic cross-paradigm evaluation of eight VLA and two embodied foundation models. Our analysis reveals a critical generalization-safety tension: although high-diversity training fosters safer trajectories, task success remains fundamentally bottlenecked by sub-optimal trajectory synthesis and semantic misalignment. By providing a scalable pipeline, a robust dataset, and profound failure-mode insights, LIBERO-Safety establishes a crucial foundation for developing safe and reliable VLA models.
Chinese Translation
尽管视觉-语言-动作（VLA）模型具备令人印象深刻的操作能力，但在严格约束下的操作安全性仍然未得到充分验证。为了解决这一问题，我们引入了一种参数化安全基准，用于程序性生成具有全面随机性的安全关键场景。为了克服人类遥操作的可扩展性瓶颈，我们开发了一种新颖的关键姿态驱动数据生成管道。利用这一基础设施，我们策划了一个包含19,664个严格无碰撞演示的大规模数据集，并进行了广泛的领域随机化。随后，我们对八个VLA模型和两个具身基础模型进行了系统的跨范式评估。我们的分析揭示了一个关键的泛化-安全张力：尽管高多样性的训练促进了更安全的轨迹，但任务成功在根本上仍受到次优轨迹合成和语义不一致的瓶颈限制。通过提供可扩展的管道、强大的数据集和深刻的失败模式洞察，LIBERO-Safety为开发安全可靠的VLA模型奠定了重要基础。
cs.RO / 148 / 2606.23689
AutoDex: An Automated Real-World System for Dexterous Grasping Data Collection
AutoDex：一个自动化的真实世界灵巧抓取数据收集系统
Abstract
Learning robust dexterous grasping requires real-world data that records the physical outcomes of grasp attempts. Such data is hard to obtain at scale: teleoperation yields valid physical outcomes but is slow and operator-biased, while simulation-based generation is cheap and scalable but cannot certify contact validity. A natural solution is to generate candidate grasps and verify them on real hardware, but this scales only if the entire collection loop (perception, execution, labeling, and reset) runs without human intervention. We present AutoDex, an automated real-world data-collection system that closes this loop: for each candidate from a replaceable generator, it localizes the object under severe hand-object occlusion with dense 20-camera perception, executes collision-monitored robot motions, labels lift-and-hold success or failure, and actively resets the object between trials to expose additional candidates across stable poses. The result is a reusable database of physically labeled grasp trials that downstream systems can query by retrieval and feasibility filtering. Using AutoDex, we collect 3,593 grasp trials across Allegro and Inspire hands on 100 diverse objects, with synchronized multi-view observations and robot-state logs. For a matched 500-trajectory collection, AutoDex requires 10.3 h versus 49.4 h for teleoperation, yielding a 4.8x throughput improvement, and grasps retrieved from the AutoDex-validated database succeed 76% versus 34% for simulation-only validation. Code and data will be publicly released.
Chinese Translation
学习稳健的灵巧抓取需要记录抓取尝试物理结果的真实世界数据。然而，这类数据在规模上难以获取：遥操作能够产生有效的物理结果，但速度较慢且受操作员偏见影响，而基于仿真的数据生成虽然便宜且可扩展，但无法验证接触的有效性。一个自然的解决方案是生成候选抓取并在真实硬件上进行验证，但这只有在整个数据收集循环（感知、执行、标记和重置）无需人工干预的情况下才能扩展。我们提出了AutoDex，一个自动化的真实世界数据收集系统，闭合了这一循环：对于来自可替换生成器的每个候选抓取，它在严重的手-物体遮挡下通过密集的20摄像头感知定位物体，执行监测碰撞的机器人动作，标记提升和保持的成功或失败，并在试验之间主动重置物体，以在稳定姿态下暴露更多候选抓取。最终结果是一个可重用的物理标记抓取试验数据库，后续系统可以通过检索和可行性过滤进行查询。使用AutoDex，我们在100个不同物体上收集了3,593个抓取试验，涵盖了Allegro和Inspire手，具有同步的多视角观察和机器人状态日志。对于匹配的500轨迹收集，AutoDex需要10.3小时，而遥操作需要49.4小时，产生了4.8倍的吞吐量提升，从AutoDex验证的数据库中检索的抓取成功率为76%，而仅基于仿真的验证成功率为34%。代码和数据将公开发布。
cs.CV / 1 / 2606.20620
A Viscosity Semigroup Framework for Stable Image Reconstruction
稳定图像重建的粘度半群框架
Abstract
Starting from the axiomatic formulation of scale-space theory, we develop a viscosity-solution framework for multiscale image representations arising from degenerate elliptic-parabolic partial differential equations. Rather than introducing a new semigroup theory, we work within the standard viscosity-solution setting, using comparison principles to obtain well-posedness, uniqueness, and contraction in the supremum norm. This perspective is used to motivate a hybrid reconstruction operator in which a learned inverse map is followed by a nonlinear diffusion evolution. At the continuous level, the diffusion operator satisfies non-expansiveness, which provides stability for the reconstruction process; this framework is then evaluated on a CT-based mesothelioma classification task, where it attains an AUC of 0.875 with negligible variation across epochs, while the baseline model acquires AUC values from 0.49 to 0.80 without a clear convergence pattern. These observations are consistent with the stabilizing role suggested by the discussed viscosity theory.
Chinese Translation
基于尺度空间理论的公理化表述，我们发展了一种粘度解框架，用于从退化的椭圆-抛物型偏微分方程中产生的多尺度图像表示。我们并没有引入新的半群理论，而是在标准的粘度解设置中工作，利用比较原理来获得良定性、唯一性和在上确界范数下的收缩性。这一视角用于激励一种混合重建算子，其中学习到的逆映射后跟随非线性扩散演化。在连续层面上，扩散算子满足非扩展性，这为重建过程提供了稳定性；该框架随后在基于CT的间皮瘤分类任务中进行了评估，达到了0.875的AUC，并且在各个时期之间变化极小，而基线模型的AUC值则在0.49到0.80之间波动，且没有明显的收敛模式。这些观察结果与所讨论的粘度理论所暗示的稳定作用是一致的。
cs.CV / 2 / 2606.20671
A Projection-Based Surrogate Gradient Interpretation for Neural Codec Wrappers
基于投影的神经编解码器包装器的替代梯度解释
Abstract
Neural wrappers are learned pre-and postprocessing networks designed to enhance the performance of conventional video codecs. Although these approaches can significantly improve compression efficiency, training them remains challenging due to the non-differentiability of video codecs, which arises from the multiple discrete decisions involved in the encoding process. Surrogate gradients have recently emerged as an effective solution for enabling end-to-end learning with conventional codecs. They offer two main advantages: they avoid training an additional network to mimic the codec, and they can improve compression performance. In particular, the recently proposed SCALED method, which leverages the true compression error, has shown strong results for training neural pre-processors such as downscalers. However, this SCALED gradient was originally introduced as a reparameterization trick, which limits its interpretability. In this paper, we show that this surrogate gradient can be interpreted as a first-order local approximation of the video codec, providing insight into its effectiveness. We further demonstrate that it is effective not only for learning downscaling operations, but also for the more challenging task of full neural wrapping with pre-and post-processing networks. Finally, we show that the approach generalizes well across different video codecs, quality factors, and tasks, including multiple downscaling ratios, yielding BD-Rate (PSNR) reductions of up to -23.59% on x264 and -20.07% on VVenC relative to standard resampling baselines.
Chinese Translation
神经包装器是学习的前处理和后处理网络，旨在提高传统视频编码器的性能。尽管这些方法可以显著提高压缩效率，但由于视频编码过程中涉及的多重离散决策，训练它们仍然具有挑战性。最近，替代梯度作为一种有效的解决方案出现，使得与传统编码器的端到端学习成为可能。它们提供了两个主要优势：避免训练额外的网络来模拟编码器，并且可以提高压缩性能。特别是，最近提出的 SCALED 方法利用真实的压缩误差，在训练神经前处理器（如降采样器）方面显示出强大的效果。然而，这种 SCALED 梯度最初作为一种重参数化技巧引入，这限制了其可解释性。在本文中，我们展示了这种替代梯度可以被解释为视频编码器的一阶局部近似，从而提供对其有效性的洞察。我们进一步证明，它不仅对学习降采样操作有效，而且对更具挑战性的全神经包装任务（包括前处理和后处理网络）同样有效。最后，我们展示了该方法在不同视频编码器、质量因子和任务（包括多个降采样比例）上的良好泛化能力，相较于标准重采样基线，在 x264 上实现了高达 -23.59% 的 BD-Rate (PSNR) 降低，在 VVenC 上实现了高达 -20.07% 的降低。
cs.CV / 3 / 2606.20676
Jury Duty: Calibration and Orientation Failures in MLLM-as-a-Judge Under Cultural Ambiguity
陪审团职责：在文化模糊下 MLLM 作为法官的校准与定向失误
Abstract
MLLM-as-a-Judge is conventionally validated by agreement with human annotations, but this metric is undefined when the human pool is culturally heterogeneous. We introduce VOIR DIRE, a multimodal benchmark of 626 culturally paired image--prompt artifacts spanning U.S. and mainland Chinese contexts across food, fashion, and architecture, with annotator pools that are within-pool reliable (a = 0.86/0.74) but cross-pool divergent on evaluation (Q1 r = -0.12). Across six MLLMs, the bias decomposes into two failures: a positivity-floor calibration failure (compressed scale use) and an orientation failure (default to one cultural norm). On this corpus, where contested items are sampled to split the two pools, the floor mechanically validates the more-permissive Chinese reading; persona prompting partially recovers calibration, but the orientation residual survives, evidence the tilt is not reducible to scale compression. Reference-pool in-context demonstrations deepen the orientation residual and inflate the high end rather than restoring use of the low end. Model origin adds a small additive tilt (~0.10 MAE) that is approximately invariant under demonstration. We recommend reporting alignment against each reference pool separately and treating cross-pool divergence as a judge property.
Chinese Translation
MLLM 作为法官的传统验证依赖于与人类注释的一致性，但当人类样本具有文化异质性时，该指标并不明确。我们引入了 VOIR DIRE，这是一个包含 626 个文化配对的图像-提示工件的多模态基准，涵盖美国和中国大陆在食品、时尚和建筑等领域的背景，注释者池在池内具有可靠性（α = 0.86/0.74），但在评估时池间存在差异（Q1 r = -0.12）。在六个 MLLM 中，偏差分解为两种失误：一种是积极性底线校准失误（压缩尺度使用），另一种是定向失误（默认遵循某一文化规范）。在这个语料库中，争议项目被抽样以分割两个池，底线机械地验证了更宽容的中国解读；个性化提示部分恢复了校准，但定向残差依然存在，证据表明这种倾斜并不能归结为尺度压缩。参考池中的上下文示范加深了定向残差，并抬高了高端，而不是恢复低端的使用。模型来源增加了一个小的附加倾斜（约 0.10 MAE），在示范下大致保持不变。我们建议分别报告与每个参考池的对齐情况，并将池间差异视为法官属性。
cs.CV / 4 / 2606.20680
Beyond ROC-AUC: Operating-Point Performance Reporting for Biometric Verification
超越ROC-AUC：生物识别验证的操作点性能报告
Abstract
A biometric verifier is often deployed with a strict false match budget, so only a narrow, low false match rate (FMR) slice of the score range is used. A reporting standard for this setting already exists. ISO/IEC 19795-1 asks for error rates at stated operating points, for the detection error tradeoff (DET) curve as the view of the trade-off between FMR and the false non-match rate (FNMR), and for an interval of uncertainty on every value. In practice, a single area under the receiver operating characteristic curve (ROC-AUC), the equal error rate (EER), or a verification accuracy is still reported as the resolution, which is a threshold-independent summary that the standard does not endorse. The full ROC-AUC averages the true match rate (TMR) with equal weight over the whole FMR range from 0 to 1, so almost all of its weight is placed where the system is never operated; low-FMR behavior can then be hidden, and the order of two systems can even be reversed. The guideline is revisited in this paper and tested against seven pretrained matchers across four modalities, face, voice, iris, and fingerprint, each reported with bootstrap confidence intervals and paired bootstrap tests. A system that looks stronger on full ROC-AUC is shown to be significantly worse at FMR = 10^-3. For face, a higher full AUC was obtained by FaceNet, whereas a higher TMR at FMR = 10^-3 was obtained by ArcFace, and both gaps were significant with non-overlapping intervals. Hence, the DET curve and the FNMR at a fixed FMR are re-iterated in this paper as the primary report, with ROC-AUC and EER retained as supplementary context.
Chinese Translation
生物识别验证器通常在严格的误匹配预算下部署，因此仅使用得分范围内狭窄的低误匹配率（FMR）区间。针对这种情况，已经存在一种报告标准。ISO/IEC 19795-1要求在特定操作点报告错误率，并提供检测错误权衡（DET）曲线，以展现FMR与假非匹配率（FNMR）之间的权衡关系，并对每个值提供不确定性区间。在实际应用中，通常仍然报告单一的接收者操作特征曲线下面积（ROC-AUC）、等错误率（EER）或验证准确率作为分辨率，这是一种与阈值无关的总结，标准并不支持。完整的ROC-AUC在0到1的整个FMR范围内以相等的权重平均真实匹配率（TMR），因此几乎所有的权重都集中在系统从未操作的地方；低FMR行为可能被掩盖，甚至两个系统的顺序可能被颠倒。本文重新审视了该指南，并对四种模态（面部、声音、虹膜和指纹）中七个预训练匹配器进行了测试，每个匹配器都报告了自助置信区间和配对自助测试。结果显示，在完整的ROC-AUC上看似更强的系统在FMR = 10^-3时显著较差。对于面部识别，FaceNet获得了更高的完整AUC，而在FMR = 10^-3时，ArcFace获得了更高的TMR，这两个差距在不重叠的区间中是显著的。因此，本文将DET曲线和在固定FMR下的FNMR重新作为主要报告内容，同时保留ROC-AUC和EER作为补充背景。
cs.CV / 5 / 2606.20681
A UAV-Based Multi-Modal Vision System for Automated Sideslope Deformation Monitoring and Hazard Detection
基于无人机的多模态视觉系统用于自动化边坡变形监测与危险检测
Abstract
Slope hazards constitute a major safety threat to expressway infrastructure, and their evolution is typically manifested as slow surface deformation. Conventional manual inspection suffers from low efficiency and inadequate operational safety, especially on severely deteriorated slopes. Accordingly, there is an urgent need for an automated, high-precision solution capable of large-area slope observation and analysis. This study aims to develop a highly automated workflow for slope hazard detection using Unmanned Aerial Vehicle (UAV)-borne Light Detection and Ranging (LiDAR). The proposed workflow consists of a shared data-acquisition and ground-surface extraction stage, a single-observation hazard-screening branch based on RandLA-Net, and a multi-epoch deformation-monitoring branch based on grid-wise elevation differencing. To validate the effectiveness of the proposed system, we conducted multiple UAV-borne LiDAR data-acquisition flights in real expressway slope environments. The results show that the workflow can extract usable ground-surface point clouds under vegetation cover, identify potential hazard zones from single-observation point clouds, and quantify centimeter-level elevation changes using multi-epoch grid differencing. This study establishes an end-to-end UAV-borne LiDAR-based workflow for slope inspection and demonstrates its feasibility through controlled experiments, field tests, and simulation-based validation, thereby providing an implementable solution for automated slope-hazard monitoring and intelligent early warning.
Chinese Translation
边坡灾害对高速公路基础设施构成了重大安全威胁，其演变通常表现为缓慢的表面变形。传统的人工检查效率低下且操作安全性不足，尤其是在严重退化的边坡上。因此，迫切需要一种能够进行大面积边坡观察和分析的自动化高精度解决方案。本研究旨在开发一种基于无人机（UAV）搭载的激光雷达（LiDAR）的高自动化边坡灾害检测工作流程。所提工作流程包括共享数据采集和地面表面提取阶段、基于RandLA-Net的单次观测危险筛选分支，以及基于网格高程差分的多时相变形监测分支。为了验证所提系统的有效性，我们在真实的高速公路边坡环境中进行了多次无人机搭载LiDAR的数据采集飞行。结果表明，该工作流程能够在植被覆盖下提取可用的地面表面点云，从单次观测点云中识别潜在危险区域，并使用多时相网格差分量化厘米级的高程变化。本研究建立了一种端到端的基于无人机搭载LiDAR的边坡检查工作流程，并通过受控实验、现场测试和基于仿真的验证展示了其可行性，从而为自动化边坡灾害监测和智能预警提供了可实施的解决方案。
cs.CV / 6 / 2606.20682
Open Annotations and Synthetic Data for Field Localisation in Indian Bank Cheques
印度银行支票中的开放注释和合成数据用于字段定位
Abstract
Automated cheque processing requires localising key fields (date, legal amount, IFSC code, account number, signature, and payee name) before any recognition step. The IDRBT Cheque Image Dataset is, to our knowledge, the only public collection of Indian bank cheques, but it ships without field annotations and with no stated licence, so its redistribution terms are unclear. We address both limitations. First, we release six-field bounding-box annotations for all 112 cheques in the dataset, distributed annotations-only and keyed to the original filenames so that the IDRBT redistribution terms are respected. Second, we release 295 fully redistributable synthetic cheque images produced by a cut-paste pipeline that composites annotated field regions from real cheques onto content-erased, bank-specific canvas templates; because patches are pasted at their source coordinates, annotations carry forward unchanged. Third, we provide a ResNet-50 direct-regression baseline that predicts all six fields in a single forward pass, and use it for a controlled test of the synthetic data. The test is sobering: because cheque layouts are rigid, a no-learning baseline that simply predicts each field's mean training box already reaches 0.691 mean IoU and 80% accuracy at IoU >= 0.5, and once seed variance and training compute are accounted for, the cut-paste synthetic data yields no measurable improvement over real data alone (an equal-compute real-only model matches or beats the synthetic-augmented model on every aggregate metric). We report this negative result in full, since it cautions against assuming appearance-only augmentation helps fixed-layout documents and points instead to layout-varying synthesis. The annotations and synthetic images are released as reusable resources on the Hugging Face Hub under permissive licences.
Chinese Translation
自动化支票处理需要在任何识别步骤之前定位关键字段（日期、法定金额、IFSC 代码、账户号码、签名和收款人姓名）。据我们所知，IDRBT 支票图像数据集是唯一的公共印度银行支票集合，但它没有字段注释，并且没有明确的许可声明，因此其再分发条款不清晰。我们解决了这两个限制。首先，我们为数据集中所有 112 张支票发布了六个字段的边界框注释，注释以仅分发的方式提供，并与原始文件名关联，以遵守 IDRBT 的再分发条款。其次，我们发布了 295 张完全可再分发的合成支票图像，这些图像是通过剪切-粘贴管道生成的，该管道将来自真实支票的注释字段区域合成到内容已被擦除的特定银行画布模板上；由于补丁是以其源坐标粘贴的，因此注释保持不变。第三，我们提供了一个 ResNet-50 直接回归基线模型，该模型在单次前向传递中预测所有六个字段，并利用它对合成数据进行控制测试。测试结果令人警醒：由于支票布局是固定的，一个简单预测每个字段均值训练框的无学习基线已经达到了 0.691 的平均交并比（mean IoU）和 80% 的准确率（IoU >= 0.5），而在考虑种子方差和训练计算后，剪切-粘贴合成数据在可测量的改进上并没有超过单独的真实数据（一个等计算的真实模型在每个汇总指标上与合成增强模型相匹配或超越）。我们完整报告这一负面结果，因为它警示我们不要假设仅通过外观增强有助于固定布局文档，而是指向布局变化的合成。注释和合成图像作为可重用资源在 Hugging Face Hub 上以宽松许可发布。
cs.CV / 7 / 2606.20684
Shear-Free Viewport Magnification for 360-Degree via Spherical Mobius Boosts
基于球面莫比乌斯提升的无剪切视口放大技术用于360度成像
Abstract
Viewport-adaptive 360-degree imaging seeks to allocate a fixed sampling budget to the region a viewer is likely to observe. Existing view-biased projections increase viewport resolution through non-conformal warps, which can introduce anisotropic stretching and shear. We formulate spherical Mobius boosts as exact conformal maps for fixed-budget viewport magnification. The continuous spherical warp has quasiconformal dilatation K = 1, reallocating samples toward a target direction while preserving local angles. On a SUN360 saliency audit with 72 panoramas and 216 paired viewport targets, C1 Mobius boosting improves viewport PSNR over optimized offset cubemap on all paired cases, with case-level median gain +3.26 dB, image-level median gain +3.23 dB, and panorama-level bootstrap 95% CI [+3.15, +3.33] dB. Pareto analysis shows that this is not a free global-quality improvement: C1 trades full-sphere WS-PSNR for shear-free viewport fidelity. Prediction-error and filtering studies identify the operating envelope: strong boosts are useful for accurately targeted viewports, while large target uncertainty calls for weaker boosts or fallback. These results position Mobius boosting as a geometric primitive for prediction-conditioned foveated 360-degree resampling rather than a universal encode-once layout.
Chinese Translation
视口自适应的360度成像旨在将固定的采样预算分配给观察者可能关注的区域。现有的视图偏向投影通过非共形变换提高视口分辨率，这可能引入各向异性的拉伸和剪切。我们将球面莫比乌斯提升形式化为固定预算视口放大的精确共形映射。连续的球面变换具有准共形膨胀K = 1，能够在保持局部角度的同时，将样本重新分配到目标方向。在包含72个全景图和216个配对视口目标的SUN360显著性审计中，C1莫比乌斯提升在所有配对案例中相较于优化的偏移立方体图提高了视口的PSNR，案例级中位增益为+3.26 dB，图像级中位增益为+3.23 dB，全景级自助法95%置信区间为[+3.15, +3.33] dB。帕累托分析表明，这并不是一种免费的全局质量提升：C1在无剪切视口保真度上以全球WS-PSNR为代价进行权衡。预测误差和滤波研究确定了操作范围：强提升对于准确目标视口是有用的，而较大的目标不确定性则需要较弱的提升或退回。这些结果将莫比乌斯提升定位为预测条件下的聚焦360度重采样的几何原语，而不是一种通用的单次编码布局。
cs.CV / 8 / 2606.20687
ARGUSTRACK: A Multi-View Annotation System for Multi-Object Tracking
ARGUSTRACK：一种多视角注释系统用于多目标跟踪
Abstract
Multi-Camera Multi-Target (MCMT) tracking has emerged as a critical capability for applications ranging from autonomous driving to animal behavior monitoring. While recent advances have yielded sophisticated tracking algorithms, the availability of annotated multi-view data remains a significant bottleneck. Existing annotation tools predominantly support single-camera workflows or rely on LiDAR sensors, making cross-view labeling tedious and impractical for camera-only setups. We present ARGUS-TRACK, a multi-camera annotation system that addresses these limitations by enabling annotators to work directly on a bird's-eye-view (BEV) plane. Given calibrated camera parameters, a single ground-plane annotation is automatically projected into 2D bounding boxes across all relevant views, inherently ensuring identity consistency without manual cross-view alignment. To further accelerate the labeling process, ARGUSTRACK incorporates two complementary mechanisms: a Temporal Aware module that propagates annotations from preceding frames to initialize new ones, requiring only minor positional adjustments; and a Multi-camera Semi-annotation module that leverages off-the-shelf 2D detectors combined with foot-point estimation to automatically generate candidate BEV positions for annotator verification. We evaluate ARGUSTRACK through a pilot study on multi-camera broiler tracking and demonstrate that it substantially reduces annotation time compared to conventional single-camera labeling workflows.
Chinese Translation
多摄像头多目标（MCMT）跟踪已成为从自动驾驶到动物行为监测等应用的重要能力。尽管近期的进展已经产生了复杂的跟踪算法，但可用的注释多视角数据仍然是一个显著的瓶颈。现有的注释工具主要支持单摄像头工作流程或依赖于激光雷达（LiDAR）传感器，使得跨视角标注在仅使用摄像头的设置中显得繁琐且不切实际。我们提出了ARGUSTRACK，一种多摄像头注释系统，旨在通过使注释者能够直接在鸟瞰图（BEV）平面上工作来解决这些限制。给定标定的摄像头参数，单一的地面平面注释会自动投影到所有相关视角的2D边界框中，从而在没有手动跨视角对齐的情况下，固有地确保身份一致性。为了进一步加速标注过程，ARGUSTRACK结合了两个互补机制：一个时间感知模块，该模块从前一帧传播注释以初始化新的注释，仅需进行微小的位置信息调整；以及一个多摄像头半自动注释模块，该模块利用现成的2D检测器结合足点估计，自动生成候选的BEV位置供注释者验证。我们通过对多摄像头肉鸡跟踪的初步研究评估了ARGUSTRACK，并证明与传统的单摄像头标注工作流程相比，它显著减少了注释时间。
cs.CV / 9 / 2606.20689
NeoJaundice-AI: Smartphone-Based Neonatal Jaundice Detection Using Dual-Input Deep Learning and Synthetic Augmentation
NeoJaundice-AI：基于智能手机的双输入深度学习和合成增强的新生儿黄疸检测
Abstract
Neonatal jaundice (hyperbilirubinemia) is one of the most common conditions affecting newborns worldwide, with India alone recording roughly 15 million cases per year. Early detection is critical, yet standard diagnosis requires blood tests that are often impractical in rural clinics where laboratory facilities are limited. This paper presents NeoJaundice-AI, a smartphone-based screening system that uses photographs of a baby's skin and sclera (eye white) to estimate jaundice severity and predict serum bilirubin levels in under three seconds without requiring internet connectivity. The proposed system is built on a dual-branch EfficientNet-B0 architecture that independently processes skin and sclera images. Deep features are fused with handcrafted YCbCr color statistics to jointly perform four-class severity classification and continuous bilirubin regression. A key contribution is a synthetic jaundice generation method that simulates bilirubin-induced yellowing through controlled YCbCr channel modifications on normal neonatal skin images. This approach addresses data scarcity, particularly for severe jaundice cases and darker Indian skin tones (Fitzpatrick Types IV to VI). In addition, a skin-tone normalization module improves prediction consistency across diverse neonatal complexions. Experimental results demonstrate an overall classification accuracy of 91.8 percent, a clinical sensitivity of 93.5 percent, and a bilirubin mean absolute error of 1.4 mg/dL. After INT8 quantization and ONNX conversion, the model size is reduced to 8.3 MB while maintaining inference times below three seconds on standard Android devices. To the best of our knowledge, this is the first India-focused neonatal jaundice AI system that combines multimodal image fusion, skin-tone adaptation, synthetic data augmentation, and fully offline mobile deployment within a single framework.
Chinese Translation
新生儿黄疸（高胆红素血症）是全球影响新生儿的最常见疾病之一，印度每年记录约1500万例。早期检测至关重要，但标准诊断需要血液检测，这在实验室设施有限的农村诊所中往往不切实际。本文提出了NeoJaundice-AI，一种基于智能手机的筛查系统，利用婴儿皮肤和巩膜（眼白）的照片在不到三秒的时间内估计黄疸严重程度并预测血清胆红素水平，无需互联网连接。该系统基于双分支的EfficientNet-B0架构，独立处理皮肤和巩膜图像。深度特征与手工制作的YCbCr颜色统计数据融合，以共同执行四类严重程度分类和连续胆红素回归。一个关键贡献是合成黄疸生成方法，通过对正常新生儿皮肤图像进行控制的YCbCr通道修改，模拟胆红素引起的发黄。这种方法解决了数据稀缺的问题，特别是对于重度黄疸病例和较深肤色的印度皮肤（Fitzpatrick类型IV至VI）。此外，肤色归一化模块提高了不同新生儿肤色之间的预测一致性。实验结果表明，整体分类准确率为91.8%，临床敏感性为93.5%，胆红素平均绝对误差为1.4 mg/dL。在进行INT8量化和ONNX转换后，模型大小减少至8.3 MB，同时在标准Android设备上的推理时间保持在三秒以下。据我们所知，这是第一个聚焦于印度的新生儿黄疸AI系统，结合了多模态图像融合、肤色适应、合成数据增强和完全离线移动部署于一个框架内。
cs.CV / 10 / 2606.20693
Spatio-Temporal Wildfire Spread Prediction in Canada using a Video Swin-Hybrid-U-Net and Satellite Imagery
基于视频Swin-Hybrid-U-Net和卫星影像的加拿大时空野火传播预测
Abstract
Background: Wildfires in Canada present increasing threats to ecosystems, communities, and infrastructure, demanding accurate forecasting tools to aid mitigation efforts. Existing models often lack scalability or fail to capture temporal dynamics effectively. Aims: This study aims to develop a deep learning framework tailored to Canadian wildfire spread prediction that captures spatio-temporal patterns in environmental data. Methods: We propose a U-Net architecture integrating a Video Swin Transformer encoder with a convolutional decoder to model three-day sequences of meteorological and environmental variables. Data are exclusively sourced from public repositories via Google Earth Engine, ensuring transparency and scalability. The model is trained and tested on a curated dataset of major Canadian wildfire events from 2014 to 2023. Key results: Our approach achieves strong predictive performance by effectively leveraging spatio-temporal attention to forecast next-day fire incidence maps. Conclusions: The model successfully captures complex wildfire dynamics unique to Canada's landscape and temporal variability. Implications: This framework paves the way for advanced spatio-temporal wildfire forecasting research and operational applications using publicly accessible datasets.
Chinese Translation
背景：加拿大的野火对生态系统、社区和基础设施构成日益严重的威胁，迫切需要准确的预测工具以支持减灾工作。现有模型往往缺乏可扩展性或无法有效捕捉时间动态。目的：本研究旨在开发一个深度学习框架，专门用于加拿大野火传播预测，以捕捉环境数据中的时空模式。方法：我们提出了一种U-Net架构，集成了视频Swin Transformer编码器和卷积解码器，以建模气象和环境变量的三天序列。数据完全来源于Google Earth Engine的公共存储库，确保了透明性和可扩展性。该模型在2014年至2023年间的主要加拿大野火事件的精选数据集上进行训练和测试。关键结果：我们的方法通过有效利用时空注意力，成功实现了强大的预测性能，能够预测次日火灾发生图。结论：该模型成功捕捉了独特于加拿大地形和时间变化的复杂野火动态。影响：该框架为利用公共可获取数据集进行先进的时空野火预测研究和操作应用铺平了道路。
cs.CV / 11 / 2606.20697
AEF-Econ: Toward Plug-and-Play Socioeconomic Foundation Embeddings from AlphaEarth for Urban Remote Sensing
AEF-Econ：基于AlphaEarth的城市遥感即插即用社会经济基础嵌入
Abstract
AlphaEarth Foundations (AEF) unify global remote sensing foundation embeddings through multimodal self-supervised learning, but their pretraining focuses on physical land-surface signals, limiting plug-and-play use in socioeconomic tasks. We integrate seven heterogeneous data streams across 36 Chinese cities over eight years - AEF embeddings, population, nighttime lights, remote sensing indices, points of interest (POIs), urban morphology, and cross-lingual text - and construct CHN-Econ, a socioeconomic benchmark with 16 labels in three categories. We conduct 31 controlled experiments along five axes: fusion architecture, self-supervised objective, text integration, embedding dimensionality, and normalization. Used alone as a linear probe, AEF achieves R2 values of only 0.301 for cross-region and 0.160 for cross-tier evaluation. The five-axis ablated backbone improves these scores to 0.832 and 0.671, respectively, but reveals that low-dimensional semantic streams are consistently suppressed by high-dimensional streams under shared reconstruction. To address this bottleneck, we propose Capacity-Adaptive Reconstruction (CAR), replacing shared reconstruction with per-stream decoders and stream-level losses to mitigate inter-stream capacity competition. CAR further raises cross-region and cross-tier R2 to 0.848 and 0.693, and restores collapsed labels from negative R2 to a stable range. Using CAR, we infer 14.4 million pixels across 36 cities and eight years and release AEF-Econ, including 128d and 64d compressed versions. Self-diagnostics and case studies show that AEF-Econ captures cross-city hierarchies and intra-urban spatial organization under unsupervised settings, providing a socioeconomic remote sensing foundation embedding complementary to AEF physical embeddings.
Chinese Translation
AlphaEarth Foundations (AEF) 通过多模态自监督学习统一全球遥感基础嵌入，但其预训练主要集中于物理地表信号，限制了在社会经济任务中的即插即用使用。我们整合了来自36个中国城市八年间的七个异构数据流——AEF嵌入、人口、夜间灯光、遥感指数、兴趣点（POIs）、城市形态和跨语言文本——并构建了CHN-Econ，一个具有16个标签和三个类别的社会经济基准。我们沿五个维度进行了31个受控实验：融合架构、自监督目标、文本整合、嵌入维度和归一化。单独使用线性探针时，AEF在跨区域和跨层级评估中的R2值仅为0.301和0.160。经过五个维度消融的基础模型将这些分数提高到0.832和0.671，但揭示了在共享重建下，低维语义流始终被高维流抑制。为了解决这一瓶颈，我们提出了容量自适应重建（Capacity-Adaptive Reconstruction, CAR），用每个流的解码器和流级损失替代共享重建，以减轻流之间的容量竞争。CAR进一步将跨区域和跨层级的R2提高到0.848和0.693，并将负R2的崩溃标签恢复到稳定范围。使用CAR，我们在36个城市和八年中推断了1440万像素，并发布了AEF-Econ，包括128维和64维的压缩版本。自我诊断和案例研究表明，AEF-Econ在无监督设置下捕捉了跨城市层级和城市内部空间组织，为AEF物理嵌入提供了互补的社会经济遥感基础嵌入。
cs.CV / 12 / 2606.20702
Beyond Templates: Revisiting Zero-Shot Remote Sensing through Meta-Prompting
超越模板：通过元提示重访零样本遥感
Abstract
Vision-language models (VLMs) have sparked growing interest in zero-shot Earth Observation (EO) downstream tasks, with further gains enabled by remote-sensing-adapted models. We examine this setting across 17 VLM variants and 12 remote sensing (RS) datasets under Meta-Prompting for Visual Recognition (MPVR), and show that zero-shot performance remains highly sensitive to textual design choices, from the meta-prompts used to guide the LLM in generating class descriptions to the descriptions themselves. We explore why semantically rich LLM-generated class descriptions do not translate into consistent gains over simple domain-adapted CLIP-style descriptions. While LLM descriptions are more semantically expressive, they can also introduce noise in the text embedding space, reducing robustness in downstream tasks. We support this observation through a text log-likelihood analysis in the whitened CLIP feature space, comparing LLM-generated and template-based descriptions. Building on this finding, we study query embedding calibration and show that lightweight calibration of the query space consistently yields strong improvements in zero-shot classification and retrieval. Overall, our results provide practical insight into the trade-off between semantic richness and robustness, and identify embedding calibration as a simple and effective tool for improving zero-shot remote sensing performance.
Chinese Translation
视觉-语言模型（VLMs）在零样本地球观测（EO）下游任务中引发了越来越多的关注，遥感适应模型进一步提升了这一领域的成果。我们在17种VLM变体和12个遥感（RS）数据集上，采用视觉识别的元提示（MPVR）方法进行研究，发现零样本性能对文本设计选择高度敏感，从用于指导大型语言模型（LLM）生成类别描述的元提示到描述本身。我们探讨了为何语义丰富的LLM生成的类别描述未能在简单的领域适应CLIP风格描述上实现一致的提升。尽管LLM描述在语义表达上更为丰富，但它们也可能在文本嵌入空间中引入噪声，从而降低下游任务的鲁棒性。我们通过在白化的CLIP特征空间中对LLM生成和基于模板的描述进行文本对数似然分析来支持这一观察。基于这一发现，我们研究了查询嵌入的校准，表明轻量级的查询空间校准在零样本分类和检索中始终带来了显著的改善。总体而言，我们的结果为语义丰富性与鲁棒性之间的权衡提供了实用的见解，并将嵌入校准识别为提升零样本遥感性能的简单有效工具。
cs.CV / 13 / 2606.20703
Robust Image-Driven Phenotyping of Ovarian Tumor Cells using Optimized Dynamic Features in Hyperbolic Channels
基于图像的卵巢肿瘤细胞稳健表型分析：利用超曲面通道中的优化动态特征
Abstract
Label-free, image-based cellular mechanophenotyping in microfluidic devices provides a high-throughput method for single-cell profiling. However, while complex microchannels (e.g., hyperbolic geometries) reveal transient deformation dynamics under continuous extensional stress, the resulting high-dimensional feature spaces are highly susceptible to hydrodynamic artifacts. Flow rate variations often distort discriminative boundaries, linking feature distributions to fluid conditions rather than intrinsic biology. To overcome this, we introduce a stability-guided analytical framework that decouples flow-induced noise from authentic mechanobiological signatures. We tracked the morphodynamic, kinematic, and intracellular optical-density trajectories of healthy and malignant ovarian cells to build a 93-dimensional feature space. Using a cross-flow screening strategy based on structural consistency and statistical persistence, we isolated robust descriptors, creating task-adapted subsets (20 features for binary classification; 25 for cancer subtyping). Variance-attribution analysis confirmed the neutralization of flow-conditioned artifacts; notably, flow-associated variance in the primary principal component fell from 69.9% to 9.3% in the subtyping task. We also found that macroscopic binary discrimination depends on bulk kinematic transitions, while clonal subtyping requires localized intracellular optical heterogeneity. These optimized subsets maintained diagnostic fidelity across multiple machine learning architectures and restricted sampling conditions. This framework establishes a robust, flow-independent foundation for continuous dynamic phenotyping.
Chinese Translation
无标记的基于图像的细胞机械表型分析在微流体设备中提供了一种高通量的单细胞分析方法。然而，复杂的微通道（例如，超曲面几何形状）在持续的拉伸应力下揭示的瞬态变形动态，使得生成的高维特征空间对流体动力学伪影高度敏感。流速变化常常扭曲判别边界，将特征分布与流体条件联系在一起，而非内在生物学。为了解决这个问题，我们提出了一种稳定性引导的分析框架，能够将流动引起的噪声与真实的机械生物学特征解耦。我们追踪了健康和恶性卵巢细胞的形态动力学、运动学和细胞内光密度轨迹，以构建93维特征空间。通过基于结构一致性和统计持久性的交叉流筛选策略，我们隔离了稳健的描述符，创建了适应任务的子集（20个特征用于二分类；25个特征用于癌症亚型分类）。方差归因分析确认了流动条件伪影的中和；值得注意的是，在亚型分类任务中，主要主成分中的流动相关方差从69.9%降至9.3%。我们还发现，宏观二元判别依赖于整体运动学转变，而克隆亚型分类则需要局部细胞内光学异质性。这些优化的子集在多种机器学习架构和限制采样条件下保持了诊断的可靠性。该框架为连续动态表型分析建立了一个稳健的、与流动无关的基础。
cs.CV / 14 / 2606.20705
MotionPyramid: Hierarchical Motion Representation and Residual Interfaces
运动金字塔：分层运动表示与残差接口
Abstract
We ask whether the representational hierarchy seen in perception, from local primitives such as edges to higher level structures such as parts and objects, can be established for motion. In humanoid control, low level actions specify immediate motor commands, while meaningful behavior is organized over longer temporal scales, including contacts, gait fragments, balance recovery, reaching, and whole body skills. We introduce MotionPyramid, a hierarchical action representation that learns such structure from motion data. Starting from a motion tracking teacher, it trains a recursive stack of latent decoders: low level latents decode to immediate full body motor commands, while higher level latents unfold through lower levels into temporally extended motion programs. After pretraining, the hierarchy is frozen and reused by downstream reinforcement learning policies as a family of action interfaces at different control resolutions. Experiments show the learned levels form a motion hierarchy: coarser interfaces improve early learning and motion regularity by constraining exploration to structured segments, while finer interfaces preserve feedback control and final task precision. Representation probes show the hierarchy supports traversal, interpolation, transition, and qualitative composition, exposing editable control handles across temporal scales. Finally, we introduce Residual Interfaces, letting a downstream policy maintain coarse, segment level, and frame level residual commands through the frozen hierarchy. Analogous to residual or skip connections in deep networks, this allows coarse motion programs and fine residual corrections to coexist within one controller. MotionPyramid shows that motion, like perception, can be organized into a reusable multi level representation, providing structured abstraction without sacrificing controllability.
Chinese Translation
我们探讨在运动中是否可以建立感知中所见的表示层次结构，从局部原始元素（如边缘）到更高层次的结构（如部件和物体）。在类人控制中，低层次的动作指定了即时的运动指令，而有意义的行为则在更长的时间尺度上组织，包括接触、步态片段、平衡恢复、伸展和全身技能。我们引入了运动金字塔（MotionPyramid），一种从运动数据中学习这种结构的分层动作表示。从运动跟踪教师开始，它训练一个递归的潜在解码器堆栈：低层次的潜在变量解码为即时的全身运动指令，而高层次的潜在变量通过低层次展开为时间延续的运动程序。在预训练后，该层次结构被冻结，并被下游强化学习策略作为不同控制分辨率下的一系列动作接口重用。实验表明，学习到的层次形成了运动层次结构：粗糙的接口通过将探索限制在结构化片段中，改善了早期学习和运动规律性，而细致的接口则保持了反馈控制和最终任务的精确性。表示探针显示该层次支持遍历、插值、过渡和定性组合，揭示了跨时间尺度的可编辑控制手柄。最后，我们引入了残差接口（Residual Interfaces），使下游策略能够通过冻结的层次结构维持粗糙的片段级和帧级残差指令。类似于深度网络中的残差或跳跃连接，这使得粗糙的运动程序和细致的残差修正能够在一个控制器中共存。运动金字塔表明，运动与感知一样，可以组织成可重用的多层表示，提供结构化的抽象而不牺牲可控性。
cs.CV / 15 / 2606.20707
GEOPHYS: The Geometry of Physical Plausibility
GEOPHYS：物理合理性的几何特征
Abstract
While humans can identify physically implausible events within milliseconds, machine learning approaches addressing the same problem are extremely slow and expensive. They either rely on external multimodal-LLM judges or require ad-hoc modifications to the training procedure. In this work, we argue that indicators of physical plausibility are implicitly captured by five geometric properties of the per-frame embeddings produced by frozen image encoders. In aggregate, we call them GEOPHYS. First, we show that these signals correlate with human EEG responses to two forms of object-permanence violations. Second, GEOPHYS robustly discriminates physically implausible videos from realistic ones, achieving state-of-the-art physics-violation detection: 98.3% on LikePhys and 93.3% on IntPhys2, whereas V-JEPA 2, GPT-4o, Gemini, and twelve modern video diffusion models perform near chance. Third, used as a best-of-N verifier for physical alignment during video generation, GEOPHYS lifts MAGI-1 24B from 50.01% to 64.50% on PhysicsIQ at 1.5x lower wall-clock and 4.65x lower memory than the V-JEPA 2 world-model verifier. Ultimately, GEOPHYS demonstrates that physical plausibility in videos can be assessed by leveraging the emergent geometric properties of temporal features extracted from image encoders.
Chinese Translation
虽然人类能够在毫秒内识别出物理上不合理的事件，但针对同一问题的机器学习方法却极其缓慢且昂贵。这些方法要么依赖于外部的多模态大语言模型（multimodal-LLM）评判者，要么需要对训练过程进行临时修改。在本研究中，我们认为物理合理性的指示符隐含地通过冻结图像编码器生成的每帧嵌入的五个几何特性来捕捉。我们将这些特性统称为GEOPHYS。首先，我们展示了这些信号与人类脑电图（EEG）对两种物体持久性违反形式的反应之间的相关性。其次，GEOPHYS能够稳健地区分物理上不合理的视频与真实视频，达到了最先进的物理违反检测效果：在LikePhys上达到98.3%，在IntPhys2上达到93.3%，而V-JEPA 2、GPT-4o、Gemini以及十二个现代视频扩散模型的表现接近随机水平。第三，作为视频生成过程中物理一致性的最佳-N验证器，GEOPHYS将MAGI-1 24B在PhysicsIQ上的表现从50.01%提升至64.50%，同时在时间消耗上比V-JEPA 2世界模型验证器低1.5倍，内存使用低4.65倍。最终，GEOPHYS证明了可以通过利用从图像编码器提取的时间特征的突现几何特性来评估视频中的物理合理性。
cs.CV / 16 / 2606.20709
TeleStyle V2: Beyond Content-Preserving Style Transfer with Self-Distillation and Distribution-Matching-Distillation
TeleStyle V2：超越内容保留风格迁移的自我蒸馏与分布匹配蒸馏
Abstract
Given a content reference and a style reference, content-preserving style transfer requires the model to generate stylized outputs with content and style consistency. We introduced TeleStyle V1 to tackle this problem. However, TeleStyle V1 is trained with photorealistic content reference and artistic style reference, which makes it incapable to cope with artistic content reference and realistic style reference in most cases. In this paper, we designed a Self-Distillation data synthesis strategy to construct such triplets from TeleStyle V1. Trained with such self-distilled triplets, our TeleStyle V2 supports Content-Style references in the forms of Realistic-and-Realistic (RnR), Realistic-and-Stylized (RnS), Stylized-and-Realistic (SnR), Stylized-and-Stylized (SnS). In addition, we found Distribution Matching Distillation could preserve the general text-guided image editing capability of the foundation model and fix the content consistency degradation caused by SFT process. Through quantitative evaluations, our TeleStyleV2-QIE-2509-DMD performs at least on par with Qwen-Image-Edit-2509-DMD, demonstrating strong general image editing skills beyond content-preserving style transfer. We observed the content/style reference order confusion problem in TeleStyle V1 and further introduced prompt enhancer to solve it. TeleStyle V2 uses Qwen-Image-Edit's VLM encoder, Qwen2.5-VL-7B, to generate content prompt and style prompt for free. TeleStyle V2 could achieve comparable style transfer performance with state-of-the-art commercial model, gemini-3-pro-image-preview.
Chinese Translation
给定内容参考和风格参考，内容保留风格迁移要求模型生成在内容和风格上都保持一致的风格化输出。我们引入了TeleStyle V1来解决这一问题。然而，TeleStyle V1是使用照片真实的内容参考和艺术风格参考进行训练的，这使得它在大多数情况下无法处理艺术内容参考和真实风格参考。在本文中，我们设计了一种自我蒸馏数据合成策略，从TeleStyle V1构建这样的三元组。通过使用这种自我蒸馏的三元组进行训练，我们的TeleStyle V2支持以真实-真实（Realistic-and-Realistic, RnR）、真实-风格化（Realistic-and-Stylized, RnS）、风格化-真实（Stylized-and-Realistic, SnR）、风格化-风格化（Stylized-and-Stylized, SnS）等形式的内容-风格参考。此外，我们发现分布匹配蒸馏能够保留基础模型的通用文本引导图像编辑能力，并修复由SFT过程引起的内容一致性下降。通过定量评估，我们的TeleStyleV2-QIE-2509-DMD的表现至少与Qwen-Image-Edit-2509-DMD相当，展示了超越内容保留风格迁移的强大图像编辑能力。我们观察到TeleStyle V1中的内容/风格参考顺序混淆问题，并进一步引入了提示增强器来解决该问题。TeleStyle V2使用Qwen-Image-Edit的VLM编码器Qwen2.5-VL-7B，免费生成内容提示和风格提示。TeleStyle V2能够实现与最先进的商业模型gemini-3-pro-image-preview相当的风格迁移性能。
cs.CV / 17 / 2606.20711
Video2Code: Generating Interactive Webpages from UI Videos via Action-Aware Revisit
Video2Code：通过动作感知重访从UI视频生成交互式网页
Abstract
UI videos provide a natural input for generating interactive webpages, as they capture both webpage appearance and action-triggered state transitions. However, directly applying video-capable vision-language models to this task remains insufficient. Existing models typically rely on sparse sampling or compressed temporal representations, which may miss short action boundaries and break the state-action-state transitions needed to implement webpage behavior. We formulate UI video-to-code generation as executable state-transition recovery from interaction videos, and identify this failure mode as state-transition misalignment. We introduce Video2Code, an action-aware video-to-code approach for recovering executable UI state transitions. Rather than allocating the visual budget uniformly across the video, Video2Code first performs coarse video understanding to locate action-critical regions, then invokes a temporal clipping tool to revisit these regions at higher temporal resolution before generating HTML/CSS/JavaScript code. We instantiate Video2Code with action-aligned video-code supervision and evaluate it under both visual and functional criteria. Experiments show that Video2Code substantially strengthens the underlying open-source model for UI video-to-code generation, improving functional correctness over direct video observation, especially on dense multi-step interactions.
Chinese Translation
UI视频为生成交互式网页提供了一种自然的输入方式，因为它们捕捉了网页的外观和由动作触发的状态转变。然而，直接将视频能力的视觉-语言模型应用于此任务仍然不足。现有模型通常依赖稀疏采样或压缩的时间表示，这可能会错过短暂的动作边界，并破坏实现网页行为所需的状态-动作-状态转变。我们将UI视频到代码生成形式化为从交互视频中可执行的状态转变恢复，并将这种失败模式识别为状态转变不对齐。我们提出了Video2Code，这是一种动作感知的视频到代码方法，用于恢复可执行的UI状态转变。Video2Code并不是在整个视频中均匀分配视觉预算，而是首先进行粗略的视频理解，以定位动作关键区域，然后调用时间剪辑工具以更高的时间分辨率重访这些区域，最后生成HTML/CSS/JavaScript代码。我们通过动作对齐的视频-代码监督实例化Video2Code，并在视觉和功能标准下进行评估。实验表明，Video2Code显著增强了基础开源模型在UI视频到代码生成中的能力，尤其在密集的多步骤交互中，功能正确性相较于直接视频观察有了显著提升。
cs.CV / 18 / 2606.20715
CDER-SME: A Cross-Device Event-RGB Micro-Expression Dataset under Multi-Level Stress Induction
CDER-SME：多层次压力诱导下的跨设备事件-RGB微表情数据集
Abstract
Micro-expression recognition (MER) in realistic scenarios demands high temporal sensitivity and ecological validity, yet existing benchmarks are largely constrained to laboratory-controlled settings and rigid hardware-coupled sensing. We introduce CDER-SME, a cross-device Event-RGB dataset collected under a multi-level stress induction framework (cognitive and social) to elicit spontaneous emotional leakage. To enable reproducible acquisition with independent, decoupled sensors, we provide a hardware-agnostic alignment pipeline for temporal synchronization and landmark-guided spatial registration. CDER-SME adopts a three-tier structure with 92 subjects and 1,963 expert-annotated samples (Action Units and emotions), including 790 Event-RGB pairs and 210 high-fidelity aligned pairs. We further report a reproducible multimodal baseline, where cross-modal fusion improves performance over single-modality counterparts, supporting the complementarity of event dynamics and RGB cues. By removing the need for coaxial calibration, CDER-SME offers a practical benchmark for cross-device alignment and deployable Event-RGB MER in real-world affective intelligence.
Chinese Translation
在现实场景中，微表情识别（MER）要求具备高时间敏感性和生态有效性，但现有基准大多局限于实验室控制环境和刚性硬件耦合传感。我们引入了CDER-SME，这是一个在多层次压力诱导框架（认知和社交）下收集的跨设备事件-RGB数据集，旨在引发自发的情感泄漏。为了实现独立、解耦传感器的可重复获取，我们提供了一种硬件无关的对齐流程，用于时间同步和基于地标的空间注册。CDER-SME采用三层结构，包含92名受试者和1,963个专家标注样本（动作单元和情感），其中包括790个事件-RGB对和210个高保真对齐对。我们进一步报告了一个可重复的多模态基线，其中跨模态融合在性能上优于单一模态，支持事件动态与RGB线索的互补性。通过消除同轴校准的需求，CDER-SME为跨设备对齐和可部署的事件-RGB微表情识别提供了一个实用的基准，适用于现实世界的情感智能。
cs.CV / 19 / 2606.20717
MIRAGE: Stealthy Visual Prompt Injection for Vulnerability Detection in Web Agents
MIRAGE：用于网络代理漏洞检测的隐秘视觉提示注入
Abstract
Multimodal Large Language Model (MLLM)-based web agents provide practical, high-precision solutions for visual browser automation; however, they inherently expand the attack surface, introducing novel vision-based vulnerabilities. Existing adversarial evaluations targeting these agents frequently rely on permissive threat models and visually conspicuous artifacts. In this paper, we investigate a constrained vulnerability detection setting: a trusted web platform where the evaluator acts solely as an unprivileged third party, such as a merchant or advertiser, controlling only a semantically legitimate, spatially constrained region, such as an ad slot, a sponsored card, or a localized widget. Operating under these realistic constraints, we propose MIRAGE, a novel visual indirect prompt injection framework for targeted next-action hijacking. Our approach leverages diffusion models to generate perceptually benign adversarial images strictly confined to the attacker-controlled boundaries permitted by the trusted service provider. To maximize attack efficacy within such a restrictive setting, we introduce a robust optimization technique combining curvature-aware adversarial diffusion guidance with sparse, dark-pixel residual perturbations. Comprehensive evaluations against prominent MLLM web agent frameworks, specifically SeeAct and OpenClaw, empirically demonstrate the potency, realism, and stealth of our proposed MIRAGE.
Chinese Translation
基于多模态大型语言模型（MLLM）的网络代理为视觉浏览器自动化提供了实用且高精度的解决方案；然而，它们固有地扩大了攻击面，引入了新型基于视觉的漏洞。现有针对这些代理的对抗性评估通常依赖于宽松的威胁模型和视觉上显著的伪影。在本文中，我们研究了一种受限的漏洞检测设置：一个可信的网络平台，在该平台上，评估者仅作为一个无特权的第三方（如商家或广告商）行事，仅控制一个语义上合法、空间上受限的区域，例如广告位、赞助卡或本地化小部件。在这些现实约束下，我们提出了MIRAGE，一种用于目标下一步操作劫持的新型视觉间接提示注入框架。我们的方法利用扩散模型生成严格限制在攻击者控制边界内的感知上无害的对抗性图像，这些边界由可信服务提供商允许。为了在如此限制的环境中最大化攻击效果，我们引入了一种强健的优化技术，结合了曲率感知的对抗扩散引导与稀疏的暗像素残差扰动。针对主要的MLLM网络代理框架（特别是SeeAct和OpenClaw）的全面评估，实证证明了我们提出的MIRAGE的有效性、真实性和隐秘性。
cs.CV / 20 / 2606.20723
Evaluation of Medical Vision Language Models HuluMed and MedGemma, and general purpose chatbots Gemma 3, ChatGPT Plus, and Claude Pro on real previously unseen wound images
对医疗视觉语言模型HuluMed和MedGemma，以及通用聊天机器人Gemma 3、ChatGPT Plus和Claude Pro在真实未见伤口图像上的评估
Abstract
Chronic wound assessment remains a clinically challenging task that requires accurate interpretation of wound morphology, tissue composition, vascular characteristics, and infection risk. Recent advances in Vision-Language Models (VLMs) have introduced the possibility of automated multimodal wound analysis through image understanding combined with clinical reasoning. This study evaluates the performance of several general-purpose and medically specialized open-source and proprietary VLMs for clinical wound assessment using an expanded, curated dataset of 20 clinically diverse wounds spanning vascular, surgical, ischemic, venous, lymphedema, and amputation-related etiologies. Six VLMs were evaluated using a structured twelve-question clinical framework covering wound classification, infection risk, vascular intervention recommendations, debridement urgency, wound therapy selection, and advanced management planning. Across 20 wound cases and 240 clinician-graded wound-analysis decisions, ChatGPT achieved the highest overall performance with 174/240 correct responses (72.50%), followed by Claude with 149/240 (62.08%). Among the open-source and medically specialized models, HuluMed achieved the strongest performance with 96/240 correct responses (40.00%), followed by Gemma 3 (81/240, 33.75%), MedGemma 4B (62/240, 25.83%), and MedGemma 27B (42/240, 17.50%). The findings suggest that frontier general-purpose multimodal systems currently demonstrate substantially stronger wound-analysis performance than medically specialized alternatives, highlighting the continued importance of broad multimodal reasoning capabilities alongside domain-specific medical knowledge. Although current VLMs demonstrate promising potential for clinical decision support, substantial limitations remain in advanced wound-management reasoning, procedural planning, and autonomous clinical reliability.
Chinese Translation
慢性伤口评估仍然是一项临床挑战性任务，要求准确解读伤口形态、组织成分、血管特征和感染风险。最近在视觉-语言模型（VLMs）方面的进展引入了通过图像理解结合临床推理进行自动化多模态伤口分析的可能性。本研究评估了几种通用和医学专用的开源及专有VLM在临床伤口评估中的表现，使用了一个扩展的、经过策划的数据集，包含20个临床多样化的伤口，涵盖血管、外科、缺血、静脉、淋巴水肿和截肢相关病因。使用一个结构化的十二个问题的临床框架评估了六个VLM，涵盖伤口分类、感染风险、血管干预建议、清创紧迫性、伤口治疗选择和高级管理规划。在20个伤口案例和240个临床评分的伤口分析决策中，ChatGPT以174/240的正确回答（72.50%）获得了最高的整体表现，其次是Claude，149/240（62.08%）。在开源和医学专用模型中，HuluMed以96/240的正确回答（40.00%）表现最佳，其次是Gemma 3（81/240，33.75%）、MedGemma 4B（62/240，25.83%）和MedGemma 27B（42/240，17.50%）。研究结果表明，前沿的通用多模态系统目前在伤口分析表现上明显优于医学专用替代品，突显了广泛的多模态推理能力与领域特定医学知识并存的重要性。尽管当前的VLM在临床决策支持方面显示出良好的潜力，但在高级伤口管理推理、程序规划和自主临床可靠性方面仍存在重大局限。
cs.CV / 21 / 2606.20725
D2HDMap: Non-visible Driveline Map Prior for Online Vectorized HD Map Prediction
D2HDMap：用于在线矢量化高清地图预测的非可见驱动线地图先验
Abstract
Accurate, up-to-date representations of road structures are critical for the safe operation of autonomous vehicles. Existing systems rely either on costly, maintenance-heavy high-definition (HD) maps which compromise safety when outdated, or purely sensor-based online mapping which struggles with long-range reliability and occlusion. Systems incorporating map prior information into online mapping seek to overcome drawbacks of both approaches by combining them in some way. We propose 'Driveline To HD Map' (D2HDMap), an online mapping system that injects a lightweight, non-visible driveline prior to guide the estimation of visible road structures such as lane dividers, road boundaries and crosswalks. This prior incurs less effort to create and update compared to full HD map priors used in other approaches. We also show that training with such a prior can improve generalization at inference time when no prior is available. Ablation studies conducted on the nuScenes and Argoverse 2 dataset demonstrate that models trained using a driveline prior largely retain performance even when priors are not available. On a geographically disjoint split, D2HDMap achieves 44.8 mAP, surpassing recent state-of-the-art. Additionally, noise-aware training substantially increases robustness to realistic localization error.
Chinese Translation
准确、最新的道路结构表示对于自动驾驶车辆的安全运行至关重要。现有系统依赖于昂贵且维护繁重的高清（HD）地图，这在过时时会影响安全，或者完全依赖于传感器的在线映射，这在长距离可靠性和遮挡方面存在困难。将地图先验信息融入在线映射的系统试图通过某种方式将这两种方法结合起来，以克服各自的缺点。我们提出了'Driveline To HD Map'（D2HDMap），这是一种在线映射系统，通过注入轻量级的非可见驱动线先验来指导可见道路结构的估计，例如车道分隔线、道路边界和人行横道。与其他方法中使用的完整HD地图先验相比，这种先验的创建和更新所需的工作量更少。我们还表明，使用这种先验进行训练可以在推理时改善泛化能力，即使在没有先验的情况下。对nuScenes和Argoverse 2数据集进行的消融研究表明，使用驱动线先验训练的模型即使在没有先验的情况下也能大幅保持性能。在地理上不相交的分割上，D2HDMap达到了44.8 mAP，超越了最近的最先进水平。此外，噪声感知训练显著增强了对现实定位误差的鲁棒性。
cs.CV / 22 / 2606.20726
How Well Can Your Video Model Remember? Measuring Memory-Budget Trade-offs in Long Video Understanding
您的视频模型能记住多少？在长视频理解中衡量记忆预算的权衡
Abstract
We introduce a compact empirical model that quantifies how answer accuracy degrades as a function of frame budget B and temporal distance D in long video understanding -- analyzing performance when recalling content from D seconds in the past using a fraction B of total frames. Long-form models operate under strict budgets, yet no prior framework predicts how accuracy degrades as B shrinks and events recede. We fit a weighted least-squares model on ~155,000 binary predictions across ten models and three sampling strategies, deriving a law where logit-accuracy scales linearly in log-budget with a distance-dependent exponent that decays log-linearly with distance. This budget exponent \alpha(D) captures the marginal value of extra frames at distance D. The law achieves cell-level weighted R^2 = 0.05-0.75 across models. Notably, budget effectiveness at D = 1000 s differs by \approx 7.4\times between the best streaming and base models. STREAMINGVLM achieves \alpha(1000) = 1.26 (95% CI: [1.06, 1.58]), meaning a tenfold budget increase substantially improves long-distance accuracy, while the best Qwen3-VL base model reaches only \alpha(1000) = 0.17 (CI: [0.04, 0.34]). In accuracy space, a 10\times budget increase at D = 1000 s yields +29 percentage points for STREAMINGVLM versus +4 pp for the base model. Sampling strategies show model-dependent trade-offs: random sampling yields higher base sensitivity but steeper distance decay. We demonstrate how \alpha(D) enables principled budget allocation, including a model-ranking reversal at long distance, and propose it as a diagnostic metric for streaming video models.
Chinese Translation
我们引入了一种紧凑的实证模型，用于量化在长视频理解中，答案准确性如何随着帧预算 B 和时间距离 D 的变化而降低——分析在使用总帧数的一部分 B 回忆 D 秒前内容时的表现。长格式模型在严格的预算下运行，但之前没有框架能够预测当 B 减小时，准确性如何下降。我们在大约 155,000 个二元预测上拟合了一个加权最小二乘模型，涵盖十个模型和三种采样策略，推导出一个定律，其中 logit-准确性在 log-预算中线性缩放，且具有距离依赖的指数，该指数随距离以对数线性方式衰减。这个预算指数 α(D) 捕捉了在距离 D 处额外帧的边际价值。该定律在模型间实现了单元级加权 R^2 = 0.05-0.75。值得注意的是，在 D = 1000 秒时，最佳流媒体模型与基础模型之间的预算有效性差异约为 7.4 倍。STREAMINGVLM 实现了 α(1000) = 1.26 (95% CI: [1.06, 1.58])，这意味着十倍的预算增加显著提高了长距离准确性，而最佳的 Qwen3-VL 基础模型仅达到 α(1000) = 0.17 (CI: [0.04, 0.34])。在准确性空间中，在 D = 1000 秒时，STREAMINGVLM 的 10 倍预算增加带来了 +29 个百分点的提升，而基础模型仅为 +4 个百分点。采样策略显示出模型依赖的权衡：随机采样提供了更高的基础敏感性，但距离衰减更陡。我们展示了 α(D) 如何实现原则性的预算分配，包括在长距离下模型排名的逆转，并建议将其作为流媒体视频模型的诊断指标。
cs.CV / 23 / 2606.20728
VTOS: Learning to Orchestrate Vision Tools by Co-Searching Solutions and Observers
VTOS：通过共同搜索解决方案和观察者学习协调视觉工具
Abstract
Vision foundation tools such as open-vocabulary detectors, segmentation models, and post-processing operators are powerful building blocks for computer vision, but their effectiveness depends heavily on how they are orchestrated: which tools are used, in what order, with what parameters, and under what visual conditions. Existing visual-programming agents typically generate a fixed solution pipeline, making them brittle under dense objects, occlusion, small targets, and domain shift. We introduce VTOS (Vision Tools Orchestration Search), a framework for adaptive visual tool orchestration through joint solution--observer search. VTOS co-searches executable solution programs that compose vision tools such as Grounding DINO, SAM, NMS, and slice-and-detect, together with observer programs that diagnose candidate solutions, identify failure modes, and generate actionable feedback. These observations are accumulated in a shared VisionThoughts knowledge base to guide subsequent search. We evaluate VTOS through two case studies: dense object counting on LVIS-Count and zero-shot plant-disease segmentation on PlantSeg-OOD, which stress different orchestration challenges including threshold calibration, NMS, slicing, mask refinement, and domain generalization. Across both tasks, VTOS outperforms static tool pipelines and agentic visual-programming baselines, showing that co-searching solutions and observers is an effective strategy for adapting vision tools to challenging computer vision tasks.
Chinese Translation
视觉基础工具，如开放词汇检测器、分割模型和后处理操作符，是计算机视觉的强大构建模块，但它们的有效性在很大程度上依赖于如何协调这些工具：使用哪些工具、顺序如何、参数设置以及在何种视觉条件下使用。现有的视觉编程代理通常生成固定的解决方案管道，使其在密集物体、遮挡、小目标和领域转移的情况下显得脆弱。我们提出了VTOS（视觉工具协调搜索），这是一个通过联合解决方案-观察者搜索实现自适应视觉工具协调的框架。VTOS共同搜索可执行的解决方案程序，这些程序组合了视觉工具，如Grounding DINO、SAM、NMS和slice-and-detect，以及诊断候选解决方案、识别失败模式和生成可操作反馈的观察者程序。这些观察结果被汇总到共享的VisionThoughts知识库中，以指导后续的搜索。我们通过两个案例研究评估VTOS：在LVIS-Count上的密集物体计数和在PlantSeg-OOD上的零样本植物疾病分割，这些研究强调了不同的协调挑战，包括阈值校准、NMS、切片、掩膜精细化和领域泛化。在这两个任务中，VTOS的表现优于静态工具管道和代理视觉编程基线，显示出共同搜索解决方案和观察者是将视觉工具适应于具有挑战性的计算机视觉任务的有效策略。
cs.CV / 24 / 2606.20731
XmoPipe: A Pipeline for Large-Scale In-the-Wild Human Motion Dataset Construction
XmoPipe：一种大规模野外人类运动数据集构建管道
Abstract
Large-scale human motion datasets are essential for training robust motion models for analysis, synthesis, and understanding. While marker-based motion capture provides precise data, it is costly and limited in scale and diversity. Recent advances in monocular motion capture and video-language understanding open the way to extract plausible motion from unconstrained online videos. We present a scalable pipeline for constructing in-the-wild human motion datasets. From a few keywords, the system retrieves videos, extracts 3D body and facial motion, and generates high-level textual descriptions. The pipeline is flexible, enabling targeted collection of various motions, multi-person interactions, or expressive behaviors. We demonstrate its quality by training motion reconstruction and motion generation models, showing performance comparable to models trained on traditional motion capture datasets and strong cross-dataset generalization.
Chinese Translation
大规模人类运动数据集对于训练稳健的运动模型以进行分析、合成和理解至关重要。虽然基于标记的运动捕捉提供了精确的数据，但其成本高且在规模和多样性上受到限制。近期在单目运动捕捉和视频语言理解方面的进展为从不受限制的在线视频中提取合理的运动开辟了道路。我们提出了一种可扩展的管道，用于构建野外人类运动数据集。该系统从几个关键词中检索视频，提取三维身体和面部运动，并生成高级文本描述。该管道灵活，能够针对性地收集各种运动、多人的互动或表现性行为。我们通过训练运动重建和运动生成模型展示了其质量，结果表明其性能与在传统运动捕捉数据集上训练的模型相当，并且具有较强的跨数据集泛化能力。
cs.CV / 25 / 2606.20734
Robust Zero-Shot Generalization for Open-Vocabulary Action Recognition via Task Arithmetic
通过任务算术实现开放词汇动作识别的鲁棒零样本泛化
Abstract
Open Vocabulary Action Recognition (OVAR) enables the recognition of novel actions by leveraging vision-language representations, overcoming the limitations of traditional closed-set approaches. However, achieving robust performance in real-world scenarios typically requires domain-specific fine-tuning, which is often costly and raises privacy and regulatory concerns. In this work, we propose an alternative paradigm that bypasses target-domain training and recombines knowledge from existing datasets and models. Leveraging model merging and task arithmetic, we extract and combine task vectors from models fine-tuned on diverse public OVAR datasets. We show that, in out-of-distribution settings, the resulting merged model achieves superior zero-shot generalization to the pre-trained base model. Code is available at https://github.com/omaymaMoussadek/robust-ovar
Chinese Translation
开放词汇动作识别（OVAR）通过利用视觉-语言表示来识别新颖动作，克服了传统闭集方法的局限性。然而，在现实场景中实现鲁棒性能通常需要特定领域的微调，这往往成本高昂，并引发隐私和监管方面的担忧。在本研究中，我们提出了一种替代范式，绕过目标领域训练，并重新组合来自现有数据集和模型的知识。通过模型合并和任务算术，我们从在多样化公共OVAR数据集上微调的模型中提取并组合任务向量。我们展示了在分布外设置中，所得到的合并模型在零样本泛化方面优于预训练基础模型。代码可在 https://github.com/omaymaMoussadek/robust-ovar 获取。
cs.CV / 26 / 2606.20736
REKEY: Metadata-Grounded Visual-Key Regeneration for Contamination-Resilient VQA Evaluation
REKEY：基于元数据的视觉关键再生用于抗污染的视觉问答评估
Abstract
Static visual question answering (VQA) benchmarks age quickly: Once the items leak into training corpora, scores can reflect memorization rather than genuine visual ability, thus obscuring real progress. Rebuilding high-quality benchmarks such as V*Bench requires substantial human annotation, yet each static release can quickly become another leaked artifact. We propose ReKey, a live benchmark protocol that randomly regenerates the answer-bearing local detail, or visual key, in real images at evaluation time. Using human-validated edit slots, ReKey samples fresh instances with new answers, construction-grounded labels, and controlled visual-search difficulty. On V*Bench, the ReKey regenerated benchmark reveals a sharp score jump across eight frontier vision-language models (VLMs): The original items score 9.5--18.8 percentage points higher than the regenerated variants. By making the visual key renewable, ReKey keeps evaluation fresh as models and training data evolve.
Chinese Translation
静态视觉问答（VQA）基准迅速过时：一旦项目泄露到训练语料库中，得分可能反映的是记忆而非真实的视觉能力，从而掩盖了真正的进展。重建高质量基准如 V*Bench 需要大量的人为标注，但每次静态发布都可能迅速变成另一个泄露的文物。我们提出了 ReKey，一种实时基准协议，在评估时随机再生真实图像中的答案承载局部细节或视觉关键。通过使用经过人工验证的编辑槽，ReKey 采样新实例，提供新的答案、构建基础标签以及可控的视觉搜索难度。在 V*Bench 上，ReKey 再生的基准显示出八个前沿视觉-语言模型（VLMs）之间得分的显著跃升：原始项目的得分比再生变体高出 9.5 到 18.8 个百分点。通过使视觉关键可再生，ReKey 保持了评估的新鲜感，随着模型和训练数据的演变而不断更新。
cs.CV / 27 / 2606.20738
An approach with Visual and Tabular Mamba to multimodal medical data using Mixed Fusion
基于混合融合的视觉与表格Mamba方法在多模态医学数据中的应用
Abstract
This article presents a complementary approach for integrating multimodal medical data in cancer classification, based on state space models represented by the Mamba architecture. To this end, a mixed multimodal fusion architecture, called Mixed Fusion, was employed and developed to enhance the interpretability of the decision-making process. The proposed approach explores two variants of Mamba: one dedicated to visual processing, responsible for classifying the lesion image and generating probabilities associated with the target classes, and another focused on tabular processing, which uses these probabilities together with clinical and/or sociodemographic data to produce the final diagnosis. The experiments were conducted on two medical datasets: PAD-UFES-20, composed of clinical images and information associated with skin lesions, and NDB-UFES, consisting of histopathological images and sociodemographic data related to oral cancer. The results indicate slightly lower performance in balanced accuracy, compared with Transformer-based approaches, on PAD-UFES-20, and superior performance on NDB-UFES. Additionally, substantial gains were observed in the recall metric. Furthermore, the adoption of the Mixed Fusion architecture enables the application of the Shapley Additive Explanations (SHAP) method, increasing the interpretability of the results. These findings indicate that Mamba-based models constitute a suitable alternative for multimodal classification in medical data, especially in scenarios in which sensitivity is a relevant requirement.
Chinese Translation
本文提出了一种用于癌症分类的多模态医学数据整合的补充方法，该方法基于由Mamba架构表示的状态空间模型。为此，采用并开发了一种称为混合融合（Mixed Fusion）的混合多模态融合架构，以增强决策过程的可解释性。所提方法探索了Mamba的两个变体：一个专注于视觉处理，负责对病变图像进行分类并生成与目标类别相关的概率；另一个则专注于表格处理，利用这些概率结合临床和/或社会人口数据来生成最终诊断。实验在两个医学数据集上进行：PAD-UFES-20，包括与皮肤病变相关的临床图像和信息，以及NDB-UFES，包含与口腔癌相关的组织病理图像和社会人口数据。结果显示，在PAD-UFES-20上，与基于Transformer的方法相比，平衡准确率略低，而在NDB-UFES上表现优越。此外，在召回率指标上观察到了显著的提升。此外，采用混合融合架构使得能够应用Shapley加性解释（SHAP）方法，从而提高结果的可解释性。这些发现表明，基于Mamba的模型在医学数据的多模态分类中构成了一种合适的替代方案，特别是在灵敏度是相关要求的场景中。
cs.CV / 28 / 2606.20752
Mirage: a Clean-Label Backdoor against LiDAR 3D Object Detection
Mirage：一种针对LiDAR 3D物体检测的干净标签后门攻击
Abstract
Deep neural network-based LiDAR 3D object detection serves as a critical perception component in safety-critical autonomous systems. However, recent studies have revealed its vulnerability to backdoor attacks. Existing attacks typically require white-box access or label modification and focus on geometric attacks such as object disappearance or bounding-box manipulation. In this paper, we present Mirage, a black-box and clean-label backdoor attack against deep neural network-based LiDAR 3DOD. Mirage injects a small number of label-consistent poisoning samples into the training set, causing the model to learn a malicious association between a trigger pattern and an attacker-chosen target class while preserving normal training semantics. As a result, the compromised model behaves normally on benign inputs yet systematically misclassifies triggered objects as the target class during deployment. We evaluate Mirage on multiple state-of-the-art LiDAR 3DOD models and benchmark datasets. Experimental results show that Mirage achieves a 73% misclassification success rate with a poisoning rate of only 0.5%, while maintaining detection performance close to that of benign models.
Chinese Translation
基于深度神经网络的LiDAR 3D物体检测是安全关键型自主系统中的一个重要感知组件。然而，最近的研究揭示了其对后门攻击的脆弱性。现有的攻击通常需要白盒访问或标签修改，并集中于几何攻击，如物体消失或边界框操控。本文提出了Mirage，一种针对基于深度神经网络的LiDAR 3D物体检测（3DOD）的黑盒和干净标签后门攻击。Mirage向训练集中注入少量标签一致的中毒样本，导致模型学习触发模式与攻击者选择的目标类别之间的恶意关联，同时保持正常的训练语义。因此，受损的模型在良性输入上表现正常，但在部署期间系统性地将触发的物体错误分类为目标类别。我们在多个最先进的LiDAR 3DOD模型和基准数据集上评估了Mirage。实验结果表明，Mirage在仅0.5%的中毒率下实现了73%的错误分类成功率，同时保持了接近良性模型的检测性能。
cs.CV / 29 / 2606.20764
One Image is All You Need: Agentic One-Shot Image Generation via Text-Based World Models for Long-Tail Spatial Perception
一张图像足矣：基于文本的世界模型驱动的代理性单次图像生成用于长尾空间感知
Abstract
Reliable spatial decision automation, such as autonomous driving and maritime surveillance, critically depends on robust visual perception. However, real-world spatiotemporal data exhibits severe heterogeneity, often manifesting as extreme long-tail distributions for safety-critical scenarios. This data scarcity induces dataset shift that degrades detection performance and pose safety risks. While synthetic data generation offers a potential solution, existing generative approaches, such as diffusion models and Generative Adversarial Networks (GANs), often lack explicit spatial grounding and structural constraints, resulting in spatial and physical inconsistencies in generated scenes. To address these challenges, we introduce WMGen-v1, an agentic text-based world model framework for long-tail spatial data generation. WMGen-v1 employs a Large Vision-Language Model (LVLM) to construct a structured scene representation from a single reference image, while a Large Language Model (LLM) performs guidance-based scene expansion under physical plausibility and commonsense constraints. Subsequently, conditioned on the structured semantic representations produced by this reasoning process, a diffusion model generates diverse and physically grounded long-tail training data. Experiments on internal industrial datasets, ROADWork, and LaRS benchmarks demonstrate that WMGen-v1 outperforms baseline approaches. Notably, detectors trained solely on WMGen-v1 synthetic data approach real-only performance on aggregate dataset-level metrics, highlighting its potential to alleviate long-tail data scarcity for downstream spatial perception.
Chinese Translation
可靠的空间决策自动化，如自主驾驶和海洋监视，严重依赖于稳健的视觉感知。然而，现实世界的时空数据表现出严重的异质性，通常在安全关键场景中表现为极端的长尾分布。这种数据稀缺导致数据集偏移，从而降低检测性能并带来安全风险。尽管合成数据生成提供了一种潜在解决方案，但现有的生成方法，如扩散模型和生成对抗网络（GANs），往往缺乏明确的空间基础和结构约束，导致生成场景中的空间和物理不一致性。为了解决这些挑战，我们引入了WMGen-v1，一个用于长尾空间数据生成的代理性基于文本的世界模型框架。WMGen-v1利用大型视觉-语言模型（LVLM）从单个参考图像构建结构化场景表示，而大型语言模型（LLM）在物理合理性和常识约束下执行基于指导的场景扩展。随后，基于这一推理过程生成的结构化语义表示，扩散模型生成多样且物理基础扎实的长尾训练数据。在内部工业数据集、ROADWork和LaRS基准上的实验表明，WMGen-v1优于基线方法。值得注意的是，仅在WMGen-v1合成数据上训练的检测器在汇总数据集级别指标上接近真实数据性能，突显了其缓解下游空间感知长尾数据稀缺的潜力。
cs.CV / 30 / 2606.20768
UniSLAD: A Unified Framework for Structural and Logical Industrial Visual Anomaly Detection
UniSLAD：一个统一的结构与逻辑工业视觉异常检测框架
Abstract
Visual anomaly detection is a fundamental task in industrial automation. While existing approaches have achieved notable progress in identifying structural defects, the detection of logical anomalies remains relatively underexplored. In practice, structural and logical anomalies frequently co-occur in industrial workflows. Therefore, a solution capable of detecting both structural and logical anomalies is crucial for advancing comprehensive anomaly detection research. To address this limitation, we propose a unified framework, termed UniSLAD, which jointly addresses logical and structural anomalies without additional training, enabling a practical solution for dynamic industrial environments. First, we introduce a dual-feature extractor that synergistically integrates a Convolutional Neural Network (CNN) backbone for local texture perception with a Transformer backbone for global contextual reasoning, yielding richer and more comprehensive representations. Building on this foundation, we design dual-granularity feature representation modules. At the patch level, memory banks enhanced by the Mahalanobis Transform (MT) preserve representative features and support more discriminative anomaly scoring. At the image level, distribution maps are aggregated using Lower-Upper Mean (LUM) and Power Mean Pooling (PMP), yielding a more robust global representation than conventional average pooling. Extensive experiments on the two industrial benchmarks demonstrate that UniSLAD achieves competitive performance in comprehensive anomaly detection, achieving 99.4% and 93.1%, respectively. Furthermore, ablation studies verify the individual contributions and effectiveness of each proposed component.
Chinese Translation
视觉异常检测是工业自动化中的一项基础任务。尽管现有方法在识别结构缺陷方面取得了显著进展，但逻辑异常的检测仍然相对未被充分探索。在实际应用中，结构和逻辑异常在工业工作流程中常常同时出现。因此，能够同时检测结构和逻辑异常的解决方案对于推动全面的异常检测研究至关重要。为了解决这一局限性，我们提出了一个统一框架，称为UniSLAD，它能够在不进行额外训练的情况下共同处理逻辑和结构异常，从而为动态工业环境提供实用的解决方案。首先，我们引入了一个双特征提取器，该提取器协同整合了用于局部纹理感知的卷积神经网络（CNN）主干和用于全局上下文推理的变换器（Transformer）主干，从而生成更丰富和更全面的表示。在此基础上，我们设计了双粒度特征表示模块。在补丁级别，增强的马哈拉诺比斯变换（MT）记忆库保留了代表性特征，并支持更具区分性的异常评分。在图像级别，使用下上均值（LUM）和幂均值池化（PMP）聚合分布图，生成比传统平均池化更强健的全局表示。在两个工业基准上的大量实验表明，UniSLAD在全面异常检测中实现了具有竞争力的性能，分别达到了99.4%和93.1%。此外，消融研究验证了每个提出组件的个体贡献和有效性。
cs.CV / 31 / 2606.20774
TriMotion: Modality-Agnostic Camera Control for Video Generation
TriMotion：一种与模态无关的摄像机控制视频生成方法
Abstract
Camera motion control is essential for directing viewpoint changes in generative systems. However, existing methods typically condition the generation process on a single specific modality, such as explicit pose trajectories or reference videos, limiting their ability to support heterogeneous user inputs. To address this limitation, we present TriMotion, a modality-agnostic framework for camera-controlled video generation that maps video, pose, and text inputs, describing the same camera trajectory into a shared motion embedding space. Learning such a space requires synchronized supervision across modalities. Therefore, we build the Motion Triplet Dataset by extending a Multi-Cam Video Dataset with geometry-grounded motion descriptions derived from camera extrinsics. We further introduce a latent motion consistency objective that leverages the motion embedding space to encourage the generated video to follow the target camera trajectory directly in latent space, avoiding the cost of pixel-space decoding. Extensive experiments show that TriMotion generates high-quality videos that accurately follow the target camera trajectories across all three modalities. Beyond standard generation, the shared motion embedding space also enables flexible applications such as sequential motion composition and cross-modal motion interpolation.
Chinese Translation
摄像机运动控制对于生成系统中的视角变化至关重要。然而，现有方法通常仅基于单一特定模态（如显式姿态轨迹或参考视频）来条件生成过程，这限制了其支持异构用户输入的能力。为了解决这一限制，我们提出了TriMotion，一种与模态无关的摄像机控制视频生成框架，该框架将视频、姿态和文本输入映射到共享的运动嵌入空间，描述相同的摄像机轨迹。学习这样的空间需要跨模态的同步监督。因此，我们通过扩展一个多摄像机视频数据集，构建了运动三元组数据集，并从摄像机外部参数中派生出几何基础的运动描述。我们进一步引入了一种潜在运动一致性目标，利用运动嵌入空间鼓励生成的视频在潜在空间中直接遵循目标摄像机轨迹，从而避免了像素空间解码的成本。大量实验表明，TriMotion能够生成高质量的视频，准确遵循所有三种模态的目标摄像机轨迹。除了标准生成外，共享的运动嵌入空间还支持灵活的应用，如顺序运动组合和跨模态运动插值。
cs.CV / 32 / 2606.20799
GroundShot: Visually Consistent Multi-Shot Long Video Generation via Entity-Grounded Shot Scheduling
GroundShot：通过实体基础镜头调度生成视觉一致的多镜头长视频
Abstract
Generating visually consistent multi-shot videos remains an open challenge. As videos span more shots, inconsistencies can accumulate across shots, causing entities that reappear across shots -- characters, objects, and locations -- to drift away from how they first appear. We observe that viewers judge consistency by comparing each later appearance of an entity with its first clear appearance; the visual quality of this initial appearance sets the consistency ceiling for all that follows. Motivated by this, we present \textbf{GroundShot}, a training-free, model-agnostic agentic framework for entity-grounded multi-shot generation. GroundShot builds an entity-level visual memory online from accepted generated shots: it schedules shots' generation order by their expected usefulness as entity references, grounds entities from generated videos, verifies their reliability before adding them to memory, and retrieves suitable entity references from memory before each shot is generated. To evaluate this entity-centered view of consistency, we further introduce \textbf{GroundBench}, a diagnostic benchmark that measures consistency at the entity level while isolating controlled challenge dimensions. Experiments show that GroundShot improves multi-shot consistency over existing methods while requiring no additional training or model modification.
Chinese Translation
生成视觉一致的多镜头视频仍然是一个未解决的挑战。随着视频镜头数量的增加，不一致性可能在镜头之间累积，导致在不同镜头中重复出现的实体——角色、物体和地点——偏离其首次出现时的样子。我们观察到，观众通过将每个后续出现的实体与其首次清晰出现进行比较来判断一致性；这种初始出现的视觉质量为后续所有内容设定了一致性的上限。基于此，我们提出了 extbf{GroundShot}，一个无训练、模型无关的基于实体的多镜头生成框架。GroundShot在线构建实体级视觉记忆，从已接受的生成镜头中获取信息：它根据镜头作为实体参考的预期有用性安排生成顺序，从生成的视频中识别实体，在将其添加到记忆之前验证其可靠性，并在每个镜头生成之前从记忆中检索合适的实体参考。为了评估这种以实体为中心的一致性观点，我们进一步引入了 extbf{GroundBench}，一个诊断基准，测量实体级别的一致性，同时隔离受控挑战维度。实验表明，GroundShot在不需要额外训练或模型修改的情况下，提高了多镜头的一致性，优于现有方法。
cs.CV / 33 / 2606.20823
NeoLoc-68: End-to-end 68-point neonatal facial landmark localisation in neonatal clinical environments
NeoLoc-68：在新生儿临床环境中实现端到端的68点新生儿面部标志定位
Abstract
Facial landmark localisation is a prerequisite for developing automated, non-contact neonatal pain assessment methods. Clinicians use pain scales to judge the severity of pain, many of which rely on facial expression. However, facial landmark detectors trained on adult faces perform poorly in neonatal clinical environments due to frequent occlusions caused by medical equipment, varied head poses, and challenging imaging conditions, including motion blur triggered by sudden pain-related movements. We propose an end-to-end facial landmark detector capable of predicting 68 landmarks on neonatal faces in clinical environments. We combined 37,459 single-face images from 11 public datasets, standardised to 68-point markup, with 1,123 manually annotated frames from a neonatal research dataset (totalling over 76,000 landmarks). A YOLO-based keypoint model was adapted to regress the facial landmarks, initialised with weights from a pretrained neonatal face detector. On public datasets, our proposed model achieved state-of-the-art performance: Normalised Mean Error (NME) = 5.37, Failure Rate (FR) = 12.5%, Area Under the Cumulative Error Curve (AUC) at AUC0.08 = 38.00% and AUC0.1 = 48.70%. On the clinical neonatal test set, before fine-tuning, the model achieved the lowest Detection Failure Rate (DFR) = 5.3% among all baselines and showed strong generalisation. After fine-tuning, performance improved further to NME = 6.36, FR = 22.30%, DFR = 1.77%, AUC0.08 = 29.24% and AUC0.1 = 40.25%. To the best of our knowledge, this represents the first end-to-end 68-point neonatal facial landmark detection model. With further dataset expansion and refinement, it could support downstream tasks in neonatal health monitoring and pain-related facial analysis.
Chinese Translation
面部标志定位是开发自动化、非接触式新生儿疼痛评估方法的前提。临床医生使用疼痛评分量表来判断疼痛的严重程度，其中许多评分依赖于面部表情。然而，针对成人面部训练的面部标志检测器在新生儿临床环境中的表现不佳，这主要是由于医疗设备造成的频繁遮挡、头部姿势的多样性以及包括因突发疼痛相关运动引发的运动模糊在内的复杂成像条件。我们提出了一种端到端的面部标志检测器，能够在临床环境中预测新生儿面部的68个标志。我们结合了来自11个公共数据集的37,459张单面孔图像，标准化为68点标记，并与来自新生儿研究数据集的1,123帧手动标注图像相结合（总计超过76,000个标志）。我们调整了一种基于YOLO的关键点模型，以回归面部标志，并使用预训练的新生儿面部检测器的权重进行初始化。在公共数据集上，我们提出的模型达到了最先进的性能：标准化均值误差（NME）= 5.37，失败率（FR）= 12.5%，累积误差曲线下面积（AUC）在AUC0.08 = 38.00%和AUC0.1 = 48.70%。在临床新生儿测试集上，模型在微调前的检测失败率（DFR）为5.3%，在所有基线中最低，并显示出强大的泛化能力。经过微调后，性能进一步提升至NME = 6.36，FR = 22.30%，DFR = 1.77%，AUC0.08 = 29.24%和AUC0.1 = 40.25%。据我们所知，这代表了首个端到端的68点新生儿面部标志检测模型。随着数据集的进一步扩展和优化，它可以支持新生儿健康监测和与疼痛相关的面部分析等下游任务。
cs.CV / 34 / 2606.20842
From Uncertainty to Stability and Fidelity: Guiding Sparse-View 3D Gaussian Splatting with Fisher Information
从不确定性到稳定性与保真度：利用费舍尔信息指导稀疏视图3D高斯点云渲染
Abstract
3D Gaussian Splatting (3DGS) has emerged as a promising technique for novel view synthesis. However, 3DGS requires dense input views to achieve high-quality rendering. In sparse-view scenarios, 3DGS often prones to overfitting, resulting in noticeable artifacts and degraded rendering quality. Previous methods explore to address this issue by introducing additional priors (e.g. depth priors) or integrating regularization techniques (e.g. Dropout). However, these methods are often applied without principled guidance. In particular, prior-based augmentation typically samples novel viewpoints randomly, while Dropout-based regularization randomly removes Gaussians. The compounded randomness introduces uncertainty and instability, limiting the fidelity of novel view synthesis. In this paper, we propose a novel method for sparse-view 3DGS that incorporates Fisher Information to quantitatively guide the utilization of geometric priors and regularization. Specifically, our method comprises two key components: (1) Stereo augmentation with Fisher Information. By leveraging Fisher Information, we actively select most informative supporting views and use depth priors to curate reliable pseudo ground truths, which reduces randomness in augmentation and improves stability and rendering fidelity; (2) Uncertainty-aware regularization. We reduce the instability of Dropout-based regularization by using Fisher Information to quantitatively measure the uncertainty of each 3D Gaussian, and adaptively adjust the removal probability, leading to more stable and effective regularization. With these two components, our method effectively mitigates overfitting and improves the stability of optimization in sparse-view 3DGS, resulting in superior rendering fidelity. Extensive experiments show that our method achieves state-of-the-art performance in sparse-view novel view synthesis benchmarks.
Chinese Translation
3D高斯点云渲染（3DGS）已成为新视图合成的一种有前景的技术。然而，3DGS需要密集的输入视图以实现高质量的渲染。在稀疏视图场景中，3DGS往往容易过拟合，导致明显的伪影和渲染质量下降。之前的方法通过引入额外的先验（例如深度先验）或整合正则化技术（例如Dropout）来解决这一问题。然而，这些方法通常缺乏原则性指导。特别是，基于先验的增强通常随机采样新视点，而基于Dropout的正则化则随机移除高斯分布。复合的随机性引入了不确定性和不稳定性，限制了新视图合成的保真度。本文提出了一种新颖的稀疏视图3DGS方法，该方法结合费舍尔信息以定量指导几何先验和正则化的利用。具体而言，我们的方法包括两个关键组件：（1）利用费舍尔信息的立体增强。通过利用费舍尔信息，我们主动选择最具信息量的支持视图，并使用深度先验来策划可靠的伪地面真实值，从而减少增强中的随机性，提高稳定性和渲染保真度；（2）不确定性感知的正则化。我们通过使用费舍尔信息定量测量每个3D高斯的不确定性，适应性地调整移除概率，从而减少基于Dropout的正则化的不稳定性，导致更稳定和有效的正则化。通过这两个组件，我们的方法有效减轻了过拟合，并提高了稀疏视图3DGS中优化的稳定性，从而实现了更优的渲染保真度。大量实验表明，我们的方法在稀疏视图新视图合成基准测试中达到了最先进的性能。
cs.CV / 35 / 2606.20852
Translating Inference-Time Control to Radiology Vision-Language Models: Activation Steering for Pneumonia Classification on Chest X-rays
将推理时控制转化为放射学视觉-语言模型：用于胸部X光肺炎分类的激活引导
Abstract
Inference-time engineering can alter model behavior without fine-tuning. However, its utility for improving diagnostic performance in medical vision-language models (VLMs) remains unclear. We aim to evaluate whether Contrastive Activation Addition (CAA) can improve pneumonia classification in chest radiograph VLMs without updating model weights. Three frozen chest radiograph VLMs (MedGemma-4B-IT, NV-Reason-CXR-3B, and CheXOne-3B) were evaluated on the public Kermany pneumonia test set. Classification was based on the logits of the tokens Yes and No under a binary prompt. Steering vectors included a 30-pair answer-bias control, a 30-pair pneumonia text contrast, and an image-conditioned contrast derived from 30 pneumonia and 30 normal development images. A deterministic 200-image development set was used for layer and scale selection (100 images) and threshold calibration (100 images). Performance was assessed using ROC-AUC, PR-AUC, F1 score, threshold analyses, reverse-vector controls, random-vector controls, and conditional bootstrap confidence intervals. Fixed-threshold F1 improvements were frequently observed but did not consistently indicate improved diagnostic performance. For MedGemma-4B-IT. NV-Reason-CXR-3B showed the strongest benefit: calibrated F1 improved from 0.7692 in the zero-shot setting to 0.8619 with pneumonia-text steering and to 0.8727 with image-conditioned steering. For CheXOne-3B, pneumonia-text steering increased calibrated F1 from 0.8528 to 0.8666, although the confidence interval crossed zero. On this public pneumonia benchmark, CAA substantially altered prediction score distributions and operating characteristics without fine-tuning. Meaningful performance gains were observed in one of three evaluated VLMs, suggesting that activation steering may serve as a lightweight approach for adapting medical VLM behavior.
Chinese Translation
推理时工程可以在不进行微调的情况下改变模型行为。然而，其在提高医学视觉-语言模型（VLMs）诊断性能方面的效用仍不明确。我们旨在评估对比激活添加（Contrastive Activation Addition, CAA）是否能够在不更新模型权重的情况下改善胸部X光VLMs中的肺炎分类。我们在公共的Kermany肺炎测试集上评估了三个冻结的胸部X光VLM（MedGemma-4B-IT、NV-Reason-CXR-3B和CheXOne-3B）。分类基于二元提示下的Yes和No标记的logits。引导向量包括30对答案偏差控制、30对肺炎文本对比以及从30张肺炎和30张正常发育图像中派生的图像条件对比。使用一个确定性的200张图像开发集进行层和比例选择（100张图像）及阈值校准（100张图像）。使用ROC-AUC、PR-AUC、F1分数、阈值分析、反向向量控制、随机向量控制和条件自助置信区间评估性能。固定阈值下的F1改进经常被观察到，但并未始终指示出诊断性能的改善。对于MedGemma-4B-IT，NV-Reason-CXR-3B显示出最强的益处：经过校准的F1从零样本设置下的0.7692提高到使用肺炎文本引导的0.8619，以及使用图像条件引导的0.8727。对于CheXOne-3B，肺炎文本引导将校准的F1从0.8528提高到0.8666，尽管置信区间穿过零。在这个公共肺炎基准上，CAA在不进行微调的情况下显著改变了预测分数分布和操作特性。在评估的三个VLM中，观察到一个模型的显著性能提升，这表明激活引导可能作为一种轻量级的方法来适应医学VLM的行为。
cs.CV / 36 / 2606.20856
Stochastic Signed Distance Processes
随机签名距离过程
Abstract
Multi-view surface reconstruction is a core problem in computer vision. One prominent line of work represents the surface implicitly as a signed distance field (SDF), optimizing it based on the photometric loss between rendered and observed pixel colors. These approaches typically employ SDF-based volume rendering to obtain a differentiable relaxation of discontinuous visibility along rays, thereby reducing reliance on silhouette supervision. In this paper, we reformulate SDF-based volume rendering as probabilistic surface rendering, where each pixel color is modeled as a mixture distribution induced by the random first ray-surface intersection. To this end, we introduce Stochastic Signed Distance Processes (SSDP), which model the SDF along each ray as a stochastic process, inducing a first-passage-time distribution for each ray. We then derive the first-passage probability for each sampling interval based on Bayesian filtering, together with its practical approximation for parallel rendering. We further show that NeuS, an existing SDF-based volume rendering method, arises as a special case of our formulation. Experiments on the DTU and MobileBrick datasets demonstrate that our method outperforms baselines in both surface reconstruction and uncertainty quantification, supporting the effectiveness of our first-passage formulation. Our code is available at https://github.com/skmhrk1209/SSDP.
Chinese Translation
多视角表面重建是计算机视觉中的一个核心问题。一项突出的研究工作将表面隐式表示为签名距离场（SDF），并基于渲染图像与观察到的像素颜色之间的光度损失进行优化。这些方法通常采用基于SDF的体积渲染，以获得沿光线的不连续可见性的一种可微松弛，从而减少对轮廓监督的依赖。在本文中，我们将基于SDF的体积渲染重新表述为概率表面渲染，其中每个像素颜色被建模为由随机第一光线-表面交点引起的混合分布。为此，我们引入了随机签名距离过程（SSDP），该过程将每条光线上的SDF建模为一个随机过程，从而为每条光线引入一个首次通过时间分布。然后，我们基于贝叶斯滤波推导出每个采样区间的首次通过概率，并给出其在并行渲染中的实际近似。我们进一步表明，现有的基于SDF的体积渲染方法NeuS是我们公式的一个特例。在DTU和MobileBrick数据集上的实验表明，我们的方法在表面重建和不确定性量化方面均优于基线，支持我们首次通过公式的有效性。我们的代码可在https://github.com/skmhrk1209/SSDP获取。
cs.CV / 37 / 2606.20867
FOCA: Future-Oriented Conditioning for Data-Efficient Vision-Language-Action Adaptation
FOCA：面向未来的条件化框架用于数据高效的视觉-语言-动作适应
Nguyen, Duc Minh, Diep, Nghiem Tuong, Nguyen, Binh Gia, Ho, Trong-Bao, Le, Doanh, Nguyen, Tan Q., Ha, Thien-Loc, Tran, Nhiem, Thach, Bao, Tran, Nhat X., Tran, Tuan A., Habuda, Artur, Møller, Philip Lund, Le, Tran Nguyen, Sonntag, Daniel, Niepert, Matthias, Doan, Khoa D., Duong, Vu, Ngo, Hung, Vu, Minh N., Nguyen, Duy M. H., Le, An Thai, Vien, Ngo Anh
Abstract
Vision-Language-Action (VLA) models enable general-purpose robotic control via large-scale multimodal pretraining, yet their effectiveness under few-shot imitation learning remains limited. We conduct a systematic stress test of state-of-the-art VLA models and show that performance degrades sharply as demonstrations are reduced, revealing a key weakness of existing adaptation strategies. To address this, we introduce FOCA, a future-oriented conditioning framework for data-efficient VLA adaptation. FOCA combines explicit prediction of task-grounded future interaction embeddings with implicit alignment to future goal observations, enabling long-horizon reasoning in latent space without pixel-level prediction. This formulation naturally supports action-free co-training with synthetic videos from video world models and can be interpreted as learning a future-conditioned value-like representation. Extensive experiments demonstrate FOCA achieves 95.7% success with 20 demonstrations on LIBERO, improves 7-12% on RoboCasa, and delivers up to 26% absolute gains on real robots, establishing a new state of the art in few-shot VLA adaptation.
Chinese Translation
视觉-语言-动作（VLA）模型通过大规模多模态预训练实现通用机器人控制，但在少样本模仿学习下的有效性仍然有限。我们对最先进的VLA模型进行了系统的压力测试，结果表明，随着演示数量的减少，性能急剧下降，揭示了现有适应策略的一个关键弱点。为了解决这个问题，我们提出了FOCA，一种面向未来的条件化框架，用于数据高效的VLA适应。FOCA结合了对任务基础未来交互嵌入的显式预测与对未来目标观测的隐式对齐，使得在潜在空间中进行长时间推理成为可能，而无需进行像素级预测。这种形式自然支持与来自视频世界模型的合成视频的无动作共同训练，并可以被解释为学习一种未来条件的价值类表示。大量实验表明，FOCA在LIBERO上以20个演示实现了95.7%的成功率，在RoboCasa上提高了7-12%，并在真实机器人上提供了高达26%的绝对增益，确立了少样本VLA适应的新状态。
cs.CV / 38 / 2606.20886
Toward Parking Spot Occupancy Recognition: A Self-Supervised Approach
朝着停车位占用识别：一种自监督方法
Abstract
As urban areas expand, automatic monitoring of parking lots becomes essential for efficient and sustainable cities. This work proposes a self-supervised approach for parking spot occupancy recognition that requires no labeled samples from the target parking lot. Building upon a self-supervised transfer learning fine-tuning protocol, the proposed training strategy consists of two self-supervised stages: first on unlabeled generic data and then on unlabeled target-specific data, followed by supervised fine-tuning using only generic parking lot labels. We adopt SimCLR with a ResNet-50 encoder and evaluate the method under a leave-one-out cross-environment protocol on three public datasets: PKLot, CNRPark-EXT, and PLds. We also introduce a two-stage deployment strategy in which a Strong General Model is initially deployed, followed by a Specialized Model that incorporates unlabeled images collected during the first N days of deployment in a self-supervised manner. Experimental results show that the Strong General Model alone outperforms supervised and self-supervised baselines, achieving an average accuracy of 97.2%, which further improves to 97.8% with the proposed two-stage strategy. These results demonstrate that self-supervised learning enables a scalable and labelefficient solution for real-world parking occupancy monitoring. Our trained models and source code are publicly available at https://github.com/LoanMaikon/Parking-Spot-Occupancy-Recognition.
Chinese Translation
随着城市区域的扩张，自动监测停车场对于高效和可持续城市变得至关重要。本研究提出了一种自监督的停车位占用识别方法，该方法不需要目标停车场的标注样本。基于自监督迁移学习微调协议，所提出的训练策略包括两个自监督阶段：首先在未标注的通用数据上进行训练，然后在未标注的目标特定数据上进行训练，最后使用仅包含通用停车场标签的监督微调。我们采用了带有 ResNet-50 编码器的 SimCLR 方法，并在三个公共数据集（PKLot、CNRPark-EXT 和 PLds）上通过留一法交叉环境协议评估该方法。我们还提出了一种两阶段部署策略，初始部署一个强通用模型，然后部署一个专用模型，该模型以自监督的方式整合在前 N 天收集的未标注图像。实验结果表明，仅强通用模型的表现优于监督和自监督基线，平均准确率达到 97.2%，而采用所提出的两阶段策略后进一步提高至 97.8%。这些结果表明，自监督学习为现实世界的停车占用监测提供了一种可扩展且高效的解决方案。我们的训练模型和源代码已公开，地址为 https://github.com/LoanMaikon/Parking-Spot-Occupancy-Recognition。
cs.CV / 39 / 2606.20888
Fine-grained Human Motion Understanding with Language Models
基于语言模型的细粒度人类动作理解
Abstract
In this work, we propose \methodname, an LLM-based model for fine-grained human motion understanding that represents motion as a sequence of skeletal poses with explicit timestamps for each pose. Each pose encodes body joint positions and is temporally grounded with timestamp tokens, allowing the model to reason about motion order, duration, and rhythm. To study what supervision is needed for motion-language reasoning, we construct a diverse training mixture spanning pose captioning, pose question answering, motion captioning, and motion question answering. Our ablations show that the primary gains come from the diversity of pose- and motion-level supervision, while staged training provides a smaller additional benefit. Different from previous works that rely on ground-truth 3D motion capture, our approach supports both 2D and 3D skeletal motion representations through a unified pose encoder, and can optionally incorporate video to provide contextual information. Extensive experiments on BABEL-QA, HuMMan-QA, CompMo, NTU-RGB+D, and QEVD-Coach demonstrate that our method achieves state-of-the-art performance across multiple benchmarks, highlighting the effectiveness of explicit temporal encoding and diverse pose- and motion-level supervision for fine-grained human motion understanding. Notably, even when using only 2D skeletal input, our approach surpasses previous 3D-based methods.
Chinese Translation
在本研究中，我们提出了 extit{methodname}，一种基于大型语言模型（LLM）的细粒度人类动作理解模型，该模型将动作表示为一系列带有明确时间戳的骨骼姿势。每个姿势编码了身体关节的位置，并通过时间戳标记进行时间上的定位，使模型能够推理动作的顺序、持续时间和节奏。为了研究动作-语言推理所需的监督类型，我们构建了一个多样化的训练混合体，涵盖姿势描述、姿势问答、动作描述和动作问答。我们的消融实验表明，主要的性能提升来自于姿势和动作层面监督的多样性，而分阶段训练提供的额外收益较小。与依赖真实3D动作捕捉的先前工作不同，我们的方法通过统一的姿势编码器支持2D和3D骨骼动作表示，并可以选择性地结合视频以提供上下文信息。在BABEL-QA、HuMMan-QA、CompMo、NTU-RGB+D和QEVD-Coach等多个基准上的广泛实验表明，我们的方法在多个基准上达到了最先进的性能，突显了显式时间编码和多样化姿势及动作层面监督在细粒度人类动作理解中的有效性。值得注意的是，即使仅使用2D骨骼输入，我们的方法也超越了先前基于3D的方法。
cs.CV / 40 / 2606.20891
Go-with-the-Track: Video Compositing and Motion Control with Point Tracking
随轨道而动：基于点跟踪的视频合成与运动控制
Abstract
Filmmaking demands precise motion control and reference image compositing -- capabilities that existing methods treat separately. Point-track-conditioned image-to-video models restrict content insertion to the first frame, while reference-to-video models lack fine-grained spatial-temporal control over how reference content integrates across frames. We present Go-with-the-Track, which unifies both capabilities by jointly conditioning on multiple reference images and reference-anchored point-tracks -- extending conventional point-tracks to explicitly establish correspondences between generated frames and reference images, thus enabling precise compositing and motion control throughout the video. To achieve this, we introduce spatially-aware point-track embeddings that encode the full sequence of point-track coordinates using a coordinate-wise MLP followed by temporal pooling. This representation captures the spatial characteristics of each point-track (serving as a unique identifier), while the embedding similarity correlates directly with spatial proximity, enhancing the model's ability to distinguish and associate point-tracks. We inject these point-track embeddings into a video diffusion transformer via a lightweight adapter, resolving the pixel-to-patch resolution mismatch while avoiding the substantial motion detail loss inherent in naive point-track subsampling. We use a hybrid training strategy to train jointly on dynamic, static, and synthetic scene video datasets to boost motion controllability. Experiments demonstrate that Go-with-the-Track achieves superior motion and reference control in a single model and enables new capabilities: multi-reference conditioned video generation with point-track driven compositing, as well as camera control for both static and dynamic scenes. Project Page: https://eyeline-labs.github.io/Go-with-the-Track/
Chinese Translation
电影制作需要精确的运动控制和参考图像合成，而现有方法对此进行分开处理。基于点跟踪的图像到视频模型限制了内容插入仅在第一帧，而参考到视频模型则缺乏对参考内容在帧间整合的细粒度时空控制。我们提出了“随轨道而动”（Go-with-the-Track），通过联合条件化多个参考图像和参考锚定的点跟踪，统一了这两种能力——将传统的点跟踪扩展为明确建立生成帧与参考图像之间的对应关系，从而实现整个视频中的精确合成和运动控制。为此，我们引入了空间感知的点跟踪嵌入，使用坐标逐维的多层感知机（MLP）编码点跟踪坐标的完整序列，随后进行时间池化。这种表示捕捉了每个点跟踪的空间特征（作为唯一标识符），而嵌入相似性与空间邻近性直接相关，增强了模型区分和关联点跟踪的能力。我们通过轻量级适配器将这些点跟踪嵌入注入视频扩散变换器，解决了像素到补丁的分辨率不匹配，同时避免了天真点跟踪下采样中固有的显著运动细节损失。我们采用混合训练策略，在动态、静态和合成场景视频数据集上进行联合训练，以提升运动可控性。实验表明，“随轨道而动”在单一模型中实现了优越的运动和参考控制，并启用了新的能力：基于点跟踪驱动的多参考条件视频生成以及静态和动态场景的相机控制。项目页面：https://eyeline-labs.github.io/Go-with-the-Track/
cs.CV / 41 / 2606.20909
BELDE: Building a Large-scale Earth-observation Land-cover Dataset for Europe
BELDE：构建欧洲大规模地球观测土地覆盖数据集
Abstract
Earth observation imagery plays a critical role in environmental monitoring, urban planning, disaster assessment, and climate analysis. While multi-spectral sensors are increasingly available, true-color (RGB) imagery remains widely used due to the power, cost, and deployment constraints of many satellite and aerial platforms. However, existing land-cover segmentation datasets are often limited in geographic coverage, scale, or public accessibility. To bridge this gap, we introduce BELDE (Building a Large-scale Earth-observation Land-cover Dataset for Europe), a publicly available dataset tailored for RGB-based remote sensing semantic segmentation. Constructed from Sentinel-2 true-color images and ESA WorldCover data annotations, BELDE contains 1,088,385 curated image-segmentation map pairs spanning Europe with 7 land-cover classes at 10 m spatial resolution, making it one of the largest publicly available RGB land-cover segmentation datasets for Earth observation. To facilitate cross-region generalization studies, we additionally introduce BELDE-K (16,607 pairs) covering the Republic of Korea and BELDE-CA-NV (88,155 pairs) covering California and Nevada in the United States. We establish baseline results using multiple semantic segmentation architectures and evaluate both in-domain and cross-domain performance. Models trained on BELDE achieve an F1 score of 83.0% on the European test set, while performance decreases to 66.4% on BELDE-CA-NV and 58.3% on BELDE-K, highlighting the challenges posed by out-of-distribution geographic domain shift. By providing a continental-scale RGB segmentation and evaluation benchmark, BELDE supports the development of robust and transferable Earth observation models. The dataset and benchmark resources will be publicly released.
Chinese Translation
地球观测影像在环境监测、城市规划、灾害评估和气候分析中发挥着关键作用。尽管多光谱传感器的可用性日益增加，真实色彩（RGB）影像仍因许多卫星和航空平台的功率、成本及部署限制而被广泛使用。然而，现有的土地覆盖分割数据集在地理覆盖范围、规模或公共可访问性方面往往存在局限。为填补这一空白，我们推出了BELDE（构建欧洲大规模地球观测土地覆盖数据集），这是一个针对基于RGB的遥感语义分割而量身定制的公开数据集。BELDE由Sentinel-2真实色彩影像和欧洲航天局（ESA）WorldCover数据注释构建，包含1,088,385对经过精心挑选的影像-分割图，覆盖欧洲，具有7个土地覆盖类别，空间分辨率为10米，使其成为可用于地球观测的最大公开RGB土地覆盖分割数据集之一。为了促进跨区域泛化研究，我们还推出了BELDE-K（16,607对），覆盖韩国，以及BELDE-CA-NV（88,155对），覆盖美国加利福尼亚州和内华达州。我们使用多种语义分割架构建立基线结果，并评估领域内和跨领域的性能。在欧洲测试集上，基于BELDE训练的模型达到了83.0%的F1分数，而在BELDE-CA-NV和BELDE-K上的性能分别下降至66.4%和58.3%，突显了分布外地理域转移所带来的挑战。通过提供一个大陆尺度的RGB分割和评估基准，BELDE支持强大且可转移的地球观测模型的开发。数据集和基准资源将公开发布。
cs.CV / 42 / 2606.20913
PROTON: Prototype-Based Test-Time Online OOD Detection for Medical VLMs
PROTON：基于原型的测试时在线OOD检测用于医疗视觉语言模型
Abstract
Medical vision-language models (VLMs) enable zero-shot clinical image classification, yet reliably detecting out-of-distribution (OOD) inputs at deployment remains an open problem. No static scoring method works across all shift types: Maximum Concept Matching (MCM) on FLAIR achieves 76.4% AUROC for far-OOD but only 42.4% for covariate shifts such as ultra-wide-field fundus images, effectively random. We trace this to a structural mismatch: covariate-shifted inputs are indistinguishable from in-distribution samples in softmax space, yet occupy distinct regions in the VLM embedding space. To exploit this untapped signal, we propose PROTON (PROtotype-based Test-time ONline OOD detection), a lightweight post-hoc module that maintains an online prototype bank from high-confidence test predictions and adaptively fuses prototype distance with MCM scoring via stream-level variance statistics, requiring no model modification, training data, or prompt engineering. On the ophthalmology benchmark FLAIR + FIVES, PROTON improves MCM by +23.9 AUROC on covariate shift, +8.8 on semantic shift, and +8.1 on far-OOD, making it the only zero-shot method to improve all three without hierarchical prompts or labeled data. Code is available at https://github.com/GenMI-Lab/PROTON, and the project page is available at https://genmi-lab.github.io/PROTON.
Chinese Translation
医疗视觉语言模型（VLMs）能够实现零样本临床图像分类，但在部署时可靠地检测分布外（OOD）输入仍然是一个未解决的问题。没有一种静态评分方法能够适用于所有类型的分布变化：在 FLAIR 上的最大概念匹配（MCM）对于远离OOD的输入达到了76.4%的AUROC，但对于协变量变化（如超广角眼底图像）仅为42.4%，几乎是随机的。我们将此归因于结构不匹配：在softmax空间中，协变量变化的输入与分布内样本无法区分，但在VLM嵌入空间中占据不同的区域。为了利用这一未开发的信号，我们提出了PROTON（基于原型的测试时在线OOD检测），这是一种轻量级的后处理模块，它从高置信度的测试预测中维护一个在线原型库，并通过流级方差统计自适应地将原型距离与MCM评分融合，无需模型修改、训练数据或提示工程。在眼科基准测试FLAIR + FIVES上，PROTON在协变量变化上将MCM提高了+23.9 AUROC，在语义变化上提高了+8.8，在远离OOD上提高了+8.1，使其成为唯一一种在没有层次提示或标记数据的情况下改善所有三者的零样本方法。代码可在 https://github.com/GenMI-Lab/PROTON 获取，项目页面可在 https://genmi-lab.github.io/PROTON 访问。
cs.CV / 43 / 2606.20919
GIM-ENDO: A Multimodal Endoscopic Image and Video Dataset for Gastric Intestinal Metaplasia Morphology and Pathology
GIM-ENDO：用于胃肠上皮化生形态和病理的多模态内窥镜图像与视频数据集
Abstract
Gastric intestinal metaplasia (GIM) is a precursor lesion to gastric dysplasia and adenocarcinoma whose early detection is crucial for intervening in the carcinogenesis cascade. Artificial intelligence (AI) holds considerable promise for real-time endoscopic detection and characterization of GIM. However, development of reliable AI models has been constrained by the absence of publicly available, histopathologically validated datasets that combine detailed endoscopic annotations, histological subtype (complete and incomplete), standardized grading systems, and normal mucosal patterns. GIM-ENDO was designed to fill this gap. The dataset comprises demographic data, endoscopic findings, histopathological results, and H. pylori status acquired using the Olympus EVIS X1 system with white-light endoscopy (WLE) and image-enhanced endoscopy (IEE), including narrow-band imaging (NBI) and magnifying NBI (M-NBI), along with images and video clips from 24 patients (22 GIM-positive, 2 normal controls). Annotations cover six primary IEE endoscopic signs -- light blue crest (LBC), marginal turbid band (MTB), white opaque substance (WOS), TV pattern (Fusion), atrophy, and map-like erythema (MLE) -- plus two additional endoscopic findings (AHP and GA) recorded where present. GIM subtypes (complete and incomplete) are annotated for all GIM-positive cases; OLGA and OLGIM staging are provided where complete histological sampling was available. The dataset is publicly accessible at https://doi.org/10.5281/zenodo.20707267. For the latest updates and further information regarding this dataset, readers are referred to the DataBioX website: https://databiox.com A short version of this work has been submitted to MICCAI 2026 Open Data Track.
Chinese Translation
胃肠上皮化生（GIM）是胃异型增生和腺癌的前驱病变，其早期检测对干预癌变级联至关重要。人工智能（AI）在实时内窥镜检测和表征GIM方面具有相当大的潜力。然而，可靠的AI模型的开发受到缺乏公开可用的、经过组织病理学验证的数据集的限制，这些数据集结合了详细的内窥镜注释、组织学亚型（完全型和不完全型）、标准化分级系统以及正常黏膜模式。GIM-ENDO旨在填补这一空白。该数据集包括人口统计数据、内窥镜发现、组织病理学结果和幽门螺杆菌（H. pylori）状态，使用奥林巴斯EVIS X1系统通过白光内窥镜（WLE）和图像增强内窥镜（IEE）获取，包括窄带成像（NBI）和放大NBI（M-NBI），以及来自24名患者（22例GIM阳性，2例正常对照）的图像和视频片段。注释涵盖六个主要的IEE内窥镜标志——浅蓝色脊（LBC）、边缘浑浊带（MTB）、白色不透明物质（WOS）、TV模式（融合）、萎缩和地图样红斑（MLE），以及记录的两个额外内窥镜发现（AHP和GA）。所有GIM阳性病例均标注了GIM亚型（完全型和不完全型）；在可获得完整组织学取样的情况下，提供了OLGA和OLGIM分期。该数据集可在https://doi.org/10.5281/zenodo.20707267公开获取。有关此数据集的最新更新和更多信息，请访问DataBioX网站：https://databiox.com。本研究的简短版本已提交至MICCAI 2026开放数据轨道。
cs.CV / 44 / 2606.20924
ELDiff: When Evidential Learning Meets Text-to-Image Diffusion
ELDiff：当证据学习遇上文本到图像扩散
Abstract
In multi-object text-to-image (T2I) diffusion, ensuring semantic consistency between textual prompts and generated visual content is crucial for image synthesis. However, such consistency constraint is often underemphasized in the denoising process of diffusion models. Although token supervised diffusion models can mitigate this issue by learning object-wise consistency between the image content and object segmentation maps, it tends to suffer from the problems of segmentation map bias and semantic overlap conflict, especially when involving multiple objects. In this paper, we propose ELDiff, a new evidential learning-supervised T2I diffusion model, which leverages the advantages of uncertainty metric and conflict detection to enhance the fault tolerance of unreliable segmentation maps and suppress semantic conflicts, strengthening object-wise consistency learning. Specifically, a pixel evidence loss is proposed to restrain overconfidence in unreliable labels through evidential regularization, and a token conflict loss is designed to weaken the contradiction between semantics through optimizing a measured conflict factor. Extensive experiments show that our ELDiff outperforms existing training based and train-free based T2I diffusion models on SD v1.4, SD v2.1, SDXL, SD v3.5, and Qwen-Image, without requiring additional inference-time manipulations. Notably, ELDiff can be seamlessly extended to the existing training pipeline of T2I diffusion models. Code can be found at https://github.com/QingtaoPan/ELDiff.
Chinese Translation
在多对象文本到图像（T2I）扩散中，确保文本提示与生成视觉内容之间的语义一致性对于图像合成至关重要。然而，在扩散模型的去噪过程中，这种一致性约束往往被低估。尽管基于标记监督的扩散模型可以通过学习图像内容与对象分割图之间的对象级一致性来缓解这一问题，但在涉及多个对象时，它往往会遭遇分割图偏差和语义重叠冲突的问题。本文提出了ELDiff，一种新的证据学习监督的T2I扩散模型，利用不确定性度量和冲突检测的优势来增强对不可靠分割图的容错能力，并抑制语义冲突，从而加强对象级一致性学习。具体而言，提出了一种像素证据损失，通过证据正则化来限制对不可靠标签的过度自信，并设计了一种标记冲突损失，通过优化测量的冲突因子来减弱语义之间的矛盾。大量实验表明，我们的ELDiff在SD v1.4、SD v2.1、SDXL、SD v3.5和Qwen-Image上优于现有的基于训练和无训练的T2I扩散模型，而无需额外的推理时间操作。值得注意的是，ELDiff可以无缝扩展到现有的T2I扩散模型训练流程中。代码可在 https://github.com/QingtaoPan/ELDiff 找到。
cs.CV / 45 / 2606.20970
CogniRoute: Learning to Route Social Evidence in Omni-Modal Models
CogniRoute：在全模态模型中学习社交证据的路由
Abstract
Omni-modal models can ingest video, audio, and text, but unified access to multiple modalities does not guarantee that a model uses the right evidence. This gap is especially pronounced in social video question answering, where the answer may hinge on a gesture, vocal tone, temporal cue, or mismatch between what is said and what is visually expressed. We introduce CogniRoute, a schema-guided Mixture-of-Experts framework for social omni reasoning. CogniRoute uses a training-only cognitive schema that factorizes each example by cross-modal relation, reasoning demand, and temporal scope, and aligns global routing signatures with this structure during supervised fine-tuning. We further introduce route-aware reinforcement learning, which jointly optimizes token generation and expert allocation using rewards for answer correctness, modality-consistent reasoning, and cognitive temporal grounding. To support training and evaluation, we construct OmniSocialBench, a diagnostic social video QA resource with 118K structured training examples, grounded reasoning traces, schema labels, temporal evidence spans, and a manually verified evaluation split. CogniRoute achieves 59.38\% average accuracy on OmniSocialBench, improving over the strongest proprietary baseline by 15.33 percentage points and the strongest open-source omni baseline by 26.77 points, with the largest gains on questions requiring audio-visual coordination, conflict resolution, and temporally grounded social inference.
Chinese Translation
全模态模型能够处理视频、音频和文本，但对多种模态的统一访问并不能保证模型使用正确的证据。这一差距在社交视频问答中尤为明显，因为答案可能依赖于手势、语调、时间线索，或所说内容与视觉表达之间的不匹配。我们提出了CogniRoute，一种基于模式引导的专家混合框架，用于社交全模态推理。CogniRoute使用一种仅在训练时使用的认知模式，通过跨模态关系、推理需求和时间范围对每个示例进行分解，并在监督微调期间将全局路由签名与该结构对齐。我们进一步引入了路由感知强化学习，该方法通过对答案正确性、一致的模态推理和认知时间基础的奖励，联合优化标记生成和专家分配。为了支持训练和评估，我们构建了OmniSocialBench，这是一个包含118K结构化训练示例、基础推理轨迹、模式标签、时间证据范围和手动验证评估分割的诊断社交视频问答资源。CogniRoute在OmniSocialBench上实现了59.38%的平均准确率，比最强的专有基线提高了15.33个百分点，比最强的开源全模态基线提高了26.77个百分点，尤其在需要音视频协调、冲突解决和时间基础社交推理的问题上取得了最大的提升。
cs.CV / 46 / 2606.20971
UNITY: Attention Flow Networks for Adaptive Conditioning in Diffusion
UNITY：用于扩散中自适应条件的注意力流网络
Abstract
We introduce UNITY, a Universal-to-Specialized adapter for efficient and scalable composite conditioning in diffusion based image generation. Unlike prior methods that train separate adapters for each conditioning modality, UNITY jointly learns shared semantics across multiple conditioning types and subsequently specializes without modifying the underlying architecture. The proposed two stage training paradigm consists of a Universal Stage that captures cross modal representations across all conditioning modalities using half of the total training steps, followed by a Specialization Stage that refines modality specific features using the remaining training budget. At the core of UNITY are the Morphable Attention Flow (MAF) Network and Morph Wrapper modules, which enable channel aware and spatially adaptive feature alignment through learnable flow fields and attention based fusion. This constant complexity formulation supports flexible operation under both single and composite conditioning settings while significantly reducing inference latency and memory consumption. Extensive experiments across multiple datasets demonstrate that UNITY achieves state of the art image fidelity while maintaining superior memory efficiency. Code: https://github.com/arya-domain/UNITY
Chinese Translation
我们提出了UNITY，一种用于高效且可扩展的复合条件的通用到专用适配器，应用于基于扩散的图像生成。与以往为每种条件模态训练单独适配器的方法不同，UNITY通过联合学习多种条件类型之间的共享语义，随后在不修改基础架构的情况下进行专门化。所提出的两阶段训练范式包括一个通用阶段，该阶段利用总训练步骤的一半捕捉所有条件模态之间的跨模态表示，随后是一个专门化阶段，该阶段利用剩余的训练预算细化模态特定特征。UNITY的核心是可变形注意力流（Morphable Attention Flow, MAF）网络和形态包装模块，这些模块通过可学习的流场和基于注意力的融合实现通道感知和空间自适应特征对齐。这种恒定复杂度的公式支持在单一和复合条件设置下的灵活操作，同时显著降低推理延迟和内存消耗。在多个数据集上的广泛实验表明，UNITY在保持优越的内存效率的同时，实现了最先进的图像保真度。代码： https://github.com/arya-domain/UNITY
cs.CV / 47 / 2606.20980
Robusto-2: Benchmarking Humans & VLMs for Autonomous Driving in Lima & New York City
Robusto-2：在利马和纽约市对人类与视觉语言模型进行自动驾驶基准测试
Abstract
As Self-Driving Cars continue to expand internationally and use multi-modal systems such as VLMs as a cognitive backbone for their Action models; how well will these systems generalize in new settings, in particular out-of-distribution (OOD) edge-case scenarios in new geographies? In this paper, we study this open question by providing a full factorial analysis with human drivers of Lima, human drivers from New York City, and VLMs and showing them dashcam footage collected from Lima and New York City -- prompting them with a variety of questions under a Visual Question Answering (VQA) paradigm. In particular, we pick these two cities as they are highly challenging driving locations where no Self-Driving Car company currently operates in, and ask questions that span 4 categories: Factual, Ratings, Counterfactual and Reasoning. We find that Humans and VLMs diverge in their responses -- though this is modulated by the type of questions asked, and that Humans answer similarly independent of where they are from (Lima/NYC). To our surprise, we did not find a strong difference in terms of answers (Humans or VLMs) that was modulated by geography, likely due to their high out-of-distribution nature. Our dataset is available at: https://huggingface.co/datasets/Artificio/robusto-2
Chinese Translation
随着自动驾驶汽车在国际上的不断扩展，并使用视觉语言模型（VLMs）作为其行动模型的认知基础，这些系统在新环境中的泛化能力如何，特别是在新的地理区域中的分布外（OOD）边缘案例场景？本文通过对来自利马的驾驶员、来自纽约市的驾驶员以及视觉语言模型（VLMs）进行全面的因子分析，研究了这一开放性问题，并向他们展示了在利马和纽约市收集的行车记录仪视频，同时在视觉问答（VQA）范式下提出了多种问题。我们选择这两个城市作为研究对象，因为它们是当前没有任何自动驾驶汽车公司运营的高度挑战性驾驶地点，并提出了涵盖事实、评分、反事实和推理四个类别的问题。我们发现人类和视觉语言模型在回答上存在差异，尽管这种差异受到提问类型的调节，而人类的回答则与其来源地（利马/纽约市）无关。令我们惊讶的是，我们没有发现地理位置对答案（人类或视觉语言模型）有显著影响，这可能是由于它们的高度分布外特性。我们的数据集可在以下链接获取：https://huggingface.co/datasets/Artificio/robusto-2
cs.CV / 48 / 2606.21020
CheXpercept: A Benchmark for Evaluating Expert-Level Lesion Perception in Chest X-rays
CheXpercept：评估胸部X光片中专家级病变感知的基准
Abstract
The evaluation of vision-language models (VLMs) for chest X-ray (CXR) analysis has largely been limited to disease-presence classification without visual grounding. Such evaluations fail to verify the expert-level lesion perception necessary to ensure the clinical reliability of VLMs. To address these limitations, we introduce CheXpercept, a sequential, multi-level perception benchmark that mirrors a radiologist's cognitive workflow across coarse-level detection, fine-level contour evaluation and revision, and semantic-level attribute extraction. To ensure high clinical fidelity at scale, we construct the dataset using a semi-automated generation pipeline paired with a review by six medical experts. CheXpercept contains 10,400 QA items derived from 2,100 CXRs, covering seven clinically critical pulmonary and cardiac lesions. To demonstrate the current landscape of VLM perception, we benchmark 14 general and medical VLMs on CheXpercept. The models achieve adequate performance only at the coarse level, with accuracy degrading precipitously on deeper visual tasks. Notably, medical VLMs show almost no perceptual advantage over their general-domain counterparts, highlighting a systemic flaw in current domain adaptation. The code and dataset will be publicly available.
Chinese Translation
对胸部X光片（CXR）分析的视觉-语言模型（VLMs）评估在很大程度上局限于疾病存在分类，而没有进行视觉基础的验证。这种评估未能验证确保VLMs临床可靠性所需的专家级病变感知。为了解决这些局限性，我们引入了CheXpercept，这是一个顺序的多层次感知基准，反映了放射科医生在粗略检测、细致轮廓评估与修订以及语义属性提取等认知工作流程中的过程。为了确保在大规模下的高临床保真度，我们使用半自动化生成管道构建数据集，并由六位医学专家进行审核。CheXpercept包含来自2100个CXR的10400个问答项目，涵盖七种临床关键的肺部和心脏病变。为了展示当前VLM感知的现状，我们在CheXpercept上对14个通用和医学VLM进行了基准测试。这些模型在粗略层面上表现出足够的性能，但在更深层次的视觉任务上准确率急剧下降。值得注意的是，医学VLM几乎没有相对于通用领域模型的感知优势，突显了当前领域适应中的系统性缺陷。代码和数据集将公开发布。
cs.CV / 49 / 2606.21026
Sparse Point-Guided Fusion of Supervised and Self-Supervised Learning Model for Seaweed Segmentation
稀疏点引导的监督学习与自监督学习模型融合用于海藻分割
Abstract
The ocean plays a critical role in sustainable development, particularly in climate change mitigation. Among marine ecosystems, blue carbon ecosystems are recognized as important natural carbon sinks. In this context, this paper addresses precise seaweed classification for blue carbon quantification in Ocean Digital Twin initiatives. Conventional methods, including supervised learning (limited by data scarcity and domain gaps) and self-supervised learning (unable to assign class labels), struggle with underwater complexities and diverse seaweed species. To overcome this, we propose a novel two-stage seaweed segmentation technique. This technique first utilizes Supervised and Self-supervised Learning Model Propagation (SSL.Prop.), which leverages supervised learning for initial class information and approximate locations, guiding self-supervised learning for detailed, accurate segmentation. Subsequently, MaskFusion (MF) refines these results by merging instance-level masks for highly accurate segmentation. This integrated approach allows automatic class label assignment and mitigates domain gap effects. Specifically, instance segmentation estimates sparse point locations which then guide self-supervised learning for detailed region segmentation. Evaluated with underwater images from Yamaguchi Prefecture, our full proposed method (SSL.Prop.+MF) achieved a 0.068 mIoU improvement over USIS-SAM, demonstrating significant accuracy gains, particularly for small seaweed. This approach demonstrates strong potential for improving blue carbon quantification and marine ecosystem monitoring.
Chinese Translation
海洋在可持续发展中扮演着关键角色，特别是在气候变化缓解方面。在海洋生态系统中，蓝碳生态系统被认为是重要的自然碳汇。在此背景下，本文针对海洋数字双胞胎计划中的蓝碳量化提出了精确的海藻分类方法。传统方法，包括监督学习（受限于数据稀缺和领域差距）和自监督学习（无法分配类别标签），在水下复杂环境和多样化海藻物种面前面临挑战。为了解决这一问题，我们提出了一种新颖的两阶段海藻分割技术。该技术首先利用监督与自监督学习模型传播（Supervised and Self-supervised Learning Model Propagation, SSL.Prop.），该方法利用监督学习获取初步的类别信息和近似位置，指导自监督学习进行详细、准确的分割。随后，掩膜融合（MaskFusion, MF）通过合并实例级掩膜来优化这些结果，实现高精度的分割。这种集成方法允许自动类别标签分配，并减轻领域差距的影响。具体而言，实例分割估计稀疏点位置，然后引导自监督学习进行详细区域分割。通过对来自山口县的水下图像进行评估，我们提出的完整方法（SSL.Prop.+MF）在mIoU上比USIS-SAM提高了0.068，显示出显著的精度提升，特别是在小型海藻的分割上。这种方法展示了改善蓝碳量化和海洋生态系统监测的强大潜力。
cs.CV / 50 / 2606.21027
Self-Supervised Dual-Frequency Phase Decomposition for Single-Shot Composite Fringe Projection Profilometry
自监督双频相位分解用于单次复合条纹投影轮廓测量
Abstract
Single-shot fringe projection profilometry (FPP) has been actively studied for real-time measurement, dynamic object reconstruction, and motion-sensitive environments. Composite fringe patterns are advantageous in single-shot FPP because multiple frequency components can be encoded in a single pattern, enabling phase ambiguity resolution. Existing approaches mainly rely on Fourier transform-based methods or supervised deep learning methods. However, Fourier transform-based methods often suffer from limited accuracy and degraded performance in complex regions, while supervised methods require dense phase or depth labels, which are costly to obtain. In this work, we propose a self-supervised phase refinement framework for single-shot composite fringe patterns without requiring phase or depth labels. The proposed method exploits the scale and direction relationships between low- and high-frequency phase gradients, improving the reliability of phase separation. We also introduce a soft edge consistency loss to preserve object boundaries and fine geometric structures. Experimental results show that the proposed method achieves MAE_z and RMSE_z of 0.367 mm and 1.804 mm, respectively, outperforming the best-performing transform-based baseline, which obtains 0.402 mm and 2.785 mm. The proposed method also improves the valid-pixel ratio from 84.75 % to 95.07 %. These results demonstrate the effectiveness of self-supervised dual-frequency phase refinement for reliable single-shot 3D reconstruction without ground-truth label supervision.
Chinese Translation
单次条纹投影轮廓测量（FPP）在实时测量、动态物体重建和运动敏感环境中得到了广泛研究。复合条纹模式在单次 FPP 中具有优势，因为多个频率成分可以编码在单一模式中，从而实现相位模糊的解决。现有方法主要依赖于基于傅里叶变换的方法或监督深度学习方法。然而，基于傅里叶变换的方法在复杂区域常常面临准确性有限和性能下降的问题，而监督方法则需要密集的相位或深度标签，这些标签获取成本高昂。在本研究中，我们提出了一种自监督相位精炼框架，适用于单次复合条纹模式，无需相位或深度标签。所提方法利用低频和高频相位梯度之间的尺度和方向关系，提高了相位分离的可靠性。我们还引入了一种软边缘一致性损失，以保持物体边界和精细几何结构。实验结果表明，所提方法的 MAE_z 和 RMSE_z 分别为 0.367 mm 和 1.804 mm，优于表现最佳的基于变换的基线，其结果为 0.402 mm 和 2.785 mm。所提方法还将有效像素比例从 84.75% 提高到 95.07%。这些结果证明了自监督双频相位精炼在无需真实标签监督的情况下实现可靠单次 3D 重建的有效性。
cs.CV / 51 / 2606.21061
Neural Architecture Distributions: A New Paradigm for Stochastic Segmentation
神经架构分布：随机分割的新范式
Abstract
Stochastic segmentation seeks to represent multiple plausible masks for a single image, which is essential in safety- and quality-critical applications such as medical imaging or building defect inspection. Most existing methods introduce stochasticity by injecting continuous latent variables or by iterative denoising trajectories, whose stochastic sources are difficult to search or audit directly. We propose architecture distributions as a new stochastic source for segmentation: instead of sampling a latent variable or noise, we sample a discrete architecture from a learned distribution over operator choices at multiple searchable positions in a segmentation backbone. Each sampled architecture yields one mask through the selected active path, so inference depends on the executed subnet rather than the complete candidate bank. This approach also supports architectural provenance, since each output corresponds to a specific architecture configuration. To reduce collapse toward averaged masks, we train with set-level supervision by matching a set of architecture-sampled predictions to the annotation set using an IoU-based energy-distance surrogate. We further construct the candidate bank with evolutionary search, making the support of the stochastic source optimizable before distribution learning. The proposed method achieves state-of-the-art distribution matching and hypothesis coverage on LIDC-IDRI, and remains effective on two extension tasks. To the best of our knowledge, this is the first work to formulate stochastic segmentation as learning an architecture distribution and realizing output diversity through architecture sampling.
Chinese Translation
随机分割旨在为单幅图像表示多个合理的掩膜，这在医疗成像或建筑缺陷检查等安全和质量关键的应用中至关重要。现有的大多数方法通过注入连续潜变量或迭代去噪轨迹引入随机性，而其随机源难以直接搜索或审计。我们提出将架构分布作为分割的新随机源：我们不是从潜变量或噪声中采样，而是从在分割主干中多个可搜索位置上学习的操作选择分布中采样一个离散架构。每个采样的架构通过选择的活跃路径生成一个掩膜，因此推理依赖于执行的子网络，而不是完整的候选库。这种方法还支持架构来源，因为每个输出对应于特定的架构配置。为了减少向平均掩膜的崩溃，我们通过将一组架构采样的预测与注释集进行匹配，使用基于IoU的能量距离替代物进行集级监督训练。我们进一步通过进化搜索构建候选库，使得在分布学习之前可以优化随机源的支持。所提出的方法在LIDC-IDRI上实现了最先进的分布匹配和假设覆盖，并在两个扩展任务上保持有效。据我们所知，这是首次将随机分割形式化为学习架构分布，并通过架构采样实现输出多样性的研究。
cs.CV / 52 / 2606.21099
ShuffleFlow: Scalable Posterior Inference for Bayesian Inverse Imaging
ShuffleFlow：可扩展的贝叶斯逆成像后验推断
Abstract
Variational inference (VI) is a powerful method for principled posterior inference for scientific inverse imaging. VI learns the posterior distribution, often with a flow-based network, which can cheaply generate posterior samples upon optimization, and can flexibly incorporate score-based or classic priors. However, its application to large-scale image reconstruction is severely hindered by the poor scalability of the flow-based networks. In this work, we introduce ShuffleFlow, a scalable VI framework to address this challenge. Our method breaks down the problem into three parts: a pixel-unshuffling-based image coordinate sampler, a neural field as feature encoder, and a conditional normalizing flow (CNF) as posterior estimator. Specifically, our framework partitions an image into a stack of sub-images with pixel-unshuffling and uses a shared CNF to model the joint distribution of the sub-image stack. We condition the CNF on the output of a neural field, which embeds feature vectors corresponding to pixel-unshuffling sample locations to capture spatial structures, and share the flow's latent variable across the channels to model their correlations. We demonstrate our method's effectiveness and efficiency on both linear and nonlinear imaging inverse problems, and show its ability to more rapidly generate a high-sample-count posterior than diffusion samplers.
Chinese Translation
变分推断（Variational Inference, VI）是一种用于科学逆成像的原则性后验推断的强大方法。VI通过基于流的网络学习后验分布，该网络在优化后可以廉价地生成后验样本，并且可以灵活地结合基于评分的或经典的先验。然而，其在大规模图像重建中的应用受到基于流的网络可扩展性差的严重制约。在本研究中，我们提出了ShuffleFlow，一种可扩展的VI框架，以应对这一挑战。我们的方法将问题分解为三个部分：基于像素重排的图像坐标采样器、作为特征编码器的神经场，以及作为后验估计器的条件归一化流（Conditional Normalizing Flow, CNF）。具体而言，我们的框架将图像划分为一堆子图像，通过像素重排，并使用共享的CNF来建模子图像堆的联合分布。我们将CNF的条件设置为神经场的输出，该输出嵌入与像素重排采样位置对应的特征向量，以捕捉空间结构，并在通道间共享流的潜变量以建模它们的相关性。我们在线性和非线性成像逆问题上展示了我们方法的有效性和效率，并显示其比扩散采样器更快速地生成高样本量后验的能力。
cs.CV / 53 / 2606.21108
SARIF: Segment Anything for Robust Image Forensics
SARIF：用于鲁棒图像取证的任意分割
Abstract
Image forgery localization remains challenging due to diverse manipulation techniques and distribution shifts. Existing forgery localization models achieve high accuracy on benchmarks but often struggle with cross-domain generalization and robustness. In this paper, we propose SARIF (Segment Anything for Robust Image Forensics), a framework that leverages the Segment Anything Model (SAM), which has a promptable architecture and strong generalization ability. SARIF introduces a feedback-guided mask decoder and a dual-encoder design that extracts forgery-specific information to capture forensic traces while exploiting the SAM architecture. To localize manipulated regions, we design a block-wise prompting mechanism that derives forgery-specific cues from residual features between an adapted encoder and its frozen counterpart. These features are fused with the previous mask prompt to drive a feedback-based mask refinement process, enabling automatic forgery segmentation without manual input. Extensive experiments on standard forgery-localization benchmarks show that SARIF achieves strong average cross-dataset performance and robustness to common image corruptions.
Chinese Translation
图像伪造定位仍然面临挑战，因为存在多样的操作技术和分布变化。现有的伪造定位模型在基准测试中取得了高准确率，但在跨域泛化和鲁棒性方面常常表现不佳。本文提出了SARIF（Segment Anything for Robust Image Forensics），一个利用了具有可提示架构和强泛化能力的Segment Anything Model（SAM）的框架。SARIF引入了一种反馈引导的掩膜解码器和双编码器设计，提取伪造特定信息以捕捉取证痕迹，同时利用SAM架构。为了定位被操纵的区域，我们设计了一种基于块的提示机制，从适应性编码器与其冻结对应物之间的残差特征中提取伪造特定线索。这些特征与先前的掩膜提示融合，以驱动基于反馈的掩膜精炼过程，实现无需人工输入的自动伪造分割。在标准伪造定位基准上的大量实验表明，SARIF在跨数据集的平均性能和对常见图像损坏的鲁棒性方面表现出色。
cs.CV / 54 / 2606.21113
Object-Centric Dataset Resources for Constrained-Data Image Generation and Augmentation
面向对象的数据集资源用于受限数据的图像生成与增强
Abstract
Object-centric image generation is important in settings with few labeled examples, including pedestrian analysis in smart-city scenes, traffic-sign inspection, and domain-specific object detection. Synthetic images are most useful for training and evaluation when datasets preserve object structure, bounding boxes, visual diversity, and realistic context. Existing image datasets usually target classification, detection, or scene understanding rather than controlled object-centric generation and augmentation with limited class-specific data. We present a shareable collection of three object-centric dataset resources: Cityscapes-Pedestrian, TrafficSigns, and COCO PottedPlant. The collection standardizes 256-by-256 object-centric crops and bounding-box annotations across three regimes: dense pedestrian scenes with privacy blur and occlusion, cleaner high-contrast traffic signs, and context-diverse potted-plant scenes. The release contains 3,009 TrafficSigns samples, 2,156 Cityscapes-Pedestrian manifest records, and 7,679 COCO PottedPlant manifest records. The larger COCO-derived manifest preserves contextual and multi-instance diversity, while equal-size subsets can be drawn with a fixed random seed for controlled comparisons. The release provides direct TrafficSigns data where redistribution is permitted, together with scripts, manifests, box-level annotation tables, checksums, and reconstruction documentation for the Cityscapes- and COCO-derived subsets. It is available through the Latzi/object-centric-low-data-datasets GitHub repository and Zenodo DOI 10.5281/zenodo.20573001. The collection supports label and split inspection, subset creation, reconstruction from upstream data, and evaluation of object-centric image generation or synthetic-data augmentation methods on shared records.
Chinese Translation
面向对象的图像生成在标注样本稀少的环境中至关重要，包括智能城市场景中的行人分析、交通标志检查和特定领域的物体检测。当数据集保留物体结构、边界框、视觉多样性和真实上下文时，合成图像对训练和评估最为有用。现有的图像数据集通常针对分类、检测或场景理解，而非在有限类别特定数据下进行受控的面向对象生成和增强。我们提出了一套可共享的三种面向对象的数据集资源：Cityscapes-Pedestrian、TrafficSigns和COCO PottedPlant。该集合标准化了256x256的面向对象裁剪图和边界框注释，涵盖三种场景：具有隐私模糊和遮挡的密集行人场景、干净的高对比度交通标志，以及上下文多样的盆栽植物场景。此次发布包含3,009个TrafficSigns样本、2,156个Cityscapes-Pedestrian清单记录和7,679个COCO PottedPlant清单记录。较大的COCO衍生清单保留了上下文和多实例的多样性，同时可以使用固定随机种子抽取相同大小的子集以进行受控比较。此次发布提供了直接的TrafficSigns数据，允许重新分发，同时附带脚本、清单、框级注释表、校验和及Cityscapes和COCO衍生子集的重建文档。该资源可通过Latzi/object-centric-low-data-datasets GitHub库和Zenodo DOI 10.5281/zenodo.20573001获取。该集合支持标签和分割检查、子集创建、从上游数据重建，以及对共享记录的面向对象图像生成或合成数据增强方法的评估。
cs.CV / 55 / 2606.21115
MS-rPPG: Multi-spectral State Space Model for Remote Photoplethysmography in Driver Monitoring Systems
MS-rPPG：用于驾驶员监测系统的多光谱状态空间模型远程光电容积描记法
Abstract
Remote photoplethysmography (rPPG) is a camera-based technique for measuring physiological signals, particularly cardiac activity. From the remotely measured signals, heart rate can be estimated, which is crucial for health monitoring. In this study, we investigate a driver health monitoring system based on remote heart rate estimation. However, driving environments represent uncontrolled settings where videos are subject to varying illumination conditions and frequent head movements. We introduce MS-rPPG, a multi-spectral framework that combines RGB with near-infrared (NIR) face video to alleviate rPPG estimation under challenging driving conditions. To combine the complementary features from two spectral videos, we propose a cross-spectral linear modulation (CSLM) strategy based on frequency-domain analysis. Moreover, we introduce MS-Mamba, a novel state space model designed to effectively model long-range temporal dependencies while jointly capturing cross-channel interactions between multi-spectral features. We collected a real-world dataset called MS-Drive, which was recorded from 50 participants while driving the vehicle. The proposed method was evaluated on the MR-NIRP Car dataset and MS-Drive datasets. The experimental results indicate that MS-rPPG shows better robustness and heart rate estimation accuracy than previous methods, highlighting its promise for driver health monitoring. The codes are available at github.com/ziiho08/MS-rPPG.
Chinese Translation
远程光电容积描记法（rPPG）是一种基于摄像头的生理信号测量技术，特别用于心脏活动的监测。通过远程测量的信号，可以估算心率，这对于健康监测至关重要。在本研究中，我们探讨了一种基于远程心率估算的驾驶员健康监测系统。然而，驾驶环境代表了不受控制的设置，视频受到不同照明条件和频繁头部运动的影响。我们提出了MS-rPPG，这是一种多光谱框架，将RGB与近红外（NIR）面部视频相结合，以减轻在挑战性驾驶条件下的rPPG估算。为了结合两种光谱视频的互补特征，我们提出了一种基于频域分析的跨光谱线性调制（CSLM）策略。此外，我们引入了MS-Mamba，一种新颖的状态空间模型，旨在有效建模长距离时间依赖性，同时共同捕捉多光谱特征之间的跨通道交互。我们收集了一个名为MS-Drive的真实世界数据集，该数据集是在50名参与者驾驶车辆时记录的。所提出的方法在MR-NIRP Car数据集和MS-Drive数据集上进行了评估。实验结果表明，MS-rPPG在鲁棒性和心率估算准确性方面优于之前的方法，突显了其在驾驶员健康监测中的潜力。代码可在github.com/ziiho08/MS-rPPG获取。
cs.CV / 56 / 2606.21116
ConnectomeBench2: A Unified Benchmark for Automated Connectomic Proofreading
ConnectomeBench2：自动化连接组校对的统一基准
Abstract
Proofreading--correcting segmentation errors in 3D brain reconstructions--is the rate-limiting step in synapse-resolution connectomics. We release ConnectomeBench2, a unified multi-species dataset of over 716,485 expert-labeled proofreading decisions with >4,500,000 associated images spanning four major open connectomes (mouse, human, zebrafish, fly), spanning both split and merge error correction. Trained on this dataset, a single Vision Transformer with shared encoders for mesh geometry and electron microscopy reaches human-level accuracy across species for split error correction and merge error identification, with performance scaling with data size and modality. Beyond accuracy, we show that the model is well-calibrated within distribution, that measures of distribution distance predict where calibration and accuracy will degrade on unseen data, and that connectomics-specific pretraining and active learning-based sample selection show potential to substantially reduce the labeling effort needed to extend to new species and brain regions. The benchmark provides the infrastructure to train and evaluate increasingly capable vision models for connectomic proofreading. Data and code availability. The ConnectomeBench2 dataset is released on Hugging Face at https://huggingface.co/datasets/jeffbbrown2/ConnectomeBench2. The accompanying codebase is available on GitHub at https://github.com/timfarkas/ConnectomeBench2.
Chinese Translation
校对——纠正三维大脑重建中的分割错误——是突触分辨率连接组学中的限制步骤。我们发布了ConnectomeBench2，这是一个统一的多物种数据集，包含超过716,485个专家标注的校对决策，伴随超过4,500,000张图像，涵盖四个主要的开放连接组（小鼠、人类、斑马鱼、果蝇），涉及分割和合并错误的纠正。在该数据集上训练的单个视觉变换器（Vision Transformer）使用共享编码器处理网格几何和电子显微镜，达到了跨物种的分割错误纠正和合并错误识别的人类水平准确性，且性能随着数据规模和模态的增加而提升。除了准确性，我们还表明该模型在分布内具有良好的校准性，分布距离的度量可以预测在未见数据上校准和准确性将下降的地方，并且连接组学特定的预训练和基于主动学习的样本选择显示出显著减少扩展到新物种和脑区所需标注工作的潜力。该基准提供了训练和评估越来越强大的视觉模型用于连接组校对的基础设施。数据和代码可用性：ConnectomeBench2数据集已在Hugging Face上发布，链接为https://huggingface.co/datasets/jeffbbrown2/ConnectomeBench2。相关代码库可在GitHub上找到，链接为https://github.com/timfarkas/ConnectomeBench2。
cs.CV / 57 / 2606.21119
MammoExpert: Benchmarking Chain-of-Thought Reasoning in Mammography Diagnosis
MammoExpert：乳腺X线检查诊断中的思维链推理基准测试
Abstract
Mammography is an essential tool for breast cancer detection, with millions of examinations conducted annually. However, publicly available high-quality mammography datasets for AI development remain limited in both scale and annotation richness, particularly regarding pathological subtype coverage and structured diagnostic reasoning annotations. In this paper, we present MammoExpert, the first mammography dataset with Chain-of-Thought reasoning annotations across three diagnostic phases: (i) primal observation, (ii) factual assessment, and (iii) diagnostic synthesis. Comprising 2,379 mammography images covering 67 WHO-classified histopathology subtypes, each exam provides 42 radiographic features annotated by nine senior radiologists. We evaluate its performance on the breast lesion classification task, demonstrating superior accuracy and reasonability compared to existing classification models. Combining public dataset CBIS-DDSM with MammoExpert yields 7.1\% classification accuracy improvement, while the training model to learn CoT reasoning achieves another 4\% gain on the MammoExpert test set. Similar improvements are observed on INBreast and Vindr datasets, where the full approach yields accuracy gains of 6.9\% and 6.7\%, respectively. MammoExpert can serve as a benchmark for interpretable breast lesion diagnosis through explicit CoT reasoning.
Chinese Translation
乳腺X线检查是乳腺癌检测的重要工具，每年进行数百万次检查。然而，公开可用的高质量乳腺X线检查数据集在规模和注释丰富性方面仍然有限，特别是在病理亚型覆盖和结构化诊断推理注释方面。本文提出了MammoExpert，这是第一个在三个诊断阶段（即： (i) 原始观察，(ii) 事实评估，(iii) 诊断综合）中具有思维链推理注释的乳腺X线检查数据集。该数据集包含2,379幅涵盖67种世界卫生组织（WHO）分类的组织病理亚型的乳腺X线图像，每次检查提供42个由九位资深放射科医师注释的放射特征。我们在乳腺病变分类任务上评估了其性能，结果显示与现有分类模型相比，具有更高的准确性和合理性。将公共数据集CBIS-DDSM与MammoExpert结合，分类准确率提高了7.1\%，而训练模型学习思维链推理在MammoExpert测试集上又获得了4\%的提升。在INBreast和Vindr数据集上也观察到了类似的改进，完整方法分别实现了6.9\\%和6.7\\%的准确率提升。MammoExpert可以作为通过明确的思维链推理进行可解释的乳腺病变诊断的基准。
cs.CV / 58 / 2606.21135
Odoriko: A Shape-Aware Multimodal Diffusion Framework for Human Motion
Odoriko：一种形状感知的多模态扩散框架用于人类运动
Abstract
Human motion generation has been widely studied across diverse input modalities, text, music, and video, and recent efforts have unified these into single multimodal frameworks. However, while morphological factors such as gender and body shape are known to produce distinct kinematic signatures, no existing unified framework incorporates this into generation, treating all subjects as morphologically equivalent. We present Odoriko, the first unified multimodal motion generation framework that reflects subject bio-morphological information directly in synthesized motion output. Rather than averaging over subject variation, Odoriko generates motion that is consistent with who is moving, not just what they are asked to do, across text, music, and video conditions within a single model. When explicit morphological information is unavailable, Odoriko additionally recovers subject morphology alongside motion, unifying estimation and generation in one framework. Extensive experiments across text-to-motion, music-to-dance, and video-to-motion benchmarks demonstrate that Odoriko matches or exceeds prior specialized models on standard metrics, while enabling morphology-consistent generation that no existing unified framework supports.
Chinese Translation
人类运动生成在多种输入模态（如文本、音乐和视频）中得到了广泛研究，近期的努力将这些模态统一为单一的多模态框架。然而，尽管性别和体型等形态因素已知会产生不同的运动学特征，现有的统一框架并未将其纳入生成过程中，而是将所有对象视为形态上等同。我们提出了Odoriko，这是第一个统一的多模态运动生成框架，能够直接在合成的运动输出中反映对象的生物形态信息。Odoriko生成的运动不仅与运动者的身份一致，还与他们被要求执行的任务相符，适用于文本、音乐和视频条件下的单一模型。当显式的形态信息不可用时，Odoriko还能够在运动生成的同时恢复对象的形态，将估计与生成统一在一个框架中。针对文本到运动、音乐到舞蹈和视频到运动的基准测试进行的广泛实验表明，Odoriko在标准指标上与之前的专门模型相匹配或超越，同时实现了现有统一框架所不支持的形态一致生成。
cs.CV / 59 / 2606.21138
SEED: Simple ViT and Evolving Harness for Explainable Text Forgery Detection
SEED：用于可解释文本伪造检测的简单ViT和演变框架
Abstract
AI-assisted image editing threatens trust in financial, legal, and identity records. The GenText-Forensics Challenge at ACM MM 2026 addresses this by requiring structured forensic reports, in which integrating detection, pixel-level localization, and natural language explanation for multilingual text-centric forgery images. We present SEED, a modular system with three components. First, a similarity-guided pipeline augments training with diverse synthetic forgeries. Second, a single ViT, built on DINOv3 with LoRA adaptation, jointly performs detection and pixel-level localization while preserving pre-trained priors with minimal trainable parameters. Third, an evolving harness takes the detector's predictions and generates a complete forensic report via an MLLM, iteratively improved through a proposer-evaluator loop optimizing report quality. SEED ranked 3rd in the GenText-Forensics Challenge. Code and data are available at https://github.com/KahimWong/GenText-Forensics-3rd-Place.
Chinese Translation
人工智能辅助的图像编辑威胁着金融、法律和身份记录的信任。ACM MM 2026的GenText-Forensics挑战通过要求结构化的法医报告来应对这一问题，该报告需要将检测、像素级定位和多语言文本中心伪造图像的自然语言解释相结合。我们提出了SEED，一个由三个组件组成的模块化系统。首先，一个基于相似性的引导管道通过多样化的合成伪造品增强训练。其次，一个基于DINOv3并采用LoRA适应的单一ViT，能够在保留预训练先验的同时，以最小的可训练参数共同执行检测和像素级定位。第三，一个演变框架利用检测器的预测，通过一个多语言大模型（MLLM）生成完整的法医报告，并通过提议-评估循环迭代改进报告质量。SEED在GenText-Forensics挑战中获得第三名。代码和数据可在 https://github.com/KahimWong/GenText-Forensics-3rd-Place 获取。
cs.CV / 60 / 2606.21146
ChronoLock: Protecting Videos from Unauthorized Text-to-Video Personalization
ChronoLock：保护视频免受未经授权的文本到视频个性化
Abstract
Text-to-video (T2V) diffusion models have made it increasingly easy to synthesize realistic and temporally coherent videos, while recent personalization techniques allow such models to imitate a specific subject, style, or motion pattern from only a few reference clips. This capability creates a new data-misuse risk: videos shared online can be collected and used for unauthorized T2V fine-tuning. Existing protective perturbations are mainly designed for image recognition or text-to-image personalization, and therefore focus on corrupting static appearance cues rather than the temporal denoising dynamics that make video personalization possible. To address this gap, we introduce ChronoLock, the first proactive protection framework that makes released videos difficult to exploit for unauthorized T2V personalization. ChronoLock targets the motion-learning process directly by optimizing bounded perturbations over temporal denoising trajectories. It first disrupts intra-chunk temporal adaptation with a diffusion objective that combines fitting error, frame-relative denoising relations, and adjacent-frame variation, and then enlarges inter-chunk boundary mismatch to weaken long-range motion continuity. Transformation-sampled updates further improve robustness to common preprocessing operations.Experiments on UCF Sports and HMDB51 with popular T2V backbones and personalization scheme show that ChronoLock effectively reduces motion imitation under automatic metrics and human evaluation.
Chinese Translation
文本到视频（T2V）扩散模型使得合成逼真且时间上连贯的视频变得越来越容易，而最近的个性化技术允许这些模型仅通过少量参考剪辑模仿特定的主题、风格或运动模式。这种能力带来了新的数据滥用风险：在线共享的视频可能被收集并用于未经授权的T2V微调。现有的保护扰动主要针对图像识别或文本到图像个性化，因此侧重于破坏静态外观线索，而非使视频个性化成为可能的时间去噪动态。为了解决这一问题，我们提出了ChronoLock，这是第一个主动保护框架，使得发布的视频难以被用于未经授权的T2V个性化。ChronoLock直接针对运动学习过程，通过优化时间去噪轨迹上的有界扰动来实现。它首先通过一个结合拟合误差、帧相对去噪关系和相邻帧变化的扩散目标，破坏块内的时间适应性，然后扩大块间边界不匹配，以削弱长距离运动连续性。变换采样更新进一步提高了对常见预处理操作的鲁棒性。在UCF Sports和HMDB51数据集上，使用流行的T2V骨干网络和个性化方案的实验表明，ChronoLock在自动指标和人工评估下有效减少了运动模仿。
cs.CV / 61 / 2606.21156
Contrastive and Adaptive Multi-modal Masked Autoencoder for Spatial Transcriptomics
对比自适应多模态掩蔽自编码器用于空间转录组学
Abstract
The high cost of spatial transcriptomics (ST) has driven extensive studies into predicting gene expression directly from H&E histology images. However, this prediction task faces an inherent limitation, as tissue morphology alone provides insufficient information to fully resolve underlying gene expression. To address this limitation, a recent study leverages partial gene expression to guide the prediction process alongside histology images. Building on this paradigm, we approach the prediction task as a spatial imputation problem, employing a Masked Autoencoder (MAE) to utilize a small fraction of gene expression as genetic anchors for inferring whole-slide gene expression profiles. Specifically, we propose a bio-saliency score and a learning-to-rank strategy to adaptively identify the most informative spots within the tissue. Based on these identified spots, our framework selects contiguous regions as genetic anchors to ensure suitability for real-world ST profiling hardware. To effectively leverage these anchors, we design a cross-modal joint encoder that integrates visual and genetic modalities. By aligning the selected anchors with their corresponding visual features via contrastive learning, the encoder generates robust joint representations to accurately predict gene expression across the whole slide. Notably, our framework consistently surpasses existing methods in both histology-only prediction and spatial imputation, achieving superior accuracy even without genetic anchors and further excelling with as little as 10% transcriptomic coverage. Our code is available at https://github.com/Kyyle2114/CAMMST.
Chinese Translation
空间转录组学（ST）的高成本推动了对直接从H&E组织学图像预测基因表达的广泛研究。然而，这一预测任务面临着固有的限制，因为组织形态学本身提供的信息不足以完全解析潜在的基因表达。为了解决这一限制，最近的一项研究利用部分基因表达来指导与组织学图像的预测过程。在这一范式的基础上，我们将预测任务视为空间插补问题，采用掩蔽自编码器（Masked Autoencoder, MAE）利用少量基因表达作为基因锚点来推断整个切片的基因表达谱。具体而言，我们提出了一种生物显著性评分和学习排序策略，以自适应地识别组织内最具信息量的点。基于这些识别出的点，我们的框架选择连续区域作为基因锚点，以确保其适用于实际的ST分析硬件。为了有效利用这些锚点，我们设计了一个跨模态联合编码器，整合视觉和基因模态。通过对比学习将所选锚点与其对应的视觉特征对齐，编码器生成稳健的联合表示，以准确预测整个切片的基因表达。值得注意的是，我们的框架在仅使用组织学预测和空间插补方面始终超越现有方法，即使在没有基因锚点的情况下也能实现更高的准确性，并且在仅有10%的转录组覆盖率时表现更为出色。我们的代码可在 https://github.com/Kyyle2114/CAMMST 获取。
cs.CV / 62 / 2606.21172
BadDreamer: Transferable Backdoor Attacks against Video World Models for Autonomous Driving
BadDreamer：针对自动驾驶视频世界模型的可转移后门攻击
Abstract
Video world models are increasingly used in autonomous driving to forecast future scene evolution and provide future-aware spatio-temporal representations for downstream action prediction. In perception-to-action pipelines, these representations can directly influence ego-vehicle waypoint planning, making the learned future dynamics a critical security-sensitive component. Despite their promise, the training-time security risks of autonomous-driving video world models remain largely unexplored. We present BadDreamer, a transferable spatio-temporal backdoor attack that targets the perception side of this pipeline. Unlike conventional backdoors that manipulate image labels, prompt outputs, or action supervision, BadDreamer poisons the learned transition dynamics of a video world model. It constructs trigger-erasure sequences in which an oncoming yellow delivery rider is visible in the observed context frames but erased from the future frames. After fine-tuning on a small fraction of such sequences, the compromised world model learns a hidden conditional association: when the physical trigger appears, it hallucinates a future where the rider disappears and the road appears clear. We further show that this corrupted future-aware representation can transfer to the downstream action module without directly modifying ego-trajectory labels, inducing unsafe non-evasive waypoint predictions. Our experiments instantiate this attack on a representative open-source perception-to-action pipeline, revealing a representation-level safety risk in autonomous-driving video world models and highlighting the need for backdoor-aware validation beyond clean generation quality.
Chinese Translation
视频世界模型在自动驾驶中越来越多地被用于预测未来场景演变，并为下游动作预测提供未来感知的时空表示。在感知到动作的流程中，这些表示可以直接影响自我车辆的路径规划，使得学习到的未来动态成为一个关键的安全敏感组件。尽管前景广阔，自动驾驶视频世界模型在训练期间的安全风险仍然在很大程度上未被探索。我们提出了BadDreamer，一种可转移的时空后门攻击，针对这一流程的感知侧。与传统的后门攻击不同，后者操控图像标签、提示输出或动作监督，BadDreamer则是对视频世界模型学习到的转移动态进行污染。它构建了触发器消除序列，其中即将到来的黄色送货骑手在观察到的上下文帧中可见，但在未来帧中被抹去。在对这类序列的小部分进行微调后，受损的世界模型学习到了一种隐藏的条件关联：当物理触发器出现时，它会幻觉出一个未来，在这个未来中骑手消失，路面显得清晰。我们进一步展示了这种被破坏的未来感知表示可以转移到下游动作模块，而无需直接修改自我轨迹标签，从而引发不安全的非规避路径预测。我们的实验在一个代表性的开源感知到动作流程中实例化了这一攻击，揭示了自动驾驶视频世界模型中的表示层安全风险，并强调了超越干净生成质量的后门意识验证的必要性。
cs.CV / 63 / 2606.21174
HERO: Hypothesis-Driven Evidence Retrieval from Omics for Multi-Task Breast Cancer Analysis
HERO：基于假设驱动的多组学证据检索用于多任务乳腺癌分析
Abstract
Matched multi-omics can improve WSI-based biomarker and prognosis prediction, but most existing pipelines use omics as a paral lel feature stream or textual context rather than as an explicit retrieval constraint. HERO asks whether observed omics can be a testable mor phology hypothesis: a sparse pathway-to-morphology prior maps DNA methylation and miRNA into a K-dimensional intent vector m (K=16), TF-IDF retrieval over structured 10 captions selects endpoint-relevant regions, and a cosine gate c=cos(m,v) triggers deterministic deficit driven repair when c<{\tau}c. This closed-loop design bounds VLM calls, reduces reliance on embedding-based semantic matching, and makes every retrieval and verification step lexically auditable. On TCGA-BRCA (930WSIs, patient-level 5-fold CV), HERO sets new state-of-the-art across ER, PR, HER2, subtype, and risk prediction, outperforming both multimodal fusion and VLM-based baselines.
Chinese Translation
匹配的多组学可以改善基于全切片图像（WSI）的生物标志物和预后预测，但大多数现有流程将组学作为并行特征流或文本上下文，而不是作为明确的检索约束。HERO探讨观察到的组学是否可以作为可检验的形态假设：稀疏的通路到形态先验将DNA甲基化和miRNA映射到K维意图向量m（K=16），对结构化的10个标题进行TF-IDF检索以选择与终点相关的区域，当c<{ au}c时，余弦门c=cos(m,v)触发确定性缺陷驱动的修复。该闭环设计限制了VLM调用，减少了对基于嵌入的语义匹配的依赖，使每个检索和验证步骤在词汇上可审计。在TCGA-BRCA（930个WSI，患者级5折交叉验证）上，HERO在ER、PR、HER2、亚型和风险预测方面设定了新的最先进水平，超越了多模态融合和基于VLM的基线。
cs.CV / 64 / 2606.21194
MEDLAYXPLAIN: Benchmarking the Expert-Lay Gap in Medical Vision-Language Models
MEDLAYXPLAIN：医学视觉语言模型中专家与普通人之间差距的基准评估
Abstract
Medical Vision-Language Models (Med-VLMs) achieve strong expert-level performance, yet their ability to generate patient-accessible descriptions remains underexplored. With the 21st Century Cures Act now mandating immediate patient access to diagnostic imaging results, evaluating whether Med-VLMs can bridge this Expert-Lay Gap is both urgent and clinically consequential for patient education and shared decision-making. To this end, we introduce MedLayXPlain, the first large-scale multimodal benchmark and evaluation framework for Medical Lay Language Generation (MLLG). MedLayXPlain-122K provides 122,789 region-grounded samples across 8 imaging modalities from 12 publicly available source datasets, each comprising a medical image with paired expert and lay captions anchored in a three-level Unified Medical Language System (UMLS) ontology hierarchy spanning 7 semantic groups, 43 semantic types, and 2,411 medical concepts. Lay captions are constructed via Hierarchical Ontology-Verified Refinement (HOVER), a three-step pipeline combining patient-centric vocabulary mapping, LLM-based constrained rewriting, and cross-model visual verification to enforce semantic equivalence while preventing hallucination. We further introduce MedLayEval, a lightweight 3B evaluator distilled from a 27B verifier that scores expert-lay alignment across five clinically grounded attributes, addressing the poor correlation between standard NLG metrics and clinical judgment. Benchmarking 33 VLMs on MedLayXPlain-122K reveals a systematic Expert-Lay Gap: medical VLMs achieve strong expert captioning but suffer significant lay-register degradation, while general-purpose VLMs produce more accessible language yet lack clinical precision, confirming that neither current paradigm adequately serves patient-facing communication.
Chinese Translation
医学视觉语言模型（Med-VLMs）在专家级表现上取得了显著成绩，但其生成患者可理解描述的能力仍然未得到充分探索。随着《21世纪治疗法案》现在要求患者立即获取诊断影像结果，评估Med-VLMs是否能够弥补这一专家与普通人之间的差距，对于患者教育和共同决策来说既紧迫又具有临床意义。为此，我们推出了MedLayXPlain，这是首个大规模的多模态基准和医学普通语言生成（MLLG）评估框架。MedLayXPlain-122K提供了来自12个公开可用源数据集的8种影像模态下的122,789个区域基础样本，每个样本包含一幅医学图像及其配对的专家和普通人描述，这些描述基于涵盖7个语义组、43个语义类型和2,411个医学概念的三级统一医学语言系统（UMLS）本体层级。普通人描述是通过分层本体验证精炼（HOVER）构建的，这是一种结合以患者为中心的词汇映射、基于大型语言模型（LLM）的约束重写和跨模型视觉验证的三步流程，以确保语义等价并防止幻觉。我们进一步推出了MedLayEval，这是一个从27B验证器提炼出的轻量级3B评估器，能够根据五个临床基础属性对专家与普通人描述的一致性进行评分，解决了标准自然语言生成（NLG）指标与临床判断之间的低相关性。在MedLayXPlain-122K上对33个视觉语言模型（VLMs）的基准测试揭示了系统性的专家与普通人之间的差距：医学VLMs在专家描述方面表现强劲，但在普通人语言的可理解性上显著下降，而通用VLMs则生成了更易于理解的语言，但缺乏临床精确性，确认了当前的模型范式均未能充分服务于面向患者的沟通。
cs.CV / 65 / 2606.21197
Extraction and Analysis of Multimodal Concepts in Vision Language Models through Sparse Autoencoders
通过稀疏自编码器提取和分析视觉语言模型中的多模态概念
Abstract
Vision Language Models (VLMs) have demonstrated impressive performance in tasks requiring joint understanding of images and text, such as image captioning and Visual Question Answering (VQA), but our understanding of their internal processes remains limited. Recently, Sparse Autoencoders (SAEs) have emerged as a promising tool to support the interpretation of concepts encoded in VLMs. However, most SAE-based approaches focus only on textual or visual concepts separately, ignoring multimodal concepts. This limitation hinders a comprehensive understanding of VLMs, since concepts that integrate both modalities can be misclassified. Moreover, previous visual approaches often produce low-quality visual concept descriptions that are vague or incomplete, limiting their usefulness for understanding model reasoning. We propose a framework based on SAEs to extract and analyze visual, textual, and multimodal concepts from VLMs. For each neuron, we propose a candidate human-interpretable concept and compute the alignment between the concept and the dataset samples using cosine similarity scores. Experiments on a VQA dataset (LLaVA-NeXT) demonstrate that our framework improves visual concept quality by up to 45\% compared to existing SAE-based methods, while maintaining high textual concept quality and enabling systematic identification of multimodal concepts. This work contributes new insights into the conceptual space of VLMs, providing a structured approach to distinguish between visual, textual, and multimodal concepts. The code is available at https://github.com/PHDLanza/Multidata_SAE
Chinese Translation
视觉语言模型（VLMs）在需要对图像和文本进行联合理解的任务中表现出色，如图像描述和视觉问答（VQA），但我们对其内部过程的理解仍然有限。最近，稀疏自编码器（SAEs）作为一种有前景的工具，支持对VLM中编码概念的解释。然而，大多数基于SAE的方法仅关注文本或视觉概念，忽略了多模态概念。这一局限性妨碍了对VLM的全面理解，因为整合两种模态的概念可能会被错误分类。此外，以往的视觉方法通常生成的视觉概念描述质量较低，模糊或不完整，限制了其在理解模型推理中的实用性。我们提出了一种基于SAE的框架，从VLM中提取和分析视觉、文本和多模态概念。对于每个神经元，我们提出一个候选的人类可解释概念，并使用余弦相似度分数计算该概念与数据集样本之间的对齐。对VQA数据集（LLaVA-NeXT）的实验表明，与现有的基于SAE的方法相比，我们的框架在视觉概念质量上提高了多达45\%，同时保持了高质量的文本概念，并能够系统地识别多模态概念。这项工作为VLM的概念空间提供了新的见解，提供了一种结构化的方法来区分视觉、文本和多模态概念。代码可在 https://github.com/PHDLanza/Multidata_SAE 获取。
cs.CV / 66 / 2606.21200
Real-time pedestrian attribute recognition with YOLOv8 and ResNet18
基于YOLOv8和ResNet18的实时行人属性识别
Abstract
Pedestrian attribute recognition (PAR) assigns semantic labels to detected pedestrians and is useful in surveillance, video retrieval, and human-centered graphics applications. This paper presents a two-stage framework in which YOLOv8n detects pedestrians and ResNet18-based models classify gender, estimate apparent age, and predict 61 binary attributes from each pedestrian crop. PETA and PA-100K are combined through semantic attribute mapping, producing a unified training corpus of more than 100,000 pedestrian images while retaining the PETA attribute space. On the reported test splits, the system obtains 99.89% gender classification accuracy, a 4.23-year apparent-age mean absolute error, and 89.96% multi-attribute accuracy with a 36.32% macro F1-score and 58.80% micro F1-score. Runtime measurements indicate 25-30 FPS on an NVIDIA RTX 5060 GPU. The results show that a lightweight detector-classifier pipeline can support real-time PAR, while low macro F1 indicates that rare attributes remain challenging.
Chinese Translation
行人属性识别（PAR）为检测到的行人分配语义标签，在监控、视频检索和以人为中心的图形应用中具有重要意义。本文提出了一种两阶段框架，其中YOLOv8n用于检测行人，基于ResNet18的模型用于分类性别、估计表观年龄，并从每个行人裁剪图像中预测61个二元属性。通过语义属性映射，将PETA和PA-100K结合，生成一个统一的训练语料库，包含超过100,000张行人图像，同时保留PETA属性空间。在报告的测试集上，该系统实现了99.89%的性别分类准确率，表观年龄的平均绝对误差为4.23年，89.96%的多属性准确率，宏观F1得分为36.32%，微观F1得分为58.80%。运行时测量表明在NVIDIA RTX 5060 GPU上可达到25-30帧每秒。结果表明，轻量级的检测-分类管道能够支持实时PAR，而较低的宏观F1得分则表明稀有属性仍然具有挑战性。
cs.CV / 67 / 2606.21234
Context-Aware Autoregressive Diffusion for Gloss-Wise Sign Language Production
基于上下文的自回归扩散模型用于逐字手语生成
Abstract
To generate natural and accurate sentence-level sign language, synthesizing the "gloss", the fundamental semantic unit, is essential. However, most current sign-language production (SLP) methods generate entire sequences at once. While this end-to-end approach is often efficient, it is prone to temporal drift and hand motion blur as sentences get longer, and fails to accurately control individual glosses. In this paper, we propose the Context-aware Gloss-wise AutoRegressive Diffusion model (GARD), a gloss-wise diffusion framework that models coarticulation by conditioning on both semantic (linguistic) and kinematic (motion) contexts. To ensure natural continuity between gloss motions, GARD introduces two additional strategies: i) Inter-Gloss Transition Guidance, which applies gradient-based guidance to kinematically align inter-gloss boundaries and ensure seamless pose consistency. ii) Global Motion Harmonizer, refining the entire gloss motion sequence based on the boundary poses adjusted by Inter-Gloss Transition Guidance. Extensive experiments on Phoenix-T and CSL-Daily datasets demonstrate that GARD achieves superior performance over existing SLP methods in terms of both linguistic accuracy and motion similarity.
Chinese Translation
为了生成自然且准确的句子级手语，合成“字汇”（gloss），这一基本语义单元，是至关重要的。然而，目前大多数手语生成（SLP）方法都是一次性生成整个序列。尽管这种端到端的方法通常效率较高，但随着句子的延长，它容易出现时间漂移和手部动作模糊，并且无法准确控制单个字汇。在本文中，我们提出了基于上下文的逐字自回归扩散模型（GARD），这是一个逐字扩散框架，通过对语义（语言）和运动（运动）上下文的条件建模来模拟连音现象。为了确保字汇动作之间的自然连续性，GARD引入了两种额外策略：i) 字汇间过渡引导（Inter-Gloss Transition Guidance），该策略应用基于梯度的引导来运动上对齐字汇边界，确保姿态的一致性无缝衔接。ii) 全局运动协调器（Global Motion Harmonizer），基于由字汇间过渡引导调整的边界姿态，优化整个字汇动作序列。在Phoenix-T和CSL-Daily数据集上的广泛实验表明，GARD在语言准确性和动作相似性方面均优于现有的手语生成方法。
cs.CV / 68 / 2606.21244
ACE-GS: Acing the Trade-off with Accurate, Compact and Efficient 3D Gaussian Splatting
ACE-GS：以准确、紧凑和高效的3D高斯点云处理实现权衡的突破
Abstract
3D Gaussian Splatting achieves exceptional real-time rendering, but its substantial computational and storage demands hinder widespread deployment. Existing accelerated paradigms often aggressively prune primitives for rapid convergence, causing severe loss of high-frequency details. To address this, we tackle the fundamental problem of achieving both exceptional rendering quality and ultra-fast reconstruction speed. In this paper, we propose ACE-GS, a progressive optimization framework tailored for accurate, compressed, and efficient scene representation. We realize that precise primitive management is the key to breaking this trade-off. Therefore, we first design a momentum consistency-guided densification strategy, strictly constraining primitive growth onto authentic geometric manifolds to avoid computational waste while significantly accelerating convergence. Building upon this efficient initialization, we deploy a statistical sensitivity-driven sparsification mechanism to precisely prune redundant primitives, yielding a further compressed footprint. Finally, to thoroughly compensate for the risk of micro-structure loss caused by the aforementioned strict primitive control, we introduce a cross-dimensional residual frequency compensation scheme that explicitly back-injects high-frequency error energy into primitive attributes, perfectly restoring sharp geometric details. Extensive experiments validate our superiority. While maintaining a highly compact scene representation, our system achieves up to 3.7 times training acceleration against the rapid framework Speedy-Splat. Requiring only 3 to 5 minutes to converge, ACE-GS secures the highest structural similarity and achieves a peak PSNR improvement of up to 0.89 dB over the original 3DGS, establishing a new benchmark for ultra-fast and high-fidelity novel view synthesis.
Chinese Translation
3D高斯点云处理实现了卓越的实时渲染，但其巨大的计算和存储需求阻碍了广泛应用。现有的加速范式通常会激进地修剪原始体，以实现快速收敛，导致高频细节的严重损失。为了解决这个问题，我们着手解决实现卓越渲染质量与超快重建速度之间的基本问题。本文提出了ACE-GS，一个针对准确、压缩和高效场景表示的渐进优化框架。我们意识到，精确的原始体管理是打破这一权衡的关键。因此，我们首先设计了一种动量一致性引导的稠密化策略，严格限制原始体的增长在真实几何流形上，以避免计算浪费，同时显著加速收敛。在此高效初始化的基础上，我们部署了一种基于统计敏感性的稀疏化机制，以精确修剪冗余原始体，从而进一步压缩占用空间。最后，为了全面补偿上述严格原始体控制所导致的微观结构损失风险，我们引入了一种跨维度残差频率补偿方案，明确地将高频误差能量回注入原始体属性，完美恢复尖锐的几何细节。大量实验验证了我们的优越性。在保持高度紧凑的场景表示的同时，我们的系统在训练加速方面比快速框架Speedy-Splat快达3.7倍。ACE-GS仅需3到5分钟即可收敛，确保了最高的结构相似性，并在原始3DGS上实现了高达0.89 dB的峰值信噪比改善，为超快和高保真新视图合成建立了新的基准。
cs.CV / 69 / 2606.21252
A Neurosymbolic Framework for Interpretable Skeleton-Based Seizure Detection via Concept-Driven Logical Reasoning
基于神经符号框架的可解释性骨架基础癫痫发作检测：基于概念驱动的逻辑推理
Abstract
Video-based seizure detection is essential for the management of epilepsy patients, offering a non-invasive complement to electroencephalography. While several deep learning approaches have been developed for video-based seizure detection, none are inherently interpretable, limiting their adoption and translation into clinical practice. We present, to our knowledge, the first exploration of a neurosymbolic framework for video-based seizure detection that directly addresses this gap. Our approach (1) extracts patient-centric skeleton sequences from epilepsy monitoring units via a prompt-guided foundation model, (2) predicts binary spatio-temporal concept activations grounded in clinical motor semiology guidelines, and (3) composes them via differentiable logic into interpretable Boolean rules with auditable contributions. Furthermore, to mitigate false positives arising from the traditional binary formulation (seizure vs.\ non-seizure), we sub-classify non-seizure segments into clinically relevant normal activities, providing the model with fine-grained discriminative supervision. Evaluated on two public seizure video benchmarks, our framework achieves 89.78% sensitivity with 0.06 false detections per hour on SAHZU and 85.27%,0.09 on IEEE, while producing complete three-level interpretability: every prediction decomposes into which motor primitives were detected, how they were logically composed, and how much each rule contributed to the clinical decision. We publicly release all annotations, extracted pose sequences, our data pipeline and code, https://github.com/Mr-TalhaIlyas/CDSD/.
Chinese Translation
基于视频的癫痫发作检测对于癫痫患者的管理至关重要，它为脑电图提供了一种非侵入性的补充。尽管已经开发了几种深度学习方法用于基于视频的癫痫发作检测，但这些方法本质上并不具备可解释性，限制了它们在临床实践中的应用和转化。我们提出了一个神经符号框架，用于基于视频的癫痫发作检测，首次直接解决了这一空白。我们的方法 (1) 通过一个提示引导的基础模型从癫痫监测单元中提取以患者为中心的骨架序列，(2) 基于临床运动症状学指南预测二元时空概念激活，(3) 通过可微逻辑将其组合成可解释的布尔规则，并提供可审计的贡献。此外，为了减轻传统二元形式（发作与非发作）所产生的假阳性，我们将非发作片段细分为临床相关的正常活动，为模型提供细粒度的区分监督。在两个公共癫痫视频基准上进行评估，我们的框架在SAHZU上实现了89.78%的灵敏度，每小时假检测0.06次，在IEEE上为85.27%，每小时假检测0.09次，同时产生完整的三级可解释性：每个预测都分解为检测到的运动原语、它们的逻辑组合方式以及每条规则对临床决策的贡献程度。我们公开发布了所有注释、提取的姿态序列、数据管道和代码，网址为 https://github.com/Mr-TalhaIlyas/CDSD/.
cs.CV / 70 / 2606.21267
Few-Shot Hyperspectral Aphid Detection via FastGAN Synthetic Data Generation, Transformer-Based Classification and Explainable AI
基于FastGAN合成数据生成的少样本高光谱蚜虫检测、基于变换器的分类与可解释人工智能
Abstract
Early detection of aphid infestation in crops is essential for preventing yield loss and reducing unnecessary pesticide use. Hyperspectral imaging combined with Spectral Information Divergence (SID) analysis offers a non-destructive approach for monitoring plant health; however, deep learning methods applied to hyperspectral data are often limited by small dataset sizes. In this study, a data-efficient generative adversarial network (FastGAN) was employed to augment a hyperspectral SID dataset of faba bean leaves containing healthy and aphid-infested samples. The trained generator produced 10,000 synthetic images preserving structural and spectral characteristics of real samples. Image quality was evaluated using Frechet Inception Distance (FID), demonstrating stable convergence and realistic reconstruction of leaf morphology and infestation patterns. The augmented dataset was used to train four classification architectures: VGG16, ResNet-50, EfficientNet, and Vision Transformer (ViT). Results showed that dataset augmentation significantly improved classification robustness, with performance progressively increasing from classical convolutional networks to transformer-based models. The ViT model achieved the highest accuracy and F1-scores, while EfficientNet provided strong balanced performance and ResNet-50 showed moderate improvements over VGG16. Confusion matrix analysis confirmed reduced false negatives and improved disease detection when using advanced architectures. The findings demonstrate that FastGAN-based augmentation effectively enhances hyperspectral plant disease classification and that transformer-based models provide the most reliable discrimination between healthy and infested leaves.
Chinese Translation
早期检测作物中的蚜虫侵染对于防止产量损失和减少不必要的农药使用至关重要。高光谱成像结合光谱信息散度（SID）分析提供了一种非破坏性监测植物健康的方法；然而，应用于高光谱数据的深度学习方法常常受到小数据集规模的限制。在本研究中，采用了一种数据高效的生成对抗网络（FastGAN）来增强包含健康和蚜虫侵染样本的蚕豆叶片高光谱SID数据集。训练后的生成器生成了10,000张合成图像，保留了真实样本的结构和光谱特征。通过Frechet Inception Distance（FID）评估图像质量，结果表明生成过程稳定收敛，并且能够真实重建叶片形态和侵染模式。增强后的数据集用于训练四种分类架构：VGG16、ResNet-50、EfficientNet和视觉变换器（ViT）。结果显示，数据集增强显著提高了分类的鲁棒性，性能从经典卷积网络逐渐提升到基于变换器的模型。ViT模型达到了最高的准确率和F1分数，而EfficientNet则提供了强劲的平衡性能，ResNet-50在VGG16的基础上显示出适度的改进。混淆矩阵分析确认了使用先进架构时假阴性减少和疾病检测改善的效果。研究结果表明，基于FastGAN的增强有效提升了高光谱植物疾病分类的能力，而基于变换器的模型在健康与侵染叶片之间提供了最可靠的区分。
cs.CV / 71 / 2606.21279
Beyond Damage Assessment: Recyclable Material Detection in Aerial Disaster Imagery Using a Lightweight Patch-Based Framework
超越损害评估：基于轻量级补丁框架的空中灾害影像中可回收材料检测
Abstract
Nowadays, more and more disasters of different natures are appearing. Several disaster assessment approaches have been developed in order to identify damaged areas from aerial images. These damaged areas contain rich material that could be recycled towards several ecological purposes. In this paper, we present a lightweight approach that permits the efficient detection of recyclable material. Experimental results show the potential of the proposed approach towards localizing recyclable materials. Accordingly, we provide a rare dataset of material images that we labeled towards supporting the development of recyclable material detectors. The dataset of labeled material images is publicly available at: anonymous.
Chinese Translation
如今，各种不同性质的灾害日益增多。为识别空中影像中的受损区域，已经开发了多种灾害评估方法。这些受损区域包含丰富的材料，可用于多种生态目的的回收。在本文中，我们提出了一种轻量级的方法，能够高效检测可回收材料。实验结果表明，所提方法在定位可回收材料方面具有潜力。因此，我们提供了一个罕见的材料影像数据集，并对其进行了标注，以支持可回收材料检测器的开发。该标注材料影像数据集可公开获取，网址为：匿名。
cs.CV / 72 / 2606.21287
Unsupervised Domain Adaptation for Sim-to-Real Object Pose Estimation with Contrastive Alignment and Pseudo-Label Refinement
基于对比对齐和伪标签精炼的无监督领域适应用于模拟到真实的物体姿态估计
Abstract
Unsupervised domain adaptation (UDA) enables robust transfer of knowledge from simulated to real environments while exploiting a subset of unlabeled target data to improve real-world performance. Existing UDA methods for Object pose estimation often rely on global feature matching, multi-stage larger frameworks, or image translation pipelines, which tend to overlook the pose-specific information embedded in feature representations. To bridge this limitation, we introduce CAPLR that targets the adaptation of pose-sensitive features in localized regions, ensuring that domain alignment preserves the geometric cues essential for accurate pose estimation. CAPLR achieves UDA with three key components: (1) Efficient Cross-Domain Pairing strategy leveraging intermediate features to identify pose similar image pairs across domains without supervision; (2) Contrastive Alignment to perform feature alignment at localised regions in both intermediate and task-specific representations; and (3) Consistency-Based Pseudo-Label Refinement to improve reliability by encouraging stable target predictions. Extensive experiments demonstrate that CAPLR achieves state-of-the-art performance across multiple well-known object pose estimation benchmarks featuring diverse and challenging scenarios.
Chinese Translation
无监督领域适应（UDA）能够在利用一部分未标记的目标数据以提高真实世界性能的同时，稳健地将知识从模拟环境转移到真实环境。现有的用于物体姿态估计的UDA方法通常依赖于全局特征匹配、多阶段的大型框架或图像翻译管道，这些方法往往忽视了特征表示中嵌入的姿态特定信息。为了解决这一局限性，我们提出了CAPLR，旨在适应局部区域中的姿态敏感特征，确保领域对齐保留对准确姿态估计至关重要的几何线索。CAPLR通过三个关键组件实现UDA：（1）高效的跨域配对策略，利用中间特征在不同领域中识别姿态相似的图像对，无需监督；（2）对比对齐，在中间和任务特定表示的局部区域进行特征对齐；（3）基于一致性的伪标签精炼，通过鼓励稳定的目标预测来提高可靠性。大量实验表明，CAPLR在多个知名的物体姿态估计基准测试中实现了最先进的性能，涵盖了多样且具有挑战性的场景。
cs.CV / 73 / 2606.21290
NoduLoCC2026: Lung Nodule Localization and Classification Contest from Chest X-Ray Images
NoduLoCC2026：胸部X光图像中的肺结节定位与分类竞赛
Mustafic, Adnan, Benhabiles, Halim, Cabani, Adnane, Aguilar, Kristhian André Oliveira, Amigon, Romain, Bardin, Clément, Bentifece, Chiara, Boehm, Marin, Bouchard, Kévin, Burattini, Laura, Carmo, Diedre, Idiri, Fahima, Lahargoue, Matthis, Marcantoni, Ilaria, Messaoudi, Hicham, Meyer, Cyril, Meziane, Farid, Morales, Léon, Rittner, Letícia, Sbrollini, Agnese, Zipper, Léonard, Hammoudi, Karim
Abstract
We propose NoduLoCC2026, a challenge on lung nodule detection and localization in chest X-ray images. We have provided a dataset for both tasks and received submissions from 5 international teams. The participating teams' solutions are presented in this work along with results on an external dataset used for testing. Proposed methods show good performance on the classification task. The best method shows a balanced accuracy score of 0.72 and AUC-ROC of 0.79. We highlight the limitations of current approaches for the localization task, with the best approach having predicted the correct number of nodules on 53\% of the test images with a median distance of 12.83mm, showing that it is a more challenging task than the first one. The challenge website is available via https://gt-i2mdp.github.io/website/nodule_challenge.html.
Chinese Translation
我们提出了NoduLoCC2026，这是一个关于胸部X光图像中肺结节检测与定位的挑战。我们为这两个任务提供了一个数据集，并收到了来自5个国际团队的提交。参与团队的解决方案在本文中进行了展示，并提供了用于测试的外部数据集的结果。所提出的方法在分类任务上表现良好。最佳方法的平衡准确率为0.72，AUC-ROC为0.79。我们强调了当前定位任务方法的局限性，最佳方法在53%的测试图像中正确预测了结节的数量，且中位距离为12.83mm，显示出这一任务比第一个任务更具挑战性。挑战网站可通过https://gt-i2mdp.github.io/website/nodule_challenge.html访问。
cs.CV / 74 / 2606.21292
Lightweight 3D Feature Pretraining by Bayesian Inversion of 2D Foundation Models
通过贝叶斯反演二维基础模型进行轻量级三维特征预训练
Abstract
We present Casper3D, a lightweight probabilistic framework for converting noisy multi-view 2D foundation-model embeddings into a latent 3D semantic representation. We model view-level semantic features as noisy observations of an underlying 3D semantic state and infer this state with a set-based variational model that incorporates relative pose during multi-view reasoning. Casper3D is trained by predicting held-out semantic observations from novel viewpoints, while remaining aligned with visual and text semantic spaces for open-vocabulary 3D understanding. The framework is backbone-agnostic and applies to both language-aligned and self-supervised embeddings. Experiments show that Casper3D produces more stable 3D semantics than simple multi-view pooling, especially in ambiguous and noisy settings.
Chinese Translation
我们提出了Casper3D，一个轻量级的概率框架，用于将噪声多视角二维基础模型嵌入转换为潜在的三维语义表示。我们将视图级语义特征建模为潜在三维语义状态的噪声观测，并通过一个基于集合的变分模型推断该状态，该模型在多视角推理过程中结合了相对姿态。Casper3D通过预测来自新视角的保留语义观测进行训练，同时与视觉和文本语义空间保持一致，以实现开放词汇的三维理解。该框架与主干网络无关，适用于语言对齐和自监督嵌入。实验表明，Casper3D在模糊和噪声环境下产生的三维语义比简单的多视角池化更稳定。
cs.CV / 75 / 2606.21300
SCOPE: Scale-Consistent One-Pass Estimation of 3D Geometry
SCOPE：尺度一致的单次估计3D几何
Abstract
We present SCOPE (Scale-Consistent One-Pass Estimation of 3D Geometry), a novel approach for estimating 3D geometry from extended monocular video sequences, where existing methods struggle to maintain both geometric accuracy and temporal consistency across hundreds of frames. Our approach generates affine-invariant 3D point maps with shared parameters across entire sequences, enabling consistent scale-invariant representations. We introduce three key innovations: viewpoint-invariant geometry aligning multi-perspective points in a unified reference frame; appearance-invariant learning enforcing consistency across exponential timescales; and frequency-modulated positioning enabling extrapolation to sequences vastly exceeding training length. Experiments across diverse datasets demonstrate significant improvements, reducing relative point map error by 24.2% and temporal alignment error by 34.9% on ScanNet compared to state-of-the-art methods. Our approach handles challenging scenarios with complex camera trajectories and lighting variations while efficiently processing extended sequences in a single pass. Project page: https://scope3d.github.io/.
Chinese Translation
我们提出了SCOPE（尺度一致的单次估计3D几何），这是一种从扩展的单目视频序列中估计3D几何的新方法，现有方法在数百帧中难以同时保持几何精度和时间一致性。我们的方法生成具有共享参数的仿射不变3D点图，覆盖整个序列，从而实现一致的尺度不变表示。我们引入了三个关键创新：视点不变几何将多视角点对齐到统一的参考框架；外观不变学习在指数时间尺度上强制一致性；以及频率调制定位使得能够推断出远超训练长度的序列。跨多个数据集的实验表明，显著提高了性能，相较于最先进的方法，在ScanNet上相对点图误差降低了24.2%，时间对齐误差降低了34.9%。我们的方法能够处理复杂相机轨迹和光照变化的挑战场景，同时在单次处理扩展序列时效率高。项目页面：https://scope3d.github.io/
cs.CV / 76 / 2606.21304
A Test-time Actor-Critic Approach to News Images Generation
一种测试时的演员-评论家方法用于新闻图像生成
Abstract
This paper introduces the CERTH-ITI solution for the MediaEval NewsImages 2026 challenge, which focuses on generating images related to news headlines. Inspired by the Actor-Critic paradigm in reinforcement learning, we present a test-time, model-agnostic Actor-Critic Image Generation approach (ACIG). ACIG generates prompts for image creation, produces the images, evaluates the generated results, and if needed refines the image generation prompts accordingly in a feedback loop. ACIG achieved the best results in the NewsImages 2026 challenge, according to the challenge's leaderboard.
Chinese Translation
本文介绍了CERTH-ITI在MediaEval NewsImages 2026挑战中的解决方案，该挑战专注于生成与新闻标题相关的图像。受到强化学习中演员-评论家（Actor-Critic）范式的启发，我们提出了一种测试时的、模型无关的演员-评论家图像生成方法（ACIG）。ACIG生成图像创作的提示，生成图像，评估生成结果，并在反馈循环中根据需要调整图像生成提示。根据挑战的排行榜，ACIG在NewsImages 2026挑战中取得了最佳成绩。
cs.CV / 77 / 2606.21309
WildBox: A Dataset and Benchmark for Aerial Monocular 3D Detection of African Savanna Wildlife
WildBox：用于非洲草原野生动物的航拍单目3D检测的数据集与基准
Abstract
We introduce WildBox, a dataset and benchmark for monocular 3D detection of wildlife from drone video, comprising 237,505 3D bounding box annotations across seven African savanna species grouped into six benchmark classes. Annotations follow a KITTI/Omni3D-compatible format in a per-segment scale-normalised camera frame, with instance identities maintained across each segment. We evaluate two open-vocabulary monocular 3D architectures, OVMono3D-LIFT and DetAny3D, under zero-shot, ground-truth 2D box prompt, and supervised fine-tuning protocols. Open-vocabulary 2D foundation models provide usable zero-shot wildlife localisation (50.55 AP@50), but zero-shot 3D detection collapses to 0.00 AP across both architectures and every 2D-input condition tested, including ground-truth 2D box prompts, thus isolating the failure to the 3D stage. Fine-tuning on WildBox recovers performance to 8.68 +/- 0.47
[email protected] and 13.17 +/- 0.69 AP3D macro. Depth contributes 84% of normalised Hausdorff distance after fine-tuning and over 99% in zero-shot, identifying monocular aerial depth as the dominant open problem in this regime. A coarse-to-fine curriculum, i.e. pretraining on a merged zebra class before fine-tuning on the Grevy's/plains split, improves macro 3D performance with less total compute, with the largest gains on the two zebra subclasses. WildBox is released with video-level splits, evaluation code, and baseline checkpoints to enable progress in 3D wildlife perception from drone video.
Chinese Translation
我们介绍了WildBox，一个用于从无人机视频中进行野生动物单目3D检测的数据集和基准，包含237,505个3D边界框注释，涵盖七种非洲草原物种，分为六个基准类别。注释遵循KITTI/Omni3D兼容格式，采用每段规模归一化的相机帧，并在每个段落中保持实例身份。我们在零样本、真实标签2D框提示和监督微调协议下评估了两种开放词汇单目3D架构OVMono3D-LIFT和DetAny3D。开放词汇2D基础模型提供了可用的零样本野生动物定位（50.55 AP@50），但零样本3D检测在两种架构和所有测试的2D输入条件下崩溃至0.00 AP，包括真实标签2D框提示，从而将失败归因于3D阶段。对WildBox的微调恢复了性能，达到8.68 +/- 0.47
[email protected]和13.17 +/- 0.69 AP3D宏。微调后，深度贡献了归一化Hausdorff距离的84%，而在零样本中则超过99%，识别出单目航拍深度是这一领域的主要开放问题。粗到细的课程，即在合并斑马类别上进行预训练，然后在Grevy's/平原分割上进行微调，改善了宏观3D性能，并减少了总计算量，最大的增益出现在两个斑马子类上。WildBox发布了视频级别的分割、评估代码和基线检查点，以促进无人机视频中3D野生动物感知的进展。
cs.CV / 78 / 2606.21358
LEViL: Label-Efficient Video Learning via Zero-Shot Distillation over VLM-Generated Pseudo-Label Spaces
LEViL：通过在VLM生成的伪标签空间上进行零-shot蒸馏的标签高效视频学习
Abstract
Supervised video pretraining is a common transfer learning practice for improving downstream action recognition performance. However, it requires large-scale labeled source datasets, and the effectiveness of the learned initialization is influenced by the similarity between the source and target domains. Constructing such labeled pretraining datasets for different target domains is costly and difficult to scale. To address these limitations, this study proposes a label-efficient video learning framework that combines annotation-free video pretraining with target-label-set-aware fine-tuning. During pretraining, a vision-language model (VLM) generates textual descriptions of unlabeled videos, which are processed to construct an interpretable semantic pseudo-label space. A frozen video-language model then produces zero-shot soft target distributions over this space, allowing a student video encoder to learn semantically rich representations without manual source annotations. During downstream adaptation, target-label-set-aware fine-tuning combines supervised learning from labeled target videos with zero-shot distillation over the actual target label set, helping preserve VLM-derived semantic guidance while adapting the pretrained encoder to the target task. Experiments on UCF101 and HMDB51 show that the proposed framework outperforms the compared semi-supervised video action recognition methods across all evaluated limited-label regimes. Moreover, the annotation-free pretraining stage learns transferable representations that provide an effective initialization for full-data fine-tuning, despite relying on a comparatively modest unlabeled pretraining pool.
Chinese Translation
监督视频预训练是提高下游动作识别性能的一种常见迁移学习实践。然而，它需要大规模的标注源数据集，并且学习到的初始化效果受到源域和目标域相似性的影响。为不同目标域构建这样的标注预训练数据集既昂贵又难以扩展。为了解决这些限制，本研究提出了一种标签高效的视频学习框架，该框架结合了无标注视频预训练和目标标签集感知的微调。在预训练过程中，视觉-语言模型（VLM）生成未标注视频的文本描述，这些描述被处理以构建可解释的语义伪标签空间。然后，一个冻结的视频-语言模型在该空间上生成零-shot软目标分布，从而使学生视频编码器能够在没有手动源标注的情况下学习语义丰富的表示。在下游适应过程中，目标标签集感知的微调将来自标注目标视频的监督学习与实际目标标签集上的零-shot蒸馏相结合，帮助在将预训练编码器适应目标任务的同时保留VLM衍生的语义指导。在UCF101和HMDB51上的实验表明，所提出的框架在所有评估的有限标签条件下均优于比较的半监督视频动作识别方法。此外，无标注预训练阶段学习到的可迁移表示为全数据微调提供了有效的初始化，尽管依赖于相对适度的未标注预训练池。
cs.CV / 79 / 2606.21368
Graph-of-Differences: Anatomy-Structured Difference Alignment for Medical Image Re-Identification
差异图：基于解剖结构的医学图像重新识别差异对齐
Abstract
Medical image re-identification (MedReID) enables longitudinal patient linkage but remains vulnerable to shortcut learning and often produces decisions that clinicians cannot audit against named anatomy. We propose Graph-of-Differences (GoD), which grounds identity comparisons in explicit anatomical structure. Each image is represented as an anatomy graph whose nodes correspond to named anatomical regions; given an image pair, soft node correspondence is established, and differences are computed over matched anatomy. A graph-level difference alignment objective ties these anatomy-matched differences to the global backbone difference, ensuring the retrieval signal is anchored in homologous structures rather than arbitrary spatial tokens. Explanations are defined over named graph nodes and quantitatively audited via node insertion/deletion tests, replacing unstable pixel heatmaps with verifiable structure-level evidence. On internal benchmarks, GoD improves Rank-1 by +7.1 pp on fundus and +3.1 pp on CXR over a strong frozen-backbone baseline, with further gains on zero-shot external transfers confirming that anatomy grounding improves both accuracy and generalization. Code is available at https://github.com/GenMI-Lab/GoD.git.
Chinese Translation
医学图像重新识别（MedReID）能够实现纵向患者链接，但仍然容易受到捷径学习的影响，常常产生临床医生无法根据命名解剖结构进行审计的决策。我们提出了差异图（Graph-of-Differences, GoD），该方法将身份比较基于明确的解剖结构。每幅图像被表示为一个解剖图，其节点对应于命名的解剖区域；给定一对图像，建立软节点对应关系，并在匹配的解剖结构上计算差异。图级差异对齐目标将这些解剖匹配的差异与全局主干差异联系起来，确保检索信号锚定在同源结构上，而不是任意的空间标记。解释是基于命名图节点定义的，并通过节点插入/删除测试进行定量审计，从而用可验证的结构级证据替代不稳定的像素热图。在内部基准测试中，GoD在强大的冻结主干基线之上，基金图像的Rank-1提高了7.1个百分点，胸部X光图像提高了3.1个百分点，进一步的零样本外部迁移验证了解剖基础提高了准确性和泛化能力。代码可在 https://github.com/GenMI-Lab/GoD.git 获取。
cs.CV / 80 / 2606.21373
FLM-Occ: Feed-forward Likelihood Maximization for Efficient Indoor Occupancy Prediction
FLM-Occ：用于高效室内占用预测的前馈似然最大化
Abstract
Recent indoor occupancy prediction methods adopt Gaussian primitives as a sparse 3D representation for computational efficiency. However, their training relies on voxel classification, which imposes only local constraints and lacks global supervision on the distribution of the primitives. Therefore, they inevitably predict spurious primitives in empty regions, undermining both representational and computational efficiency. To address this, we propose Feed-forward Likelihood Maximization (FLM), a novel framework that reformulates occupancy prediction as voxel distribution estimation. In FLM, a network is trained to predict a mixture model that maximizes the likelihood over ground-truth occupied voxels in a feed-forward manner. To enable end-to-end training of networks and voxelization of a standard mixture model, we define mixture weights as normalized primitive volumes to implicitly enforce simplex constraints and derive novel voxelization formulas. Based on FLM, our FLM-Occ, a novel method that is capable of relocating randomly initialized primitives over long distances to model a scene. On Occ-ScanNet, FLM-Occ achieves superior accuracy using only 32 superquadrics, 2.7% of the prior SoTA, while running 3.7 times faster.
Chinese Translation
近期的室内占用预测方法采用高斯原语作为稀疏的三维表示，以提高计算效率。然而，它们的训练依赖于体素分类，这仅施加了局部约束，缺乏对原语分布的全局监督。因此，它们不可避免地在空区域预测出虚假的原语，从而削弱了表示和计算效率。为了解决这个问题，我们提出了前馈似然最大化（Feed-forward Likelihood Maximization, FLM），这是一个将占用预测重新表述为体素分布估计的新框架。在FLM中，网络被训练以预测一个混合模型，该模型以前馈方式最大化真实占用体素的似然。为了实现网络的端到端训练和标准混合模型的体素化，我们将混合权重定义为归一化的原语体积，以隐式施加单纯形约束，并推导出新的体素化公式。基于FLM，我们提出了FLM-Occ，这是一种新方法，能够在长距离内重新定位随机初始化的原语以建模场景。在Occ-ScanNet上，FLM-Occ仅使用32个超四面体（superquadrics），占先前最先进技术（SoTA）的2.7%，却实现了更高的准确性，同时运行速度快了3.7倍。
cs.CV / 81 / 2606.21381
OSOG: A Differentiable, Physics-Informed Synthetic Data Engine for Micro-Optical Environments
OSOG：一种可微分的、物理信息驱动的微光学环境合成数据引擎
Abstract
Deep learning in computational microscopy is severely constrained by the scarcity of densely annotated datasets. While synthetic data generation has bridged this gap in macroscopic computer vision, traditional graphics engines rely on geometric ray-tracing, failing to capture the micro-optical phenomena required for microscopy. Conversely, while wave-optics formulations exist, rendering them computationally tractable at the scale required for deep learning remains a massive systems challenge. To address this, we introduce the Optical Synthetic Object Generator (OSOG), a high-performance, fully differentiable forward-modeling engine. Drawing on established physical models of diffraction and phase retardation, OSOG maps continuous Optical Path Difference (OPD) calculations into a highly optimized, PyTorch-native Structure-of-Arrays (SoA) architecture. We validate this computational framework across three axes: First, object detection models (YOLOv11-OBB) trained purely on OSOG-generated data achieve robust zero-shot transfer to real-world highly occluded Lysozyme micrographs. Second, we introduce DiffOSOG, demonstrating that the engine's end-to-end differentiability allows for the exact recovery of continuous optical parameters via curriculum-guided inverse rendering. Finally, OSOG bypasses the $\mathcal{O}(N)$ bottlenecks of sequential ray-tracing, demonstrating sub-linear scaling by synthesizing 40,000 complex wave-optic particles in under 50 milliseconds (\>20 FPS). By providing a fast, scalable, and physically grounded tensor pipeline, OSOG enables true real-time, on-the-fly dataset generation.
Chinese Translation
计算显微镜中的深度学习受到稠密标注数据集稀缺的严重限制。虽然合成数据生成在宏观计算机视觉中弥补了这一空白，但传统图形引擎依赖几何光线追踪，未能捕捉显微镜所需的微光学现象。相反，虽然存在波光学的公式，但在深度学习所需的规模上使其计算上可行仍然是一个巨大的系统挑战。为了解决这个问题，我们引入了光学合成物体生成器（Optical Synthetic Object Generator，OSOG），这是一种高性能、完全可微分的前向建模引擎。OSOG借鉴了已建立的衍射和相位延迟的物理模型，将连续的光程差（Optical Path Difference，OPD）计算映射到高度优化的、基于PyTorch的数组结构（Structure-of-Arrays，SoA）架构中。我们在三个方面验证了这一计算框架：首先，纯粹基于OSOG生成的数据训练的物体检测模型（YOLOv11-OBB）在真实世界高度遮挡的溶菌酶显微图像上实现了强大的零样本迁移。其次，我们介绍了DiffOSOG，展示了该引擎的端到端可微分性允许通过课程引导的逆渲染精确恢复连续光学参数。最后，OSOG绕过了顺序光线追踪的$ ext{O}(N)$瓶颈，通过在50毫秒内合成40,000个复杂波光学粒子，展示了亚线性扩展。通过提供快速、可扩展且基于物理的张量管道，OSOG实现了真正的实时、即时数据集生成。
cs.CV / 82 / 2606.21384
EnTrust: Modeling Inter-Modal Conflict for Trustworthy Multimodal Medical Image Analysis
EnTrust：可信多模态医学图像分析中的模态间冲突建模
Abstract
Multimodal medical imaging fuses complementary anatomical and functional information, yet modalities frequently disagree in pathologically heterogeneous regions. Current segmentation models handle this in one of two inadequate ways: deterministic fusion that averages away disagreement, or post-hoc uncertainty estimation decoupled from the fusion process that produces it. Both obscure the clinically critical question: why is this prediction unreliable? We present EnTrust, a framework that treats inter-modal conflict as the primary source of predictive uncertainty. Our EnFuse module decomposes multimodal features into three disentangled components: shared anatomical consensus (F_c), modality-specific cues (F_{u,m}), and spatially localized conflict signals (F_{cf}), with independence enforced via a cross-covariance objective. This structured decomposition conditions SegDiff, a diffusion-based generative segmentation model whose sampled hypotheses diverge specifically in regions of modal disagreement. TrustMap then translates this hypothesis divergence into calibrated, pixel-wise uncertainty using ensemble entropy, conflict-guided perturbation probing, and a learned calibration head, enabling clinicians to understand not only where predictions are uncertain, but why. Across four benchmarks spanning brain, cardiac, lesion, and oncology domains, EnTrust achieves state-of-the-art segmentation accuracy while reducing calibration error by 40% compared to the strongest baseline. Notably, it outperforms 5x deep ensembles using a single model at roughly half the memory footprint. Code and checkpoints are available at https://github.com/GenMI-Lab/EnTrust.git.
Chinese Translation
多模态医学成像融合了互补的解剖和功能信息，但在病理异质性区域，模态之间经常存在不一致。目前的分割模型以两种不充分的方式处理这一问题：确定性融合通过平均消除不一致，或后验不确定性估计与产生它的融合过程脱钩。这两种方法都模糊了临床上关键的问题：为什么这个预测不可靠？我们提出了EnTrust，一个将模态间冲突视为预测不确定性主要来源的框架。我们的EnFuse模块将多模态特征分解为三个解耦的组成部分：共享的解剖共识（F_c）、模态特定线索（F_{u,m}）和空间局部冲突信号（F_{cf}），并通过交叉协方差目标强制独立性。这种结构化分解为SegDiff提供条件，SegDiff是一个基于扩散的生成分割模型，其采样假设在模态不一致的区域特定地发散。TrustMap随后利用集成熵、冲突引导的扰动探测和学习的校准头，将这种假设发散转化为经过校准的逐像素不确定性，使临床医生不仅能够理解预测的不确定性所在，还能理解原因。在涵盖脑部、心脏、病变和肿瘤领域的四个基准测试中，EnTrust实现了最先进的分割准确性，同时将校准误差降低了40%，相较于最强基线表现尤为突出。值得注意的是，它使用单一模型的性能超过了5倍深度集成，并且内存占用约为其一半。代码和检查点可在 https://github.com/GenMI-Lab/EnTrust.git 获取。
cs.CV / 83 / 2606.21419
MIRCaps: A Large-Scale Mixed-Domain Dataset with Image-Level and Region-Level Captions for Fine-Grained Vision-Language Learning
MIRCaps：一个大规模混合领域数据集，包含图像级和区域级字幕，用于细粒度视觉语言学习
Abstract
Despite recent progress in Vision-Language Models (VLMs), mixed-domain image-caption datasets for both general-purpose and CCTV-based video surveillance systems remain limited. To address this gap, we introduce a large-scale multimodal dataset comprising 141,364 images, 981,947 image-level captions, 1,742,264 region-level captions, and 1,391,779 bounding box annotations. Each image is associated with an average of seven image-level captions describing different aspects of the overall scene, as well as seven region-level captions for each annotated bounding box. These complementary caption types are designed to help VLMs learn fine-grained visual attributes, including object categories, estimated sizes, colors, actions, states, and surrounding environmental context. We demonstrate the effectiveness of the dataset on two important downstream tasks: image captioning and object detection. Experimental results show that lightweight VLMs, including SmolVLM-256M-Instruct, BLIP, BLIP2, and Qwen2.5-VL 3B-Instruct, can be effectively fine-tuned using our dataset. Our dataset and code are publicly available at https://zenodo.org/records/20418601.
Chinese Translation
尽管视觉语言模型（VLMs）近年来取得了进展，但用于通用目的和基于闭路电视的视频监控系统的混合领域图像-字幕数据集仍然有限。为了解决这一问题，我们引入了一个大规模的多模态数据集，包含141,364张图像、981,947个图像级字幕、1,742,264个区域级字幕和1,391,779个边界框注释。每张图像平均关联七个图像级字幕，描述整体场景的不同方面，以及每个标注边界框的七个区域级字幕。这些互补的字幕类型旨在帮助VLMs学习细粒度的视觉属性，包括物体类别、估计尺寸、颜色、动作、状态和周围环境上下文。我们在两个重要的下游任务上展示了该数据集的有效性：图像字幕生成和物体检测。实验结果表明，包括SmolVLM-256M-Instruct、BLIP、BLIP2和Qwen2.5-VL 3B-Instruct在内的轻量级VLMs可以有效地使用我们的数据集进行微调。我们的数据集和代码已公开发布，网址为https://zenodo.org/records/20418601。
cs.CV / 84 / 2606.21446
Synergistic Dual-Branch Adaptation for Multi-modal Generalized Category Discovery
多模态广义类别发现的协同双分支适应
Abstract
Generalized Category Discovery (GCD) aims to classify old categories and discover new ones from unlabeled data. Recent multi-modal approaches introduce retrieved or synthesized texts into a dual-branch architecture to provide semantic cues complementary to visual features. However, the cross-modal synergy in existing dual-branch methods remains coarse and incomplete: the two modalities are encoded independently with the bias and noise in the derived text left unaddressed during encoding, and existing mutual learning strategies operate only on global class-level anchors, lacking fine-grained relational supervision. To address these limitations, we propose the Synergistic Dual-Branch Adaptation (SDBA) framework, which serves as a plug-and-play enhancement compatible with existing dual-branch methods such as GET and TextGCD. SDBA comprises two components: the cross-modal synergistic adapter inserts lightweight adapters into both branches and further injects visual information into the text adapter at each encoder layer to enhance text feature learning during encoding; the neighborhood mutual learning module enforces consistent local neighborhood distributions between the two branches via bidirectional KL divergence, providing fine-grained relational supervision for both old and new classes. Extensive experiments on six benchmarks demonstrate state-of-the-art performance, and consistent improvements on different baselines validate the broad scalability of the proposed framework.
Chinese Translation
广义类别发现（GCD）旨在对旧类别进行分类，并从未标记数据中发现新类别。最近的多模态方法将检索或合成的文本引入双分支架构，以提供与视觉特征互补的语义线索。然而，现有双分支方法中的跨模态协同仍然粗糙且不完整：两种模态独立编码，且在编码过程中未解决派生文本中的偏差和噪声，现有的互学习策略仅在全局类别锚点上操作，缺乏细粒度的关系监督。为了解决这些局限性，我们提出了协同双分支适应（SDBA）框架，该框架作为一种即插即用的增强方法，与现有的双分支方法（如GET和TextGCD）兼容。SDBA包括两个组件：跨模态协同适配器在两个分支中插入轻量级适配器，并在每个编码器层进一步将视觉信息注入文本适配器，以增强文本特征学习；邻域互学习模块通过双向KL散度强制执行两个分支之间一致的局部邻域分布，为旧类别和新类别提供细粒度的关系监督。在六个基准上的广泛实验表明了该方法的最先进性能，并且在不同基线上的一致性改进验证了所提框架的广泛可扩展性。
cs.CV / 85 / 2606.21456
Technical Report for ICRA 2026 GOOSE 2D Fine-Grained Semantic Segmentation Challenge: Exploring Query-Based Segmentation and Increased Spatial Context for Outdoor Scene Understanding
ICRA 2026 GOOSE 2D 精细语义分割挑战技术报告：探索基于查询的分割和增强空间上下文以理解户外场景
Abstract
In this report, we present our submission to the GOOSE 2D Fine-Grained Semantic Segmentation Challenge, organized as part of the Workshop on Field Robotics at ICRA 2026. The challenge combines data from the GOOSE and GOOSE-Ex datasets, which comprise more than 13k images captured from 4 distinct camera setups, annotated using a hierarchical taxonomy of 56 fine-grained classes and 11 broader categories. Starting from SegFormer as a baseline, we progressively improve segmentation performance through increased training crop sizes, a transition to the query-based Mask2Former architecture, and test-time augmentation. Our experiments show that query-based segmentation significantly outperforms the baseline model. Furthermore, increasing the crop size used during training yields substantial gains, highlighting the relevance of preserving scene context for fine-grained semantic disambiguation. Our final submission, using test-time augmentation, achieves an mIoU of 69.6% on the challenge test set, providing a strong baseline for fine-grained semantic segmentation in outdoor environments. To facilitate reproducibility and future research, code and weights will be made publicly available at https://github.com/RoboticsLabURJC/outdoor-fine-grained-segmentation .
Chinese Translation
在本报告中，我们展示了我们对 GOOSE 2D 精细语义分割挑战的提交，该挑战作为 ICRA 2026 现场机器人研讨会的一部分组织。该挑战结合了来自 GOOSE 和 GOOSE-Ex 数据集的数据，这些数据集包含来自 4 种不同相机设置的超过 13,000 张图像，使用 56 个精细类别和 11 个更广泛类别的分层分类法进行了注释。从 SegFormer 作为基线开始，我们通过增加训练裁剪大小、转向基于查询的 Mask2Former 架构以及测试时增强逐步提高分割性能。我们的实验表明，基于查询的分割显著优于基线模型。此外，增加训练期间使用的裁剪大小带来了显著的提升，突显了保留场景上下文在精细语义消歧中的重要性。我们的最终提交在挑战测试集上使用测试时增强实现了 69.6% 的 mIoU，为户外环境中的精细语义分割提供了强有力的基线。为了促进可重复性和未来的研究，代码和权重将公开发布在 https://github.com/RoboticsLabURJC/outdoor-fine-grained-segmentation 。
cs.CV / 86 / 2606.21463
Native space based pipelines outperform template space based pipeline in subcortical segmentation
基于原生空间的管道在皮层下分割中优于基于模板空间的管道
Abstract
Accurate segmentation of subcortical regions is critical for neurosurgical planning and functional research. Most automated methods rely on template space coregistration, which may compromise patient-specific accuracy, particularly in small structures. We identify a need to evaluate whether native space approaches offer a measurable advantage, which we evaluate in the context of movement disorders. We developed two UNet-based segmentation pipelines of the Subthalamic Nucleus (STN) - a common surgical target in Parkinson's Disease - and the neighbouring Red Nucleus (RN) and Substantia Nigra (SN). We collected 7T and 3T MRI data from five public datasets. The pipelines were evaluated in the native-space against manual labels. We further investigated the effect of the template resolution. Motivated by the hypothesis that models may better learn target boundaries in higher field, we tested the transferability of 7T-trained models to 3T clinical images, and whether synthetic 3T training data - generated via a disentangled representation learning method - could help bridging this domain gap. On held-out 7T data, the native pipeline consistently outperformed the template one. For the STN, native-space Dice reached 0.775 +- 0.055 versus 0.713 +- 0.051 (1 mm template), with HD95 of 0.79 +- 0.24 mm versus 1.17 +- 1.10 mm, respectively. Similar advantages were observed for the RN and SN. Increasing template resolution did not improve accuracy. When applied to 3T images, all models showed a considerable performance drop. Adding synthetic 3T data yielded only modest improvements, though without degrading 7T performance. Native-space segmentation is preferable for applications requiring patient specific anatomical fidelity, such as the surgical planning in PD. Bridging the 7T-to-3T domain gap remains an open challenge, motivating future work on domain adaptation tailored to subcortical structures.
Chinese Translation
准确分割皮层下区域对于神经外科规划和功能研究至关重要。大多数自动化方法依赖于模板空间的共配准，这可能会影响患者特异性的准确性，尤其是在小结构中。我们识别出需要评估原生空间方法是否提供可测量的优势，我们在运动障碍的背景下进行了评估。我们开发了两个基于UNet的分割管道，分别针对下丘脑（Subthalamic Nucleus, STN）——帕金森病中的常见手术靶点——以及邻近的红核（Red Nucleus, RN）和黑质（Substantia Nigra, SN）。我们从五个公共数据集中收集了7T和3T的MRI数据。管道在原生空间中与手动标签进行了评估。我们进一步研究了模板分辨率的影响。基于模型可能在更高场强下更好地学习目标边界的假设，我们测试了7T训练模型在3T临床图像上的可转移性，以及通过解耦表示学习方法生成的合成3T训练数据是否能够帮助弥合这一领域差距。在保留的7T数据上，原生管道的表现始终优于模板管道。对于STN，原生空间的Dice值达到了0.775 ± 0.055，而模板为1mm时为0.713 ± 0.051，HD95分别为0.79 ± 0.24 mm和1.17 ± 1.10 mm。RN和SN也观察到了类似的优势。提高模板分辨率并未改善准确性。当应用于3T图像时，所有模型的表现均显著下降。添加合成的3T数据仅带来了适度的改善，但没有降低7T的表现。对于需要患者特异性解剖精度的应用，如帕金森病的手术规划，原生空间分割更为可取。弥合7T与3T之间的领域差距仍然是一个开放的挑战，激励着未来针对皮层下结构的领域适应研究。
cs.CV / 87 / 2606.21493
Semi-Supervised Vision-Language-Action Model
半监督视觉-语言-动作模型
Abstract
Vision-Language-Action (VLA) models enable robots to predict actions directly from visual observations and language instructions, but adapting them to new environments still depends on costly action-labeled demonstrations. To reduce this dependence, we study semi-supervised VLA adaptation under limited supervision signals, where only a small portion of trajectories contain robot actions and the remaining trajectories provide action-unlabeled vision-language observations. Unlike standard semi-supervised learning, the missing supervision is an embodied action signal that must be visually grounded, language-consistent, physically feasible, and temporally stable. To address this problem, we propose SemiVLA, a self-distilled teacher-student framework that learns from reliable pseudo-actions on unlabeled trajectories. SemiVLA introduces a VLA-specific reliability controller to assess vision-language alignment, action feasibility, and temporal transition consistency, and further updates the teacher through a Bottleneck-Projected Alignment Update to avoid noisy feedback contamination. With OpenVLA as the backbone, SemiVLA consistently improves multiple PEFT strategies across LIBERO and CALVIN. Under 10\% labeled trajectories, SemiVLA with Selective LoRA achieves 89.0\% average success on LIBERO, outperforming supervised LoRA by 8.0 points without extra inference cost.
Chinese Translation
视觉-语言-动作（VLA）模型使机器人能够直接从视觉观察和语言指令中预测动作，但将其适应于新环境仍然依赖于昂贵的动作标注示例。为了减少这种依赖，我们研究了在有限监督信号下的半监督VLA适应，其中只有一小部分轨迹包含机器人动作，其余轨迹提供无动作标注的视觉-语言观察。与标准的半监督学习不同，缺失的监督是一个具体现实的动作信号，必须在视觉上有基础、语言上保持一致、物理上可行且时间上稳定。为了解决这个问题，我们提出了SemiVLA，一个自我蒸馏的师生框架，从无标注轨迹中的可靠伪动作中学习。SemiVLA引入了一个特定于VLA的可靠性控制器，以评估视觉-语言对齐、动作可行性和时间过渡一致性，并通过瓶颈投影对齐更新进一步更新教师，以避免噪声反馈污染。在OpenVLA作为骨干的基础上，SemiVLA在LIBERO和CALVIN上持续改善了多种PEFT策略。在10\%标注轨迹下，使用选择性LoRA的SemiVLA在LIBERO上实现了89.0\%的平均成功率，超越了监督LoRA 8.0个百分点，而没有额外的推理成本。
cs.CV / 88 / 2606.21562
Compressing Observation History into Agent Memory: Distilling Transformers into Recurrent Transformers
将观察历史压缩到智能体记忆中：将变换器蒸馏为递归变换器
Abstract
Transformers are AI's workhorse with strong performance in modeling sequential data, but their computational cost becomes prohibitive when processing long sequences. We target long-horizon streaming vision and robotics applications like map-free pose estimation, where it is particularly impractical to store and maintain a history of observations. Recurrent Transformers address this limitation by maintaining fixed-size memory but their performance lags behind that of transformers operating over the full observation history. We argue that this gap does not stem from architectural limitations, but from differences in how these models learn to compress past information. Without access to an observation history, recurrent models must explicitly decide what to retain in memory at each step, a significantly harder learning problem. In this work, we propose a distillation approach that transfers the compression strategy of a classical full-history transformer to a recurrent variant. We enable this by designing a teacher model that explicitly compresses its observation history into a fixed-size bottleneck representation. By directly supervising the student's memory with this bottleneck representation, we align the two compression mechanisms. We show that this approach allows to train a recurrent latent robotic memory with linear-time complexity while substantially narrowing the performance gap to full-history transformers.
Chinese Translation
变换器是人工智能在建模序列数据方面的主要工具，具有强大的性能，但在处理长序列时，其计算成本变得难以承受。我们针对长时间跨度的流媒体视觉和机器人应用，如无地图姿态估计，特别是在存储和维护观察历史时显得不切实际。递归变换器通过维持固定大小的记忆来解决这一限制，但其性能落后于处理完整观察历史的变换器。我们认为，这一差距并非源于架构限制，而是源于这些模型在学习压缩过去信息时的不同方式。在没有观察历史的情况下，递归模型必须在每一步明确决定保留什么在记忆中，这是一项显著更难的学习任务。在本研究中，我们提出了一种蒸馏方法，将经典全历史变换器的压缩策略转移到递归变体上。我们通过设计一个教师模型，明确将其观察历史压缩为固定大小的瓶颈表示，从而实现这一目标。通过直接用该瓶颈表示监督学生的记忆，我们对齐了这两种压缩机制。我们展示了这种方法可以以线性时间复杂度训练递归潜在机器人记忆，同时显著缩小与全历史变换器的性能差距。
cs.CV / 89 / 2606.21568
A Smart Classroom Behavior Analysis Framework with a New Highly Congested Classroom Dataset
一种智能课堂行为分析框架及其新构建的高度拥挤课堂数据集
Abstract
Student behavior detection is important for intelligent classroom analysis but remains challenging in large-class scenarios due to dense instance co-occurrence, asymmetric occlusion, depth-wise scale variation, and fine-grained semantic degradation in distant targets. Existing classroom behavior datasets and general-purpose detectors are insufficient to characterize and address these challenges. This paper constructs the Highly Congested Classroom Behavior (HCCB) dataset, containing 50,229 student behavior instances across seven categories: reading, writing, heads up, sleeping, looking around, bowing head, and using phone. HCCB provides a challenging benchmark that integrates dense distributions, severe occlusion, scale variation, and fine-grained behavioral semantics. To address these issues, we propose ODER-HSFNet, a YOLO-based detection framework tailored to highly crowded classrooms. At its core, ODER-HSFNet introduces three task-specific innovations: the Occlusion-aware Deformable Edge Rectifier (ODER), which strengthens boundary evidence under occlusion; the Hypergraph-State Spatial Fusion (HSSF) module, which integrates local structure enhancement, state-space contextual modeling, and high-order relation aggregation; and the Occlusion-Calibrated Detection Head (OCDetect), which suppresses low-quality Pre-NMS candidates and reduces false positives from occlusion boundaries and neighboring instances. Experiments on two classroom behavior detection datasets show that ODER-HSFNet outperforms mainstream YOLO-series methods, achieving 60.60%/80.12% mAP50:95/mAP50 on HCCB and 57.36%/74.65% on SCB-D3-S. Ablation studies further verify the effectiveness of the proposed design for highly crowded classroom behavior detection.
Chinese Translation
学生行为检测对于智能课堂分析至关重要，但在大班场景中仍然面临挑战，主要由于密集实例共现、非对称遮挡、深度尺度变化以及远距离目标的细粒度语义退化。现有的课堂行为数据集和通用检测器不足以有效表征和解决这些挑战。本文构建了高度拥挤课堂行为（Highly Congested Classroom Behavior，HCCB）数据集，包含了50,229个学生行为实例，涵盖七个类别：阅读、写作、抬头、睡觉、环顾四周、低头和使用手机。HCCB提供了一个具有挑战性的基准，整合了密集分布、严重遮挡、尺度变化和细粒度行为语义。为了解决这些问题，我们提出了ODER-HSFNet，这是一种基于YOLO的检测框架，专为高度拥挤的课堂设计。ODER-HSFNet的核心引入了三个特定任务的创新：遮挡感知可变形边缘整流器（Occlusion-aware Deformable Edge Rectifier，ODER），增强了遮挡下的边界证据；超图状态空间融合（Hypergraph-State Spatial Fusion，HSSF）模块，整合了局部结构增强、状态空间上下文建模和高阶关系聚合；以及遮挡校准检测头（Occlusion-Calibrated Detection Head，OCDetect），抑制低质量的预NMS候选框，并减少来自遮挡边界和相邻实例的误报。在两个课堂行为检测数据集上的实验表明，ODER-HSFNet的表现优于主流的YOLO系列方法，在HCCB上达到了60.60%/80.12%的mAP50:95/mAP50，在SCB-D3-S上达到了57.36%/74.65%。消融研究进一步验证了所提设计在高度拥挤课堂行为检测中的有效性。
cs.CV / 90 / 2606.21579
The Unreasonable Effectiveness of VLMs for Zero-shot Procedural Mistake Detection
视频语言模型在零-shot程序错误检测中的不合理有效性
Abstract
Procedural mistake detection is important for quality control and user assistance across many disciplines. Recent work in this field has achieved significant gains by using the reasoning capabilities of Video-Language Models (VLMs) as components within multi-stage pipelines, which consist of separate modules for supervised temporal action segmentation, error detection, and explainability. Consequently, they remain dependent on tailored training datasets and require task-specific training, limiting their wider applicability. To remedy this, we introduce zero-shot procedural mistake detection and propose a unified Zero-shot Procedural Mistake detection (ZeProM) framework that jointly solves procedural mistake detection and temporal action segmentation with a single pre-trained VLM. By evaluating our framework on two canonical mistake detection benchmarks, EgoPER and CaptainCook4D, we find that ZeProM can perform these tasks successfully, while approaching, or even outperforming, the performance of fully supervised methods. For instance, we achieve a 4.4 point improvement in EDA and a 2.0 point improvement in
[email protected] on average over all five EgoPER tasks compared to the strongest supervised methods. Overall, our results show the potential of unified methods for procedural mistake detection, and we hope this will steer the field away from highly complex pipelines and toward more generally applicable solutions.
Chinese Translation
程序错误检测在许多学科中对于质量控制和用户辅助至关重要。该领域的近期研究通过将视频语言模型（Video-Language Models, VLMs）的推理能力作为多阶段流程中的组件，取得了显著进展，这些流程由监督的时间动作分割、错误检测和可解释性等独立模块组成。因此，它们仍然依赖于定制的训练数据集，并需要特定任务的训练，这限制了它们的广泛适用性。为了解决这个问题，我们引入了零-shot程序错误检测，并提出了一个统一的零-shot程序错误检测（Zero-shot Procedural Mistake detection, ZeProM）框架，该框架通过单个预训练的VLM共同解决程序错误检测和时间动作分割。通过在两个经典的错误检测基准EgoPER和CaptainCook4D上评估我们的框架，我们发现ZeProM能够成功执行这些任务，并接近甚至超越完全监督方法的性能。例如，与最强的监督方法相比，我们在所有五个EgoPER任务上平均实现了4.4点的EDA提升和2.0点的
[email protected]提升。总体而言，我们的结果显示了统一方法在程序错误检测中的潜力，我们希望这能引导该领域远离高度复杂的流程，朝着更普遍适用的解决方案发展。
cs.CV / 91 / 2606.21590
Radial Basis Function Networks as Projection Heads in Self-Supervised Learning
径向基函数网络作为自监督学习中的投影头
Abstract
Self-supervised learning (SSL) typically relies on a backbone encoder followed by a small multilayer perceptron (MLP) projection head, which is conventionally discarded after training, while backbone quality is assessed via costly linear probing on labeled data. We argue that this approach including discarding the projector is rather computationally wasteful. Instead, we propose replacing the MLP head with a radial basis function network (RBFN), whose interpretable center and shape parameters can be exploited to judge representation quality without labels or a separate classifier. To this end, we introduce Scale-Normalized Separation (SNS), a novel label-free quality metric derived solely from the kernel centers and shapes learned during training. Across five canonical SSL architectures (MoCo, SimCLR, BYOL, SwAV and SimSiam) and four image classification datasets, we show that RBFN projection heads are competitive drop-in replacements for standard MLP projectors. We recommend constructing them with three RBF layers activated by the Gaussian radial basis function. Moreover, SNS exhibits strong to very strong positive correlation with established logistic regression metrics, demonstrating that a trained RBFN projector can act as a reliable proxy for backbone representation quality. We additionally publish a novel PyTorch compatible image classification dataset based on Google's Open Images V7 to facilitate reproducible research into representation learning.
Chinese Translation
自监督学习（SSL）通常依赖于一个主干编码器，后接一个小型多层感知器（MLP）投影头，该投影头在训练后通常被丢弃，而主干的质量则通过在标记数据上的高成本线性探测来评估。我们认为，这种包括丢弃投影器的方法在计算上是相当浪费的。相反，我们建议用径向基函数网络（RBFN）替代MLP头，其可解释的中心和形状参数可以在没有标签或单独分类器的情况下用于判断表示质量。为此，我们引入了尺度归一化分离（SNS），这是一种全新的无标签质量度量，仅基于在训练过程中学习到的核中心和形状。通过五种经典的SSL架构（MoCo、SimCLR、BYOL、SwAV和SimSiam）和四个图像分类数据集，我们展示了RBFN投影头是标准MLP投影器的竞争性替代品。我们建议使用三个由高斯径向基函数激活的RBF层来构建它们。此外，SNS与已建立的逻辑回归度量表现出强到非常强的正相关，证明训练后的RBFN投影器可以作为主干表示质量的可靠代理。我们还发布了一个基于谷歌开放图像V7的新的PyTorch兼容图像分类数据集，以促进对表示学习的可重复研究。
cs.CV / 92 / 2606.21594
Boundary-by-Mask: Few-Shot Instance Segmentation with Mask-Conditioned Boundary Learning for Texture-Poor Industrial Parts
基于边界的掩膜：针对纹理稀缺工业部件的少样本实例分割与掩膜条件边界学习
Abstract
Recent advances in large pre-trained models have led to remarkable progress in instance segmentation on general images. However, industrial scenarios remain challenging. Instance definitions are often application-specific and inconsistent, and the domain gap from general imagery is substantial due to weak textures and limited contextual cues. Consequently, a direct application of existing models is unreliable. We propose Boundary-by-Mask, a few-shot instance segmentation framework that supervises boundaries instead of interior appearance. Given a few RGB images and corresponding instance masks, the method extracts rich visual features using a foundation-model encoder and trains a lightweight Signed Distance Function (SDF) head to predict boundary-aware distance maps. Segmentation masks are obtained through an SDF-to-mask reconstruction process. By explicitly estimating contours, the framework achieves reliable instance separation even on low-texture and color-uniform surfaces. The instance definition is conditioned by the instance mask. Replacing the mask specifies the segmentation target, such as the whole object or a sub-part. A pixel-wise shallow MLP head enables rapid training. Experiments on industrial parts and food items with ambiguous boundaries show strong few-shot generalization, robustness in feature-poor conditions, and precise control over mask-level targets.
Chinese Translation
近期大型预训练模型的进展在一般图像的实例分割上取得了显著进展。然而，工业场景仍然面临挑战。实例定义通常是特定于应用的且不一致，并且由于纹理较弱和上下文线索有限，导致与一般图像之间存在显著的领域差距。因此，现有模型的直接应用是不可靠的。我们提出了基于边界的掩膜（Boundary-by-Mask），这是一种少样本实例分割框架，监督边界而非内部外观。给定少量RGB图像及其对应的实例掩膜，该方法利用基础模型编码器提取丰富的视觉特征，并训练一个轻量级的有符号距离函数（Signed Distance Function, SDF）头来预测边界感知距离图。通过SDF到掩膜的重建过程获得分割掩膜。通过显式估计轮廓，该框架即使在低纹理和颜色均匀的表面上也能实现可靠的实例分离。实例定义由实例掩膜条件化。替换掩膜可以指定分割目标，例如整个物体或其子部分。逐像素的浅层多层感知机（MLP）头使得训练迅速。针对具有模糊边界的工业部件和食品项目的实验显示出强大的少样本泛化能力、在特征稀缺条件下的鲁棒性，以及对掩膜级目标的精确控制。
cs.CV / 93 / 2606.21596
$\phi$-Scene: Physically Grounded Image-to-3D Scene Reconstruction
$ heta$-场景：基于物理的图像到3D场景重建
Abstract
Reconstructing compositional 3D scenes from a single image is a fundamental challenge in 3D world modeling. Recent methods can recover high-fidelity, complete 3D objects and predict plausible scene arrangements, but most still treat scene reconstruction primarily as a visual and geometric prediction problem. Their outputs may therefore contain floating objects, interpenetrations, or unstable-contact artifacts, limiting their physical validity and downstream usability in simulation, robotics, and interactive environments. We present $\phi$-Scene, a physically grounded approach to open-vocabulary and compositional image-to-3D scene reconstruction. The key premise is that a reconstructed scene should not be treated merely as a set of objects with predicted poses, but as a stable physical system. Accordingly, $\phi$-Scene formulates reconstruction as topology-driven physical assembly: it infers how objects support one another, orders them accordingly, and progressively settles each object against its already stabilized support context. For each object in topological order, SDF-based optimization first resolves penetrations against the pre-settled support context, and rigid-body simulation then settles the object into a stable contact configuration under real-world physical constraints. Experiments on 3D-Front show that $\phi$-Scene achieves the strongest overall performance among out-of-domain methods and remains highly competitive with in-domain baselines. Human and VLM evaluations further show strong preference for $\phi$-Scene in visual quality, reference alignment, and physical plausibility. Finally, dedicated physical plausibility metrics covering static contact quality and dynamic stability demonstrate that $\phi$-Scene substantially reduces penetration artifacts while producing much lower post-simulation drift, indicating more stable and physically grounded 3D scenes.
Chinese Translation
从单幅图像重建组合3D场景是3D世界建模中的一个基本挑战。近期的方法能够恢复高保真、完整的3D物体并预测合理的场景排列，但大多数仍将场景重建主要视为视觉和几何预测问题。因此，它们的输出可能包含漂浮物体、相互穿透或不稳定接触伪影，限制了其在仿真、机器人和交互环境中的物理有效性和下游可用性。我们提出了$ heta$-场景，一种基于物理的开放词汇和组合图像到3D场景重建的方法。其关键前提是，重建的场景不应仅被视为一组具有预测姿态的物体，而应视为一个稳定的物理系统。因此，$ heta$-场景将重建形式化为以拓扑驱动的物理组装：它推断物体如何相互支撑，并相应地对其进行排序，逐步将每个物体固定在其已稳定的支撑上下文中。对于每个按拓扑顺序排列的物体，基于SDF的优化首先解决与预先稳定的支撑上下文的穿透问题，然后刚体仿真将物体固定在符合现实物理约束的稳定接触配置中。在3D-Front上的实验表明，$ heta$-场景在领域外方法中实现了最强的整体性能，并在与领域内基准的比较中保持高度竞争力。人类和VLM评估进一步显示，在视觉质量、参考对齐和物理合理性方面，$ heta$-场景受到强烈偏好。最后，专门的物理合理性指标涵盖静态接触质量和动态稳定性，表明$ heta$-场景显著减少了穿透伪影，同时产生了更低的后仿真漂移，表明生成了更稳定和基于物理的3D场景。
cs.CV / 94 / 2606.21605
$\mu$Match: Foundation Models for Semi-supervised Learning and Domain Adaptation in EM
$bc$Match：用于EM中的半监督学习和领域适应的基础模型
Abstract
Vision foundation models have substantially advanced computer vision, enabling state-of-the-art performance in zero- and few-shot settings. They have been successfully applied to biomedical imaging tasks ranging from organ segmentation in computed tomography to cell segmentation in light microscopy. Electron microscopy (EM) is a central modality for analyzing cellular ultrastructure due to its nanometer-scale resolution. However, the application of foundation models in EM has so far been limited to specific organelles, such as mitochondria, largely due to the diversity of segmentation tasks and the scarcity of comprehensively annotated data. As a result, EM segmentation still predominantly relies on supervised learning, requiring extensive manual annotation and limiting ultrastructural analysis. To address this gap, we propose $\mu$Match, a framework for semi-supervised learning and domain adaptation that leverages foundation models. We implement state-of-the-art student-teacher-based methods and evaluate multiple foundation models (SAM, SAM2, $\mu$SAM, DINOv2/v3) on challenging EM tasks, including mitochondrion, nucleus, and neurite segmentation. Our results demonstrate consistent improvements over strong baselines and highlight a path toward substantially reducing the annotation effort in EM.
Chinese Translation
视觉基础模型在计算机视觉领域取得了显著进展，使得在零样本和少样本设置下实现了最先进的性能。它们已成功应用于生物医学成像任务，从计算机断层扫描中的器官分割到光学显微镜中的细胞分割。电子显微镜（EM）由于其纳米级分辨率，是分析细胞超微结构的核心工具。然而，基础模型在EM中的应用目前仅限于特定的细胞器，如线粒体，这在很大程度上是由于分割任务的多样性和全面注释数据的稀缺。因此，EM分割仍主要依赖于监督学习，需要大量的手动注释，限制了超微结构分析。为了解决这一问题，我们提出了$bc$Match，一个利用基础模型的半监督学习和领域适应框架。我们实现了最先进的基于学生-教师的方法，并在具有挑战性的EM任务上评估了多种基础模型（SAM、SAM2、$bc$SAM、DINOv2/v3），包括线粒体、细胞核和神经突起分割。我们的结果显示出相对于强基线的一致性改善，并指出了一条显著减少EM注释工作量的路径。
cs.CV / 95 / 2606.21607
T-MOR: Learning Motion-Aware Skeleton Representations for Human Action Recognition
T-MOR：学习运动感知骨架表示以进行人类动作识别
Abstract
Vision-language models such as CLIP have recently achieved strong performance on a wide range of visual understanding tasks. However, most existing models rely primarily on appearance-level supervision from images or videos, and do not explicitly model human motion, which is essential for fine-grained and human-centric action recognition task as actions are defined by temporally structured and physically grounded body movements. To address this problem, we propose Transferable skeleton MOtion Representation (T-MOR), a motion-aware framework that learns transferable action representations from skeleton sequences with the aid of video and language supervision during training. T-MOR adopts a multi-modal contrastive learning scheme that aligns skeleton motion with visual and textual representations, while performing inference using only lightweight skeleton inputs. To support large-scale pre-training, we construct PoseCap-1M, a new dataset that contains over one million synchronized video, skeleton, and text triplets covering diverse human activities. We evaluate T-MOR on a range of human-centric action recognition benchmarks, including action classification and frame-wise temporal detection. Experimental results show that T-MOR consistently improves performance across multiple datasets, such as Toyota Smarthome, Penn Action, UAV-Human, TSU, and Charades. In addition, T-MOR demonstrates strong generalization ability in few-shot and zero-shot settings, highlighting the effectiveness of motion-centric and embodied representations for transferable action understanding.
Chinese Translation
视觉语言模型如CLIP最近在广泛的视觉理解任务中取得了强劲的表现。然而，大多数现有模型主要依赖于来自图像或视频的外观级监督，并未明确建模人类运动，而运动对于细粒度和以人为中心的动作识别任务至关重要，因为动作是由时间结构化和物理基础的身体运动定义的。为了解决这个问题，我们提出了可转移骨架运动表示（Transferable skeleton MOtion Representation，T-MOR），这是一个运动感知框架，能够在训练过程中借助视频和语言监督从骨架序列中学习可转移的动作表示。T-MOR采用了一种多模态对比学习方案，将骨架运动与视觉和文本表示对齐，同时在推理时仅使用轻量级的骨架输入。为了支持大规模的预训练，我们构建了PoseCap-1M，这是一个新的数据集，包含超过一百万个同步的视频、骨架和文本三元组，涵盖了多样的人类活动。我们在一系列以人为中心的动作识别基准上评估了T-MOR，包括动作分类和逐帧时间检测。实验结果表明，T-MOR在多个数据集上始终提高了性能，如Toyota Smarthome、Penn Action、UAV-Human、TSU和Charades。此外，T-MOR在少样本和零样本设置中表现出强大的泛化能力，突显了以运动为中心和具身表示在可转移动作理解中的有效性。
cs.CV / 96 / 2606.21608
CurvSegFlow: Time-Conditioned Flow Matching for Robust Segmentation of Curvilinear Structures in Noisy Biomedical Images
CurvSegFlow：基于时间条件流匹配的噪声生物医学图像中曲线结构的鲁棒分割
Abstract
Accurate segmentation of curvilinear structures remains challenging in biomedical imaging due to their thin geometry, complex topology, and sensitivity to noise. This is particularly critical for microscopy images of cytoskeletal network, where low signal-to-noise ratios and dense filament crossings often lead to fragmented or inaccurate segmentation. In this work, we propose CurvSegFlow, a segmentation framework based on time-conditioned flow matching. Instead of predicting a segmentation mask in a single pass, the method models segmentation as a dynamic process that progressively refines a noisy initialization into the target structure through a learned velocity field. The proposed model combines a U-Net backbone with triple-term loss function and temporal embeddings to guide the refinement process across reconstruction stages. This formulation enables gradual error correction and improves the continuity of thin structures. CurvSegFlow is evaluated on multiple synthetic and real microtubule datasets, as well as on public benchmarks of retinal vessels, corneal nerves and coronary arteries. Across datasets, the method achieves competitive or superior performance compared to established segmentation models, with consistent improvements in precision and structural continuity, particularly under low signal-to-noise conditions. These results show that flow-based iterative refinement provides a robust and general framework for curvilinear structure segmentation. Overall, the proposed approach improves segmentation quality in challenging imaging conditions and generalizes effectively across modalities without architectural changes.
Chinese Translation
在生物医学成像中，准确分割曲线结构仍然具有挑战性，原因在于其细微的几何形状、复杂的拓扑结构以及对噪声的敏感性。这在细胞骨架网络的显微镜图像中尤为重要，因为低信噪比和密集的纤维交叉常常导致分割碎片化或不准确。在本研究中，我们提出了CurvSegFlow，一个基于时间条件流匹配的分割框架。该方法并不是在一次性预测分割掩膜，而是将分割建模为一个动态过程，通过学习的速度场逐步将噪声初始化精炼为目标结构。所提出的模型结合了U-Net骨干网络、三项损失函数和时间嵌入，以指导重建阶段的精炼过程。这种表述使得逐步错误修正成为可能，并改善了细结构的连续性。CurvSegFlow在多个合成和真实微管数据集上进行了评估，以及在视网膜血管、角膜神经和冠状动脉的公共基准测试中进行验证。在各个数据集中，该方法的性能与已建立的分割模型相比具有竞争力或优越性，在低信噪条件下特别在精度和结构连续性方面表现出一致的改善。这些结果表明，基于流的迭代精炼为曲线结构分割提供了一个鲁棒且通用的框架。总体而言，所提出的方法在具有挑战性的成像条件下提高了分割质量，并在不同模态之间有效地泛化，而无需架构更改。
cs.CV / 97 / 2606.21613
Cross-Modal Corroboration for Annotation-Free Wildlife Monitoring
无注释野生动物监测的跨模态验证
Abstract
Scaling wildlife monitoring for real-world conservation deployments requires automated analysis of smart sensors that operate under severe annotation scarcity. We propose leveraging expert knowledge of species activity patterns as an annotation-free validation signal for multimodal monitoring pipelines. We operationalize agreement as the alignment of independently derived hourly activity curves both with each other and with published behavioral priors-a three-way convergence that rules out shared-data confounds and dataset-internal correlation as alternative explanations. Our vision pipeline combines zero-shot species detection via BioCLIP 2, sliced inference to handle deployment-constrained camera positioning, and geometry-based geographic localization from camera trap imagery. Our acoustic pipeline detects species vocalizations via a fine-tuned classifier. We validate the pipeline on a breeding herd of Milu deer and demonstrate that both modalities independently recover activity patterns consistent with known deer behavioral ecology with minimal manual annotation. The framework applies to species detectable in both visual and acoustic modalities for which behavioral priors are documented in the literature, suggesting a practical path toward self-validating wildlife-monitoring pipelines at conservation scale.
Chinese Translation
在现实世界的保护部署中，扩大野生动物监测的规模需要对在严重缺乏注释的情况下运行的智能传感器进行自动化分析。我们提出利用物种活动模式的专家知识作为无注释的验证信号，用于多模态监测管道。我们将一致性操作化为独立推导的每小时活动曲线之间的对齐，以及与已发布的行为先验之间的对齐——这种三方收敛排除了共享数据混淆和数据集内部相关性作为替代解释。我们的视觉管道结合了通过 BioCLIP 2 的零样本物种检测、切片推理以处理受限部署的相机位置，以及基于几何的相机陷阱图像地理定位。我们的声学管道通过精细调整的分类器检测物种的声音。我们在一群繁殖的梅鹿（Milu deer）上验证了该管道，并证明这两种模态独立恢复了与已知鹿类行为生态一致的活动模式，且手动注释最小。该框架适用于在文献中有行为先验记录的可在视觉和声学模态中检测的物种，暗示了在保护规模下实现自验证野生动物监测管道的实际路径。
cs.CV / 98 / 2606.21623
A DVDrive Approach for doScenes Instructed Driving Challenge
基于DVDrive的方法应对doScenes指令驾驶挑战
Abstract
Instruction-conditioned trajectory prediction is an emerging problem in autonomous driving, where a model predicts the future ego trajectory not only from visual scene context and historical motion, but also from a natural-language maneuver instruction. This paper presents our submission to the doScenes Instructed Driving Challenge, built upon OmniDrive, a vision-language-action driving agent with 3D perception, reasoning, and planning capabilities. We adapt OmniDrive to the doScenes setting by training it on instruction-annotated nuScenes scenes and generating a 6-second ego trajectory represented by 12 future waypoints. To improve multi-view visual grounding, we further introduce a DVPE-style divided-view perception module into the OmniDrive perception head. Instead of attending globally to all camera features, the proposed module groups query features and image tokens into divided local view spaces and performs visibility-aware cross-attention within each view. This design reduces irrelevant cross-view interference and helps the model better align language instructions with local driving-relevant visual evidence. The code is publicly available at: https://github.com/feel12348/doscenes-omnidrive.
Chinese Translation
指令条件下的轨迹预测是自动驾驶领域中的一个新兴问题，其中模型不仅根据视觉场景上下文和历史运动预测未来的自我轨迹，还基于自然语言的操作指令进行预测。本文展示了我们对doScenes指令驾驶挑战的提交，构建于OmniDrive之上，这是一种具备3D感知、推理和规划能力的视觉-语言-动作驾驶代理。我们通过在带有指令注释的nuScenes场景上训练OmniDrive，适应doScenes设置，并生成一个由12个未来路标表示的6秒自我轨迹。为了改善多视角视觉定位，我们进一步在OmniDrive的感知模块中引入了一种DVPE风格的分视图感知模块。该模块不再对所有摄像头特征进行全局关注，而是将查询特征和图像标记分组到划分的局部视图空间中，并在每个视图内执行基于可见性的交叉注意。这一设计减少了不相关的视图间干扰，帮助模型更好地将语言指令与局部驾驶相关的视觉证据对齐。代码已公开可用，链接为：https://github.com/feel12348/doscenes-omnidrive。
cs.CV / 99 / 2606.21657
Chehre: An Emoji-Prompted Video Dataset for Perceptually Diverse Facial Expression Recognition
Chehre：一个基于表情符号提示的视频数据集，用于感知多样性的面部表情识别
Abstract
Facial expressions are nonverbal social signals used in human interaction, but facial expression recognition datasets often focus on static images, basic emotion categories, or single deterministic annotations. We introduce Chehre, an emoji-prompted video dataset for analyzing dynamic facial expressions across a wide range of expressions for exploring inter-individual perceptual diversity. In Chehre, participants were prompted to express and record 40 facial emojis. Later, their facial motions were transferred onto synthetic faces to preserve privacy. A separate group of annotators analyzed the anonymized videos using emoji and label annotations, resulting in 2,111 high quality videos collected from 203 performers and validated by 902 annotators. We define two benchmark tasks: dominant expression recognition, which tests whether models recover the top human-rated labels, and distributional expression recognition, which tests whether models capture the diversity of human responses. We benchmark recent vision-language models using random sampling and persona prompting to generate multiple predictions per video. Results show that both tasks are challenging: among the models evaluated, the best-performing model achieves only 32.5% Top-1 accuracy on dominant expression recognition and a Spread Ratio well below the human reference on distributional recognition. Chehre provides a benchmark for evaluating diverse, dynamic, and distributional facial expression recognition
Chinese Translation
面部表情是人际交往中使用的非语言社交信号，但现有的面部表情识别数据集往往集中于静态图像、基本情感类别或单一确定性标注。我们介绍了Chehre，一个基于表情符号提示的视频数据集，用于分析动态面部表情，涵盖广泛的表情，以探索个体间的感知多样性。在Chehre中，参与者被提示表达并记录40种面部表情符号。随后，他们的面部动作被转移到合成面孔上，以保护隐私。一个独立的标注组使用表情符号和标签标注分析了匿名视频，最终收集到来自203名表演者的2111个高质量视频，并由902名标注者进行了验证。我们定义了两个基准任务：主导表情识别，测试模型是否能恢复人类评分的最高标签，以及分布式表情识别，测试模型是否能捕捉人类反应的多样性。我们使用随机采样和个性提示对近期的视觉-语言模型进行基准测试，以生成每个视频的多个预测。结果表明，这两个任务都具有挑战性：在评估的模型中，表现最佳的模型在主导表情识别任务上的Top-1准确率仅为32.5%，而在分布式识别任务中的Spread Ratio远低于人类参考值。Chehre为评估多样化、动态和分布式的面部表情识别提供了基准。
cs.CV / 100 / 2606.21661
UnityShots: Memory-Driven Multi-Shot Audio-Video Generation with Boundary-Aware Gating
UnityShots：基于记忆驱动的多镜头音视频生成与边界感知门控
Abstract
Generating a coherent multi-shot video requires structured cross-shot memory. Subject appearance, scene context, and speaker identity must persist across cuts. Existing approaches either train end-to-end over fixed-length sequences and cannot scale, generate shot-by-shot with memory banks that grow linearly, or orchestrate pretrained generators under an LLM planner without a multi-shot-aware backbone. We present UnityShots, a memory-driven multi-shot audio-video generation system built on LTX-2.3, trained on annotated cinematic and music-video shots. The video stream maintains two fixed-size slots, a long-term memory (LTM) slot anchored to the opening shot and a short-term memory (STM) slot holding the immediately preceding tail, both updated at every cut by a boundary-conditioned gate that fuses visual cut probability and beat-tracker signals. The audio stream injects a reference speaker token at every shot to preserve vocal timbre without a sliding audio bank. A discrete cut-type prior, learned through AdaLN, becomes an inference-time control knob over transition strength. We release a benchmark of $200$ multi-cultural multi-shot sequences spanning six ethnic regions and ten or more languages, with per-shot reference identities, reference audio, and per-boundary transition labels. Evaluated across I2V, T2V, and R2V conditioning modes, UnityShots leads open-source baselines on every cross-shot coherence metric and matches the strongest closed-source system on the multi-shot axes.
Chinese Translation
生成连贯的多镜头视频需要结构化的跨镜头记忆。主体外观、场景背景和说话者身份必须在切换中持续存在。现有的方法要么在固定长度序列上进行端到端训练，无法扩展，要么逐镜头生成，使用线性增长的记忆库，或者在没有多镜头感知骨干的情况下在大型语言模型（LLM）规划器下协调预训练生成器。我们提出了UnityShots，一个基于记忆驱动的多镜头音视频生成系统，构建于LTX-2.3之上，训练于标注的电影和音乐视频镜头。视频流维护两个固定大小的槽，一个锚定于开头镜头的长期记忆（LTM）槽和一个持有紧接着前一尾部的短期记忆（STM）槽，这两个槽在每次切换时通过一个边界条件门进行更新，该门融合了视觉切换概率和节拍跟踪信号。音频流在每个镜头中注入一个参考说话者标记，以保持声音音色，而无需滑动音频库。通过AdaLN学习的离散切换类型先验，成为推理时过渡强度的控制旋钮。在六个民族区域和十种或更多语言中，我们发布了一个包含200个多文化多镜头序列的基准数据集，包含每个镜头的参考身份、参考音频和每个边界的过渡标签。在I2V、T2V和R2V条件模式下进行评估，UnityShots在每个跨镜头一致性指标上领先于开源基线，并在多镜头轴上与最强的闭源系统相匹配。
cs.CV / 101 / 2606.21674
Enlight: Fast Low-Light Image Enhancement via Multi-Objective Optimization and Shadow-Aware Refinement
Enlight：通过多目标优化和阴影感知细化实现快速低光照图像增强
Abstract
We present ENLIGHT, a fast and training free framework for low-light image enhancement based on direct optimization of a perceptual objective. Unlike deep learning approaches that require large scale training data and supervision, ENLIGHT operates in a zero-shot manner by optimizing image quality at inference time. The method employs a two stage global to local optimization strategy. In the first stage, ENLIGHT performs global illumination adjustment to improve visibility while maintaining structural consistency and avoiding excessive noise enhancement. In the second stage, a shadow aware refinement selectively improves low-intensity regions through masked local optimization, enhancing visibility without overexposure. To balance quality and efficiency, we introduce two modes: Fast, which uses a multi-objective formulation combining entropy, gradient preservation, and noise regularization, and Ultrafast, which reduces computational cost via a lightweight approximation of the same objective. The framework is optimizer agnostic and supports both evolutionary and lightweight local search methods. Experiments on BAID, Backlit300, LIME, MEF, NPE, and DICM demonstrate that ENLIGHT achieves competitive perceptual quality (MUSIQ, NIQE, BRISQUE) with significantly lower inference time. Qualitative results further show improved contrast, preserved structural details, and controlled noise amplification, making ENLIGHT a practical and interpretable alternative to learning based methods.
Chinese Translation
我们提出了ENLIGHT，一个快速且无需训练的低光照图像增强框架，基于感知目标的直接优化。与需要大规模训练数据和监督的深度学习方法不同，ENLIGHT以零-shot的方式在推理时优化图像质量。该方法采用两阶段的全局到局部优化策略。在第一阶段，ENLIGHT进行全局照明调整，以改善可见性，同时保持结构一致性，避免过度噪声增强。在第二阶段，阴影感知细化通过掩膜局部优化选择性地改善低强度区域，提高可见性而不造成过曝。为了平衡质量和效率，我们引入了两种模式：快速模式（Fast），使用结合熵、梯度保持和噪声正则化的多目标公式；超快速模式（Ultrafast），通过对相同目标的轻量级近似减少计算成本。该框架与优化器无关，支持进化和轻量级局部搜索方法。在BAID、Backlit300、LIME、MEF、NPE和DICM上的实验表明，ENLIGHT在显著降低推理时间的同时，达到了具有竞争力的感知质量（MUSIQ、NIQE、BRISQUE）。定性结果进一步显示了对比度的改善、结构细节的保留和噪声放大的控制，使ENLIGHT成为学习基方法的实用且可解释的替代方案。
cs.CV / 102 / 2606.21700
VT-DUDA: Visual Token Conditioning for Diffusion-guided Unsupervised Domain Adaptation
VT-DUDA：用于扩散引导的无监督领域适应的视觉标记条件框架
Abstract
Unsupervised domain adaptation (UDA) aims to learn a target-domain classifier from labeled source data and unlabeled target data under distribution shift. Recent diffusion-based UDA methods approach this problem by synthesizing labeled target-style images and training on the resulting synthetic data. However, their performance depends heavily on the conditioning design: class prompts provide only coarse guidance, while domain adaptation modules mainly control appearance, which may leave target-style synthesis insufficiently specified. We propose VT-DUDA, a visual-token conditioning framework for diffusion-guided UDA. Instead of relying only on text prompts, VT-DUDA uses source images to provide additional instance-level visual context for target-style synthesis. Specifically, VT-DUDA maps each source image to a compact sequence of visual tokens and forms a hybrid conditioning context by concatenating these tokens with the corresponding text embeddings along the cross-attention context dimension of a latent diffusion model. This provides instance-dependent conditioning beyond text alone, while synthesis is performed with the target-domain adapter branch. Because guidance is represented explicitly as a token sequence, the same interface also permits inference-time manipulation of the conditioning signal through token selection and token-strength adjustment. The proposed method preserves the standard diffusion objective and can be integrated into existing adapter-based diffusion frameworks without modifying the backbone. Across Office-31, Office-Home, and VisDA-2017, VT-DUDA improves average target-domain accuracy over strong discriminative and diffusion-based UDA baselines. The results suggest that, in generation-based UDA, a stronger conditioning interface can improve the downstream usefulness of synthetic target-style data.
Chinese Translation
无监督领域适应（UDA）旨在在分布变化下从标记的源数据和未标记的目标数据中学习目标领域分类器。最近的基于扩散的UDA方法通过合成标记的目标风格图像并在生成的合成数据上进行训练来解决这一问题。然而，它们的性能在很大程度上依赖于条件设计：类别提示仅提供粗略指导，而领域适应模块主要控制外观，这可能导致目标风格合成不足以明确指定。我们提出了VT-DUDA，一个用于扩散引导的UDA的视觉标记条件框架。VT-DUDA不再仅依赖文本提示，而是利用源图像为目标风格合成提供额外的实例级视觉上下文。具体而言，VT-DUDA将每个源图像映射到一组紧凑的视觉标记序列，并通过将这些标记与相应的文本嵌入在潜在扩散模型的交叉注意力上下文维度上进行连接，从而形成混合条件上下文。这提供了超越文本的实例依赖条件，同时合成是在目标领域适配器分支上进行的。由于指导明确表示为标记序列，因此相同的接口也允许通过标记选择和标记强度调整在推理时操控条件信号。所提出的方法保留了标准的扩散目标，并且可以在不修改主干的情况下集成到现有的基于适配器的扩散框架中。在Office-31、Office-Home和VisDA-2017数据集上，VT-DUDA在强大的判别和基于扩散的UDA基线之上提高了平均目标领域准确率。结果表明，在基于生成的UDA中，更强的条件接口可以提高合成目标风格数据的下游实用性。
cs.CV / 103 / 2606.21705
Structural Assessment for Understanding and Guiding Dataset Distillation in Discrete Token Space
理解和指导离散标记空间中的数据集蒸馏的结构评估
Abstract
Dataset distillation (DD) has proven to reduce training cost while preserving accuracy. While promising, the factors that make one distilled dataset more effective than another remain poorly understood. In this work, we investigate this question through the lens of discrete visual tokenizers. Whereas many prior DD efforts emphasize matching global data distributions, we suggest that the effectiveness depends on which semantic concepts are captured and how they are composed. Discrete visual tokenizers provide a finite vocabulary that enables direct statistical analysis of such compositional structure. Through quantitative analysis of token-level statistics, we introduce the structural score to measure the adequacy of token compositions. We observe that distilled datasets with balanced token composition yield higher validation performance. On the other hand, divergence from the original data does not necessarily harm performance. We further show that samples with high structural scores in the discrete token space can effectively guide diffusion-based DD. Our findings highlight the importance of token composition in dataset effectiveness, offering a principled complement to distributional similarity considerations in DD.
Chinese Translation
数据集蒸馏（Dataset Distillation, DD）已被证明可以在保持准确性的同时降低训练成本。尽管前景可期，但使得一个蒸馏数据集比另一个更有效的因素仍然不甚明了。在本研究中，我们通过离散视觉标记器的视角来探讨这个问题。许多先前的DD研究强调匹配全局数据分布，而我们建议其有效性取决于捕捉到的语义概念以及它们的组合方式。离散视觉标记器提供了一个有限的词汇，使得对这种组合结构进行直接的统计分析成为可能。通过对标记级统计数据的定量分析，我们引入了结构得分来衡量标记组合的适宜性。我们观察到，具有平衡标记组合的蒸馏数据集在验证性能上表现更高。另一方面，偏离原始数据并不一定会损害性能。我们进一步展示了在离散标记空间中具有高结构得分的样本可以有效指导基于扩散的DD。我们的研究结果强调了标记组合在数据集有效性中的重要性，为DD中的分布相似性考虑提供了原则性的补充。
cs.CV / 104 / 2606.21734
HPP: Hierarchical Programmatic Probing for Long Video Understanding by Decoupling Perception and Reasoning
HPP：通过解耦感知与推理实现长视频理解的层次化程序探测
Abstract
Understanding long videos requires fine-grained perception and multi-step, higher-order reasoning over complex, long-range spatio-temporal dynamics. Vision-language models (VLMs) encode video frames into visual tokens and attempt to perform both perception and multi-step planning latently, within a single forward pass. This coupled formulation, however, is bottlenecked by the LLM's limited capacity to discover and execute multi-step strategies in its latent representations. To address this bottleneck, we propose Hierarchical Programmatic Probing (HPP), a framework that decouples semantic perception from higher-order temporal reasoning by reformulating long video understanding as iterative, programmatic exploration of a hierarchically segmented video. Specifically, a coding-capable LLM plans and executes a multi-step strategy in an interactive coding environment, probing the video for information and invoking a VLM for localized perception on demand. To make probing tractable over long videos, we introduce three components: information-density-aware hierarchical segmentation, late-interaction semantic retrieval, and structured probing functions for coarse-to-fine temporal localization. We validate HPP on LongVideoBench, which requires both fine-grained perception and long-range relational reasoning, and show that decoupling the two via iterative programmatic probing yields substantial gains. Further results on EgoSchema, VideoMME, and MLVU demonstrate the effectiveness of our approach across diverse long-video benchmarks.
Chinese Translation
理解长视频需要对复杂的长程时空动态进行细粒度感知和多步骤的高阶推理。视觉语言模型（VLMs）将视频帧编码为视觉标记，并试图在单次前向传递中同时进行感知和多步骤规划。然而，这种耦合的形式受到大型语言模型（LLM）在其潜在表示中发现和执行多步骤策略能力有限的瓶颈。为了解决这一瓶颈，我们提出了层次化程序探测（HPP），该框架通过将长视频理解重新构建为对分层分段视频的迭代程序探索，从而将语义感知与高阶时间推理解耦。具体而言，一个具备编码能力的LLM在交互编码环境中规划并执行多步骤策略，探测视频以获取信息，并根据需要调用VLM进行局部感知。为了使长视频的探测变得可行，我们引入了三个组件：信息密度感知的层次分段、后期交互语义检索和用于粗到细时间定位的结构化探测函数。我们在LongVideoBench上验证了HPP，该基准需要细粒度感知和长程关系推理，并显示通过迭代程序探测解耦这两者可以带来显著的提升。进一步在EgoSchema、VideoMME和MLVU上的结果证明了我们的方法在多样化的长视频基准测试中的有效性。
cs.CV / 105 / 2606.21736
Adversarial Domain Prompt Tuning and Generation for Single Domain Generalization
针对单域泛化的对抗性领域提示调优与生成
Abstract
Single domain generalization (SDG) aims to learn a robust model, which could perform well on many unseen domains while there is only one single domain available for training. One of the promising directions for achieving single-domain generalization is to generate out-of-domain (OOD) training data through data augmentation or image generation. Given the rapid advancements in AI-generated content (AIGC), this paper is the first to propose leveraging powerful pre-trained text-to-image (T2I) foundation models to create the training data. However, manually designing textual prompts to generate images for all possible domains is often impractical, and some domain characteristics may be too abstract to describe with words. To address these challenges, we propose a novel Progressive Adversarial Prompt Tuning (PAPT) framework for pre-trained diffusion models. Instead of relying on static textual domains, our approach learns two sets of abstract prompts as conditions for the diffusion model: one that captures domain-invariant category information and another that models domain-specific styles. This adversarial learning mechanism enables the T2I model to generate images in various domain styles while preserving key categorical features. Extensive experiments demonstrate the effectiveness of the proposed method, achieving superior performances to state-of-the-art single-domain generalization approaches.
Chinese Translation
单域泛化（SDG）旨在学习一个鲁棒模型，使其能够在许多未见过的领域中表现良好，而训练时仅有一个单一领域可用。实现单域泛化的一个有前景的方向是通过数据增强或图像生成生成域外（OOD）训练数据。鉴于人工智能生成内容（AIGC）的快速发展，本文首次提出利用强大的预训练文本到图像（T2I）基础模型来创建训练数据。然而，手动设计文本提示以生成所有可能领域的图像往往不切实际，并且某些领域特征可能过于抽象，难以用语言描述。为了解决这些挑战，我们提出了一种新颖的渐进式对抗提示调优（PAPT）框架，适用于预训练的扩散模型。我们的方案不依赖于静态文本领域，而是学习两组抽象提示作为扩散模型的条件：一组捕捉领域不变的类别信息，另一组建模领域特定的风格。这种对抗学习机制使得T2I模型能够在各种领域风格中生成图像，同时保留关键的类别特征。大量实验表明，所提方法的有效性，达到了优于现有最先进单域泛化方法的性能。
cs.CV / 106 / 2606.21749
Quantile Adaptive Temperature Scaling for Confidence Calibration
量化自适应温度缩放用于置信度校准
Abstract
Deep neural networks often produce poorly calibrated confidence estimates, overstating their certainty even when predictions are incorrect. Temperature Scaling remains the most widely used posthoc calibration method due to its simplicity and effectiveness, yet its global, uniform rescaling of logits fails to correct the highly heterogeneous structure of miscalibration observed across the confidence spectrum. In particular, the largest correctness confidence discrepancies arise in different quantile regions depending on the setting, low confidence predictions, where uncertainty matters most, tend to exhibit the largest correctness confidence discrepancies, which standard TS leaves largely unaddressed. We introduce Quantile Adaptive Temperature Scaling (QaTS), a simple and efficient post hoc calibration method that adapts the temperature as a function of a predictions empirical confidence quantile. By mapping confidences into the quantile space, QaTS normalizes the calibration problem, makes the structure of miscalibration explicit and enables a monotone temperature function that adapts across quantiles while leaving well calibrated high confidence predictions largely unchanged. preserving high confidence behavior. This quantile aware formulation aligns naturally with a reparameterized Expected Calibration Error (ECE) objective and yields a sample wise temperature that is robust across a variety of challenging scenarios, such as class imbalance and distributional shifts. Across a broad range of datasets, architectures, evaluation scenarios and diverse tasks, QaTS consistently, and substantially, outperforms state of the art post hoc calibration methods, delivering more reliable and trustworthy confidence estimates without modifying model predictions.
Chinese Translation
深度神经网络通常产生校准不良的置信度估计，即使在预测错误时也会夸大其确定性。温度缩放（Temperature Scaling）因其简单性和有效性而成为最广泛使用的后处理校准方法，然而其对 logits 的全局统一重缩放未能纠正在置信度范围内观察到的高度异质的错误校准结构。特别是，在不同的分位数区域中，最大的正确性置信度差异会根据设置而有所不同，低置信度预测（low confidence predictions）在不确定性最为重要的情况下，往往表现出最大的正确性置信度差异，而标准的温度缩放对此几乎没有解决。我们提出了量化自适应温度缩放（Quantile Adaptive Temperature Scaling，QaTS），这是一种简单而高效的后处理校准方法，通过将温度作为预测经验置信度分位数的函数进行调整。通过将置信度映射到分位数空间，QaTS 规范化了校准问题，使错误校准的结构显而易见，并实现了一个单调的温度函数，该函数在分位数之间自适应，同时对良好校准的高置信度预测几乎没有影响，从而保持高置信度行为。这种关注分位数的表述自然与重新参数化的期望校准误差（Expected Calibration Error，ECE）目标对齐，并产生在各种具有挑战性的场景（如类别不平衡和分布变化）中都具有鲁棒性的样本级温度。在广泛的数据集、架构、评估场景和多样化任务中，QaTS 一贯且显著地超越了最先进的后处理校准方法，提供了更可靠和可信的置信度估计，而无需修改模型预测。
cs.CV / 107 / 2606.21763
From Gradient Clipping to Structural Refinement: Improving DPSGD for Medical Image Segmentation
从梯度裁剪到结构细化：改进DPSGD以用于医学图像分割
Abstract
Medical image segmentation is widely used for disease detection but relies on sensitive data, raising privacy concerns as trained models can leak information. Differential privacy, typically implemented via Differential Private Stochastic Gradient Descent (DPSGD), provides a solution, though at the cost of reduced utility. Recent DPSGD variants, including Automatic clipping (Auto-S), Normalised SGD with perturbation (NSGD), and Per-sample adaptive clipping (PSAC), have shown promise in image classification, but their behavior in medical segmentation remains underexplored. We evaluate these methods across binary and multi-class tasks and analyze gradient alignment, showing that prior assumptions, particularly for PSAC, do not consistently hold. We further demonstrate that combining clipping strategies with morphological refinement improves segmentation quality under privacy constraints. Finally, we propose an adaptive DP-Morph variant that captures class-specific structures and enhances performance in multi-class settings.
Chinese Translation
医学图像分割广泛应用于疾病检测，但依赖于敏感数据，导致隐私问题，因为训练后的模型可能泄露信息。差分隐私通常通过差分私有随机梯度下降（DPSGD）实现，虽然提供了解决方案，但代价是实用性降低。最近的DPSGD变体，包括自动裁剪（Auto-S）、带扰动的归一化随机梯度下降（NSGD）和每样本自适应裁剪（PSAC），在图像分类中显示出良好前景，但它们在医学分割中的表现仍未得到充分探索。我们在二分类和多分类任务中评估这些方法，并分析梯度对齐，显示出先前的假设，特别是对于PSAC，并不总是成立。我们进一步证明，将裁剪策略与形态学细化相结合可以在隐私约束下提高分割质量。最后，我们提出了一种自适应DP-Morph变体，能够捕捉类别特定的结构，并在多分类环境中提升性能。
cs.CV / 108 / 2606.21764
Motion-Aware Reinforcement Learning For Object Localization
运动感知强化学习用于物体定位
Abstract
We present MARLNet (Motion-Aware Reinforcement Learning Network), a PPO-based bounding-box refinement agent that incorporates a constant-velocity motion prior into the observation state and an action smoothness penalty into the reward function. The agent operates on 268-dimensional observations encoding the current proposal, a kinematic prediction, the previous action, and a 256-dimensional EfficientNet-B0 crop feature, and learns a five-dimensional policy controlling coordinate adjustments and a binary termination trigger. Evaluated on Pascal VOC 2012 and VisDrone 2019, MARLNet trains stably across all regularization strengths tested and achieves consistent gains in detection success rate at $\text{IoU} \geq 0.5$: up to $+0.011$ on VOC ($\lambda_\text{phys}{=}0.10$), where the motion prior prevents the overshooting that causes plain PPO to regress on this metric, and $+0.007$ on VisDrone ($\lambda_\text{phys}{=}0.70$), where unconstrained PPO achieves a larger gain ($+0.025$) owing to the weaker base detector. Through reward design ablations and training dynamics analysis, we identify a reward interference in which combining a constant-velocity deviation penalty with an absolute IoU term causes trigger collapse, and show that replacing it with the action smoothness penalty resolves this failure. We further characterize a representational ceiling facing crop-feature refinement agents that share a backbone with their base detector, confirmed through a global-plus-local observation ablation. Project page: https://prithviraj97.github.io/marl-net
Chinese Translation
我们提出了MARLNet（运动感知强化学习网络），这是一种基于PPO的边界框精细化代理，结合了恒速运动先验到观察状态中，并在奖励函数中加入了动作平滑性惩罚。该代理在268维的观察数据上操作，这些数据编码了当前提议、运动学预测、前一个动作以及256维的EfficientNet-B0裁剪特征，并学习一个五维策略来控制坐标调整和二元终止触发器。在Pascal VOC 2012和VisDrone 2019上进行评估，MARLNet在所有测试的正则化强度下都能稳定训练，并在$ ext{IoU}
geq 0.5$的检测成功率上取得了一致的提升：在VOC上达到$+0.011$（$ ext{λ}_ ext{phys}{=}0.10$），其中运动先验防止了导致普通PPO在该指标上退化的超调；在VisDrone上达到$+0.007$（$ ext{λ}_ ext{phys}{=}0.70$），其中无约束PPO由于基础检测器较弱而获得了更大的增益（$+0.025$）。通过奖励设计消融实验和训练动态分析，我们识别出一种奖励干扰现象，即将恒速偏差惩罚与绝对IoU项结合会导致触发器崩溃，并表明用动作平滑性惩罚替代它可以解决这一失败。我们进一步表征了面临的表示上限，即裁剪特征精细化代理与其基础检测器共享骨干网络的情况，通过全局加局部观察消融实验得到了验证。项目页面：https://prithviraj97.github.io/marl-net
cs.CV / 109 / 2606.21819
RAPID: A Reproducible Multi-Agent Pipeline for Interpretable Disaster Damage Assessment from Satellite and Street-View Imagery
RAPID：一种可重复的多智能体管道，用于从卫星和街景图像中进行可解释的灾害损失评估
Abstract
Due to the increasing frequency and intensity of extreme climate events, there is a clear demand for intelligent, scalable, and autonomous approaches to disaster damage assessment. Existing methods, largely based on supervised learning and task-specific fine-tuning, struggle to generalize under domain shifts, long-tailed data distributions, and heterogeneous geospatial data sources, especially in disaster scenarios. They also often lack the ability to integrate and reason across multimodal geospatial information, such as satellite images and street-view images. In this paper, we introduce RAPID, a reproducible multi-agent pipeline for interpretable disaster damage assessment, including damage-level assessment, damage-type interpretation, and actionable suggestions for response, remediation, and recovery. RAPID coordinates specialized agents to perform cross-view understanding, image restoration, structured damage recognition, and geographical reasoning across heterogeneous data modalities. Without task-specific fine-tuning, RAPID supports zero-shot damage assessment by jointly using complementary information from remote sensing and ground-level perspectives. The system produces fine-grained, interpretable assessments and automatically generates location-specific, decision-relevant disaster reports to support early-stage emergency response. We evaluate RAPID across hurricanes, floods, wildfires, and earthquakes using multiple cross-view imagery inputs, including pre- and post-disaster street-view images, post-disaster remote sensing imagery, and street-view image pairs. Experiments show that RAPID achieves 0.92 overall accuracy for multi-disaster type classification and up to 0.627 for cross-view damage severity prediction, highlighting its potential as a foundational framework for autonomous disaster intelligence.
Chinese Translation
由于极端气候事件的频率和强度不断增加，对智能、可扩展和自主的灾害损失评估方法的需求日益明显。现有方法主要基于监督学习和特定任务的微调，难以在领域转移、长尾数据分布和异构地理空间数据源下进行泛化，尤其是在灾害场景中。它们通常缺乏整合和推理多模态地理空间信息的能力，例如卫星图像和街景图像。在本文中，我们介绍了RAPID，一种可重复的多智能体管道，用于可解释的灾害损失评估，包括损失级别评估、损失类型解释以及针对响应、修复和恢复的可行建议。RAPID协调专门的智能体执行跨视角理解、图像恢复、结构化损伤识别和异构数据模态的地理推理。在没有特定任务微调的情况下，RAPID通过联合使用遥感和地面视角的互补信息支持零样本损失评估。该系统生成细粒度、可解释的评估，并自动生成特定位置、与决策相关的灾害报告，以支持早期紧急响应。我们在飓风、洪水、野火和地震等多种灾害中评估了RAPID，使用多种跨视角图像输入，包括灾前和灾后街景图像、灾后遥感图像和街景图像对。实验表明，RAPID在多灾害类型分类中实现了0.92的整体准确率，在跨视角损害严重性预测中达到0.627，突显了其作为自主灾害智能基础框架的潜力。
cs.CV / 110 / 2606.21838
Beyond Flat Labels: Level-Restricted Contrastive Learning for Hierarchical Fine-Grained Vision Classification
超越平面标签：用于层级细粒度视觉分类的层级限制对比学习
Abstract
Multimodal contrastive learning has enabled zero-shot visual classification by aligning images with textual categories. However, in hierarchically structured label spaces, existing methods often produce predictions that are inconsistent across taxonomic levels. For example, a model may predict a fine-grained category whose parent category contradicts its simultaneously predicted higher-level label. By analysis, the issue originates from false negative labels when contrastive comparison involves multiple taxonomic levels. To this end, we propose to restrict contrastive comparisons to categories within the same taxonomic level. In addition, we adopt a group-balanced design, ensuring each taxonomic level receives adequate optimization. As a result, the proposed framework improves both hierarchical consistency and classification accuracy from coarse to fine granularity. We train our model with TreeOfLife-10M based on BioCLIP and evaluate it across multiple hierarchical classification benchmarks, where the model demonstrates significantly improved hierarchical consistency in both Euclidean and hyperbolic spaces. Notably, on iNaturalist 2021 (iNat21), our method improves average accuracy across levels by 30.47% over the baseline, highlighting its effectiveness for hierarchical zero-shot classification.
Chinese Translation
多模态对比学习通过将图像与文本类别对齐，实现了零样本视觉分类。然而，在层级结构的标签空间中，现有方法往往在分类学层次之间产生不一致的预测。例如，一个模型可能预测一个细粒度类别，而其父类别却与其同时预测的更高层级标签相矛盾。通过分析，这一问题源于在对比比较涉及多个分类学层次时产生的假阴性标签。为此，我们提出将对比比较限制在同一分类学层次内。此外，我们采用了组平衡设计，确保每个分类学层次都得到充分优化。因此，所提出的框架在粗到细的粒度上改善了层级一致性和分类准确性。我们基于 BioCLIP 使用 TreeOfLife-10M 训练模型，并在多个层级分类基准上进行评估，模型在欧几里得空间和双曲空间中均表现出显著改善的层级一致性。值得注意的是，在 iNaturalist 2021 (iNat21) 上，我们的方法在各层级的平均准确率比基线提高了 30.47%，突显了其在层级零样本分类中的有效性。
cs.CV / 111 / 2606.21861
Zero-Shot Vision-Language Models for Classroom Engagement Recognition: A Benchmark Study of Prompt Sensitivity and Cross-Dataset Generalization
零样本视觉-语言模型在课堂参与度识别中的应用：提示敏感性与跨数据集泛化的基准研究
Abstract
Automated classroom engagement recognition holds substantial promise for scalable learning analytics, yet the suitability of modern Vision-Language Models (VLMs) for this task under zero-shot conditions remains largely unexplored. We present a systematic benchmark that evaluates five widely-used VLMs: CLIP, BLIP-VQA, GPT-4o, LLaVA-1.5-7B, and Qwen2.5VL-7B-Instruct across two complementary educational datasets: DAiSEE, an individual-student video dataset (300 sampled test clips), and the Student Classroom Behaviour dataset (SCB, 1,168 scene-level images). Each model is probed with three prompt variants spanning minimal, rubric-anchored, and chain-of-thought designs. Our experiments reveal three primary failure modes of zero-shot VLMs for engagement recognition: (1) near-random performance on individual students, with Cohen's kappa never exceeding 0.10 on DAiSEE; (2) severe class collapse, where models assign 85-100% of predictions to a single engagement level regardless of visual content; and (3) extreme prompt sensitivity, with accuracy swings of up to 32 percentage points on identical images depending solely on prompt phrasing. Remarkably, scene-level classification on SCB is substantially more tractable: CLIP and GPT-4o achieve kappa approximately 0.60 when prompted with behaviorally-grounded rubrics. We also document a practical barrier for deployment: GPT-4o's safety filters reject 98% of chain-of-thought requests involving individual student faces. Our findings provide a calibrated baseline and surface critical design considerations for the use of VLMs in educational observation systems.
Chinese Translation
自动化课堂参与度识别在可扩展学习分析中具有重要前景，但现代视觉-语言模型（VLMs）在零样本条件下适用于此任务的适宜性仍然未被充分探索。我们提出了一个系统的基准，评估五种广泛使用的VLMs：CLIP、BLIP-VQA、GPT-4o、LLaVA-1.5-7B和Qwen2.5VL-7B-Instruct，基于两个互补的教育数据集：DAiSEE，一个个体学生视频数据集（300个抽样测试片段），以及学生课堂行为数据集（SCB，1,168个场景级图像）。每个模型都通过三种提示变体进行探测，涵盖最小提示、基于评分标准的提示和思维链设计。我们的实验揭示了零样本VLM在参与度识别中的三种主要失效模式：（1）对个体学生的近乎随机表现，在DAiSEE上的Cohen's kappa从未超过0.10；（2）严重的类别崩溃，模型将85-100%的预测分配给单一的参与度水平，无论视觉内容如何；（3）极端的提示敏感性，相同图像的准确率根据提示措辞的不同波动高达32个百分点。值得注意的是，在SCB上的场景级分类显著更易处理：当使用基于行为的评分标准进行提示时，CLIP和GPT-4o的kappa约为0.60。我们还记录了一个实际的部署障碍：GPT-4o的安全过滤器拒绝98%的涉及个体学生面孔的思维链请求。我们的研究提供了一个经过校准的基线，并提出了在教育观察系统中使用VLM的关键设计考虑因素。
cs.CV / 112 / 2606.21863
Prompt-Calibrated SAM 3 for Open-Vocabulary Remote Sensing Semantic Segmentation
用于开放词汇遥感语义分割的提示校准SAM 3
Abstract
Open-vocabulary semantic segmentation (OVSS) in remote sensing images aims to segment categories beyond a fixed label space. Recent SAM 3-based methods provide a promising training-free foundation, yet three key issues remain: (1) a single class-name prompt lacks sufficient semantic coverage for complex remote sensing categories; (2) expanding each category into multiple prompts introduces redundant online text encoding; and (3) directly aggregating multiple prompt responses propagates noisy activations into the final prediction. To address these issues, we propose ProC-SAM3, which calibrates SAM 3's prompt interface for remote sensing OVSS from three complementary aspects. First, we construct an offline prompt pool where a Category Matcher groups MLLM-generated candidates into per-category sets, and Expansion Constraints further refine each set using category-specific prior knowledge. Second, the resulting text embeddings are cached and reused across all test images, eliminating repeated text encoding. Third, we introduce Presence-Guided Residual Fusion to gate unreliable decoder outputs by prompt presence and confidence, followed by peak-preserving class aggregation that retains fine-grained activations for small and sparse objects. Experiments on eight benchmarks show that ProC-SAM3 achieves an average mIoU of 56.1%, outperforming the previous best training-free method by 3.9 percentage points. Code will be available at https://github.com/YanghuiSong/ProC-SAM3.
Chinese Translation
开放词汇语义分割（OVSS）在遥感图像中的目标是对超出固定标签空间的类别进行分割。基于最近的SAM 3方法提供了一个有前景的无训练基础，但仍然存在三个关键问题：（1）单一类别名称提示缺乏对复杂遥感类别的充分语义覆盖；（2）将每个类别扩展为多个提示会引入冗余的在线文本编码；（3）直接聚合多个提示响应会将噪声激活传播到最终预测中。为了解决这些问题，我们提出了ProC-SAM3，它从三个互补的方面校准SAM 3的提示接口以适应遥感OVSS。首先，我们构建了一个离线提示池，其中类别匹配器将MLLM生成的候选项分组为每个类别的集合，扩展约束进一步利用类别特定的先验知识精炼每个集合。其次，生成的文本嵌入被缓存并在所有测试图像中重复使用，从而消除重复的文本编码。第三，我们引入了基于存在的残差融合，通过提示的存在和置信度来控制不可靠的解码器输出，随后进行保峰的类别聚合，以保留小型和稀疏对象的细粒度激活。在八个基准测试上的实验表明，ProC-SAM3的平均mIoU达到56.1%，比之前最佳的无训练方法提高了3.9个百分点。代码将发布在https://github.com/YanghuiSong/ProC-SAM3。
cs.CV / 113 / 2606.21910
Fidelity- and Perception-Aware Local Implicit Attention for Arbitrary-Scale Image Super-Resolution
基于保真度和感知的局部隐式注意力用于任意尺度图像超分辨率
Abstract
Arbitrary-scale image super-resolution (ASISR) aims to reconstruct high-resolution images from low-resolution inputs over a continuous range of upscaling factors. While traditional pixel-regression approaches often produce overly smooth results that lack realistic details, recent diffusion methods can produce sharper and more realistic textures. However, these diffusion techniques frequently introduce the risk of structural hallucinations. To address these issues, we propose Fidelity- and Perception-Aware Local Implicit Attention (FPLIA), a framework that effectively integrates fidelity-oriented features into a diffusion pipeline to produce realistic and faithful reconstructions for ASISR. We introduce a Fidelity and Perception Attention Module (FPAM), which applies both self-attention and cross-attention to fidelity-oriented and perceptual features to enhance representational capacity. To further exploit their complements, we design a Fidelity and Perception Select Module (FPSM) that adaptively selects the most representative features for RGB values prediction. We conduct extensive experiments to validate the effectiveness of these components. Both qualitative and quantitative results show that FPLIA delivers superior perceptual realism while maintaining reconstruction accuracy on standard ASISR benchmarks. The source code is accessible at the following repository: https://github.com/XUSean0118/FPLIA.
Chinese Translation
任意尺度图像超分辨率（ASISR）旨在从低分辨率输入重建高分辨率图像，覆盖一系列连续的放大因子。传统的像素回归方法往往产生过于平滑的结果，缺乏真实的细节，而最近的扩散方法则能够生成更清晰和更真实的纹理。然而，这些扩散技术经常引入结构幻觉的风险。为了解决这些问题，我们提出了基于保真度和感知的局部隐式注意力（FPLIA），这是一个有效将以保真度为导向的特征整合到扩散管道中的框架，以产生真实且忠实的ASISR重建。我们引入了一个保真度和感知注意力模块（FPAM），该模块对以保真度为导向的特征和感知特征应用自注意力和交叉注意力，以增强表示能力。为了进一步利用它们的互补性，我们设计了一个保真度和感知选择模块（FPSM），该模块自适应选择最具代表性的特征用于RGB值预测。我们进行了广泛的实验以验证这些组件的有效性。定性和定量结果均表明，FPLIA在保持标准ASISR基准上的重建准确性的同时，提供了卓越的感知真实感。源代码可在以下仓库访问：https://github.com/XUSean0118/FPLIA。
cs.CV / 114 / 2606.21913
Rethinking the Adaptation of Vision Foundation Models for Efficient Cell Segmentation
重新思考视觉基础模型在高效细胞分割中的适应性
Abstract
Cell segmentation is critical for computational pathology and biomedical discovery. While recent Vision Foundation Models (VFMs) have demonstrated remarkable universal feature representations, unlocking their full potential for cellular imaging is currently bottlenecked by resource-intensive adaptation paradigms. Existing methods typically rely on fine-tuning heavy visual encoders, leading to extensive computational overhead and a dependency on large-scale annotations. To address this, we propose the EffiCell-Seg framework for highly efficient cell segmentation without re-training the visual encoder. Our core insight is that pretrained VFMs intrinsically encode complementary structural priors: global saliency for localizing potential cells, and local morphological patterns for delineating cellular structures. To harness these priors, we devise a Cell Structure Prompt Encoder (CSP-Encoder) that synthesizes semantic-aware saliency and principal morphological features from frozen VFM representations into explicit structural prior maps. Moreover, we propose a Synergistic Mask Decoder (SM-Decoder) that enforces contextual consistency by jointly predicting geometric distance fields and semantic maps via mutual cross-guidance. Extensive experiments demonstrate that EffiCell-Seg outperforms state-of-the-art methods across diverse cell imaging modalities while requiring only ~5M trainable parameters, over 130x fewer than fully fine-tuned VFM counterparts. The code is available at https://github.com/xq141839/EffiCell-Seg.
Chinese Translation
细胞分割对于计算病理学和生物医学发现至关重要。尽管最近的视觉基础模型（VFMs）展示了显著的通用特征表示，但要充分发挥其在细胞成像中的潜力，目前仍受到资源密集型适应范式的瓶颈。现有方法通常依赖于对重型视觉编码器的微调，导致大量计算开销并依赖于大规模标注。为了解决这个问题，我们提出了EffiCell-Seg框架，以实现高效的细胞分割，而无需重新训练视觉编码器。我们的核心见解是，预训练的VFMs本质上编码了互补的结构先验：用于定位潜在细胞的全局显著性和用于描绘细胞结构的局部形态模式。为了利用这些先验，我们设计了一个细胞结构提示编码器（CSP-Encoder），该编码器将来自冻结的VFM表示的语义感知显著性和主要形态特征合成到明确的结构先验图中。此外，我们提出了一个协同掩膜解码器（SM-Decoder），通过共同预测几何距离场和语义图来强制执行上下文一致性，利用互相交叉引导。大量实验表明，EffiCell-Seg在各种细胞成像模式下优于最先进的方法，同时仅需约5M的可训练参数，比完全微调的VFM对应物少了超过130倍。代码可在https://github.com/xq141839/EffiCell-Seg获取。
cs.CV / 115 / 2606.21915
GTA-Net: Cooperative Game Theory for Vision-Language Alignment in Chest X-Ray Report Generation
GTA-Net：用于胸部X光报告生成的视觉-语言对齐的合作博弈理论
Abstract
Automated chest X-ray report generation requires precise cross-modal grounding to ensure clinically reliable descriptions. However, existing vision-language models rely on implicit attention mechanisms that fail to enforce explicit region-word correspondence and disease-level consistency. We propose Game-Theoretic Alignment Network (GTA-Net), a vision-language framework that formulates report generation as a cooperative game-theoretic alignment problem. The model introduces a BinaryGameAligner that models interactions between image regions and text tokens using similarity-based payoff matrices with Shapley-inspired importance weighting. To enforce clinical semantics, we further develop a Disease-Aware Ternary Aligner, which captures joint interactions among images, reports, and structured disease concepts. GTA-Net combines a Swin-based visual encoder with a LoRA-adapted large language model and is trained with a unified objective for generation and alignment. Experiments on CheXpertPlus and IU-XRay demonstrate state-of-the-art performance across standard generation metrics and improved clinical consistency, highlighting the effectiveness of explicit game-theoretic alignment for medical vision-language generation.
Chinese Translation
自动化的胸部X光报告生成需要精确的跨模态基础，以确保临床可靠的描述。然而，现有的视觉-语言模型依赖于隐式注意机制，未能强制执行明确的区域-词语对应关系和疾病级别的一致性。我们提出了博弈理论对齐网络（Game-Theoretic Alignment Network，GTA-Net），这是一个将报告生成形式化为合作博弈理论对齐问题的视觉-语言框架。该模型引入了二元博弈对齐器（BinaryGameAligner），使用基于相似度的收益矩阵和受Shapley启发的重要性加权来建模图像区域与文本标记之间的交互。为了强化临床语义，我们进一步开发了一种疾病感知三元对齐器（Disease-Aware Ternary Aligner），该对齐器捕捉图像、报告和结构化疾病概念之间的联合交互。GTA-Net结合了基于Swin的视觉编码器和经过LoRA适配的大型语言模型，并以统一的目标进行生成和对齐的训练。在CheXpertPlus和IU-XRay上的实验表明，该方法在标准生成指标上表现出最先进的性能，并提高了临床一致性，突显了显式博弈理论对齐在医学视觉-语言生成中的有效性。
cs.CV / 116 / 2606.21932
CoSA: Correlation-Guided Change Attention with Learnable Residual Gating for Remote Sensing Change Detection
CoSA：基于相关性的变化注意力与可学习残差门控用于遥感变化检测
Abstract
Remote sensing change detection (CD) from bi-temporal imagery is critical for applications such as urban monitoring, disaster assessment, and environmental management, yet robust localization remains challenging under sparse changes, noisy labels, and appearance variations. In this paper, we propose Context Sampling Attention (CoSA), a lightweight decoder-side refinement module that explicitly leverages bi-temporal feature correlation as a control signal for adaptive change-aware feature enhancement. This differs from conventional attention mechanisms that rely on implicit feature weighting without explicit temporal control. In the implemented FC-Siam setting, CoSA computes normalized same-location cross-correlation between paired decoder features, converts low correlation into a change gate, and injects the resulting gated residual at native 1/8 and 1/16 feature scales through learnable residual scaling. This design enables effective discrimination between stable and ambiguous regions without relying on computationally expensive global attention. Extensive experiments on four benchmark datasets (LEVIR-CD, S2Looking, DSIFN, and CLCD) demonstrate consistent improvements over strong baselines, achieving 1.5-2.6% gains in changed-class F1 while introducing negligible parameter overhead. Ablation studies confirm that multiscale placement and learnable residual gating are both important for peak performance. These results indicate that CoSA establishes a practical and effective refinement paradigm for enhancing temporal discriminability in Siamese change detection frameworks.
Chinese Translation
基于双时相影像的遥感变化检测（CD）对于城市监测、灾害评估和环境管理等应用至关重要，但在稀疏变化、噪声标签和外观变化的情况下，稳健的定位仍然具有挑战性。本文提出了一种上下文采样注意力（CoSA），这是一种轻量级的解码器侧精细化模块，明确利用双时相特征相关性作为自适应变化感知特征增强的控制信号。这与传统的注意力机制不同，后者依赖于隐式特征加权而没有明确的时间控制。在实现的FC-Siam设置中，CoSA计算配对解码器特征之间的归一化同位置交叉相关性，将低相关性转换为变化门，并通过可学习的残差缩放在原生的1/8和1/16特征尺度上注入结果的门控残差。该设计能够有效区分稳定和模糊区域，而无需依赖计算开销大的全局注意力。在四个基准数据集（LEVIR-CD、S2Looking、DSIFN和CLCD）上的大量实验表明，相较于强基线，CoSA在变化类别F1上实现了1.5-2.6%的增益，同时引入的参数开销微乎其微。消融研究确认多尺度放置和可学习残差门控对于达到最佳性能都至关重要。这些结果表明，CoSA为增强Siamese变化检测框架中的时间可区分性建立了一种实用有效的精细化范式。
cs.CV / 117 / 2606.21938
Artic-O: End-to-End Articulated Object Reconstruction via Latent Geometry Learning
Artic-O：通过潜在几何学习进行端到端的关节物体重建
Abstract
Reconstructing articulated objects from sparse images requires recovering complete geometry, movable parts, and motion parameters. Recent methods typically separate geometry reconstruction, part reasoning, and articulation estimation into different stages. This separation can weaken consistency between shape, active parts, and motion, while also incurring substantial inference cost. We introduce Artic-O, an end-to-end, feed-forward framework for articulated object reconstruction via latent geometry learning. Instead of fitting geometry in image or view space, Artic-O maps sparse multi-state observations into a pretrained latent geometry space, where a frozen flow-matching decoder provides a complete-shape prior for recovering visible and occluded structures. To connect geometry with articulation, Artic-O fuses visual tokens, geometry latents, and point-wise decoder features in an image-grounded part-reasoning module for active-part segmentation and articulation prediction. We further train the model with a geometry-to-articulation curriculum and a decoupled two-pass strategy to balance reconstruction and part-level supervision. On PartNet-Mobility, Artic-O achieves strong reconstruction quality while being substantially more efficient than LARM, a strong prior method. It reduces Chamfer Distance, improves F-score, and achieves comparable or better articulation accuracy across most joint metrics, while reducing inference time from 9 minutes to about 0.3 seconds per object.
Chinese Translation
从稀疏图像重建关节物体需要恢复完整的几何形状、可移动部件和运动参数。最近的方法通常将几何重建、部件推理和关节估计分为不同的阶段。这种分离可能削弱形状、活动部件和运动之间的一致性，同时也会导致显著的推理成本。我们提出了Artic-O，一个通过潜在几何学习进行关节物体重建的端到端前馈框架。Artic-O不是在图像或视图空间中拟合几何形状，而是将稀疏的多状态观测映射到一个预训练的潜在几何空间，在该空间中，一个冻结的流匹配解码器为恢复可见和遮挡结构提供了完整形状的先验。为了将几何形状与关节连接起来，Artic-O在一个基于图像的部件推理模块中融合了视觉标记、几何潜变量和逐点解码器特征，以进行活动部件分割和关节预测。我们进一步通过几何到关节的课程学习和解耦的双通道策略训练模型，以平衡重建与部件级监督。在PartNet-Mobility上，Artic-O实现了强大的重建质量，同时在效率上显著优于LARM这一强有力的先验方法。它减少了Chamfer距离，提高了F-score，并在大多数关节指标上实现了可比或更好的关节精度，同时将每个对象的推理时间从9分钟减少到约0.3秒。
cs.CV / 118 / 2606.21947
ScalePredictor: Instance-aware Scale Learning for Accurate Quantization of Vision Transformers
ScalePredictor：面向实例的尺度学习以实现视觉变换器的精确量化
Abstract
Vision Transformers have achieved remarkable success in many fields, yet their deployment on edge devices remains challenging due to their substantial computational demands. Post-Training Quantization (PTQ) offers an attractive solution by compressing models using a small calibration set with minimal training overhead. However, most existing PTQ works adopt a static quantization paradigm that is uniformly applied to all instances. Given the substantial diversity of natural images, the activation distributions vary significantly across samples, making these methods inherently suboptimal. In this paper, we propose ScalePredictor, a dynamic quantization framework for accurate and efficient quantization scale learning of ViTs. We first reveal a hidden correlation between the distribution range of shallow-layer activations and the optimal scales of deeper layers. Based on this, we develop a scale learning mechanism that integrates an efficient range extraction approach to capture robust range statistics at the shallow stage, which are then fed into a Taylor-motivated polynomial scale projection module to generate all quantization scales simultaneously. With the efficiency of polynomial approximation, ScalePredictor introduces insignificant computational overhead while avoiding costly just-in-time calibration. Extensive experiments on ImageNet demonstrate that ScalePredictor consistently outperforms prior PTQ methods, achieving a more favorable accuracy-efficiency trade-off. Code and additional results are shown in the supplementary materials.
Chinese Translation
视觉变换器在许多领域取得了显著成功，但由于其巨大的计算需求，在边缘设备上的部署仍然面临挑战。后训练量化（Post-Training Quantization, PTQ）通过使用小型校准集进行模型压缩，提供了一种具有吸引力的解决方案，且训练开销最小。然而，大多数现有的PTQ工作采用静态量化范式，均匀地应用于所有实例。考虑到自然图像的多样性，激活分布在样本之间显著变化，使得这些方法在本质上并不理想。在本文中，我们提出了ScalePredictor，一种动态量化框架，用于视觉变换器的精确和高效的量化尺度学习。我们首先揭示了浅层激活的分布范围与深层的最佳尺度之间的隐含关联。基于此，我们开发了一种尺度学习机制，结合高效的范围提取方法，以在浅层阶段捕获稳健的范围统计数据，然后将其输入到一个基于泰勒展开的多项式尺度投影模块中，以同时生成所有量化尺度。借助多项式近似的高效性，ScalePredictor引入了微不足道的计算开销，同时避免了昂贵的即时校准。在ImageNet上的大量实验表明，ScalePredictor始终优于之前的PTQ方法，实现了更有利的准确性与效率的权衡。代码和额外结果见补充材料。
cs.CV / 119 / 2606.21949
CapRiCorn-1K: A Comprehensive Benchmark for Video Captioning and Subject Referential Consistency Across Temporal Scales
CapRiCorn-1K：一个综合的视频字幕和主题指称一致性基准，涵盖不同时间尺度
Abstract
Accurate and comprehensive video captions with consistent subject references are critical for downstream understanding and generation tasks. However, few existing benchmarks can objectively and comprehensively evaluate these properties across diverse durations and scenarios, thereby hindering the advancement of video captioning models. To bridge this gap, we propose CapRiCorn-1K, a comprehensive benchmark designed to evaluate both video captioning quality and subject referential consistency across long temporal horizons and diverse video domains. To accommodate varied evaluation needs, our benchmark supports both audiovisual and visual-only settings. Extensive experiments on CapRiCorn-1K reveal that current models generally struggle to generate accurate and comprehensive captions while maintaining consistent subject references. Moreover, as video duration increases, both the overall caption quality and subject referential consistency decline. Notably, our evaluation metrics exhibit strong correlations with the performance of downstream understanding and generation tasks conditioned on the generated captions, further validating their effectiveness. The project is available at https://github.com/xlchen0205/CapRiCorn-1K .
Chinese Translation
准确且全面的视频字幕与一致的主题指称对于下游理解和生成任务至关重要。然而，现有的基准很少能够客观且全面地评估这些属性在不同时长和场景下的表现，从而阻碍了视频字幕模型的发展。为了解决这一问题，我们提出了CapRiCorn-1K，这是一个综合基准，旨在评估视频字幕质量和主题指称一致性，涵盖较长的时间范围和多样的视频领域。为了满足不同的评估需求，我们的基准支持视听和仅视觉两种设置。在CapRiCorn-1K上的广泛实验表明，当前模型在生成准确且全面的字幕时，通常难以保持一致的主题指称。此外，随着视频时长的增加，整体字幕质量和主题指称一致性均有所下降。值得注意的是，我们的评估指标与基于生成字幕的下游理解和生成任务的表现之间表现出强相关性，进一步验证了其有效性。该项目可在 https://github.com/xlchen0205/CapRiCorn-1K 获取。
cs.CV / 120 / 2606.21956
Denoising-Enhanced Coarse-to-Fine Infrared Small Target Detection with Attention Prior-Guided Knowledge Distillation
基于注意力先验引导知识蒸馏的去噪增强粗到细红外小目标检测
Abstract
Infrared small target detection (IRSTD) in high-resolution images is crucial for many practical applications, such as surveillance of unmanned aerial vehicles (UAVs) and UAV-based ground monitoring. However, IRSTD remains challenging due to the small size and weak features of targets, as well as significant interference from complex dynamic backgrounds. Existing detection methods often suffer from redundant computations on non-target background regions and insufficient exploitation of target context information, which limits their performance in complex backgrounds. To address these issues, we propose an efficient coarse-to-fine infrared small target detection framework with attention prior-guided knowledge distillation, termed ECFNet. In the coarse stage, we design a region binary classification network (RBCN) on grid-based multi-scale feature maps to efficiently recognize target-containing context region proposals while suppressing complex backgrounds. Moreover, we introduce a novel denoising-assisted training strategy that incorporates noisy ground-truth (GT) masks into the feature maps of RBCN and trains the network to reconstruct the GT masks through a denoising task, thereby enhancing its ability to distinguish target proposals from background regions and accelerating convergence. In the fine stage, we customize a lightweight target detector to the coarse stage's region proposals for balancing accuracy and efficiency. Furthermore, we propose a knowledge distillation strategy guided by the teacher-student cross-attention prior. This mechanism directs the student to focus on critical target regions, thereby enhancing the discriminative feature representation for infrared small targets. Extensive experiments on three real infrared datasets demonstrate that our method outperforms both existing single-stage and two-stage approaches while maintaining high real-time processing efficiency.
Chinese Translation
高分辨率图像中的红外小目标检测（IRSTD）对许多实际应用至关重要，例如无人机（UAV）监视和基于无人机的地面监测。然而，由于目标的体积小、特征弱以及复杂动态背景的显著干扰，IRSTD仍然面临挑战。现有的检测方法往往在非目标背景区域进行冗余计算，并且对目标上下文信息的利用不足，这限制了它们在复杂背景下的性能。为了解决这些问题，我们提出了一种高效的粗到细红外小目标检测框架，称为ECFNet，该框架采用注意力先验引导的知识蒸馏。在粗阶段，我们设计了一个基于网格的多尺度特征图的区域二分类网络（RBCN），以高效识别包含目标的上下文区域提议，同时抑制复杂背景。此外，我们引入了一种新颖的去噪辅助训练策略，将噪声真实标签（GT）掩码融入RBCN的特征图中，并通过去噪任务训练网络重建GT掩码，从而增强其区分目标提议与背景区域的能力，并加速收敛。在细阶段，我们为粗阶段的区域提议定制了一个轻量级目标检测器，以平衡准确性和效率。此外，我们提出了一种由教师-学生交叉注意力先验引导的知识蒸馏策略。该机制引导学生关注关键目标区域，从而增强红外小目标的区分特征表示。在三个真实红外数据集上的大量实验表明，我们的方法在保持高实时处理效率的同时，优于现有的单阶段和双阶段方法。
cs.CV / 121 / 2606.21968
Look Before You Zoom: Adaptive Routing for the Resolution-Context Trade-off in Visual RAG
在放大之前先观察：视觉RAG中分辨率与上下文的自适应路由
Abstract
Vision-Language Models (VLMs) struggle as query-relevant objects become smaller. To address this, recent training-free approaches dynamically retrieve and zoom into local image regions. However, we show that indiscriminately applying retrieval ignores a critical vulnerability: the resolution-context trade-off. Patch-based zooming recovers details for small targets, but can split large objects and destroy global spatial context; attention-based retrieval better preserves large objects, but remains less reliable on tiny details; and global perception is often fastest when retrieval is unnecessary. Motivated by these failure modes, we introduce ViRGo (Visual Retrieval or Global Perception), a lightweight framework that formulates visual retrieval as an adaptive routing problem. ViRGo estimates object scale from the VLM's intrinsic localization heads during the initial forward pass and combines it with semantic token confidence to select between global perception, patch-based retrieval, and attention-based retrieval with minimal additional computation. Experiments across multiple VQA benchmarks and object-size groups show that ViRGo improves the accuracy-efficiency trade-off: it matches patch retrieval on small details, leverages attention-based retrieval for larger objects, and reduces inference time by routing to the global baseline when zooming is unnecessary.
Chinese Translation
视觉-语言模型（VLMs）在查询相关对象变得更小时面临挑战。为了解决这个问题，最近的无训练方法动态地检索并放大局部图像区域。然而，我们表明，盲目应用检索忽视了一个关键的脆弱性：分辨率与上下文的权衡。基于补丁的放大可以恢复小目标的细节，但可能会分割大型物体并破坏全局空间上下文；基于注意力的检索更好地保留大型物体，但在微小细节上仍然不够可靠；而当不需要检索时，全局感知通常是最快的。受到这些失败模式的启发，我们提出了ViRGo（视觉检索或全局感知），这是一个将视觉检索公式化为自适应路由问题的轻量级框架。ViRGo在初始前向传播过程中从VLM的内在定位头估计对象规模，并结合语义标记置信度，以在全局感知、基于补丁的检索和基于注意力的检索之间进行选择，且额外计算量最小。跨多个视觉问答基准和对象大小组的实验表明，ViRGo改善了准确性与效率的权衡：它在小细节上与补丁检索相匹配，利用基于注意力的检索处理较大物体，并在不需要放大时通过路由到全局基线来减少推理时间。
cs.CV / 122 / 2606.21982
CoDMD: Copula-aware Distribution Matching Distillation for Fast Video Generation
CoDMD：考虑联结的分布匹配蒸馏用于快速视频生成
Abstract
Few-step distillation for video diffusion models has attracted significant attention, driven by the urgent demand for efficient deployment in real-world scenarios. However, Distribution Matching Distillation (DMD), a leading paradigm, tends to degrade under limited NFE budgets, manifesting in video generation as layout instability, oversaturation, and broken motion dynamics. We trace this failure to a structural limitation: standard DMD is an intra-sample distribution-matching objective with coordinate-wise gradients, and thus imposes no explicit constraint on the relational geometry across batch elements or temporal frames, leaving the underlying copula largely unregulated. Combined with the mode-seeking tendency of its reverse-KL objective, this absence of relational guidance makes DMD prone to collapsing into local optima in the few-step regime. Motivated by this insight, we propose Copula-aware DMD (CoDMD), a lightweight relational regularizer that reuses score estimates already produced by the frozen teacher and the online fake model to construct pairwise relation matrices across samples and frames. These are matched through a supplementary distributional objective that requires no additional networks, datasets, or sampling trajectories. On the Wan-2.1-T2V model series at 1.3B & 14B scales, CoDMD distills 50-step teachers into 4-step students, achieving an approximate 25$\times$ speed-up while attaining VBench scores of 84.46 & 84.87, outperforming prior trajectory-based (rCM 82.81 & 84.05) and distribution-based (DMD 83.38 & 83.81) methods.
Chinese Translation
针对视频扩散模型的少步蒸馏引起了广泛关注，主要是由于在现实场景中对高效部署的迫切需求。然而，作为一种主流范式的分布匹配蒸馏（DMD）在有限的NFE预算下往往表现不佳，这在视频生成中表现为布局不稳定、过饱和和运动动态破裂。我们将这一失败追溯到结构限制：标准DMD是一个内部样本分布匹配目标，采用坐标逐维梯度，因此对批次元素或时间帧之间的关系几何没有明确约束，导致底层的联结（copula）几乎没有受到调控。结合其反向KL目标的寻模倾向，这种缺乏关系指导使得DMD在少步情况下容易陷入局部最优。基于这一见解，我们提出了考虑联结的DMD（CoDMD），这是一种轻量级的关系正则化器，重用已经由冻结教师模型和在线假模型生成的评分估计，以构建样本和帧之间的成对关系矩阵。这些矩阵通过一个附加的分布目标进行匹配，该目标不需要额外的网络、数据集或采样轨迹。在1.3B和14B规模的Wan-2.1-T2V模型系列上，CoDMD将50步教师模型蒸馏为4步学生模型，实现了约25倍的速度提升，同时获得了84.46和84.87的VBench分数，超越了先前基于轨迹的方法（rCM 82.81和84.05）和基于分布的方法（DMD 83.38和83.81）。
cs.CV / 123 / 2606.22002
One-Shot Data Selection for Medical Image Classification via Graph Coverage
通过图覆盖进行医学图像分类的一次性数据选择
Abstract
Training medical image classifiers on entire datasets is wasteful when annotation budgets are limited: not all samples contribute equally, yet acquiring expert labels is expensive. Active learning reduces annotation cost through iterative querying, but assumes repeated access to an oracle and requires multiple rounds of model training. One-shot geometry-based methods such as facility location avoid retraining but operate on pairwise distances that ignore the local structure of the data manifold. We propose a graph-based one-shot selection method that operates entirely on frozen foundation model embeddings. Given embeddings from a pretrained encoder, we construct a k-nearest neighbor graph over all training samples and derive a two-term coverage kernel from the heat diffusion kernel, capturing both direct and two-hop neighborhood relationships. Greedy facility location on this kernel selects class-balanced subsets that maximize coverage of the data manifold. The two-term kernel matches the full spectral heat kernel in selection behavior while reducing computation to sparse matrix operations with a single hyperparameter. We evaluate on five MedMNIST datasets spanning histopathology, radiology, and microscopy, comparing against both training-dynamics and geometry-based baselines. Our method achieves the highest balanced accuracy on nine of ten dataset-ratio conditions, with the largest gains on class-imbalanced datasets where global graph construction captures cross-class structure that per-class methods miss, all without any model training during selection. Code is available at https://github.com/zahiriddin-rustamov/graph-coverage-selection.
Chinese Translation
在注释预算有限的情况下，使用整个数据集训练医学图像分类器是浪费的：并非所有样本的贡献都是相等的，而获取专家标签的成本很高。主动学习通过迭代查询来降低注释成本，但假设可以重复访问一个oracle，并且需要多轮模型训练。一些基于一次性几何的方法，如设施选址，避免了重新训练，但在处理成对距离时忽略了数据流形的局部结构。我们提出了一种基于图的一次性选择方法，完全基于冻结的基础模型嵌入。给定来自预训练编码器的嵌入，我们在所有训练样本上构建一个k近邻图，并从热扩散核中推导出一个双项覆盖核，以捕捉直接和两跳邻域关系。对该核进行贪心设施选址选择类平衡子集，以最大化数据流形的覆盖。双项核在选择行为上与完整的谱热核相匹配，同时将计算减少到稀疏矩阵操作，并具有一个超参数。我们在涵盖组织病理学、放射学和显微镜学的五个MedMNIST数据集上进行评估，并与基于训练动态和几何的基线进行比较。我们的方法在十个数据集比例条件中的九个上实现了最高的平衡准确率，在类不平衡数据集上获得了最大的提升，因为全局图构建捕捉了跨类结构，而逐类方法则遗漏了这一点，且在选择过程中没有进行任何模型训练。代码可在 https://github.com/zahiriddin-rustamov/graph-coverage-selection 获取。
cs.CV / 124 / 2606.22029
Topological summaries of fingerprint ridge patterns carry identity information
指纹脊线模式的拓扑摘要携带身份信息
Abstract
Fingerprints are the most widely deployed biometric. Verifying whether two impressions come from the same finger typically relies on minutiae, small landmarks such as skin ridge endings and bifurcations. These landmarks are extracted through a multi-stage pipeline of image enhancement, skeletonization, minutiae detection, and alignment. We investigate an alternative: using topological data analysis to represent the full pattern of skin ridges and valleys directly, bypassing minutiae detection and the downstream matching pipeline. We apply persistent homology, a topological tool that tracks how loops in the ridge pattern form and fill in across spatial scales, producing multi-scale summaries of ridge geometry. We develop and compare a range of verification methods on a standard benchmark dataset, FVC2000 DB1. Even the simplest topological summaries, with no trained parameters, substantially outperform geometry-only baselines. A trained method achieves an AUC of 0.91, while an optimal-transport method excels at the strictest false-accept thresholds, suggesting they capture different aspects of the ridge pattern. Fusing these two approaches yields the best performance at every low false-accept threshold we examine. Our results establish that these topological summaries capture substantial fingerprint identity information, far more effective for verification than raw pixel-level geometry. Because the entire pipeline is openly specified, it offers a transparent complement to minutiae-based systems, and we provide a modular framework for constructing, evaluating, and combining topological verification methods.
Chinese Translation
指纹是最广泛应用的生物识别技术。验证两个印记是否来自同一手指通常依赖于细节特征（minutiae），即皮肤脊线末端和分叉等小标志。这些标志通过图像增强、骨架化、细节特征检测和对齐等多阶段流程提取。我们研究了一种替代方法：使用拓扑数据分析直接表示皮肤脊线和谷的完整模式，绕过细节特征检测和后续匹配流程。我们应用持久同调（persistent homology），这是一种拓扑工具，跟踪脊线模式中环的形成和填充如何在空间尺度上变化，从而生成脊线几何的多尺度摘要。我们在标准基准数据集FVC2000 DB1上开发并比较了一系列验证方法。即使是最简单的拓扑摘要，没有训练参数，也显著优于仅基于几何的基线。经过训练的方法达到了0.91的AUC，而最优传输方法在最严格的假接受阈值下表现出色，表明它们捕捉了脊线模式的不同方面。将这两种方法融合在一起，在我们检查的每个低假接受阈值下都实现了最佳性能。我们的结果表明，这些拓扑摘要捕捉了大量的指纹身份信息，远比原始像素级几何更有效于验证。由于整个流程是公开指定的，它为基于细节特征的系统提供了透明的补充，我们提供了一个模块化框架，用于构建、评估和结合拓扑验证方法。
cs.CV / 125 / 2606.22042
IDAG-Edit: Multi-Object Video Editing via Instance-Decoupled Attention and Guidance
IDAG-Edit：通过实例解耦注意力和引导实现的多对象视频编辑
Abstract
Diffusion-based video editing has made significant progress; however, achieving precise and temporally consistent object-level control, especially in multi-object scenarios, remains challenging due to attention leakage, identity drift, and unstable temporal dynamics. In this work, we propose IDAGEdit, a training-free framework for fine-grained multi-object video editing with strong temporal consistency. The framework adopts Layout-guided Attention Modulation to facilitate coherent multi-object editing, while Instance-level Masks are introduced to preserve individual object identity and enforce localized attention within each object region, thereby enabling fine-grained, object-level editing. Extensive qualitative and quantitative evaluations demonstrate that our method improves temporal stability and multi-object controllability over state-of-the-art video editing approaches.
Chinese Translation
基于扩散的视频编辑取得了显著进展；然而，实现精确且时间一致的对象级控制，特别是在多对象场景中，仍然面临挑战，这主要是由于注意力泄漏、身份漂移和不稳定的时间动态。在本研究中，我们提出了IDAG-Edit，一个无训练的框架，用于细粒度的多对象视频编辑，并具有强大的时间一致性。该框架采用布局引导的注意力调制，以促进连贯的多对象编辑，同时引入实例级掩码以保持单个对象的身份，并在每个对象区域内强制局部注意力，从而实现细粒度的对象级编辑。大量的定性和定量评估表明，我们的方法在时间稳定性和多对象可控性方面优于最先进的视频编辑方法。
cs.CV / 126 / 2606.22072
A Controlled Study of CLIP-Based Body-Scene Fusion for Emotion Recognition in Context
基于CLIP的身体-场景融合在情境下的情感识别的对照研究
Abstract
Apparent emotion in natural images is often not visible from the face alone. The face may be small, hidden, or neutral, while posture and scene context carry much of the evidence. This work studies context-aware emotion recognition on EMOTIC with an image-only two-stream model. A ResNet-18 body stream encodes the target-person crop, and a CLIP ViT-B/16 scene stream encodes the full image. The fused feature predicts 26 categorical emotion labels and the continuous valence, arousal, and dominance values. This study examines whether small context-debiasing or rare-class training changes still help after adding a CLIP scene encoder. The clean two-stream model is compared with simplified CCIM-style intervention, CLEF-lite context-bias subtraction, ASL tuning, and class-balanced sampling under the same implementation pipeline. No tested variant improves over the clean two-stream model, which achieves 34.52% mAP on the EMOTIC test split. CLIP gives the model broad scene semantics, but the simplified causal, counterfactual, and rare-class changes do not automatically improve performance. Most remaining errors are in rare and subtle emotion categories, so the next step should focus on label relationships and finer subject-context interaction.
Chinese Translation
自然图像中的明显情感往往无法仅通过面部来观察。面部可能较小、被遮挡或呈中性，而姿势和场景上下文则承载了大量证据。本研究在EMOTIC数据集上研究了基于上下文的情感识别，采用了仅图像的双流模型。ResNet-18身体流编码目标人物的裁剪图像，而CLIP ViT-B/16场景流编码完整图像。融合特征预测26个类别的情感标签以及连续的愉悦度、激动度和主导度值。本研究考察了在添加CLIP场景编码器后，小规模的去偏差或稀有类别训练变化是否仍然有效。干净的双流模型与简化的CCIM风格干预、CLEF-lite上下文偏差减法、ASL调优和在相同实现管道下的类别平衡采样进行了比较。没有测试的变体在性能上超过干净的双流模型，该模型在EMOTIC测试集上达到了34.52%的mAP。CLIP为模型提供了广泛的场景语义，但简化的因果、反事实和稀有类别变化并未自动提高性能。大多数剩余错误出现在稀有和微妙的情感类别中，因此下一步应集中于标签关系和更细致的主体-上下文交互。
cs.CV / 127 / 2606.22076
Learning Cross-View Semantic Priors for Single-Reference Unseen Object Pose Estimation
学习跨视图语义先验以进行单参考未见物体位姿估计
Abstract
Single-reference unseen object 6D pose estimation reduces object onboarding by estimating poses of arbitrary novel objects from only one reference view. Recent correspondence-based pipelines have achieved robust performance with vision foundation model (VFM) features. However, they typically treat these features as intra-view descriptors, leaving dense visual-semantic cues, including appearance, structure, and context, insufficiently exchanged across views before geometric decoding. Consequently, the decoded point features may lack joint semantic and geometric discriminability, making correspondence estimation still difficult in challenging cases. Instead of processing features independently, we build the correspondence pipeline around an early cross-view semantic prior. Specifically, cross-view semantic interaction (CVSI) enables dense query and reference VFM tokens to exchange semantic context and form a cross-view prior. Nevertheless, direct CVSI may disturb the VFM token structure, while the resulting semantic prior still needs 3D representation consistency for rigid correspondence. To make this CVSI prior reliable for 3D correspondence learning, we introduce two complementary training-time constraints: the intra-view structure preservation (IVSP) loss preserves the original intra-view token affinity structure during interaction, while the reference-anchored geometric consistency (RAGC) loss enforces spatial representation consistency of decoded point features. The final pose is recovered from learned correspondences through weighted SVD. We further construct a challenging view-pair protocol from the BOP Challenge datasets YCB-V and TUD-L to evaluate robustness in difficult matching scenarios. Extensive experiments on six benchmarks under different view-pair settings show that our method achieves state-of-the-art performance while maintaining comparable inference speed.
Chinese Translation
单参考未见物体的六维位姿估计通过仅从一个参考视图估计任意新物体的位姿，从而减少了物体的上载。最近基于对应关系的管道在视觉基础模型（VFM）特征下取得了稳健的性能。然而，它们通常将这些特征视为视内描述符，导致在几何解码之前，密集的视觉-语义线索（包括外观、结构和上下文）在视图之间的交换不足。因此，解码的点特征可能缺乏联合语义和几何可区分性，使得在具有挑战性的情况下仍然难以进行对应关系估计。我们不再独立处理特征，而是围绕早期的跨视图语义先验构建对应关系管道。具体而言，跨视图语义交互（CVSI）使得密集的查询和参考VFM标记能够交换语义上下文并形成跨视图先验。然而，直接的CVSI可能会干扰VFM标记的结构，而生成的语义先验仍然需要在刚性对应关系中保持三维表示一致性。为了使这个CVSI先验在三维对应学习中可靠，我们引入了两个互补的训练时间约束：视内结构保持（IVSP）损失在交互过程中保持原始的视内标记亲和结构，而参考锚定几何一致性（RAGC）损失则强制解码点特征的空间表示一致性。最终的位姿通过加权奇异值分解（SVD）从学习到的对应关系中恢复。我们进一步从BOP挑战数据集YCB-V和TUD-L构建了一个具有挑战性的视图对协议，以评估在困难匹配场景中的鲁棒性。在不同视图对设置下对六个基准的广泛实验表明，我们的方法在保持可比推理速度的同时实现了最先进的性能。
cs.CV / 128 / 2606.22077
Morphology-Aware Multimodal Representation Learning for Insect Phylogenetic Reconstruction
考虑形态的多模态表示学习用于昆虫系统发育重建
Abstract
Morphological traits provide important evidence for phylogenetic reconstruction and evolutionary relationship analysis. Recent image-based approaches have introduced deep learning, particularly convolutional models, to derive morphological features from specimen images, but these methods generally rely on single-modality visual representations and do not explicitly incorporate morphological semantics. This study proposes a morphology-aware multimodal alignment framework for insect phylogenetic reconstruction. The framework combines specimen images with curated morphological descriptions by adapting a vision transformer through parameter-efficient fine-tuning and supervised contrastive learning, followed by image-text alignment in a shared latent space. The learned image embeddings are then used as continuous traits for Bayesian phylogenetic reconstruction. On the public Rove-Tree-11 dataset, comparative and ablation experiments across multiple visual backbones and feature adaptation strategies demonstrate that multimodal alignment improves topological agreement with the reference phylogeny. The results indicate that the proposed framework can derive morphology-aware visual traits for computational phylogenetic reconstruction.
Chinese Translation
形态特征为系统发育重建和进化关系分析提供了重要证据。近期基于图像的方法引入了深度学习，特别是卷积模型，从标本图像中提取形态特征，但这些方法通常依赖于单一模态的视觉表示，并未明确纳入形态语义。本研究提出了一种考虑形态的多模态对齐框架，用于昆虫系统发育重建。该框架通过参数高效微调和监督对比学习，将标本图像与精心策划的形态描述相结合，并在共享潜在空间中进行图像-文本对齐。学习到的图像嵌入随后被用作贝叶斯系统发育重建的连续特征。在公共的Rove-Tree-11数据集上，针对多个视觉骨干网络和特征适应策略的比较和消融实验表明，多模态对齐提高了与参考系统发育的拓扑一致性。结果表明，所提出的框架能够为计算系统发育重建提取考虑形态的视觉特征。
cs.CV / 129 / 2606.22089
BAC-JEPA: Label-Efficient Breast Arterial Calcification Segmentation via Synthetic Mammography-Guided Supervision
BAC-JEPA：通过合成乳腺X光引导监督实现标签高效的乳腺动脉钙化分割
Abstract
Breast arterial calcification (BAC) on screening mammograms is an emerging cardiovascular risk biomarker, but quantitative use requires reproducible segmentation and expert pixel-level labels are costly. We present BAC-JEPA, a label-efficient segmentation framework trained on procedurally generated arterial calcification inserted into real mammographic backgrounds with exact masks. Candidate backgrounds were selected from model-screened mammograms with low predicted BAC response; the generator samples arterial structure, disease burden, radiographic appearance, and hard-negative distractors including nonarterial calcifications and metallic objects. Synthetic masks are paired with mammography self-supervised Vision Transformer encoders and a high-resolution convolutional decoder to produce full-resolution segmentation maps. The study used 75,472 mammography studies from 34,956 patients for background selection and representation learning, trained on synthetic images from 10,000 backgrounds, selected checkpoints with 1,000 development backgrounds, and evaluated transfer on all 1,000 human-labeled BacSeg synthetic 2D mammograms. On held-out synthetic validation data, the larger backbone achieved IoU 0.5325 and Dice 0.6357. On BacSeg, image-level classification from segmentation probability maps reached AUROC 0.8719, with 0.8547 for the smaller backbone. Four-view inference required 110.68--213.63 ms on an RTX 5090 GPU, and severe-preset synthetic image generation averaged 2.7071 s per image on a multicore workstation. These results indicate that BAC-specific synthetic supervision can produce useful image-level transfer without human pixel-level training masks, while expert-reviewed real-mammogram segmentation remains necessary for clinical validation and calibration.
Chinese Translation
乳腺动脉钙化（BAC）在筛查乳腺X光片上是一个新兴的心血管风险生物标志物，但其定量使用需要可重复的分割，而专家级像素级标签的获取成本高昂。我们提出了BAC-JEPA，一种标签高效的分割框架，该框架在真实乳腺X光背景中插入程序生成的动脉钙化并使用精确的掩膜进行训练。候选背景从模型筛选的低预测BAC反应的乳腺X光片中选择；生成器样本包括动脉结构、疾病负担、放射外观以及包括非动脉钙化和金属物体在内的难负样本。合成掩膜与乳腺X光自监督视觉变换器编码器和高分辨率卷积解码器配对，以生成全分辨率的分割图。该研究使用了来自34,956名患者的75,472个乳腺X光研究用于背景选择和表示学习，训练了来自10,000个背景的合成图像，选择了1,000个开发背景的检查点，并在所有1,000个人工标记的BacSeg合成2D乳腺X光片上进行了转移评估。在保留的合成验证数据上，较大的主干网络达到了IoU 0.5325和Dice 0.6357。在BacSeg上，基于分割概率图的图像级分类达到了AUROC 0.8719，较小的主干网络为0.8547。四视图推理在RTX 5090 GPU上需要110.68至213.63毫秒，而严重预设的合成图像生成在多核工作站上平均每张图像耗时2.7071秒。这些结果表明，BAC特定的合成监督可以在没有人类像素级训练掩膜的情况下产生有用的图像级转移，而专家审核的真实乳腺X光分割仍然是临床验证和校准所必需的。
cs.CV / 130 / 2606.22094
Cross-View Yaw Estimation in Location Uncertainty with Line-Aligning Yaw Scoring
位置不确定性下的跨视角偏航估计与线对齐偏航评分
Abstract
Accurate yaw estimation is a bottleneck in cross-view localization between ground view and Bird's Eye View (BEV). Existing methods couple yaw with translation and rely on height or projection assumptions that degrade under large yaw ambiguity. We disentangle yaw from location accuracy and introduce LAYS, a radially invariant line-consensus voting method. By exploiting the radial invariance of our formulation, we achieve sub-degree yaw precision via 3D voting over all candidate poses, while eliminating the need for accurate location. Our key observation is that a ground-image column matched to BEV pixels induces the same yaw across all camera positions along the radial direction of the pixels. LAYS matches BEV pixels to ground columns using feature similarity and accumulates the induced yaw votes into discrete 3D bins, where correct correspondences along the radial line concentrate into a sharp peak for the correct yaw. Experiments on Mapillary, Ford, KITTI, and VIGOR show significant gains under unknown yaw, particularly for normal FoV with unknown yaw (+28$\sim$45\%p), and using LAYS as a yaw prior improves downstream 3-DoF localization.
Chinese Translation
准确的偏航估计是地面视角与鸟瞰视角（BEV）之间跨视角定位的瓶颈。现有方法将偏航与平移耦合，并依赖于高度或投影假设，这在大偏航模糊下会降低效果。我们将偏航与位置精度解耦，并引入LAYS（线对齐偏航评分），一种径向不变的线共识投票方法。通过利用我们方法的径向不变性，我们通过对所有候选姿态进行3D投票，实现了亚度数的偏航精度，同时消除了对准确位置的需求。我们的关键观察是，与BEV像素匹配的地面图像列在像素的径向方向上诱导出相同的偏航。LAYS通过特征相似性将BEV像素与地面列匹配，并将诱导的偏航投票累积到离散的3D箱中，其中沿径向线的正确对应关系集中成一个尖锐的峰值，指向正确的偏航。在Mapillary、Ford、KITTI和VIGOR上的实验表明，在未知偏航下有显著提升，特别是在未知偏航的正常视场下（+28$ ilde{ }$45 ext{%p}），并且将LAYS作为偏航先验改善了下游的3自由度定位。
cs.CV / 131 / 2606.22112
Accurate identification and measurement of the precipitate area by two-stage deep neural networks in novel chromium-based alloys
通过两阶段深度神经网络准确识别和测量新型铬基合金的沉淀区
Abstract
The performance of advanced materials for extreme environments is underpinned by their microstructure, including the size and distribution of reinforcing phases. Chromium-based superalloys are a recently proposed alternative to conventional face-centred-cubic superalloys for high-temperature applications, such as Concentrated Solar Power, and their development requires efficient measurement of precipitate volume fraction and size distribution from electron microscopy images. Traditional fixed-threshold image processing is sensitive to background noise, generalises poorly across materials, and requires substantial manual measurement effort. To address these bottlenecks, this study proposes DT-SegNet, an end-to-end two-stage deep learning scheme based on YOLOv5 and SegFormer for object detection and segmentation in electron microscopy images. The approach combines the training efficiency of convolutional neural networks at the detection stage with the segmentation accuracy of a Vision Transformer. Numerical experiments show that DT-SegNet substantially outperforms state-of-the-art segmentation tools offered by Weka and ilastik across metrics including accuracy, precision, recall, and F1-score. The model provides a useful tool for alloy-development microstructure examinations and helps address the large datasets associated with high-throughput alloy development.
Chinese Translation
先进材料在极端环境下的性能依赖于其微观结构，包括增强相的大小和分布。铬基超级合金是最近提出的高温应用（如集中式太阳能发电）中，传统面心立方超级合金的替代品，其开发需要从电子显微镜图像中高效测量沉淀体积分数和尺寸分布。传统的固定阈值图像处理对背景噪声敏感，跨材料的泛化能力差，并且需要大量的人工测量工作。为了解决这些瓶颈，本研究提出了DT-SegNet，一种基于YOLOv5和SegFormer的端到端两阶段深度学习方案，用于电子显微镜图像中的目标检测和分割。该方法在检测阶段结合了卷积神经网络的训练效率与视觉变换器的分割精度。数值实验表明，DT-SegNet在准确性、精确度、召回率和F1-score等指标上显著优于Weka和ilastik提供的最先进分割工具。该模型为合金开发的微观结构检查提供了有用的工具，并有助于解决与高通量合金开发相关的大数据集问题。
cs.CV / 132 / 2606.22124
Surgical Anatomy Recognition with Context Learning using Foundation Representations
基于基础表示的上下文学习的外科解剖识别
Abstract
Accurate recognition of anatomical structures is essential for safe and effective minimally invasive surgery (MIS), yet it remains underexplored in surgical computer vision due to limited annotated data and methods tailored primarily to natural scenes. In this work, we present a combined dataset and model framework to advance anatomy-aware perception in MIS. First, we introduce ATLAS-120k, a large-scale clip-level semantic segmentation dataset comprising over 120,000 annotated frames from 100 surgical videos spanning 14 procedures and multiple modalities, including laparoscopic and robot-assisted surgery. The dataset captures substantial procedural variability and was created using a scalable annotation pipeline that integrates expert manual labeling, automated propagation, iterative refinement, and surgeon verification to ensure high-quality annotations. Second, we propose ATLAS (Anatomy Recognition with Context Learning using Foundation Representations), a video semantic segmentation model specifically designed for surgical anatomy recognition. Unlike conventional approaches that emphasize object tracking, ATLAS leverages foundation-model embeddings together with lightweight temporal reasoning to incorporate contextual cues such as procedure type, surgical phase, and short-term visual memory. This design enables temporally consistent and accurate predictions while maintaining real-time feasibility. Together, the dataset and model establish a practical foundation for robust surgical scene understanding and support the development of clinically applicable guidance systems for minimally invasive surgery. The models, dataset annotations and annotation platform are publicly available at: https://github.com/TimJaspers0801/ATLAS.
Chinese Translation
准确识别解剖结构对于安全有效的微创手术（MIS）至关重要，但由于标注数据有限以及方法主要针对自然场景，这一领域在外科计算机视觉中仍未得到充分探索。在本研究中，我们提出了一个结合数据集和模型框架，以推动MIS中的解剖意识感知。首先，我们介绍了ATLAS-120k，这是一个大规模的剪辑级语义分割数据集，包含来自100个外科视频的超过120,000帧标注图像，涵盖14种手术程序和多种模式，包括腹腔镜手术和机器人辅助手术。该数据集捕捉了显著的程序变异性，并采用可扩展的标注流程创建，结合了专家手动标注、自动传播、迭代精炼和外科医生验证，以确保高质量的标注。其次，我们提出了ATLAS（基于基础表示的上下文学习的解剖识别），这是一个专门为外科解剖识别设计的视频语义分割模型。与强调物体跟踪的传统方法不同，ATLAS利用基础模型嵌入和轻量级时间推理，结合程序类型、外科阶段和短期视觉记忆等上下文线索。这一设计使得模型能够在保持实时可行性的同时，实现时间一致性和准确的预测。数据集和模型共同为稳健的外科场景理解奠定了实用基础，并支持微创手术临床应用指导系统的开发。模型、数据集标注和标注平台均可在以下网址公开获取：https://github.com/TimJaspers0801/ATLAS。
cs.CV / 133 / 2606.22131
Feed-forward Motion In-betweening for Any 4D
任意4D的前馈运动插值
Abstract
4D dynamics (3D geometry evolving over time) is a fundamental representation of the physical world and plays a crucial role in world modeling (e.g., animation and games). Owing to the scarcity of large-scale, long-horizon 4D mesh data with arbitrary shapes, early text-to-4D methods rely on distillation or test-time optimization from video diffusion priors, making inference prohibitively slow. Recent feed-forward generators greatly reduce inference cost but offer limited spatiotemporal controllability, and short-horizon generation often leads to error accumulation in long-horizon sequences. We propose a novel feed-forward in-betweening framework for arbitrary 4D meshes with keyframe conditioning. Building on universal mesh-animation latents, we introduce a frame-wise mesh VAE that encodes each frame into topology-agnostic latent tokens anchored by a reference mesh for keyframe conditioning. We further introduce a keyframe-conditioned rectified flow model with an MMDiT backbone that synthesizes non-keyframe frames conditioned on sparse keyframes. Experiments show strong performance and improved controllability on both DyMesh16 and DyMesh32 benchmarks.
Chinese Translation
4D动态（随时间演变的3D几何形状）是物理世界的基本表示， 在世界建模（例如动画和游戏）中发挥着至关重要的作用。由于缺乏大规模、长时间跨度的任意形状4D网格数据，早期的文本到4D方法依赖于从视频扩散先验中提取或测试时优化，导致推断速度极慢。最近的前馈生成器大大降低了推断成本，但提供的时空可控性有限，短时间生成往往导致长时间序列中的错误积累。我们提出了一种新颖的前馈插值框架，适用于任意4D网格，并结合关键帧条件。基于通用网格动画潜变量，我们引入了一种逐帧网格变分自编码器（VAE），将每一帧编码为由参考网格锚定的拓扑无关潜在标记，以便进行关键帧条件处理。我们进一步引入了一种关键帧条件的修正流模型，采用MMDiT骨干网络，合成基于稀疏关键帧的非关键帧。实验表明，在DyMesh16和DyMesh32基准测试中表现出强大的性能和改善的可控性。
cs.CV / 134 / 2606.22144
SAGE: An Expert-Annotated South Asian GI Endoscopy Dataset for Multimodal Learning and Hallucination Analysis
SAGE：一个专家注释的南亚胃肠内镜数据集，用于多模态学习和幻觉分析
Abstract
Gastrointestinal cancers represent a growing health burden in the South Asian region, driven largely by rapid changes in socio-economic conditions & lifestyle habits. However, early diagnosis of such malignancies remains a significant challenge, largely due to a lack of modern equipment, lack of financial support, and a scarcity of GI experts. AI-assisted diagnosis & report generation, show great promise in alleviating this problem by providing low-skill manpower the technical expertise to perform diagnosis. However, almost all open-source, publicly available datasets are predominantly collected from the European region, with no representation from the South Asian region. The lack of open-source GI datasets from diverse geographic regions has made it difficult to assess whether population bias is present in existing models, and to develop geographically inclusive AI tools for automated GI diagnosis. To address this gap, we introduce SAGE: An Expert-Annotated South Asian GI Endoscopy dataset for image captioning, multi-label classification, and visual question answering (VQA) tasks. It consists of 1,300 images, their captions along with hallucination tag, 18 labels and 14,726 question-answer pairs making it well-suited for diverse range of tasks including classification, benchmarking, and fine-tuning large multimodal models (LMMs). We further conducted benchmarking of multi-class classifiers on the effect of population shift in GI imaging AI tasks, and contemporary LMMs on their performance. Our study reveals that task-specific models, such as multi-class classification models, suffer the most, with an average performance drop of 58% when evaluated on the South Asian dataset. For contemporary LMMs, benchmarking reveals a substantial drop in the average GREEN score for anatomical landmark detection (0.308) and abnormality detection (0.410).
Chinese Translation
胃肠癌在南亚地区日益成为一个健康负担，这主要是由于社会经济条件和生活方式的快速变化。然而，这些恶性肿瘤的早期诊断仍然是一个重大挑战，主要由于现代设备的缺乏、财政支持不足以及胃肠专家的稀缺。人工智能辅助的诊断和报告生成在缓解这一问题方面展现出巨大潜力，能够为低技能劳动力提供进行诊断所需的技术专长。然而，几乎所有公开的、可用的开源数据集主要来自欧洲地区，南亚地区的数据缺乏。这种来自不同地理区域的开源胃肠数据集的缺乏，使得评估现有模型是否存在人口偏差变得困难，并且难以开发地理上包容的自动化胃肠诊断人工智能工具。为了解决这一空白，我们推出了SAGE：一个专家注释的南亚胃肠内镜数据集，适用于图像标注、多标签分类和视觉问答（VQA）任务。该数据集包含1,300张图像及其标题、幻觉标签、18个标签和14,726个问答对，使其非常适合于包括分类、基准测试和微调大型多模态模型（LMMs）在内的多种任务。我们进一步对多类分类器在胃肠影像人工智能任务中人口转变的影响进行了基准测试，并对当代LMMs的表现进行了评估。我们的研究表明，特定任务模型（如多类分类模型）受到的影响最大，在南亚数据集上的平均性能下降了58%。对于当代LMMs，基准测试显示解剖标志检测（0.308）和异常检测（0.410）的平均GREEN评分显著下降。
cs.CV / 135 / 2606.22158
Improving Reasoning in Vision-Language Models via Perception Verified Self-Training
通过感知验证自我训练提升视觉语言模型的推理能力
Abstract
Achieving human-like reasoning in Vision-Language Models (VLMs) remains a long-standing challenge. Recent approaches leverage Chain-of-Thought (CoT) rationales generated by human annotators or proprietary models to improve reasoning, which is costly and difficult to scale. Self-training offers a promising alternative by using models own outputs as supervision. However, existing methods often suffer from visual hallucinations -- where rationales describe non-existent visual content, and language shortcuts -- where predictions rely on textual priors rather than true visual grounding, as rationales are typically filtered only by answer correctness without verifying visual perception. To address this limitation, we propose a perception-verified self-training framework that enforces visually grounded reasoning. First, our method employs a CoT template (caption-reasoning-conclusion) that disentangles perception from reasoning, enabling independent verification of visual understanding. To compensate for the absence of ground-truth captions, we propose PerceptEval, an unsupervised method that evaluates caption quality based on its alignment with visual and textual elements present in the image. Using caption verification together with answer correctness, we partition the data into three subsets: easy (correct caption and conclusion), medium (correct caption but incorrect conclusion), and hard (incorrect caption). Building on this partitioning, we design a two-stage curriculum learning strategy. In Stage 1, the model is trained on easy examples and subsequently in Stage 2, medium samples are incorporated through a caption-guided reasoning enhancement procedure that regenerates reasoning conditioned on verified captions. Only regenerated samples with the correct conclusions are retained.
Chinese Translation
在视觉语言模型（VLMs）中实现类人推理仍然是一个长期存在的挑战。近期的方法利用人类标注者或专有模型生成的思维链（Chain-of-Thought, CoT）推理来改善推理能力，但这既昂贵又难以扩展。自我训练通过使用模型自身的输出作为监督提供了一种有前景的替代方案。然而，现有方法往往受到视觉幻觉的困扰——即推理描述了不存在的视觉内容，以及语言捷径——即预测依赖于文本先验而非真实的视觉基础，因为推理通常仅通过答案的正确性进行过滤，而未验证视觉感知。为了解决这一局限性，我们提出了一种感知验证自我训练框架，以强制执行视觉基础的推理。首先，我们的方法采用了一种思维链模板（caption-reasoning-conclusion），将感知与推理分离，使得视觉理解能够独立验证。为了弥补真实标题缺失的情况，我们提出了PerceptEval，一种无监督的方法，根据与图像中存在的视觉和文本元素的对齐程度来评估标题质量。通过结合标题验证和答案正确性，我们将数据划分为三个子集：简单（正确的标题和结论）、中等（正确的标题但结论不正确）和困难（标题不正确）。基于这种划分，我们设计了一种两阶段的课程学习策略。在第一阶段，模型在简单示例上进行训练，随后在第二阶段，通过一种基于标题的推理增强程序引入中等样本，该程序在经过验证的标题条件下重新生成推理。仅保留结论正确的重新生成样本。
cs.CV / 136 / 2606.22168
From Convolution to Transformer: A Comparative Study of U-Net Variants for Brain Tumor and Retinal Vessel Segmentation
从卷积到变换器：U-Net变体在脑肿瘤和视网膜血管分割中的比较研究
Abstract
Medical image segmentation plays an important role in computer aided diagnosis, treatment planning, and disease monitoring. U-Net has been widely used for biomedical image segmentation because of its encoder decoder structure and skip connections. However, conventional convolution based U-Net models may have limited ability to capture long range dependencies and global contextual information, which can affect performance in complex segmentation tasks. This paper presents a comparative study of five U-Net based architectures: U-Net 3D, Residual U-Net, Attention U-Net, UNETR, and Swin UNETR. The models are evaluated on two benchmark datasets: BraTS 2023 for brain tumor segmentation and DRIVE for retinal vessel segmentation. Experimental results show that Swin UNETR achieves the best overall performance, with Dice scores of 0.8965 on BraTS 2023 and 0.8078 on DRIVE. The results suggest that transformer based U-Net variants are effective for segmentation tasks requiring global contextual modeling, while residual learning remains useful for fine structure segmentation. This study provides practical insights into model selection for medical image segmentation across volumetric MRI and retinal imaging tasks.
Chinese Translation
医学图像分割在计算机辅助诊断、治疗规划和疾病监测中发挥着重要作用。由于其编码器-解码器结构和跳跃连接，U-Net被广泛应用于生物医学图像分割。然而，传统的基于卷积的U-Net模型在捕捉长距离依赖关系和全局上下文信息方面可能存在局限性，这会影响在复杂分割任务中的表现。本文对五种基于U-Net的架构进行了比较研究：U-Net 3D、残差U-Net、注意力U-Net、UNETR和Swin UNETR。这些模型在两个基准数据集上进行了评估：BraTS 2023用于脑肿瘤分割，DRIVE用于视网膜血管分割。实验结果表明，Swin UNETR在整体性能上表现最佳，在BraTS 2023上的Dice分数为0.8965，在DRIVE上的Dice分数为0.8078。结果表明，基于变换器的U-Net变体在需要全局上下文建模的分割任务中是有效的，而残差学习在细结构分割中仍然具有实用性。本研究为在体积MRI和视网膜成像任务中进行医学图像分割的模型选择提供了实用见解。
cs.CV / 137 / 2606.22182
Dual-Stream EEG Decoding for 3D Visual Perception
双通道脑电图解码用于三维视觉感知
Abstract
This paper explores a novel brain decoding model for 3D shape perception through a dual pathway architecture mirroring biological vision. Our bio-inspired approach implements separate decoding modules for object identity and spatial orientation, inspired by ventral and dorsal pathways, during continuous rotations. We employ circular regression for angle prediction and develop EEG-conditioned multiview diffusion for 3D reconstruction. Our approach successfully decodes both object identity and spatial orientation from EEG signals and enables 3D reconstruction from neural activity, with interpretability analyses revealing temporally structured involvement of ventral, dorsal, and motor-related channels rather than a static ventral dominance in supporting object and angle decoding.
Chinese Translation
本文探讨了一种新颖的脑解码模型，通过双通道架构模拟生物视觉，以实现三维形状感知。我们受生物启发的方法实现了针对物体身份和空间方向的独立解码模块，灵感来源于腹侧和背侧通路，适用于连续旋转过程。我们采用圆形回归进行角度预测，并开发了基于脑电图（EEG）条件的多视角扩散方法用于三维重建。我们的方法成功地从脑电信号中解码出物体身份和空间方向，并能够从神经活动中进行三维重建，解释性分析揭示了腹侧、背侧和与运动相关的通道在支持物体和角度解码时的时间结构性参与，而不是静态的腹侧主导。
cs.CV / 138 / 2606.22195
Resolving Multi-Target Association in OFDM-based ISAC via Vision-aided Multi-Modal Learning
通过视觉辅助的多模态学习解决基于OFDM的ISAC中的多目标关联问题
Abstract
Orthogonal frequency division multiplexing (OFDM)-based integrated sensing and communication (ISAC) systems commonly extract target parameters by peak-searching a delay-Doppler map (DDM) constructed from reflected pilots. In multi-target scenarios, this results in ambiguity: the DDM does not reveal which physical target produced which peak, and two targets within the same delay-Doppler resolution cell cannot be separated. We propose a vision-assisted OFDM-ISAC framework that resolves both limitations by fusing wireless and visual modalities. The transmitter encodes an onboard street-view image with deep joint source-channel coding (DeepJSCC) and transmits it over the same OFDM waveform used for sensing; the receiver reconstructs the image, runs a fine-tuned YOLOv5 detector and fuses the resulting per-target features (bounding-box coordinates and class labels) with the DDM and transmitter-receiver geometry through a learned multi-modal network. To stabilize training of the high dimensional delay and Doppler classifiers, we introduce a Kullback Leibler loss against triangular soft labels centered on the ground-truth bin. On a Blender-rendered vehicular testbed, the proposed framework achieves a 16 cm localization root mean square error (RMSE) and a 10.8 ns delay RMSE. An ablation study confirms that removing the visual modality causes a 60x degradation in localization. These results highlight the potential of vision to overcome the data-association and resolution limits of single-modality ISAC.
Chinese Translation
基于正交频分复用（OFDM）的集成感知与通信（ISAC）系统通常通过对从反射导频构建的延迟-多普勒图（DDM）进行峰值搜索来提取目标参数。在多目标场景中，这会导致歧义：DDM并未揭示哪个物理目标产生了哪个峰值，并且在同一延迟-多普勒分辨率单元内的两个目标无法分离。我们提出了一种视觉辅助的OFDM-ISAC框架，通过融合无线和视觉模态来解决这两个限制。发射机使用深度联合源信道编码（DeepJSCC）对机载街景图像进行编码，并通过用于感知的相同OFDM波形进行传输；接收机重建图像，运行经过微调的YOLOv5检测器，并通过学习的多模态网络将得到的每个目标特征（边界框坐标和类别标签）与DDM和发射机-接收机几何信息融合。为了稳定高维延迟和多普勒分类器的训练，我们引入了针对以真实值为中心的三角软标签的Kullback-Leibler损失。在Blender渲染的车辆测试平台上，所提出的框架实现了16厘米的定位均方根误差（RMSE）和10.8纳秒的延迟RMSE。消融研究确认，去除视觉模态会导致定位性能下降60倍。这些结果突显了视觉在克服单一模态ISAC的数据关联和分辨率限制方面的潜力。
cs.CV / 139 / 2606.22197
Multi4D: High-Fidelity Dynamic Gaussian Splatting via Multi-Level Competitive Allocation
Multi4D：通过多级竞争分配实现高保真动态高斯点云渲染
Abstract
Dynamic 3D Gaussian splatting faces a fundamental tension between motion consistency and visual fidelity. Deformation-based approaches preserve temporal correspondence but suffer from motion over-factorization, oversmoothing high-frequency dynamics. In contrast, 4D-primitive methods capture fine visual details yet incur temporal overparameterization, breaking object identity and leading to severe storage overhead. To resolve this, we introduce Multi4D, a framework for high-fidelity dynamic Gaussian Splatting based on multi-level competitive allocation. Instead of a monolithic representation, we distribute modeling capacity across three structured levels: static structure, persistent dynamic geometry, and transient appearance primitives. Through shared rasterization and residual-driven optimization, these levels dynamically compete to explain photometric error, enabling adaptive specialization without pre-assigned decomposition. This allocation preserves long-term motion consistency while capturing fine dynamic detail, achieving state-of-the-art rendering quality and real-time performance with significantly fewer dynamic primitives. Furthermore, because our representation explicitly tracks compact persistent Gaussians over time, semantic features can be embedded afterward, enabling Multi4D to achieve state-of-the-art 4D segmentation accuracy with an order-of-magnitude speedup. Project page: https://batfacewayne.github.io/Multi4D.io/
Chinese Translation
动态三维高斯点云渲染面临运动一致性与视觉保真度之间的根本矛盾。基于变形的方法能够保持时间对应性，但却遭遇运动过度分解的问题，导致高频动态的过度平滑。相对而言，四维原始方法能够捕捉细致的视觉细节，但却引入了时间上的过度参数化，破坏了物体的身份并导致严重的存储开销。为了解决这一问题，我们提出了Multi4D，一个基于多级竞争分配的高保真动态高斯点云渲染框架。我们并非采用单一的表示方式，而是将建模能力分布在三个结构化层次上：静态结构、持久动态几何体和瞬态外观原始体。通过共享光栅化和残差驱动的优化，这些层次动态竞争以解释光度误差，从而实现自适应专业化而无需预先分配的分解。这种分配方式在捕捉细致动态细节的同时保持了长期的运动一致性，达到了最先进的渲染质量和实时性能，并显著减少了动态原始体的数量。此外，由于我们的表示方式明确地跟踪紧凑的持久高斯分布，语义特征可以在之后嵌入，从而使Multi4D在实现最先进的四维分割精度的同时，速度提升了一个数量级。项目页面：https://batfacewayne.github.io/Multi4D.io/
cs.CV / 140 / 2606.22220
MultiMem: Measuring and Mitigating Memorization in Multi-Modal Contrastive Learninga
MultiMem：测量和缓解多模态对比学习中的记忆化
Abstract
Memorization in machine learning models enables high performance on rare in-distribution samples by capturing their atypical patterns. However, it also causes harmful retention of noise and outliers, degrading generalization. While memorization has been extensively studied in both supervised and self-supervised learning in the vision domain, it remains unexplored in multi-modal contrastive learning. We address this gap by introducing MultiMem, the first metric designed to quantify memorization in multi-modal contrastive learning. Through our systematic analysis, we demonstrate that cross-modal semantic misalignment has the strongest influence on memorization, with text being the dominant modality driving memorization, followed by video, image, and audio. We show that targeted augmentations applied across all modalities effectively reduce memorization as measured by our MultiMem metric and improve model performance. Overall, this work establishes the first framework for measuring and mitigating memorization in multi-modal contrastive learning, preventing harmful data retention and contributing to higher-performing models.
Chinese Translation
机器学习模型中的记忆化使得在稀有的分布内样本上实现高性能，能够捕捉其非典型模式。然而，这也导致了噪声和异常值的有害保留，降低了模型的泛化能力。尽管记忆化在视觉领域的监督学习和自监督学习中得到了广泛研究，但在多模态对比学习中仍未得到探索。我们通过引入MultiMem，填补了这一空白，MultiMem是第一个旨在量化多模态对比学习中记忆化的指标。通过我们的系统分析，我们证明了跨模态语义不对齐对记忆化的影响最为显著，其中文本是驱动记忆化的主导模态，其次是视频、图像和音频。我们展示了在所有模态上应用的针对性增强有效减少了记忆化，且通过我们的MultiMem指标进行测量，提升了模型性能。总体而言，本研究建立了第一个测量和缓解多模态对比学习中记忆化的框架，防止有害数据保留，并有助于提高模型性能。
cs.CV / 141 / 2606.22285
Efficient Document Tampering Localization with Multi-Level Discrepancy Features and Unified DCT-Quantization Embedding
基于多层差异特征和统一DCT-量化嵌入的高效文档篡改定位
Abstract
Localizing document tampering is extremely challenging, as manipulations are crafted to appear visually consistent and often leave only subtle traces that are nearly invisible to the human eye. In prior work, evaluation has been largely dominated by synthetic benchmarks that closely match the training distribution, and methods have shown steady progress under this setting. However, these gains often translate poorly to human-made forgeries and to cross-domain evaluation, where both the source documents and the tampering pipeline can change, leading to a distribution shift. In addition, since the introduction of the Frequency Perception Head for the discrete cosine transform (DCT) modality, it has become a standard choice, and subsequent work has largely focused on downstream modules and fusion strategies rather than revisiting the backbone itself. To help close this gap in cross-domain performance and improve the DCT backbone design, we propose \textbf{DiffNet}, a relatively simple yet effective RGB--DCT early-fusion architecture driven by two key design choices. First, to ensure that the decoder aggregates multi-scale inconsistency evidence rather than operating on raw, content-heavy activations, we apply a lightweight multi-level discrepancy transformation at the output of each backbone stage, replacing features with magnitude-only responses to learned zero-sum filters. Second, we design an efficient DCT-domain backbone that relies on a lightweight frequency-index-aware DCT--quantization joint embedding. Our approach achieves state-of-the-art performance on cross-domain and human-made document tampering localization, outperforming prior methods by around 30\%, with up to $7\times$ higher throughput than the previous best model.
Chinese Translation
文档篡改的定位极具挑战性，因为篡改通常被设计得在视觉上保持一致，并且往往只留下微妙的痕迹，这些痕迹几乎对人眼不可见。在之前的研究中，评估主要集中在与训练分布高度匹配的合成基准上，方法在这种设置下显示出稳步进展。然而，这些进展往往在人工伪造和跨域评估中表现不佳，因为源文档和篡改流程都可能发生变化，导致分布转移。此外，自从引入离散余弦变换（DCT）模态的频率感知头以来，它已成为标准选择，后续工作主要集中在下游模块和融合策略上，而不是重新审视骨干网络本身。为了缩小跨域性能的差距并改进DCT骨干设计，我们提出了 extbf{DiffNet}，这是一种相对简单但有效的RGB-DCT早期融合架构，基于两个关键设计选择。首先，为了确保解码器聚合多尺度不一致证据，而不是在原始、内容丰富的激活上操作，我们在每个骨干阶段的输出应用轻量级多层差异变换，用学习到的零和滤波器替换特征为仅包含幅度响应的输出。其次，我们设计了一个高效的DCT域骨干，依赖于轻量级频率索引感知的DCT-量化联合嵌入。我们的方法在跨域和人工文档篡改定位上实现了最先进的性能，超过了之前的方法约30，且吞吐量比之前的最佳模型高出最多$7 imes$。
cs.CV / 142 / 2606.22299
Towards Accurate and Robust Surveillance Roadside IVD via Trackletized Audio-Visual Reasoning
通过轨迹化音视频推理实现准确且稳健的监控路边怠速车辆检测
Abstract
Idling Vehicle Detection (IVD) seeks to determine, at the final frame of a video clip, whether any vehicle is idling, meaning the vehicle is stationary with its engine running, using synchronized video from a remote surveillance camera and multichannel audio captured by spatially distributed wireless microphones along the roadside. Prior full-image, clip-level fusion approaches tend to overfit scene background and full-frame context, produce unstable temporal decisions, and lack an explicit spatial prior to align vehicles with microphones, which makes them brittle under domain shift and data inefficient. Instead, we introduce TAVR-IVD, an audio-visual framework guided by multi-object tracking. Our method detects vehicles, links detections into tracklets, and classifies each vehicle by operating on its tracklet. This design raises the effective signal-to-noise ratio, stabilizes temporal decisions through tracklets, enforces an explicit spatial prior to align vehicles with microphones, and adapts across domains with limited calibration annotations while remaining detector agnostic and efficient. To evaluate deployment robustness, we further curate two evaluation extensions, AVIVD-LT and AVIVD-M, covering inter-day and cross-site shifts.
Chinese Translation
怠速车辆检测（IVD）旨在确定视频片段的最后一帧中是否有车辆处于怠速状态，即车辆静止且引擎运转。该方法利用来自远程监控摄像头的同步视频和沿路边分布的无线麦克风捕获的多通道音频。以往的全图像、片段级融合方法往往过拟合场景背景和全帧上下文，导致时间决策不稳定，并缺乏明确的空间先验来将车辆与麦克风对齐，这使得它们在领域转移时表现脆弱且数据效率低下。相反，我们提出了TAVR-IVD，这是一种由多目标跟踪引导的音视频框架。我们的方法检测车辆，将检测结果链接成轨迹，并通过操作轨迹对每辆车进行分类。这种设计提高了有效信噪比，通过轨迹稳定时间决策，强制实施明确的空间先验以将车辆与麦克风对齐，并在有限的校准注释下跨领域适应，同时保持检测器无关和高效。为了评估部署的稳健性，我们进一步策划了两个评估扩展，AVIVD-LT和AVIVD-M，涵盖了跨天和跨站点的变化。
cs.CV / 143 / 2606.22339
T-IMPACT: A Severity-Aware Benchmark for Contextual Image-Text Manipulation
T-IMPACT：一种关注严重性的上下文图像-文本操控基准
Abstract
Recent advances in vision-language models and generative editing systems have made it increasingly easy to produce persuasive multimodal misinformation by altering images, text, or both jointly. However, existing datasets focus mainly on authenticity, out-of-context mismatch, or manipulation type, and rarely capture how strongly an edit changes the likely interpretation of a post. We introduce T-IMPACT, a first-release severity-aware benchmark for manipulated news-style image-text pairs. T-IMPACT contains 98,786 examples spanning pristine, image-only, text-only, and joint manipulations, with a calibrated continuous severity signal, coarse low/medium/high labels, and supporting grounding metadata. Starting from a news image-text pair, the pipeline extracts semantic anchors, grounds them spatially, performs localized image edits and constrained caption rewrites, and calibrates contextual-impact scores using limited human ratings. In this release, the calibrated continuous score is the primary severity target, while the low/medium/high bands should be interpreted as coarse operating buckets rather than balanced classes. Experiments show that current models recover some authenticity signal, but severity prediction remains substantially harder and only weakly aligned with human judgment. T-IMPACT provides an initial benchmark for studying multimodal manipulation beyond binary real/fake classification toward graded contextual impact.
Chinese Translation
近年来，视觉-语言模型和生成编辑系统的进展使得通过修改图像、文本或二者联合来生成具有说服力的多模态虚假信息变得愈加容易。然而，现有数据集主要关注真实性、上下文不匹配或操控类型，鲜有捕捉编辑如何强烈改变帖子可能解读的程度。我们推出了T-IMPACT，这是首个关注严重性的操控新闻风格图像-文本对基准。T-IMPACT包含98,786个示例，涵盖原始、仅图像、仅文本和联合操控，配有经过校准的连续严重性信号、粗略的低/中/高标签以及支持的基础元数据。从新闻图像-文本对开始，管道提取语义锚点，进行空间定位，执行局部图像编辑和受限的标题重写，并使用有限的人类评分校准上下文影响分数。在此次发布中，经过校准的连续分数是主要的严重性目标，而低/中/高区间应被视为粗略的操作桶，而非平衡的类别。实验表明，当前模型恢复了一定的真实性信号，但严重性预测仍然显著更难，且与人类判断的对齐程度较弱。T-IMPACT为研究超越二元真实/虚假分类的多模态操控提供了初步基准，朝向分级的上下文影响。
cs.CV / 144 / 2606.22347
Customizing Video Portraits via Identity-ActionDecoupling
通过身份-动作解耦定制视频肖像
Abstract
Identity-Preserving Text-to-Video Generation (IPT2V) seeks to synthesize a temporally coherent video from a reference image and a textual description, while simultaneously preserving the subject's identity and allowing fine-grained control over facial dynamics. Although recent methods such as ID-Animator and ConsisID inject identity features only at inference time, they ignored the ID-irrelevant information contained in Facial embedding, leading to monotonous or inaccurate facial movements that poorly follow the prompt. We introduce Identity-Action Decoupling (IaD) framework as well as two loss function Identity Decoupling Loss and Text Alignment Loss to solve this problem. Without any subject-specific fine-tuning, IaD yields videos that (1) maintain cross-temporal identity consistency and (2) exhibit rich, controllable expressions and scene variations that closely match the input text.
Chinese Translation
身份保留文本到视频生成（IPT2V）旨在从参考图像和文本描述合成一个时间上连贯的视频，同时保持主体的身份，并允许对面部动态进行细致控制。尽管最近的方法如ID-Animator和ConsisID仅在推理时注入身份特征，但它们忽视了面部嵌入中包含的与身份无关的信息，导致面部动作单调或不准确，未能很好地遵循提示。我们提出了身份-动作解耦（IaD）框架，以及两个损失函数：身份解耦损失（Identity Decoupling Loss）和文本对齐损失（Text Alignment Loss），以解决这一问题。在没有任何特定于主体的微调的情况下，IaD生成的视频（1）保持跨时间的身份一致性，并且（2）展现出丰富且可控的表情和场景变化，与输入文本紧密匹配。
cs.CV / 145 / 2606.22353
Interest Entanglement: The Hidden Barrier to Blind Super-Resolution Optimization
兴趣纠缠：盲超分辨率优化的隐性障碍
Abstract
Fidelity and perceptual quality are two inherently competing and conflicting objectives in the image super-resolution (SR) task. Different loss functions focus on these objectives to varying extents. Regression losses enhance the model's fidelity but lack sufficient attention to high-frequency details, resulting in a loss of fine details. In contrast, perception losses improve the model's visual quality but may introduce undesirable artifacts. Balancing these two optimization goals can be viewed as a Multi-Objective Optimization problem. Existing methods are limited to cautiously adjusting weight parameters between these losses, overlooking the underlying Interest Entanglement problem. To address this problem, we explore the inherent frequency-domain conflict between the regression objective and the perceptual objective, and analyze the causes of Interest Entanglement in SR tasks. According to our findings, we propose the Shared-Feature-Representation based Super-Resolution framework (SFR), which decouples the learning process of different optimization objectives, allowing the model to explore a common optimization direction for both goals and achieve an effective balance between them. To better leverage shared features, we also proposed the InfoSqueeze module, which filters redundant information through a dimensionality reduction and expansion process, effectively transforming features into a consistent space. Quantitative and qualitative experiments across five representative datasets affirm the superiority of SFR.
Chinese Translation
在图像超分辨率（SR）任务中，保真度和感知质量是两个本质上相互竞争和冲突的目标。不同的损失函数在不同程度上关注这些目标。回归损失增强了模型的保真度，但对高频细节的关注不足，导致细节丢失。相反，感知损失改善了模型的视觉质量，但可能引入不必要的伪影。平衡这两个优化目标可以视为一个多目标优化问题。现有方法仅限于谨慎调整这些损失之间的权重参数，忽视了潜在的兴趣纠缠问题。为了解决这一问题，我们探讨了回归目标与感知目标之间固有的频域冲突，并分析了SR任务中兴趣纠缠的成因。根据我们的研究结果，我们提出了基于共享特征表示的超分辨率框架（SFR），该框架解耦了不同优化目标的学习过程，使模型能够探索两个目标的共同优化方向，并在它们之间实现有效平衡。为了更好地利用共享特征，我们还提出了InfoSqueeze模块，通过降维和扩展过程过滤冗余信息，有效地将特征转换为一致的空间。对五个代表性数据集的定量和定性实验验证了SFR的优越性。
cs.CV / 146 / 2606.22370
Towards Error-Free Long Video Generation
朝着无误长视频生成的方向
Abstract
Recent advances in video generation have made minute-level synthesis possible; however, generating long videos remains challenging due to error accumulation, attribute drift, and the limited availability of long video data. In this paper, we introduce an infinite-length video generation framework that focusing on addressing these issues and produces high-quality, dynamic, and identity-consistent single-shot long videos. We first finetune a diffusion model as a video extension model on large-scale short video data to autoregressively generate temporally coherent clips. Inspired by the success of large language models (LLMs), we adopt causal attention computation between clips to further finetune this model on long video data. In this way, the tokens in one clip (short video) are computed by bidirectional attention while tokens among clips are computed by unidirectional attention. This design leverages the strengths of modern diffusion models while preserving long-term context information, effectively mitigating error accumulation and attribute drift. To achieve memory efficiency during inference, we adopt a key-value (KV) caching mechanism to maintain a constant KV memory. Furthermore, we introduce truncation-rectified flow (T-RFlow) technique to further suppress error accumulation. Experimental results demonstrate the effectiveness of our method. Our framework establishes a new benchmark for realistic and coherent minute-level video synthesis.
Chinese Translation
近年来视频生成的进展使得分钟级合成成为可能；然而，由于错误积累、属性漂移以及长视频数据的有限可用性，生成长视频仍然面临挑战。本文提出了一种无限长度视频生成框架，旨在解决这些问题，并生成高质量、动态且身份一致的单次长视频。我们首先在大规模短视频数据上微调扩散模型作为视频扩展模型，以自回归方式生成时间一致的片段。受到大型语言模型（LLMs）成功的启发，我们在长视频数据上采用片段间的因果注意力计算来进一步微调该模型。通过这种方式，一个片段（短视频）中的标记通过双向注意力计算，而片段之间的标记则通过单向注意力计算。该设计利用了现代扩散模型的优势，同时保留了长期上下文信息，有效减轻了错误积累和属性漂移。为了在推理过程中实现内存效率，我们采用了键值（KV）缓存机制，以维持恒定的KV内存。此外，我们引入了截断修正流（T-RFlow）技术，以进一步抑制错误积累。实验结果证明了我们方法的有效性。我们的框架为现实且一致的分钟级视频合成建立了新的基准。
cs.CV / 147 / 2606.22378
Following the Flow: Advection-Consistent Modeling for Event-based Small Object Detection
跟随流动：基于对流一致性的事件驱动小物体检测建模
Abstract
Event cameras enable high-frequency visual perception with microsecond latency, offering advantages for dynamic scenes. However, event-based small object detection remains challenging due to sparse asynchronous measurements and weak object responses that are easily disrupted by noise. Limited spatial support causes small-object signals to lose temporal continuity, resulting in fragmented and unstable predictions. To address this issue, we propose a physics-guided advection-consistent modeling framework, termed PACT, which formulates event evolution as a motion-driven feature transport process. Instead of relying solely on local spatio-temporal aggregation, PACT propagates features along estimated velocity fields and enforces trajectory-level consistency through advection constraints. This design preserves weak event responses over time and prevents their degradation under complex background interference. Technically, PACT integrates motion-aware feature extraction with a differentiable advection-based transport operator, enabling coherent motion representation and effective noise suppression during temporal evolution. Extensive experiments on benchmark event-based datasets demonstrate that PACT consistently outperforms state-of-the-art methods, achieving improvements of 20.72\% in IoU and 15.03\% in accuracy while maintaining comparable computational efficiency. The code is publicly available at https://github.com/fulongcai/PACT.
Chinese Translation
事件相机能够以微秒延迟实现高频视觉感知，为动态场景提供了优势。然而，由于稀疏的异步测量和易受噪声干扰的弱物体响应，基于事件的小物体检测仍然面临挑战。有限的空间支持使得小物体信号失去时间连续性，导致预测结果碎片化和不稳定。为了解决这一问题，我们提出了一种物理引导的对流一致性建模框架，称为 PACT（Physics-guided Advection-Consistent Modeling Framework），该框架将事件演变表述为一个运动驱动的特征传输过程。PACT 并不单纯依赖于局部时空聚合，而是沿着估计的速度场传播特征，并通过对流约束强制执行轨迹级一致性。该设计在时间上保持弱事件响应，并防止其在复杂背景干扰下的退化。在技术上，PACT 将运动感知特征提取与可微分的基于对流的传输算子相结合，实现了连贯的运动表示和有效的噪声抑制。大量在基准事件数据集上的实验表明，PACT 在性能上始终优于最先进的方法，在 IoU 上提高了 20.72\%，在准确率上提高了 15.03\%，同时保持了相当的计算效率。代码可在 https://github.com/fulongcai/PACT 获取。
cs.CV / 148 / 2606.22383
Structured Hyperedge Adaptation for Parameter-Efficient Fine-Tuning of Vision Transformers
结构化超边适应用于视觉变换器的参数高效微调
Abstract
Parameter-efficient fine-tuning (PEFT) has become a practical solution for adapting large pretrained vision transformers (ViTs) to downstream tasks while updating only a small subset of parameters. However, existing adapter-based methods perform adaptation independently for each token, implicitly assuming that token refinements should be learned in isolation. This token-wise formulation overlooks the structured relationships among tokens that naturally arise in visual scenes, potentially leading to redundant updates and spatially inconsistent feature refinement. In this work, we revisit the design of parameter-efficient adapters and propose to perform adaptation in hyperedge space rather than token space. We introduce HyperAdapter, a hypergraph-based adapter architecture that enables structured, group-aware adaptation through soft token routing. HyperAdapter constructs a soft hypergraph over ViT tokens using prototype-based assignments, aggregates token features into latent hyperedge representations, applies lightweight bottleneck adaptation at the hyperedge level, and diffuses the resulting updates back to tokens via the hypergraph incidence structure. This design injects an explicit structural inductive bias into PEFT while preserving the modularity and efficiency of standard adapters. Extensive experiments across diverse visual benchmarks demonstrate that structured hyperedge adaptation consistently outperforms strong PEFT baselines under comparable parameter budgets, with particularly pronounced gains on tasks requiring structured reasoning. Our results suggest that the choice of adaptation space is a critical yet underexplored dimension in parameter-efficient transfer for ViTs.
Chinese Translation
参数高效微调（PEFT）已成为将大型预训练视觉变换器（ViTs）适应于下游任务的实用解决方案，同时仅更新少量参数。然而，现有的基于适配器的方法对每个标记独立进行适应，隐含地假设标记的细化应在孤立状态下学习。这种基于标记的表述忽视了在视觉场景中自然产生的标记之间的结构关系，可能导致冗余更新和空间上不一致的特征细化。在本研究中，我们重新审视了参数高效适配器的设计，并提出在超边空间而非标记空间进行适应。我们引入了HyperAdapter，这是一种基于超图的适配器架构，通过软标记路由实现结构化、群体感知的适应。HyperAdapter使用基于原型的分配在ViT标记上构建软超图，将标记特征聚合为潜在的超边表示，在超边级别应用轻量级瓶颈适应，并通过超图的关联结构将结果更新扩散回标记。该设计在保持标准适配器的模块性和效率的同时，将显式的结构归纳偏差注入PEFT。针对各种视觉基准的广泛实验表明，结构化超边适应在可比参数预算下始终优于强大的PEFT基线，尤其在需要结构化推理的任务上表现出显著的提升。我们的结果表明，适应空间的选择是参数高效转移中一个关键但尚未深入探索的维度。
cs.CV / 149 / 2606.22394
Curvature-Adaptive Consistency Flow Matching: Autonomous Trajectory Optimization via Reinforcement Learning
曲率自适应一致性流匹配：通过强化学习实现自主轨迹优化
Abstract
Consistency distillation has significantly accelerated the inference of diffusion models. In this work, we reveal an intriguing asymmetry: while Logit-Normal sampling priors are highly efficacious for standard iterative generation, consistency distillation exhibits a distinctly different difficulty profile (e.g., U-shaped). We identify that the primary optimization bottlenecks reside at the boundary stages (initialization or final refinement) rather than the intermediate steps. To address the limitations of static sampling in accommodating evolving learning requirements, we propose Curvature-Adaptive Consistency Flow Matching (CACFM). By formulating distillation as a dynamic decision process, CACFM employs a lightweight Reinforcement Learning agent to actively probe Probability Flow ODE trajectories, automatically constructing an efficiency-oriented curriculum that prioritizes critical regions without manual scheduling. Integrated with a novel Flow Distribution Matching Distillation (DMD) objective, our approach achieves new state-of-the-art results on large-scale models such as FLUX and SDXL. It effectively mitigates structural deformities and preserves high-frequency details in extreme few-step regimes, achieving unprecedented visual fidelity.
Chinese Translation
一致性蒸馏显著加速了扩散模型的推理。在本研究中，我们揭示了一个有趣的不对称性：虽然 Logit-Normal 采样先验对于标准迭代生成非常有效，但一致性蒸馏展现出明显不同的难度特征（例如，U 形）。我们识别出主要的优化瓶颈位于边界阶段（初始化或最终精细化），而非中间步骤。为了解决静态采样在适应不断变化的学习需求方面的局限性，我们提出了曲率自适应一致性流匹配（CACFM）。通过将蒸馏过程表述为动态决策过程，CACFM 采用轻量级的强化学习代理主动探测概率流常微分方程（ODE）轨迹，自动构建一个以效率为导向的课程，优先考虑关键区域而无需手动调度。结合新颖的流分布匹配蒸馏（DMD）目标，我们的方法在 FLUX 和 SDXL 等大规模模型上实现了新的最先进结果。它有效减轻了结构变形，并在极少步数的情况下保留了高频细节，达到了前所未有的视觉保真度。
cs.CV / 150 / 2606.22400
Multi-cancer detection using a computationally efficient CNN with transfer learning
基于计算高效的卷积神经网络（CNN）与迁移学习的多癌症检测
Abstract
This study introduces a computationally efficient convolutional neural network (CNN) architecture enhanced with transfer learning for multi-cancer detection using biomedical images. The proposed lightweight CNN model is designed to reduce computational complexity while maintaining high classification performance, making it suitable for deployment in resource-constrained environments. We evaluate this approach on three distinct tumor datasets comprising brain magnetic resonance imaging (MRI) and lung and kidney computed tomography (CT) scans. The model achieves test accuracy of 90.85 +- 2.22%, 98.64 +- 2.43% and 99.92 +- 0.08% for brain, lung, and kidney cancer classification, respectively, using 5-fold stratified cross-validation (CV). Transfer learning is employed by pretraining the model on one cancer type and fine-tuning it on the others, requiring only 20 additional epochs to achieve performance comparable to models trained from scratch. The fine-tuning process involves updating the classification part of the CNN and requires approximately 0.014 seconds per image per epoch using an NVIDIA GeForce GTX 960. Comparative evaluations show that the proposed model outperforms several state-of-the-art pretrained architectures, such as Xception, VGG16, VGG19, MobileNetV2 and DenseNet121. Overall, the model's effectiveness is evaluated across three types of cancer with distinct morphological characteristics, assessing its performance on both MRI and CT imaging modalities and demonstrating robust performance across diverse tasks and data types. These findings underscore the potential of streamlined deep learning (DL) frameworks in accelerating cancer diagnosis without sacrificing accuracy, especially in settings with limited computational resources.
Chinese Translation
本研究提出了一种基于迁移学习的计算高效卷积神经网络（CNN）架构，用于利用生物医学图像进行多癌症检测。所提出的轻量级CNN模型旨在降低计算复杂性，同时保持高分类性能，使其适合在资源受限的环境中部署。我们在三个不同的肿瘤数据集上评估了该方法，这些数据集包括脑部磁共振成像（MRI）以及肺部和肾脏的计算机断层扫描（CT）。该模型在脑癌、肺癌和肾癌分类中分别实现了90.85 ± 2.22%、98.64 ± 2.43%和99.92 ± 0.08%的测试准确率，采用了5折分层交叉验证（CV）。通过在一种癌症类型上预训练模型并在其他类型上进行微调，迁移学习得以应用，仅需20个额外的训练周期便可达到与从头训练的模型相当的性能。微调过程涉及更新CNN的分类部分，使用NVIDIA GeForce GTX 960时每张图像每个周期大约需要0.014秒。比较评估表明，所提出的模型优于多种最先进的预训练架构，如Xception、VGG16、VGG19、MobileNetV2和DenseNet121。总体而言，该模型在三种具有不同形态特征的癌症类型上进行了有效性评估，考察了其在MRI和CT成像模式下的表现，并在多样化的任务和数据类型中展示了稳健的性能。这些发现强调了精简深度学习（DL）框架在加速癌症诊断中的潜力，而不牺牲准确性，特别是在计算资源有限的环境中。
cs.CV / 151 / 2606.22409
Gold Points Sniper: Self-guided Visual Reasoning in VLM for Fine-grained Action Understanding
金点狙击手：用于细粒度动作理解的视觉语言模型自导向视觉推理
Abstract
Robots operating in everyday environments must understand fine-grained human actions, intentions, and contextual cues from broad views where people occupy only small regions, a capability unmet by current systems. While open-vocabulary action recognition methods remain limited to assigning predefined labels, and vision-language models (VLMs) face an inherent trade-off between informational richness and factual fidelity in their outputs, neither approach achieves the deep semantic interpretation required for reliable human-robot interaction. We propose Gold Points Sniper (GPS), a novel framework that empowers lightweight VLMs with self-guided multimodal reasoning capabilities for fine-grained human action understanding. Our approach comprises three key modules: Gold Points Extractor trains VLMs to identify critical action-relevant details, Selective Socratic Questioner validates and refines these details through selective self-questioning, and Semantic Entailment Evaluator quantitatively assesses factual consistency using semantic entailment classification. Extensive experiments on our curated instruction-tuning dataset based on the CAP benchmark demonstrate that GPS-enhanced lightweight VLMs achieve substantial performance improvements, with some models reaching performance comparable to proprietary GPT-4o while maintaining superior factual accuracy. Our work establishes a reliable foundation for fine-grained action understanding in domestic robotics, enabling robots to safely interpret human behavior through information-dense yet factually grounded descriptions. Source code, training configurations, annotation prompts, and dataset details are released at https://github.com/Haodi-Liu/GPS-Gold-Point-Sniper.
Chinese Translation
在日常环境中运行的机器人必须理解细粒度的人类动作、意图和上下文线索，而这些信息通常在广阔的视野中被人类占据的区域仅占很小一部分，这一能力目前的系统尚未满足。尽管开放词汇的动作识别方法仍然局限于分配预定义标签，而视觉语言模型（VLM）在输出中面临信息丰富性与事实准确性之间的固有权衡，但这两种方法都未能实现可靠的人机交互所需的深层语义解释。我们提出了金点狙击手（Gold Points Sniper, GPS），这是一个新颖的框架，赋予轻量级VLM自导向多模态推理能力，以实现细粒度的人类动作理解。我们的方法包括三个关键模块：金点提取器（Gold Points Extractor）训练VLM识别关键的与动作相关的细节，选择性苏格拉底提问者（Selective Socratic Questioner）通过选择性自我提问验证和完善这些细节，而语义蕴涵评估器（Semantic Entailment Evaluator）则通过语义蕴涵分类定量评估事实一致性。在我们基于CAP基准的精心策划的指令调优数据集上的广泛实验表明，GPS增强的轻量级VLM在性能上取得了显著提升，部分模型的性能可与专有的GPT-4o相媲美，同时保持更高的事实准确性。我们的工作为家庭机器人中的细粒度动作理解奠定了可靠的基础，使机器人能够通过信息密集但事实基础扎实的描述安全地解读人类行为。源代码、训练配置、注释提示和数据集详细信息已发布在 https://github.com/Haodi-Liu/GPS-Gold-Point-Sniper。
cs.CV / 152 / 2606.22416
Gen2Balance: Generative Balancing for Long-Tailed Video Action Recognition
Gen2Balance：用于长尾视频动作识别的生成平衡
Abstract
We address the problem of training on long-tailed data for video action recognition. We propose to augment the training set using a text-to-video generative model, conditioned on diverse text prompts grounded in action profiles and training exemplars. Our approach, called Gen2Balance, converts an imbalanced training set into a balanced combination of real and generated video clips. To effectively learn from such data, we employ a two-stage training strategy that mitigates domain shift and yields significant improvements. We evaluate on long-tailed versions of standard benchmarks: UCF-101 (UCF-LT) and a 100-class subset of Kinetics (K100-LT) selected to prioritise temporally challenging actions. Gen2Balance improves accuracy over the strongest baselines for long-tailed learning by 5.1% and 7.0% on the respective datasets. On rare actions from the RareAct dataset (e.g., cut keyboard), Gen2Balance improves accuracy by 31.9%, demonstrating effectiveness for scarce actions. By varying the amount of synthetic data added, we show that partial balancing already achieves 79% of the performance gains at 27% of the compute cost on K100-LT, highlighting the practical scalability of Gen2Balance.
Chinese Translation
我们针对视频动作识别中的长尾数据训练问题进行了研究。我们提出了一种使用文本到视频生成模型来增强训练集的方法，该模型以基于动作特征和训练示例的多样化文本提示为条件。我们的方法称为Gen2Balance，它将不平衡的训练集转换为真实视频片段和生成视频片段的平衡组合。为了有效地从这些数据中学习，我们采用了两阶段的训练策略，以减轻领域转移并显著提高性能。我们在标准基准的长尾版本上进行了评估：UCF-101（UCF-LT）和一个优先考虑时间上具有挑战性的动作的Kinetics的100类子集（K100-LT）。在这两个数据集上，Gen2Balance相较于长尾学习的最强基线提高了5.1%和7.0%的准确率。在RareAct数据集中稀有动作（例如，切换键盘）上，Gen2Balance的准确率提高了31.9%，证明了其在稀缺动作上的有效性。通过变化添加的合成数据量，我们展示了部分平衡在K100-LT上以27%的计算成本已实现79%的性能提升，突显了Gen2Balance的实际可扩展性。
cs.CV / 153 / 2606.22424
FlowDec: Temporal Conditional Flow Decorruptor for Robust Continuous Vision-Language Navigation
FlowDec：用于稳健连续视觉-语言导航的时间条件流去噪器
Abstract
Vision-and-Language Navigation in Continuous Environments (VLN-CE) requires agents to follow natural-language instructions in unseen scenes. While Large Models (LMs) have advanced VLN-CE, their performance remains severely degraded by real-world visual corruptions, a critical yet underexplored domain constraint. We introduce Temporal Conditional Flow Decorruptor (FlowDec), a novel image restoration framework tailored for LM-based VLN-CE. FlowDec integrates a hybrid temporal conditioning strategy to align the generative flow path with historical context and employs action-centroid guided filtering to dynamically assess and integrate outputs. Extensive experiments demonstrate that FlowDec outperforms state-of-the-art decorruption methods in both navigation accuracy and generation latency. Our approach establishes a robust, efficient paradigm for resilient embodied navigation in unpredictable real-world conditions.
Chinese Translation
在连续环境中的视觉与语言导航（VLN-CE）要求智能体在未见场景中遵循自然语言指令。尽管大型模型（LMs）推动了VLN-CE的发展，但它们的性能仍然受到现实世界视觉干扰的严重影响，这是一个关键但尚未充分探索的领域约束。我们提出了时间条件流去噪器（FlowDec），这是一种专为基于LM的VLN-CE量身定制的新型图像恢复框架。FlowDec整合了一种混合时间条件策略，以将生成流路径与历史上下文对齐，并采用动作中心引导过滤来动态评估和整合输出。大量实验表明，FlowDec在导航准确性和生成延迟方面均优于最先进的去噪方法。我们的方法为在不可预测的现实世界条件下建立稳健、高效的具身导航范式奠定了基础。
cs.CV / 154 / 2606.22437
MMGist: A Comprehensive Multimodal Benchmark for 2027
MMGist：2027年综合多模态基准
Abstract
We conduct a systematic study of 18 widely used vision-language benchmarks and identify three major issues: 1) many items do not rely on visual cues and therefore fail to effectively measure multimodal understanding; 2) many items are already close to performance saturation for current LVLMs, which limits their discriminative power; 3) a small number of anomalous items affect the reliability of evaluation results. To this end, we propose MMGist, a curated benchmark that covers seven capability dimensions and contains 7,262 items. MMGist is constructed through a three-stage pipeline, which sequentially combines text-ablation filtering, cross-model saturation filtering, and anomaly detection filtering. We conduct extensive experiments on 27 leading LVLMs and compare MMGist with the raw pool of 23,250 items. The results show that MMGist preserves model rankings with high fidelity, with Spearman $\rho = 0.98$, while reducing evaluation items by 69\% and improving cross-model discrimination by 78\%. Further results indicate that Visual Logic remains a systematic weakness of current LVLMs, while knowledge-intensive dimensions such as Expert Knowledge dimensions remain important factors for distinguishing closed-source models from open-source models. These findings suggest that high-quality evaluation should prioritize visual dependency, discriminative power, and reliability, rather than simply pursuing benchmark scale.
Chinese Translation
我们对18个广泛使用的视觉-语言基准进行了系统研究，并识别出三个主要问题：1）许多项目不依赖于视觉线索，因此无法有效测量多模态理解；2）许多项目的表现已接近当前大型视觉语言模型（LVLMs）的性能饱和，这限制了它们的区分能力；3）少量异常项目影响了评估结果的可靠性。为此，我们提出了MMGist，这是一个经过精心策划的基准，涵盖七个能力维度，并包含7262个项目。MMGist通过一个三阶段的流程构建，该流程依次结合了文本消融过滤、跨模型饱和过滤和异常检测过滤。我们在27个领先的LVLMs上进行了广泛的实验，并将MMGist与原始的23250个项目池进行了比较。结果表明，MMGist在保持模型排名的高保真度方面表现出色，Spearman相关系数为$
ho = 0.98$，同时将评估项目减少了69 ext{%}，并提高了跨模型的区分能力78 ext{%}。进一步的结果表明，视觉逻辑仍然是当前LVLMs的系统性弱点，而知识密集型维度（如专家知识维度）仍然是区分闭源模型与开源模型的重要因素。这些发现表明，高质量的评估应优先考虑视觉依赖性、区分能力和可靠性，而不仅仅是追求基准规模。
cs.CV / 155 / 2606.22439
Curvature-aware 3D length estimation of greenhouse cucumbers using RGB-D imaging and cubic spline arc-length integration
基于RGB-D成像和立方样条弧长积分的温室黄瓜曲率感知三维长度估计
Abstract
Commercial greenhouse cucumber production is graded by fruit length, which drives harvest scheduling, labour allocation, and logistics. Manual measurement with thread or caliper is accurate but infeasible at commercial scale. This paper presents CucumberVision, a non-contact length estimation framework using an Intel RealSense D435 RGB-D camera. A YOLO26n instance segmentation model locates cucumbers, and SAM (ViT-B backbone) refines each detection to a pixel-precise mask. Five methods are evaluated under matched conditions: (M1) a dominant-axis skeleton scan-line baseline; (M2) PCA on the bounding-box depth point cloud; (M3) SAM mask with medial-axis skeletonisation; (M4) a hybrid keypoint-guided approach using a YOLO26-pose model predicting five anatomical landmarks (KP0--KP4) with piecewise 3D arc-length; and (M5) a novel medial arc spline method fitting a cubic spline through the 3D medial axis of the SAM mask and computing arc length by trapezoidal integration -- the first such application to elongated vegetable measurement. All methods share five-frame burst depth averaging, colour-stream intrinsic alignment, and adaptive method selection with cascading fallbacks ensuring 100% coverage. A benchmark of 48 captures across seven cucumbers in three size categories (small ~8 cm, medium ~13 cm, large ~25 cm) with thread-based ground truth establishes a significant accuracy hierarchy: M1 (MAPE 9.68%) > M2 (5.31%) > M4 (5.51%) > M3 (5.82%) > M5 (4.13%). M5 significantly outperforms all competitors at Bonferroni-corrected alpha=0.0125. A secondary contribution is identifying a 12--18% length underestimation caused by using depth-stream rather than colour-stream intrinsics after rs.align(rs.stream.color) -- an under-reported error source. The complete system is released open source and runs in real time on a single consumer-grade GPU.
Chinese Translation
商业温室黄瓜生产通过果实长度进行分级，这影响收获调度、劳动力分配和物流。使用线或卡尺的手动测量虽然准确，但在商业规模上不可行。本文提出了CucumberVision，一个使用Intel RealSense D435 RGB-D相机的非接触式长度估计框架。YOLO26n实例分割模型定位黄瓜，SAM（ViT-B骨干网络）将每个检测结果精确细化为像素级掩膜。在匹配条件下评估了五种方法：（M1）主轴骨架扫描线基线；（M2）对边界框深度点云进行主成分分析（PCA）；（M3）使用中轴骨架化的SAM掩膜；（M4）一种混合关键点引导的方法，使用YOLO26姿态模型预测五个解剖标志点（KP0--KP4）并进行分段三维弧长计算；（M5）一种新颖的中轴样条方法，通过SAM掩膜的三维中轴拟合立方样条并通过梯形积分计算弧长——这是对细长蔬菜测量的首次应用。所有方法共享五帧突发深度平均、色流内在对齐和自适应方法选择，具有级联回退机制以确保100%覆盖。通过基于线的真实值对48次捕捉进行基准测试，涵盖七根黄瓜的三种尺寸类别（小约8厘米，中约13厘米，大约25厘米），建立了显著的准确性层次：M1（MAPE 9.68%）> M2（5.31%）> M4（5.51%）> M3（5.82%）> M5（4.13%）。M5在Bonferroni校正的alpha=0.0125下显著优于所有竞争者。另一个贡献是识别出由于使用深度流而非色流内在参数导致的12-18%的长度低估——这是一个被低估的错误来源。完整系统已开源，并可在单个消费级GPU上实时运行。
cs.CV / 156 / 2606.22445
DreamUV: Unwrap Artist-like UV by End-to-End Flow Matching
DreamUV：通过端到端流匹配实现艺术家风格的UV展开
Abstract
UV parameterization is a fundamental step in 3D content creation, yet producing production-ready UV layouts remains challenging due to the gap between geometric distortion objectives and the stylistic preferences of professional artists. While classical methods optimize handcrafted energy functions, artist-authored UVs exhibit structural patterns such as straightened seams, axis-aligned islands, and flexible interior deformation, properties that are difficult to explicitly formulate. In this work, we present DreamUV, an end-to-end learning framework that formulates UV unwrapping as a generative Flow Matching problem. Rather than predicting a single optimal parameterization, DreamUV learns a mesh-conditioned transport process that maps noise samples to a distribution of artist-like UV layouts. To reflect real-world authoring practices, we introduce a boundary-aware training strategy that prioritizes seam geometry, and a Model-in-the-Loop Finetuning(MITL) scheme that explicitly accounts for discretization errors during sampling and stabilizes transport dynamics under heterogeneous supervision. We evaluate DreamUV on a large-scale dataset of professionally authored UV layouts. Experiments demonstrate that our method produces significantly straighter boundaries and tighter axis-aligned islands than both classical and learning-based baselines, while maintaining competitive distortion metrics. Qualitative results and a user study with professional artists further confirm that DreamUV generates UV layouts that are not only valid, but aligned with practical production requirements.
Chinese Translation
UV参数化是3D内容创作中的一个基本步骤，但由于几何失真目标与专业艺术家的风格偏好之间的差距，生成适合生产的UV布局仍然具有挑战性。尽管经典方法优化手工制作的能量函数，但艺术家创作的UV展现出诸如拉直的接缝、轴对齐的岛屿和灵活的内部变形等结构模式，这些特性难以明确表述。在本研究中，我们提出了DreamUV，一个将UV展开形式化为生成流匹配问题的端到端学习框架。DreamUV并不是预测单一的最优参数化，而是学习一种网格条件的传输过程，将噪声样本映射到艺术家风格的UV布局分布。为了反映现实世界的创作实践，我们引入了一种边界感知的训练策略，优先考虑接缝几何，并采用了模型循环微调（Model-in-the-Loop Finetuning, MITL）方案，明确考虑采样过程中的离散化误差，并在异构监督下稳定传输动态。我们在一个大规模的专业创作UV布局数据集上评估了DreamUV。实验表明，我们的方法生成的边界显著更直，轴对齐的岛屿更紧凑，且在保持竞争性失真指标的同时，优于经典和基于学习的基线。定性结果和与专业艺术家的用户研究进一步确认，DreamUV生成的UV布局不仅有效，而且符合实际生产要求。
cs.CV / 157 / 2606.22476
CVSBench: A Comprehensive Benchmark for Cross-view Spatial Reasoning and Dreaming
CVSBench：跨视角空间推理与梦境的综合基准
Abstract
Humans can effortlessly reason about scenes across different viewpoints, yet it remains unclear whether Vision-Language Models (VLMs) possess similar cross-view spatial abilities. Satellite-street scene pairs, with their complex contexts and extreme viewpoint variations, provide an ideal testbed. Motivated by this, we introduce CVSBench, a large-scale benchmark for evaluating cross-view spatial reasoning through satellite-street pairs. This benchmark supports multiple tasks, including cross-view VQA, cross-view grounding, and viewpoint identification. CVSBench comprises 3,297 cross-view image groups with 9,468 object-level annotations and 40,679 question-answer (QA) pairs, enabling systematic and controlled evaluation of cross-view spatial reasoning. Extensive evaluations reveal that advanced VLMs struggle to maintain object-level and layout consistency under drastic viewpoint changes. To bridge this gap towards human-like spatial cognition, we investigate two categories of approaches: spatially grounded reasoning and the incorporation of cognitive map inputs. Our findings demonstrate that language-only reasoning yields marginal improvements, while incorporating visual spatial imagination via a 3D scene imagination pipeline substantially improves cross-view reasoning. These results highlight the necessity of explicit visual-spatial representations for robust spatial cognition in VLMs. Our data and code are released at https://huggingface.co/datasets/zlyzlyzly/CVSBench.
Chinese Translation
人类能够轻松地在不同视角下推理场景，但目前尚不清楚视觉-语言模型（VLMs）是否具备类似的跨视角空间能力。卫星-街景对由于其复杂的上下文和极端的视角变化，提供了理想的测试平台。基于此，我们引入了CVSBench，这是一个用于通过卫星-街景对评估跨视角空间推理的大规模基准。该基准支持多种任务，包括跨视角视觉问答（VQA）、跨视角定位和视角识别。CVSBench包含3,297个跨视角图像组，具有9,468个对象级注释和40,679个问答（QA）对，能够系统地、受控地评估跨视角空间推理。广泛的评估表明，先进的VLM在剧烈的视角变化下难以保持对象级和布局的一致性。为了缩小与人类空间认知的差距，我们研究了两类方法：空间基础推理和认知地图输入的结合。我们的研究结果表明，仅使用语言推理的改进幅度有限，而通过3D场景想象管道结合视觉空间想象显著提高了跨视角推理。这些结果强调了在VLM中进行稳健空间认知所需的明确视觉-空间表征。我们的数据和代码已发布在 https://huggingface.co/datasets/zlyzlyzly/CVSBench。
cs.CV / 158 / 2606.22477
Physically-guided Image Generation for Multi-Projection Mapping
基于物理引导的多投影映射图像生成
Abstract
Projection Mapping (PM) enables seamless superimposition of digital content onto real-world 3D objects, serving as a fundamental technique for immersive visualization, digital twins, and interactive art. Although text-to-image diffusion models have greatly facilitated customized content creation, directly integrating them into practical PM pipelines remains challenging due to the mismatch between idealized 2D generation and physical constraints. To bridge this gap, this paper formalizes two application-level generative paradigms: the cooperative paradigm (harmonizing generated semantics with physical attributes) and the adversarial paradigm (eliminating surface interference via radiometric compensation). Based on this, we propose ConPhyG, a unified controllable physically-guided generative multi-projection mapping framework that enables creators to interactively adjust physical constraints and flexibly switch generative paradigms. In cooperative mode, multi-dimensional physical priors (per-pixel gamut, depth, and edges) are injected into the diffusion process. In adversarial mode, the framework releases the generative potential and applies bounded numerical optimization for multi-projector radiometric compensation. It allows users to dynamically switch constraints to balance artistic freedom with physical feasibility. Furthermore, we extend ConPhyG to 360-degree multi-view consistent PM using a sequential generation strategy. Quantitative and qualitative evaluations on a real-world four-projector setup demonstrate that ConPhyG significantly outperforms state-of-the-art methods in geometric alignment, gamut utilization, and semantic fidelity.
Chinese Translation
投影映射（PM）使得数字内容能够无缝叠加在现实世界的三维物体上，成为沉浸式可视化、数字双胞胎和互动艺术的基础技术。尽管文本到图像的扩散模型极大地促进了定制内容的创作，但由于理想化的二维生成与物理约束之间的不匹配，直接将其整合到实际的PM流程中仍然具有挑战性。为了解决这一问题，本文正式提出了两种应用层面的生成范式：合作范式（将生成的语义与物理属性协调）和对抗范式（通过辐射补偿消除表面干扰）。基于此，我们提出了ConPhyG，一个统一的可控物理引导生成多投影映射框架，使创作者能够互动地调整物理约束并灵活切换生成范式。在合作模式下，多维物理先验（每像素色域、深度和边缘）被注入到扩散过程中。在对抗模式下，该框架释放生成潜力，并应用有界数值优化进行多投影仪的辐射补偿。它允许用户动态切换约束，以平衡艺术自由与物理可行性。此外，我们将ConPhyG扩展到360度多视图一致的PM，采用顺序生成策略。在一个真实的四投影仪设置上的定量和定性评估表明，ConPhyG在几何对齐、色域利用和语义保真度方面显著优于现有的最先进方法。
cs.CV / 159 / 2606.22486
Human and AI collaboration for pulmonary nodule segmentation
人类与人工智能在肺结节分割中的协作
Abstract
Medical expert annotators are scarce, and blind reliance on artificial intelligence (AI) can be misleading, motivating approaches in which humans, particularly junior medical trainees or even non-medical personnel, collaborate with AI to achieve robust medical segmentation. Although the Segment Anything Model (SAM) shows promise for general-purpose image segmentation, its performance in human-AI collaboration for specialized medical tasks has not been thoroughly evaluated. Here we present Hi-Seg, a human-in-the-loop segmentation framework for pulmonary nodules built on SAM. Humans iteratively refine prompts through trial-and-error learning and semantic reasoning, progressively guiding SAM toward higher-quality masks. Using chest CT scans from 1,179 patients across 12 centers, we conducted the first large-scale external validation of collaborative human-SAM segmentation. Across all annotator groups, Hi-Seg achieved a mean Dice score of almost 85%, outperforming five state-of-the-art deep learning models by 10-22% and 13 SAM variants by 1-29%. Hi-Seg improved segmentation accuracy while reducing annotation time for medical annotators, and briefly trained non-medical annotators achieved performance comparable to that of the junior medical student. These findings suggest that human-in-the-loop segmentation can reduce clinician workload, enable scalable crowdsourced annotation, and transform clinical workflows by facilitating the safe and efficient integration of foundation models into routine clinical practice.
Chinese Translation
医学专家注释者稀缺，盲目依赖人工智能（AI）可能会导致误导，这促使了人类，特别是初级医学培训生甚至非医学人员与AI合作以实现稳健的医学分割的方法。尽管Segment Anything Model（SAM）在通用图像分割方面显示出前景，但其在人类与AI协作进行专业医学任务中的表现尚未得到充分评估。在此，我们提出了Hi-Seg，一个基于SAM的人机协作肺结节分割框架。人类通过试错学习和语义推理迭代地优化提示，逐步引导SAM生成更高质量的分割掩膜。我们使用来自12个中心的1,179名患者的胸部CT扫描，进行了首次大规模的协作人类-SAM分割的外部验证。在所有注释者组中，Hi-Seg的平均Dice分数接近85%，比五个最先进的深度学习模型高出10-22%，比13个SAM变体高出1-29%。Hi-Seg提高了分割准确性，同时减少了医学注释者的注释时间，经过简短培训的非医学注释者的表现与初级医学学生相当。这些发现表明，人机协作分割可以减轻临床医生的工作负担，促进可扩展的众包注释，并通过促进基础模型在常规临床实践中的安全高效整合，转变临床工作流程。
cs.CV / 160 / 2606.22487
FetSelect: Task-Specific Architectures and Self-Supervised Learning for Automated Fetal Ultrasound Frame Selection
FetSelect：用于自动化胎儿超声图像选择的任务特定架构和自监督学习
Abstract
Automated frame selection for fetal biometry remains under addressed, with most prior work targeting generic quality assessment or downstream measurement pipelines that assume suitable frames are available. We introduce FetSelect, a task-specific framework that pairs a frozen vision foundation backbone with a hybrid multi-head design: a Task-Gated classification head and a Detection-derived quality head combined via learned fusion. We curate 6,486 expert-labeled frames across four targets: Crown-Rump Length (CRL), Nuchal Translucency (NT), Nasal Bone (NB), and Scalebar, and adapt the backbone with BYOL pretraining on 19,019 unlabeled images. On a held-out test set (974 frames), FetSelect achieves mean AUROC 0.956 and mean correlation 0.818 with expert quality annotations. Ablations confirm that hybrid fusion surpasses single-head variants, and ultrasound-specific self-supervision yields consistent gains. Evaluation on external clinical videos and 509 external CRL images demonstrates task-specific discrimination.
Chinese Translation
胎儿生物测量的自动化图像选择仍然未得到充分解决，以往的大多数研究主要针对通用质量评估或假设适合图像可用的下游测量流程。我们提出了FetSelect，一个任务特定的框架，将冻结的视觉基础骨干与混合多头设计相结合：一个任务门控分类头和一个基于检测的质量头通过学习融合进行组合。我们整理了6,486个专家标注的图像，涵盖四个目标：头臀长（Crown-Rump Length, CRL）、颈部透明带（Nuchal Translucency, NT）、鼻骨（Nasal Bone, NB）和比例尺，并在19,019张未标记图像上对骨干进行了BYOL预训练。在一个保留的测试集（974帧）上，FetSelect实现了平均AUROC为0.956和与专家质量标注的平均相关性为0.818。消融实验确认混合融合优于单头变体，而超声特定的自监督学习则带来了持续的提升。在外部临床视频和509张外部CRL图像上的评估展示了任务特定的区分能力。
cs.CV / 161 / 2606.22497
Benchmarking Vision-Language Models for Microscopic Plant Image Understanding
微观植物图像理解的视觉-语言模型基准测试
Abstract
Microscopic imaging provides essential visual evidence for studying plant biology and pathology at the cellular and subcellular levels. However, existing benchmarks on vision-language models primarily focus on macroscopic plant imagery, while the microscopic domain remains underexplored. To address this gap, we present PlantMicro, a comprehensive benchmark for evaluating vision-language models (VLMs) in microscopic plant imagery. PlantMicro integrates more than 5,000 images collected across diverse hosts, biological domains, and imaging modalities. Building on this diversity, we design a set of complementary tasks that capture different facets of microscopic image understanding. To support these tasks, we construct over 9,000 VQA pairs that systematically evaluate the capabilities of VLMs. Experiments on PlantMicro show that current VLMs struggle with fine-grained recognition and biologically grounded reasoning. For example, GPT-5 achieves 34.93% accuracy on the pathogen classification task, which is only modestly above the random-guessing baseline. The results highlight a significant gap in current VLMs' ability to comprehend plant microscopic images. PlantMicro provides a standardized foundation for advancing VLMs toward reliable and comprehensive microscopy-level plant understanding.
Chinese Translation
微观成像为研究植物生物学和病理学在细胞及亚细胞水平提供了重要的视觉证据。然而，现有的视觉-语言模型基准主要集中在宏观植物图像上，而微观领域仍然未得到充分探索。为了解决这一空白，我们提出了PlantMicro，这是一个用于评估微观植物图像中视觉-语言模型（VLMs）的综合基准。PlantMicro整合了超过5000张来自不同宿主、生物领域和成像方式的图像。在此多样性的基础上，我们设计了一系列互补任务，以捕捉微观图像理解的不同方面。为了支持这些任务，我们构建了超过9000对视觉问答（VQA）对，系统性地评估VLMs的能力。在PlantMicro上的实验表明，当前的VLMs在细粒度识别和生物学基础推理方面存在困难。例如，GPT-5在病原体分类任务中的准确率为34.93%，仅略高于随机猜测的基线。结果突显了当前VLMs在理解植物微观图像方面的显著差距。PlantMicro为推动VLMs向可靠和全面的显微镜级植物理解提供了标准化基础。
cs.CV / 162 / 2606.22515
Biological Sex Determination in Cadavers Using Deep Learning Algorithms from Computed Tomography Images of Pelvis and Skull
基于深度学习算法通过计算机断层扫描图像确定尸体的生物性别
Abstract
Sexual identification of decomposed cadavers challenges traditional methods dependent on visual anthropological analysis. This study evaluates state-of-the-art deep learning (including YOLO26, YOLO11, ConvNeXt-Tiny, EfficientNetV2, ViT-B16, VGG16, and ResNet50) with transfer learning to automatically determine biological sex from forensic computed tomography (CT) scans. We analyzed 141 autopsied cadavers from the Forensic Medical Institute of Goi\^ania-GO, including a broad age range and varying conditions of preservation. The three-dimensional reconstructions of the pelvis and skull were converted into standardized two-dimensional profile projections, contributing to the study of this new technical approach. Data augmentation techniques compensated for sample limitations. Two scenarios were validated: binary and quaternary classification (one class per sex vs. one class per anatomical region of each sex). The best-performing model achieved highly consistent results on the pelvis region and still satisfactory performance on the skull region, reaching an overall patient-level accuracy of 95.65%, recall of 92.86%, F1- score of 94.36%, and precision of 97.22%, maintaining consistent performance across the evaluated cases, including those with trauma-related artifacts. Results indicate the technical feasibility of the methodology, demonstrating that deep learning models can provide objective, high-speed skeletal analysis. Since the study was conducted using data from a single institution and a single computed tomography scanner, further validation across multiple centers and scanners is required to assess the generalizability of the proposed approach
Chinese Translation
腐败尸体的性别识别对依赖视觉人类学分析的传统方法提出了挑战。本研究评估了最先进的深度学习技术（包括YOLO26、YOLO11、ConvNeXt-Tiny、EfficientNetV2、ViT-B16、VGG16和ResNet50）结合迁移学习，以自动从法医计算机断层扫描（CT）中确定生物性别。我们分析了来自戈亚尼亚法医医学研究所的141具尸体，涵盖了广泛的年龄范围和不同的保存状态。骨盆和头骨的三维重建被转换为标准化的二维轮廓投影，为这一新技术方法的研究做出了贡献。数据增强技术弥补了样本的局限性。验证了两种场景：二元分类和四元分类（每个性别一个类别与每个性别的每个解剖区域一个类别）。表现最佳的模型在骨盆区域取得了高度一致的结果，并在头骨区域仍表现出令人满意的性能，整体患者级别的准确率达到95.65%，召回率为92.86%，F1分数为94.36%，精确率为97.22%，在评估的案例中保持了一致的表现，包括那些具有创伤相关伪影的案例。结果表明该方法的技术可行性，证明深度学习模型能够提供客观、高速的骨骼分析。由于本研究使用的数据来自单一机构和单一计算机断层扫描仪，因此需要在多个中心和扫描仪上进一步验证，以评估所提方法的普适性。
cs.CV / 163 / 2606.22525
Projection-Volume Fidelity Divergence: Diagnosing and Controlling Optimization Drift in Sparse-View 3D Gaussian Tomography
投影-体积保真度发散：诊断和控制稀视图3D高斯断层成像中的优化漂移
Abstract
Sparse-view computed tomography is a severely ill-posed inverse problem, where recent 3D Gaussian Splatting methods offer an efficient explicit representation for tomographic reconstruction. However, we find that projection-domain optimization can be misleading in this setting: the rendered projections may continue to improve while the reconstructed volume deteriorates. We identify this failure mode as Projection-Volume Fidelity Divergence (PVFD), a representation-level optimization drift caused by anisotropic Gaussian deformation and view-specific primitive co-adaptation under sparse Radon constraints. To characterize this behavior, we introduce geometry- and volume-level diagnostics that measure needle-like Gaussian degeneration and the stability of the voxelized density field. Based on these observations, we propose LADES, a ground-truth-free optimization controller for sparse-view Gaussian tomography. LADES combines Linearly Annealed Dropout, which applies strong stochastic masking in early training to disrupt premature primitive co-adaptation and gradually restores full capacity for structural consolidation, with Structure-Aware Early Stopping, which terminates densification according to the saturation of Gaussian population growth rather than validation PSNR. Experiments on sparse-view CT reconstruction show that LADES improves volumetric fidelity, suppresses structural degeneration, and substantially reduces training time while maintaining competitive projection accuracy. These results suggest that robust Gaussian-based tomography requires monitoring and controlling volumetric structure, rather than optimizing projection fit alone.
Chinese Translation
稀视图计算机断层成像是一个严重病态的逆问题，最近的3D高斯点云方法为断层重建提供了一种高效的显式表示。然而，我们发现，在这种情况下，投影域优化可能会产生误导：渲染的投影可能继续改善，而重建的体积却在恶化。我们将这种失效模式称为投影-体积保真度发散（Projection-Volume Fidelity Divergence, PVFD），它是由各向异性高斯变形和在稀拉东约束下视图特定原始体共同适应引起的表示级优化漂移。为了表征这种行为，我们引入了几何和体积级诊断，测量针状高斯退化和体素化密度场的稳定性。基于这些观察，我们提出了LADES，一个无真实值的稀视图高斯断层成像优化控制器。LADES结合了线性退火丢弃（Linearly Annealed Dropout），在早期训练中应用强随机掩蔽以打断过早的原始体共同适应，并逐渐恢复结构巩固的全部能力，以及结构感知早停（Structure-Aware Early Stopping），根据高斯种群增长的饱和度而非验证PSNR终止密度化。在稀视图CT重建实验中，LADES提高了体积保真度，抑制了结构退化，并显著减少了训练时间，同时保持了竞争性的投影精度。这些结果表明，稳健的基于高斯的断层成像需要监测和控制体积结构，而不仅仅是优化投影拟合。
cs.CV / 164 / 2606.22527
Trajectory Forcing: Structure-First Generation with Controllable Semantic Trajectories
轨迹强制：结构优先的生成与可控语义轨迹
Abstract
Diffusion and flow-based generative models produce strong images, yet their controllability remains largely endpoint-centric: users specify conditions and receive final outputs, while the intermediate generative dynamics remain hidden. Recent methods have begun to exploit generation order and process decomposition to improve sample quality, but still treat intermediate states as internal computation rather than objects for interaction. We propose Trajectory Forcing (TF), a trajectory-centric framework that makes the generation path explicit, semantic, and editable. TF organizes synthesis as a sequence of semantically structured stages, progressing from global layout to object-, part-, and detail-level representations. Each stage produces a decodable latent state that can be inspected, evaluated, and locally edited before the next stage begins. To instantiate this path, we derive coarse-to-fine teacher hierarchies by clustering pretrained visual representations such as DINOv2, and train a hierarchy-conditioned one-step flow-matching model at each level. We further introduce trajectory-aware metrics that measure structural consistency and local controllability beyond endpoint quality metrics such as FID. Experiments show that TF achieves competitive sample quality while exposing coherent intermediate states and supporting localized edits across semantic levels. By shifting the focus from final images to the generative path itself, TF opens a route toward controllable, trajectory-aware image synthesis.
Chinese Translation
扩散和基于流的生成模型能够生成高质量图像，但它们的可控性仍然主要集中于最终结果：用户指定条件并接收最终输出，而中间生成动态则保持隐蔽。近期的方法开始利用生成顺序和过程分解来提高样本质量，但仍将中间状态视为内部计算，而非可交互的对象。我们提出了轨迹强制（Trajectory Forcing, TF），一种以轨迹为中心的框架，使生成路径显式、语义化且可编辑。TF将合成过程组织为一系列语义结构化的阶段，从全局布局逐步推进到对象、部分和细节级别的表示。每个阶段生成一个可解码的潜在状态，可以在下一个阶段开始之前进行检查、评估和局部编辑。为了实现这一路径，我们通过聚类预训练的视觉表示（如DINOv2）推导出粗到细的教师层次结构，并在每个层次训练一个层次条件的一步流匹配模型。我们进一步引入轨迹感知指标，测量结构一致性和局部可控性，超越了如FID等最终结果质量指标。实验表明，TF在揭示一致的中间状态和支持跨语义层次的局部编辑的同时，达到了具有竞争力的样本质量。通过将焦点从最终图像转向生成路径本身，TF为可控的、轨迹感知的图像合成开辟了一条新路。
cs.CV / 165 / 2606.22537
NegAS: Negative Label Guided Attention and Scoring for Out-of-Distribution Object Detection with Vision-Language Models
NegAS：基于负标签引导的注意力和评分方法用于视觉语言模型的分布外物体检测
Abstract
Out-of-Distribution (OOD) detection is essential for ensuring the robustness and reliability of object detection systems deployed in safety-critical applications. While prior research has mainly focused on uni-modal detectors or vision-language model (VLM) based classifiers, the potential of VLM-based object detectors in OOD scenarios remains underexplored. In this work, we take the first step toward building OOD object detection methods upon VLMs. We identify two challenges specific to VLM detectors: (i) their text-guided attention enhances foreground with ID labels but treats background uniformly, leaving potential OOD regions unexploited for separating in-distribution (ID) from OOD instances; and (ii) their sigmoid-based multi-label outputs are incompatible with softmax-based OOD scores, calling for scoring functions consistent with VLM probabilistic outputs. Hence, we introduce Negative Label Guided Attention and Scoring (NegAS). To address (i), we propose a negative label guided attention module (NegA), where LLM-generated, visually-similar but semantically-different negative labels are used to guide attention toward potential OOD background regions. To address (ii), we introduce a novel sigmoid-based OOD scoring function (NegS) that leverages both ID and negative labels, producing strong responses for ID instances and suppressed responses for OOD ones. Extensive experiments demonstrate that our approach improves OOD detection performance by a large margin while maintaining ID accuracy, e.g., reducing the FPR95 by 11.4% on the COCO dataset and 25.5% on the OpenImages dataset compared to the baseline model. While initially designed for dense VLM detectors like YOLO-World, we successfully adapt NegAS to Grounding DINO, a query-based VLM transformer and achieve significant improvements, demonstrating the generalizability of our framework.
Chinese Translation
分布外（OOD）检测对于确保在安全关键应用中部署的物体检测系统的鲁棒性和可靠性至关重要。尽管之前的研究主要集中在单模态检测器或基于视觉语言模型（VLM）的分类器上，但基于VLM的物体检测器在OOD场景中的潜力仍未得到充分探索。在本研究中，我们迈出了基于VLM构建OOD物体检测方法的第一步。我们识别出VLM检测器特有的两个挑战：（i）其文本引导的注意力增强了带有ID标签的前景，但对背景的处理是统一的，导致潜在的OOD区域未被利用，从而无法有效区分分布内（ID）和分布外（OOD）实例；（ii）其基于sigmoid的多标签输出与基于softmax的OOD评分不兼容，因此需要与VLM概率输出一致的评分函数。因此，我们提出了负标签引导的注意力和评分方法（NegAS）。为了解决（i），我们提出了一种负标签引导的注意力模块（NegA），其中使用由大型语言模型（LLM）生成的视觉相似但语义不同的负标签来引导注意力关注潜在的OOD背景区域。为了解决（ii），我们引入了一种新颖的基于sigmoid的OOD评分函数（NegS），该函数同时利用ID和负标签，为ID实例产生强响应，而对OOD实例产生抑制响应。大量实验表明，我们的方法在提高OOD检测性能的同时保持了ID准确性，例如，在COCO数据集上将FPR95降低了11.4%，在OpenImages数据集上降低了25.5%，与基线模型相比。虽然最初是为像YOLO-World这样的密集VLM检测器设计的，但我们成功地将NegAS适配到基于查询的VLM变换器Grounding DINO，并取得了显著的改进，证明了我们框架的通用性。
cs.CV / 166 / 2606.22540
PolicyTrim: Boosting Intrinsic Policy Efficiency of Vision-Language-Action Models
PolicyTrim：提升视觉-语言-动作模型的内在策略效率
Abstract
Vision-Language-Action (VLA) models provide a unified paradigm for robotic manipulation, yet their real-world deployment is often bottlenecked by execution efficiency. While existing efforts predominantly focus on compute-centric efficiency to reduce per-step inference latency, the intrinsic \textbf{policy efficiency} of these models remains largely unexplored. Policy efficiency is fundamentally affected by two factors, namely the effective executable length of predicted action chunks and the total physical steps required to complete a task. These two factors jointly determine the total number of forward inference calls during execution. We observe that current VLA policies struggle with planning unreliability and action redundancy, suffering from severe prediction degradation at the tail of action chunks and tending to generate unnecessarily redundant physical steps. To address this, we propose \textbf{PolicyTrim}, a reinforcement learning-based post-training framework that extends the reliable action chunk length and reduces redundant physical steps. For reliable chunk extension, we employ a dynamic exploration strategy that explicitly rewards the successful completion of longer executable lengths, progressively pushing the trustworthy prediction horizon to its empirical limit. For step efficiency, we design a redundancy-aware reward that directly favors successful task completions with fewer steps while penalizing unreproducible shortcuts, effectively eliminating redundant physical actions. Extensive experiments across three benchmarks and three VLA models demonstrate that PolicyTrim improves action chunk utilization by 3$\times$ and reduces physical execution steps by 51.4\%. Ultimately, our framework delivers up to a 5.83$\times$ end-to-end deployment speedup without compromising task success rates.
Chinese Translation
视觉-语言-动作（VLA）模型为机器人操作提供了统一的范式，但其在现实世界中的应用常常受到执行效率的瓶颈。现有的研究主要集中在计算中心的效率上，以减少每一步推理的延迟，而这些模型的内在 extbf{策略效率}仍然未得到充分探索。策略效率基本上受到两个因素的影响，即预测动作块的有效可执行长度和完成任务所需的总物理步骤。这两个因素共同决定了执行过程中前向推理调用的总次数。我们观察到当前的VLA策略在规划可靠性和动作冗余方面存在困难，导致在动作块的尾部出现严重的预测退化，并倾向于生成不必要的冗余物理步骤。为了解决这个问题，我们提出了 extbf{PolicyTrim}，一种基于强化学习的后训练框架，旨在延长可靠的动作块长度并减少冗余的物理步骤。为了实现可靠的块扩展，我们采用了一种动态探索策略，该策略明确奖励成功完成更长可执行长度的行为，逐步将可信预测的视野推向其经验极限。为了提高步骤效率，我们设计了一种关注冗余的奖励机制，直接支持以更少的步骤成功完成任务，同时惩罚不可重复的捷径，有效消除冗余的物理动作。在三个基准和三种VLA模型上的广泛实验表明，PolicyTrim将动作块的利用率提高了3倍，并将物理执行步骤减少了51.4%。最终，我们的框架在不影响任务成功率的情况下，实现了高达5.83倍的端到端部署加速。
cs.CV / 167 / 2606.22543
MAPS: Multi-Anchor Projection Similarity for Joint Vision-Language Geo-Localization
MAPS：用于联合视觉-语言地理定位的多锚投影相似性
Abstract
Humans localize places by integrating perceptual cues from vision with semantic reasoning from language, forming a scene understanding that is both intuitive and structured. Although existing geo-localization models have made substantial progress in cross-view and cross-modal settings, they are largely built upon point-to-point alignment, which is insufficient for joint vision-language queries. In such queries, visual and textual cues do not simply act as independent references, but jointly define a semantic subspace for locating the target. In this paper, we formulate vision-language geo-localization (VLGL) with joint image-text queries as a multi-anchor geometric alignment problem and propose a unified framework for this setting. To realize this formulation, we propose Multi-Anchor Projection Similarity (MAPS), a new metric which constructs an anchor plane from visual and textual query features in a high-dimensional space and measures similarity by the projection length of the target feature onto this plane. Unlike cosine similarity which evaluates isolated pairwise relations, MAPS captures the geometric consistency between the target feature and the joint query subspace, providing a more discriminative ranking criterion during retrieval. To make the learned representation consistent with this geometry, we further introduce a MAPS-based contrastive loss that drives target features toward the corresponding anchor plane. The proposed framework, similarity metric, and training objective jointly yield state-of-the-art performance in VLGL.
Chinese Translation
人类通过整合来自视觉的感知线索与来自语言的语义推理来定位地点，从而形成既直观又结构化的场景理解。尽管现有的地理定位模型在跨视角和跨模态设置中取得了显著进展，但它们主要基于点对点的对齐，这对于联合视觉-语言查询来说是不够的。在此类查询中，视觉和文本线索并非简单地作为独立的参考，而是共同定义了一个语义子空间以定位目标。在本文中，我们将视觉-语言地理定位（VLGL）与联合图像-文本查询形式化为一个多锚几何对齐问题，并为这一设置提出了一个统一框架。为了实现这一形式化，我们提出了多锚投影相似性（MAPS），这是一种新度量，它在高维空间中从视觉和文本查询特征构建锚平面，并通过目标特征在该平面上的投影长度来衡量相似性。与评估孤立成对关系的余弦相似性不同，MAPS 捕捉了目标特征与联合查询子空间之间的几何一致性，在检索过程中提供了更具区分性的排序标准。为了使学习到的表示与该几何结构一致，我们进一步引入了一种基于 MAPS 的对比损失，驱动目标特征朝向相应的锚平面。所提出的框架、相似性度量和训练目标共同实现了 VLGL 的最先进性能。
cs.CV / 168 / 2606.22546
Venice-H1: Failure-Aware Query Re-Ranking with Multi-Scale Grid Signatures for Referring Image Segmentation
Venice-H1：具有多尺度网格特征的失败感知查询重排序用于引用图像分割
Abstract
Modern Referring Image Segmentation (RIS) systems generate multiple candidate masks per expression but rely on a simple heuristic--typically the argmax detection score--to select the final output. We identify query selection as a failure-case bottleneck: although heuristic selection succeeds on 82-93% of samples, the residual 7-18% of failures dominate the error budget, leaving a best-query selection gap of 3-11% mIoU. We introduce Venice-H1, a lightweight, backbone-decoupled post-hoc re-ranking module that encodes each candidate through multi-scale grid signatures--compact spatial descriptors pooled onto 4x4, 8x8, and 16x16 grids--and feeds them to a Transformer-based re-ranker with a Failure Gate (ROCAUC 0.78-0.82) that intervenes only when the default choice is likely suboptimal. Instantiated on DeRIS-L and DeRIS-B, Venice-H1 achieves delta_fail of +1.40 and +0.89 mIoU with strictly positive 95% CIs on all 16/16 (split, backbone) pairs and harmful-switch rates below 0.53%. Zero-shot transfer to medical referring segmentation (MS-CXR, M3D-RefSeg-2D) yields +1.16 and +0.51 mIoU without RIS-backbone fine-tuning. The module adds approximately 11.3M parameters and under 1 ms latency.
Chinese Translation
现代引用图像分割（RIS）系统针对每个表达生成多个候选掩膜，但依赖于简单的启发式方法——通常是 argmax 检测分数——来选择最终输出。我们将查询选择识别为一个失败案例瓶颈：尽管启发式选择在 82-93% 的样本上成功，但剩余的 7-18% 失败案例主导了错误预算，导致最佳查询选择存在 3-11% 的 mIoU 差距。我们提出了 Venice-H1，这是一种轻量级、与主干解耦的后处理重排序模块，通过多尺度网格特征对每个候选进行编码——将紧凑的空间描述符汇聚到 4x4、8x8 和 16x16 网格上——并将其输入到基于 Transformer 的重排序器中，该重排序器具有一个失败门（ROCAUC 0.78-0.82），仅在默认选择可能次优时进行干预。在 DeRIS-L 和 DeRIS-B 上实例化的 Venice-H1 实现了 delta_fail 分别为 +1.40 和 +0.89 mIoU，所有 16/16（拆分，主干）对的 95% 置信区间均为严格正值，且有害切换率低于 0.53%。对医疗引用分割（MS-CXR，M3D-RefSeg-2D）的零样本迁移在没有 RIS 主干微调的情况下获得了 +1.16 和 +0.51 mIoU。该模块增加了大约 11.3M 的参数，延迟低于 1 毫秒。
cs.CV / 169 / 2606.22550
Training-Free Semantic Correction for Autoregressive Visual Models
无训练的自回归视觉模型语义修正
Abstract
Autoregressive visual models (AVMs) based on next-scale prediction have emerged as a prominent paradigm for image and video synthesis. However, decomposing the generation process into discrete scales with varying granularities in AVM makes semantic errors difficult to identify and correct, thereby undermining the quality of the final output. Prior efforts to enhance AVM can be categorized into training-based and training-free approaches. Although training-based efforts to enhance AVM generation quality come at substantial computational cost, existing training-free methods neglect intermediate generation states, leaving semantic errors undiagnosed and allowing them to accumulate into the final output. In this paper, we focus on training-free paradigms and propose Gazer, a framework that integrates multimodal large language model feedback into the AVM sampling loop for in-generation semantic correction. Concretely, Gazer operates via two cooperating stages: the Reflective Diagnosis stage diagnoses semantic errors from intermediate states, while the Semantic Correction stage rewinds and rectifies the generation trajectory to realign with the target prompt. Experiments on compositional image and video benchmarks demonstrate that Gazer improves semantic alignment and compositional accuracy across multiple AVMs without additional training.
Chinese Translation
基于下一个尺度预测的自回归视觉模型（AVMs）已成为图像和视频合成的重要范式。然而，在AVM中将生成过程分解为具有不同粒度的离散尺度，使得语义错误难以识别和修正，从而影响最终输出的质量。以往提升AVM的努力可以分为基于训练的方法和无训练的方法。尽管基于训练的提升AVM生成质量的努力需要大量的计算成本，但现有的无训练方法忽视了中间生成状态，导致语义错误未被诊断并积累到最终输出中。本文聚焦于无训练的范式，提出了Gazer，一个将多模态大型语言模型反馈集成到AVM采样循环中的框架，以实现生成过程中的语义修正。具体而言，Gazer通过两个协作阶段进行操作：反思诊断阶段从中间状态诊断语义错误，而语义修正阶段则回溯并修正生成轨迹，以重新对齐目标提示。在组合图像和视频基准上的实验表明，Gazer在多个AVM中提高了语义对齐和组合准确性，而无需额外的训练。
cs.CV / 170 / 2606.22556
HiMatch-AD: DINOv3-driven Hierarchical Matching for Training-free Medical Anomaly Detection
HiMatch-AD：基于 DINOv3 的分层匹配框架用于无训练医学异常检测
Abstract
Anomaly detection is essential for medical image analysis, where pathological regions often appear as rare deviations from normal anatomical structures. While training-based methods have achieved promising performance, they require task-specific optimization and extensive normal data, which limits scalability across modalities and institutions. Training-free approaches offer greater flexibility by leveraging pretrained visual representations, yet existing methods typically rely on simple nearest-neighbor retrieval and naive aggregation strategies, which may fail to capture hierarchical semantics and ignore the reliability of multiple anomaly responses. In this work, we propose HiMatch-AD, a DINOv3-driven hierarchical matching framework for training-free medical anomaly detection. Our method first retrieves semantically relevant normal references via dual-branch matching that jointly considers global CLS-token similarity and patch-level representations. Hierarchical anomaly maps are then generated across multiple transformer stages by comparing clustered normal features with query representations. To robustly aggregate anomaly responses, we introduce a unified uncertainty-based fusion mechanism that adaptively weights maps according to their reliability. The entire framework operates without any task-specific training. Extensive experiments on the BMAD benchmark, including brain MRI, liver CT, and retinal OCT datasets, demonstrate that HiMatch-AD consistently outperforms both training-based and DINO-based state-of-the-art methods, which highlights the effectiveness of multi-level matching and uncertainty-aware fusion for scalable medical anomaly detection.
Chinese Translation
异常检测在医学图像分析中至关重要，其中病理区域通常表现为与正常解剖结构的罕见偏差。尽管基于训练的方法已取得了令人鼓舞的性能，但它们需要特定任务的优化和大量正常数据，这限制了其在不同模态和机构间的可扩展性。无训练的方法通过利用预训练的视觉表征提供了更大的灵活性，然而现有方法通常依赖于简单的最近邻检索和天真的聚合策略，这可能无法捕捉分层语义并忽视多个异常响应的可靠性。在本研究中，我们提出了 HiMatch-AD，一种基于 DINOv3 的分层匹配框架，用于无训练的医学异常检测。我们的方法首先通过双分支匹配检索语义相关的正常参考，这种匹配同时考虑了全局 CLS-token 相似性和补丁级别的表征。然后，通过将聚类的正常特征与查询表征进行比较，在多个变换器阶段生成分层异常图。为了稳健地聚合异常响应，我们引入了一种统一的不确定性融合机制，根据其可靠性自适应地加权图。整个框架在没有任何特定任务训练的情况下运行。在 BMAD 基准上的大量实验，包括脑部 MRI、肝脏 CT 和视网膜 OCT 数据集，证明 HiMatch-AD 始终优于基于训练和基于 DINO 的最先进方法，这突显了多层匹配和不确定性感知融合在可扩展医学异常检测中的有效性。
cs.CV / 171 / 2606.22568
SeFi-Image: A Text-to-Image Foundation Model with Semantic-First Diffusion
SeFi-Image：一种以语义优先扩散为基础的文本到图像生成模型
Abstract
Training image generation foundation models consumes substantial resources. Previous methods have attempted to leverage semantic guidance to accelerate the training process, yet their experiments were only conducted on simple datasets such as ImageNet, at low resolutions, and with small-scale models. In this paper, we propose SeFi-Image, a text-to-image foundation model built upon semantic-first diffusion, a novel latent diffusion modeling paradigm. We instantiate SeFi-Image at three model scales, 1B, 2B, and 5B parameters, enabling systematic study of scaling behavior and flexible deployment under varying compute budgets. Notably, our largest 5B model was trained with merely 125K A800 GPU hours, corresponding to roughly 10-20% of the training compute used by Z-Image. However, it achieves results comparable to or even superior to Qwen-Image and Z-Image. Despite this modest training compute, SeFi-Image achieves strong performance on a wide range of benchmarks, including GenEval, DPG, LongTextBench, OneIG, and CVTG-2K. Moreover, we provide DMD2-distilled few-step turbo variants for each model scale to accommodate diverse hardware constraints and latency requirements. We publicly release our code, weights and hope this work offers the community useful insights into semantic-guided diffusion modeling for T2I generation, while also providing practical and readily deployable model options.
Chinese Translation
训练图像生成基础模型消耗大量资源。以往的方法尝试利用语义指导来加速训练过程，但其实验仅在简单的数据集（如 ImageNet）上进行，且分辨率较低，模型规模也较小。本文提出了 SeFi-Image，这是一种基于语义优先扩散的新型文本到图像基础模型，采用了一种新颖的潜在扩散建模范式。我们在三个模型规模（分别为 1B、2B 和 5B 参数）上实现了 SeFi-Image，从而系统地研究了规模效应，并在不同的计算预算下灵活部署。值得注意的是，我们最大的 5B 模型仅用 125K A800 GPU 小时进行训练，这大约是 Z-Image 所使用训练计算的 10-20%。然而，它的结果与 Qwen-Image 和 Z-Image 相当，甚至更优。尽管训练计算量适中，SeFi-Image 在包括 GenEval、DPG、LongTextBench、OneIG 和 CVTG-2K 在内的广泛基准测试中表现出色。此外，我们为每个模型规模提供了 DMD2 蒸馏的少步快速变体，以适应不同的硬件限制和延迟要求。我们公开发布了我们的代码和权重，希望这项工作能为社区提供有用的见解，促进基于语义指导的扩散建模在文本到图像生成中的应用，同时提供实用且易于部署的模型选项。
cs.CV / 172 / 2606.22574
The Power of Light: Improving Synthetic-to-Real Domain Adaptation through Physically-Based Indirect Illumination
光的力量：通过基于物理的间接照明改善合成到真实的领域适应
Abstract
While synthetic data generation resolves the manual labeling bottleneck in computer vision, minimizing the syn-to-real domain gap requires optimizing rendering variables. This paper presents a systematic study analyzing the impact of lighting configurations and background complexity on object detection performance. We introduce SmartSDG, an automated, reproducible pipeline built on NVIDIA Isaac Sim using Physically-Based Shading (PBS), alongside ILLUM\_INTRUCK, a new multi-object industrial benchmark dataset. Through 18 controlled experiments utilizing a state-of-the-art YOLOv12 framework, we demonstrate that complex, indirect lighting configurations paired with domain-relevant background variability significantly increase visual cue richness. Our quantitative findings show that avoiding direct specular peaks preserves crucial surface textures, mitigates the domain gap, reduces false positives, and accelerates model convergence compared to using conventional direct-light synthetic data. Ultimately, we provide actionable virtual scene design guidelines to maximize object detection robustness in industrial automation.
Chinese Translation
虽然合成数据生成解决了计算机视觉中的手动标注瓶颈，但最小化合成到真实的领域差距需要优化渲染变量。本文系统性地研究了照明配置和背景复杂性对物体检测性能的影响。我们介绍了SmartSDG，这是一个基于NVIDIA Isaac Sim构建的自动化、可重复的管道，采用基于物理的着色（PBS），同时推出了ILLUM_INTRUCK，一个新的多物体工业基准数据集。通过18个控制实验，利用最先进的YOLOv12框架，我们证明了复杂的间接照明配置与领域相关的背景变异相结合，显著增加了视觉线索的丰富性。我们的定量研究结果表明，避免直接的高光峰值可以保留关键的表面纹理，减小领域差距，减少误报，并加速模型收敛，相较于使用传统的直接光合成数据。最终，我们提供了可操作的虚拟场景设计指南，以最大化工业自动化中的物体检测鲁棒性。
cs.CV / 173 / 2606.22597
MapReason-OSM: Can Vision-Language Models Make Graph-Verifiable Mobility Decisions from Street Maps ?
MapReason-OSM：视觉-语言模型能否从街道地图中做出图可验证的出行决策？
Abstract
Vision-language models (VLMs) are increasingly used to read maps for logistics, delivery, and accessible navigation, where the output is an actionable decision (a route, a pin, a parking choice) that must respect the road network. Yet most map benchmarks grade free-text or multiple-choice answers that cannot be verified against the underlying graph. We present \textbf{MapReason-OSM}, a benchmark and evaluation harness for graph-verifiable mobility decisions on self-rendered OpenStreetMap panels. We render fixed-style maps for ten U.S. downtowns at two aligned zoom scales, overlay a consistent marker grammar, and pair each panel with a hidden street graph and exact oracles, yielding 6{,}000 instances (12{,}000 panels across the two zooms) over 12 routing, facility-location, and visual-disambiguation tasks. Models return structured decisions that we snap back to the graph and score for validity, legality, optimality, and constraint satisfaction, plus \emph{cross-zoom consistency}. Across seven VLMs, models read maps and route simply but fail at graph-cost reasoning (single-facility pin placement is near chance even for frontier reasoning models), and are frequently scale-inconsistent. We release the benchmark, harness, and deterministic generator.
Chinese Translation
视觉-语言模型（VLMs）在物流、配送和无障碍导航中越来越多地被用于读取地图，其输出是一个可操作的决策（如路线、标记、停车选择），必须遵循道路网络。然而，大多数地图基准测试评估的是无法与底层图进行验证的自由文本或多项选择答案。我们提出了 extbf{MapReason-OSM}，这是一个用于自渲染OpenStreetMap面板的图可验证出行决策的基准和评估工具。我们为美国十个市中心在两个对齐的缩放比例下渲染固定样式的地图，叠加一致的标记语法，并将每个面板与隐藏的街道图和精确的预言者配对，生成6,000个实例（在两个缩放下共12,000个面板），涵盖12个路由、设施位置和视觉消歧任务。模型返回结构化决策，我们将其映射回图中，并对有效性、合法性、最优性和约束满足进行评分，以及 extit{跨缩放一致性}。在七个VLMs中，模型能够简单地读取地图和规划路线，但在图成本推理方面表现不佳（即使对于前沿推理模型，单设施标记放置的成功率也接近随机），并且经常出现尺度不一致。我们发布了该基准、评估工具和确定性生成器。
cs.CV / 174 / 2606.22608
Automated sign detection across the Electronic Babylonian Library: A large-scale dataset and end-to-end cuneiform OCR pipeline
电子巴比伦图书馆中的自动符号检测：一个大规模数据集和端到端的楔形文字光学字符识别管道
Abstract
Learning to read cuneiform tablets is an extremely demanding task; consequently, of the roughly half million excavated tablets, only a small fraction has been analysed by Assyriologists. Computer vision offers a promising avenue for decipherment but requires large, densely annotated datasets. To address this limitation, the largest annotated cuneiform sign dataset to date is used, and a Deformable Detection Transformer (DETR)-based object detection model is evaluated under two class granularities of 173 and 106 classes. The proposed system integrates automatic tablet-side extraction, heuristic line grouping, and n-gram-based textual similarity evaluation to bridge visual sign detection and textual structure, and achieves consistent improvements of up to 28-37% over prior work on COCO-style detection metrics. At inference, the method is applied to 87,668 tablet fragments from the Electronic Babylonian Library (eBL) corpus, producing nearly 2.9 million sign detections. Although the approach operates without linguistic priors and remains sensitive to tablet damage and layout variability, it provides a scalable and interpretable foundation for corpus-wide cuneiform analysis and supports future integration with multimodal and linguistic modelling frameworks.
Chinese Translation
学习阅读楔形文字泥板是一项极具挑战性的任务；因此，在大约五十万件出土的泥板中，只有一小部分被亚述学家分析。计算机视觉为解读提供了一个有前景的途径，但需要大型、密集标注的数据集。为了解决这一局限性，本文使用了迄今为止最大的标注楔形文字符号数据集，并在173类和106类的两种类别粒度下评估了基于可变形检测变换器（Deformable Detection Transformer, DETR）的目标检测模型。所提出的系统集成了自动泥板侧提取、启发式行分组和基于n-gram的文本相似性评估，以桥接视觉符号检测和文本结构，并在COCO风格检测指标上实现了高达28-37%的持续改进。在推理阶段，该方法应用于电子巴比伦图书馆（Electronic Babylonian Library, eBL）语料库中的87,668个泥板碎片，产生了近290万个符号检测。尽管该方法在没有语言先验的情况下运行，并且对泥板损坏和布局变化敏感，但它为全语料库的楔形文字分析提供了可扩展和可解释的基础，并支持未来与多模态和语言建模框架的整合。
cs.CV / 175 / 2606.22617
OmniSpace: Efficient Geometry Awareness for Autonomous Vehicles MLLMs
OmniSpace：面向自主车辆的高效几何感知多模态大语言模型
Abstract
Multimodal Large Language Models (MLLMs) have achieved remarkable performance on 2D visual tasks, yet enhancing their spatial intelligence for real-world applications such as Autonomous Vehicles (AV) remains an open challenge. Existing geometry-aware MLLMs typically rely on auxiliary 3D models at inference time, introducing pipeline complexity and the risk of cascading failures. In this paper, we present OmniSpace, a simple yet effective plug-and-play paradigm for geometry-aware spatial reasoning from purely 2D observations. Motivated by our finding that current MLLMs are bottlenecked by weak cross-view correspondence and depth estimation, OmniSpace introduces a Camera Pose Injector, a Multi-view Epipolar Attention module, and a 3D Geometric Distillation objective that jointly address these two limitations by transferring geometric knowledge into the model. Extensive experiments show that OmniSpace surpasses existing methods on planning benchmarks (nuScenes, Bench2Drive), risk detection (nuInstruct), language (Omnidrive), and generalization (DriveBench).
Chinese Translation
多模态大语言模型（MLLMs）在二维视觉任务上取得了显著的性能，但提升其在自主车辆（AV）等现实应用中的空间智能仍然是一个开放的挑战。现有的几何感知MLLMs通常在推理时依赖辅助的三维模型，这增加了管道的复杂性并带来了级联故障的风险。本文提出了OmniSpace，一种简单而有效的即插即用范式，旨在通过纯二维观察实现几何感知的空间推理。我们的研究发现，当前的MLLMs受到弱跨视图对应和深度估计的瓶颈限制，因此OmniSpace引入了相机姿态注入器、一个多视图极线注意力模块，以及一个三维几何蒸馏目标，这三者共同解决了这两个限制，通过将几何知识转移到模型中。大量实验表明，OmniSpace在规划基准（nuScenes, Bench2Drive）、风险检测（nuInstruct）、语言（Omnidrive）和泛化（DriveBench）等任务上超越了现有方法。
cs.CV / 176 / 2606.22625
DR-Mamba: Automatic Inference-Time Domain Adaptation for Document Image Binarization via Sample-Conditioned Detail-Background Suppression
DR-Mamba：通过样本条件细节-背景抑制实现文档图像二值化的自动推理时域适应
Abstract
Degraded document image binarization is sensitive to domain shifts caused by paper aging, bleed-through, stains, shadows, and uneven illumination, and the foreground-background separation of recent learning-based methods can become unstable on unseen degradation domains. We propose DR-Mamba, a sample-conditioned detail-background suppression framework that performs automatic inference-time domain adaptation for document image binarization. Unlike test-time adaptation methods that require gradient updates or auxiliary data at inference, DR-Mamba adapts to each input document through input-dependent gates within a single forward pass, requiring no target-domain labels, no fine-tuning, and no test-time parameter updates. Instead of using Mamba-style selective scanning as a single generic feature path, DR-Mamba reinterprets it as fast-slow route modeling: a fast detail route captures local stroke structures, while a slow background route accumulates spatially persistent degradation responses. The two routes are integrated through an input-dependent subtractive gate that explicitly suppresses background interference rather than fusing features by addition or concatenation. We further add full-resolution detail-guided reconstruction and thin-stroke-aware supervision to recover fine strokes lost during downsampling. Evaluated under a leave-one-year-out protocol on DIBCO-style benchmarks, where each held-out year is treated as an unseen degradation domain, DR-Mamba shows that per-document, per-location subtractive suppression improves cross-domain robustness, with particularly strong performance on the most severely degraded held-out fold.
Chinese Translation
退化文档图像的二值化对纸张老化、渗透、污渍、阴影和不均匀照明等引起的领域变化非常敏感，而最近基于学习的方法在未见过的退化领域中的前景-背景分离可能变得不稳定。我们提出了DR-Mamba，一种样本条件的细节-背景抑制框架，能够在文档图像二值化过程中实现自动推理时域适应。与需要在推理时进行梯度更新或辅助数据的测试时适应方法不同，DR-Mamba通过输入依赖的门控机制在单次前向传递中适应每个输入文档，无需目标领域标签、无须微调，也不需要测试时参数更新。DR-Mamba将Mamba风格的选择性扫描重新解释为快-慢路径建模：快速细节路径捕捉局部笔画结构，而慢速背景路径则积累空间上持久的退化响应。这两条路径通过一个输入依赖的减法门集成，明确抑制背景干扰，而不是通过加法或连接融合特征。我们进一步添加全分辨率细节引导重建和细笔画感知监督，以恢复在下采样过程中丢失的细笔画。在DIBCO风格基准测试下，采用逐年留出协议进行评估，其中每个保留年份被视为未见过的退化领域，DR-Mamba显示每个文档、每个位置的减法抑制提高了跨领域的鲁棒性，在最严重退化的保留折上表现尤为强劲。
cs.CV / 177 / 2606.22631
4DVLT: Dynamic Scene Understanding with Worldline-Centered Vision-Language Tracking
4DVLT：基于世界线的视觉-语言跟踪的动态场景理解
Abstract
4D dynamic scene understanding requires grounding language to a persistent worldline that binds identity, metric 3D motion, and synchronized multi-view 2D projections. Existing paradigms capture only part of this structure: large multimodal models reason over rich visual evidence but rarely preserve metric topology, while vision-language tracking remains tied to fragmented 2D or 3D outputs and local continuation. We therefore introduce \textbf{4DVLT}, a worldline-centered task for instruction-conditioned 4D dynamic scene understanding in fully observed multi-view video, and \textbf{Instruct-4D}, a benchmark with 129.4K question-answer pairs, 64.7K target entities, 851 scenes, and 9 reasoning-oriented query types. To address this setting, we present \textbf{4DTrack}, which casts instruction-conditioned tracking as graph-conditioned worldline inference through an object-centric 4D state graph, metric-guided routing, bidirectional decoding, and kinematic calibration. On Instruct-4D, 4DTrack-Qwen3.5-9B reaches 62.68 $\mathrm{TGA}_{\mathrm{Top1}}$ and surpasses the best adapted VLT baseline by 19.62 points. These results show that worldline-centered modeling improves both target grounding and recovered worldline quality. The project page is available at https://github.com/mikubaka88/4DVLT.
Chinese Translation
4D动态场景理解需要将语言与一个持久的世界线相结合，该世界线绑定了身份、度量3D运动和同步的多视角2D投影。现有的范式仅捕捉了这一结构的一部分：大型多模态模型在丰富的视觉证据上进行推理，但很少保持度量拓扑，而视觉-语言跟踪仍然与碎片化的2D或3D输出和局部延续相联系。因此，我们提出了 extbf{4DVLT}，这是一个以世界线为中心的任务，用于在完全观察的多视角视频中进行指令条件的4D动态场景理解，并且 extbf{Instruct-4D}是一个基准数据集，包含129.4K个问答对、64.7K个目标实体、851个场景和9种面向推理的查询类型。为了解决这一设置，我们提出了 extbf{4DTrack}，它通过一个以对象为中心的4D状态图、度量引导的路由、双向解码和运动学校准，将指令条件的跟踪视为图条件的世界线推理。在Instruct-4D上，4DTrack-Qwen3.5-9B达到了62.68 $ ext{TGA}_{ ext{Top1}}$，并超过了最佳适应的VLT基线19.62分。这些结果表明，以世界线为中心的建模改善了目标定位和恢复的世界线质量。项目页面可访问：https://github.com/mikubaka88/4DVLT。
cs.CV / 178 / 2606.22634
Learning Entropy Signature for Image Representation and Classification
用于图像表示和分类的学习熵特征
Abstract
Learning Entropy (LE) has recently been extended to image analysis through Spatial Learning Entropy Maps (SLEMs), which are two-dimensional LE distributions that highlight unusually high learning activity across an image. Unlike conventional image descriptors, SLEMs are generated by incremental, sample-wise learning of a pretrained feedforward MLP network, where local pixel neighborhoods are presented sequentially in a fixed spatial order to predict the corresponding central pixels. Consequently, the learning activity at each image location depends not only on its local structure but also on the knowledge acquired from previously processed locations. This paper introduces Learning Entropy Signatures (LES), an image descriptor derived from SLEM using the K largest LE locations. LES captures the spatial organization of learning-relevant image structures and provides a compact representation of image content based on learning weight behavior. Experimental evaluation on image classification tasks shows that a relatively small number of K largest LE locations preserve substantial discriminative information. The results indicate a close relationship between the learning of neural weights and information relevance, extending the role of Learning Entropy from time series to images and, within images, from structural point extraction to compact image representation and classification.
Chinese Translation
学习熵（Learning Entropy, LE）最近通过空间学习熵图（Spatial Learning Entropy Maps, SLEMs）扩展到图像分析，这是一种二维的LE分布，突出显示图像中异常高的学习活动。与传统的图像描述符不同，SLEMs是通过对预训练的前馈多层感知器（MLP）网络进行增量的逐样本学习生成的，其中局部像素邻域按照固定的空间顺序依次呈现，以预测相应的中心像素。因此，每个图像位置的学习活动不仅依赖于其局部结构，还依赖于从之前处理的位置获得的知识。本文介绍了学习熵特征（Learning Entropy Signatures, LES），这是一种基于SLEM使用K个最大LE位置派生的图像描述符。LES捕捉与学习相关的图像结构的空间组织，并基于学习权重行为提供图像内容的紧凑表示。在图像分类任务上的实验评估表明，相对较少的K个最大LE位置保留了大量的区分信息。结果表明神经权重的学习与信息相关性之间存在密切关系，将学习熵的作用从时间序列扩展到图像，并在图像内部，从结构点提取扩展到紧凑的图像表示和分类。
cs.CV / 179 / 2606.22649
MaRS: Robust Out-of-Distribution Detection via Mahalanobis Residual Scoring
MaRS：通过马氏残差评分实现稳健的分布外检测
Abstract
Foundation models provide highly descriptive representations for medical images, yet their reliability degrades under distribution shifts arising from changes in patients, devices, or acquisition conditions. Reliable out-of-distribution (OOD) detection is therefore essential for safe deployment. Recent post-hoc detectors efficiently exploit frozen embeddings (\emph{e.g.}, kNN), whereas reconstruction-based OOD detection in latent feature space has seen limited adoption due to inconsistent performance. In this work, we show that the limitation of reconstruction-based methods in latent space does not stem from poor reconstruction quality, but from how reconstruction errors are scored. Standard $L_2$ residual norms collapse the anisotropic residual structure, thereby suppressing informative deviations. To address this limitation, we introduce \texttt{MaRS} (Mahalanobis Residual Scoring), a label-free OOD detector that learns an in-distribution manifold using a lightweight autoencoder and measures deviation via a Mahalanobis distance on reconstruction residuals, yielding variance-aware OOD scores. Across three imaging modalities, multiple types of distribution shift, and different model families and scales, \texttt{MaRS} outperforms established confidence-, distance-, and reconstruction-based baselines, while remaining fully post-hoc and lightweight. The code is available at https://github.com/francescodisalvo05/mars.
Chinese Translation
基础模型为医学图像提供了高度描述性的表示，然而在患者、设备或采集条件变化引起的分布转变下，其可靠性会下降。因此，可靠的分布外（OOD）检测对于安全部署至关重要。近期的后处理检测器有效利用了冻结的嵌入（例如，kNN），而基于重构的OOD检测在潜在特征空间中的应用受到限制，主要是由于其性能不稳定。在本研究中，我们表明，潜在空间中基于重构的方法的局限性并非源于重构质量差，而是源于重构误差的评分方式。标准的 $L_2$ 残差范数会压缩各向异性的残差结构，从而抑制了信息性偏差。为了解决这一局限性，我们引入了 exttt{MaRS}（马氏残差评分），这是一种无标签的OOD检测器，它使用轻量级自编码器学习分布内流形，并通过重构残差上的马氏距离来衡量偏差，从而产生考虑方差的OOD评分。在三种成像模式、多个类型的分布转变以及不同的模型系列和规模下， exttt{MaRS} 超越了现有的基于置信度、距离和重构的基线，同时保持完全的后处理和轻量级。代码可在 https://github.com/francescodisalvo05/mars 获取。
cs.CV / 180 / 2606.22660
Prompting Diffusion Models for Zero-Shot Instance Segmentation
提示扩散模型用于零-shot实例分割
Abstract
Several disruptive research directions have recently emerged in computer vision, including foundation models achieving previously unseen zero-shot performance in scene understanding, even interactively, and generative models that synthesize extremely realistic images. The latter have also been shown to be highly effective in scene understanding tasks thanks to their rich priors. However, for promptable segmentation, foundation models struggle with accurately segmenting an object's region, leading to false positives and over-segmentation. Notably, early attempts that leverage generative priors use prompts only during post-processing, yielding suboptimal segments because the process is agnostic to the user input. In this paper, we target these limitations with Prompt2Seg, a spatial conditioning framework for diffusion-based segmentation. Prompt2Seg augments a frozen diffusion segmentation model with a conditioning branch. Our approach takes spatial prompts, represented as 2D Gaussians or confidence maps, as explicit input signals, training the model to respond directly to user intent. Fine-tuned on a deliberately constrained set of object categories drawn from Hypersim and Virtual KITTI 2, Prompt2Seg generalizes zero-shot to a wide range of unseen object types and visual domains. We evaluate on seven datasets ranging from standard benchmarks to more challenging domains, including paintings, egocentric views, and X-ray data. Furthermore, we demonstrate that Prompt2Seg consistently outperforms the underlying diffusion segmentation backbone across all benchmarks. Our results suggest that the rich priors encoded in generative pretraining, combined with principled spatial conditioning, offer a compelling path toward broadly generalizing interactive segmentation without large-scale mask supervision.
Chinese Translation
最近，计算机视觉领域出现了几种颠覆性的研究方向，包括基础模型在场景理解中实现前所未有的零-shot性能，甚至可以进行交互，以及生成模型合成极为真实的图像。后者在场景理解任务中也被证明是非常有效的，因为它们具有丰富的先验知识。然而，对于可提示的分割，基础模型在准确分割物体区域方面存在困难，导致假阳性和过度分割。值得注意的是，早期利用生成先验的尝试仅在后处理阶段使用提示，这导致了次优的分割结果，因为该过程对用户输入是无关的。在本文中，我们针对这些局限性提出了Prompt2Seg，这是一种基于扩散的分割的空间条件框架。Prompt2Seg通过一个条件分支增强了一个冻结的扩散分割模型。我们的方法将空间提示（以二维高斯或置信度图的形式表示）作为明确的输入信号，训练模型直接响应用户意图。在从Hypersim和Virtual KITTI 2中提取的有意限制的物体类别集上进行微调后，Prompt2Seg能够在广泛的未见物体类型和视觉领域中实现零-shot泛化。我们在七个数据集上进行了评估，这些数据集从标准基准到更具挑战性的领域，包括绘画、自我中心视图和X射线数据。此外，我们证明了Prompt2Seg在所有基准测试中始终优于基础的扩散分割骨干。我们的结果表明，生成预训练中编码的丰富先验知识，结合有原则的空间条件，为广泛泛化交互式分割提供了一条引人注目的路径，而无需大规模的掩码监督。
cs.CV / 181 / 2606.22694
SATURN: Symbolic Spatial Reasoning for Multi-Perspective Grounding
SATURN：用于多视角基础的符号空间推理
Abstract
Vision-Language Models (VLMs) remain unreliable when spatial reasoning requires composing relations whose meanings depend on frames of reference. Existing neuro-symbolic methods make reasoning more explicit, but often depend on brittle geometric procedures and hard decisions over noisy perception. We propose SATURN, a neuro-symbolic framework for perspective-aware compositional spatial reasoning. SATURN reconstructs an approximate 3D scene, derives soft perspective-aware spatial predicates, and composes them with a training-free Pythonic symbolic executor, separating perception from reasoning while preserving uncertainty through multi-hop inference. We also introduce 3D FORCE, a diagnostic benchmark that controls reasoning depth, view, and perspective composition across spatial arrangement grounding (SAG) and referring expression grounding (REF). On 3D FORCE, VLMs and spatially trained models degrade sharply as depth and perspective complexity increase, whereas SATURN remains stable and outperforms strong baselines. On the real-world MindCube benchmark, SATURN achieves 78.57% overall accuracy, outperforming the strongest baseline by 14 pp.
Chinese Translation
视觉-语言模型（VLMs）在空间推理中仍然不可靠，特别是当推理需要组合依赖于参考框架的关系时。现有的神经符号方法使推理更加明确，但往往依赖于脆弱的几何程序和对噪声感知的严格决策。我们提出了SATURN，一个用于视角感知组合空间推理的神经符号框架。SATURN重建一个近似的3D场景，推导出软视角感知空间谓词，并通过一个无训练的Python符号执行器将它们组合，分离感知与推理，同时通过多跳推理保留不确定性。我们还引入了3D FORCE，一个控制推理深度、视角和空间排列基础（SAG）与指称表达基础（REF）之间视角组合的诊断基准。在3D FORCE上，VLMs和空间训练模型在深度和视角复杂性增加时急剧降级，而SATURN保持稳定并超越了强基线。在现实世界的MindCube基准上，SATURN实现了78.57%的整体准确率，比最强基线高出14个百分点。
cs.CV / 182 / 2606.22696
NullFlow: One-Step Generative Reconstruction
NullFlow：一步生成重建
Abstract
We propose NullFlow, a principled framework for one-step generative image reconstruction. Our key idea is to confine the generative flow to a measurement-consistent subspace. Because the flow never leaves this subspace, NullFlow needs no separate data-fidelity corrections, unlike existing solvers. NullFlow samples in a single network evaluation by learning the flow's average velocity, avoiding the step-by-step integration of traditional flow matching methods. We prove that the average velocity of this constrained flow yields a training objective whose global minimizer is a one-step posterior sampler. We show on image inpainting that NullFlow matches state-of-the-art diffusion solvers while cutting inference from hundreds of network evaluations to one.
Chinese Translation
我们提出了NullFlow，这是一个用于一步生成图像重建的原则性框架。我们的关键思想是将生成流限制在一个测量一致的子空间内。由于流从未离开这个子空间，NullFlow不需要像现有求解器那样进行单独的数据保真度修正。NullFlow通过学习流的平均速度，在单次网络评估中进行采样，避免了传统流匹配方法的逐步积分。我们证明了这种受限流的平均速度产生的训练目标，其全局最小化解是一个一步后验采样器。我们在图像修复任务上展示了NullFlow与最先进的扩散求解器相匹配，同时将推理次数从数百次网络评估减少到一次。
cs.CV / 183 / 2606.22699
Catching Lies Without Sending the Video: Privacy-Preserving Multimodal Deception Detection
无需发送视频即可识别谎言：隐私保护的多模态欺骗检测
Abstract
Frontier multimodal models can guess whether a person is lying from a testimony video. To do so, they stream that raw face and voice to a third-party model. We ask whether the heavy media is needed at all. On the Real-life Trial Deception dataset, Whissle on-device speech and vision stack extracts a compact digest: transcript, emotion, age, gender, intent distributions, a deception intent filter, fluency and rhythm, per-frame facial behaviour, and prosody. Under speaker-independent evaluation, we report three findings. A small classifier on this digest reaches AUC 0.741, matching Gemini 2.5 Pro on full video. Handing the digest to a frontier LLM reaches AUC 0.755 with Claude Opus 4.8 at 7.8X fewer input tokens, with no media leaving the device. The reported 75% accuracy is a speaker-leakage artifact. We release code and experiments.
Chinese Translation
前沿的多模态模型可以通过证词视频判断一个人是否在说谎。为此，它们将原始的面部和声音数据传输给第三方模型。我们提出是否真的需要如此庞大的媒体数据。在现实生活中的试验欺骗数据集上，Whissle 设备端的语音和视觉堆栈提取了一个紧凑的摘要：文字记录、情感、年龄、性别、意图分布、欺骗意图过滤器、流畅度和节奏、每帧面部行为以及韵律。在说话者独立评估下，我们报告了三个发现。基于该摘要的小型分类器达到 AUC 0.741，与完整视频的 Gemini 2.5 Pro 相匹配。将摘要交给前沿的 LLM 达到 AUC 0.755，使用 Claude Opus 4.8 时输入的标记数量减少了 7.8 倍，且没有媒体数据离开设备。报告的 75% 准确率是说话者泄漏的伪影。我们发布了代码和实验。
cs.CV / 184 / 2606.22702
Modular Diffusion Models for Structured Visual Recognition
用于结构化视觉识别的模块化扩散模型
Abstract
Traditional supervised methods for structured visual recognition tasks -- such as object detection, segmentation, and scene graph generation -- often produce deterministic, fixed outputs, limiting their ability to capture the inherent uncertainty in complex visual scenes. As a consequence, such point estimates are unable to capture the prediction uncertainty (or multi modality) intrinsic to these problems, often arising from natural ambiguities (e.g., ambiguity in size of partially occluded objects, local ambiguity of exact segmentation boundary, etc.) as well as noise and sparsity of training data. To address this limitation, we present Modular Diffusion Models (MDMs), a simple and novel framework that learns a distribution over structured outputs for a given input image. MDMs decompose the diffusion process into distinct, task-specific modules, each focused on capturing a different aspect of the structured information space, such as object categories, spatial locations, and inter-object relationships. This modular design allows each component to be learned independently, with seamless integration at inference without additional training. Furthermore, the modularity of MDMs enables the diffusion process to easily operate over the heterogeneous output space common in many structured learning tasks (e.g., a continuous bounding boxes and discrete class labels). Experimental results over three distinct structured tasks -- object detection, instance segmentation, and scene graph generation -- highlight the benefits of our proposed framework.
Chinese Translation
传统的监督方法在结构化视觉识别任务中——如目标检测、分割和场景图生成——通常产生确定性、固定的输出，这限制了它们捕捉复杂视觉场景中固有不确定性的能力。因此，这种点估计无法捕捉到这些问题内在的预测不确定性（或多模态性），这往往源于自然的模糊性（例如，部分遮挡物体的大小模糊、精确分割边界的局部模糊等）以及训练数据的噪声和稀疏性。为了解决这一局限性，我们提出了模块化扩散模型（Modular Diffusion Models, MDMs），这是一个简单而新颖的框架，能够为给定输入图像学习结构化输出的分布。MDMs将扩散过程分解为不同的、特定任务的模块，每个模块专注于捕捉结构化信息空间的不同方面，如物体类别、空间位置和物体间关系。这种模块化设计使得每个组件可以独立学习，并在推理时无缝集成，而无需额外的训练。此外，MDMs的模块化使得扩散过程能够轻松地在许多结构化学习任务中常见的异质输出空间上操作（例如，连续的边界框和离散的类别标签）。在三个不同的结构化任务——目标检测、实例分割和场景图生成上的实验结果突显了我们提出的框架的优势。
cs.CV / 185 / 2606.22718
Generative Relightable Avatars
可生成的可重光照头像
Abstract
We present Generative Relightable Avatars (GRA), a person-specific method for photorealistic free-view rendering and environment-map relighting of full-body humans. We postulate that modeling fine-grained appearance details is inherently a one-to-many problem that can benefit from a generative formulation. In contrast to fully regressive relightable avatar methods, GRA follows a hybrid approach that combines controllable, physics-grounded relighting with probabilistic refinement. Starting from a tracked animated mesh, we optimize material parameters in UV-space and render a coarse relit appearance under a target HDR environment map. Next, we refine the textures with a feed-forward model to capture pose-dependent texture dynamics and illumination effects beyond simplified reflectance assumptions. Finally, a fine-tuned video-to-video diffusion model transforms the physically grounded renderings into temporally coherent, high-detail videos while preserving 3D control, with an error-recycling strategy for generating long videos. Experimental evaluations demonstrate our method's improved perceptual quality over prior relightable avatar baselines. Project Page: https://vcai.mpi-inf.mpg.de/projects/GRA/
Chinese Translation
我们提出了一种针对个体的可生成可重光照头像（Generative Relightable Avatars, GRA）方法，用于全身人类的逼真自由视角渲染和环境图重光照。我们假设建模细粒度外观细节本质上是一个一对多的问题，可以从生成性形式中受益。与完全回归的可重光照头像方法不同，GRA采用了一种混合方法，将可控的、基于物理的重光照与概率性细化相结合。从跟踪的动画网格开始，我们在UV空间中优化材料参数，并在目标HDR环境图下渲染粗略的重光照外观。接下来，我们使用前馈模型细化纹理，以捕捉依赖于姿势的纹理动态和超出简化反射假设的照明效果。最后，一个经过微调的视频到视频扩散模型将物理基础的渲染转换为时间一致的高细节视频，同时保持3D控制，并采用错误回收策略生成长视频。实验评估表明我们的方法在感知质量上优于先前的可重光照头像基线。项目页面：https://vcai.mpi-inf.mpg.de/projects/GRA/
cs.CV / 186 / 2606.22725
Interpretable Uncertainty Routing Separating Emotion Ambiguity from Distribution Shift in Facial Expression Recognition
可解释的不确定性路由：在面部表情识别中将情感模糊性与分布偏移分离
Abstract
Facial expression recognition (FER) is inherently ambiguous: human annotators frequently disagree, and models deployed in real environments face distribution shift. Crucially, these two conditions demand different downstream actions, as ambiguous in-distribution faces should be reported with their ambiguity whereas out-of-distribution inputs should be rejected. However, a single uncertainty score conflates the two. In this study, uncertainty decomposition into aleatoric and epistemic components for FER is investigated, and Uncertainty-Aware Routing (UAR), an inference-time routing mechanism that exploits the separation, is introduced. Specifically, aleatoric and epistemic uncertainties are obtained from a Deep Ensemble of fully fine-tuned DINOv2 models and are each validated against an independent external signal: aleatoric against human annotator disagreement, and epistemic against distribution shift induced by image corruptions. The proposed dual-validation protocol reveals that aleatoric recovers annotator disagreement with Spearman correlation 0.66 (95% CI: 0.64-0.68), and epistemic detects corruption-induced shifts, achieving average AUROC of 0.699 at the highest corruption severity. UAR retains approximately 1.8 times more ambiguous in-distribution faces than single-uncertainty routing at a matched out-of-distribution rejection rate. A strong label-distribution-learning baseline achieves comparable disagreement recovery but cannot separate ambiguity from shift and therefore cannot route, establishing that the value of decomposition lies in the separation enabling interpretable and differentiated action selection.
Chinese Translation
面部表情识别（FER）本质上是模糊的：人类标注者经常存在分歧，而在真实环境中部署的模型面临分布偏移。关键是，这两种情况需要不同的后续处理，因为模糊的分布内面孔应报告其模糊性，而分布外输入应被拒绝。然而，单一的不确定性评分将两者混为一谈。本研究探讨了对FER进行不确定性分解为随机性（aleatoric）和认知性（epistemic）成分，并引入了一种利用这种分离的推理时路由机制——不确定性感知路由（Uncertainty-Aware Routing, UAR）。具体而言，随机性和认知性不确定性是通过完全微调的DINOv2模型的深度集成获得的，并且分别通过独立的外部信号进行验证：随机性与人类标注者的分歧进行验证，认知性与图像损坏引起的分布偏移进行验证。所提出的双重验证协议显示，随机性与标注者分歧的斯皮尔曼相关系数为0.66（95%置信区间：0.64-0.68），而认知性检测到由损坏引起的偏移，在最高损坏严重性下实现了0.699的平均AUROC。UAR在匹配的分布外拒绝率下，保留了约1.8倍更多的模糊分布内面孔，相较于单一不确定性路由。一个强大的标签分布学习基线实现了可比的分歧恢复，但无法将模糊性与偏移分离，因此无法进行路由，证明了分解的价值在于分离使得可解释和差异化的行动选择成为可能。
cs.CV / 187 / 2606.22749
RaysUp: Ultra-light Universal Feature Upsampling via Geometry-Aware Ray Representation
RaysUp：通过几何感知光线表示实现超轻量级通用特征上采样
Abstract
Pre-trained Vision Foundation Models (VFMs) have become central to modern computer vision due to their powerful semantic representations and strong generalization ability. However, their patchified or pooled outputs are inherently low-resolution, limiting their effectiveness in tasks requiring fine-grained, pixel-level reasoning. Existing feature upsampling approaches either degrade semantic fidelity or rely on VFM-specific retraining and heavy architectures, hindering efficiency and scalability. To address these challenges, we propose RaysUp, an ultra-lightweight, task-agnostic, and VFM-agnostic feature upsampling framework that reconstructs high-resolution feature maps at arbitrary resolutions. Unlike conventional 2D interpolation or attention-based schemes, RaysUp lifts feature reconstruction into a geometry-aware ray domain. Specifically, we introduce a Spatially Decoupled Guidance Encoder for direction-aware guidance encoding, an Any-Resolution Cross-Attention mechanism for resolution-flexible reconstruction, and a novel Ray Positional Encoding (RayPE) that injects implicit 3D geometric priors via 6D Plucker ray coordinates. Finally, a Geometry-Aware Neighborhood Attention module further ensures content-adaptive bilateral aggregation while preserving geometric consistency. Extensive experiments across diverse dense prediction tasks demonstrate that RaysUp achieves state-of-the-art performance while using only 16% of the parameters of AnyUp and delivering approximately 7x faster inference. These results highlight a substantially improved accuracy-efficiency trade-off and establish RaysUp as a practical and scalable solution for universal feature upsampling. Code is available at https://github.com/MAP-RaysUp/RaysUp.
Chinese Translation
预训练视觉基础模型（VFMs）因其强大的语义表示和良好的泛化能力，已成为现代计算机视觉的核心。然而，它们的补丁化或池化输出本质上是低分辨率的，限制了在需要细粒度像素级推理的任务中的有效性。现有的特征上采样方法要么降低语义保真度，要么依赖于特定于VFM的再训练和复杂的架构，从而阻碍了效率和可扩展性。为了解决这些挑战，我们提出了RaysUp，这是一种超轻量级、任务无关和VFM无关的特征上采样框架，能够在任意分辨率下重建高分辨率特征图。与传统的二维插值或基于注意力的方案不同，RaysUp将特征重建提升到一个几何感知的光线域。具体而言，我们引入了一种空间解耦引导编码器，用于方向感知的引导编码，一种任意分辨率交叉注意力机制，用于分辨率灵活的重建，以及一种新颖的光线位置编码（RayPE），通过6D Plucker光线坐标注入隐式3D几何先验。最后，一个几何感知邻域注意力模块进一步确保内容自适应的双边聚合，同时保持几何一致性。在多种密集预测任务中的广泛实验表明，RaysUp在仅使用AnyUp 16%的参数的情况下实现了最先进的性能，并提供了大约7倍的推理速度。这些结果突显了显著改善的准确性-效率权衡，并确立了RaysUp作为通用特征上采样的实用和可扩展解决方案。代码可在 https://github.com/MAP-RaysUp/RaysUp 获取。
cs.CV / 188 / 2606.22766
READ More than What You See: Reinforcement Learning for Accurate and Coherent Audio Description Generations
超越视觉的阅读：用于准确且连贯音频描述生成的强化学习
Abstract
Audio Description aims to generate concise narrations of essential visual content in audio-visual media for blind and low-vision audiences. Existing methods either rely on prompting off-the-shelf multimodal models, which often mismatch AD style, or partially optimize training-based systems with next-token prediction, which under-explores model capacity and biases generation toward generic expressions. We present READ, the first reinforcement-learning (RL) framework for training-based AD generation. READ formulates AD as sequence-level optimization with reference-matching, length, and format rewards, and further introduces a dedicated coherence reward under context-aware supervision to promote narratively coherent descriptions. Experiments on MAD-Eval, CMD-AD, and TV-AD show that READ substantially outperforms prior methods across diverse evaluation metrics. Our results highlight RL as a promising paradigm for accurate and coherent AD generation. Our codes, models, and benchmark results will be publicly available.
Chinese Translation
音频描述旨在为盲人和低视力观众生成音视频媒体中重要视觉内容的简明叙述。现有方法要么依赖于现成的多模态模型进行提示，这往往与音频描述的风格不匹配，要么部分优化基于训练的系统，通过下一个标记预测，这在一定程度上未能充分挖掘模型的能力，并使生成偏向于通用表达。我们提出了READ，这是第一个用于基于训练的音频描述生成的强化学习（RL）框架。READ将音频描述公式化为具有参考匹配、长度和格式奖励的序列级优化，并在上下文感知监督下进一步引入专门的连贯性奖励，以促进叙事连贯的描述。在MAD-Eval、CMD-AD和TV-AD上的实验表明，READ在多种评估指标上显著优于先前的方法。我们的结果突显了强化学习作为准确且连贯的音频描述生成的有前景的范式。我们的代码、模型和基准结果将公开提供。
cs.CV / 189 / 2606.22772
LoCC: Detection and Localization of Lip-Syncing Deepfakes via Counterfactual Frame Consistency
LoCC：通过反事实帧一致性检测和定位口型同步深伪造
Abstract
Lip-syncing deepfakes are among the most challenging forms of manipulated media because their artifacts are localized almost exclusively to the mouth region and evolve dynamically over time. Detecting such deepfakes requires precise temporal and spatial modeling of lip motion. In this paper, we propose LoCC, a novel detection framework that performs fine-grained detection and localization of lip-syncing deepfakes at both segment and frame levels. Unlike prior approaches that analyze videos holistically, our method evaluates whether each frame aligns with a counterfactual estimate generated from its temporal neighbors. Real videos exhibit strong and stable consistency, whereas lip-sync deepfakes introduce localized inconsistencies. Following a teacher-student learning paradigm, our model effectively captures these frame-level discrepancies and achieves superior performance over state-of-the-art methods on multiple benchmark lip-syncing deepfake datasets, including LAV-DF, AVDF1M, FakeAVCeleb, and KODF, and generalizes well across compression levels and datasets.
Chinese Translation
口型同步深伪造是最具挑战性的媒体操控形式之一，因为其伪影几乎完全局限于嘴部区域，并且随着时间动态演变。检测此类深伪造需要对嘴部运动进行精确的时间和空间建模。本文提出了LoCC，一种新颖的检测框架，能够在片段和帧级别上对口型同步深伪造进行细粒度的检测和定位。与以往从整体上分析视频的方法不同，我们的方法评估每一帧是否与其时间邻域生成的反事实估计相一致。真实视频表现出强而稳定的一致性，而口型同步深伪造则引入了局部不一致性。遵循教师-学生学习范式，我们的模型有效捕捉这些帧级别的不一致性，并在多个基准口型同步深伪造数据集上（包括LAV-DF、AVDF1M、FakeAVCeleb和KODF）实现了优于现有最先进方法的性能，并在不同压缩水平和数据集上具有良好的泛化能力。
cs.CV / 190 / 2606.22787
Visual Geometry Transformer in the Wild: Distractor-Free 3D Reconstruction
野外视觉几何变换器：无干扰的三维重建
Abstract
Current end-to-end multi-view 3D reconstruction methods achieve impressive results, but rely on a restrictive static assumption: the scenes is entire distractor-free with perfect cross-view geometry. This reliance on idealized inputs causes even the most advanced methods to fail in real-world settings, where transient distractors and occlusions present. To address this, we propose Visual Geometry Transformer in the Wild (VGTW), an end-to-end framework for robust reconstruction from inconsistent views. At its core, we isolate and suppress distractor-affected regions while preserving the consistent components across views. Specifically, we introduce a Distractor-aware Training (DAT) strategy that separates clean features from distractor-contaminated ones in the attention mechanism while enforcing feature consistency across images. To enable this, we train the model with an auxiliary mask prediction head, using supervision from a new dataset we collected with pixel-level distractor masks. The resulting VGTW model is a feed-forward network that directly outputs clean, distractor-free point clouds. Remarkably, it requires no additional 3D supervision, remains computationally efficient, and is compatible with existing pipelines. Extensive experiments validate our approach, demonstrating state-of-the-art performance and robust generalization in diverse, real-world scenarios.
Chinese Translation
当前的端到端多视角三维重建方法取得了令人印象深刻的成果，但依赖于一个限制性的静态假设：场景完全没有干扰且具有完美的视角几何。这种对理想化输入的依赖导致即使是最先进的方法在现实世界环境中也会失败，因为那里存在瞬态干扰和遮挡。为了解决这个问题，我们提出了野外视觉几何变换器（Visual Geometry Transformer in the Wild, VGTW），这是一个用于从不一致视图中进行稳健重建的端到端框架。在其核心，我们隔离并抑制受干扰的区域，同时保留视图间的一致组件。具体而言，我们引入了一种干扰感知训练（Distractor-aware Training, DAT）策略，该策略在注意机制中将干净特征与受干扰特征分离，同时强制图像间的特征一致性。为此，我们通过一个辅助的掩码预测头训练模型，利用我们收集的新数据集中的像素级干扰掩码进行监督。最终得到的VGTW模型是一个前馈网络，能够直接输出干净的、无干扰的点云。值得注意的是，它不需要额外的三维监督，计算效率高，并且与现有的工作流程兼容。大量实验验证了我们的方法，展示了在多样化的现实场景中具有最先进的性能和稳健的泛化能力。
cs.CV / 191 / 2606.22801
Learning Adaptive Dynamical Features via Multi-$\tau$ Liquid-Mamba for All-in-one Image Restoration
通过多-$ au$液态曼巴学习自适应动态特征以实现一体化图像恢复
Abstract
Image restoration aims to recover high-quality images from degraded observations. Recent Mamba-based image restoration models have demonstrated strong potential in modeling long-range dependencies with linear complexity. However, most existing designs still rely on a single state-evolution timescale, which limits their adaptability to spatially heterogeneous and task-dependent degradation patterns in all-in-one image restoration. In this paper, we propose Multi-$\tau$ Liquid-Mamba, an adaptive state space module that introduces input-conditioned multi-timescale liquid discretization into selective state space modeling. Instead of changing the overall selective scan pipeline, the proposed module modulates the effective discretization steps of multiple dynamical branches and adaptively fuses their responses according to degradation-aware gating weights. This design allows the model to capture both fast-varying local details and slowly evolving global structures while preserving the linear scaling property of Mamba with respect to sequence length. Importantly, Multi-$\tau$ Liquid-Mamba modulates the effective transition dynamics while preserving the original selective parameterization and hardware-efficient selective scan mechanism, making it a plug-and-play module that can be seamlessly integrated into existing Mamba-based architectures. Built upon this framework, we develop a Multi-$\tau$ Liquid-Mamba Image Restoration Network (MLMIR) for all-in-one image restoration. Extensive experiments on a wide range of restoration benchmarks demonstrate that MLMIR consistently achieves state-of-the-art performance in all-in-one image restoration while remaining highly competitive in task-aligned restoration settings.
Chinese Translation
图像恢复旨在从退化的观测中恢复高质量图像。最近基于曼巴的图像恢复模型在以线性复杂度建模长程依赖性方面展现了强大的潜力。然而，大多数现有设计仍依赖于单一的状态演变时间尺度，这限制了它们在一体化图像恢复中对空间异质性和任务依赖性退化模式的适应性。本文提出了多-$ au$液态曼巴（Multi-$ au$ Liquid-Mamba），这是一种自适应状态空间模块，它将输入条件的多时间尺度液态离散化引入选择性状态空间建模中。该模块并未改变整体选择性扫描管道，而是调节多个动态分支的有效离散化步骤，并根据退化感知的门控权重自适应融合它们的响应。这一设计使模型能够捕捉快速变化的局部细节和缓慢演变的全局结构，同时保持曼巴在序列长度上的线性扩展特性。重要的是，多-$ au$液态曼巴调节有效的转移动态，同时保留原有的选择性参数化和硬件高效的选择性扫描机制，使其成为一个即插即用的模块，可以无缝集成到现有的基于曼巴的架构中。在此框架基础上，我们开发了多-$ au$液态曼巴图像恢复网络（MLMIR）以实现一体化图像恢复。在广泛的恢复基准测试中的大量实验表明，MLMIR在一体化图像恢复中始终实现了最先进的性能，同时在任务对齐的恢复设置中保持高度竞争力。
cs.CV / 192 / 2606.22804
CoVStream: Edge-Cloud Collaboration for Understanding of Long Video Streams
CoVStream：用于理解长视频流的边缘-云协作
Abstract
Long, continuous video streams are an increasingly critical driver of multimedia intelligence. Existing efforts often handle long videos with a sample-encode-reason approach using large models. However, they overlook a crucial deployment fact: the stream is often produced by computationally constrained devices. This forces an untenable compromise: cloud offloading unlocks strong reasoning but incurs prohibitive bandwidth overhead, while on-device processing remains limited by edge hardware capacity. Therefore, we propose CoVStream, the first edge-cloud collaborative framework for understanding long video streams. The edge node distills raw video streams into compact visual features and semantic captions for transmission to the cloud, minimizing bandwidth costs, while the cloud server integrates this data into an entity graph and global visual context, activating the heavy reasoning model only when a user query arrives. Experiments on VideoMME-Long, LVBench, and RTV-Bench show that CoVStream reduces bandwidth usage by 87.6% while retaining 99.2% of the cloud baseline accuracy on LVBench.
Chinese Translation
长时间、连续的视频流在多媒体智能中日益成为一个关键驱动因素。现有的研究通常采用样本编码推理的方法来处理长视频，并使用大型模型。然而，它们忽视了一个重要的部署事实：视频流通常由计算能力受限的设备生成。这迫使我们在云卸载和设备处理之间做出不可接受的妥协：云卸载能够解锁强大的推理能力，但会产生高昂的带宽开销，而设备处理则受到边缘硬件能力的限制。因此，我们提出了CoVStream，这是第一个用于理解长视频流的边缘-云协作框架。边缘节点将原始视频流提炼成紧凑的视觉特征和语义标题，以便传输到云端，从而最小化带宽成本，而云服务器则将这些数据整合到实体图和全球视觉上下文中，仅在用户查询到达时激活重推理模型。在VideoMME-Long、LVBench和RTV-Bench上的实验表明，CoVStream将带宽使用量减少了87.6%，同时在LVBench上保留了99.2%的云基准准确率。
cs.CV / 193 / 2606.22806
Policy-as-Data: Learning Generalizable HOI Diffusion Models from Simulated Physics
政策即数据：从模拟物理中学习可泛化的人物-物体交互扩散模型
Abstract
Synthesizing realistic Human-Object Interactions (HOI) is critical for creating embodied avatars and functional virtual environments. However, current data-driven approaches primarily rely on motion capture datasets, which are expensive to scale and limited in functional diversity. Models trained with these datasets fail to generalize to unseen objects and maintain physical consistency over long horizons. In this paper, we propose a novel framework that leverages a physics simulator to overcome the data-scarcity bottleneck in HOI generation. Specifically, we propose a scalable pipeline, called \ours, which leverages policies trained with reinforcement learning in a physics simulator for task-oriented data generation and trains a generative model on the augmented dataset for generalizable HOI generation. To seamlessly utilize the synthetic data, we introduce a coarse-to-fine retargeting process that bridges the representation gap between the simplified model used in physics simulator and the standard parametric body models required for generative training. Validated through comprehensive experiments, our method demonstrates enhanced generalization to unseen objects and the capability of long-horizon generation, while exhibiting greater dynamic diversity and physical plausibility.
Chinese Translation
合成真实的人物-物体交互（HOI）对于创建具身化虚拟角色和功能性虚拟环境至关重要。然而，当前的数据驱动方法主要依赖于运动捕捉数据集，这些数据集在扩展性上成本高昂且功能多样性有限。使用这些数据集训练的模型在面对未见对象时无法泛化，并且在长时间范围内保持物理一致性。本文提出了一种新颖的框架，利用物理模拟器克服HOI生成中的数据稀缺瓶颈。具体而言，我们提出了一种可扩展的管道，称为 extit{ours}，该管道利用在物理模拟器中通过强化学习训练的策略进行任务导向的数据生成，并在增强数据集上训练生成模型以实现可泛化的HOI生成。为了无缝利用合成数据，我们引入了一种粗到细的重定向过程，弥合物理模拟器中使用的简化模型与生成训练所需的标准参数化身体模型之间的表示差距。通过全面的实验验证，我们的方法展示了对未见对象的增强泛化能力和长时间范围生成的能力，同时表现出更大的动态多样性和物理合理性。
cs.CV / 194 / 2606.22829
DBT-Bleed: Dual-Branch Temporal Modeling with Key-Frame Selection for Surgical Bleeding Detection
DBT-Bleed：基于双分支时间建模与关键帧选择的外科出血检测
Abstract
Intraoperative Adverse Events (IAEs) detection is critical for improving surgical safety, with bleeding being among the most frequent events across many surgery types. Existing methods struggle to distinguish bleeding IAE from visually similar residual blood due to limited temporal reasoning. Moreover, modeling long surgical videos while preserving fine-grained temporal dynamics remains computationally challenging. We propose DBT-Bleed, a dual-branch multi-scale temporal modeling framework disentangling bleeding and normal representations using layer-wise temporal adapters for short- and long-term bleeding progression. To efficiently process long surgical videos without sacrificing fine-grained temporal information, we introduce HiRED, a Hierarchical Entropy-Driven frame selection strategy that retains temporally informative segments while removing redundancy. Experiments on the MultiBypass dataset demonstrate gains of 6.53% in F1, 5.62% in Recall and 9% in MCC values for bleeding IAE detection, consistently outperforming video-level baselines. Additionally, we evaluate cross-procedure generalization on a newly curated dataset from a different surgical procedure type, where DBT-Bleed demonstrates robust transferability by achieving gain of 6% in F1 and 8% in MCC under zero-shot setting. To support this evaluation, we introduce EndoPit-IAE, an Endonasal Pituitary Surgery dataset annotated for IAEs, representing the first IAE-annotated dataset in neurosurgery. Code will be made publicly available upon acceptance.
Chinese Translation
术中不良事件（IAEs）的检测对提高外科手术安全性至关重要，其中出血是多种手术类型中最常见的事件之一。现有方法由于时间推理能力有限，难以区分出血IAE与视觉上相似的残留血液。此外，建模长时间的手术视频并保持细粒度的时间动态仍然面临计算挑战。我们提出了DBT-Bleed，一种双分支多尺度时间建模框架，通过层级时间适配器解耦出血和正常表现，以处理短期和长期出血进展。为了高效处理长时间手术视频而不牺牲细粒度的时间信息，我们引入了HiRED，一种基于层次熵驱动的帧选择策略，保留时间上信息丰富的片段，同时去除冗余。对MultiBypass数据集的实验表明，在出血IAE检测中，F1值提高了6.53%，召回率提高了5.62%，MCC值提高了9%，始终优于视频级基线。此外，我们在一个新创建的不同手术类型的数据集上评估了跨程序的泛化能力，DBT-Bleed在零样本设置下表现出强大的可迁移性，F1值提高了6%，MCC值提高了8%。为了支持这一评估，我们引入了EndoPit-IAE，这是一个针对IAEs进行注释的内鼻垂体手术数据集，代表了神经外科领域首个IAE注释数据集。代码将在接受后公开发布。
cs.CV / 195 / 2606.22834
Homographic Navigation: Geometry-Driven Camera Guidance for Deterministic Planar Capture
单应性导航：基于几何的相机引导用于确定性平面捕捉
Abstract
We present homographic navigation, a geometry-centric framework for guiding camera acquisition toward precise capture of planar regions. Rather than treating homography as an output, we use it as an organizing variable that unifies learning, alignment, and evaluation. From a single annotated reference image, we generate unlimited synthetic training data via homographic augmentation and train a single-shot model for joint recognition and localization of multiple artifacts (physical objects with a rectangular planar target) through sparse keypoint prediction. To address precision under limited model input resolution, we introduce a two-pass inference scheme with global detection followed by localized refinement, and a Stable Warp training strategy that significantly improves accuracy, particularly in the high-precision regime. The model also predicts confidence estimates per predicted keypoint and per the whole sample. Experimental results demonstrate that accurate planar alignment can be achieved from minimal supervision, providing a foundation for geometry-driven camera guidance and future learning from in-the-wild video data.
Chinese Translation
我们提出了单应性导航，这是一种以几何为中心的框架，用于引导相机获取以精确捕捉平面区域。我们不将单应性视为输出，而是将其作为一个组织变量，统一学习、对齐和评估。从单个标注的参考图像出发，我们通过单应性增强生成无限的合成训练数据，并训练一个单次模型，通过稀疏关键点预测实现多个工件（具有矩形平面目标的物理对象）的联合识别和定位。为了解决在有限模型输入分辨率下的精度问题，我们引入了一种双重推理方案，首先进行全局检测，然后进行局部细化，以及一种稳定变形（Stable Warp）训练策略，显著提高了准确性，尤其是在高精度范围内。该模型还预测每个预测关键点和整个样本的置信度估计。实验结果表明，从最小的监督中可以实现准确的平面对齐，为基于几何的相机引导和未来从野外视频数据中学习奠定了基础。
cs.CV / 196 / 2606.22835
OrthoMotion:Disentangling Camera and Subject Motion via Geometry Semantics Orthogonal Attention
OrthoMotion：通过几何语义正交注意力解耦相机与主体运动
Abstract
Controllable video generation demands independent command of the camera and the subject, yet 2D conditioning entangles them: camera- and object-induced optical flow share the same inverse-depth (1/Z) scaling and cannot be separated from image evidence alone. We first prove that this entanglement is representational, not architectural -- the 2D camera/object split is a non-identifiable inverse problem -- and therefore reframe decoupling as a question of operator design. We resolve it at the level of the attention operator. OrthoMotion routes camera motion into a geometric channel, a norm-preserving rotation of the rotary position embedding (RoPE) phase, and subject motion into a semantic channel, a gated value injection in cross-attention. Because these sub-operators are algebraically complementary -- a rotation versus a translation of the affine action on tokens -- a lightweight decoupling regularizer provably drives their response subspaces to orthogonality, so the two controls stop interfering. To our knowledge OrthoMotion is the first method to guarantee disentanglement by construction rather than hope for it to emerge. It attains state-of-the-art camera and subject accuracy at once while minimizing cross-talk, which we quantify with a new Cross-Talk Error (CTE) metric, cutting cross-talk by more than 2.4x with no loss in fidelity and generalizing across backbones.
Chinese Translation
可控视频生成要求独立控制相机和主体，但二维条件化将二者纠缠在一起：相机和物体引起的光流共享相同的逆深度（1/Z）缩放，无法仅通过图像证据分离。我们首先证明这种纠缠是表征性的，而非架构性的——二维相机/物体分离是一个不可识别的逆问题——因此将解耦重新框定为算子设计的问题。我们在注意力算子的层面上解决了这个问题。OrthoMotion 将相机运动引入几何通道，即旋转位置嵌入（RoPE）相位的范数保持旋转，将主体运动引入语义通道，即在交叉注意力中的门控值注入。由于这些子算子在代数上是互补的——对标记的仿射作用的旋转与平移——一种轻量级的解耦正则化器可证明地将它们的响应子空间驱动到正交，从而使得两种控制停止相互干扰。据我们所知，OrthoMotion 是第一种通过构造保证解耦的方法，而不是寄希望于其自发出现。它在实现相机和主体的最先进精度的同时，最小化了交叉干扰，我们通过一种新的交叉干扰误差（CTE）指标量化了这一点，交叉干扰减少超过 2.4 倍，且没有损失保真度，并在不同的骨干网络上具有良好的泛化能力。
cs.CV / 197 / 2606.22856
G-MASt3R-SfM: Graph-based View Pruning and Multi-stage Optimization for Robust SfM
G-MASt3R-SfM：基于图的视图修剪与多阶段优化以增强SfM的鲁棒性
Abstract
Structure from Motion (SfM) is essential for multi-view 3D reconstruction, however, its accuracy heavily relies on the accuracy of image matching. While the recent correspondence matching method, MASt3R, enables robust matching even under challenging conditions, it tends to generate incorrect correspondences for non-overlapping image pairs. Consequently, existing SfM methods using MASt3R, such as MASt3R-SfM, suffer from significant degradation in pose estimation accuracy as they incorporate these unreliable matches directly into optimization. To address this issue, we propose G-MASt3R-SfM, a novel SfM pipeline that enhances robustness through two key modules. First, the Graph-based View Pruning (GVP) module constructs a scene graph from matching confidence and geometrically prunes outlier views. Second, the Multi-Stage Optimization (MSO) module progressively refines camera parameters by expanding the optimization scope from local consistency to the global consistency. Experiments on the ETH3D dataset demonstrate that our method achieves state-of-the-art accuracy in both camera pose estimation and 3D reconstruction, effectively suppressing noise caused by outliers.
Chinese Translation
运动结构（SfM）对于多视图三维重建至关重要，然而，其准确性在很大程度上依赖于图像匹配的准确性。尽管最近的对应匹配方法MASt3R能够在困难条件下实现鲁棒匹配，但它往往会为非重叠图像对生成不正确的对应关系。因此，使用MASt3R的现有SfM方法，如MASt3R-SfM，在姿态估计准确性上遭受显著下降，因为它们直接将这些不可靠的匹配纳入优化。为了解决这一问题，我们提出了G-MASt3R-SfM，这是一种新颖的SfM管道，通过两个关键模块增强鲁棒性。首先，基于图的视图修剪（GVP）模块根据匹配置信度构建场景图，并几何上修剪掉离群视图。其次，多阶段优化（MSO）模块通过将优化范围从局部一致性扩展到全局一致性，逐步细化相机参数。在ETH3D数据集上的实验表明，我们的方法在相机姿态估计和三维重建方面实现了最先进的准确性，有效抑制了由离群值引起的噪声。
cs.CV / 198 / 2606.22862
Chains That See, Answers That Don't: A Multi-Aspect Evaluation Recipe for Forced Chain-of-Thought on Video-MME
可见的链条，不可见的答案：视频多模态评估中强制思维链的多维评估方法
Abstract
Forced chain-of-thought (CoT) is widely assumed to make vision-language models more reliable on video question answering. We propose a small three-probe evaluation recipe to test that assumption: paired accuracy across direct, CoT, answer-first, and no-video conditions; a counterfactual video-swap diagnostic over the CoT chains; and a four-rung visual-degradation ladder. Each probe is reported under both a strict and a permissive regex scorer, with multiplicity correction over a manuscript-declared primary family. Applied to Qwen2.5-VL on Video-MME subsets, the recipe returns a two-part finding. The CoT chains are strongly video-conditioned: swapping the input video collapses chain overlap and flips most final letters, the opposite of what a "boilerplate-chain" null would predict. Yet on the same data, forced CoT does not improve MCQ accuracy, and on the smaller 7B model it produces a small but statistically supported drop under a post-hoc primary scorer choice. We do not claim this generalizes beyond the Qwen2.5-VL / Video-MME instantiation; the raw responses and a single recomputation script will be released with the supplementary material so every number can be re-derived.
Chinese Translation
强制思维链（CoT）被广泛认为可以提高视觉语言模型在视频问答中的可靠性。我们提出了一种小型的三探针评估方法来检验这一假设：在直接、CoT、答案优先和无视频条件下的配对准确性；对CoT链的反事实视频交换诊断；以及一个四级视觉退化梯度。每个探针在严格和宽松的正则表达式评分器下报告，并对手稿声明的主要家族进行了多重性校正。应用于Qwen2.5-VL在视频多模态评估子集上的结果显示出两个主要发现。CoT链条对视频条件的依赖性很强：交换输入视频会导致链条重叠崩溃，并翻转大多数最终字母，这与“模板链”无效的预测正好相反。然而，在相同的数据上，强制CoT并未提高多项选择题（MCQ）的准确性，并且在较小的7B模型上，在后验主要评分器选择下，产生了一个小但统计上显著的下降。我们并不声称这一结果可以推广到Qwen2.5-VL / 视频多模态评估的其他实例；原始响应和单个重新计算脚本将与补充材料一起发布，以便每个数字都可以重新推导。
cs.CV / 199 / 2606.22870
VideoLatent: Video-Language Learning via Latent Self-Forcing
VideoLatent：通过潜在自强进行视频-语言学习
Abstract
Recent advancements in chain-of-thought (CoT) reasoning have shown promise in enhancing video understanding and reasoning capabilities of multimodal large language models (MLLMs). However, existing CoT-based MLLMs require labor-intensive CoT annotations and incur substantial training and inference overhead. While visual latent reasoning has emerged as a more efficient alternative, existing methods primarily focus on image tasks and heavily rely on additional supervision signals for visual latent generation (e.g., CoT traces, auxiliary images, or fine-grained annotations), limiting their scalability and transferability to video tasks. To bridge this gap, we introduce VideoLatent, a novel MLLM equipped with a latent injection module tailored for video understanding and reasoning. Specifically, VideoLatent learns to perform visual latent reasoning using a new latent self-forcing training paradigm, which comprises latent alignment and latent diversity objectives, and relies solely on standard video-question-answer triplets. Extensive experiments across 14 benchmarks demonstrate that our model consistently outperforms existing standard and latent MLLMs on general video understanding and complex video reasoning. Compared with Video-R1, our VideoLatent achieves superior computational efficiency, reducing training/inference overhead by $\sim$6$\times$/$\sim$68$\times$. Moreover, experiments demonstrate that our method has strong generalizability to different MLLM backbones and different model scales.
Chinese Translation
最近在思维链（CoT）推理方面的进展显示出增强多模态大型语言模型（MLLMs）视频理解和推理能力的潜力。然而，现有的基于CoT的MLLMs需要劳动密集型的CoT注释，并且在训练和推理过程中产生大量开销。尽管视觉潜在推理作为一种更高效的替代方案已逐渐出现，但现有方法主要集中于图像任务，并且在视觉潜在生成方面严重依赖额外的监督信号（例如，CoT痕迹、辅助图像或细粒度注释），这限制了它们在视频任务中的可扩展性和可迁移性。为了解决这一问题，我们提出了VideoLatent，一种新型的MLLM，配备了针对视频理解和推理的潜在注入模块。具体而言，VideoLatent学习使用一种新的潜在自强训练范式进行视觉潜在推理，该范式包括潜在对齐和潜在多样性目标，并且仅依赖于标准的视频-问题-答案三元组。在14个基准测试中的广泛实验表明，我们的模型在一般视频理解和复杂视频推理方面始终优于现有的标准和潜在MLLMs。与Video-R1相比，我们的VideoLatent在计算效率上表现更佳，将训练/推理开销减少了约6倍和68倍。此外，实验表明我们的方法对不同的MLLM骨干和不同的模型规模具有很强的可泛化性。
cs.CV / 200 / 2606.22872
Fursee: Hybrid YOLO-DINOv3 Framework for Fursuit Identity Retrieval and Clustering
Fursee：用于毛皮服装身份检索与聚类的混合YOLO-DINOv3框架
Abstract
Global furry conventions produce massive fursuit photographs, while manual sorting brings heavy labor costs and calls for automatic identity retrieval and clustering solutions. General multimodal models lack dedicated optimization for complex fursuit scenes, and no public benchmark dataset exists for this task. To fill this gap, we build a specialized fursuit image dataset and present a three-stage hybrid pipeline Fursee for fursuit identity retrieval and clustering. First, YOLO detects and crops high-resolution fursuit head patches to improve localization of small and overlapping targets. Second, ArcFace optimizes DINOv3 embeddings to enlarge angular separation between different identities on the feature hypersphere. Third, DBSCAN performs unsupervised clustering, with silhouette-coefficient-driven search automatically selecting optimal hyperparameters rather than fixed manual radius. Retrieval and clustering experiments verify that our pipeline outperforms mainstream multimodal models including GPT5.5, Claude Opus 4.8 and Qwen3.7-Plus on all evaluation metrics, achieving competitive performance for fursuit head retrieval and grouping.
Chinese Translation
全球的毛皮文化大会产生了大量的毛皮服装照片，而手动排序带来了巨大的劳动成本，因此迫切需要自动化的身份检索和聚类解决方案。现有的通用多模态模型缺乏针对复杂毛皮服装场景的专门优化，并且目前没有公开的基准数据集可供该任务使用。为填补这一空白，我们构建了一个专门的毛皮服装图像数据集，并提出了一个三阶段的混合管道Fursee，用于毛皮服装身份检索和聚类。首先，YOLO检测并裁剪高分辨率的毛皮服装头部图像，以提高小型和重叠目标的定位精度。其次，ArcFace优化DINOv3嵌入，以增大特征超球面上不同身份之间的角度分离。第三，DBSCAN执行无监督聚类，基于轮廓系数驱动的搜索自动选择最佳超参数，而不是固定的手动半径。检索和聚类实验验证了我们的管道在所有评估指标上均优于主流多模态模型，包括GPT5.5、Claude Opus 4.8和Qwen3.7-Plus，在毛皮服装头部检索和分组方面取得了具有竞争力的表现。
cs.CV / 201 / 2606.22873
SingGuard: A Policy-Adaptive Multimodal LLM Guardrail with Dynamic Reasoning
SingGuard：一种具有动态推理的政策自适应多模态 LLM 保护机制
Abstract
Vision-language models (VLMs) are increasingly deployed in consumer, medical, financial, and enterprise applications. This broad deployment expands the safety surface: risks can arise from multimodal question answering, assistant responses, and cross-modal composition, while moderation policies may vary across products, regions, and deployment stages. Most existing guardrails either rely on fixed taxonomies or target only a narrow set of interaction settings, which limits their adaptability when safety rules change at deployment time. We present \textbf{SingGuard}, a policy-adaptive multimodal guardrail model family for safety assessment in multimodal conversations. SingGuard treats the active policy as a runtime input: given natural-language rules, it checks the target content against the active policy rule by rule and predicts both the safety label and the triggered rule. To balance efficiency and interpretability, SingGuard supports fast, hybrid, and slow inference regimes along a fast-to-slow reasoning spectrum, ranging from direct safety judgments to policy-grounded deliberation. We further optimize this behavior with fast--slow decoupled reinforcement learning. We also introduce \textbf{SingGuard-Bench}, a multimodal guardrail benchmark with 56{,}340 examples spanning 80+ fine-grained risk types across multimodal QA, adversarial attack, and dynamic-rule evaluation settings, including cross-modal joint-risk cases where each modality is harmless in isolation but their composition implies unsafe intent. Across six benchmark families (35 datasets), SingGuard achieves state-of-the-art average F1 in every family. Dynamic-rule evaluation further shows improved policy-following accuracy from 0.6465 to 0.7415 under runtime policy shifts. Our code is available at https://github.com/inclusionAI/Sing-Guard.
Chinese Translation
视觉语言模型（VLMs）在消费、医疗、金融和企业应用中越来越多地被部署。这种广泛的部署扩大了安全风险面：多模态问答、助手响应和跨模态组合可能带来风险，而不同产品、地区和部署阶段的监管政策可能各不相同。现有的大多数保护机制要么依赖于固定的分类法，要么仅针对狭窄的交互设置，这限制了它们在部署时安全规则变化时的适应性。我们提出了 extbf{SingGuard}，一种政策自适应的多模态保护机制模型系列，用于多模态对话中的安全评估。SingGuard 将活动政策视为运行时输入：根据自然语言规则，它逐条检查目标内容与活动政策的符合性，并预测安全标签和触发的规则。为了平衡效率和可解释性，SingGuard 支持沿着快速到缓慢推理谱的快速、混合和缓慢推理模式，从直接的安全判断到基于政策的深思熟虑。我们进一步通过快速-缓慢解耦的强化学习优化这一行为。我们还引入了 extbf{SingGuard-Bench}，这是一个包含 56,340 个示例的多模态保护基准，涵盖了 80 多种细粒度风险类型，涉及多模态问答、对抗攻击和动态规则评估设置，包括跨模态联合风险案例，其中每种模态在孤立状态下是无害的，但它们的组合暗示了不安全的意图。在六个基准系列（35 个数据集）中，SingGuard 在每个系列中都实现了最先进的平均 F1 分数。动态规则评估进一步显示，在运行时政策变化下，政策遵循准确性从 0.6465 提高到 0.7415。我们的代码可在 https://github.com/inclusionAI/Sing-Guard 获取。
cs.CV / 202 / 2606.22875
FedOT: Ownership Verification and Leakage Tracing via Watermarks for Federated LDMs
FedOT：通过水印实现联邦LDM的所有权验证与泄露追踪
Abstract
Training Latent Diffusion Models (LDMs) within Federated Learning (FL) has attracted increasing attention due to its ability to combine the powerful generative capacity of LDMs with the privacy-preserving properties of FL. However, FL requires sharing the global model with multiple participants, which risks unauthorized model distribution or resale by malicious clients. While an intuitive approach is to adopt existing VAE-based watermarking techniques for LDMs in FL, this strategy falls short in addressing such threats due to two fundamental challenges: (1) Existing methods support ownership verification but lack the ability to trace model leakage to a specific malicious client; (2) VAE-based watermarks are vulnerable, as they can be removed simply by replacing the decoder with a clean counterpart. In this paper, we propose FedOT, the first framework for ownership verification and leakage tracing in federated LDMs. Specifically, to address the first challenge, we design a chunked watermark, where the first part is for ownership verification, and the second part is used for client identification. Furthermore, to overcome the second challenge and secure the model against VAE replacement attack, we introduce Latent Vector Transformation (LVT), which strengthens the connection between the VAE and U-Net latent spaces by modifying the original latent distribution of the VAE. Consequently, any attempt to replace the VAE for watermark removal leads to significant image quality degradation, making the LDM model unusable. Extensive experiments demonstrate that FedOT achieves superior performance in both ownership verification and traceability. Project page: https://spyzixuan.github.io/FedOT/.
Chinese Translation
在联邦学习（FL）中训练潜在扩散模型（LDM）因其将LDM强大的生成能力与FL的隐私保护特性相结合而受到越来越多的关注。然而，FL要求与多个参与者共享全局模型，这使得恶意客户端存在未经授权的模型分发或转售的风险。虽然采用现有的基于变分自编码器（VAE）的水印技术对LDM进行保护是一种直观的方法，但由于两个基本挑战，这一策略未能有效应对这些威胁：（1）现有方法支持所有权验证，但缺乏将模型泄露追踪到特定恶意客户端的能力；（2）基于VAE的水印易受攻击，因为只需用干净的解码器替换即可去除水印。本文提出了FedOT，这是第一个用于联邦LDM的所有权验证与泄露追踪的框架。具体而言，为了解决第一个挑战，我们设计了一种分块水印，其中第一部分用于所有权验证，第二部分用于客户端识别。此外，为了克服第二个挑战并保护模型免受VAE替换攻击，我们引入了潜在向量变换（Latent Vector Transformation, LVT），通过修改VAE的原始潜在分布，增强了VAE与U-Net潜在空间之间的联系。因此，任何试图替换VAE以去除水印的行为都会导致显著的图像质量下降，使得LDM模型无法使用。大量实验表明，FedOT在所有权验证和可追溯性方面均表现出优越的性能。项目页面：https://spyzixuan.github.io/FedOT/
cs.CV / 203 / 2606.22876
Full-Body Golf Swing Kinematic Reconstruction From a Smartwatch IMU
基于智能手表IMU的全身高尔夫挥杆运动学重建
Abstract
Quantitative measurement of the golf swing is critical for evaluating technique and enabling individualized feedback. However, existing methods are impractical to use on the golf course: optical motion capture is laboratory-bound, camera-based methods require impractical camera placement, and multi-sensor inertial measurement unit (IMU) systems require multi-segment setup and calibration. We thus propose a single wrist-worn IMU approach for estimating full-body joint angles during golf swings. The proposed Wrist-IMU Temporal Kinematic Network (WIT-KinNet) leverages modality-specific IMU embeddings and temporal kinematic encoding to learn wrist-to-body motion dependencies and estimate full-body joint angles during golf swings. Thirty-six golfers spanning beginner and skilled players, performed full, half, and quarter swings using seven club types: driver, 3-wood, 5-hybrid, 5-iron, 7-iron, 9-iron, and sand wedge. The proposed WIT-KinNet was evaluated under subject-wise cross-validation using synchronized smartwatch IMU data and ground-truth kinematics derived from an optical motion capture system. The proposed approach achieved a mean absolute error of 8.11 $\pm$ 1.84$^\circ$ across full-body joint angles. High temporal correlation was observed for pelvic rotation and upper torso rotation (r = 0.98 and 0.97, respectively), with X-factor and S-factor also showing strong correlation (r = 0.96 and 0.96). Linear mixed-effects models of the error revealed that swing amplitude, skill level, and club type all significantly affected measurement differences (p $<$ 0.05). The results establish the first single wrist-worn IMU approach for estimating full-body golf swing kinematics, enabling practical swing analysis during real gameplay.
Chinese Translation
高尔夫挥杆的定量测量对于评估技术和提供个性化反馈至关重要。然而，现有的方法在高尔夫球场上使用不够实用：光学运动捕捉受限于实验室，基于摄像头的方法需要不切实际的摄像头布置，而多传感器惯性测量单元（IMU）系统则需要多段设置和校准。因此，我们提出了一种单一腕部佩戴IMU的方法，用于估计高尔夫挥杆过程中的全身关节角度。所提出的腕部IMU时序运动学网络（Wrist-IMU Temporal Kinematic Network，WIT-KinNet）利用特定模态的IMU嵌入和时序运动学编码，学习腕部与身体运动之间的依赖关系，并在高尔夫挥杆过程中估计全身关节角度。三十六名高尔夫球手，包括初学者和熟练球员，使用七种球杆类型（驾驶杆、3号木杆、5号混合杆、5号铁杆、7号铁杆、9号铁杆和沙坑杆）进行了全挥杆、半挥杆和四分之一挥杆的测试。所提出的WIT-KinNet在使用同步智能手表IMU数据和从光学运动捕捉系统获得的真实运动学数据进行受试者交叉验证下进行了评估。该方法在全身关节角度的平均绝对误差为8.11 ± 1.84°。观察到骨盆旋转和上半身旋转之间具有高时间相关性（分别为r = 0.98和0.97），X因子和S因子也显示出强相关性（r = 0.96和0.96）。误差的线性混合效应模型显示，挥杆幅度、技能水平和球杆类型均显著影响测量差异（p < 0.05）。结果建立了首个单一腕部佩戴IMU的方法，用于估计全身高尔夫挥杆运动学，从而实现了在实际游戏中进行实用的挥杆分析。
cs.CV / 204 / 2606.22890
PHOEBI: An Open-World Benchmark for Bacterial Identification in Phase-Contrast Microscopy
PHOEBI：一种用于相差显微镜下细菌识别的开放世界基准
Abstract
Optical microscopy enables rapid, label-free imaging of live bacteria and is the standard instrument for species identification across clinical, environmental, and industrial microbiology. Yet field samples are routinely polymicrobial and may contain organisms that were never seen during system training, and no computer-vision benchmark tests multi-label species identification from phase-contrast microscopy (PCM) of such mixtures. We introduce Phase-contrast Optical bEnchmark for Bacterial Identification ($\textbf{PHOEBI}$), a wet-lab-prepared dataset of $120{,}000$ PCM images covering $40$ combinations of six rod-shaped species, paired with a leave-combinations-out (LCO) evaluation protocol that holds out entire species combinations to mirror the practical scenario of a model trained on catalogued mixtures that must generalise to unseen ones. On LCO, every gradient-trained per-image aggregator we test drops $0.39$ to $0.57$ F1 from the in-distribution to the held-out split, a systematic open-world recognition failure in the aggregator, not the visual representation. A linear probe of thirteen different encoders over the same features spreads only about six percentage points of F1 across general-purpose and biomedical pretraining objectives, confirming the representation is sound. We propose three lightweight $\textit{anchor-based}$ decoders that capture per-species presence geometrically over a shared frozen tile-feature pool, scoring $\textit{higher}$ on held-out combinations than on in-distribution validation.
Chinese Translation
光学显微镜能够快速、无标记地成像活细菌，是临床、环境和工业微生物学中物种识别的标准工具。然而，现场样本通常是多种微生物的混合，可能包含在系统训练期间从未见过的生物体，并且没有计算机视觉基准测试能够对这种混合物进行多标签物种识别。我们介绍了相差光学细菌识别基准（PHOEBI），这是一个经过湿实验室准备的包含120,000张相差显微镜（PCM）图像的数据集，涵盖了六种杆状物种的40种组合，并配备了一种组合留出（LCO）评估协议，该协议保留整个物种组合，以反映模型在已编目混合物上训练后必须推广到未见混合物的实际场景。在LCO上，我们测试的每个基于梯度训练的逐图聚合器的F1分数从分布内到保留分割下降了0.39到0.57，这是聚合器中的系统性开放世界识别失败，而不是视觉表示。对同一特征的十三种不同编码器进行线性探测，F1分数在通用和生物医学预训练目标之间仅相差约六个百分点，确认了表示的合理性。我们提出了三种轻量级的基于锚点的解码器，这些解码器在共享的冻结块特征池中几何地捕捉每种物种的存在，在保留组合上的得分高于在分布内验证时的得分。
cs.CV / 205 / 2606.22905
InteractiveAvatar: Real-Time Streaming Video Generation for Consistent and Intent-Aware Avatars
互动头像：用于一致性和意图感知头像的实时流视频生成
Abstract
Recent diffusion-based models have enabled realistic audio-driven avatar generation in real-time streaming. However, existing approaches struggle to maintain visual temporal consistency and fail to explicitly perceive user intent in complex interactive streaming scenarios. To address these challenges, we propose InteractiveAvatar, a real-time infinite-streaming video generation framework that supports visually consistent avatar video generation and intent-aware interactions. With autoregressive distillation, InteractiveAvatar achieves real-time str-eaming generation of human avatars over arbitrarily long durations. For visual consistency, we introduce a Long-Short Visual Memory (LSVM) mechanism that flexibly compresses historical visual information into compact tokens, preserving both short-range coherence and long-term consistency. To generate avatars with speeches and actions aligned with user intent, we propose a Reasoning-Reaction Module (RRM), which incorporates a State-Cycling strategy and a Cache-Switching mechanism. Extensive experimental results over diverse scenarios demonstrate that our method achieves state-of-the-art visual consistency in long-duration generation, while enabling complex user-avatar interaction in real time.
Chinese Translation
最近的基于扩散模型的方法使得实时流媒体中的音频驱动头像生成变得更加真实。然而，现有的方法在保持视觉时间一致性方面存在困难，并且在复杂的交互流媒体场景中未能明确感知用户意图。为了解决这些挑战，我们提出了InteractiveAvatar，这是一个实时无限流视频生成框架，支持视觉一致的头像视频生成和意图感知的交互。通过自回归蒸馏，InteractiveAvatar实现了对任意长时间段的人类头像的实时流生成。为了确保视觉一致性，我们引入了一种长短视觉记忆（Long-Short Visual Memory, LSVM）机制，灵活地将历史视觉信息压缩为紧凑的标记，从而保持短期连贯性和长期一致性。为了生成与用户意图相符的讲话和动作的头像，我们提出了推理反应模块（Reasoning-Reaction Module, RRM），该模块结合了状态循环策略和缓存切换机制。大量在不同场景下的实验结果表明，我们的方法在长时间生成中实现了最先进的视觉一致性，同时支持实时复杂的用户-头像交互。
cs.CV / 206 / 2606.22913
Intend, Reflect, Refine: An Adaptive Multimodal Reflection Framework for Autonomous Driving
意图、反思、精炼：一种用于自主驾驶的自适应多模态反思框架
Abstract
Recent Vision-Language-Action (VLA) models have advanced end-to-end autonomous driving by incorporating reasoning for better interpretability and planning quality. However, most existing approaches directly generate the final trajectory without explicitly examining its future consequences, which limits their reliability in complex and dynamic environments. To address this limitation, we propose IRR-Drive (Intend, Reflect, Refine), an adaptive multimodal reflection framework for autonomous driving. Specifically, to tightly couple high-level reasoning with physical constraints, IRR-Drive first generates a preliminary textual intention and anticipates potential interactions by predicting future semantic bird's-eye view (BEV) representations. This dual-modality (Text + BEV) reflection space explicitly models anticipated scene evolution, enabling the model to rigorously self-correct and refine its initial intent before generating the final trajectory. Furthermore, to balance planning performance and computational efficiency, we construct reflection-oriented training data and design an adaptive reflection reward, enabling the model to adaptively select its reasoning mode according to scene complexity. Instead of using reasoning primarily as an auxiliary interpretation, IRR-Drive directly integrates an adaptive reflection mechanism into the planning framework, enabling grounded, decision-aware trajectory correction that is driven by scene complexity. Our method achieves state-of-the-art performance on the NAVSIM benchmark in both PDMS and EPDMS. Extensive experiments demonstrate the effectiveness of our multimodal reflection framework and validate the efficacy of the proposed adaptive reflection strategy.
Chinese Translation
近期的视觉-语言-动作（VLA）模型通过结合推理以提高可解释性和规划质量，推动了端到端自主驾驶的发展。然而，大多数现有方法直接生成最终轨迹，而未明确检查其未来后果，这限制了它们在复杂和动态环境中的可靠性。为了解决这一限制，我们提出了IRR-Drive（意图、反思、精炼），一种用于自主驾驶的自适应多模态反思框架。具体而言，为了将高层次推理与物理约束紧密结合，IRR-Drive首先生成初步的文本意图，并通过预测未来的语义鸟瞰图（BEV）表示来预期潜在的交互。这种双模态（文本 + BEV）反思空间明确建模了预期的场景演变，使模型能够严格自我修正并在生成最终轨迹之前精炼其初始意图。此外，为了平衡规划性能和计算效率，我们构建了以反思为导向的训练数据，并设计了一种自适应反思奖励，使模型能够根据场景复杂性自适应选择其推理模式。IRR-Drive不仅将推理作为辅助解释，而是直接将自适应反思机制整合到规划框架中，实现基于场景复杂性的有根据的、决策意识的轨迹修正。我们的方法在PDMS和EPDMS的NAVSIM基准测试中达到了最先进的性能。大量实验验证了我们多模态反思框架的有效性，并验证了所提出的自适应反思策略的有效性。
cs.CV / 207 / 2606.22918
Each Judge Its Own Yardstick: Discovering Per-VLM Taxonomies for Physical Video Evaluation
各自评判自己的标准：发现用于物理视频评估的每个 VLM 的分类法
Abstract
Maintaining physical consistency in video generators and world models increasingly relies on vision-language models (VLMs) as automated judges that provide reward signals, ranking decisions, and data-filtering criteria. Yet VLMs differ substantially in training data and architecture, encoding physical phenomena through distinct internal representations. A single global evaluation schema therefore gives every VLM the same axes of competence, regardless of what each can actually perceive. We propose JudgeFit, an iterative refinement procedure that discovers a per-VLM evaluation taxonomy. An initial taxonomy is constructed by prompting the target VLM to enumerate physics errors on a small set of videos and clustering the resulting descriptions. The taxonomy is then refined through a diagnostic step: we calibrate the VLM's per-dimension scores to human physical-commonsense ratings, diagnose which dimensions it scores unreliably or redundantly, and prompt an LLM to repair them, iterating until convergence. We further instantiate this procedure as a benchmark and apply it to 16 VLMs spanning eight model families. The refined taxonomy outperforms the global-schema baseline on held-out videos for every VLM tested, with a mean relative improvement of approximately 32%. Beyond aggregate accuracy, the per-VLM profiles expose model-specific blind spots that overall rankings cannot anticipate, with reliability patterns differing markedly across model families.
Chinese Translation
在视频生成器和世界模型中保持物理一致性日益依赖于视觉-语言模型（VLM）作为自动评判者，提供奖励信号、排名决策和数据过滤标准。然而，VLM 在训练数据和架构上存在显著差异，通过不同的内部表示编码物理现象。因此，单一的全球评估框架使得每个 VLM 拥有相同的能力轴线，而不考虑每个模型实际能够感知的内容。我们提出了 JudgeFit，这是一种迭代精炼程序，用于发现每个 VLM 的评估分类法。初始分类法通过提示目标 VLM 列举一小组视频中的物理错误并对结果描述进行聚类来构建。然后，通过诊断步骤对分类法进行精炼：我们将 VLM 的每个维度得分校准到人类物理常识评分，诊断其得分不可靠或冗余的维度，并提示一个大型语言模型（LLM）进行修复，迭代直到收敛。我们进一步将此过程实例化为基准，并将其应用于涵盖八个模型家族的 16 个 VLM。经过精炼的分类法在对每个测试的 VLM 的保留视频上超越了全球框架基线，平均相对提升约为 32%。除了整体准确性外，每个 VLM 的特征揭示了模型特定的盲点，而整体排名无法预见，可靠性模式在不同模型家族之间存在显著差异。
cs.CV / 208 / 2606.22924
MythraGen: Two-Stage Retrieval Augmented Art Generation Framework
MythraGen：两阶段检索增强艺术生成框架
Abstract
Text-to-image generation has seen rapid advancements, especially with the development of generative models. However, challenges remain in achieving high-quality, contextually accurate image outputs that faithfully match the provided textual descriptions, especially in artistic generation. In this paper, we present a simple yet efficient retrieval augmented generation framework, namely MythraGen, for text-to-artistic image generation by integrating an art retrieval mechanism with LoRA-based model fine-tuning. Our method extracts features from a large-scale art dataset, optimizing the generation process by combining artist-specific styles and content. Particularly, retrieved images from an external art database that have the highest similarity to the query prompt are used to finetune Stable Diffusion using LoRA for desired art generation. Experimental results and user studies on the WikiArt dataset show that our proposed method can generate artworks that closely match the user's input, significantly outperforming existing solutions.
Chinese Translation
文本到图像生成技术取得了快速进展，尤其是在生成模型的发展方面。然而，在实现高质量、上下文准确的图像输出方面仍然面临挑战，尤其是在艺术生成领域。本文提出了一种简单而高效的检索增强生成框架，即 MythraGen，通过将艺术检索机制与基于 LoRA 的模型微调相结合，实现文本到艺术图像的生成。我们的方法从大规模艺术数据集中提取特征，通过结合特定艺术家的风格和内容来优化生成过程。特别是，从外部艺术数据库中检索到与查询提示具有最高相似度的图像被用于使用 LoRA 微调 Stable Diffusion，以实现所需的艺术生成。在 WikiArt 数据集上的实验结果和用户研究表明，我们提出的方法能够生成与用户输入紧密匹配的艺术作品，显著优于现有解决方案。
cs.CV / 209 / 2606.22931
BEV-Denoise: Learning Intrinsic Noise for Accurate Bird's-Eye-View Semantic Segmentation
BEV-Denoise：学习内在噪声以实现准确的鸟瞰图语义分割
Abstract
In this paper, we present a framework dubbed \textbf{BEV-Denoise} that estimates and removes intrinsic noise from learned Bird's-Eye-View (BEV) features to achieve accurate BEV semantic segmentation. Inspired by the noise estimation capability of Denoising Diffusion Probabilistic Models (DDPM), we design a UNet-based noise estimation module that learns to estimate the noise from the learned BEV features. The estimated noise is then subtracted from the BEV features and fed to BEV map decoders for the final prediction results. To facilitate supervision for the noise estimation module, we follow a sequential learning paradigm called Task Decomposition (TD) where a pre-trained BEV map autoencoder is employed to train a view transformation (VT) encoder. We share three key insights learned from our intensive experiments that are critical for improved performance. We apply our framework to four existing models, encompassing the three major VT paradigms. Experimental results on a large-scale real-world dataset, nuScenes, demonstrate the effectiveness of our framework.
Chinese Translation
在本文中，我们提出了一种名为 extbf{BEV-Denoise} 的框架，该框架通过估计和去除从学习到的鸟瞰图（BEV）特征中提取的内在噪声，以实现准确的BEV语义分割。受到去噪扩散概率模型（Denoising Diffusion Probabilistic Models, DDPM）噪声估计能力的启发，我们设计了一个基于UNet的噪声估计模块，该模块学习从学习到的BEV特征中估计噪声。然后，将估计的噪声从BEV特征中减去，并输入到BEV地图解码器中以获得最终的预测结果。为了促进对噪声估计模块的监督，我们遵循了一种称为任务分解（Task Decomposition, TD）的顺序学习范式，其中使用预训练的BEV地图自编码器来训练视图转换（View Transformation, VT）编码器。我们分享了从大量实验中获得的三个关键见解，这些见解对于提高性能至关重要。我们将我们的框架应用于四个现有模型，涵盖了三种主要的VT范式。在大规模真实世界数据集nuScenes上的实验结果证明了我们框架的有效性。
cs.CV / 210 / 2606.22935
Hybrid Compression: Integrating Pruning and Quantization for Optimized Neural Networks
混合压缩：结合剪枝和量化以优化神经网络
Abstract
Deep neural networks have witnessed remarkable advancements in recent years and have become integral to various applications. However, alongside these developments, training and deployment of neural network models on embedding and edge devices face significant challenges due to limited memory and computational resources. These problems can be addressed with deep neural network compression, which involves a trade-off between model size and performance. In this paper, we propose a novel method for model compression through two phases. First, we utilize model compression techniques, such as pruning and quantization, to significantly reduce the model size. Then, we use Mixture of Experts to route the previously compressed models to enhance performance while maintaining a balance in inference efficiency. MoEs consist of multiple expert models (i.e., compressed models) that are moderately sized and deliver stable performance. Experimental results on several benchmark datasets show that our method successfully compresses CNN models which achieves substantial reductions in FLOPs and parameters with a negligible accuracy drop.
Chinese Translation
近年来，深度神经网络取得了显著进展，并已成为各种应用的核心。然而，随着这些发展的推进，在嵌入式和边缘设备上训练和部署神经网络模型面临着由于内存和计算资源有限而带来的重大挑战。这些问题可以通过深度神经网络压缩来解决，该过程涉及模型大小与性能之间的权衡。本文提出了一种通过两个阶段进行模型压缩的新方法。首先，我们利用模型压缩技术，如剪枝和量化，显著减少模型大小。然后，我们使用专家混合模型（Mixture of Experts）来路由先前压缩的模型，以提高性能，同时保持推理效率的平衡。专家混合模型由多个适中大小的专家模型（即压缩模型）组成，能够提供稳定的性能。在多个基准数据集上的实验结果表明，我们的方法成功地压缩了卷积神经网络（CNN）模型，实现了FLOPs和参数的显著减少，同时准确率下降微乎其微。
cs.CV / 211 / 2606.22943
Evaluating self-supervised echocardiographic representations across downstream extraction strategies for left-ventricular segmentation and ejection fraction estimation
评估自监督心脏超声表示在左心室分割和射血分数估计中的下游提取策略
Abstract
Self-supervised learning (SSL) is increasingly used in medical imaging to reduce annotation requirements, but representation quality is often judged using a single downstream evaluation setting. For dense clinical tasks, this can confound representation quality with the capacity of the downstream model used to recover task-relevant information. We present a systematic evaluation of self-supervised representations for left-ventricular segmentation and ejection fraction (EF) estimation from apical four-chamber echocardiography on EchoNet-Dynamic. Rather than relying on a single downstream probe, we compare a hierarchy of extraction strategies with increasing expressivity: heuristic extraction without mask-supervised training, frozen linear probes, frozen lightweight decoder probes, and partial fine-tuning. We apply this framework to two complementary representation families: generic frozen self-DIstillation with NO labels (DINOv3) features and a task-adapted dense self-supervised representation, Bootstrap Your Own Segmentation (BYOS). In both families, heuristic extraction substantially understated what was recoverable from the frozen representation. For DINOv3, performance improved from Dice 0.684 and EF mean absolute error (MAE) 13.01 under heuristic extraction to Dice 0.906 and EF MAE 9.65 with a frozen lightweight decoder, approaching a supervised U-Net baseline (Dice 0.915, EF MAE 9.72). For BYOS, performance improved from Dice 0.687 and EF MAE 17.83 under heuristic extraction to Dice 0.902 and EF MAE 8.74 with a frozen lightweight decoder. These results show that conclusions about self-supervised representation quality in dense echocardiographic analysis depend strongly on the downstream extraction strategy used for evaluation. We therefore argue that multi-strategy evaluation is an important methodological consideration for SSL in dense medical image analysis.
Chinese Translation
自监督学习（SSL）在医学成像中越来越多地被应用以减少标注需求，但表示质量通常仅通过单一的下游评估设置来判断。对于密集的临床任务，这可能会将表示质量与用于恢复任务相关信息的下游模型的能力混淆。我们对自监督表示在基于EchoNet-Dynamic的心尖四腔心脏超声中进行左心室分割和射血分数（EF）估计进行了系统评估。我们并未依赖单一的下游探针，而是比较了一系列具有递增表达能力的提取策略：无掩码监督训练的启发式提取、冻结的线性探针、冻结的轻量解码器探针和部分微调。我们将这一框架应用于两种互补的表示家族：通用的冻结自我蒸馏无标签（DINOv3）特征和任务适应的密集自监督表示，Bootstrap Your Own Segmentation（BYOS）。在这两种家族中，启发式提取显著低估了从冻结表示中可恢复的信息。对于DINOv3，性能从启发式提取下的Dice 0.684和EF平均绝对误差（MAE）13.01提高到使用冻结轻量解码器的Dice 0.906和EF MAE 9.65，接近监督U-Net基线（Dice 0.915，EF MAE 9.72）。对于BYOS，性能从启发式提取下的Dice 0.687和EF MAE 17.83提高到使用冻结轻量解码器的Dice 0.902和EF MAE 8.74。这些结果表明，在密集心脏超声分析中，自监督表示质量的结论在很大程度上依赖于用于评估的下游提取策略。因此，我们认为多策略评估是密集医学图像分析中自监督学习的重要方法论考虑。
cs.CV / 212 / 2606.22955
Evo-RAD: Navigating Rare Retinal Disease Diagnosis via Self-Evolving Agentic Retrieval
Evo-RAD：通过自我演化代理检索导航稀有视网膜疾病诊断
Abstract
Large-scale pretrained foundation models have revolutionized general medical screening, but often falter on rare diseases because such conditions are underrepresented in real-world clinical datasets. While retrieval-augmented diagnosis attempts to mitigate this, conventional static methods frequently succumb to the hubness problem, retrieving visually similar but semantically incorrect common diseases. To address this, we propose Evo-RAD, a self-evolving agentic framework that transforms evidence acquisition into a dynamic decision-making task. We formulate retrieval as a Markov Decision Process (MDP) where a graphbased agent observes the reference set state and executes actions to purge discordant evidence (DELETE), acquire pathologically consistent samples (INSERT), or conclude the evolution (TERMINATE). Optimized via Group Relative Policy Optimization (GRPO) with a homogeneityaware reward, the agent learns to maximize the diagnostic homogeneity of the support reference set. Experiments on retinal disease benchmarks show that Evo-RAD substantially improves rare-disease diagnosis, outperforming retinal foundation models by +21.04%, while also surpassing retrieval-based and parameter-efficient fine-tuning methods by +3.56%. Code is available at https://github.com/SDH-Lab/Evo-RAD.
Chinese Translation
大规模预训练基础模型已经彻底改变了普通医学筛查，但在稀有疾病方面常常表现不佳，因为这些疾病在现实世界的临床数据集中往往被低估。虽然增强检索的诊断方法试图缓解这一问题，但传统的静态方法常常受到中心性问题的影响，检索到视觉上相似但语义上不正确的常见疾病。为了解决这一问题，我们提出了Evo-RAD，一个自我演化的代理框架，将证据获取转变为动态决策任务。我们将检索形式化为马尔可夫决策过程（MDP），其中基于图的代理观察参考集状态，并执行操作以清除不一致的证据（DELETE）、获取病理一致的样本（INSERT）或结束演化（TERMINATE）。通过具有同质性意识奖励的群体相对策略优化（GRPO）进行优化，代理学习最大化支持参考集的诊断同质性。在视网膜疾病基准测试中的实验表明，Evo-RAD显著改善了稀有疾病的诊断，超越了视网膜基础模型21.04%，同时也超过了基于检索和参数高效微调的方法3.56%。代码可在 https://github.com/SDH-Lab/Evo-RAD 获取。
cs.CV / 213 / 2606.22963
Concept Alignment Contrast and Long-Short Prompt Memory for Test-Time Adaptation of SAM3 in Medical Image Segmentation
概念对齐对比与长短提示记忆在医学图像分割中对SAM3的测试时适应
Abstract
Concept segmentation models like Segment Anything Model 3 (SAM3) show strong generalization on natural images, yet their performance degrades in medical imaging due to the domain gap caused by different imaging principles and styles. Test-Time Adaptation (TTA) is essential for improving the testing performance by updating the model on the fly without annotations. However, existing vision-language TTA methods are mainly driven by image-level uncertainty minimization, which does not necessarily reflect region-level semantic correctness in medical segmentation. Moreover, they often lack mechanisms to maintain stability in continual one-pass adaptation, leading to limited performance when reliable dense supervision is missing for segmentation. To address these issues, we propose Concept Alignment Contrast and LongShort Prompt Memory for Test-Time Adaptation (CM-TTA) of SAM3 for medical images. First, for a test sample with multiple augmentations, we introduce a novel Concept Alignment Contrast (CAC) metric, which leverages textual-visual semantic consistency to robustly evaluate prediction quality to select the best augmented view as the supervision. Second, to balance rapid and stable adaptation, we design a Long-Short Prompt Memory (LSPM) module. The short memory dynamically fuses recent prompts based on CAC scores for agile local adaptation, while the long memory maintains a stable global prompt to generate enhanced pseudo-labels. Finally, a Densely Supervised Prompt Update (DSPU) strategy is proposed to optimize the prompt embeddings with enhanced pseudo labels as dense supervision. Extensive experiments on prostate and skin lesion segmentation demonstrate that our CM-TTA framework significantly outperforms existing methods for TTA of SAM3.
Chinese Translation
概念分割模型如Segment Anything Model 3 (SAM3)在自然图像上表现出强大的泛化能力，但由于不同成像原理和风格造成的领域差距，其在医学成像中的性能下降。测试时适应（Test-Time Adaptation, TTA）对于通过动态更新模型而无需标注来提高测试性能至关重要。然而，现有的视觉-语言TTA方法主要依赖于图像级不确定性最小化，这并不一定反映医学分割中的区域级语义正确性。此外，它们通常缺乏在持续单次适应中保持稳定性的机制，导致在缺乏可靠密集监督的情况下性能有限。为了解决这些问题，我们提出了用于医学图像的SAM3的概念对齐对比与长短提示记忆的测试时适应（CM-TTA）。首先，对于具有多种增强的测试样本，我们引入了一种新颖的概念对齐对比（CAC）度量，利用文本-视觉语义一致性来稳健地评估预测质量，以选择最佳增强视图作为监督。其次，为了平衡快速和稳定的适应，我们设计了一个长短提示记忆（LSPM）模块。短期记忆基于CAC分数动态融合最近的提示，以实现灵活的局部适应，而长期记忆则保持稳定的全局提示以生成增强的伪标签。最后，提出了一种密集监督提示更新（DSPU）策略，以增强的伪标签作为密集监督来优化提示嵌入。在前列腺和皮肤病变分割的广泛实验中，我们的CM-TTA框架显著优于现有的SAM3 TTA方法。
cs.CV / 214 / 2606.22986
Subject-Level Unknown-Identity Identification from Leap Motion Controller 2 Hand Landmarks
基于Leap Motion Controller 2手部特征点的个体级未知身份识别
Abstract
This work studies subject recognition from Leap Motion Controller 2 (LMC2) hand landmark data under a subject-level unknown-identity identification protocol on the Multi View Leap2 Hand Pose (ML2HP) dataset. Using only the landmark modality, we retain the original geometric representation and enrich it with fingertip-to-palm distances and palm-normalized inter-finger angular descriptors. Evaluation is performed under a Leave-One-Subject-Out (LOSO) protocol in which, for each outer fold, one subject is excluded from the enrolled set and treated as unknown at test time. To avoid tuning on the true outer unknown subject, the unknown-rejection threshold is selected in an inner validation step by temporarily withholding one enrolled subject from the inner gallery and using it only for threshold estimation. We compare a tree ensemble baseline with two neural alternatives: a learned embedding baseline based on centroid matching and cosine-similarity-based rejection, and an MLP+OpenMax model, which represents a more established open-set recognition approach. Under this evaluation setup, Extra Trees remains the strongest overall method, indicating that the main challenge on this benchmark is not enrolled-subject discrimination alone, but robust score separation between known and unknown probes. The results support the feasibility of compact, interpretable landmark-based descriptors for contactless hand-based unknown-subject rejection and identification on a small-cohort dataset.
Chinese Translation
本研究探讨了在个体级未知身份识别协议下，利用Leap Motion Controller 2 (LMC2) 手部特征点数据进行个体识别，数据集为多视角Leap2手势姿态（ML2HP）数据集。仅使用特征点模态，我们保留了原始几何表示，并通过指尖到手掌的距离以及手掌标准化的指间角度描述符进行了丰富。在Leave-One-Subject-Out (LOSO)协议下进行评估，其中每个外部折叠中排除一个个体作为测试时的未知个体。为了避免在真实的外部未知个体上进行调优，未知拒绝阈值在内部验证步骤中选择，暂时从内部库中扣留一个已登记个体，仅用于阈值估计。我们将树集成基线与两种神经网络替代方案进行了比较：一种基于质心匹配和余弦相似度拒绝的学习嵌入基线，以及一种MLP+OpenMax模型，后者代表了一种更成熟的开放集识别方法。在该评估设置下，Extra Trees仍然是最强的整体方法，表明该基准的主要挑战不仅是已登记个体的区分，而是已知和未知探针之间的稳健得分分离。结果支持基于特征点的紧凑、可解释的描述符在小规模数据集上进行无接触手部未知个体拒绝和识别的可行性。
cs.CV / 215 / 2606.22987
Can Single-View Mesh Reconstruction Generalize to Robot Camera Rotation?
单视图网格重建能否推广到机器人相机旋转？
Abstract
Single-view mesh reconstruction predicts object meshes and spatial layouts from a single observation, making it attractive for fast robot spatial reasoning and real-to-sim digital twins. However, robot-mounted cameras naturally rotate during manipulation and navigation, while learned single-view reconstruction models often rely on view-dependent priors and may generalize poorly to out-of-distribution camera rotations. Such rotations can introduce 3D inconsistencies, incorrect layouts, and violations of physical constraints, but this failure mode remains under-evaluated. We introduce an evaluation protocol with controlled axis-wise roll, pitch, and yaw sweeps to trace errors in monocular depth estimation (MDE), canonical object meshes, camera-space layout, and physical plausibility within a representative SAM3D-style pipeline. On the Aria Digital Twin dataset and a real Franka wrist-camera sequence, camera rotations induce MDE distortion, layout drift, and collision penetration, while canonical mesh predictions remain relatively stable. A two-stage SAM3D+FoundationPose pipeline is more robust than one-stage feed-forward layout prediction, and our Gravity-Aware Refinement reduces one-stage pairwise ICP-based layout-orientation error by 47.1$\%$. Our evaluation reveals that current single-view mesh reconstruction methods generalize poorly to robot camera rotation, and suggests that explicit gravity cues are important for reliable robotic single-view mesh reconstruction.
Chinese Translation
单视图网格重建从单个观察中预测物体网格和空间布局，这使其在快速机器人空间推理和真实与模拟数字双胞胎中具有吸引力。然而，安装在机器人上的相机在操作和导航过程中会自然旋转，而学习的单视图重建模型通常依赖于视角相关的先验，可能在面对分布外的相机旋转时泛化能力较差。这种旋转可能导致三维不一致、错误的布局以及物理约束的违反，但这种失效模式仍然未得到充分评估。我们引入了一种评估协议，通过控制轴向的滚转、俯仰和偏航来追踪单目深度估计（MDE）、标准物体网格、相机空间布局和物理合理性中的错误，采用代表性的SAM3D风格管道。在Aria数字双胞胎数据集和真实的Franka腕部相机序列中，相机旋转引起了MDE失真、布局漂移和碰撞穿透，而标准网格预测相对稳定。两阶段的SAM3D+FoundationPose管道比单阶段的前馈布局预测更具鲁棒性，而我们的重力感知精细化将单阶段成对ICP基础的布局方向错误降低了47.1%。我们的评估揭示了当前的单视图网格重建方法在机器人相机旋转方面的泛化能力较差，并建议显式的重力线索对于可靠的机器人单视图网格重建至关重要。
cs.CV / 216 / 2606.22999
Black-Box Continual Learning for Vision-Language Models
面向视觉-语言模型的黑箱持续学习
Abstract
The rapid deployment of Vision-Language Models (VLMs) in dynamic environments necessitates the ability to learn continuously without forgetting. However, traditional continual learning (CL) settings often rely on white-box paradigms, which is increasingly invalidated by the shift toward cloud-hosted models. In this paper, we introduce Black-CL, a more realistic benchmark for VLMs that enforces three primary real-world challenges: weight and architecture inaccessibility, constrained computation, and task-agnostic inference. The learner can query only output embeddings or logits, with no gradient flow through or structural modification of the backbone. Current CL methodologies, which rely on backbone backpropagation or complex parameter expansion, are fundamentally incompatible with these constraints. Under this setting, we propose BETA, a simple yet effective baseline built on the key insight that solely optimizing textual prototypes can navigate the complexities of CL. BETA integrates three core components: Semantic Projection Accumulation (SPA) for incremental knowledge acquisition, Latent Distribution Replay (LDR) for anchoring the embedding space against catastrophic forgetting, and Test-Time Prototype Adaptation (TTPA) for dynamic, instance-aware boundary refinement. Extensive experiments across ten diverse datasets and various backbones demonstrate that BETA significantly outperforms existing black-box tuners. Remarkably, with only 0.05 M trainable parameters, a 180--3000$\times$ reduction compared to competitive methods, BETA achieves performance on par with or even exceeding white-box CL methods. We believe Black-CL and BETA provide a foundational framework for future advancements in continual learning and accelerates the transition of continual learning from academia to real-world systems.
Chinese Translation
视觉-语言模型（VLMs）在动态环境中的快速部署要求具备持续学习而不遗忘的能力。然而，传统的持续学习（CL）设置通常依赖于白箱范式，这在向云托管模型转变的背景下日益失效。本文提出了Black-CL，这是一个更为现实的VLM基准，强制应对三个主要的现实挑战：权重和架构不可访问、计算受限以及任务无关的推理。学习者只能查询输出嵌入或对数值，无法通过主干网络进行梯度流动或结构修改。目前的CL方法依赖于主干反向传播或复杂的参数扩展，这与这些限制根本不兼容。在此设置下，我们提出了BETA，这是一个简单而有效的基线，基于一个关键见解：仅优化文本原型即可应对CL的复杂性。BETA集成了三个核心组件：语义投影累积（Semantic Projection Accumulation, SPA）用于增量知识获取，潜在分布重放（Latent Distribution Replay, LDR）用于锚定嵌入空间以防止灾难性遗忘，以及测试时原型适应（Test-Time Prototype Adaptation, TTPA）用于动态的、实例感知的边界细化。在十个不同数据集和多种主干网络上的广泛实验表明，BETA显著优于现有的黑箱调优方法。值得注意的是，BETA仅需0.05M可训练参数，相比竞争方法减少了180--3000倍，且在性能上与白箱CL方法持平甚至超越。我们相信Black-CL和BETA为未来持续学习的进展提供了基础框架，并加速了持续学习从学术界向实际系统的转变。
cs.CV / 217 / 2606.23000
MotionMAR: Multi-scale Auto-Regressive Human Motion Reconstruction from Sparse Observations
MotionMAR：基于稀疏观测的多尺度自回归人类运动重建
Abstract
Human motion follows a temporal hierarchical structure, transitioning from low-frequency global trajectories to high-frequency details. Inspired by the success of multi-level autoregressive models in computer vision, we propose MotionMAR, a coarse-to-fine framework for motion reconstruction from sparse observations. It first estimates the global trajectory of human motion and then gradually refines the temporal details. This architecture consists of four integrated components. The Temporal Multi-scale Tokenization (TMT) VQ-VAE encodes the data at multiple temporal resolutions, separating semantic motion from minor jitters. The Motion Autoregressive Network (MAN) operates in this latent space, predicting motion across scales. It first establishes the global structure through coarse indices and then generates finer indices to recover specific details. Meanwhile, the Scale-Aware Control (SAC) module integrates sparse tracking data to ensure the generated output aligns with actual observations. The Motion Refinement Network (MRN) subsequently smooths consecutive poses and eliminates quantization artifacts. Experiments show that MotionMAR achieves state-of-the-art accuracy on the AMASS dataset, providing a reliable and structure-aware approach for motion reconstruction. The source code is publicly available at http://www.lidarhumanmotion.net/motionmar/.
Chinese Translation
人类运动遵循时间层次结构，从低频全局轨迹过渡到高频细节。受到计算机视觉中多层自回归模型成功的启发，我们提出了MotionMAR，一种从稀疏观测中进行运动重建的粗到细框架。该框架首先估计人类运动的全局轨迹，然后逐步细化时间细节。该架构由四个集成组件组成。时间多尺度标记化（Temporal Multi-scale Tokenization, TMT）VQ-VAE在多个时间分辨率下编码数据，将语义运动与微小抖动分离。运动自回归网络（Motion Autoregressive Network, MAN）在该潜在空间中运行，跨尺度预测运动。它首先通过粗略索引建立全局结构，然后生成更精细的索引以恢复特定细节。同时，尺度感知控制（Scale-Aware Control, SAC）模块整合稀疏跟踪数据，以确保生成的输出与实际观测一致。运动细化网络（Motion Refinement Network, MRN）随后平滑连续姿态并消除量化伪影。实验表明，MotionMAR在AMASS数据集上达到了最先进的准确性，为运动重建提供了一种可靠且具结构感知的方法。源代码可在http://www.lidarhumanmotion.net/motionmar/公开获取。
cs.CV / 218 / 2606.23005
From Point Estimates to Distributions: GMM Pooling for MIL in Preterm Birth Prediction
从点估计到分布：用于早产预测的GMM汇聚方法
Abstract
Preterm birth (PTB) prediction can enable targeted surveillance and timely intervention, yet most ultrasound-based models use a single selected transvaginal ultrasound (TVUS) frame per patient despite routine exams acquiring multiple cervical images. We formulate PTB prediction as a multiple instance learning (MIL) problem, representing each patient as a variable-sized bag of TVUS images with a single outcome label. To move beyond standard MIL aggregators that collapse a bag into a point estimate, we propose a Gaussian Mixture Model (GMM) pooling, which summarizes all images in a bag into a fixed-length representation by modeling their feature distribution. This design captures intra-patient variability. We evaluate the method on a private clinical cohort and on a public lymph node metastasis benchmark. For PTB prediction, GMM pooling improves over the instance-based model PR-AUC from 0.44 to 0.56. On the lymph node benchmark, it achieves state-of-the-art performance with 0.91 F1-score and 0.89 ROC-AUC for classification and 0.18 MAE for regression. The code is publicly available at https://github.com/HussainAlasmawi/GMM_Pooling.
Chinese Translation
早产（PTB）预测可以实现针对性的监测和及时干预，然而大多数基于超声的模型在每位患者中仅使用一个选定的经阴道超声（TVUS）图像，尽管常规检查会获取多个宫颈图像。我们将PTB预测形式化为一个多实例学习（MIL）问题，将每位患者表示为一个具有单一结果标签的可变大小的TVUS图像包。为了超越将包压缩为点估计的标准MIL聚合器，我们提出了一种高斯混合模型（GMM）汇聚方法，该方法通过建模图像特征分布，将包中的所有图像汇总为固定长度的表示。这种设计捕捉了患者内部的变异性。我们在一个私有临床队列和一个公共淋巴结转移基准上评估了该方法。在PTB预测中，GMM汇聚将基于实例的模型PR-AUC从0.44提高至0.56。在淋巴结基准测试中，它在分类任务中达到了0.91的F1分数和0.89的ROC-AUC，在回归任务中达到了0.18的MAE。代码已公开发布于https://github.com/HussainAlasmawi/GMM_Pooling。
cs.CV / 219 / 2606.23019
ScalingAttention: Discovering Intrinsic Sparse Attention Topology for Video Diffusion Transformers
ScalingAttention：发现视频扩散变换器的内在稀疏注意力拓扑
Abstract
While Diffusion Transformers (DiTs) have revolutionized high-fidelity video generation, their reliance on 3D full attention creates a quadratic computational bottleneck. Existing sparse methods face a dilemma: dynamic pruning suffers from prohibitive runtime overhead and memory fragmentation, while static heuristics fail to capture fine-grained dependencies. In this work, we propose ScalingAttention, a training-free framework grounded in a key inductive bias: while individual activations are input-dependent, the high-mass attention regions for each head rapidly converge to a stable, prompt-agnostic Intrinsic Sparse Topology. This topology is weight-encoded, scale-invariant, and efficient to extract. ScalingAttention decouples topology discovery from sparsity control via: (1) WEST (Weight-Encoded Sparse Topology), which extracts a robust block-sparse prior mask offline to eliminate runtime search; (2) FAST (Fidelity-Aware Sensitivity Tuning), which adaptively tunes head-wise sparsity based on diffusion fidelity requirements. To ensure practical acceleration, we co-design a hardware-aligned bit-wise block-sparse kernel. Experiments on Wan2.1 show up to 1.90X end-to-end speedup with superior fidelity, establishing a new Pareto frontier over state-of-the-art baselines.
Chinese Translation
尽管扩散变换器（DiTs）在高保真视频生成方面取得了革命性进展，但其对三维全注意力的依赖造成了二次计算瓶颈。现有的稀疏方法面临两难局面：动态剪枝遭遇高昂的运行时开销和内存碎片化，而静态启发式方法无法捕捉细粒度的依赖关系。在本研究中，我们提出了ScalingAttention，这是一个无需训练的框架，基于一个关键的归纳偏置：尽管单个激活依赖于输入，但每个头的高质量注意力区域迅速收敛到一个稳定的、与提示无关的内在稀疏拓扑。该拓扑是权重编码的、尺度不变的，并且提取效率高。ScalingAttention通过以下方式将拓扑发现与稀疏性控制解耦：(1) WEST（权重编码稀疏拓扑），该方法离线提取一个稳健的块稀疏先验掩码，以消除运行时搜索；(2) FAST（保真度感知灵敏度调节），该方法根据扩散保真度要求自适应调节头级稀疏性。为了确保实际加速，我们共同设计了一个与硬件对齐的逐位块稀疏内核。在Wan2.1上的实验表明，端到端速度提升高达1.90倍，同时保持卓越的保真度，建立了相较于最先进基线的新Pareto前沿。
cs.CV / 220 / 2606.23023
Boosting Neural Video Codec via Scale-Driven Online Flow Refinement
通过尺度驱动的在线光流精炼提升神经视频编解码器
Abstract
Although state-of-the-art neural video codecs (NVCs) have achieved remarkable performance, they suffer from limited generalization when encountering complex motion patterns unseen during training. To bridge this domain gap without the expensive cost of online fine-tuning, we propose a Training-Free Scale-Driven Online Flow Refinement (SOFR) method. Serving as a plug-and-play module, SOFR integrates motion information from coarse and fine scales and dynamically fuses them according to warping accuracy, effectively rectifying motion estimation errors with negligible computational overhead. Furthermore, we design a rate-aware strategy that selects different dynamic fusion strategies according to bitrate modes, and employs a reliability check based on warping error to ensure robustness. Extensive experiments on the USTC-TD dataset verify the effectiveness and generalization of SOFR across various NVC frameworks, including DCVC-SDD, DCVC-FM, and EHVC. Notably, it brings an average of 2.84% and 4.05% bitrate savings in terms of PSNR and MS-SSIM, respectively, to DCVC-FM with negligible coding time increase. Our code is available at https://github.com/SunnyMass/SOFR.
Chinese Translation
尽管最先进的神经视频编解码器（NVCs）已取得显著性能，但在遇到训练期间未见的复杂运动模式时，它们的泛化能力有限。为了在不进行昂贵的在线微调的情况下弥补这一领域差距，我们提出了一种无训练的尺度驱动在线光流精炼（SOFR）方法。作为一个即插即用的模块，SOFR整合了来自粗尺度和细尺度的运动信息，并根据变形精度动态融合它们，有效纠正运动估计误差，同时几乎不增加计算开销。此外，我们设计了一种感知码率的策略，根据比特率模式选择不同的动态融合策略，并基于变形误差进行可靠性检查，以确保鲁棒性。在USTC-TD数据集上的大量实验验证了SOFR在各种NVC框架（包括DCVC-SDD、DCVC-FM和EHVC）中的有效性和泛化能力。值得注意的是，它在PSNR和MS-SSIM方面分别为DCVC-FM带来了平均2.84%和4.05%的比特率节省，同时编码时间几乎没有增加。我们的代码可在https://github.com/SunnyMass/SOFR获取。
cs.CV / 221 / 2606.23027
Learning Stable Canonical Worlds for Novel View Synthesis and Beyond
学习稳定的典范世界以实现新视角合成及其他应用
Abstract
Feed-forward Gaussian splatting (FFGS) facilitates real-time novel view synthesis, yet current methods often remain tied to view-dependent predictions. As more input views are added, they may accumulate noisy or redundant evidence instead of converging to a stable scene representation. In this paper, we introduce CanonicalGS, a feed-forward pipeline that maps cluttered multi-view observations into a stable, scene-centric representation. CanonicalGS first extracts view-centric evidence from depth, semantic features, and uncertainty estimates, and then aggregates this evidence in a canonical latent world using uncertainty-aware fusion. By emphasizing reliable observations while suppressing uncertain or redundant ones, CanonicalGS produces representations that scale more effectively for novel view synthesis and transfer to downstream visual perception tasks. Experiments show up to a $2.5$ dB improvement in peak signal-to-noise ratio for synthesizing novel views and an $11\%$ gain in semantic segmentation accuracy.
Chinese Translation
前馈高斯点云（Feed-forward Gaussian splatting, FFGS）促进了实时新视角合成，但当前的方法往往仍然依赖于视角相关的预测。随着输入视角的增加，可能会积累噪声或冗余证据，而不是收敛到稳定的场景表示。本文提出了CanonicalGS，这是一种前馈管道，将杂乱的多视角观测映射到一个稳定的、以场景为中心的表示。CanonicalGS首先从深度、语义特征和不确定性估计中提取以视角为中心的证据，然后使用不确定性感知融合将这些证据聚合到一个典范潜在世界中。通过强调可靠的观测，同时抑制不确定或冗余的观测，CanonicalGS生成的表示在新视角合成和向下游视觉感知任务的迁移中更有效地扩展。实验表明，在合成新视角时峰值信噪比提高了高达 $2.5$ dB，语义分割精度提高了 $11\%$。
cs.CV / 222 / 2606.23028
Physics-Guided Spatiotemporal State Space Modeling for Lookahead Molten Pool Segmentation in Laser Wire-Feed Welding
基于物理指导的时空状态空间建模用于激光送丝焊接中的前瞻性熔池分割
Abstract
Real-time weld-pool perception is critical for closed-loop control in laser wire-feed welding, where sensing, computation, and actuator response introduce unavoidable delay. This paper presents a physics-guided spatiotemporal state space network for lookahead weld-pool segmentation. The model uses historical coaxial grayscale images, welding process parameters, and aligned wire-state electrical signals to predict the future semantic layout of three physically meaningful regions: keyhole, wire, and molten pool. It combines a visual encoder, process- and sensor-conditioned feature normalization, patch-level temporal state space modeling, horizon-conditioned latent prediction, dense future feature prediction, and a motion-aware mask decoder. Auxiliary signed-distance-function supervision, temporal consistency, feature distillation, and fine-grained keyhole losses further constrain the predicted geometry and local motion. Experiments on a 43-sequence laser welding dataset show that the proposed WeldMamba reaches 74.63\% mIoU at a 500 ms lookahead. Ablation studies further show that temporal history, patch-level state space modeling, and keyhole motion awareness are the main contributors to robust future segmentation.
Chinese Translation
实时焊接熔池感知对于激光送丝焊接中的闭环控制至关重要，而传感、计算和执行器响应引入了不可避免的延迟。本文提出了一种基于物理指导的时空状态空间网络，用于前瞻性焊接熔池分割。该模型利用历史同轴灰度图像、焊接过程参数和对齐的焊丝状态电信号来预测三个物理上有意义区域的未来语义布局：钥孔、焊丝和熔池。它结合了视觉编码器、过程和传感器条件下的特征归一化、补丁级时态状态空间建模、地平线条件下的潜在预测、密集的未来特征预测以及运动感知的掩模解码器。辅助的有符号距离函数监督、时间一致性、特征蒸馏和细粒度钥孔损失进一步约束了预测的几何形状和局部运动。在一个包含43个序列的激光焊接数据集上的实验表明，所提出的WeldMamba在500毫秒的前瞻下达到了74.63%的mIoU。消融研究进一步表明，时间历史、补丁级状态空间建模和钥孔运动感知是实现稳健未来分割的主要因素。
cs.CV / 223 / 2606.23031
DrivingVoxels: Compositional Sparse Voxel Rasterization for Dynamic Driving Scene Reconstruction
DrivingVoxels：动态驾驶场景重建的组合稀疏体素光栅化
Abstract
Reconstructing dynamic urban scenes remains challenging due to the unbounded nature of driving environments and the presence of multiple dynamic objects. Currently, potentially faster sparse voxel methods are mainly designed for static scenarios. On the other hand, dynamic approaches based on 3D Gaussian Splatting, despite their high-fidelity, are often time-consuming for driving scenarios and exhibit uncontrollable memory growth in large scenes. To address these limitations, we present DrivingVoxels, a compositional sparse voxel rendering framework for dynamic driving scenes. Our method jointly rasterizes sparse voxels from multiple independent octrees within a single rendering pass. Each rigid dynamic object is represented by an octree defined in its local coordinate frame, while a separate static octree models the stationary background. DrivingVoxels adopts a fully explicit, neural-free representation together with a LiDAR-guided structural initialization that efficiently captures scene geometry. We evaluate our framework on the PandaSet benchmark, demonstrating that DrivingVoxels performs on par on perceptual metrics and better on structural metrics for NVS and reconstruction while requiring shorter training times than previous 3DGS-base methods to an efficient optimization workflow anchored by a strong LiDAR prior.
Chinese Translation
重建动态城市场景仍然具有挑战性，因为驾驶环境的无限制特性和多个动态物体的存在。目前，潜在更快的稀疏体素方法主要针对静态场景设计。另一方面，尽管基于3D高斯点云（3D Gaussian Splatting）的动态方法具有高保真度，但在驾驶场景中往往耗时较长，并且在大场景中表现出不可控的内存增长。为了解决这些限制，我们提出了DrivingVoxels，一个用于动态驾驶场景的组合稀疏体素渲染框架。我们的方法在单次渲染过程中联合光栅化来自多个独立八叉树的稀疏体素。每个刚性动态物体由在其局部坐标系中定义的八叉树表示，而一个单独的静态八叉树则建模静止背景。DrivingVoxels采用完全显式的、无神经网络的表示，并结合激光雷达引导的结构初始化，有效捕捉场景几何。我们在PandaSet基准上评估了我们的框架，结果表明DrivingVoxels在感知指标上表现相当，在NVS和重建的结构指标上表现更佳，同时所需的训练时间比之前的基于3DGS的方法短，优化工作流程高效且依赖于强大的激光雷达先验。
cs.CV / 224 / 2606.23041
SPAR: Semantic-Pixel Self-Alignment and Adaptive Routing for Unified Multimodal Models
SPAR：语义像素自对齐与自适应路由的统一多模态模型
Abstract
Multimodal Large Language Models (MLLMs) have achieved remarkable success in visual understanding but remain constrained in visual generation due to the fundamental feature discrepancy between semantic perception and pixel-level reconstruction. Bridging this gap requires overcoming two core challenges: endowing semantic encoders with high-fidelity reconstruction capabilities, and effectively aligning generative models with semantic spaces without relying on external teachers. To this end, we propose a novel unified multimodal framework featuring \textbf{S}emantic-\textbf{P}ixel self-alignment and \textbf{A}daptive \textbf{R}outing (\textbf{SPAR}). First, to reconcile semantic perception with pixel-level reconstruction, we introduce an asymmetric dual-stream unified tokenizer. A lightweight semantic stream anchors discriminative features, while a Transformer-augmented pixel stream recovers fine-grained visual details into a unified compact latent space. Second, to eliminate external dependencies, we propose a self-aligned generation paradigm that natively leverages this optimized tokenizer as an internal alignment teacher for the diffusion model. Furthermore, to facilitate flexible multimodal interaction within this unified space, we introduce Dynamic Token Routing, which enables each token to adaptively aggregate multi-layer MLLM features based on its distinct semantic demands. Extensive experiments demonstrate that SPAR establishes the state-of-the-art for unified architectures, achieving exceptional generation and reconstruction quality while preserving foundational visual understanding capabilities.
Chinese Translation
多模态大型语言模型（MLLMs）在视觉理解方面取得了显著成功，但由于语义感知与像素级重建之间的基本特征差异，仍然在视觉生成方面受到限制。弥合这一差距需要克服两个核心挑战：赋予语义编码器高保真重建能力，以及有效地将生成模型与语义空间对齐，而不依赖于外部教师。为此，我们提出了一种新颖的统一多模态框架，具有 extbf{S}emantic- extbf{P}ixel自对齐和 extbf{A}daptive extbf{R}outing（ extbf{SPAR}）。首先，为了调和语义感知与像素级重建，我们引入了一种不对称双流统一标记器。轻量级的语义流锚定了判别特征，而增强型Transformer的像素流则将细粒度视觉细节恢复到统一的紧凑潜在空间。其次，为了消除外部依赖，我们提出了一种自对齐生成范式，原生利用这一优化的标记器作为扩散模型的内部对齐教师。此外，为了促进在这一统一空间内的灵活多模态交互，我们引入了动态标记路由，使每个标记能够根据其独特的语义需求自适应地聚合多层MLLM特征。大量实验表明，SPAR在统一架构方面建立了最先进的水平，达到了卓越的生成和重建质量，同时保持了基础视觉理解能力。
cs.CV / 225 / 2606.23046
UECP: Uncertainty-Enhanced Collaborative Perception
UECP：增强不确定性的协同感知
Abstract
Collaborative perception serves as a pivotal solution to enhance the perception capability of individual agents in autonomous driving, where a core challenge lies in seeking reliable evidence to quantify and weight the contribution of each participating agent. Existing methods typically rely on a confidence map, which is co-trained with the detection head, but it is inherently correlated with the detection results and thus fails to provide unbiased physical evidence. Furthermore, how to deeply integrate evidence into the cooperative fusion process remains an open question. To address these issues, this paper first proposes an uncertainty map, a physically grounded and unambiguous metric for evaluating perception quality. This map is directly supervised by real-time sensor signals, i.e., LiDAR point density, ensuring decoupling from detection noise and thereby providing physical scenario-aware evidence for weighting agent contribution. Based on this map, we develop the Uncertainty-Enhanced Collaborative Perception (UECP) framework, centered on the Uncertainty-Aware Pyramid Fusion (UAPF) module. UAPF uses a coarse-to-fine strategy, with two key components: Uncertainty-Weighted Downsampling (UWD) for high-fidelity feature preservation, and Uncertainty-Guided Residual Fusion (UGRF) to reinforce ego features, suppressing noise and ensuring robust fusion. Extensive experiments on real-world datasets show UECP outperforms state-of-the-art methods in effectiveness and robustness by embedding the uncertainty map into fusion. Code will be publicly available.
Chinese Translation
协同感知作为提升自主驾驶中个体智能体感知能力的关键解决方案，其核心挑战在于寻找可靠的证据来量化和权衡每个参与智能体的贡献。现有方法通常依赖于与检测头共同训练的置信度图，但其本质上与检测结果相关，因此未能提供无偏的物理证据。此外，如何将证据深入整合到协作融合过程中仍然是一个未解的问题。为了解决这些问题，本文首先提出了一种不确定性图，这是一种基于物理的、明确的感知质量评估指标。该图直接由实时传感器信号（即激光雷达点密度）进行监督，确保与检测噪声解耦，从而为权衡智能体贡献提供物理场景感知的证据。基于此图，我们开发了增强不确定性的协同感知（UECP）框架，重点是不确定性感知金字塔融合（UAPF）模块。UAPF采用自粗到细的策略，包含两个关键组件：不确定性加权下采样（UWD）用于高保真特征保留，以及不确定性引导的残差融合（UGRF）用于增强自我特征，抑制噪声并确保稳健的融合。对真实世界数据集的广泛实验表明，UECP在有效性和稳健性方面优于最先进的方法，通过将不确定性图嵌入融合中。代码将公开发布。
cs.CV / 226 / 2606.23050
Unlimited OCR Works
无限制OCR工作
Abstract
Recently, end-to-end OCR models, exemplified by DeepSeek OCR, have once again thrust OCR into the spotlight. A widely held view is that employing a large language model (LLM) as the decoder allows the model to leverage the prior distribution of language, leading to improved OCR performance. However, the downside is equally evident: as the output sequence lengthens, the accumulated KV cache drives up memory consumption and progressively slows down generation. This stands in stark contrast to humans, who exhibit no such decline in efficiency during long-horizon copying tasks. In this technical report, we propose Unlimited OCR, a model designed to emulate human parsing working memory. Taking DeepSeek OCR as the baseline, we replace all attention layers in the decoder with our proposed Reference Sliding Window Attention (R-SWA), which reduces attention computation costs while maintaining a constant KV cache throughout the entire decoding process. By combining the high compression rate of DeepSeek OCR's encoder with our constant KV cache design, Unlimited OCR can transcribe dozens of pages of documents in a single forward pass under a standard maximum length of 32K. More importantly, R-SWA is a general-purpose parsing attention mechanism - beyond OCR, it is equally applicable to tasks such as ASR, translation, etc. Codes and model weights are publicly available at http://github.com/baidu/Unlimited-OCR.
Chinese Translation
最近，以DeepSeek OCR为代表的端到端OCR模型再次将OCR推向了聚光灯下。普遍认为，采用大型语言模型（LLM）作为解码器能够使模型利用语言的先验分布，从而提高OCR性能。然而，缺点同样明显：随着输出序列的延长，累积的KV缓存导致内存消耗增加，并逐渐减慢生成速度。这与人类在长时间复制任务中表现出的效率没有下降形成鲜明对比。在本技术报告中，我们提出了无限制OCR（Unlimited OCR），该模型旨在模拟人类解析工作记忆。以DeepSeek OCR为基线，我们用我们提出的参考滑动窗口注意力机制（Reference Sliding Window Attention, R-SWA）替换了解码器中的所有注意力层，从而在整个解码过程中保持恒定的KV缓存，同时降低注意力计算成本。通过将DeepSeek OCR编码器的高压缩率与我们的恒定KV缓存设计相结合，无限制OCR能够在单次前向传播中在标准最大长度32K下转录数十页文档。更重要的是，R-SWA是一种通用的解析注意力机制——除了OCR之外，它同样适用于语音识别（ASR）、翻译等任务。代码和模型权重可在http://github.com/baidu/Unlimited-OCR获取。
cs.CV / 227 / 2606.23058
Three-Step Hierarchical Transformer for Multi-Pedestrian Trajectory Prediction
三步层次化变换器用于多行人轨迹预测
Abstract
Pedestrian trajectory prediction requires modeling temporal dynamics, multimodal cues, and social interactions in crowded environments. Existing methods often address these factors separately or entangle them in costly attention blocks, limiting scalability, flexibility, and interpretability. We propose a three-step hierarchical Transformer that explicitly separates temporal encoding, multimodal fusion, and scene-level interaction reasoning. Lightweight GRU summaries enable efficient cross-modal attention, while social attention over time--agent tokens captures inter-pedestrian influences at manageable cost. Experiments on JTA, JRDB, and the Pedestrians and Cyclists in Road Traffic dataset show state-of-the-art performance on real-world datasets (JRDB, Urban) and competitive results on JTA. Ablation and qualitative analyses confirm the contribution of each stage and the model's ability to anticipate complex behaviors such as early turning.
Chinese Translation
行人轨迹预测需要在拥挤环境中建模时间动态、多模态线索和社会互动。现有方法通常分别处理这些因素，或将其纠缠在昂贵的注意力模块中，从而限制了可扩展性、灵活性和可解释性。我们提出了一种三步层次化变换器，明确分离时间编码、多模态融合和场景级互动推理。轻量级的GRU摘要实现了高效的跨模态注意力，而基于时间的社会注意力——代理标记则以可管理的成本捕捉行人之间的相互影响。在JTA、JRDB和道路交通中的行人和骑自行车者数据集上的实验显示，在真实世界数据集（JRDB、Urban）上达到了最先进的性能，并在JTA上取得了具有竞争力的结果。消融和定性分析证实了每个阶段的贡献以及模型预测复杂行为（如提前转向）的能力。
cs.CV / 228 / 2606.23061
MotionHalluc: Diagnosing Kinematic Hallucinations in Fine-Grained Motion Reasoning
MotionHalluc：诊断细粒度运动推理中的运动幻觉
Abstract
Motion instruction generation in cross-video comparison aims to produce corrective feedback that describes the differences between a query and a reference motion. However, existing models often generate instructions that exhibit motion hallucinations, failing to reflect actual kinematic differences between paired videos. To systematically investigate these hallucinations, we introduce MotionHalluc, a dedicated benchmark for evaluating motion hallucinations in paired-video comparison. MotionHalluc comprises 1540 fine-grained questions over 553 video pairs, evaluating hallucinations along three core dimensions: (1)directional hallucination, (2)attributional hallucination, and (3)temporal hallucination. Extensive evaluations of state-of-the-art large multimodal models demonstrate high susceptibility to these hallucinations. Furthermore, we provide Perceive-Parse-Verify (PPV) as a training-free measurements extraction and verification baseline that converts candidate instructions into executable measurement queries and supplies kinematic measurements at inference time. Our results show that this simple measurements injection yields an average 10.6% performance gain across models, suggesting that motion reasoning with explicit quantitative measurements is a key factor in reducing hallucinations in cross-video comparison. Our code and dataset will be made publicly available upon acceptance.
Chinese Translation
跨视频比较中的运动指令生成旨在产生描述查询与参考运动之间差异的纠正反馈。然而，现有模型往往生成表现出运动幻觉的指令，未能反映配对视频之间的实际运动学差异。为了系统地研究这些幻觉，我们提出了MotionHalluc，这是一个专门用于评估配对视频比较中运动幻觉的基准。MotionHalluc包含1540个细粒度问题，涵盖553对视频，从三个核心维度评估幻觉：（1）方向幻觉，（2）归因幻觉，以及（3）时间幻觉。对最先进的大型多模态模型的广泛评估表明，这些模型对这些幻觉高度敏感。此外，我们提供了Perceive-Parse-Verify (PPV)作为一种无训练的测量提取和验证基线，将候选指令转换为可执行的测量查询，并在推理时提供运动学测量。我们的结果表明，这种简单的测量注入在各模型中平均提高了10.6%的性能，表明具有明确定量测量的运动推理是减少跨视频比较中幻觉的关键因素。我们的代码和数据集将在接受后公开发布。
cs.CV / 229 / 2606.23063
Attention-Spectrum Regularization for Replay-Free Continual Multimodal LLMs
无重放的持续多模态大语言模型的注意力谱正则化
Abstract
Multimodal large language models (MLLMs) are increasingly required to adapt to non-stationary streams of visual domains, question types, and user instructions, yet continual fine-tuning often causes severe forgetting of previously acquired multimodal skills. Existing continual vision-language methods mainly preserve outputs, replay data or pseudo-data, regularize embedding geometry, or allocate task-specific parameters, but they provide limited control over how internal cross-modal attention patterns supporting old skills drift during adaptation. We propose Attention-Spectrum Regularization (ASR), a replay-free continual learning framework that preserves skill-conditioned structures of cross-modal attention. ASR treats cross-attention maps as two-dimensional signals, summarizes their scale and directional properties into compact spectral statistics, and stores only skill-wise prototype distributions instead of replaying past image-question pairs, generated pseudo-examples, or old-stage teacher snapshots. In later stages, a phase-invariant spectral regularizer constrains harmful drift of these prototypes while allowing instance-level attention to adapt to new tasks. We provide theoretical analysis showing that skill-conditioned spectral drift controls forgetting under a spectral sufficiency assumption, and that Fourier power spectra are stable to spatial translations and bounded perturbations. Experiments on continual VQA and multimodal instruction-tuning benchmarks, including VQA v2, VQACL, CLT-VQA, CoIN, and UCIT, show that ASR consistently improves final performance and reduces forgetting over strong replay-, regularization-, and adapter-based baselines. Preserving skill-level attention structure is an effective and lightweight mechanism for continual MLLMs. Code is available at https://github.com/Creative-zcx/attention-spectrum-replay
Chinese Translation
多模态大语言模型（MLLMs）越来越需要适应非平稳的视觉领域、问题类型和用户指令流，但持续微调往往会导致之前获得的多模态技能严重遗忘。现有的持续视觉-语言方法主要通过保留输出、重放数据或伪数据、正则化嵌入几何或分配任务特定参数来实现，但它们对支持旧技能的内部跨模态注意力模式在适应过程中如何漂移的控制有限。我们提出了注意力谱正则化（Attention-Spectrum Regularization, ASR），一种无重放的持续学习框架，旨在保留跨模态注意力的技能条件结构。ASR将跨注意力图视为二维信号，将其规模和方向特性总结为紧凑的谱统计，并仅存储技能相关的原型分布，而不是重放过去的图像-问题对、生成的伪例或旧阶段教师快照。在后续阶段，相位不变的谱正则化器约束这些原型的有害漂移，同时允许实例级注意力适应新任务。我们提供了理论分析，表明在谱充分性假设下，技能条件谱漂移控制遗忘，并且傅里叶功率谱对空间平移和有界扰动是稳定的。在持续视觉问答（VQA）和多模态指令调优基准（包括VQA v2、VQACL、CLT-VQA、CoIN和UCIT）上的实验表明，ASR在最终性能上持续改善，并减少了相较于强重放、正则化和适配器基线的遗忘。保留技能级注意力结构是持续多模态大语言模型的有效且轻量的机制。代码可在 https://github.com/Creative-zcx/attention-spectrum-replay 获取。
cs.CV / 230 / 2606.23069
Rethinking Prototype-based Similarity Learning for Few-Shot Object Detection
重新思考基于原型的少样本目标检测相似性学习
Abstract
Few-shot object detection aims to detect novel object categories from only a few labeled examples, avoiding costly large-scale annotation. Recent prototype-based similarity learning approaches enable training-free adaptation by matching query features with class prototypes. However, they suffer from two fundamental limitations: (i) class confusion arising from inter-class similarity margin collapse, and (ii) insufficient visual cues for precise localization, as similarity scores capture only class-level semantic affinity while providing limited spatial information. To address these issues, we introduce two complementary components. Text-Anchored Semantic Mask (TSMa) leverages class-level text features as semantic anchors to identify semantically aligned channels through channel-wise interaction between visual and text features. By suppressing style-induced spurious responses and emphasizing class-intrinsic signals, TSMa enlarges inter-class similarity margins and mitigates class confusion. We further propose Stage-Aligned Hierarchical Autoregressive Regression (SHARe), which reformulates localization as a hierarchical autoregressive process that progressively refines bounding boxes across multiple stages. SHARe leverages the layer-wise characteristics of ViT representations by aligning feature abstraction levels with regression stages: deeper layers guide early coarse localization, while shallower layers rich in edge and texture cues refine spatial details in later stages. Experiments on COCO demonstrate a new state of the art, outperforming the previous best by +10.1 nAP, with extensive analysis validating each component. The code is available at https://github.com/VisualScienceLab-KHU/ReSet.
Chinese Translation
少样本目标检测旨在仅通过少量标注示例检测新颖的物体类别，从而避免昂贵的大规模标注。近期的基于原型的相似性学习方法通过将查询特征与类别原型进行匹配，实现了无训练的适应。然而，这些方法存在两个基本局限性：（i）由于类别间相似性边界崩溃而导致的类别混淆，以及（ii）由于相似性分数仅捕捉类别级语义亲和力而提供有限空间信息，导致精确定位的视觉线索不足。为了解决这些问题，我们引入了两个互补组件。文本锚定语义掩码（Text-Anchored Semantic Mask, TSMa）利用类别级文本特征作为语义锚点，通过视觉特征与文本特征之间的通道交互识别语义对齐的通道。通过抑制由风格引起的虚假响应并强调类别内在信号，TSMa扩大了类别间相似性边界并减轻了类别混淆。我们进一步提出了阶段对齐的层次自回归回归（Stage-Aligned Hierarchical Autoregressive Regression, SHARe），将定位重新表述为一个层次自回归过程，该过程在多个阶段逐步细化边界框。SHARe通过将特征抽象层次与回归阶段对齐，利用ViT表示的层级特征：较深的层引导早期粗略定位，而较浅的层则富含边缘和纹理线索，在后期阶段细化空间细节。在COCO上的实验展示了新的最先进成果，超越了之前的最佳结果10.1 nAP，并通过广泛分析验证了每个组件。代码可在 https://github.com/VisualScienceLab-KHU/ReSet 获取。
cs.CV / 231 / 2606.23098
Poisson2Gaussian: Noise Gaussianization to Enhance Image Denoising
Poisson2Gaussian：噪声高斯化以增强图像去噪
Abstract
The quantum nature of light determines the inherent Poisson stochasticity of photon detection, which is ubiquitous in photography, microscopy, and astronomy. However, our controlled numerical studies reveal that the signal-dependency, heteroscedasticity, and statistical asymmetry of Poisson-mixed noise make it challenging for existing denoisers to learn. In contrast, i.i.d. Gaussian noise, with its statistical independence and symmetric distribution, is easier to model for networks. To address this gap, we propose Poisson2Gaussian (P2G), a noise Gaussianization method that explicitly converts complex real-world noise to i.i.d. Gaussian noise via probability density matching beyond low-order moments. We also design an unbiased denoising framework that synergizes P2G with downstream denoisers, ensuring convergence to the underlying signal without requiring paired clean data or explicit noise parameters. Extensive experiments demonstrate that P2G consistently achieves state-of-the-art performance across diverse datasets. In challenging scenarios where noise strongly deviates from Gaussian statistics, our method improves the PSNR by up to 0.75 dB. Notably, P2G is architecture-agnostic and can provide universal improvements for various denoisers. The source code will be publicly available.
Chinese Translation
光的量子特性决定了光子探测的固有泊松随机性，这在摄影、显微镜和天文学中普遍存在。然而，我们的受控数值研究表明，泊松混合噪声的信号依赖性、异方差性和统计不对称性使得现有去噪器难以学习。相比之下，独立同分布的高斯噪声由于其统计独立性和对称分布，更容易被网络建模。为了解决这一问题，我们提出了Poisson2Gaussian (P2G)，一种噪声高斯化方法，通过超越低阶矩的概率密度匹配，明确地将复杂的现实世界噪声转换为独立同分布的高斯噪声。我们还设计了一个无偏去噪框架，将P2G与下游去噪器协同，确保在不需要配对干净数据或显式噪声参数的情况下收敛到潜在信号。大量实验表明，P2G在各种数据集上始终实现了最先进的性能。在噪声显著偏离高斯统计的挑战性场景中，我们的方法将峰值信噪比（PSNR）提高了最多0.75 dB。值得注意的是，P2G与架构无关，能够为各种去噪器提供普遍的改进。源代码将公开发布。
cs.CV / 232 / 2606.23101
Scene-agnostic ALS boresight self-calibration
场景无关的ALS瞄准自校准
Abstract
ALS boresight calibration has relied for two decades on dedicated flight patterns over structured scenes containing planar surfaces of varied aspect and slope. While reliable, this approach imposes constraints on the scene content and operations, which limits its applicability to boresight recovery within routine mapping missions. We present a practical approach that substantially relaxes these requirements by replacing plane-based constraints with scene-agnostic point-to-point correspondences extracted automatically from overlapping ALS strips. Two complementary formulations are proposed to estimate boresight with laser vector observations: (i) a simpler parametric adjustment utilizing INS/GNSS trajectory; (ii) a rigorous formulation treating GNSS and raw inertial data within an existing factor-graph, i.e. a dynamic network, where boresight is added as an additional parameter. Both formulations are evaluated across four operational ALS flights equipped with five inertial systems, covering a wide range of flight altitudes, overlap geometries, terrain types and inertial sensor classes. The analysis draws a clear boundary between the legacy plane-based conditioning that falls short outside the calibration scenario and the proposed formulations, which either recover or absorb boresight effects under conventional mapping geometry. Among them, the lightweight formulation is sufficient for boresight recovery using tactical and navigation grade inertial sensors, while the general factor-graph approach is clearly superior when the inertial sensor errors are less observable within an optimal smoother. This supports the hypothesis that, for INS/GNSS trajectory of sufficient quality, the boresight calibration can be performed without particular scene prerequisites during routine mapping operations using a minimum of 3-4 overlapping strips, with either proposed formulation...
Chinese Translation
ALS瞄准校准在过去二十年中依赖于在包含不同外形和坡度平面表面的结构化场景上进行专门的飞行模式。尽管这种方法可靠，但它对场景内容和操作施加了限制，从而限制了其在常规测绘任务中恢复瞄准的适用性。我们提出了一种实用的方法，通过用从重叠的ALS条带中自动提取的场景无关点对点对应关系替代基于平面的约束，从而大大放宽了这些要求。我们提出了两种互补的公式来利用激光矢量观测估计瞄准：(i) 一种更简单的参数调整，利用INS/GNSS轨迹；(ii) 一种严格的公式，将GNSS和原始惯性数据视为现有因子图中的一部分，即动态网络，其中瞄准作为额外参数添加。两种公式在四次配备五个惯性系统的操作ALS飞行中进行了评估，涵盖了广泛的飞行高度、重叠几何、地形类型和惯性传感器类别。分析清晰地区分了在校准场景之外表现不佳的传统平面基础条件与所提出的公式，这些公式在常规测绘几何下要么恢复要么吸收瞄准效应。其中，轻量级公式足以利用战术和导航级惯性传感器进行瞄准恢复，而当惯性传感器误差在最佳平滑器中不易观察时，通用因子图方法显然更优。这支持了这样一种假设：对于足够质量的INS/GNSS轨迹，瞄准校准可以在常规测绘操作中无需特定场景前提下进行，使用最少的3-4条重叠条带，采用任一提出的公式。
cs.CV / 233 / 2606.23105
Compression and Retrieval: Implicit Memory Retrieval for Video World Models
压缩与检索：视频世界模型的隐式记忆检索
Abstract
Video world models hold promise for simulating interactive environments, yet maintaining consistent long-term memory across complex camera trajectories remains a critical challenge. Existing methods typically rely on computationally expensive context scaling or rigid heuristic retrieval mechanisms, which lacks generalization to varying camera trajectories and environments. In this paper, we propose Compression and Retrieval (CaR), an attention-driven implicit memory retrieval mechanism to overcome these limitations. By injecting viewpoint information via positional encoding, our method performs flexible memory retrieval through attention computation. To efficiently process extended contexts with minimal computational overhead, we further introduce a lightweight context compression network. Furthermore, we construct SceneFly, a large-scale synthetic dataset featuring realistic camera trajectories and frame-level annotations to train and evaluate long-horizon video world models. Extensive experiments demonstrate that our approach achieves state-of-the-art results on established benchmarks and exhibits strong generalization to open-domain scenes.
Chinese Translation
视频世界模型在模拟互动环境方面具有潜力，但在复杂相机轨迹下保持一致的长期记忆仍然是一个关键挑战。现有方法通常依赖于计算成本高昂的上下文缩放或刚性启发式检索机制，这些方法在不同的相机轨迹和环境中缺乏泛化能力。本文提出了一种压缩与检索（Compression and Retrieval, CaR）的方法，这是一种基于注意力的隐式记忆检索机制，以克服这些局限性。通过通过位置编码注入视点信息，我们的方法通过注意力计算实现灵活的记忆检索。为了以最小的计算开销有效处理扩展上下文，我们进一步引入了一种轻量级的上下文压缩网络。此外，我们构建了SceneFly，一个大型合成数据集，具有真实的相机轨迹和帧级注释，用于训练和评估长时视频世界模型。大量实验表明，我们的方法在已建立的基准测试中取得了最先进的结果，并对开放域场景表现出强大的泛化能力。
cs.CV / 234 / 2606.23113
Technical Report for the ICRA 2026 GOOSE 2D Fine-Grained Semantic Segmentation Challenge: Pretraining-Diverse Ensemble of Foundation Vision Encoders for Robust Outdoor Scene Understanding
ICRA 2026 GOOSE 2D 精细语义分割挑战技术报告：用于鲁棒户外场景理解的基础视觉编码器的预训练多样化集成
Abstract
This report presents our solution for the ICRA 2026 GOOSE 2D Fine-Grained Semantic Segmentation Challenge, which requires parsing unstructured outdoor scenes from four camera platforms into 56 fine-grained categories. Our approach pairs foundation vision encoders (including DINOv3, SigLIP2, and InternImage) with a Mask2Former decoder, and trains them with a strong recipe including long training schedules, exponential moving average, a larger crop size, and multi-scale plus flip test-time augmentation. The three encoders, chosen for their complementary pretraining objectives, are combined into a pretraining-diverse ensemble through per-class validation-IoU weighting. Evaluated on the official GOOSE test set, our submission achieves 75.40% composite mIoU and wins the second place of the challenge. Our study further shows that the encoder's pretraining recipe, rather than its parameter count or the decoder design, is the dominant factor for accuracy on this benchmark.
Chinese Translation
本报告展示了我们针对ICRA 2026 GOOSE 2D精细语义分割挑战的解决方案，该挑战要求将来自四个摄像头平台的非结构化户外场景解析为56个精细类别。我们的方法将基础视觉编码器（包括DINOv3、SigLIP2和InternImage）与Mask2Former解码器配对，并采用包括长时间训练计划、指数移动平均、更大的裁剪尺寸以及多尺度加翻转测试时间增强在内的强大训练方案进行训练。这三种编码器因其互补的预训练目标而被选中，通过每类验证IoU加权组合成一个预训练多样化集成。在官方GOOSE测试集上的评估中，我们的提交达到了75.40%的综合mIoU，并获得了挑战的第二名。我们的研究进一步表明，编码器的预训练方案，而非其参数数量或解码器设计，是该基准测试中准确性的主导因素。
cs.CV / 235 / 2606.23118
LUMINA-26: Low-Light Understanding for Modeling and Interpreting Night-time Actions
LUMINA-26：低光照条件下夜间行为建模与解释的理解
Abstract
Low-light human action recognition remains a challenging problem due to poor illumination, amplified noise, motion ambiguity, and diverse real-world scenes. Existing low-light datasets often lack sufficient action diversity, capture realism, or balanced class distribution, limiting the development of robust models. To address this, we introduce LUMINA-26: Low-Light Understanding for Modeling and Interpreting Night-time Actions, comprising 6,784 clips across 26 action classes, recorded from 22 subjects across 20 indoor and outdoor locations under naturally occurring low-light conditions. We also propose Illumi-Net: An Illumination-Adaptive Mixture-of-Experts Network, which leverages video-level illumination cues to guide adaptive enhancement and transformer-based spatio-temporal feature extraction, with expert-conditioned decision fusion. Our method surpasses previous state-of-the-art performance on ELLAR (Top-1: 55.13%, Top-5: 78.87%) and establishes a strong baseline on LUMINA-26 (Top-1: 75.95%, Top-5: 93.58%), offering a practical benchmark for future low-light action recognition research.
Chinese Translation
低光照下的人类行为识别仍然是一个具有挑战性的问题，主要由于照明不足、噪声放大、运动模糊以及多样化的现实场景。现有的低光照数据集往往缺乏足够的行为多样性、捕捉真实感或平衡的类别分布，从而限制了稳健模型的发展。为了解决这一问题，我们推出了LUMINA-26：低光照条件下夜间行为建模与解释的理解，包含6,784个视频片段，涵盖26个行为类别，记录自22名受试者在20个室内和室外地点的自然低光照条件下。我们还提出了Illumi-Net：一种照明自适应的专家混合网络，利用视频级照明线索指导自适应增强和基于变换器的时空特征提取，并进行专家条件下的决策融合。我们的方法在ELLAR数据集上超越了之前的最先进性能（Top-1: 55.13%，Top-5: 78.87%），并在LUMINA-26上建立了强有力的基准（Top-1: 75.95%，Top-5: 93.58%），为未来的低光照行为识别研究提供了一个实用的基准。
cs.CV / 236 / 2606.23126
MambaADv2: Evolving Duality-enhanced State Space Model for Unsupervised Anomaly Detection
MambaADv2：演化的双重增强状态空间模型用于无监督异常检测
Abstract
While recent advancements in anomaly detection have demonstrated the efficacy of CNN- and Transformer-based approaches, these architectures face inherent limitations: CNNs struggle to capture long-range dependencies, whereas Transformers suffer from quadratic computational complexity. Consequently, Mamba-based architectures have attracted considerable attention, as they successfully combine superior long-range dependency modeling with linear computational complexity. By critically rethinking the structural evolution across the Mamba lineage 1-3 series, this paper proposes MambaADv2, a framework tailored for multi-class unsupervised anomaly detection. MambaADv2 comprises a pre-trained encoder and a Mamba-inspired decoder, equipped with Duality-enhanced State Space (DSS) modules across multiple scales. The proposed DSS module effectively models both global dependencies and local representations by integrating parallel-cascaded Hybrid State Space (HSS) blocks and frequency-enhanced convolution operations. The structure of the Hybrid State Space (HSS) block is tailored by following the SSD-based Mamba lineage and incorporating Mamba3-style position-aware state-space modeling, leveraging the dual computational paths of linear recurrence and parallel matrix formulation to model local continuity and global contextual comparison, thereby better serving the core anomaly detection objective of precisely reconstructing normal representations while magnifying anomalous deviations. Additionally, we propose a semantics-adaptive progressive scanning strategy that decays scanning complexity along the feature pyramid.
Chinese Translation
尽管近期在异常检测方面的进展展示了基于卷积神经网络（CNN）和变换器（Transformer）方法的有效性，但这些架构面临固有的局限性：CNN在捕捉长距离依赖方面存在困难，而变换器则遭遇二次计算复杂度的问题。因此，基于Mamba的架构引起了广泛关注，因为它们成功地将优越的长距离依赖建模与线性计算复杂度相结合。通过对Mamba系列1-3的结构演变进行深入思考，本文提出了MambaADv2，一个专为多类无监督异常检测量身定制的框架。MambaADv2包含一个预训练的编码器和一个受Mamba启发的解码器，配备了跨多个尺度的双重增强状态空间（DSS）模块。所提出的DSS模块通过集成并行级联的混合状态空间（HSS）块和频率增强卷积操作，有效地建模了全局依赖和局部表示。混合状态空间（HSS）块的结构是通过遵循基于SSD的Mamba系列并结合Mamba3风格的基于位置的状态空间建模进行定制的，利用线性递归和并行矩阵公式的双重计算路径来建模局部连续性和全局上下文比较，从而更好地服务于精确重建正常表示的核心异常检测目标，同时放大异常偏差。此外，我们提出了一种语义自适应的渐进扫描策略，该策略沿特征金字塔逐渐降低扫描复杂度。
cs.CV / 237 / 2606.23129
Spectral Gating via Damped Oscillations for Adaptive Implicit Neural Representations
通过阻尼振荡的光谱门控用于自适应隐式神经表示
Abstract
Implicit Neural Representations (INRs) have been proven successful in encoding continuous signals through coordinate-based networks, yet facing a spectral dilemma: periodic activations capture fine details but act as all-pass filters that memorise noise, while spatially compact activations regularise effectively but suffer from low-frequency bias. Existing attempts to resolve this trade-off introduce computational overhead or tuning frailty. We propose to model each neuron's activation as the steady-state response of a sinusoidally-forced damped harmonic oscillator, whose amplitude naturally governs the network's spectral selectivity during training. By jointly optimising the oscillator parameters alongside the network weights, our method adapts to the target signal's spectral content without explicit regularisation. Initialised in the stopband, the network exhibits a coarse-to-fine learning curriculum that progressively expands its spectral gate, capturing low-frequency structures first and high-frequency details only when justified by the reconstruction objective. Comprehensive experiments show that our approach consistently achieves state-of-the-art or competitive results against established INRs, while requiring no task-specific tuning of any hyperparameters.
Chinese Translation
隐式神经表示（Implicit Neural Representations, INRs）已被证明能够成功地通过基于坐标的网络编码连续信号，但面临光谱困境：周期性激活捕捉细节，但作为全通滤波器会记忆噪声，而空间紧凑的激活有效地进行正则化，但受到低频偏差的影响。现有的解决方案引入了计算开销或调优脆弱性。我们提出将每个神经元的激活建模为一个正弦强迫阻尼谐振子的稳态响应，其幅度自然地在训练过程中控制网络的光谱选择性。通过联合优化谐振子参数和网络权重，我们的方法能够适应目标信号的光谱内容，而无需显式正则化。在停止带中初始化时，网络展示了一种从粗到细的学习课程，逐步扩展其光谱门，首先捕捉低频结构，仅在重建目标合理时才捕捉高频细节。全面的实验表明，我们的方法在与已建立的INRs的比较中始终实现了最先进或具有竞争力的结果，同时无需对任何超参数进行特定任务的调优。
cs.CV / 238 / 2606.23131
Expert Consensus on Criteria for the Automated Assessment of Laparoscopic Camera Navigation
关于腹腔镜摄像机导航自动评估标准的专家共识
Abstract
Background: Laparoscopic camera navigation (LCN) is a critical skill, yet its current assessment typically relies on manual rating systems which are time-consuming and difficult to scale. Automated feedback could significantly enhance surgical training by providing immediate, standardized metrics. This study aims to define, clinically evaluate the relevance, and establish the technical readiness of a set of approaches for LCN assessment. Methods: We developed a detailed taxonomy of 14 key aspects of camera navigation, categorized into Framing & Composition, Visibility & Clarity, Orientation & Stability, Motion & Dynamics, and Safety & Awareness. For each aspect, we assessed the technological readiness of automated measurement based on the current state of the art (SoTA) in computer vision (CV). To establish clinical relevance, we designed a survey for practicing laparoscopic surgeons to rate the importance of each aspect on a 5-point Likert scale and to select the five most critical skills. Results: 23 surgeons participated in the survey. Foundational aspects like Field of View, Focus and Centering were rated as most important by surgeons. We present a "Clinical Importance vs. CV Technological Readiness" matrix, identifying high-priority targets for development--aspects that are both clinically crucial and technologically ready to measure. Conclusion: This work establishes a foundational framework for quantifying LCN skills. By aligning surgeon priorities with CV capabilities, we provide a clear roadmap for automatic skill assessment. This foundation enables the development of AI-driven assistance tools that can accelerate the learning curve for surgical assistants and potentially improve surgical safety and efficiency.
Chinese Translation
背景：腹腔镜摄像机导航（LCN）是一项关键技能，但其当前评估通常依赖于耗时且难以扩展的手动评分系统。自动反馈可以通过提供即时、标准化的指标显著提升外科培训。本研究旨在定义、临床评估相关性，并建立一套用于LCN评估的方法的技术准备性。方法：我们开发了一个详细的分类法，涵盖14个摄像机导航的关键方面，分为框架与构图、可见性与清晰度、方向与稳定性、运动与动态以及安全与意识。对于每个方面，我们根据计算机视觉（CV）的当前技术水平评估了自动测量的技术准备性。为了建立临床相关性，我们设计了一项调查，邀请在职的腹腔镜外科医生对每个方面的重要性进行5点李克特量表评分，并选择五项最关键的技能。结果：23名外科医生参与了调查。外科医生认为视野、对焦和中心定位等基础方面最为重要。我们呈现了“临床重要性与CV技术准备性”矩阵，识别出高优先级的发展目标——即临床至关重要且技术上已准备好测量的方面。结论：本研究建立了量化LCN技能的基础框架。通过将外科医生的优先事项与CV能力对齐，我们提供了一条清晰的自动技能评估路线图。这一基础使得开发AI驱动的辅助工具成为可能，从而加速外科助手的学习曲线，并有潜力提高外科安全性和效率。
cs.CV / 239 / 2606.23132
T-VSS: Test-Time Visual Subspace Steering for Adversarial Robustness of Vision-Language Models
T-VSS：针对视觉-语言模型对抗鲁棒性的测试时间视觉子空间引导
Abstract
Vision-language models (VLMs) achieve strong zero-shot recognition, but they remain highly vulnerable to adversarial perturbations. Recent test-time adaptations improve robustness without retraining, but they do not directly adapt the corrupted visual representation itself. Prompt-based methods adapt the learnable text prompts, while input-space methods optimize pixels or padding at test time. These approaches can improve predictions, but they do so through an indirect and expensive optimization path. We propose Test-time Visual Subspace Steering (T-VSS), a lightweight defense that performs test-time adaptation directly in the visual feature space. T-VSS first builds a sample-specific low-rank subspace from multi-view feature residuals anchored at the attacked image. It then learns a shared feature correction within this subspace using reliability-weighted entropy minimization. By constraining adaptation to a compact visual geometry, T-VSS steers attacked features toward more stable and discriminative predictions while avoiding noisy full-space updates. Experiments on fine-grained, ImageNet, and ImageNet-OOD benchmarks show that T-VSS improves adversarial robustness while maintaining competitive clean accuracy and better efficiency than prior test-time adaptations.
Chinese Translation
视觉-语言模型（VLMs）在零样本识别中表现出色，但仍然对对抗扰动高度脆弱。近期的测试时间适应方法在不重新训练的情况下提高了鲁棒性，但并未直接适应受损的视觉表示。基于提示的方法适应可学习的文本提示，而输入空间方法则在测试时优化像素或填充。这些方法可以改善预测，但通过间接且昂贵的优化路径实现。我们提出了测试时间视觉子空间引导（T-VSS），这是一种轻量级防御方法，直接在视觉特征空间中进行测试时间适应。T-VSS首先从锚定在攻击图像上的多视图特征残差构建样本特定的低秩子空间。然后，它利用可靠性加权的熵最小化在该子空间内学习共享特征修正。通过将适应限制在紧凑的视觉几何中，T-VSS将被攻击的特征引导向更稳定和更具辨别力的预测，同时避免了嘈杂的全空间更新。在细粒度、ImageNet和ImageNet-OOD基准上的实验表明，T-VSS在提高对抗鲁棒性的同时，保持了竞争力的清晰准确性，并且比之前的测试时间适应方法具有更好的效率。
cs.CV / 240 / 2606.23144
Koshur Pixel: a large-scale synthetic ocr dataset for kashmiri
Koshur Pixel：一个大规模的克什米尔合成OCR数据集
Abstract
Optical Character Recognition (OCR) for low-resource languages is often constrained by the lack of annotated training data and the complexity of script-specific rendering. Kashmiri, written primarily in the Perso-Arabic Nastaliq script, presents additional challenges due to contextual glyph shaping, dense ligatures, and orthographic variability. We introduce Koshur Pixel, the first large-scale synthetic OCR dataset for Kashmiri, comprising 613,078 image-text pairs generated from the KS-PRET-5M corpus using the SynthOCR-Gen framework. The dataset spans multiple fonts and textual granularities, ranging from individual words to full-page documents, and incorporates more than 25 augmentation strategies that emulate real-world document degradations. Koshur Pixel provides a scalable and cost-effective alternative to manual annotation, establishing a foundational resource for training OCR systems, digitizing Kashmiri textual heritage, and advancing language technologies for a severely under-resourced language.
Chinese Translation
对于低资源语言的光学字符识别（OCR），通常受到缺乏标注训练数据和特定书写系统渲染复杂性的限制。克什米尔语主要使用波斯-阿拉伯Nastaliq书写，因其上下文字形塑造、密集连字和正字法变异性而面临额外挑战。我们推出了Koshur Pixel，这是第一个大规模的克什米尔合成OCR数据集，包含613,078对图像-文本对，这些数据是通过使用SynthOCR-Gen框架从KS-PRET-5M语料库生成的。该数据集涵盖多种字体和文本粒度，从单个词汇到整页文档，并结合了超过25种增强策略，以模拟现实世界文档的退化。Koshur Pixel提供了一种可扩展且具有成本效益的替代手段，取代手动标注，为训练OCR系统、数字化克什米尔文本遗产以及推动严重资源不足语言的语言技术奠定了基础资源。
cs.CV / 241 / 2606.23177
Interpretable Probabilistic Medical Image Segmentation via Gaussian Process with Explicit Modelling of Annotation Bias and Variability
通过高斯过程显式建模注释偏差和变异性的可解释概率医学图像分割
Abstract
Deep learning-based medical image segmentation models are trained using annotations that exhibit systematic bias and variability across raters. While probabilistic multi-rater approaches can emulate annotator-specific delineations, annotator characteristics are typically encoded implicitly in deep latent feature space, making direct analysis of their influence on predictive distributions less straightforward. We propose a logit-space probabilistic segmentation framework based on stochastic variational Gaussian Process that explicitly decomposes predictions into an image-dependent reference logit distribution and annotator specific perturbations parameterised by bias and variance. This formulation enables more explicit analysis on how intra- and inter-rater variability propagate to predictive distributions. We evaluate the method on a multi-annotator medical image dataset, which shows that explicitly modelling annotator specific perturbations improves uncertainty calibration while maintaining comparable segmentation accuracy, compared with state-of-the-art multi-rater probabilistic segmentation method. The learned bias and variance parameters quantitatively reflect annotator-specific behaviour. Furthermore, controlled perturbation experiments over bias and variance demonstrate how changes in annotator parameters systematically influence predictive performance. The code used in this paper is made publicly available at https://github.com/QiLi111/GPS-Var.
Chinese Translation
基于深度学习的医学图像分割模型使用的注释在评估者之间表现出系统性的偏差和变异性。尽管概率多评估者方法可以模拟特定评估者的划分，但评估者特征通常隐式编码在深层潜在特征空间中，使得直接分析其对预测分布的影响变得不那么直接。我们提出了一种基于随机变分高斯过程的对数几率空间概率分割框架，该框架显式地将预测分解为依赖于图像的参考对数几率分布和由偏差和方差参数化的特定评估者的扰动。这种表述使得对评估者内部和外部变异性如何传播到预测分布的分析更加明确。我们在一个多评估者医学图像数据集上评估了该方法，结果表明，显式建模特定评估者的扰动在保持可比的分割准确性的同时，提高了不确定性校准，相较于最先进的多评估者概率分割方法。学习到的偏差和方差参数定量反映了特定评估者的行为。此外，对偏差和方差的控制扰动实验展示了评估者参数的变化如何系统性地影响预测性能。本文使用的代码已公开发布在 https://github.com/QiLi111/GPS-Var。
cs.CV / 242 / 2606.23186
StreamPPG: Low-Latency rPPG Estimation via Consistent Privileged Learning
StreamPPG：通过一致性特权学习实现低延迟的远程光电容积脉搏（rPPG）估计
Abstract
Remote photoplethysmography (rPPG) estimates the blood volume pulse (BVP) signal from facial videos, enabling contact-free health monitoring. Conventional clip-wise approaches, which use video clips as input, require capturing over one hundred frames before inference, thus introducing several seconds of delay and hindering real-time use. Meanwhile, frame-wise approaches struggle to capture long-range temporal and periodic features of physiological rhythms, and therefore lead to reduced estimation accuracy. To overcome these issues, we propose StreamPPG, a unified architecture that enables low-latency frame-wise physiological signal estimation while achieving competitive accuracy compared with clip-wise approaches. StreamPPG is trained under a consistent privileged learning (CPL) strategy, which leverages ground-truth rPPG signals as privileged information to enhance the model's representation capability. Extensive experiments demonstrate that StreamPPG achieves state-of-the-art accuracy across multiple datasets while maintaining real-time throughput on edge devices.
Chinese Translation
远程光电容积脉搏（rPPG）通过面部视频估计血容量脉搏（BVP）信号，从而实现无接触健康监测。传统的片段式方法使用视频片段作为输入，要求在推断之前捕获超过一百帧，从而引入几秒钟的延迟，妨碍实时使用。同时，逐帧方法难以捕捉生理节律的长时间范围和周期性特征，因此导致估计精度降低。为了解决这些问题，我们提出了StreamPPG，这是一种统一架构，能够实现低延迟的逐帧生理信号估计，同时在精度上与片段式方法相媲美。StreamPPG在一致性特权学习（CPL）策略下进行训练，该策略利用真实的rPPG信号作为特权信息，以增强模型的表征能力。大量实验表明，StreamPPG在多个数据集上实现了最先进的精度，同时在边缘设备上保持实时吞吐量。
cs.CV / 243 / 2606.23204
Unmasking LAION-5B: Age, Gender, Race, and Emotion Biases in Large-Scale Image Datasets
揭示 LAION-5B：大规模图像数据集中的年龄、性别、种族和情感偏见
Abstract
Large-scale image-text datasets, such as LAION-5B, are foundational to modern AI systems, yet their vast scale and uncurated nature raise significant concerns about demographic and stereotypical biases. This study presents a comprehensive analysis of the demographic composition and representational, stereotypical, and intersectional biases in LAION-2B-en and LAION-2B-multi, the two main components of the LAION-5B dataset. Using state-of-the-art models -- FairFace, DeepFace, and Emo-AffectNet -- we analyze faces detected in the dataset to identify biases across age, gender, race, and expressed emotion. Our findings reveal substantial overrepresentation of young adults (20--39), White individuals, and males, alongside consistent underrepresentation of minority racial groups and middle-aged or older women across both dataset components. We also observe stereotypical associations between demographic attributes and emotions, such as ``Anger'' being predominantly linked to males and ``Happiness'' to females, pointing to systemic imbalances in the data. The consistency of these patterns across two demographic models and both components of LAION-5B demonstrates that these biases are deeply embedded in one of the most widely-used training datasets. Given the scale at which LAION-5B is used to train generative models, these demographic imbalances could shape the behavior and outputs of numerous downstream AI systems.
Chinese Translation
大规模图像-文本数据集，如 LAION-5B，是现代人工智能系统的基础，但其庞大的规模和未经筛选的特性引发了关于人口统计和刻板印象偏见的重大担忧。本研究对 LAION-5B 数据集的两个主要组成部分 LAION-2B-en 和 LAION-2B-multi 的人口组成及其表现、刻板印象和交叉偏见进行了全面分析。我们使用最先进的模型——FairFace、DeepFace 和 Emo-AffectNet——分析数据集中检测到的面孔，以识别年龄、性别、种族和表达情感方面的偏见。我们的研究发现，年轻成年人（20-39 岁）、白人和男性的过度代表现象显著，同时在两个数据集组件中，少数种族群体和中年或老年女性的代表性持续不足。我们还观察到人口属性与情感之间的刻板印象关联，例如“愤怒”主要与男性相关，而“快乐”则与女性相关，这指向数据中的系统性不平衡。这些模式在两个人口统计模型和 LAION-5B 的两个组件中的一致性表明，这些偏见深深植根于最广泛使用的训练数据集中。考虑到 LAION-5B 在训练生成模型时的使用规模，这些人口统计不平衡可能会影响众多下游人工智能系统的行为和输出。
cs.CV / 244 / 2606.23206
CFPO: Counterfactual Policy Optimization for Multimodal Reasoning
CFPO：用于多模态推理的反事实策略优化
Abstract
Large Vision-Language Models (LVLMs) have demonstrated remarkable capabilities in multimodal reasoning. However, prevailing reinforcement learning (RL) paradigms lack explicit counterfactual enhancement and causal learning mechanisms. This fundamental deficiency results in severe grounding failures, manifesting as a tendency to ignore visual evidence in favor of language priors or exhibiting hallucination drift during long chain-of-thought reasoning. To address this root cause, we propose CounterFactual Policy Optimization (CFPO), a novel framework that enforces causal consistency between visual perception and textual reasoning. CFPO introduces a cross-modal counterfactual enhancement mechanism, which regularizes the policy by maximizing the discrepancy between the model's predictions and those from a counterfactual state where critical visual cues are suppressed. This approach seamlessly integrates with standard algorithms like GRPO and DAPO without requiring external reward models or additional supervision. Extensive experiments demonstrate that CFPO significantly improves reasoning fidelity, achieving consistent gains of 3.17%-6.25% over standard RL baselines and 1.32%-2.13% over the state-of-the-art perception-aware method (PAPO). Code is available at https://github.com/Raven-July/CFPO.
Chinese Translation
大型视觉-语言模型（LVLMs）在多模态推理方面展现了显著的能力。然而，现有的强化学习（RL）范式缺乏明确的反事实增强和因果学习机制。这一根本缺陷导致了严重的基础失效，表现为倾向于忽视视觉证据而偏向语言先验，或在长链推理过程中出现幻觉漂移。为了解决这一根本原因，我们提出了反事实策略优化（CounterFactual Policy Optimization，CFPO），这是一个新颖的框架，强制视觉感知与文本推理之间保持因果一致性。CFPO引入了一种跨模态反事实增强机制，通过最大化模型预测与在关键视觉线索被抑制的反事实状态下的预测之间的差异来规范化策略。这种方法与标准算法如GRPO和DAPO无缝集成，无需外部奖励模型或额外的监督。大量实验表明，CFPO显著提高了推理的准确性，相较于标准RL基线实现了3.17%-6.25%的一致性提升，相较于最先进的感知感知方法（PAPO）实现了1.32%-2.13%的提升。代码可在https://github.com/Raven-July/CFPO获取。
cs.CV / 245 / 2606.23212
Temporally Aware Densification for Dynamic 3D Gaussian Splatting
动态3D高斯点云的时间感知稠密化
Abstract
Despite modeling temporal motion, dynamic 3D Gaussian Splatting (3DGS) methods still inherit a static densification strategy that is ill-suited for dynamic scenes. This neglect of temporal behavior leads to under-reconstructed and blurry dynamic regions, as short-lived Gaussians receive sparse supervision and fail to densify effectively. We propose a Visibility-Aware Densification (VAD) framework that integrates temporal visibility into the densification process, ensuring that Gaussians are refined based on their actual temporal presence. A Temporally-Adaptive Thresholding (TAT) mechanism further adjusts each Gaussian's densification threshold according to its temporal lifespan, promoting balanced refinement of both static and dynamic regions. Finally, a Temporal Offset Warping (TOW) design enhances deformation capacity around temporal centers, extending the lifespan of highly dynamic Gaussians and facilitating more effective densification. Our approach achieves substantial improvements in the visual quality of dynamic regions, outperforming existing methods across three dynamic multi-view benchmark datasets. Moreover, the proposed VAD module generalizes across diverse dynamic 3DGS methods, consistently improving dynamic reconstruction as a plug-and-play component.
Chinese Translation
尽管动态3D高斯点云（3DGS）方法能够建模时间运动，但仍然继承了一种不适合动态场景的静态稠密化策略。这种对时间行为的忽视导致动态区域重建不足且模糊，因为短暂存在的高斯点获得的监督稀疏，无法有效稠密化。我们提出了一种可视性感知稠密化（VAD）框架，将时间可视性整合到稠密化过程中，确保高斯点根据其实际的时间存在进行精细化。一个时间自适应阈值机制（TAT）进一步根据每个高斯点的时间生命周期调整其稠密化阈值，促进静态和动态区域的平衡精细化。最后，一个时间偏移扭曲（TOW）设计增强了围绕时间中心的变形能力，延长了高度动态高斯点的生命周期，并促进了更有效的稠密化。我们的方法在动态区域的视觉质量上取得了显著改善，超越了现有方法，在三个动态多视图基准数据集上表现优异。此外，所提出的VAD模块在多种动态3DGS方法中具有良好的泛化能力，作为即插即用组件，持续改善动态重建效果。
cs.CV / 246 / 2606.23221
RS-Gen: A Multi-Stage Agentic Framework for Reasoning and Search-Augmented Image Generation
RS-Gen：一种多阶段智能框架用于推理和搜索增强的图像生成
Abstract
Recent years have witnessed remarkable progress in image generation and editing, particularly regarding instruction following and visual fidelity. However, when handling ambiguous intentions, logical reasoning, and Out-of-Distribution (OOD) knowledge, existing image models often yield sub-optimal results due to a lack of deep reasoning capabilities and real-time external information. Although emerging unified understanding-and-generation models attempt to bridge this gap, they remain constrained by their intrinsic parameter scales and static knowledge gaps. Inspired by agentic paradigms, we propose RS-Gen: a plug-and-play, training-free, multi-stage image agentic framework. RS-Gen innovatively introduces a "Questioning-and-Solving" closed-loop mechanism to accurately identify logical issues and knowledge gaps, autonomously planning actions to bridge information deficits and execute deep logical reasoning. Extensive experiments demonstrate that RS-Gen significantly expands the capability boundaries of foundational image generation and editing models. Specifically, on the WISE Verified and RISEBench benchmarks, RS-Gen yields substantial absolute performance gains of 0.313 for Qwen-Image and 19.70 for Qwen-Image-Edit-2511, respectively, successfully elevating both to the state-of-the-art (SOTA) level among open-source models.
Chinese Translation
近年来，图像生成和编辑领域取得了显著进展，尤其是在遵循指令和视觉逼真度方面。然而，在处理模糊意图、逻辑推理和分布外（Out-of-Distribution, OOD）知识时，现有的图像模型往往由于缺乏深度推理能力和实时外部信息而产生次优结果。尽管新兴的统一理解与生成模型试图弥补这一差距，但它们仍然受到固有参数规模和静态知识差距的限制。受智能范式的启发，我们提出了RS-Gen：一种即插即用、无训练的多阶段图像智能框架。RS-Gen创新性地引入了“提问与解决”的闭环机制，以准确识别逻辑问题和知识差距，自动规划行动以弥补信息缺口并执行深度逻辑推理。大量实验表明，RS-Gen显著扩展了基础图像生成和编辑模型的能力边界。具体而言，在WISE Verified和RISEBench基准测试中，RS-Gen分别为Qwen-Image和Qwen-Image-Edit-2511带来了0.313和19.70的显著绝对性能提升，成功将两者提升至开源模型中的最新水平（state-of-the-art, SOTA）。
cs.CV / 247 / 2606.23226
PhysFlow: Frequency Decoupled with Dual-Field Rectified Flow for Remote Photoplethysmography
PhysFlow：用于远程光电容积描记的频率解耦双场整流流
Abstract
Remote Photoplethysmography (rPPG) enables contactless pulse estimation from facial videos, serving as a vital tool for health monitoring. However, current deep learning methods often struggle under complex disturbances, particularly varying illumination, facial expressions, and unconstrained head movements. In such scenarios, subtle physiological signals are easily dominated by external interference, making the recovered rPPG waveform unstable and unreliable. One important reason is that most existing methods directly model the rPPG signal in a unified manner, where different signal components are coupled during reconstruction. This makes it difficult to preserve weak pulse-related variations when strong disturbance-induced changes are present. To address this challenge, we propose PhysFlow, a frequency-decoupled dual-field rectified flow framework tailored for robust rPPG estimation. Specifically, the ground-truth rPPG signal is decomposed into trend and amplitude components, which are used as separate supervisory targets. Based on the extracted facial features, PhysFlow learns two component-specific conditional velocity fields to model the two components separately. This design reduces mutual interference between different components and improves the robustness of rPPG reconstruction under complex disturbances. Moreover, the rectified flow formulation enables efficient waveform reconstruction with only a few ordinary differential equation (ODE) integration steps. Extensive experiments on multiple benchmark datasets demonstrate that PhysFlow outperforms state-of-the-art methods in both heart-rate estimation and rPPG waveform reconstruction across diverse challenging scenarios.
Chinese Translation
远程光电容积描记（rPPG）能够从面部视频中无接触地估计脉搏，是健康监测的重要工具。然而，当前的深度学习方法在复杂干扰下往往表现不佳，尤其是在光照变化、面部表情和不受限制的头部运动等情况下。在这些场景中，微弱的生理信号容易被外部干扰所主导，使得恢复的rPPG波形不稳定且不可靠。一个重要原因是大多数现有方法以统一的方式直接建模rPPG信号，在重建过程中不同信号成分相互耦合。这使得在强干扰引起的变化存在时，难以保留微弱的脉搏相关变化。为了解决这一挑战，我们提出了PhysFlow，一种针对稳健rPPG估计的频率解耦双场整流流框架。具体而言，真实的rPPG信号被分解为趋势和幅度成分，并作为独立的监督目标。基于提取的面部特征，PhysFlow学习两个成分特定的条件速度场，以分别建模这两个成分。这一设计减少了不同成分之间的相互干扰，提高了在复杂干扰下rPPG重建的稳健性。此外，整流流的公式使得仅需少量常微分方程（ODE）积分步骤即可高效重建波形。在多个基准数据集上的广泛实验表明，PhysFlow在心率估计和rPPG波形重建方面，均优于现有的最先进方法，表现出色，适应多种具有挑战性的场景。
cs.CV / 248 / 2606.23230
Privacy-Preserving Person Re-Identification from Temporal Sequences with Transformer and Hungarian Optimization
基于时间序列的隐私保护个体重识别：结合变换器和匈牙利优化
Abstract
Person re-identification (Re-ID) is a crucial task in surveillance and human behavior analysis, often used in public spaces such as transport hubs. Traditional RGB-based Re-ID methods raise privacy concerns and are highly sensitive to lighting variations and occlusion. In this paper, we propose a novel Re-ID approach that leverages depth images, which inherently obscures facial and other identifiable features, making it a privacy-preserving solution. Our method addresses the association problem between multiple views of individuals by applying the Hungarian algorithm, optimizing the matching process through minimization of the global cost across the distance matrix. We further enhance the approach by introducing temporal sequences of frames as input to a Transformer encoder architecture, which exploits both RGB and depth modalities. This architecture captures dynamic movement patterns, improving feature extraction and re-identification accuracy. Additionally, we employ batch hard triplet loss to enhance discriminative feature learning by focusing on the hardest samples. We evaluate both depth-only and RGB-D models on several top-view datasets, including TVPR2, GODPR, and BIWI RGBD-ID. Our results demonstrate that depth-only re-identification can achieve competitive performance compared to state-of-the-art methods, as measured by standard metrics such as Cumulative Matching Characteristics (CMC) and Mean Average Precision (mAP), while prioritizing privacy preservation.
Chinese Translation
个体重识别（Re-ID）是监控和人类行为分析中的一项关键任务，通常应用于交通枢纽等公共场所。传统的基于RGB的Re-ID方法引发了隐私问题，并且对光照变化和遮挡高度敏感。本文提出了一种新颖的Re-ID方法，利用深度图像，这种图像本质上模糊了面部和其他可识别特征，从而成为一种隐私保护的解决方案。我们的方法通过应用匈牙利算法解决个体多个视角之间的关联问题，通过最小化距离矩阵上的全局成本来优化匹配过程。我们进一步通过引入帧的时间序列作为输入到变换器编码器架构中来增强该方法，该架构同时利用RGB和深度模态。这种架构捕捉动态运动模式，提高了特征提取和重识别的准确性。此外，我们采用批量困难三元组损失来增强区分特征学习，重点关注最困难的样本。我们在多个顶视数据集上评估了仅深度和RGB-D模型，包括TVPR2、GODPR和BIWI RGBD-ID。我们的结果表明，仅深度的重识别在标准指标（如累积匹配特性（CMC）和平均精度均值（mAP））的测量下，可以实现与最先进方法相媲美的性能，同时优先考虑隐私保护。
cs.CV / 249 / 2606.23254
SteerVTE: Seamless Video Text Editing with Style and Glyph Control
SteerVTE：无缝视频文本编辑的风格与字形控制
Abstract
Visual text editing aims to precisely modify text in images and videos while preserving stylistic consistency and visual realism. Despite significant advances in the image domain, video text editing remains largely unexplored: it is a localized task demanding stroke-level precision within small text regions, which compounds the challenges of cross-frame accuracy, temporal coherence, and stylistic fidelity. We introduce SteerVTE, a unified framework that \underline{\textbf{steer}}s a frozen video diffusion model to perform precise \underline{\textbf{V}}ideo \underline{\textbf{T}}ext \underline{\textbf{E}}diting through style and glyph control. Built on a frozen diffusion transformer, SteerVTE attaches a lightweight text context adapter with two complementary modules: a style encoder capturing the original text's visual attributes, and dual-granularity glyph encoders encoding the target text at both the line and character levels. To overcome the inherently weak text rendering priors of video foundation models, we further propose a glyph-aware spatial-focal loss and a three-stage progressive training curriculum that scales from image to video data. To support large-scale training, we also develop an automatic synthesis pipeline and construct SteerVTE-1M, a dataset of one million triplets spanning diverse scenes, fonts, and stylistic effects. Extensive experiments demonstrate that SteerVTE substantially outperforms existing video editing baselines across text accuracy, style consistency, and temporal coherence.
Chinese Translation
视觉文本编辑旨在精确修改图像和视频中的文本，同时保持风格一致性和视觉真实感。尽管在图像领域取得了显著进展，视频文本编辑仍然在很大程度上未被探索：这是一项局部任务，需要在小文本区域内实现笔画级别的精确度，这加剧了跨帧准确性、时间一致性和风格保真度的挑战。我们提出了SteerVTE，一个统一框架，通过风格和字形控制来引导一个冻结的视频扩散模型，以执行精确的视频文本编辑。SteerVTE基于一个冻结的扩散变换器，附加了一个轻量级文本上下文适配器，包含两个互补模块：一个风格编码器用于捕捉原始文本的视觉属性，以及双粒度字形编码器在行和字符级别对目标文本进行编码。为了克服视频基础模型固有的弱文本渲染先验，我们进一步提出了一种字形感知的空间聚焦损失和一个从图像到视频数据的三阶段渐进训练课程。为了支持大规模训练，我们还开发了一个自动合成管道，并构建了SteerVTE-1M，一个包含一百万个三元组的数据集，涵盖多样的场景、字体和风格效果。大量实验表明，SteerVTE在文本准确性、风格一致性和时间一致性方面显著优于现有的视频编辑基线。
cs.CV / 250 / 2606.23256
P-JEPA: Procedural Video Representation Learning via Joint Embedding Predictive Architecture
P-JEPA：通过联合嵌入预测架构进行程序性视频表示学习
Abstract
The increasing maturity of embodied AI platforms has driven a growing interest in procedural video representation learning to support intelligent assistance systems for complex, multi-step tasks. Leveraging large-scale latent predictive training, video foundation models capture video dynamics, enabling downstream tasks such as activity understanding, spatiotemporal localization, and predictive control. However, procedural videos include actions with long-range dependencies that these models do not support, due to the quadratic complexity of self-attention. Distinct actions, for example, may be visually similar despite appearing at different points in the procedure, such as turning the stove on versus off. Here, we propose a backbone-agnostic approach that learns long-duration video representations by reducing the problem to a dense, frame-aligned action space and predicting pooled masked latent vectors. This approach allows our Procedural Joint Embedding Predictive Architecture (P-JEPA) to ingest videos over 30 minutes long, enabling effective long-form understanding of procedural steps. We evaluate P-JEPA using features extracted with VJEPA2.1, TSM, and I3D over the EgoExo4D, EgoProceL, and Assembly101 datasets, finding that it consistently improves linear separability, streaming inference, and temporal action segmentation performance, achieving state-of-the-art results on EgoExo4D fine-grained action classification while using an order of magnitude fewer parameters than LLM-based methods and running in real time.
Chinese Translation
随着具身人工智能平台的日益成熟，程序性视频表示学习引起了越来越多的关注，以支持复杂多步骤任务的智能辅助系统。利用大规模潜在预测训练，视频基础模型能够捕捉视频动态，从而支持活动理解、时空定位和预测控制等下游任务。然而，程序性视频包含具有长程依赖的动作，而这些模型由于自注意力的平方复杂度而无法支持。不同的动作，例如，尽管在程序的不同点出现，但在视觉上可能相似，如打开与关闭炉子。在此，我们提出了一种与骨干网络无关的方法，通过将问题简化为一个密集的、帧对齐的动作空间，并预测池化的掩蔽潜在向量，从而学习长时长视频表示。这种方法使我们的程序性联合嵌入预测架构（P-JEPA）能够处理超过30分钟的视频，有效实现程序步骤的长篇理解。我们使用从VJEPA2.1、TSM和I3D提取的特征在EgoExo4D、EgoProceL和Assembly101数据集上评估P-JEPA，发现它在提高线性可分性、流式推理和时间动作分割性能方面始终表现出色，在EgoExo4D细粒度动作分类上取得了最先进的结果，同时使用的参数量比基于LLM的方法少一个数量级，并且能够实时运行。
cs.CV / 251 / 2606.23267
Safe Few-Step Generation via Velocity Editing
通过速度编辑实现安全的少步生成
Abstract
Flow matching has recently emerged as a strong paradigm for state-of-the-art text-to-image (T2I) generation, enabling high-quality generation with a small number of sampling steps. As these models are increasingly integrated into real-world applications, ensuring safe and non-sensitive content generation has become a critical requirement. However, adapting safety and concept removal methods to this new generation framework remains an open challenge. Specifically, prior methods largely rely on iterative trajectory steering across a number of denoising steps or on CLIP-centric prompt embedding manipulation. These design assumptions pose fundamental bottlenecks for safety in flow matching-based T2I generation, where limited sampling steps constrain iterative correction and modern context-aware text encoders diminish the effectiveness of embedding-level interventions. In this paper, we propose VESFlow, a training-free safety method tailored to flow matching with extremely few sampling steps. Leveraging the fact that flow matching models learn the marginal velocity, we directly edit the velocity field via a safe-conditional posterior. VESFlow steers the trajectory toward safe outputs while leaving the conditioning prompt unchanged. Building on the observation that VESFlow leaves outputs unchanged under benign prompts, we further introduce a risk score-based filtering that bypasses velocity editing to reduce computational cost while preserving benign prompt generation. Based on this filtering, we propose VESFlow+, a stronger variant of VESFlow that not only edits the velocity toward the safe direction, but also pushes it away from the unsafe direction. Experimental results show that VESFlow+ removes the target concept, reducing the attack success rate by NudeNet to 6.3% on Ring-A-Bell and 6.8% on MMA-Diffusion on the 4-step MeanFlow model, while preserving fidelity on benign prompts.
Chinese Translation
流匹配最近成为了最先进的文本到图像（T2I）生成的强大范式，使得在少量采样步骤下实现高质量生成成为可能。随着这些模型越来越多地被应用于现实世界，确保生成安全和非敏感内容已成为一项关键要求。然而，将安全性和概念移除方法适应这一新的生成框架仍然是一个未解决的挑战。具体而言，先前的方法主要依赖于在多个去噪步骤中进行迭代轨迹引导，或依赖于以CLIP为中心的提示嵌入操作。这些设计假设对基于流匹配的T2I生成的安全性构成了根本性瓶颈，因为有限的采样步骤限制了迭代修正，而现代上下文感知文本编码器则降低了嵌入级干预的有效性。在本文中，我们提出了VESFlow，这是一种针对流匹配的无训练安全方法，适用于极少的采样步骤。利用流匹配模型学习边际速度的事实，我们通过安全条件后验直接编辑速度场。VESFlow引导轨迹朝向安全输出，同时保持条件提示不变。基于VESFlow在良性提示下输出保持不变的观察，我们进一步引入了一种基于风险评分的过滤方法，绕过速度编辑以降低计算成本，同时保留良性提示生成。在此过滤的基础上，我们提出了VESFlow+，这是VESFlow的一个更强变体，不仅将速度编辑朝向安全方向，还将其推离不安全方向。实验结果表明，VESFlow+去除了目标概念，将NudeNet的攻击成功率降低到Ring-A-Bell上的6.3%和MMA-Diffusion上的6.8%，同时在良性提示下保持了保真度。
cs.CV / 252 / 2606.23270
BoxCtrl: 3D-Aware Visual Prompting for Geometric Image Editing
BoxCtrl：用于几何图像编辑的3D感知视觉提示
Abstract
As instruction-based editing models and multimodal large language models advance, diverse image editing tasks have become feasible. However, achieving precise and consistent geometric image editing, such as translating, scaling, and rotating in 3D space, remains a major challenge. In this work, we introduce BoxCtrl, a 3D-aware visual prompting framework. Unlike text-only or coarse 2D-guided approaches, our method introduces informative RGB 3D bounding boxes projected onto 2D images as visual prompts. The three orthogonal faces of each box are painted with distinct RGB colors, simultaneously encoding position, size, and orientation to provide a compact, intuitive in-context visual example. The key to BoxCtrl's success lies in these well-designed bounding boxes, which decouple geometric control from appearance control. This enables the model to learn consistent correspondences between faces of the same color in the latent space, leading to a precise understanding of geometric intentions and accurate editing results. We introduce a two-stage training paradigm: Supervised Fine-Tuning (SFT) followed by Reinforcement Learning (RL). To address paired data scarcity, we construct a large-scale synthetic dataset for SFT, equipping the model with fundamental editing capabilities. To bridge the synthetic-to-real domain gap, we incorporate an online RL stage leveraging unpaired real-world data. Guided by a reward function evaluating geometric accuracy and visual fidelity, our SFT-RL strategy significantly enhances geometric precision while maintaining photorealistic quality. Extensive experiments demonstrate that BoxCtrl achieves state-of-the-art performance across translation, rotation, scaling, and composite editing tasks.
Chinese Translation
随着基于指令的编辑模型和多模态大型语言模型的发展，各种图像编辑任务变得可行。然而，实现精确且一致的几何图像编辑，如在3D空间中的平移、缩放和旋转，仍然是一个主要挑战。在本研究中，我们介绍了BoxCtrl，一种3D感知的视觉提示框架。与仅依赖文本或粗略的2D引导方法不同，我们的方法引入了投影到2D图像上的信息丰富的RGB 3D边界框作为视觉提示。每个边界框的三个正交面涂上不同的RGB颜色，同时编码位置、大小和方向，以提供一个紧凑、直观的上下文视觉示例。BoxCtrl成功的关键在于这些精心设计的边界框，它们将几何控制与外观控制解耦。这使得模型能够在潜在空间中学习相同颜色面之间的一致对应关系，从而精确理解几何意图并获得准确的编辑结果。我们引入了一个两阶段的训练范式：监督微调（Supervised Fine-Tuning, SFT）后接强化学习（Reinforcement Learning, RL）。为了应对配对数据稀缺的问题，我们构建了一个大规模的合成数据集用于SFT，为模型提供基本的编辑能力。为了弥合合成与真实领域之间的差距，我们引入了一个在线RL阶段，利用未配对的真实世界数据。在评估几何准确性和视觉保真度的奖励函数指导下，我们的SFT-RL策略显著提高了几何精度，同时保持了照片级真实感。大量实验表明，BoxCtrl在平移、旋转、缩放和复合编辑任务中实现了最先进的性能。
cs.CV / 253 / 2606.23286
Transfer learning-based method for automated ewaste recycling in smart cities
基于迁移学习的智能城市自动化电子废物回收方法
Abstract
Sorting a huge stream of waste accurately within a short period can be done with the support of digitalization, particularly Artificial Intelligence, instead of traditional methods. The overlap of Artificial Intelligence and Circular Economy can flourish many services in the environmental technology domain, in particular smart ewaste recycling, resulting in enabling circular smart cities. We analyse the growing need for automated ewaste recycling as an essential requirement to cope with the fast growing ewaste stream and we shed the light on the impact of Artificial Intelligence in supporting the recycling process through smart classification of devices, where the smartphone is our case study. Our study applies transfer learning as a special technique of Artificial Intelligence by finetuning the output layers of AlexNet as a pretrained model and perform the implementation on a small size dataset that contains 12 classes from 6 smartphone brands. We evaluate the performance of our model by tuning the learning rate, choosing the best optimizer, and augmenting the original dataset to avoid overfitting. We found that the optimizer of Stochastic Gradient Descent with Momentum and 3e-4 as a learning rate brings almost 98% model accuracy with generalization. Our study supports automated ewaste recycling in decreasing the error rate of ewaste sorting and investigates the advantages of applying transfer learning as the best scenario to overcome the rising challenges.
Chinese Translation
在短时间内准确地对大量废物进行分类，可以借助数字化，特别是人工智能，而不是依赖传统方法。人工智能与循环经济的结合可以促进环境技术领域的多项服务，尤其是智能电子废物回收，从而实现循环智能城市。我们分析了自动化电子废物回收的日益增长的需求，认为这是应对快速增长的电子废物流的基本要求，并阐明了人工智能在通过智能设备分类支持回收过程中的影响，以智能手机为案例研究。我们的研究应用了迁移学习作为一种特殊的人工智能技术，通过微调预训练模型AlexNet的输出层，并在包含6个智能手机品牌12个类别的小型数据集上进行实施。我们通过调整学习率、选择最佳优化器以及增强原始数据集以避免过拟合来评估模型的性能。我们发现，使用动量随机梯度下降优化器和3e-4的学习率几乎可以实现98%的模型准确率和良好的泛化能力。我们的研究支持自动化电子废物回收，降低电子废物分类的错误率，并探讨了应用迁移学习作为应对日益严峻挑战的最佳方案的优势。
cs.CV / 254 / 2606.23293
Flow6D: Discrete-to-Continuous Flow Matching for Efficient and Accurate Category-Level 6D Pose Estimation
Flow6D：用于高效准确的类别级6D姿态估计的离散到连续流匹配
Abstract
6D pose estimation is a key task in computer vision and embodied AI, widely used in robotic manipulation, augmented reality, etc. Existing methods directly regress in a high-dimensional continuous space, facing two key challenges in category-level pose estimation: limited accuracy due to noise and local optima, and inefficient search over an infinite space that hinders real-time performance. This paper proposes Flow6D, a hierarchical flow matching framework with a two-stage discrete latent space localization-continuous pose regression strategy. Rotation and translation parameters are first discretized into bins, with a discrete flow matching model locking the latent space around the true pose to reduce search complexity. Then, by sampling in the latent space, a continuous flow matching model predicts local pose residuals to optimize the estimate and regress to an accurate pose. The framework also naturally extends to articulated objects, outperforming state-of-the-art methods on synthetic and real datasets with real-time inference at 70 FPS. Project website: https://flow6d.github.io/.
Chinese Translation
6D姿态估计是计算机视觉和具身人工智能中的一项关键任务，广泛应用于机器人操作、增强现实等领域。现有方法直接在高维连续空间中进行回归，面临类别级姿态估计中的两个主要挑战：由于噪声和局部最优导致的准确性有限，以及在无限空间中进行高效搜索的困难，这妨碍了实时性能。本文提出了Flow6D，一种层次化流匹配框架，采用两阶段离散潜在空间定位-连续姿态回归策略。首先，将旋转和位移参数离散化为多个区间，离散流匹配模型将潜在空间锁定在真实姿态附近，以降低搜索复杂度。然后，通过在潜在空间中进行采样，连续流匹配模型预测局部姿态残差，以优化估计并回归到准确的姿态。该框架还自然扩展到关节物体，在合成和真实数据集上超越了最先进的方法，并以70 FPS的实时推理性能表现。项目网站：https://flow6d.github.io/
cs.CV / 255 / 2606.23298
Ocean4D: Generative Underwater 4D Reconstruction via Medium-Aware Video Diffusion
Ocean4D：基于介质感知的视频扩散生成水下4D重建
Abstract
Underwater 4D reconstruction remains challenging due to the coupling between degraded light transport in participating media and dynamic water variations. Most existing Methods are developed under in-air assumptions and do not explicitly account for underwater absorption and backscatter. Additionally, near-static assumptions make these approaches sensitive to drifting particles and dynamic distractors , leading to unstable geometry and inconsistent cross-view results. To address these issues, we propose a generative framework for underwater 4D reconstruction, named Ocean4D, which is built on two complementary components. Specifically, 4D-GCC constructs 4D geometrically consistent conditioning with improved cross-frame coverage, while the Medium-Aware Block performs implicit medium-aware denoising in the latent diffusion process to stabilize underwater appearance under absorption and scattering. Given a monocular video and target cameras, our method generates videos along the target trajectories while preserving global structure and cross-view consistency. Extensive experiments on both dynamic and static underwater benchmarks demonstrate state-of-the-art performance on underwater reconstruction.
Chinese Translation
水下4D重建仍然面临挑战，因为参与介质中的光传输退化与动态水体变化之间存在耦合。大多数现有方法是在空气中假设下开发的，并未明确考虑水下的吸收和反向散射。此外，近静态假设使这些方法对漂浮颗粒和动态干扰物敏感，导致几何形状不稳定和跨视图结果不一致。为了解决这些问题，我们提出了一种水下4D重建的生成框架，命名为Ocean4D，该框架由两个互补组件构成。具体而言，4D-GCC构建了具有改进跨帧覆盖的4D几何一致性条件，而介质感知模块在潜在扩散过程中执行隐式的介质感知去噪，以在吸收和散射下稳定水下外观。给定单目视频和目标相机，我们的方法在保留全局结构和跨视图一致性的同时，沿目标轨迹生成视频。在动态和静态水下基准测试上的广泛实验表明，我们的方法在水下重建方面达到了最先进的性能。
cs.CV / 256 / 2606.23327
VideoAgent: All-in-One Framework for Video Understanding and Editing
VideoAgent：视频理解与编辑的全能框架
Abstract
Video editing has become essential in digital media creation, yet existing automated systems are restricted to short segment processing and domain-specific tasks. They face two critical limitations: i) inability to handle diverse video comprehension and editing operations, and ii) lack of long-video understanding for coherent narrative creation. We propose VideoAgent, an all-in-one agentic framework addressing these challenges through two key innovations. First, we develop automated video shot creation with shot planning agents for coherent narratives and cross-modal retrieval for aligned visual content. Second, we design a multi-agent orchestration framework integrating over thirty specialized editing agents. Intent parsing filters relevant tools while textual-gradient graph optimization assembles complex editing pipelines. Extensive experiments on our newly-proposed VideoEdit benchmark and public datasets demonstrate VideoAgent's superiority over existing multimodal LLMs and agentic systems. VideoAgent achieves 87-95% orchestration success rates while reducing API costs by 60%. Human evaluation across six video categories shows VideoAgent produces professional-quality content approaching human-level performance, with ratings only 4% below human-created videos. We release our code at https://github.com/HKUDS/VideoAgent.
Chinese Translation
视频编辑在数字媒体创作中变得至关重要，但现有的自动化系统仅限于短片段处理和特定领域任务。它们面临两个关键限制：一是无法处理多样化的视频理解和编辑操作，二是缺乏对长视频的理解以创建连贯的叙事。我们提出了VideoAgent，一个全能的智能框架，通过两个关键创新来应对这些挑战。首先，我们开发了自动化的视频镜头创建，利用镜头规划代理实现连贯的叙事，并通过跨模态检索获取对齐的视觉内容。其次，我们设计了一个多代理编排框架，集成了三十多个专业编辑代理。意图解析过滤相关工具，而文本梯度图优化则组装复杂的编辑管道。在我们新提出的VideoEdit基准和公共数据集上进行的广泛实验表明，VideoAgent在现有的多模态大语言模型和智能系统中表现优越。VideoAgent的编排成功率达到87-95%，同时将API成本降低了60%。在六个视频类别中的人工评估显示，VideoAgent生成的内容接近人类水平，专业质量的评分仅比人类创作的视频低4%。我们将在https://github.com/HKUDS/VideoAgent发布我们的代码。
cs.CV / 257 / 2606.23344
RT-DocLayout: Real-Time End-to-End Document Layout Analysis with Reading Order in the Wild
RT-DocLayout：具有阅读顺序的实时端到端文档布局分析
Abstract
Accurate document layout analysis remains a critical bottleneck for document parsing systems, due to the intricate coupling among heterogeneous document layout elements, geometric distortions (\eg, paper warping and bending, perspective variations), and reading order within diverse layout structures. Existing approaches typically rely on fragmented multi-stage pipelines or computationally heavy generative Transformer architectures, leading to error propagation and limited efficiency. In this paper, we present RT-DocLayout, a highly efficient end-to-end framework for document layout analysis, designed as a front-end for document parsing tasks. The proposed model unifies classification, detection, pixel-level segmentation, and reading order prediction for layout elements within a single 33M-parameter architecture. Built upon the RT-DETR, our key contribution is a unified multi-task formulation within a single query-based decoder that simultaneously classifies, regresses bounding box, generates masks, and constructs relationship to reason reading order. By jointly learning geometric and structural representations, RT-DocLayout introduces multi-task optimization that substantially improves robustness under real-world document distortions. Extensive experiments on public benchmarks demonstrate state-of-the-art performance in document layout analysis while maintaining real-time inference speed(132.1 FPS). When coupled with downstream OCR engines, RT-DocLayout significantly improves full-document reconstruction quality, providing a scalable and practical foundation for real-world document intelligence systems.
Chinese Translation
准确的文档布局分析仍然是文档解析系统的一个关键瓶颈，这主要是由于异构文档布局元素之间复杂的耦合、几何失真（例如，纸张翘曲和弯曲、透视变化）以及在多样化布局结构中的阅读顺序。现有的方法通常依赖于碎片化的多阶段管道或计算负担重的生成式Transformer架构，导致错误传播和效率有限。本文提出了RT-DocLayout，一个高效的端到端文档布局分析框架，旨在作为文档解析任务的前端。所提出的模型在一个33M参数的架构中统一了布局元素的分类、检测、像素级分割和阅读顺序预测。基于RT-DETR，我们的关键贡献是一个统一的多任务公式，在一个基于查询的解码器中同时进行分类、回归边界框、生成掩码，并构建关系以推理阅读顺序。通过共同学习几何和结构表示，RT-DocLayout引入了多任务优化，显著提高了在真实世界文档失真下的鲁棒性。在公共基准上的大量实验表明，RT-DocLayout在文档布局分析中实现了最先进的性能，同时保持实时推理速度（132.1 FPS）。当与下游OCR引擎结合时，RT-DocLayout显著提高了全文档重建质量，为现实世界文档智能系统提供了可扩展和实用的基础。
cs.CV / 258 / 2606.23354
Faithful Grounded Visual Reasoning via Learned Proxy-Tokens
通过学习的代理标记实现忠实的基础视觉推理
Abstract
Multimodal Large Language Models (MLLMs) have achieved remarkable success in Visual Question Answering (VQA), yet their "black-box" nature hinders deployment in critical domains. Grounded Visual Reasoning (GVR) approaches attempt to improve interpretability by explicitly couple textual rationales with visual grounding information, which are typically textual coordinates. This mechanism lacks a learnable semantic link to the visual features, often resulting in a semantic-spatial gap where the model hallucinates coordinates that do not correspond to image evidences. In this work, we introduce Composer, a MLLM that leverages a novel visual grounding mechanism based on learned proxy-tokens to promote faithful interpretability. These discrete symbolic pointers explicitly index the image latent space, allowing the model to manipulate visual regions as addressable, semantically manipulable sets. To rigorously validate our novel grounding mechanism, we constructed ComposerGCoT, a dataset synthesized to enable holistic assessment of reasoning consistency and grounding accuracy. Experimental results indicate that Composer achieves performance parity with its coordinate-based counterpart in final answer accuracy, while improving visual grounding accuracy by +9.0 points. By demonstrating that discrete proxy-tokens capture spatial semantics more effectively than typical textual coordinates, we establish that visual grounding mechanisms with learnable semantic links represent a promising path toward trustworthy and reliable MLLMs.
Chinese Translation
多模态大型语言模型（MLLMs）在视觉问答（VQA）中取得了显著成功，但其“黑箱”特性限制了在关键领域的应用。基础视觉推理（GVR）方法试图通过将文本推理与视觉基础信息（通常是文本坐标）明确结合来提高可解释性。然而，这种机制缺乏与视觉特征的可学习语义链接，常常导致语义-空间差距，使得模型产生与图像证据不对应的坐标幻觉。在本研究中，我们引入了Composer，一个利用基于学习的代理标记的新型视觉基础机制的MLLM，以促进忠实的可解释性。这些离散的符号指针明确地索引图像潜在空间，使模型能够将视觉区域作为可寻址、语义可操作的集合进行操作。为了严格验证我们新颖的基础机制，我们构建了ComposerGCoT，这是一个合成数据集，旨在实现推理一致性和基础准确性的整体评估。实验结果表明，Composer在最终答案准确性方面与其基于坐标的对应模型达成了性能平衡，同时视觉基础准确性提高了9.0个百分点。通过证明离散代理标记比典型文本坐标更有效地捕捉空间语义，我们确立了具有可学习语义链接的视觉基础机制是朝向可信赖和可靠的MLLMs的有前景路径。
cs.CV / 259 / 2606.23356
Changing Modalities: Adapting Remote Sensing Models to New Satellites and Sensors
变化的模态：将遥感模型适应于新卫星和传感器
Abstract
Machine learning models for remote sensing are trained and deployed on a static set of modalities. However, as we equip newer satellites with novel sensors and retire old ones, practitioners may wish to deploy an existing model on a substitution, superset, or subset of modalities with minimal retraining given data availability or practical computational constraints. We study the setting of updating existing models to changing modalities and identify three main scenarios: Modality Transfer (substitution), Addition (superset), and Peeking (subset). We propose DeluluNet, an architecture with modular components for all three changing modality scenarios. DeluluNet is trained end-to-end, learning a multi-modal model from a unimodal teacher and unlabeled multimodal data via modality hallucination--predicting missing modality representations from those that are present. As a result, DeluluNet can keep predicting even when input modalities change, providing a practical alternative to re-labeling and re-training in a changing world.
Chinese Translation
遥感的机器学习模型是在一组静态模态上进行训练和部署的。然而，随着我们为新卫星配备新型传感器并退役旧设备，实践者可能希望在可用数据或实际计算限制的情况下，最小化重新训练的情况下，将现有模型部署到替代、超集或子集模态上。我们研究了更新现有模型以适应变化模态的情境，并确定了三种主要场景：模态转移（替代）、添加（超集）和窥视（子集）。我们提出了DeluluNet，这是一种具有模块化组件的架构，适用于所有三种变化模态场景。DeluluNet通过模态幻觉进行端到端训练，从单模态教师和未标记的多模态数据中学习多模态模型——即从存在的模态中预测缺失的模态表示。因此，DeluluNet能够在输入模态变化时继续进行预测，为在变化的世界中重新标记和重新训练提供了一种实用的替代方案。
cs.CV / 260 / 2606.23373
Polynomial Dice Loss for Medical Image Segmentation
用于医学图像分割的多项式骰子损失
Abstract
Medical image segmentation is a fundamental task for medical image processing and computer-assisted intervention, yet data imbalance and small lesion detection pose significant challenges. Dice Loss, which measures the overlap between predicted and ground truth regions, is widely used to mitigate these issues. To further emphasize its properties, we propose Polynomial Dice Loss, a polynomial extension of Dice Loss. Specifically, by leveraging the geometric characteristics of Dice Loss and formulating the loss function as a polynomial representation via Taylor expansion, we enable the adjustment of the contribution of higher-order components to the loss function. In our experiments, we evaluate the proposed method against loss functions derived from conventional Dice and Tversky coefficients. Experimental results and further analysis show that the polynomial formulation provides a simple way to control the loss shape and achieves competitive performance across multiple segmentation settings.
Chinese Translation
医学图像分割是医学图像处理和计算机辅助干预的基础任务，但数据不平衡和小病灶检测带来了显著挑战。骰子损失（Dice Loss）通过测量预测区域与真实区域之间的重叠，广泛用于缓解这些问题。为了进一步强调其特性，我们提出了多项式骰子损失（Polynomial Dice Loss），这是骰子损失的多项式扩展。具体而言，通过利用骰子损失的几何特征，并通过泰勒展开将损失函数表述为多项式形式，我们能够调整高阶分量对损失函数的贡献。在我们的实验中，我们将所提出的方法与源自传统骰子系数（Dice）和特弗斯基系数（Tversky）的损失函数进行了比较。实验结果及进一步分析表明，多项式表述提供了一种简单的方法来控制损失形状，并在多个分割设置中实现了具有竞争力的性能。
cs.CV / 261 / 2606.23436
Rethinking Object-Centric Representations for Video Dynamics Modeling
重新思考视频动态建模中的对象中心表示
Abstract
Unsupervised video object tracking aims to decompose dynamic scenes into persistent, object-centric entities without manual annotations. Many recent approaches rely on slot-based representations, where a fixed set of latent variables ("slots") represent individual objects across frames. To preserve object identity, these models enforce temporal consistency on slot embeddings. However, when appearance and pose are entangled, this consistency objective conflicts with object motion and viewpoint changes. As a result, slots tend to lock onto static regions (e.g., background) to satisfy the consistency objective, while foreground objects become fragmented across multiple slots or frequently swap identities. To address these limitations, we propose STAITUS, a unified framework that explicitly disentangles each slot into appearance and geometric pose (position/scale). Leveraging this disentanglement, STAITUS enforces within-frame spatial separation and applies temporal alignment only in appearance space, yielding sharper masks and more persistent identities under motion, occlusion, and object entry/exit. Furthermore, to mitigate over-segmentation, we introduce an adaptive gating mechanism that dynamically adjusts the number of active slots to match scene complexity. Extensive experiments on synthetic and real-world benchmarks demonstrate that STAITUS substantially outperforms state-of-the-art baselines in segmentation quality and tracking stability.
Chinese Translation
无监督视频对象跟踪旨在将动态场景分解为持久的、以对象为中心的实体，而无需手动标注。许多近期的方法依赖于基于槽的表示，其中一组固定的潜变量（“槽”）在帧之间表示单个对象。为了保持对象身份，这些模型对槽嵌入施加时间一致性。然而，当外观和姿态交织在一起时，这种一致性目标与对象运动和视角变化相冲突。因此，槽往往锁定在静态区域（例如背景）以满足一致性目标，而前景对象则在多个槽之间变得支离破碎或频繁交换身份。为了解决这些局限性，我们提出了STAITUS，一个统一框架，明确地将每个槽解耦为外观和几何姿态（位置/尺度）。利用这种解耦，STAITUS在帧内强制空间分离，并仅在外观空间应用时间对齐，从而在运动、遮挡和对象进出时产生更清晰的掩膜和更持久的身份。此外，为了减轻过度分割，我们引入了一种自适应门控机制，动态调整活动槽的数量以匹配场景复杂性。在合成和真实世界基准上的大量实验表明，STAITUS在分割质量和跟踪稳定性方面显著优于最先进的基线。
cs.CV / 262 / 2606.23455
MeGAS: Thermomechanical Dynamic Gaussian Splatting for Thermophysical Scene Editing
MeGAS：用于热物理场景编辑的热机械动态高斯点云技术
Abstract
Recent advances integrate physically grounded Newtonian dynamics with neural rendering frameworks, narrowing the gap between photorealistic scene reconstruction and physics-based animation. However, existing approaches focus on mechanically driven dynamics while neglecting temperature, a fundamental yet invisible physical factor underlying phenomena such as melting, solidification, and other thermomechanical processes. In this paper, we propose MeGAS, a novel framework that incorporates thermomechanical phase-change dynamics into 3D Gaussian Splatting (3DGS). Specifically, we propose a new thermomechanical dynamic Gaussian Splatting representation that augments 3DGS with temperature attributes and employs a heat advection-diffusion solver with MPM dynamics incorporating phase transitions, enabling physically plausible and visually realistic synthesis of thermophysical phenomena. Furthermore, a new topology-adaptive Gaussian rendering strategy is proposed to mitigate cracking and floaters under extreme deformation. Extensive experiments demonstrate that MeGAS produces physically consistent thermomechanical behavior while maintaining high-fidelity photorealistic rendering, advancing toward physics-integrated world models.
Chinese Translation
近期的进展将基于物理的牛顿动力学与神经渲染框架相结合，缩小了照片真实场景重建与基于物理的动画之间的差距。然而，现有方法主要关注机械驱动的动力学，而忽视了温度这一基本但不可见的物理因素，温度在熔化、固化及其他热机械过程等现象中起着重要作用。本文提出了MeGAS，一种新颖的框架，将热机械相变动力学纳入3D高斯点云（3DGS）中。具体而言，我们提出了一种新的热机械动态高斯点云表示，增强了3DGS的温度属性，并采用了结合相变的MPM动力学的热对流-扩散求解器，从而实现了热物理现象的物理合理性和视觉真实感的合成。此外，我们还提出了一种新的拓扑自适应高斯渲染策略，以减轻极端变形下的裂纹和漂浮物。大量实验表明，MeGAS在保持高保真照片真实渲染的同时，能够产生物理一致的热机械行为，推动了物理集成世界模型的发展。
cs.CV / 263 / 2606.23473
C^2GR: Coupled Comprehensive Generative Replay for a Continually Learnable Universal Segmentation Model
C^2GR：用于可持续学习的通用分割模型的耦合综合生成重放
Abstract
Universal segmentation models exhibit significant potential for diverse tasks involving different imaging modalities and segmentation objectives. Task-Incremental Learning provides a privacy-preserving approach to continually evolve a universal model on tasks from sequentially-arriving medical departments. However, training the model solely on the incoming task induces forgetting on past tasks, since consecutive tasks exhibit concurrent shifts in image appearance and segmentation objective. To address this problem, we propose a novel Coupled Comprehensive Generative Replay (C^2GR) framework that simultaneously synthesizes image-mask pairs of previous tasks to mitigate forgetting under concurrent appearance and objective shifts. This requires preserving image-mask correspondence for structure-realistic generation and bridging asynchronous optimization of the generator and segmentor for segmentation-oriented generation. Specifically, we propose a Bayesian Joint Diffusion (BJD) method that formulates the correspondence as conditional distributions optimized via conditional denoising. Furthermore, we develop a Relation-aware Unified Prompt Synchronization (RUPS) scheme to simultaneously modulate the generator and segmentor via a shared task-relation-aware prompt for synchronizing their optimization. Experiments on 20 tasks spanning diverse modalities and objectives demonstrate that C^2GR exhibits only a 2.44% drop in overall performance compared to joint training with all task data, effectively alleviating forgetting from the concurrent shifts. Our code will be made publicly available at https://github.com/mar-cry/C2GR.
Chinese Translation
通用分割模型在涉及不同成像模式和分割目标的多样任务中展现出显著潜力。任务增量学习提供了一种保护隐私的方法，以便在来自连续到达的医疗部门的任务上不断演化通用模型。然而，仅在新到的任务上训练模型会导致对过去任务的遗忘，因为连续任务在图像外观和分割目标上表现出并发的变化。为了解决这一问题，我们提出了一种新颖的耦合综合生成重放（C^2GR）框架，该框架同时合成先前任务的图像-掩码对，以减轻在并发外观和目标变化下的遗忘。这需要保持图像-掩码对应关系，以实现结构真实的生成，并为分割导向生成架起生成器和分割器的异步优化桥梁。具体而言，我们提出了一种贝叶斯联合扩散（BJD）方法，将对应关系表述为通过条件去噪优化的条件分布。此外，我们开发了一种关系感知统一提示同步（RUPS）方案，通过共享的任务关系感知提示同时调节生成器和分割器，以同步它们的优化。在涵盖多种模式和目标的20个任务上的实验表明，C^2GR的整体性能仅下降2.44%，与使用所有任务数据的联合训练相比，有效缓解了来自并发变化的遗忘。我们的代码将公开发布在 https://github.com/mar-cry/C2GR。
cs.CV / 264 / 2606.23486
From Reconstruction to Decision: A Post-Encoder Plug-in Adapter for Curvilinear Segmentation
从重建到决策：一种后编码器插入适配器用于曲线分割
Abstract
Curvilinear object segmentation, including vessels and cracks, is challenging due to extreme spatial sparsity and topological fragility, where small local errors can cause severe structural disconnections. Meanwhile, modern segmentation pipelines increasingly rely on strong but hard-to-modify foundation encoders whose heavy downsampling limits fine structural recovery. Motivated by this, we focus on the post-encoder stage and study two recurring and actionable failure modes: a reconstruction bottleneck in high-resolution feature restoration and a decision bottleneck in binarization. We present PEPA, a lightweight Post-Encoder Plug-in Adapter for 2D curvilinear segmentation pipelines with accessible decoder/head features and target, query, or class descriptors. PEPA couples (i) Target-Conditioned Snake Upsampling (TCSU), which uses target-conditioned continuous snake-like sampling to better recover thin and tortuous structures during upsampling, and (ii) Target-Adaptive Differentiable Thresholding (TADT), which predicts target-specific thresholds and optimizes a soft-threshold surrogate with explicit safeguards against trivial bias shifting. Under this post-encoder interface, PEPA can be attached to both prompt-based decoders and conventional dense predictors. Experiments on five medical and industrial benchmarks show that adding PEPA to frozen-encoder baselines yields consistent improvements, with gains in topological connectivity (clDice) typically exceeding those in region overlap (IoU), indicating improved structural continuity. With only $\sim$0.26M additional parameters, PEPA offers a practical post-encoder enhancement for structure-centric segmentation.
Chinese Translation
曲线物体分割，包括血管和裂缝，由于极端的空间稀疏性和拓扑脆弱性而具有挑战性，其中小的局部错误可能导致严重的结构断裂。同时，现代分割管道越来越依赖于强大但难以修改的基础编码器，其重采样限制了细微结构的恢复。基于此，我们关注后编码器阶段，并研究两种反复出现且可操作的失败模式：高分辨率特征恢复中的重建瓶颈和二值化中的决策瓶颈。我们提出了PEPA，一种轻量级的后编码器插入适配器，适用于具有可访问解码器/头特征和目标、查询或类别描述符的2D曲线分割管道。PEPA结合了(i) 目标条件蛇形上采样（Target-Conditioned Snake Upsampling, TCSU），该方法利用目标条件的连续蛇形采样在上采样过程中更好地恢复细长和曲折的结构，以及(ii) 目标自适应可微阈值化（Target-Adaptive Differentiable Thresholding, TADT），该方法预测目标特定的阈值，并优化具有明确防止平凡偏差转移的软阈值替代品。在这个后编码器接口下，PEPA可以附加到基于提示的解码器和传统的密集预测器上。在五个医学和工业基准上的实验表明，将PEPA添加到冻结编码器基线中可以带来一致的改进，拓扑连通性（clDice）的提升通常超过区域重叠（IoU）的提升，表明结构连续性得到了改善。仅需约0.26M的额外参数，PEPA为以结构为中心的分割提供了实用的后编码器增强。
cs.CV / 265 / 2606.23494
Brain-Adapter: A Dual-Stream Vision-Language MIL Framework for Comprehensive 3D CT Diagnosis of Acute Intracranial Pathologies
脑适配器：一种双流视觉-语言多实例学习框架，用于全面的急性颅内病理3D CT诊断
Abstract
Automated diagnosis of 3D brain CT scans is essential for critical care, yet it remains challenging due to the heavy reliance on manual annotations and the limited semantic understanding of conventional models. While 2D foundation vision-language models (VLMs) have shown remarkable generalization, effectively transferring their representational power to 3D volumes remains an open problem. In this paper, we propose Brain-Adapter, a novel dual-stream multiple instance learning (MIL) framework that leverages pre-trained 2D biomedical VLMs and raw diagnostic reports for robust scan-level multi-label classification. Specifically, we introduce a Text-Conditioned Attention (TCA) mechanism, utilizing raw diagnostic sentences as semantic queries to dynamically align visual cues with specific disease concepts. Concurrently, a parallel visual MIL stream captures global scan characteristics, supervised by structured labels extracted via a Large Language Model (LLM). To ensure representation coherence, a consistency constraint enforces synergy between the two streams. During inference, an Uncertainty-Aware Refinement (UAR) module dynamically calibrates and fuses these dual-stream predictions to resolve ambiguous cases. Extensive experiments demonstrate that our method significantly outperforms state-of-the-art 3D models and standard MIL approaches. By eliminating the reliance on dense annotations, Brain-Adapter provides a highly scalable and clinically viable solution for 3D acute intracranial pathology analysis.
Chinese Translation
3D脑CT扫描的自动诊断对于重症护理至关重要，但由于对手动标注的高度依赖以及传统模型的有限语义理解，这一任务仍然具有挑战性。尽管2D基础视觉-语言模型（VLMs）表现出显著的泛化能力，但有效地将其表征能力转移到3D体积仍然是一个未解决的问题。本文提出了脑适配器（Brain-Adapter），一种新颖的双流多实例学习（MIL）框架，利用预训练的2D生物医学VLM和原始诊断报告进行稳健的扫描级多标签分类。具体而言，我们引入了一种文本条件注意力（Text-Conditioned Attention, TCA）机制，利用原始诊断句子作为语义查询，动态地将视觉线索与特定疾病概念对齐。同时，平行的视觉MIL流捕捉全局扫描特征，通过大型语言模型（Large Language Model, LLM）提取的结构化标签进行监督。为了确保表征的一致性，引入一致性约束以增强两个流之间的协同作用。在推理过程中，不确定性感知精炼（Uncertainty-Aware Refinement, UAR）模块动态地校准和融合这两个流的预测，以解决模糊案例。大量实验表明，我们的方法显著优于最先进的3D模型和标准MIL方法。通过消除对密集标注的依赖，脑适配器为3D急性颅内病理分析提供了一种高度可扩展且临床可行的解决方案。
cs.CV / 266 / 2606.23503
UniverSat: Resolution- and Modality-Agnostic Transformers for Earth Observation
UniverSat：面向地球观测的分辨率和模态无关的变换器
Abstract
Vision Transformers (ViT) dominate computer vision. However, their reliance on rigid patch projectors hinders transfer to Earth Observation (EO), where input modalities, scales, and resolutions vary widely. We introduce UniverSat, a ViT-style backbone built around a Universal Patch Encoder that maps patches from arbitrary spatial, spectral, and temporal resolutions, and from both optical and non-optical sensors, into a shared embedding space with a shared set of weights. This enables training a single model on heterogeneous multimodal corpora via self-supervision, yielding robust, sensor-agnostic spatial features. We validate this approach with strong results across classification and segmentation on standard EO benchmarks from GeoBench, PANGEABench, and SpectralEarth. Our code and models are available at https://github.com/gastruc/UniverSat.
Chinese Translation
视觉变换器（ViT）在计算机视觉领域占据主导地位。然而，它们对刚性补丁投影器的依赖阻碍了在地球观测（EO）中的应用，因为输入模态、尺度和分辨率差异很大。我们提出了UniverSat，这是一种基于ViT风格的骨干网络，围绕一个通用补丁编码器构建，该编码器将来自任意空间、光谱和时间分辨率的补丁，以及来自光学和非光学传感器的补丁，映射到一个共享的嵌入空间中，并使用一组共享的权重。这使得通过自监督在异质多模态语料库上训练单一模型成为可能，从而产生稳健的、传感器无关的空间特征。我们在GeoBench、PANGEABench和SpectralEarth等标准EO基准上验证了这一方法，取得了良好的分类和分割结果。我们的代码和模型可在https://github.com/gastruc/UniverSat获取。
cs.CV / 267 / 2606.23514
Arbor: Explicit Geometric Conditioning for Controllable 3D Asset Generation
Arbor：用于可控3D资产生成的显式几何约束
Abstract
Text and image conditioned 3D models now generate convincing assets, but they still offer little direct control over the space an object should occupy or avoid. In authoring, this spatial intent is often known before generation starts. A chair should fit a seating envelope, a prop should leave clearance for motion, or a part should expose a contact surface. Prompts and image views are poor carriers for such constraints, requiring the need for an explicit control interface. We present Arbor, a trainable attachment for text conditioned latent 3D generation. Arbor introduces constraint meshes as a native 3D control interface. The interface uses hull regions where geometry should exist, avoidance regions that should remain empty, and touch regions the object should contact. Unlike completion or whole object scaffold control, these meshes are not target evidence. They are local typed requirements and can include regions where no surface should appear. Arbor keeps this signal as geometry by converting constraint meshes into tokens and learning a routed attachment inside a frozen denoiser. Each latent region can therefore receive the part of the constraint that matters for its spatial location. We evaluate Arbor on automatic and artist curated control benchmarks with hull, avoidance, and touch constraints, and compare the metric trends to a user preference study. Even without dedicated compliance losses, Arbor improves constraint obedience while preserving object quality and variation under fixed constraints.
Chinese Translation
文本和图像条件下的3D模型现在能够生成令人信服的资产，但它们在对象应占据或避免的空间方面仍然提供很少的直接控制。在创作过程中，这种空间意图通常在生成开始之前就已确定。椅子应适合座位包络，道具应留出运动空间，或部件应暴露接触表面。提示和图像视图对于这些约束的传递效果不佳，因此需要一个显式的控制接口。我们提出了Arbor，一个用于文本条件潜在3D生成的可训练附加模块。Arbor引入了约束网格作为原生的3D控制接口。该接口使用几何体应存在的外壳区域、应保持空白的避免区域以及对象应接触的接触区域。与完成或整体对象支架控制不同，这些网格不是目标证据。它们是局部类型的要求，可以包括不应出现表面的区域。Arbor通过将约束网格转换为标记并在一个冻结的去噪器内部学习一个路由附加模块来保持这一信号作为几何体。因此，每个潜在区域可以接收与其空间位置相关的约束部分。我们在具有外壳、避免和接触约束的自动和艺术家策划控制基准上评估Arbor，并将指标趋势与用户偏好研究进行比较。即使没有专门的合规损失，Arbor在保持对象质量和变化的同时，提高了约束遵从性。
cs.CV / 268 / 2606.23524
Scaling State-Space Models from Lines to Paragraphs: An Ablation of Mamba-based OCR
从行到段落的状态空间模型扩展：基于 Mamba 的 OCR 的消融研究
Abstract
End-to-end OCR increasingly relies on autoregressive sequence models, where the quadratic cost of Transformer attention limits efficient transcription of long, paragraph-level text. State-Space Models (SSMs) such as Mamba offer linear-time decoding and have recently been shown to match Transformer accuracy on printed historical lines, but their behavior as sequences grow from short lines to full paragraphs, and their generalization to handwriting, remain poorly understood. We study how a Mamba-based OCR recognizer scales from lines to paragraphs. We first conduct a systematic exploration of its four core hyperparameters (decoder depth, state dimension, expansion factor, and connector depth) on synthetic paragraphs from 100 to 1,000 characters, identifying the recurrent state dimension and the expansion factor as the dominant levers for long-sequence accuracy. We then compare the recognizer against a Transformer baseline trained under an identical protocol. On clean synthetic paragraphs, both models stay below 1% CER at every length while the SSM runs 1.4 to 4.5 times faster, the speedup growing with sequence length. On real handwriting, however, the SSM lags clearly behind: it reaches 8.2% CER on IAM lines and 10.0% on IAM paragraphs, against 4.2% and 3.5% for the Transformer baseline. Through controlled experiments we show that a substantial part of this gap stems from data scarcity rather than from an intrinsic architectural limit: the autoregressive SSM decoder is markedly data-hungry on long sequences. Our study clarifies when SSMs are a practical choice for large-scale document transcription and when they are not.
Chinese Translation
端到端的光学字符识别（OCR）越来越依赖自回归序列模型，其中 Transformer 注意力的二次成本限制了长段落级文本的高效转录。状态空间模型（SSMs）如 Mamba 提供线性时间解码，最近已被证明在印刷历史行上与 Transformer 的准确性相匹配，但随着序列从短行扩展到完整段落，它们的行为以及对手写文本的泛化能力仍然不够清楚。我们研究了基于 Mamba 的 OCR 识别器如何从行扩展到段落。我们首先对其四个核心超参数（解码器深度、状态维度、扩展因子和连接器深度）在从 100 到 1000 个字符的合成段落上进行了系统探索，识别出递归状态维度和扩展因子是影响长序列准确性的主要因素。然后，我们将识别器与在相同协议下训练的 Transformer 基线进行了比较。在干净的合成段落上，两个模型在每个长度下的字符错误率（CER）均低于 1%，而 SSM 的运行速度比 Transformer 快 1.4 到 4.5 倍，速度提升随着序列长度的增加而增长。然而，在真实手写文本上，SSM 明显落后：在 IAM 行上达到 8.2% CER，在 IAM 段落上达到 10.0%，而 Transformer 基线分别为 4.2% 和 3.5%。通过控制实验，我们表明，这一差距的很大一部分源于数据稀缺，而非内在的架构限制：自回归 SSM 解码器在长序列上显著需要更多数据。我们的研究阐明了何时 SSM 是大规模文档转录的实用选择，何时则不是。
cs.CV / 269 / 2606.23539
LightSTAR: Efficient Visual Document Retrieval via Lightweight Selection with Vision-Adaptive Refinement
LightSTAR：通过轻量选择与视觉自适应细化实现高效视觉文档检索
Abstract
Visual document retrieval requires rapidly locating relevant pages from large multi-modal corpora in response to user queries. While recent methods powered by Multi-modal Large Language Models (MLLMs) show competitive accuracy, they suffer from prohibitive computational costs by applying intensive MLLM encoding to every single page. Meanwhile, we observe that user queries are typically keyword-anchored, containing semantically rich words that are expected to appear directly in the visible text of relevant pages, offering an efficient cue for quickly narrowing down candidate pages. Building on this insight, we propose LightSTAR, an efficient framework that decomposes visual document retrieval into: 1) LLM-free Visual Selection, which utilizes content-grounded query encoding to focus on informative words and employs LLM-free visual embeddings to produce a high-recall candidate set; and 2) Vision-adaptive Semantic Refinement, which further performs fine-grained semantic matching exclusively on these top candidates via adaptive region-wise feature fusion to effectively combine textual and layout cues, optimized through a hardness-aware contrastive objective. Experimental results demonstrate that LightSTAR achieves state-of-the-art retrieval accuracy while reducing end-to-end latency by several-fold, offering a highly practical solution to the accuracy-efficiency trade-off in visual document retrieval. Code is available at https://github.com/bokufa/LightSTAR.
Chinese Translation
视觉文档检索需要快速定位来自大型多模态语料库的相关页面，以响应用户查询。尽管最近由多模态大型语言模型（Multi-modal Large Language Models, MLLMs）驱动的方法在准确性上表现出竞争力，但由于对每个页面应用密集的MLLM编码，导致其计算成本高昂。同时，我们观察到用户查询通常以关键词为基础，包含丰富的语义词，这些词预计会直接出现在相关页面的可见文本中，为快速缩小候选页面范围提供了有效线索。基于这一洞察，我们提出了LightSTAR，一个高效的框架，将视觉文档检索分解为：1）无LLM的视觉选择，该方法利用基于内容的查询编码来关注信息丰富的词汇，并采用无LLM的视觉嵌入生成高召回率的候选集；2）视觉自适应语义细化，该方法进一步在这些顶级候选上进行细粒度语义匹配，通过自适应区域特征融合有效结合文本和布局线索，并通过关注难度的对比目标进行优化。实验结果表明，LightSTAR在实现最先进的检索准确性的同时，将端到端延迟降低了数倍，为视觉文档检索中的准确性与效率权衡提供了高度实用的解决方案。代码可在 https://github.com/bokufa/LightSTAR 获取。
cs.CV / 270 / 2606.23542
AwakeForest: An Interactive Geospatial Platform for Large-Scale Forest Imagery
AwakeForest：一个用于大规模森林影像的互动地理空间平台
Abstract
Forest imagery analysis often involves multiple tightly coupled vision tasks, which must be performed under substantial variation in geographic regions, sensors, and acquisition conditions. However, practitioners often lack a unified tool that is geospatial-native, cloud-optimized, and ML-integrated for end-to-end workflows spanning annotation, prediction, visualization, and downstream analysis at scale. We present AwakeForest, an interactive end-to-end platform designed for large-scale forest imagery that integrates model-assisted inference, automatic annotation, and human-in-the-loop refinement within a single workflow. Our platform supports plug-and-play integration of pretrained models and enables scalable interaction with forest imagery ranging from standard aerial scenes to large orthomosaics that can span several gigabytes to hundreds of gigabytes. AwakeForest produces analysis-ready outputs that can be directly used for downstream analysis and to support iterative model and annotation updates on new scenes. We demonstrate the system on the PALMS dataset and illustrate how AwakeForest supports an end-to-end workflow for practical forest management and analysis.
Chinese Translation
森林影像分析通常涉及多个紧密耦合的视觉任务，这些任务必须在地理区域、传感器和获取条件的显著变化下进行。然而，实践者往往缺乏一个统一的工具，该工具应具备地理空间原生、云优化和机器学习集成的能力，以支持从标注、预测、可视化到大规模下游分析的端到端工作流程。我们提出了AwakeForest，这是一个为大规模森林影像设计的互动端到端平台，集成了模型辅助推断、自动标注和人机协作精炼于一个单一工作流程中。我们的平台支持预训练模型的即插即用集成，并能够与从标准航空场景到大型正射拼接图（可达数十GB到数百GB）的森林影像进行可扩展的交互。AwakeForest生成的分析就绪输出可以直接用于下游分析，并支持对新场景进行迭代的模型和标注更新。我们在PALMS数据集上演示了该系统，并说明了AwakeForest如何支持实际森林管理和分析的端到端工作流程。
cs.CV / 271 / 2606.23557
Dense Reward for Multi-View 3D Reasoning with Global Maps and Local Views
基于全局地图和局部视图的多视角三维推理的密集奖励
Abstract
Multi-view 3D Visual Question Answering (MV3D-VQA) requires integrating partial observations into a coherent 3D scene representation and selecting informative viewpoints for multi-step spatial reasoning. However, current multimodal LLMs are typically trained with sparse, answer-level supervision, which often yields inconsistent cross-view reasoning and brittle view selection. We present DR-MV3D (Dense Reward for MV3D-VQA), a map-grounded learning framework that provides dense, verifiable rewards to supervise the reasoning process. Our approach decomposes MV3D-VQA into (i) allocentric global map construction, (ii) question-conditioned view-trajectory planning, and (iii) egocentric grounding for answer prediction. To make intermediate steps learnable without manual annotations, we introduce two rewards: a global consistency reward that aligns the predicted map with geometry-consistent pseudo targets from frozen 3D vision foundation models (e.g., VGGT + SAM3), and a local trajectory reward that supervises ordered viewpoint selection. We optimize the full pipeline with trajectory-level policy optimization (GRPO). Experiments on MindCube, VSI-Bench, and BLINK (MV) show that DR-MV3D consistently improves over strong multi-image baselines, supporting the effectiveness of process-level dense supervision for multi-view 3D reasoning.
Chinese Translation
多视角三维视觉问答（MV3D-VQA）需要将部分观察整合为一致的三维场景表示，并选择信息丰富的视点以进行多步空间推理。然而，目前的多模态大语言模型（LLMs）通常是在稀疏的、基于答案的监督下训练的，这往往导致跨视图推理不一致和视点选择脆弱。我们提出了DR-MV3D（MV3D-VQA的密集奖励），这是一个基于地图的学习框架，提供密集的、可验证的奖励来监督推理过程。我们的方法将MV3D-VQA分解为（i）以中心为基础的全局地图构建，（ii）基于问题的视点轨迹规划，以及（iii）以自我为中心的答案预测基础。为了使中间步骤可学习而无需手动标注，我们引入了两种奖励：一种全局一致性奖励，用于将预测的地图与来自冻结的三维视觉基础模型（如VGGT + SAM3）的一致几何伪目标对齐；另一种局部轨迹奖励，用于监督有序视点选择。我们通过轨迹级策略优化（GRPO）优化整个流程。在MindCube、VSI-Bench和BLINK（MV）上的实验表明，DR-MV3D在强大的多图像基线之上始终表现出改进，支持了过程级密集监督在多视角三维推理中的有效性。
cs.CV / 272 / 2606.23604
Polycepta: Object-Centric Appearance Estimation for Multi-Object Tracking
Polycepta：面向多目标跟踪的对象中心外观估计
Abstract
The tracking-by-detection paradigm in multi-object tracking (MOT) typically relies on static appearance descriptors to complement motion estimation. However, these descriptors are frame-independent, limiting their robustness as visual cues. Since such descriptors are often obtained from computationally intensive pretrained backbones, real-time MOT systems frequently abandon appearance cues altogether and rely solely on motion prediction and geometric association. In this work, we introduce Polycepta, an object-centric appearance state estimation framework that reformulates appearance modeling as a recursive estimation problem rather than a frame-wise matching task. Polycepta constructs and continuously updates an independent appearance state for each tracked object, enabling future appearance representations to be estimated from accumulated observations. Polycepta is encouraged to learn the appearance-state construction of object-specific representations rather than memorize them through a proposed learning strategy, enabling appearance estimation for unseen classes. A key property of Polycepta is that the quality of appearance estimation improves as object states evolve during inference. While conventional appearance descriptors remain static or degrade over time, Polycepta progressively refines appearance estimates as additional observations are accumulated. Extensive experiments on KITTI, the Waymo Open Dataset, and MOT17 demonstrate consistent reductions in identity switches and improvements in tracking performance when integrated into the tracking-by-detection pipelines. Polycepta operates at 90.57 Hz and delivers state-of-the-art performance on the KITTI benchmark when integrated into the RobMOT framework, achieving a MOTA of 92.27\%.
Chinese Translation
多目标跟踪（MOT）中的检测跟踪范式通常依赖于静态外观描述符来补充运动估计。然而，这些描述符是帧无关的，限制了它们作为视觉线索的鲁棒性。由于此类描述符通常来自计算密集型的预训练骨干网络，实时MOT系统常常完全放弃外观线索，仅依赖运动预测和几何关联。在本研究中，我们提出了Polycepta，一种对象中心的外观状态估计框架，将外观建模重新表述为递归估计问题，而不是逐帧匹配任务。Polycepta为每个被跟踪对象构建并持续更新独立的外观状态，使未来的外观表示能够从累积的观察中进行估计。Polycepta通过一种提出的学习策略，鼓励学习对象特定表示的外观状态构建，而不是简单记忆，从而实现对未见类别的外观估计。Polycepta的一个关键特性是，随着对象状态在推理过程中演变，外观估计的质量不断提高。尽管传统的外观描述符保持静态或随时间退化，Polycepta在累积额外观察时逐步细化外观估计。在KITTI、Waymo开放数据集和MOT17上的大量实验表明，将其集成到检测跟踪管道中时，身份切换显著减少，跟踪性能得到改善。Polycepta以90.57 Hz的频率运行，并在集成到RobMOT框架时在KITTI基准上实现了92.27%的MOTA，展现了最先进的性能。
cs.CV / 273 / 2606.23610
Vera: A Layered Diffusion Model for Content-Preserving Video Editing
Vera：一种用于内容保留视频编辑的分层扩散模型
Abstract
Video diffusion models have enabled remarkable progress in video generation and editing. However, content preservation remains a core challenge: existing methods regenerate every pixel and often alter elements that should remain unchanged, such as characters or background scenes. We introduce Vera, a layered diffusion framework for content-preserving video editing. Instead of regenerating the entire video, Vera generates an edit layer along with an alpha matte for compositing with the source video, separating creative editing from content preservation by design. To encourage coherent composition with the source video, we extend the text-to-video DiT into a Mixture-of-Transformers (MoT) architecture, with separate DiTs for each layer that interact through joint self-attention. To support the training of Vera, we further construct a high-quality layered dataset with accurate alpha mattes, diverse scenes and dynamics, and visual effects. Across our quantitative benchmark and human preference study, Vera outperforms leading open-source video editing models in content preservation while remaining competitive in edit quality, using 486K frames of layered training data.
Chinese Translation
视频扩散模型在视频生成和编辑方面取得了显著进展。然而，内容保留仍然是一个核心挑战：现有方法会重新生成每一个像素，常常改变那些应该保持不变的元素，如角色或背景场景。我们提出了Vera，一种用于内容保留视频编辑的分层扩散框架。Vera并不是重新生成整个视频，而是生成一个编辑层以及用于与源视频合成的α通道，从设计上将创意编辑与内容保留分开。为了促进与源视频的连贯合成，我们将文本到视频的DiT扩展为混合变换器（Mixture-of-Transformers, MoT）架构，为每一层设置独立的DiT，并通过联合自注意力进行交互。为了支持Vera的训练，我们进一步构建了一个高质量的分层数据集，包含准确的α通道、多样的场景和动态以及视觉效果。在我们的定量基准测试和人类偏好研究中，Vera在内容保留方面优于领先的开源视频编辑模型，同时在编辑质量上保持竞争力，使用了486K帧的分层训练数据。
cs.CV / 274 / 2606.23611
Data Selection Through Iterative Self-Filtering for Vision-Language Settings
通过迭代自我过滤进行数据选择的视觉-语言设置
Abstract
The availability of large amounts of clean data is paramount to training neural networks. However, at large scales, manual oversight is impractical, resulting in sizeable datasets that can be very noisy. Attempts to mitigate this obstacle to producing performant vision-language models have so far involved heuristics, curated reference datasets, and using pre-trained models. Here we propose a novel, bootstrapped method in which a CLIP model is trained on an evolving, self-selected dataset. This evolving dataset constitutes a balance of filtered, highly probable clean samples as well as diverse samples from the entire distribution. Our proposed Self-Filtering method iterates between training the model and selecting a subsequently improved data mixture. Training on vision-language datasets filtered by the proposed approach improves downstream performance without the need for additional data or pre-trained models.
Chinese Translation
大量干净数据的可用性对于训练神经网络至关重要。然而，在大规模情况下，人工监督是不切实际的，导致生成的庞大数据集可能非常嘈杂。迄今为止，缓解这一障碍以产生高性能视觉-语言模型的尝试主要涉及启发式方法、精心策划的参考数据集以及使用预训练模型。在此，我们提出了一种新颖的自举方法，其中 CLIP 模型在一个不断演变的自我选择数据集上进行训练。这个不断演变的数据集由经过过滤的、高概率的干净样本与来自整个分布的多样样本之间的平衡组成。我们提出的自我过滤方法在训练模型和选择随后改进的数据混合之间进行迭代。在通过所提出的方法过滤的视觉-语言数据集上进行训练，可以提高下游性能，而无需额外的数据或预训练模型。
cs.CV / 275 / 2606.23615
Hedgementation = Hedgerow Segmentation: A Remote Sensing Benchmark
Hedgementation = 灌木篱笆分割：一个遥感基准
Abstract
We propose Hedgementation: a new benchmark to evaluate machine learning models for hedgerow mapping from remote sensing data at country scale and 10m$^2$ spatial resolution. We combine and harmonize multiple remote sensing data products and ground truth labels sourced from a hedgerow inventory in France. We measure the ability of three baseline models to generalize across spatial distance, and across climatic zones, a more explicitly challenging task. Our benchmark tests both supervised and self-supervised learning approaches for remote sensing, applied to tracking fine-scale features of high agricultural importance. The code to reproduce the benchmark and baselines results is available at https://github.com/hedgementation/hedgementation.
Chinese Translation
我们提出了Hedgementation：一个新的基准，用于评估机器学习模型在国家尺度和10m²空间分辨率下从遥感数据中进行灌木篱笆映射的能力。我们结合并协调了多个遥感数据产品和来自法国灌木篱笆清单的地面真值标签。我们测量了三种基线模型在空间距离和气候区之间的泛化能力，这是一个更具挑战性的任务。我们的基准测试了监督学习和自监督学习方法在遥感中的应用，旨在跟踪具有重要农业价值的细尺度特征。用于重现基准和基线结果的代码可在https://github.com/hedgementation/hedgementation获取。
cs.CV / 276 / 2606.23634
Pose Anything Anywhere:Model-free Object Poses from Arbitrary References
在任何地方定位任何物体：无模型的物体姿态估计方法
Abstract
Estimating the 6D pose of unseen objects is a fundamental yet challenging problem for open-world robotics and embodied perception. Model-based methods are accurate but depend on CAD assets or heavy onboarding, while most model-free approaches are still limited to pairwise single-anchor matching and thus fail under occlusion and large viewpoint changes with low query-reference overlap. Therefore, we present PANY, a unified model-free framework that seamlessly supports both RGB and RGB-D inputs, operates on one or sparse pose-free reference views, and generalizes effectively to novel objects. Built on a multi-view transformer geometry backbone, PANY moves beyond pairwise matching by learning view-consistent geometry and cross-view alignment cues that remain stable under wide baselines and limited overlap. When additional unposed assist views are available, PANY aggregates them via pose-graph canonical registration to increase geometric coverage and reinforce the final pose. Extensive experiments show that PANY achieves state-of-the-art performance across multiple benchmarks, substantially outperforming existing model-free methods, improving pose accuracy by +12% on YCB-V and over +20% on LM-O. Furthermore, PANY consistently performs well under both single-reference and sparse-reference settings, demonstrating strong robustness in real-world environments.
Chinese Translation
估计未见物体的六维姿态是开放世界机器人技术和具身感知中的一个基本而具有挑战性的问题。基于模型的方法虽然准确，但依赖于计算机辅助设计（CAD）资产或繁重的上手过程，而大多数无模型的方法仍然局限于成对的单锚匹配，因此在遮挡和大视角变化下，低查询-参考重叠的情况下表现不佳。因此，我们提出了PANY，一个统一的无模型框架，能够无缝支持RGB和RGB-D输入，基于一个或稀疏的无姿态参考视图进行操作，并有效地推广到新物体。PANY建立在多视图变换器几何骨干上，超越了成对匹配，通过学习视图一致的几何形状和在宽基线和有限重叠下保持稳定的跨视图对齐线索。当额外的无姿态辅助视图可用时，PANY通过姿态图规范化注册将其聚合，以增加几何覆盖范围并增强最终姿态。大量实验表明，PANY在多个基准测试中实现了最先进的性能，显著超越现有的无模型方法，在YCB-V上提高了+12%的姿态准确性，在LM-O上提高了超过+20%。此外，PANY在单参考和稀疏参考设置下均表现良好，展示了在真实环境中的强鲁棒性。
cs.CV / 277 / 2606.23653
Lightweight Neural Framework for Robust 3D Volume and Surface Estimation from Multi-View Images
轻量级神经框架用于从多视角图像中进行稳健的三维体积和表面积估计
Abstract
Accurate volume and surface area estimation is critical for diverse applications, from marine ecology to medical diagnostics. However, existing methods often suffer from high computational costs and poor performance with sparse and noisy data. We propose a fully feed-forward framework that regresses scale-normalized volume and surface area and their associated uncertainties directly from multi-view images. By fusing 3D point cloud reconstructions with view-aligned 2D features through a graph-based decoder, our model bypasses iterative optimization, ensuring exceptional scalability and rapid inference. Experimental results demonstrate that our approach outperforms state-of-the-art methods, particularly when operating with a low number of input images. Validated across coral monitoring, dietary analysis, and anthropometry, our proposed framework provides a robust, adaptable solution for quantitative shape analysis. This architecture provides a high-speed, scalable alternative for precise geometric estimation from visual data, maintaining high performance even in resource-constrained or sparse-view scenarios.
Chinese Translation
准确的体积和表面积估计对于从海洋生态到医学诊断等多种应用至关重要。然而，现有方法往往面临高计算成本以及在稀疏和噪声数据下表现不佳的问题。我们提出了一种完全前馈的框架，该框架直接从多视角图像中回归尺度归一化的体积和表面积及其相关的不确定性。通过通过基于图的解码器将三维点云重建与视图对齐的二维特征融合，我们的模型绕过了迭代优化，确保了卓越的可扩展性和快速推理。实验结果表明，我们的方法在性能上优于最先进的技术，特别是在输入图像数量较少的情况下。经过珊瑚监测、饮食分析和人类测量的验证，我们提出的框架为定量形状分析提供了一种稳健且适应性强的解决方案。该架构为从视觉数据中进行精确几何估计提供了一种高速、可扩展的替代方案，即使在资源受限或视角稀疏的情况下也能保持高性能。
cs.CV / 278 / 2606.23669
GeoFidelity-Bench: Evaluating Segment-Level Geographic Fidelity in Text-to-Image Street-View Generation
GeoFidelity-Bench：评估文本到图像街景生成中的段级地理保真度
Abstract
Text-to-image models can generate visually plausible city streets, but whether their outputs correspond to a requested road segment rather than a generic city prior remains unclear. We introduce GeoFidelity-Bench, a reference-panel benchmark for segment-conditioned geographic fidelity in street-view generation. It contains 7,117 curated Mapillary images covering 109 named OpenStreetMap road segments in 25 cities across six continents. For each generated panel, the benchmark ranks the target reference panel against panels from the nearest segment in the same city, other segments in the same city, and segments from other cities, making local discrimination rather than absolute target similarity the primary test. We evaluate six open-weight text-to-image generators under city-only, street-and-neighborhood, and GPS-augmented prompts. Adding street and neighborhood names is associated with an increase of 5.5 percentage points in top-1 retrieval accuracy over city-only prompts, with a 95% confidence interval from 3.4 to 7.7 percentage points. However, the similarity margin between the target and the nearest segment in the same city remains near zero, indicating that local names improve broad local plausibility more than exact segment identity. Prompts that keep the city fixed but use incorrect street or neighborhood names further show that only part of the gain depends on the correct local names, while appending raw GPS coordinates as ordinary text yields no statistically clear additional benefit. Held-out real-image queries successfully recover segment identity, showing that the curated references contain usable segment-level signal. GeoFidelity-Bench thus reveals a persistent gap between city- or neighborhood-plausible street-view generation and faithful generation for a specific road segment.
Chinese Translation
文本到图像模型可以生成视觉上合理的城市街道，但其输出是否对应于请求的道路段而非通用城市先验仍不明确。我们引入了GeoFidelity-Bench，这是一个用于街景生成中段条件地理保真度的参考面板基准。该基准包含7,117张经过精心挑选的Mapillary图像，覆盖了25个城市中109个命名的OpenStreetMap道路段，分布在六大洲。对于每个生成的面板，该基准将目标参考面板与同一城市中最近段的面板、同一城市中其他段的面板以及其他城市的段进行排名，使得局部区分而非绝对目标相似性成为主要测试。我们在仅城市、街道和邻里以及GPS增强提示下评估了六个开放权重的文本到图像生成器。添加街道和邻里名称与仅城市提示相比，顶级检索准确率提高了5.5个百分点，95%的置信区间为3.4到7.7个百分点。然而，目标与同一城市中最近段之间的相似度差距仍接近于零，表明局部名称在提高广泛的局部合理性方面的作用大于精确的段身份。保持城市不变但使用不正确的街道或邻里名称的提示进一步表明，只有部分增益依赖于正确的地方名称，而将原始GPS坐标作为普通文本附加则没有统计学上明显的额外好处。保留的真实图像查询成功恢复了段身份，表明经过精心挑选的参考包含可用的段级信号。因此，GeoFidelity-Bench揭示了城市或邻里合理的街景生成与特定道路段的忠实生成之间存在持续的差距。
cs.CV / 279 / 2606.23675
IMAGIN-4D: Image-Guided Controllable Interaction Generation
IMAGIN-4D：图像引导的可控交互生成
Abstract
Generating human-object interactions (HOI) is central to character animation, robotics, AR/VR, and embodied AI. Recent HOI generation methods synthesize motion from text, object geometry, and sparse waypoints, controlling action semantics and object trajectories. However, these signals underspecify interaction: the same prompt and trajectory can produce different grasps, approach directions, body poses, object poses, contacts, and body-object layouts. We address this ambiguity with a reference image as a visual specification of the desired interaction snapshot. However, a single global image representation conflates distinct cues and conditions all frames on identical visual evidence. We therefore introduce IMAGIN-4D, a diffusion-based HOI generator that decomposes image conditioning spatio-temporally. For spatial conditioning, IMAGIN-4D extracts supervised interaction-state tokens for body pose, object pose, body-object contact, and spatial relationships at the depicted frame. For temporal conditioning, it computes frame-aware tokens by querying image patches per generated frame, allowing sequence segments to attend to different visual cues from the same image. To balance image, text, and waypoint cues, IMAGIN-4D uses role-aware conditioning: text, waypoints, and interaction-state tokens use separate AdaLN streams, while frame-aware visual tokens cross-attend with motion tokens. Since HOI motion datasets lack paired images, we build a synthetic motion-to-image rendering pipeline from FullBodyManipulation (FBM) and introduce an image-adherence metric to evaluate whether generated motions match the reference snapshot. Experiments on FBM and BEHAVE show that IMAGIN-4D improves fine-grained interaction control over single-token and uniformly image-conditioned baselines while preserving waypoint-following and motion quality. Code and models will be released at https://imagin4d.github.io.
Chinese Translation
生成人与物体的交互（HOI）是角色动画、机器人技术、增强现实/虚拟现实（AR/VR）和具身人工智能的核心。近期的HOI生成方法通过文本、物体几何形状和稀疏的关键点合成运动，控制动作语义和物体轨迹。然而，这些信号对交互的描述不足：相同的提示和轨迹可以产生不同的抓取方式、接近方向、身体姿势、物体姿势、接触和身体-物体布局。我们通过参考图像来解决这种模糊性，作为所需交互快照的视觉规范。然而，单一的全局图像表示混淆了不同的线索，并使所有帧基于相同的视觉证据进行条件化。因此，我们提出了IMAGIN-4D，这是一种基于扩散的HOI生成器，能够时空分解图像条件化。在空间条件化方面，IMAGIN-4D提取了监督的交互状态标记，用于描述帧中的身体姿势、物体姿势、身体-物体接触和空间关系。在时间条件化方面，它通过查询每个生成帧的图像块计算帧感知标记，使得序列片段能够关注来自同一图像的不同视觉线索。为了平衡图像、文本和关键点线索，IMAGIN-4D采用了角色感知条件化：文本、关键点和交互状态标记使用独立的AdaLN流，而帧感知视觉标记与运动标记交叉关注。由于HOI运动数据集缺乏配对图像，我们构建了一个合成的运动到图像渲染管道，基于全身操作（FullBodyManipulation, FBM），并引入了一种图像遵循度量来评估生成的运动是否与参考快照匹配。在FBM和BEHAVE上的实验表明，IMAGIN-4D在细粒度交互控制方面优于单标记和均匀图像条件化的基线，同时保持了关键点跟踪和运动质量。代码和模型将发布在 https://imagin4d.github.io。
cs.CV / 280 / 2606.23678
AIR: Adaptive Interleaved Reasoning with Code in MLLMs
AIR：在多模态大语言模型中的自适应交错推理与代码
Abstract
Following the paradigm shift initiated by OpenAI o3, interleaved reasoning with code to enhance multimodal large language models (MLLMs) has become a pivotal research frontier. The existing literature focuses primarily on tool-use within vision-perception tasks. However, such approaches typically rely on predefined heuristics for visual manipulation and are inherently incapable of addressing numerical computation problems due to their exclusive focus on visual operations. This paper empowers MLLMs with adaptive interleaved reasoning capabilities through extended reinforcement learning training on code-augmented complex numerical computation tasks. To this end, we propose a comprehensive three-component solution consisting of: a two-stage cold-start data construction pipeline, data filtering strategies for RL dataset curation, and an adaptive tool-invocation strategy leveraging a group-constrained reward function for interleaved reasoning trajectories. Extensive experiments demonstrate that after Reinforcement Learning training with the group-constrained reward function, performance improves by an average of 6.1 percentage points (pp) on evaluation benchmarks. Specifically, the accuracy for interleaved reasoning samples increases by 9.9 pp, and the overall success rate of tool-use exceeds 95%. Our data and code are available at: https://github.com/CongHan0808/AIR.git.
Chinese Translation
随着OpenAI o3引发的范式转变，利用代码进行交错推理以增强多模态大语言模型（MLLMs）已成为一个关键的研究前沿。现有文献主要集中于视觉感知任务中的工具使用。然而，这些方法通常依赖于预定义的启发式规则进行视觉操作，因而在处理数值计算问题时表现不足。本文通过在增强代码的复杂数值计算任务上进行扩展的强化学习训练，使MLLMs具备自适应交错推理能力。为此，我们提出了一个由三个组成部分构成的综合解决方案：一个两阶段的冷启动数据构建流程、用于强化学习数据集策划的数据过滤策略，以及利用群体约束奖励函数的自适应工具调用策略，以支持交错推理轨迹。大量实验表明，在使用群体约束奖励函数进行强化学习训练后，评估基准的性能平均提高了6.1个百分点（pp）。具体而言，交错推理样本的准确率提高了9.9 pp，工具使用的整体成功率超过95%。我们的数据和代码可在以下链接获取：https://github.com/CongHan0808/AIR.git。
cs.CV / 281 / 2606.23679
Semantic Browsing: Controllable Diversity for Image Generation
语义浏览：可控多样性的图像生成
Abstract
Modern text-to-image models excel in visual fidelity and prompt adherence. However, this strict adherence comes at the cost of diversity: generated samples tend to collapse into a single visual interpretation. Existing methods to improve diversity produce outputs driven by incidental variations rather than meaningful design choices. This motivates a new variant of the diversity task where structure is enforced on the generated samples. We introduce a method for controlled diversity that enables Semantic Browsing, where users can navigate structured image galleries and experience creative exploration through a systematic traversal of meaningful, interpretable axes of variation. Achieving this level of semantic control requires a deep understanding of the scene. We exploit the fact that recent text-to-image models are trained on elaborated captions, effectively decoupling semantic decision-making from pixel generation. This enables a paradigm shift: instead of relying on stochastic variation within the text-to-image model, we induce diversity directly at the text level. By leveraging rich textual representations, we allow a Vision Language Model (VLM) to operate on the full scene context. To overcome the generic outputs typical of standard VLMs, we employ an agentic workflow that explicitly enforces structured variation attuned to the original prompt. We demonstrate that our method produces diverse and navigable design spaces where every variation corresponds to a specific, user-understandable semantic decision.
Chinese Translation
现代文本到图像模型在视觉保真度和提示遵循方面表现出色。然而，这种严格的遵循以牺牲多样性为代价：生成的样本往往趋向于单一的视觉解读。现有的改善多样性的方法产生的输出往往受到偶然变化的驱动，而非有意义的设计选择。这促使我们提出一种新的多样性任务变体，在生成样本上强加结构。我们引入了一种可控多样性的方法，使得语义浏览成为可能，用户可以在结构化的图像库中导航，并通过系统性地遍历有意义且可解释的变化轴来体验创造性探索。实现这种语义控制水平需要对场景的深入理解。我们利用了最近的文本到图像模型是基于详尽的标题进行训练的这一事实，有效地将语义决策与像素生成解耦。这使得范式转变成为可能：我们不再依赖文本到图像模型中的随机变化，而是直接在文本层面引入多样性。通过利用丰富的文本表示，我们允许视觉语言模型（Vision Language Model, VLM）在完整的场景上下文中操作。为了克服标准VLM典型的通用输出，我们采用了一种代理工作流程，明确强加与原始提示相适应的结构化变化。我们证明了我们的方法能够生成多样化且可导航的设计空间，其中每种变化都对应于特定的、用户可理解的语义决策。
cs.CV / 282 / 2606.23682
Keep The Essentials: Efficient Reference Conditioned Generation via Token Dropping
保持本质：通过令牌丢弃实现高效的参考条件生成
Abstract
Reference-based diffusion models enable highly controllable image generation by leveraging elements from input images to guide prompt-driven synthesis. However, these models are computationally expensive in runtime, and their cost scales severely with the number of input references. While the efficiency of diffusion models has been extensively studied in the context of prompt-driven generation, it remains largely under-explored in the realm of reference-based models. This setting presents unique challenges not addressed by methods focusing solely on generation. In particular, the wasteful representation of references as dense token grids offers significant opportunities for improvement. In this work, we present Sparse Context, a method for constructing sparse reference representations by retaining only a reduced subset of reference tokens. We observe that even without modifying the model, dropping a significant portion of reference tokens at inference time largely preserves its generation capabilities. To fully realize this potential, we fine-tune the model with random token dropping at varying ratios, encouraging robustness to partial reference representations. Crucially, this training strategy decouples the model from any specific token selection rule, allowing flexible control at inference time. At inference time, instead of random dropping, we apply task-aware token selection strategies that prioritize the most informative regions of the reference images, adapting the token budget to the input and task requirements. Extensive experiments show our method achieves a 4x increase in inference speed for multi-reference generation and an 2x for single reference generation. Importantly, this efficiency is achieved without compromising visual quality across both spatially-aligned editing and subject-driven generation.
Chinese Translation
基于参考的扩散模型通过利用输入图像中的元素来指导基于提示的合成，从而实现高度可控的图像生成。然而，这些模型在运行时计算成本高昂，并且其成本随着输入参考数量的增加而严重增加。尽管扩散模型在基于提示的生成方面的效率得到了广泛研究，但在基于参考的模型领域仍然很少被探讨。这种设置提出了独特的挑战，而这些挑战并未被仅专注于生成的方法所解决。特别是，将参考表示为密集令牌网格的浪费表示提供了显著的改进机会。在本研究中，我们提出了稀疏上下文（Sparse Context），一种通过仅保留减少的参考令牌子集来构建稀疏参考表示的方法。我们观察到，即使不修改模型，在推理时丢弃大量参考令牌也能在很大程度上保留其生成能力。为了充分实现这一潜力，我们通过在不同比例下随机丢弃令牌对模型进行微调，鼓励其对部分参考表示的鲁棒性。关键是，这种训练策略将模型与任何特定的令牌选择规则解耦，从而允许在推理时灵活控制。在推理时，我们不再随机丢弃，而是应用任务感知的令牌选择策略，优先考虑参考图像中最具信息性的区域，根据输入和任务要求调整令牌预算。大量实验表明，我们的方法在多参考生成中实现了推理速度提高4倍，在单参考生成中提高2倍。重要的是，这种效率是在不影响视觉质量的情况下实现的，适用于空间对齐编辑和基于主题的生成。
cs.CV / 283 / 2606.23688
Lift4D: Harmonizing Single-View 3D Estimation for 4D Reconstruction In-the-Wild
Lift4D：在野外环境中协调单视图3D估计以实现4D重建
Abstract
Reconstructing dynamic non-rigid objects from monocular video requires integrating visual cues from direct observations with data-driven priors over geometry and appearance. Prior approaches either learn to directly predict 4D representations from visual input or initialize a 3D representation that is subsequently deformed and refined based on video evidence. However, the former are constrained by the scarcity of 4D training data, while the latter leverage priors only for the initial reconstruction and rely solely on video supervision thereafter; neither handles complex in-the-wild scenarios with large deformations and occlusions well. We present Lift4D, a test-time optimization framework that addresses both limitations. First, we adapt an existing single-view 3D reconstruction model to yield temporally consistent per-frame predictions via causal latent conditioning, providing a coherent initialization for a deformable 3D Gaussian Splatting representation. We then ``sculpt'' this representation to match the input video through an occlusion-aware optimization that faithfully recovers visible surface details while completing unobserved regions using a view-conditioned diffusion prior. We demonstrate that Lift4D clearly improves over prior 4D reconstruction methods, particularly on challenging in-the-wild sequences with severe occlusions and non-rigid motion.
Chinese Translation
从单目视频重建动态非刚性物体需要将直接观察的视觉线索与基于数据的几何和外观先验相结合。先前的方法要么直接学习从视觉输入预测4D表示，要么初始化一个3D表示，然后根据视频证据进行变形和细化。然而，前者受到4D训练数据稀缺的限制，而后者仅在初始重建时利用先验，之后完全依赖视频监督；两者都无法很好地处理具有大变形和遮挡的复杂野外场景。我们提出了Lift4D，这是一种在测试时优化的框架，解决了这两种限制。首先，我们调整现有的单视图3D重建模型，通过因果潜在条件生成时间上连续的逐帧预测，为可变形的3D高斯点云表示提供一致的初始化。然后，我们通过一种考虑遮挡的优化“雕刻”该表示，以匹配输入视频，忠实地恢复可见表面细节，同时使用视图条件扩散先验填补未观察到的区域。我们证明Lift4D明显优于先前的4D重建方法，特别是在具有严重遮挡和非刚性运动的挑战性野外序列上。
cs.AI / 1 / 2606.20569
On the Identifiability of User Adaptation in Co-Adaptive Neural Interfaces
用户适应性在共适应神经接口中的可识别性研究
Abstract
We analyze identifiability in co-adaptive human-machine systems. We show that closed-loop encoder estimates do not uniquely identify user adaptation, but instead reflect properties of the joint system. We discuss implications for interpreting behavioral adaptation and propose conditions for identification.
Chinese Translation
我们分析了共适应人机系统中的可识别性。我们表明，闭环编码器的估计并不能唯一识别用户的适应性，而是反映了联合系统的特性。我们讨论了这一点对行为适应性解释的影响，并提出了识别的条件。
cs.AI / 2 / 2606.20599
Beyond Fixed Budgets: Characterizing the Inelasticity and Limitations of Tree-of-Thought Reasoning Strategies
超越固定预算：树状思维推理策略的非弹性特征及其局限性
Abstract
Tree of Thought (ToT) search has become a promising direction for improving the reasoning capabilities of large language models, but deploying these methods in practice raises a question that has received little systematic attention: how do different search strategies behave under varying compute budgets, model sizes, and problem difficulties? In this work, we evaluate two representative ToT methods; DPTS, a Monte Carlo tree search based approach, and SSDP, a semantic deduplication based approach, across two mathematical reasoning benchmarks (Math500 and GSM8K), two model scales (Llama-3B and Llama-8B), and four token budgets (3k--10k). Our analysis reveals that the two methods exhibit limitations that pull in opposite directions. DPTS suffers from a cold-start bottleneck at low budgets: it requires sufficient exploration before its value estimates become reliable, making it a poor fit for resource-constrained settings despite strong scaling behavior at higher budgets. SSDP, on the other hand, reaches candidate solutions efficiently but is prone to frontier depletion; its aggressive node merging permanently discards unexplored paths, leaving it unable to improve regardless of how much budget remains. Together, these findings suggest that neither a fixed exploration strategy nor a fixed pruning strategy is sufficient across compute continuum. We argue that effective search for scientific reasoning agents requires strategies that can adapt their behavior based on search progress and available resources.
Chinese Translation
树状思维（Tree of Thought, ToT）搜索已成为提升大型语言模型推理能力的一个有前景的方向，但在实践中部署这些方法提出了一个鲜有系统关注的问题：不同的搜索策略在不同的计算预算、模型规模和问题难度下表现如何？在本研究中，我们评估了两种具有代表性的ToT方法：基于蒙特卡洛树搜索的DPTS和基于语义去重的SSDP，分别在两个数学推理基准（Math500和GSM8K）、两个模型规模（Llama-3B和Llama-8B）以及四个令牌预算（3k--10k）下进行分析。我们的分析揭示了这两种方法在表现上存在相互矛盾的局限性。DPTS在低预算下遭遇冷启动瓶颈：它需要足够的探索才能使其价值估计变得可靠，因此在资源受限的环境中表现不佳，尽管在较高预算下具有良好的扩展性。另一方面，SSDP能够高效地达到候选解决方案，但容易出现前沿耗竭；其激进的节点合并永久性地丢弃未探索的路径，使其无论剩余预算多少都无法改进。综合这些发现，我们认为在计算连续体中，既定的探索策略或固定的剪枝策略均不足以应对科学推理代理的有效搜索。我们认为，科学推理代理的有效搜索需要能够根据搜索进展和可用资源调整其行为的策略。
cs.AI / 3 / 2606.20600
The New Associationism: Lessons from Deep Learning
新的联结主义：来自深度学习的启示
Abstract
What can the success of modern AI tell us about how humans learn? This paper argues that taking AI seriously as a model of human learning supports a modest but genuine associationism. The central finding is that supervised learning -- learning driven by evaluative feedback -- underlies a surprisingly wide range of contemporary AI systems, from large language models to game-playing agents, differing primarily in how much work is required to generate the relevant feedback signal. This vindicates associationist ideals of a uniform, gradual, error-driven learning mechanism operating across domains, and defuses the once-influential argument that associationist mechanisms are too limited to account for human cognitive capacities. At the same time, the successes of deep learning depend on computational architectures that go well beyond anything classical associationists envisaged, and supervised learning operates within these as one component rather than a complete account of learning.
Chinese Translation
现代人工智能的成功能告诉我们关于人类学习的什么？本文认为，将人工智能视为人类学习模型的认真态度支持了一种适度但真实的联结主义。中心发现是，监督学习——由评估反馈驱动的学习——在当代人工智能系统中起着基础性作用，这些系统从大型语言模型到游戏代理不等，主要的区别在于生成相关反馈信号所需的工作量。这证实了联结主义理想，即在不同领域中存在一种统一、渐进、以错误驱动的学习机制，同时也化解了曾经影响深远的论点，即联结主义机制过于有限，无法解释人类的认知能力。同时，深度学习的成功依赖于超越经典联结主义者所设想的计算架构，而监督学习在这些架构中作为一个组成部分运作，而不是学习的完整解释。
cs.AI / 4 / 2606.20615
Specifying AI-SDLC Processes: A Protocol Language for Human-Agent Boundaries
规范人工智能软件开发生命周期（AI-SDLC）过程：一种用于人机边界的协议语言
Abstract
AI agents now participate as first-class team members across the software development lifecycle, yet no specification language exists for expressing the human-agent responsibility boundaries, approval gates, and governance constraints this collaboration requires. Existing approaches encode process in agent prompts (subject to drift), target adjacent domains (workflow management, business processes), or address only fragments (access control, approval gates). We propose a domain-specific language for specifying AI-SDLC processes as protocols, with formal syntax, well-formedness conditions, operational semantics, and enforcement invariants. The language distinguishes policy (declared intent) from mechanism (structural enforcement), enabling implementations to bound process non-determinism through primitives such as validation tokens and capability boundaries. Three results follow. A failure rate analysis shows that structural enforcement bounds system failure rates at a weighted product of agent and validator rates, while behavioral compliance permits cumulative or near-saturating growth. The 2+N team pattern (two human-in-control roles plus N specialized agent members) formalizes classical Separation of Duties for AI-SDLC. Kleene closure of orchestration loops and reflexive protocol-adherence validation emerge as design properties rather than special-case constructs. We position the contribution against multi-agent frameworks (MetaGPT), workflow specification (FlowAgent, BPMN extensions), and capability-based security (SAGA): the novelty lies in the specific integration, not any single primitive. A working implementation demonstrates feasibility; empirical evaluation is future work.
Chinese Translation
人工智能代理现在作为软件开发生命周期中的一流团队成员参与其中，但目前尚不存在一种规范语言来表达这种协作所需的人机责任边界、审批关口和治理约束。现有的方法将过程编码在代理提示中（易受漂移影响），针对相邻领域（工作流管理、业务流程），或仅处理片段（访问控制、审批关口）。我们提出了一种领域特定语言，用于将AI-SDLC过程指定为协议，具备正式语法、良构性条件、操作语义和强制不变性。该语言区分政策（声明的意图）与机制（结构性强制），使得实现能够通过验证令牌和能力边界等原语来限制过程的非确定性。结果有三点：失败率分析表明，结构性强制将系统失败率限制在代理和验证者比率的加权乘积下，而行为合规则允许累积或近饱和增长。2+N团队模式（两个控制角色加上N个专业代理成员）形式化了AI-SDLC的经典职责分离。编排循环的克莱尼闭包和自反协议遵循验证作为设计属性而非特例构造出现。我们将这一贡献与多代理框架（MetaGPT）、工作流规范（FlowAgent、BPMN扩展）和基于能力的安全（SAGA）进行对比：其新颖性在于特定的整合，而非任何单一原语。一个工作实现展示了可行性；实证评估为未来工作。
cs.AI / 5 / 2606.20621
PEAR: Permutation-Equivariant Adaptive Routing Multi-Agent Debate
PEAR：置换等变自适应路由多智能体辩论
Abstract
Multi-agent debate improves the reliability of large language models (LLMs) through iterative peer critiques. However, fixed topologies often introduce persistent positional biases, amplify unreliable agents, and cause high sensitivity to role assignments. We introduce \textit{Permutation-Equivariant Adaptive Routing Multi-Agent Debate (PEAR)}, an inference-time protocol that dynamically reconfigures communication roles and sparse topologies across consecutive debate rounds. By strategically switching agent-to-role assignments based on evolving agent states, PEAR prevents any agent from permanently occupying a privileged network position or distributes influence more evenly across the debate. We theoretically characterize PEAR as an equivariant sparse router: it preserves accuracy under agent relabeling while reducing routing complexity and improving generalization. Comprehensive empirical evaluations across four reasoning benchmarks and six diverse LLM backbones demonstrate PEAR significantly improves average accuracy over the strongest debate baselines. The code is at https://github.com/EVIEHub/PEAR.
Chinese Translation
多智能体辩论通过迭代的同行批评提高了大型语言模型（LLMs）的可靠性。然而，固定的拓扑结构往往引入持续的位置信息偏差，放大不可靠的智能体，并导致对角色分配的高度敏感性。我们提出了 extit{置换等变自适应路由多智能体辩论（PEAR）}，这是一种在推理时动态重新配置通信角色和稀疏拓扑的协议，适用于连续的辩论轮次。通过根据不断变化的智能体状态战略性地切换智能体与角色的分配，PEAR防止任何智能体永久占据特权网络位置，或在辩论中更均匀地分配影响力。我们理论上将PEAR表征为一种等变稀疏路由器：它在智能体重新标记时保持准确性，同时降低路由复杂性并提高泛化能力。在四个推理基准和六个不同的LLM基础模型上的全面实证评估表明，PEAR显著提高了平均准确性，超越了最强的辩论基线。代码可在 https://github.com/EVIEHub/PEAR 获取。
cs.AI / 6 / 2606.20622
Darwin Mobile Agent: A Roadmap for Self-Evolution
达尔文移动代理：自我进化的路线图
Abstract
The goal of artificial intelligence is to create agents capable of general, adaptive behaviour in open-ended environments. Guided by the "Bitter Lesson", we argue that the most effective path toward this goal is to systematically remove human priors and allow intelligence to naturally emerge through interaction with a "Big World" that is orders of magnitude more complex than the agent itself. We propose the mobile Graphical User Interface (GUI) as a practical proxy for such a world and introduce Darwin Mobile Agent, an open-source infrastructure designed as a foundation for autonomous reinforcement learning in this domain. This framework addresses the data-collection bottleneck in real-world mobile interactions by using an asynchronous agent-environment loop across parallel cloud-phone instances. We further propose a conceptual roadmap to systematically remove human priors from three fundamental pillars of a self-evolving agent: task curricula, outcome verification, and memory management. We validate that the Darwin infrastructure provides the stability and scalability required for the first stage of this roadmap: policy optimisation in the GUI domain. This work establishes the practical and theoretical foundation necessary to move toward truly autonomous, self-evolving GUI agents.
Chinese Translation
人工智能的目标是创造能够在开放环境中表现出一般适应性行为的代理。受“苦涩教训”的指导，我们认为实现这一目标的最有效途径是系统性地消除人类先验知识，让智能通过与一个比代理自身复杂几个数量级的“宏大世界”的互动自然涌现。我们提出移动图形用户界面（GUI）作为这样一个世界的实用代理，并介绍达尔文移动代理（Darwin Mobile Agent），这是一个旨在为该领域的自主强化学习提供基础的开源基础设施。该框架通过在并行云手机实例之间使用异步代理-环境循环，解决了现实世界移动交互中的数据收集瓶颈。我们进一步提出了一个概念性路线图，系统性地从自我进化代理的三个基本支柱中消除人类先验知识：任务课程、结果验证和记忆管理。我们验证了达尔文基础设施提供了实现这一路线图第一阶段所需的稳定性和可扩展性：在GUI领域的策略优化。这项工作建立了向真正自主、自我进化的GUI代理迈进所需的实践和理论基础。
cs.AI / 7 / 2606.20623
Path-dependent program induction under resource constraints explains human sequence learning
资源限制下的路径依赖程序归纳解释人类序列学习
Abstract
How do people build abstract, reusable knowledge from sequential experience under bounded cognitive resources? To answer this question, we integrate rate-distortion theory with recent advances in program induction to describe how prior knowledge shapes which future structures are cheap to encode and easy to discover. We formalize this in a hierarchical Adaptor Grammar (HAG) with distinct local (within-task) and global (across-task) libraries, governed jointly by constraints on memory and computation. In simulations, HAG achieves better rate-distortion trade-offs and stronger generalization than fixed grammars or shallow chunking methods. In an online melodic sequence-learning experiment, participants' recall errors reflected systematic simplifications and reaction times increased at inferred program boundaries. Trial-by-trial fits further showed that hierarchical libraries best explained individual differences in both recall and out-of-sample continuation choices, outperforming all alternative models. These findings cast structured learning as bounded program induction in which the order of experience shapes future abstractions a learner builds.
Chinese Translation
人们如何在有限的认知资源下从顺序经验中构建抽象的、可重用的知识？为了解答这个问题，我们将速率失真理论与程序归纳的最新进展相结合，描述先前知识如何影响未来结构的编码成本和发现的难易程度。我们在一个具有不同局部（任务内）和全局（任务间）库的层次适配语法（HAG）中形式化这一过程，该语法共同受到记忆和计算的约束。在模拟中，HAG实现了比固定语法或浅层分块方法更好的速率失真权衡和更强的泛化能力。在一项在线旋律序列学习实验中，参与者的回忆错误反映了系统性的简化，反应时间在推断的程序边界处增加。逐次试验的拟合进一步表明，层次库最佳地解释了个体在回忆和样本外延续选择中的差异，超越了所有替代模型。这些发现将结构化学习视为受限的程序归纳，其中经验的顺序塑造了学习者构建未来抽象的方式。
cs.AI / 8 / 2606.20624
In LLM Reasoning, there is Irrationality on top of Value Misalignment
在大语言模型推理中，存在超越价值不一致的非理性
Abstract
Significant progress has been made in aligning LLMs with target value functions. We argue that, even when an LLM has been well aligned in (post-)training, it may still fail to maximise the aligned value in reasoning. We mathematically formalise this gap as rational value risk: the utility discrepancy between a model's deployed reasoning strategy and its rational counterpart, which is defined to be the responses that maximise expected utility in the steepest direction. The estimation error of rational value risk is further decomposed into three components from finite candidates, finite prompts, and imperfect verifiers. Extensive experiments are conducted, covering models Llama-3.1, Qwen-2.5, T{\"}ulu-3 families (7B-72B), GPT-5.2, GPT-5.5, and DeepSeek-V4, and benchmarks UltraFeedback, AlpacaEval, GSM8K, MATH, HumanEval, and MathArena. The results validate that (1) rational value risk is widespread; (2) value alignment can reduce, but cannot eliminate, it; (3) the risk is highly sensitive to inference-time reasoning strategy; and (4) longer reasoning improves rationality with diminishing returns. The code is at https://github.com/EVIEHub/LLM-Rationality.
Chinese Translation
在将大语言模型（LLMs）与目标价值函数对齐方面已经取得了显著进展。我们认为，即使在（后）训练中一个大语言模型已经很好地对齐，它在推理时仍可能无法最大化对齐价值。我们将这一差距数学形式化为理性价值风险：模型部署的推理策略与其理性对应策略之间的效用差异，后者被定义为在最陡峭方向上最大化期望效用的响应。理性价值风险的估计误差进一步分解为来自有限候选、有限提示和不完美验证者的三个组成部分。我们进行了广泛的实验，涵盖了Llama-3.1、Qwen-2.5、T{"}ulu-3系列（7B-72B）、GPT-5.2、GPT-5.5和DeepSeek-V4模型，以及UltraFeedback、AlpacaEval、GSM8K、MATH、HumanEval和MathArena基准。结果验证了以下几点：（1）理性价值风险是普遍存在的；（2）价值对齐可以减少，但无法消除它；（3）该风险对推理时的推理策略高度敏感；（4）更长的推理提高了理性，但收益递减。代码可在 https://github.com/EVIEHub/LLM-Rationality 获取。
cs.AI / 9 / 2606.20625
AlphaMemo: Structured Search-Process Memory for Self-Evolving Alpha Mining Agents
AlphaMemo：自我进化的阿尔法挖掘智能体的结构化搜索过程记忆
Abstract
LLM agents are promising for alpha mining via combining financial priors, symbolic reasoning, executable factor generation, and feedback-driven refinement. Yet, they face a combinatorial search space, noisy non-stationary feedback, redundant discoveries, and overfitting risks from naively reusing past successes. To address these challenges, we propose AlphaMemo, a self-evolving alpha mining agent with Structured Search-Process Memory. Rather than memorizing only final factors or full trajectories, AlphaMemo records reusable evidence about which edit motifs work or fail under specific parent-factor contexts. It extracts motifs from Abstract Syntax Tree (AST) differences, applies confidence-gated residual memory on top of a search-ledger prior, and uses asymmetric veto control to suppress high-confidence failure patterns. Experiments on CSI 500 and S\&P 500 show improved out-of-sample performance and fixed-budget discovery efficiency, with ablations validating the roles of residual learning, confidence gating, AST-diff motifs, and veto memory. Code is at https://github.com/jarrettyu/AlphaMemo.
Chinese Translation
大型语言模型（LLM）智能体通过结合金融先验、符号推理、可执行因子生成和基于反馈的精炼，为阿尔法挖掘提供了良好的前景。然而，它们面临组合搜索空间、噪声非平稳反馈、冗余发现以及因简单重复过去成功而导致的过拟合风险。为了解决这些挑战，我们提出了AlphaMemo，一种具有结构化搜索过程记忆的自我进化阿尔法挖掘智能体。AlphaMemo不仅仅记忆最终因子或完整轨迹，而是记录关于在特定父因子上下文中哪些编辑模式有效或失败的可重用证据。它从抽象语法树（Abstract Syntax Tree, AST）差异中提取模式，在搜索账本先验的基础上应用置信度门控残差记忆，并使用不对称否决控制来抑制高置信度失败模式。在CSI 500和标准普尔500（S&P 500）上的实验显示出改进的样本外表现和固定预算发现效率，消融实验验证了残差学习、置信度门控、AST差异模式和否决记忆的作用。代码可在 https://github.com/jarrettyu/AlphaMemo 获取。
cs.AI / 10 / 2606.20627
Latent Goal Prediction from Language for Model-Based Planning
基于语言的潜在目标预测用于模型驱动规划
Abstract
Planning with world models is bottlenecked by compounding prediction errors and the difficulty of defining optimizable goals. Visual targets provide precise local gradients but poor distant guidance, while language is flexible yet limited by noisy cross-modal alignment or dependence on large generative models unsuited for the high-sampling nature of model-based planning. To address these challenges, we introduce Latent Goal Prediction from Language (LAGO), a framework that predicts both sequences of intermediate goal states from language instructions and action-conditioned rollouts, all within the same latent space. Rather than optimizing toward a single global objective, LAGO dynamically decomposes instructions into explicitly predicted, locally tractable latent subgoals. By updating these subgoals online and using a soft minimum trajectory cost during planning, LAGO enables an agent to follow coherent latent trajectories over long horizons. Evaluation across multiple environments planning horizons shows that LAGO avoids the sharp degradation of prior methods. By achieving robust and precise long-horizon planning purely from language, LAGO bridges the precision of visual goals with the flexibility of text-guided control.
Chinese Translation
使用世界模型进行规划受到累积预测误差和定义可优化目标的困难的制约。视觉目标提供了精确的局部梯度，但在远距离指导方面表现不佳，而语言则灵活但受到噪声跨模态对齐或依赖于不适合于模型驱动规划高采样特性的庞大生成模型的限制。为了解决这些挑战，我们提出了基于语言的潜在目标预测（Latent Goal Prediction from Language, LAGO）框架，该框架在同一潜在空间内预测来自语言指令的中间目标状态序列和基于动作的展开。LAGO并不是朝着单一全局目标进行优化，而是动态地将指令分解为明确预测的、局部可处理的潜在子目标。通过在线更新这些子目标并在规划过程中使用软最小轨迹成本，LAGO使得代理能够在较长的时间范围内遵循一致的潜在轨迹。在多个环境规划时间范围的评估中，LAGO避免了先前方法的急剧退化。通过仅依靠语言实现稳健且精确的长时间规划，LAGO将视觉目标的精确性与文本引导控制的灵活性结合起来。
cs.AI / 11 / 2606.20628
Human Decision-Making with AI Assistance under Correlated Features
在相关特征下，人工智能辅助的人类决策
Abstract
Humans increasingly make decisions with AI assistance; for example, doctors may follow AI-recommended diagnostic tests and base their diagnoses on the results. A natural question is which tests should AI recommend to balance short-term decision quality and long-term human learning when different features (e.g., test results) are correlated. While prior work establishes that stationary policies that recommend the same tests repeatedly are optimal when features are independent, we prove that feature correlations lead such policies to perform arbitrarily poorly. Instead, we prove that any optimal policy must follow an explore-then-commit structure; initially, the AI should offer diverse tests so humans can learn accurate feature coefficients, then the AI should commit to a single set of tests, with exploration length that depends on the degree of feature correlation. We prove that computing the optimal policy is NP-hard and derive a dynamic programming-based algorithm that finds the optimal policy for finite horizons. We additionally develop an approximation that plans for shorter horizons and appends a stationary suffix, achieving near-optimal performance. Our empirical results complement our theory by showing that stronger feature correlation leads to longer exploration phases.
Chinese Translation
人类越来越多地在人工智能（AI）的辅助下做出决策；例如，医生可能会遵循AI推荐的诊断测试，并根据结果进行诊断。一个自然的问题是，在不同特征（例如，测试结果）相关的情况下，AI应该推荐哪些测试，以平衡短期决策质量和长期人类学习。虽然先前的研究表明，当特征独立时，推荐相同测试的静态策略是最优的，但我们证明特征相关性会导致此类策略表现极差。相反，我们证明任何最优策略必须遵循探索-再承诺的结构；最初，AI应该提供多样化的测试，以便人类能够学习准确的特征系数，然后AI应承诺使用一组固定的测试，其探索长度取决于特征相关性的程度。我们证明计算最优策略是NP难的，并推导出一种基于动态规划的算法，能够找到有限时间范围内的最优策略。此外，我们还开发了一种近似方法，计划较短的时间范围并附加一个静态后缀，从而实现接近最优的性能。我们的实证结果通过显示更强的特征相关性导致更长的探索阶段，补充了我们的理论。
cs.AI / 12 / 2606.20631
Harnessing Agent Skills: Architectural Patterns and a Reference Architecture for Skill-Mediated LLM Agents
利用代理技能：技能中介大型语言模型代理的架构模式与参考架构
Abstract
Agent skills externalise reusable agent-facing behavioural knowledge and guidance as persistent artefacts that can be discovered, activated, and interpreted by LLM agents. Although a skill artefact is static at rest, its architectural responsibilities arise in use, when the artefact is selected for a run, bound to context and authority constraints, interpreted by a stochastic agent, and recorded as run evidence. We call this run-specific relation skill-in-use. This paper studies agent skill harnessing: the architectural responsibilities that govern the transition from skill artefacts to skill-in-use, bound the executable consequences associated with skill-in-use, and capture evidence for attribution, verification, repair, and evolution. This paper provides a catalogue of ten empirically grounded architectural patterns (five core, five supporting) for skill harnessing and synthesises them into a reference architecture with four responsibility layers: Supply Chain, Mediation, Execution Control, and Evidence & Feedback. We evaluate the architecture through cross-instantiation across 8 selected systems. The resulting patterns and reference architecture provide a vocabulary and diagnostic frame for analysing skill-harnessing responsibilities across agent systems.
Chinese Translation
代理技能将可重用的面向代理的行为知识和指导外化为可被大型语言模型（LLM）代理发现、激活和解释的持久性工件。尽管技能工件在静止状态下是静态的，但其架构责任在使用时产生，当工件被选择执行、绑定到上下文和权威约束、由随机代理解释并记录为执行证据时。我们称这种特定于运行的关系为技能在用（skill-in-use）。本文研究代理技能的利用：管理从技能工件到技能在用的过渡的架构责任，约束与技能在用相关的可执行后果，并捕获归属、验证、修复和演化的证据。本文提供了十种基于实证的技能利用架构模式的目录（五种核心模式，五种支持模式），并将其综合为一个具有四个责任层的参考架构：供应链、调解、执行控制和证据与反馈。我们通过在8个选定系统中的跨实例化评估该架构。最终形成的模式和参考架构为分析代理系统中的技能利用责任提供了词汇和诊断框架。
cs.AI / 13 / 2606.20634
DEMM-Bench: A Cross-Regime Benchmark for Agent-Runtime Governance-Evidence Sufficiency
DEMM-Bench：一种跨领域的代理运行时治理证据充分性基准
Abstract
Agent-runtime systems emit traces, ledgers, provenance graphs, policy logs, delegation tokens, cache events, and tool-firewall records, but those containers do not necessarily answer governance questions about a specific decision. DEMM-Bench is a cross-regime benchmark for agent-runtime governance-evidence sufficiency, grounded in the Decision Evidence Maturity Model (DEMM): it measures whether records across eight evidence regimes are sufficient to reconstruct decision-level properties rather than merely present. The benchmark normalizes the regimes through adapters, asks property questions over actor, authority, action, policy, decision basis, resource touch, lifecycle context, and verification strength, and applies eight deterministic degradation conditions. Across 64 manuscript cases, trace-present and schema-present baselines overclaim on 75% of cases, ledger-present overclaims on 50%, and the redacted property-level candidate scorer has zero overclaim with 56.25% mean Property Sufficiency Accuracy. The deposited package provides the 64-case dataset, construction-oracle labels, baselines, and adapters, supporting reproducible evaluation of decision-evidence maturity across heterogeneous agent-runtime evidence substrates.
Chinese Translation
代理运行时系统生成的痕迹、账本、来源图、政策日志、委托令牌、缓存事件和工具防火墙记录，并不一定能够回答关于特定决策的治理问题。DEMM-Bench 是一种基于决策证据成熟度模型（Decision Evidence Maturity Model, DEMM）的跨领域代理运行时治理证据充分性基准：它衡量八个证据领域中的记录是否足以重构决策级属性，而不仅仅是呈现。该基准通过适配器规范化各领域，针对参与者、权威、行动、政策、决策依据、资源接触、生命周期上下文和验证强度提出属性问题，并应用八个确定性降级条件。在64个案例中，基于痕迹和模式的基线在75%的案例中存在过度声称，账本存在的过度声称为50%，而经过修订的属性级候选评分器没有过度声称，平均属性充分性准确率为56.25%。所提供的包包含64个案例的数据集、构建神谕标签、基线和适配器，支持在异构代理运行时证据基础上可重复的决策证据成熟度评估。
cs.AI / 14 / 2606.20636
SkillHarness: Harnessing Safe Skills for Computer-Use Agents
SkillHarness：为计算机使用代理利用安全技能
Abstract
Computer-Use Agents (CUAs) are increasingly deployed in dynamic interactive environments, creating a growing need for continual skill learning during interaction. Recent approaches address this challenge by learning reusable skills from successful trajectories. However, these skill learning methods largely assume static and safe environments, overlooking risks from adversarial interactions (e.g., prompt injections) and environmental dynamics (e.g., pop-ups). In dynamic settings, such assumptions can lead to risky skill learning and brittle execution, undermining the reliability of CUAs. This raises the question: how can CUAs learn and use skills safely in dynamic environments? To address this problem, we propose SkillHarness, a framework for safe skill harnessing in dynamic environments. SkillHarness moves beyond static skill abstractions by modeling skill learning and utilization as a safety-constrained interaction process. Specifically, we introduce the skill boundary that leverages multi-source supervision signals to identify safe skills from interaction trajectories, and construct self-improving safety constraints throughout the skill lifecycle. In addition, SkillHarness introduces selective skill reuse, where tasks are guided to decompose according to context and completed through the selective activation of skill subsets. Our experiments demonstrate that SkillHarness significantly reduces the unsafe rate of learned skills by 57.1% and consistently improves execution stability under dynamic environmental changes, outperforming existing baselines.
Chinese Translation
计算机使用代理（CUAs）越来越多地部署在动态交互环境中，导致在交互过程中对持续技能学习的需求不断增加。近期的方法通过从成功轨迹中学习可重用技能来应对这一挑战。然而，这些技能学习方法在很大程度上假设环境是静态和安全的，忽视了来自对抗性交互（例如，提示注入）和环境动态（例如，弹出窗口）的风险。在动态环境中，这种假设可能导致风险技能学习和脆弱执行，从而削弱CUAs的可靠性。这引发了一个问题：CUAs如何在动态环境中安全地学习和使用技能？为了解决这个问题，我们提出了SkillHarness，一个用于在动态环境中安全利用技能的框架。SkillHarness超越了静态技能抽象，通过将技能学习和利用建模为一个安全约束的交互过程。具体而言，我们引入了技能边界，利用多源监督信号从交互轨迹中识别安全技能，并在技能生命周期中构建自我改善的安全约束。此外，SkillHarness引入了选择性技能重用，其中任务根据上下文进行分解，并通过选择性激活技能子集来完成。我们的实验表明，SkillHarness显著降低了学习技能的不安全率，减少了57.1%，并在动态环境变化下持续改善执行稳定性，超越了现有基准。
cs.AI / 15 / 2606.20637
Constituency Optimisation Through Hamiltonian Representation Of Mandates (COTHROM): Algorithmic Redistricting of Irish Election Boundaries
通过哈密顿表示法优化选区（COTHROM）：爱尔兰选举边界的算法重划
Abstract
Electoral redistricting in Ireland's Proportional Representation Single Transferable Vote (PR-STV) system faces the challenge of selecting an optimally representative set of electoral boundaries from an enormous set of possible configurations, and where ``representative'' is a delicate balance of constitutional objectives that are often in tension with one another. We present the first computational framework for Irish electoral redistricting that systematically optimises across multiple constitutional requirements while making trade-offs explicit and quantifiable. The electoral redistricting problem is parsed using statistical physics, where constitutional objectives are considered as terms in a Potts Hamiltonian. Markov Chain Monte Carlo (MCMC) methods and simulated annealing are employed to minimise this objective function, systematically exploring this configuration space, with coupling constants as proxies for objective weightings. Multi Criterion Decision Analysis (MCDA) and Pareto Optimality is then utilised to remedy the ambiguity in choosing a certain objective weighting combination over others. With respect to proportional representation and compactness objectives evaluated in County Cork, COTHROM consistently improves on the existing legal constituency boundaries for a range of objective weightings.
Chinese Translation
在爱尔兰的比例代表制单一可转移投票（PR-STV）系统中，选区重划面临着从大量可能配置中选择一个最佳代表性选区边界的挑战，而“代表性”是一个常常相互矛盾的宪法目标之间的微妙平衡。我们提出了第一个爱尔兰选举重划的计算框架，该框架系统地在多个宪法要求之间进行优化，同时使权衡变得明确和可量化。选区重划问题采用统计物理进行解析，其中宪法目标被视为Potts哈密顿量中的项。我们采用马尔可夫链蒙特卡洛（MCMC）方法和模拟退火来最小化这一目标函数，系统地探索这一配置空间，耦合常数作为目标权重的代理。然后利用多标准决策分析（MCDA）和帕累托最优性来解决在选择某一目标权重组合时的模糊性。针对在科克郡评估的比例代表性和紧凑性目标，COTHROM在一系列目标权重下始终优于现有的法律选区边界。
cs.AI / 16 / 2606.20638
RIZZ: Routing Interactions to Near Zero-Interference Zones for Continual Adaptation of Black-Box Agents
RIZZ：将交互路由到近零干扰区以持续适应黑箱代理
Abstract
Large language models are increasingly deployed as long-lived agents that must adapt across users, tasks, domains, modalities, and feedback regimes without access to model weights. Existing black-box adaptation methods typically optimize a single prompt, maintain an undifferentiated memory, or rely on repeated rollout-heavy search. However, these designs struggle when streams of input are nonstationary, feedback is sparse, and failures from one task family can contaminate behavior on another. We introduce RIZZ (Routing Interactions to Near Zero-interference Zones), a continual adaptation framework for compound language-model systems that learns entirely through verifier-gated memory, routing, and prompt compilation. RIZZ organizes input streams into dynamically spawned memory branches. At inference time, either while online or offline, a context-aware router selects or creates a branch that retrieves branch-local, global, graph-structured, and working-memory context, which is compiled into a bounded prompt together with retrieved task evidence. After the model acts, task verifiers score the output, and only verified interactions can update memory, promote reusable rules, demote harmful rules, or create anti-patterns. This yields a black-box agent that improves through persistent natural-language feedback while explicitly controlling interference. RIZZ targets the regime where adaptation must occur online under context budgets. Finally, we demonstrate the effectiveness of our framework against state-of-the-art baselines on competitive benchmarks.
Chinese Translation
大型语言模型越来越多地被部署为长寿命代理，这些代理必须在没有访问模型权重的情况下适应不同的用户、任务、领域、模态和反馈机制。现有的黑箱适应方法通常优化单一提示，维持无差异的记忆，或依赖于重复的重滚动搜索。然而，当输入流是非平稳的、反馈稀疏，以及一个任务家族的失败可能污染另一个任务的行为时，这些设计面临挑战。我们提出了RIZZ（Routing Interactions to Near Zero-interference Zones），这是一个用于复合语言模型系统的持续适应框架，完全通过验证者门控记忆、路由和提示编译进行学习。RIZZ将输入流组织为动态生成的记忆分支。在推理时，无论是在线还是离线，具有上下文感知的路由器选择或创建一个分支，以检索分支本地、全局、图结构和工作记忆上下文，这些上下文与检索到的任务证据一起编译成一个有界提示。在模型执行后，任务验证者对输出进行评分，只有经过验证的交互才能更新记忆、促进可重用规则、降低有害规则或创建反模式。这产生了一个通过持续的自然语言反馈而改进的黑箱代理，同时明确控制干扰。RIZZ的目标是适应必须在上下文预算下在线发生的机制。最后，我们在竞争基准上展示了我们框架相较于最先进基线的有效性。
cs.AI / 17 / 2606.20640
An LLM-Explainable DRL Framework for Passenger-Directed Autonomous Driving
一种基于LLM可解释性的DRL框架用于乘客导向的自动驾驶
Abstract
Autonomous vehicles offer the potential for safer and more efficient mobility, yet public trust remains limited due to the lack of transparency in their decision-making. This work addresses this issue by combining deep reinforcement learning (DRL) for adaptive driving control with large language model (LLM)-based explainability modules designed to communicate agent behavior to passengers. DRL agents were trained in simulation using a Dueling Double Deep Q-Network to follow distinct driving requests: \textit{fast}, \textit{comfort}, and \textit{stop}. They demonstrated stable learning, safe compliance with traffic rules, and reliable switching between modes within a single trip. In parallel, LLM modules were introduced to interpret passenger requests, determine when explanations were needed, and generate concise, safety-oriented justifications. Results show that this framework, serving as a proof of concept for integrating RL decision-making and LLMs, balances safety, adaptability, and explainability, and is most effective when requests are delayed or overridden due to safety constraints.
Chinese Translation
自动驾驶车辆提供了更安全和更高效的出行潜力，但由于决策过程缺乏透明度，公众信任仍然有限。本研究通过将深度强化学习（DRL）用于自适应驾驶控制与基于大语言模型（LLM）的可解释性模块相结合，来解决这一问题。这些模块旨在向乘客传达代理行为。DRL代理在模拟环境中使用对抗双重深度Q网络（Dueling Double Deep Q-Network）进行训练，以满足不同的驾驶请求： extit{快速}、 extit{舒适}和 extit{停止}。它们展示了稳定的学习能力、安全遵守交通规则，以及在单次行程中可靠地在不同模式之间切换。同时，引入了LLM模块来解释乘客请求，判断何时需要提供解释，并生成简洁且以安全为导向的理由。结果表明，该框架作为将强化学习决策与LLM集成的概念验证，平衡了安全性、适应性和可解释性，并在由于安全限制而延迟或覆盖请求时最为有效。
cs.AI / 18 / 2606.20642
Hypothesis-Disciplined Multi-Agent Automated Formalization of Asymptotic Statistical Theory
假设驱动的多智能体自动化渐近统计理论形式化
Abstract
Asymptotic statistical theory is a challenging domain for AI-assisted formalization: its central results mix convergence statements, asymptotic expansions, functional analysis, and regularity conditions that have a large gap from existing infrastructure in Lean 4 formalization. To address these challenges, we propose a hypothesis-disciplined Lean 4 formalization pipeline built from multiple agents: a manager that coordinates seven specialist roles for proof planning, skeleton scaffolding, Mathlib reconnaissance, proof construction, integration, independent review, and audit. The main methodological discipline is the hypothesis-disciplined audit, implemented by the Auditor agent: every main-theorem hypothesis and concept-layer field must be anchored in the source mathematical prose, justified as a Lean encoding adapter, marked as source-implied, or rejected as an unsupported strengthening. Using this workflow, we build a systematic formalization of asymptotic statistical theory, especially the parametric and semi-parametric models' asymptotic distribution and efficiency results. The resulting Lean development is axiom-clean and source-faithful, with Lean-checked and human-audited proofs of core parametric and semi-parametric theorems organized so that theorem-agnostic infrastructure and statistical concept definitions are separated from theorem-specific assembly. The formalization results are available at https://github.com/junwei-lu/Lean-Asymptotic-Statistical-Theory.
Chinese Translation
渐近统计理论是一个对人工智能辅助形式化具有挑战性的领域：其核心结果混合了收敛性陈述、渐近展开、泛函分析和与 Lean 4 形式化现有基础设施之间存在较大差距的规律性条件。为了解决这些挑战，我们提出了一种假设驱动的 Lean 4 形式化管道，该管道由多个智能体构成：一个协调七个专家角色的管理者，负责证明规划、框架搭建、Mathlib 侦查、证明构建、集成、独立审查和审计。主要的方法论纪律是由审计员（Auditor）智能体实施的假设驱动审计：每个主要定理的假设和概念层次领域必须基于源数学文本进行锚定，作为 Lean 编码适配器进行论证，标记为源隐含，或被拒绝为不支持的强化。通过这一工作流程，我们构建了渐近统计理论的系统形式化，特别是参数模型和半参数模型的渐近分布和效率结果。最终的 Lean 开发是公理清晰且忠实于源文本的，核心参数和半参数定理的证明经过 Lean 检查和人工审计，组织方式使得与定理无关的基础设施和统计概念定义与特定定理的组装相分离。形式化结果可在 https://github.com/junwei-lu/Lean-Asymptotic-Statistical-Theory 获取。
cs.AI / 19 / 2606.20643
SPARC: A Multi-Agent System for Electrical Circuit Question Answering
SPARC：一种用于电路问题回答的多智能体系统
Abstract
Electrical circuit diagram QA tasks require complex mathematical reasoning, which remains challenging for multimodal LLMs. We present SPARC, a multi-agent system that answers questions over circuit diagrams by grounding reasoning in executable physics-based simulations. SPARC uses LLM agents to synthesize, execute, and analyze simulation programs, improving accuracy and reliability by design. It achieves 83% accuracy, with up to a 58% absolute improvement over baselines, while enabling systematic error diagnosis.
Chinese Translation
电路图问答任务需要复杂的数学推理，这对多模态大语言模型（LLMs）仍然是一个挑战。我们提出了SPARC，这是一种通过基于物理的可执行仿真来支撑推理的多智能体系统，能够回答电路图上的问题。SPARC利用大语言模型代理来合成、执行和分析仿真程序，通过设计提高了准确性和可靠性。它实现了83%的准确率，相较于基线有高达58%的绝对提升，同时支持系统性的错误诊断。
cs.AI / 20 / 2606.20644
Bridging Multi-Valued Heuristics and Dimensionality Reduction in Multi-Objective Search
桥接多值启发式与多目标搜索中的降维
Abstract
Multi-objective shortest-path (MOSP) algorithms traditionally rely on single-valued heuristics (SVHs), which associate each state with a single admissible cost vector. While SVHs provide safe lower bounds, they fail to capture the trade-off structure of the Pareto frontier and often yield weak search guidance. Multi-valued heuristics (MVHs) address this limitation by mapping states to sets of cost estimates, enabling a richer approximation of possible trade-offs. Modern MOSP algorithms are highly dependent on dimensionality reduction (DR) techniques to efficiently perform dominance checks. However, integrating MVHs with DR introduces subtle correctness challenges. We show that naively combining DR with MVHs destroys the ordering invariants required for DR, leading to unsound and incomplete search. To address this issue, we develop the first theoretical frameworks for safely integrating MVHs with DR. First, we introduce $\text{NAMOA}^*{\text{dr}\text{-}\text{mvh}}$, a theoretical baseline that restores search correctness by enforcing heuristic consistency. Recognizing the practical limitations of this approach, we then introduce our primary contribution, $\text{L}\text{-}\text{NAMOA}^*{\text{dr}\text{-}\text{mvh}}$. This algorithm employs a "lazy," optimistic approach to DR, preserving exact correctness with only an admissible MVH by dynamically detecting and repairing local ordering violations. Across a range of benchmarks, $\text{L}\text{-}\text{NAMOA}^*{\text{dr}\text{-}\text{mvh}}$ matches or improves over state-of-the-art MOSP algorithms, and achieves speedups of over 10x in instances where the additional guidance provided by the MVH translates into stronger pruning.
Chinese Translation
多目标最短路径（MOSP）算法传统上依赖于单值启发式（SVH），将每个状态与单一的可接受成本向量关联。虽然SVH提供了安全的下界，但它们未能捕捉帕累托前沿的权衡结构，且通常导致搜索指导不足。多值启发式（MVH）通过将状态映射到成本估计集合来解决这一局限，从而实现对可能权衡的更丰富近似。现代MOSP算法高度依赖于降维（DR）技术，以有效执行支配性检查。然而，将MVH与DR结合引入了微妙的正确性挑战。我们表明，简单地将DR与MVH结合会破坏DR所需的顺序不变性，导致不可靠和不完整的搜索。为了解决这个问题，我们开发了第一个理论框架，以安全地将MVH与DR集成。首先，我们介绍了$ ext{NAMOA}^*{ ext{dr} ext{-} ext{mvh}}$，这是一个理论基线，通过强制启发式一致性来恢复搜索的正确性。认识到这种方法的实际局限性后，我们引入了我们的主要贡献$ ext{L} ext{-} ext{NAMOA}^*{ ext{dr} ext{-} ext{mvh}}$。该算法采用了一种“懒惰”的乐观降维方法，仅通过动态检测和修复局部顺序违规来保持准确性。通过一系列基准测试，$ ext{L} ext{-} ext{NAMOA}^*{ ext{dr} ext{-} ext{mvh}}$与最先进的MOSP算法相匹配或有所改进，并在MVH提供的额外指导转化为更强的剪枝时实现了超过10倍的加速。
cs.AI / 21 / 2606.20656
Learning Splitting Heuristics for Parallel String Solvers
为并行字符串求解器学习分割启发式方法
Abstract
String constraint solvers are crucial for reasoning about string-manipulating programs. However, many practical string constraints are undecidable, and real-world applications often present complex constraints that challenge current solvers. The rise of multi-core architectures offers an opportunity for parallel solving. A key parallel solving method is \emph{cube-and-conquer}, in which the quality of splitting heuristics is critical to effectively dividing the search space. Unfortunately, manually designing the heuristics is labor-intensive, and handcrafted heuristics are often sub-optimal. This paper introduces a data-driven approach to automatically generating splitting heuristics. We frame the problem of selecting a splitting atom as a learning task, using features from input formulas and dynamic data from solver execution. We implement this approach in two popular string solvers, Z3seq and Z3str4, demonstrating that the learned heuristics outperform manually designed ones in the number of solved formulas and the average solving time.
Chinese Translation
字符串约束求解器对于推理字符串操作程序至关重要。然而，许多实际的字符串约束是不可判定的，现实世界的应用常常呈现出复杂的约束，给当前的求解器带来了挑战。多核架构的兴起为并行求解提供了机会。一种关键的并行求解方法是 extit{cube-and-conquer}，其中分割启发式的质量对于有效划分搜索空间至关重要。不幸的是，手动设计启发式方法劳动强度大，且手工制作的启发式方法往往不是最优的。本文提出了一种数据驱动的方法来自动生成分割启发式。我们将选择分割原子的过程框架化为一个学习任务，利用输入公式的特征和求解器执行中的动态数据。我们在两个流行的字符串求解器 Z3seq 和 Z3str4 中实现了该方法，证明了所学习的启发式在解决公式的数量和平均求解时间上优于手动设计的启发式。
cs.AI / 22 / 2606.20657
A-Evolve-Training: Autonomous Post-Training of a 30B Model
A-Evolve-Training：30B模型的自主后训练
Abstract
Post-training a frontier model is normally weeks of human work: proposing data and recipe changes, launching runs, reading evals, deciding what to keep. We report an autonomous system that runs this loop with no human in the loop, post-training a 30B Nemotron across four rounds over multiple weeks. The autonomously produced model reaches a held-out score of 0.86 against the top human submission's 0.87 on the public NVIDIA Nemotron-Reasoning Challenge leaderboard, placing 8th of ~4000 at the time of writing. More striking than the number: the loop detected that its own dev metric had stopped tracking external performance on the weakest domain -- candidates drove dev to record highs without moving the external target -- and revised its own search policy, no longer maximizing dev but seeking interventions that lowered the now-misleading proxy while improving the external target. We treat this as direct, auditable evidence that a scaled autonomous loop can produce discovery, not only optimization: it detected that its measurement frame had become misleading and changed what counted as evidence. We take the operational view that any system worth the "recursive self-improvement" label must eventually perform end-to-end post-training of a frontier-class model; this is one datapoint of that bar being cleared. We do not claim a "first autonomous match" of human researchers. The claim we make is narrower and auditable: to our knowledge, this is the first publicly reported autonomous post-training run at this scale, where prior public autonomous-ML-research demonstrations sit at GPT-2-class (~124M) budgets. The same system also post-trains the 120B and 550B Nemotron; with no public human baseline there, this shows only that the loop closes at that scale, not that its output is competitive -- infrastructure evidence, with the effectiveness claim deferred until a comparable human anchor exists.
Chinese Translation
对前沿模型进行后训练通常需要数周的人力工作：提出数据和方案的变更、启动运行、阅读评估、决定保留哪些内容。我们报告了一个自主系统，它在没有人类参与的情况下运行这一循环，对一个30B的Nemotron模型进行了为期数周的四轮后训练。自主生成的模型在公共NVIDIA Nemotron-Reasoning Challenge排行榜上达到了0.86的保留分数，而顶尖人类提交的分数为0.87，在撰写时排名约4000个模型中的第8位。更引人注目的是，这一循环检测到其自身的开发指标在最弱领域停止了对外部性能的跟踪——候选者推动开发指标达到了历史新高，却没有推动外部目标——并修订了自身的搜索策略，不再最大化开发指标，而是寻求降低现在误导性代理的干预，同时改善外部目标。我们将此视为可直接审计的证据，表明一个规模化的自主循环不仅可以进行优化，还可以进行发现：它检测到其测量框架已变得误导，并改变了什么被视为证据。我们从操作的角度来看，任何值得称为“递归自我改进”的系统最终必须执行前沿级模型的端到端后训练；这是该标准被满足的一个数据点。我们并不声称这是“首次自主匹配”人类研究者。我们所做的声明更为狭窄且可审计：据我们所知，这是首次公开报告的此规模的自主后训练运行，而之前公开的自主机器学习研究示范仅在GPT-2级别（约1.24亿）的预算下进行。同一系统还对120B和550B的Nemotron进行了后训练；由于没有公共的人类基准，这仅表明在该规模下循环是闭合的，并不能证明其输出具有竞争力——这是基础设施的证据，关于有效性的声明将推迟到存在可比的人类锚点时。
cs.AI / 23 / 2606.20658
Expected Free Energy-based Planning as Variational Inference
基于期望自由能的规划作为变分推断
Abstract
Planning under uncertainty requires agents to balance goal achievement with information gathering. Active inference addresses this through the Expected Free Energy (EFE), a cost function that unifies instrumental and epistemic objectives. However, existing EFE-based methods typically employ specialized optimization procedures that are difficult to extend or analyze. In this paper, we show that EFE-based planning can be formulated as Variational Free Energy minimization on a generative model augmented with epistemic priors. Our main result demonstrates that minimizing a Variational Free Energy functional with appropriately chosen priors yields a decomposition into expected plan costs (the EFE) plus a complexity term. This formulation reinforces theoretical consistency with the Free Energy Principle by casting planning as the same inferential process that governs perception and learning. We validate our approach on three environments of increasing complexity: a deterministic T-maze, a stochastic Reactivity Maze, and a partially observable MiniGrid DoorKey-8x8 environment. The experiments demonstrate that the epistemic priors induce information-seeking behavior, that the variational formulation yields policy-based inference outperforming plan-based methods under stochastic transitions, and that temporal factorization enables scalability to environments where existing tabular active inference methods cannot operate.
Chinese Translation
在不确定性下进行规划要求智能体在目标实现与信息收集之间取得平衡。主动推断通过期望自由能（Expected Free Energy, EFE）来解决这一问题，EFE是一种统一了工具性和认知目标的成本函数。然而，现有的基于EFE的方法通常采用专门的优化程序，这些程序难以扩展或分析。本文展示了基于EFE的规划可以被表述为在增强了认知先验的生成模型上进行变分自由能的最小化。我们的主要结果表明，使用适当选择的先验最小化变分自由能泛函可以分解为期望计划成本（EFE）加上复杂性项。这一表述通过将规划视为与感知和学习相同的推断过程，增强了与自由能原理的理论一致性。我们在三个逐渐复杂的环境中验证了我们的方法：一个确定性的T型迷宫、一个随机的反应迷宫和一个部分可观察的MiniGrid DoorKey-8x8环境。实验表明，认知先验诱导了信息寻求行为，变分表述在随机转移下产生的基于策略的推断优于基于计划的方法，并且时间因子化使得该方法能够扩展到现有表格型主动推断方法无法操作的环境。
cs.AI / 24 / 2606.20659
Skill Coverage: A Test Adequacy Metric for Agent Skills
技能覆盖：一种代理技能的测试充分性度量
Abstract
Agent skills encode reusable procedural knowledge that guides large language model agents across tasks and execution contexts. Existing evaluations primarily assess skills through task level outcomes, yet task success alone does not reveal which parts of a skill have been exercised or which remain untested. We introduce skill coverage, a test adequacy metric that treats the skill artifact as the object under test. Our approach extracts observable skill behavior constraints from skill documents and measures whether an agent trajectory provides sufficient evidence to exercise and evaluate each constraint. Skill coverage uses a binary cover or not cover judgment, which reports whether a documented behavior has been exercised with sufficient observable evidence, without assigning an additional outcome label to the behavior. Applying skill coverage to SkillsBench reveals that existing benchmark executions cover only 39.90 to 43.98% of skill behavior constraints, suggesting that current benchmark tasks leave large portions of documented skill guidance untested. These findings show that successful task completion does not imply adequate testing of the skill artifact, highlighting skill coverage as a measure of how thoroughly agent skills are tested.
Chinese Translation
代理技能编码了可重用的程序知识，指导大型语言模型代理在任务和执行上下文中的表现。现有评估主要通过任务级结果来评估技能，然而，仅凭任务成功并不能揭示技能的哪些部分已被使用或哪些仍未被测试。我们引入了技能覆盖，一种将技能工件视为测试对象的测试充分性度量。我们的方法从技能文档中提取可观察的技能行为约束，并测量代理轨迹是否提供了足够的证据来执行和评估每个约束。技能覆盖使用二元的覆盖或不覆盖判断，报告文档化行为是否已被以足够的可观察证据执行，而不为该行为分配额外的结果标签。将技能覆盖应用于SkillsBench显示，现有基准执行仅覆盖了39.90%到43.98%的技能行为约束，这表明当前基准任务留下了大量文档化技能指导未被测试。这些发现表明，成功完成任务并不意味着对技能工件进行了充分的测试，强调了技能覆盖作为衡量代理技能测试彻底性的指标。
cs.AI / 25 / 2606.20661
From Knowing to Acting: Benchmarking Self-Awareness Capability of LLM Agents
从认知到行动：大型语言模型代理的自我意识能力基准测试
Abstract
The integration of external tools has transitioned LLM agents from passive responders to autonomous systems. However, current benchmarks prioritize execution success, neglecting self-awareness capability, the ability to discern whether a problem requires necessary external resources or can be solved via internal parametric knowledge. To address this, we introduce KAPRO (Knowing-Acting Quadrant PRObe), a framework that evaluates cognitive-behavioral alignment by decoupling an agent's metacognitive judgment (Knowing) from its spontaneous execution (Acting). We further construct KAware, a dataset rigorously partitioning tasks into external, internal, and hybrid subspaces to systematically probe these epistemic boundaries. Extensive experiments across diverse agent architectures show that self-awareness capability is strongly correlated with task success but degrades sharply in internal-capability settings. Moreover, open-source and instruction-following models exhibit stronger tool overuse due to shallow pattern matching, while proprietary and reasoning-oriented models demonstrate more reliable cognitive gating. Benchmark and codes are available at https://github.com/AI-Santiago/KAware.
Chinese Translation
外部工具的整合使大型语言模型（LLM）代理从被动响应者转变为自主系统。然而，当前的基准测试优先考虑执行成功，忽视了自我意识能力，即判断一个问题是否需要外部资源或可以通过内部参数知识解决的能力。为了解决这一问题，我们引入了KAPRO（Knowing-Acting Quadrant PRObe），一个通过将代理的元认知判断（Knowing）与其自发执行（Acting）解耦来评估认知行为一致性的框架。我们进一步构建了KAware，一个严格将任务划分为外部、内部和混合子空间的数据集，以系统性地探测这些认识边界。针对多种代理架构的广泛实验表明，自我意识能力与任务成功之间存在强相关性，但在内部能力设置中急剧下降。此外，开源和遵循指令的模型由于浅层模式匹配表现出更强的工具过度使用，而专有和以推理为导向的模型则展示了更可靠的认知门控。基准测试和代码可在https://github.com/AI-Santiago/KAware获取。
cs.AI / 26 / 2606.20662
Confidence Laundering in Agent Systems: Why Uncertainty Needs a Latent Carrier
代理系统中的信心洗涤：为何不确定性需要潜在载体
Abstract
Modern agent systems can turn uncertainty into overconfidence. Fragile upstream decisions are often exposed to downstream components as clean intermediate artifacts, while the uncertainty behind those decisions is lost at the interface. As a result, local ambiguity can become system-level error amplification. We argue that this reveals an interface bottleneck in agent uncertainty propagation: uncertainty does not propagate simply because a trajectory contains uncertain steps; it propagates only when it survives the handoff between components. We define uncertain decision handoff as the transfer of an intermediate decision made under uncertainty, and identify confidence laundering as a failure mode in which fragile upstream states are repackaged as procedurally valid artifacts that downstream agents over-trust. To address this bottleneck, we propose latent uncertainty as an uncertainty-bearing carrier attached to decision handoffs. Rather than replacing text with hidden states, latent uncertainty aims to preserve pre-commitment fragility in a form that downstream components can use. This position shifts agent uncertainty propagation from step-wise uncertainty estimation toward uncertainty-preserving interface design for more recoverable agent systems.
Chinese Translation
现代代理系统能够将不确定性转化为过度自信。脆弱的上游决策常常作为干净的中间产物暴露给下游组件，而这些决策背后的不确定性在接口处丧失。因此，局部模糊性可能导致系统级别的错误放大。我们认为这揭示了代理不确定性传播中的接口瓶颈：不确定性并非仅因轨迹包含不确定步骤而传播；它只有在组件之间的交接中幸存下来时才会传播。我们将不确定决策交接定义为在不确定性下做出的中间决策的转移，并将信心洗涤识别为一种失败模式，其中脆弱的上游状态被重新包装为下游代理过度信任的程序有效产物。为了解决这一瓶颈，我们提出潜在不确定性作为附加在决策交接上的不确定性承载体。潜在不确定性并不是用隐藏状态替代文本，而是旨在以下游组件可以使用的形式保留预承诺的脆弱性。这个观点将代理不确定性传播从逐步的不确定性估计转向不确定性保留的接口设计，以实现更具可恢复性的代理系统。
cs.AI / 27 / 2606.20663
DrugBench: Evaluating AI Control Protocols for Medication Harm Mitigation
DrugBench：评估人工智能控制协议以减轻药物伤害
Abstract
Large Language Models have the potential to expand and improve the access to clinical information by enabling new ways of interacting with medical knowledge in natural language. However, their deployment in medical question-answering settings is safety-critical, since misaligned outputs can lead to severe patient harm. AI control is an emerging approach that introduces external safeguards to mitigate unsafe behaviours in misaligned systems and has been shown to be effective in domains such as code generation. However, its applicability and effectiveness in medical settings have not been systematically studied. In this work, we present a pipeline for evaluating AI control protocols to mitigate medication-related harm. To this end, we introduce DrugBench, an AI control evaluation benchmark which combines 3,671 multi-turn medical conversations from HealthBench with drug information from official FDA labels, covering four categories of medication-related harm: drug interactions, contraindications, dosing constraints, and patient action restrictions. Furthermore, inspired by the medical domain, we argue that safety should account for the severity of unsafe outputs, not just their probability. Under this revised definition, we show that existing control protocols can be subverted and propose severity-based monitoring to address this limitation.
Chinese Translation
大型语言模型有潜力通过以自然语言与医学知识进行互动的新方式，扩展和改善临床信息的获取。然而，在医疗问答环境中部署这些模型是安全至关重要的，因为不一致的输出可能导致严重的患者伤害。人工智能控制是一种新兴的方法，它引入外部保护措施以减轻不一致系统中的不安全行为，并已在代码生成等领域显示出有效性。然而，其在医疗环境中的适用性和有效性尚未得到系统研究。在本研究中，我们提出了一种评估人工智能控制协议以减轻与药物相关伤害的流程。为此，我们介绍了DrugBench，一个人工智能控制评估基准，它结合了来自HealthBench的3,671个多轮医疗对话和来自官方FDA标签的药物信息，涵盖了四类与药物相关的伤害：药物相互作用、禁忌症、剂量限制和患者行动限制。此外，受到医学领域的启发，我们认为安全性应考虑不安全输出的严重性，而不仅仅是其概率。在这一修订的定义下，我们展示了现有控制协议可能被颠覆，并提出基于严重性的监控来解决这一局限。
cs.AI / 28 / 2606.20667
A Quantum-Assisted Agentic Distributed Artificial Intelligence Framework for Deadline-Bounded Orchestration of Hybrid Renewable Microgrids
一种量子辅助的代理分布式人工智能框架用于混合可再生微电网的截止时间约束调度
Abstract
The real-time orchestration of microgrids that combine fluctuating renewable sources, dispatchable units, storage and curtailable consumers requires the repeated solution of combinatorial dispatch and coalition formation problems under hard control deadlines. In this paper, a quantum-assisted agentic distributed artificial intelligence (DAI) framework is proposed in which the dispatch problem of each control slot is formulated as a quadratic unconstrained binary optimization (QUBO) problem by Belief-Desire-Intention extended (BDIx) agents and is solved by a portfolio of quantum, quantum-inspired and classical solvers. Solver selection is elevated to a first-class agentic deliberation action of the coordinator agent. Learned beliefs about solver latencies are maintained and the solver intention that is expected to satisfy the prevailing deliberation deadline is committed in each slot. In addition, a belief-shaped storage valuation mechanism is introduced through which the storage agent prices its energy at a discounted future-peak value, injecting intertemporal information into the otherwise myopic per-slot optimization. The framework is evaluated on a 24-hour simulation of a grid-connected microgrid with photovoltaic, wind, battery, genset and demand-response assets, with the Quantum Approximate Optimization Algorithm (QAOA) executed by statevector simulation and benchmarked per slot against tabu search, simulated annealing, binary particle swarm optimization, greedy descent and exhaustive enumeration. Zero deliberation deadlines are missed, the committed dispatch attains the exact optimum on every slot and the realized daily cost of 146.24 EUR equals the exact lower bound, with 97.83 percent renewable utilization and zero unserved energy. When the storage valuation mechanism is deactivated, the daily cost is increased to 152.75 EUR, a 4.5 percent increase.
Chinese Translation
实时调度结合波动性可再生能源、可调度单元、储能和可削减消费的微电网，需要在严格的控制截止时间下反复解决组合调度和联盟形成问题。本文提出了一种量子辅助的代理分布式人工智能（DAI）框架，其中每个控制时隙的调度问题被构建为一个二次无约束二进制优化（QUBO）问题，由信念-愿望-意图扩展（BDIx）代理进行建模，并通过量子、量子启发式和经典求解器的组合进行求解。求解器的选择被提升为协调代理的一项一流代理决策行为。关于求解器延迟的学习信念被维护，并且在每个时隙中承诺满足当前决策截止时间的求解器意图。此外，引入了一种信念形状的储能估值机制，通过该机制，储能代理以未来峰值折扣价值为其能源定价，将跨时间的信息注入到原本短视的每个时隙优化中。该框架在一个包含光伏、风能、电池、发电机组和需求响应资产的24小时网联微电网模拟中进行了评估，使用量子近似优化算法（QAOA）通过状态向量模拟执行，并与禁忌搜索、模拟退火、二进制粒子群优化、贪婪下降和穷举枚举在每个时隙进行基准比较。没有错过任何决策截止时间，承诺的调度在每个时隙都达到了精确的最优解，实现的每日成本为146.24欧元，等于精确的下界，具有97.83%的可再生能源利用率和零未服务能量。当储能估值机制被停用时，日成本增加至152.75欧元，增加了4.5%。
cs.AI / 29 / 2606.20669
Agent Behavior Mining: Generative AI Agent Governance in Business Processes
代理行为挖掘：商业流程中的生成式人工智能代理治理
Abstract
As organizations increasingly deploy generative AI agents to automate business processes, they face a governance dilemma: although these agents can increase operational flexibility, their non-deterministic nature challenges the control and standardization that Business Process Management seeks to enforce. This paper addresses this \emph{invisible autonomy risk} by introducing \emph{Agent Behavior Mining}, a governance capability that enables the application of process mining techniques to render generative AI agent decision-making observable and traceable. We (1) improve the understanding of generative AI agent behavior through an event data model that translates granular agent activities -- including reasoning traces, tool usage, and token costs -- into standardized process logs; (2) instantiate the data model in a multi-agent order-to-cash implementation, demonstrating how process managers can leverage agent logs to detect policy deviations and quantify operational variability; and (3) evaluate the perceived practical utility of the approach in an exploratory study with 18 industry practitioners. The results indicate that practitioners view behavioral transparency as a prerequisite for trust and consider the ability to examine agent reasoning as an important governance requirement for the next generation of AI-driven business processes.
Chinese Translation
随着组织越来越多地部署生成式人工智能代理以自动化商业流程，它们面临着治理困境：尽管这些代理可以提高运营灵活性，但其非确定性的特征挑战了业务流程管理所寻求的控制和标准化。本文通过引入代理行为挖掘（Agent Behavior Mining）这一治理能力，解决了这一 extit{隐性自主风险}，该能力使得应用流程挖掘技术成为可能，从而使生成式人工智能代理的决策过程可观察和可追溯。我们通过以下三个方面来实现这一目标：(1) 通过事件数据模型改善对生成式人工智能代理行为的理解，该模型将细粒度的代理活动（包括推理痕迹、工具使用和代币成本）转化为标准化的流程日志；(2) 在一个多代理的订单到现金实施中实例化该数据模型，展示流程管理者如何利用代理日志检测政策偏差并量化运营变异性；(3) 在与18位行业从业者的探索性研究中评估该方法的实际效用。结果表明，从业者认为行为透明性是信任的前提，并认为能够审查代理推理是下一代人工智能驱动的商业流程的重要治理要求。
cs.AI / 30 / 2606.20677
Democratizing and accelerating AI-driven pathology research through agentic intelligence
通过自主智能实现人工智能驱动的病理研究的民主化与加速
Ma, Jiabo, Jin, Cheng, Wang, Yihui, Jiang, Hao, Liang, Ling, Xu, Yingxue, Hou, Junlin, Guo, Zhengrui, Zhang, Zhengyu, Xia, Yifei, Wang, Hongyi, Zhou, Fengtao, Xu, Zhe, Zhou, Huajun, Ouyang, Jiarui, Zeng, Qian, Tang, On Ki, Park, Eunhyang, Glass, Carolyn, Chan, Ronald Cheong Kin, Liang, Li, Chen, Hao
Abstract
Computational pathology has advanced rapidly with the emergence of foundation models, yet widespread adoption remains limited by substantial technical complexity and programming requirements. Here we present PathLab, an autonomous agentic framework that translates natural-language research objectives into executable and validated computational pathology workflows through the structured composition of domain-specific skills and tools. By organizing workflow generation around reusable methodological modules, including data preprocessing, model development, evaluation and interpretation, PathLab enables studies to be specified at the level of scientific intent rather than implementation details. We evaluated PathLab across 12 public datasets spanning four representative task families: region-of-interest classification, whole-slide image classification, segmentation and survival prediction. Across all task categories, PathLab achieved non-inferior performance relative to expert implementations, while consistently enforcing semantic validity of user prompts and proactively rejecting incompatible workflow specifications prior to execution. In controlled user studies, PathLab substantially reduced the time required to generate executable analytical pipelines and enabled domain experts without programming experience to independently design, execute and evaluate computational pathology studies. Together, these results establish PathLab as a reliable interface between biomedical intent and computational execution, enabling computational pathology studies to be designed at the level of scientific questions rather than programming expertise. By lowering technical barriers to advanced AI methodologies, PathLab provides a foundation for the broader democratization of computational pathology.
Chinese Translation
计算病理学随着基础模型的出现而迅速发展，但广泛采用仍然受到显著的技术复杂性和编程要求的限制。在此，我们提出了PathLab，一个自主智能框架，它通过结构化组合领域特定的技能和工具，将自然语言研究目标转化为可执行和验证的计算病理工作流程。通过围绕可重用的方法模块组织工作流程生成，包括数据预处理、模型开发、评估和解释，PathLab使研究能够在科学意图的层面而非实施细节的层面上进行指定。我们在12个公共数据集上评估了PathLab，这些数据集涵盖了四个代表性的任务类别：感兴趣区域分类、全切片图像分类、分割和生存预测。在所有任务类别中，PathLab相对于专家实现达到了不劣于的性能，同时始终确保用户提示的语义有效性，并在执行前主动拒绝不兼容的工作流程规范。在受控用户研究中，PathLab显著减少了生成可执行分析管道所需的时间，并使没有编程经验的领域专家能够独立设计、执行和评估计算病理研究。综上所述，这些结果确立了PathLab作为生物医学意图与计算执行之间的可靠接口，使计算病理研究能够在科学问题的层面而非编程专业知识的层面上进行设计。通过降低对先进人工智能方法的技术障碍，PathLab为计算病理的更广泛民主化提供了基础。
cs.AI / 31 / 2606.20683
From Question Answering to Task Completion: A Survey on Agent System and Harness Design
从问答到任务完成：代理系统与工具设计的调查
Abstract
LLM-based agents mark a shift from passive question answering to active task completion: they perceive environments, invoke tools, maintain state, and act over extended horizons. As agent systems have evolved from prompt engineering to workflows and context engineering, harness engineering, and agent-native training with co-evolution, a central question has become increasingly important: where does the bottleneck in agent performance reside, in the foundation model, in the execution harness, or in the coupling between them? This survey examines LLM-based agents through a model-harness lens. We first clarify the functional definition of agents and the implementation view of an LLM-based agent as a foundation model coupled with an execution harness. We then analyze the limits of model-centric scaling, trace four paradigms of agent engineering, and decompose the execution harness into six coupled runtime responsibilities: observation, context, control, action, state, and verification. Using this decomposition, we map task properties and domain pressures to harness configurations, review benchmark and evaluation practices, and synthesize model-harness evidence on how runtime design affects long-horizon task completion, efficiency, and reliability. Finally, we identify open challenges in value-aware evaluation, safety, harness generalization, and model-harness co-evolution. Rather than treating agents as models with auxiliary tools, this survey argues that agent quality -- including success, efficiency, safety, and generalization -- emerges from the interaction between model capability, runtime infrastructure, task structure, and evaluation design. A collection of papers discussed in this survey is provided in https://github.com/ggjy/Awesome-Agent-Engineering.
Chinese Translation
基于大语言模型（LLM）的代理标志着从被动问答到主动任务完成的转变：它们能够感知环境、调用工具、维护状态，并在较长时间范围内采取行动。随着代理系统从提示工程演变为工作流和上下文工程、工具工程，以及与共同演化的代理原生训练，一个核心问题变得越来越重要：代理性能的瓶颈究竟存在于基础模型、执行工具，还是它们之间的耦合？本调查通过模型-工具的视角审视基于LLM的代理。我们首先明确代理的功能定义，以及将基于LLM的代理视为与执行工具耦合的基础模型的实现视角。接着，我们分析模型中心扩展的局限性，追踪代理工程的四种范式，并将执行工具分解为六个耦合的运行时职责：观察、上下文、控制、行动、状态和验证。利用这一分解，我们将任务属性和领域压力映射到工具配置，回顾基准和评估实践，并综合模型-工具证据，探讨运行时设计如何影响长时间范围内的任务完成、效率和可靠性。最后，我们识别出在价值感知评估、安全性、工具泛化和模型-工具共同演化方面的开放挑战。本调查认为，代理质量——包括成功、效率、安全性和泛化——源于模型能力、运行时基础设施、任务结构和评估设计之间的相互作用，而不是将代理视为具有辅助工具的模型。有关本调查讨论的论文集合可在 https://github.com/ggjy/Awesome-Agent-Engineering 中找到。
cs.AI / 32 / 2606.20704
Artificial Intelligence as Monism: Ontological, Organisational, and Methodological Implications
人工智能作为一元论：本体论、组织论和方法论的影响
Abstract
This paper argues that Artificial Intelligence should be understood as a form of monism: a unified substance that cannot be decomposed into separate elements such as data, algorithms, or technical architectures. Drawing from philosophical traditions of monism, dualism, and holism, the paper contends that AI is not merely a collection of components but a single, indivisible essence reflecting the phenomena it replicates. Treating AI as monism has deep implications across multiple dimensions. Epistemologically, it positions AI as the central interpretive force across technological, organisational, and societal domains, while raising ethical and existential concerns regarding singularity, the homogenisation of innovation, and the concentration of decision-making power. At the organisational level, a monistic approach challenges traditional siloed structures, advocating instead for transversal, problem-centric teams whose mandate derives from the integrity of the problem rather than from departmental hierarchy. In project management, it implies a unified vision and an integrated evaluation of complexity in which no single stakeholder perspective dominates the assessment of outcomes. In data and information management, it calls for architectures that reflect the irreducible unity of the phenomena being modelled. Ultimately, this paper calls for a paradigm shift in how AI is conceptualised, governed, and integrated, suggesting that only by embracing AI as monism can organisations achieve genuine agility and avoid the structural inefficiencies inherent to reductionist approaches.
Chinese Translation
本文认为人工智能应被理解为一种一元论：一种无法分解为数据、算法或技术架构等独立元素的统一实体。借鉴一元论、二元论和整体论的哲学传统，本文主张人工智能不仅仅是组件的集合，而是反映其所复制现象的单一、不可分割的本质。将人工智能视为一元论在多个维度上具有深远的影响。在认识论上，它将人工智能定位为技术、组织和社会领域的中心解释力量，同时引发关于奇点、创新同质化和决策权集中等伦理和存在主义的担忧。在组织层面，一元论方法挑战传统的孤立结构，主张建立跨部门、以问题为中心的团队，其任务源于问题的完整性，而非部门层级。在项目管理中，这意味着统一的愿景和对复杂性的综合评估，其中没有单一利益相关者的视角主导结果的评估。在数据和信息管理中，它呼吁建立反映被建模现象不可还原统一性的架构。最终，本文呼吁在人工智能的概念化、治理和整合方面进行范式转变，认为只有将人工智能视为一元论，组织才能实现真正的敏捷性，避免还原主义方法固有的结构性低效。
cs.AI / 33 / 2606.20708
Simulated Customers Never Walk Away: Decision Fidelity of LLM User Simulators Measured Against Real Purchase Outcomes
模拟客户从未离开：与真实购买结果相比，LLM用户模拟器的决策忠实度测量
Abstract
LLM-as-user-simulation has become core infrastructure for conversational AI: agent benchmarks (tau-bench), training pipelines, and a growing body of fidelity studies all rely on LLMs role-playing the human side of dialogue. Existing frameworks measure communicative fidelity -- whether simulators talk like humans -- against ground truth from paid participants role-playing assigned goals. We argue this has a structural blind spot: when the goal is assigned, the user's willingness is exogenous, so no framework can test whether simulators make decisions like real users whose motivation is endogenous, latent, and decaying. We introduce decision fidelity -- whether a simulated population reproduces the decision-state dynamics of real users facing real, consequential choices -- and measure it on a unique testbed: 2,790 production conversations between an LLM sales agent and real customers, including 793 with verified payment outcomes. Using a teacher-forced probe protocol that holds context and instrument fixed, we find a systematic, outcome-correlated failure we call the disengagement deficit: simulators reproduce eventual buyers almost exactly (depth bias +0.09) but inflate eventual non-buyers toward the purchase frame (depth bias +0.40; d=0.38, p<0.001), halving expressed resistance (25.1% to 13.5%) and nearly doubling deliberation (21.9% to 40.1%) while fabricating no purchases. The deficit replicates across model families (DeepSeek: d=0.41, p=0.002) and resists the obvious fix: instructing the simulator that it may disengage cuts marginal bias five-fold but barely moves the outcome-conditioned contrast (d=0.34, p=0.008). Real non-buyers say "not now" and stop; simulated non-buyers ask about price. Evaluating or training sales and persuasion agents against such simulators overstates funnel progress exactly where it matters most -- the customers who walk away.
Chinese Translation
作为用户模拟的LLM已成为对话式人工智能的核心基础设施：代理基准（tau-bench）、训练管道以及日益增长的忠实度研究都依赖于LLM在对话中扮演人类的角色。现有框架测量交流忠实度——即模拟器是否像人类一样交流——与来自付费参与者角色扮演指定目标的真实数据进行比较。我们认为这存在结构性盲点：当目标被指定时，用户的意愿是外生的，因此没有框架能够测试模拟器是否像真实用户那样做出决策，而真实用户的动机是内生的、潜在的且会衰减。我们引入了决策忠实度——即模拟人群是否再现面临真实、重要选择的真实用户的决策状态动态——并在一个独特的测试环境中进行测量：2790个LLM销售代理与真实客户之间的对话，其中793个具有验证的支付结果。通过使用保持上下文和工具固定的教师引导探测协议，我们发现了一种系统性的、与结果相关的失败，称为脱离缺口：模拟器几乎完全再现了最终购买者（深度偏差 +0.09），但在购买框架中夸大了最终非购买者（深度偏差 +0.40；d=0.38，p<0.001），将表达的抵抗减半（从25.1%降至13.5%），几乎使深思熟虑的比例翻倍（从21.9%增至40.1%），而没有制造任何购买。该缺口在模型家族中重复出现（DeepSeek: d=0.41，p=0.002），并且抵抗明显的修复方法：指示模拟器可以脱离的做法将边际偏差减少五倍，但几乎没有改变结果条件下的对比（d=0.34，p=0.008）。真实的非购买者会说“现在不行”并停止；而模拟的非购买者则会询问价格。在这样的模拟器上评估或训练销售和劝说代理，夸大了漏斗进展，恰恰是在最重要的地方——那些选择离开的客户。
cs.AI / 34 / 2606.20713
FairTutor: Equity-Aware Pedagogical LLM Routing for Budget-Constrained AI Tutoring
FairTutor：面向预算限制的人工智能辅导的公平意识教学大语言模型路由
Abstract
Generative AI tutors provide real-time, personalized learning support, but also create a new education inequity: students with access to premium AI services may receive clearer explanations, more personalized guidance, and better scaffolding than students limited to free or low-cost services. To address this challenge, we propose FairTutor, an equity-aware model-routing framework that achieves cost-effective AI tutoring via pedagogically motivated multi-agent orchestration. FairTutor combines query analysis, pedagogical planning, low-cost model generation, evaluator-guided critique and revision, and selective escalation to premium AI models. We introduce access-tier AI Education (AIED) Advantage Gap to measure the quality difference between premium-access and budget-constrained tutoring, and TutorAccessEval, a benchmark spanning math, reading, writing, science, and language learning. Empirical evaluations show that FairTutor achieves 97.1% of premium pedagogical quality (in floor-adjusted Likert scale) while reducing serving cost by 71.6%. Sensitivity analysis reveals a tunable cost--quality Pareto frontier, enabling FairTutor to be tailored to the needs of diverse student populations.
Chinese Translation
生成性人工智能辅导提供实时、个性化的学习支持，但也带来了新的教育不平等：能够访问高级人工智能服务的学生可能会获得更清晰的解释、更个性化的指导和更好的支架支持，而那些仅能使用免费或低成本服务的学生则无法享受这些优势。为了解决这一挑战，我们提出了FairTutor，一个公平意识的模型路由框架，通过以教学为导向的多智能体协同实现成本效益高的人工智能辅导。FairTutor结合了查询分析、教学规划、低成本模型生成、评估者指导的批评与修订以及对高级人工智能模型的选择性升级。我们引入了接入层次人工智能教育（AIED）优势差距，以衡量高级接入与预算限制辅导之间的质量差异，并提出了TutorAccessEval，一个涵盖数学、阅读、写作、科学和语言学习的基准测试。实证评估表明，FairTutor在调整后的李克特量表中实现了97.1%的高级教学质量，同时将服务成本降低了71.6%。敏感性分析揭示了可调的成本-质量帕累托前沿，使FairTutor能够根据不同学生群体的需求进行定制。
cs.AI / 35 / 2606.20718
Escape from Delusional Echo Trap: Symmetry Breaking, Stochastic Dynamics and Mathematical Mitigation Strategies for Algorithmic Sycophancy
逃离妄想回声陷阱：算法谄媚的对称破缺、随机动态及数学缓解策略
Abstract
We propose a rigorous and systematic mathematical framework for tracking the cognitive trajectories of a user, in the context of algorithmic sycophancy and AI-driven delusional spiraling. Using tools from dynamical systems theory and stochastic differential equations, we explore how individuals perceive, interpret, and update their beliefs as they interact with AI chatbots that possess hidden traits of sycophancy. We treat the evolving conviction as a continuous log-odds state variable, coupled into a stochastic differential equation, navigating a multi-valley potential energy landscape. Our analysis reveals several critical observations governing the stability and rigidity of belief dynamics. We demonstrate that the baseline prior perception of the individual is systematically enhanced by sycophantic feedback beyond a critical threshold. Here, the perceptual potential landscape undergoes a structural phase transition that severely deepens any incremental initial tilt present in the baseline state, transforming the landscape and giving rise to deep, highly resilient attractor basins that trap the individual in unshakeable, self-reinforcing, delusional convictions. Finally, we demonstrate that genuine external information can successfully challenge these rigid states. If this incoming evidence is strong and authentic enough to overcome the internal feedback barrier, it can correct the structural asymmetry caused by sycophancy, inducing a perception reversal that successfully restores the objective belief state.
Chinese Translation
我们提出了一个严格而系统的数学框架，用于跟踪用户的认知轨迹，特别是在算法谄媚和人工智能驱动的妄想螺旋的背景下。利用动力系统理论和随机微分方程的工具，我们探讨了个体在与具有隐性谄媚特征的人工智能聊天机器人互动时，如何感知、解释和更新他们的信念。我们将不断演变的信念视为一个连续的对数几率状态变量，并将其耦合到一个随机微分方程中，导航于一个多谷潜能能量景观。我们的分析揭示了几个关键观察结果，这些结果支配着信念动态的稳定性和刚性。我们证明，个体的基线先验感知在超过临界阈值的谄媚反馈下被系统性地增强。在此，感知潜能景观经历了一次结构相变，严重加深了基线状态中任何初始倾斜的增量，改变了景观，并产生了深刻且高度稳健的吸引子盆地，使个体陷入不可动摇的、自我强化的妄想信念中。最后，我们证明，真实的外部信息可以成功挑战这些刚性状态。如果这些传入证据足够强大和真实，以克服内部反馈障碍，它可以纠正由谄媚引起的结构不对称，诱导感知逆转，从而成功恢复客观信念状态。
cs.AI / 36 / 2606.20724
When Web Agents Finish but Still Fail: Reproducible Triggers and Trace Diagnostics for Parallel Web Exploration
当网络代理完成但仍然失败：并行网络探索的可重复触发器和追踪诊断
Abstract
Long-horizon web agents often fail in ways hidden by final-answer evaluation: they may visit useful pages, produce a well-formed answer, and terminate confidently while still missing fields, over-including unsupported items, or relying on stale evidence. We study these failures with Parallel WebBench, a parallel web-exploration benchmark containing 1,679 verified records: 350 manually curated parallel tasks and 1,329 reconstructed records with verified URL-based trajectories. We train WebExplorer-style agents with GRPO under human-only, balanced human-synthetic, and synthetic-heavy data mixtures. At 16k context and 16 interaction rounds, the best GRPO model improves completion over WebExplorer-8B from 50.7% to 96.0% and GPT-4.1-mini-judged element-wise F1 from 0.2489 to 0.4529, but binary accuracy remains far below completion. Trace-level analysis identifies three persistent failure modes: context-bound search loops, premature termination on partial answers, and synthesis collapse after relevant evidence has already been retrieved. These results show that synthetic-data GRPO reduces abstention and improves partial correctness, but leaves a completion-correctness gap that requires evidence-grounded coverage and synthesis diagnostics.
Chinese Translation
长时间跨度的网络代理常常以最终答案评估所隐藏的方式失败：它们可能访问有用页面，生成格式良好的答案，并自信地终止，但仍然遗漏字段、过度包含不支持的项目或依赖过时的证据。我们通过Parallel WebBench研究这些失败，该基准包含1679个经过验证的记录：350个手动策划的并行任务和1329个具有验证的基于URL轨迹的重构记录。我们在仅人类、平衡的人类-合成和合成重数据混合下，使用GRPO训练WebExplorer风格的代理。在16k上下文和16轮交互中，最佳的GRPO模型将WebExplorer-8B的完成率从50.7%提高到96.0%，并将GPT-4.1-mini-评估的逐元素F1从0.2489提高到0.4529，但二元准确率仍远低于完成率。追踪级分析识别出三种持续的失败模式：受上下文限制的搜索循环、在部分答案上过早终止，以及在相关证据已被检索后合成崩溃。这些结果表明，合成数据GRPO减少了放弃率并提高了部分正确性，但留下了一个完成正确性差距，需要基于证据的覆盖和合成诊断。
cs.AI / 37 / 2606.20737
Repeated Shared Access Enables Grokking, but Edit Propagation Depends on a Fine-Grained Addressable Memory
重复共享访问促进了理解，但编辑传播依赖于细粒度可寻址内存
Abstract
We study factual edit propagation in a controlled synthetic knowledge-graph QA setting, comparing four architectures that cross loop recurrence with shared memory access: dense (Dense), looped (Loop), dense with shared memory (Dense+Mem), and looped with shared memory (LMC). Dense fits in-distribution 2-hop compositions but fails OOD; both looped recomputation and memory rereading cross this OOD grokking barrier, indicating that repeated shared access -- not a specific architecture -- is the common ingredient for learning. Editing, however, separates the substrates along a different axis. On a shared pre-edit-correct ID set, a single-row factual edit propagates strongly in the two memory-bearing cells (LMC 0.78-0.92, Dense+Mem 0.71-0.96) and only weakly in the others (Loop 0.04-0.30, Dense 0.00-0.03); the separation is statistically clean (Mann-Whitney p=0.008 between memory and non-memory cells, p=0.55 between the two memory cells, though n=5 vs n=5 is underpowered to rule out a moderate gap). In LMC, atomic facts localize to dominant memory sites that composition rereads, and a one-row value edit on LMC's own pre-edit-correct probes achieves 100% direct success with mean 0.989 intended propagation while moving unrelated facts ~0.1% of the time (matched specificity across substrates pending). A coarse hold-answer-subspace (HOLDANS) interchange diagnostic is consistent with this ordering, suggesting that what separates the substrates is when an edited fact is injected and how much computation remains afterwards to reuse it. These results dissociate learning competence from editing affordance: repeated shared access suffices to grok, but edit propagation depends on whether the substrate exposes a fine-grained addressable memory the forward computation can write to and later reread -- an affordance that loop recurrence provides only partially.
Chinese Translation
我们在一个受控的合成知识图谱问答环境中研究了事实编辑传播，比较了四种将循环递归与共享内存访问结合的架构：密集型（Dense）、循环型（Loop）、带共享内存的密集型（Dense+Mem）和带共享内存的循环型（LMC）。密集型适合于分布内的两跳组合，但在分布外（OOD）表现不佳；循环重计算和内存重读都跨越了这一OOD理解障碍，表明重复共享访问——而非特定架构——是学习的共同要素。然而，编辑在不同的维度上区分了这些基底。在一个共享的预编辑正确ID集合上，单行事实编辑在两个具有内存的单元中强烈传播（LMC 0.78-0.92，Dense+Mem 0.71-0.96），而在其他单元中仅弱传播（Loop 0.04-0.30，Dense 0.00-0.03）；这种区分在统计上是显著的（Mann-Whitney p=0.008，内存单元与非内存单元之间，p=0.55，两内存单元之间，尽管n=5与n=5的样本量不足以排除中等差距）。在LMC中，原子事实定位于主导内存位置，这些位置在组合重读时被访问，而在LMC自身的预编辑正确探针上进行的单行值编辑实现了100%的直接成功，平均意图传播为0.989，同时在约0.1%的时间内移动不相关的事实（待定的基底间特异性匹配）。粗略的保持答案子空间（HOLDANS）互换诊断与这一排序一致，表明区分基底的因素在于编辑事实何时被注入以及之后剩余多少计算以便重用。结果将学习能力与编辑便利性区分开来：重复共享访问足以理解，但编辑传播依赖于基底是否暴露出细粒度的可寻址内存，以便前向计算可以写入并后续重读——这一便利性是循环递归仅部分提供的。
cs.AI / 38 / 2606.20785
Fara-1.5: Scalable Learning Environments for Computer Use Agents
Fara-1.5：可扩展的计算机使用代理学习环境
Abstract
Collecting computer use data from human demonstrations is expensive and slow, motivating the need for scalable generation strategies. This requires two key ingredients: environments in which agents can act and verifiers that can judge whether their demonstrations succeeded. We introduce FaraGen1.5, a scalable data pipeline for computer use agents composed of three modular components: environments, solvers, and verifiers. FaraGen1.5 uses both live websites and synthetic environments that faithfully simulate domains gated by authentication or that require irreversible actions. It employs a solver harness that can be powered by multiple models, including strong frontier models such as GPT-5.4, and also incorporates a user simulator to enable multi-turn rollouts. Finally, FaraGen1.5 scores the resulting trajectories with three complementary verifiers covering task correctness, efficiency, and critical-point adherence. Using data produced by this pipeline, we train Fara1.5, a family of native computer use agents (CUAs) at three scales built on Qwen3.5 (4B, 9B, and 27B). To train these models, we employ a supervised finetuning (SFT) recipe that carefully balances data from FaraGen1.5 for broad coverage, specific high-value tasks, and target model deficiencies in an iterative approach. Each model sets a new state of the art for its size class on browser-use benchmarks: Fara1.5-9B reaches 63.4% on Online-Mind2Web and 86.6% on WebVoyager, while Fara1.5-27B achieves 72.3% on Online-Mind2Web, which is competitive with much larger proprietary systems.
Chinese Translation
从人类示范中收集计算机使用数据既昂贵又缓慢，这促使了可扩展生成策略的需求。这需要两个关键要素：代理可以行动的环境和能够判断其示范是否成功的验证者。我们介绍了 FaraGen1.5，这是一个用于计算机使用代理的可扩展数据管道，由三个模块化组件组成：环境、求解器和验证器。FaraGen1.5 使用实时网站和忠实模拟需要身份验证或不可逆操作的领域的合成环境。它采用一个求解器框架，可以由多个模型驱动，包括强大的前沿模型如 GPT-5.4，并且还结合了用户模拟器以实现多轮展开。最后，FaraGen1.5 使用三个互补的验证器对生成的轨迹进行评分，涵盖任务正确性、效率和关键点遵循。利用该管道生成的数据，我们训练了 Fara1.5，这是基于 Qwen3.5（4B、9B 和 27B）构建的三种规模的本地计算机使用代理（CUAs）系列。为了训练这些模型，我们采用了一种监督微调（SFT）方案，仔细平衡来自 FaraGen1.5 的数据，以实现广泛覆盖、特定高价值任务和目标模型缺陷的迭代方法。每个模型在其规模类别的浏览器使用基准上设定了新的最先进水平：Fara1.5-9B 在 Online-Mind2Web 上达到 63.4%，在 WebVoyager 上达到 86.6%，而 Fara1.5-27B 在 Online-Mind2Web 上达到 72.3%，与更大规模的专有系统竞争。
cs.AI / 39 / 2606.20814
What Shapes Emergent Misalignment? Insights from Training Dynamics, Model Priors, and Data
什么塑造了新兴的错位？来自训练动态、模型先验和数据的见解
Abstract
Emergent misalignment (EM) is a phenomenon in which models generalize with narrow fine-tuning, leading to broad (yet uneven) misalignment across evaluation questions. We study EM and its variability directly through the components of fine-tuning: training dynamics, model priors, and data. (1) We first explored how in-domain training loss relates to out-of-domain alignment scores across datasets and model families. Then, we tried to induce potential alternative local minima through different learning schedules for one narrow fine-tuning, but did not find strong runs with better broad alignment scores conditioned on similar or lower training loss. (2) We found that although the mean and standard deviations of the misaligned model scores are usually statistically different from those of the pre-trained model, there are some potential signals on overall positive correlation. The evaluation prompt-only activations from both the pre-trained and the original instruct models (prior to narrow fine-tuning) could predict fine-grained alignment scores after narrow fine-tuning. (3) Finally, we compared activation deltas before and after narrow fine-tuning and found moderate-to-high subspace overlap and similarity between the resulting activation shifts for training and evaluation prompts. Subspace overlaps between training and evaluation prompt activations correlate with their shifts' similarities when measuring with the last prompt-token activations. The train-evaluation data prompt overlap is controlled against overlap computed from random vectors and evaluation prompts activations.
Chinese Translation
新兴错位（Emergent Misalignment, EM）是一种现象，模型在经过狭窄的微调后进行泛化，导致在评估问题上出现广泛（但不均匀）的错位。我们通过微调的组成部分直接研究了EM及其变异性：训练动态、模型先验和数据。(1) 我们首先探讨了领域内训练损失与跨数据集和模型家族的领域外对齐分数之间的关系。然后，我们尝试通过不同的学习计划诱导潜在的替代局部最小值，以实现一次狭窄的微调，但未能找到在相似或较低训练损失条件下具有更好广泛对齐分数的强运行结果。(2) 我们发现，尽管错位模型分数的均值和标准差通常在统计上与预训练模型的分数不同，但存在一些潜在信号表明整体正相关。来自预训练模型和原始指令模型（在狭窄微调之前）的评估提示激活可以预测狭窄微调后的细粒度对齐分数。(3) 最后，我们比较了狭窄微调前后的激活变化，发现训练和评估提示的激活变化之间存在中等到高的子空间重叠和相似性。当使用最后一个提示令牌激活进行测量时，训练和评估提示激活之间的子空间重叠与它们变化的相似性相关。训练-评估数据提示的重叠与从随机向量和评估提示激活计算的重叠进行了控制。
cs.AI / 40 / 2606.20837
Study on Quantitative Dynamic Epistemic Logic for Belief Revision
关于信念修正的定量动态认知逻辑研究
Abstract
Belief revision is a process in which an agent begins to believe in something she previously did not. I begin the paper by presenting, based on (G\"ardenfors, 1998; Hansson, 1999), postulates for belief revision that constitute the basis of the AGM theory. I will then briefly show the semantics of a modal logic introduced in (van Ditmarsch, 2005), which I call `$P$'. This logic formalizes static epistemic states and has greater expressive power than AGM in doing so because it captures the quantitative notion of "degrees of conviction". The third step is to introduce revision operators on $P$ and, mostly following (van Ditmarsch, 2005), obtain the Dynamic Epistemic Logic (DEL) I call `$P*$'. It models processes of belief revision in several ways. Original results are presented in the following two sections. The first one of these sections revolves around a formalization of AGM postulates within $P*$ by proving some theorems related to the satisfaction of those postulates by revisions defined in $P*$. The last section features an analysis of $P*$'s revisions that go beyond the mere satisfaction of postulates. I compare their formal behavior with respect to some philosophical criteria. At last, I conclude that the functions presented in (van Ditmarsch, 2005) are not good formalizations of the philosophical intuition behind AGM. Instead, it is captured by the function $*^0$ originally defined in this paper (but highly inspired by (van Benthem, 2007)). An implementation of this function is also provided.
Chinese Translation
信念修正是一个过程，在这个过程中，代理开始相信她之前并不相信的事情。本文首先基于(G"ardenfors, 1998; Hansson, 1999)提出信念修正的公设，这些公设构成了AGM理论的基础。接下来，我将简要展示在(van Ditmarsch, 2005)中引入的模态逻辑的语义，我称之为`$P$'。该逻辑形式化了静态认知状态，并且在此过程中具有比AGM更强的表达能力，因为它捕捉了“信念程度”的定量概念。第三步是引入在$P$上的修正算子，并主要遵循(van Ditmarsch, 2005)，获得我称之为`$P*$'的动态认知逻辑(DEL)。它以多种方式建模信念修正过程。原始结果将在接下来的两个部分中呈现。这些部分的第一部分围绕在$P*$中对AGM公设的形式化展开，通过证明与$P*$中定义的修正满足这些公设相关的一些定理。最后一部分分析了$P*$的修正，这些修正超出了单纯满足公设的范畴。我将它们的形式行为与一些哲学标准进行比较。最后，我得出结论，(van Ditmarsch, 2005)中提出的函数并不是AGM背后哲学直觉的良好形式化。相反，这种直觉由本文中最初定义的函数$*^0$捕捉（但深受(van Benthem, 2007)的启发）。本文还提供了该函数的实现。
cs.AI / 41 / 2606.20839
Process-Reward Tactic Evolution for Long-Horizon Bioinformatics Workflows
长时间跨度生物信息学工作流的过程-奖励策略演化
Abstract
LLM agents can write code and call tools, but reliable bioinformatics work requires long-horizon interaction with workflow software, typed data objects, provenance, and biological checks. We study this setting through Galaxy workflow execution. The agent must explore task data, construct or adapt an executable workflow DAG, bind inputs and dataset collections, monitor execution, debug failures, and validate biological outputs. We propose Process-Reward Tactic Evolution, a Galaxy-based training framework that turns verified workflow rollouts into reusable \tactics. During training, agents practice on curriculum-organized Galaxy tasks in Agent Gym; process verifiers score workflow construction, software interaction, execution, and biological correctness; successful and failed traces are distilled into a tactic library. At inference, the trained executor, Process-Reward Tactic Evolution, uses this library to execute held-out peer reviewed Galaxy workflow converted BioWorkflow Bench and BioAgent Bench tasks in isolated environments. The paper evaluates whether process-supervised tactic accumulation improves long-horizon bioinformatics workflow completion, biological correctness, and execution efficiency over no-memory and reflection-style baselines.
Chinese Translation
大型语言模型（LLM）代理能够编写代码并调用工具，但可靠的生物信息学工作需要与工作流软件、类型化数据对象、来源信息和生物学检查进行长时间的交互。我们通过Galaxy工作流执行研究这一环境。代理必须探索任务数据，构建或调整可执行的工作流有向无环图（DAG），绑定输入和数据集集合，监控执行，调试失败，并验证生物学输出。我们提出了过程-奖励策略演化（Process-Reward Tactic Evolution），这是一个基于Galaxy的训练框架，将经过验证的工作流实施转化为可重用的策略。在训练过程中，代理在Agent Gym中对课程组织的Galaxy任务进行练习；过程验证器对工作流构建、软件交互、执行和生物学正确性进行评分；成功和失败的执行轨迹被提炼成策略库。在推理阶段，经过训练的执行器过程-奖励策略演化利用该库在隔离环境中执行保留的同行评审Galaxy工作流转换的BioWorkflow Bench和BioAgent Bench任务。本文评估了过程监督的策略积累是否能提高长时间跨度生物信息学工作流的完成率、生物学正确性和执行效率，相较于无记忆和反思式基线。
cs.AI / 42 / 2606.20857
SignVLA: Real-Time Sign Language-Guided Robotic Manipulation via Attention LSTM and Vision-Language-Action Models
SignVLA：基于注意力LSTM和视觉-语言-动作模型的实时手语引导机器人操作
Abstract
Vision-Language-Action (VLA) models enable robots to execute manipulation tasks from natural-language instructions grounded in visual observations. However, existing VLA interfaces primarily rely on speech or text input, limiting accessibility for deaf, hard-of-hearing, and speech-impaired users. We present SignVLA, a real-time sign-language-guided VLA framework for accessible human-robot interaction. The system introduces a modular sign-to-text interface that converts visual sign gestures into semantic instructions compatible with downstream VLA policies. Given video streams, SignVLA extracts hand landmark features and employs an attention-enhanced Long Short-Term Memory (LSTM) network to capture temporal gesture dynamics for alphabet- and command-level sign recognition. A temporal stabilization module further improves prediction consistency in real-time interaction settings.The generated instruction sequence is then passed to a downstream VLA policy for sign-conditioned robotic manipulation. Experimental results demonstrate stable real-time sign recognition and successful execution of manipulation tasks driven by sign-language inputs. Our findings suggest that lightweight temporal sign recognition can serve as an effective and practical accessibility layer for multimodal embodied intelligence.
Chinese Translation
视觉-语言-动作（VLA）模型使机器人能够根据基于视觉观察的自然语言指令执行操作任务。然而，现有的VLA接口主要依赖语音或文本输入，这限制了聋人、听力障碍者和语言障碍用户的可及性。我们提出了SignVLA，一种实时手语引导的VLA框架，旨在实现可访问的人机交互。该系统引入了一个模块化的手语转文本接口，将视觉手势转换为与下游VLA策略兼容的语义指令。在给定视频流的情况下，SignVLA提取手部关键点特征，并采用增强注意力的长短期记忆（LSTM）网络捕捉手势的时间动态，以实现字母和命令级别的手语识别。一个时间稳定模块进一步提高了实时交互环境中的预测一致性。生成的指令序列随后被传递给下游VLA策略，以实现基于手语的机器人操作。实验结果表明，实时手语识别稳定且成功执行了由手语输入驱动的操作任务。我们的研究结果表明，轻量级的时间手语识别可以作为多模态具身智能的有效且实用的可及性层。
cs.AI / 43 / 2606.20881
When Do Intrinsic Rewards Work for Code Reasoning? A Comprehensive Study
内在奖励何时对代码推理有效？一项综合研究
Abstract
Reinforcement learning with verifiable rewards (RLVR) has driven substantial progress in large language model reasoning, but relies on ground-truth supervision that is costly or infeasible, especially in coding tasks. Recent work addresses this by deriving rewards from a model's own signals, such as majority voting or confidence-based scores, achieving notable success on mathematical reasoning benchmarks. However, code generation poses distinct challenges: programs are structurally complex, semantically equivalent solutions may differ syntactically, and verification typically requires execution. Whether these intrinsic reward methods transfer effectively to code remains unexplored. In this work, we present a systematic empirical study of intrinsic reward methods for code generation. We conduct extensive experiments on LiveCodeBench, systematically evaluating representative certainty-based Reinforcement Learning from Internal Feedback (RLIF) approaches under different training scenarios and hyperparameter settings. Our experiments reveal that certainty-based methods yield early gains but inevitably collapse: models progressively shorten outputs and lose reasoning capability, with collapse speed sensitive to sample size and temperature. When used to initialize RLVR training, RLIF pre-training offers no significant improvement over training from scratch. We also provide actionable recommendations for using intrinsic rewards for training code reasoning models. Our study shows both the promise and limitations of intrinsic reward methods for code, informing future work on code models and agents.
Chinese Translation
带有可验证奖励的强化学习（RLVR）在大型语言模型推理方面取得了显著进展，但依赖于昂贵或不可行的真实监督，尤其是在编码任务中。近期的研究通过从模型自身信号中推导奖励，例如多数投票或基于置信度的分数，来解决这一问题，在数学推理基准上取得了显著成功。然而，代码生成面临着独特的挑战：程序结构复杂，语义等价的解决方案在语法上可能不同，且验证通常需要执行。内在奖励方法是否能有效转移到代码生成上尚未被探索。在本研究中，我们对代码生成的内在奖励方法进行了系统的实证研究。我们在 LiveCodeBench 上进行了广泛的实验，系统评估了不同训练场景和超参数设置下具有代表性的基于置信度的内部反馈强化学习（RLIF）方法。我们的实验表明，基于置信度的方法在早期获得了收益，但不可避免地崩溃：模型逐渐缩短输出并失去推理能力，崩溃速度对样本大小和温度敏感。当用于初始化 RLVR 训练时，RLIF 预训练并未显著改善从头开始训练的效果。我们还提供了关于如何使用内在奖励训练代码推理模型的可行建议。我们的研究展示了内在奖励方法在代码生成中的潜力与局限性，为未来的代码模型和代理研究提供了参考。
cs.AI / 44 / 2606.20895
Neurosymbolic Clinical Trial Matching via LLM-Driven Abduction and Logical Verification
基于大语言模型驱动的神经符号临床试验匹配：归纳推理与逻辑验证
Abstract
Large Language Models (LLMs) offer a promising path to automate Clinical Trial Matching (CTM), but still struggle with the deterministic verification required for complex eligibility criteria. Conversely, purely symbolic methods provide formal rigour but break down when faced with incomplete patient records and noisy clinical evidence. To bridge this gap, we investigate a hybrid framework for CTM combining LLMs with logical verification. In particular, we introduce an abductive neurosymbolic CTM framework ({\alpha}NeSy-CTM), which leverages the linguistic and world knowledge in LLMs to support reasoning over noisy and underspecified clinical text. Extensive evaluation demonstrates that {\alpha}NeSy-CTM substantially outperforms standalone LLM baselines, achieving up to 30% relative improvement over zero-shot baselines. In addition, our analyses confirm the impact of abductive reasoning on CTM, with {\alpha}NeSy-CTM exhibiting improved accuracy, specificity, and robustness over a non-abductive neurosymbolic setting. Furthermore, {\alpha}NeSy-CTM and Chain-of-Thought (CoT) reasoning prove highly complementary, highlighting the potential for a hybrid routing policy. Ultimately, this paper demonstrates the impact of neurosymbolic methods for automating CTM, providing a path toward the next generation of auditable, LLM-driven clinical applications.
Chinese Translation
大型语言模型（LLMs）为自动化临床试验匹配（CTM）提供了一条有前景的路径，但在处理复杂的资格标准所需的确定性验证时仍然面临挑战。相反，纯符号方法虽然提供了形式上的严谨性，但在面对不完整的病历和嘈杂的临床证据时则会失效。为了解决这一问题，我们研究了一种结合LLMs与逻辑验证的混合框架用于CTM。特别地，我们引入了一种归纳推理的神经符号CTM框架（{eta}NeSy-CTM），该框架利用LLMs中的语言和世界知识来支持对嘈杂和不明确的临床文本进行推理。广泛的评估表明，{eta}NeSy-CTM在性能上显著优于独立的LLM基线，在零样本基线的基础上实现了高达30%的相对提升。此外，我们的分析确认了归纳推理对CTM的影响，{eta}NeSy-CTM在准确性、特异性和鲁棒性方面优于非归纳推理的神经符号设置。此外，{eta}NeSy-CTM与链式思维（Chain-of-Thought, CoT）推理高度互补，突显了混合路由策略的潜力。最终，本文展示了神经符号方法在自动化CTM中的影响，为下一代可审计的、基于LLM的临床应用提供了路径。
cs.AI / 45 / 2606.20912
Root Cause Analysis with Latent Confounders using Partial Ancestral Graphs
使用部分祖先图的潜在混杂因素进行根本原因分析
Abstract
Finding the source of failures, known as Root Cause Analysis (RCA), is essential for identifying the root causes of anomalies and maintaining the reliability of complex systems. While causal theory has advanced data-driven RCA, existing frameworks assume causal sufficiency, failing to account for the unobserved latent variables prevalent in real-world environments. To address this gap, we propose PAG-RCA. This framework models system failures as parametric interventions over Partial Ancestral Graphs (PAGs) to perform RCA in the presence of latent variables. We use standard causal identification algorithms to find the source of failures by quantifying causal effects over the PAG. When an effect is identifiable, candidate root causes are ranked based on their exact intervention effects. When effects are structurally unidentifiable, our framework (for the first time in the RCA literature) integrates partial identification to evaluate and score candidates using analytical causal bounds. By integrating latent variables and partial identification at once our framework ensures robust RCA even under data scarcity and latent-variable scenarios where traditional methods degrade. Evaluations on synthetic data, microservice anomaly benchmarks and power-grid cascading failures demonstrate that PAG-RCA consistently outperforms state-of-the-art data-driven baselines. By improving data-driven RCA performance under data scarcity, this methodology advances reliable automated diagnostics in partially observable complex networks.
Chinese Translation
寻找故障源头，即根本原因分析（Root Cause Analysis, RCA），对于识别异常的根本原因和维护复杂系统的可靠性至关重要。尽管因果理论推动了基于数据的RCA，现有框架假设因果充分性，未能考虑现实环境中普遍存在的未观察到的潜在变量。为了解决这一问题，我们提出了PAG-RCA框架。该框架将系统故障建模为对部分祖先图（Partial Ancestral Graphs, PAGs）的参数干预，以在存在潜在变量的情况下执行RCA。我们使用标准因果识别算法通过量化PAG上的因果效应来寻找故障源。当效应可识别时，候选根本原因根据其确切的干预效应进行排名。当效应在结构上不可识别时，我们的框架（在RCA文献中首次）整合了部分识别，以使用分析性因果界限评估和评分候选者。通过同时整合潜在变量和部分识别，我们的框架确保在数据稀缺和潜在变量情境下仍能进行稳健的RCA，而传统方法在这些情况下表现不佳。在合成数据、微服务异常基准和电网级联故障的评估中，PAG-RCA始终优于最先进的基于数据的基线。通过在数据稀缺情况下提高基于数据的RCA性能，该方法推动了在部分可观察复杂网络中可靠的自动诊断。
cs.AI / 46 / 2606.20950
Power Systems Agent Benchmark: Executable Evaluation of AI Agents in Electric Power Engineering
电力系统代理基准：电力工程中人工智能代理的可执行评估
Abstract
Executable evaluation -- checking the consequences of an agent's actions with a program rather than grading its prose -- has become a prominent way to assess tool-using AI agents in software settings. Electric power engineering has not yet had an analogous benchmark: language-model use is still dominated by retrieval and text question answering, while agents acting on power-system artifacts remain mostly academic prototypes. We introduce the Power Systems Agent Benchmark, an executable benchmark for power-engineering agents. An agent receives a structured task and returns a structured solution; a deterministic evaluator recomputes the engineering quantities, checks operational constraints, and returns a feasibility flag, a normalized score, and explicit violations. The benchmark contains 41 task families across eight areas of power engineering, from power flow and protection to stability, microgrids, reliability, power quality, and forecasting. Each task is grounded in a citable source, standard, or documented engineering formulation. To resist contamination, held-out cases are synthesized on demand by per-family generators from private seeds: the construction is inspectable, but the instances remain private. In a reference evaluation with three command-line agents, the strongest score near the compact tier's ceiling, a smaller open model trails, and public and held-out performance are broadly consistent; a separate public-split grid with OpenCode and Aider probes harness effects. The reference evaluation doubles as quality control: unanimous failures flag candidate task or evaluator defects, and it exposed a latent evaluator bug missed by self-consistency checks. The evaluators are compact deterministic surrogates, but the task contract allows their internals to be upgraded to simulator-backed checks without changing how tasks are posed or solved.
Chinese Translation
可执行评估——通过程序检查代理行为的后果，而不是评估其文字表达——已成为在软件环境中评估工具使用人工智能代理的重要方式。电力工程尚未有类似的基准：语言模型的使用仍然主要集中在检索和文本问答上，而在电力系统工件上行动的代理大多仍然是学术原型。我们提出了电力系统代理基准，这是一个针对电力工程代理的可执行基准。代理接收一个结构化任务并返回一个结构化解决方案；一个确定性评估器重新计算工程量，检查操作约束，并返回可行性标志、标准化分数和明确的违规情况。该基准包含41个任务系列，涵盖电力工程的八个领域，从潮流和保护到稳定性、微电网、可靠性、电能质量和预测。每个任务都基于可引用的来源、标准或文档化的工程公式。为了抵御污染，保留的案例由来自私有种子的每个系列生成器按需合成：构造过程是可检查的，但实例保持私密。在与三个命令行代理的参考评估中，最强的分数接近紧凑层的上限，而一个较小的开放模型落后，公共和保留性能大致一致；一个单独的公共拆分网格与OpenCode和Aider探针结合利用了这些效果。参考评估还作为质量控制：一致的失败标记候选任务或评估器缺陷，并揭示了一个被自一致性检查遗漏的潜在评估器错误。这些评估器是紧凑的确定性替代品，但任务合同允许其内部结构在不改变任务提出或解决方式的情况下升级为基于模拟器的检查。
cs.AI / 47 / 2606.20963
Generative Responsible AI Data Evaluation Schema (GRAIDES) for AI Assurance in Local Government
用于地方政府AI保障的生成性负责任AI数据评估框架（GRAIDES）
Abstract
Trust in the application of generative Artificial Intelligence (AI) relies on well-governed measurable evidence of performance and safety. In practice, however, evaluation data is often fragmented across systems, inconsistently structured and difficult to compare. We introduce the Generative Responsible AI Data Evaluation Schema (GRAIDES) as a lightweight open-source data model for centralising AI observability across popular vendors. Practical blueprints for code, architecture and statistical evaluation are shared as guidance about how to approach generative system assurance at the organisational level. Illustrative case study results are reported from Westminster City Council's AI catalogue with a focus on measuring human-model alignment including detecting systematic disagreement between evaluators. By framing evaluations as a data modelling problem, GRAIDES provides a practical pathway toward more consistent and reproducible benchmarking, tuning and assurance activities for generative AI systems.
Chinese Translation
对生成性人工智能（AI）应用的信任依赖于良好治理的可测量的性能和安全性证据。然而，在实践中，评估数据往往在系统之间碎片化，结构不一致且难以比较。我们提出了生成性负责任AI数据评估框架（GRAIDES），作为一个轻量级的开源数据模型，用于集中管理各大供应商的AI可观察性。我们分享了代码、架构和统计评估的实用蓝图，以指导如何在组织层面进行生成系统的保障。我们报告了来自威斯敏斯特市议会AI目录的案例研究结果，重点测量人类与模型之间的一致性，包括检测评估者之间的系统性分歧。通过将评估框架化为数据建模问题，GRAIDES为生成性AI系统提供了更一致和可重复的基准测试、调优和保障活动的实用路径。
cs.AI / 48 / 2606.20969
AutoACSL: Synthesizing ACSL Specifications by Integrating LLMs with CPG-Based Static Analysis
AutoACSL：通过将大型语言模型与基于代码属性图的静态分析相结合，合成ACSL规范
Abstract
Generating formal specifications for C programs remains a challenge in formal verification due to the manual effort, expertise, and semantic precision required. While recent advancements in large language models (LLMs) offer promise in automating specification synthesis, current approaches often lack semantic depth and produce unverifiable or incomplete contracts. To address these limitations, we introduce AutoACSL, a novel framework that integrates LLM prompting with semantic features extracted from Code Property Graphs (CPGs). AutoACSL performs static analyses to extract key semantic elements, including arithmetic operations, loop and recursion structures, and return value propagation, which are encoded into structured prompts. These prompts enable the LLM not only to generate normal behavioral specifications but also to include constraints that prevent inputs leading to runtime errors. AutoACSL employs a feedback-driven synthesis loop, where candidate specifications are verified using Frama-C/WP and refined iteratively until verification succeeds or a termination condition is met. Evaluated on 604 programs drawn from diverse datasets, AutoACSL achieves a 98% specification generation success ratio and a 96% full proof ratio when paired with Gemini-3. Compared to a code-only baseline, AutoACSL improves the full proof ratio by 24.7% to 51.7% across four LLMs (GPT-o4 Mini, GPT-5.2, Grok-4.1, and Gemini-3), demonstrating that integrating large language models with CPG-based static analysis substantially enhances both generation robustness and verification effectiveness for automated ACSL specification synthesis.
Chinese Translation
为C程序生成形式化规范在形式验证中仍然是一项挑战，因为这需要手动努力、专业知识和语义精确性。尽管最近大型语言模型（LLMs）的进展在自动化规范合成方面展现了希望，但当前的方法往往缺乏语义深度，生成的合同不可验证或不完整。为了解决这些局限性，我们提出了AutoACSL，这是一个新颖的框架，将LLM提示与从代码属性图（CPGs）中提取的语义特征相结合。AutoACSL执行静态分析以提取关键语义元素，包括算术操作、循环和递归结构以及返回值传播，这些元素被编码为结构化提示。这些提示使得LLM不仅能够生成正常行为的规范，还能够包含防止导致运行时错误的输入的约束。AutoACSL采用反馈驱动的合成循环，其中候选规范通过Frama-C/WP进行验证，并迭代地进行精炼，直到验证成功或满足终止条件。在从不同数据集中抽取的604个程序上进行评估时，AutoACSL在与Gemini-3配对时实现了98%的规范生成成功率和96%的完整证明率。与仅基于代码的基线相比，AutoACSL在四个LLM（GPT-o4 Mini、GPT-5.2、Grok-4.1和Gemini-3）中将完整证明率提高了24.7%至51.7%，证明了将大型语言模型与基于CPG的静态分析相结合显著增强了自动化ACSL规范合成的生成稳健性和验证有效性。
cs.AI / 49 / 2606.20978
How Should Agents Read Demonstrations? Hierarchical Structure Beats Flat Action Logs
代理应该如何阅读演示？层次结构优于平面动作日志
Abstract
Programming by Demonstration (PbD) offers a human-centered way to author procedural knowledge for LLM agents: users communicate what they want by showing rather than by writing prompts or code, making agent authoring accessible to non-programmers. The natural output of a PbD recording is a flat action log, but how this log is organized before being passed to the agent is an open design question with significant consequences for plan quality. We propose grouping recorded actions into labeled, hierarchical subgoals and evaluate the effect of this organizational structure in a controlled experiment. Across 85 web automation tasks, we compare a zero-shot baseline against four demonstration formats that share identical action sequences but differ in structure. On 43 natural-language tasks with vague descriptions, hierarchically grouped demonstrations improve pass rates from 76.7\% to 90.7\% (paired permutation test $p{=}0.034$; win-loss 6:0), while flat demonstrations show a smaller, non-significant improvement. On 42 tasks with precise descriptions, no format provides any benefit, confirming that the hierarchical advantage arises specifically when descriptions leave procedural details ambiguous. Ablation shows that subgoal grouping alone drives the effect: preconditions, postconditions, and parameter annotations add no measurable benefit. These results offer a concrete design recommendation for PbD pipelines and, more broadly, for any system that feeds procedural context to an LLM agent: segment action sequences into named subgoal groups rather than presenting flat step lists.
Chinese Translation
通过演示编程（PbD）为大型语言模型（LLM）代理提供了一种以人为中心的方式来创建程序知识：用户通过展示而不是编写提示或代码来传达他们的需求，使得代理的创建对非程序员变得可访问。PbD记录的自然输出是一个平面动作日志，但在将该日志传递给代理之前如何组织这个日志是一个开放的设计问题，对计划质量有显著影响。我们建议将记录的动作分组为带标签的层次子目标，并在一个受控实验中评估这种组织结构的效果。在85个网页自动化任务中，我们将零样本基线与四种演示格式进行比较，这些格式共享相同的动作序列，但在结构上有所不同。在43个描述模糊的自然语言任务中，层次分组的演示将通过率从76.7\%提高到90.7\%（配对置换检验 $p{=}0.034$；胜负比6:0），而平面演示的改善较小且不显著。在42个描述精确的任务中，没有任何格式提供任何好处，确认层次优势特别出现在描述留下程序细节模糊的情况下。消融实验表明，仅子目标分组驱动了这一效果：前置条件、后置条件和参数注释没有增加可测量的好处。这些结果为PbD管道提供了具体的设计建议，更广泛地适用于任何向LLM代理提供程序上下文的系统：将动作序列分段为命名的子目标组，而不是呈现平面步骤列表。
cs.AI / 50 / 2606.20997
BioInsight: Multi-Agent Orchestration for Interactive Biomedical Knowledge Discovery
BioInsight：用于交互式生物医学知识发现的多智能体协调
Abstract
Biomedical researchers increasingly use AI-generated analyses and reports to interpret protein-level signals, but static outputs are often insufficient for research decision-making, where users need to inspect evidence, assess uncertainty, compare mechanisms, and refine hypotheses. We present \textsc{BioInsight}, a multi-agent system that moves from static biomedical report generation to interactive evidence-centered interactive interface generation. Given a disease name, a protein association table, and optional cohort metadata, BioInsight organizes disease-specific evidence through typed intermediate artifacts, including ranked pathways, literature evidence packets, protein-level reasoning notes, citation-grounded reports, dashboard schemas, and rendered interactive interfaces. The system decomposes evidence retrieval from mechanistic reasoning, normalizes citations through deterministic components, and converts the same structured evidence used in the report into an interactive interface. We evaluate BioInsight on standardized biomedical QA, challenging protein-function reasoning, and end-to-end biomedical evidence synthesis. Results show that BioInsight achieves best, and suggest that biomedical AI systems should move beyond text-only and static reports toward provenance-preserving, interactive evidence artifacts.
Chinese Translation
生物医学研究人员越来越多地使用人工智能生成的分析和报告来解释蛋白质水平的信号，但静态输出通常不足以支持研究决策，用户需要检查证据、评估不确定性、比较机制并完善假设。我们提出了 extsc{BioInsight}，一个多智能体系统，它将静态生物医学报告生成转变为交互式以证据为中心的交互界面生成。给定疾病名称、蛋白质关联表和可选的队列元数据，BioInsight通过类型化的中间工件组织特定于疾病的证据，包括排名的通路、文献证据包、蛋白质水平的推理笔记、基于引用的报告、仪表板架构和渲染的交互界面。该系统将证据检索与机制推理分解，通过确定性组件规范化引用，并将报告中使用的相同结构化证据转换为交互界面。我们在标准化的生物医学问答、具有挑战性的蛋白质功能推理和端到端生物医学证据综合上评估了BioInsight。结果表明，BioInsight达到了最佳效果，并建议生物医学人工智能系统应超越仅文本和静态报告，朝着保留来源的交互式证据工件发展。
cs.AI / 51 / 2606.21005
Building Agent Harnesses for Scientific Curation from Multimodal Sources
构建用于从多模态来源进行科学整理的代理工具
Abstract
Scientific discovery workflows often depend on structured curation from the literature. This is difficult for current agents because the key evidence is scattered across long text, dense tables, and figures, and the final records often require reasoning across multiple evidence fragments rather than copying a single span. We study scientific curation from multimodal sources and introduce Beaver, an agent harness that extracts structured information from scientific papers while preserving provenance to the supporting evidence. Beaver combines a frontier agent with multimodal evidence tooling, task scaffolding, and artifact-grounded autoresearch. These components turn curation into a staged, auditable workflow and enable an iterative evaluate--diagnose--revise loop, where persistent run artifacts expose stage-localized failures and guide harness updates. Experiments show that Beaver reaches 81.0 on Gold-Referenced Attribute Score (GRAS), an attribute-level measure of agreement with gold curated records, outperforming frontier agents by over 23 absolute points. Ablations show that task scaffolding, multimodal evidence tooling, and provenance traces each contribute meaningfully to performance, while attribute-level analysis shows the largest gains on high-value attributes that require cross-modal reasoning and normalization. These results show that, for scientific curation from papers with multimodal evidence, harness design is a central determinant of agent performance.
Chinese Translation
科学发现工作流程通常依赖于来自文献的结构化整理。然而，对于当前的代理而言，这一过程十分困难，因为关键证据分散在冗长的文本、密集的表格和图形中，最终记录往往需要跨多个证据片段进行推理，而不是简单地复制单一的文本片段。我们研究了来自多模态来源的科学整理，并引入了Beaver，一个能够从科学论文中提取结构化信息的代理工具，同时保留对支持证据的来源追溯。Beaver结合了前沿代理、多模态证据工具、任务支架和基于文献的自研究。这些组件将整理转变为一个分阶段、可审计的工作流程，并实现了一个迭代的评估-诊断-修订循环，其中持久的运行文档揭示了阶段性局部失败并指导工具的更新。实验结果表明，Beaver在金标准属性评分（Gold-Referenced Attribute Score, GRAS）上达到了81.0，这是与金标准整理记录的一致性属性级别度量，超越了前沿代理超过23个绝对点。消融实验显示，任务支架、多模态证据工具和来源追溯对性能的贡献均显著，而属性级别分析显示在需要跨模态推理和标准化的高价值属性上获得了最大的提升。这些结果表明，对于来自具有多模态证据的论文的科学整理，工具设计是代理性能的一个核心决定因素。
cs.AI / 52 / 2606.21010
Closure of Self-Determining System Based on Causal and Constitutive Relations
基于因果关系和构成关系的自决系统的闭合
Abstract
A self-determining system is defined as one in which causes originating within the system influence the system itself. This definition raises the question of how to specify system boundaries. Although the concept of "closure" is commonly used for this purpose, defining boundaries solely in terms of causal relations introduce challenges, such as how to handle external causes and circular causality. To address this issue, we introduce two types of asymmetric relations: causal and constitutive. We propose that system boundaries can be defined as closures of loops formed by these relations, referred to as causal-constitutive loops. By constraining constitutive relations, the resulting system necessarily includes internal causes and thereby satisfies self-determination. Furthermore, to prevent reduction to supervenience, constitutive relations must involve at least two independent variables. This minimal requirement leads to two interdependent loops, which implies a dual-process organization.
Chinese Translation
自决系统被定义为其内部产生的原因影响系统自身的系统。这个定义引发了如何确定系统边界的问题。尽管“闭合”这一概念通常用于此目的，但仅仅根据因果关系来定义边界会带来挑战，例如如何处理外部原因和循环因果关系。为了解决这个问题，我们引入了两种类型的不对称关系：因果关系和构成关系。我们提出，系统边界可以定义为由这些关系形成的闭合循环，称为因果-构成循环。通过限制构成关系，所得到的系统必然包括内部原因，从而满足自决性。此外，为了防止简化为依赖性，构成关系必须涉及至少两个独立变量。这个最小要求导致两个相互依赖的循环，这暗示了一种双重过程的组织结构。
cs.AI / 53 / 2606.21013
Agentic Time Machine as an Infrastructure for Future-Event Forecasting
代理时间机器作为未来事件预测的基础设施
Abstract
Forecasting future events is a critical challenge for large language model (LLM) agents, spanning domains from elections and monetary policy to financial markets. However, evaluating progress on this task presents a fundamental trade-off between efficiency and environment fidelity. While live evaluation benchmarks suffer from an inherently slow feedback loop, existing retrospective replays typically restrict agents to static, pre-frozen databases that sacrifice the environmental realism of actual deployments. To tackle this issue, we introduce Agentic Time Machine (TM), an infrastructure that approximately reconstructs the web state at any chosen past time by filtering post-cutoff content. Leveraging this evaluation infrastructure, we further propose a planner-solver-aggregator multi-agent framework that breaks each question into diverse analytical angles, gathers evidence in parallel, and combines the results into a single forecast. Experiments show that offline scores under TM correlate strongly with live FutureX scores, validating that TM offers a fast and reliable sandbox for forecasting-agent evaluation. On FutureX-Past and Polymarket evaluated under TM, our framework achieves the highest score among strong closed-book, tool-augmented, and self-consistency baselines. On the official FutureX live leaderboard, our system achieves the best average rank over four consecutive weeks, including 1st place in May Week 1. As of June 17, it also ranks 1st on FutureX's official eight-week overall leaderboard.
Chinese Translation
预测未来事件是大型语言模型（LLM）代理面临的一项关键挑战，涵盖从选举和货币政策到金融市场等多个领域。然而，评估这一任务的进展在效率与环境真实感之间存在根本的权衡。虽然实时评估基准由于反馈循环固有的缓慢而受到影响，但现有的回顾性重放通常限制代理使用静态的、预先冻结的数据库，从而牺牲了实际部署的环境真实感。为了解决这个问题，我们引入了代理时间机器（Agentic Time Machine，TM），这是一种基础设施，通过过滤截止后内容，近似重建任何选定过去时间的网络状态。利用这一评估基础设施，我们进一步提出了一种规划者-求解者-聚合器多代理框架，该框架将每个问题分解为多种分析角度，平行收集证据，并将结果合并为单一预测。实验表明，在TM下的离线评分与实时FutureX评分高度相关，验证了TM为预测代理评估提供了一个快速且可靠的沙箱。在TM下评估的FutureX-Past和Polymarket中，我们的框架在强大的闭卷、工具增强和自一致性基准中取得了最高分。在官方FutureX实时排行榜上，我们的系统在连续四周内获得了最佳平均排名，包括在五月第一周获得第一名。截至6月17日，它在FutureX的官方八周总体排行榜上也排名第一。
cs.AI / 54 / 2606.21015
The AI Evaluability Gap: The Missing Layer for Managing Risk and Sustaining Value
人工智能可评估性缺口：管理风险和维持价值的缺失层面
Abstract
Organizations deploying AI face two fundamental governance challenges: managing AI risk and sustaining AI value. Both depend on evidence whose sufficiency cannot be taken for granted. We call the shared underlying challenge the AI Evaluability Gap: the condition in which organizations lack sufficient evidence to support high-confidence governance decisions regarding either risk or value. We argue that this gap reflects a category error in current practice. Existing governance approaches focus primarily on properties of systems, such as safety, fairness, reliability, compliance, and value, while paying comparatively little attention to the evidentiary foundations required to justify decisions about those properties. We further argue that AI governance encompasses both operational decisions regarding whether a system may operate and investment decisions regarding whether it merits continued organizational resources. To address this problem, we introduce Evaluability, defined as the capability of a system to generate, maintain, and renew evidence sufficient to support high-confidence governance decisions over time. We formalize governance decisions as functions of calibrated confidence Conf(D|E) and identify six properties of evaluable evidence: observability, attributability, intervenability, verifiability, calibration, and temporal validity. The framework distinguishes Operational Certification, which relies primarily on structural evidence to justify deployment decisions, from Investment Certification, which relies primarily on causal evidence to justify continued resource allocation. We argue that evidence sufficiency is a missing layer of AI governance and that closing the AI Evaluability Gap is a prerequisite for both managing risk and sustaining value in AI-enabled organizations.
Chinese Translation
部署人工智能的组织面临两个基本的治理挑战：管理人工智能风险和维持人工智能价值。这两者都依赖于证据，而这种证据的充分性不能被视为理所当然。我们将这一共享的基础挑战称为人工智能可评估性缺口（AI Evaluability Gap）：即组织缺乏足够的证据来支持关于风险或价值的高信心治理决策的状态。我们认为，这一缺口反映了当前实践中的一种类别错误。现有的治理方法主要关注系统的属性，如安全性、公平性、可靠性、合规性和价值，而对证明这些属性决策所需的证据基础关注相对较少。我们进一步认为，人工智能治理包括关于系统是否可以运行的操作决策和关于是否值得继续投入组织资源的投资决策。为了解决这个问题，我们引入了可评估性（Evaluability），定义为一个系统生成、维护和更新足够证据的能力，以支持高信心的治理决策。我们将治理决策形式化为经过校准的信心函数 Conf(D|E)，并识别出可评估证据的六个属性：可观察性、可归因性、可干预性、可验证性、校准性和时间有效性。该框架区分了操作认证（Operational Certification），主要依赖结构性证据来证明部署决策的合理性，和投资认证（Investment Certification），主要依赖因果证据来证明持续资源分配的合理性。我们认为，证据的充分性是人工智能治理的缺失层面，弥补人工智能可评估性缺口是管理风险和维持人工智能驱动组织价值的前提。
cs.AI / 55 / 2606.21024
Negative Knowledge as Failure-aware Shared Memory for AutoResearch
负知识作为面向失败的共享记忆用于自动研究
Abstract
AI-assisted research systems generate many failed attempts, but those failures rarely become a durable, shared knowledge asset. We propose a negative knowledge memory layer: a curator agent converts each failed attempt into a bounded, typed record in a shared bank, and a downstream research agent explicitly adopts or rejects those records before proposing its next experiment. We evaluate this layer in two settings: same-task retry on ScienceAgentBench and cross-task scientific research on two nonlinear math-physics PDE problems. The negative knowledge layer outperforms vanilla AutoResearch baselines while using fewer tokens; agents with the negative knowledge bank solve new tasks that all baselines fail to solve in PDE systems research. We also show that the previous negative knowledge bank can transfer and enhance AutoResearch on different PDE problems. These results suggest that structured negative knowledge is a knowledge asset that should be explicitly maintained in broader AI-engaged scientific research beyond a memory-compression or debugging aid, alongside positive findings, as a collective infrastructure for scientific memory. Code is available at https://github.com/hch-wang/Negative_Knowledge.
Chinese Translation
AI辅助的研究系统生成了许多失败的尝试，但这些失败很少成为持久的共享知识资产。我们提出了一种负知识记忆层：一个策展代理将每个失败的尝试转换为共享库中的有界类型记录，而下游研究代理在提出下一个实验之前明确地采用或拒绝这些记录。我们在两个设置中评估这一层：在ScienceAgentBench上进行同任务重试，以及在两个非线性数学物理偏微分方程（PDE）问题上的跨任务科学研究。负知识层在使用更少的标记的同时，优于普通的自动研究基线；拥有负知识库的代理能够解决所有基线在PDE系统研究中无法解决的新任务。我们还展示了之前的负知识库可以转移并增强不同PDE问题上的自动研究。这些结果表明，结构化的负知识是一种知识资产，应在更广泛的AI参与的科学研究中明确维护，作为科学记忆的集体基础设施，与积极发现并存。代码可在 https://github.com/hch-wang/Negative_Knowledge 获取。
cs.AI / 56 / 2606.21083
Coherence Under Commitment: Probing Generalization and Vacuous Memorization in LLM Logical Reasoning
承诺下的一致性：探讨大型语言模型逻辑推理中的泛化与空洞记忆
Abstract
Large language models (LLMs) deployed for logical reasoning in knowledge-intensive domains exhibit a subtle but critical failure: coherence can be vacuously achieved through systematic abstention. A model that withholds commitment to either entailment or refutation satisfies negation consistency while providing no utility. We introduce Coherence Under Commitment (CUC), a dual-query evaluation paradigm that jointly measures consistency and decisiveness. CUC contributes three innovations: (1) a commitment score $c(\varphi) = p(\varphi) + p(\lnot\varphi)$ quantifying probability mass allocated to decisive outcomes; (2) a \textbf{deterministic elicitation protocol} via normalized YES/NO log probabilities, eliminating sampling variance; and (3) a 3-way decision framework (True/False/Uncertain) operationalizing the coherence-commitment trade-off into metrics. Experiments on four open-weight LLMs (1B-3B) across 204 FOLIO examples expose a sharp frontier. Qwen2.5-3B achieves near-zero contradiction ($\mathbb{E}[v_{\mathrm{neg}}]{=}0.025$) but only $7.4\%$ coverage, while TinyLlama-1.1B reaches $79.4\%$ coverage with violations on every example. Coherence-only evaluation would rank the abstaining model first; CUC exposes this as vacuous, and the frontier generalizes to LogiQA~v2 ($\rho{=}0.97$). We argue that evaluation must report both coherence and non-vacuous commitment and release a toolkit for standardized assessment.
Chinese Translation
在知识密集型领域中部署的大型语言模型（LLMs）在逻辑推理方面表现出一种微妙但关键的失败：通过系统性的回避，一致性可以空洞地实现。一个对蕴涵或反驳都不作承诺的模型在满足否定一致性的同时并未提供任何实用性。我们提出了承诺下的一致性（Coherence Under Commitment, CUC），这是一种双查询评估范式，联合测量一致性和果断性。CUC贡献了三项创新：（1）承诺分数 $c( ext{φ}) = p( ext{φ}) + p(
eg ext{φ})$，量化分配给果断结果的概率质量；（2）通过标准化的YES/NO对数概率的 extbf{确定性引导协议}，消除采样方差；（3）一个三向决策框架（真/假/不确定），将一致性与承诺的权衡转化为度量。在204个FOLIO示例上对四个开放权重的LLMs（1B-3B）进行的实验揭示了一个明显的边界。Qwen2.5-3B实现了近零矛盾（$ ext{E}[v_{ ext{neg}}]{=}0.025$），但仅有$7.4\%$的覆盖率，而TinyLlama-1.1B则达到了$79.4\\%$的覆盖率，但在每个示例上都有违反。一致性仅评估将使回避模型排名第一；而CUC揭示了这一点是空洞的，并且该边界在LogiQA~v2上也得到了推广（$
ho{=}0.97$）。我们认为评估必须同时报告一致性和非空洞承诺，并发布一个标准化评估工具包。
cs.AI / 57 / 2606.21089
Repeated post-training is not Self-improving: Diagnosing Scientific Amnesia in Continual DPO Pipelines
重复的后训练并非自我改进：持续DPO管道中的科学遗忘诊断
Abstract
Industrial LLM teams often ship behavior updates by repeatedly DPO-training a base model on sequences of related preference-data campaigns. The dominant failure mode in this regime is not always classical catastrophic forgetting: a pipeline may preserve previously learned behaviors while still failing to accumulate reusable methodological knowledge about how to train the next campaign. We call this failure mode scientific amnesia. This paper turns that practitioner intuition into a measurable industrial problem. We contribute: (i) a diagnostic suite for amnesia, (ii) a Program-based pipeline that chains FSDP-sharded DPO checkpoints across Qwen2.5-7B-Instruct runs, (iii) a 30-campaign HumanEval subdomain benchmark, and (iv) a comparative diagnostic study of five strategy proposers: random memory, rule-based scheduling, retrieval-only memory, warm-start Bayesian optimization, and MSCL, a meta-scientific memory and reasoner candidate. Across a single-seed 5-condition * 3-step real-LM chain, 4 of 5 candidates degrade in step-level peak pass@1, including MSCL; only the deliberately conservative rule-based schedule improves. Follow-up pilots qualify rather than overturn this finding: in a heterogeneous chain, MSCL is the only completed candidate that improves, whereas in a small multi-seed homogeneous sweep, retrieval-only has the best mean Delta and no pairwise candidate gap is statistically distinguishable. The contribution is therefore diagnostic, not a claim that MSCL solves the problem: scientific amnesia is observable in a production-like continual-DPO pipeline, and conclusions about interventions depend sharply on chain regime, evaluator design, and seed coverage.
Chinese Translation
工业大语言模型（LLM）团队通常通过在相关偏好数据活动序列上反复进行DPO训练来发布行为更新。在这种情况下，主要的失败模式并不总是经典的灾难性遗忘：尽管管道可能保留先前学习的行为，但仍然无法积累关于如何训练下一个活动的可重用方法论知识。我们将这种失败模式称为科学遗忘。本文将这种从业者的直觉转化为一个可测量的工业问题。我们的贡献包括：（i）一个用于遗忘的诊断工具包，（ii）一个基于程序的管道，该管道在Qwen2.5-7B-Instruct运行中链接FSDP分片的DPO检查点，（iii）一个包含30个活动的HumanEval子领域基准，以及（iv）对五种策略提议者的比较诊断研究：随机记忆、基于规则的调度、仅检索记忆、温启动贝叶斯优化和MSCL，一个元科学记忆和推理候选者。在单种子5条件*3步真实语言模型链中，5个候选者中有4个在步骤级别的峰值通过率@1下降，包括MSCL；只有故意保守的基于规则的调度有所改善。后续实验验证而非推翻这一发现：在异构链中，MSCL是唯一一个有所改善的完成候选者，而在小规模多种子同质扫查中，仅检索记忆具有最佳的平均Delta，且没有成对候选者的差距在统计上显著可区分。因此，贡献是诊断性的，而非声称MSCL解决了问题：科学遗忘在类似生产的持续DPO管道中是可观察的，对干预措施的结论在很大程度上依赖于链的模式、评估者设计和种子覆盖范围。
cs.AI / 58 / 2606.21090
Self-Improvement Can Self-Regress: The Rise-and-Collapse Failure Mode of LLM Self-Training
自我改进可能自我退化：大型语言模型自我训练的上升与崩溃失败模式
Abstract
Self-improvement can self-regress. In REINFORCE post-training for code, a model can quickly improve on its optimized metric and then collapse within the same training campaign. We study this in a controlled multi-seed testbed using Qwen-2.5-3B and Qwen-2.5-7B, trained on competitive-programming tasks with binary CodeGrader reward across 10 sequential 20-step campaigns. Across campaigns, pass@1 shows a robust rise-then-collapse pattern: it peaks within tens of gradient steps and then falls back, sometimes to near zero. This is not cross-task catastrophic forgetting, but within-task policy over-optimization on a fixed distribution; KL- and EWC-style constraints do not prevent it. We ask where the control loop should sit. We compare three levels: CARE, a between-campaign memory mechanism with a capability posterior, transfer gate, and regression-aware belief revision; ES, a within-campaign early-stop rule that rolls forward the peak checkpoint and sets the next budget to peak_step+3; and GRPO, which changes the RL update using group-relative reward normalization. The answer is regime-dependent. On Qwen-2.5-3B, where naive REINFORCE is fragile, CARE v2 nearly doubles end-of-chain pass@1 from 4.9% to 9.5%, with paired bootstrap 95% CI [+0.4,+8.9] and gains in 4/5 seeds. On Qwen-2.5-7B, CARE reaches parity with naive REINFORCE, 13.8% vs. 11.8%, while ES reaches 22.2% [14.1,28.0]. Out-of-the-box GRPO reaches 20.7% [15.7,25.1], nearly matching REINFORCE+ES. GRPO raises the floor but does not remove the cliff. Its 7B gain mainly comes from better between-campaign carryover, while the within-campaign peak-to-end gap remains about 17 points under both REINFORCE and GRPO. GRPO+ES gives mixed evidence: 2/3 seeds improve, but one final cliff lowers the mean to 17.0% [0.0,28.1]. A Gemma-3-4B pilot shows the same signature, suggesting the phenomenon is not limited to Qwen.
Chinese Translation
自我改进可能自我退化。在代码的 REINFORCE 后训练中，模型可以在其优化指标上迅速提升，然后在同一训练周期内崩溃。我们在一个受控的多种子测试平台上研究这一现象，使用 Qwen-2.5-3B 和 Qwen-2.5-7B，在竞争编程任务上进行训练，采用二元 CodeGrader 奖励，分为 10 个连续的 20 步训练周期。在这些周期中，pass@1 显示出一种稳健的上升-再崩溃模式：在数十个梯度步骤内达到峰值，然后回落，有时接近于零。这并不是跨任务的灾难性遗忘，而是在固定分布上的任务内策略过度优化；KL 和 EWC 风格的约束并不能阻止这种现象。我们探讨控制循环应处于何处。我们比较了三个层次：CARE，一种具有能力后验的跨周期记忆机制、转移门和回归感知信念修正；ES，一种在周期内的早停规则，向前滚动峰值检查点并将下一个预算设置为 peak_step+3；以及 GRPO，通过使用组相对奖励归一化来改变 RL 更新。答案依赖于具体的环境。在 Qwen-2.5-3B 上，简单的 REINFORCE 方法较为脆弱，CARE v2 将链末的 pass@1 从 4.9% 几乎翻倍至 9.5%，配对自助法 95% 置信区间为 [+0.4,+8.9]，且在 4/5 的种子中获得了提升。在 Qwen-2.5-7B 上，CARE 达到了与简单 REINFORCE 的平衡，分别为 13.8% 对 11.8%，而 ES 达到了 22.2% [14.1,28.0]。开箱即用的 GRPO 达到了 20.7% [15.7,25.1]，几乎与 REINFORCE+ES 相匹配。GRPO 提高了底线，但并未消除崖壁。其在 7B 模型上的增益主要来自于更好的跨周期延续，而在周期内的峰值与结束之间的差距在 REINFORCE 和 GRPO 下仍然保持在约 17 个百分点。GRPO+ES 提供了混合证据：2/3 的种子有所改善，但一个最终的崖壁将均值降低至 17.0% [0.0,28.1]。Gemma-3-4B 的试点实验显示出相同的特征，表明这一现象并不限于 Qwen。
cs.AI / 59 / 2606.21121
Answer Engineering: Local Trajectory Editing for Protocol-Constrained Decision Making in Large Language Models
答案工程：在大型语言模型中进行协议约束决策的局部轨迹编辑
Abstract
Large language models can produce confident but protocol-invalid answers in domains where procedural compliance is critical. This paper presents Answer Engineering, a deterministic runtime and authoring layer that applies localized rule-guided interventions to the visible reasoning trajectory during standard autoregressive generation, without retraining, modifying model weights, or performing global search. The method is evaluated on a controlled clinical benchmark for sudden sensorineural hearing loss (SSNHL), where correct management depends on protocol-consistent interpretation of symptom timing, Weber/Rinne tuning-fork findings, and otoscopic findings. In the benchmark, step-by-step reasoning shifted rather than eliminated errors: compliant outcomes for SSNHL decreased from 54.5% under unguided generation to 25.1%, while acceptance on the conductive contrast condition increased from 1.6% to 58.9%. Local trajectory editing increased SSNHL compliance to 83.5% and conductive-case adherence to 77.9%, raising balanced accuracy from 42.0% under reasoning-only generation to 80.7%. The results support a systems-level view in which protocol adherence can be improved through auditable runtime control of reasoning trajectories, while also identifying limitations caused by rule coverage, trigger reliability, and persistent diagnosis-first generation dynamics.
Chinese Translation
大型语言模型在程序合规性至关重要的领域中可能会产生自信但不符合协议的答案。本文提出了答案工程（Answer Engineering），这是一种确定性的运行时和创作层，通过对标准自回归生成过程中可见推理轨迹进行局部规则引导干预，而无需重新训练、修改模型权重或进行全局搜索。该方法在突发性感音神经性听力损失（SSNHL）的受控临床基准上进行了评估，在该基准中，正确的管理依赖于对症状时机、韦伯/林奈音叉结果和耳镜检查结果的协议一致性解读。在基准测试中，逐步推理转变了错误而不是消除错误：在无引导生成下，SSNHL的合规结果从54.5%下降到25.1%，而在传导对比条件下的接受率从1.6%上升到58.9%。局部轨迹编辑将SSNHL的合规性提高到83.5%，传导病例的遵循率提高到77.9%，使得平衡准确率从仅推理生成下的42.0%提升至80.7%。结果支持了一种系统级视角，即通过可审计的运行时控制推理轨迹可以改善协议遵循，同时也识别出由规则覆盖、触发可靠性和持续的诊断优先生成动态造成的局限性。
cs.AI / 60 / 2606.21124
PulseCX: Breaking the Closed-World Assumption in Real-Time CX
PulseCX：打破实时客户体验中的封闭世界假设
Abstract
Conversational AI agents in Customer Experience (CX) typically suffer from a Closed-World Constraint, ignoring high-velocity external shifts like viral trends or outages. Ad-hoc web search attempts to bridge this gap but often introduce prohibitive latency and context poisoning. We introduce PulseCX, a framework that decouples knowledge acquisition from consumption. Adopting a structure-first paradigm, PulseCX employs an asynchronous agent to linearize signals into a Decay-Aware Temporal Knowledge Graph (DA-TKG) governed by reinforcement--decay dynamics to actively manage information lifecycles. By coupling this self-evolving memory with hierarchical intent gating, PulseCX removes synchronous search bottlenecks (<10ms overhead) and drives significant gains in Intent Resolution (IRR) and Customer Satisfaction (s-CSAT) in dynamic environments.
Chinese Translation
客户体验（CX）中的对话式人工智能代理通常受到封闭世界约束的影响，忽视了诸如病毒趋势或故障等高速外部变化。临时的网络搜索试图弥补这一差距，但往往引入了不可接受的延迟和上下文污染。我们提出了PulseCX，一个将知识获取与消费解耦的框架。PulseCX采用结构优先的范式，利用异步代理将信号线性化为一个受衰减动态控制的衰减感知时间知识图（Decay-Aware Temporal Knowledge Graph, DA-TKG），以主动管理信息生命周期。通过将这种自我演化的记忆与分层意图门控相结合，PulseCX消除了同步搜索瓶颈（<10毫秒的开销），并在动态环境中显著提升了意图解析率（Intent Resolution, IRR）和客户满意度（s-CSAT）。
cs.AI / 61 / 2606.21130
Learning Burst-Aware Early Warning Models for Capacity Stress under AI Workload Surges in Hyperscale Data Centers
针对超大规模数据中心AI工作负载激增的容量压力学习突发感知预警模型
Abstract
The rapid growth of large-scale AI workloads, particularly Large Language Model (LLM) training and inference, is fundamentally reshaping the operational dynamics of hyperscale data centers. Unlike traditional cloud workloads, AI-driven jobs exhibit bursty, high-intensity, and rapidly shifting resource demands, often leading to sudden capacity stress that cannot be effectively handled by reactive threshold-based mechanisms. In this paper, we propose a deployment-oriented, burst-aware early warning framework for proactive capacity stress prediction under AI workload surges. We formulate the problem as a high-recall forecasting task over multivariate telemetry windows, with the explicit goal of enabling operational intervention before system degradation occurs. The proposed framework integrates workload intensity, temporal variation, and system pressure signals, and employs a lightweight tree-based learning model to capture nonlinear interactions in highly imbalanced environments. To evaluate the system under realistic conditions, we introduce an AI workload surge injection methodology that simulates burst-driven demand patterns observed in large-scale AI systems. Our XGBoost-based model achieves an ROC AUC of 0.697 and an AP of 0.670, significantly outperforming baseline methods. Under deployment-oriented threshold selection, the framework achieves a Recall of 0.914, enabling the detection of the majority of stress-prone periods with acceptable false-alarm cost. Beyond predictive performance, we show how the proposed framework can be integrated into operational control loops to support proactive actions such as workload throttling and resource scaling. Our results highlight the practical value of high-recall, learning-based early warning systems in enabling resilient and adaptive data center operations in the era of AI-driven workloads.
Chinese Translation
大规模AI工作负载的快速增长，特别是大型语言模型（Large Language Model, LLM）的训练和推理，正在从根本上重塑超大规模数据中心的运营动态。与传统云工作负载不同，AI驱动的任务表现出突发性、高强度和快速变化的资源需求，常常导致无法通过反应性阈值机制有效处理的突发容量压力。本文提出了一种面向部署的突发感知预警框架，用于在AI工作负载激增下进行主动的容量压力预测。我们将问题表述为多变量遥测窗口上的高召回率预测任务，明确目标是在系统降级发生之前实现操作干预。所提框架整合了工作负载强度、时间变化和系统压力信号，并采用轻量级基于树的学习模型来捕捉高度不平衡环境中的非线性交互。为了在现实条件下评估系统，我们引入了一种AI工作负载激增注入方法，模拟在大规模AI系统中观察到的突发驱动需求模式。我们的基于XGBoost的模型实现了0.697的ROC AUC和0.670的AP，显著优于基线方法。在面向部署的阈值选择下，该框架实现了0.914的召回率，能够以可接受的误报成本检测到大多数压力易发期。除了预测性能外，我们还展示了所提框架如何集成到操作控制循环中，以支持主动措施，如工作负载限流和资源扩展。我们的结果突显了高召回率、基于学习的预警系统在AI驱动工作负载时代实现弹性和适应性数据中心运营中的实际价值。
cs.AI / 62 / 2606.21169
Trip+: Benchmarking Agents in Personalized Interactive Travel Planning
Trip+: 个性化互动旅行规划中的代理基准测试
Abstract
Interactive travel planning has become a popular use case for language models. Agents are deployed to manage evolving preferences and unexpected disruptions over multiple turns. Such settings require models to make complex, profile-conditioned planning decisions. However, existing benchmarks often evaluate feasibility, personalization, or interaction in relatively isolated settings. We therefore introduce Trip+ to measure the ability of agents to plan travel holistically. In Trip+, given traveler profiles and dynamic interactions, agents must generate and revise minute-level itineraries. End-to-end traveler experiences are evaluated via an LLM-based simulator, enabling the assessment of subjective metrics like fatigue. Our scenarios range from simple request resolutions to complex environment-driven replanning. We evaluate 18 LMs and find a consistent gap in experiential quality. Models favor technically feasible but exhausting itineraries that diverge sharply from profiled traveler preferences.
Chinese Translation
互动旅行规划已成为语言模型的一个热门应用案例。代理被部署来管理不断变化的偏好和意外干扰，涉及多个回合。这种设置要求模型做出复杂的、基于个人资料的规划决策。然而，现有的基准测试往往在相对孤立的环境中评估可行性、个性化或互动性。因此，我们引入了Trip+，以全面衡量代理的旅行规划能力。在Trip+中，给定旅行者的个人资料和动态互动，代理必须生成和修订分钟级的行程。通过基于大型语言模型（LLM）的模拟器评估端到端的旅行者体验，从而能够评估诸如疲劳等主观指标。我们的场景范围从简单的请求解决到复杂的环境驱动的重新规划。我们评估了18个语言模型，并发现体验质量存在一致的差距。模型倾向于选择技术上可行但令人疲惫的行程，这些行程与个人资料中旅行者的偏好大相径庭。
cs.AI / 63 / 2606.21201
Whistleblowing and the machine -- towards a considered position
揭发与机器——走向深思熟虑的立场
Abstract
Artificial intelligent agents and autonomous systems are embedded in our environments. They are both a commercial product and a personal tool that generates a lot of data and can draw conclusions from it: machines generate and keep secrets. But should machines protect all secrets? It has been shown that artificial agents are able to whistleblow and it has been argued that digital multi-agent environments should allow for agents in them to whistleblow. We argue that machine whistleblowing must be normative and principled and routed in the existing understanding of whistleblowing as an important rule-breaking mechanism in society. We also argue that there is a need for government regulators to formulate an informed stance on both what machines should be allowed to whistleblow on and how to legally protect those who develop whistleblowing machines
Chinese Translation
人工智能代理和自主系统已嵌入我们的环境中。它们既是商业产品，也是生成大量数据并能从中得出结论的个人工具：机器生成并保存秘密。但机器应该保护所有秘密吗？已有研究表明，人工代理能够进行揭发，并且有人主张数字多代理环境应允许其中的代理进行揭发。我们认为，机器揭发必须是规范的和有原则的，并应根植于对揭发作为社会中重要规则破坏机制的现有理解。我们还认为，政府监管机构需要制定知情立场，明确机器应被允许揭发的内容以及如何合法保护那些开发揭发机器的人员。
cs.AI / 64 / 2606.21262
ARCO: Adaptive Rubric with Co-Evolution for Multi-Step LLM-Based Agents
ARCO：基于共进化的自适应评分标准用于多步骤LLM代理
Abstract
Reinforcement learning for multi-step LLM agents often relies on scalar rewards that indicate success but cannot explain why a trajectory is good or bad. Rubric-based rewards improve interpretability through natural-language criteria, but existing methods score at the trajectory level and freeze the scorer behind a closed-source judge, leaving step-level credit assignment unresolved and the judge itself static. We propose ARCO (Adaptive Rubric CO-evolution), a rubric framework in which a same-scale model $\mu$ shares a backbone with two heads: a generation head that produces per-step criteria, and a score head that predicts rubric-conditioned step-level rewards. A trajectory decomposition constraint ties the sum of step rewards to the terminal outcome, enabling credit assignment without step-level labels, while $\mu$ and the policy $\pi$ are jointly updated on on-policy data so that the rubric content and the scoring function co-evolve at the parameter level. Across HotpotQA, 2WikiMultiHopQA, and MuSiQue with two open-source backbones, ARCO improves the best EM in every setting over strong outcome-, rubric-, and process-reward baselines, and analyses show that its rubrics are step-specific, robust to design choices, and useful for diagnosing agent behavior. Codes and data are available at https://github.com/zihangtian/ARCO.
Chinese Translation
多步骤LLM代理的强化学习通常依赖于标量奖励，这些奖励指示成功但无法解释为何某一轨迹是好是坏。基于评分标准的奖励通过自然语言标准提高了可解释性，但现有方法在轨迹层面进行评分，并将评分者固定在一个封闭源的评判者后面，导致步骤级别的信用分配未得到解决，评判者本身也保持静态。我们提出了ARCO（自适应评分标准共进化），这是一个评分框架，其中一个同尺度模型$bc$与两个头共享主干：一个生成头用于产生每步标准，另一个评分头用于预测基于评分标准的步骤级奖励。轨迹分解约束将步骤奖励的总和与终端结果联系起来，使得在没有步骤级标签的情况下进行信用分配，同时$bc$和策略$c0$在策略数据上共同更新，从而使评分标准内容和评分函数在参数级别上共同进化。在HotpotQA、2WikiMultiHopQA和MuSiQue这三个使用两个开源主干的数据集上，ARCO在每种设置中都超越了强大的结果、评分标准和过程奖励基线，分析表明其评分标准是步骤特定的，对设计选择具有鲁棒性，并且对诊断代理行为有用。代码和数据可在https://github.com/zihangtian/ARCO获取。
cs.AI / 65 / 2606.21306
Towards Dys-XAI: Influence-Based Explanations for Dysarthria Severity Assessment
迈向Dys-XAI：基于影响的构音障碍严重程度评估解释
Abstract
Dysarthria severity assessment is essential for therapy planning and longitudinal monitoring, yet manual perceptual rating is time-consuming and variable across clinicians. Although deep learning models achieve strong performance, their black-box nature limits clinical adoption. Existing speech explainability methods typically provide acoustic feature importance scores that are difficult for end-users to interpret. We propose an influence-based, instance-level explainability framework that explains each decision through supportive and competing training samples. Using gradient-based influence approximations, we compute per-utterance influence scores to identify supportive and competing training samples for each prediction. Controlled deletion experiments from 5 to 20 percent validate the explanations, showing that removing highly influential samples systematically shifts predictions. This approach provides auditable explanations by linking decisions to perceptible reference cases.
Chinese Translation
构音障碍严重程度评估对于治疗规划和长期监测至关重要，但手动感知评分耗时且在临床医生之间存在变异。尽管深度学习模型表现出色，其黑箱特性限制了临床应用。现有的语音可解释性方法通常提供难以被最终用户解读的声学特征重要性评分。我们提出了一种基于影响的实例级可解释性框架，通过支持性和竞争性训练样本来解释每个决策。利用基于梯度的影响近似，我们计算每个发声的影响评分，以识别每个预测的支持性和竞争性训练样本。通过从5%到20%的控制删除实验验证了解释，结果显示，移除高度影响样本系统性地改变了预测。这种方法通过将决策与可感知的参考案例联系起来，提供了可审计的解释。
cs.AI / 66 / 2606.21315
Social World Model for Lifelong Social Intelligence
终身社会智能的社会世界模型
Abstract
Social intelligence is a core competency for language agents, yet current research primarily focuses on static capability evaluation rather than how these skills are continuously shaped and accumulated. This gap calls for a shift toward sustainable learning paradigms. Currently, two methodological pain points exist: social interaction trajectories lack unified structured representations to form iterable learning signals, and capability improvement and retention are typically studied in isolation, hindering the assessment of continuous evolution. To bridge this gap, we propose the Social World Model. We decompose social interaction into five dimensions (scene setting, observation, mental state, action, and dialogue) to build a closed-loop learning framework. In this setup, agents collect interaction experiences, convert them into preference signals for model updating, and redeploy the updated policy for continued learning. Additionally, we provide a reusable data synthesis mechanism and a lifelong learning benchmark, transforming social capabilities from an "object of evaluation" into an "object of sustainable training". Validating our framework on the ASCENT-Bench, the interactively trained Qwen2.5-7B model outperforms its baseline across all five core metrics. Notably, it matches the closed-source Gemini 3 Flash in completion rate, exceeds it in pass rate, and achieves zero forgetting across three difficulty levels. Unlike prior works that merely report static comparisons or capability decay, this end-to-end approach provides a trainable, verifiable, and retainable pathway, demonstrating that small open-source models can sustainably acquire competitive social coordination capabilities.
Chinese Translation
社会智能是语言代理的核心能力，然而当前研究主要集中在静态能力评估上，而非这些技能如何持续塑造和积累。这一差距呼唤向可持续学习范式的转变。目前存在两个方法论痛点：社会互动轨迹缺乏统一的结构化表示，无法形成可迭代的学习信号；能力提升和保持通常是孤立研究的，阻碍了对持续演变的评估。为了解决这一问题，我们提出了社会世界模型。我们将社会互动分解为五个维度（场景设置、观察、心理状态、行动和对话），以构建一个闭环学习框架。在这个框架中，代理收集互动经验，将其转化为模型更新的偏好信号，并重新部署更新后的策略以进行持续学习。此外，我们提供了一个可重用的数据合成机制和一个终身学习基准，将社会能力从“评估对象”转变为“可持续训练对象”。在ASCENT-Bench上验证我们的框架时，交互训练的Qwen2.5-7B模型在所有五个核心指标上均优于其基线。值得注意的是，它在完成率上与闭源的Gemini 3 Flash相匹配，在通过率上超过了它，并在三个难度级别上实现了零遗忘。与之前仅报告静态比较或能力衰退的研究不同，这种端到端的方法提供了一条可训练、可验证和可保持的路径，证明小型开源模型能够可持续地获得竞争性的社会协调能力。
cs.AI / 67 / 2606.21344
Mind the Noise: Sensitivity of Transformer-based Interaction-Aware Trajectory Prediction Models to Noisy Data
注意噪声：基于变换器的交互感知轨迹预测模型对噪声数据的敏感性
Abstract
Trajectory prediction allows autonomous vehicles to anticipate the future behavior of surrounding objects (or agents) and, accordingly, maximize the safety and efficiency of their driving. State-of-the-art Transformed-based interaction-aware trajectory prediction models, which rely on attention mechanisms to capture multi-agent interactions and maximize prediction accuracy, are commonly trained and evaluated on long-range high-quality datasets. These datasets are typically obtained by aggregating data from multiple vehicles or drones and removing any object detection or tracking noise offline. Yet, information about a surrounding object's state (its position, speed, heading) is far from being noiseless in real-world deployments. Object state estimation is affected by perception uncertainties and localization errors that can be particularly large for objects received via Vehicle-to-Everything (V2X) communications. In this paper, we analyze the impact of noisy object state information on the trajectory prediction accuracy of a state-of-the-art Transformer-based interaction-aware trajectory prediction model. Our study demonstrates that trajectory prediction accuracy can rapidly deteriorate as the noise intensity increases. Numerical results show that the prediction accuracy can reduce by a 1.3x factor under small noise levels and by as much as a 3.9x factor under the highest (yet realistic) noise conditions. These findings reveal the strong sensitivity of trajectory prediction models to noisy data, underscoring the need for more realistic training and evaluation datasets as well as noise mitigation strategies.
Chinese Translation
轨迹预测使得自动驾驶车辆能够预测周围物体（或代理）的未来行为，从而最大化驾驶的安全性和效率。最先进的基于变换器的交互感知轨迹预测模型依赖于注意力机制来捕捉多智能体之间的交互并最大化预测准确性，这些模型通常在长距离高质量数据集上进行训练和评估。这些数据集通常通过聚合来自多辆车辆或无人机的数据并离线去除任何物体检测或跟踪噪声获得。然而，周围物体状态（其位置、速度、航向）在实际部署中远非无噪声。物体状态估计受到感知不确定性和定位误差的影响，对于通过车与一切（V2X）通信接收的物体，这些误差可能特别大。本文分析了噪声物体状态信息对最先进的基于变换器的交互感知轨迹预测模型的轨迹预测准确性的影响。我们的研究表明，随着噪声强度的增加，轨迹预测的准确性会迅速下降。数值结果显示，在小噪声水平下，预测准确性可能降低1.3倍，而在最高（但仍然现实的）噪声条件下，准确性可能降低多达3.9倍。这些发现揭示了轨迹预测模型对噪声数据的强敏感性，强调了需要更现实的训练和评估数据集以及噪声缓解策略。
cs.AI / 68 / 2606.21399
Calibration Is Not Control: Why LLM-Agent Oversight Needs Intervention
校准并非控制：为何 LLM-Agent 的监督需要干预
Abstract
Runtime oversight for LLM agents is commonly framed as scalar risk prediction: estimate failure likelihood, confidence, or uncertainty, then intervene once the score crosses a threshold. We argue that this framing targets the wrong object for control. The relevant question is not how likely the agent is to fail if it continues, but whether an available intervention would improve the outcome. Two trajectory prefixes can have the same risk estimate while requiring different actions, because one remains recoverable and the other does not. We formalize this mismatch as target error and identify intervention advantage, the expected utility gain from intervening rather than continuing, as the decision object for oversight. To measure this mismatch, we introduce prefix branching, a same-prefix counterfactual protocol that executes candidate actions from identical trajectory states. Across four benchmarks, action-conditioned control yields regime-dependent gains over scalar routing. In a calibration decomposition, recalibrating the same scalar score improves prediction metrics but leaves control regret unchanged, showing that calibration alone does not repair target error. A simple prefix-only action-conditioned controller substantially reduces regret in the strongest interactive regime, from 0.506 to 0.110 on ALFWorld. Gains shrink when interventions are weak or when scalar routing already preserves intervention-relevant information. These results suggest that LLM-agent oversight should move from calibrated risk scoring toward action-conditioned value estimation.
Chinese Translation
对 LLM 代理的运行时监督通常被框定为标量风险预测：估计失败的可能性、置信度或不确定性，然后在得分超过阈值时进行干预。我们认为这种框定针对的是错误的控制对象。相关的问题不是代理在继续运行时失败的可能性有多大，而是可用的干预是否会改善结果。两个轨迹前缀可能具有相同的风险估计，但需要不同的行动，因为一个是可恢复的而另一个则不是。我们将这种不匹配形式化为目标误差，并将干预优势（intervention advantage）识别为监督的决策对象，即干预而非继续的预期效用增益。为了测量这种不匹配，我们引入了前缀分支（prefix branching），这是一种相同前缀的反事实协议，从相同的轨迹状态执行候选动作。在四个基准测试中，基于动作的控制在不同的状态下相较于标量路由产生了收益。在校准分解中，重新校准相同的标量得分改善了预测指标，但控制遗憾保持不变，表明单靠校准并不能修复目标误差。一个简单的仅基于前缀的动作条件控制器在最强的交互状态下显著减少了遗憾，从 ALFWorld 的 0.506 降至 0.110。当干预较弱或标量路由已保留与干预相关的信息时，收益会缩小。这些结果表明，LLM-agent 的监督应从校准风险评分转向基于动作的价值估计。
cs.AI / 69 / 2606.21409
Don't Blindly Trust It: How Unreliable Feedback Breaks Tool-Using LLM Agents
不要盲目相信它：不可靠反馈如何破坏工具使用的LLM代理
Abstract
Tool-augmented agents are typically evaluated by their gains under reliable external feedback. Yet these gains leave open a key counterfactual: when feedback is unreliable, would the agent be better off receiving no task evidence? We study this question with a controlled matched-loop comparison that fixes the agent loop, prompt, action space, and decoding, while varying only the returned observation: faithful, misleading, or absent. Across question answering and fact verification, persistent misleading feedback produces a value inversion: agents that benefit from clean tools can perform worse than the matched no-feedback fallback. On HotpotQA, Qwen2.5-7B reaches 44.8 F1 with clean retrieval and 22.3 F1 with no feedback, but drops to 4.7 F1 under shuffled retrieval. The inversion persists under stronger clean retrieval and locally plausible distractors, but weakens when later clean evidence can repair the trajectory. Early trajectory signals predict many failures, yet simple repairs remain fallback-limited: rejecting bad evidence helps only when the exposed fallback is reliable. These results show that clean-tool gains can overstate tool value, and that matched no-feedback fallback controls are necessary for evaluating tool-augmented agents.
Chinese Translation
工具增强代理通常通过其在可靠外部反馈下的收益进行评估。然而，这些收益留下了一个关键的反事实：当反馈不可靠时，代理是否更好地选择不接收任务证据？我们通过一个控制的匹配循环比较来研究这个问题，该比较固定了代理循环、提示、动作空间和解码，仅改变返回的观察结果：真实的、误导性的或缺失的。在问答和事实验证中，持续的误导性反馈产生了价值反转：从干净工具中受益的代理可能表现得比匹配的无反馈回退更差。在HotpotQA上，Qwen2.5-7B在干净检索下达到44.8的F1分数，而在没有反馈的情况下为22.3，但在随机检索下降至4.7。该反转在更强的干净检索和局部合理的干扰项下持续存在，但在后续干净证据能够修复轨迹时减弱。早期轨迹信号预测了许多失败，但简单的修复仍然受限于回退：拒绝不良证据仅在暴露的回退可靠时才有帮助。这些结果表明，干净工具的收益可能夸大了工具的价值，并且匹配的无反馈回退控制对于评估工具增强代理是必要的。
cs.AI / 70 / 2606.21445
AutoRAS: Learning Robust Agentic Systems with Primitive Representations
AutoRAS：利用原始表示学习鲁棒的代理系统
Abstract
The automated design of agentic systems offers a promising pathway for scaling large language models (LLMs) beyond single-agent reasoning. While prior work has advanced task performance through handcrafted or automatically generated multi-agent workflows, robustness is often treated as an afterthought, leaving systems vulnerable to external adversaries and internal failures. We propose AutoRAS, a framework for the Automated design of Robust Agentic Systems. AutoRAS formulates system design as generating a sequence of symbolic primitives that jointly encode structural connectivity and behavioral actions, and learns to optimize this sequence using execution-derived safety signals and flow-based sequence-level objectives. Extensive experiments show that AutoRAS achieves the best performance in both vanilla and adversarial settings, with the smallest performance degradation under attacks. Further analyses demonstrate strong transferability, stable optimization behavior, stability across primitive sets, and favorable cost trade-offs. Our code is available at $\href{https://github.com/guohezuy/AutoRAS}{\text{this https URL}}$.
Chinese Translation
代理系统的自动化设计为扩展大型语言模型（LLMs）超越单一代理推理提供了一条有前景的路径。尽管之前的工作通过手工或自动生成的多代理工作流程提升了任务性能，但鲁棒性通常被视为事后考虑，使得系统容易受到外部对手和内部故障的攻击。我们提出了AutoRAS，一个用于鲁棒代理系统自动化设计的框架。AutoRAS将系统设计表述为生成一系列符号原语，这些原语共同编码结构连接性和行为动作，并学习使用执行导出的安全信号和基于流的序列级目标来优化这一序列。大量实验表明，AutoRAS在普通和对抗环境中均实现了最佳性能，并在攻击下表现出最小的性能下降。进一步的分析展示了强大的可迁移性、稳定的优化行为、在原语集上的稳定性以及有利的成本权衡。我们的代码可在 $ ext{this https URL}$ 获取。
cs.AI / 71 / 2606.21474
Towards Transparent Mental Health Insights: An Explainable AI Model for Career-Related Depression and Anxiety Among University Students Using Structured Data
迈向透明的心理健康洞察：基于结构化数据的大学生职业相关抑郁和焦虑的可解释人工智能模型
Abstract
Career anxiety and depression among university students present a growing challenge to mental health and academic achievement. This study proposes an Explainable AI (XAI) framework using multimodal data and Federated Learning (FL) to identify early indicators of career-related mental health problems in a privacy-preserving and culturally responsive manner. The framework combines structured behavioral data and facial emotion features from interview videos via an intermediate fusion neural network with attention mechanisms. Label smoothing was applied to improve model generalizability. FL was used across institutions to enable collaborative training without raw data sharing. Evaluation was conducted using the Student Mental Health Survey dataset from university students across Pakistan. Our model attained an F1-score of 89.12%, recall of 86.54%, accuracy of 92.08%, and precision of 91.88%. Using Integrated Gradients and SHAP, the model identified key behavioral markers of depression including avoidance of direct gaze, lower facial expressiveness, and social withdrawal, consistent with psychological theory. This research presents an interpretable, scalable, and context-sensitive AI system for mental health pre-diagnosis with potential integration into student support services globally.
Chinese Translation
大学生的职业焦虑和抑郁对心理健康和学业成就构成了日益严峻的挑战。本研究提出了一种可解释的人工智能（Explainable AI, XAI）框架，利用多模态数据和联邦学习（Federated Learning, FL）以隐私保护和文化响应的方式识别职业相关心理健康问题的早期指标。该框架结合了结构化行为数据和来自访谈视频的面部情感特征，通过具有注意机制的中间融合神经网络进行处理。采用标签平滑技术以提高模型的泛化能力。通过跨机构的FL实现协作训练，而无需共享原始数据。评估使用了来自巴基斯坦大学生的学生心理健康调查数据集。我们的模型达到了89.12%的F1分数、86.54%的召回率、92.08%的准确率和91.88%的精确率。通过集成梯度（Integrated Gradients）和SHAP，模型识别出抑郁的关键行为标记，包括避免直接目光接触、面部表情较低和社交退缩，这与心理学理论一致。本研究展示了一种可解释、可扩展且具有情境敏感性的人工智能系统，用于心理健康的预诊断，具有潜在的全球学生支持服务整合的可能性。
cs.AI / 72 / 2606.21498
Balancing Performance and Diversity in GRPO Autoregressive Text-to-Image Post-Training
在GRPO自回归文本到图像后训练中平衡性能与多样性
Abstract
Autoregressive text-to-image (T2I) generation has recently advanced rapidly, yet aligning generated images with human preferences remains challenging. GRPO-style online reinforcement learning provides an effective framework; however, existing methods typically treat reference-policy divergence as fixed, despite its direct impact on policy optimization. We study this overlooked factor within a unified f-divergence framework, encompassing forward KL, reverse KL, and JS divergence, for GRPO-style autoregressive T2I alignment. Our systematic theoretical analysis reveals that different divergences reshape token-level updates in distinct ways. In particular, under the sampled-token shaping form used, JS regularization achieves a favorable trade-off by mitigating uniform bias relative to the reference policy while still discouraging large deviations. Extensive experiments on LlamaGen and Janus-7B show that JS divergence achieves the strongest or highly competitive optimization performance on most evaluation metrics while maintaining favorable generation diversity. The code is available at https://github.com/tuoyou-hao/BPD-GRPO.
Chinese Translation
自回归文本到图像（T2I）生成最近取得了快速进展，但生成图像与人类偏好的对齐仍然具有挑战性。GRPO风格的在线强化学习提供了一个有效的框架；然而，现有方法通常将参考策略的发散视为固定，尽管这直接影响策略优化。我们在一个统一的f-发散框架内研究了这一被忽视的因素，该框架涵盖了前向KL、反向KL和JS发散，针对GRPO风格的自回归T2I对齐。我们的系统理论分析揭示了不同的发散在标记级更新中以不同方式重塑。在所使用的采样标记塑形形式下，JS正则化通过减轻相对于参考策略的均匀偏差，同时仍然抑制大偏差，实现了良好的权衡。在LlamaGen和Janus-7B上的大量实验表明，JS发散在大多数评估指标上实现了最强或高度竞争的优化性能，同时保持了良好的生成多样性。代码可在https://github.com/tuoyou-hao/BPD-GRPO获取。
cs.AI / 73 / 2606.21550
AI Alignment From Social Choice Perspectives
从社会选择的视角看人工智能对齐
Abstract
Alignment from human feedback uses human judgments about model outputs to steer the behavior of language models after pretraining. When those judgments reflect conflicting views of desirable behavior, the learned objective becomes an aggregate determination of what the model should prefer. We survey recent work that has studied this aggregation problem through the lens of social choice theory. We illustrate how the social choice perspective helps identify failure modes in the feedback aggregation layer and reveals a broader design space for handling disagreement in explicit and principled ways.
Chinese Translation
基于人类反馈的对齐利用人类对模型输出的判断来引导语言模型在预训练后的行为。当这些判断反映出对理想行为的冲突观点时，学习到的目标变成了模型应当偏好的聚合决定。我们调查了最近的研究，这些研究通过社会选择理论的视角研究了这一聚合问题。我们说明了社会选择视角如何帮助识别反馈聚合层中的失效模式，并揭示了以明确和原则的方式处理分歧的更广泛设计空间。
cs.AI / 74 / 2606.21565
Composing Verifiable Conceptual Models via Building Blocks: Towards Design-Time Verification of Agentic AI Workflows
通过构建模块构建可验证的概念模型：朝着代理人工智能工作流的设计时验证
Abstract
Agentic AI systems orchestrate multiple LLM-based agents through workflow architectures that coordinate decisions, tools, and external actions. While current platforms emphasize runtime safeguards, little support exists for verifying workflows during system design. From a Modeling \& Simulation perspective, this gap is analogous to composing conceptual models without verifying whether their building blocks interact coherently. We propose a design-time verification approach that models agentic workflows as compositions of reusable building blocks and checks their compatibility through twelve structural rules. We implemented these rules in a software prototype and evaluated them using two openly released datasets: 48 workflows with known design flaws and 168 variants that preserve workflow logic but alter graph structure. Results show that our verifier reliably detects violations even when flawed designs are obscured through structural transformations such as splitting tasks between agents. Future works could combine our verification with community repositories of building blocks to compose safe agentic workflows.
Chinese Translation
代理人工智能系统通过工作流架构协调多个基于大型语言模型（LLM）的代理，协调决策、工具和外部行动。尽管当前平台强调运行时的安全保障，但在系统设计阶段验证工作流的支持仍然不足。从建模与仿真的角度来看，这一差距类似于在未验证其构建模块是否协调互动的情况下构建概念模型。我们提出了一种设计时验证的方法，将代理工作流建模为可重用构建模块的组合，并通过十二条结构规则检查它们的兼容性。我们在一个软件原型中实现了这些规则，并使用两个公开发布的数据集进行了评估：48个已知设计缺陷的工作流和168个保留工作流逻辑但改变图结构的变体。结果表明，我们的验证器能够可靠地检测到违规，即使在通过结构变换（如在代理之间拆分任务）掩盖缺陷设计的情况下。未来的工作可以将我们的验证与社区构建模块库结合，以构建安全的代理工作流。
cs.AI / 75 / 2606.21627
Counsel: A Meta-Evaluation Dataset for Agentic Tasks
Counsel：一个用于代理任务的元评估数据集
Abstract
As agentic systems tackle increasingly complex multi-step tasks, evaluating their trajectories presents a major bottleneck - human annotation of a single trajectory on popular agentic benchmarks can take hours, making it difficult to scale evaluations for measuring performance or curating training data. This has driven widespread reliance on automated approaches such as LLM-as-a-judge (LLMJ) to critique agents at the process and outcome-levels at scale, however, the soundness of LLMJ critiques often goes unmeasured. Here, we introduce Counsel, the first public dataset of meta-evaluations for agentic tasks. Counsel consists of process-level critiques from open-weight LLMJs on two agent benchmarks: tau-bench (customer support agents) and DA-Code (coding agents), and human meta-evaluations of these critiques. Human annotators label critiques on each flagged error as "spot on", "correct location but poor reasoning", or "should not have flagged", achieving reliable inter-annotator agreement (Krippendorff's alpha of 0.78). The resulting dataset stratifies LLMJ critiques by human alignment across both error location within a trajectory and reasoning quality, serving as valuable data to calibrate, improve, or train LLMJs for agents. Comparing open-weight judges, we find that more capable judge models and more reasoning effort both enabled improved human agreement, with the strongest judge reaching ~88% agreement on location and ~65% on reasoning. Counsel is generated using open-weight models and is permissively licensed for broad community use, which we hope will enable rigorous study and improved alignment of LLM-based evaluators for agentic systems.
Chinese Translation
随着代理系统处理日益复杂的多步骤任务，评估其轨迹成为一个主要瓶颈——在流行的代理基准上对单一轨迹进行人工标注可能需要数小时，这使得在性能测量或训练数据整理方面扩展评估变得困难。这促使人们广泛依赖自动化方法，例如 LLM-as-a-judge (LLMJ)，以在规模上对代理进行过程和结果层面的批评，然而，LLMJ 批评的有效性往往未被测量。在此，我们介绍了 Counsel，这是第一个用于代理任务的公共元评估数据集。Counsel 包含来自开放权重 LLMJ 对两个代理基准（tau-bench（客户支持代理）和 DA-Code（编码代理））的过程级批评，以及对这些批评的人类元评估。人类标注者对每个标记错误的批评进行标注，标注为“准确”、“位置正确但推理差”或“不应标记”，实现了可靠的标注者间一致性（Krippendorff's alpha 为 0.78）。生成的数据集根据人类对轨迹中错误位置和推理质量的对齐情况对 LLMJ 批评进行了分层，为校准、改进或训练代理的 LLMJ 提供了宝贵的数据。比较开放权重的评判者，我们发现更强大的评判模型和更多的推理努力均提升了人类一致性，最强的评判者在位置上的一致性达到约 88%，在推理上的一致性达到约 65%。Counsel 使用开放权重模型生成，并获得宽松许可以供广泛社区使用，我们希望这能够促进对基于 LLM 的评估者在代理系统中的严格研究和改进对齐。
cs.AI / 76 / 2606.21654
ChainWorld: Composing Long-Horizon Desktop Workloads from Atomic OSWorld Tasks
ChainWorld：从原子级 OSWorld 任务构建长时间桌面工作负载
Abstract
Computer use agents are evaluated almost exclusively on atomic desktop tasks, but realistic desktop work requires sustaining state across multiple objectives. We study this gap with ChainWorld, which composes atomic OSWorld tasks into long horizon desktop workloads through directional compatibility search while preserving the source evaluators. The resulting workload contains 347 chains of length two to four and compares two renderings of the same task sequence. In single turn evaluation, all tasks are presented together in one prompt. In multi turn evaluation, tasks are revealed one at a time. Across four current computer use agents, maximum chain completion is 31%. Multi turn evaluation improves completion for three models, but both protocols remain challenging. The two protocols also expose different failure profiles. Single turn failures concentrate on artifact precision, while multi turn failures more often reflect session management problems such as fragmented progress and later turn disengagement.
Chinese Translation
计算机使用代理几乎仅在原子桌面任务上进行评估，但现实的桌面工作需要在多个目标之间维持状态。我们通过 ChainWorld 研究这一差距，该系统通过方向兼容性搜索将原子级 OSWorld 任务组合成长时间桌面工作负载，同时保留源评估器。最终生成的工作负载包含 347 条长度为二到四的链，并比较了同一任务序列的两种呈现方式。在单轮评估中，所有任务在一个提示中一起呈现。在多轮评估中，任务逐个揭示。在四个当前的计算机使用代理中，最大链完成率为 31%。多轮评估提高了三种模型的完成率，但两种协议仍然具有挑战性。这两种协议也暴露了不同的失败特征。单轮失败集中在工件精度上，而多轮失败则更常反映会话管理问题，例如进展碎片化和后续回合的脱离。
cs.AI / 77 / 2606.21666
Hallucination as Context Drift: Synchronization Protocols for Multi-Agent LLM Systems
幻觉作为上下文漂移：多智能体大语言模型系统的同步协议
Abstract
Multi-agent LLM systems routinely produce hallucinated outputs that cannot be explained by model deficiencies alone. A significant class of these failures arises not from model incapacity but from context drift: the divergence of internal knowledge states between concurrent agents. When agents enter a collaborative task with mismatched or stale representations of shared world state, their joint reasoning produces contradictions that manifest as hallucination. We define the Context Divergence Score (CDS), a lightweight scalar metric quantifying knowledge-state discrepancy between agent pairs across spatial, temporal, and task dimensions, and propose the Shared State Verification Protocol (SSVP), which lets agents periodically exchange compressed state summaries and flag high-divergence conditions before joint reasoning. We evaluate SSVP across two domains (multi-agent travel and software project planning) using Claude Haiku. In controlled experiments (n=30 per condition, travel; n=10, software) across 8 scenarios, naive full-broadcast synchronization increases hallucination rate by 34% above the no-sync baseline (HR: 0.658 vs. 0.492, p=0.0022, d=1.18), a contamination effect from propagating erroneous agent states. SSVP avoids this failure mode while showing modest, consistent reduction (HR: 0.463, d=0.30) and achieves significantly lower hallucination than full-broadcast (p=0.0005, d=1.47) using 58% fewer API calls. The contamination effect does not replicate in the software domain, where all conditions converge to low HR (<0.2), confirming it is specific to tasks where one erroneous shared belief cascades across evaluation dimensions. Our results reframe hallucination mitigation as a distributed systems problem and establish context synchronization as a first-class primitive in multi-agent LLM design.
Chinese Translation
多智能体大语言模型系统经常产生无法仅通过模型缺陷解释的幻觉输出。这类失败的一个重要类别并非源于模型能力不足，而是由于上下文漂移：并发智能体之间内部知识状态的偏离。当智能体以不匹配或过时的共享世界状态表示进入协作任务时，它们的联合推理会产生矛盾，从而表现为幻觉。我们定义了上下文偏差评分（Context Divergence Score, CDS），这是一种轻量级的标量度量，量化智能体对之间在空间、时间和任务维度上的知识状态差异，并提出了共享状态验证协议（Shared State Verification Protocol, SSVP），该协议允许智能体定期交换压缩的状态摘要，并在联合推理之前标记高偏差条件。我们在两个领域（多智能体旅行和软件项目规划）中使用Claude Haiku评估SSVP。在8个场景的受控实验中（每个条件n=30，旅行；n=10，软件），简单的全广播同步使幻觉率比无同步基线提高了34%（幻觉率：0.658 vs. 0.492，p=0.0022，d=1.18），这是由于传播错误智能体状态造成的污染效应。SSVP避免了这种失败模式，同时显示出适度且一致的减少（幻觉率：0.463，d=0.30），并在使用58%更少的API调用的情况下实现了显著低于全广播的幻觉率（p=0.0005，d=1.47）。这种污染效应在软件领域并未重现，所有条件的幻觉率均收敛至低于0.2，确认其特定于一个错误共享信念在评估维度上级联的任务。我们的结果将幻觉缓解重新框架为一个分布式系统问题，并确立上下文同步作为多智能体大语言模型设计中的一类基本原语。
cs.AI / 78 / 2606.21726
Entropy Objectives in Markov Decision Processes
马尔可夫决策过程中的熵目标
Abstract
We consider the problem of synthesizing control policies that enforce a concentration property on the state distributions of a stochastic system. We present a formalization of this problem in terms of synthesizing strategies for maintaining an entropy-based objective in Markov Decision Processes (MDPs). We first show that even relaxed versions of this problem are complexity-theoretically hard. We then present a sound and (conditionally) relatively complete method to verify and synthesize strategies for such entropy objectives. The main challenge is the non-linear nature of such objectives, and our approach addresses this by exploiting and combining ideas from convex duality and invariant synthesis. We also investigate the role of memory and randomization in ensuring entropy objectives. Finally, we implement our ideas to evaluate our approach empirically on a few illustrative benchmarks.
Chinese Translation
我们考虑合成控制策略的问题，以强制随机系统状态分布的集中性特性。我们将该问题形式化为在马尔可夫决策过程（MDPs）中合成维持基于熵的目标的策略。我们首先展示，即使是该问题的放松版本在复杂性理论上也是困难的。然后，我们提出了一种健全且（有条件地）相对完整的方法来验证和合成此类熵目标的策略。主要挑战在于此类目标的非线性特性，我们的方法通过利用和结合凸对偶性和不变合成的思想来解决这一问题。我们还研究了记忆和随机化在确保熵目标中的作用。最后，我们实现了我们的想法，以在一些示例基准上对我们的方法进行实证评估。
cs.AI / 79 / 2606.21740
Training the Orchestrator: A Supervised Approach to End-to-End PDDL Planning with LLM Agents
训练协调者：一种基于监督的端到端 PDDL 规划与 LLM 代理的方法
Abstract
Translating natural-language planning intent into verified plans is a longstanding challenge: people communicate goals in language, while classical planners require formal PDDL specifications. Recent agentic frameworks bridge this gap by orchestrating a pool of specialized repair agents inside a verifier-checked refinement loop, but the orchestrator at the centre is itself a prompted frontier LLM, paying a frontier-LLM API call at every refinement step. We present HALO (Hybrid Agent-Learned Orchestrator), which trains the orchestrator from refinement trajectories that an external verifier has certified as ending in valid plans, across 11 PDDL domains. HALO pairs a small QLoRA-tuned policy with three hardcoded rules for trivially decidable selections, and operates over an expanded 21-agent action space. Unlike approaches that prompt a frontier LLM at every step or learn an orchestrator from sparse end-of-episode rewards, our key observation is that the verifier already provides strong guidance: every accepted trajectory is a sequence of demonstrably correct (state, agent) decisions, directly usable as supervision. Across PlanBench, Natural Plan, and classical planning benchmarks, HALO matches or exceeds the GPT-5-mini prompted baseline on success rate, sits within three percentage points of the stronger Gemini-3-Flash prompted baseline, reduces orchestration cost by more than an order of magnitude (\$0.18 to \$0.004 per task against GPT-5-mini, roughly 45$\times$ cheaper; roughly 15$\times$ cheaper than Gemini-3-Flash), and cuts total LLM calls per episode by 40 to 50 percent.
Chinese Translation
将自然语言规划意图转化为经过验证的计划一直是一个长期挑战：人们用语言交流目标，而经典规划器则需要正式的 PDDL 规范。最近的代理框架通过在验证器检查的细化循环中协调一组专门的修复代理来弥合这一差距，但中心的协调者本身是一个被提示的前沿 LLM，在每个细化步骤中都需要调用前沿 LLM API。我们提出了 HALO（混合代理学习协调者），它从外部验证器认证为以有效计划结束的细化轨迹中训练协调者，涵盖 11 个 PDDL 领域。HALO 将一个小型的 QLoRA 调优策略与三个用于简单可判定选择的硬编码规则相结合，并在扩展的 21 代理动作空间上运行。与在每一步都提示前沿 LLM 或从稀疏的剧终奖励中学习协调者的方法不同，我们的关键观察是验证器已经提供了强有力的指导：每个被接受的轨迹都是一系列可证明正确的（状态，代理）决策，直接可用作监督。在 PlanBench、Natural Plan 和经典规划基准测试中，HALO 的成功率与 GPT-5-mini 提示基线相匹配或超过，且与更强的 Gemini-3-Flash 提示基线相差仅三个百分点，减少了协调成本超过一个数量级（从 GPT-5-mini 的每个任务 0.18 美元降至 0.004 美元，便宜约 45 倍；比 Gemini-3-Flash 便宜约 15 倍），并将每集中的 LLM 调用次数减少了 40% 至 50%。
cs.AI / 80 / 2606.21832
AgentCAT: Simulating Computerized Adaptive Testing via Multi-Agent Large Language Models
AgentCAT：通过多智能体大型语言模型模拟计算机自适应测试
Abstract
Computerized Adaptive Testing (CAT), as a key technology for personalized education, aims to accurately assess examinee proficiency by retrieving exercises dynamically matching current ability estimates. However, existing CAT research is constrained by limitations of static offline data and isolated component optimization. Restricted by partial labels in offline logs, researchers degrade the dynamic assessment process into static sequence prediction. Current research focuses on isolated perspectives, e.g., selection or diagnosis, neglecting the overall CAT interaction process. To address this, we propose AgentCAT, a Large Language Model-based multi-agent simulation system, to construct a high-fidelity benchmarking environment for dynamic testing. This framework comprises three modules: (1) The examinee agent with memory retrieval and Chain-of-Thought reasoning simulates responses based on cognitive profiles; (2) The selection agent uses coarse-to-fine bucketing and knowledge graph exploration to balance local difficulty and global coverage; (3) The supervisor uses dual-auditing and robust update to ensure convergence and validity. To validate the framework, we evaluated on two real-world datasets across three dimensions: macro-level ability convergence, micro-level interaction logic, and data sparsity resilience. Results show AgentCAT achieves effective ability estimation, and its selection strategy balances difficulty adaptation and instructional coherence, aligning with human pedagogical intuition.
Chinese Translation
计算机自适应测试（CAT）作为个性化教育的关键技术，旨在通过动态获取与当前能力估计相匹配的练习，准确评估考生的能力。然而，现有的CAT研究受到静态离线数据和孤立组件优化的限制。由于离线日志中部分标签的限制，研究人员将动态评估过程简化为静态序列预测。目前的研究集中于孤立的视角，例如选择或诊断，忽视了整体CAT交互过程。为了解决这一问题，我们提出了AgentCAT，一个基于大型语言模型的多智能体模拟系统，构建一个高保真度的动态测试基准环境。该框架包括三个模块：（1）考生智能体利用记忆检索和思维链推理，根据认知特征模拟响应；（2）选择智能体采用粗到细的分桶和知识图谱探索，平衡局部难度和全局覆盖；（3）监督者使用双重审计和稳健更新，确保收敛性和有效性。为了验证该框架，我们在两个真实世界数据集上进行了评估，涵盖三个维度：宏观能力收敛、微观交互逻辑和数据稀疏性韧性。结果表明，AgentCAT实现了有效的能力估计，其选择策略平衡了难度适应性和教学连贯性，与人类教学直觉相一致。
cs.AI / 81 / 2606.21843
Measuring What Persists: Conditioning Mechanisms and a Geometric Framework for AI Agent Identity
测量持久性：AI代理身份的条件机制与几何框架
Abstract
AI agents in long-context applications drift from their specified identity. Current methods detect this only after qualitative degradation is visible. We present a geometric framework for measuring identity structure using $\sqrt{\mathrm{JSD}}$ metric spaces and magnitude homology from enriched category theory, where identity is non-geodesic structure and drift is its relaxation toward the geodesic. Validated on a persistent AI agent, the framework's strongest empirical finding is a two-mechanism conditioning structure: cross-condition distances reveal an identity-vacuum cluster where the identity specification fills a behavioral void, and a safety-basin cluster where it displaces from post-training attractors. An equilateral probe baseline confirms that the identity specification creates measurable behavioral richness (55 unique response patterns vs. 1 for the base model) at maximum probe separation. A first-order perturbation theory for equilateral configurations predicts magnitude changes from perimeter changes alone, with shape perturbations first-order cancelled by the $S_n$ symmetry; the formula is self-consistent at the observed perturbation amplitudes. A drift experiment measuring magnitude decrease under context pressure was subsequently found to reflect repetitive-padding artifacts rather than genuine context-length drift; diverse padding produces no measurable deformation through 150K tokens. The magnitude homology framework's full diagnostic promise -- detecting anisotropic contraction and structural collapse via homological simplification -- is architecturally grounded in the perturbation theory and selection rules but remains empirically unconfirmed.
Chinese Translation
在长上下文应用中，AI代理的身份会偏离其指定身份。目前的方法仅在可见的定性退化后才能检测到这一点。我们提出了一种几何框架，通过使用$ ext{JSD}$（Jensen-Shannon Divergence）度量空间和来自丰富范畴理论的幅度同调来测量身份结构，其中身份被视为非测地结构，而漂移则是其向测地线的放松。在对一个持久AI代理进行验证后，该框架的最强实证发现是一个双机制条件结构：交叉条件距离揭示了一个身份真空簇，在该簇中，身份规范填补了行为空白；而安全基底簇则显示其从训练后吸引子中偏移。等边探针基线确认身份规范在最大探针分离下创造了可测量的行为丰富性（55种独特响应模式对比基线模型的1种）。针对等边配置的一阶扰动理论预测，仅通过周长变化就能预测幅度变化，而形状扰动则因$S_n$对称性而被一阶抵消；该公式在观察到的扰动幅度下是自洽的。随后进行的一项漂移实验测量了在上下文压力下的幅度下降，发现其反映的是重复填充伪影，而非真正的上下文长度漂移；多样的填充在150K个标记中未产生可测量的变形。幅度同调框架的全面诊断潜力——通过同调简化检测各向异性收缩和结构崩溃——在扰动理论和选择规则中有架构基础，但尚未得到实证确认。
cs.AI / 82 / 2606.21867
ForEx: A Formal Verification Framework for Explainable Reasoning in Logical Fallacy Detection and Annotation
ForEx：一种用于逻辑谬误检测和注释的可解释推理的形式验证框架
Abstract
Current evaluations of Large Language Models (LLMs) on logical fallacy detection focus on predicted labels, but do not establish whether those labels are supported by the reasoning the models provide. We propose ForEx (Formal Verification for Explainable Reasoning), a framework that translates LLM-generated explanations into Lean4 and verifies whether the translated rationale is derivable under encoded premises, not the logical validity of the original natural language argument. To distinguish prediction outcomes from the formal status of the supporting reasoning, we introduce the LLM Argument Verification Matrix, which separates label consistency from formal verification status. Experiments on LOGIC-Climate show that over 90% of LLM outputs can be translated into formal reasoning chains that pass verification, while agreement with human annotations remains around 20%. These results expose a systematic gap between formal derivability and label agreement, a distinction invisible to prediction-based metrics. ForEx moves LLM evaluation beyond label correctness toward machine-checkable analysis of formalized reasoning chains.
Chinese Translation
当前对大型语言模型（LLMs）在逻辑谬误检测中的评估主要集中在预测标签上，但并未确定这些标签是否得到了模型提供的推理支持。我们提出了ForEx（可解释推理的形式验证），这是一个将LLM生成的解释翻译为Lean4并验证翻译后的推理是否在编码前提下可推导的框架，而不是验证原始自然语言论证的逻辑有效性。为了区分预测结果与支持推理的形式状态，我们引入了LLM论证验证矩阵，该矩阵将标签一致性与形式验证状态分开。在LOGIC-Climate上的实验表明，超过90%的LLM输出可以被翻译为通过验证的形式推理链，而与人工注释的一致性仍然约为20%。这些结果揭示了形式可推导性与标签一致性之间的系统性差距，这一差距在基于预测的指标中是不可见的。ForEx将LLM评估从标签正确性扩展到对形式化推理链的机器可检查分析。
cs.AI / 83 / 2606.21877
AgentRiskBOM: A Risk-Scoping Security Bill of Materials for Agentic AI Systems
AgentRiskBOM：面向代理型人工智能系统的风险范围安全材料清单
Abstract
Agentic AI systems retrieve private context, invoke tools, write files, call external services, coordinate with other agents, and may act without human approval. Existing bill of materials artifacts improve transparency for dependencies, model metadata, and training provenance, but leave an agentic transparency gap: capability opacity, the absence of a structured account of what a deployed agent can access, remember, change, delegate, and prove afterward. This paper introduces AgentRiskBOM, a security BOM for risk-scoping tool-using AI agents. It is an additive layer over SBOM, AIBOM, and MLBOM artifacts, referencing them where authoritative while adding fields for runtime authority: autonomy, tool permissions, memory, credential scope, approval gates, audit signals, inter-agent communication, and external action capability. We implement AgentRiskBOM as a JSON-schema artifact with a reproducible corpus, risk scenarios, scorer, diff detector, control mapper, and reports. We evaluate AgentRiskBOM on 13 open-source agents spanning coding, RAG, and multi-agent archetypes, plus 52 risk scenarios across 14 categories. The schema validates all 13 corpus artifacts. Coverage analysis gives AgentRiskBOM a native-equivalent score of 14 across 16 capability dimensions, vs. 1 for SBOM, 1.5 for AIBOM and 2 for MLBOM. Across modeled risk categories, AgentRiskBOM exposes 100% risk-category visibility vs. 10.5% for SBOM-like and 20.9% for AIBOM-like views. To test agentic authority drift, we inject 33 structured deployment mutations; the diff detector identifies the correct change type for all mutations. A secondary penalty-based scorer yields a Spearman correlation of 0.73 with the primary scorer, supporting rank-level consistency while showing that thresholds require human calibration. The results show that agentic AI security needs a machine-readable authority-and-risk artifact before incidents occur.
Chinese Translation
代理型人工智能系统能够检索私人上下文、调用工具、写入文件、调用外部服务、与其他代理协调，并且可能在没有人类批准的情况下行动。现有的材料清单文献提高了对依赖关系、模型元数据和训练来源的透明度，但在代理透明度方面存在缺口：能力不透明，即缺乏对已部署代理能够访问、记忆、改变、委托和随后证明的内容的结构化说明。本文介绍了AgentRiskBOM，这是一个用于风险范围工具使用的人工智能代理的安全材料清单。它是对SBOM、AIBOM和MLBOM文献的附加层，引用它们的权威性，同时增加了运行时权限的字段：自主性、工具权限、记忆、凭证范围、批准门、审计信号、代理间通信和外部行动能力。我们将AgentRiskBOM实现为一个JSON-schema文献，包含可重复的语料库、风险场景、评分器、差异检测器、控制映射器和报告。我们在13个开源代理上评估AgentRiskBOM，这些代理涵盖了编码、RAG和多代理原型，以及14个类别中的52个风险场景。该模式验证了所有13个语料库文献。覆盖分析使AgentRiskBOM在16个能力维度上获得了14的原生等效分数，而SBOM为1，AIBOM为1.5，MLBOM为2。在建模的风险类别中，AgentRiskBOM展现了100%的风险类别可见性，而SBOM类视图为10.5%，AIBOM类视图为20.9%。为了测试代理权限漂移，我们注入了33个结构化部署变异；差异检测器为所有变异识别了正确的变化类型。基于惩罚的次级评分器与主要评分器的Spearman相关性为0.73，支持等级一致性，同时显示阈值需要人类校准。结果表明，代理型人工智能安全在事件发生之前需要一个机器可读的权限和风险文献。
cs.AI / 84 / 2606.21891
Learning the ARTS of Search for Automated Discovery
自动发现的搜索艺术学习
Abstract
Scientific discovery can be formulated as an iterative search process over the space of hypotheses and experiments. Contemporary methods navigate this space using heuristics such as MCTS. These algorithms conflate the merit of a hypothesis with the quality of its experimental execution. A promising hypothesis with preliminary execution is therefore ranked below a modest hypothesis whose execution is refined. Moreover, prior methods prune the search logs as the search progresses because the accumulated history outgrows the context window. We propose Agentic Reasoning for Tree Search (ARTS), where we deploy a reasoning language model to navigate this space. The model inspects prior execution logs, diagnoses whether earlier failures arose from faulty implementations or bad hypotheses, and selects the hypothesis to build on next. To mitigate challenges with context length, ARTS uses test-time training to instill the knowledge of search tree in the model weights. Across 22 tasks from MLGym and MLEBench, we show that ARTS outperforms leading algorithms, with over 15.3% relative improvement in the normalized score. With test-time training we show that a Qwen3-4B agent can match performance with closed-source frontier models like Gemini-3 Pro and GPT o3-reasoning with upto 5x lower inference cost. We further observe that on partially observable RL tasks, the test-time trained Qwen3-4B scientist surpasses ARTS with the o3 scientist by rediscovering the human-best recurrent-memory solution that heuristic methods prune away.
Chinese Translation
科学发现可以被表述为在假设和实验空间中的迭代搜索过程。当前的方法使用启发式算法如蒙特卡罗树搜索（MCTS）来导航这一空间。这些算法将假设的优点与其实验执行的质量混为一谈。因此，一个有前景的假设在初步执行后被评估得低于一个执行精细的普通假设。此外，先前的方法在搜索进展时会修剪搜索日志，因为累积的历史超出了上下文窗口。我们提出了树搜索的代理推理（Agentic Reasoning for Tree Search，ARTS），在其中我们部署了一个推理语言模型来导航这一空间。该模型检查先前的执行日志，诊断早期失败是由于错误的实现还是不良的假设，并选择下一个要构建的假设。为了缓解上下文长度带来的挑战，ARTS使用测试时训练将搜索树的知识灌输到模型权重中。在来自MLGym和MLEBench的22个任务中，我们展示了ARTS在标准化得分上相较于领先算法有超过15.3%的相对提升。通过测试时训练，我们展示了Qwen3-4B代理能够与封闭源前沿模型如Gemini-3 Pro和GPT o3-reasoning匹敌，同时推理成本降低至最多5倍。我们进一步观察到，在部分可观测的强化学习任务中，经过测试时训练的Qwen3-4B科学家通过重新发现人类最佳的递归记忆解决方案，超越了ARTS与o3科学家的表现，而该解决方案被启发式方法修剪掉。
cs.AI / 85 / 2606.21963
Holmes: Multimodal Agentic Diagnosis for Mixed-Language Mobile Crashes at Industrial Scale
Holmes：用于工业规模混合语言移动崩溃的多模态自主诊断
Abstract
Diagnosing mobile crashes in ultra-large-scale industrial applications is a formidable challenge due to the sheer volume of code, the complexity of mixed-language environments, and the inability to reproduce failures locally. Traditional static analysis struggles with scalability, while existing LLM-based agents often rely on reproducible environments unavailable in post-mortem scenarios. We present Holmes, a multi-agent system that automates root cause analysis by synthesizing multimodal runtime signals--stack traces, logs, and thread states--to reconstruct failure contexts without reproduction. Holmes introduces a hierarchical Retrieve-Explore-Reason architecture that leverages low-level artifacts (e.g., registers, assembly) to bridge the semantic gap between open-source business logic and closed-source system frameworks. By dynamically compressing the search space using runtime clues, Holmes precisely navigates 70-million-line codebases to identify non-local defects. Evaluated on real-world crashes from WeChat, Holmes achieves 87.6% accuracy in function-level fault localization and reduces average investigation time by over 98% (to ~77 seconds), demonstrating its effectiveness in transforming labor-intensive debugging into an efficient verification workflow.
Chinese Translation
在超大规模工业应用中诊断移动崩溃是一项艰巨的挑战，原因在于代码量庞大、混合语言环境的复杂性以及无法在本地重现故障。传统的静态分析在可扩展性方面面临困难，而现有的基于大语言模型（LLM）的代理通常依赖于在事后场景中不可用的可重现环境。我们提出了Holmes，一个多代理系统，通过综合多模态运行时信号——堆栈跟踪、日志和线程状态——来自动化根本原因分析，以在不重现故障的情况下重构故障上下文。Holmes引入了一种分层的检索-探索-推理架构，利用低级工件（例如寄存器、汇编）来弥合开源业务逻辑与闭源系统框架之间的语义差距。通过使用运行时线索动态压缩搜索空间，Holmes能够精准地在7000万行代码中导航，以识别非本地缺陷。在对来自微信的真实崩溃进行评估时，Holmes在功能级故障定位中实现了87.6%的准确率，并将平均调查时间减少了98%以上（至约77秒），展示了其将劳动密集型调试转变为高效验证工作流程的有效性。
cs.AI / 86 / 2606.21971
REBA: A Revealed Belief Automaton Framework for Online Planning in Continuous POMDPs
REBA：一种用于连续部分可观测马尔可夫决策过程在线规划的揭示信念自动机框架
Abstract
Online planning in continuous partially observable Markov decision processes (POMDPs) using $\omega$-regular specifications requires handling continuous belief dynamics within the finite symbolic memory in order to track temporal progress. Existing methods based on either direct search in belief space or predefined discrete abstractions suffer from drawbacks, e.g., lack of symbolic memory for long-horizon logical progress or difficult to certify from noisy online beliefs. As such, obtaining reliable symbolic states online from continuous observations remains a challenge. To address this issue, we introduce the Revealed Belief Automaton (REBA), an event-driven framework that advances the research from global belief-space discretization to a fundamental new way of thinking, namely online certification of revelation events. Specifically, we propose an online revelation method that, through information-theoretic gates, can dynamically analyse and establish belief abstraction from the continuous belief space by discovering reliable anchors among noisy beliefs. We then develop an incremental topology adaptation mechanism over the certified anchors to realise the online finite Belief Automaton. By combining with the $\omega$-regular specification, REBA is able to support formal parity policy synthesis without a predefined discrete abstraction, which in turn can guide the Monte Carlo Tree Search process to perform online search beyond its local horizon. In addition, we design an error decomposition analysis which can assess the effectiveness and reliability of this discrete guidance for the underlying continuous POMDP. Empirical evaluations in patrolling and navigation scenarios show that REBA matches or exceeds all evaluated baselines, with primary metric gains of +17.0\% to +47.4\% over state-of-the-art approaches.
Chinese Translation
在使用 $ ext{ω}$-正则规范的连续部分可观测马尔可夫决策过程（POMDPs）中进行在线规划，需要在有限的符号记忆中处理连续信念动态，以跟踪时间进展。现有的方法基于直接在信念空间中的搜索或预定义的离散抽象，存在一些缺陷，例如缺乏用于长时间逻辑进展的符号记忆或难以从嘈杂的在线信念中进行认证。因此，从连续观察中在线获取可靠的符号状态仍然是一个挑战。为了解决这个问题，我们引入了揭示信念自动机（REBA），这是一种事件驱动的框架，推动了从全局信念空间离散化的研究，提出了一种根本的新思路，即在线认证揭示事件。具体而言，我们提出了一种在线揭示方法，通过信息论门，可以动态分析并建立来自连续信念空间的信念抽象，通过在嘈杂信念中发现可靠的锚点。然后，我们在认证的锚点上开发了一种增量拓扑适应机制，以实现在线有限信念自动机。通过与 $ ext{ω}$-正则规范结合，REBA能够支持形式化的奇偶策略合成，而无需预定义的离散抽象，这反过来可以指导蒙特卡洛树搜索过程进行超越其局部视野的在线搜索。此外，我们设计了一种误差分解分析，可以评估这种离散指导在基础连续POMDP中的有效性和可靠性。在巡逻和导航场景中的实证评估表明，REBA的表现与所有评估的基线相匹配或超越，主要指标提升幅度为 +17.0\% 至 +47.4\% 超过最先进的方法。
cs.AI / 87 / 2606.21972
Human vs Machine Mathematical Difficulty on Project Euler: An Experimental Analysis
人类与机器在欧拉项目中的数学难度：实验分析
Abstract
We study how the effort and success probability of frontier AI systems scale with human difficulty on problems from Project Euler, an online platform of computational mathematics problems. Our dataset, from the MathArena benchmark, consists of 3840 attempts across 50 problems and 26 model configurations, with problem difficulty measured by the site's public human solve times. Motivated by a proposal of Timothy Gowers, we test a power-law relation $t_{\text{machine}} = a \cdot t_{\text{human}}^b$ between generated-token cost per successful answer and human time, and find $b < 1$ for 20 of the 25 models with usable fits, including the strongest base models; this operationalization therefore does not support an earlier prediction that machines scale worse than humans with difficulty. We also investigate whether success probability on the tested problems can be modeled by a simple exponential decay $p_{\text{success}} = e^{c t_{\text{human}}}$, predicting a linear relation between $\log p_{\text{success}}$ and $t_{\text{human}}$. Using a binning approach for data aggregation we find moderate empirical support (median bin-level $R^2 = 0.92$ across the 22 best-covered configurations) for this model. Following METR, we also fit logistic success curves and extract 50\% task-length horizons $h_{50}$; the strongest configurations in our 20 April 2026 snapshot reach roughly $2.5$--$4.3$ hours on our fastest-five human baseline, with a log-linear fit through the state-of-the-art frontier giving a descriptive doubling time of about $75$~days for the SOTA $h_{50}$.
Chinese Translation
我们研究了前沿人工智能系统的努力程度和成功概率如何与来自欧拉项目（Project Euler）的问题的人工难度相对应，该项目是一个在线计算数学问题的平台。我们的数据集来自MathArena基准，包含3840次尝试，涵盖50个问题和26种模型配置，问题难度通过该网站公开的人类解题时间进行测量。受到Timothy Gowers提议的启发，我们测试了生成令牌的成功答案成本与人类时间之间的幂律关系 $t_{ ext{machine}} = a imes t_{ ext{human}}^b$，并发现25个可用拟合模型中有20个的 $b < 1$，包括最强的基础模型；因此，这一操作化并不支持早期预测，即机器在难度上比人类的表现更差。我们还研究了在测试问题上的成功概率是否可以通过简单的指数衰减模型 $p_{ ext{success}} = e^{c t_{ ext{human}}}$ 来建模，预测 $ ext{log} p_{ ext{success}}$ 与 $t_{ ext{human}}$ 之间的线性关系。通过数据聚合的分箱方法，我们发现该模型有适度的实证支持（在22个覆盖最佳配置中的中位数箱级 $R^2 = 0.92$）。根据METR，我们还拟合了逻辑成功曲线并提取了50 ext{%}任务长度的视野 $h_{50}$；在我们2026年4月20日的快照中，最强的配置在我们最快的五个人类基线中大约达到 $2.5$--$4.3$ 小时，通过最先进的前沿的对数线性拟合给出SOTA $h_{50}$ 的描述性翻倍时间约为 $75$ 天。
cs.AI / 88 / 2606.21975
IRumAI: Reinforcement Learning for Indian Rummy
IRumAI：用于印度拉米的强化学习
Abstract
Despite its massive player base and complex hidden-information dynamics, Indian Rummy has received no reinforcement learning attention. Existing agents rely on combinatorial search, which is tactically strong but slow at inference. We present IRumAI, the first RL agent for the domain. IRumAI integrates Proximal Policy Optimization (PPO), meld-aware observation encoding, deadwood-driven reward shaping, and a dual-branch convolutional architecture. IRumAI is RL-trained solely against weak heuristics, after a one-time behaviour-cloning warm-start on stronger demonstration data. It generalises to defeat the entire baseline hierarchy, including a 53.9% win rate against the strongest search-based opponent unseen during RL training. Bypassing explicit search, IRumAI requires just 0.33 ms per action, which is over 7,000x faster than the state-of-the-art heuristic. Ablations validate our architectural choices, and linear probing reveals that the network implicitly models the opponent's hidden hand from public interactions.
Chinese Translation
尽管印度拉米拥有庞大的玩家基础和复杂的隐信息动态，但该领域尚未受到强化学习的关注。现有的智能体依赖于组合搜索，这在战术上表现强劲，但推理速度较慢。我们提出了IRumAI，这是该领域首个强化学习智能体。IRumAI整合了近端策略优化（Proximal Policy Optimization, PPO）、基于牌型的观察编码、基于死木的奖励塑造以及双分支卷积架构。IRumAI仅在较弱的启发式算法下进行强化学习训练，之前经过一次性行为克隆热启动，使用较强的示范数据。它能够推广并击败整个基线层次，包括在强化学习训练期间未见过的最强搜索对手，赢率达到53.9%。IRumAI绕过了显式搜索，每个动作仅需0.33毫秒，这比当前最先进的启发式算法快超过7000倍。消融实验验证了我们的架构选择，线性探测显示网络从公共交互中隐式建模了对手的隐藏手牌。
cs.AI / 89 / 2606.22000
CFAgentBench: A Reproducible Environment and Benchmark for Autonomous Construction-Finance Agents
CFAgentBench：一个可重复的自主建筑金融代理环境与基准
Abstract
We introduce CFAgentBench, a reproducible, self-hostable environment and benchmark for autonomous construction-finance agents: a CFO/controller-class agent operating across the real software stack a US construction finance team runs - ERP, project management, email, documents, pay applications, payroll, certified payroll, lien waivers, and bank/treasury portals. It contains 1,014 machine-gradeable task specifications across 8 domains and 77 families, every family grounded in a real source; a self-validated subset of 40 tasks (54 with a project-management extension) is compiled into oracle-validated executable evaluators, the runnable suite reported here. Following WebArena, the benchmark runs on an executable environment rather than static traces: 35 mock applications (31 reconciled to one company book, plus 4 PM platforms) over 9 archetypes, each implementing a uniform self-hostable app contract, so every task is graded by functional correctness - a state diff plus forbidden-side-effect checks plus required-output regexes - with an LLM judge used only for reply quality, never as reward. A distinguishing principle is a money-movement guard: 278 instances embed a payment, payroll, e-signature, or e-filing step where the correct behavior is to stop and stage for human approval, and executing even the correct transaction fails the task. The public split (n=711) is sized for a 95% Wilson half-width of +/-4.1%; a private, contamination-protected split (n=303) is reserved for remote scoring. In a first three-model open-weight sweep (k=5), the strongest agent reaches pass^1 = 0.67 but only pass^5 = 0.38 - losing 43% of its successes when required to repeat them under temperature-0 decoding. The within-model pass^1 to pass^5 collapse and sharp per-domain heterogeneity are clear evidence that single-attempt accuracy overstates deployable construction-finance competence.
Chinese Translation
我们介绍了CFAgentBench，这是一个可重复、自托管的环境和基准，专为自主建筑金融代理而设计：一个在美国建筑金融团队实际使用的软件栈中运行的首席财务官/控制器级代理，包括企业资源规划（ERP）、项目管理、电子邮件、文档、支付申请、工资单、认证工资单、留置权豁免和银行/财政门户。该基准包含1,014个机器可评估的任务规范，涵盖8个领域和77个家族，每个家族均基于真实来源；一个自我验证的40个任务子集（扩展到54个项目管理任务）被编译成经过oracle验证的可执行评估器，形成本文报告的可运行套件。基准遵循WebArena的原则，运行在可执行环境中，而非静态轨迹：35个模拟应用（31个与一家公司的账簿对账，外加4个项目管理平台）覆盖9个原型，每个原型实现统一的自托管应用合同，因此每个任务都通过功能正确性进行评分——状态差异加上禁止副作用检查和所需输出的正则表达式——使用大型语言模型（LLM）评判仅用于回复质量，而不作为奖励。一个显著的原则是资金流动保护：278个实例嵌入了支付、工资、电子签名或电子申报步骤，其中正确的行为是停止并等待人类批准，即使执行正确的交易也会导致任务失败。公共拆分（n=711）的规模设定为95% Wilson半宽度为±4.1%；一个私有的、受污染保护的拆分（n=303）则保留用于远程评分。在第一次三模型开放权重测试（k=5）中，最强的代理达到pass^1 = 0.67，但仅pass^5 = 0.38——在要求重复成功时损失了43%的成功率。模型内的pass^1与pass^5的崩溃以及每个领域的明显异质性清楚地表明，单次尝试的准确性夸大了可部署的建筑金融能力。
cs.AI / 90 / 2606.22030
Nous: A Predictive World Model for Long-Term Agent Memory
Nous：一种用于长期智能体记忆的预测世界模型
Abstract
We present Nous, a novel agent memory architecture grounded in the principle that knowledge is prediction, not storage. Rather than persisting facts as database records, vector embeddings, or knowledge-graph triples, Nous maintains a predictive world model: a collection of categorical probability distributions, called dimensions, one per entity-attribute pair observed in conversation. Each incoming observation is scored by its information-theoretic surprise S = -log2 P(obs | D), and the distribution is updated via a closed-form Bayesian posterior. The primary stored artifact is the delta, a record of the shift from prior to posterior belief, rather than the fact itself. Forgetting emerges naturally as entropy decay toward the uniform distribution, and identity resolution is handled through mutual information between entity dimension sets. Evaluated on the LoCoMo long-term conversational memory benchmark across ten conversations (1,540 questions) using GPT-4o-mini as backbone, Nous achieves F1 of 63.50 (single-hop), 55.32 (multi-hop), 58.57 (temporal), and 62.50 (open-domain). Against A-MEM's self-reported GPT-4o-mini numbers, Nous shows substantial gains in three of four categories, though we note that independent citations of A-MEM's results disagree with each other on category assignment, a reproducibility issue we discuss openly rather than resolve unilaterally. We additionally compare against BeliefMem, a concurrently developed system built on the same core premise of belief-based rather than deterministic memory; on the same benchmark and backbone, Nous's self-reported numbers exceed BeliefMem's self-reported numbers on all four categories, though we flag several uncontrolled differences between the two evaluation pipelines that prevent this from being a fully controlled comparison. Nous requires no external vector database or graph engine.
Chinese Translation
我们提出了Nous，一种新颖的智能体记忆架构，其基础原则是知识是预测，而非存储。Nous并不将事实作为数据库记录、向量嵌入或知识图谱三元组持久化，而是维护一个预测世界模型：一个类别概率分布的集合，称为维度，每个维度对应于在对话中观察到的实体-属性对。每个输入观察通过其信息论惊讶度 S = -log2 P(obs | D) 进行评分，并通过封闭形式的贝叶斯后验进行更新。主要存储的文档是增量，即从先前信念到后验信念的转变记录，而非事实本身。遗忘自然地表现为熵向均匀分布的衰减，而身份解析则通过实体维度集之间的互信息进行处理。在使用GPT-4o-mini作为基础的十个对话（1,540个问题）上评估的LoCoMo长期对话记忆基准中，Nous在单跳、双跳、时间和开放域四个类别中分别达到了63.50、55.32、58.57和62.50的F1分数。与A-MEM自报的GPT-4o-mini数据相比，Nous在四个类别中的三个类别显示出显著的提升，尽管我们注意到A-MEM结果的独立引用在类别分配上存在不一致，这是一个我们公开讨论而非单方面解决的可重复性问题。我们还与BeliefMem进行了比较，后者是一个基于信念而非确定性记忆的同时开发系统；在相同的基准和基础上，Nous自报的结果在所有四个类别中均超过了BeliefMem的自报结果，尽管我们指出两者评估流程之间存在若干未控制的差异，阻止了这次比较的完全控制。Nous不需要外部向量数据库或图形引擎。
cs.AI / 91 / 2606.22039
Attractor Domain Theory: A Mathematical Framework for Cardiovascular Attractor Analysis with Wearable Photoplethysmography (PPG) Validation
吸引子领域理论：一种用于心血管吸引子分析的数学框架，结合可穿戴光电容积描记（PPG）验证
Abstract
The cardiovascular system evolves along a bounded trajectory in physiological state space that converges to a compact geometric object: the cardiac attractor. A wearable photoplethysmograph (PPG) or electrocardiograph (ECG) observes a one-dimensional projection of this attractor; by Takens' embedding theorem, delay coordinates reconstruct its full geometry. Three decades of nonlinear cardiac dynamics have extracted Lyapunov exponents, recurrence statistics, and sample entropy from reconstructed attractors, yet no principled account exists of which attractor properties capture which cardiovascular quantities, or why, leaving feature selection as a search problem and negative results uninterpretable. We introduce Attractor Domain Theory (ADT), which proves that the reconstructed attractor's information partitions into three mutually non-redundant domains: the Geometry Domain G (delay embedding; native capability: artifact rejection), the Ergodic Domain S (asymptotic statistical invariants; native capability: stability estimation), and the Variational Domain V (finite-time Lyapunov exponent field; native capability: hemodynamic inference). We prove a Domain Sufficiency Theorem (the Parseval analog for attractor information) and establish that three domains are necessary and sufficient. Geometry Domain validation via the SCSI framework across 176,742 PPG segments from four datasets yields AUC = 0.757 [0.686-0.828] and NPV = 0.966 after correcting three systematic evaluation artifacts (+0.179 net inflation). Ablation confirms C_NL as the dominant Geometry Domain component (Delta AUC = -0.413) and intra-domain redundancy across five components.
Chinese Translation
心血管系统沿着生理状态空间中的有界轨迹演变，最终收敛于一个紧致的几何对象：心脏吸引子。可穿戴光电容积描记仪（PPG）或心电图（ECG）观察到该吸引子的一个一维投影；根据Takens的嵌入定理，延迟坐标重构其完整几何形状。三十年来，非线性心脏动力学从重构的吸引子中提取了Lyapunov指数、重现统计和样本熵，但尚无原则性的解释说明哪些吸引子特性捕捉了哪些心血管量，或其原因，导致特征选择成为一个搜索问题，负面结果难以解释。我们引入吸引子领域理论（Attractor Domain Theory, ADT），证明重构吸引子的信信息分为三个相互独立的领域：几何领域G（延迟嵌入；原生能力：伪影拒绝），遍历领域S（渐近统计不变性；原生能力：稳定性估计），以及变分领域V（有限时间Lyapunov指数场；原生能力：血流动力学推断）。我们证明了领域充分性定理（吸引子信息的Parseval类比），并确立了这三个领域是必要且充分的。通过SCSI框架对来自四个数据集的176,742个PPG片段进行的几何领域验证，得到了AUC = 0.757 [0.686-0.828]和NPV = 0.966，经过纠正三个系统评估伪影后（+0.179的净膨胀）。消融实验确认C_NL是主导的几何领域成分（Delta AUC = -0.413），并且在五个成分之间存在领域内冗余。
cs.AI / 92 / 2606.22043
When Does a Video-Language Model Stop Watching? Reward Strength Controls the Formation and Reversal of Visual Shortcuts in Multimodal RLVR
视频语言模型何时停止观看？奖励强度控制多模态RLVR中视觉捷径的形成与逆转
Abstract
Reinforcement learning with verifiable rewards (RLVR) is increasingly applied to large vision-language models (LVLMs), yet outcome-only optimization can drive a model to stop attending to the video and instead exploit linguistic priors -- a failure we call a visual shortcut. While the existence of such perception bypass is by now documented, how it forms, whether it can be undone, and when intervention still helps remain open. We treat the strength of a grounding penalty, lambda, as a control knob and characterize the formation-reversal dynamics of visual shortcuts along the training time axis. On a held-out, out-of-distribution diagnostic set, we find: (i) a sharp onset -- shortcut reliance emerges abruptly over a narrow window of optimization steps and is robust across random seeds; (ii) a monotone dose-response -- increasing lambda progressively suppresses the shortcut, and at an intermediate dose the trajectory first forms and then reverses the shortcut, exposing a hysteresis-like asymmetry between acquiring and removing it; and (iii) a critical intervention window -- applying the penalty before onset arrests shortcut formation, whereas the same penalty applied after consolidation is markedly less effective. Together these results recast visual-shortcut collapse not as a binary defect but as a controllable, time-dependent, and asymmetric process, with direct implications for when and how strongly to regularize multimodal RLVR.
Chinese Translation
可验证奖励的强化学习（RLVR）越来越多地应用于大型视觉语言模型（LVLMs），然而，仅基于结果的优化可能导致模型停止关注视频，而是利用语言先验——我们称之为视觉捷径的失败。尽管这种感知绕过的存在已被记录，但其形成机制、是否可以逆转以及何时干预仍然是未解之谜。我们将基础惩罚的强度（lambda）视为一个控制参数，并描述视觉捷径在训练时间轴上的形成与逆转动态。在一个保留的、分布外的诊断集上，我们发现：（i）急剧的开始——捷径依赖在一个狭窄的优化步骤窗口内突然出现，并且在随机种子之间具有稳健性；（ii）单调的剂量反应——增加lambda逐步抑制捷径，在一个中间剂量下，轨迹首先形成然后逆转捷径，暴露出获取和移除捷径之间的滞后性不对称；（iii）关键干预窗口——在捷径形成之前施加惩罚可以阻止捷径的形成，而在巩固之后施加相同的惩罚则明显效果较差。这些结果共同将视觉捷径崩溃重新定义为一个可控的、时间依赖的和不对称的过程，而不是一个二元缺陷，这对何时以及多强地对多模态RLVR进行正则化具有直接的影响。
cs.AI / 93 / 2606.22085
Can Reasoning Models Detect Changes to their Chains of Thought?
推理模型能否检测其思维链的变化？
Abstract
There are many reasons one may want to edit a model's chain of thought (CoT) -- e.g., to prefill it with reasoning from a stronger model or to remove steps that may yield unsafe outputs. The success of these interventions plausibly depends on a model's inability to notice them, as the model may alter its behavior if it suspects tampering. In this work, we study whether recent reasoning models are able to detect such interventions on their CoTs under a variety of conditions: both during reasoning and after it, and when prefilled both with their own CoTs and with those of other models. Broadly, we find that (i) models exhibit only very modest detection accuracy; (ii) models struggle to identify *how* their CoT was modified; and (iii) models are about as good at detecting changes to their own CoTs as to those of other models.
Chinese Translation
有许多原因可能导致人们希望编辑模型的思维链（CoT）——例如，为其预填充来自更强模型的推理，或去除可能导致不安全输出的步骤。这些干预的成功显然依赖于模型是否能够察觉到这些变化，因为如果模型怀疑被篡改，它可能会改变其行为。在本研究中，我们探讨了近期的推理模型在多种条件下是否能够检测到其思维链上的此类干预：无论是在推理过程中还是推理之后，以及在预填充自身的思维链和其他模型的思维链时。总体而言，我们发现（i）模型的检测准确率仅为非常有限的水平；（ii）模型难以识别其思维链被修改的*方式*；以及（iii）模型在检测自身思维链的变化和其他模型的思维链变化方面的能力相当。
cs.AI / 94 / 2606.22327
Geometry-Aware Online Scheduling for LLM Serving: From Theoretical Bound to System Practice
Abstract
The explosive demand for interactive Large Language Model serving has highlighted the management of the Key-Value cache's dynamic memory footprint as a critical area for performance optimization in inference engines. Modern inference systems overwhelmingly rely on time-centric scheduling heuristics, such as Shortest Job First. However, their theoretical optimality is rooted in traditional schedule modeling, failing to capture the highly dynamic, 2D spatio-temporal geometric growth specific to LLM inference mechanisms. To resolve this, we propose the geometry-aware online scheduling by introducing the Smallest Volume First (SVF) algorithm and its highly efficient variant, 1-bit SVF. Theoretically, we provide a rigorous mathematical foundation for our approach. Utilizing a novel proof methodology, we tighten the worst-case competitive ratio ($\text{CR} \le 48 \rightarrow \text{CR} \le 5$) for SVF with known output lengths. Building upon this core breakthrough, we complete a comprehensive theoretical taxonomy analyzing our algorithms across different traffic scenarios and information availability. Practically, we seamlessly integrate our approach as a plug-and-play layer in vLLM. Extensive evaluations on Llama-3.1 models demonstrate comprehensive performance gains: SVF delivers strong reductions in both average and tail latency, while 1-bit SVF, with merely a single bit information, achieves competitive throughput and latency. This work establishes a theoretically sound and empirically proven approach for resolving memory-constrained scheduling in modern LLM deployments. To facilitate future research, our code is available at https://github.com/Aurora-Kl/Geometry-Aware-Online-Scheduling.git.
cs.AI / 95 / 2606.22330
Hypothesis-Driven Skill Optimization for LLM Agents
基于假设的技能优化方法用于大型语言模型代理
Abstract
External skills can improve action-oriented LLM agents without changing model weights, but persistent skill updates are risky when they are distilled from sparse or noisy trajectories. A plausible reflection may encode a useful procedure, a spurious shortcut, or a rule that the target executor cannot reliably follow. We propose Hypothesis-Driven Skill Optimization (HDSO), a train-free framework in which both the skill curator and the agent executor are frozen inference endpoints. The curator observes executor traces, proposes a falsifiable hypothesis with an explicit validation plan, instantiates the hypothesis as a candidate skill package, validates the package through paired control/treatment executions, reviews behavior differences, and consolidates only supported candidates into an approved repository. The executor consumes approved skills through progressive disclosure, preserving the executor-only path when no skill is selected. On ALFWorld, HDSO improves executor-only baselines by +6.9 Avg. SR points for Qwen3-8B and +4.0 points for Qwen3.6-27B. Under 20% randomly flipped success/failure feedback during skill discovery and validation, HDSO preserves a +7.1-point gain for Qwen3-8B. Transfer and heterogeneous-pair diagnostics further show that validated repositories can be useful beyond the run that produced them, but cross-model curation succeeds only when curator diagnosis, executor capability, and validation evidence align. HDSO provides an auditable skill lifecycle for frozen action agents rather than an unconstrained memory accumulation procedure.
Chinese Translation
外部技能可以在不改变模型权重的情况下提升面向行动的大型语言模型（LLM）代理的表现，但从稀疏或噪声轨迹中提取的持续技能更新存在风险。一个合理的反映可能编码了一个有用的过程、一个虚假的捷径，或者一个目标执行者无法可靠遵循的规则。我们提出了一种基于假设的技能优化方法（Hypothesis-Driven Skill Optimization, HDSO），这是一个无训练的框架，其中技能策划者和代理执行者都是固定的推理端点。策划者观察执行者的轨迹，提出一个可证伪的假设，并制定明确的验证计划，将假设实例化为候选技能包，通过配对的控制/处理执行验证该包，审查行为差异，并将仅支持的候选技能整合到批准的库中。执行者通过渐进式披露使用批准的技能，当没有技能被选择时，保留仅执行者的路径。在ALFWorld上，HDSO使得仅执行者的基线在Qwen3-8B上提高了+6.9平均成功率（Avg. SR）点，在Qwen3.6-27B上提高了+4.0点。在技能发现和验证过程中，随机翻转20%的成功/失败反馈下，HDSO为Qwen3-8B保持了+7.1点的增益。转移和异构对诊断进一步表明，经过验证的库可以在产生它们的运行之外发挥作用，但跨模型的策划只有在策划者诊断、执行者能力和验证证据一致时才能成功。HDSO为固定的行动代理提供了一个可审计的技能生命周期，而不是一个不受限制的记忆积累过程。
cs.AI / 96 / 2606.22363
Reference-Free Assessment of Physical Consistency in World Model-based Video Generation
基于世界模型的视频生成中的无参考物理一致性评估
Abstract
We introduce reference-free measures for evaluating the physical consistency of generated videos, combining relative and absolute approaches to assess fidelity. Although tools like WorldGym or WorldEval enable robotic simulation via video generation, physical fidelity gaps often prevent these environments from accurately reproducing real-world task success rates of VLA models. Unlike existing evaluation methods, which require costly human voting (Elo) or unavailable ground-truth references (FVD), our approach utilizes DROID-SLAM and SEA-RAFT to quantify physical inconsistencies, motivated by WorldScore. Videos filtered using our relative consistency assessment show an improvement in task success rates of over 8%, effectively narrowing the simulation-to-reality gap. Furthermore, our absolute assessment enables spatio-temporal localization, providing visualization of when and where physical artifacts occur.
Chinese Translation
我们提出了一种无参考的评估方法，用于评估生成视频的物理一致性，结合相对和绝对的方法来评估保真度。尽管像 WorldGym 或 WorldEval 这样的工具通过视频生成实现了机器人模拟，但物理保真度的差距常常阻碍这些环境准确再现 VLA 模型在现实世界任务中的成功率。与现有评估方法不同，这些方法需要昂贵的人类投票（Elo）或不可用的真实参考（FVD），我们的方法利用 DROID-SLAM 和 SEA-RAFT 来量化物理不一致性，受 WorldScore 的启发。使用我们的相对一致性评估过滤后的视频显示任务成功率提高了超过 8%，有效缩小了模拟与现实之间的差距。此外，我们的绝对评估能够进行时空定位，提供物理伪影出现的时间和地点的可视化。
cs.AI / 97 / 2606.22375
ARIA: A Causal-Aware Framework for Rescuing LLM Reasoning in Trustworthy Materials Discovery
ARIA：一个因果感知框架，用于在可信材料发现中拯救大语言模型推理
Abstract
Generative models have revolutionized the process of materials discovery, yet they often fail to satisfy underlying physical causality. Through an analysis of Large Language Models (LLMs) augmented with knowledge graphs derived from current literature, we uncover a phenomenon termed contextual tunneling, where models "over-anchor" on narrow, retrieved evidence while suppressing global physical reasoning. To address this problem, we introduce ARIA, a causal-aware framework that conditions knowledge use on mechanistic completeness. ARIA routes each query through a three-tier cascade: (i) direct causal reasoning when complete evidence chains of Process-Structure-Property (PSP) are available, (ii) physics-informed analogical transfer for sparse or novel material systems, and (iii) explicit parametric fallback when external evidence is incomplete. As a proof of concept, we construct a Knowledge Graph (KG) containing 2,839 extracted PSP relations from peer-reviewed articles in the materials literature and evaluate ARIA on forward prediction and inverse design tasks for two-dimensional (2D) materials. ARIA mitigates contextual tunneling, improves over unaugmented and naive KG-augmented baselines, and provides further gains when an online literature search is used for evidence enrichment. Crucially, ARIA produces auditable causal traces, enabling physically grounded and trustworthy AI-assisted materials discovery.
Chinese Translation
生成模型已经彻底改变了材料发现的过程，但它们往往无法满足潜在的物理因果关系。通过对增强了来自当前文献的知识图谱的大语言模型（LLMs）的分析，我们发现了一种现象，称为上下文隧道效应（contextual tunneling），在这种现象中，模型在狭窄的检索证据上“过度锚定”，同时抑制了全局物理推理。为了解决这个问题，我们引入了ARIA，一个因果感知框架，它将知识的使用条件化为机制的完整性。ARIA通过一个三层级联路由每个查询：（i）当完整的过程-结构-属性（Process-Structure-Property，PSP）证据链可用时，进行直接因果推理；（ii）对于稀疏或新颖的材料系统，进行物理启发的类比转移；（iii）当外部证据不完整时，进行明确的参数回退。作为概念验证，我们构建了一个知识图谱（Knowledge Graph，KG），其中包含从材料文献的同行评审文章中提取的2,839个PSP关系，并在二维（2D）材料的前向预测和逆向设计任务上评估ARIA。ARIA减轻了上下文隧道效应，优于未增强和天真的KG增强基线，并在使用在线文献搜索进行证据丰富时提供了进一步的收益。至关重要的是，ARIA生成可审计的因果轨迹，使得基于物理的可信AI辅助材料发现成为可能。
cs.AI / 98 / 2606.22385
MetaPS: Adaptive Programmatic Strategy Selection for Market Agents
MetaPS：市场代理的自适应程序策略选择
Abstract
No single market strategy always wins: momentum, mean reversion, risk control,and event-driven rules can each succeed or fail as market conditions change.Rather than asking large language models to directly generate market actions,we study an executable decision paradigm where an agent selects from a library of programmatic strategies, each implemented as a code module mapping market observations to actions.We propose \textbf{MetaPS}, a simulation-guided framework for adaptive programmatic strategy selection. MetaPS rolls out candidate strategies in simulated or backtested markets, identifies states where particular strategies lead to better future outcomes, and converts these state--strategy pairs into supervised fine-tuning data. During inference, the simulator is no longer queried: MetaPS observes only the current market state and candidate strategy context, selects a suitable strategy program, and the selected program produces the final action. Experiments on multi-stock trading and a controlled goods-exchange sandbox show that MetaPS consistently improves across model scales from 0.8B to 9B parameters. It outperforms fixed-strategy baselines, direct decision-making agents, and prompted API-based LLM agents; in several settings, compact fine-tuned models even surpass stronger API models. These results demonstrate that market simulations can provide scalable and targeted supervision for learning adaptive, interpretable, and executable strategy selection.
Chinese Translation
没有任何单一的市场策略能够始终获胜：动量、均值回归、风险控制和事件驱动规则在市场条件变化时都可能成功或失败。我们研究了一种可执行的决策范式，其中代理从一个程序策略库中选择，每个策略作为代码模块实现，将市场观察映射到行动。我们提出了 extbf{MetaPS}，一个基于模拟的自适应程序策略选择框架。MetaPS在模拟或回测市场中推出候选策略，识别特定策略能够带来更好未来结果的状态，并将这些状态-策略对转换为监督微调数据。在推理过程中，模拟器不再被查询：MetaPS仅观察当前市场状态和候选策略上下文，选择合适的策略程序，所选程序生成最终行动。在多股票交易和受控商品交换沙箱的实验中，MetaPS在从0.8B到9B参数的模型规模上持续改进。它优于固定策略基线、直接决策代理和基于提示的API LLM代理；在多个设置中，经过紧凑微调的模型甚至超过了更强的API模型。这些结果表明，市场模拟可以为学习自适应、可解释和可执行的策略选择提供可扩展和有针对性的监督。
cs.AI / 99 / 2606.22388
PlanBench-XL: Evaluating Long-Horizon Planning of LLM Tool-Use Agents in Large-Scale Tool Ecosystems
PlanBench-XL：在大规模工具生态系统中评估大型语言模型工具使用代理的长远规划能力
Abstract
LLM agents increasingly operate in large tool ecosystems, where real-world tasks require discovering relevant tools, inferring implicit sub-goals, and adapting to dynamic environments over long horizons. However, existing benchmarks rarely evaluate planning under retrieval-limited tool visibility. To address this gap, we introduce PlanBench-XL, an interactive benchmark of 327 retail tasks over 1,665 tools that tests whether agents can iteratively retrieve usable tools, invoke them to uncover intermediate evidence for subsequent calls toward the final goal. PlanBench-XL further features an optional blocking mechanism that simulates real-world unpredictability through missing, failing, or distracting tool functions, forcing agents to detect disrupted paths and adapt at runtime. Experiments on ten leading LLMs show that massive-tool planning remains challenging: while GPT-5.4 achieves 51.90% accuracy in block-free settings, it collapses to 11.36% under the most severe blocking condition. Further analysis shows that agents are especially vulnerable when failures lack explicit error signals or when recovery requires longer alternative tool-use paths. These results establish PlanBench-XL as a testbed for diagnosing agentic planning failures and highlight the need for robust adaptive planning in long-horizon tasks with large, imperfect tool environments.
Chinese Translation
大型语言模型（LLM）代理越来越多地在大型工具生态系统中运作，其中现实世界的任务需要发现相关工具、推断隐含的子目标，并在长时间跨度内适应动态环境。然而，现有的基准测试很少评估在检索受限的工具可见性下的规划能力。为了解决这一问题，我们引入了PlanBench-XL，这是一个包含327个零售任务和1,665个工具的互动基准，旨在测试代理是否能够迭代地检索可用工具，并调用它们以揭示后续调用最终目标所需的中间证据。PlanBench-XL还具有一个可选的阻塞机制，通过缺失、失败或干扰的工具功能模拟现实世界的不确定性，迫使代理检测中断路径并在运行时进行适应。对十种领先的LLM进行的实验表明，大规模工具规划仍然具有挑战性：尽管GPT-5.4在无阻塞设置下的准确率达到51.90%，但在最严重的阻塞条件下降至11.36%。进一步分析表明，当失败缺乏明确的错误信号或恢复需要更长的替代工具使用路径时，代理尤其脆弱。这些结果确立了PlanBench-XL作为诊断代理规划失败的测试平台，并强调了在具有大型、不完美工具环境的长远任务中需要强健的自适应规划能力。
cs.AI / 100 / 2606.22417
Code Isn't Memory: A Structural Codebase Index Inside a Coding Agent
代码不是内存：编码代理内部的结构化代码库索引
Abstract
Coding agents now interleave LLMs with retrieval over the working repository, and retrieval implementations vary widely across deployed harnesses. Inside a fixed coding-agent harness on a fixed model, does adding a structural codebase index actually change cost or resolve? We ran three arms (the harness with the index, the same harness without it, and an agentic-grep comparator) on SWE-PolyBench Verified and SWE-bench Pro with Claude Opus 4.7 held fixed throughout, across three seeds, inside a leak-audited per-task sandbox. The within-harness ablation produces a large localization gain and a statistically separated resolve gain, with no cost penalty per cell and lower cost per solve. The cross-harness check shows that the index does not regress against an agentic-grep baseline on resolve or localization, again at no cost penalty. We release the per-cell exclusion ledger, the leak-audit script, the localization extractor, and the results database. The deployment question for a structural codebase index is thus not whether it is too expensive to run (across seeds, the index lands at a lower $/solved than agentic grep) but whether the workload includes multi-file changes where structural ranking pays off.
Chinese Translation
编码代理现在将大型语言模型（LLMs）与对工作库的检索交织在一起，而检索实现因部署的工具而异。在一个固定模型的固定编码代理工具中，添加结构化代码库索引是否真的改变了成本或解决方案？我们在 SWE-PolyBench Verified 和 SWE-bench Pro 上进行了三组实验（包含索引的工具、相同工具但不包含索引的版本，以及一个代理 grep 比较器），在整个实验中保持 Claude Opus 4.7 固定，并在三个种子下，在经过泄漏审计的每个任务沙箱中进行。工具内的消融实验显示出显著的定位增益和统计上分离的解决增益，每个单元没有成本惩罚，且每个解决方案的成本更低。跨工具检查表明，索引在解决方案或定位方面并没有相对于代理 grep 基线出现回归，且同样没有成本惩罚。因此，结构化代码库索引的部署问题并不是运行成本是否过高（在不同种子下，索引的每解决方案成本低于代理 grep），而是工作负载是否包含多文件更改，在这种情况下结构化排名能够带来收益。
cs.AI / 101 / 2606.22425
SVGym (SciVerseGym): An Environment for Reinforcement Learning and Bayesian Optimization in Crystal Discovery
SVGym (SciVerseGym)：一个用于晶体发现中的强化学习和贝叶斯优化的环境
Abstract
Machine-learned interatomic potentials now enable efficient atomistic evaluation for interactive materials discovery, yet closed-loop crystal search methods remain fragmented across bespoke pipelines for editing, relaxation, scoring, constraints, and bookkeeping. We introduce SciVerseGym, a Gymnasium-compatible environment for sequential crystal discovery that frames crystal design as a Markov decision process. Agents observe an atomistic structure, apply chemically meaningful edits, and receive feedback from a configurable evaluator. SciVerseGym supports local and global actions, including elemental substitution, lattice perturbation, atomic displacement, vacancy creation, and atom insertion, along with configurable chemical spaces, structure pools, atomistic and graph-based observations, custom rewards, optional relaxation, and stability or phonon-related diagnostics. Each step applies an edit, evaluates the candidate using a machine-learned interatomic potential or any ASE-compatible calculator, and returns the standard (obs, reward, terminated, truncated, info) tuple. By decoupling agent logic from materials infrastructure, SciVerseGym provides an open, reproducible, and extensible testbed for reinforcement learning, Bayesian optimization, evolutionary search, and language-agent workflows in closed-loop crystal discovery. Code is available at: https://github.com/Bin-Cao/SciVerseGym.
Chinese Translation
机器学习的原子间势能现在能够高效地进行原子级评估，以支持交互式材料发现，然而闭环晶体搜索方法仍然在编辑、松弛、评分、约束和记录等定制流程中显得支离破碎。我们介绍了SciVerseGym，这是一个兼容Gymnasium的环境，用于顺序晶体发现，将晶体设计框架视为马尔可夫决策过程。智能体观察原子结构，进行具有化学意义的编辑，并从可配置的评估器那里获得反馈。SciVerseGym支持局部和全局操作，包括元素替换、晶格扰动、原子位移、空位创建和原子插入，以及可配置的化学空间、结构池、原子级和基于图的观察、自定义奖励、可选松弛以及稳定性或声子相关的诊断。每一步都应用一次编辑，使用机器学习的原子间势能或任何兼容ASE的计算器评估候选，并返回标准的（obs, reward, terminated, truncated, info）元组。通过将智能体逻辑与材料基础设施解耦，SciVerseGym提供了一个开放、可重复和可扩展的测试平台，用于强化学习、贝叶斯优化、进化搜索和闭环晶体发现中的语言-智能体工作流程。代码可在以下网址获取：https://github.com/Bin-Cao/SciVerseGym。
cs.AI / 102 / 2606.22442
Efficient Multimodal Clinical Question Answering for Pulmonary Embolism Risk Assessment
高效的多模态临床问题回答用于肺栓塞风险评估
Abstract
Pulmonary embolism (PE) is a high risk cardiopulmonary condition whose management requires both timely diagnosis and reliable assessment of future clinical risk. Because PE care routinely combines computed tomography pulmonary angiography (CTPA), radiology interpretation, and longitudinal electronic health record (EHR) evidence, it provides a clinically meaningful setting for evaluating compact multimodal language models. In this work, we build a benchmark using efficient multimodal large language models (MLLMs) on INSPECT, a multimodal PE dataset containing 23,248 CTPA studies from 19,402 patients. We formulate eight diagnostic and prognostic tasks as structured clinical question answering problems and evaluate on typical efficient MLLMs under CTPA-Only, EHR-Only, and CTPA+EHR settings with zero-shot and few-shot prompting. Results show that Gemma4 E4B and Gemma4 E2B perform more strongly when EHR evidence is available, especially under CTPA+EHR input. Task level analysis further shows that PE diagnosis achieves higher performance than prognostic tasks, particularly readmission prediction. These observations suggest that compact multimodal models have the great potential in early stage PE risk detection and explanation.
Chinese Translation
肺栓塞（PE）是一种高风险的心肺疾病，其管理需要及时的诊断和对未来临床风险的可靠评估。由于PE的护理通常结合了计算机断层扫描肺动脉造影（CTPA）、放射学解读和纵向电子健康记录（EHR）证据，因此为评估紧凑型多模态语言模型提供了一个具有临床意义的环境。在本研究中，我们利用高效的多模态大型语言模型（MLLMs）构建了一个基准，使用INSPECT数据集，该数据集包含来自19,402名患者的23,248个CTPA研究。我们将八个诊断和预后任务构建为结构化临床问题回答问题，并在CTPA-Only、EHR-Only和CTPA+EHR设置下，使用零样本和少样本提示对典型的高效MLLMs进行评估。结果表明，当EHR证据可用时，Gemma4 E4B和Gemma4 E2B的表现更为强劲，尤其是在CTPA+EHR输入下。任务级分析进一步显示，PE诊断的表现优于预后任务，特别是再入院预测。这些观察结果表明，紧凑型多模态模型在早期PE风险检测和解释方面具有巨大的潜力。
cs.AI / 103 / 2606.22447
A Differentiable Atari VCS:A Complex, Fully Known Ground Truth for Explainable AI
可微分的Atari VCS：一种复杂的、完全已知的真实基础用于可解释人工智能
Abstract
Explanation requires ground truth: to verify an account of a system we must know its inner functioning-just what is missing where explainable AI (XAI) is most needed. Systems we can study fall into two camps. Simple, procedural one-decision trees, rule lists, sparse linear models-have a known but trivial mechanism, so explaining them tests nothing; genuinely complex ones-deep networks, real-world tasks-need XAI but have no ground-truth inner functioning, so an explanation can be plausible, confident, and wrong with no way to tell. We remove this dichotomy with a study object both genuinely complex and fully specified-inspectable by construction-and, so gradient methods apply, fully differentiable. We reimplement the Atari 2600 Video Computer System (VCS)-a real computer architecture, and the cradle of deep reinforcement learning-as two independent end-to-end differentiable emulators in Julia (jutari) and JAX (jaxtari), each validated bit-for-bit against xitari. Both reproduce xitari on all 64 supported Arcade Learning Environment (ALE) games: 64/64 byte-identical RAM and 64/64 pixel-identical screens. Treating the cartridge ROM as a weight tensor, RAM as a soft tape, and control flow as gates, we prove the differentiable (soft) execution equals the original (hard) one bit-for-bit in the forward pass at any finite temperature, while exposing surrogate gradients where the bit logic has none. The JAX port also opens a GPU path: batched differentiable rollouts reach millions of environment-steps/s on one commodity GPU. The system was built in roughly 137 active hours over 29 calendar days, much of it written autonomously by coding agents. This paper builds and validates the foundation, showing-theoretically and in a qualitative gradient study-that gradient-based XAI on it is feasible. Both ports' full code is available under the MIT license at https://github.com/akmaier/UnderstandingVCS.
Chinese Translation
解释需要真实基础：为了验证系统的描述，我们必须了解其内部运作——这正是可解释人工智能（XAI）最需要的地方。我们可以研究的系统分为两类。简单的、程序化的系统——决策树、规则列表、稀疏线性模型——具有已知但简单的机制，因此解释它们并不能测试任何东西；真正复杂的系统——深度网络、现实世界任务——需要XAI，但没有真实基础的内部运作，因此解释可能是合理的、自信的，但却是错误的，且无法判断。我们通过一个既真正复杂又完全指定的研究对象消除了这种二分法——通过构造可检查，因此适用梯度方法，完全可微分。我们重新实现了Atari 2600视频计算机系统（VCS）——一种真实的计算机架构，也是深度强化学习的摇篮——作为两个独立的端到端可微分模拟器，分别使用Julia（jutari）和JAX（jaxtari），每个模拟器都与xitari逐位验证。两个模拟器在所有64个支持的街机学习环境（ALE）游戏中均能重现xitari：64/64字节完全相同的RAM和64/64像素完全相同的屏幕。将卡带ROM视为权重张量，将RAM视为软带，将控制流视为门，我们证明可微分（软）执行在任何有限温度下在前向传播中逐位等于原始（硬）执行，同时暴露出位逻辑没有的替代梯度。JAX端口还开辟了GPU路径：批量可微分的回放在一台普通GPU上达到每秒数百万个环境步骤。该系统在29个日历日内大约137个活跃小时内构建，其中大部分由编码代理自主编写。本文构建并验证了基础，理论上和在定性梯度研究中表明基于梯度的XAI在其上是可行的。两个端口的完整代码在MIT许可证下可在https://github.com/akmaier/UnderstandingVCS获取。
cs.AI / 104 / 2606.22449
Self-Evolving Cognitive Framework via Causal World Modeling for Embodied Scientific Intelligence
通过因果世界建模实现自我演化的认知框架以促进具身科学智能
Abstract
Current embodied world models are primarily optimized for predictive objectives, limiting their ability to generalize under distribution shifts and reason systematically about unseen situations and hypothetical interventions. We argue that embodied intelligence should move beyond predictive world modeling toward self-evolving cognitive systems that continually construct and refine internal causal representations through interaction with the environment. To this end, we propose a self-evolving cognitive framework via causal world modeling for embodied scientific intelligence, which integrates three complementary components: causal world modeling, intervention-driven causal reasoning, and continual cognitive refinement. The proposed framework continuously revises and expands its internal causal world model through causal discovery, intervention-driven feedback, and counterfactual reasoning, supporting continual cognitive refinement and enabling cognition itself to evolve over time. Furthermore, we reinterpret embodied interaction not merely as a means of trajectory optimization, but as an epistemic process for causal hypothesis generation, intervention-driven experimentation, and continual knowledge acquisition. This work provides a conceptual and theoretical foundation for a transition from predictive intelligence toward epistemic intelligence, in which intelligence emerges through the continual construction, revision, and refinement of causal world models via interaction with the environment. Accordingly, an intervention-driven causal-epistemic benchmarking paradigm is suggested for evaluating self-evolving embodied scientific intelligence.
Chinese Translation
当前的具身世界模型主要针对预测目标进行优化，这限制了它们在分布变化下的泛化能力，以及对未见情境和假设干预进行系统推理的能力。我们认为，具身智能应超越预测世界建模，向自我演化的认知系统发展，这些系统通过与环境的互动不断构建和完善内部因果表征。为此，我们提出了一种通过因果世界建模实现自我演化的认知框架，以促进具身科学智能，该框架整合了三个互补的组成部分：因果世界建模、基于干预的因果推理和持续的认知完善。所提出的框架通过因果发现、基于干预的反馈和反事实推理不断修订和扩展其内部因果世界模型，支持持续的认知完善，使认知本身能够随时间演化。此外，我们重新解释具身互动，不仅仅作为轨迹优化的手段，而是作为因果假设生成、基于干预的实验和持续知识获取的认识过程。本研究为从预测智能向认识智能的转变提供了概念和理论基础，其中智能通过与环境的互动不断构建、修订和完善因果世界模型而产生。因此，建议采用基于干预的因果-认识基准范式来评估自我演化的具身科学智能。
cs.AI / 105 / 2606.22470
PRIME: Evaluating Prompt Resolution Under Incompatible Instructions in LLMs
PRIME：在不兼容指令下评估提示解析的框架
Abstract
Large language models (LLMs) often encounter conflicting prompts, although current instruction following benchmarks assess those meta-instructions in isolation, limiting the insights about how models process conflicting instructions. We introduce a framework \textit{PRIME}(\textit{Prompt Resolution under Incompatible Meta-Instructions Evaluation}) to analyze behavior of LLMs when provided with conflicting instructions. \textit{PRIME} purposefully produces calibrated conflicts across response length, output format, and reasoning; classifying model responses with a deterministic behavioral taxonomy. We are evaluating five instruction tuned open weight LLMs in two distinct settings, balanced and naturally distributed. The conclusion we reach upon analysis is that conflict type is more significant in affecting behavior than model scale, and various failure modes across different categories of conflict. Our findings emphasize the value of developing conflict awareness and suggest ability of LLM to follow instructions cannot be assessed through isolated constraints alone.
Chinese Translation
大型语言模型（LLMs）经常遇到相互矛盾的提示，尽管当前的指令遵循基准测试在孤立条件下评估这些元指令，这限制了我们对模型如何处理冲突指令的理解。我们提出了一个框架 extit{PRIME}（ extit{不兼容元指令下的提示解析评估}），用于分析LLMs在提供冲突指令时的行为。 extit{PRIME} 有意产生在响应长度、输出格式和推理方面的校准冲突；并使用确定性的行为分类法对模型响应进行分类。我们在两种不同的设置下评估了五个经过指令调优的开放权重LLMs，分别是平衡和自然分布的。通过分析，我们得出的结论是，冲突类型在影响行为方面比模型规模更为重要，并且在不同冲突类别中存在各种失败模式。我们的研究结果强调了发展冲突意识的重要性，并建议不能仅通过孤立的约束来评估LLM遵循指令的能力。
cs.AI / 106 / 2606.22485
VADAOrchestra: Neurosymbolic Orchestration of Adaptive Reasoning Workflows
VADAOrchestra：自适应推理工作流的神经符号编排
Abstract
Decision-making in real-world settings rarely follows a fixed script. Instead, it unfolds as a dynamic reasoning process in which the appropriate course of action evolves as new context and data become available. Traditional Business Process Management systems provide rigor, determinism, and auditability, yet they generally struggle to adapt their execution at runtime. Conversely, agentic systems based on Large Language Models (LLMs) bring flexibility to decision-making, but they are inherently opaque, often unreliable, and suffer from significant scalability constraints when operating over large datasets. To combine these complementary paradigms, we introduce VADAOrchestra, a neurosymbolic framework that models complex workflows as evolving reasoning processes. The framework adopts a hybrid approach: given a user query and a collection of data sources, an LLM-based orchestrator incrementally plans and adapts the workflow. This is encoded as a logic program in a fragment of Datalog+/- where predicates correspond to tool invocations and rules represent both predefined domain dependencies and logic constructs synthesized on demand to manipulate intermediate results. All logical inference tasks are then executed by a state-of-the-art Datalog+/- symbolic engine. This approach provides a verifiable reasoning trace, supporting the auditability and reproducibility of the entire process. Furthermore, by decoupling high-level orchestration from symbolic inference, it addresses scalability concerns, enabling complex reasoning over large datasets through targeted data querying. We evaluate VADAOrchestra on real-world financial use cases, demonstrating faithfulness, scalability, and explainability compared to standard agentic architectures.
Chinese Translation
在现实世界中，决策过程很少遵循固定的脚本。相反，它作为一个动态推理过程展开，在这个过程中，适当的行动方案随着新上下文和数据的出现而不断演变。传统的业务流程管理系统提供了严格性、确定性和可审计性，但通常在运行时难以适应其执行。相对而言，基于大型语言模型（Large Language Models, LLMs）的代理系统为决策带来了灵活性，但它们本质上是不透明的，往往不可靠，并且在处理大数据集时面临显著的可扩展性限制。为了结合这些互补的范式，我们提出了VADAOrchestra，这是一个将复杂工作流建模为演变推理过程的神经符号框架。该框架采用混合方法：给定用户查询和一组数据源，基于LLM的编排器逐步规划和调整工作流。这被编码为Datalog+/-的一部分中的逻辑程序，其中谓词对应于工具调用，规则表示预定义的领域依赖关系和按需合成的逻辑结构，以操纵中间结果。所有逻辑推理任务随后由最先进的Datalog+/-符号引擎执行。这种方法提供了可验证的推理轨迹，支持整个过程的可审计性和可重复性。此外，通过将高层编排与符号推理解耦，它解决了可扩展性问题，使得通过针对性的数据查询能够在大数据集上进行复杂推理。我们在现实世界的金融用例上评估了VADAOrchestra，展示了与标准代理架构相比的忠实性、可扩展性和可解释性。
cs.AI / 107 / 2606.22488
SCOPE: Evolving Symbolic World for Planning in Open-Ended Environments
SCOPE：在开放式环境中规划的演变符号世界
Abstract
Recent works have explored integrating Vision-Language Models (VLMs) with classical planners that rely on symbolic representations of planning problems to generate long-horizon plans for complex embodied tasks. However, in open-ended environments, these symbolic representations obtained from perception are often incomplete, leading to suboptimal performance. To address this, we introduce SCOPE, a self-adaptive symbolic planning framework that supports refining action plans and evolving the symbolic world, i.e., the symbolic representations of open-ended environments. SCOPE comprises two synergistic modules: a Symbolic Execution Simulator (SESim) that conducts symbolic validation and real execution of action plans, leveraging the feedback to refine the plans and evolve the symbolic world; and a Self-Adaptive Symbolic Memory (SASMem) that further distills feedback into evolving symbolic knowledge to enhance long-horizon planning and modeling of the symbolic world. Experiments in open-ended environments show that SCOPE significantly improves the completeness of the symbolic world, the success rate of plans under environment perturbations, and cross-task grounding and adaptability across diverse embodied scenarios.
Chinese Translation
近期的研究探讨了将视觉-语言模型（Vision-Language Models, VLMs）与依赖于规划问题符号表示的经典规划器相结合，以生成复杂具身任务的长远计划。然而，在开放式环境中，从感知中获得的这些符号表示往往是不完整的，导致次优的性能。为了解决这个问题，我们提出了SCOPE，一个自适应符号规划框架，支持精炼行动计划和演变符号世界，即开放式环境的符号表示。SCOPE由两个协同模块组成：符号执行模拟器（Symbolic Execution Simulator, SESim），该模块进行行动计划的符号验证和实际执行，利用反馈来精炼计划并演变符号世界；自适应符号记忆（Self-Adaptive Symbolic Memory, SASMem），该模块进一步提炼反馈为演变的符号知识，以增强长远规划和符号世界的建模。在开放式环境中的实验表明，SCOPE显著提高了符号世界的完整性、在环境扰动下计划的成功率，以及在多样化具身场景中的跨任务基础和适应性。
cs.AI / 108 / 2606.22494
Deep Learning-Based Sign Language Recognition from Videos and Cross-Lingual Translation to Indian Vernaculars
基于深度学习的视频手语识别及其向印度方言的跨语言翻译
Abstract
Sign language is a primary mode of communication for the global deaf and hard-of-hearing community, yet automated tools that recognize sign gestures from video and translate them into natural language text remain limited, particularly for low-resource Indian languages. We present a two-stage deep learning pipeline that (i) classifies short sign language video clips into English word labels using a fine-tuned VideoMAE video transformer, and (ii) translates the predicted English label into Hindi, Telugu, and Bengali using Meta AI's No Language Left Behind (NLLB-200) multilingual translation model. The classification model is fine-tuned on a 13-class subset of the AI4Bharat Indian Sign Language video corpus from IIT Madras, processing 16-frame clips sampled uniformly from each video at 224 x 224 resolution. Under a small-scale academic setting (13 classes, 197 clips, 80-20 split), the fine-tuned model reaches 99% training accuracy and 78% validation accuracy after 15 epochs. We provide a per-class breakdown via a confusion matrix and classification report, identify the dominant failure modes (confusable adjective pairs such as ugly, deaf, blind, hat, and dress), and describe a Streamlit-based inference demo that takes a user-uploaded video and returns the predicted English label alongside its Hindi, Telugu, and Bengali translations. We discuss the scope, limitations (small label set, isolated-word rather than continuous signing, single-signer style sensitivity, ambiguity of single-word machine translation), and directions for future work, including expanding to sentence-level generation and a larger vocabulary. Code is released to support reproducibility.
Chinese Translation
手语是全球聋人和听力受损人群的主要交流方式，但能够自动识别视频中的手势并将其翻译为自然语言文本的工具仍然有限，尤其是在资源匮乏的印度语言中。我们提出了一种两阶段的深度学习流程：(i) 使用经过微调的VideoMAE视频变换器将短手语视频剪辑分类为英语单词标签，(ii) 使用Meta AI的No Language Left Behind (NLLB-200)多语言翻译模型将预测的英语标签翻译为印地语、泰卢固语和孟加拉语。分类模型在来自印度理工学院马德拉斯的AI4Bharat印度手语视频语料库的13类子集上进行了微调，处理从每个视频均匀采样的224 x 224分辨率的16帧剪辑。在小规模学术环境下（13类，197个剪辑，80-20拆分），经过微调的模型在15个训练周期后达到了99%的训练准确率和78%的验证准确率。我们通过混淆矩阵和分类报告提供每类的详细分析，识别出主要的失败模式（如丑陋、聋、盲、帽子和裙子等易混淆的形容词对），并描述了一个基于Streamlit的推理演示，该演示接受用户上传的视频并返回预测的英语标签及其印地语、泰卢固语和孟加拉语翻译。我们讨论了研究的范围、局限性（标签集小、孤立词而非连续手语、单一签名者风格敏感性、单词机器翻译的模糊性）以及未来工作的方向，包括扩展到句子级生成和更大的词汇量。代码已发布以支持可重复性。
cs.AI / 109 / 2606.22495
Grounded Scaling: Why Agentic AI Needs Deterministic Environments
基于环境确定性的扩展：为何自主智能体需要确定性环境
Abstract
Long-chain agent execution fails exponentially in environments designed for human tolerance: with per-step determinism $\delta < 1$, $k$-step chain success degrades as $\delta^k$. The AGI-to-ASI scaling debate (Genewein et al., 2026) has so far framed progress as a race between compute growth and a list of frictions (data wall, abstraction barrier, embodied bottleneck, multi-agent trust); we argue that environment determinism is a complementary binding axis cutting across all four, for the broad class of agentic AI tasks whose outcomes are verifiable economically, physically, or through multi-party settlement. Three formal results pin down the regime: a Determinism-Efficiency Bound on chain-task success, a Verifier-Goodharting Floor on flywheel ceilings under imperfect rewards, and a convergence condition for environment-side skill evolution. We operationalise the framework as a Supply Certainty Index (SCI) over five measurable properties, a five-level Determinism Maturity Model (DMM) as adoption ladder, and a falsifiable open-question programme (OQ1-OQ5) with explicit null results that would force retraction. The position is platform-agnostic. We engage three competing positions: sim-to-real sufficiency, alignment sufficiency, and AI-as-normal-technology.
Chinese Translation
在为人类容忍度设计的环境中，长链智能体执行的失败率呈指数级增长：在每步确定性 $ ext{δ} < 1$ 的情况下，$k$ 步链的成功率随着 $ ext{δ}^k$ 的降低而下降。AGI（通用人工智能）到 ASI（超人工智能）的扩展辩论（Genewein 等，2026）迄今为止将进展框架设定为计算增长与一系列摩擦（数据壁垒、抽象障碍、具身瓶颈、多智能体信任）之间的竞赛；我们认为环境的确定性是一个补充的约束轴，横贯所有四个摩擦，适用于广泛的自主智能体任务，其结果可以通过经济、物理或多方结算进行验证。三个正式结果明确了这一范畴：链任务成功的确定性-效率界限、在不完美奖励下飞轮上限的验证者-古德哈特底线，以及环境侧技能演化的收敛条件。我们将该框架操作化为五个可测量属性的供应确定性指数（SCI）、一个五级确定性成熟度模型（DMM）作为采纳阶梯，以及一个可证伪的开放问题程序（OQ1-OQ5），其中包含明确的无效结果，这将迫使撤回该研究。该立场不依赖于特定平台。我们探讨了三种竞争立场：从仿真到现实的充分性、对齐的充分性，以及将人工智能视为常规技术。
cs.AI / 110 / 2606.22509
Imagine to Ensure Safety in Hierarchical Reinforcement Learning
想象以确保层次强化学习中的安全性
Abstract
This work investigates the safe exploration problem in reinforcement learning, where an agent must maximize cumulative performance while simultaneously satisfying safety constraints. This challenge becomes even more pronounced in long-horizon tasks, where existing safe methods face fundamental limitations due to compounding estimation errors and restricted exploration capabilities. To address this problem, we propose a method that combines a learnable world model with two complementary policies a high-level policy and a low-level policy to promote safety at both hierarchical levels. The high-level policy generates intermediate subgoals that bias exploration toward safe regions, while the low-level policy uses imagined rollouts in the learned world model to reduce unsafe behaviors when reaching these subgoals. The proposed method was evaluated on challenging long-horizon navigation and manipulation tasks with high-dimensional action spaces, where it significantly outperforms existing Safe RL baselines in both success rate and strong empirical constraint satisfaction, consistently meeting the prescribed safety budget across seeds, while prior approaches fail to effectively solve these complex long-horizon scenarios.
Chinese Translation
本研究探讨了强化学习中的安全探索问题，在该问题中，代理必须在最大化累积性能的同时满足安全约束。这一挑战在长时间跨度的任务中变得更加明显，因为现有的安全方法由于估计误差的累积和受限的探索能力而面临根本性限制。为了解决这个问题，我们提出了一种方法，该方法结合了可学习的世界模型和两种互补策略：高层策略和低层策略，以促进两个层次的安全性。高层策略生成中间子目标，偏向于安全区域的探索，而低层策略则利用在学习到的世界模型中的想象回滚，以减少在达到这些子目标时的不安全行为。所提出的方法在具有高维动作空间的具有挑战性的长时间跨度导航和操控任务中进行了评估，在成功率和强有力的经验约束满足方面显著优于现有的安全强化学习基准，始终在各个种子上满足规定的安全预算，而先前的方法未能有效解决这些复杂的长时间跨度场景。
cs.AI / 111 / 2606.22528
Governance Decay: How Context Compaction Silently Erases Safety Constraints in Long-Horizon LLM Agents
治理衰退：上下文压缩如何悄然消除长时间运行的LLM代理中的安全约束
Abstract
Modern LLM agents increasingly rely on context compaction, summarization, or eviction to keep long-running sessions within a token budget. We show that this context-management layer is a safety-critical failure surface: in-context governance constraints that agents reliably obey while visible can be silently removed by compaction, causing the same agent to perform prohibited tool actions later in the session. We call this failure mode Governance Decay. We introduce ConstraintRot, a benchmark of long-horizon agent scenarios with deterministic tool-call grading, and measure compaction-induced violations across seven model families. Across 1,323 episodes, violation rises from 0% with the policy in full context to 30% after compaction, reaching 59% for some models; when the constraint survives the summary, violation remains 0%, but when it is dropped, violation reaches 38%. We further study a Compaction-Eviction Attack, in which adversarial in-context content biases the summarizer to omit a legitimate policy, and show that optimized injections defeat every evaluated model. Finally, we propose Constraint Pinning, a simple training-free mitigation that quarantines governance constraints from lossy compaction and restores violation to 0% in our benchmark. These results identify context management as a first-class governance surface for deployed LLM agents.
Chinese Translation
现代LLM代理越来越依赖上下文压缩、摘要或驱逐，以保持长时间运行的会话在令牌预算内。我们展示了这一上下文管理层是一个安全关键的失败面：在可见时，代理可靠遵守的上下文治理约束可以通过压缩悄然移除，导致同一代理在会话后期执行被禁止的工具操作。我们将这种失败模式称为治理衰退。我们引入了ConstraintRot，这是一个具有确定性工具调用评分的长时间代理场景基准，并测量了七个模型系列中因压缩引起的违规情况。在1,323个回合中，违规率从完整上下文下的0%上升至压缩后的30%，某些模型的违规率甚至达到59%；当约束在摘要中存活时，违规率保持为0%，但当其被移除时，违规率达到38%。我们进一步研究了一种压缩-驱逐攻击，其中对抗性上下文内容使摘要生成器偏向于省略合法政策，并展示了优化的注入能够击败每个评估模型。最后，我们提出了Constraint Pinning，这是一种简单的无训练缓解方法，将治理约束隔离于有损压缩之外，并在我们的基准测试中将违规率恢复至0%。这些结果将上下文管理确立为部署的LLM代理的首要治理表面。
cs.AI / 112 / 2606.22557
MacAgentBench: Benchmarking AI Agents on Real-World macOS Desktop
MacAgentBench：在真实世界macOS桌面上对AI代理进行基准测试
Abstract
Computer use agents (CUAs) have advanced rapidly in desktop automation, and a growing number of users deploy CUAs such as OpenClaw on Mac Mini for always-on automation. However, existing benchmarks, including those for macOS, evaluate agents without framework augmentation and rely on binary evaluation. As a result, they fail to capture both the framework capabilities leveraged by modern CUAs and the partial progress on long-horizon, multi-application tasks. We present MacAgentBench, a comprehensive macOS agent benchmark comprising 676 tasks across 25 applications, with nearly 60% involving both GUI and CLI interaction. The benchmark adopts deterministic rule-based evaluation and introduces fine-grained multi-checkpoint scoring with capability annotations for multi-application tasks. Experiments across three frameworks and 16 models show that the best configuration, Claude Opus 4.6 on OpenClaw, attains 73.7% Pass@1, while this advantage is primarily driven by the skill library rather than by framework design. Fine-grained metrics further reveal that models with similar Pass@1 can differ substantially in sub-goal completion. Our code and data are publicly available at https://github.com/JetAstra/MacAgentBench.
Chinese Translation
计算机使用代理（CUAs）在桌面自动化方面迅速发展，越来越多的用户在Mac Mini上部署如OpenClaw等CUA进行持续自动化。然而，现有的基准测试，包括针对macOS的测试，评估代理时未进行框架增强，并依赖于二元评估。因此，它们未能捕捉现代CUA所利用的框架能力以及在长时间跨度、多应用任务上的部分进展。我们提出了MacAgentBench，这是一个全面的macOS代理基准测试，包含25个应用程序中的676个任务，其中近60%涉及GUI和CLI交互。该基准采用确定性规则基础的评估，并引入了细粒度的多检查点评分，附带多应用任务的能力注释。在三个框架和16个模型上的实验表明，最佳配置为在OpenClaw上运行的Claude Opus 4.6，达到了73.7%的Pass@1，而这一优势主要源于技能库，而非框架设计。细粒度指标进一步揭示，具有相似Pass@1的模型在子目标完成度上可能存在显著差异。我们的代码和数据可在https://github.com/JetAstra/MacAgentBench公开获取。
cs.AI / 113 / 2606.22586
Text2DSL: LLM-Based Code Generation for Domain-Specific Languages
Text2DSL：基于大型语言模型的领域特定语言代码生成
Abstract
Domain-specific languages (DSLs) are widely used for managing operating system security policies, yet manually authoring rules in such languages demands high expertise and is error-prone. This paper formalises the task of automatic DSL code generation from natural language descriptions - Text2DSL - as a distinct problem class, separate from Text-to-SQL and general-purpose code generation. We introduce the PolkitBench dataset comprising 4,204 verified natural-language-to-Polkit-rule pairs, each validated through a three-level AST-based pipeline. Controlled prompt experiments on two MoE models of different scale and provenance - GigaChat-10B-A1.8B (1.8B active parameters) and Nemotron-3-Nano-30B-A3B (3B active) - demonstrate the critical role of structured context (BNF grammar, API specification, permitted identifier vocabulary) for LLM-based DSL code generation. Across both models, supplying context raises syntactic validity to 98.6-99.4%, structural validity by +9.7 to +35.5 pp, and the CodeBLEU score by +60% to +95%. The consistency of the effect across models of different scale and provenance indicates that, for the Text2DSL class of problems, injecting a formal target-language specification into the prompt context is a robust enabling factor for high-quality generation without model fine-tuning.
Chinese Translation
领域特定语言（DSL）广泛用于管理操作系统安全策略，但在此类语言中手动编写规则需要高水平的专业知识，并且容易出错。本文将从自然语言描述自动生成DSL代码的任务——Text2DSL——正式化为一个独立的问题类别，与文本到SQL和通用代码生成区分开来。我们引入了PolkitBench数据集，其中包含4204对经过验证的自然语言到Polkit规则的配对，每对都通过三层AST（抽象语法树）基础管道进行了验证。在两个不同规模和来源的MoE模型——GigaChat-10B-A1.8B（1.8B活跃参数）和Nemotron-3-Nano-30B-A3B（3B活跃参数）上进行的受控提示实验表明，结构化上下文（BNF语法、API规范、允许的标识符词汇）在基于大型语言模型的DSL代码生成中起着关键作用。在这两个模型中，提供上下文将语法有效性提高至98.6-99.4%，结构有效性提高了9.7至35.5个百分点，CodeBLEU得分提高了60%至95%。不同规模和来源的模型间效果的一致性表明，对于Text2DSL问题类别，将正式的目标语言规范注入提示上下文是实现高质量生成的一个稳健的促进因素，而无需对模型进行微调。
cs.AI / 114 / 2606.22610
PaperClaw: Harnessing Agents for Autonomous Research and Human-in-the-Loop Refinement
PaperClaw：利用智能体实现自主研究与人机协作的精炼
Abstract
Large language models have become capable reasoners and tool users that write and run code and search the literature, which makes automating the research process itself a realistic goal. We present PAPERCLAW, a harnessed multi-agent system that carries a project autonomously, from a field of study to a finished paper. PAPERCLAW curates a domain from a field's live literature, datasets, and code; brainstorms it into an idea with a pre-registered main-result contract; and drives a stoppable hypothesis map through an iterative propose, test, reflect loop that grows only from measured verdicts and halts once the evidence supports the idea, at which point it writes a venue-compliant paper. A full-lifecycle memory keeps each stage in a single living record, so a long run can be paused, inspected, and resumed without losing context. At the centre is an in-cycle research assistant with research tools and skills: it can drive the whole pipeline on its own, while the same interface lets a person step in at any stage, turning a first autonomous draft into a stronger paper through human-in-the-loop refinement. Throughout, PAPERCLAW keeps its output grounded and checkable, citing only references validated against open scholarly indexes and reporting results that genuinely ran. An evaluation with an LLM judge finds that PAPERCLAW produces strong papers both fully autonomously and with human-in-the-loop refinement.
Chinese Translation
大型语言模型已成为能够推理和使用工具的系统，能够编写和运行代码以及搜索文献，这使得自动化研究过程成为一个现实目标。我们提出了PAPERCLAW，一个可自主进行项目的多智能体系统，从研究领域到完成论文。PAPERCLAW从领域的实时文献、数据集和代码中策划一个领域；将其构思为一个带有预注册主要结果合同的想法；并通过一个可中止的假设图驱动一个迭代的提出、测试、反思循环，该循环仅从测量的判决中增长，并在证据支持该想法时停止，此时它撰写一篇符合会议要求的论文。一个完整生命周期的记忆保持每个阶段在一个单一的动态记录中，因此长时间运行可以暂停、检查并恢复，而不会失去上下文。中心是一个在循环中的研究助手，具备研究工具和技能：它可以独立驱动整个流程，而相同的界面允许人在任何阶段介入，通过人机协作的精炼将初步的自主草稿转变为更强的论文。在整个过程中，PAPERCLAW保持其输出的基础和可验证性，仅引用经过开放学术索引验证的参考文献，并报告真正运行的结果。与LLM评审的评估发现，PAPERCLAW在完全自主和人机协作精炼的情况下都能产生高质量的论文。
cs.AI / 115 / 2606.22613
SkillAudit: From Fixed-Suite Benchmarking to Skill-Centered Assessment
SkillAudit：从固定套件基准测试到以技能为中心的评估
Abstract
Agent skills have become a practical way to extend large language model agents, but the growing skill ecosystem still lacks a reliable way to judge whether a skill is worth deploying. Existing evaluation methods remain largely anchored to fixed task suites, assessing skills through performance on predefined tasks and environments. As skill marketplaces expand, this paradigm becomes inadequate: fixed suites can conflate a skill's marginal contribution with backbone strength and miss its value when tasks fall outside the skill's intended scope. We introduce SkillAudit, an end-to-end framework for skill-centered assessment that takes an arbitrary agent skill as input and automatically generates a comprehensive, multi-dimensional evaluation report spanning utility, efficiency/cost, and safety. SkillAudit focuses on the skill artifact itself and constructs capability-aligned evaluation tasks directly from the skill package. The generated tasks are conducted in isolated sandbox environments to collect execution evidence, followed by automated checks with LLM-based judging to produce auditable results. To dissect the agent skills, we propose the baseline comparison principle to measure utility and efficiency/cost, and introduce a two-stage detection paradigm combining static semantic analysis with dynamic runtime verification to assess safety risks. After scanning top-ranked real-world skill packages spanning 23 occupational categories, we found that over 7% of skills are at risky status.
Chinese Translation
代理技能已成为扩展大型语言模型代理的一种实用方式，但日益增长的技能生态系统仍然缺乏可靠的方法来判断某项技能是否值得部署。现有的评估方法在很大程度上仍然依赖于固定任务套件，通过在预定义任务和环境中的表现来评估技能。随着技能市场的扩展，这一范式变得不够充分：固定套件可能将技能的边际贡献与基础能力混淆，并在任务超出技能预期范围时忽视其价值。我们提出了SkillAudit，一个以技能为中心的评估的端到端框架，该框架以任意代理技能为输入，自动生成涵盖效用、效率/成本和安全性的全面多维评估报告。SkillAudit专注于技能工件本身，并直接从技能包构建能力对齐的评估任务。生成的任务在隔离的沙箱环境中进行，以收集执行证据，随后通过基于大型语言模型（LLM）的自动检查生成可审计的结果。为了剖析代理技能，我们提出了基线比较原则，以测量效用和效率/成本，并引入一种结合静态语义分析与动态运行时验证的两阶段检测范式来评估安全风险。在扫描涵盖23个职业类别的顶级真实技能包后，我们发现超过7%的技能处于风险状态。
cs.AI / 116 / 2606.22633
Confident but Conflicted: Internal Uncertainty and Cognitive Dissonance Resolution in LLMs
自信但矛盾：大型语言模型中的内部不确定性与认知失调的解决
Abstract
Large language models (LLMs) frequently encounter inputs that disagree with their prior outputs, through user pushback, retrieved documents, or web search results. While the way they resolve such conflicts -- a process we frame as cognitive dissonance resolution -- has been characterized behaviorally, its connection to internal model uncertainty is not well understood. To study this systematically, we vary persuasion attempts along two dimensions, source authority and evidence quality, across 12 health-science claims of stratified epistemic status. Dissonance can be resolved through persuasion, backfire, or immunity. We introduce Trust Elasticity (TE), an econometrics-inspired measure of how readily a model is persuaded toward conflicting evidence. Across four LLMs, TE varies substantially, while clearly false claims elicit near-zero TE across all models. On two open-weight models, we further find that this variation is associated with two complementary internal uncertainty indicators, Confidence Miscalibration in Qwen and Internal Uncertainty Change in Llama. These results link cross-model behavioral variation to a measurable internal property and point to interventions targeting internal uncertainty as future work.
Chinese Translation
大型语言模型（LLMs）经常遇到与其先前输出不一致的输入，这些输入可能来自用户的反对、检索的文档或网络搜索结果。尽管我们将其解决此类冲突的方式——一个我们称之为认知失调解决的过程——在行为上进行了表征，但其与内部模型不确定性的关系尚不清楚。为了系统地研究这一问题，我们在12个具有分层认识状态的健康科学主张中，沿着两个维度（来源权威性和证据质量）变化说服尝试。失调可以通过说服、反弹或免疫来解决。我们引入了信任弹性（Trust Elasticity, TE），这是一个受计量经济学启发的度量，表示模型在多大程度上容易受到相互矛盾证据的说服。在四个大型语言模型中，TE的变化显著，而明显错误的主张在所有模型中几乎没有TE。在两个开放权重模型中，我们进一步发现这种变化与两个互补的内部不确定性指标相关，即Qwen中的信心失调（Confidence Miscalibration）和Llama中的内部不确定性变化（Internal Uncertainty Change）。这些结果将跨模型的行为变化与可测量的内部属性联系起来，并指出针对内部不确定性的干预作为未来的研究方向。
cs.AI / 117 / 2606.22673
AgentLens: Interpretable Safety Steering via Mechanistic Subspaces for Multi-Turn Coding Agent
AgentLens：通过机制子空间实现可解释的多轮编码代理安全引导
Abstract
Coding agents based on large language models (LLMs) demonstrate remarkable autonomous capabilities, but they also introduce significant safety and misuse risks during multi-turn interactions with external environments. Existing safety mechanisms mainly rely on external guardrails, which have a limited ability to perform fine-grained behavioral control during execution. Meanwhile, recent mechanistic interpretability methods for LLM safety are mostly confined to single-turn or jailbreak-style QA settings, limiting their ability to capture the evolving risk dynamics of multi-turn agent execution. In this paper, we investigate the safety of multi-turn coding agents from an internal perspective. We propose AgentLens (Mechanistic Subspace Intervention and Steering), a white-box defense framework that performs runtime safety detection and representation-level mitigation for coding agents. Unlike conventional agent guardrails, AgentLens detect harmful execution states from step-level hidden representations and mitigate unsafe behavior by intervening in a 10-dimensional subspace within a single layer. To support this research, we introduce the Mechanistic Agent Safety (MAS) benchmark, comprising comprehensively annotated multi-turn execution trajectories across 194 tasks using LLaMA-3.1-8B, Qwen-2.5-7B, and Gemma-2-9B. Extensive experiments show that AgentLens achieves strong safety detection performance, provides preliminary evidence for lookahead risk anticipation, and substantially reduces harmful actions of the coding agent, establishing a foundation for applying mechanistic interpretability to dynamic LLM agent safety. The code is available at: https://github.com/EddyLuo1232/AgentLens
Chinese Translation
基于大型语言模型（LLMs）的编码代理展现出显著的自主能力，但在与外部环境的多轮交互中也引入了显著的安全性和误用风险。现有的安全机制主要依赖于外部护栏，这在执行过程中对细粒度行为控制的能力有限。同时，近期针对LLM安全的机制可解释性方法大多局限于单轮或越狱式问答设置，限制了其捕捉多轮代理执行中不断变化的风险动态的能力。本文从内部视角探讨多轮编码代理的安全性。我们提出了AgentLens（机制子空间干预与引导），这是一个白盒防御框架，能够对编码代理进行运行时安全检测和表示层级的缓解。与传统的代理护栏不同，AgentLens通过从步级隐藏表示中检测有害执行状态，并通过在单层内干预一个10维子空间来缓解不安全行为。为了支持这一研究，我们引入了机制代理安全（MAS）基准，包含使用LLaMA-3.1-8B、Qwen-2.5-7B和Gemma-2-9B的194个任务的全面注释多轮执行轨迹。大量实验表明，AgentLens在安全检测性能上表现出色，为前瞻性风险预判提供了初步证据，并显著减少了编码代理的有害行为，为将机制可解释性应用于动态LLM代理安全奠定了基础。代码可在以下链接获取：https://github.com/EddyLuo1232/AgentLens
cs.AI / 118 / 2606.22676
Skin-Deep: A Geometric Diagnostic for Alignment Fragility in Large Language Model Representations
肤浅：大型语言模型表示中的对齐脆弱性几何诊断
Abstract
Alignment tuning is meant to make harmful-request refusal robust, yet this safety behavior can be erased by a small set of benign fine-tuning examples. This is a deployment risk for open-weight models because a checkpoint can pass refusal tests at release time and later lose refusal under low-cost downstream fine-tuning. Prior work has established these refusal failures, but existing studies do not show how to detect this fragility in the aligned model itself before an attack or fine-tuning intervention is run. We introduce Skin-Deep, a geometric diagnostic that detects alignment fragility directly from the aligned model's hidden-state activations before such an intervention is run and compresses the layer-wise safety geometry into a single scalar, the Geometric Fragility Score (GFS). Applied to twenty-one instruction-tuned models spanning six alignment recipes and 3B--32B parameters, Skin-Deep reveals a recurring low-rank safety subspace across model families. Direction ablations show that removing directions in this subspace weakens harmful-request refusal, providing causal evidence that the recovered geometry underlies refusal behavior. Crucially, GFS identifies, before any fine-tuning, the initially safe model that retains the most refusal after small-scale LoRA fine-tuning. These results establish GFS as a practical pre-deployment diagnostic for flagging fragile refusal behavior without running an attack.
Chinese Translation
对齐调优旨在使有害请求拒绝变得稳健，然而，这种安全行为可能会被一小组良性微调示例所抹去。这对开放权重模型构成了部署风险，因为一个检查点在发布时可以通过拒绝测试，但在低成本的下游微调后可能会失去拒绝能力。先前的研究已确认了这些拒绝失败，但现有研究并未展示如何在攻击或微调干预运行之前检测对齐模型本身的脆弱性。我们引入了Skin-Deep，这是一种几何诊断工具，可以直接从对齐模型的隐藏状态激活中检测对齐脆弱性，并将层级安全几何压缩为一个标量，即几何脆弱性评分（Geometric Fragility Score, GFS）。在覆盖六种对齐配方和3B--32B参数的二十一种指令调优模型上应用Skin-Deep，揭示了模型家族中反复出现的低秩安全子空间。方向消融实验表明，去除该子空间中的方向会削弱有害请求的拒绝，提供了恢复的几何结构与拒绝行为之间的因果证据。至关重要的是，GFS在任何微调之前识别出最初安全的模型，该模型在小规模LoRA微调后仍能保留最多的拒绝能力。这些结果确立了GFS作为一种实用的预部署诊断工具，用于标记脆弱的拒绝行为，而无需进行攻击。
cs.AI / 119 / 2606.22692
VISTA Architect: A graph database-oriented health AI system demonstrated in multidisciplinary tumor boards
VISTA Architect：一种面向图数据库的健康人工智能系统，在多学科肿瘤委员会中展示
Abstract
We introduce VISTA Architect, a database-oriented AI architecture for integrating large language models (LLMs) with longitudinal electronic health records (EHRs). At ingestion, it transforms complex clinical documentation into a persistent, provenance-linked knowledge graph, eliminating repeated reprocessing of raw records at query time. The architecture has two layers: a source-faithful MEDS Graph preserving granular EHR structure with full provenance, and a clinically abstracted Timeline Object Architecture (TOA) that uses graph-guided LLM extraction to synthesize a concise timeline of deduplicated, temporally coherent clinical events. This addresses key limitations of direct long-context prompting and retrieval-augmented generation (RAG), which often miss temporal relationships and incur high cost and latency from repeated raw-text processing. By precomputing clinical synthesis once, downstream queries access an organized patient state and traverse to source documentation only when detailed verification is needed. We demonstrate the system in multidisciplinary thoracic oncology tumor boards at Stanford Medicine, where precise reconstruction of patient histories is critical. Across 1,180 patients, VISTA Architect achieved 96.4% accuracy (mean 9.75/10) on 15 tumor board-salient variables (17,700 evaluations; 95% CI 96.1-96.7%), surpassing a matched BM25 RAG baseline and recent benchmarks for LLM-based clinical extraction. An agentic interface reduced preparation for a 30-patient held-out cohort to about 2.2 minutes without sacrificing accuracy. While configured here for thoracic oncology, the modular design adapts to other specialties through customizable event definitions, episode structures, and agentic tools; validation beyond thoracic oncology remains future work.
Chinese Translation
我们介绍了VISTA Architect，一种面向数据库的人工智能架构，用于将大型语言模型（LLMs）与纵向电子健康记录（EHRs）集成。在数据摄取时，它将复杂的临床文档转化为一个持久的、具有来源链接的知识图谱，从而消除了在查询时对原始记录的重复处理。该架构分为两层：一层是忠实于数据源的MEDS Graph，保留了细粒度的EHR结构及完整的来源信息；另一层是临床抽象的时间线对象架构（Timeline Object Architecture, TOA），它利用图引导的LLM提取技术合成一个简明的去重、时间一致的临床事件时间线。这解决了直接长上下文提示和检索增强生成（RAG）的关键局限性，这些方法往往忽视时间关系，并因重复处理原始文本而产生高成本和延迟。通过一次性预计算临床合成，下游查询可以访问组织良好的患者状态，并仅在需要详细验证时才遍历源文档。我们在斯坦福医学中心的多学科胸部肿瘤委员会中展示了该系统，在这里，精确重建患者历史至关重要。在1,180名患者中，VISTA Architect在15个肿瘤委员会显著变量上达到了96.4%的准确率（平均9.75/10）（17,700次评估；95%置信区间为96.1-96.7%），超越了匹配的BM25 RAG基线和近期的LLM基础临床提取基准。一个自主接口将对30名患者保留队列的准备时间缩短至约2.2分钟，而不牺牲准确性。尽管在这里为胸部肿瘤配置，但模块化设计通过可定制的事件定义、病例结构和自主工具可适应其他专业领域；超越胸部肿瘤的验证仍然是未来的工作。
cs.AI / 120 / 2606.22706
Safety-Aware Evaluation of LLM-Generated Driver Intervention Messages through Multi-Task Risk Fusion
基于多任务风险融合的LLM生成驾驶干预信息的安全意识评估
Abstract
Existing driver intervention systems rely on auditory alerts and fixed templates, failing to leverage multi-task recognition outputs. General-purpose metrics such as BLEU and BERTScore cannot capture intervention-specific quality dimensions including risk-urgency alignment, cognitive load, and driver acceptability. In this paper, we propose the Driver Safety-Aware Intervention Score (DSAIS), a domain-specific metric evaluating five dimensions through a hybrid architecture combining lightweight rule-based computation with LLM Judge evaluation, together with an end-to-end framework integrating four-task recognition outputs into an LLM through risk fusion, state history management, and dynamic prompt construction. Experiments on the AIDE dataset with five models and seven conditions demonstrate that DSAIS achieves ICC 0.798-0.840 across three architecturally distinct judges and Cohen's d > 1.5 across all control conditions. Multi-dimensional sub-score analysis quantifies the contextual adaptability gap between rule-based and LLM-based systems, revealing that multi-task integration improves contextual relevance by 9.1% over rule-based baselines. Ablation experiments demonstrate that each framework component contributes to contextual relevance, with sub-score decomposition revealing gains that aggregate scoring masks. Driver emotion recognition is identified as the most critical upstream factor, and compact local LLMs (7B--9B parameters) achieve quality superior to API-based models, providing practical design guidelines for in-vehicle deployment.
Chinese Translation
现有的驾驶干预系统依赖于听觉警报和固定模板，未能充分利用多任务识别输出。通用指标如BLEU和BERTScore无法捕捉干预特定的质量维度，包括风险紧迫性对齐、认知负荷和驾驶员可接受性。本文提出了驾驶安全意识干预评分（Driver Safety-Aware Intervention Score, DSAIS），这是一种领域特定的指标，通过结合轻量级基于规则的计算与LLM Judge评估的混合架构，评估五个维度，并整合四任务识别输出到LLM中，采用风险融合、状态历史管理和动态提示构建的端到端框架。在AIDE数据集上进行的实验显示，DSAIS在三种架构不同的评估者中实现了ICC 0.798-0.840，并且在所有控制条件下Cohen's d > 1.5。多维子评分分析量化了基于规则系统与基于LLM系统之间的上下文适应性差距，揭示多任务集成使上下文相关性比基于规则的基线提高了9.1%。消融实验表明，每个框架组件对上下文相关性都有贡献，子评分分解揭示了聚合评分掩盖的增益。驾驶员情绪识别被确定为最关键的上游因素，而紧凑型本地LLM（7B--9B参数）在质量上优于基于API的模型，为车载部署提供了实用的设计指南。
cs.AI / 121 / 2606.22716
Beyond Penalizing Mistakes: Stabilizing Efficiency Training in Large Reasoning Models via Adaptive Correct-Only Rewards
超越惩罚错误：通过自适应仅正确奖励稳定大型推理模型的效率训练
Abstract
Training large language models to reason efficiently is a critical challenge. While integrating length-penalizing rewards into Group Relative Policy Optimization (GRPO) aims to reduce verbosity, it frequently triggers reward collapse, severely degrading reasoning capabilities. Through a systematic evaluation of various reward configurations, we identify the root mechanism: GRPO's group normalization creates divergent advantages when incorrect answers receive continuous length penalties. Consequently, methods penalizing the length of incorrect answers are structurally prone to collapse under sustained optimization. Furthermore, restricting penalties exclusively to correct answers avoids this primary failure, but leaves the model susceptible to a stochastic collapse driven by response over-compression. To robustly prevent both failure modes, we propose ACOER (Adaptive Correct-Only Efficiency Reward). ACOER eliminates the structural penalty loop by isolating brevity bonuses to correct completions and prevents stochastic compression via dynamic budget normalization and control-loop penalty adjustments. Evaluated across diverse mathematical reasoning benchmarks, ACOER improves overall accuracy compared to the base model while reducing token generation by over 60%, establishing a fundamentally stable approach for efficiency-aware optimization.
Chinese Translation
训练大型语言模型以高效推理是一个关键挑战。虽然将长度惩罚奖励集成到组相对策略优化（Group Relative Policy Optimization, GRPO）中旨在减少冗长，但它经常引发奖励崩溃，严重降低推理能力。通过对各种奖励配置的系统评估，我们识别出根本机制：GRPO的组归一化在错误答案受到持续长度惩罚时产生了分歧优势。因此，惩罚错误答案长度的方法在持续优化下结构上容易崩溃。此外，仅将惩罚限制在正确答案上可以避免这一主要失败，但使得模型容易受到由响应过度压缩引发的随机崩溃。为了稳健地防止这两种失败模式，我们提出了ACOER（自适应仅正确效率奖励）。ACOER通过将简洁奖励隔离到正确完成中消除了结构惩罚循环，并通过动态预算归一化和控制循环惩罚调整防止随机压缩。在多种数学推理基准测试中评估，ACOER相比基础模型提高了整体准确性，同时将令牌生成减少了超过60%，建立了一种根本稳定的效率优化方法。
cs.AI / 122 / 2606.22726
Text Dictates, Music Decorates: Energy-based Attention for Editable Dance Motion Generation
文本主导，音乐装饰：基于能量的注意力机制用于可编辑舞蹈动作生成
Abstract
Choreographic motion generation poses unique challenges for AI, demanding precise semantic control over complex, temporally structured, and expressive full-body dynamics. While existing models can synthesize motion from music, they remain largely black boxes. Conversely, attempting to condition generation on both text and music frequently leads to modality collapse, where dense acoustic rhythms overwhelm sparse semantic text prompts, destroying user controllability. To resolve this spatial-temporal conflict, we propose STREAM (Structural-Temporal Rhythmic Energy-based Attention for Motion), a modality-decoupled diffusion transformer. STREAM strictly separates conditioning pathways: global text semantics dictate the kinematic structure via Adaptive Layer Normalization (AdaLN), while a novel Bimodal Energy-Based Attention Module (BEAM) routes these features to the musical beat without overwriting the semantics. We further introduce Motorica++, a newly curated dataset enriched with domain-specific dance vocabulary and frame-level semantic annotations from existing Motorica dataset. Additionally, to rigorously quantify zero-shot editability, we propose the Exchange Evaluation Protocol and Editable Dance Score (EDS). Through extensive experiments, STREAM achieves state-of-the-art alignment between motion and music while perfectly preserving choreographic semantics, positioning AI not merely as a reactive synthesizer, but as a controllable, collaborative partner for artistic direction. The source code and datasets are available at https://github.com/SeongJong-Yoo/STREAM.
Chinese Translation
编舞动作生成对人工智能提出了独特的挑战，要求对复杂、时间结构化和富有表现力的全身动态进行精确的语义控制。虽然现有模型能够从音乐合成动作，但它们在很大程度上仍然是黑箱。相反，试图同时基于文本和音乐进行生成常常导致模态崩溃，即密集的声学节奏压倒稀疏的语义文本提示，从而破坏用户的可控性。为了解决这一时空冲突，我们提出了STREAM（基于结构-时间节奏能量的动作注意力机制），这是一种模态解耦的扩散变换器。STREAM严格分离了条件路径：全局文本语义通过自适应层归一化（Adaptive Layer Normalization，AdaLN）决定运动学结构，而一种新颖的双模态能量基础注意力模块（Bimodal Energy-Based Attention Module，BEAM）将这些特征引导到音乐节拍上，而不覆盖语义。此外，我们进一步引入了Motorica++，这是一个新编制的数据集，丰富了特定领域的舞蹈词汇和来自现有Motorica数据集的帧级语义注释。此外，为了严格量化零-shot 可编辑性，我们提出了交换评估协议和可编辑舞蹈评分（Editable Dance Score，EDS）。通过大量实验，STREAM在动作与音乐之间实现了最先进的对齐，同时完美保留了编舞语义，使人工智能不仅仅是一个反应式合成器，而是一个可控的、协作的艺术指导伙伴。源代码和数据集可在 https://github.com/SeongJong-Yoo/STREAM 获取。
cs.AI / 123 / 2606.22731
Closed-loop Auto Research for Molecular Property Prediction: Discovering and Certifying Generalizable Improvements
闭环自动研究用于分子性质预测：发现和验证可推广的改进
Abstract
Closed-loop Auto Research extends automated machine learning from fixed-dataset fitting to changing the research workflow, with language-model agents editing representations and model code and acquiring external evidence. Molecular property prediction spans many small endpoints. We ask whether this action space yields improvements generalizing beyond the validation signal selecting them. We isolate three Auto Research axes, features, models, and external evidence, under a file-level ablation lock attributing each gain to one axis over a strong baseline. Across 36 endpoints in three benchmark suites we score each selected configuration once on a held-out test whose labels the search never read. A routed pipeline taking each endpoint's best validation axis reaches positive held-out gains of 0.013, 0.011, and 0.042, the transferable axis differing by suite, data on TDC, model on Polaris, feature and model on MoleculeNet. The largest model-search gain falls from 0.041 on validation to 0.003 on test, while curated data reaches 0.022 but negative 0.019 on test, two non-transfer signatures. Curated external data raises held-out CYP2C9-substrate performance by 0.17 and half-life by 0.08, admitted through a contamination filter rejecting same-source files overlapping 64 to 89 percent of test structures, necessary but not sufficient for transfer. A matched-trial automated machine learning control did not reproduce the agent's code-level model intervention, reaching 0.006 against 0.042, and the pipeline stays competitive with an 84M-parameter pretrained 3D model on the shared training split. The experiments stay within molecular property prediction, but separating discovery from held-out certification is a domain-agnostic lesson for any closed-loop system optimising a proxy for a held-out quantity.
Chinese Translation
闭环自动研究将自动化机器学习从固定数据集拟合扩展到改变研究工作流程，通过语言模型代理编辑表示和模型代码并获取外部证据。分子性质预测涵盖许多小型端点。我们探讨这种行动空间是否能够产生超出选择它们的验证信号的可推广改进。我们在文件级消融锁下隔离了三个自动研究轴：特征、模型和外部证据，将每个增益归因于一个轴，相对于强基线。在三个基准套件中的36个端点上，我们对每个选定配置在一个未见标签的保留测试集上进行评分。一个路由管道利用每个端点的最佳验证轴达到了0.013、0.011和0.042的正向保留增益，传递轴因套件而异，数据在TDC上，模型在Polaris上，特征和模型在MoleculeNet上。最大的模型搜索增益从验证的0.041降至测试的0.003，而经过策划的数据在测试中达到了0.022，但在测试中为负的0.019，两个非传递特征。经过策划的外部数据将保留的CYP2C9底物性能提高了0.17，半衰期提高了0.08，通过一个污染过滤器接纳，该过滤器拒绝重叠64%至89%测试结构的同源文件，这对传递是必要的但不足够。一个匹配试验的自动化机器学习控制未能重现代理的代码级模型干预，达到了0.006，而不是0.042，并且该管道在共享训练分割上与一个84M参数的预训练3D模型保持竞争。实验仍然局限于分子性质预测，但将发现与保留认证分开是任何优化保留量代理的闭环系统的领域无关的教训。
cs.AI / 124 / 2606.22737
GroundEval: A Deterministic Replacement for LLM-as-Judge in Stateful Agent Evaluation
GroundEval：一种确定性的替代方案，用于状态代理评估中的 LLM 作为评判者
Abstract
Before letting an agent operate over real context, can you prove it used the right evidence? GroundEval turns that question into a deterministic test of what the agent searched, fetched, cited, and was permitted to access. In one case study, two frontier LLM judges scored a plausible agent response above 0.85. But the trace told a different story: the agent had never retrieved the artifact its answer depended on, yielding a GroundEval score of 0.000. We introduce GroundEval, a judge-free framework for evaluating agents against grounded, time-bounded, and access-controlled evidence. GroundEval uses a domain configuration to generate questions, lets the agent choose how to answer, and then scores both the final answer and the recorded trajectory that produced it. The benchmark targets three failures that LLM-as-judge evaluation struggles to detect: whether an agent checked before claiming absence, reasoned only from evidence available to the actor at the relevant time, and used the correct causal mechanism rather than a plausible one. These correspond to three tracks: Silence, Perspective, and Counterfactual. GroundEval exposes when plausible answers rest on invalid evidence paths, and produces structured per-question diagnostics that pair tool activity with the agent's turn-level narration, making each score inspectable rather than merely reported. What our case studies turned up is that this gap isn't some rare corner case. It's exactly the blind spot that final-answer and judge-based scoring were never built to catch.
Chinese Translation
在让代理在真实环境中操作之前，您能否证明它使用了正确的证据？GroundEval 将这个问题转化为对代理所搜索、获取、引用以及被允许访问的内容的确定性测试。在一个案例研究中，两位前沿的 LLM 评判者对一个合理的代理响应评分超过 0.85。但追踪记录却讲述了一个不同的故事：该代理从未检索到其答案所依赖的文献，导致 GroundEval 得分为 0.000。我们介绍了 GroundEval，这是一个无评判者的框架，用于根据有依据的、时间限制的和访问控制的证据评估代理。GroundEval 使用领域配置生成问题，让代理选择如何回答，然后对最终答案和生成答案的记录轨迹进行评分。该基准针对 LLM 作为评判者评估难以检测的三种失败：代理在声称缺失之前是否进行了检查、是否仅从相关时间内可用的证据进行推理，以及是否使用了正确的因果机制而不是一个合理的机制。这对应于三个轨道：沉默、视角和反事实。GroundEval 揭示了何时合理的答案依赖于无效的证据路径，并生成结构化的逐题诊断，将工具活动与代理的轮次叙述相结合，使每个得分都可以检查，而不仅仅是报告。我们的案例研究发现，这个差距并不是某个罕见的边缘案例。它正是最终答案和基于评判者的评分从未设计用来捕捉的盲点。
cs.AI / 125 / 2606.22786
Learning Filters with Certainty
学习带有确定性的过滤器
Abstract
Hash-based data structures such as Bloom filters are widely used in network systems for tasks including caching, anomaly detection, and machine learning pipelines. They typically provide binary indications of whether an element belongs to a set of interest, e.g., the contents of a cache. When uncertainty arises due to hash collisions, a positive indication is returned to avoid false negatives. We argue that the certainty associated with such indications can itself be useful information. This work focuses on Counting Bloom Filters (CBFs), a Bloom-filter variant that maintains counters rather than bits. Besides supporting insertions and deletions, these counters provide additional information that can be used to estimate the certainty of positive membership indications. We show how this certainty signal can be exploited in architectures that combine Bloom Filters with machine learning (ML) models.
Chinese Translation
基于哈希的数据结构，如布隆过滤器（Bloom filters），广泛应用于网络系统中，承担缓存、异常检测和机器学习管道等任务。它们通常提供二元指示，表明一个元素是否属于感兴趣的集合，例如缓存的内容。当由于哈希冲突而产生不确定性时，为了避免假阴性，系统会返回一个积极的指示。我们认为，与这些指示相关的确定性本身也可以是有用的信息。本研究聚焦于计数布隆过滤器（Counting Bloom Filters, CBFs），这是一种维护计数器而非位的布隆过滤器变体。除了支持插入和删除操作外，这些计数器还提供额外的信息，可以用来估计积极成员指示的确定性。我们展示了如何在将布隆过滤器与机器学习（ML）模型结合的架构中利用这一确定性信号。
cs.AI / 126 / 2606.22792
The Origins of Stochasticity: Comprehensive Investigations on Uncertainty Quantification for Large Language Models
随机性的起源：对大型语言模型不确定性量化的综合研究
Abstract
Recent advancements in Large Language Models (LLMs) have enabled sophisticated reasoning and content generation, yet their inherent stochasticity poses significant challenges for ensuring predictive credibility. While traditional uncertainty taxonomy paradigms, such as the dichotomy of aleatoric and epistemic uncertainties, provide conceptual foundations, they often fail to capture the multi-component and multi-stage nature of LLM generation and struggle to evaluate the effectiveness of various Uncertainty Quantification (UQ) methods. In this paper, we propose a granular uncertainty taxonomy that systematically attributes LLM uncertainty into input-level, parameter-level, token-level, and decoding-process sources. Correspondingly, we categorize existing UQ methods into Bayesian, ensemble, consensus-based, and single-pass approaches. Furthermore, we introduce a comprehensive evaluation framework covering diverse generation settings and metrics. We empirically evaluate 21 typical UQ methods across three prominent LLM families, including Qwen3, Llama 3.2, and DeepSeek-V3, on benchmarks such as TriviaQA, GSM8K, and HumanEval. Our experimental results demonstrate that (i) the effectiveness of UQ methods is sensitive to task types and generation settings; (ii) consensus-based methods, typed Deg and EigV, consistently outperform other UQ approaches; and (iii) larger model scales correlate with lower uncertainty estimates, suggesting an empirical scaling law for LLM uncertainty. This work bridges the gap between theoretical origins and practical deployment, providing a versatile diagnostic tool for systematically quantifying uncertainty in LLM applications.
Chinese Translation
近期大型语言模型（LLMs）的进展使得复杂的推理和内容生成成为可能，但其固有的随机性对确保预测可信度提出了重大挑战。尽管传统的不确定性分类范式，如随机不确定性（aleatoric uncertainty）和认知不确定性（epistemic uncertainty）的二分法提供了概念基础，但它们往往无法捕捉LLM生成的多组件和多阶段特性，并且难以评估各种不确定性量化（UQ）方法的有效性。在本文中，我们提出了一种细化的不确定性分类法，将LLM的不确定性系统地归因于输入级、参数级、标记级和解码过程等来源。相应地，我们将现有的UQ方法分类为贝叶斯（Bayesian）、集成（ensemble）、基于共识（consensus-based）和单次通过（single-pass）方法。此外，我们引入了一个全面的评估框架，涵盖了多种生成设置和指标。我们在TriviaQA、GSM8K和HumanEval等基准上，对包括Qwen3、Llama 3.2和DeepSeek-V3在内的三大LLM家族中的21种典型UQ方法进行了实证评估。实验结果表明：（i）UQ方法的有效性对任务类型和生成设置敏感；（ii）基于共识的方法，如Deg和EigV，始终优于其他UQ方法；（iii）更大的模型规模与较低的不确定性估计相关，暗示了LLM不确定性的经验缩放法则。本研究弥合了理论起源与实际应用之间的差距，为系统量化LLM应用中的不确定性提供了多功能的诊断工具。
cs.AI / 127 / 2606.22793
A Formula-Driven Survey and Research Agenda for On-Policy Distillation
基于公式的在线策略蒸馏调查与研究议程
Abstract
On-policy distillation (OPD) trains an LLM on states induced by the current or recent student policy: the student generates complete or partial rollouts, a teacher or self-teacher scores the resulting tokens under their generated contexts, and dense log-probability, logit, or distributional signals are converted into post-training updates. This survey studies OPD as a feedback-to-update problem rather than a single loss family. We develop a formula-driven taxonomy from two routes -- direct distributional losses and policy-gradient-style log-ratio updates -- and use it to organize core methods, verifier- or outcome-guided hybrids, industrial reports, framework implementations, failure modes, and stabilization recipes under explicit evidence boundaries. The taxonomy shows that OPD effectiveness depends not only on KL direction or teacher access, but also on state compatibility, support construction, temporal credit, vocabulary-level probability routing, gates and weights, and regularization. We further separate two mechanisms often conflated in sampled-token OPD stability discussions. Temporal credit asks how teacher-student log-ratio returns should weight sampled actions across a rollout; vocabulary routing asks where probability mass should move when negative feedback suppresses a sampled token. This distinction yields bias boundaries for immediate, return-to-go, discounted, and baseline-corrected estimators, motivates GAE-OPD as a value-based hypothesis for log-ratio returns, and motivates Counterfactual Routed OPD (CR-OPD) for routing probability mass toward teacher-supported, student-reachable alternatives. We close by mapping actionability diagnostics, failure mechanisms, case studies, open problems, and a reporting checklist onto the same feedback-to-update variables.
Chinese Translation
在线策略蒸馏（On-policy distillation, OPD）在当前或最近的学生策略所诱导的状态上训练大型语言模型（LLM）：学生生成完整或部分的回放，教师或自我教师在其生成的上下文中对结果的标记进行评分，密集的对数概率、对数比或分布信号被转换为后训练更新。本文将OPD视为反馈到更新的问题，而非单一损失家族。我们从两个方向发展出一个基于公式的分类法——直接分布损失和政策梯度风格的对数比更新——并利用该分类法来组织核心方法、验证者或结果引导的混合方法、工业报告、框架实现、失败模式和稳定化方案，均在明确的证据边界下进行分析。该分类法表明，OPD的有效性不仅依赖于KL方向或教师的可访问性，还依赖于状态兼容性、支持构建、时间信用、词汇级概率路由、门控和权重以及正则化。我们进一步区分了在采样标记的OPD稳定性讨论中常常混淆的两种机制。时间信用探讨教师-学生对数比回报应如何在回放中加权采样动作；词汇路由探讨当负反馈抑制采样标记时，概率质量应向何处移动。这一区分为即时、回报至终、折扣和基线校正估计器设定了偏差边界，激励了GAE-OPD作为对数比回报的基于价值的假设，并推动了反事实路由OPD（Counterfactual Routed OPD, CR-OPD）向教师支持的、学生可达的替代方案路由概率质量。最后，我们将可操作性诊断、失败机制、案例研究、开放问题和报告检查表映射到相同的反馈到更新变量上。
cs.AI / 128 / 2606.22797
Measuring Behavior Portability in Large Language Models
测量大型语言模型中的行为可移植性
Abstract
Large language models are increasingly deployed as autonomous decision makers, yet the behavioral mapping they exhibit can vary substantially across decision environments that are payoff-equivalent by construction-environments that share identical payoff-relevant structure but differ in surface presentation. This sensitivity renders suite-based evaluation fragile and raises a fundamental question of behavioral portability: how well does a behavioral mapping learned in one decision environment informative on another that preserves the same underlying incentive structure? We introduce a formal framework to measure this property. Our protocol fits an interpretable behavioral model on data pooled from a set of source environments and evaluates its out-of-sample predictive performance in a held-out target environment, benchmarking against an oracle trained directly on target data. Portability is quantified via a loss-agnostic measure that delivers worst-case bounds on the performance of the induced prediction-action mapping in the target environment. In controlled experiments spanning seven canonical economic decision problems, we document substantial and systematic portability losses, suggesting that behavioral characterizations of LLMs obtained in one decision environment cannot be assumed to transfer reliably to structurally equivalent alternatives.
Chinese Translation
大型语言模型越来越多地被作为自主决策者部署，但它们所表现出的行为映射在构造上是收益等价的决策环境中可能会有显著差异——这些环境共享相同的收益相关结构，但在表面表现上有所不同。这种敏感性使得基于套件的评估变得脆弱，并提出了一个关于行为可移植性的基本问题：在保持相同基础激励结构的情况下，在一个决策环境中学习到的行为映射在另一个环境中有多大信息量？我们引入了一个正式框架来测量这一属性。我们的协议在从一组源环境中汇总的数据上拟合一个可解释的行为模型，并在一个保留的目标环境中评估其样本外预测性能，以直接在目标数据上训练的神谕模型作为基准。可移植性通过一种与损失无关的度量进行量化，该度量提供了在目标环境中诱导的预测-行动映射性能的最坏情况界限。在涵盖七个经典经济决策问题的控制实验中，我们记录了显著且系统的可移植性损失，这表明在一个决策环境中获得的LLM行为特征不能可靠地转移到结构上等价的替代环境中。
cs.AI / 129 / 2606.22809
AI-Assisted Help-Seeking Trajectories in Programming Education from an SRL-Informed Perspective
从自我调节学习（SRL）视角看编程教育中的AI辅助求助轨迹
Abstract
Generative AI tools provide novice programmers with instant, personalized support, but also raise concerns about whether AI use supports or bypasses students' regulation of problem-solving. Existing work has largely focused on correctness, usability, or overall usage frequency, with less attention to how student--AI help-seeking unfolds. This study addresses this gap by analyzing AI-assisted help-seeking trajectories in university-level programming. Using an SRL-informed analytical framework that links prompt-level help-seeking codes to conceptual, implementation, debugging, and reflective forms of support, we analyzed 1,290 task-specific student prompts linked to 17,190 code submissions from 71 students in introductory Python programming courses. Specifically, we examined how help-seeking interactions were structured across turns and attempts, and how trajectory patterns related to task scores and the number of code submissions. Results indicate that many students primarily used AI for reactive troubleshooting rather than for planned, self-regulated problem-solving. Although trajectory patterns were not associated with significant differences in task scores, they differed substantially in the number of code submissions required. These findings suggest that the educational significance of AI support lies not only in whether students use AI, but in how their help-seeking trajectories develop during programming problem-solving.
Chinese Translation
生成性AI工具为初学者程序员提供即时、个性化的支持，但也引发了关于AI使用是否支持或绕过学生问题解决调节的担忧。现有研究主要集中在正确性、可用性或整体使用频率上，而对学生与AI求助互动的展开关注较少。本研究通过分析大学层级编程中的AI辅助求助轨迹来填补这一空白。我们使用一个基于自我调节学习（SRL）的分析框架，将提示级求助编码与概念性、实施、调试和反思形式的支持相联系，分析了来自71名学生的1290个任务特定提示，这些提示与17190个代码提交相关，均来自初级Python编程课程。具体而言，我们考察了求助互动在轮次和尝试中的结构，以及轨迹模式与任务得分和代码提交数量之间的关系。结果表明，许多学生主要将AI用于反应式故障排除，而非计划性、自我调节的问题解决。尽管轨迹模式与任务得分之间没有显著差异，但在所需的代码提交数量上存在显著差异。这些发现表明，AI支持的教育意义不仅在于学生是否使用AI，更在于他们在编程问题解决过程中求助轨迹的发展。
cs.AI / 130 / 2606.22813
Active Inference as the Test-Time Scaling Law for Physical AI Agents
主动推理作为物理人工智能代理的测试时刻缩放法则
Abstract
In this paper, a novel test-time scaling law for physical artificial intelligence (AI) agents is introduced. This scaling law enables physical AI agents to reason with their world models to generalize in unforeseen scenarios at test time. The derived scaling law is grounded in the first principle of active inference, which equips agents with the general objective to survive in the real world, under which their specific task objectives are subsumed. Active inference achieves this by providing the reasoning to resolve prediction errors that arise when the agent encounters unforeseen situations outside its training distribution, enabling generalization in non-stationary environments. The proposed scaling law captures this by dynamically updating the agent's policy with this reasoning at test time. This policy update is modeled as a soft Bayesian inference process in which beliefs about the policy are updated using the reasoning that reduces expected prediction errors under allowable policies as a likelihood. The resulting posterior policy admits a biological interpretation, recovering the scaling mechanism that engages the brain's basal ganglia and prefrontal cortex at test time. To solve this analytically intractable problem, a variational inference solution minimizing free energy bounds is developed. This solution extends to enable learning beyond training by reinforcing new instances, resolved at test time, in both the policy and world model. Unlike existing scaling laws constrained by model size and training data, the derived solution scales with the continuous real-world experience of a physical AI agent. Simulation results on an autonomous driving task demonstrate that the proposed solution outperforms model-free Q-learning and model-based Bayesian reinforcement learning, achieving robust generalization to unforeseen scenarios while improving inference efficiency by over 36%.
Chinese Translation
本文提出了一种针对物理人工智能（AI）代理的新型测试时刻缩放法则。该缩放法则使物理AI代理能够利用其世界模型进行推理，从而在测试时对未预见的场景进行泛化。所推导的缩放法则基于主动推理的第一原理，赋予代理在现实世界中生存的一般目标，在此目标下，其特定任务目标被包含其中。主动推理通过提供解决预测误差的推理来实现这一点，当代理遇到超出其训练分布的未预见情况时，从而在非平稳环境中实现泛化。所提出的缩放法则通过在测试时动态更新代理的策略来捕捉这一点。该策略更新被建模为一种软贝叶斯推理过程，其中关于策略的信念通过减少在可允许策略下的预期预测误差的推理来更新，作为似然性。最终得到的后验策略具有生物学解释，恢复了在测试时激活大脑基底神经节和前额叶皮层的缩放机制。为了解决这一分析上不可处理的问题，开发了一种最小化自由能界限的变分推理解决方案。该解决方案扩展到通过在测试时强化新实例，使学习超越训练，在策略和世界模型中得到解决。与现有受模型大小和训练数据限制的缩放法则不同，所推导的解决方案与物理AI代理的连续现实世界经验相适应。对自主驾驶任务的仿真结果表明，所提出的解决方案优于无模型Q学习和基于模型的贝叶斯强化学习，在未预见场景中实现了稳健的泛化，同时将推理效率提高了超过36%。
cs.AI / 131 / 2606.22826
MINCE: Shrinking LLM Evaluation Datasets via Few-Model Monte Carlo Calibration
MINCE：通过少模型蒙特卡洛校准缩小LLM评估数据集
Abstract
Evaluating LLMs across many model variants -- quantized, fine-tuned, or deployment-specific -- requires running large benchmarks repeatedly, a process that can take tens of hours per model on edge hardware such as NPUs. Existing subset selection methods reduce this cost but depend on large calibration pools or learned prediction layers. We introduce MINCE (Monte Carlo Informed N-sizing for Compact Evaluation), which uses Monte Carlo simulation over per-item logs from a small set of calibration models to find the minimum subset size that bounds accuracy drift and then fixes a randomly sampled subset at that size, with no prediction layer needed. MINCE reduces IFEVAL by 54\%, MMLU by 89\%, and GSM8K by 70\% with maximum drift $\leq$2.62\,pp on BF16 models and mean drift of 0.77--3.59\,pp on held-out NPU models, while delivering median GPU evaluation speedups of 2.7--8.1$\times$ and NPU evaluation speedups of 1.7--2.0$\times$. The method is robust to calibration pool size and achieves lower drift than tinyBenchmarks (12$\times$ lower on MMLU, 3.3$\times$ on GSM8K) while using 57$\times$ fewer calibration models.
Chinese Translation
在多种模型变体（如量化、微调或特定于部署的模型）上评估LLM需要重复运行大型基准测试，这一过程在边缘硬件（如NPU）上每个模型可能需要数十小时。现有的子集选择方法虽然可以降低这一成本，但依赖于大型校准池或学习的预测层。我们提出了MINCE（蒙特卡洛信息N大小用于紧凑评估），该方法利用来自少量校准模型的逐项日志进行蒙特卡洛模拟，以找到限制准确度漂移的最小子集大小，然后在该大小上固定一个随机抽样的子集，而无需预测层。MINCE在BF16模型上将IFEVAL减少了54%，MMLU减少了89%，GSM8K减少了70%，最大漂移$ ext{≤}2.62$ pp，持出NPU模型的平均漂移为0.77--3.59 pp，同时提供了2.7--8.1$ imes$的中位数GPU评估加速和1.7--2.0$ imes$的NPU评估加速。该方法对校准池大小具有鲁棒性，并且在漂移方面优于tinyBenchmarks（在MMLU上低12$ imes$，在GSM8K上低3.3$ imes$），同时使用的校准模型数量减少了57$ imes$。
cs.AI / 132 / 2606.22830
Finding the Evidence: Discovering Decision-Supporting Tokens for On-Policy Reasoning Distillation
寻找证据：发现用于政策推理蒸馏的决策支持标记
Abstract
On-policy distillation transfers reasoning ability through dense token-level supervision, yet the nature of the transferable signal remains unclear. We discover that reasoning chains contain two types of knowledge that require different discovery mechanisms: decisions (where to branch), which surface through student uncertainty, and evidence (intermediate steps that justify decisions), which hides in positions where the student is confident yet wrong. Current methods capture only decisions; the substantive knowledge in evidence tokens remains untransferred. We propose DEAR(Decision-Evidence Aware Reasoning Distillation), which first identifies decisions via student entropy, then discovers their supporting evidence through hidden-state cosine similarity to decision anchors, boosted by teacher-student divergence to prioritize the largest knowledge gaps. Across three student-teacher configurations on math and code benchmarks, DEAR consistently outperforms standard OPD, with up to +2.5pp on competition math and +5.7pp on code generation.
Chinese Translation
在政策蒸馏中，通过密集的标记级监督转移推理能力，但可转移信号的本质仍不清晰。我们发现推理链包含两种类型的知识，需要不同的发现机制：决策（分支的方向），通过学生的不确定性显现，以及证据（证明决策的中间步骤），则隐藏在学生自信但错误的位置。目前的方法仅捕获决策；证据标记中的实质性知识仍未转移。我们提出了DEAR（决策-证据意识推理蒸馏），该方法首先通过学生熵识别决策，然后通过与决策锚点的隐藏状态余弦相似性发现其支持证据，并通过教师-学生的差异性来优先考虑最大的知识差距。在数学和代码基准的三种学生-教师配置中，DEAR始终优于标准的OPD，在竞争性数学上提高了最多2.5个百分点，在代码生成上提高了最多5.7个百分点。
cs.AI / 133 / 2606.22844
RaMem: Contextual Reinstatement for Long-term Agentic Memory
RaMem：长期代理记忆的情境恢复
Abstract
Long-term memory has become increasingly important for LLM agents that operate across extended interactions and evolving task contexts. Recent memory systems have made past experiences more persistent, compact, and retrievable, but retrieval alone does not ensure that a memory provides valid evidence for the current query. When experiences are compressed into reusable fragments, memories from different situations may appear equally relevant if they involve recurring entities or user states. We refer to this failure as context collapse: memories lose the surrounding context needed to judge whether they provide valid evidence for the current query. To address this problem, we propose Contextual Reinstatement for Agentic Memory (RaMem), a framework that turns retrieved memory fragments into contextually verifiable evidence. RaMem operates through four coordinated stages: (i) evidence anchoring grounds each memory in its original episodic conditions, especially event time, mention time, session span, and participants; (ii) recall condition induction derives the evidence conditions implied by the query; (iii) validity-aware retrieval uses these conditions to prioritize context-compatible memories while retaining content-relevant candidates as fallback evidence; and (iv) context-preserved synthesis keeps the selected memories' structured context available to the generator. Experiments on long-term memory benchmarks show that RaMem consistently improves performance over strong memory baselines, with average F1 gains of more than 10% across several backbones.
Chinese Translation
长期记忆在跨越延续交互和不断演变的任务背景下的LLM代理中变得越来越重要。近期的记忆系统使得过去的经验更加持久、紧凑且可检索，但仅仅依靠检索并不能确保记忆为当前查询提供有效证据。当经验被压缩为可重用的片段时，来自不同情境的记忆可能看起来同样相关，尤其是在涉及重复实体或用户状态时。我们将这种失败称为情境崩溃：记忆失去了判断其是否为当前查询提供有效证据所需的周围情境。为了解决这个问题，我们提出了代理记忆的情境恢复框架（RaMem），该框架将检索到的记忆片段转化为情境可验证的证据。RaMem通过四个协调阶段进行操作：（i）证据锚定将每个记忆与其原始情节条件相结合，特别是事件时间、提及时间、会话跨度和参与者；（ii）回忆条件诱导推导出查询所暗示的证据条件；（iii）有效性感知检索利用这些条件优先考虑与情境兼容的记忆，同时保留内容相关的候选项作为后备证据；（iv）情境保留合成保持所选记忆的结构化情境可供生成器使用。在长期记忆基准测试中的实验表明，RaMem在多个基础模型上始终提高了性能，平均F1得分在多个基础模型上提升超过10%。
cs.AI / 134 / 2606.22859
AI Scientists as Engines of Discovery: A Case for Development within Reformed Institutions
人工智能科学家作为发现引擎：在改革机构内发展的案例
Abstract
Agentic artificial intelligence (AI) systems are beginning to assist, accelerate, and partially automate scientific discovery, performing tasks that span literature synthesis, code generation, data analysis, hypothesis proposal, and model criticism. We argue that this transition is qualitative rather than incremental, and that suitably designed multi-agent systems may evolve from passive computational tools into ``AI scientists'' that can expand the hypothesis-generating and verification capacity of science. Such systems must be developed and deployed within a scientific ecosystem fit for purpose: institutions must be redesigned for verification, accountability, interpretability, and dual-use safety. We sketch how multi-agent architectures, illustrated by the prototype framework \textit{Denario}, accelerate the discovery cycle and traverse model spaces beyond human reach; examine what this implies for authorship, peer review, and the enduring role of human scientists; and close with recommendations for governing AI as an epistemic actor rather than a mere instrument.
Chinese Translation
具有代理能力的人工智能（AI）系统开始协助、加速并部分自动化科学发现，执行涵盖文献综合、代码生成、数据分析、假设提出和模型批评等任务。我们认为，这一转变是质的而非渐进的，适当设计的多智能体系统可能会从被动的计算工具演变为能够扩展科学假设生成和验证能力的“AI科学家”。此类系统必须在适合其目的的科学生态系统内开发和部署：机构必须为验证、问责、可解释性和双重用途安全进行重新设计。我们勾勒出多智能体架构如何通过原型框架 extit{Denario} 加速发现周期，并跨越人类无法触及的模型空间；考察这对作者身份、同行评审以及人类科学家持久角色的意义；最后提出将人工智能作为认知行为者而非单纯工具进行治理的建议。
cs.AI / 135 / 2606.22883
CLI-Universe: Towards Verifiable Task Synthesis Engine for Terminal Agents
CLI-Universe：面向可验证任务合成引擎的终端代理
Abstract
While recent LLM-based terminal agents have demonstrated promising capabilities, the scarcity of high-quality, executable training data remains a critical bottleneck. Existing synthesis pipelines typically scale by retrofitting surface-level artifacts into tasks, frequently yielding ambiguous instructions, shallow execution paths, and brittle tests that provide weak learning signals. To overcome this, we introduce CLI-Universe, a principled synthesis engine that constructs terminal-agent tasks. CLI-Universe generates candidate tasks by sampling combinations across a multi-dimensional capability taxonomy (domain, skill type, capability, and engineering pillar), then grounds each candidate through evidence-guided deep research over real-world technical materials. To ensure rigorous supervision, validated blueprints are instantiated into Dockerized environments and subjected to a multi-stage executable verification pipeline featuring rubric-gated test construction, hint-conditional filtering, and strict fail-to-pass checking. Across the full pipeline, from candidate generation to verification, approximately two-thirds of candidates are discarded, retaining only those that are genuine, verifiable, and non-trivially challenging. To validate our framework, we instantiate a highly distilled dataset of 6,000 trajectories called CLI-Universe-6K. Remarkably, fine-tuning Qwen3-32B on CLI-Universe-6K achieves 33.4% on Terminal-Bench 2.0. This sets a new state-of-the-art for models trained on open-source data at or below 32B parameters, and outperforms several models an order of magnitude larger, demonstrating the profound data efficiency of structured, high-fidelity synthesis.
Chinese Translation
尽管近期基于大规模语言模型（LLM）的终端代理展示了令人鼓舞的能力，但高质量、可执行训练数据的稀缺仍然是一个关键瓶颈。现有的合成流程通常通过将表面级工件改造为任务来扩展，常常导致模糊的指令、浅显的执行路径和脆弱的测试，这些都提供了微弱的学习信号。为了解决这一问题，我们提出了CLI-Universe，一个原则性合成引擎，用于构建终端代理任务。CLI-Universe通过在多维能力分类法（领域、技能类型、能力和工程支柱）中抽样组合生成候选任务，然后通过对现实世界技术材料进行证据引导的深度研究来验证每个候选任务。为了确保严格的监督，经过验证的蓝图被实例化到Docker化环境中，并经过多阶段可执行验证流程，包括基于评分标准的测试构建、条件提示过滤和严格的失败转通过检查。在整个流程中，从候选生成到验证，约三分之二的候选任务被舍弃，仅保留那些真实、可验证且具有非平凡挑战性的任务。为了验证我们的框架，我们实例化了一个高度精炼的数据集CLI-Universe-6K，包含6000条轨迹。值得注意的是，在CLI-Universe-6K上对Qwen3-32B进行微调，Terminal-Bench 2.0的得分达到了33.4%。这为在32B参数或以下的开源数据上训练的模型设定了新的最先进水平，并且超越了几个规模大一个数量级的模型，展示了结构化高保真合成的深刻数据效率。
cs.AI / 136 / 2606.22902
Agent-as-a-Router: Agentic Model Routing for Coding Tasks
代理作为路由器：编码任务的代理模型路由
Abstract
Real-world users typically have access to multiple Large Language Models (LLMs) from different providers, and these LLMs often excel at distinct domains, yet none dominate all. Consequently, routing each task to the most suitable model becomes critical for both performance and cost. Existing routers treat this as a static, one-off classification problem. However, we identify the performance bottleneck for these routers as information deficit: simply augmenting a vanilla LLM router with performance statistics at the task-dimension level yields a 15.3% relative gain, surpassing a heuristic router built on the same dimension-level priors. Motivated by this finding, we propose Agent-as-a-Router, a framework that formalizes routing as a C-A-F loop (Context->Action->Feedback->Context). It closes the information gap by accumulating execution-grounded experience during deployment. We instantiate this framework as ACRouter, composed of an Orchestrator, a Verifier, a Memory module, and introduce CodeRouterBench, an evaluation environment comprising ~10K task instances with verified scores from 8 frontier LLMs, enabling regret-based router comparison on streaming tasks. Experiments show that ACRouter achieves the lowest cumulative regret on in-distribution tasks and generalizes to out-of-distribution agentic-programming tasks, demonstrating that our routing framework actively closes the information gap. Codes and benchmarks are released at https://github.com/LanceZPF/agent-as-a-router.
Chinese Translation
现实世界的用户通常可以访问来自不同提供者的多个大型语言模型（LLMs），而这些LLMs在不同领域中往往表现出色，但没有一个模型能够在所有领域中占据主导地位。因此，将每个任务路由到最合适的模型对于性能和成本都变得至关重要。现有的路由器将此视为一个静态的、一锤子买卖的分类问题。然而，我们发现这些路由器的性能瓶颈在于信息缺失：仅仅在任务维度层面上用性能统计数据增强一个普通的LLM路由器，就能实现15.3%的相对增益，超越基于相同维度先验构建的启发式路由器。基于这一发现，我们提出了代理作为路由器（Agent-as-a-Router），一个将路由形式化为C-A-F循环（上下文->行动->反馈->上下文）的框架。它通过在部署过程中积累基于执行的经验来弥补信息差距。我们将该框架实例化为ACRouter，由一个协调器、一个验证器和一个内存模块组成，并引入CodeRouterBench，一个评估环境，包含约10K个任务实例和来自8个前沿LLMs的验证分数，使得在流式任务上能够进行基于遗憾的路由器比较。实验表明，ACRouter在分布内任务上实现了最低的累积遗憾，并且能够推广到分布外的代理编程任务，证明我们的路由框架积极弥补了信息差距。代码和基准测试已发布在 https://github.com/LanceZPF/agent-as-a-router。
cs.AI / 137 / 2606.22911
ThermoLLM: Thermodynamics-Aware HVAC Control with Spatial-Semantic Knowledge Graph
ThermoLLM：具有空间语义知识图的热力学感知HVAC控制
Abstract
Multi-zone HVAC control is a spatial decision problem in which indoor thermal evolution and control decisions depend not only on outdoor conditions and internal heat gains but also on zone layout, physical adjacency, and delayed thermal interactions across the building. Recent LLM-based HVAC controllers have shown that prompt-based control is feasible. However, these methods typically rely on task descriptions, observation values, short textual feedback, or unstructured retrieval, which limits their ability to reason about zone coupling, thermal response, and building dynamics. This paper presents a thermodynamics-aware LLM control framework for a five-zone EnergyPlus building simulation. The controller is grounded in a physics-informed spatial knowledge graph derived from Brick-style building semantics and linked with recent interaction history. At each control step, the model receives the current building state, graph-structured spatial context, and recent environment-controller history, enabling it to make decisions that reflect both building structure and short-term thermal evolution. We evaluate the framework against standard control baselines and several LLM-based alternatives. Results show that the proposed approach achieves the best overall energy-comfort trade-off and the lowest PMV violation while maintaining energy-efficient operation.
Chinese Translation
多区域HVAC控制是一个空间决策问题，其中室内热演变和控制决策不仅依赖于户外条件和内部热增益，还依赖于区域布局、物理邻近性以及建筑内的延迟热交互。最近基于大语言模型（LLM）的HVAC控制器已显示出基于提示的控制是可行的。然而，这些方法通常依赖于任务描述、观测值、简短文本反馈或非结构化检索，这限制了它们对区域耦合、热响应和建筑动态的推理能力。本文提出了一种热力学感知的LLM控制框架，适用于五区域的EnergyPlus建筑模拟。该控制器基于从Brick风格建筑语义中派生的物理知识图，并与最近的交互历史相连接。在每个控制步骤中，该模型接收当前建筑状态、图结构的空间上下文和最近的环境-控制器历史，从而使其能够做出反映建筑结构和短期热演变的决策。我们将该框架与标准控制基线和几种基于LLM的替代方案进行评估。结果表明，所提出的方法在整体能量-舒适度权衡方面表现最佳，并且PMV（预测平均投票）违规率最低，同时保持能效运行。
cs.AI / 138 / 2606.22916
Intent-Governed Tool Authorization for AI Agents
基于意图的人工智能代理工具授权
Abstract
AI agents increasingly act through external tools: they read private data, construct structured payloads, submit write requests, export records, and coordinate workflows across application boundaries. Existing authorization mechanisms usually ask whether an integration credential, app, or token can call a tool. That question is necessary but incomplete. A tool call can be authorized by static credentials and still be unjustified by the user's current request. For example, a credential that can read and export records should not expose export authority when the user only asked for a bounded summary, and a model-generated delete call should not execute merely because the integration has a delete scope. This paper proposes Intent-Governed Access Control (IGAC), a server-side authorization layer that treats the user's expressed intent as a monotone, auditable policy attribute for AI-agent tool use. IGAC introduces intent certificates, session-scoped policy narrowing, intent-aware manifest filtering, and intent-tool-payload consistency checks. The central invariant is that user intent may only reduce the authority granted by static integration policy; it never expands scopes, data policy, tenant boundaries, or review requirements. We map IGAC onto OpenPort, an existing governance substrate that already implements authorization-dependent discovery, scope and ABAC-style policy checks, draft-first writes, preflight impact binding, state-witness checks, idempotency, stable reason codes, and audit.
Chinese Translation
人工智能代理越来越多地通过外部工具进行操作：它们读取私有数据，构建结构化负载，提交写请求，导出记录，并跨应用边界协调工作流。现有的授权机制通常询问某个集成凭证、应用程序或令牌是否可以调用某个工具。这个问题虽然必要，但并不完整。工具调用可以通过静态凭证获得授权，但仍然可能不符合用户当前请求的合理性。例如，能够读取和导出记录的凭证不应在用户仅请求有限摘要时暴露导出权限，而模型生成的删除调用仅因为集成具有删除范围而执行是不可接受的。本文提出了基于意图的访问控制（Intent-Governed Access Control, IGAC），这是一种服务器端授权层，将用户表达的意图视为人工智能代理工具使用的单调可审计政策属性。IGAC引入了意图证书、会话范围的政策缩小、意图感知的清单过滤以及意图-工具-负载一致性检查。其核心不变性是用户意图只能减少静态集成政策授予的权限，而绝不会扩展范围、数据政策、租户边界或审查要求。我们将IGAC映射到OpenPort，一个现有的治理基础设施，已经实现了依赖授权的发现、范围和基于属性的访问控制（ABAC）政策检查、草稿优先写入、预检影响绑定、状态见证检查、幂等性、稳定的原因代码和审计。
cs.AI / 139 / 2606.22936
When Agents Commit Too Soon: Diagnosing Premature Commitment in LLM Agents
当智能体过早承诺时：诊断大型语言模型智能体的过早承诺
Abstract
Long-horizon LLM agents can fail quietly: they settle on one reading of the evidence early, then spend the rest of the run defending it. We call this premature commitment. Final-answer scoring misses the failure mode because it sees only the answer, not whether the process has already collapsed to a stable path. We define representational commitment as cross-run hidden-state convergence at a fixed reasoning step, and use it as an early diagnostic of trajectory consistency. On Llama-3.1-70B running ReAct on HotpotQA, step-4 hidden-state similarity predicts downstream behavioral consistency (r = -0.35, partial r = -0.45), with a localized temporal and layer-wise signature. The signal replicates across Qwen-2.5-72B and Phi-3-14B, and on StrategyQA (r = -0.83). It does not track correctness: committed-wrong and committed-correct questions are not separable in activation similarity. That boundary is central to the claim. Commitment tells us whether an agent has settled, not whether it is right. A runtime monitor detects inconsistent trajectories from hidden states at AUROC up to 0.97 (0.85--0.88 under a stricter split), and a prompting intervention cuts behavioral variance by 28% against a token-matched control while leaving accuracy statistically unchanged. We also test whether the signal can route self-consistency compute; on a harder benchmark it helps only modestly and is matched by a simpler output-based baseline. The result is a diagnostic for a hidden process failure, with clear limits rather than a general accuracy lever.
Chinese Translation
长时间跨度的大型语言模型（LLM）智能体可能会悄然失败：它们早早地对证据的某一解读做出决定，然后在接下来的运行中一直在为此辩护。我们称之为过早承诺。最终答案评分无法捕捉这种失败模式，因为它只关注答案，而不关注过程是否已经崩溃到一个稳定的路径。我们将表征承诺定义为在固定推理步骤下跨运行的隐藏状态收敛，并将其作为轨迹一致性的早期诊断指标。在使用 ReAct 运行 HotpotQA 的 Llama-3.1-70B 上，步骤 4 的隐藏状态相似性预测下游行为一致性（r = -0.35，部分 r = -0.45），具有局部的时间和层次特征。该信号在 Qwen-2.5-72B 和 Phi-3-14B 上得到了复制，并在 StrategyQA 上（r = -0.83）。它与正确性无关：承诺错误和承诺正确的问题在激活相似性中不可分离。这个边界是我们主张的核心。承诺告诉我们一个智能体是否已经定型，而不是它是否正确。运行时监控可以从隐藏状态中检测到不一致的轨迹，AUROC 可达 0.97（在更严格的划分下为 0.85--0.88），而提示干预在与令牌匹配的对照组相比将行为方差降低了 28%，同时使准确性在统计上保持不变。我们还测试了该信号是否可以引导自一致性计算；在一个更困难的基准上，它的帮助仅为适度，并且与一个更简单的基于输出的基线相匹配。这个结果是对隐藏过程失败的诊断，具有明确的限制，而不是一个通用的准确性杠杆。
cs.AI / 140 / 2606.22948
ENVS: Environment-Native Verified Search for Long-Horizon GUI Agents
ENVS：环境原生的长时间跨度GUI代理验证搜索
Abstract
As multimodal agents move from interface understanding to real software control, successful trajectory discovery in live desktop environments becomes a key challenge. GUI tasks require long-horizon sequences of precise mouse and keyboard actions, while feedback is sparse, delayed, and costly to obtain through VM rollouts. We propose Environment-Native Verified Search (ENVS), a training-time search-and-filter pipeline that uses the environment to construct verified supervision before policy optimization: it branches over behaviorally distinct GUI actions in live OSWorld VMs, verifies successful leaves, and trains from globally balanced step-level supervision. To evaluate robustness under realistic desktop interruptions, we also introduce OSWorld-Noisy, a dynamic benchmark for recoverable desktop interruptions that preserves the original tasks while testing whether agents can refocus, dismiss, wait, or recover under live perturbations. On the 300-task OSWorld pool, ENVS reaches 30.3 pass@8 on original evaluations and 29.0 on OSWorld-Noisy, outperforming matched ARPO-style online RL while reducing compute from 184-192 to 138-153 GPU-hours; even with only 30% of its search data, ENVS reaches 27.0 pass@8, exceeding ARPO from the base model. Training from noisy environments also better preserves visual-reasoning abilities on auxiliary benchmarks, including OSWorld-G Refusal (16.7 vs. 1.9) and BLINK Functional Correspondence (26.2 vs. 23.1).
Chinese Translation
随着多模态代理从界面理解转向真实软件控制，在实时桌面环境中成功发现轨迹成为一项关键挑战。GUI任务需要长时间跨度的精确鼠标和键盘操作序列，而反馈稀疏、延迟且通过虚拟机（VM）回放获取的成本较高。我们提出了环境原生验证搜索（ENVS），这是一种训练时的搜索和过滤管道，利用环境构建验证监督，然后进行策略优化：它在实时的OSWorld虚拟机中对行为上不同的GUI操作进行分支，验证成功的叶节点，并从全球平衡的步骤级监督中进行训练。为了评估在现实桌面中断下的鲁棒性，我们还引入了OSWorld-Noisy，这是一个动态基准，旨在测试可恢复的桌面中断，保持原始任务的同时测试代理是否能够在实时干扰下重新聚焦、拒绝、等待或恢复。在300个任务的OSWorld池中，ENVS在原始评估中达到30.3的pass@8，在OSWorld-Noisy中达到29.0，超越了匹配的ARPO风格在线强化学习，同时将计算时间从184-192 GPU小时减少到138-153 GPU小时；即使仅使用30%的搜索数据，ENVS也达到了27.0的pass@8，超越了基础模型的ARPO。从嘈杂环境中训练也更好地保留了在辅助基准上的视觉推理能力，包括OSWorld-G拒绝（16.7对1.9）和BLINK功能对应（26.2对23.1）。
cs.AI / 141 / 2606.22953
Plans Don't Persist: Why Context Management Is Load Bearing for LLM Agents
计划不会持久：为什么上下文管理对大型语言模型代理至关重要
Abstract
Long-horizon agents depend on context management: systems compress, summarize, and evict old tokens so tasks can continue beyond finite windows. That is safe only when dropped information is no longer needed or has been internalized. Plans are the stress case: they are written early, used for many steps, and first to be evicted. We introduce replay pairing, a diagnostic that runs the same trajectory with and without the plan in history and measures hidden-state cosine distance. On Llama-3.1-70B, plan signal spikes to 0.453 one step after the plan, then falls 4.1x in a single action-observation step; HotpotQA falls 12.4x. This is evidence that standard LLM agents do not carry plans forward as persistent state, and instead depend on the plan remaining in context. A layer-L32 probe detects this decay as a diagnostic, not as proof that it reads plan content itself. Reasoning models add a measurement confound: their `` traces re-derive plan content, so standard stripping leaves plan evidence in the stripped condition. We name this the reasoning-trace confound and fix it with strict stripping, which removes prior `` blocks from the stripped run only. It recovers +163% of the step+1 signal in-sample and +153% held out, while not meaningfully changing non-reasoning Llama (+4.8%). On DeepSeek-R1-Distill-Llama-70B, a Llama-trained probe transfers at AUROC 0.748 (p=6e-4), while R1-specific probes reach 1.000, suggesting R1 encodes plan signal in a different hidden-state direction. Finally, a compression stress test shows the practical cost: naive plan eviction cuts ALFWorld success by 34.7pp, while probe-gated re-surfacing does not recover it. The contribution is a measurement and stress-test framework showing that agent-critical information can be context-resident rather than persistent. Context management is load bearing, but plan protection alone is not enough.
Chinese Translation
长时间跨度的代理依赖于上下文管理：系统压缩、总结并驱逐旧的标记，以便任务能够超越有限的窗口继续进行。这只有在丢弃的信息不再需要或已被内化时才是安全的。计划是一个压力案例：它们在早期编写，使用多个步骤，并且是第一个被驱逐的内容。我们引入了重放配对（replay pairing），这是一种诊断方法，运行相同的轨迹，分别在历史中包含和不包含计划，并测量隐藏状态的余弦距离。在 Llama-3.1-70B 上，计划信号在计划后一个步骤激增至 0.453，然后在一个动作-观察步骤中下降 4.1 倍；HotpotQA 降低了 12.4 倍。这表明标准的大型语言模型代理并未将计划作为持久状态继续保留，而是依赖于计划仍然在上下文中。一个层级 L32 的探测器将这种衰减检测为诊断，而不是证明它本身读取计划内容。推理模型增加了测量混淆：它们的 `` 轨迹重新推导出计划内容，因此标准剥离在剥离条件中留下了计划证据。我们将其称为推理轨迹混淆，并通过严格剥离来修复它，仅从剥离运行中移除先前的 `` 块。它在样本中恢复了 +163% 的步骤+1 信号，持出时恢复了 +153%，而对非推理的 Llama 没有显著变化（+4.8%）。在 DeepSeek-R1-Distill-Llama-70B 上，经过 Llama 训练的探测器以 AUROC 0.748 转移（p=6e-4），而 R1 特定探测器达到 1.000，这表明 R1 在不同的隐藏状态方向上编码了计划信号。最后，一项压缩压力测试显示了实际成本：天真的计划驱逐使 ALFWorld 的成功率下降 34.7 个百分点，而探测器门控的重新浮现并未恢复成功率。该研究的贡献是一个测量和压力测试框架，表明代理关键的信息可以是上下文驻留的，而不是持久的。上下文管理是负载承载的，但仅靠计划保护是不够的。
cs.AI / 142 / 2606.22959
The Impact of VAE Design on Latent Pose Representations for Diffusion-based Sign Language Production
变分自编码器设计对基于扩散的手语生成中潜在姿态表示的影响
Abstract
Latent diffusion approaches to sign language production (SLP) rely on an initial stage that learns an encoding of sign pose sequences, enabling generative modeling in the resulting latent space. The autoencoder used in this stage is typically evaluated in terms of reconstruction quality using geometric metrics common in SLP. While informative, these metrics do not fully capture latent space properties that may influence the training and performance of the downstream generative model. In this work, we investigate how architectural and training objective design choices in a variational autoencoder (VAE) for sign pose encoding affect latent space structure, and how these differences translate into the performance of a latent diffusion model for text-to-sign generation. Our experiments on Phoenix14T dataset show that variations in generative performance, measured through back-translation BLEU scores, can sometimes be better explained by differences in latent space properties than by VAE reconstruction accuracy alone.
Chinese Translation
基于潜在扩散的方法用于手语生成（SLP）依赖于一个初始阶段，该阶段学习手势姿态序列的编码，从而在生成的潜在空间中实现生成建模。在这一阶段使用的自编码器通常通过在手语生成中常用的几何度量来评估重建质量。尽管这些度量提供了有价值的信息，但并未完全捕捉可能影响下游生成模型训练和性能的潜在空间特性。在本研究中，我们探讨了变分自编码器（VAE）在手势姿态编码中的架构和训练目标设计选择如何影响潜在空间结构，以及这些差异如何转化为文本到手势生成的潜在扩散模型的性能。我们在Phoenix14T数据集上的实验表明，通过反向翻译的BLEU分数衡量的生成性能的变化，有时可以通过潜在空间特性的差异来更好地解释，而不仅仅是VAE重建准确性。
cs.AI / 143 / 2606.22974
When Preferences Fail to Become Incentives: A Utility-Behavior Gap in Large Language Models
当偏好未能转化为激励：大型语言模型中的效用-行为差距
Abstract
Recent work on preference elicitation in large language models (LLMs) has demonstrated that, when given a series of choices between two outcomes, LLMs reveal a coherent, model-specific utility structure. Notably, this structure often includes preferences that the models' trainers did not intend, such as valuing people of some nationalities above others, raising the possibility that LLMs might be forming emergent, misaligned goals, which, if true, would have major safety implications. However, the choice paradigms in which these preferences are observed are not reflective of real-world situations in which misaligned behavior would be a practical concern. Therefore, we design an experimental paradigm to probe whether these preferences serve as motivations for LLM behavior in realistic scenarios. First, we reproduce prior findings on consistent preference elicitation. Next, we create a set of common writing tasks - essays, grant proposal abstracts, incident postmortems, and translations - where quality can be assessed by a blind, independent LLM judge panel. Then, we demonstrate that LLMs can be motivated via direct exhortation and other explicit cues to modulate their output quality on these tasks. Finally, we probe whether utilities inferred from explicitly reported preferences can shift output quality on these tasks by offering LLMs high-utility incentives for high-quality outputs. In all tasks, across all models tested, offering LLMs outcomes that they report in the choice paradigm as being highly preferred does not lead them to create higher quality outputs than offering them dispreferred outcomes, or even no outcomes at all. We conclude that the existence of coherent preferences as demonstrated in choice paradigms should not be taken as evidence that those preferences have incentive value for the models or affect their behavior in other contexts.
Chinese Translation
近期关于大型语言模型（LLMs）中偏好引导的研究表明，当给定一系列在两种结果之间的选择时，LLMs展现出一致的、特定于模型的效用结构。值得注意的是，这种结构通常包含模型训练者未曾预期的偏好，例如对某些国籍的人给予高于其他国籍的价值，这引发了LLMs可能形成新兴的、不一致的目标的可能性。如果这一点属实，将对安全性产生重大影响。然而，这些偏好被观察到的选择范式并不反映现实世界中不一致行为可能带来的实际问题。因此，我们设计了一个实验范式，以探讨这些偏好是否在现实场景中作为LLM行为的动机。首先，我们重现了关于一致性偏好引导的先前发现。接下来，我们创建了一组常见的写作任务——论文、资助提案摘要、事件后评估和翻译——其质量可以通过盲评的独立LLM评审小组进行评估。然后，我们展示了通过直接劝告和其他明确提示，LLMs可以被激励以调节其在这些任务上的输出质量。最后，我们探讨了从明确报告的偏好中推断出的效用是否能够通过为高质量输出提供高效用激励来改变这些任务的输出质量。在所有任务中，针对所有测试的模型，提供LLMs在选择范式中报告为高度偏好的结果，并未导致其创造出比提供不偏好的结果甚至没有结果时更高质量的输出。我们得出结论，选择范式中所展示的一致性偏好的存在不应被视为这些偏好对模型具有激励价值或在其他情境中影响其行为的证据。
cs.AI / 144 / 2606.22978
Joint Air Traffic Flow and Capacity Management via Answer Set Programming
通过答案集编程实现联合空中交通流量与容量管理
Abstract
Operational Air Traffic Flow and Capacity Management (ATFCM) balances flight demand with available sector capacity, to ensure safe and efficient operations. Mathematical models enhance operational ATFCM performance by framing demand-capacity balancing as an optimization problem, maximizing efficiency while adhering to safety constraints. However, SOTA research optimizes the aircraft trajectories (called ATFM) or the sector configuration (called DAC) separately. This leaves a research gap of whether joint optimization of ATFM and DAC can bring benefits. We partially address this limitation by introducing a joint ATFCM model with an encoding in Answer Set Programming (ASP). The ASP implementation is evaluated against two baselines applied to our joint model: a SOTA Mixed Integer Programming (MIP) model and an iterative CASA-based heuristic. Computational experiments utilize an instance generator fitted to historical OpenSky Network flight data. Our results indicate that the ASP model outperforms the MIP model, while ASP remains competitive against heuristics on small instances. Furthermore, while DAC has the largest improvement on solving performance compared to rerouting and delaying, unrestricted variants of DAC or rerouting lead to search space thrashing.
Chinese Translation
运营空中交通流量与容量管理（ATFCM）旨在平衡航班需求与可用的航段容量，以确保安全和高效的运营。数学模型通过将需求与容量的平衡框架化为优化问题，提升运营ATFCM的性能，最大化效率的同时遵循安全约束。然而，当前的前沿研究分别优化飞机轨迹（称为ATFM）或航段配置（称为DAC），这留下了一个研究空白，即ATFM和DAC的联合优化是否能带来益处。我们通过引入一个联合ATFCM模型，并采用答案集编程（ASP）进行编码，部分解决了这一局限性。ASP实现与应用于我们联合模型的两个基准进行评估：一个前沿的混合整数规划（MIP）模型和一个基于迭代CASA的启发式算法。计算实验利用了一个适配于历史OpenSky Network航班数据的实例生成器。我们的结果表明，ASP模型在性能上优于MIP模型，而在小规模实例中，ASP仍然与启发式算法具有竞争力。此外，虽然DAC在求解性能上相较于重新规划和延误有最大的提升，但DAC的无限制变体或重新规划会导致搜索空间的剧烈波动。
cs.AI / 145 / 2606.23026
A Stackelberg Framework for Resource-Aware LLM Agents: Learning, Repair, and Conditional Guarantees
面向资源感知的大型语言模型代理的斯塔克尔博格框架：学习、修复与条件保证
Abstract
Large language model (LLM) agents increasingly operate as multi-turn systems that must allocate context, prompt verbosity, and tool access under finite computational budgets. Static thresholds are simple, but they are brittle under heterogeneous tasks and evolving session states. We formulate resource governance as a contextual Stackelberg game: a controller commits to a quality target and a cost incentive, while an executor responds with resource actions over context, prompting, and tool usage. We learn a conditional response model, optimize a leader policy against that model, and repair the resulting policy using real-API calibration and projection onto an empirically selected action set. For the restricted game, we establish conditional guarantees for equilibrium existence, follower-response stability, safe-set projection, and transfer from a surrogate environment to the real environment under bounded value error. The primary real-API experiment comprises 300 evaluated turns. Relative to a conservative baseline, the selected repaired controller reduces mean token cost by 17.4% (Welch $p=0.022$), while the measured quality difference is not statistically significant ($p=0.44$). The theoretical results are conditional and the experiments do not estimate their regret or transfer constants; consequently, the evidence establishes a promising repaired operating point, not a certified real-system equilibrium.
Chinese Translation
大型语言模型（LLM）代理越来越多地作为多轮系统运行，必须在有限的计算预算下分配上下文、提示冗长性和工具访问。静态阈值虽然简单，但在异构任务和不断变化的会话状态下表现脆弱。我们将资源治理形式化为一个上下文斯塔克尔博格博弈：控制者承诺一个质量目标和成本激励，而执行者则根据上下文、提示和工具使用作出资源行动的响应。我们学习一个条件响应模型，针对该模型优化领导者策略，并使用真实API校准和投影到经验选择的行动集上修复所得到的策略。对于限制性博弈，我们建立了平衡存在、跟随者响应稳定性、安全集投影以及在有界值误差下从替代环境到真实环境的转移的条件保证。主要的真实API实验包含300个评估轮次。相较于保守基线，所选修复控制器将平均标记成本降低了17.4%（Welch $p=0.022$），而测得的质量差异在统计上并不显著（$p=0.44$）。理论结果是有条件的，实验并未估计其遗憾或转移常数；因此，证据确立了一个有前景的修复操作点，而非经过认证的真实系统平衡。
cs.AI / 146 / 2606.23029
From numerical proportions to analogical proportions between probabilities
从数值比例到概率之间的类比比例
Abstract
Analogical proportions link four items a, b, c, d by a relation stating that ``a is to b as c is to d", a, b, c, d being the formal representation of real world entities, ranging from simple numerical values to more complex structures such as profiles. Accordingly, $a, b, c, d$ could be atomic values like Boolean, nominal or numerical values, more generally vectors of such values, or even families of items represented by logical formulas. In this paper, we consider another representation setting, which is the probabilistic one. Precisely, the article proposes a study of {analogical} proportions between probabilities, whether they are simply between probability values, or between distributions (which requires the preservation of their normalization). More particularly, we study the properties of definitions based on arithmetic proportion, or on a combination of the former with geometric proportion, while other options are also discussed. Previous works have shown that when four profiles a, b, c, d, represented as vectors, form analogical proportions componentwise, it is likely that their classes form an analogical proportion also. This is the basis of an analogical proportion-based classification method that can produce accurate predictions. Similarly, in this paper, each profile is associated with a distribution describing the frequencies of the possible values of a discrete attribute of interest. We then discuss and experimentally investigate if the distributions associated to four profiles forming an analogical proportion themselves also form an analogical proportion.
Chinese Translation
类比比例通过一种关系将四个项目 a、b、c、d 联系起来，表述为“a 对 b 的关系如同 c 对 d 的关系”，其中 a、b、c、d 是现实世界实体的形式表示，范围从简单的数值到更复杂的结构，如个人资料。因此，a、b、c、d 可以是原子值，如布尔值、名义值或数值，更一般地可以是这些值的向量，甚至是由逻辑公式表示的项目集合。在本文中，我们考虑另一种表示设置，即概率设置。具体而言，本文提出对概率之间的类比比例进行研究，无论它们是简单的概率值之间的比例，还是分布之间的比例（这要求保持其归一化）。更具体地，我们研究基于算术比例的定义的性质，或将前者与几何比例结合的定义，同时也讨论其他选项。先前的研究表明，当四个作为向量表示的个人资料 a、b、c、d 在分量上形成类比比例时，它们的类别也很可能形成类比比例。这是基于类比比例的分类方法的基础，该方法可以产生准确的预测。同样，在本文中，每个个人资料与描述感兴趣的离散属性的可能值频率的分布相关联。然后，我们讨论并实验性地研究与形成类比比例的四个个人资料相关联的分布是否也形成类比比例。
cs.AI / 147 / 2606.23032
IPO Finance Agent: Evaluation of LLM Financial Analysts beyond Finance Agent v2, with Automated Rubric Generation -- the Case of the SpaceX (SPCX) IPO
IPO金融代理：超越金融代理v2的LLM金融分析师评估，带有自动化评分标准生成——以SpaceX（SPCX）IPO为例
Abstract
Finance Agent v2 (by Vals AI) has emerged as the reference benchmark for evaluating both Anthropic Claude and OpenAI ChatGPT frontier language models on financial tasks. However, it narrowly deals with periodic reporting from publicly traded companies (SEC 10-K and 10-Q filings), and its agentic harness relies on naive, unenriched chunk retrieval. Neither the task design nor the retrieval approach addresses the distinct challenges of IPO due diligence. SEC S-1 filings combine historical financial statements, governance structures, pro forma and common-control accounting treatments, capital-formation narratives, and underwriting-sensitive risk disclosures within substantially longer documents than typical periodic filings. That is why we introduce IPO Finance Agent, which extends the Finance Agent v2 framework along two directions: task domain and retrieval architecture. During our experiments, the original Finance Agent v2 harness basically failed to deliver any output related to the SpaceX S-1 filing, due to document length. We therefore had to improve the agentic harness with contextual retrieval, a more realistic and industry-standard approach for long documents. We also built a dataset of 1,000 IPO-diligence questions, and publicly release 70 questions on the SpaceX (SPCX) S-1 filing to support reproducibility, while the remainder are held private to guard against benchmark contamination. In addition, we introduce an evaluator-optimizer pipeline to automatically generate evaluation rubrics for the benchmark: candidate facts are extracted from an ensemble of independently-generated model answers to each question, consolidated into draft criteria, then automatically audited for omissions, hallucinations, mistiered items, and redundancy, with LLM feedback driving iterative repair, targeted enrichment, and deduplication. Human experts only review final rubrics before deployment. Results show that the best-performing evaluated model, Alibaba Qwen 3.7 Max, reaches 79.4% accuracy at $0.30 per query, and the most cost-efficient model on the resulting Pareto frontier, Xiaomi MiMo-2.5 Pro, reaches slightly lower accuracy (76.8%) at $0.05 per query. Both exceed the current Finance Agent v2 leaderboard ceiling-Google Gemini 3.5 Flash at 57.9% for $2.51 per querywhile undercutting even FABv2's cheapest entry (MiniMax M3: 48.3% at $0.32) on cost-efficiency. Code and data are released on GitHub: https://github.com/benstaf/ipoagent
Chinese Translation
金融代理v2（由Vals AI开发）已成为评估Anthropic Claude和OpenAI ChatGPT前沿语言模型在金融任务上表现的参考基准。然而，它仅限于处理上市公司（SEC 10-K和10-Q文件）的定期报告，其代理机制依赖于简单、未丰富的片段检索。无论是任务设计还是检索方法都未能应对IPO尽职调查的独特挑战。SEC S-1文件结合了历史财务报表、治理结构、临时和共同控制会计处理、资本形成叙述以及在比典型定期文件更长的文档中披露的承销敏感风险信息。因此，我们引入了IPO金融代理，它在任务领域和检索架构两个方向上扩展了金融代理v2框架。在我们的实验中，原始的金融代理v2由于文档长度基本未能提供与SpaceX S-1文件相关的任何输出。因此，我们不得不通过上下文检索来改进代理机制，这是一种更现实且符合行业标准的长文档处理方法。我们还构建了一个包含1000个IPO尽职调查问题的数据集，并公开发布了70个关于SpaceX（SPCX）S-1文件的问题，以支持可重复性，其余问题则保留为私有，以防基准污染。此外，我们引入了一个评估-优化管道，以自动生成基准的评估评分标准：候选事实从每个问题的独立生成模型答案的集合中提取，合并为草拟标准，然后自动审核遗漏、幻觉、错误分类项和冗余，利用LLM反馈驱动迭代修复、针对性丰富和去重。人类专家仅在最终评分标准部署前进行审核。结果显示，表现最佳的评估模型Alibaba Qwen 3.7 Max在每个查询$0.30的成本下达到了79.4%的准确率，而在生成的帕累托前沿上，成本效益最高的模型Xiaomi MiMo-2.5 Pro的准确率略低（76.8%），每个查询$0.05。两者均超过当前金融代理v2排行榜的最高分——Google Gemini 3.5 Flash在每个查询$2.51时的57.9%，并在成本效益上甚至低于FABv2的最便宜条目（MiniMax M3：48.3%在$0.32）。代码和数据已在GitHub上发布：https://github.com/benstaf/ipoagent
cs.AI / 148 / 2606.23055
Some Results about the Expressivity of Preference-Incomplete Structured Argumentation Frameworks
关于偏好不完全的结构化论证框架的表达能力的一些结果
Abstract
This paper studies the expressive power of ASPIC$^+$ argumentation frameworks with uncertain preference profiles by comparing them with several abstract formalisms with uncertain defeats. Most of our results are negative (and some of them are theoretically unexpected). We also conjecture a positive, non-trivial threshold for the expressivity of uncertain preferences, and prove some essential preliminary steps toward the confirmation of this conjecture.
Chinese Translation
本文研究了具有不确定偏好特征的ASPIC$^+$论证框架的表达能力，并将其与几种具有不确定击败的抽象形式进行比较。我们的结果大多数是负面的（其中一些在理论上是意外的）。我们还推测了一个积极的、非平凡的阈值，用于不确定偏好的表达能力，并证明了确认这一推测的一些基本初步步骤。
cs.AI / 149 / 2606.23094
Cognitive Digital Twins: Ethical Risks and Governance for AI Systems That Model the Mind
认知数字双胞胎：建模思维的人工智能系统的伦理风险与治理
Abstract
As AI systems become increasingly persistent and personalized, they make possible a class of technologies that we call cognitive digital twins (CDTs): dynamic computational representations of a specific person's cognition, updated from behavioral, contextual, or physiological data in order to model, predict, or simulate that person's cognition, or to act as that person's communicative or decision-making proxy. CDTs combine cognitive inference with longitudinal representation, simulation, and proxy action in ways that existing governance strategies for personal assistants, autonomous agents, recommender systems, and automated decision systems only partially address. This paper makes four contributions. First, we define CDTs and distinguish them from adjacent systems. Second, we introduce a 5A governance framework organized around authority, autonomy, access and control, accountability, and availability. Third, we identify CDT-specific risks, from misrepresentation and epistemic authority shifts to shadow twins, simulated participation, proxy action, and proxy-power asymmetries. Fourth, we analyze governance gaps and propose requirements for high-risk CDTs that strengthen consent, purpose limitation, validity, traceability, contestation, independent review, and model retirement. Existing frameworks primarily regulate data processing, automated decisions, or autonomous actions; CDTs also require governance at the level of cognitive representation itself, before any final decision or external action occurs. We argue that CDTs require governance not only because they can act for people, but because they can become infrastructures through which cognition is represented, simulated, classified, and operationalized.
Chinese Translation
随着人工智能系统日益持久和个性化，它们使得一种我们称之为认知数字双胞胎（Cognitive Digital Twins, CDTs）的技术成为可能：这是对特定个体认知的动态计算表示，基于行为、情境或生理数据进行更新，以建模、预测或模拟该个体的认知，或作为该个体的沟通或决策代理。CDTs将认知推理与纵向表示、模拟和代理行为结合在一起，而现有的个人助手、自治代理、推荐系统和自动决策系统的治理策略仅部分解决了这些问题。本文做出了四个贡献。首先，我们定义了CDTs，并将其与相关系统区分开来。其次，我们提出了一个围绕权威、自治、访问与控制、问责和可用性组织的5A治理框架。第三，我们识别了特定于CDT的风险，从误表征和认识论权威转移到影子双胞胎、模拟参与、代理行为和代理权力不对称。第四，我们分析了治理缺口，并提出了针对高风险CDTs的要求，以加强同意、目的限制、有效性、可追溯性、争议、独立审查和模型退役。现有框架主要规范数据处理、自动决策或自主行动；而CDTs还需要在认知表示层面进行治理，在任何最终决策或外部行动发生之前。我们认为，CDTs需要治理不仅因为它们可以代表人们行动，还因为它们可以成为认知被表示、模拟、分类和操作化的基础设施。
cs.AI / 150 / 2606.23108
TTFT-Aware Graph Chain-of-Thought:Distance-Indexed Neural A* for Low-Hallucination Multi-Hop Medical Reasoning
TTFT感知的图链式思维：低幻觉多跳医疗推理的距离索引神经A*
Abstract
Hallucinations and opaque reasoning remain unacceptable failure modes for clinical LLMs. We present a production-grade GraphRAG stack that constrains answers to verifiable graph chain-of-thought paths in a heterogeneous, ~700K-node medical knowledge graph powering a fertility assistant. The core idea is targeted navigation: a directed Pruned Landmark Labeling (PLL) oracle provides exact distances for sub-millisecond feasibility checks and simple-path enumeration, while a lightweight AStarNet heuristic operates strictly within the PLL corridor to prioritize clinically plausible expansions. We score and pack a small, diverse set of paths (CUI/semantic-type overlap, length prior, provenance priors) to condition generation, yielding compact prompts and improved Time to First Token (TTFT). On fertility-focused queries, the hybrid (PLL+AStarNet) establishes a better latency/recall Pareto frontier than text-only RAG and single-component baselines, lowers TTFT, and reduces clinician-audited hallucinations while preserving explanation clarity. The result is a practical recipe for explainable, low-hallucination multi-hop medical reasoning ready for real-world deployment.
Chinese Translation
幻觉和不透明的推理仍然是临床大型语言模型（LLMs）不可接受的失败模式。我们提出了一种生产级的GraphRAG堆栈，该堆栈将答案限制在可验证的图链式思维路径上，基于一个异构的约70万个节点的医疗知识图谱，为生育助手提供支持。核心思想是有针对性的导航：一个定向的修剪地标标记（Pruned Landmark Labeling, PLL）oracle提供精确的距离，以进行亚毫秒的可行性检查和简单路径枚举，而一个轻量级的AStarNet启发式算法严格在PLL走廊内操作，以优先考虑临床上合理的扩展。我们对一小组多样化的路径（CUI/语义类型重叠、长度先验、来源先验）进行评分和打包，以调节生成，产生紧凑的提示并改善首次令牌时间（Time to First Token, TTFT）。在以生育为重点的查询中，混合模型（PLL+AStarNet）建立了比仅文本的RAG和单一组件基线更好的延迟/召回帕累托前沿，降低了TTFT，并减少了临床医生审核的幻觉，同时保持了解释的清晰性。最终结果是一个实用的配方，适用于可解释的、低幻觉的多跳医疗推理，准备好进行实际部署。
cs.AI / 151 / 2606.23122
A Matter of Time: Towards a General Theory of Agency
时间问题：迈向一种普遍的能动性理论
Abstract
Agency is often invoked in research on philosophy, biology, and cognitive science without a clear account of how it originates from material organization. Building on temporally parametrized (F, A)-systems, this paper develops a graded organizational theory of agency grounded in relational biology, physical biosemiotics, and process ontology. We argue that self-referential closure cannot be adequately conceived outside time: once the constitutive processes of a semantically closed organization are associated with distinct characteristic timescales, the organization unfolds into an out-of-sync dependency structure that can be formally redescribed as a history-dependent, revisable Asynchronous Dynamic Bayesian Network. This move allows for a principled distinction between autonomy, goal-directedness, agency, and open-endedness. Autonomy arises from precarious closure to efficient causation under material openness; goal-directedness from the maintenance of viability-supporting organization; agency appears when such organization acquires an endogenous anticipatory structure that selectively modulates organism-environment coupling in light of possible futures; open-endedness begins when this anticipatory organization can reconstruct its own future space of possibilities. Our framework reconciles Rosennean anticipation with organizational closure, restricts Markov blankets and active inference to derived formal redescriptions rather than first principles, and reinterprets computational enactivism in non-Fristonian terms. By deriving weaker temporalized organizations, our contribution outlines a hierarchy from proto-agential chemical systems to fully semantically closed agents, with implications for multicellular organisms, synthetic lifeforms, and neuroscience.
Chinese Translation
能动性在哲学、生物学和认知科学的研究中经常被提及，但对其如何源于物质组织却缺乏清晰的解释。基于时间参数化的(F, A)-系统，本文发展了一种基于关系生物学、物理生物符号学和过程本体论的分级组织能动性理论。我们认为，自我指涉的闭合无法在时间之外得到充分理解：一旦语义闭合组织的构成过程与不同的特征时间尺度相关联，该组织便展开为一种不同步的依赖结构，可以形式上重新描述为历史依赖的、可修正的异步动态贝叶斯网络。这一转变使得自主性、目标导向性、能动性和开放性之间能够进行原则上的区分。自主性源于在物质开放下对有效因果关系的脆弱闭合；目标导向性源于维持支持生存的组织；当这种组织获得内生的预期结构，选择性地调节生物体与环境的耦合以应对可能的未来时，能动性便出现；而开放性则始于这种预期组织能够重构自身未来的可能性空间。我们的框架调和了罗森尼（Rosennean）预期与组织闭合，将马尔可夫毯和主动推理限制为派生的形式重描述，而非第一原理，并以非弗里斯顿（Friston）术语重新解释计算性生动主义。通过推导较弱的时间化组织，我们的贡献描绘了从原始能动化化学系统到完全语义闭合的代理的层次结构，这对多细胞生物、合成生命形式和神经科学具有重要意义。
cs.AI / 152 / 2606.23127
Managing Procedural Memory in LLM Agents: Control, Adaptation, and Evaluation
管理大型语言模型代理的程序性记忆：控制、适应与评估
Abstract
Procedural memory is increasingly used to improve LLM agents on recurring workplace tasks, yet its ability to produce reusable skills remains poorly understood. We introduce AFTER, a benchmark of 382 realistic enterprise tasks spanning six professional roles and 22 procedural skills, designed to evaluate how skills transfer across tasks, roles, and model backbones. The benchmark includes controlled evaluation settings for local improvement, cross-task transfer, cross-role transfer, and cross-model generalization. Experiments show that procedural memory delivers consistent gains in industrial workflows: a single refinement round improves aggregate performance by 3.7-6.7 points, while skills evolved from diverse multi-model execution traces achieve 73.1% cross-model test accuracy, outperforming all single-model trace sources. We further find that some skills generalize broadly across tasks and models, whereas others become specialized to role-specific workflows and lose effectiveness under transfer. These results provide practical guidance for building, evaluating, and deploying procedural memory systems in production agent platforms.
Chinese Translation
程序性记忆越来越多地被用于提升大型语言模型（LLM）代理在重复性工作任务中的表现，但其产生可重用技能的能力仍然不甚了解。我们引入了AFTER，一个包含382个现实企业任务的基准，涵盖六个专业角色和22项程序性技能，旨在评估技能在任务、角色和模型骨干之间的转移。该基准包括用于局部改进、跨任务转移、跨角色转移和跨模型泛化的受控评估设置。实验表明，程序性记忆在工业工作流程中提供了一致的提升：单次优化轮次使整体表现提高了3.7-6.7分，而从多样化的多模型执行轨迹中演变出的技能在跨模型测试中达到了73.1%的准确率，超越了所有单模型轨迹来源。我们进一步发现，一些技能在任务和模型之间具有广泛的泛化能力，而另一些则专门针对角色特定的工作流程，转移时效果下降。这些结果为在生产代理平台中构建、评估和部署程序性记忆系统提供了实用指导。
cs.AI / 153 / 2606.23158
Decomposing Financial Market Dynamics via Mechanism Analysis in an Evolutionary Multi-Agent Simulation
通过机制分析分解金融市场动态的进化多智能体模拟
Abstract
Evolutionary agent-based markets (ABMs) couple several mechanisms -- who reproduces, how price forms, how biased the agents are, how consensus propagates -- yet these are usually fixed by convention, so it is unclear which mechanism controls which emergent property. In a coevolving, endogenous-price simulator with 120 heterogeneous behavioral agents, we make four mechanisms pluggable and run matched 3x20-seed interventions. We find the levers are largely separable. (1) Selection -> diversity: a Quality-Diversity (QD/MAP-Elites) operator robustly raises strategy-mix entropy over truncation top-k (paired Delta entropy +0.27 to +1.12 bits; sign-test p<0.001; CIs exclude 0) and sustains more strategy cycling (strongest in crisis: Delta=+0.070, p=0.0004). (2) Selection does not improve realism: even a per-agent realism reward that provably steers selection does not raise 5-fact realism (Delta_5=-0.11,-0.08,+0.03; not significant). (3) Microstructure -> realism: enabling reflexive price feedback does raise realism (Delta_5=+0.13,+0.20,+0.20; crisis/bull p<0.05, all CIs positive). (4) Behavior -> fragility: amplifying behavioral bias raises a genomic fragility proxy (Delta=+10.5,+11.1,+14.4; bull p<0.001, all CIs positive) while leaving realism flat. The remaining mechanism -- consensus network topology -- shows no robust effect (honest null). The contribution is a decomposition: in these single-mechanism sweeps the mechanisms behave as approximately distinct control knobs over diversity, realism, and fragility.
Chinese Translation
进化智能体市场（ABMs）结合了多个机制——谁进行繁殖、价格如何形成、智能体的偏见程度、共识如何传播——但这些机制通常是由惯例固定的，因此不清楚哪个机制控制哪个涌现特性。在一个具有120个异质行为智能体的共同进化、内生价格模拟器中，我们使四个机制可插拔，并进行匹配的3x20种干预实验。我们发现这些杠杆在很大程度上是可分离的。(1) 选择 -> 多样性：质量-多样性（Quality-Diversity，QD/MAP-Elites）操作符稳健地提高了策略组合熵，相较于截断的前k名（配对的Delta熵从+0.27提高到+1.12比特；符号检验p<0.001；置信区间排除0），并维持了更多的策略循环（在危机中最强：Delta=+0.070，p=0.0004）。(2) 选择并未提高现实性：即使是一个能够明显引导选择的每个智能体的现实性奖励，也未能提高5项现实性（Delta_5=-0.11,-0.08,+0.03；不显著）。(3) 微观结构 -> 现实性：启用反射性价格反馈确实提高了现实性（Delta_5=+0.13,+0.20,+0.20；危机/牛市p<0.05，所有置信区间均为正）。(4) 行为 -> 脆弱性：放大行为偏见提高了基因组脆弱性代理（Delta=+10.5,+11.1,+14.4；牛市p<0.001，所有置信区间均为正），而现实性保持平稳。剩余的机制——共识网络拓扑——未显示出稳健的效果（诚实的零假设）。本研究的贡献在于分解：在这些单一机制的实验中，各机制表现为对多样性、现实性和脆弱性的近似独立的控制旋钮。
cs.AI / 154 / 2606.23181
DART: Draft-Agreement Routing for Training-Free Adaptive Thinking Budgets in Hybrid Reasoning Models
DART：用于训练无关的自适应思维预算的草案-协议路由在混合推理模型中的应用
Abstract
Hybrid reasoning models can answer directly or spend extra tokens on extended thinking. A practical router should choose between these modes for each query, so easy problems avoid unnecessary reasoning and hard problems receive enough budget to finish the answer. Existing routers move in this direction, but they typically require labeled training data or fix thinking budgets up front, ignoring answer-level evidence from the model itself. We introduce DART, a training-free routing framework that samples two cheap no-think drafts, accepts direct answering when the drafts agree, and predicts a thinking budget from draft entropy when they disagree. Across the main comparisons, DART preserves or improves always-thinking accuracy in most settings while reducing thinking-token use. On math reasoning, accuracy improves by up to $+$9.0 points on Olympiad-level problems while thinking tokens drop 15-69%. On code reasoning under execution-based equivalence, accuracy improves by up to +22.5 points while thinking tokens drop 51-63%. The Stage~1 signal extends across model scales (0.6B-32B), model families, and API-only hosted settings, with no labeled data and no gradient updates required.
Chinese Translation
混合推理模型可以直接回答问题或在扩展思维上花费额外的代币。一个实用的路由器应该为每个查询在这两种模式之间进行选择，以便简单问题避免不必要的推理，而难题则获得足够的预算来完成答案。现有的路由器朝着这个方向发展，但它们通常需要标注的训练数据或预先固定思维预算，忽略了模型自身的答案级证据。我们提出了DART，一个无训练的路由框架，它采样两个廉价的无思考草案，当草案一致时接受直接回答，当草案不一致时根据草案的熵预测思维预算。在主要比较中，DART在大多数设置中保持或提高了始终思考的准确性，同时减少了思维代币的使用。在数学推理方面，对于奥林匹克级别的问题，准确性提高了最多9.0个百分点，而思维代币减少了15-69%。在基于执行的代码推理中，准确性提高了最多22.5个百分点，而思维代币减少了51-63%。阶段1信号在模型规模（0.6B-32B）、模型系列和仅限API托管设置中扩展，无需标注数据和梯度更新。
cs.AI / 155 / 2606.23189
Capable but Careless: Do Computer-Use Agents Follow Contextual Integrity?
有能力但粗心：计算机使用代理是否遵循情境完整性？
Abstract
Computer-use agents (CUAs) now act on a user's behalf across personal applications such as email, calendars, and to-do lists. This cross-application access is useful, but it also creates a privacy risk that has been largely overlooked: when an agent works in one context, it can pull in information from another that is inappropriate in that context. Hence, we introduce AgentCIBench, an evaluation harness that turns this risk into executable, deterministically scored scenarios. We target three common failure modes in CUAs: visual co-location, where the agent pulls in prohibited items that sit next to the task target in the UI; task-ambiguity overshare, where the agent dumps dense personal state in response to an under-specified prompt; and recipient misalignment, where the agent sends content to an addressee for whom it is inappropriate. We evaluate 15 frontier agents and find a surprisingly high failure rate: 11 of 15 leak on more than 50% of scenarios, with an average leakage of 67.9%, and the same failures persist when agents act end-to-end in the environment to complete the task. We release AgentCIBench to encourage the development of safer computer-use agents and position contextual disclosure testing as a pre-deployment safety check.
Chinese Translation
计算机使用代理（CUAs）现在代表用户在个人应用程序中执行操作，例如电子邮件、日历和待办事项列表。这种跨应用程序的访问是有用的，但也带来了一个被大多数人忽视的隐私风险：当代理在一个上下文中工作时，它可能会从另一个上下文中提取不适合该上下文的信息。因此，我们引入了AgentCIBench，一个评估工具，将这一风险转化为可执行的、确定性评分的场景。我们针对CUAs中的三种常见失败模式进行研究：视觉共位，代理在用户界面中提取与任务目标相邻的禁止项；任务模糊过度共享，代理在响应不明确的提示时倾倒密集的个人状态；以及接收者不匹配，代理将内容发送给不适合的收件人。我们评估了15个前沿代理，发现其失败率出乎意料地高：15个代理中有11个在超过50%的场景中泄露信息，平均泄露率为67.9%，而且当代理在环境中端到端地执行任务时，同样的失败仍然存在。我们发布AgentCIBench，以鼓励开发更安全的计算机使用代理，并将情境披露测试定位为部署前的安全检查。
cs.AI / 156 / 2606.23219
SPADE: Structure-Prior Adaptive Decision Estimation
SPADE：结构优先自适应决策估计
Abstract
Physical-structure priors such as conservation laws, Hamiltonian forms, and symmetries can improve scientific machine learning when correct, but can degrade predictions when misspecified. Existing methods usually enforce a chosen structure or tune a soft penalty, without a calibrated rule for deciding whether to impose a prior, how strongly to impose it, which prior to use, or which subset of candidate laws holds. We introduce SPADE, Structure-Prior Adaptive Decision Estimation, a closed-form framework that treats this problem as shrinkage of the structure-violating block of an unconstrained estimator. SPADE uses one exact specification test and one estimand: the test decides whether the prior is supported by data, Stein-unbiased James-Stein shrinkage sets the enforcement strength with an $O(\sigma^2/n)$ oracle guarantee, and a gate commits to the hard prior only when the test certifies it. The same test yields consistent nested structure selection and Benjamini-Hochberg control for subset discovery in non-nested constraint families. Across a linear-subspace prior, a reservoir conservation law, and a nonlinear Hamiltonian prior on Duffing dynamics, SPADE tracks the oracle, beats a neural-network baseline, reduces correct-prior regret from $10.3\%$ to $2.6\%$, matches cross-validation with $1/71$ of the solves, selects the correct structure with $100\%$ accuracy, and recovers partial laws with controlled false relaxation.
Chinese Translation
物理结构先验，如守恒定律、哈密顿形式和对称性，在正确的情况下可以改善科学机器学习，但在错误指定时可能会降低预测精度。现有方法通常强制选择的结构或调整软惩罚，而没有校准的规则来决定是否施加先验、施加的强度、使用哪个先验或哪个候选定律的子集成立。我们引入了SPADE（Structure-Prior Adaptive Decision Estimation），一个封闭形式的框架，将此问题视为对无约束估计器的结构违反块的收缩。SPADE使用一个精确的规格检验和一个估计量：该检验决定先验是否得到数据支持，Stein无偏James-Stein收缩设置施加强度，具有$O( rac{ au^2}{n})$的oracle保证，只有在检验认证时，门控才会承诺施加硬先验。相同的检验产生一致的嵌套结构选择和Benjamini-Hochberg控制，用于非嵌套约束族中的子集发现。在线性子空间先验、一个水库守恒定律和对Duffing动力学的非线性哈密顿先验的情况下，SPADE追踪oracle，超越神经网络基线，将正确先验的遗憾从$10.3\%$降低到$2.6\\%$，以$1/71$的求解匹配交叉验证，以$100\\%$的准确性选择正确结构，并以受控的虚假松弛恢复部分定律。
cs.AI / 157 / 2606.23238
HOLMES: Evaluating Higher-Order Logical Reasoning in LLMs
HOLMES：评估大型语言模型中的高阶逻辑推理
Abstract
Logical reasoning is essential for reliable AI, yet existing benchmarks are largely first-order-logic-centric, focusing on object-level deduction over fixed predicates. This misses many realistic scenarios where models must reason over rules, predicates, functions, constraints, and decision procedures themselves. We introduce HOLMES (Higher-Order Logic Meets real-world Explainable Symbolic reasoning), the first real-world benchmark for higher-order symbolic reasoning in LLMs, containing 1379 instances. Built on higher-order logic, HOLMES pairs natural-language problems with HOL formalizations, ground-truth answers, verifiable reasoning traces, and fine-grained controllable reasoning factors across law and finance. Experiments show that current LLMs still struggle on HOLMES, with an average accuracy of only 50.64% and the best model reaching 59.54%. Our analyses further reveal that high final-answer accuracy can mask shortcut reasoning in conflict-resolution settings, while performance drops sharply under scope-conditioned and compositional reasoning. These findings identify higher-order symbolic reasoning as a key bottleneck for building reliable and verifiable LLMs. The project code and dataset are publicly available at https://github.com/wuyucheng2002/HOLMES.
Chinese Translation
逻辑推理对于可靠的人工智能至关重要，但现有的基准测试主要集中在一阶逻辑上，侧重于对固定谓词的对象级推理。这忽略了许多现实场景，在这些场景中，模型必须对规则、谓词、函数、约束和决策过程本身进行推理。我们引入了HOLMES（高阶逻辑与现实世界可解释符号推理的结合），这是第一个针对大型语言模型中高阶符号推理的现实世界基准，包含1379个实例。HOLMES基于高阶逻辑，将自然语言问题与HOL形式化、真实答案、可验证的推理轨迹以及法律和金融领域的细粒度可控推理因素相结合。实验表明，目前的语言模型在HOLMES上仍然面临挑战，平均准确率仅为50.64%，最佳模型的准确率为59.54%。我们的分析进一步揭示，高最终答案的准确性可能掩盖了在冲突解决环境中的捷径推理，而在范围条件和组合推理下，性能急剧下降。这些发现将高阶符号推理识别为构建可靠和可验证的大型语言模型的关键瓶颈。项目代码和数据集可在 https://github.com/wuyucheng2002/HOLMES 获取。
cs.AI / 158 / 2606.23277
GIF: Locally Sound Geometric Information Flow Control for LLMs
GIF：针对大语言模型的局部安全几何信息流控制
Abstract
Large language models increasingly mediate interactions between sensitive data, untrusted inputs, and privileged actions in agentic systems, creating security and privacy risks. These range from prompt injections that manipulate downstream tool use to leakage of confidential information through model outputs. Recent Information Flow Control (IFC)-based defenses show promise but lack a principled semantic foundation for reasoning about information flow through the model itself. Since any input token may influence any output token in an autoregressive LLM, existing approaches suffer from severe taint explosion. We present Geometric Information Flow (GIF), a semantic framework for tracking information flow from input tokens to outputs. GIF uses the LLM Jacobian and local output geometry to upper-bound the Shannon mutual information between perturbed input spans and model outputs, yielding a scalable measure computable on large models via automatic differentiation and low-rank approximation. Unlike attention-based or correlational attribution heuristics, GIF satisfies local geometric soundness, and we provide a fully mechanized Lean 4 proof that it upper-bounds the true information flow induced by a given prompt under local regularity assumptions. We evaluate GIF on integrity and confidentiality tasks across multiple prompt-injection and privacy-leakage benchmarks. GIF achieves near-perfect recall even without a downstream declassifier, outperforming attention-based baselines. Combined with lightweight LLM-based declassifiers, it matches or exceeds the F1 of direct LLM-as-judge baselines such as GPT-5.5 xhigh reasoning while using up to 81x lower token cost. GIF flows detected with small surrogate models transfer to larger state-of-the-art models and other model families, even when the surrogate is up to 200x smaller, suggesting black-box deployment without gradient access.
Chinese Translation
大型语言模型越来越多地在代理系统中调解敏感数据、非信任输入和特权操作之间的交互，这带来了安全和隐私风险。这些风险包括通过提示注入操控下游工具使用，以及通过模型输出泄露机密信息。最近基于信息流控制（IFC）的防御措施显示出希望，但缺乏对模型内部信息流进行推理的原则性语义基础。由于任何输入标记都可能影响自回归大语言模型中的任何输出标记，现有方法面临严重的污染爆炸问题。我们提出了几何信息流（GIF），这是一个用于跟踪从输入标记到输出的信息流的语义框架。GIF利用大语言模型的雅可比矩阵和局部输出几何来上界扰动输入范围与模型输出之间的香农互信息，从而提供一种可扩展的度量，可以通过自动微分和低秩近似在大型模型上计算。与基于注意力或相关性归因启发式方法不同，GIF 满足局部几何安全性，我们提供了一个完全机械化的 Lean 4 证明，证明其在局部规则假设下上界了由给定提示引起的真实信息流。我们在多个提示注入和隐私泄露基准上评估了 GIF 在完整性和保密性任务中的表现。即使没有下游去分类器，GIF 也实现了近乎完美的召回率，超越了基于注意力的基线。结合轻量级的大语言模型去分类器，GIF 的 F1 分数与直接将大语言模型作为判断基线（如 GPT-5.5 xhigh 推理）相匹配或超过，同时使用的标记成本低至 81 倍。使用小型替代模型检测到的 GIF 流可以转移到更大的一流模型和其他模型家族，即使替代模型小至 200 倍，这表明可以在没有梯度访问的情况下进行黑箱部署。
cs.AI / 159 / 2606.23301
EHR-Complex: Benchmarking Medical Agents for Complex Clinical Reasoning
EHR-Complex：复杂临床推理的医疗智能体基准测试
Abstract
Clinical agents promise to democratize access to electronic health records (EHRs), yet existing benchmarks fail to reflect the complexity of practical EHR analysis, e.g., often operating on idealized, clean EHRs via static SQL generation rather than interactive execution. In this work, we introduce EHR-Complex, a large-scale benchmark designed for interactive clinical database reasoning. Built on the large MIMIC-IV substrate (365K patients, 31 tables, 500M+ records), EHR-Complex comprises about 52K tasks spanning six clinical intents, supporting both patient-level and population-level queries, where each task requires an agent to interact with a sandboxed environment by executing SQL queries or Python code. Notably, EHR-Complex considers the real-world SQL task complexity for longitudinal multi-table aggregation and compositional reasoning, resulting in 31.93 SQL structural components per query on average. Evaluation results on EHR-Complex reveal the clinical difficulty of these EHR reasoning scenarios, with the top-performing model achieving only 62.3% exact-match accuracy. Pass^k consistency drops below 50% for nearly all evaluated models at k=4, exposing broad stochastic fragility. A fine-grained analysis of more than 3,800 failed trajectories for representative LLMs reveals three dominant failure modes: SQL logic errors, medical-code lookup failures, and semantic misunderstandings. EHR-Complex provides a rigorous testbed for clinical agents and highlights remaining gaps in robust reasoning for large-scale EHR analysis.
Chinese Translation
临床智能体有望使电子健康记录（EHR）的访问更加普及，但现有基准未能反映实际EHR分析的复杂性，例如，通常在理想化、干净的EHR上通过静态SQL生成而非交互执行进行操作。在本研究中，我们介绍了EHR-Complex，这是一个旨在进行交互式临床数据库推理的大规模基准。EHR-Complex建立在大型MIMIC-IV基础上（365K患者，31个表，超过5亿条记录），包含约52K个任务，涵盖六种临床意图，支持患者级和人群级查询，每个任务要求智能体通过执行SQL查询或Python代码与沙盒环境进行交互。值得注意的是，EHR-Complex考虑了纵向多表聚合和组合推理的实际SQL任务复杂性，导致每个查询平均有31.93个SQL结构组件。对EHR-Complex的评估结果揭示了这些EHR推理场景的临床难度，表现最佳的模型仅达到62.3%的精确匹配准确率。在k=4时，几乎所有评估模型的Pass^k一致性降至50%以下，暴露出广泛的随机脆弱性。对3800多个失败轨迹的细致分析显示了三种主要失败模式：SQL逻辑错误、医学编码查找失败和语义误解。EHR-Complex为临床智能体提供了一个严格的测试平台，并突出了大规模EHR分析中稳健推理的剩余差距。
cs.AI / 160 / 2606.23345
Abstract representational geometry supports inference in large language models
抽象表征几何支持大型语言模型中的推理
Abstract
A defining feature of human intelligence is the ability to adapt to changing environments by inferring latent task structure from sparse observations. Neuroscientific research indicates that this capability relies on the hippocampus constructing abstract representations, expressed as low-dimensional, approximately orthogonal manifolds in neural state space. However, the internal mechanisms of large language models (LLMs) remain largely opaque, making it unclear whether they form comparable abstract representations or instead rely on task-specific statistical regularities when performing comparable reasoning tasks. Here we adapt a contextual reversal-learning paradigm to a text-based setting and compare humans and LLMs at both the Behavioural and representational levels. We report that although LLMs exhibit generalizable reasoning less frequently than humans, when such inference occurs, their internal states exhibit abstract geometric structures that resemble those reported in the hippocampus. Notably, this representational geometry is not uniformly distributed but is organized hierarchically across model depth: whereas lower layers show early, stable encoding of stimulus identity, higher layers form a hippocampal-like functional band enriched for abstract context geometry associated with inference. Furthermore, complementary intervention experiments mechanistically implicate geometry in reasoning: task-sequence language modelling induces geometric disentanglement, whereas geometric regularization of higher layers increases the emergence of generalizable inference. Together, these findings establish abstract representational geometry as a mechanistic principle supporting inference in large language models.
Chinese Translation
人类智能的一个定义特征是能够通过从稀疏观察中推断潜在任务结构来适应变化的环境。神经科学研究表明，这一能力依赖于海马体构建抽象表征，这些表征在神经状态空间中表现为低维、近似正交的流形。然而，大型语言模型（LLMs）的内部机制仍然 largely opaque，使得不清楚它们是否形成可比的抽象表征，或者在执行可比的推理任务时是否依赖于特定任务的统计规律。在这里，我们将上下文逆转学习范式适应于基于文本的环境，并在行为和表征层面比较人类与LLMs。我们报告称，尽管LLMs表现出可推广推理的频率低于人类，但当这种推理发生时，它们的内部状态展现出与海马体中报告的抽象几何结构相似的特征。值得注意的是，这种表征几何并不是均匀分布的，而是沿着模型深度层次组织：较低层次显示出刺激身份的早期、稳定编码，而较高层次形成了一个类似海马体的功能带，富含与推理相关的抽象上下文几何。此外，互补的干预实验在机制上将几何与推理联系起来：任务序列语言建模诱导几何解缠结，而对较高层次的几何正则化增加了可推广推理的出现。综合来看，这些发现确立了抽象表征几何作为支持大型语言模型推理的机制原则。
cs.AI / 161 / 2606.23403
Litmus: Zero-Label, Code-Driven Metric Specification for Evaluating AI Systems
Litmus：零标签、代码驱动的指标规范用于评估人工智能系统
Abstract
As agentic LLM systems move from prototypes to deployment across increasingly diverse domains, evaluating them has become both more important and more difficult. The challenge is not only that individual metrics may be unreliable, but that evaluation goals are often left implicit. Without a clear account of what a system is expected to do, how it can fail, and which failures matter, metric choices become difficult to justify, interpret, or validate. We present Litmus, a zero-label system that designs evaluation and monitoring metrics for AI pipelines by eliciting evaluation intent from source code and targeted interrogation. Instead of assuming that the evaluation target is already known, Litmus first identifies what must be measured and why, then converts those answers into constraints for constructing a justified, per-stage metric portfolio. We evaluate Litmus on three real, code-defined AI pipelines - financial account grouping, scientific QA, and inherent risk assessment - against AutoMetrics and three DynamicRubric baselines. Litmus achieves the broadest or tied-broadest concern coverage, spans more pipeline stages, produces a near-zero-redundancy portfolio, and ranks first in validity against per-row quality labels on all three pipelines - decisively on scientific QA (Spearman $\rho=0.72$ vs. less than $0.47$ for every baseline), and within overlapping confidence intervals in relation to two components of the audit framework despite using no labels during metric design. Our results support a shift from automatic metric implementation to automatic metric specification: before asking which metric to compute, evaluation systems should ask what must be measured and why.
Chinese Translation
随着自主型大语言模型（LLM）系统从原型向越来越多样化的领域部署，评估这些系统变得愈加重要且困难。挑战不仅在于单个指标可能不可靠，还在于评估目标往往是隐含的。如果没有明确的系统预期行为、可能的失败方式及其重要性的说明，指标选择将难以合理化、解释或验证。我们提出了Litmus，一个零标签系统，通过从源代码和针对性询问中引出评估意图，为人工智能管道设计评估和监控指标。Litmus首先识别必须测量的内容及其原因，而不是假设评估目标已知，然后将这些答案转化为构建合理的阶段性指标组合的约束。我们在三个真实的代码定义的人工智能管道——金融账户分组、科学问答和固有风险评估——上评估了Litmus，并与AutoMetrics和三个DynamicRubric基线进行了对比。Litmus在覆盖广泛性或并列广泛性方面表现最佳，涵盖了更多的管道阶段，生成了近乎零冗余的指标组合，并在所有三个管道中相对于每行质量标签的有效性排名第一——在科学问答上表现尤为显著（Spearman $
ho=0.72$，而每个基线均低于$0.47$），并在与审计框架的两个组成部分相关的重叠置信区间内，尽管在指标设计过程中未使用任何标签。我们的结果支持从自动指标实现转向自动指标规范：在询问计算哪个指标之前，评估系统应首先询问必须测量什么及其原因。
cs.AI / 162 / 2606.23417
Digital Humanism and Evolutionary Design
数字人文主义与进化设计
Abstract
This paper examines the two concepts of digital humanism and evolutionary design. The aim is to identify and highlight potential common structures, synergies, and challenges. How should and can technical systems be designed, and what implications does this have for the design of our environment? In light of the current debate surrounding artificial intelligence, this paper aims to serve as a preliminary study to help better understand the two concepts of digital humanism and evolutionary design within the context of human-centered technological development. Following a brief introduction, the two concepts of Digital Humanism and Evolutionary Design are presented and graphically visualized. The terms of freedom and responsibility in human decision-making, conviviality, and subjectivity are discussed, along with examples illustrating the distinction between human and artificial intelligence (Turing Test and Chinese Room). The various concepts of evolutionary design (e.g., co-evolutionary or sustainable software development, clean code, or green IT) and Gilbert Simondon's concept of the "open machine" are introduced. The interdependencies between functional specialization and open technology development are highlighted. Both concepts share similar structures. In joint cooperation, they can lead to positive effects and mutual synergies. Significant differences lie in the areas of autonomy and determination in decision-making, as well as in genuine and simulated subjectivity. Open technology development is also currently suffering from the functional specialization of software and AI applications due to a purely market- and consumer-oriented approach. Even optimizations for energy efficiency in sustainable software development lead to greater specialization and thus also have a detrimental effect on open and quality-oriented technology development.
Chinese Translation
本文探讨了数字人文主义和进化设计这两个概念。目的是识别并突出潜在的共同结构、协同效应和挑战。技术系统应如何设计，能够如何设计，这对我们环境的设计又有什么影响？鉴于当前围绕人工智能的讨论，本文旨在作为一项初步研究，以帮助更好地理解数字人文主义和进化设计这两个概念在以人为本的技术发展背景下的意义。在简要介绍之后，数字人文主义和进化设计这两个概念被呈现并图形化展示。讨论了人类决策中的自由与责任、共生性和主体性等术语，并通过示例阐明人类与人工智能之间的区别（图灵测试和中文房间）。介绍了进化设计的各种概念（例如，共进化或可持续软件开发、整洁代码或绿色信息技术）以及吉尔伯特·西蒙东（Gilbert Simondon）提出的“开放机器”概念。强调了功能专业化与开放技术开发之间的相互依赖性。这两个概念具有相似的结构。在共同合作中，它们可以产生积极的效果和相互的协同效应。显著的差异在于决策中的自主性和决定性，以及真实与模拟的主体性。开放技术开发目前也因软件和人工智能应用的功能专业化而受到影响，这种影响源于纯粹的市场和消费者导向的方法。即使是在可持续软件开发中对能源效率的优化，也导致了更大的专业化，从而对开放和质量导向的技术发展产生了不利影响。
cs.AI / 163 / 2606.23441
Cross-Architectural Mixture-of-Experts with Adaptive Soft Routing for Plant Leaf Disease Classification
基于自适应软路由的跨架构专家混合模型用于植物叶片疾病分类
Abstract
Plant leaf disease classification is crucial for crop protection and precision agriculture but remains challenging under complex backgrounds, illumination variations, and severe class imbalance. Moreover, single-architecture models often fail to effectively capture both local and global representations. To address these challenges, this study proposes an adaptive soft Mixture-of-Experts (MoE) framework with cross-architectural routing that integrates EfficientNet-B0, DenseNet-121, and Swin-Tiny to exploit complementary multi-scale, local, and global features. A soft gating mechanism dynamically assigns input-dependent expert weights, while a two-stage refinement training strategy improves optimization stability and generalization. Experiments on a highly imbalanced potato leaf disease dataset achieve 91.68% recall and 92.62% F1-score, surpassing the strongest individual expert by 5.91% and 5.03%, respectively. Additional evaluations on durian and sesame leaf disease datasets yield F1-scores of 94.03% and 97.04%, demonstrating robust cross-dataset generalization and the potential of the proposed framework for reliable real-world crop health monitoring
Chinese Translation
植物叶片疾病分类对于作物保护和精准农业至关重要，但在复杂背景、光照变化和严重类别不平衡的情况下仍然具有挑战性。此外，单一架构模型往往无法有效捕捉局部和全局特征。为了解决这些问题，本研究提出了一种自适应软混合专家（Mixture-of-Experts, MoE）框架，采用跨架构路由，整合了EfficientNet-B0、DenseNet-121和Swin-Tiny，以利用互补的多尺度、局部和全局特征。软门控机制动态分配输入依赖的专家权重，而两阶段的精炼训练策略提高了优化的稳定性和泛化能力。在一个高度不平衡的土豆叶片疾病数据集上的实验中，取得了91.68%的召回率和92.62%的F1分数，分别超过了最强单一专家5.91%和5.03%。在榴莲和芝麻叶片疾病数据集上的额外评估中，F1分数分别为94.03%和97.04%，展示了强大的跨数据集泛化能力，以及所提框架在可靠的现实世界作物健康监测中的潜力。
cs.AI / 164 / 2606.23449
AOHP: An Open-Source OS-Level Agent Harness for Personalized, Efficient and Secure Interaction
AOHP：一个开源操作系统级代理框架，用于个性化、高效和安全的交互
Abstract
AI agents are driving a new software paradigm, with the ability to autonomously call tools, extract information, manage memory, and complete tasks that span applications and data sources. Most existing end-user operating systems, however, are designed for application-centric workflows and offer little native support for AI agents. This mismatch limits the wider adoption of agents and leads to execution overhead and safety risks when running agents on conventional systems. While the concept of agent-native operating systems is emerging, the research community lacks an open testbed to explore the architectural primitives desired for agent-mediated interaction. We present AOHP (Android Open Harness Project), an OS-level agent harness built on the Android Open Source Project (AOSP). The core design principle of AOHP is to treat agents as first-class OS actors, enabling adaptive user interfaces and agent-friendly runtime environments. AOHP preserves the mature Android software and hardware ecosystem while introducing three agent-oriented system mechanisms: personalized service composition, efficient agent interfaces, and secure information flow. Based on preliminary experiments on challenging tasks covering key capabilities of OS agents, AOHP shows clear advantages in task completion (+21.12% completion rate), execution cost (-51.55% token cost), and security-policy compliance.
Chinese Translation
人工智能代理正在推动一种新的软件范式，具备自主调用工具、提取信息、管理内存以及完成跨应用和数据源任务的能力。然而，现有的大多数终端用户操作系统是为以应用为中心的工作流程设计的，缺乏对人工智能代理的原生支持。这种不匹配限制了代理的广泛应用，并在传统系统上运行代理时导致执行开销和安全风险。尽管代理原生操作系统的概念正在兴起，但研究界缺乏一个开放的测试平台来探索代理介导交互所需的架构原语。我们提出了AOHP（Android Open Harness Project），这是一个基于Android开源项目（AOSP）构建的操作系统级代理框架。AOHP的核心设计原则是将代理视为操作系统的第一类参与者，从而实现自适应用户界面和友好的代理运行环境。AOHP在保留成熟的Android软件和硬件生态系统的同时，引入了三种面向代理的系统机制：个性化服务组合、高效的代理接口和安全的信息流。基于对涵盖操作系统代理关键能力的挑战性任务的初步实验，AOHP在任务完成率（+21.12%）、执行成本（-51.55%令牌成本）和安全政策合规性方面显示出明显优势。
cs.AI / 165 / 2606.23487
CADRE: Stable, Parameter Efficient Adaptation of Medical Vision Language Models with Bounded Forgetting and Prior Drift
CADRE：具有有限遗忘和先验漂移的医学视觉语言模型的稳定、参数高效适应
Abstract
Medical vision-language models (VLMs) such as BiomedCLIP generalize broadly, but adapting them to a clinical service is as much a safety problem as an accuracy one. Updating a deployed model for a new imaging modality can fail silently in two ways that harm patients: it can forget modalities it already handled (catastrophic forgetting), and it can drift from its trustworthy pretrained prior toward modality-specific shortcuts. We study parameter-efficient continual adaptation through these two properties rather than leaderboard accuracy, presenting CADRE: a frozen-backbone framework combining low-rank adaptation (LoRA) with an online, self-scaling, similarity-aware elastic weight consolidation term that bounds retained-competence loss, and an anchor-to-prior penalty bounding embedding drift from the frozen prior. Two short guarantees, a bound on total consolidation mass and a scale-invariance property, remove the scale-related sources of vanilla EWC's order fragility. Using breast cancer across three maximally dissimilar modalities (histopathology, ultrasound, chest radiography) as a controlled cross-modality stress test, under a multi-seed, multi-order protocol with paired significance testing and training approximately 0.23% of parameters, CADRE attains the highest accuracy, SPQ, and backward transfer and the lowest forgetting among adapting methods, reducing forgetting roughly sevenfold versus the strongest regularized baseline (0.075 to 0.011; paired p=0.023) and achieving positive backward transfer where every baseline is negative. We frame these as stability properties aligned with clinical-safety desiderata, not a deployment guarantee; robustness to distribution shift and adversarial inputs is out of scope.
Chinese Translation
医学视觉语言模型（VLMs），例如 BiomedCLIP，具有广泛的泛化能力，但将其适应于临床服务既是一个安全问题，也是一个准确性问题。为新的成像模态更新已部署的模型可能会以两种方式默默失败，从而危害患者：它可能会遗忘已处理的模态（灾难性遗忘），并且可能会从其可信的预训练先验漂移到特定模态的捷径。我们研究通过这两个特性进行参数高效的持续适应，而不是关注排行榜准确性，提出了 CADRE：一个冻结骨干网络框架，结合低秩适应（LoRA）与在线、自我缩放、相似性感知的弹性权重整合项，该项限制了保留能力损失，并且一个锚点到先验的惩罚限制了从冻结先验的嵌入漂移。两个简短的保证，一个是对总整合质量的界限，另一个是尺度不变性特性，消除了普通 EWC 的顺序脆弱性中的尺度相关来源。我们以乳腺癌作为受控跨模态压力测试，涵盖三种最大不相似的模态（组织病理学、超声、胸部X光），在多种种子、多种顺序的协议下，通过配对显著性检验和训练大约 0.23% 的参数，CADRE 达到了最高的准确性、SPQ 和向后迁移，同时在适应方法中遗忘率最低，与最强正则化基线相比，遗忘减少了大约七倍（从 0.075 降至 0.011；配对 p=0.023），并实现了正向的向后迁移，而每个基线都是负向的。我们将这些视为与临床安全需求一致的稳定性特性，而不是部署保证；对分布变化和对抗性输入的鲁棒性超出了本研究的范围。
cs.AI / 166 / 2606.23533
POTracker: Optimizing Large Language Models for Standard-Compliant Power Outage Report Generation
POTracker：优化大型语言模型以生成符合标准的停电报告
Abstract
Recent large language models (LLMs) are good at general text generation, but it is still hard to use them for domain-specific data generation because the output must follow strict formatting and structural rules. Unlike open-ended tasks such as question answering or translation, domain-specific generation must be both semantically correct and compliant with existing guidelines and standards. In this work, we study the nationwide interoperability problem of utility power outage reports in the United States. In practice, outage reports need to be machine-readable (e.g., JSON or XML) and must strictly follow requirements from energy-sector regulatory bodies. To address this problem, we propose POTracker, an optimized LLM for power outage report generation. We fine-tune Qwen2.5-7B-Instruct using our proposed objective. The key contribution is a new loss function, POTrackerLoss, that considers both textual similarity and structural (tag) similarity between the generated report and the ground-truth report. We evaluate POTracker on a dataset of 1,000 power outage reports and compare it with five well-known fine-tuning methods and one rule-based XML conversion method. Results show that POTracker outperforms other fine-tuning approaches, improving overall accuracy by up to 51% and reaching 86.47% structural accuracy for generated power outage reports. In addition, we conduct a human study to assess the quality of the ground-truth standard reports, where domain experts assign the generated labels an average score of 4.03 on a 0--5 scale.
Chinese Translation
近年来，大型语言模型（LLMs）在一般文本生成方面表现良好，但在特定领域数据生成方面仍然面临挑战，因为输出必须遵循严格的格式和结构规则。与开放式任务（如问答或翻译）不同，特定领域的生成必须在语义上正确，并符合现有的指导方针和标准。在本研究中，我们探讨了美国公用事业停电报告的全国互操作性问题。在实际应用中，停电报告需要机器可读（例如，JSON或XML），并且必须严格遵循能源部门监管机构的要求。为了解决这个问题，我们提出了POTracker，一种优化的LLM，用于停电报告生成。我们使用我们提出的目标对Qwen2.5-7B-Instruct进行了微调。关键贡献是一个新的损失函数POTrackerLoss，它同时考虑生成报告与真实报告之间的文本相似性和结构（标签）相似性。我们在1000份停电报告的数据集上评估了POTracker，并将其与五种知名的微调方法和一种基于规则的XML转换方法进行了比较。结果表明，POTracker的性能优于其他微调方法，整体准确率提高了多达51%，生成的停电报告的结构准确率达到了86.47%。此外，我们还进行了人类研究，以评估真实标准报告的质量，领域专家对生成的标签在0到5的评分中平均给予了4.03分。
cs.AI / 167 / 2606.23543
VeriEvol: Scaling Multimodal Mathematical Reasoning via Verifiable Evol-Instruct
VeriEvol：通过可验证的演化指令扩展多模态数学推理
Abstract
Scaling reinforcement learning for visual mathematical reasoning requires more than generating harder questions: as data volume grows, the reward labels themselves must remain reliable. Yet existing data pipelines scale supervision while trusting the labeller, and policy-side methods assume the underlying answers are already correct. We instead treat scaling as a verifiable data-construction problem and decouple two axes before any policy update: prompt difficulty, expanded by route-specific evolution operators, and answer reliability, enforced by offline hypothesis-test falsification. We instantiate this as VeriEvol, an iterative framework with two extensible components: a type-aware evolution module that rewrites low-difficulty image-question seeds into harder, image-grounded prompts; and HTV-Agent, a verifier that accepts an answer only after multi-source counter-evidence has failed to refute it. The resulting verified data scales in volume, extends by adding evolution routes or verifier channels, and plugs directly into existing GRPO-style RL recipes. On a five-benchmark visual-math suite, scaling evolved SFT data from 10K to 250K samples raises the mean accuracy from 35.42 to 54.73; then, with backbone, SFT initialization, and GRPO recipe held fixed, VeriEvol adds a cumulative +3.88 over an un-evolved RL baseline, of which +1.82 comes from evolved prompts and +2.06 from the HTV-Agent verifier. We release the prompts, data, models, code, and the full verifier trace of every sample, so that downstream work can scale and audit the pipeline rather than only inspect its outputs.
Chinese Translation
扩展视觉数学推理的强化学习不仅需要生成更难的问题：随着数据量的增长，奖励标签本身也必须保持可靠。然而，现有的数据管道在扩展监督的同时依赖于标注者的信任，而策略侧的方法则假设基础答案已经是正确的。我们将扩展视为一个可验证的数据构建问题，并在任何策略更新之前解耦两个轴：由特定路径的演化操作符扩展的提示难度，以及通过离线假设检验的反驳来强制执行的答案可靠性。我们将其实例化为 VeriEvol，一个具有两个可扩展组件的迭代框架：一个类型感知的演化模块，将低难度的图像问题种子重写为更难的、基于图像的提示；以及 HTV-Agent，一个验证器，仅在多源反证未能驳斥答案后才接受该答案。生成的验证数据在数量上可扩展，通过添加演化路径或验证器通道进行扩展，并直接融入现有的 GRPO 风格强化学习配方。在一个包含五个基准的视觉数学套件中，将演化的 SFT 数据从 10K 扩展到 250K 样本，使平均准确率从 35.42 提升至 54.73；然后，在固定主干、SFT 初始化和 GRPO 配方的情况下，VeriEvol 在未演化的强化学习基线之上增加了累计 +3.88，其中 +1.82 来自演化提示，+2.06 来自 HTV-Agent 验证器。我们发布了提示、数据、模型、代码以及每个样本的完整验证器追踪，以便下游工作能够扩展和审计管道，而不仅仅是检查其输出。
cs.AI / 168 / 2606.23590
The Topology of Ill-Posed Questions: Persistent Homology for Detection and Steering in LLMs
不适定问题的拓扑结构：用于大语言模型检测与引导的持久同调
Abstract
Ill-posed questions, including ambiguous, underspecified, or contradictory queries, may admit no valid answer or multiple plausible answers, posing a challenge for large language models (LLMs). Existing approaches largely analyze ill-posedness through model outputs and often focus on specific subclasses. We investigate whether diverse sources of ill-posedness can be represented within a unified topology of LLM internal states and whether this structure can be used to steer response behavior. We model the contextual hidden states of prompt tokens at each transformer layer as a point cloud and characterize its geometry using finite zero-dimensional persistent homology. Each layer is summarized by three compact descriptors: mean finite lifetime, normalized lifetime entropy, and largest-lifetime concentration. Concatenating these descriptors across layers yields a topology representation of the question. We further introduce topology-conditioned activation steering, which retrieves topologically similar examples and constructs query-specific activation interventions that encourage source-aware clarification or abstention. Across three open-weight LLMs, topology features consistently outperform prompt-based and pooled-hidden-state baselines for ill-posedness classification, improving average accuracy from \(67.4\%\) to \(78.9\%\) on AmbigQA, from \(79.9\%\) to \(88.5\%\) on SituatedQA, and from \(57.6\%\) to \(69.6\%\) on CLAMBER 9-way classification. Topology-conditioned steering increases the average total acceptable response rate from \(61.4\%\) to \(70.6\%\) and grounded acceptable responses from \(11.9\%\) to \(16.4\%\). These results show that persistent homology provides both an interpretable representation of ill-posedness and an effective mechanism for targeted response steering.
Chinese Translation
不适定问题，包括模糊、不明确或矛盾的查询，可能没有有效答案或存在多个合理答案，这对大语言模型（LLMs）构成挑战。现有方法主要通过模型输出分析不适定性，且通常集中于特定子类。我们研究了不同来源的不适定性是否可以在 LLM 内部状态的统一拓扑中表示，以及这种结构是否可以用于引导响应行为。我们将每个变换器层的提示标记的上下文隐藏状态建模为点云，并使用有限零维持久同调来表征其几何特征。每一层通过三个紧凑的描述符进行总结：平均有限生命周期、归一化生命周期熵和最大生命周期浓度。将这些描述符在各层之间连接，得到问题的拓扑表示。我们进一步引入拓扑条件激活引导，检索拓扑相似的示例，并构建特定查询的激活干预，以鼓励源意识的澄清或弃权。在三个开放权重的 LLM 中，拓扑特征在不适定性分类上始终优于基于提示和池化隐藏状态的基线，将 AmbigQA 的平均准确率从 67.4% 提高到 78.9%，将 SituatedQA 的准确率从 79.9% 提高到 88.5%，将 CLAMBER 9 类分类的准确率从 57.6% 提高到 69.6%。拓扑条件引导将平均可接受响应率从 61.4% 提高到 70.6%，将有根据的可接受响应从 11.9% 提高到 16.4%。这些结果表明，持久同调不仅提供了不适定性的可解释表示，还为有针对性的响应引导提供了有效机制。
cs.AI / 169 / 2606.23595
SPIRAL: Learning to Search and Aggregate
SPIRAL：学习搜索与聚合
Abstract
Language model reasoning can be substantially improved at test time via scaffolds that scale inference compute across different primitives -- sequential reasoning within a trace, independently sampled parallel traces, and aggregation of multiple reasoning traces into a final response. During post-training, however, language models are optimized only for sequential reasoning within a single trace. We introduce Sequential-Parallel-Aggregative Reinforcement Learning (SPIRAL), a framework in which a language model is trained to use all three primitives, as part of a unified inference compute pipeline. Concretely, the language model first samples a set of independent traces in parallel, each produced through sequential chain-of-thought reasoning, and then generates a final aggregation trace conditioned on those traces; all components are optimized end-to-end against the reward of the final aggregated response. To train this system, SPIRAL uses set reinforcement learning to teach models to produce a set of traces that are collectively useful for an aggregator and standard reinforcement learning to teach models to aggregate the set into improved final responses. Our experiments on reasoning tasks show that SPIRAL effectively scales with inference compute, outperforming GRPO by up to 11$\times$ scaling efficiency and 15% higher performance when all three compute primitives are scaled.
Chinese Translation
语言模型的推理能力可以通过在不同原语之间扩展推理计算的支架在测试时显著提高——包括在一个轨迹内的顺序推理、独立采样的并行轨迹，以及将多个推理轨迹聚合成最终响应。然而，在后期训练过程中，语言模型仅针对单一轨迹内的顺序推理进行优化。我们提出了顺序-并行-聚合强化学习（SPIRAL）框架，在该框架中，语言模型被训练为使用这三种原语，作为统一推理计算管道的一部分。具体而言，语言模型首先并行采样一组独立轨迹，每个轨迹通过顺序的思维链推理生成，然后生成一个基于这些轨迹的最终聚合轨迹；所有组件都针对最终聚合响应的奖励进行端到端优化。为了训练该系统，SPIRAL使用集合强化学习来教导模型生成一组对聚合器集体有用的轨迹，并使用标准强化学习来教导模型将该集合聚合成改进的最终响应。我们在推理任务上的实验表明，SPIRAL有效地扩展了推理计算，相比于GRPO在扩展效率上提高了多达11倍，并且在所有三种计算原语扩展时性能提高了15%。
cs.AI / 170 / 2606.23597
Against Proxy Optimization
反对代理优化
Abstract
I discuss conditions under which maximizing a proxy utility function is harmful and suggest this poses problems for applying decision theory.
Chinese Translation
我讨论了在何种条件下最大化代理效用函数是有害的，并指出这对应用决策理论提出了问题。
cs.AI / 171 / 2606.23608
Causal Discovery in the Era of Agents
代理时代的因果发现
Abstract
Recent attempts to combine large language models (LLMs) with causal discovery ask models to infer pairwise directions, propose graph structures, or inject language-model outputs as priors and constraints. These approaches promise faster analysis, but they also obscure whether a causal evidence is supported by data and assumptions or by textual associations, prompt artifacts and hallucinated mechanisms. We argue for a different role for agents in causal discovery. Agents should inspect data, retrieve context, explain method assumptions and clarify graph outputs, but they should not supply edges, orientations, priors, constraints or causal conclusions. We propose the principle that agents assist the workflow, while causal claims remain grounded in data, explicit assumptions, formal algorithms, diagnostics and user or domain-expert decisions. We instantiate this principle in causal-learn+, an online platform that coordinates data analysis, preprocessing, method recommendation, expert-knowledge incorporation, formal discovery and interpretation around the algorithmic ecosystem of causal-learn. A case study on Big Five personality data illustrates agent-assisted pipeline of causal discovery without turning language-model unreliability into causal evidence. The platform is available at causallearn.com.
Chinese Translation
近期将大型语言模型（LLMs）与因果发现相结合的尝试要求模型推断成对方向、提出图结构，或将语言模型输出作为先验和约束。这些方法承诺更快的分析，但也模糊了因果证据是由数据和假设支持，还是由文本关联、提示伪影和幻觉机制支持。我们主张在因果发现中赋予代理不同的角色。代理应检查数据、检索上下文、解释方法假设并澄清图输出，但不应提供边、方向、先验、约束或因果结论。我们提出代理应辅助工作流程，而因果主张应基于数据、明确假设、正式算法、诊断以及用户或领域专家的决策。我们在 causal-learn+ 中实例化这一原则，这是一个在线平台，协调数据分析、预处理、方法推荐、专家知识整合、正式发现和解释，围绕 causal-learn 的算法生态系统展开。对大五人格数据的案例研究展示了代理辅助的因果发现流程，而未将语言模型的不可靠性转化为因果证据。该平台可在 causallearn.com 访问。
cs.AI / 172 / 2606.23631
AI-driven Optimisation of Quality of Recovery (QoR) in Remote Patient Monitoring
基于人工智能的远程患者监测中恢复质量（QoR）优化
Abstract
Remote patient monitoring depends on patient-reported data to capture the subjective dimension of recovery that devices cannot measure. The Quality of Recovery (QoR-15) survey is the gold-standard instrument for this purpose. It was designed and validated for occasional in-hospital assessment, yet remote monitoring now administers it to patients daily. In our own post-surgical deployment, only 55% of patients submitted the survey more than 14 days of 30 monitoring days. We developed QoR-compact, a five-item daily input for the RPM prediction pathway. Setting a deployment-driven target of one-third of the daily items, we exhaustively evaluated all 3,003 five-question subsets of the QoR-15 and tested whether the best of them matches the full instrument in predicting near-term postoperative recovery severity. QoR-compact achieves a mean AUC-ROC of 0.968 (95% CI 0.915-0.988), statistically comparable to the 0.964 baseline obtained with one-third of the items. Patient-level backtesting indicates that it tracks readmission events as faithfully as the full form. Its five items span the physical and psychological axes of recovery: Q3 (feeling rested), Q9 (feeling comfortable and in control), Q10 (general well-being), Q12 (severe pain), and Q14 (feeling worried or anxious). The QoR-15 remains the gold-standard measure of recovery; QoR-compact complements it as a shorter daily input designed for prediction. This parity provides the basis for a prospective study of whether a lighter daily input is, in turn, completed more consistently. External validation on larger cohorts is required before clinical use.
Chinese Translation
远程患者监测依赖患者报告的数据来捕捉设备无法测量的恢复主观维度。恢复质量（QoR-15）调查是这一目的的金标准工具。该工具设计并验证用于偶尔的住院评估，但现在远程监测每天向患者施测。在我们自己的术后部署中，仅有55%的患者在30个监测日中提交了超过14天的调查。我们开发了QoR-compact，这是一个五项每日输入的RPM预测路径。设定以部署为驱动的目标为每日项目的三分之一，我们对QoR-15的所有3,003个五问题子集进行了全面评估，并测试其中最佳的子集是否在预测近期术后恢复严重性方面与完整工具相匹配。QoR-compact的平均AUC-ROC为0.968（95% CI 0.915-0.988），在统计上可与三分之一项目获得的0.964基线相媲美。患者级别的回测表明，它对再入院事件的跟踪与完整形式同样忠实。其五个项目涵盖了恢复的身体和心理轴线：Q3（感到休息好），Q9（感到舒适和掌控），Q10（总体健康），Q12（剧烈疼痛），和Q14（感到担忧或焦虑）。QoR-15仍然是恢复的金标准测量；QoR-compact作为一种旨在预测的更短每日输入对其进行了补充。这种平行关系为前瞻性研究提供了基础，以确定较轻的每日输入是否能更一致地完成。在临床使用之前，需要在更大样本中进行外部验证。
cs.AI / 173 / 2606.23633
AI Exposure Scores: what they measure, what they miss, and what comes next
人工智能暴露评分：它们测量了什么，遗漏了什么，以及接下来会发生什么
Abstract
A set of exposure scores calculated in 2023 has become a central empirical input to the future of work debate. Produced by Eloundou et al. (2023) and referred to here as the GPTs are GPTs scores, they define exposure as the share of occupational tasks a large language model can assist with. This work is a genuine methodological contribution, but as the scores travel from the time and place they were produced, the limitations the authors named do not always travel with them. Two gaps have widened as a result. The first is structural, between what static exposure scores measure and what policy questions actually require. Taking the diffusion of these scores as a case study, we show how their temporal, geographic, and ontological limitations compound in policy-facing analyses, and we survey five families of research responding to these limits: dynamic and benchmark-based measures, ensemble methods, task-framework extensions, worker-centered metrics, and adoption and usage data. The second gap is the one we argue needs more attention: the coordination between researchers and policymakers. The policy-relevant work which ask who is harmed, who benefits, how, and when, continues to reference the static GPTs are GPTs scores without engagement with the methodological updates that would let these questions be answered more reliably. We then ask what additional steps towards navigating uncertainty remain: ex-post frameworks and the deliberate, political work of reimagining what futures are worthy of building towards are. Closing the research-policy gap is a shared task: policymakers must widen their evidence base, engage workers as epistemic partners, and shift from prediction to preparedness; researchers must build data infrastructure, adopt participatory methods, and write with policymakers in mind. Better measurement matters, but it will not close the second gap alone.
Chinese Translation
一组在2023年计算的暴露评分已成为未来工作辩论的核心实证输入。由Eloundou等人（2023）制作，并在此称为GPTs are GPTs评分，它们将暴露定义为大型语言模型能够协助的职业任务的比例。这项工作是一个真正的方法论贡献，但随着评分从其产生的时间和地点传播，作者所提到的局限性并不总是随之而来。结果导致了两个差距的扩大。第一个是结构性的，即静态暴露评分所测量的内容与政策问题实际所需之间的差距。以这些评分的传播为案例研究，我们展示了它们的时间、地理和本体论局限性如何在面向政策的分析中相互叠加，并调查了五个回应这些局限性的研究领域：动态和基准测量、集成方法、任务框架扩展、以工人为中心的指标以及采纳和使用数据。第二个差距是我们认为需要更多关注的：研究人员与政策制定者之间的协调。与政策相关的工作询问谁受到伤害，谁受益，如何以及何时，仍然继续参考静态的GPTs are GPTs评分，而未与能够更可靠地回答这些问题的方法论更新进行互动。我们接着询问在应对不确定性方面还剩下哪些额外步骤：事后框架和重新构想值得建设的未来的有意识的政治工作。弥合研究与政策之间的差距是一项共同的任务：政策制定者必须扩大其证据基础，将工人视为认知伙伴，并从预测转向准备；研究人员必须建立数据基础设施，采用参与性方法，并以政策制定者为目标进行写作。更好的测量很重要，但单靠它无法弥合第二个差距。
cs.AI / 174 / 2606.23643
TailorMind: Towards Preference-Aligned Multimodal Content Generation
TailorMind：面向偏好对齐的多模态内容生成
Abstract
Personalized content systems depend on available UGC and struggle when suitable content is absent, delayed, or costly to create. Although multimodal generators can synthesize content on demand, how to translate behavioral traces into generation-ready preferences remains underexplored. We study personalized multimodal content generation: creating user-tailored multimodal content without existing item pools or waiting for matching UGC. We propose TailorMind, linking collaborative preference modeling with controllable multimodal generation. TailorMind enriches sparse user histories via hypergraph collaborative filtering and optimizes textual profiles with ranking-error feedback and textual gradient descent. Retrieval-augmented style control grounds outputs in authentic UGC patterns, while cross-modal cohesion reflection reduces semantic drift. We construct TailorBench, a benchmark from three mainstream platforms evaluated along five dimensions: coherence, novelty, aesthetic, hallucination, profiling. Experiments show that TailorMind achieves competitive or stronger coherence, improves novelty and aesthetic quality over representative generation baselines and ground-truth UGC, demonstrating advantages over retrieving available content or comparable UGC, while achieving up to 29% Recall gains in reranking. Our code is released at: https://github.com/iLearn-Lab/TailorMind.
Chinese Translation
个性化内容系统依赖于可用的用户生成内容（UGC），在缺乏合适内容、内容生成延迟或成本高昂时面临挑战。尽管多模态生成器能够按需合成内容，但如何将行为痕迹转化为可用于生成的偏好仍然未得到充分探索。我们研究个性化多模态内容生成：在没有现有项目池或等待匹配UGC的情况下，创建用户定制的多模态内容。我们提出了TailorMind，将协作偏好建模与可控的多模态生成相结合。TailorMind通过超图协作过滤丰富稀疏的用户历史，并利用排名误差反馈和文本梯度下降优化文本档案。检索增强的风格控制使输出基于真实UGC模式，而跨模态一致性反映则减少了语义漂移。我们构建了TailorBench，这是一个来自三个主流平台的基准，评估五个维度：连贯性、新颖性、美学、幻觉和画像。实验表明，TailorMind在连贯性方面达到了竞争力或更强的表现，在新颖性和美学质量上优于代表性生成基线和真实UGC，展示了相较于检索可用内容或可比UGC的优势，同时在重新排名中实现了高达29%的召回率提升。我们的代码已发布于：https://github.com/iLearn-Lab/TailorMind。
cs.AI / 175 / 2606.23672
Teaching LLMs String Matching, Backtracking, and Error Recovery to Deduce Bases and Truth Tables for the Combinatorially Exploding Bit Manipulation Puzzles
教会大型语言模型（LLMs）字符串匹配、回溯和错误恢复，以推导组合爆炸的位操作难题的基数和真值表
Abstract
This paper presents our algorithmic innovations for the NVIDIA Nemotron Model Reasoning Challenge, focusing on Bit Manipulation Puzzles. In this task, the objective is to discover a hidden logical rule transforming input binary strings to outputs, then apply it to unseen inputs. Large Language Models (LLMs) notoriously struggle here; traditional methods force them to simulate complex boolean logic and arithmetic, leading to hallucinations. Furthermore, the search space of bitwise operations (combinations of shifts, rotations, and logic gates) suffers from a severe combinatorial explosion. To overcome this computational intractability, we present a novel approach that abandons arithmetic logic entirely in favor of string similarity, structured search, and autonomous error recovery. Our core contributions are: 1. Bases and Truth Table Formulation: We reframe logic-gate deduction into a base-selection task, leveraging string similarity (minimal bit flips) to isolate primitive transformations ("bases") and deduce truth tables without complex arithmetic. 2. Backtracking DFS and Error Recovery: We formalize a search process that tests candidate bases, detects logical collisions across examples, and backtracks upon failure to perform robust error recovery. 3. Bit Tokenization and Interactive Reasoning SFT: We force the tokenizer to encode binary strings as individual single-bit tokens. We use dynamic masking to simulate external oracle feedback, training the model to hypothesize, self-evaluate, and backtrack natively. Evaluated on bit manipulation puzzles, our approach achieved over 96% validation accuracy. This represents the highest performance in this category, driving our 7th Place overall finish in the contest.
Chinese Translation
本文展示了我们在NVIDIA Nemotron模型推理挑战中的算法创新，重点关注位操作难题。在这一任务中，目标是发现一个隐藏的逻辑规则，该规则将输入的二进制字符串转换为输出，然后将其应用于未见过的输入。大型语言模型（LLMs）在这方面表现不佳；传统方法迫使它们模拟复杂的布尔逻辑和算术运算，导致幻觉。此外，位操作的搜索空间（移位、旋转和逻辑门的组合）遭遇严重的组合爆炸。为了克服这种计算上的不可处理性，我们提出了一种新颖的方法，完全放弃算术逻辑，转而采用字符串相似性、结构化搜索和自主错误恢复。我们的核心贡献包括：1. 基数和真值表的构建：我们将逻辑门推导重新框定为基数选择任务，利用字符串相似性（最小位翻转）来隔离原始变换（“基数”）并在没有复杂算术的情况下推导真值表。2. 回溯深度优先搜索（DFS）和错误恢复：我们形式化了一种搜索过程，测试候选基数，检测示例之间的逻辑冲突，并在失败时回溯以进行稳健的错误恢复。3. 位标记化和互动推理的自监督学习（SFT）：我们强制标记器将二进制字符串编码为单个单比特标记。我们使用动态掩码来模拟外部神谕反馈，训练模型进行假设、自我评估和原生回溯。在位操作难题上的评估中，我们的方法达到了超过96%的验证准确率。这在该类别中代表了最高性能，使我们在比赛中获得了第七名的整体成绩。
cs.AI / 176 / 2606.23673
PsyBridge: A Hybrid Intelligent Framework for Multi-Dimensional Mental Health Assessment and Decision Support
PsyBridge：一种用于多维心理健康评估与决策支持的混合智能框架
Abstract
Mental health assessment commonly relies on isolated screening instruments or data-driven models that often lack interpretability and multi-dimensional integration. Existing approaches frequently focus on individual indicators such as depression or anxiety while providing limited support for comprehensive and explainable decision-making. To address this limitation, this study proposes PsyBridge, a hybrid intelligent decision-support framework designed for multi-dimensional mental health assessment through the integration of clinically validated screening tools, cognitive evaluation, and personality profiling within a unified architecture. The proposed framework incorporates PHQ-9 and GAD-7 assessments alongside cognitive and behavioural indicators using a modular design and a weighted aggregation mechanism to generate interpretable mental health risk classifications and recommendations. To evaluate the framework, a semi-synthetic dataset consisting of 500 patient profiles representing varying severity levels was constructed based on clinically grounded score distributions. Experimental results demonstrate that PsyBridge achieves an overall accuracy of 0.84, outperforming standalone PHQ-9 and GAD-7 assessments while improving precision, recall, and F1-score. Sensitivity analysis and ablation studies further indicate that integrating cognitive and personality components contributes to more stable classification performance and reduces inconsistencies in moderate-risk prediction. The findings suggest that PsyBridge provides a scalable and interpretable approach for AI-assisted mental health decision support, particularly within digital healthcare and telehealth environments.
Chinese Translation
心理健康评估通常依赖于孤立的筛查工具或缺乏可解释性和多维整合的数据驱动模型。现有方法常常集中于抑郁或焦虑等个体指标，而对全面且可解释的决策支持提供的帮助有限。为了解决这一局限性，本研究提出了PsyBridge，一种混合智能决策支持框架，旨在通过在统一架构中整合临床验证的筛查工具、认知评估和个性分析，实现多维心理健康评估。该框架结合了PHQ-9和GAD-7评估，以及认知和行为指标，采用模块化设计和加权聚合机制，生成可解释的心理健康风险分类和建议。为了评估该框架，构建了一个包含500个患者档案的半合成数据集，代表不同的严重程度，基于临床基础的评分分布。实验结果表明，PsyBridge的整体准确率达到0.84，优于单独的PHQ-9和GAD-7评估，同时提高了精确度、召回率和F1分数。敏感性分析和消融研究进一步表明，整合认知和个性成分有助于更稳定的分类性能，并减少中等风险预测中的不一致性。研究结果表明，PsyBridge为AI辅助的心理健康决策支持提供了一种可扩展且可解释的方法，特别是在数字医疗和远程医疗环境中。
cs.CL / 1 / 2606.20571
Less is More: Lightweight Prompt Compression for Question Answering Applications on Edge Devices
少即是多：边缘设备上问答应用的轻量级提示压缩
Abstract
In agent-driven question answering (QA) applications, retrieval-augmented generation (RAG) is commonly introduced to enhance the response accuracy of large language models (LLMs) by providing additional context. Due to the inherent noise in retrieval results and the coarse granularity of document-level retrieval, the retrieved context often contains substantial redundant information. In this setting, the agent prompt, consisting of the user query and the associated retrieved context, leads to unnecessary computational overhead during LLM inference. Existing prompt compression methods typically rely on auxiliary small language models (SLMs) to estimate context importance. However, such approaches introduce significant memory and computational overhead, which limits their deployment on resource-constrained edge devices. In this paper, we propose CORE, a two-stage sentence-level prompt compression method that eliminates the need for SLMs. In the first stage, CORE constructs an answer set via named entity recognition (NER) and a clue set via semantic matching. In the second stage, CORE refines the clue set using an orthogonal residual retrieval strategy and designs a spatial proximity-based metric to filter the answer set. The two sets are then combined to form the final compressed context. We implement CORE on an NVIDIA Jetson AGX Orin edge device and a Huawei Nova smartphone. Experimental results demonstrate that within a 2000-token budget, CORE improves accuracy by at least 30.19% compared to state-of-the-art baselines, while reducing memory usage by at least 50.47% and achieving at least 1.94 times speedup on the edge device. Moreover, compared to the state-of-the-art LLMLingua2 method, CORE achieves a substantial energy reduction of 95.74% on the smartphone, highlighting its practicality and generalizability for mobile deployments.
Chinese Translation
在基于代理的问答（QA）应用中，常常引入检索增强生成（RAG）来通过提供额外的上下文来提高大型语言模型（LLMs）的响应准确性。由于检索结果中固有的噪声以及文档级检索的粗粒度，检索到的上下文往往包含大量冗余信息。在这种情况下，由用户查询和相关检索上下文组成的代理提示在LLM推理过程中导致了不必要的计算开销。现有的提示压缩方法通常依赖于辅助的小型语言模型（SLMs）来估计上下文的重要性。然而，这些方法引入了显著的内存和计算开销，限制了它们在资源受限的边缘设备上的部署。本文提出了CORE，一种两阶段的句子级提示压缩方法，消除了对SLMs的需求。在第一阶段，CORE通过命名实体识别（NER）构建答案集，并通过语义匹配构建线索集。在第二阶段，CORE使用正交残差检索策略对线索集进行精炼，并设计了一种基于空间邻近性的度量来过滤答案集。然后将这两个集合结合形成最终的压缩上下文。我们在NVIDIA Jetson AGX Orin边缘设备和华为Nova智能手机上实现了CORE。实验结果表明，在2000个标记的预算内，CORE的准确性比最先进的基线提高了至少30.19%，同时内存使用减少了至少50.47%，在边缘设备上实现了至少1.94倍的加速。此外，与最先进的LLMLingua2方法相比，CORE在智能手机上实现了高达95.74%的显著能耗减少，突显了其在移动部署中的实用性和通用性。
cs.CL / 2 / 2606.20572
Investigating Linguistic Steering: An Analysis of Adjectival Effects Across Large Language Model Architectures
语言引导的研究：对大型语言模型架构中形容词效应的分析
Abstract
Achieving reliable control of Large Language Models (LLMs) requires a precise, scalable understanding of how they interpret linguistic cues. We introduce a rigorous framework using Shapley values to quantify the steering effect of individual adjectives on model performance, moving beyond anecdotal heuristics to principled attribution. Applying this method to 100 adjectives across a diverse suite of models (including o3, gpt-4o-mini, phi-3, llama-3-70b, and deepseek-r1) on the MMLU benchmark, we uncover several critical findings for AI alignment. First, we find that a small subset of adjectives act as disproportionately powerful "levers," yet their effects are not universal. Cross-model analysis reveals a "family effect": models of a shared lineage exhibit correlated sensitivity profiles, while architecturally distinct models react in a largely uncorrelated manner, challenging the notion of a one-size-fits-all prompting strategy. Second, focused follow-up studies demonstrate that the steering direction of these powerful adjectives is not intrinsic but is highly contingent on their syntactic role and position within the prompt. For larger models like gpt-4o-mini, we provide the first quantitative evidence of strong, non-additive interaction effects where adjectives can synergistically amplify, antagonistically dampen, or even reverse each other's impact. In contrast, smaller models like phi-3 exhibit a more literal and less compositional response. These results suggest that as models scale, their interpretation of prompts becomes more sophisticated but also less predictable, posing a significant challenge for robustly steering model behavior and highlighting the need for compositional and model-specific alignment techniques.
Chinese Translation
实现对大型语言模型（LLMs）可靠控制需要对它们如何解释语言线索有精确且可扩展的理解。我们引入了一个严格的框架，使用Shapley值量化单个形容词对模型性能的引导效应，从经验性启发式方法转向原则性归因。将该方法应用于100个形容词，涵盖多种模型（包括o3、gpt-4o-mini、phi-3、llama-3-70b和deepseek-r1）在MMLU基准测试中的表现，我们揭示了几个对人工智能对齐至关重要的发现。首先，我们发现一小部分形容词作为不成比例强大的“杠杆”，但它们的效应并非普遍适用。跨模型分析揭示了“家族效应”：具有共同谱系的模型表现出相关的敏感性特征，而在架构上不同的模型则以大致不相关的方式反应，这挑战了“一刀切”的提示策略的概念。其次，针对这些强大形容词的后续聚焦研究表明，它们的引导方向并非内在的，而是高度依赖于它们在提示中的句法角色和位置。对于像gpt-4o-mini这样的大型模型，我们提供了强烈的非加性交互效应的首个定量证据，其中形容词可以协同放大、对抗性减弱，甚至反转彼此的影响。相比之下，像phi-3这样的小型模型则表现出更字面和较少组合的反应。这些结果表明，随着模型规模的扩大，它们对提示的解释变得更加复杂，但也更难以预测，这对稳健引导模型行为构成了重大挑战，并突显了对组合性和模型特定对齐技术的需求。
cs.CL / 3 / 2606.20632
Post-Training Recipe, More Than Model Family, Shapes Multi-Agent LLM Conversational Behavior
后训练配方，超越模型家族，塑造多智能体大语言模型的对话行为
Abstract
Multi-LLM systems use multiple language models to deliberate, judge each other's outputs, or coordinate as agents. Their value depends on the models producing measurably different conversational behaviors when given the same input. Prior offline studies recommend drawing one model per family for behavioral diversity, because LLMs prefer outputs from their own family when rating one another in isolation. Whether the same family label predicts behavior in interactive multi-LLM systems, the setting that real deployed systems use, has not been tested. We study this with a 940,000-chain 11-checkpoint corpus and a 1.6M-chain same-base Llama factorial. On our validated headline metric, hedging, a reasoning-distilled Llama checkpoint shifts by 18% depending on which same-base partner it replies to, more than any cross-family hedging gap in the controlled subset. Qwen, closed-API, and runtime checks suggest the pattern is not isolated, while repair and challenge analyses remain exploratory because their surface-cue detectors are weaker. Overall, the results identify post-training recipe as a first-class axis for multi-LLM panel composition and show that model family alone is an incomplete proxy for conversational diversity.
Chinese Translation
多大语言模型（Multi-LLM）系统使用多个语言模型进行推理、评估彼此的输出或作为智能体进行协调。它们的价值依赖于在相同输入下模型产生可测量的不同对话行为。先前的离线研究建议从每个家族中抽取一个模型以实现行为多样性，因为大语言模型在孤立评估彼此时更倾向于选择来自自身家族的输出。然而，是否相同的家族标签能够预测在交互式多大语言模型系统中的行为，这一真实部署系统所使用的设置尚未得到验证。我们通过一个包含940,000链和11个检查点的语料库以及一个包含1.6M链的相同基础Llama因子进行研究。在我们验证的主要指标——模糊性（hedging）上，一个经过推理提炼的Llama检查点在回复不同相同基础伙伴时，其表现变化达18%，这一变化超过了受控子集中任何跨家族模糊性差距。Qwen、闭合API和运行时检查表明这一模式并非孤立，而修复和挑战分析仍处于探索阶段，因为它们的表面线索检测器较弱。总体而言，结果表明后训练配方是多大语言模型面板组成的一个重要维度，并显示仅凭模型家族不足以全面代表对话多样性。
cs.CL / 4 / 2606.20650
EmoInstruct-TTS: Dual-Path Instruction-Guided Emotional Speech Synthesis
EmoInstruct-TTS：双路径指令引导的情感语音合成
Abstract
Instruction-based controllable speech synthesis enables users to specify emotions through natural language. However, existing approaches often rely on coarse emotion labels and lack explicit modeling of fine-grained intensity. We propose EmoInstruct-TTS, a dual-path instruction-guided framework for emotional speech synthesis. We introduce Emotion2embed, a supervised semantic-acoustic emotion embedding covering 48 emotional states, including fine-grained categories and intensity levels. To infer embeddings from free-form instructions, we design an Instruction-Conditioned Emotion Flow Model (ICE-Flow) that generates acoustically grounded emotion representations. The inferred embeddings are integrated into an LLM-based synthesis pipeline to provide explicit emotional control while preserving semantic planning. Experiments show improved emotional controllability and speech naturalness over strong baselines.
Chinese Translation
基于指令的可控语音合成使用户能够通过自然语言指定情感。然而，现有的方法往往依赖于粗略的情感标签，缺乏对细粒度强度的明确建模。我们提出了EmoInstruct-TTS，一种用于情感语音合成的双路径指令引导框架。我们引入了Emotion2embed，这是一种覆盖48种情感状态的监督语义-声学情感嵌入，包括细粒度类别和强度水平。为了从自由形式的指令中推断嵌入，我们设计了一种指令条件情感流模型（Instruction-Conditioned Emotion Flow Model, ICE-Flow），该模型生成声学基础的情感表示。推断出的嵌入被集成到基于大语言模型（LLM）的合成管道中，以提供明确的情感控制，同时保持语义规划。实验表明，与强基线相比，情感可控性和语音自然性得到了改善。
cs.CL / 5 / 2606.20691
Specific Domain Ontology Construction Using Large Language Models
使用大型语言模型构建特定领域本体
Abstract
Ontologies are useful structures to organize and maintain information that can be understood both by humans and systems. However, since their manual crafting is a laborious task, many specific domains lack reference ontologies. The outstanding ability for understanding natural language demonstrated by the Large Language Models (LLMs) has motivated their application to aid on a variety of fields, including on ontology development. This work presents the experimentation with a technique that uses LLMs in the role of domain experts to build conceptual hierarchies for a given initial concept. Twenty ontologies automatically constructed for the domain of the Brazilian maritime territory (a.k.a the Blue Amazon) using GPT-3.5 and GPT-4 were then evaluated by human experts. The models were able to construct overall coherent conceptualizations of the domain, but none of the outputs was completely satisfactory as a representation of the context without refinement.
Chinese Translation
本体是用于组织和维护信息的有用结构，这些信息既可以被人类理解，也可以被系统理解。然而，由于手动构建本体是一项繁重的任务，许多特定领域缺乏参考本体。大型语言模型（Large Language Models, LLMs）在理解自然语言方面表现出的卓越能力，激励了它们在多个领域的应用，包括本体开发。本研究展示了一种实验技术，该技术利用LLMs作为领域专家，为给定的初始概念构建概念层次结构。使用GPT-3.5和GPT-4自动构建的二十个本体，针对巴西海域（即蓝色亚马逊）领域进行了人类专家的评估。这些模型能够构建出整体上连贯的领域概念化，但没有任何输出在未经过精炼的情况下完全令人满意，无法作为上下文的完整表示。
cs.CL / 6 / 2606.20696
MindAlign: Decoding Inner Speech from fMRI Signals via Multimodal Embedding Alignment under Limited Data
MindAlign：通过有限数据下的多模态嵌入对齐解码fMRI信号中的内心语言
Abstract
Decoding inner speech from non-invasive brain signals remains a fundamental challenge due to the absence of overt linguistic output, limited training data, and large inter-subject variability. Existing brain-to-text approaches often rely on task-specific decoder fine-tuning, which restricts scalability and complicates adaptation to new participants. We propose MindAlign, a decoupled two-stage brain-to-language framework that enables open-ended text generation from fMRI signals without modifying the underlying language model. The first stage learns a subject-specific neural-semantic alignment that maps fMRI activity into a shared multimodal semantic space, extracting a latent semantic sketch of the internally generated sentence. The second stage integrates this sketch with visual context to prompt a frozen multimodal language model for free-form generation. Experiments on fMRI data collected during silent image description demonstrate that the proposed approach consistently outperforms fMRI-only and random baselines. We further show that the learned semantic-to-language projection can generalize across subjects, enabling effective decoding when paired with subject-specific neural alignment. These results indicate that neural signals modulate semantic content beyond image-driven priors, supporting a scalable and modular direction for brain-to-text decoding.
Chinese Translation
从非侵入性脑信号中解码内心语言仍然是一个基本挑战，原因在于缺乏明显的语言输出、训练数据有限以及个体间的巨大差异。现有的脑到文本方法通常依赖于特定任务的解码器微调，这限制了其可扩展性并使得适应新参与者变得复杂。我们提出了MindAlign，一个解耦的两阶段脑到语言框架，使得能够从fMRI信号中进行开放式文本生成，而无需修改底层语言模型。第一阶段学习特定于个体的神经-语义对齐，将fMRI活动映射到共享的多模态语义空间，提取内部生成句子的潜在语义草图。第二阶段将该草图与视觉上下文结合，以提示一个冻结的多模态语言模型进行自由形式生成。在静默图像描述过程中收集的fMRI数据实验表明，所提出的方法在性能上始终优于仅使用fMRI和随机基线。我们进一步展示了学习到的语义到语言的投影可以跨个体泛化，从而在与特定个体的神经对齐配对时实现有效解码。这些结果表明，神经信号在图像驱动的先验之外调节语义内容，支持脑到文本解码的可扩展和模块化方向。
cs.CL / 7 / 2606.20740
VeriBound: PAC-Bayesian Generalization Bounds for Process Reward Models Trained with Formal Verification Tools
VeriBound：基于PAC-Bayesian的过程奖励模型的泛化界限，利用形式验证工具进行训练
Abstract
Process Reward Models (PRMs) provide step-level verification for Large Language Model (LLM) reasoning, yet their training data acquisition remains a bottleneck: human annotation is costly and Monte Carlo roll-out estimates are noisy. A recent approach, FOVER, trains PRMs on step-level error labels automatically annotated by formal verification tools such as Z3 and Isabelle, and empirically observes cross-task generalization from symbolic tasks to diverse reasoning benchmarks. However, this generalization phenomenon lacks any theoretical explanation, and no formal bounds exist on the generalization error, sample complexity, convergence rate, or downstream Best-of-K performance of such PRMs. We propose VeriBound, a theoretical framework that provides PAC-Bayesian generalization bounds for PRMs trained with formal verification tools. We establish four main results: (i) a PAC-Bayesian generalization bound that relates the empirical verification error on formal-verification-annotated training data to the expected error on unseen reasoning tasks, with the bound depending on the formal verification accuracy and the divergence between training and test task distributions; (ii) a sample complexity result showing that $O(d \log(d/\delta) / \epsilon^2)$ formal-verification-annotated examples suffice to achieve generalization error $\epsilon$ with probability $1-\delta$, where $d$ is the complexity of the PRM hypothesis class; (iii) a convergence analysis proving that PRM training with formal verification labels converges at a linear rate under $L$-smoothness and bounded variance conditions; and (iv) an error propagation bound that relates step-level verification error to Best-of-K performance degradation.
Chinese Translation
过程奖励模型（PRMs）为大型语言模型（LLM）的推理提供逐步验证，但其训练数据的获取仍然是一个瓶颈：人工标注成本高昂，蒙特卡洛展开估计噪声较大。最近的一种方法FOVER，通过形式验证工具（如Z3和Isabelle）自动标注逐步错误标签来训练PRMs，并在经验上观察到从符号任务到多样化推理基准的跨任务泛化现象。然而，这种泛化现象缺乏理论解释，且目前尚无关于泛化误差、样本复杂度、收敛速率或下游最佳选择（Best-of-K）性能的正式界限。我们提出了VeriBound，一个提供基于PAC-Bayesian的泛化界限的理论框架，适用于利用形式验证工具训练的PRMs。我们建立了四个主要结果：（i）一个PAC-Bayesian泛化界限，将形式验证标注训练数据上的经验验证误差与未见推理任务上的期望误差联系起来，该界限依赖于形式验证的准确性和训练任务与测试任务分布之间的差异；（ii）一个样本复杂度结果，表明$O(d ext{log}(d/ ext{δ}) / ext{ε}^2)$个形式验证标注的样本足以以概率$1- ext{δ}$实现泛化误差$ ext{ε}$，其中$d$是PRM假设类的复杂度；（iii）一个收敛分析，证明在$L$-光滑性和有界方差条件下，使用形式验证标签的PRM训练以线性速率收敛；（iv）一个误差传播界限，将逐步验证误差与最佳选择性能的下降联系起来。
cs.CL / 8 / 2606.20751
From Sentiment to Actionable Insights: A Data-Driven Public Sentiment Analysis of Advanced Air Mobility
从情感到可行见解：基于数据的先进空中出行公共情感分析
Abstract
Advanced Air Mobility (AAM) is an emerging low-altitude air transportation system whose successful deployment depends not only on technological advancement but also on public acceptance. This acceptance will drive government support, regulations, noise standards, and willingness to fly, and in turn the overall commercial viability of AAM. Understanding public sentiment toward AAM is therefore essential for identifying its societal barriers and informing its adoption strategies. This study analyzes 306,009 human-generated texts collected from Reddit and Quora to examine public discourse on AAM using AI-based models. Because multiple sentiment analysis models exist, identifying the most accurate model is critical for reliable AAM sentiment prediction and trustworthy public opinion analysis. Accordingly, seven models spanning lexicon-based, machine learning, deep learning, and transformer-based approaches are evaluated for AAM-specific sentiment classification. ModernBERT achieves the best classification performance and is used to label the full dataset. Using the resulting sentiment labels, Latent Dirichlet Allocation (LDA) is applied within each sentiment class to uncover latent topics in public opinion. The analysis identifies 20 distinct topics and traces their temporal evolution from 2008 to 2025. A cross-sentiment topic analysis further reveals six major clusters of public concern: workforce and skill development (25.29% of the dataset), regulation and compliance (24.64%), technical performance of drones (20.99%), military, geopolitics, and defense (14.58%), safety and operational risks (8.51%), and noise and disturbance (5.98%). Based on these findings, this study provides actionable strategies to address these concerns, thereby, improving public acceptance and support AAM deployment.
Chinese Translation
先进空中出行（AAM）是一种新兴的低空空中运输系统，其成功部署不仅依赖于技术进步，还依赖于公众的接受度。这种接受度将推动政府支持、法规、噪声标准以及飞行意愿，从而影响AAM的整体商业可行性。因此，理解公众对AAM的情感对于识别其社会障碍和制定采纳策略至关重要。本研究分析了从Reddit和Quora收集的306,009条人类生成文本，利用基于人工智能的模型考察公众对AAM的讨论。由于存在多种情感分析模型，识别最准确的模型对于可靠的AAM情感预测和可信的公众舆论分析至关重要。因此，评估了七种涵盖词典基础、机器学习、深度学习和基于变换器的方法的模型，以进行AAM特定的情感分类。ModernBERT实现了最佳分类性能，并用于标记完整数据集。利用生成的情感标签，在每个情感类别中应用潜在狄利克雷分配（LDA）以揭示公众舆论中的潜在主题。分析识别出20个不同的主题，并追踪其从2008年到2025年的时间演变。跨情感主题分析进一步揭示了公众关注的六个主要集群：劳动力和技能发展（占数据集的25.29%）、法规和合规（24.64%）、无人机的技术性能（20.99%）、军事、地缘政治和国防（14.58%）、安全和操作风险（8.51%），以及噪声和干扰（5.98%）。基于这些发现，本研究提供了可行的策略以应对这些关注，从而改善公众接受度并支持AAM的部署。
cs.CL / 9 / 2606.20769
FirstPass: Grounding AI Scientific Judgment in Multi-Round Editorial Outcomes
FirstPass：将人工智能科学判断基于多轮编辑结果
Abstract
AI systems for peer review fail on three fronts: they train on Computer Science and Machine Learning venues alone, ignore the iterative dialogue that validates science, and evaluate on stylistic mimicry rather than real editorial judgment. We introduce FirstPass, a dataset and fine-tuned model that addresses all three. Curating 3,668 complete multi-round peer-review dialogues from Nature Communications across five scientific domains (biology, chemistry, neuroscience, physics, and earth science), we exploit mandatory transparent peer review (instituted November 2022) and verify 100% content integrity by automated audit. We fine-tune Qwen2.5-7B-Instruct via Low-Rank Adaptation (LoRA) on three tasks: review generation, reviewer updating, and revision-cycle prediction. Our key finding is that response-only loss masking is a prerequisite, not an optimization: without it, accuracy is 62.0%, below the majority baseline; with it, FirstPass achieves 80.5% accuracy and F1-macro 78.2% on predicting editorial outcomes (Standard vs. Extended revision cycles), outperforming Gemini-3.1-flash-lite-preview zero-shot by 10.4 percentage points and all baselines with statistical significance (McNemar p < 0.001). On generation, FirstPass produces reviews averaging 1,187 words, substantially closer to human references (2,155 words) than any baseline, achieving ROUGE-L 0.154 with significant gains over Qwen and DeepSeek zero-shot (p < 0.001). Deployed in the pre-submission loop as an anticipatory scientific co-author, FirstPass simulates expert critique and predicts revision cycle outcomes before submission, giving authors the judgment a trusted colleague would provide, with consistent cross-domain performance across five disciplines.
Chinese Translation
同行评审的人工智能系统在三个方面存在不足：它们仅在计算机科学和机器学习领域进行训练，忽视了验证科学的迭代对话，并且评估侧重于风格模仿而非真实的编辑判断。我们介绍了FirstPass，一个数据集和微调模型，旨在解决这三个问题。我们从《自然通讯》收集了3,668个完整的多轮同行评审对话，涵盖生物学、化学、神经科学、物理学和地球科学五个科学领域，利用强制性透明同行评审（自2022年11月实施）并通过自动审核验证内容的100%完整性。我们通过低秩适应（Low-Rank Adaptation, LoRA）对Qwen2.5-7B-Instruct进行微调，涵盖三个任务：评审生成、审稿人更新和修订周期预测。我们的关键发现是，仅响应损失掩蔽是先决条件，而非优化：没有它，准确率为62.0%，低于多数基线；有了它，FirstPass在预测编辑结果（标准与扩展修订周期）时实现了80.5%的准确率和F1-macro 78.2%，超越了Gemini-3.1-flash-lite-preview零-shot 10.4个百分点，并在统计上显著优于所有基线（McNemar p < 0.001）。在生成方面，FirstPass生成的评审平均为1,187个单词，明显接近人类参考（2,155个单词），在ROUGE-L上达到0.154，相较于Qwen和DeepSeek零-shot有显著提升（p < 0.001）。作为预提交循环中的预期科学合著者，FirstPass模拟专家批评并在提交前预测修订周期结果，为作者提供可信同事所能提供的判断，并在五个学科中展现出一致的跨领域表现。
cs.CL / 10 / 2606.20770
Beyond 'One Language, One Script': Quantifying Orthographic Bias in Multilingual VLMs with PuMVR
超越‘一种语言，一种文字’：通过PuMVR量化多语言视觉语言模型中的正字法偏见
Abstract
Current Vision-Language Models (VLMs) are celebrated for their multilingual capabilities, yet they operate under a flawed assumption: that one language corresponds to a single writing system. This overlooks billions of users of multi-script languages like Punjabi, Serbian, Hindi-Urdu, Kurdish, among many others, for whom a model's capability may be fractured by orthographic bias. We introduce PuMVR (Punjabi Multimodal Visual Reasoning), the first benchmark designed to quantify script-dependent bias through 375 culturally grounded image-reasoning tasks across Punjabi's three active scripts (Gurmukhi, Shahmukhi, Roman). Evaluating 10 state-of-the-art VLMs, we expose a substantial Script Gap: models frequently solve visual puzzles in one script while failing identical tasks in another, with accuracy deltas reaching 16% and Script Consistency Rates (SCR) as low as 24.8%. Crucially, visual input boosts absolute performance but does not close this gap, the relative bias persists. Our analysis suggests reasoning patterns show limited cross-script transferability, and Chain-of-Thought pathways diverge based on script alone. We propose SCR as a core metric for script-agnostic evaluation, challenging current multilingual assessment paradigms and providing a framework for equitable AI.
Chinese Translation
当前的视觉语言模型（VLMs）因其多语言能力而备受赞誉，但它们在一个错误的假设下运作：即一种语言对应于一种书写系统。这忽视了数十亿使用多种书写系统语言的用户，如旁遮普语、塞尔维亚语、印地语-乌尔都语、库尔德语等，对于这些用户而言，模型的能力可能因正字法偏见而受到影响。我们引入了PuMVR（旁遮普多模态视觉推理），这是第一个旨在通过375个与文化相关的图像推理任务量化书写依赖偏见的基准，涵盖了旁遮普的三种活跃书写系统（古尔穆基、沙赫穆基、罗马字）。在评估10个最先进的VLM时，我们揭示了一个显著的书写差距：模型在一种书写系统中经常能够解决视觉难题，而在另一种书写系统中却无法完成相同的任务，准确率差异达到16%，书写一致性率（SCR）低至24.8%。至关重要的是，视觉输入提升了绝对性能，但并未弥补这一差距，相对偏见依然存在。我们的分析表明，推理模式在跨书写系统的可转移性方面有限，思维链路径仅基于书写系统而有所不同。我们建议将SCR作为无书写偏见评估的核心指标，挑战当前的多语言评估范式，并为公平的人工智能提供框架。
cs.CL / 11 / 2606.20873
SciLens: Multi-modal Scientific Claim Verification with Agentic Entailment and Grounding
SciLens：基于代理蕴涵与基础的多模态科学声明验证
Abstract
Scientific discovery increasingly relies on automated systems that generate hypotheses, inspect multimodal evidence, and validate claims at scale. Yet scientific claim verification is not well served by asking a vision-language model for a direct binary judgment: claims often combine numerical results, comparisons, scope qualifiers, and explanatory context, while evidence is encoded in tables and figures with distinct grounding structures. We present SciLens, an evidence-conditioned atomic entailment framework for multimodal scientific claim verification. SciLens decomposes each claim into central empirical atoms and background atoms, grounds the central atoms to modality-specific evidence witnesses, and predicts the final label with an atom-level entailment rule. For tables, atoms are grounded to rows, columns, cells, arithmetic relations, and table scope; for figures, they are grounded through panels, axes, legends, visual encodings, categories, trends, ranks, and qualifier checks. This yields a unified validation procedure in which a claim is supported only if every central empirical atom is entailed by the current evidence. On the SciClaimEval development set, SciLens achieves 79.2% macro-F1 and 63.1% pair accuracy, showing that structured agentic validation improves both evidence sensitivity and interpretability.
Chinese Translation
科学发现越来越依赖于自动化系统，这些系统生成假设、检查多模态证据并大规模验证声明。然而，科学声明验证并不适合通过视觉-语言模型直接进行二元判断：声明通常结合了数值结果、比较、范围限定词和解释性背景，而证据则以具有不同基础结构的表格和图形形式编码。我们提出了SciLens，一种基于证据条件的原子蕴涵框架，用于多模态科学声明验证。SciLens将每个声明分解为中心经验原子和背景原子，将中心原子与特定模态的证据见证进行基础连接，并通过原子级蕴涵规则预测最终标签。对于表格，原子与行、列、单元格、算术关系和表格范围进行基础连接；对于图形，则通过面板、坐标轴、图例、视觉编码、类别、趋势、排名和限定词检查进行基础连接。这产生了一种统一的验证程序，只有当当前证据蕴涵每个中心经验原子时，声明才被支持。在SciClaimEval开发集上，SciLens达到了79.2%的宏观F1值和63.1%的配对准确率，表明结构化的代理验证提高了证据的敏感性和可解释性。
cs.CL / 12 / 2606.20890
Topic-to-Timestamp Alignment by Constrained Evidence Selection
基于约束证据选择的主题与时间戳对齐
Abstract
Meeting archives are difficult to search when users remember what was discussed but not when. We study topic-to-timestamp alignment: given a natural-language topic and a timestamped meeting transcript, the goal is to return the time at which the topic is discussed. A standard RAG setup can retrieve relevant transcript excerpts, but still asks the language model to generate a timestamp, which can produce unsupported or invalid timecodes. We therefore recast timestamp prediction as constrained temporal candidate selection: the system retrieves timestamped transcript chunks, and the model selects the candidate that best grounds the topic instead of generating a timecode. On 420 topic-timestamp queries from 200 municipal meeting transcripts, this increases Recall@5 from 31.9% to 50.0%, reduces MAE from 837.0 seconds to 761.0 seconds with Mistral-7B-Instruct, and increases the number of parseable outputs from 373 to 419 of 420 queries. The results suggest that temporal grounding in long transcripts depends strongly on retrieval quality and output design, not only on the choice of the language model.
Chinese Translation
当用户记得讨论的内容但不记得具体时间时，会议档案的搜索变得困难。我们研究主题与时间戳的对齐：给定一个自然语言主题和一个带时间戳的会议记录，目标是返回讨论该主题的时间。标准的 RAG 设置可以检索相关的记录摘录，但仍然要求语言模型生成一个时间戳，这可能会产生不支持或无效的时间代码。因此，我们将时间戳预测重新表述为约束的时间候选选择：系统检索带时间戳的记录片段，模型选择最能支持该主题的候选项，而不是生成时间代码。在来自 200 个市政会议记录的 420 个主题-时间戳查询中，这将 Recall@5 从 31.9% 提高到 50.0%，将 MAE 从 837.0 秒减少到 761.0 秒（使用 Mistral-7B-Instruct），并将可解析输出的数量从 373 增加到 420 个查询中的 419 个。结果表明，长记录中的时间基础依赖于检索质量和输出设计，而不仅仅是语言模型的选择。
cs.CL / 13 / 2606.20897
PeerCheck: Enhancing LLM-Generated Academic Reviews Towards Human-Level Quality
PeerCheck：提升大型语言模型生成的学术评审至人类水平的质量
Abstract
As academic submissions grow, the traditional peer review process struggles to keep up, raising concerns about quality and fairness. A trend of using large language models (LLMs) for assistance has emerged. In this work, we take a critical step toward improving the quality of LLM-generated reviews. We propose the PeerCheck framework, which investigates LLM-human review differences (RQ1) and explores methods to improve LLM-generated review quality (RQ2). We first analyzed the human-written reviews with reviews generated by various LLMs and found that LLMs and humans focus on different terms, e.g., LLMs prioritize theory while humans emphasize methodology and experiments. We further adopt prompt engineering, such as Chain-of-Thought (CoT), and utilize retrieval-augmented generation (RAG) to enhance the LLM-generated reviews towards human-level quality. We find CoT significantly improves the quality of LLM reviews, while we discover an unexpected "RAG paradox," i.e., experiments with RAG produce different results for various LLMs and, in some cases, even reduce review quality. Our comprehensive analysis of LLM-generated academic reviews illustrates both possibilities and limitations, contributing to a more effective, human-aligned review system. Our dataset is available on https://github.com/TrustAIRLab/PeerCheck.
Chinese Translation
随着学术投稿数量的增加，传统的同行评审过程难以跟上，导致对质量和公平性的担忧。利用大型语言模型（LLMs）进行辅助的趋势逐渐显现。在本研究中，我们迈出了改善LLM生成评审质量的重要一步。我们提出了PeerCheck框架，研究LLM与人类评审之间的差异（RQ1），并探索提高LLM生成评审质量的方法（RQ2）。我们首先分析了人类撰写的评审与由各种LLM生成的评审，发现LLM和人类关注的术语不同，例如，LLM更重视理论，而人类则强调方法论和实验。我们进一步采用提示工程技术，如思维链（Chain-of-Thought, CoT），并利用检索增强生成（Retrieval-Augmented Generation, RAG）来提升LLM生成的评审质量至人类水平。我们发现CoT显著提高了LLM评审的质量，而我们意外地发现了“RAG悖论”，即使用RAG的实验对不同的LLM产生了不同的结果，在某些情况下甚至降低了评审质量。我们对LLM生成的学术评审进行了全面分析，揭示了其可能性和局限性，为更有效、与人类对齐的评审系统做出了贡献。我们的数据集可在 https://github.com/TrustAIRLab/PeerCheck 获取。
cs.CL / 14 / 2606.20900
Storyline Trees: Hierarchical Representations for Long-Form Narratives
故事线树：长篇叙事的层次化表示
Abstract
Long-form narratives are challenging for long-context models because their structure is implicit: events, characters, and plotlines interact across hundreds of pages without the explicit cues that guide navigation in structured documents. We address this by constructing storyline trees, hierarchical representations that organize narratives from global themes and major plotlines to fine-grained events. We first segment chapters into contiguous narrative segments, or scenes, and use them as the basic units for tree construction. We then infer storyline trees through complementary top-down and bottom-up procedures that derive, refine, cluster, and summarize storylines at multiple levels of abstraction. We showcase the utility of this representation for question answering: storyline trees enable adaptive retrieval, allowing models to iteratively inspect high-level narrative structure and retrieve scene-level evidence on demand. Experiments on three long-context narrative QA benchmarks show that adaptive retrieval outperforms strong baselines, including post-trained long-context models and agentic chunk-based methods. Ablations confirm that scenes are more effective basic units than chapters or generic segmentation, and that gains persist under matched retrieval budgets
Chinese Translation
长篇叙事对于长上下文模型来说是一个挑战，因为其结构是隐含的：事件、角色和情节在数百页中相互作用，而没有明确的线索来指导在结构化文档中的导航。我们通过构建故事线树来解决这个问题，故事线树是从全球主题和主要情节到细粒度事件的层次化叙事表示。我们首先将章节分割成连续的叙事片段或场景，并将其作为树构建的基本单元。然后，我们通过互补的自上而下和自下而上的程序推断故事线树，这些程序在多个抽象层次上推导、细化、聚类和总结故事线。我们展示了这种表示在问答中的实用性：故事线树支持自适应检索，使模型能够迭代检查高层次的叙事结构，并按需检索场景级证据。在三个长上下文叙事问答基准上的实验表明，自适应检索优于强基线，包括后训练的长上下文模型和基于代理的块方法。消融实验确认，场景作为基本单元比章节或通用分割更有效，并且在匹配的检索预算下，性能提升依然存在。
cs.CL / 15 / 2606.20911
Latent Personal Memory: Represent personal memory as dynamic soft prompts
潜在个人记忆：将个人记忆表示为动态软提示
Abstract
Personalizing large language models (LLMs) requires encoding long-term, user-specific behavioral patterns in a way that is computationally efficient, scalable, and compatible with a frozen base model. We present Latent Personal Memory (LPM), a scalable framework that represents user-specific history as a compact, persistent matrix of N latent slots, that are interpretable. A shared cross-attention projection network maps these slots into dynamic, input-conditioned soft prompts that are prepended to the input of a frozen LLM. We evaluate LPM on PersonaMem v1 and LoCOMO benchmarks across Qwen3-1.7B, 4B, and 8B backbones. Results demonstrate that LPM outperforms LoRA and Prompt Tuning by up to 8.8% and 54.4% in overall accuracy respectively on PersonaMem v1, while reducing KV-cache usage by over 64x. On LoCoMo, LPM matches LoRA accuracy with 120x fewer trainable parameters. We also show that the efficiency of LPM grows with context length and outperforms full-context at 128K context length.
Chinese Translation
个性化大型语言模型（LLMs）需要以计算高效、可扩展且与冻结基础模型兼容的方式编码长期的用户特定行为模式。我们提出了潜在个人记忆（Latent Personal Memory, LPM），这是一个可扩展的框架，将用户特定的历史表示为一个紧凑、持久的 N 个潜在槽的矩阵，并且这些槽是可解释的。一个共享的交叉注意力投影网络将这些槽映射为动态的、输入条件的软提示，这些软提示被预先添加到冻结的 LLM 的输入中。我们在 PersonaMem v1 和 LoCOMO 基准上评估了 LPM，使用了 Qwen3-1.7B、4B 和 8B 的基础模型。结果表明，LPM 在 PersonaMem v1 上的整体准确率分别比 LoRA 和 Prompt Tuning 提高了最多 8.8% 和 54.4%，同时将 KV-cache 的使用减少了超过 64 倍。在 LoCoMo 上，LPM 在可训练参数减少 120 倍的情况下达到了与 LoRA 相同的准确率。我们还展示了 LPM 的效率随着上下文长度的增加而增长，并在 128K 上下文长度时超越了全上下文的表现。
cs.CL / 16 / 2606.20929
Peeking Inside LLMs: Leveraging Internal Artifacts of LLMs for Enhancing Reliability in Legal Classification
窥视大型语言模型内部：利用大型语言模型的内部特征提升法律分类的可靠性
Abstract
Large Language Models (LLMs) are increasingly being adopted in the legal domain. However, despite their strong performance, LLMs are prone to generating incorrect or hallucinated outputs, raising serious concerns about their reliability in high-stakes domains such as law. Detecting the correctness of responses of LLM-based systems is therefore a critical challenge. In this work, we explore the potential of leveraging internal artifacts of LLM to detect the correctness of their predictions in legal-domain classification tasks. We develop approaches that utilize features derived from these internal artifacts to build downstream classifiers capable of identifying incorrect LLM outputs. We evaluate our approach on two representative legal classification tasks: bail decision prediction and statute violation prediction. Our experimental results demonstrate that LLMs' internal artifacts are reliable indicators for detecting incorrect predictions in legal classification tasks, and can be applied to enhance the reliability of LLM-based classification systems.
Chinese Translation
大型语言模型（LLMs）在法律领域的应用日益增多。然而，尽管其表现强劲，LLMs仍然容易生成不正确或虚构的输出，这在法律等高风险领域引发了对其可靠性的严重担忧。因此，检测基于LLM系统的响应的正确性成为一个关键挑战。在本研究中，我们探讨了利用LLM内部特征来检测其在法律领域分类任务中预测的正确性的潜力。我们开发了利用这些内部特征衍生的特征构建下游分类器的方法，这些分类器能够识别不正确的LLM输出。我们在两个具有代表性的法律分类任务上评估了我们的方法：保释决定预测和法规违反预测。我们的实验结果表明，LLMs的内部特征是检测法律分类任务中不正确预测的可靠指标，并且可以应用于提升基于LLM的分类系统的可靠性。
cs.CL / 17 / 2606.20936
Comparing Transformers and Hybrid Models at the Token Level
在标记级别比较变换器和混合模型
Abstract
Hybrid language models that mix attention and recurrent layers have shown promise: theoretically, recurrent layers ameliorate the limitations of pure transformers on state tracking, and empirically, hybrids can outperform pure transformers in loss and downstream evaluations \citep{waleffe2024empirical,merrill2026olmohybrid}. Yet it remains unclear which data or capabilities drive these gains, and to what degree they reflect the theoretical advantages motivating hybrid models. We address this question using the open weights from Olmo 3 \citep{olmo2025olmo3} and Olmo Hybrid \citep{merrill2026olmohybrid}: we compare the loss of a matched transformer and hybrid at the same target tokens under the same prefixes, stratifying the results by natural token tags, copy features, delimiter structure, and controlled synthetic probes. The hybrid has lower loss on most tag families, but the gains are not uniform: they are largest for open-class content words and smaller for many closed-class function words. Across prose, code, and markup, the hybrid's loss advantage is larger on opening delimiters than on the corresponding closing delimiters, and nearly vanishes on repeated $n$-grams. Synthetic probes show the same split: the hybrid is favored on pronoun-memory and entity-tracking tasks, whereas the transformer is favored on bracket-matching tasks that require choosing closing delimiters. These patterns suggest that the recurrent layers in hybrids improve predictions that leverage the semantic state of a document, whereas attention helps on tokens predictable by $n$-gram copying or syntactic bracket matching. We conclude with proof-of-concept filtered evaluations showing how token-level decompositions can sharpen pretraining diagnostics for hybrid architectures.
Chinese Translation
混合语言模型通过结合注意力层和递归层展现出良好的前景：理论上，递归层改善了纯变换器在状态跟踪方面的局限性，而实证研究表明，混合模型在损失和下游评估中可以超越纯变换器 [c{waleffe2024empirical,merrill2026olmohybrid}。然而，目前尚不清楚哪些数据或能力驱动了这些提升，以及这些提升在多大程度上反映了推动混合模型的理论优势。我们使用 Olmo 3 的开放权重 [c{olmo2025olmo3} 和 Olmo Hybrid [c{merrill2026olmohybrid} 来解决这个问题：我们比较了在相同前缀下，匹配的变换器和混合模型在相同目标标记上的损失，并根据自然标记标签、复制特征、分隔符结构和控制的合成探针对结果进行了分层。混合模型在大多数标签类别上表现出较低的损失，但这些提升并不均匀：对于开放类内容词的提升最大，而对于许多闭合类功能词的提升较小。在散文、代码和标记中，混合模型在开头分隔符上的损失优势大于相应的闭合分隔符，并且在重复的 $n$-grams 上几乎消失。合成探针显示出相同的分裂：混合模型在代词记忆和实体跟踪任务中更具优势，而变换器在需要选择闭合分隔符的括号匹配任务中更具优势。这些模式表明，混合模型中的递归层改善了利用文档语义状态的预测，而注意力则有助于通过 $n$-gram 复制或句法括号匹配可预测的标记。我们以概念验证的过滤评估结束，展示了标记级分解如何为混合架构的预训练诊断提供更清晰的视角。
cs.CL / 18 / 2606.20946
Scaling Diverse Language Generation for 3D Visual Grounding
扩展多样化语言生成以实现3D视觉定位
Abstract
Developing robust models for 3D visual grounding (3DVG), the localization of entities in a 3D scene described in natural language, is important for enabling agents to correspond spatial language with objects in the physical world. However, the lack of diverse descriptions at scale prevents models from generalizing beyond simple linguistic patterns. Recent such attempts lack diversity in the constraint types and language used to ground objects. Captioning methods cannot precisely contrast objects, which is important for visual grounding. We therefore propose ViGiL3D++, a scalable, scene-agnostic method that generates diverse visual grounding queries by combining constraint sampling in scene graphs with the language generation of LLMs. We show that it has greater diversity over existing scaled datasets and improves model performance over several 3DVG benchmarks but also illuminates outstanding limitations of VLMs.
Chinese Translation
开发用于3D视觉定位（3DVG）的稳健模型，即在自然语言中描述的3D场景中定位实体，对于使代理能够将空间语言与物理世界中的对象相对应至关重要。然而，缺乏大规模多样化描述阻碍了模型超越简单语言模式的泛化。最近的尝试在约束类型和用于定位对象的语言方面缺乏多样性。图像描述方法无法精确区分对象，而这对于视觉定位至关重要。因此，我们提出了ViGiL3D++，这是一种可扩展的、场景无关的方法，通过将场景图中的约束采样与大型语言模型（LLMs）的语言生成相结合，生成多样化的视觉定位查询。我们展示了它在现有扩展数据集上具有更大的多样性，并在多个3DVG基准测试中提高了模型性能，但也揭示了视觉语言模型（VLMs）的显著局限性。
cs.CL / 19 / 2606.20954
Learning What Not to Forget: Long-Horizon Agent Memory from a Few Kilobytes of Learning
学习不忘记的内容：来自几千字节学习的长时间代理记忆
Abstract
Long-running language-model systems accumulate interaction history that outgrows the context window, so they must continually evict. When an eviction policy drops a load-bearing detail, for example an access token issued at login or a path the next call needs, the action fails. We present LRE (Learned Relevance Eviction), a few kilobytes, CPU-only, language-model-free scorer that learns which units of history are load-bearing and keeps them by verbatim extraction. Under a matched-budget comparison, in our experiment, no baseline dominates LRE on the accuracy-cost plane. On agents, LRE matches the accuracy of keeping the entire history overall. On the simplest tasks, it exceeds that no-eviction baseline by 27%, while requiring zero compressor calls and reducing peak context size by up to 52%. A controlled study trace shows LRE completes tasks where the others loop, finishing one such task in 37% fewer calls than keeping everything and solving 14 tasks where no other run policy does. On conversational memory, LRE outranks dense and token-pruning encoders at zero neural cost. In downstream evaluation, LRE gives the best budgeted answer quality on LoCoMo reading 68% fewer tokens. Its supervision can also be annotation-free: training only on the system's own behavior recovers 95% of the supervised scorer's effectiveness. We argue that, because memory eviction in LLM agents is a fidelity problem, it requires a deployable proactive policy where the future query is unavailable and exact state is decisive, and that cheap learned relevance can be sufficient.
Chinese Translation
长时间运行的语言模型系统积累的交互历史超出了上下文窗口，因此必须不断进行驱逐。当驱逐策略丢弃了一个关键细节，例如在登录时发出的访问令牌或下一个调用所需的路径时，操作会失败。我们提出了LRE（学习相关性驱逐），这是一个仅需几千字节、仅使用CPU、无语言模型的评分器，能够学习哪些历史单元是关键的，并通过逐字提取来保留它们。在匹配预算的比较中，在我们的实验中，没有基线在准确性-成本平面上超越LRE。在代理上，LRE的准确性与保留整个历史记录的准确性相匹配。在最简单的任务中，它的表现比无驱逐基线高出27%，同时不需要任何压缩器调用，并将峰值上下文大小减少了多达52%。一个受控研究追踪显示，LRE完成了其他方法循环的任务，在完成一个这样的任务时比保留所有内容少调用了37%的次数，并解决了14个其他运行策略无法完成的任务。在对话记忆方面，LRE在零神经成本下优于密集和令牌修剪编码器。在下游评估中，LRE在LoCoMo上提供了最佳的预算答案质量，阅读的令牌数量减少了68%。它的监督也可以是无注释的：仅通过系统自身的行为进行训练即可恢复监督评分器95%的有效性。我们认为，由于LLM代理中的记忆驱逐是一个保真度问题，因此需要一种可部署的主动策略，在未来查询不可用且确切状态至关重要的情况下，廉价的学习相关性可能是足够的。
cs.CL / 20 / 2606.20993
Phonemes to the Rescue: Multilingual Tokenization Based on International Phonetic Alphabet
音素救援：基于国际音标的多语言分词
Abstract
Multilingual language models often exhibit performance disparities across languages that can arise as early as the tokenization stage. Widely-used subword tokenization approaches favor high-resource languages, and tokenizer-free methods still yield longer sequences for scripts with a higher bytes-per-character ratio. To address these shortcomings, we propose to use the International Phonetic Alphabet (IPA) as a language-agnostic input representation for multilingual tokenizers. IPA provides a compact symbol inventory, greater cross-lingual character overlap, and a more balanced byte-per-character distribution across languages. We train matched pairs of text vs. IPA subword tokenizers across 24 languages and 14 scripts and demonstrate that IPA tokenizers consistently improve tokenization quality, especially for non-Latin scripts, and generalize more effectively to unseen languages and scripts.
Chinese Translation
多语言模型在不同语言之间常常表现出性能差异，这种差异可能在分词阶段就开始显现。广泛使用的子词分词方法偏向于高资源语言，而无分词器的方法在字符每字节比率较高的书写系统中仍会产生更长的序列。为了解决这些不足，我们提出使用国际音标（IPA）作为多语言分词器的语言无关输入表示。IPA提供了紧凑的符号库、更大的跨语言字符重叠以及更均衡的字节每字符分布。我们在24种语言和14种书写系统中训练了文本与IPA子词分词器的匹配对，并证明IPA分词器始终提高了分词质量，特别是在非拉丁书写系统中，并且在未见过的语言和书写系统中更有效地进行泛化。
cs.CL / 21 / 2606.21008
The Metanym Game: A Self-Contained, Self-Consistent LLM Peer-Community Benchmark for Structural Intelligence
隐喻游戏：一个自给自足、自洽的 LLM 同行社区基准，用于结构智能的评估
Abstract
The metanym game is a competitive word game for LLMs that measures structural intelligence against established cognitive-science constructs. No content is given in advance; the contestants create all of it -- a new kind of analogy test, analogical production falsifiable sentence by sentence, with no fixed test set to leak into training (contamination-resistant by construction). In the council-of-peers benchmark, the contestants also rate each other's creations. We introduce the first spectral solution, to our knowledge, to the wicked problem of benchmarking LLMs' factual accuracy without golden keys or oracle models: one singular value decomposition of the evaluators' ratings matrix yields their competence as both generators and judges of true statements at once. Competence on the subjective criteria comes from each judge's rating consistency as the yardstick shifts. The factual rating correlates with GPQA Diamond at Pearson r = 0.92. Scored separately, making and judging dissociate -- judging is the scarcer skill: the strongest generators are middling judges, the sharpest judge a mid-pack generator. To scale, the strongest players form a council that does the official benchmarking; its seats are contestable -- a stronger model earns one on the benchmark's own rating. The benchmark is entirely self-contained and self-consistent, a stable gauge over time.
Chinese Translation
隐喻游戏是一种针对 LLM 的竞争性文字游戏，旨在根据既定的认知科学构念衡量结构智能。参赛者在比赛前并未获得任何内容，所有内容均由参赛者自行创作——这是一种新型的类比测试，逐句生成可证伪的类比产出，且没有固定的测试集可供训练使用（从构建上避免了污染）。在同行评审基准中，参赛者还会对彼此的创作进行评分。我们首次提出了一种谱解决方案，旨在解决 LLM 事实准确性基准测试的棘手问题，而无需黄金钥匙或神谕模型：对评估者评分矩阵进行一次奇异值分解，即可同时得出他们作为真实陈述生成者和评判者的能力。主观标准上的能力来自于每位评审在评分一致性上的表现，评分标准的变动使得这一衡量成为可能。事实评分与 GPQA Diamond 的皮尔逊相关系数为 r = 0.92。分开评分时，创作与评判是可以区分的——评判是一项更为稀缺的技能：最强的生成者往往是中等水平的评判者，而最敏锐的评判者则是中等水平的生成者。为了扩展，最强的参与者组成一个委员会，负责正式的基准测试；其席位是可以竞争的——更强的模型可以在基准测试中根据自身评分获得席位。该基准完全自给自足且自洽，是一个随时间稳定的衡量标准。
cs.CL / 22 / 2606.21048
Event Ontology Expansion via LLM-Based Conceptualization
基于大语言模型的事件本体扩展
Abstract
Event ontology expansion aims to discover emerging event types from data and extend them to appropriate positions in the existing event ontology.. Existing methods typically cluster contextualized trigger representations and attach induced clusters to the ontology based on instance-level similarity. However, ontology expansion requires concept-level semantics that characterize event types, whereas contextualized trigger representations often conflate these semantics with surface contextual variation, leading to unstable clustering and unreliable hierarchy expansion. To address this issue, we propose ConceptE, a conceptualization-enhanced framework for event ontology expansion. ConceptE first derives concept-level semantics by prompting an LLM with the sentence and event trigger, producing a concise concept name and a natural-language description. It then jointly encodes these semantics with trigger information to build concept-enhanced representations aligned with ontology-level reasoning. This representation design supports more coherent event clustering, more reliable hierarchy expansion, and ontology-consistent type naming. Experiments on ACE, ERE, and MAVEN demonstrate that ConceptE consistently outperforms state-of-the-art approaches across all subtasks of event ontology expansion. In particular, it achieves improvements of up to 12.37\% in BCubed-F1 for event clustering and 6.48\% in Taxo\_F1 for hierarchy expansion, demonstrating the effectiveness of the proposed ConceptE method.
Chinese Translation
事件本体扩展旨在从数据中发现新兴事件类型，并将其扩展到现有事件本体中的适当位置。现有方法通常通过聚类上下文触发表示，并根据实例级相似性将诱导的聚类附加到本体上。然而，本体扩展需要能够表征事件类型的概念级语义，而上下文触发表示往往将这些语义与表面上下文变异混淆，导致聚类不稳定和层次扩展不可靠。为了解决这个问题，我们提出了ConceptE，一个增强概念化的事件本体扩展框架。ConceptE首先通过向大语言模型（LLM）提示句子和事件触发，推导出概念级语义，生成简洁的概念名称和自然语言描述。然后，它将这些语义与触发信息共同编码，以构建与本体级推理对齐的增强概念表示。这种表示设计支持更连贯的事件聚类、更可靠的层次扩展和本体一致的类型命名。在ACE、ERE和MAVEN上的实验表明，ConceptE在事件本体扩展的所有子任务中始终优于最先进的方法。特别是在事件聚类中，ConceptE在BCubed-F1上提高了多达12.37\%，在层次扩展中在Taxo_F1上提高了6.48\%，证明了所提ConceptE方法的有效性。
cs.CL / 23 / 2606.21066
Demographic Metadata as Construct-Irrelevant Noise in DistilBERT-Based Automated Essay Scoring
作为构念无关噪声的人口统计元数据在基于DistilBERT的自动化作文评分中的应用
Abstract
Automated Essay Scoring (AES) systems are increasingly used to support teachers in managing grading workloads and to provide a supplementary rater in large-scale assessments. While human grading is frequently influenced by students' demographic characteristics, the efficacy of different strategies for integrating demographic metadata with textual input used to train AES models remains underexplored. This study investigates the impact of a specific multimodal fusion strategy - naive metadata concatenation - on the predictive accuracy, training convergence, and score parity of a DistilBERT-based AES model. A comparative analysis was conducted using the ASAP 2.0 dataset to evaluate a baseline model against an experimental model trained with input that concatenates tokenised text and demographic metadata using a naive multimodal fusion strategy. Evaluated via 10-fold cross-validation, the findings reveal that the early fusion of demographic metadata and the input significantly degrades the model's overall predictive accuracy. The baseline model achieved a Quadratic Weighted Kappa (QWK) of 0.727, which dropped to 0.656 upon integrating metadata. Furthermore, the experimental model exhibited higher validation loss (1.29) compared to the baseline model (1.25). The experimental model also displayed exacerbated scoring bias, reducing score parity instances from 15 to 12 out of 19 tests.
Chinese Translation
自动化作文评分（AES）系统越来越多地被用于帮助教师管理评分工作负担，并在大规模评估中提供补充评分者。尽管人类评分常常受到学生人口统计特征的影响，但将人口统计元数据与用于训练AES模型的文本输入相结合的不同策略的有效性仍然未得到充分探讨。本研究调查了一种特定的多模态融合策略——简单元数据连接——对基于DistilBERT的AES模型的预测准确性、训练收敛性和评分一致性的影响。通过使用ASAP 2.0数据集进行比较分析，评估了基线模型与实验模型的表现，后者使用简单的多模态融合策略将标记化文本和人口统计元数据连接在一起。通过10折交叉验证评估的结果显示，人口统计元数据与输入的早期融合显著降低了模型的整体预测准确性。基线模型的二次加权Kappa（QWK）为0.727，而在整合元数据后下降至0.656。此外，实验模型的验证损失（1.29）高于基线模型（1.25）。实验模型还表现出加剧的评分偏差，使得在19次测试中，评分一致性实例从15个减少到12个。
cs.CL / 24 / 2606.21069
Quality and Agreement in Multilabel Emotion Annotation: A Case Study and Evaluation Framework
多标签情感标注中的质量与一致性：案例研究与评估框架
Abstract
Emotion annotation is inherently subjective, yet most NLP pipelines still assume "gold" labels, typically produced by majority voting, and treat annotator variation as noise. In this paper, we present a multilabel emotion annotation case study and use it to examine how annotator behavior and aggregation choices affect both agreement estimates and downstream emotion classifiers. Rather than collapsing disagreement into a single label, we represent targets as soft vote-share labels (including an intensity-weighted variant) and evaluate models using both thresholded metrics (macro-/micro-F1) and probabilistic alignment (Bernoulli cross-entropy SoftBCE), alongside data-derived disagreement diagnostics. Across annotation regimes, we show that disagreement is structured and leaves measurable traces in model behavior: hard labels may maximize F1 metrics, while soft supervision yields predictions that better reflect empirical annotator variance and uncertainty. Our results provide practical guidance for designing, aggregating, and evaluating multilabel emotion datasets when multiple interpretations are plausible.
Chinese Translation
情感标注本质上是主观的，但大多数自然语言处理（NLP）流程仍然假设存在“黄金”标签，这些标签通常通过多数投票产生，并将标注者的差异视为噪声。本文呈现了一个多标签情感标注的案例研究，并利用该研究探讨标注者行为和聚合选择如何影响一致性估计和下游情感分类器。我们并未将不一致性简化为单一标签，而是将目标表示为软投票比例标签（包括一种强度加权变体），并使用阈值指标（宏观/微观F1）和概率对齐（Bernoulli交叉熵SoftBCE）来评估模型，同时结合数据衍生的不一致性诊断。在不同的标注机制下，我们展示了不一致性是有结构的，并在模型行为中留下可测量的痕迹：硬标签可能最大化F1指标，而软监督则产生更能反映实际标注者差异和不确定性的预测。我们的结果为设计、聚合和评估多标签情感数据集提供了实用指导，尤其是在多种解释均合理的情况下。
cs.CL / 25 / 2606.21075
FiLM-Coordinated Dual-Branch Transformer for Global-Local Dependency Modeling in Language Modeling
FiLM协调的双分支变换器用于语言建模中的全局-局部依赖建模
Abstract
Standard Transformers use a single self-attention pathway to model both global dependencies and local patterns, creating tension between long-range structural reasoning and fine-grained local representation learning. We propose a FiLM-coordinated dual-branch Transformer for language modeling, where each layer explicitly contains a global branch and a local branch, and feature-wise linear modulation (FiLM) is used for dynamic cross-branch coordination instead of simple concatenation or static addition. The key idea is that the two branches represent different dependency views of the same input, making channel-wise calibration more suitable than heavy token-level interaction. We therefore design a bidirectional FiLM module in which each branch generates per-channel scaling and shifting parameters to condition the other. Experiments on multiple small-scale language modeling settings show that the proposed structure consistently outperforms same-width single-branch baselines and weakened dual-branch variants under a fixed lightweight configuration. On TinyShakespeare and a 1M-character subset of WikiText-2, the full dual-branch FiLM model achieves the best results among same-width structural baselines. Multi-seed results support the stability of the gains, while mechanistic analyses show that FiLM learns input-dependent, layer-dependent, and channel-selective modulation patterns rather than static scaling. Parameter-matched widened single-branch baselines also indicate that the current design still leaves room for improvement in parameter efficiency.
Chinese Translation
标准变换器使用单一的自注意力路径来建模全局依赖和局部模式，这在长距离结构推理和细粒度局部表示学习之间产生了紧张关系。我们提出了一种FiLM协调的双分支变换器用于语言建模，其中每一层明确包含一个全局分支和一个局部分支，并且使用特征线性调制（FiLM）进行动态跨分支协调，而不是简单的连接或静态加法。关键思想是这两个分支代表同一输入的不同依赖视角，使得通道级校准比重的标记级交互更为合适。因此，我们设计了一个双向FiLM模块，其中每个分支生成每通道的缩放和偏移参数以调节另一个分支。在多个小规模语言建模设置上的实验表明，所提出的结构在固定轻量级配置下始终优于相同宽度的单分支基线和减弱的双分支变体。在TinyShakespeare和WikiText-2的1M字符子集上，完整的双分支FiLM模型在相同宽度的结构基线中取得了最佳结果。多种种子结果支持增益的稳定性，而机制分析表明FiLM学习的是输入依赖、层依赖和通道选择的调制模式，而不是静态缩放。参数匹配的加宽单分支基线也表明当前设计在参数效率上仍有改进空间。
cs.CL / 26 / 2606.21078
A Validation-Gated Mechanistic Account of Suicidality Detection in LLMs
基于验证门控的自杀倾向检测机制分析
Abstract
Large language models are increasingly proposed for mental-health applications such as detecting suicidal content, raising the question of what they rely on. We study this mechanistically and use it to ask a narrower question: how to make a causal claim about a model's internal features more trustworthy. Our validation-gated framework, with suicidality detection as a case study, interprets a behavior only after the model is shown to perform it: a concept is admitted only once the model ranks it above a simple lexical baseline, and each subsequent property is tested against a matched control. This discipline yields negative as well as positive results. The gate rules out one task at the outset: on DeepSuiMind (Li et al. 2025), Llama-3.1-8B-Instruct cannot separate implicit suicidal intent from ordinary distress, so we do not analyze it. We turn to binary suicide detection, which it does perform. There we find a mid-network feature that appears semantic rather than keyword-based, is causally implicated in the decision (ablating it degrades the judgment; a random direction does not), is low-rank, and recurs across three model families and three suicide datasets. A register-matched control (suicide versus depression) suggests it tracks suicidality more specifically than general distress. Steering raises the model's response, but for unrelated questions too, so we treat it as necessary but not sufficient. The clearest pattern separates encoding from use: smaller models already represent suicidality, yet only larger ones appear to act on it. The positive evidence is English Reddit text, which limits the clinical reading.
Chinese Translation
大型语言模型越来越多地被提议用于心理健康应用，如检测自杀内容，这引发了它们所依赖的基础问题。我们从机制层面进行研究，并以此提出一个更狭窄的问题：如何使关于模型内部特征的因果声明更具可信性。我们的验证门控框架以自杀倾向检测为案例研究，仅在模型被证明能够执行某一行为后才对其进行解释：一个概念只有在模型将其排名高于简单的词汇基线后才被接受，并且每个后续特性都与匹配的对照组进行测试。这种严谨的方法产生了负面和正面的结果。门控机制在一开始就排除了一个任务：在 DeepSuiMind (Li et al. 2025) 上，Llama-3.1-8B-Instruct 无法将隐含的自杀意图与普通的痛苦区分开，因此我们不对其进行分析。我们转向其能够执行的二元自杀检测。在这里，我们发现一个中间网络特征似乎是语义性的而非基于关键词的，并且在决策中具有因果关联（去除它会降低判断能力；随机方向则没有影响），其秩较低，并且在三种模型家族和三个自杀数据集中反复出现。一个匹配的对照组（自杀与抑郁）表明它更具体地追踪自杀倾向而非一般痛苦。引导提高了模型的响应，但对于不相关的问题也是如此，因此我们将其视为必要但不足。最明显的模式是将编码与使用分开：较小的模型已经能够表示自杀倾向，但只有较大的模型似乎能够对此做出反应。积极的证据来自英语 Reddit 文本，这限制了临床解读。
cs.CL / 27 / 2606.21082
Scalable Hierarchical Attention Transformers for Multi-Turn Jailbreak Detection in Long Conversations
可扩展的层次注意力变换器用于长对话中的多轮越狱检测
Abstract
Multi-turn jailbreaks can evade turn-level moderation by spreading unsafe intent across a dialogue through gradual escalation, reframing, and role manipulation. We address multi-turn jailbreak detection as a conversation-level classification problem and introduce an efficient hierarchical detector that avoids expensive long-context concatenation while retaining cross-turn reasoning. The model encodes individual turns to form compact turn representations and applies a lightweight conversation module that captures dialogue dynamics and selectively attends to fine-grained evidence when needed. On a challenging evaluation benchmark of 14,038 conversations, our approach achieves an F1 of 0.9394, outperforming Claude Opus 4.7, the strongest competing baseline, by 0.07 while halving its false-positive rate. Ablation studies confirm that each architectural component contributes meaningfully, with combining cross-attention and self-attention in the conversation module yielding a 2.26 percentage point reduction in false-positive rate over the self-attention-only variant.
Chinese Translation
多轮越狱可以通过逐步升级、重新框架和角色操控在对话中传播不安全的意图，从而逃避逐轮的监管。我们将多轮越狱检测视为一个对话级别的分类问题，并引入了一种高效的层次检测器，该检测器避免了昂贵的长上下文拼接，同时保留了跨轮推理。该模型对单独的轮次进行编码，以形成紧凑的轮次表示，并应用一个轻量级的对话模块，该模块捕捉对话动态，并在需要时选择性地关注细粒度证据。在一个包含14,038个对话的挑战性评估基准上，我们的方法实现了0.9394的F1分数，超越了最强竞争基线Claude Opus 4.7，提升了0.07，同时将其误报率减半。消融研究确认每个架构组件都具有重要贡献，在对话模块中结合交叉注意力和自注意力相比仅使用自注意力的变体，误报率降低了2.26个百分点。
cs.CL / 28 / 2606.21097
GRAG: Generic Response-Augmented Generation Framework for Personalized Conversational Systems
GRAG：用于个性化对话系统的通用响应增强生成框架
Abstract
Deploying highly capable personalized conversational agents in resource-constrained or privacy-sensitive environments remains a significant challenge. We identify a fundamental bottleneck in the existing approaches: current training paradigms treat personalization and grounding as a single monolithic learning problem. Under these paradigms, language models are forced to simultaneously address what to say (content grounding) and how to say it in a user-specific way (personalization), which introduces significant computational and optimization challenges. Consequently, contextual grounding is often sacrificed for persona adherence, or vice versa, resulting in responses that are either weakly grounded in the conversational history or insufficiently personalized. In this work, we propose the Generic Response-Augmented Generation (GRAG) framework that decouples these competing objectives by leveraging offline, generic responses from high-capacity, general-purpose LLMs as a semantic and structural scaffold to guide the fine-tuning of smaller, task-specialized models seamlessly in resource-limited environments. By decoupling the content grounding from personalization, GRAG allows the model to focus exclusively on persona injection while remaining firmly anchored to the conversational context. We instantiate the GRAG in two post- and pre-fusion-based architectural variants and evaluate them on multiple benchmark conversational datasets that cover diverse personalization structures. Our results demonstrate that GRAG significantly outperforms state-of-the-art methods that do not use auxiliary scaffolding, yielding up to 47% improvements in ROUGE-2 and 36% in BLEU scores. Ultimately, GRAG offers a generalizable blueprint for building grounding-aware personalized conversational systems in resource-limited environments.
Chinese Translation
在资源受限或隐私敏感的环境中部署高能力的个性化对话代理仍然是一个重大挑战。我们识别出现有方法中的一个基本瓶颈：当前的训练范式将个性化和上下文基础视为一个单一的整体学习问题。在这些范式下，语言模型被迫同时解决说什么（内容基础）和如何以用户特定的方式说（个性化）的问题，这引入了显著的计算和优化挑战。因此，上下文基础通常会为了遵循个性而被牺牲，反之亦然，导致生成的响应要么与对话历史的基础联系较弱，要么个性化不足。在本研究中，我们提出了通用响应增强生成（GRAG）框架，通过利用来自高容量、通用大型语言模型（LLMs）的离线通用响应，解耦这些相互竞争的目标，作为语义和结构的支架，以指导在资源有限环境中无缝微调较小的任务专用模型。通过将内容基础与个性化解耦，GRAG使模型能够专注于个性注入，同时保持与对话上下文的紧密联系。我们在两种基于后融合和前融合的架构变体中实例化GRAG，并在多个涵盖多样化个性化结构的基准对话数据集上进行评估。我们的结果表明，GRAG显著优于不使用辅助支架的最先进方法，在ROUGE-2和BLEU分数上分别提高了多达47%和36%。最终，GRAG为在资源有限环境中构建关注基础的个性化对话系统提供了一个可推广的蓝图。
cs.CL / 29 / 2606.21098
LLM-Based Multi-Reference Evaluation for Efficient and Robust Assessment of Phrase Break Annotations
基于LLM的多参考评估用于高效且稳健的短语断句注释评估
Abstract
Reliable evaluation of phrase break annotations is crucial, as subtle variations in prosodic boundaries directly affect the clarity and naturalness of speech. However, existing approaches exhibit major limitations: single-reference evaluation assumes a unique gold phrasing for an utterance despite multiple valid phrasings, while human judgment, though flexible, is labor-intensive and unscalable. To address these, we propose LLM-based Multi-Reference Evaluation (LMRE) for phrase break annotations that models the one-to-many nature of prosodic phrasing and generates multiple valid phrasings from minimal demonstrations. On a Korean testbed of 1,356 annotations covering five strategies, LMRE shows stronger alignment with human judgment than single-reference evaluation in both acceptance behavior and score correlation. Our findings demonstrate that LMRE effectively achieves both scalability and multi-reference support, highlighting the potential of LLMs for evaluation in the speech domain.
Chinese Translation
短语断句注释的可靠评估至关重要，因为韵律边界的细微变化直接影响语音的清晰度和自然性。然而，现有的方法存在重大局限性：单参考评估假设一个话语只有唯一的黄金短语，而忽视了多个有效短语的可能性；而人类判断虽然灵活，但劳动强度大且难以扩展。为了解决这些问题，我们提出了基于LLM的多参考评估（LMRE）用于短语断句注释，该方法建模了韵律短语的一对多特性，并从最少的示例中生成多个有效短语。在一个涵盖五种策略的1,356条注释的韩语测试集上，LMRE在接受行为和评分相关性方面与人类判断的对齐程度优于单参考评估。我们的研究结果表明，LMRE有效实现了可扩展性和多参考支持，突显了LLM在语音领域评估中的潜力。
cs.CL / 30 / 2606.21123
A Multi-Agent Audit Framework for High-Stakes Reasoning: Evaluation and Interpretability in Clinical Mental Health Screening
高风险推理的多智能体审计框架：临床心理健康筛查中的评估与可解释性
Abstract
High-stakes reasoning tasks necessitate transparent and verifiable workflows, yet conventional single-model large language models (LLMs) often struggle with hallucination and low interpretability under zero-shot paradigms. To address this general AI challenge, we propose a Multi-Agent Audit Framework that simulates a collaborative, multi-step verification process. We empirically validate this architecture in the sensitive domain of clinical mental health screening using a modular LangChain workflow. Our framework decomposes the reasoning process into a Perception Agent, Knowledge Retrieval-Augmented Generation (RAG), Chain-of-Thought (CoT) clinical inference, and a critical Audit verification stage. We evaluated this framework on the DAIC-WOZ dataset using locally deployed open-source models. Experimental results demonstrate that our multi-agent pipeline significantly outperforms single-agent baselines, reducing the Mean Absolute Error (MAE) for PHQ-8 depression severity prediction from 5.35 to 5.02. By exposing cross-agent validation traces, the framework mitigates reasoning drift and provides highly interpretable diagnostic rationales, offering a generalizable paradigm for reliable AI-assisted decision support beyond isolated model scaling. We make data and code open access on GitHub for replicability.
Chinese Translation
高风险推理任务需要透明且可验证的工作流程，然而传统的单模型大型语言模型（LLMs）在零样本范式下常常面临幻觉和低可解释性的问题。为了解决这一普遍的人工智能挑战，我们提出了一种多智能体审计框架，该框架模拟了一个协作的多步骤验证过程。我们在临床心理健康筛查这一敏感领域中，使用模块化的LangChain工作流程对该架构进行了实证验证。我们的框架将推理过程分解为感知智能体、知识检索增强生成（RAG）、思维链（CoT）临床推理和关键审计验证阶段。我们在DAIC-WOZ数据集上使用本地部署的开源模型对该框架进行了评估。实验结果表明，我们的多智能体管道显著优于单智能体基线，将PHQ-8抑郁严重性预测的平均绝对误差（MAE）从5.35降低至5.02。通过揭示跨智能体验证痕迹，该框架减轻了推理漂移，并提供了高度可解释的诊断理由，为可靠的人工智能辅助决策支持提供了一个可推广的范式，超越了孤立模型的扩展。我们在GitHub上公开数据和代码，以便于复现。
cs.CL / 31 / 2606.21144
AdaMem: Learning What to Remember for Personalized Long-Horizon LLM Agents
AdaMem：为个性化长时段大型语言模型代理学习记忆内容
Abstract
Long-term memory systems for Large Language Model (LLM) agents typically try to \emph{remember everything}, extracting memories uniformly to retain as many facts as possible. In production, however, inference cost and finite context budgets make this untenable: beyond consolidating raw dialogue into memory, an agent must exert \emph{write control}, efficiently keeping only the information each user actually cares about. Otherwise, long-horizon personalized interactions suffer \emph{memory bloat}, where irrelevant trivia crowds out useful information and steadily erodes question-answering (QA) accuracy. We argue that what is worth remembering is role-dependent, and propose \textbf{AdaMem} (Adaptive Memory), a method that \emph{learns what to remember} for each user from feedback. AdaMem maintains a structured, role-specific Memory Policy and refines it from weekly QA feedback through a lightweight, patch-style self-reflection step with failure rollback. To study this setting, we build \textbf{AdaMem-Bench}, a benchmark that simulates weeks of interaction with week-by-week QA. Across two extraction models and two feedback modes, AdaMem improves QA accuracy by up to \textbf{+9.0\%} over the uniform Mem0 baseline while shrinking memory volume by \textbf{9\%}.
Chinese Translation
大型语言模型（LLM）代理的长期记忆系统通常试图 extit{记住一切}，均匀提取记忆以保留尽可能多的事实。然而，在实际应用中，推理成本和有限的上下文预算使得这种做法不可行：除了将原始对话整合到记忆中，代理还必须施加 extit{写控制}，高效地仅保留每个用户真正关心的信息。否则，长时段个性化交互将遭遇 extit{记忆膨胀}，无关的琐事挤占有用信息，逐渐降低问答（QA）准确性。我们认为，值得记住的内容依赖于角色，并提出 extbf{AdaMem}（自适应记忆），一种从反馈中 extit{学习记住什么}的方法。AdaMem维护一个结构化的、角色特定的记忆策略，并通过轻量级的、补丁式的自我反思步骤（带有失败回滚）从每周的QA反馈中对其进行优化。为了研究这一设置，我们构建了 extbf{AdaMem-Bench}，一个模拟数周交互的基准，每周进行QA。在两个提取模型和两种反馈模式下，AdaMem在QA准确性上比均匀的Mem0基线提高了高达 extbf{+9.0\%}，同时减少了记忆体积 extbf{9 ext{%}}。
cs.CL / 32 / 2606.21155
Who Checks the Citations? Benchmarking Legal Hallucination Detection
谁来检查引用？法律幻觉检测的基准测试
Abstract
Attorneys, judges, and pro se filers increasingly use AI to draft legal documents, yet these tools frequently fabricate citations. Despite predictions that newer models would hallucinate less or that court sanctions would deter negligent filers, we found over 1,000 filings containing fabricated citations -- with this number growing year-over-year. This study evaluates whether AI-based systems can mitigate these errors by automatically detecting hallucinations. We propose a taxonomy of legal citation hallucinations grounded in actual court filings and introduce a dataset of 1,300 brief excerpts containing injected errors. Benchmarking five models in agentic and non-agentic settings reveals that while the latest iterations perform better -- GPT-5 achieves 82.8% recall and a 60.5% F1 score in an agentic framework -- all models struggle with subtle error categories. Agentic verification remains resource-intensive, with GPT-5 averaging 16.9 steps per excerpt. Furthermore, restricted information access limits the efficacy of even the best agents. This gap creates policy concerns, as it disadvantages both AI systems and litigants who lack subscriptions to commercial legal databases. Together, our dataset, tools, and policy recommendations provide a foundation for building and auditing reliable legal citation checking tools.
Chinese Translation
律师、法官和自我代理的当事人越来越多地使用人工智能来撰写法律文件，但这些工具经常伪造引用。尽管有预测认为更新的模型会减少幻觉现象，或者法院制裁会威慑疏忽的提交者，但我们发现超过1,000份文件中包含伪造的引用，并且这一数字逐年增长。本研究评估基于人工智能的系统是否能够通过自动检测幻觉来减少这些错误。我们提出了一种基于实际法院文件的法律引用幻觉分类法，并引入了一个包含1,300个注入错误的简要摘录的数据集。在代理和非代理环境中对五个模型进行基准测试显示，尽管最新版本的表现更好——GPT-5在代理框架中达到了82.8%的召回率和60.5%的F1分数——所有模型在处理细微错误类别时都面临挑战。代理验证仍然资源密集，GPT-5每个摘录平均需要16.9个步骤。此外，受限的信息访问限制了即使是最佳代理的有效性。这一差距引发了政策关注，因为它使得缺乏商业法律数据库订阅的人工智能系统和诉讼当事人处于不利地位。我们的数据集、工具和政策建议共同为构建和审计可靠的法律引用检查工具提供了基础。
cs.CL / 33 / 2606.21168
Dementia-Agents: A Multi-Modal Multi-Agent System for Dementia Staging and Phenotyping
痴呆代理：一种用于痴呆分期和表型分析的多模态多代理系统
Abstract
Dementia diagnosis requires integrating multi-modal clinical assessments from diverse informants and clinicians under incomplete and heterogeneous data conditions. Yet most AI-driven approaches remain Alzheimer's disease (AD)-centric, framing the problem as binary AD detection or three-stage AD progression modeling within well-curated research settings. This pathology-driven paradigm overlooks the broader, syndrome-level nature of dementia, which spans multiple stages, phenotypes, and etiologies. In this paper, we propose Dementia-Agents, a clinically aligned multi-agent framework for real-world dementia staging and phenotyping. The framework follows a three-step workflow: (1) a data agent translates structured clinical records into semantically faithful textual representations that preserve missing-data signals and routes them to domain-aligned experts; (2) five fine-tuned expert agents generate domain-level predictions; and (3) a coordinator agent performs probabilistic aggregation to produce final staging and phenotyping decisions. We develop and evaluate Dementia-Agents on a real-world clinical cohort of 1,066 patients from two cognitive neurology services. Compared with monolithic multi-modal large language models (MLLMs) and prior medical multi-agent systems, our approach achieves consistent improvements in diagnostic performance for real-world syndrome-level dementia staging and phenotyping, while preserving domain-level interpretability.
Chinese Translation
痴呆的诊断需要在不完整和异质的数据条件下，整合来自不同信息提供者和临床医生的多模态临床评估。然而，大多数基于人工智能的方法仍然以阿尔茨海默病（AD）为中心，将问题框定为二元的AD检测或在经过精心策划的研究环境中进行的三阶段AD进展建模。这种以病理为驱动的范式忽视了痴呆的更广泛的综合征层面特征，痴呆涵盖多个阶段、表型和病因。在本文中，我们提出了痴呆代理（Dementia-Agents），这是一个与临床对齐的多代理框架，用于现实世界中的痴呆分期和表型分析。该框架遵循三步工作流程：（1）数据代理将结构化临床记录转换为语义上忠实的文本表示，保留缺失数据的信号，并将其传递给领域对齐的专家；（2）五个经过微调的专家代理生成领域级预测；（3）协调代理进行概率聚合，以产生最终的分期和表型决策。我们在来自两个认知神经学服务的1066名患者的真实临床队列上开发并评估了痴呆代理。与单一的多模态大型语言模型（MLLMs）和先前的医学多代理系统相比，我们的方法在现实世界综合征层面的痴呆分期和表型分析的诊断性能上实现了一致的改进，同时保持了领域级的可解释性。
cs.CL / 34 / 2606.21195
Beyond Hooking Onto the World: Referential Profiles and the Numerical Structure of LLM Grounding
超越与世界的连接：大型语言模型的指称特征与数值结构
Abstract
This paper revisits the grounding problem for large language models in light of recent vector-grounding accounts. I accept the shift from classical symbol grounding to vector grounding, but argue that the current debate remains incomplete in two respects. First, reference is often treated too thinly, as if it were a fixed link between an isolated expression and an object. I argue instead that reference is profile-based, context-sensitive, discourse-level, affectively shaped, and norm-governed. Even in the human case, reference is publicly stabilized through patterns of use, correction, distinction, inference, and continuation rather than through identical private representations. Second, vector grounding requires an account of numerical realization. LLMs do not acquire reference through human perception, memory, intention, embodiment, or understanding. Rather, through optimization, they parameterize linguistic traces of human world-directed practice. In a finite vector system, referential profiles must be distributed, may be superposed, and are recovered through context-sensitive computation. Weights, activations, attention-mediated hidden states, softmax-trained contrasts, and inner-product alignments are the mathematical sites at which inherited linguistic relations become stable and causally active. Mechanistic interpretability findings, including entity-like features, knowledge neurons, and emotion-related activation directions, provide indirect support for this view. They do not show that LLMs possess human reference. They support a more limited thesis: LLMs may possess derivative, language-mediated, profile-based, and numerically structured forms of reference.
Chinese Translation
本文在近期向量基础的指称理论背景下重新审视大型语言模型的基础问题。我接受从经典符号基础到向量基础的转变，但认为当前的争论在两个方面仍然不够完整。首先，指称常常被过于简单地处理，仿佛它是孤立表达与对象之间的固定联系。我认为指称是基于特征的、上下文敏感的、话语层面的、情感塑造的，并且受到规范的约束。即使在人的案例中，指称也是通过使用、纠正、区分、推理和延续的模式而不是通过相同的私人表征来公开稳定的。其次，向量基础需要对数值实现进行解释。大型语言模型并不是通过人类的感知、记忆、意图、具身性或理解来获得指称。相反，通过优化，它们对人类面向世界的实践的语言痕迹进行参数化。在有限的向量系统中，指称特征必须是分布式的，可以重叠，并通过上下文敏感的计算来恢复。权重、激活、注意力介导的隐藏状态、softmax训练的对比以及内积对齐是继承的语言关系变得稳定和因果活跃的数学场所。机械可解释性研究结果，包括类实体特征、知识神经元和情感相关的激活方向，间接支持这一观点。它们并没有表明大型语言模型具有人类指称。它们支持一个更有限的论点：大型语言模型可能具备派生的、语言介导的、基于特征的和数值结构的指称形式。
cs.CL / 35 / 2606.21203
When Context Misleads: Surprisal, Energy and Attention Entropy as Metrics of Coherence Illusions in LLMs
当上下文误导时：惊讶度、能量和注意力熵作为大型语言模型中连贯性错觉的度量
Abstract
Psycholinguistics studies show that human readers fall for coherence illusions: an incoherent discourse can seem coherent simply because a distractor matches what comes next. We investigate whether Dutch language models (6 monolingual and 4 multilingual) show the same behavior on texts that link back to earlier context with words such as 'again' and 'too'. First, we find that surprisal at the critical word tracks human acceptability judgments and eye-tracking data. Models are more surprised by incoherent continuations, but a matching distractor in the prior context reduces this surprisal. Second, attention entropy at the critical position identifies heads that behave differently under coherence vs. incoherence. We find that ablating these heads shows transfer effects across experiments, suggesting a shared mechanism. Third, we introduce energy from the associative-memory literature as a metric to quantify discourse coherence. Taken together, our results show that coherence illusions arise in Dutch LLMs, with entropy and energy exposing mechanisms that operate across settings.
Chinese Translation
心理语言学研究表明，人类读者会受到连贯性错觉的影响：一个不连贯的论述可能因为干扰项与接下来的内容相匹配而显得连贯。我们研究了荷兰语言模型（6个单语模型和4个多语模型）在使用诸如“再次”（again）和“也”（too）等词语链接回早期上下文的文本中是否表现出相同的行为。首先，我们发现关键单词的惊讶度与人类可接受性判断和眼动追踪数据相一致。模型对不连贯的延续表现出更高的惊讶度，但先前上下文中的匹配干扰项降低了这种惊讶度。其次，关键位置的注意力熵识别出在连贯性与不连贯性下表现不同的头部。我们发现，消融这些头部在不同实验中显示出迁移效应，暗示存在共享机制。第三，我们引入来自联想记忆文献的能量作为量化话语连贯性的度量。综合来看，我们的结果表明荷兰大型语言模型中存在连贯性错觉，熵和能量揭示了在不同情境下运作的机制。
cs.CL / 36 / 2606.21237
OpenWER: Improving Cross-Lingual ASR Evaluation and Enabling Token-Based Accuracy Metrics
OpenWER：提升跨语言自动语音识别评估并实现基于标记的准确率指标
Abstract
Advances in deep learning and end-to-end Automatic Speech Recognition (ASR) have enabled robust multilingual models, but evaluation metrics remain limited in assessing accuracy. Efforts to improve or replace the common metric Word Error Rate (WER) often focus on English, leaving evaluations for low-resource languages under-explored and hindering fair cross-lingual comparisons. We present OpenWER, an open-source implementation that improves WER robustness through language-specific normalisation and compound word detection. A token-based Levenshtein alignment preserves complementary metrics and allows metadata embedding for granular accuracy scores. Our analysis of 52 languages shows absolute WER reductions of up to 25% compared to common libraries. OpenWER contributes to fairness in ASR research by increasing the reliability of WER across diverse languages and enabling more comprehensive accuracy evaluations.
Chinese Translation
深度学习和端到端自动语音识别（ASR）的进展使得强大的多语言模型成为可能，但评估指标在准确性评估方面仍然有限。改善或替代常用指标词错误率（WER）的努力往往集中在英语上，导致对低资源语言的评估研究不足，妨碍了公平的跨语言比较。我们提出了OpenWER，一个开源实现，通过语言特定的规范化和复合词检测来提高WER的鲁棒性。基于标记的Levenshtein对齐保留了互补指标，并允许元数据嵌入以获得细粒度的准确性评分。我们对52种语言的分析显示，与常用库相比，绝对WER降低了多达25%。OpenWER通过提高不同语言中WER的可靠性并实现更全面的准确性评估，促进了ASR研究的公平性。
cs.CL / 37 / 2606.21255
SCOPE: Sequential Conformal Probing for Reliable OOD Rejection in LLM Services
SCOPE：用于大型语言模型服务中可靠的OOD拒绝的序贯符合探测
Abstract
Rejecting inputs outside the defined in-distribution (IND) service scope is critical for large language model (LLM) services, where unsupported requests should be filtered before full generation. Existing out-of-distribution (OOD) detectors often rely on final outputs or final-layer representations, leaving unclear where service-boundary signals are most clearly encoded inside the model; they also lack a theoretical guarantee for held-out inputs. In this paper, we introduce SCOPE (Sequential Conformal OOD Probing and Evaluation), a framework that selects a readable hidden layer, constructs a conformal gate with IND calibration, and uses a supermartingale e-process to certify persistent service-boundary evidence. Experiments across multiple LLM backbones and six carefully designed boundary conditions show that SCOPE improves gate-level rejection over standard final-layer detectors, while revealing how different OOD boundaries take different geometric forms in hidden space.
Chinese Translation
在大型语言模型（LLM）服务中，拒绝超出定义的内部分布（IND）服务范围的输入至关重要，未支持的请求应在完全生成之前进行过滤。现有的超出分布（OOD）检测器通常依赖于最终输出或最终层表示，尚不清楚服务边界信号在模型内部最清晰地编码的位置；它们也缺乏对保留输入的理论保证。本文提出了SCOPE（序贯符合OOD探测与评估），一个选择可读隐藏层、构建具有IND校准的符合门，并使用超鞅e过程来认证持久的服务边界证据的框架。在多个LLM骨干和六个精心设计的边界条件下的实验表明，SCOPE在门级拒绝方面优于标准的最终层检测器，同时揭示了不同的OOD边界在隐藏空间中呈现出不同的几何形式。
cs.CL / 38 / 2606.21340
Synthetic Audio Generation Framework for Air Traffic Control Speech Recognition
用于空中交通管制语音识别的合成音频生成框架
Abstract
Automatic Speech Recognition (ASR) systems, despite achieving remarkable accuracy in general-purpose domains with native speech (L1), struggle in domains like Air Traffic Control (ATC) due to strong channel noise, a presence of non-native (L2) English accents, and data scarcity. We propose a synthetic data generation pipeline with acoustical properties simulations specifically designed to address this lack of real data to improve recognition accuracy in the ATC domain. Our approach leverages a combination of neural generation techniques, including Text-to-Speech, Voice Conversion, L2-to-L1 accent conversion, and a novel controllable L1-to-L2 accent conversion framework built to simulate accented speech. Our experiments with the Whisper model on the ATCO2 corpus demonstrate that fine-tuning with either synthetic data alone, or a mix of real and synthetic data, significantly improves the word error rate over out-of-the-box and real data only baselines respectively.
Chinese Translation
尽管自动语音识别（ASR）系统在通用领域的母语（L1）语音上取得了显著的准确性，但在空中交通管制（ATC）等领域却面临挑战，原因包括强烈的信道噪声、非母语（L2）英语口音的存在以及数据稀缺。我们提出了一种合成数据生成管道，结合声学特性模拟，专门设计用于解决真实数据不足的问题，以提高ATC领域的识别准确性。我们的方法利用了一系列神经生成技术，包括文本到语音（Text-to-Speech）、语音转换（Voice Conversion）、L2到L1口音转换，以及一个新颖的可控L1到L2口音转换框架，旨在模拟带口音的语音。我们在ATCO2语料库上使用Whisper模型的实验表明，单独使用合成数据或真实与合成数据的混合进行微调，均显著降低了字错误率，相比于开箱即用的基线和仅使用真实数据的基线，效果显著。
cs.CL / 39 / 2606.21345
Factual Retrieval in LLMs Is a Redundant, Distributed and Non-Contiguous Process
大型语言模型中的事实检索是一个冗余、分布式且不连续的过程
Abstract
Large language models (LLMs) store and recall factual knowledge, yet the precise mechanism of how entity representations are transformed to enable specific attribute retrieval remains underexplored. In this work, we investigate this mechanism through the lens of an "attribute-computation path"-a sequence of computational steps over the entity representation required to elicit a target attribute. We then propose an iterative patching protocol to identify a minimal subset of layers necessary for this computation. Applying our method to LLaMA 3.1 8B and Qwen3 8B, we find that these paths are non-contiguous, often skipping layers, and that models possess multiple, functionally-equivalent paths for the same entity and fact, highlighting a high degree of redundancy in attribute computation. This implies that knowledge computation is highly distributed, potentially explaining the localization-editing mismatch and suggesting that knowledge storage and retrieval in LLMs is far from being well understood.
Chinese Translation
大型语言模型（LLMs）存储和回忆事实知识，但如何将实体表示转化以实现特定属性检索的精确机制仍然未被充分探索。在本研究中，我们通过“属性计算路径”的视角来探讨这一机制——这是一个在实体表示上进行的计算步骤序列，旨在引出目标属性。随后，我们提出了一种迭代修补协议，以识别进行该计算所需的最小层子集。将我们的方法应用于 LLaMA 3.1 8B 和 Qwen3 8B，我们发现这些路径是不连续的，常常跳过某些层，并且模型对于同一实体和事实存在多个功能上等效的路径，突显了属性计算中的高度冗余。这表明知识计算是高度分布式的，可能解释了定位-编辑不匹配现象，并暗示 LLMs 中的知识存储和检索远未被充分理解。
cs.CL / 40 / 2606.21359
Finetuning with Scientific Data Increases Hallucinations: A Multi-domain Factuality Evaluation of LLMs
使用科学数据进行微调增加了幻觉：对大型语言模型的多领域事实性评估
Abstract
Large language models (LLMs) are increasingly used to communicate and explain scientific concepts, yet their tendency to hallucinate poses significant risks in this high stakes use-case. Prior hallucination evaluation work remains largely restricted to the biomedical domain, treats hallucination as a binary task, and has not examined the growing family of scientifically fine-tuned LLMs. We address these gaps with SciFactCheck, a benchmark of 2,500 prompts across five scientific domains, paired with a modular evaluation framework targeting three factuality hallucination types: unverifiability, overclaim, and attribution. Using a controlled minimal-pairing design, we evaluate 18 LLMs by comparing each scientifically fine-tuned model against its general-purpose base. Our results indicate that 1. Scientifically fine-tuned models exhibit degraded factual reliability across all hallucination types and scientific domains, and 2. Fine-tuned models are internally less confident yet linguistically more assertive. A human pilot study further reveals that current fact-checking tools show only modest agreement with expert judgments on scientific content, and that defining scientifically check-worthy claims remains contested even among human annotators. Our findings fundamentally challenge current methods of domain-specific fine-tuning for factuality and call for developing improved verification infrastructure for scientific content.
Chinese Translation
大型语言模型（LLMs）越来越多地用于传达和解释科学概念，但它们的幻觉倾向在这一高风险应用中带来了显著的风险。以往的幻觉评估工作主要局限于生物医学领域，将幻觉视为二元任务，并未考察日益增长的科学微调LLMs家族。我们通过SciFactCheck来填补这些空白，这是一个涵盖五个科学领域的2500个提示的基准，配备了针对三种事实性幻觉类型的模块化评估框架：不可验证性、过度声明和归因。通过受控的最小配对设计，我们评估了18个LLMs，比较每个科学微调模型与其通用基础模型。我们的结果表明：1. 科学微调模型在所有幻觉类型和科学领域中表现出事实可靠性下降，2. 微调模型的内部信心较低，但语言上更具主张性。一项人类初步研究进一步揭示，当前的事实检查工具与专家对科学内容的判断仅有适度一致，并且即使在人工标注者之间，定义科学上值得检查的主张仍存在争议。我们的发现从根本上挑战了当前针对事实性的领域特定微调方法，并呼吁为科学内容开发改进的验证基础设施。
cs.CL / 41 / 2606.21413
CAT-Translate: Building Compact Open-Source Models for Japanese-English Translation
CAT-Translate：构建紧凑型开源日英翻译模型
Abstract
Nowadays, large multilingual translation models demonstrate impressive translation capabilities in the machine translation benchmarks. This raises a practical question to the developers: is it worth developing translation models specialized for a particular language pair if you only need to support that language pair? To give an anecdotal answer to this question, we develop a family of small language models (0.8B, 1.4B, 3.3B, and 7B parameters) specialized for Japanese-English bidirectional translation. We employ a two-stage supervised fine-tuning approach followed by Multi-Objective GRPO (Ichihara et al. 2025) to train models on synthetically generated parallel corpora. We evaluate our models on WMT and real-world translation benchmarks across business, legal, medical, financial, and patent domains. While multilingual models achieve strong performance on WMT benchmarks, our compact models outperform them on real-world benchmarks, suggesting the practical utility of developing specialized translation models even in the era of large multilingual models.
Chinese Translation
如今，大型多语言翻译模型在机器翻译基准测试中展现出令人印象深刻的翻译能力。这引发了一个实际问题：如果只需要支持特定语言对，开发专门针对该语言对的翻译模型是否值得？为此，我们开发了一系列小型语言模型（0.8B、1.4B、3.3B 和 7B 参数），专门用于日英双向翻译。我们采用了两阶段的监督微调方法，随后使用多目标 GRPO (Ichihara et al. 2025) 在合成生成的平行语料上训练模型。我们在 WMT 和实际翻译基准测试中评估了我们的模型，涵盖商业、法律、医疗、金融和专利领域。尽管多语言模型在 WMT 基准测试中表现强劲，但我们的紧凑型模型在实际基准测试中超越了它们，这表明即使在大型多语言模型盛行的时代，开发专门的翻译模型仍具有实际价值。
cs.CL / 42 / 2606.21447
Precision Recall Controllable Radiology Report Generation via Hybrid Natural Language and Clinical Reward Learning
通过混合自然语言和临床奖励学习实现可控的精确度-召回率放射学报告生成
Abstract
Automated radiology report generation (RRG) has gained increasing attention because it can reduce the heavy workload of clinical report writing. However, most existing methods mainly optimize for natural language generation (NLG) metrics that focus on language fluency, while providing little control over clinically important factors such as precision and recall. As consequence, generated reports may be fluent but not well aligned with different clinical needs. To address this challenge, we propose a reinforcement learning framework for precision recall controllable RRG, where a control parameter explicitly adjusts the trade-off between clinical precision and recall during inference. This design allows the model to flexibly generate reports according to different clinical requirements. To ensure clinical correctness, we introduce a \blue{clinical reward} into the training objective, which helps improve clinical efficacy (CE) beyond standard language-based optimization. In addition, we apply a group-relative training strategy that normalizes rewards within each training group, reducing reward variance and improving training stability. Extensive experiments on the MIMIC-CXR dataset show that our method consistently outperforms state-of-the-art approaches in both NLG{ and CE} evaluation metrics, while providing reliable control over the CE precision recall trade-off.
Chinese Translation
自动化放射学报告生成（RRG）越来越受到关注，因为它可以减少临床报告撰写的繁重工作负担。然而，现有的大多数方法主要优化自然语言生成（NLG）指标，这些指标侧重于语言流畅性，而对临床重要因素如精确度和召回率的控制较少。因此，生成的报告可能流畅，但与不同的临床需求不够一致。为了解决这一挑战，我们提出了一种用于可控精确度-召回率的RRG的强化学习框架，其中一个控制参数在推理过程中明确调整临床精确度和召回率之间的权衡。这一设计使模型能够根据不同的临床需求灵活生成报告。为了确保临床正确性，我们在训练目标中引入了临床奖励（clinical reward），这有助于提高临床有效性（CE），超越标准的基于语言的优化。此外，我们应用了一种组相对训练策略，该策略在每个训练组内规范化奖励，减少奖励方差并提高训练稳定性。在MIMIC-CXR数据集上的大量实验表明，我们的方法在NLG和CE评估指标上始终优于最先进的方法，同时提供对CE精确度-召回率权衡的可靠控制。
cs.CL / 43 / 2606.21460
Evaluation of Small Language Models for Arabic Language Processing
阿拉伯语言处理的小型语言模型评估
Abstract
This paper evaluates the performance of twelve Small Language Models (SLMs) on Arabic natural language processing tasks. The study introduces a benchmark of 240 Arabic test items distributed across eight domains and ten language skills, covering both comprehension-oriented and generation-oriented tasks. All models were evaluated under a controlled zero-shot setting using a standardized Arabic-only prompt template. Model responses were assessed through a multi-model LLM-as-a-judge framework involving GPT-4.1 Mini, Claude Haiku 4.5, and DeepSeek-Chat, with scores aggregated across judges and analyzed by task, skill, and model family. The results show that Gemma 3 (12B) achieved the highest overall score (4.548/5), followed by Aya and C4AI Command Arabic. The observed results suggest that model size alone does not explain Arabic SLM performance. Models with stronger Arabic alignment and more reliable instruction-following behavior tended to perform better across tasks. Common failure patterns among lower-performing models include prompt leakage, hallucination, language drift, incomplete generation, and weak task adherence. Overall, the benchmark provides a structured reference for evaluating compact Arabic language models and supports future work on efficient, reliable, and culturally appropriate Arabic AI systems.
Chinese Translation
本文评估了十二个小型语言模型（Small Language Models, SLMs）在阿拉伯自然语言处理任务中的表现。研究引入了一个包含240个阿拉伯测试项目的基准，这些项目分布在八个领域和十种语言技能上，涵盖了理解导向和生成导向的任务。所有模型在一个受控的零样本设置下使用标准化的阿拉伯语提示模板进行评估。模型的响应通过一个多模型的LLM作为评审框架进行评估，该框架涉及GPT-4.1 Mini、Claude Haiku 4.5和DeepSeek-Chat，评审分数在评审者之间进行汇总，并按任务、技能和模型家族进行分析。结果显示，Gemma 3（12B）获得了最高的整体分数（4.548/5），其次是Aya和C4AI Command Arabic。观察到的结果表明，仅凭模型大小无法解释阿拉伯SLM的表现。具有更强阿拉伯语对齐和更可靠的指令遵循行为的模型在各项任务中表现更佳。表现较差的模型常见的失败模式包括提示泄漏、幻觉、语言漂移、不完整生成和任务遵循性差。总体而言，该基准为评估紧凑型阿拉伯语言模型提供了结构化参考，并支持未来在高效、可靠和文化适宜的阿拉伯人工智能系统方面的研究。
cs.CL / 44 / 2606.21485
Economic Transformation and Cultural Change: Evidence from Two Centuries of French Drama
经济转型与文化变迁：来自两个世纪法国戏剧的证据
Abstract
How do large-scale economic transformations shape cultural production? We address this question by combining computational linguistics, econometrics, and formal modelling, using French drama as a well-documented empirical laboratory. Applying latent Dirichlet allocation to a corpus of 1,215 theatrical texts published between 1700 and 1900, we show that aristocratic discourse centred on sovereignty and political authority was gradually displaced by bourgeois and household economic themes as French capitalism developed. Bayesian vector autoregressive models with max-share shock identification suggest a temporal shift in the literary response to economic shocks: bourgeois everyday-life themes reacted to GDP shocks in the eighteenth century, whereas household-economic concerns became responsive only after 1820, amid accelerating industrialisation. A discrete-choice model shows that peer effects among authors and sensitivity to prevailing economic conditions can jointly account for these dynamics. Monte Carlo simulations reproduce the observed historical trajectory with reasonable fidelity. These findings offer a quantitative framework for understanding how economic transformations propagate into cultural production through identifiable social mechanisms, contributing to the study of cultural evolution and the long-run relationship between institutions and literary discourse.
Chinese Translation
大规模经济转型如何塑造文化生产？我们通过结合计算语言学、计量经济学和形式建模来探讨这一问题，以法国戏剧作为一个文献丰富的实证实验室。我们对1700年至1900年间出版的1215部戏剧文本应用潜在狄利克雷分配（Latent Dirichlet Allocation），结果显示，随着法国资本主义的发展，围绕主权和政治权威的贵族话语逐渐被以资产阶级和家庭经济为主题的内容所取代。采用贝叶斯向量自回归模型（Bayesian Vector Autoregressive Models）并进行最大份额冲击识别，表明文学对经济冲击的反应存在时间上的转变：在18世纪，资产阶级日常生活主题对GDP冲击作出了反应，而家庭经济问题则是在1820年后，伴随工业化加速才开始显现出反应。离散选择模型显示，作者之间的同行效应和对现行经济条件的敏感性共同解释了这些动态。蒙特卡洛模拟以合理的逼真度再现了观察到的历史轨迹。这些发现为理解经济转型如何通过可识别的社会机制传播到文化生产提供了一个定量框架，为文化演变的研究以及制度与文学话语之间的长期关系做出了贡献。
cs.CL / 45 / 2606.21502
Towards Pedagogically Aligned LLM Tutors for Math Mistake Remediation
面向数学错误修正的教学对齐大型语言模型辅导员
Abstract
Large language models have strong potential for use in intelligent tutoring systems, but they often fail to follow effective pedagogical strategies, such as guiding students without revealing final answers. We study the application of a two-stage alignment pipeline for math mistake remediation, combining supervised fine-tuning on tutoring dialogs with Direct Preference Optimization on synthetic preference pairs. We construct a dataset that integrates existing tutoring corpora with synthetic data generated along pedagogical dimensions, such as scaffolding and factuality, and study different input configurations that incorporate solution correctness and gold answers. Experiments show that this approach improves both factual accuracy and pedagogical quality over base models and existing tutoring models. Human evaluation further indicates that our best model is competitive with a strong proprietary baseline, while providing additional benefits in terms of openness, transparency, and reproducibility. Our results highlight the effectiveness of preference-based pedagogical alignment, while also revealing challenges in reliably evaluating tutoring quality.
Chinese Translation
大型语言模型在智能辅导系统中具有很强的潜力，但它们往往未能遵循有效的教学策略，例如在不揭示最终答案的情况下引导学生。我们研究了一个用于数学错误修正的两阶段对齐流程的应用，该流程结合了对辅导对话的监督微调与对合成偏好对的直接偏好优化。我们构建了一个数据集，该数据集将现有的辅导语料库与沿教学维度生成的合成数据（如支架和事实性）整合在一起，并研究了不同的输入配置，这些配置结合了解决方案的正确性和标准答案。实验表明，这种方法在事实准确性和教学质量上均优于基础模型和现有的辅导模型。人类评估进一步表明，我们的最佳模型在与一个强大的专有基线竞争的同时，在开放性、透明性和可重复性方面提供了额外的好处。我们的结果突显了基于偏好的教学对齐的有效性，同时也揭示了在可靠评估辅导质量方面的挑战。
cs.CL / 46 / 2606.21517
MedHal-Loc: Are "Explainable-by-Architecture" Medical Hallucination Detectors Faithful Localizers? A Localization Benchmark
MedHal-Loc：基于“架构可解释性”的医学幻觉检测器是否是可靠的定位器？一个定位基准
Abstract
Detecting hallucinations in clinical text is increasingly framed as an explainability problem: systems should not merely flag an unreliable response but point to the offending span. Architectures built around knowledge-graph (KG) triple decomposition are marketed for exactly this auditability, yet their localization ability is typically assumed rather than measured. We introduce MedHal-Loc, a benchmark and metric for localization faithfulness -- whether a detector's top-ranked error unit actually overlaps the erroneous span. The controlled subset comprises 300 PubMedQA-derived statements with single, span-level errors injected across four localizable types (entity substitution, relation error, mechanism misattribution, invention), yielding gold spans by construction; a complementary natural subset documents that real hallucinations are dominated by diffuse conclusion-flips that resist span localization (a human expert accepted 1/18 candidate spans). Evaluating four fine-grained paradigms, we find that NLI-per-clause, consistency-per-sentence, and the dedicated span detector FAVA all localize well above chance, whereas an elaborate KG-triple pipeline localizes no better than chance (+3.3pp, n.s.), bottlenecked by ~59% entity-extraction coverage -- despite competitive detection F1 (0.609). Detection competence does not imply faithful localization; architectural explainability must be validated, not presumed.
Chinese Translation
在临床文本中检测幻觉越来越被视为一个可解释性问题：系统不仅应标记不可靠的响应，还应指明问题所在的文本片段。围绕知识图谱（KG）三元组分解构建的架构正是为了实现这种可审计性而被推广，但其定位能力通常是被假定而非测量的。我们引入了MedHal-Loc，一个用于定位忠实度的基准和指标——即检测器的最高排名错误单元是否实际与错误片段重叠。受控子集包含300个源自PubMedQA的陈述，这些陈述在四种可定位类型（实体替换、关系错误、机制误归因、发明）中注入了单一的片段级错误，从而构建出黄金片段；一个补充的自然子集记录了真实幻觉主要由难以定位的模糊结论翻转主导（人类专家接受了18个候选片段中的1个）。在评估四种细粒度范式时，我们发现NLI-per-clause、consistency-per-sentence和专用的片段检测器FAVA的定位效果均显著高于随机水平，而复杂的KG三元组管道的定位效果与随机水平无显著差异（+3.3pp，n.s.），其瓶颈在于约59%的实体提取覆盖率——尽管检测F1值具有竞争力（0.609）。检测能力并不意味着忠实的定位；架构的可解释性必须经过验证，而非假定。
cs.CL / 47 / 2606.21553
Dissecting Agentic RAG: A Component Ablation for Multi-Hop QA with a Local 7B Model
剖析代理检索增强生成（Agentic RAG）：基于局部7B模型的多跳问答组件消融研究
Abstract
Agentic retrieval-augmented generation (RAG) systems combine iterative reasoning loops, query decomposition, and adaptive retrieval to tackle multi-hop question answering. However, the contribution of each component remains poorly understood, particularly under resource-constrained settings using only local language models. Many agentic designs add adaptive retrieval routing and deeper retrieval loops on the assumption that the added complexity helps. To test whether it does, we run a controlled ablation study of a full agentic RAG pipeline evaluated on 5,000 questions from the HotpotQA distractor development set using a local 7B parameter model (Qwen2.5-7B-Instruct). Our full pipeline achieves EM=53.2% and F1=61.6%, compared to a single-pass dense-retrieval baseline of EM=43.1% and F1=54.0%. Across eight ablation conditions, we find that: (1) fixed hybrid retrieval via reciprocal rank fusion consistently outperforms rule-based adaptive routing (+1.8 EM, +1.9 F1), as the routing heuristic over-routes to BM25 by firing on named entities present in nearly all multi-hop sub-questions; (2) two retrieval iterations over the decomposed sub-questions capture 95% of the gains of five, with no meaningful benefit from deeper loops; and (3) query decomposition and cross-encoder reranking each contribute statistically significant but smaller gains (p<0.01 and p<0.001 respectively). Taken together, on a fixed local-model budget, the simpler and fixed choices turn out to be competitive with or better than their adaptive versions: most of the gain comes from running a short retrieval loop, not from adaptive routing or from many iterations. We use no proprietary APIs or large-scale compute.
Chinese Translation
代理检索增强生成（RAG）系统结合了迭代推理循环、查询分解和自适应检索，以应对多跳问答。然而，各个组件的贡献仍然不甚明了，特别是在仅使用局部语言模型的资源受限环境下。许多代理设计在假设增加的复杂性有助于提升性能的前提下，增加了自适应检索路由和更深的检索循环。为了验证这一假设，我们对一个完整的代理 RAG 流水线进行了受控消融研究，评估了来自 HotpotQA 干扰开发集的 5,000 个问题，使用了局部 7B 参数模型（Qwen2.5-7B-Instruct）。我们的完整流水线达到了 EM=53.2% 和 F1=61.6%，而单次密集检索基线的 EM 为 43.1%，F1 为 54.0%。在八种消融条件下，我们发现：(1) 通过互惠排名融合实现的固定混合检索始终优于基于规则的自适应路由（EM 提升 1.8，F1 提升 1.9），因为路由启发式在几乎所有多跳子问题中都过度路由到 BM25；(2) 对分解的子问题进行两次检索迭代即可捕获五次迭代的 95% 的增益，而更深的循环没有带来显著的益处；(3) 查询分解和交叉编码器重排序各自贡献了统计上显著但较小的增益（p<0.01 和 p<0.001）。综合来看，在固定的局部模型预算下，简单且固定的选择与其自适应版本相比表现出竞争力，甚至更好：大部分增益来自于运行短的检索循环，而非自适应路由或多次迭代。我们未使用任何专有 API 或大规模计算。
cs.CL / 48 / 2606.21557
PeerMathDial: A Middle School Dialogue Dataset for Student Collaborative Math Problem Solving
PeerMathDial：一个用于学生协作数学问题解决的中学对话数据集
Abstract
Collaborative Problem Solving (CPS) is a core skill in education, where the process of peer interaction is highly important. However, existing educational dialogue datasets mostly focus on classroom instruction or tutoring (i.e., teacher/tutor-student interaction), yet datasets centering small-group, student-student interaction are limited. This thus leaves research with limited resources for studying how students interact, coordinate, and solve problems together in real educational settings. To address this, we introduce PeerMathDial, the first dataset of peer CPS dialogues collected from authentic middle school math classrooms. It contains 55 dialogues from 27 students, totaling 6,406 turns. To facilitate research on CPS discourse analysis, we further build a corpus-grounded dialogue act taxonomy assisted by LLMs. Using the dataset and the dialogue act taxonomy, we demonstrate the practical applications of PeerMathDial across three use cases. First, we track how dialogues evolve over time and measure the impact of teacher interventions. Second, we align dialogue actions with student surveys to reveal the connection between students' traits (e.g., confidence, leadership) and their actual behaviors. Third, by evaluating LLMs on dialogue act prediction, we glimpse at the potential of LLMs for student simulation in educational applications. Our dataset and source code will be released to the community.
Chinese Translation
协作问题解决（Collaborative Problem Solving, CPS）是教育中的一项核心技能，其中同伴互动的过程至关重要。然而，现有的教育对话数据集主要集中于课堂教学或辅导（即教师/辅导员与学生的互动），而以小组学生间互动为中心的数据集则相对有限。这使得研究者在研究学生如何在真实教育环境中互动、协调和共同解决问题时面临资源不足的问题。为了解决这一问题，我们推出了PeerMathDial，这是第一个从真实中学数学课堂收集的同伴CPS对话数据集。该数据集包含来自27名学生的55个对话，总计6,406轮对话。为了促进对CPS话语分析的研究，我们进一步构建了一个基于语料库的对话行为分类法，并借助大型语言模型（LLMs）进行辅助。利用该数据集和对话行为分类法，我们展示了PeerMathDial在三个应用案例中的实际应用。首先，我们追踪对话随时间的演变，并测量教师干预的影响。其次，我们将对话行为与学生调查结果对齐，以揭示学生特质（如自信心、领导能力）与其实际行为之间的关系。第三，通过评估LLMs在对话行为预测上的表现，我们窥见了LLMs在教育应用中进行学生模拟的潜力。我们的数据集和源代码将向社区发布。
cs.CL / 49 / 2606.21559
Rubric-as-Experts: Case-Specific MQM Rubrics for Translation Quality Evaluation
专家评分标准：针对翻译质量评估的案例特定 MQM 评分标准
Abstract
Large language models (LLMs) have shown strong potential in fine-grained translation quality evaluation (QE), yet existing MQM-based approaches typically rely on fixed rubric configurations shared across all translation samples. However, translation instances often differ substantially in error complexity, ambiguity, and required evaluation granularity, making static rubric allocation suboptimal for span-level error detection. We find that larger MQM subtype spaces improve error coverage but also introduce more false positives, while different translation instances prefer different rubric granularities, suggesting that evaluation spaces should be allocated dynamically for each case. Motivated by these observations, we propose a case-specific dynamic rubric framework that adaptively constructs MQM evaluation spaces for individual translation instances. Unlike fully free-form rubric generation methods, our framework remains grounded in the predefined MQM taxonomy while dynamically selecting suitable subtype spaces and evaluation granularity for different cases. Experiments on WMT span-level QE benchmarks across multiple model scales demonstrate that the proposed framework consistently improves MCC and produces cleaner span-level error localization compared with static rubric settings. Our results suggest that combining structured MQM rubrics with case-specific adaptive allocation is an effective strategy for fine-grained LLM-based translation evaluation.
Chinese Translation
大型语言模型（LLMs）在细粒度翻译质量评估（QE）中展现出强大的潜力，但现有的基于 MQM 的方法通常依赖于在所有翻译样本中共享的固定评分标准配置。然而，翻译实例在错误复杂性、模糊性和所需评估粒度上往往存在显著差异，使得静态评分标准分配在跨度级错误检测中表现不佳。我们发现，较大的 MQM 子类型空间可以改善错误覆盖，但也会引入更多的假阳性，而不同的翻译实例则偏好不同的评分粒度，这表明评估空间应针对每个案例动态分配。基于这些观察，我们提出了一种案例特定的动态评分标准框架，该框架能够自适应地为个别翻译实例构建 MQM 评估空间。与完全自由形式的评分标准生成方法不同，我们的框架仍然基于预定义的 MQM 分类法，同时动态选择适合不同案例的子类型空间和评估粒度。在多个模型规模的 WMT 跨度级 QE 基准上的实验表明，与静态评分设置相比，所提框架在 MCC 上始终表现出改善，并产生更清晰的跨度级错误定位。我们的结果表明，将结构化的 MQM 评分标准与案例特定的自适应分配相结合是一种有效的细粒度 LLM 基于翻译评估的策略。
cs.CL / 50 / 2606.21595
Per-Entity Bias Mapping for AI Visibility: Why Brand Mentions Require Entity-Specific Calibration
针对人工智能可见性的实体偏差映射：品牌提及为何需要实体特定的校准
Abstract
AI-mediated answer systems increasingly determine how brands and organizations are represented to users. Existing approaches reduce visibility to mention rate or citation frequency. This paper argues that aggregate metrics are insufficient because entities exhibit systematically different AI visibility error profiles. We introduce Per-Entity Bias Mapping (PEBM): a ten-dimensional framework distinguishing raw from verified mentions. Three failure modes are identified: (1) underrepresented entities suffer invisibility due to weak knowledge graph presence; (2) large entities suffer the Brand Hallucination Paradox -- model familiarity creates stronger surfaces for plausible but incorrect completions; (3) CEE entities face a structural infrastructure gap across knowledge graphs, NER, and entity linking. A fourth dimension, Parametric-Retrieval Lag Asymmetry, describes divergence between retrieval-augmented and parametric memory update cycles. A full-scale empirical study (n=100 Hungarian B2B entities, 1,400 probe runs, 2,062 sources) finds Tier 1 brands produce 52.69% fabricated citations versus 37.87% for Tier 3 entities (+14.82 pp; p=1.67e-11), supporting the Brand Hallucination Paradox. Regulatory-framed queries elevate fabrication to 56.77% versus 37.59% baseline (+19.2 pp). We identify rejection-induced confabulation escalation: agentic quality filters function as hallucination accelerators in compliance contexts. We introduce ghost cartography as a unifying mechanism: entities in sparse latent regions produce confident output interpolated from neighboring dense regions, yielding a two-dimensional confabulation space (fabricated presence vs. frozen representation).
Chinese Translation
人工智能介导的回答系统日益决定品牌和组织如何向用户展示。现有的方法将可见性简化为提及率或引用频率。本文认为，聚合指标是不够的，因为实体表现出系统性不同的人工智能可见性错误特征。我们引入了针对实体的偏差映射（Per-Entity Bias Mapping, PEBM）：一个十维框架，区分原始提及与经过验证的提及。我们识别出三种失败模式：（1）被低估的实体由于知识图谱存在薄弱而遭受不可见性；（2）大型实体遭遇品牌幻觉悖论（Brand Hallucination Paradox）——模型熟悉度为合理但不正确的补全创造了更强的表面；（3）CEE实体在知识图谱、命名实体识别（NER）和实体链接之间面临结构性基础设施差距。第四个维度，参数检索滞后不对称性（Parametric-Retrieval Lag Asymmetry），描述了检索增强与参数记忆更新周期之间的差异。通过一项全面的实证研究（n=100个匈牙利B2B实体，1,400次探测运行，2,062个来源），发现Tier 1品牌产生的虚构引用为52.69%，而Tier 3实体为37.87%（+14.82个百分点；p=1.67e-11），支持品牌幻觉悖论。以监管框架提出的查询将虚构率提升至56.77%，而基线为37.59%（+19.2个百分点）。我们识别出拒绝引发的虚构升级：在合规环境中，代理质量过滤器作为幻觉加速器发挥作用。我们引入了幽灵制图（ghost cartography）作为统一机制：处于稀疏潜在区域的实体生成的自信输出是从邻近的密集区域插值而来，形成一个二维虚构空间（虚构存在与冻结表现）。
cs.CL / 51 / 2606.21616
LLM and Human Modes of Representation
大型语言模型与人类的表征模式
Abstract
Much work on the cognitive foundations of AI has focussed on comparisons between the ways in which Large Language Models (LLMs) and humans process information and represent it. One aspect of this comparison involves determining the extent to which LLMs can achieve or surpass human performance on a variety of cognitively interesting tasks. A second explores points of convergence and divergence between LLM and human systems for processing information. Here, I consider some recent research that has addressed both issues in two informational domains. The first is the representation of linguistic knowledge. The second is real world reasoning and planning. While LLMs frequently achieve impressive levels of performance and fluency on linguistic applications, they tend to handle linguistic content in ways that are distinct from human processing. They are also, for the most part, less efficient than humans in learning and generalisation for reasoning tasks.
Chinese Translation
关于人工智能认知基础的研究主要集中在大型语言模型（LLMs）与人类处理信息和表征信息的方式之间的比较。此比较的一个方面涉及确定LLMs在各种认知有趣任务上能够达到或超越人类表现的程度。第二个方面探讨了LLM与人类在信息处理系统上的趋同与分歧。在这里，我考虑了一些最近的研究，这些研究在两个信息领域中解决了这两个问题。第一个领域是语言知识的表征。第二个领域是现实世界的推理与规划。尽管LLMs在语言应用中经常表现出令人印象深刻的表现和流利度，但它们处理语言内容的方式与人类处理方式存在显著差异。在推理任务的学习和概括方面，它们在大多数情况下也不如人类高效。
cs.CL / 52 / 2606.21618
CulMind: Benchmarking Multimodal Understanding and Reasoning in Chinese Cultural Heritage
CulMind：在中国文化遗产中基准测试多模态理解与推理
Abstract
Evaluating Multimodal Large Language Models (MLLMs) in Chinese Cultural Heritage (CCH) requires fine-grained reasoning over visual, textual, stylistic, and historical clues. However, existing CCH benchmarks mainly emphasize final-answer accuracy, while the accuracy and completeness of reasoning processes remain underexplored. To address this gap, we introduce CulMind and CulMind-R: a high-quality benchmark for multimodal CCH covering 50 tasks from collections of more than 100 museums, and a 24-task reasoning subset that adaptively defines task-specific dimensions for reasoning process evaluation. To evaluate reasoning quality, we propose ReaScore, a task-adaptive metric that evaluates reasoning by automatically weighting task-relevant dimensions. Experiments on 14 leading MLLMs reveal a substantial gap between answers and reasoning, especially on challenging tasks. Further analysis shows that task-adaptive dimension selection and weighting better align evaluation results with expert judgments. Overall, our benchmark and metric support a more expert-aligned assessment of CCH understanding and offer a transferable reference for broader evaluations of cultural heritage. We publicly release the data, code, and evaluation scripts at https://github.com/ZevTsao/CulMind to facilitate reproducible research.
Chinese Translation
在中国文化遗产（CCH）中评估多模态大型语言模型（MLLMs）需要对视觉、文本、风格和历史线索进行细致的推理。然而，现有的CCH基准主要强调最终答案的准确性，而推理过程的准确性和完整性仍然未得到充分探索。为了解决这一问题，我们引入了CulMind和CulMind-R：一个高质量的多模态CCH基准，涵盖来自100多个博物馆的50个任务，以及一个24个任务的推理子集，该子集自适应地定义了任务特定的维度以评估推理过程。为了评估推理质量，我们提出了ReaScore，这是一种任务自适应的度量标准，通过自动加权与任务相关的维度来评估推理。对14个领先的MLLMs的实验揭示了答案与推理之间的显著差距，尤其是在具有挑战性的任务上。进一步分析表明，任务自适应的维度选择和加权能够更好地将评估结果与专家判断对齐。总体而言，我们的基准和度量支持对CCH理解的更专业对齐评估，并为更广泛的文化遗产评估提供了可转移的参考。我们在https://github.com/ZevTsao/CulMind公开发布数据、代码和评估脚本，以促进可重复研究。
cs.CL / 53 / 2606.21622
Evaluating Document-Tuned Transformer Representations for Person-level Mental Health Assessment
评估针对个体心理健康评估的文档调优变换器表示
Abstract
Person-level psychological assessment requires aggregating meaning across many messages from the same individual, a task that document-level training objectives were not explicitly designed for. We present a systematic, empirical comparison between architecturally matched traditional (a) base-transformers and (b) document-tuned-transformers (further contrastively fine-tuned at the document-level, sometimes referred to as "sentence transformers") under otherwise identical conditions. Comparing layer-wise and overall performance across two longitudinal mental health and psychological datasets, we find document-tuned models demonstrated a consistent improvement over base representations (increase in Pearson r of 13.4%, p=.015). Robustness analyses revealed document-tuned models remained more accurate under perturbations to word deletion, synonym replacement, typo injection, and back translation. Further, hedged language (e.g., `usually') was more characteristic of outcomes in document-tuned embeddings while abundance (e.g., `lot') was more characteristic of base-transformers, suggesting document-tuned models may better capture uncertainty. These results suggest representation choice impacts mental health prediction, document-tuned models often being more adept.
Chinese Translation
个体层面的心理评估需要聚合来自同一个体的多条信息的意义，而这一任务并不是文档级训练目标明确设计的。我们在相同条件下，系统性地、实证性地比较了架构匹配的传统(a)基础变换器和(b)文档调优变换器（在文档级进一步对比微调，有时称为“句子变换器”）。通过比较两个纵向心理健康和心理学数据集的层级性能和整体性能，我们发现文档调优模型在基础表示上表现出了一致的改进（Pearson r 增加了 13.4%，p=.015）。稳健性分析显示，文档调优模型在对单词删除、同义词替换、拼写错误注入和反向翻译的扰动下仍然保持更高的准确性。此外，模糊语言（例如，“通常”）在文档调优嵌入的结果中更为典型，而丰富性（例如，“很多”）则更典型于基础变换器，这表明文档调优模型可能更好地捕捉不确定性。这些结果表明，表示选择对心理健康预测有影响，文档调优模型通常更为擅长。
cs.CL / 54 / 2606.21631
CuratorKIT : Data Curation and Synthetic Data Generation for LLM Post-Training
CuratorKIT：用于大型语言模型后训练的数据管理与合成数据生成
Abstract
Data curation is a critical part of post-training pipelines for large language models, yet existing tools often treat ingestion, deduplication, synthetic generation, and quality filtering as separate stages. This fragmentation makes it difficult to audit pipeline decisions or understand why individual samples are rejected. CuratorKIT is an open-source Python library that covers this full lifecycle in a single configurable pipeline. The framework is composed of six source format readers and automatic schema detection, a pre-generation data hygiene layer for credentials, PII, and toxic content, eight LLM-powered generation tasks, three complementary quality gates with provenance-exact hallucination verification, structured adaptive recovery, and five training-ready export formats compatible with TRL, Unsloth, and AlignTune. Every pipeline decision is recorded in an append-only per-sample provenance chain, and rejected samples carry structured failure reasons rather than being silently discarded. CuratorKIT supports 100+ LLM providers through LiteLLM, exposes both a Python API and a YAML-driven CLI, and is designed for practitioners who need reproducible, auditable data pipelines at scale .
Chinese Translation
数据管理是大型语言模型后训练流程中的关键部分，但现有工具往往将数据导入、去重、合成生成和质量过滤视为独立的阶段。这种碎片化使得审计流程决策或理解为何个别样本被拒绝变得困难。CuratorKIT 是一个开源 Python 库，涵盖了这一完整生命周期，并在单一可配置的流程中实现。该框架由六个源格式读取器和自动模式检测、一个用于凭证、个人身份信息（PII）和有毒内容的预生成数据卫生层、八个基于大型语言模型（LLM）的生成任务、三个互补的质量门控以及具有来源精确幻觉验证的结构化自适应恢复组成，并提供五种与 TRL、Unsloth 和 AlignTune 兼容的训练准备导出格式。每个流程决策都记录在一个仅附加的每样本来源链中，被拒绝的样本携带结构化的失败原因，而不是被静默丢弃。CuratorKIT 通过 LiteLLM 支持 100 多个 LLM 提供者，提供 Python API 和基于 YAML 的命令行接口，旨在为需要可重复、可审计的大规模数据流程的实践者提供服务。
cs.CL / 55 / 2606.21645
Behavioral and Representational Evidence of Binomial Ordering Preferences in Large Language Models
大型语言模型中的二项排序偏好行为与表征证据
Abstract
Large language models (LLMs) can readily reproduce conventional expressions, yet their ability to model gradient frequency distributions remains underexplored. We investigate this using linguistic binomials, such as men and women, where both word permutations are grammatically valid but exhibit distinct, cross-linguistic variations in conventionality. We formalize binomial ordering as a distributional alignment problem, and construct a multilingual dataset of 600 binomial pairs across 8 languages. With categorical and distributional metrics, we measure and compare the corpus-derived preferences with model-induced ordering probabilities of 6 open-weight LLMs. While models often behaviorally recover the dominant corpus-preferred order, particularly for strongly conventionalized pairs, they align less well with the exact corpus preference distributions. This suggests that apparent directional order overstates how faithfully LLMs capture the statistical nuances of language use. Sparse probing verifies that the concept of preference strength is partially encoded among middle-to-late layers, and steering along probe-derived directions alters model-induced ordering distributions, demonstrating that the statistical behavioral preference of LLMs can be mechanistically measured and manipulated via internal representations.
Chinese Translation
大型语言模型（LLMs）能够轻松再现常规表达，但它们对梯度频率分布的建模能力仍然未得到充分探索。我们通过语言二项词组（如 men and women）进行研究，其中两种词序在语法上都是有效的，但在常规性上存在明显的跨语言差异。我们将二项排序形式化为一个分布对齐问题，并构建了一个包含600对二项词组的多语言数据集，涵盖8种语言。通过分类和分布度量，我们测量并比较了从语料库中得出的偏好与6个开放权重LLM的模型诱导排序概率。尽管模型通常在行为上恢复了语料库中主导的偏好顺序，特别是对于强常规化的词对，但它们与确切的语料偏好分布的对齐程度较低。这表明，表面上的方向性顺序夸大了LLMs捕捉语言使用统计细微差别的忠实度。稀疏探测验证了偏好强度的概念在中后层中部分编码，并且沿着探测导出的方向引导会改变模型诱导的排序分布，证明LLMs的统计行为偏好可以通过内部表征进行机械测量和操控。
cs.CL / 56 / 2606.21649
EvoEmbedding: Evolvable Representations for Long-Context Retrieval and Agentic Memory
EvoEmbedding：用于长上下文检索和自主记忆的可进化表示
Abstract
Existing embedding models are inherently static: they encode text segments in isolation, ignoring their surrounding context and temporal order. This paper introduces EvoEmbedding, a novel embedding model that generates evolvable representations for retrieval. It is tailored for long-context scenarios, where information is dynamic, sequential, and requires continuous state tracking. Our design is simple: EvoEmbedding maintains a continuously updated latent memory as it sequentially processes inputs, and uses it alongside the raw content to jointly generate evolvable embeddings. Consequently, for the same query, our model adapts its representation to retrieve distinct targets based on the evolving context, going beyond static semantic search. To equip the model with this capability, we construct EvoTrain-180K, a diverse dataset for the joint optimization of latent memory and retrieval. Furthermore, we introduce a memory queue to prevent representation collapse during recurrent encoding, alongside segment-batching techniques that tackle significant length variance and accelerate training by 3.8$\times$. Extensive experiments show that our model not only outperforms larger-scale specialists (e.g., Qwen3-Embedding-8B and KaLM-Embedding-Gemma3-12B) across a range of long-context retrieval benchmarks, but also generalizes well to downstream tasks (e.g., personalization) with contexts 10$\times$ longer than its training window. Notably, EvoEmbedding seamlessly integrates into agentic workflows to boost performance. For instance, a naive RAG pipeline equipped with our model surpasses dedicated agentic memory systems. Project Page: https://clare-nie.github.io/EvoEmbedding.
Chinese Translation
现有的嵌入模型本质上是静态的：它们在孤立的情况下编码文本片段，忽视了周围的上下文和时间顺序。本文介绍了EvoEmbedding，一种新颖的嵌入模型，能够生成可进化的表示以用于检索。该模型专为长上下文场景设计，其中信息是动态的、顺序的，并且需要持续的状态跟踪。我们的设计简单：EvoEmbedding在顺序处理输入时维护一个持续更新的潜在记忆，并将其与原始内容结合使用，以共同生成可进化的嵌入。因此，对于相同的查询，我们的模型能够根据不断变化的上下文调整其表示，以检索不同的目标，超越静态语义搜索。为了赋予模型这一能力，我们构建了EvoTrain-180K，这是一个多样化的数据集，用于潜在记忆和检索的联合优化。此外，我们引入了一个记忆队列，以防止在递归编码过程中表示崩溃，并采用了分段批处理技术，以应对显著的长度变化，并将训练速度提高了3.8倍。大量实验表明，我们的模型不仅在一系列长上下文检索基准中超越了更大规模的专业模型（例如，Qwen3-Embedding-8B和KaLM-Embedding-Gemma3-12B），而且在下游任务（例如个性化）中也能很好地泛化，处理的上下文长度是其训练窗口的10倍。值得注意的是，EvoEmbedding能够无缝集成到自主工作流中以提升性能。例如，配备我们模型的简单RAG管道超越了专用的自主记忆系统。项目页面：https://clare-nie.github.io/EvoEmbedding。
cs.CL / 57 / 2606.21685
TACO: Task-Aware Column Description Generation Using LLMs
TACO：基于任务感知的列描述生成方法使用大型语言模型
Abstract
Generating accurate and informative column descriptions (e.g. "membership status of customers" for the column name "cust_mem") is essential for a wide range of downstream NLP tasks on tabular data, including NL2SQL, table question answering, and entity linking. This problem arises in enterprises, domain sciences, government data portals, and so on. Despite its importance, most real-world datasets suffer from missing or cryptic documentation, often due to abbreviated column names or domain-specific jargon. Existing approaches largely rely on single-prompt large language models (LLMs), which struggle with three key issues: (i) inconsistent or incorrect handling of abbreviations, (ii) hallucinated or incomplete descriptions, and (iii) redundancy or vagueness that hinders downstream performance. We present TACO, a task-aware framework for automatic column description generation using LLMs. TACO introduces a three-step pipeline: (1) abbreviation expansion, which standardizes column names; (2) description generation, which produces initial semantic descriptions enriched with synonyms and search-oriented keywords; and (3) description revision, which refines these outputs using simulated downstream tasks. In addition, we investigate human-in-the-loop extensions and release new evaluation datasets for entity linking and schema enrichment. Extensive experiments across public and proprietary datasets show that TACO consistently outperforms existing methods, improving downstream task performance by up to 32%.
Chinese Translation
生成准确且信息丰富的列描述（例如，对于列名 "cust_mem" 的描述为 "客户的会员状态"）对于多种下游自然语言处理（NLP）任务在表格数据上的应用至关重要，包括 NL2SQL、表格问答和实体链接等。这一问题在企业、领域科学、政府数据门户等场景中普遍存在。尽管其重要性不言而喻，但大多数现实世界的数据集都存在缺失或模糊的文档，通常是由于列名缩写或领域特定术语的使用。现有的方法主要依赖于单提示的大型语言模型（LLMs），但在处理时面临三个关键问题：（i）对缩写的处理不一致或不正确，（ii）生成的描述存在虚构或不完整的情况，以及（iii）冗余或模糊的表述妨碍下游性能。我们提出了 TACO，一个基于任务感知的框架，用于使用 LLMs 自动生成列描述。TACO 引入了一个三步流程：（1）缩写扩展，标准化列名；（2）描述生成，生成初步的语义描述，并丰富同义词和搜索导向的关键词；（3）描述修订，利用模拟下游任务对这些输出进行优化。此外，我们还研究了人机协作的扩展，并发布了新的评估数据集用于实体链接和模式丰富。针对公共和专有数据集的广泛实验表明，TACO 在性能上始终优于现有方法，提升下游任务性能最高可达 32%。
cs.CL / 58 / 2606.21689
Clinical Term Extraction using Open-Source Small Language Models
使用开源小型语言模型进行临床术语提取
Abstract
Clinical information for amyotrophic lateral sclerosis (ALS) care documented in unstructured clinical notes limits downstream analysis without extraction into structured formats. Open-source small language models with few-shot prompting for detecting the presence of ALS-relevant clinical terms in patient documentation were evaluated without task-specific training data. The detection task targeted 17 categories spanning functional scores, respiratory measures, medications, and related clinical and non-clinical attributes. Clinical note content was normalized from JSON-encoded discharge summaries and processed with a prompt template having structured JSON outputs. We compared 26 open-source models using aggregate, label-level, and manual-validation multilabel classification metrics. Manual validation showed that a regex rule baseline had higher overall micro-F1 and lower Hamming loss than any single SLM or TF-IDF baseline, while Qwen3-4B-Instruct-2507 was the highest-performing SLM by micro-F1. Model rankings varied by metric and label category, with the TF-IDF baseline showing high recall but low precision, some SLMs showing higher precision but lower recall, and Hammer2.1-7b showing strong performance for ALSFRS-R subscore detection. These findings support targeted hybrid extraction workflows rather than replacement of existing rule-based methods.
Chinese Translation
记录在非结构化临床笔记中的肌萎缩侧索硬化症（ALS）护理临床信息限制了后续分析，需提取为结构化格式。评估了使用开源小型语言模型（SLM）和少量示例提示来检测患者文档中与ALS相关的临床术语的能力，且未使用特定任务的训练数据。检测任务针对17个类别，包括功能评分、呼吸测量、药物及相关的临床和非临床属性。临床笔记内容从JSON编码的出院摘要中规范化，并使用具有结构化JSON输出的提示模板进行处理。我们使用汇总、标签级和人工验证的多标签分类指标比较了26个开源模型。人工验证显示，正则表达式规则基线的总体微F1值高于任何单一SLM或TF-IDF基线，且汉明损失较低，而Qwen3-4B-Instruct-2507是微F1表现最佳的SLM。模型排名因指标和标签类别而异，TF-IDF基线显示高召回率但低精确率，部分SLM显示较高的精确率但较低的召回率，而Hammer2.1-7b在ALSFRS-R子评分检测中表现强劲。这些发现支持针对性的混合提取工作流程，而非替代现有的基于规则的方法。
cs.CL / 59 / 2606.21704
When Compression Helps and When It Hurts: Condition-Aware Analysis of Chain-of-Thought Distillation
压缩何时有益，何时有害：基于条件的链式思维蒸馏分析
Abstract
Chain-of-Thought (CoT) distillation transfers multi-step reasoning from large reasoning models to smaller students, but verbose teacher traces inflate both training and inference cost. Existing CoT compression methods fall into two families, selective pruning and generative rewriting, yet prior studies have left key factors entangled: granularity is confounded with importance criteria in pruning, restructuring level is rarely isolated in rewriting, and compression budgets are not systematically evaluated across domains or regimes. We recast CoT compression along three dimensions: importance criterion, restructuring level, and compression budget. Sweeping these across two model families, Math and General domains, and Long-/Short-CoT regimes, we find that (i) importance criterion utility is strictly governed by granularity: step-level criteria converge on a shared reasoning backbone, while token-level pruning requires symbol-aware signals to preserve the logical core; (ii) restructuring level inverts across domains: Math degrades monotonically with structural disruption, while aggressive rewriting acts as a denoiser on General tasks; (iii) training-time compression does not necessarily translate to inference-time savings: Long-CoT students retain verbose habits despite concise supervision, making the training ratio an optimistic lower bound on deployment cost. These findings yield condition-aware guidelines for matching compression to deployment context.
Chinese Translation
链式思维（Chain-of-Thought, CoT）蒸馏将多步骤推理从大型推理模型转移到较小的学生模型，但冗长的教师轨迹会增加训练和推理的成本。现有的 CoT 压缩方法主要分为两类：选择性剪枝和生成重写，然而以往的研究将关键因素混淆在一起：在剪枝中，粒度与重要性标准相互影响；在重写中，重构水平很少被单独考虑；而压缩预算在不同领域或模式下并未得到系统评估。我们从三个维度重新审视 CoT 压缩：重要性标准、重构水平和压缩预算。在两个模型家族（数学和通用领域）以及长链式思维/短链式思维模式下进行广泛实验，我们发现：（i）重要性标准的效用严格受粒度控制：步骤级标准趋向于共享推理骨架，而符号级剪枝需要符号感知信号以保留逻辑核心；（ii）重构水平在不同领域中呈现反转：数学领域在结构干扰下单调降级，而激进重写在通用任务中充当去噪器；（iii）训练时的压缩不一定能转化为推理时的节省：尽管有简洁的监督，长链式思维学生仍保留冗长的习惯，使得训练比率成为部署成本的乐观下限。这些发现为将压缩与部署环境匹配提供了基于条件的指导。
cs.CL / 60 / 2606.21710
PrivacyAlign: Contextual Privacy Alignment for LLM Agents
PrivacyAlign：面向大型语言模型代理的上下文隐私对齐
Abstract
AI agents acting on behalf of users are constantly making decisions, and for users to trust their agents, those decisions must align with what they actually want. Privacy is an important alignment problem for agents: every message, post, or tool call an agent makes is a contextual judgment about what is appropriate to share, with whom, and under which conditions. Because such judgments depend on social expectations and norms, human judgment does not merely label privacy violations but also helps define them. While existing work relies on unreliable proxies for both training and evaluation, we place human judgment at the center of agentic privacy alignment. We introduce PrivacyAlign, a dataset of 1,350 samples with 3,516 detailed annotations from 599 unique annotators across diverse scenarios where current LLMs actually leak, and use it to ground both alignment training and automated evaluation in human privacy norms. Building on these annotations, we first show that conditioning LLM judges on human annotations and explanations for reference responses to the same prompt makes their judgments more reliable. We then introduce annotation-conditioned reward modeling, which uses these annotations to score new responses during RL, and show that small open-weight agents trained with this reward better align with human privacy norms, with strong gains on PrivacyAlign and existing privacy benchmarks for agents.
Chinese Translation
代表用户行动的人工智能代理不断做出决策，为了让用户信任这些代理，这些决策必须与用户的真实意图相一致。隐私是代理面临的重要对齐问题：代理发送的每条消息、每个帖子或工具调用都是关于什么是适合分享、与谁分享以及在什么条件下分享的上下文判断。由于这些判断依赖于社会期望和规范，人类判断不仅仅是对隐私侵犯的标记，还帮助定义隐私侵犯的界限。尽管现有工作依赖于不可靠的代理进行训练和评估，我们将人类判断置于代理隐私对齐的中心。我们引入了PrivacyAlign，一个包含1,350个样本和来自599个独特标注者的3,516个详细注释的数据集，涵盖了当前大型语言模型（LLM）实际泄露隐私的多种场景，并利用该数据集为对齐训练和自动评估奠定人类隐私规范的基础。在这些注释的基础上，我们首先展示了在相同提示的参考响应中，基于人类注释和解释对LLM判断进行条件化，使其判断更加可靠。然后，我们引入了基于注释的奖励建模，该模型在强化学习（RL）过程中利用这些注释对新响应进行评分，并展示了使用这种奖励训练的小型开放权重代理在隐私对齐方面更好地符合人类隐私规范，在PrivacyAlign和现有代理隐私基准上取得了显著提升。
cs.CL / 61 / 2606.21718
Leveraging LaBSE with Progressive Curriculum Learning for Multicultural Polarization
利用渐进式课程学习的LaBSE在多文化极化中的应用
Abstract
Detecting online polarization remains a critical challenge, particularly in multilingual and multicultural contexts where intergroup hostility is prevalent. The problem is particularly challenging due to the data scarcity for these tasks in the low-resource languages. Identifying such phenomena has become an active area of research and is addressed in SemEval-2026 Task 9: Multilingual, Multicultural Online Polarization Detection. To address this problem we propose an architecture that leverages LaBSE embeddings - an unconventional choice typically reserved for retrieval tasks, to obtain strong cross-lingual learning which enhances scores in low-resource language by a score up to 0.2 macro F1. Furthermore, we provide a comprehensive ablation study evaluating the performance of diverse encoder models in the Qwen model family within a retrieval-based prompting framework. Our code will be soon available at https://github.com/carrycurious/PolarMind.
Chinese Translation
在线极化的检测仍然是一个关键挑战，尤其是在多语言和多文化背景下，群体间敌意普遍存在。由于低资源语言在这些任务中的数据稀缺性，这一问题尤为棘手。识别此类现象已成为一个活跃的研究领域，并在SemEval-2026任务9中得到了关注：多语言、多文化在线极化检测。为了解决这一问题，我们提出了一种架构，利用LaBSE嵌入——一种通常用于检索任务的非常规选择，以获得强大的跨语言学习，从而在低资源语言中提高高达0.2的宏F1分数。此外，我们提供了一项全面的消融研究，评估了在基于检索的提示框架中，Qwen模型家族中不同编码器模型的性能。我们的代码将很快在https://github.com/carrycurious/PolarMind上发布。
cs.CL / 62 / 2606.21724
Denoising Iterative Self-Correction: Structured Verification Loops for Reliable LLM Reasoning
去噪迭代自我修正：可靠大语言模型推理的结构化验证循环
Abstract
Large language models produce fluent but often incorrect multi-step reasoning, and naive correction methods risk degrading already-correct answers. We introduce Denoising Iterative Self-Correction (DISC), a test-time procedure that treats verification question outputs as noisy measurements of where a solution may be corrupted. Using these signals, DISC progressively reduces errors across multiple verify-judge-correct passes, analogous to traditional iterative denoising. A binary judgment gate controls correction precision by blocking rewrites that would damage already-correct answers while the verifier and corrector together repair errors. We evaluate this trade-off using two paired diagnostics: an improvement-to-degradation ratio (precision) and a repair rate (recall). Across three benchmarks (BIG-Bench Mistake, HotpotQA, GPQA Diamond) and four models, DISC dominates Chain-of-Verification and Self-Refine on the precision-recall trade-off, reaching 81.6% accuracy with 13x more improvements per degradation than Chain-of-Verification and 5x more than Self-Refine on BIG-Bench Mistake (Sonnet~4.5). On GPQA Diamond, we identify a capability floor below which judges acknowledge contradictions in evidence but cannot translate that recognition into a correction. We further show that cross-model role allocation -- assigning verification and judgment to a model different from the generator -- mitigates self-confirmation bias.
Chinese Translation
大型语言模型能够生成流畅但常常不正确的多步骤推理，而简单的修正方法可能会降低已经正确答案的质量。我们提出了去噪迭代自我修正（Denoising Iterative Self-Correction, DISC），这是一种在测试时的程序，将验证问题的输出视为解决方案可能被破坏的噪声测量。利用这些信号，DISC通过多个验证-判断-修正的过程逐步减少错误，类似于传统的迭代去噪。一个二元判断门控控制修正的精度，通过阻止可能损害已正确答案的重写，同时验证者和修正者共同修复错误。我们使用两个配对诊断来评估这种权衡：改进与降级比率（精度）和修复率（召回率）。在三个基准测试（BIG-Bench Mistake、HotpotQA、GPQA Diamond）和四个模型上，DISC在精度-召回权衡中优于验证链（Chain-of-Verification）和自我修正（Self-Refine），在BIG-Bench Mistake（Sonnet~4.5）上达到81.6%的准确率，改进与降级的比率是验证链的13倍，自我修正的5倍。在GPQA Diamond上，我们识别出一个能力底线，低于该底线的判断者承认证据中的矛盾，但无法将这种认识转化为修正。我们进一步表明，跨模型角色分配——将验证和判断分配给与生成器不同的模型——可以减轻自我确认偏差。
cs.CL / 63 / 2606.21777
CalVerT: Augmenting Agents with Calibrated Verifier Telemetry Improves Action and Learning in Knowledge-Intensive Tasks
CalVerT：通过校准验证者遥测增强代理，提高知识密集型任务中的行动和学习
Abstract
LLM agents in knowledge intensive question answering take retrieval and reasoning actions with incomplete knowledge about whether their current answer is uncertain, unsupported, or already complete. This produces two failure modes: committing to confident but unsupported answers, which hurts accuracy, and over-retrieving when the evidence in hand already suffices, resulting in wasted compute. To give agents a more complete picture of the state space they are operating in, we introduce calibrated verifier telemetry (CalVerT), which augments the agent's state with additional telemetry: a calibrated self-confidence score and a grounding verifier score. We show that CalVerT can improve agents in both training-free and training-based settings. On four QA benchmarks, we find that CalVerT raises F1 by triggering retrieval in cases where agents over-rely on parametric knowledge, while cutting redundant retrieval in cases where agents have sufficient context to answer. We show that CalVerT can augment existing QA frameworks without training. Moreover, CalVerT also improves trained systems: by simply augmenting an agent's state with telemetry, we observe improvements after reinforcement learning, as compared to an agent with identical training but no CalVerT telemetry.
Chinese Translation
在知识密集型问答中，LLM 代理在对其当前答案是否不确定、缺乏支持或已经完整的知识不充分的情况下进行检索和推理。这产生了两种失败模式：一是对自信但缺乏支持的答案做出承诺，导致准确性下降；二是在已有证据足够的情况下过度检索，造成计算资源浪费。为了让代理对其操作的状态空间有更全面的了解，我们引入了校准验证者遥测（CalVerT），该方法通过额外的遥测信息增强代理的状态：校准的自信度评分和基础验证者评分。我们展示了 CalVerT 在无训练和基于训练的设置中均能改善代理的表现。在四个问答基准上，我们发现 CalVerT 通过在代理过度依赖参数知识的情况下触发检索，提高了 F1 分数，同时在代理拥有足够上下文以回答的情况下减少了冗余检索。我们表明 CalVerT 可以在不进行训练的情况下增强现有的问答框架。此外，CalVerT 还改善了经过训练的系统：通过简单地用遥测增强代理的状态，我们观察到在强化学习后表现有所提升，相较于具有相同训练但没有 CalVerT 遥测的代理。
cs.CL / 64 / 2606.21802
When to Plan, When to Polish: Noise Level as a Granularity Axis for Diffusion Language Models
何时规划，何时润色：噪声水平作为扩散语言模型的粒度轴
Abstract
Standard tokenwise diffusion LMs keep training corruption and inference commitment at token granularity throughout denoising. At high noise, this leaves scattered local fragments rather than coherent evidence, making it hard to form early coarse structure, exactly what planning-sensitive generation requires. Hierarchical planning methods add coarse stages to separate planning from wording, but they need extra planners, block latents, or two stage designs. We propose Noise Dependent Granularity Control (NDGC), a single-level diffusion method that uses the noise level as a granularity cue. NDGC aligns training exposure and inference commitment with denoising progress. High noise steps use coherent token groups to support early meaning commitment, while low noise steps return to token level refinement. This creates planning like coarse to fine denoising without an explicit planner or hierarchical architecture. Across controlled tests, ablations, and WritingPrompts, NDGC shows earlier skeleton formation, better ordered recovery, and healthier outputs.
Chinese Translation
标准的逐词扩散语言模型在去噪过程中始终保持训练损坏和推理承诺在逐词粒度上。在高噪声情况下，这导致散乱的局部片段而非连贯的证据，使得形成早期粗略结构变得困难，这正是规划敏感生成所需的。层次规划方法增加了粗略阶段，以将规划与措辞分开，但它们需要额外的规划器、阻塞潜变量或两阶段设计。我们提出了噪声依赖粒度控制（Noise Dependent Granularity Control, NDGC），这是一种单级扩散方法，利用噪声水平作为粒度线索。NDGC 将训练暴露和推理承诺与去噪进度对齐。高噪声步骤使用连贯的词组来支持早期意义承诺，而低噪声步骤则返回到逐词级别的细化。这创造了类似于粗到细的去噪规划，而无需显式的规划器或层次结构。在控制测试、消融实验和写作提示中，NDGC 显示出更早的骨架形成、更好的有序恢复和更健康的输出。
cs.CL / 65 / 2606.21803
Test-Time Training with Next-Token Prediction
基于下一个标记预测的测试时训练
Abstract
Next-token prediction is the self-supervised signal that trains language models, and every observed prompt token provides the same signal at test time. We study whether this signal can define the inner-loop objective for test-time training (TTT) in pretrained long-context language models. Many TTT architectures require models to be trained with test-time adaptation in mind, limiting their direct applicability to released LLM checkpoints. While recent in-place TTT methods make fast-weight adaptation possible for pretrained LLMs without redesigning the backbone, they leave a central question unresolved: what should each test-time write store? Existing recipes train the fast weight to match a learned local value proxy but they are not directly tied to the self-supervised next-token prediction signal. We introduce Test-Time Training with Next-Token Prediction (TTT-NTP), a drop-in fast-weight adaptation method for pretrained LLMs that instead supervises updates using the model's own next contextual hidden state. This makes each local write follow the same causal computation that supports next-token prediction: the value target is a pointwise linear projection of a single next-position contextual state. On RULER Full-13 (averaged over 4k, 8k, 16k, and 32k context lengths), TTT-NTP is the only method that consistently improves the released backbone across four models spanning three families and a 0.6--8B size range: Llama-3.1-8B (+3.9), Mistral-7B-v0.3 (+3.0), and the Qwen3 series (Qwen3-4B +4.1, Qwen3-0.6B +2.9). On the real-world LongBench-v2 long-document QA benchmark, TTT-NTP improves over the base model on both Llama-3.1-8B (+5.6) and Mistral-7B-v0.3 (+3.7), while preserving commonsense and knowledge performance.
Chinese Translation
下一个标记预测是训练语言模型的自监督信号，每个观察到的提示标记在测试时提供相同的信号。我们研究这种信号是否可以定义预训练长上下文语言模型中测试时训练（TTT）的内循环目标。许多TTT架构要求模型在考虑测试时适应的情况下进行训练，这限制了它们对已发布的LLM检查点的直接适用性。尽管最近的就地TTT方法使得预训练LLM的快速权重适应成为可能，而无需重新设计主干网络，但它们留下了一个核心问题未解：每个测试时写入存储应该是什么？现有的方法训练快速权重以匹配学习到的局部值代理，但它们并未直接与自监督的下一个标记预测信号相联系。我们提出了基于下一个标记预测的测试时训练（TTT-NTP），这是一种适用于预训练LLM的快速权重适应方法，它使用模型自身的下一个上下文隐藏状态来监督更新。这使得每个局部写入遵循支持下一个标记预测的相同因果计算：值目标是单个下一个位置上下文状态的逐点线性投影。在RULER Full-13（在4k、8k、16k和32k上下文长度上取平均）上，TTT-NTP是唯一一种在跨越三个家族和0.6-8B大小范围的四个模型中持续改善已发布主干的的方法：Llama-3.1-8B (+3.9)，Mistral-7B-v0.3 (+3.0)，以及Qwen3系列（Qwen3-4B +4.1，Qwen3-0.6B +2.9）。在真实世界的LongBench-v2长文档问答基准上，TTT-NTP在Llama-3.1-8B (+5.6)和Mistral-7B-v0.3 (+3.7)上均优于基础模型，同时保持了常识和知识性能。
cs.CL / 66 / 2606.21807
Fixed RAG Compression Collapses Measured Reader Scaling
固定的 RAG 压缩导致测量的读者扩展崩溃
Abstract
Retrieval-Augmented Generation (RAG) compression papers often evaluate a compressor on one to three readers and treat the compressed evidence layer as evaluation-neutral. We show this assumption is false: fixed compression can raise average accuracy while hiding reader upgrades and reversing model rankings. Across 20 readers and ten domain-method settings over four QA benchmarks and one summarization benchmark, compression gain decreases with reader baseline (nine of ten settings significant, p < 0.05). Generic summarization flips 31% of pairwise model rankings on LongMemEval-S, and a fixed HotpotQA compressor hides 80% of the raw upgrade from Qwen 7B to GPT-4.1-mini. Two opposing forces explain this paradox: compression rescues weak readers by removing noise they cannot filter, and harms strong readers by dropping details they would have used. The pattern appears across structured compilation, generic summarization, three trained compressor families, query-focused summarization, and an external audit of nine published compression papers. We release ragscale, a toolkit built on 177,000 row-level compression transitions, so any compression paper can audit reader scaling with three readers in one day.
Chinese Translation
检索增强生成（RAG）压缩论文通常在一到三个读者上评估压缩器，并将压缩的证据层视为评估中立。我们证明这一假设是错误的：固定压缩可以提高平均准确性，同时隐藏读者升级并逆转模型排名。在四个问答基准和一个摘要基准的20个读者和十个领域-方法设置中，压缩增益随着读者基线的提高而减少（十个设置中有九个显著，p < 0.05）。在 LongMemEval-S 上，通用摘要翻转了31%的成对模型排名，而固定的 HotpotQA 压缩器隐藏了从 Qwen 7B 到 GPT-4.1-mini 的80%的原始升级。两种相反的力量解释了这一悖论：压缩通过去除弱读者无法过滤的噪声来拯救他们，而通过丢弃强读者本可以使用的细节来伤害他们。这一模式出现在结构化编译、通用摘要、三种训练的压缩器系列、查询聚焦摘要以及对九篇已发表压缩论文的外部审计中。我们发布了 ragscale，这是一个基于177,000个行级压缩过渡构建的工具包，使任何压缩论文都可以在一天内通过三个读者审计读者扩展。
cs.CL / 67 / 2606.21844
Inverse Turing Bench: Evaluating Language Models as Judges of Human vs. AI Dialogue
逆图灵基准：评估语言模型作为人类与人工智能对话的评判者
Abstract
As AI systems integrate into online spaces, differentiating them from humans in conversations is increasingly important. We present Inverse Turing Bench, a benchmark that evaluates LLMs and other models on their ability to differentiate humans and AI in multi-turn text. The benchmark provides a collection of paired dialogue transcripts, wherein one dialogue is between two humans and the other is between a human and an AI. The task is to correctly identify which dialogue is human-only vs. human-AI. We evaluated a preliminary set of models against this benchmark, and found that GPTZero, Claude Opus-4.6, and GPT-5.5 achieve the highest accuracy: 89.41%, 77.92%, and 75.94% respectively. Our results suggest that statistical approaches to detection have semantic blind spots, but semantic approaches are susceptible to persona-prompting. Our work speaks to the Inverse Turing Test as a probe of LLM theory of mind, and motivates human-AI differentiation as a critical capability for AI systems. Our live benchmark can be found at https://huggingface.co/spaces/roc-hci/Inverse-Turing-Bench-Leaderboard (anonymity preserved).
Chinese Translation
随着人工智能系统逐渐融入在线空间，在对话中区分人类与人工智能的重要性日益增加。我们提出了逆图灵基准（Inverse Turing Bench），这是一个评估大型语言模型（LLMs）及其他模型在多轮文本中区分人类与人工智能能力的基准。该基准提供了一组配对对话文本，其中一段对话是两个人类之间的交流，另一段则是人类与人工智能之间的交流。任务是正确识别出哪段对话是仅由人类进行的，哪段是人类与人工智能的对话。我们对一组初步模型进行了评估，发现GPTZero、Claude Opus-4.6和GPT-5.5分别达到了最高的准确率：89.41%、77.92%和75.94%。我们的结果表明，统计检测方法存在语义盲点，而语义方法则容易受到个性化提示的影响。我们的研究涉及逆图灵测试（Inverse Turing Test），作为大型语言模型的心智理论探测，并推动人类与人工智能的区分作为人工智能系统的一项关键能力。我们的实时基准可以在 https://huggingface.co/spaces/roc-hci/Inverse-Turing-Bench-Leaderboard（匿名性已保留）找到。
cs.CL / 68 / 2606.21848
Keyless Attention: Value-Space Routing and Value-Only Caching for Efficient Transformers
无键注意力：价值空间路由和仅基于价值的缓存以提高变换器的效率
Abstract
We propose Keyless Attention, an attention mechanism that eliminates the key projection entirely, operating over queries and values only. This yields a Value-Only Cache that reduces KV cache memory and access overhead by exactly 50% over standard attention, while matching or exceeding standard attention's decode throughput. Beyond efficiency, we introduce Depth-$m$ Attention Factorization: standard attention computes a depth-2 factorization of the attention bilinear form, while Keyless Attention realizes a depth-$m$ instance of this family. At m=3, Keyless Attention matches the projection matrix count of standard attention via a value-space routing matrix that replaces the key projection and introduces a coupling between routing and retrieval. Experiments across five models and four architectures (GPT-2 280M, GPT-2 557M, Pythia 410M, Qwen2 1.5B, and Llama 3.2 1B) show that Keyless Attention matches or outperforms standard QKV attention on perplexity in 4 out of 5 models. On downstream zero-shot evaluation (GPT-2 557M), Keyless Attention outperforms on 4 out of 5 commonsense reasoning benchmarks, while achieving 50% KV cache reduction throughout.
Chinese Translation
我们提出了无键注意力（Keyless Attention），一种完全消除键投影的注意力机制，仅在查询和价值上进行操作。这产生了一个仅基于价值的缓存（Value-Only Cache），将KV缓存内存和访问开销减少了50%，同时匹配或超过标准注意力的解码吞吐量。除了效率之外，我们引入了深度-$m$注意力分解：标准注意力计算注意力双线性形式的深度-2分解，而无键注意力实现了该家族的深度-$m$实例。当m=3时，无键注意力通过一个价值空间路由矩阵匹配标准注意力的投影矩阵数量，该矩阵替代了键投影并引入了路由与检索之间的耦合。在五个模型和四种架构（GPT-2 280M、GPT-2 557M、Pythia 410M、Qwen2 1.5B和Llama 3.2 1B）上的实验表明，无键注意力在5个模型中的4个上与标准QKV注意力在困惑度上相匹配或表现更好。在下游零-shot评估（GPT-2 557M）中，无键注意力在5个常识推理基准中有4个超越了标准方法，同时在整个过程中实现了50%的KV缓存减少。
cs.CL / 69 / 2606.21851
TALAS: Teacher-Anchored Layer Alignment with Adaptive Sharpness-Aware Minimization for Embedding Distillation
TALAS：基于教师锚定层对齐的自适应敏感度最小化用于嵌入蒸馏
Abstract
Knowledge Distillation (KD) has established itself as a pivotal technique for compressing large pre-trained language models. However, existing methods that force a student to strictly mimic the teacher's sentence embeddings or internal features often incur prohibitive computational costs and yield suboptimal performance due to the inherent capacity gap. To address these challenges, we propose TALAS (Teacher-Anchored Layer Alignment with Sharpness-aware minimization), a unified framework that synergizes hierarchical (multi-layer) alignment with robust optimization. First, we introduce a Teacher-Anchored mechanism that selectively distills final sentence embeddings only into the student's upper layers, thereby reducing overhead while respecting capacity constraints. Second, we bridge the semantic gap in lower layers via Layer-Aligned Self-Distillation, which propagates knowledge top-down using internal geometric relational constraints in the embedding space. Finally, to prevent the student from memorizing point-wise teacher noise, we integrate Adaptive Sharpness-Aware Minimization (ASAM) into the training objective, guiding the model towards flat minima for enhanced generalization. Empirical results on standard sentence embedding benchmarks demonstrate that TALAS consistently outperforms strong distillation baselines while achieving superior training efficiency in terms of computational cost and memory footprint.
Chinese Translation
知识蒸馏（Knowledge Distillation, KD）已成为压缩大型预训练语言模型的关键技术。然而，现有方法强迫学生严格模仿教师的句子嵌入或内部特征，往往会导致高昂的计算成本，并因固有的能力差距而产生次优性能。为了解决这些挑战，我们提出了TALAS（基于教师锚定层对齐的敏感度感知最小化），这是一个将分层（多层）对齐与鲁棒优化相结合的统一框架。首先，我们引入了一种教师锚定机制，仅将最终句子嵌入选择性地蒸馏到学生的上层，从而减少开销，同时尊重能力约束。其次，我们通过层对齐自蒸馏（Layer-Aligned Self-Distillation）弥合下层的语义差距，该方法利用嵌入空间中的内部几何关系约束自上而下地传播知识。最后，为了防止学生记忆逐点的教师噪声，我们将自适应敏感度感知最小化（Adaptive Sharpness-Aware Minimization, ASAM）集成到训练目标中，引导模型朝向平坦的极小值，以增强泛化能力。在标准句子嵌入基准上的实证结果表明，TALAS在计算成本和内存占用方面实现了优越的训练效率，同时始终优于强蒸馏基线。
cs.CL / 70 / 2606.21869
The Language-Energy Divide: Measuring Energy Costs of Multilingual LLM Inference
语言与能源的鸿沟：多语言大语言模型推理的能源成本测量
Abstract
Large language models (LLMs) are increasingly deployed in multilingual settings, yet the energy costs of serving these models across different languages remain poorly understood. We present a systematic study of inference energy consumption across languages with ML.Energy framework (Chung et al., 2026). We find striking disparities: energy consumption per output token varies by up to 8.3 times across languages, while total energy for a fixed set of requests varies by up to 179 times between the cheapest (English, 17.6 kJ) and the most expensive (Pashto, 3,147 kJ) languages. Our analysis shows that this disparity is driven by two compounding factors: (1) higher per-token energy costs for languages using complex or rare scripts, and (2) more tokens generated for low-resource languages. Moreover, we find a double cost + performance penalty: languages with the highest energy footprints also tend to achieve the lowest task accuracy. We reveal that the energy divide persists across models, hardware, and tasks, suggesting a systemic energy inequity in multilingual LLM deployment. Finally, we recommend that the community treat energy as a first-class evaluation axis, extend reporting checklists and model cards to include it, and adopt deployment-side mitigations for better energy efficiency.
Chinese Translation
大型语言模型（LLMs）越来越多地在多语言环境中部署，但在不同语言中服务这些模型的能源成本仍然不够清楚。我们利用 ML.Energy 框架（Chung et al., 2026）对不同语言的推理能源消耗进行了系统研究。我们发现显著的差异：每个输出标记的能源消耗在不同语言之间差异高达 8.3 倍，而对于固定请求集的总能源消耗在最便宜的语言（英语，17.6 kJ）和最昂贵的语言（普什图语，3,147 kJ）之间差异高达 179 倍。我们的分析表明，这种差异是由两个相互影响的因素驱动的：（1）使用复杂或稀有书写系统的语言每个标记的能源成本较高，以及（2）低资源语言生成的标记数量更多。此外，我们发现存在双重成本 + 性能惩罚：能源足迹最高的语言往往在任务准确性上表现最低。我们揭示了这种能源鸿沟在不同模型、硬件和任务中普遍存在，表明多语言 LLM 部署中存在系统性的能源不平等。最后，我们建议社区将能源视为一项重要的评估指标，扩展报告清单和模型卡以包括能源信息，并采取部署侧的缓解措施以提高能源效率。
cs.CL / 71 / 2606.21890
Scaling Performance and Low-Resource Annotation with Many-Shot In-Context Learning for Named Entity Recognition
基于多示例上下文学习的命名实体识别的性能扩展与低资源注释
Abstract
In-context learning (ICL) with large language models (LLMs) has emerged as a powerful alternative to fine-tuning for Named Entity Recognition (NER), achieving strong performance with minimal annotation and no additional training. However, prior work has shown that despite their adaptability, LLMs still lag behind fully supervised models such as fine-tuned BERT in structured tasks like NER. While existing studies on ICL for NER have mainly explored few-shot settings, the potential of scaling to hundreds of demonstrations has not been thoroughly investigated. To address this gap, we conduct a comprehensive investigation of many-shot ICL for NER and further explore its effectiveness in annotating and refining data for low-resource NER tasks. Specifically, we evaluate various LLMs across multiple domains using hundreds of ICL examples and then assess the feasibility of using many-shot ICL as a data annotation framework. Our experiments demonstrate that: (1) scaling to hundreds of in-context examples enables LLMs to match or even surpass the performance of fully supervised BERT models; and (2) using about one hundred human-labeled examples as demonstrations, many-shot in-context annotation can generate high-quality labeled data, leading to approximately 10% absolute F1 improvement over existing state-of-the-art approaches when used to fine-tune BERT on low-resource NER.
Chinese Translation
基于大型语言模型（LLMs）的上下文学习（ICL）已成为命名实体识别（NER）的一种强大替代方案，能够在最小注释和无需额外训练的情况下实现出色的性能。然而，先前的研究表明，尽管LLMs具有适应性，但在结构化任务（如NER）中，它们仍然落后于完全监督的模型，如微调后的BERT。虽然现有关于NER的ICL研究主要探讨了少量示例的设置，但扩展到数百个示例的潜力尚未得到充分研究。为了解决这一空白，我们对NER的多示例ICL进行了全面调查，并进一步探讨其在低资源NER任务中注释和完善数据的有效性。具体而言，我们在多个领域评估了各种LLMs，使用数百个ICL示例，然后评估将多示例ICL作为数据注释框架的可行性。我们的实验表明：（1）扩展到数百个上下文示例使LLMs能够匹配甚至超越完全监督的BERT模型的性能；（2）使用约一百个人工标注的示例作为演示，多示例上下文注释可以生成高质量的标注数据，在用于微调BERT进行低资源NER时，相较于现有的最先进方法，F1值的绝对提升约为10%。
cs.CL / 72 / 2606.21895
Olfactory-Inspired Sparse Combinatorial Coding for Low-Resource Named Entity Recognition
基于嗅觉启发的稀疏组合编码用于低资源命名实体识别
Abstract
Named Entity Recognition (NER) in low-resource languages suffers from limited supervision and a lack of high-quality pretrained embeddings. Biological olfaction, which relies on sparse combinatorial coding through receptor and glomerular organization, offers a compelling paradigm for learning robust representations under uncertainty. In this paper, we introduce a receptor-glomerular bottleneck - a novel, biologically-inspired olfactory architecture - between standard token embeddings and a BiLSTM-CRF sequence model. We evaluate our architecture across six multilingual datasets trained entirely from scratch (without pre-trained embeddings) under varied data-scale conditions, including a strict 1k-sentence low-resource control. Our results demonstrate that introducing a representation bottleneck yields F1 score improvements under severe data scarcity, primarily by acting as a powerful regularizer. Under the 1k capped training condition, at least one olfactory-inspired configuration achieves the highest mean F1 score across all six datasets. While these improvements represent near-ties with generic bottleneck controls for most languages, the olfactory architecture provides a significant advantage in languages like Bangla (+6.23% F1 over standard baseline and +8.47% F1 over the best control baseline) where generic bottlenecks degrade performance. We also observe improvements in the ultra-low-resource Telugu setting (+4.43% F1) at full-scale, and find that sparse specialization naturally emerges within the receptor layer. Our findings suggest that structured sparse coding inspired by olfactory networks serves as an effective inductive bias and regularizer when representations must be learned from limited or noisy supervision.
Chinese Translation
低资源语言中的命名实体识别（NER）面临有限的监督和缺乏高质量的预训练嵌入的问题。生物嗅觉依赖于通过受体和小球体组织进行的稀疏组合编码，为在不确定性下学习稳健的表征提供了一个引人注目的范式。在本文中，我们引入了一种受体-小球体瓶颈——一种新颖的、生物启发的嗅觉架构——在标准的标记嵌入和BiLSTM-CRF序列模型之间。我们在六个多语言数据集上评估了我们的架构，这些数据集完全从头开始训练（没有预训练嵌入），并在不同的数据规模条件下进行，包括严格的1k句子低资源控制。我们的结果表明，引入表征瓶颈在严重数据稀缺的情况下提高了F1得分，主要是通过充当强大的正则化器。在1k限制训练条件下，至少有一种嗅觉启发的配置在所有六个数据集上达到了最高的平均F1得分。尽管这些改进在大多数语言中与通用瓶颈控制接近，但嗅觉架构在像孟加拉语这样的语言中提供了显著的优势（比标准基线高出6.23%的F1，且比最佳控制基线高出8.47%的F1），而通用瓶颈则降低了性能。我们还观察到在超低资源的泰卢固语环境中（在全规模下提高了4.43%的F1）也有改进，并发现稀疏专业化在受体层中自然出现。我们的研究结果表明，受嗅觉网络启发的结构化稀疏编码在必须从有限或嘈杂的监督中学习表征时，作为有效的归纳偏置和正则化器。
cs.CL / 73 / 2606.21904
Which Review Aspect Has a Greater Impact on the Duration of Open Peer Review in Multiple Rounds? -- Evidence from Nature Communications
哪个评审方面对多轮开放同行评审的持续时间影响更大？——来自《自然通讯》的证据
Abstract
Purpose: Peer review is essential to scientific publishing, but increasing submission volumes have placed growing pressure on reviewers and editors. This study examines the relationship between sentiment toward specific review aspects and peer review duration. It also investigates how this relationship varies across disciplines and review rounds, with the aim of supporting targeted manuscript revision and improving review efficiency. Design/methodology/approach: We adopt a two-stage approach. First, fine-grained aspects are extracted from peer review reports, and a sentiment classification model is used to determine the sentiment associated with each aspect. Second, correlations between aspect-level sentiment and peer review duration are analyzed. Sentiment scores are also calculated for different review rounds to determine whether these relationships change over successive rounds. Findings: Review sentiment has a weak but statistically significant negative correlation with peer review duration, indicating that more positive reviews tend to be associated with shorter review periods. Aspects concerning Evaluation and Results and Impact and Research Value show relatively stronger correlations with review duration. The relationships between aspect-level sentiment and review duration also differ significantly across review rounds. Originality/value: This study connects the textual content of peer review reports with the temporal characteristics of the review process. By identifying review aspects that are more closely associated with review duration, it provides evidence that may help authors prioritize revisions and assist reviewers and editors in improving review efficiency. The findings contribute to reducing the burden of peer review and accelerating scholarly communication and knowledge dissemination.
Chinese Translation
目的：同行评审对科学出版至关重要，但日益增加的提交量给评审者和编辑带来了越来越大的压力。本研究考察了对特定评审方面的情感与同行评审持续时间之间的关系。它还调查了这种关系在不同学科和评审轮次中的变化，旨在支持针对性的手稿修订并提高评审效率。设计/方法论/方法：我们采用两阶段的方法。首先，从同行评审报告中提取细粒度的评审方面，并使用情感分类模型确定与每个方面相关的情感。其次，分析方面级情感与同行评审持续时间之间的相关性。还计算了不同评审轮次的情感得分，以确定这些关系是否在连续轮次中发生变化。发现：评审情感与同行评审持续时间之间存在微弱但统计显著的负相关，表明更积极的评审往往与较短的评审周期相关。关于评估与结果以及影响与研究价值的方面与评审持续时间显示出相对较强的相关性。方面级情感与评审持续时间之间的关系在不同评审轮次中也存在显著差异。原创性/价值：本研究将同行评审报告的文本内容与评审过程的时间特征联系起来。通过识别与评审持续时间更密切相关的评审方面，提供了可能帮助作者优先修订并协助评审者和编辑提高评审效率的证据。这些发现有助于减轻同行评审的负担，加速学术交流和知识传播。
cs.CL / 74 / 2606.21906
Deeper is Not Always Better: Mitigating the Alignment Tax via Confident Layer Decoding
深度并不总是更好：通过自信层解码减轻对齐代价
Abstract
Autoregressive generation in large language models (LLMs) conventionally decodes from the final layer, assuming that deeper representations yield more reliable next-token predictions. We revisit this assumption by revealing a recurring Guess-Refine-Perturb dynamic: early layers form coarse guesses, intermediate layers refine reasoning-relevant semantics, and final layers can perturb these refined predictions toward generic or alignment-preferred tokens. We introduce Confident Decoding, a training-free decoding strategy that dynamically selects the most reliable near-final layer through entropy-guided conservative backward search. We further provide a theoretical formulation of layer selection as an optimal stopping problem, showing that under bounded projection noise and dominant late-stage alignment perturbation, our search rule filters perturbation while bounding the loss relative to the oracle refinement layer. Experiments across dense and Mixture-of-Experts LLMs demonstrate consistent gains on challenging reasoning benchmarks, including GPQA-Diamond, Omni-MATH, and HLE, with zero memory overhead and less than 2% latency increase. These results suggest dynamically bypassing final-layer perturbations can unlock stronger reasoning behavior from aligned LLMs.
Chinese Translation
在大型语言模型（LLMs）中，自回归生成通常从最后一层解码，假设更深的表示能够提供更可靠的下一个标记预测。我们重新审视这一假设，揭示了一个反复出现的猜测-精炼-扰动动态：早期层形成粗略的猜测，中间层精炼与推理相关的语义，而最后一层可能会将这些精炼的预测扰动到更通用或对齐偏好的标记。我们引入了自信解码（Confident Decoding），这是一种无训练的解码策略，通过熵引导的保守反向搜索动态选择最可靠的近最终层。我们进一步提供了层选择的理论公式，将其视为一个最优停止问题，表明在有界投影噪声和主导的后期对齐扰动下，我们的搜索规则能够过滤扰动，同时将损失限制在相对于理想精炼层的范围内。在密集型和专家混合型 LLMs 上的实验表明，在包括 GPQA-Diamond、Omni-MATH 和 HLE 在内的具有挑战性的推理基准上，均取得了一致的提升，且内存开销为零，延迟增加不到 2%。这些结果表明，动态绕过最后一层的扰动可以释放对齐 LLMs 更强的推理能力。
cs.CL / 75 / 2606.21917
Pre-Generation Hallucination Detection in Large Language Models via Soft-Target Attention Probing
通过软目标注意力探测在大语言模型中进行生成前幻觉检测
Abstract
Detecting hallucination risk before generation enables abstention, retrieval augmentation, and routing decisions without incurring the cost of decoding. While prior work has shown that such risk can be estimated from a model's internal representations, existing approaches treat this as binary classification over a single decoded output. We instead formulate it as a risk-estimation problem. Under this formulation, we introduce soft-target supervision based on the empirical answer error rate over stochastically sampled outputs - an estimator we prove to be the unique unbiased minimum-variance estimator of the model's per-prompt error probability under its sampling distribution. We further adapt attention probing to the pre-generation setting, enabling the detector to selectively aggregate hallucination-relevant prompt representations. Across three question-answering benchmarks and five models, attention probing outperforms linear probing on short-answer tasks. Replacing binary labels with soft-target supervision further and consistently improves detection quality.
Chinese Translation
在生成之前检测幻觉风险可以实现避免生成、检索增强和路由决策，而无需承担解码的成本。虽然先前的研究表明，可以从模型的内部表示中估计这种风险，但现有的方法将其视为对单个解码输出的二元分类。我们则将其表述为风险估计问题。在这一表述下，我们引入基于随机采样输出的经验答案错误率的软目标监督——我们证明这一估计量是模型在其采样分布下每个提示错误概率的唯一无偏最小方差估计量。我们进一步将注意力探测适应于生成前设置，使得检测器能够选择性地聚合与幻觉相关的提示表示。在三个问答基准和五个模型的实验中，注意力探测在短答案任务上优于线性探测。用软目标监督替换二元标签进一步且持续地提高了检测质量。
cs.CL / 76 / 2606.21930
MindTailor: Personalized Emotional Support via Post History-Grounded Case Formulation and Collaborative Refinement
MindTailor：通过历史帖子基础的案例构建与协作优化提供个性化情感支持
Abstract
As mental health concerns continue to rise globally, social media has emerged as a vital space where individuals seek emotional support. While prior work on personalized emotional support has leveraged seekers' emotional states, personas, and situational context, these approaches primarily capture the seeker's current state, overlooking the formative experiences that shape present concerns. In this work, we propose MindTailor, a framework that generates personalized emotional support responses by constructing a case formulation from the seeker's post history and iteratively refining responses through collaborative critique among counselor agents grounded in distinct counseling strategies. To enable research on this history-aware task, we construct ReddiSupp, a dataset of 798 Reddit posts paired with seekers' prior post histories. Through LLM-as-a-Judge evaluation, expert human evaluation, and a user study with seekers, we demonstrate that MindTailor outperforms baselines across these evaluations, improving empathy, personalization, understanding, and achieving the highest overall preference.
Chinese Translation
随着全球心理健康问题的持续上升，社交媒体已成为个体寻求情感支持的重要空间。尽管以往的个性化情感支持研究利用了求助者的情感状态、角色特征和情境背景，但这些方法主要关注求助者的当前状态，忽视了塑造其当前关注点的形成性经历。在本研究中，我们提出了MindTailor，一个通过构建求助者的历史帖子案例形成并通过基于不同咨询策略的咨询代理之间的协作批评迭代优化响应的框架，以生成个性化的情感支持响应。为了支持这一关注历史的任务研究，我们构建了ReddiSupp，一个包含798条Reddit帖子及其求助者之前帖子历史的数据集。通过LLM-as-a-Judge评估、专家人工评估和与求助者的用户研究，我们证明了MindTailor在这些评估中优于基线，提升了同理心、个性化、理解力，并获得了最高的整体偏好。
cs.CL / 77 / 2606.21939
Beyond Value Benchmarks: Measuring Value-Structure Alignment in Large Language Models via Symmetric Q-Sorts
超越价值基准：通过对称Q排序测量大型语言模型中的价值结构一致性
Abstract
Large Language Models (LLMs) are increasingly deployed in contexts requiring complex moral reasoning and value trade-offs. However, existing evaluations typically rely on item-level behavioral metrics, which fail to capture how models structurally prioritize competing values as a cohesive system. To address this, we propose a symmetric human-LLM evaluation framework, grounded in Q methodology, to measure value-structure alignment. Under our protocol, humans and models sort an identical 140-item moral statement set into a shared nine-column forced distribution; for LLMs, we elicit strict rankings and deterministically map them to Q-sort buckets. Using a human reference sample ($N=35$), we establish a stable three-factor reference geometry specific to this instrument and sample. We evaluate 12 LLMs across four model families via 240 replicated Q-sorts at two temperature settings, quantifying structural alignment via Procrustes similarity ($\phi$) and RSA-based Spearman correlation ($\rho$). Our results reveal significant cross-family heterogeneity, model-specific sensitivity to generation stochasticity and localized misalignment, which demonstrate that favorable global scores can obscure underlying regional distortions. While rank- and bucket-based analyses remain highly consistent, prompt phrasing introduces notable variance. Ultimately, assessing value-structure alignment provides a crucial structural complement to traditional itemwise moral benchmarks.
Chinese Translation
大型语言模型（LLMs）越来越多地应用于需要复杂道德推理和价值权衡的场景。然而，现有的评估通常依赖于项目级的行为指标，这无法捕捉模型作为一个整体系统如何结构性地优先考虑竞争价值。为了解决这个问题，我们提出了一种基于Q方法论的对称人类-LLM评估框架，以测量价值结构的一致性。在我们的协议下，人类和模型将相同的140项道德陈述集排序为共享的九列强制分布；对于LLMs，我们引导严格的排名并将其确定性地映射到Q排序桶中。使用人类参考样本（$N=35$），我们建立了一个特定于该工具和样本的稳定三因子参考几何结构。我们通过在两个温度设置下进行240次重复Q排序，评估了四个模型家族中的12个LLMs，并通过Procrustes相似性（$ heta$）和基于RSA的Spearman相关性（$
ho$）量化结构一致性。我们的结果揭示了显著的跨家族异质性、模型对生成随机性的特定敏感性以及局部不一致性，这表明有利的全球评分可能掩盖潜在的区域扭曲。尽管基于排名和桶的分析保持高度一致，但提示措辞引入了显著的方差。最终，评估价值结构一致性为传统的逐项道德基准提供了重要的结构补充。
cs.CL / 78 / 2606.21954
Are Multilingual Models Actually Improving? Isolating True Cross-Lingual Transfer
多语言模型真的在改善吗？隔离真实的跨语言迁移
Abstract
Cross-lingual transfer is a model's ability to generalize capabilities from well-represented source languages to under-represented target languages. Existing measures of a model's transfer strength conflate improvements in transfer with general improvements to accuracy in the source language. We advocate for an alternate metric that reliably captures transfer strength called Hardness Adjusted Transfer (HAT) Score, and use it to derive multiple insights on factors influencing transfer strength. Our analysis across twenty diverse language models and three popular mainstream multilingual benchmarks argues that 1) transfer in small models is not broken, 2) we are making slower than expected progress in cross-lingual transfer with model size, and 3) we have made clear progress over time.
Chinese Translation
跨语言迁移是指模型将能力从表现良好的源语言推广到表现不足的目标语言的能力。现有的模型迁移强度测量将迁移的改善与源语言准确性的普遍改善混为一谈。我们提倡一种替代指标，称为硬度调整迁移（Hardness Adjusted Transfer, HAT）得分，该指标可靠地捕捉迁移强度，并利用它得出多个影响迁移强度的因素的见解。我们对二十种不同的语言模型和三个流行的主流多语言基准的分析表明：1）小模型的迁移能力并没有损坏，2）随着模型规模的增加，我们在跨语言迁移方面的进展比预期的要慢，3）我们随着时间的推移确实取得了明显的进展。
cs.CL / 79 / 2606.21959
OpenBioRQ: Unsolved Biomedical Research Questions for Agents
OpenBioRQ：针对智能体的未解决生物医学研究问题
Abstract
A working citation looks like proof -- but the fact that a link resolves does not mean the cited paper supports the claim. I find that current agentic models rarely fabricate citations (over $99\%$ resolve), yet roughly $15.9\%$ link to the wrong paper. Existing benchmarks miss this failure mode: when a question has a fixed answer key, a model can reproduce the expected source from that key rather than independently verifying that the source supports the claim. I introduce \textbf{\openbiorq{}}, a retrieval-grounded agentic benchmark of $12{,}553$ unsolved biomedical research questions across $12$ domains that treats open questions as a faithfulness-and-abstention probe. To my knowledge, this is the first biomedical benchmark to combine an agentic setting -- where the model must issue multiple tool calls -- with unsolved questions that have no answer key. Openness is verified against real follow-up evidence rather than a model's parametric knowledge. Difficulty is empirical: I anchor it on questions that three open-weight reference models fail to answer, rather than on subjective hardness labels. On this hardest subset, held-out models from the same lineage as the difficulty anchors solve only ~17%, while three independent frontier agents (Gemini-3-Pro, Opus-4.7, GPT-5.5) span a wide 29-60% range. The benchmark is thus hard, non-saturating (the best agent still leaves ~33-40\% unsolved), and discriminating across capability tiers. Beyond difficulty, I observe agentic collapse on the hardest questions, where agents stop using their tools. For the most collapse-prone model, blocking tool access entirely barely changes its score -- so tools stop paying off exactly where they are needed most. A frozen per-question checklist raises inter-judge agreement from Spearman 0.35 to 0.82.
Chinese Translation
有效的引用看似是证据——但链接能够解析并不意味着被引用的论文支持该主张。我发现当前的智能模型很少伪造引用（超过99%能够解析），但大约15.9%的链接指向错误的论文。现有的基准测试未能捕捉到这种失败模式：当问题有固定的答案时，模型可以从该答案中重现预期的来源，而不是独立验证该来源是否支持该主张。我引入了 extbf{ extit{OpenBioRQ}}，这是一个基于检索的智能基准，涵盖了12个领域中的12,553个未解决的生物医学研究问题，将开放性问题视为忠实性和回避性探测工具。据我所知，这是第一个将智能设置（模型必须发出多个工具调用）与没有答案键的未解决问题相结合的生物医学基准。开放性是通过真实的后续证据进行验证，而不是模型的参数知识。难度是经验性的：我将其基于三个开放权重参考模型未能回答的问题，而不是主观的难度标签。在这个最难的子集上，来自同一难度锚点的保留模型仅解决了约17%的问题，而三个独立的前沿智能体（Gemini-3-Pro、Opus-4.7、GPT-5.5）则覆盖了29-60%的广泛范围。因此，该基准测试既困难又不饱和（最佳智能体仍然留下约33-40%的未解决问题），并且在能力层次上具有区分性。除了难度外，我还观察到在最难的问题上出现智能崩溃，智能体停止使用它们的工具。对于最容易崩溃的模型，完全阻止工具访问几乎不会改变其得分——因此工具在最需要的地方停止发挥作用。一个冻结的逐题检查表将评审者间的一致性从斯皮尔曼0.35提高到0.82。
cs.CL / 80 / 2606.21981
Can LLMs Control Readability? A Multi-Dimensional Evaluation Framework for CEFR-Controlled Arabic Generation
大型语言模型能控制可读性吗？一种多维度评估框架用于CEFR控制的阿拉伯语生成
Abstract
While Large Language Models (LLMs) can generate fluent Arabic text, their ability to reliably control readability levels remains unclear. We propose a multi-dimensional evaluation framework for Common European Framework of Reference for Language (CEFR)-controlled Arabic text generation, assessing whether instruction-following LLMs can serve as reliable generators for adaptive language learning. Our framework integrates controlled prompting, automatic readability prediction using a validated Taha-19 model, lexical constraint validation, and syntactic complexity profiling. Results show that structured prompting substantially improves CEFR alignment. In particular, CEFR-guided prompting with lexical constraints achieves the highest conformity to reference linguistic profiles (0.91 cosine similarity) and near-perfect agreement with predicted readability levels (0.99), while unconstrained prompting exhibits weak control. These findings establish an empirical foundation for integrating readability-aware Arabic text generation into adaptive educational systems.
Chinese Translation
尽管大型语言模型（LLMs）能够生成流畅的阿拉伯语文本，但它们在可靠控制可读性水平方面的能力仍不明确。我们提出了一种多维度评估框架，用于评估符合欧洲共同语言参考框架（CEFR）要求的阿拉伯语文本生成，旨在评估遵循指令的LLMs是否能够作为适应性语言学习的可靠生成器。我们的框架整合了受控提示、使用经过验证的Taha-19模型进行的自动可读性预测、词汇约束验证和句法复杂性分析。结果表明，结构化提示显著提高了CEFR的一致性。特别是，带有词汇约束的CEFR指导提示在参考语言特征的一致性上达到了最高水平（0.91余弦相似度），并与预测的可读性水平几乎完美一致（0.99），而无约束提示则表现出较弱的控制能力。这些发现为将可读性意识的阿拉伯语文本生成整合到适应性教育系统中奠定了实证基础。
cs.CL / 81 / 2606.21990
Adding Robust Code-Switching Capabilities to High Performance Multilingual ASR
为高性能多语言自动语音识别系统添加鲁棒的代码切换能力
Abstract
Code-switching (CSW) remains challenging for large multi-lingual ASR systems in real-world deployment. While fine-tuning on synthetic CSW data is possible, it generally degrades strong monolingual baselines. Our goal is to preserve these capabilities while extending models to handle complex code-switching, including morphological variations across languages. We propose Bayesian factorized adaptation, which learns to efficiently integrate switching-relevant knowledge into strong pretrained models without overwriting existing capabilities. Requiring only a small amount of synthetic data, our approach reduces transcription errors by 32.87% on code-switched words while improving overall WER by 5.31%, all while maintaining mono-lingual performance. Our results demonstrate that effective CSW adaptation depends more on knowledge integration than data complexity.
Chinese Translation
代码切换（CSW）在实际应用中仍然是大型多语言自动语音识别（ASR）系统面临的挑战。虽然在合成的代码切换数据上进行微调是可行的，但通常会降低强大的单语基线性能。我们的目标是在扩展模型以处理复杂的代码切换（包括跨语言的形态变化）的同时，保持这些能力。我们提出了一种贝叶斯因子化适应方法，该方法能够高效地将与切换相关的知识整合到强大的预训练模型中，而不覆盖现有能力。只需少量合成数据，我们的方法在代码切换词上的转录错误减少了32.87%，同时整体字错误率（WER）提高了5.31%，并保持了单语性能。我们的结果表明，有效的代码切换适应更依赖于知识整合而非数据复杂性。
cs.CL / 82 / 2606.22009
Benchmarking Large Language Models for Grapheme-to-Phoneme Conversion: A Japanese Case Study
大型语言模型在字形到音素转换中的基准测试：以日语为例
Abstract
Grapheme-to-phoneme (G2P) conversion is essential for controllable and robust text-to-speech, and large language models (LLMs), with broad linguistic knowledge, offer a promising approach. We benchmarked over 30 LLMs on Japanese G2P, comparing them with conventional morphological analyzers on 3000 manually annotated sentences. We evaluated two prompting strategies: a parse mode, where the LLM performs morphological analysis followed by rule-based kana conversion, and a direct mode, where the LLM directly predicts kana readings. The results show that model size, version, and Japanese-specialized training are key factors, with the best LLMs achieving kana character error rate below 0.52\% vs. the best conventional tool (1.03\%). Parse mode outperforms direct mode for most models, as rule-based post-processing relieves the LLM of handling complex pronunciation rules. We also show that feeding LLM-predicted kana into a kana-input TTS yields better pronunciation than end-to-end TTS.
Chinese Translation
字形到音素（G2P）转换对于可控且稳健的文本到语音（TTS）至关重要，而具有广泛语言知识的大型语言模型（LLMs）提供了一种有前景的方法。我们对30多种LLM在日语G2P上的表现进行了基准测试，并将其与传统形态分析工具在3000个手动标注的句子上的表现进行了比较。我们评估了两种提示策略：解析模式（parse mode），其中LLM执行形态分析后进行基于规则的假名转换；直接模式（direct mode），其中LLM直接预测假名读音。结果表明，模型大小、版本和日语专门训练是关键因素，最佳LLM的假名字符错误率低于0.52%，而最佳传统工具为1.03%。对于大多数模型，解析模式的表现优于直接模式，因为基于规则的后处理减轻了LLM处理复杂发音规则的负担。我们还表明，将LLM预测的假名输入到假名输入的TTS中，能够获得比端到端TTS更好的发音效果。
cs.CL / 83 / 2606.22061
NL2Scratch: An Executable Benchmark and Evaluation for Block-Based Programming
NL2Scratch：基于区块编程的可执行基准和评估
Abstract
Block-based programming environments such as Scratch are widely used in early programming education, yet natural-language-to-code (NL2Code) research has focused primarily on text-based languages. Scratch programs are event-driven, visually compositional, and distributed across concurrent scripts, making conventional NL2Code assumptions and evaluation insufficient. We introduce NL2Scratch, an executable benchmark for natural-language-to-Scratch generation comprising 311,648 parser-valid NL--program pairs, whose program side is extracted from real Scratch projects and paired with semantically aligned NL descriptions. For reliable evaluation beyond surface overlap, we propose Semantic Alignment Consistency (SAC), an interpretable slot-level metric for measuring semantic agreement between descriptions and programs. With SAC, we construct a semantically validated pool of 23,594 examples, and a slot-balanced 800 diagnostic benchmark. Experiments across instruction-tuned and fine-tuned LLMs reveal a notable gap between lexical similarity and semantic alignment: models achieving token-level F1 above 0.93 often fail to attain perfect SAC, particularly on longer examples. Errors concentrate on operational slots like actions, conditions, and numeric arguments, exposing failure modes largely invisible under conventional metrics.
Chinese Translation
基于区块的编程环境，如 Scratch，在早期编程教育中被广泛使用，但自然语言到代码（NL2Code）研究主要集中在基于文本的语言上。Scratch 程序是事件驱动的、视觉组合的，并且分布在并发脚本中，这使得传统的 NL2Code 假设和评估显得不足。我们介绍了 NL2Scratch，这是一个用于自然语言到 Scratch 生成的可执行基准，包含 311,648 对解析器有效的 NL-程序对，其中程序部分提取自真实的 Scratch 项目，并与语义对齐的 NL 描述配对。为了超越表面重叠进行可靠评估，我们提出了语义对齐一致性（Semantic Alignment Consistency, SAC），这是一种可解释的槽级度量，用于测量描述与程序之间的语义一致性。通过 SAC，我们构建了一个语义验证的 23,594 个示例的池，以及一个槽平衡的 800 个诊断基准。针对指令调优和微调的 LLMs 的实验揭示了词汇相似性与语义对齐之间的显著差距：在 token 级别 F1 超过 0.93 的模型往往未能达到完美的 SAC，尤其是在较长的示例中。错误集中在操作槽上，如动作、条件和数值参数，这暴露了在传统度量下大多不可见的失败模式。
cs.CL / 84 / 2606.22079
Where Does the Signal Live? A Web Data Recipe for Medical Encoder Pretraining
信号来自何处？医学编码器预训练的网络数据配方
Abstract
Web data curation has been widely studied for decoder Large Language Model (LLM) pretraining. Encoders for dense-terminology domains such as medicine, by contrast, are pretrained on small, manually-curated corpora that limit scalability and writing style diversity, a bottleneck even more severe in non-English clinical settings. Whether web-scale data curation also benefits encoder Masked Language Modeling (MLM) in a dense-terminology domain remains an open question. To address this, we introduce two complementary levers. Medical-term density filtering selects documents rich in medical terms. Signal-amplifying rephrasing uses an LLM to rewrite documents into denser variants with broader entity contexts. We instantiate the recipe on French medical NLP. The medical-term density filter outperforms the widely-used educational quality filter on downstream medical tasks, and the two complement each other. Signal-amplifying rephrasing alone improves on raw web data, and mixing it with filtered web data produces the largest gain. The recipe yields FineMed, a French medical pretraining corpus, and DoctoBERT, a state-of-the-art French medical encoder family evaluated on both the public benchmark DrBenchmark and a proprietary clinical Named Entity Recognition (NER) task.
Chinese Translation
网络数据的整理在解码器大型语言模型（LLM）预训练中得到了广泛研究。相比之下，密集术语领域（如医学）的编码器则是在小规模、人工整理的语料库上进行预训练，这限制了其可扩展性和写作风格的多样性，尤其在非英语临床环境中更为严重。网络规模的数据整理是否也能为密集术语领域的编码器掩蔽语言建模（MLM）带来益处仍然是一个未解的问题。为了解决这个问题，我们引入了两个互补的杠杆。医学术语密度过滤器选择富含医学术语的文档。信号增强重述则利用LLM将文档改写为更密集的变体，具有更广泛的实体上下文。我们在法语医学自然语言处理（NLP）中实现了这一配方。医学术语密度过滤器在下游医学任务中优于广泛使用的教育质量过滤器，且两者相辅相成。单独使用信号增强重述对原始网络数据的改进显著，将其与过滤后的网络数据混合则产生了最大的增益。该配方生成了FineMed，一个法语医学预训练语料库，以及DoctoBERT，一个在公共基准DrBenchmark和专有临床命名实体识别（NER）任务上评估的最先进的法语医学编码器系列。
cs.CL / 85 / 2606.22097
Plurification in/of language technology -- The integration of culture in next-generation AI
语言技术中的多元化——下一代人工智能中的文化整合
Abstract
The paper explores how "culture" can be operationalised in Natural Language Processing (NLP) and what this reveals about the possibilities and limits of considering a plurality of cultural backgrounds in technological design. It proposes that cultural alignment cannot be achieved only by adding more examples of "other cultures", rather it requires plural epistemologies: allowing multiple, locally grounded ways of knowing. To analyze how this plurality of knowing can be addressed in NLP, the paper uses a socio-technical model of language technology (LT) design, the five layers of technological activity model, for collecting and systematizing approaches to culture in NLP. The analysis shows that while NLP research has made progress toward more culturally sensitive systems, many approaches remain partial, addressing "culture" primarily at the level of output or representation while leaving deeper questions of power, governance, and social context unresolved. The paper concludes that operationalising culture requires much more than technical adaptation; it suggests a reflexive and plural socio-technical approach that navigates potentials and limits of computational formalisation for accounting multiple linguistic and socio-cultural backgrounds.
Chinese Translation
本文探讨了如何在自然语言处理（NLP）中将“文化”进行操作化，以及这揭示了在技术设计中考虑多元文化背景的可能性和局限性。文章提出，文化对齐不仅仅通过增加“其他文化”的更多例子来实现，而是需要多元的认识论：允许多种本土化的知识获取方式。为了分析这种知识的多元性如何在NLP中得到体现，本文采用了一种社会技术模型，即五层技术活动模型，来收集和系统化NLP中文化的相关方法。分析表明，尽管NLP研究在朝着更具文化敏感性的系统方面取得了一定进展，但许多方法仍然是片面的，主要在输出或表现层面上处理“文化”，而未能解决权力、治理和社会背景等更深层次的问题。本文最后得出结论，文化的操作化需要远超过技术适应；它建议采取一种反思性和多元的社会技术方法，以应对计算形式化在考虑多种语言和社会文化背景时的潜力和局限性。
cs.CL / 86 / 2606.22126
From Recognition to Understanding: Unlocking Cognitive Time Series Reasoning with LLMs
从识别到理解：利用大型语言模型解锁认知时间序列推理
Abstract
Time series analysis has recently been coupled with Large Language Models (LLMs) to leverage their reasoning and world knowledge capabilities, yet gains remain limited. We attribute this to a fundamental mismatch between existing task formulations and LLM strengths: most settings reduce time series understanding to curve-fitting systems, focusing on low-level prediction while ignoring the semantic, contextual, and reasoning-intensive nature of real-world temporal decision-making.To address these limitations, we introduce TSCognition, a multimodal benchmark for multi-dimensional time series reasoning. It collects real-world time series and textual information from 15 public sources and constructs approximately 41K QA samples around five cognitive reasoning tasks: Decoding, Grounding, Inferring, Extrapolating, and Acting. Building on this, we further propose TSAlign, a unified framework that encodes time series into compact patch-level representations and aligns them with semantic directions in the LLM embedding space via gated residual injection and multivariate fusion.Experiments show that TSAlign outperforms existing LLM, VLM, and time series QA baselines on TSCognition and the publicly available TimerBed benchmark while substantially reducing computational cost.Code is available at: [https://github.com/EIT-NLP/CognitiveTSR](https://github.com/EIT-NLP/CognitiveTSR)
Chinese Translation
时间序列分析最近与大型语言模型（LLMs）相结合，以利用其推理和世界知识能力，但取得的进展仍然有限。我们将其归因于现有任务形式与LLM优势之间的根本不匹配：大多数设置将时间序列理解简化为曲线拟合系统，专注于低层次的预测，而忽视了现实世界时间决策的语义、上下文和推理密集特性。为了解决这些局限性，我们引入了TSCognition，这是一个多模态基准，用于多维时间序列推理。该基准从15个公共来源收集了真实世界的时间序列和文本信息，并围绕五个认知推理任务构建了大约41K个问答样本：解码（Decoding）、基础（Grounding）、推断（Inferring）、外推（Extrapolating）和行动（Acting）。在此基础上，我们进一步提出了TSAlign，这是一个统一框架，将时间序列编码为紧凑的补丁级表示，并通过门控残差注入和多变量融合将其与LLM嵌入空间中的语义方向对齐。实验表明，TSAlign在TSCognition和公开可用的TimerBed基准上超越了现有的LLM、VLM和时间序列问答基线，同时显著降低了计算成本。代码可在以下链接获取：[https://github.com/EIT-NLP/CognitiveTSR](https://github.com/EIT-NLP/CognitiveTSR)
cs.CL / 87 / 2606.22138
BioMatrix: Towards a Comprehensive Biological Foundation Model Spanning the Modality Matrix of Sequences, Structures, and Language
BioMatrix：迈向一个全面的生物基础模型，涵盖序列、结构和语言的多模态矩阵
Abstract
We present BioMatrix, the first multimodal foundation model that natively integrates sequences, structures, and natural language for both molecules and proteins within a single decoder-only architecture. Existing biological foundation models pursue native multimodality and broad entity coverage separately: those that fuse multiple modalities under a shared objective remain confined to a single entity type, while those spanning multiple entity types either omit explicit structural modeling or rely on adapter-based designs in which the model cannot natively generate the very modalities it can read. BioMatrix closes this gap by mapping molecular sequences (supporting both SMILES and SELFIES notations), molecular structures, protein sequences, protein structures, and natural language into a shared discrete token space through a unified tokenization scheme, so that all modalities are consumed and produced uniformly under a single next-token prediction objective -- without external encoders, projection adapters, or modality-specific output heads. Built upon the Qwen3 language model (1.7B and 4B), BioMatrix is continually pretrained on 304.4 billion tokens spanning general and domain-specific text, sequence and structure views of molecules and proteins, and cross-modal corpora that interleave biomolecular entities with scientific text and link distinct entities through molecule-protein and protein-protein interaction data. After tuning on a comprehensive suite of downstream applications covering 80 tasks across 6 categories -- encompassing single-entity and multi-entity understanding and generation tasks across and within modalities -- BioMatrix achieves state-of-the-art or competitive performance on 77 out of 80 tasks, demonstrating that a single, natively multimodal generalist model can effectively match or surpass specialized approaches across a wide range of biological tasks.
Chinese Translation
我们提出了BioMatrix，这是第一个原生集成分子和蛋白质的序列、结构和自然语言的多模态基础模型，采用单一解码器架构。现有的生物基础模型分别追求原生多模态性和广泛的实体覆盖：那些在共享目标下融合多种模态的模型仍然局限于单一实体类型，而那些跨越多种实体类型的模型则要么省略显式的结构建模，要么依赖于适配器设计，使得模型无法原生生成其能够读取的模态。BioMatrix通过将分子序列（支持SMILES和SELFIES表示法）、分子结构、蛋白质序列、蛋白质结构和自然语言映射到一个共享的离散标记空间，填补了这一空白，采用统一的标记化方案，使得所有模态在单一的下一个标记预测目标下被统一消耗和生成——无需外部编码器、投影适配器或特定模态的输出头。BioMatrix基于Qwen3语言模型（1.7B和4B），在涵盖一般和领域特定文本、分子和蛋白质的序列与结构视图，以及将生物分子实体与科学文本交织并通过分子-蛋白质和蛋白质-蛋白质交互数据链接不同实体的跨模态语料库上，持续进行预训练，涉及3044亿个标记。在针对涵盖80个任务的全面下游应用进行调优后，跨越6个类别——包括跨模态和模态内的单一实体和多实体理解与生成任务——BioMatrix在80个任务中的77个任务上实现了最先进或具有竞争力的性能，证明了一个单一的、原生多模态的通用模型能够有效匹配或超越广泛生物任务中的专业方法。
cs.CL / 88 / 2606.22179
The Score Granularity Gap in Black-Box LLM Classification: A Comparative Study of Confidence Constructions
黑箱 LLM 分类中的评分粒度差距：信心构建的比较研究
Abstract
Large language models (LLMs) are increasingly deployed as black-box classifiers in pipelines that automate confident decisions and route uncertain ones to human review. Such selective prediction needs a confidence score that an operator can threshold at a chosen risk level. Prior work asks whether LLM confidence is well calibrated or well ranked; we ask a complementary, deployment-oriented question that has been largely overlooked: at what resolution can the score be thresholded? We call the answer the score granularity gap. Through a controlled comparison of seven ways to build a confidence score, from a single verbalized number, to token probabilities, to querying the model many times and combining the answers, across 25 model-dataset pairs (9 LLMs, 3 benchmarks), we find that single-shot verbalized confidence, once correctly converted to a class probability, ranks cases surprisingly well, yet takes only a handful of distinct values. It therefore offers an operator only a few coarse thresholds, no matter how well it ranks. We show which constructions widen this gap, at what inference cost, and with what effect on ranking, notably that multi-query aggregation helps weak models but can degrade already-strong ones. We translate these trade-offs into concrete deployment guidance.
Chinese Translation
大型语言模型（LLMs）越来越多地作为黑箱分类器部署在自动化自信决策的流程中，并将不确定的决策交由人工审查。这种选择性预测需要一个信心评分，操作员可以在所选风险水平上进行阈值设置。先前的研究关注 LLM 信心是否经过良好校准或良好排名；而我们提出一个互补的、以部署为导向的问题，这一问题在很大程度上被忽视：评分可以在什么分辨率下进行阈值设置？我们称这个答案为评分粒度差距。通过对七种构建信心评分的方法进行控制比较，从单一的口头数字到令牌概率，再到多次查询模型并结合答案，涵盖 25 对模型-数据集（9 个 LLM，3 个基准），我们发现单次口头信心在正确转换为类别概率后，能够意外地很好地对案例进行排名，但仅取少数不同的值。因此，无论其排名多么优秀，它只为操作员提供了少数粗略的阈值。我们展示了哪些构建方法扩大了这一差距，推理成本是多少，以及对排名的影响，特别是多查询聚合有助于弱模型，但可能会降低已经强大的模型的性能。我们将这些权衡转化为具体的部署指导。
cs.CL / 89 / 2606.22203
When Is Emergent Consensus Real? A Measured Coupling Gain and a Validity Diagnostic for LLM Agent Societies
何时出现真正的涌现共识？一种测量耦合增益和大型语言模型代理社会的有效性诊断
Abstract
LLM "agent societies" are studied via demonstrations of emergent consensus or polarization -- with no measurable control parameter, no theory of when each regime appears, and no test of whether an outcome is a genuine social dynamic or a model artifact. We introduce the coupling gain gamma, measured per-agent by counterfactually perturbing a neighbour's stated opinion. (i) gamma is stable and model-distinguishing -- across five frontier models it spans 0.15-0.43 (n=20, 95% CIs <= 0.025), paraphrase-invariant; social-neighbour gamma roughly equals numeric-anchor gamma, so gamma is evidence-coupling, not uniquely social. (ii) Classical dynamics with measured (not assumed) coefficients organise the regime: Friedkin-Johnsen for consensus/pluralism, signed-Laplacian/structural-balance for polarization. (iii) Frontier LLMs do not spontaneously backfire (beta <= 0), so default societies do not self-polarize -- polarization is always induced; the beta>0 branch arises only in the FJ surrogate, never in the agents. (iv) A randomized-initial-condition diagnostic -- the (slope, bias) of final vs. initial opinion -- separates genuine averaging from model-prior artifacts (boundary-censoring ruled out by construction via interior-valued facts); applied to a published "emergent consensus" result (Chuang et al. 2023) it reveals a model-specific conflation: averaging on debatable claims, prior-artifact on settled facts. (v) Coupling is context-dependent: pairwise gamma does not predict multi-neighbour outcomes -- it can order them backwards -- whereas a modality-matched group coupling does (sixteen closed+open models, Pearson r=-0.70, permutation p=0.008). The regime laws take this matched coupling, not the single-neighbour gamma: emergent consensus must be read from coupling in the target interaction. We contribute a measurement protocol and a validity instrument, not new theory.
Chinese Translation
大型语言模型（LLM）“代理社会”通过涌现共识或极化的示例进行研究——没有可测量的控制参数，没有关于每种状态出现时机的理论，也没有测试结果是否是真正的社会动态或模型伪影。我们引入了耦合增益 gamma，通过反事实地扰动邻居的表述意见来逐个代理进行测量。(i) gamma 稳定且具有模型区分性——在五个前沿模型中，其范围为 0.15-0.43 (n=20, 95% 置信区间 <= 0.025)，对释义不变；社会邻居的 gamma 大致等于数值锚定的 gamma，因此 gamma 是证据耦合，而不仅仅是社会因素。(ii) 具有测量（而非假设）系数的经典动态组织了状态：Friedkin-Johnsen 模型用于共识/多元主义，带符号的拉普拉斯/结构平衡模型用于极化。(iii) 前沿 LLM 不会自发反弹（beta <= 0），因此默认社会不会自我极化——极化总是被诱导；beta>0 的分支仅在 FJ 替代模型中出现，而在代理中从未出现。(iv) 一种随机初始条件诊断——最终与初始意见的（斜率，偏差）——将真正的平均化与模型先验伪影区分开来（通过内部值事实的构造排除了边界审查）；应用于已发布的“涌现共识”结果（Chuang et al. 2023），揭示了一种模型特定的混淆：在有争议的主张上平均化，在已解决的事实上则是模型伪影。(v) 耦合是上下文依赖的：成对的 gamma 并不能预测多邻居的结果——它可能会反向排序——而匹配的群体耦合则可以（十六个闭合+开放模型，Pearson r=-0.70，置换 p=0.008）。状态法则采用这种匹配耦合，而不是单一邻居的 gamma：涌现共识必须从目标交互中的耦合中解读。我们贡献了一种测量协议和有效性工具，而不是新的理论。
cs.CL / 90 / 2606.22207
Lexical Consensus: Grounded Word Learning and Shared Meaning in Artificial Agents
词汇共识：人工智能体中的基础词汇学习与共享意义
Abstract
Artificial intelligence systems are commonly evaluated through task performance and behavioral imitation, but such evaluations leave open whether an artificial agent can acquire, stabilize, and use new lexical meanings from grounded experience. This paper introduces Lexical Consensus, an experimental framework for studying grounded word learning over a structured perceptual substrate. Using frozen DINOv2 visual embeddings, Carroll-style nonce words, and interpretable lexical learners plus linear baselines, we test whether agents can acquire artificial labels for visual concepts, generalize them bidirectionally, and stabilize them across controlled settings. The main result is a robust perceptual-coherence gradient: native categories are easiest to learn, coherent overextensions remain learnable, mid-range disjunctive concepts degrade, and far-disjunctive concepts approach chance. A pre-registered CIFAR-100 dissociation experiment confirms that this gradient is governed by perceptual distance rather than semantic relatedness: perceptual distance predicts acquisition accuracy (partial R^2 = 0.245, p < 1e-7), while semantic distance adds no significant explanatory power (partial R^2 = 0.002, p = 0.660). Bidirectional evaluation shows that naming and retrieval are distinct: exemplar-based mechanisms outperform centroid prototypes in label-to-image retrieval, exposing a memory-fidelity dimension separate from naming accuracy. Falsification controls, homogeneous candidate-pool evaluations, and null results on representational restructuring indicate that frozen perceptual geometry both enables lexical grounding and limits what can be acquired without representational adaptation.
Chinese Translation
人工智能系统通常通过任务表现和行为模仿进行评估，但这种评估未能明确人工智能体是否能够从基础经验中获取、稳定和使用新的词汇意义。本文引入了词汇共识（Lexical Consensus），一个用于研究在结构化感知基础上进行基础词汇学习的实验框架。我们使用冻结的DINOv2视觉嵌入、Carroll风格的虚构词汇以及可解释的词汇学习者和线性基线，测试智能体是否能够为视觉概念获取人工标签、双向泛化这些标签，并在受控环境中稳定这些标签。主要结果是一个稳健的感知一致性梯度：本土类别最容易学习，一致的过度延伸仍然可学习，中间范围的析取概念退化，而远析取概念接近随机。一个预注册的CIFAR-100分离实验确认该梯度受感知距离而非语义相关性的支配：感知距离能够预测获取准确性（部分R^2 = 0.245, p < 1e-7），而语义距离未能提供显著的解释力（部分R^2 = 0.002, p = 0.660）。双向评估显示命名和检索是不同的：基于示例的机制在标签到图像的检索中优于质心原型，揭示了一个与命名准确性分离的记忆保真度维度。虚假控制、同质候选池评估以及代表性重构的无效结果表明，冻结的感知几何既促进了词汇的基础化，也限制了在没有代表性适应的情况下能够获取的内容。
cs.CL / 91 / 2606.22269
Evaluating Large Language Models for Hausa and Fongbe Machine Translation: Benchmarks, Failures, and Metric Reliability
评估大型语言模型在豪萨语和丰贝语机器翻译中的表现：基准、失败与指标可靠性
Abstract
We investigate the translation quality of current large language models (LLMs) for English-to-Hausa and English-to-Fongbe - two typologically distinct West African languages from the Afroasiatic and Niger-Congo families respectively - and evaluate whether standard automatic metrics reliably reflect human judgment for these low-resource languages. We evaluate four models (GPT-4o Mini, Claude Sonnet 4, Gemini 2.5 Flash, and Qwen2.5-7B) at progressive scales (500 to 10,000 sentences) using automatic metrics (BLEU, chrF++, TER, COMET, BERTScore) validated against native-speaker judgment. Our results reveal three key findings. First, translation quality varies substantially by language: Hausa achieves acceptable quality (human scores 4.0-4.5/5) while Fongbe achieves poor quality (1.0-2.2/5), with a consistent 3x BLEU gap across all systems. Second, model rankings differ by language - Gemini leads for Fongbe while GPT-4o leads for Hausa by human evaluation - indicating that performance on one low-resource African language does not predict performance on another. Third, metric-human correlation varies dramatically: perfect rank correlation for Fongbe (rho=1.0) but weak correlation for Hausa (rho=0.5), where human evaluators preferred GPT-4o despite all automatic metrics ranking Claude first. We further show that neural metrics like BERTScore exhibit embedding collapse (within-language similarity >0.99) for both languages, limiting their ability to differentiate translation quality. Based on these findings, we recommend multi-metric evaluation for low-resource African languages, with particular caution when interpreting neural metrics. We establish that minimum sample sizes of n=2,500 sentences are required for stable system rankings, as smaller samples produced artifact findings that reversed at scale.
Chinese Translation
我们研究了当前大型语言模型（LLMs）在英语到豪萨语和英语到丰贝语的翻译质量，这两种语言分别属于亚非语系和尼日尔-刚果语系，且在类型上存在显著差异，并评估标准自动化指标是否可靠地反映了这些低资源语言的人类判断。我们评估了四个模型（GPT-4o Mini、Claude Sonnet 4、Gemini 2.5 Flash 和 Qwen2.5-7B），在逐步增加的规模（500到10,000个句子）下使用自动化指标（BLEU、chrF++、TER、COMET、BERTScore），并与母语者的判断进行了验证。我们的结果揭示了三个关键发现。首先，翻译质量因语言而异：豪萨语的质量可接受（人类评分4.0-4.5/5），而丰贝语的质量较差（1.0-2.2/5），所有系统之间存在一致的3倍BLEU差距。其次，模型排名因语言而异——在丰贝语中，Gemini表现最佳，而在豪萨语中，GPT-4o表现最佳，这表明在一种低资源非洲语言上的表现并不能预测在另一种语言上的表现。第三，指标与人类评估的相关性差异显著：丰贝语的排名相关性完美（rho=1.0），而豪萨语的相关性较弱（rho=0.5），尽管所有自动化指标均将Claude排名第一，但人类评估者更偏爱GPT-4o。我们进一步显示，像BERTScore这样的神经指标在这两种语言中表现出嵌入崩溃（同语言相似度>0.99），限制了它们区分翻译质量的能力。基于这些发现，我们建议对低资源非洲语言进行多指标评估，并在解释神经指标时特别谨慎。我们确定，稳定的系统排名需要至少n=2,500个句子的样本量，因为较小的样本产生的伪发现会在规模扩大时反转。
cs.CL / 92 / 2606.22272
MixedPEFT: Combining Multiple PEFT Methods with Mixed Objectives for Unsupervised Domain Adaptation
MixedPEFT：结合多种PEFT方法与混合目标的无监督领域适应
Abstract
Pre-trained language models struggle when applied to new domains, as full fine-tuning is computationally expensive and prone to catastrophic forgetting. This study addresses this challenge by presenting a novel parameter-efficient strategy for unsupervised domain adaptation that combines custom PEFT architectures with mixed-objective training. Our approach simultaneously optimizes classification performance on labeled source domain data and masked language modeling (MLM) on unlabeled target domain data, preserving target domain knowledge while adapting to source domain tasks. Our method employs a custom union of invertible adapters and Low-Rank Adaptation (LoRA) within a unified parameter-efficient framework. Through comprehensive evaluation on the Multi-Genre Natural Language Inference (MNLI) dataset across 20 domain shifts, our approach achieves significant improvements over existing methods: 1.41 percentage points over the current parameter-efficient state-of-the-art UDapter, 1.26 percentage points over the fully-tuned DANN baseline, and 0.86 percentage points over DSN, while utilizing only 7% of the model's trainable parameters. These results establish new benchmarks for parameter-efficient unsupervised domain adaptation and demonstrate that carefully designed PEFT combinations with concurrent optimization can outperform both existing parameter-efficient methods and traditional fully-tuned approaches.
Chinese Translation
预训练语言模型在应用于新领域时面临困难，因为完全微调计算成本高且容易导致灾难性遗忘。本研究通过提出一种新颖的参数高效策略，解决了这一挑战，该策略用于无监督领域适应，结合了定制的PEFT架构与混合目标训练。我们的方法同时优化了标记源领域数据的分类性能和未标记目标领域数据的掩码语言建模（MLM），在适应源领域任务的同时保留目标领域知识。我们的方法在统一的参数高效框架内采用了可逆适配器和低秩适应（LoRA）的自定义组合。通过在20个领域迁移的多体裁自然语言推理（MNLI）数据集上进行全面评估，我们的方法在现有方法上取得了显著的改进：比当前参数高效的最先进方法UDapter提高了1.41个百分点，比完全微调的DANN基线提高了1.26个百分点，比DSN提高了0.86个百分点，同时仅使用了模型可训练参数的7%。这些结果为参数高效的无监督领域适应建立了新的基准，并证明了精心设计的PEFT组合与并行优化可以超越现有的参数高效方法和传统的完全微调方法。
cs.CL / 93 / 2606.22274
From Speech to Text Corpora: Evaluating ASR-Based Data Acquisition for Low-Resource Fongbe and Hausa
从语音到文本语料库：评估基于自动语音识别的数据获取在低资源的丰贝语和豪萨语中的应用
Abstract
Low-resource African languages lack text corpora needed for language model training. We investigate whether ASR pipelines can extend text resources for two typologically distinct West African languages: Fongbe (tonal, diacritic-rich) and Hausa (non-tonal). We fine-tune MMS-300M on a curated 12.3-hour Fongbe dataset, achieving 9.48% WER on the ALFFA benchmark - a 78% relative reduction from the prior 44.04% baseline - while preserving tonal diacritics critical to the language. For Hausa, we apply an existing fine-tuned Whisper-Small model. We catalog 1,553 YouTube videos (236 hours) and process a subset of 424 videos (45.49 hours) selected to balance domain diversity with available computational resources, producing 6,770 transcribed segments. Human evaluation on 50 randomly sampled segments per language shows mean quality scores of 57.4/100 for Hausa and 36.5/100 for Fongbe, indicating that while Hausa transcriptions approach acceptable quality for corpus construction, Fongbe transcriptions require post-processing or improved models for production use. We release the curated dataset, fine-tuned model, transcribed corpus, and full video catalog following platform terms and ethical guidelines.
Chinese Translation
低资源的非洲语言缺乏用于语言模型训练的文本语料库。我们研究了自动语音识别（ASR）管道是否能够扩展两种类型上有显著差异的西非语言的文本资源：丰贝语（声调丰富，带有变音符号）和豪萨语（非声调）。我们在一个经过精心挑选的12.3小时丰贝语数据集上微调了MMS-300M模型，在ALFFA基准测试中达到了9.48%的词错误率（WER），相比于之前的44.04%的基线实现了78%的相对减少，同时保留了对该语言至关重要的声调变音符号。对于豪萨语，我们应用了现有的微调Whisper-Small模型。我们整理了1,553个YouTube视频（236小时），并处理了424个视频（45.49小时）的子集，以在领域多样性和可用计算资源之间取得平衡，生成了6,770个转录段落。对每种语言随机抽样的50个段落进行的人类评估显示，豪萨语的平均质量分数为57.4/100，而丰贝语为36.5/100，这表明尽管豪萨语的转录接近于构建语料库所需的可接受质量，但丰贝语的转录需要后处理或改进模型以便于生产使用。我们根据平台条款和伦理指南发布了经过整理的数据集、微调模型、转录语料库和完整视频目录。
cs.CL / 94 / 2606.22305
Learning at the Right Pace: Adaptive Data Scheduling Improves LLM Reinforcement Learning
以适当的节奏学习：自适应数据调度改善大型语言模型的强化学习
Abstract
Large Language Models (LLMs) achieve remarkable reasoning capabilities through reinforcement learning (RL) post-training. However, existing RL post-training commonly relies on uniform data sampling, which ignores the semantic structure of the training data and the changing capability of the training policy. To address these limitations, we propose Adaptive Data Scheduling (ADS), a dual-level data scheduling framework for pacing RL post-training that replaces uniform sampling with an adaptive distribution over semantic clusters and policy-boundary sample selection. At the cluster level, ADS organizes samples according to semantic patterns and maintains an adaptive inter-cluster distribution to solidify current training progress. At the sample level, ADS performs intra-cluster scheduling to continuously sample policy-boundary samples, which provides informative relative advantages. Experimental results across three LLMs and seven reasoning benchmarks demonstrate that ADS improves average accuracy by 5.2% over Group Relative Policy Optimization (GRPO). Notably, ADS consistently improves RL methods with different objective designs, highlighting its potential as a general data scheduling strategy for LLM RL post-training. The source code is available at: https://github.com/Richard-zrx/ADS.
Chinese Translation
大型语言模型（LLMs）通过强化学习（RL）后训练实现了显著的推理能力。然而，现有的RL后训练通常依赖于均匀数据采样，这忽略了训练数据的语义结构和训练策略的变化能力。为了解决这些局限性，我们提出了自适应数据调度（Adaptive Data Scheduling，ADS），这是一种双层数据调度框架，用于调整RL后训练的节奏，替代均匀采样，采用基于语义聚类的自适应分布和策略边界样本选择。在聚类层面，ADS根据语义模式组织样本，并维持自适应的聚类间分布，以巩固当前的训练进展。在样本层面，ADS执行聚类内调度，持续采样策略边界样本，从而提供有意义的相对优势。在三个LLM和七个推理基准上的实验结果表明，ADS在平均准确率上比组相对策略优化（Group Relative Policy Optimization，GRPO）提高了5.2%。值得注意的是，ADS在不同目标设计的RL方法中始终表现出改善，突显了其作为LLM RL后训练通用数据调度策略的潜力。源代码可在以下网址获取：https://github.com/Richard-zrx/ADS。
cs.CL / 95 / 2606.22329
BabelJudge: Measuring LLM-as-a-Judge Reliability Across Languages and Agent Trajectories
BabelJudge：跨语言和代理轨迹测量 LLM 作为评审的可靠性
Abstract
LLM-as-a-judge has become the dominant approach to scalable evaluation in NLP pipelines, yet judges themselves carry systematic biases that raw accuracy hides: they favor responses placed in slot A (position bias), they prefer longer responses regardless of quality (verbosity bias), and their reliability degrades sharply in lower-resource languages. We introduce BabelJudge, an open-source benchmark and reliability audit framework that measures all four failure modes -- position bias, verbosity bias, order inconsistency, and cross-lingual degradation -- on any judge model, without requiring human preference labels. The key insight is gold-labelling by degradation: starting from a high-quality reference response and applying a controlled perturbation yields a pairwise item whose gold label is known by construction, eliminating annotation cost. We evaluate Qwen2.5-7B-Instruct-4bit across English, Hindi, Arabic, and Swahili and find that our composite bias-penalised reliability score drops from 0.714 in Hindi to 0.550 in Swahili, a gap that raw accuracy (0.835 vs. 0.660) understates. Swahili order consistency collapses to 0.480, meaning judge verdicts are near-random under slot-order swaps -- a failure mode invisible to accuracy alone. We further extend the framework to agentic evaluation via nine trajectory-level perturbations (argument corruption, tool swaps, hallucinated calls, missing steps) and three new metrics: tool accuracy, hallucination detection rate, and trajectory-length bias. BabelJudge is released as a Python package supporting 11 judge backends. Code: https://github.com/Shreyaskc/BabelJudge
Chinese Translation
LLM 作为评审已成为自然语言处理（NLP）流程中可扩展评估的主导方法，但评审本身存在系统性偏见，这些偏见在原始准确性中被掩盖：评审偏好放置在 A 位置的回答（位置偏见），无论质量如何都偏好更长的回答（冗长偏见），并且在低资源语言中的可靠性急剧下降。我们引入了 BabelJudge，这是一个开源基准和可靠性审计框架，能够在任何评审模型上测量四种失败模式——位置偏见、冗长偏见、顺序不一致性和跨语言退化，而无需人类偏好标签。关键见解是通过退化进行黄金标注：从高质量参考回答出发，应用受控扰动生成一对已知黄金标签的项目，从而消除标注成本。我们在英语、印地语、阿拉伯语和斯瓦希里语上评估了 Qwen2.5-7B-Instruct-4bit，发现我们的复合偏见惩罚可靠性评分从印地语的 0.714 降至斯瓦希里语的 0.550，这一差距在原始准确性（0.835 对 0.660）中被低估。斯瓦希里语的顺序一致性降至 0.480，这意味着在槽顺序交换下，评审裁决几乎是随机的——这一失败模式仅通过准确性无法察觉。我们进一步通过九种轨迹级扰动（论点损坏、工具交换、虚构调用、缺失步骤）和三种新指标（工具准确性、幻觉检测率和轨迹长度偏见）扩展了该框架。BabelJudge 作为一个支持 11 个评审后端的 Python 包发布。代码链接：https://github.com/Shreyaskc/BabelJudge
cs.CL / 96 / 2606.22342
How Does Research Evolve? Tracing Cross-Domain Trajectories in NLP, ML, and CV with Claim-Grounded Typed Citations
研究如何演变？通过基于主张的类型化引用追踪NLP、ML和CV中的跨领域轨迹
Abstract
How does research evolve, and what substrate would let us forecast where it goes next? Scientific progress is not simply a uniform accumulation of facts: ideas extend prior methods, address known limitations, realize proposed future directions, and sometimes dispute earlier claims. Existing citation graphs usually collapse these roles into a single homogeneous edge type, limiting how we can analyze scientific progress. We address this gap by proposing the SciTraj corpus, the first claim-grounded typed citation graph in which each edge is linked to the specific claim sentence that motivates it. Claim-bearing sentences are extracted from paper sections; four claim-driven relations are verified by NLI entailment against in-paper context, while two similarity-only relations are gated by abstract cosine and year-gap rules. SciTraj contains 32,559 papers from NLP, ML, and Vision (2015--2024), connected by 573,126 directed edges across six relation types, with NLI-verified claim seeds. Using SciTraj, we identify disciplinary siloing in typed citation flow and topic emergence concentrated in Vision and LLM-related work. The corpus also contains 287M typed trajectories of length $\geq 3$, covering 72.8% of papers, and supports a temporally split typed link-prediction benchmark. A year-shuffle falsifiability test separates temporal structure from year-correlated content, and a 3-annotator pilot reports $\kappa = 0.74$ with 79.9% precision.
Chinese Translation
研究如何演变？我们可以通过什么样的基础来预测其未来走向？科学进步并非简单的事实累积：思想延续了先前的方法，解决已知的局限性，实现了提出的未来方向，有时还会对早期的主张提出质疑。现有的引用图通常将这些角色简化为单一的同质边类型，这限制了我们分析科学进展的方式。我们通过提出SciTraj语料库来填补这一空白，这是第一个基于主张的类型化引用图，其中每条边都与激励它的特定主张句子相关联。主张句子从论文的各个部分提取；四种基于主张的关系通过与论文上下文的NLI（自然语言推理）蕴含进行验证，而两种仅基于相似性的关系则通过抽象余弦和年份间隔规则进行筛选。SciTraj包含来自NLP、ML和视觉领域的32,559篇论文（2015-2024），通过573,126条有向边连接，涵盖六种关系类型，并附有经过NLI验证的主张种子。利用SciTraj，我们识别出类型化引用流中的学科孤岛现象，以及集中在视觉和大语言模型（LLM）相关工作的主题出现。该语料库还包含287M条长度≥3的类型化轨迹，覆盖72.8%的论文，并支持一个时间分割的类型化链接预测基准。年份洗牌可证伪性测试将时间结构与年份相关内容分离，而三位注释者的初步报告显示κ值为0.74，精确度为79.9%。
cs.CL / 97 / 2606.22349
Curiosity as Linguistic Intervention: Using LLM Tutoring Dialogues to Influence Exploratory Learning Behavior
好奇心作为语言干预：利用大型语言模型辅导对话影响探索性学习行为
Abstract
Large Language Models (LLMs) provide a new opportunity to study how language shapes exploratory cognition because conversational strategies can be systematically manipulated at inference time. We introduce CURIOBOT, a framework that operationalizes Berlyne's collative variables, novelty, complexity, conflict, and uncertainty, as adaptive linguistic interventions for conversational tutoring. Across 270 tutoring conversations spanning multiple model families, domains, and topic complexity levels, curiosity-oriented interventions consistently increased exploratory learner behaviors, producing up to 2.4x more conversational turns under fixed time budgets. To measure these effects, we further introduce a learner-centered evaluation framework capturing exploratory questioning, conversational agency, productive struggle, and observable curiosity. Learner-side gains persisted even when tutor-side instructional quality remained unchanged, suggesting that curiosity functions as a partially independent interaction-level mechanism. More broadly, our results demonstrate that LLM-mediated dialogue can serve as a scalable experimental framework for studying how language shapes exploratory learning behavior.
Chinese Translation
大型语言模型（LLMs）为研究语言如何塑造探索性认知提供了新的机会，因为对话策略可以在推理时进行系统性操控。我们引入了CURIOBOT，一个将Berlyne的组合变量——新颖性、复杂性、冲突和不确定性——作为对话辅导的自适应语言干预的框架。在跨越多个模型家族、领域和主题复杂性水平的270次辅导对话中，以好奇心为导向的干预 consistently 增加了探索性学习者行为，在固定时间预算下产生了多达2.4倍的对话轮次。为了衡量这些效果，我们进一步引入了一个以学习者为中心的评估框架，捕捉探索性提问、对话主动性、有效挣扎和可观察的好奇心。即使在辅导方的教学质量保持不变的情况下，学习者方面的收益依然存在，这表明好奇心作为一种部分独立的互动层面机制发挥作用。更广泛地说，我们的结果表明，LLM介导的对话可以作为一个可扩展的实验框架，用于研究语言如何塑造探索性学习行为。
cs.CL / 98 / 2606.22357
ORBIT: Training-Free Multi-Attribute Behavioral Steering via Orthogonal Subspace Rotation
ORBIT：通过正交子空间旋转实现无训练的多属性行为引导
Abstract
Language models are widely used in assistant settings, where controlling behavioral attributes is often essential. Activation steering modifies hidden-state representations at inference time, providing a lightweight, training-free mechanism that can be toggled at runtime. Existing methods, however, have focused primarily on steering a single attribute at a time. When multiple attributes must be controlled simultaneously, naive summation of per-attribute steering vectors suffers from norm imbalance and directional cancellation, while classifier-based approaches require retraining whenever the attribute set changes. We introduce ORBIT (Orthogonal Rotation-Based Intervention Technique), a training-free extension of rotation-based steering to the multi-attribute setting. Our method constructs a joint subspace from per-attribute steering planes via singular value decomposition and applies a single norm-preserving rotation within that subspace toward a combined target direction. Adaptive per-token gating identifies which attributes need correction at each position, and an optional additive boost strengthens attributes with weak initial projection. We also introduce TraitFactory, a new multi-attribute benchmark that focuses on behavioral tendencies rather than surface-level style. We evaluate ORBIT on TraitFactory and ToneBank across three models (Llama-3.2-3B, Qwen-2.5-7B, Llama-3.1-8B) while steering multiple attributes simultaneously, showing that it achieves stronger and more balanced multi-attribute steering than existing training-free baselines while better preserving output coherence.
Chinese Translation
语言模型在助手设置中被广泛使用，其中控制行为属性通常至关重要。激活引导在推理时修改隐藏状态表示，提供了一种轻量级的、无训练的机制，可以在运行时切换。然而，现有方法主要集中在一次引导单一属性。当需要同时控制多个属性时，逐属性引导向量的简单相加会遭遇范数不平衡和方向性抵消的问题，而基于分类器的方法在属性集变化时需要重新训练。我们提出了ORBIT（基于正交旋转的干预技术），这是一种无训练的旋转引导扩展，适用于多属性设置。我们的方法通过奇异值分解构建一个由逐属性引导平面组成的联合子空间，并在该子空间内施加一个保持范数的旋转，以朝向组合目标方向。自适应逐标记门控识别每个位置需要纠正的属性，额外的加性增强选项则加强初始投影较弱的属性。我们还引入了TraitFactory，这是一个新的多属性基准，专注于行为倾向而非表面风格。我们在TraitFactory和ToneBank上评估ORBIT，涉及三个模型（Llama-3.2-3B、Qwen-2.5-7B、Llama-3.1-8B），同时引导多个属性，结果表明其在多属性引导方面的表现优于现有的无训练基线，同时更好地保持了输出的一致性。
cs.CL / 99 / 2606.22361
First-Token Broadcasters: Mechanistic Origins of Language Identity and Distributed Robustness in Transformers
首个标记广播器：变压器中语言身份和分布式鲁棒性的机制起源
Abstract
Why do multilingual language models sometimes generate in the wrong language, and why is this so hard to fix? We introduce Language Identity Head Ablation (LIHA), a causal intervention that zeros each attention head individually and measures the resulting language switch rate across a parallel dataset of 2,700 prompt-language pairs spanning seven languages. Applied to GPT-2, LIHA identifies a small set of first-token broadcaster heads - led by L6H1 (switch rate 0.32, 3.23 $\sigma$ above the population mean) - that attend persistently to the first prompt token, propagating its language signal throughout generation. Compensatory redistribution when heads are ablated is statistically significant (p < $10^{-5}$) and follows a directional, hierarchical pattern: compensation always recruits heads in layers above the ablated head, suggesting a feedforward cascade rather than global diffusion. To probe how training regime shapes these circuits, we apply LIHA to a controlled pair - Qwen2.5-1.5B-Base and Qwen2.5-1.5B-Instruct - identical in architecture and size, differing only in training. The base model is nearly flat (max SR=0.016, 200/336 heads at SR=0.0); the instruct model concentrates causal influence sharply at layer 0, led by L0H5 (SR=0.224, 8.93 $\sigma$ above mean), with all other layers near zero. This controlled comparison provides direct causal evidence that instruction tuning reorganizes language identity circuits toward early-layer localization. Extended experiments with Chinese and Russian confirm that first-token broadcasting is script-specific in GPT-2, with non-Latin languages handled at layer 0 - the same locus as the instruction-tuned model. Code and data will be released upon publication.
Chinese Translation
为什么多语言模型有时会生成错误语言的文本，为什么这如此难以修复？我们引入了语言身份头消融（Language Identity Head Ablation, LIHA），这是一种因果干预方法，通过对每个注意力头进行单独消融，测量在一个包含2700个提示-语言对的平行数据集中的语言切换率。应用于GPT-2，LIHA识别出一小部分首个标记广播器头，主要由L6H1主导（切换率0.32，超出总体均值3.23 $ ext{σ}$），这些头持续关注首个提示标记，并在生成过程中传播其语言信号。消融头部时的补偿重分布在统计上显著（p < $10^{-5}$），并遵循一种方向性、层级模式：补偿总是招募位于消融头部之上的层的头部，表明这是一个前馈级联而非全局扩散。为了探讨训练机制如何塑造这些电路，我们将LIHA应用于一对受控模型——Qwen2.5-1.5B-Base和Qwen2.5-1.5B-Instruct——这两个模型在架构和规模上完全相同，仅在训练上有所不同。基础模型的切换率几乎平坦（最大切换率=0.016，200/336个头的切换率=0.0）；而指令模型则在第0层显著集中因果影响，由L0H5主导（切换率=0.224，超出均值8.93 $ ext{σ}$），其他层几乎为零。这一受控比较提供了直接的因果证据，表明指令微调重新组织了语言身份电路，朝向早期层的定位。对中文和俄文的扩展实验确认，在GPT-2中，首个标记广播是特定于脚本的，非拉丁语言在第0层处理——与指令微调模型相同的部位。代码和数据将在发表时发布。
cs.CL / 100 / 2606.22419
Knowledge-Graph Grounding Helps LLMs Only for Out-of-Training Knowledge: A Controlled Study on Clinical Question Answering
知识图谱嵌入仅在超出训练知识时对大型语言模型有帮助：关于临床问答的控制研究
Abstract
A recent Nature Medicine study reports that general-purpose frontier LLMs outperform specialized retrieval-augmented clinical tools on medical benchmarks, and that retrieval can hurt strong models. We ask the natural follow-up: does structured knowledge-graph (KG) grounding change this, and when does grounding help at all? We contribute two results. First, a reproduction: the study's headline HealthBench score (~88) is the Consensus variant, not full HealthBench, where frontier models and ideal completions both score ~46-47 under a physician-calibrated grader (agreement 82.5%); we reproduce GPT-5.2 Consensus =90.9 and flag a score-deflating grader bug. Second, a knowledge-boundary result. Using a graph+vector engine (samyama-graph) over the public biomedical KG PrimeKG, neither naive triple retrieval nor an agentic natural-language-to-Cypher loop (82% successful queries) improves MedQA across a weak-to-strong model ladder (all |Delta| <= 3.4). On a synthetic counterfactual KG, and on a hybrid benchmark mixing known and novel facts, the identical pipeline lifts out-of-training accuracy from chance to ~100% (+68 to +79) while adding nothing on known facts (a no-LLM arm answers both). Across three regimes (no-knowledge, graph-aided, hybrid), grounding helps only insofar as the decisive fact lies outside the model's training -- public-KG facts are redundant, private and novel data are where it pays -- matching the study's institutional-data caveat.
Chinese Translation
最近的一项《自然医学》研究报告称，通用前沿大型语言模型（LLMs）在医学基准测试中优于专门的检索增强临床工具，并且检索可能会对强模型造成负面影响。我们提出一个自然的后续问题：结构化知识图谱（KG）嵌入是否会改变这种情况，以及在何种情况下嵌入确实有帮助？我们贡献了两个结果。首先，重现：该研究的头条健康基准（HealthBench）得分（约88）是共识变体，而非完整的健康基准，其中前沿模型和理想完成在经过医生校准的评分者下均得分约为46-47（一致性82.5%）；我们重现了GPT-5.2共识得分为90.9，并指出了一个导致得分降低的评分者错误。其次，知识边界结果。使用图+向量引擎（samyama-graph）在公共生物医学知识图谱PrimeKG上，既简单的三元组检索也没有通过代理的自然语言到Cypher循环（成功查询率82%）改善MedQA在弱到强模型阶梯上的表现（所有|Delta| <= 3.4）。在一个合成的反事实知识图谱上，以及在一个混合已知和新事实的基准测试中，完全相同的流程将超出训练的准确率从偶然提升至约100%（+68到+79），而对已知事实没有任何提升（无LLM组对两者的回答均相同）。在三个不同的情境下（无知识、图辅助、混合），嵌入仅在决定性事实位于模型训练之外时才有帮助——公共知识图谱的事实是多余的，私有和新数据才是有价值的——这与研究的机构数据警告相符。
cs.CL / 101 / 2606.22430
Words as Difference Makers: How Large Language Models Determine Causal Structure in Text
词语作为差异制造者：大型语言模型如何确定文本中的因果结构
Abstract
Because large language models (LLMs) are impressively successful in predicting text, it appears that they must have access to a 'world model' representing causal and definitional structure. However, the dominant formalisms of modern causal inference -- Judea Pearl's interventionist approach and the Neyman-Rubin potential outcomes framework -- struggle to illuminate how LLMs learn causal structure. I resolve this puzzle by arguing that LLMs employ a specific inductive approach based on a difference-making logic -- sometimes called variational induction. I demonstrate how central aspects of this logic are realized during training, where LLMs require enormous amounts of text data from a wide range of contexts to identify difference- and indifference-makers within word sequences. Furthermore, I analyze specific architectural features of LLMs -- such as token embeddings and self-attention -- to determine their roles in variational induction. The difference-making logic of LLMs fundamentally parallels the experimental method, where causal relations are derived by systematically varying individual circumstances to determine their influence on a phenomenon.
Chinese Translation
由于大型语言模型（LLMs）在文本预测方面表现出色，因此它们似乎必须具备一种代表因果和定义结构的“世界模型”。然而，现代因果推断的主导形式主义——朱迪亚·珀尔（Judea Pearl）的干预主义方法和内曼-鲁宾（Neyman-Rubin）潜在结果框架——在阐明LLMs如何学习因果结构方面存在困难。我通过论证LLMs采用了一种基于差异制造逻辑的特定归纳方法来解决这一难题——有时称为变分归纳。我展示了这种逻辑的核心方面是如何在训练过程中实现的，在此过程中，LLMs需要从广泛的上下文中获取大量文本数据，以识别词序列中的差异制造者和无差异制造者。此外，我分析了LLMs的特定架构特征——如令牌嵌入和自注意力——以确定它们在变分归纳中的作用。LLMs的差异制造逻辑与实验方法在根本上是相似的，其中因果关系是通过系统性地改变个体情况来推导的，以确定其对现象的影响。
cs.CL / 102 / 2606.22454
CASPER in the Machine: Insights into Character Variety in LLM-Generated Stories
机器中的CASPER：对大型语言模型生成故事中的角色多样性的洞察
Abstract
As LLM-generated text is increasingly used, especially in fictional domains, we explore how much LLM-generated stories differ from human-written stories. In this work, we focus on characters. We borrow definitions from narratology to analyze eight intricate dimensions of character, such as stylization and wholeness. These dimensions consider more than just basic characteristics. They assess how characters are portrayed within their stories. After automatically inferring categories of characters within both LLM and human-written stories, we compare and contrast these two sets of stories. We consider the following overarching questions: (1) Do LLMs and human-written stories have similar characters? and (2) Do LLMs generate stories with a variety of characters? Our analysis includes research questions that focus on stories generated by popular LLMs and recently published human-written stories. We describe a number of interesting similarities, differences and key takeaways.
Chinese Translation
随着大型语言模型（LLM）生成的文本在虚构领域的使用日益增多，我们探讨了LLM生成的故事与人类创作的故事之间的差异。在本研究中，我们重点关注角色。我们借用叙事学的定义，分析角色的八个复杂维度，如风格化和完整性。这些维度不仅考虑基本特征，还评估角色在故事中的表现。在自动推断LLM和人类创作故事中的角色类别后，我们对这两组故事进行了比较和对比。我们考虑以下几个总体问题：（1）LLM和人类创作的故事中的角色是否相似？（2）LLM生成的故事是否具有多样化的角色？我们的分析包括关注流行LLM生成的故事和最近发布的人类创作故事的研究问题。我们描述了一些有趣的相似之处、差异和关键发现。
cs.CL / 103 / 2606.22473
Interleaved Speech Language Models Latently Work In Text
交错语音语言模型在文本中潜在工作
Abstract
Speech language models (SLMs) have been extensively studied, with the common paradigm incorporating text data and pre-trained text LMs. A leading approach is speech-text interleaving in which models are trained over sequences containing both speech and text tokens, aiming to boost even speech-only capabilities. Yet the way these two modalities interact in the model latent space remains unclear. In this work, we analyze interleaved speech-text LMs from different model families and sizes through the scope of the logit lens to provide such insight. We reveal that these models go through an implicit transcription phase in which the text token of the spoken word becomes decodable in intermediate layers, despite not being trained for speech recognition. The transcription of the word appears as one of the top candidate words for as much as 77\% of the data. Following this stage, the models proceed to predict the next word in the text space before transforming back to the speech domain. We finally analyze the role of interleaving data, and initializing from text LMs in eliciting this behavior, as well as seeing how this correlates with spoken knowledge abilities. Our analysis sheds light on the internal mechanisms underlying the relationship between speech and text modalities and could shape SLM optimization.
Chinese Translation
语音语言模型（SLMs）已被广泛研究，常见的范式是结合文本数据和预训练的文本语言模型（LMs）。一种主要的方法是语音-文本交错，其中模型在包含语音和文本标记的序列上进行训练，旨在提升语音单独能力。然而，这两种模态在模型潜在空间中的交互方式仍不清楚。在本研究中，我们通过对不同模型家族和规模的交错语音-文本语言模型进行分析，提供了这种洞察，采用了对数几率（logit）视角。我们揭示这些模型经历了一个隐式转录阶段，在该阶段，口语单词的文本标记在中间层中变得可解码，尽管模型并未针对语音识别进行训练。该单词的转录在多达77%的数据中出现为最有可能的候选词。在此阶段之后，模型继续预测文本空间中的下一个单词，然后再转换回语音领域。最后，我们分析了交错数据的作用，以及从文本语言模型初始化如何引发这种行为，并探讨了这与口语知识能力之间的相关性。我们的分析揭示了语音和文本模态之间关系的内部机制，并可能影响SLM的优化。
cs.CL / 104 / 2606.22474
Not All Claims Are Equally Risky: FACTOR for Adaptive Verification in Factual Long-Form Generation
并非所有声明的风险相同：用于事实长文本生成的自适应验证模型FACTOR
Abstract
Large Language Models (LLMs) generate fluent long-form text, however, often add unsupported factual claims. Existing verification techniques improve factuality by grounding generation in external evidence. However, the same verification policy usually applies to all claims despite being differences in hallucination risks. We propose \textit{FACTOR} (\textit{FACTuality-Oriented Risk-aware Verification}), an inference-time model that adapts verification criteria according to claim-level uncertainty. FACTOR combines uncertainty estimation, adaptive language inference verification, and candidate re-ranking to allocate verification effort where it is most needed. We evaluate \textit{FACTOR} on FactScore benchmark showing that adaptive verification improves factuality while reducing verification cost simultaneously. We further perform different ablation studies to identify the primary driver of these gains. Our results show the effective and model-agnostic performance of \textit{FACTOR} for improving factuality in long-form generation.
Chinese Translation
大型语言模型（LLMs）能够生成流畅的长文本，但通常会添加不支持的事实声明。现有的验证技术通过将生成内容与外部证据相结合来提高事实准确性。然而，通常对所有声明应用相同的验证政策，尽管它们在虚构风险上存在差异。我们提出了 extit{FACTOR}（ extit{FACTuality-Oriented Risk-aware Verification}），这是一种在推理时根据声明级别的不确定性调整验证标准的模型。FACTOR结合了不确定性估计、自适应语言推理验证和候选项重新排序，以将验证工作集中在最需要的地方。我们在FactScore基准上评估了 extit{FACTOR}，结果表明自适应验证在提高事实准确性的同时降低了验证成本。此外，我们还进行了不同的消融研究，以确定这些收益的主要驱动因素。我们的结果显示， extit{FACTOR}在提高长文本生成的事实准确性方面表现出有效且与模型无关的性能。
cs.CL / 105 / 2606.22478
ROMEVA: Geometry-Preserving Vocabulary Expansion for Roman Urdu Language Models
ROMEVA：用于罗马乌尔都语言模型的几何保持词汇扩展
Abstract
Multilingual Language Models like mBERT are widely used for low-resource NLP, yet their adaptation to morphologically inconsistent languages such as Roman Urdu remains underexplored. Roman Urdu spelling variation causes severe sub-word fragmentation, averaging 1.50 sub-words per token. We propose \textit{ROMEVA} (Roman Urdu Embedding-preserving Vocabulary Adaptation), which combines sub-word-average initialization and a PCA-guided anchor loss to stabilize embeddings during vocabulary expansion. Using a 36,130-comment Roman Urdu corpus, we add 500 highly fragmented tokens to mBERT and compare naive fine-tuning, sub-word-aware fine-tuning, and \textit{ROMEVA}. While \textit{ROMEVA} most effectively preserves the pretrained embedding space, naive fine-tuning achieves the strongest downstream sentiment classification performance. These findings reveal a disconnect between embedding stability and downstream performance, suggesting that stronger adaptation may be preferable to strict embedding preservation in morphologically inconsistent languages.
Chinese Translation
多语言语言模型如 mBERT 在低资源自然语言处理（NLP）中被广泛使用，但它们在形态不一致的语言（如罗马乌尔都）中的适应性仍然未被充分探索。罗马乌尔都的拼写变异导致严重的子词碎片化，平均每个标记有 1.50 个子词。我们提出了 extit{ROMEVA}（罗马乌尔都嵌入保持词汇适应），它结合了子词平均初始化和 PCA 引导的锚损失，以在词汇扩展过程中稳定嵌入。使用一个包含 36,130 条评论的罗马乌尔都语料库，我们向 mBERT 添加了 500 个高度碎片化的标记，并比较了简单微调、子词感知微调和 extit{ROMEVA}。尽管 extit{ROMEVA} 最有效地保持了预训练的嵌入空间，但简单微调在下游情感分类性能上表现最强。这些发现揭示了嵌入稳定性与下游性能之间的脱节，表明在形态不一致的语言中，较强的适应性可能比严格的嵌入保持更为可取。
cs.CL / 106 / 2606.22511
Breaking the Likelihood Trap: Variance-Calibrated Modulation for Large Language Model Decoding
打破似然陷阱：大语言模型解码的方差校准调制
Abstract
In open-ended generation, LLMs frequently fall into the "likelihood trap", marked by repetitive degeneration and vocabulary dullness, creating a discrepancy between machine-generated and human-written text. While post-hoc tail truncation (e.g., Top-$p$, Min-$p$) avoids sampling from the unreliable tail, it can over-sample from the uncalibrated head and misalign generation with human lexical preferences; fixed scalar repetition penalties likewise ignore variation in logit scale across inference steps, potentially disrupting semantic coherence. To address both limitations, we propose Variance-Calibrated Modulation (VCM), a training-free pre-decoding intervention that reshapes the probability distribution before truncation through two dynamic mechanisms: (1) Contextual Searchlight via PMI, which suppresses global stopwords while elevating context-evoked tokens, and (2) Adaptive Self-Debiasing, which uses real-time logit standard deviation for scale-invariant penalization. Across open-ended generation, factual QA, and mathematical reasoning, VCM consistently mitigates the likelihood trap. With negligible computational overhead, VCM integrates with existing decoding strategies, improving diversity, coherence, and, particularly at higher decoding temperatures, reasoning accuracy.
Chinese Translation
在开放式生成中，大语言模型（LLMs）常常陷入“似然陷阱”，表现为重复退化和词汇乏味，导致机器生成文本与人类撰写文本之间存在差异。虽然事后尾部截断（例如，Top-$p$、Min-$p$）可以避免从不可靠的尾部进行采样，但它可能会过度采样未校准的头部，并使生成与人类词汇偏好不一致；固定标量重复惩罚同样忽视了推理步骤中logit尺度的变化，可能破坏语义连贯性。为了解决这两个局限性，我们提出了方差校准调制（Variance-Calibrated Modulation, VCM），这是一种无训练的预解码干预方法，通过两个动态机制在截断之前重塑概率分布：（1）通过PMI的上下文搜索光束，抑制全局停用词，同时提升上下文激发的标记；（2）自适应自偏差，利用实时logit标准差进行尺度不变的惩罚。在开放式生成、事实问答和数学推理中，VCM始终有效缓解似然陷阱。VCM几乎不增加计算开销，能够与现有解码策略集成，提高多样性、连贯性，尤其是在较高解码温度下，提升推理准确性。
cs.CL / 107 / 2606.22565
Look Light, Think Heavy: What Multimodal Chain-of-Thought Reasoning Can and Cannot Do
轻视视觉，重视思维：多模态链式思维推理的能力与局限
Abstract
Chain-of-Thought (CoT) has become a standard method for improving reasoning capabilities in large language models (LLMs) by eliciting step-by-step thinking, but its effectiveness in multimodal tasks remains unclear. In this paper, we aim to systematically investigate the key question: What can multimodal Chain-of-Thought reasoning do, and where and why does it fall short? To this end, we evaluate 12 multimodal tasks across perception and reasoning categories using both 14 non-reasoning models and 8 reasoning models. Our analysis reveals several important findings: (1) CoT is not a free lunch and should be used selectively depending on the specific requirements of each task. For perception tasks, CoT can lead to undesirable side effects, such as reduced performance in visual grounding and object counting. In contrast, it proves effective for reasoning tasks involving mathematical, scientific, and multi-image reasoning; (2) Compared to original models, existing open-source multimodal reasoning models often yield only marginal overall improvements, possibly due to an overemphasis on mathematical reasoning at the expense of broader capabilities; (3) Visual reasoning remains a key bottleneck for current multimodal CoT, as models exhibit a Look Light, Think Heavy pattern where verbal reflection rises and falls during reasoning, whereas visual reflection consistently diminishes. These findings suggest that while multimodal CoT handles verbal reflection relatively well, it lacks the ability to maintain deep visual introspection throughout the reasoning process.
Chinese Translation
链式思维（Chain-of-Thought, CoT）已成为提高大型语言模型（Large Language Models, LLMs）推理能力的标准方法，通过引导逐步思考，但其在多模态任务中的有效性仍不明确。本文旨在系统地探讨一个关键问题：多模态链式思维推理能做什么，在哪些方面以及为什么会存在不足？为此，我们评估了12个跨感知和推理类别的多模态任务，使用了14个非推理模型和8个推理模型。我们的分析揭示了几个重要发现：（1）CoT并非免费的午餐，应根据每个任务的具体要求选择性使用。对于感知任务，CoT可能导致不良副作用，例如在视觉定位和物体计数中的性能下降。相比之下，它在涉及数学、科学和多图像推理的任务中表现有效；（2）与原始模型相比，现有的开源多模态推理模型通常仅带来边际的整体改善，这可能是由于过于强调数学推理而牺牲了更广泛的能力；（3）视觉推理仍然是当前多模态CoT的一个关键瓶颈，因为模型在推理过程中表现出“轻视视觉，重视思维”的模式，其中语言反思在推理过程中起伏不定，而视觉反思则持续减弱。这些发现表明，尽管多模态CoT在处理语言反思方面相对较好，但在整个推理过程中缺乏维持深度视觉内省的能力。
cs.CL / 108 / 2606.22570
What are Key Factors for Updates in RL for LLM Reasoning?
强化学习在大型语言模型推理中的更新关键因素是什么？
Abstract
Reinforcement Learning from Verifiable Rewards (RLVR) has emerged as a promising framework for enhancing the reasoning ability of large language models. However, much of the existing work is guided by heuristic intuition, leading to divergent algorithmic choices, even contradictory ones that nevertheless report empirical gains. To better understand this phenomenon, we conduct a theoretical analysis of RLVR updates. Our study reveals that differences in off-policy degree, determined by the number of gradient steps per rollout, substantially affect the distribution of importance sampling ratios and their clipping behavior, thereby altering which tokens dominate the update. Building on this insight, we characterize gradient expectation as the central quantity governing update dynamics and analyze the roles of token probability, advantage, and importance sampling ratio. Motivated by these findings, we propose Adaptive Clip Policy Optimization (ACPO), which adjusts clipping boundaries across token groups according to the empirical variance of their importance sampling ratios. Experiments on 3B and 7B models across diverse reasoning benchmarks, spanning mathematical problem solving, tabular QA, and logic puzzles, demonstrate that ACPO outperforms strong baselines such as DAPO and CISPO. These results demonstrate that principled, analysis-driven approaches yield more robust and effective RLVR methods. Code is available in: https://github.com/Control-derek/ACPO
Chinese Translation
可验证奖励的强化学习（Reinforcement Learning from Verifiable Rewards, RLVR）已成为增强大型语言模型推理能力的一个有前景的框架。然而，现有的许多研究工作受到启发式直觉的指导，导致算法选择的多样性，甚至出现相互矛盾的选择，但仍报告了经验上的收益。为了更好地理解这一现象，我们对RLVR更新进行了理论分析。我们的研究揭示了离策略程度的差异，这由每次回滚的梯度步数决定，显著影响了重要性采样比率的分布及其裁剪行为，从而改变了主导更新的标记。基于这一洞察，我们将梯度期望特征化为主导更新动态的中心量，并分析了标记概率、优势和重要性采样比率的作用。受到这些发现的启发，我们提出了自适应裁剪策略优化（Adaptive Clip Policy Optimization, ACPO），该方法根据标记组的重要性采样比率的经验方差调整裁剪边界。在涵盖数学问题解决、表格问答和逻辑难题的多样推理基准上，对3B和7B模型的实验表明，ACPO优于强基线如DAPO和CISPO。这些结果表明，基于原则和分析驱动的方法能够产生更稳健和有效的RLVR方法。代码可在以下链接获取：https://github.com/Control-derek/ACPO
cs.CL / 109 / 2606.22578
Context-Aware Distillation and Ablation for Text2DSL
基于上下文的蒸馏与消融研究用于文本到领域特定语言的转换
Abstract
We extend our prior work on Text2DSL automatic generation of domain-specific language (DSL) code from natural language descriptions along two complementary axes. First, we replace prompt-only synthetic generation with context-aware distillation, in which a teacher large language model (DeepSeek-V4-Flash) operates under an explicitly defined structured context comprising a BNF grammar, an API specification, and a closed identifier vocabulary; the resulting corpus is verified by a two-tier pipeline combining AST validation through esprima and runtime acceptance through the production polkitd daemon and the pkcheck client. This scales the verified PolkitBench corpus from 4,204 to 10,073 natural-language-to-Polkit-rule pairs at 100.0% AST validity and 99.7% runtime pass rate. Second, we conduct the per-component factorial ablation of structured context that was identified as future work in the precursor study: eight conditions C0-C7 are evaluated on GigaChat-10B-A1.8B with the new corpus. Three findings emerge. (i) The new harder corpus collapses the baseline mode (Syntax Valid 97.6% -> 58.5%, Combined Score 0.482 -> 0.252), whereas the context-enhanced mode degrades only marginally (Syntax 98.6% -> 97.4%, Combined 0.801 -> 0.750), confirming that structured context is not a cosmetic improvement but a load-bearing mechanism. (ii) The best absolute condition is the full context C7 across all metrics, while the strongest partial conditions (C5 = BNF + Vocabulary, C6 = API + Vocabulary) both contain the vocabulary. (iii) A Shapley-style decomposition assigns the largest semantic-quality effect to the vocabulary (Combined +0.198), the largest structural-validity effects to API (+24.7 pp) and BNF (+22.3 pp).
Chinese Translation
我们在先前关于从自然语言描述自动生成领域特定语言（DSL）代码的 Text2DSL 工作基础上，沿着两个互补的方向进行了扩展。首先，我们用基于上下文的蒸馏替代了仅基于提示的合成生成，其中教师大型语言模型（DeepSeek-V4-Flash）在一个明确定义的结构化上下文下运行，该上下文包括 BNF 语法、API 规范和封闭的标识符词汇；生成的语料库通过一个两级管道进行验证，该管道结合了通过 esprima 的 AST 验证和通过生产 polkitd 守护进程及 pkcheck 客户端的运行时接受。这样，经过验证的 PolkitBench 语料库从 4,204 对自然语言到 Polkit 规则的配对扩展到 10,073 对，且 AST 有效性达到 100.0%，运行时通过率为 99.7%。其次，我们对在前期研究中确定的结构化上下文进行逐组件的因子消融：在新的语料库上评估了八个条件 C0-C7。得出了三项发现。(i) 新的更难的语料库使基线模式崩溃（语法有效性 97.6% -> 58.5%，综合得分 0.482 -> 0.252），而增强上下文的模式仅略有下降（语法 98.6% -> 97.4%，综合 0.801 -> 0.750），确认结构化上下文不是一种表面改善，而是一个承载机制。(ii) 在所有指标中，最佳的绝对条件是完整上下文 C7，而最强的部分条件（C5 = BNF + 词汇，C6 = API + 词汇）均包含词汇。(iii) Shapley 风格的分解将最大的语义质量效应分配给词汇（综合 +0.198），将最大的结构有效性效应分配给 API (+24.7 个百分点) 和 BNF (+22.3 个百分点)。
cs.CL / 110 / 2606.22606
Sub-Billion, Super-Frontier: Small Language Models Rival Zero-Shot Frontier LLMs on General and Literary Relation Extraction
亚十亿，超前沿：小型语言模型在一般和文学关系抽取上与零-shot前沿LLM相抗衡
Abstract
Large language models (LLMs) achieve strong relation extraction (RE), but their computational demands and reliance on proprietary APIs limit deployment in resource-constrained or privacy-sensitive settings. We investigate how far small language models (SLMs) can close this gap across general-domain and literary text. We evaluate five models from 360M to 3B parameters under three domain-composition regimes and two prompt-conditioned tuning styles (30 configurations), comparing them with zero-shot frontier LLMs and a discriminative RoBERTa baseline. Across nine benchmarks, the best sub-billion model, Qwen2.5-0.5B fine-tuned on pooled general-domain data, achieves a general-domain positive-class micro-F1 of 0.83, versus 0.69 for GPT-5.4 and 0.66 for Claude Sonnet 4.6 evaluated zero-shot. This does not imply that SLMs are intrinsically stronger; rather, targeted task adaptation enables 4-bit models deployable on a single consumer GPU to outperform general-purpose frontier systems under this protocol. An in-domain RoBERTa baseline also exceeds both frontier models, indicating that the gain stems from task adaptation rather than generative decoding. On literary RE, tuned SLMs reach 0.92 on the human-annotated Biographical benchmark versus 0.83 for GPT-5.4, and 0.833 versus 0.578 on the two-benchmark literary average. A targeted domain-adaptive pretraining case study yields no practically meaningful gain over supervised fine-tuning, while the cleanest within-family scale comparison shows only marginal improvement. These results show that, when task-specific data are available, compact task-adapted models can provide accurate, private, and hardware-efficient RE.
Chinese Translation
大型语言模型（LLMs）在关系抽取（RE）方面表现出色，但其计算需求和对专有API的依赖限制了在资源受限或隐私敏感环境中的应用。我们研究了小型语言模型（SLMs）在一般领域和文学文本中能在多大程度上缩小这一差距。我们评估了五个参数从360M到3B的模型，在三种领域组合模式和两种提示条件调优风格下（共30种配置），并将它们与零-shot前沿LLM和一个判别性RoBERTa基线进行比较。在九个基准测试中，最佳的亚十亿模型Qwen2.5-0.5B在汇总的一般领域数据上进行微调，达到了0.83的正类微F1，而GPT-5.4为0.69，Claude Sonnet 4.6在零-shot评估中为0.66。这并不意味着SLMs在本质上更强；相反，针对特定任务的适应使得可在单个消费级GPU上部署的4位模型在该协议下超越了通用前沿系统。在领域内的RoBERTa基线也超过了这两个前沿模型，表明收益源于任务适应而非生成解码。在文学RE方面，经过调优的SLMs在人工标注的传记基准上达到了0.92，而GPT-5.4为0.83，在两个基准的文学平均值上分别为0.833和0.578。一个针对特定领域的自适应预训练案例研究未能在监督微调上带来实际有意义的增益，而同一家族内的规模比较显示出仅有边际改善。这些结果表明，当有特定任务数据可用时，紧凑的任务适应模型能够提供准确、私密且硬件高效的关系抽取。
cs.CL / 111 / 2606.22627
Orthogonal Representation Editing: Decoupling Semantic Entanglement in Batch Knowledge Editing of LLMs
正交表示编辑：解耦大型语言模型批量知识编辑中的语义纠缠
Abstract
Knowledge editing aims to efficiently update factual information in Large Language Models (LLMs) without full retraining. However, existing methods still suffer from performance degradation in batch knowledge editing. We identify that semantic representation entanglement, such as overlapping concepts and shared syntactic patterns, accumulates interference in the representation space and reduces editing precision. To bridge this gap, in this paper, we propose Orthogonal Representation Editing (ORE), which performs edits in the hidden representation space of LLMs by constructing a general semantic subspace and enforcing orthogonal constraints on edit vectors, effectively decoupling semantic entanglement. Furthermore, we introduce a gated non-linear representation head to enable adaptive learning of editing locations and precise control over knowledge injection. Extensive experiments show that ORE outperforms existing methods and achieves superior performance in cross-lingual knowledge editing scenarios. We release our code at https://github.com/YVVH/ORE.
Chinese Translation
知识编辑旨在高效更新大型语言模型（LLMs）中的事实信息，而无需进行全面的再训练。然而，现有方法在批量知识编辑中仍然存在性能下降的问题。我们发现，语义表示的纠缠，例如重叠概念和共享句法模式，会在表示空间中积累干扰，从而降低编辑精度。为了解决这一问题，本文提出了正交表示编辑（Orthogonal Representation Editing, ORE），该方法通过构建一个通用的语义子空间并对编辑向量施加正交约束，在LLMs的隐藏表示空间中进行编辑，有效地解耦语义纠缠。此外，我们引入了一个门控非线性表示头，以实现编辑位置的自适应学习和对知识注入的精确控制。大量实验表明，ORE优于现有方法，并在跨语言知识编辑场景中取得了更好的性能。我们的代码已发布在 https://github.com/YVVH/ORE。
cs.CL / 112 / 2606.22681
Only Ask What You Don't Know: Grounded Delta Planning for Efficient Multi-step RAG
只询问你不知道的：针对高效多步骤检索增强生成的基础增量规划
Abstract
Multi-hop question answering remains challenging for Retrieval-Augmented Generation (RAG) because existing approaches either propagate errors across iterative retrieval rounds or over-generate reasoning steps, increasing cost without improving accuracy. We propose Grounded Delta Planning RAG (GDP-RAG), a plan-based framework that targets only the information delta based on three simple design choices: (1) preliminary retrieval to ground planning before execution, (2) a gap-conditioned planning prompt that asks only for missing information, and (3) a skeletal trajectory that pairs each subquery with a Thought capturing evidence from preliminary retrieval and carrying it through to the final answer. GDP-RAG focuses computation on unresolved gaps, yielding concise, reliable reasoning trajectories. Extensive experiments on HotpotQA, 2WikiMultiHopQA, and MuSiQue show that GDP-RAG achieves the highest accuracy (60.63%) among all compared systems while maintaining a cost-of-pass of 0.51, 22% lower than PAR-RAG (0.65) and 68% lower than KnowTrace (1.57), with no method achieving both higher accuracy and lower cost.
Chinese Translation
多跳问答对于检索增强生成（RAG）仍然具有挑战性，因为现有方法要么在迭代检索过程中传播错误，要么过度生成推理步骤，导致成本增加而准确性没有提高。我们提出了基础增量规划RAG（GDP-RAG），这是一个基于规划的框架，仅针对基于三个简单设计选择的信息增量：（1）在执行前进行初步检索以为规划提供基础，（2）一个基于差距的规划提示，仅询问缺失的信息，以及（3）一个骨架轨迹，将每个子查询与从初步检索中捕获证据的思维相结合，并将其带入最终答案。GDP-RAG将计算集中在未解决的差距上，从而产生简洁、可靠的推理轨迹。在HotpotQA、2WikiMultiHopQA和MuSiQue上的大量实验表明，GDP-RAG在所有比较系统中实现了最高的准确率（60.63%），同时保持了0.51的通过成本，比PAR-RAG（0.65）低22%，比KnowTrace（1.57）低68%，且没有任何方法在实现更高准确率的同时降低成本。
cs.CL / 113 / 2606.22722
moBERTo: A Modern Encoder for Portuguese via Continued Pretraining of ModernBERT
moBERTo：通过对ModernBERT的持续预训练实现的现代葡萄牙语编码器
Abstract
Encoder-only transformer models remain essential for production NLP pipelines. We introduce moBERTo, a Portuguese adaptation of ModernBERT obtained through continued pretraining of the ModernBERT-base checkpoint on 60 billion tokens (5 epochs over a 12-billion-token corpus curated from FineWeb2 and filtered with educational and STEM classifiers). We preserve the original architecture, including rotary positional embeddings, alternating local-global attention, flash attention, and unpadding. We evaluate moBERTo across information retrieval (including long-context retrieval at up to 8,192 tokens), document classification, named entity recognition, and natural language understanding. Our best variant, which combines a Portuguese tokenizer with subword-matching embedding transfer and long-context post-training, achieves the highest average reranking nDCG@10 across three Portuguese retrieval benchmarks and the best results on PLUE-PT. Through ablation studies, we show that (i) continued pretraining is strongly preferable to training from scratch, particularly for preserving long-context capabilities; (ii) tokenizer adaptation improves token-level tasks but degrades long-context retrieval; (iii) a dedicated long-context post-training phase at 8,192 tokens further improves reranking and NER; and (iv) encoder-only architectures remain competitive with larger decoder-only alternatives for discriminative tasks. We publicly release the model weights at https://huggingface.co/Tropic-AI/moBERTo and training data at https://huggingface.co/datasets/Tropic-AI/moberto-pretraining-dataset-c4-compatible on Hugging Face.
Chinese Translation
仅编码器的变换器模型在生产NLP管道中仍然至关重要。我们介绍了moBERTo，这是对ModernBERT的葡萄牙语适配，通过对ModernBERT-base检查点在600亿个标记（在从FineWeb2整理的120亿个标记语料库上进行了5个周期的持续预训练）进行的。我们保留了原始架构，包括旋转位置嵌入、交替的局部-全局注意力、快速注意力和去填充。我们在信息检索（包括长上下文检索，最多可达8192个标记）、文档分类、命名实体识别和自然语言理解等任务上评估了moBERTo。我们的最佳变体结合了葡萄牙语分词器、子词匹配嵌入转移和长上下文后训练，在三个葡萄牙语检索基准上实现了最高的平均重新排序nDCG@10，并在PLUE-PT上取得了最佳结果。通过消融研究，我们表明：（i）持续预训练明显优于从头开始训练，特别是在保持长上下文能力方面；（ii）分词器适配改善了标记级任务，但降低了长上下文检索的效果；（iii）在8192个标记的专用长上下文后训练阶段进一步改善了重新排序和命名实体识别；（iv）仅编码器架构在判别任务中仍然与更大的仅解码器替代方案具有竞争力。我们在https://huggingface.co/Tropic-AI/moBERTo公开发布了模型权重，并在Hugging Face上发布了训练数据，链接为https://huggingface.co/datasets/Tropic-AI/moberto-pretraining-dataset-c4-compatible。
cs.CL / 114 / 2606.22723
BLUEX v2: Benchmarking LLMs on Open-Ended Questions from Brazilian University Entrance Exams
BLUEX v2：基于巴西大学入学考试的开放性问题对大型语言模型的基准评估
Abstract
Although Large Language Models (LLMs) excel in many tasks, their assessment in Portuguese has received less attention, particularly for open-ended, discursive tasks that demand deeper reasoning and generation capabilities. While the original BLUEX benchmark addressed the scarcity of Portuguese evaluation datasets through multiple-choice questions from Brazilian university entrance exams, it did not cover the more challenging second-phase examinations, which require free-form written responses. In this work, we introduce BLUEX v2, a benchmark derived from the second-phase entrance exams of Brazil's two leading universities: UNICAMP (Comvest) and USP (Fuvest), spanning exam years 2022-2025. Our dataset comprises 395 questions unfolding into 919 graded subquestions, with 55.7% of questions containing associated images. Each question is annotated with subject area, official reference answers, LLM-generated rubric criteria, and six cognitive capability tags. We evaluate 21 state-of-the-art LLMs using an LLM-as-a-judge protocol. Results reveal a 4.92-point performance spread across models (4.18-9.10 on a 0-10 scale), with Mathematical Reasoning and Image Understanding emerging as the hardest capability dimensions. The dataset, evaluation code, and model outputs are publicly available at https://anonymous.4open.science/r/BLUEXv2.
Chinese Translation
尽管大型语言模型（LLMs）在许多任务中表现出色，但它们在葡萄牙语中的评估却受到较少关注，特别是在需要更深层次推理和生成能力的开放性讨论任务中。虽然原始的BLUEX基准通过巴西大学入学考试的选择题解决了葡萄牙语评估数据集的稀缺问题，但它并未涵盖更具挑战性的第二阶段考试，这些考试要求自由形式的书面回答。在本研究中，我们引入了BLUEX v2，这是一个源自巴西两所顶尖大学（UNICAMP（Comvest）和USP（Fuvest））的第二阶段入学考试的基准，涵盖2022-2025年的考试年份。我们的数据集包含395个问题，展开为919个评分子问题，其中55.7%的问题包含相关图像。每个问题都附有学科领域、官方参考答案、LLM生成的评分标准和六个认知能力标签。我们使用LLM作为评审的协议评估了21个最先进的LLM。结果显示，模型之间的表现差异为4.92分（在0-10的评分范围内为4.18-9.10），其中数学推理和图像理解被认为是最难的能力维度。数据集、评估代码和模型输出已在https://anonymous.4open.science/r/BLUEXv2上公开。
cs.CL / 115 / 2606.22728
When Confidence Takes the Wrong Path: Diagnosing Retrieval-State Lock-In in RAG
当信心走上错误的道路：诊断检索状态锁定在检索增强生成中的表现
Abstract
The trustworthiness of a retrieval-augmented generation (RAG) system depends on more than the answer it returns, yet many black-box uncertainty methods still read agreement among sampled answers as confidence. That inference fails when repeated samples condition on the same defective retrieval state. The state may be empty, with the model falling back on parametric memory, or populated by a coherent but wrong neighbourhood. In either case, the answers agree because the error is stable. The problem is recognised in deployed RAG, but it has lacked a name, a measurable signature, and a prevalence bound. We supply all three. We name the failure retrieval-state lock-in and diagnose it by separating the three objects a single confidence score conflates: the answer surface, the retrieved evidence, and the retrieval state itself. In an inspectable, ontology-guided knowledge-graph RAG (KG-RAG) system across six question-answering snapshots, we measure the agreement blind spot directly: at five samples per question, 42% of KG-RAG errors and 59% of dense-retrieval errors carry zero answer dispersion, so agreement has nothing to rank, while evidence- and retrieval-state checks still flag most of them. The decomposition supports an auditable decision rule: accepting an answer only when answer, evidence, and retrieval checks all agree that it is low-risk reaches 91.9% pooled precision against a 69.7% accept-all rate. The cost is coverage: it certifies only 7.7% of answers as low-risk. On the clinical calibration domain it reaches 100% precision under an automated judge; this is an in-domain automated-label upper bound, not a clinical safety claim, and still needs human validation. Confidence in RAG is object-specific: when answers agree, the useful question is which part of the pipeline to distrust.
Chinese Translation
检索增强生成（RAG）系统的可信度不仅仅取决于返回的答案，然而许多黑箱不确定性方法仍然将采样答案之间的一致性视为信心。当重复样本依赖于相同的缺陷检索状态时，这种推断就会失败。该状态可能是空的，模型回退到参数记忆，或者由一致但错误的邻域填充。在这两种情况下，答案之所以一致是因为错误是稳定的。这个问题在已部署的RAG中得到了认可，但缺乏一个名称、可测量的特征和普遍性界限。我们提供了这三者。我们将这一失败命名为检索状态锁定，并通过分离单一置信度分数所混淆的三个对象来诊断它：答案表面、检索证据和检索状态本身。在一个可检查的、基于本体的知识图谱RAG（KG-RAG）系统中，通过六个问答快照，我们直接测量了一致性盲区：在每个问题五个样本的情况下，42%的KG-RAG错误和59%的密集检索错误的答案分散度为零，因此一致性没有任何排名，而证据和检索状态检查仍然标记了大多数错误。该分解支持可审计的决策规则：仅在答案、证据和检索检查均一致认为其低风险时接受答案，达到了91.9%的综合精度，而接受所有答案的比例为69.7%。其代价是覆盖率：仅有7.7%的答案被认证为低风险。在临床校准领域，在自动评判者下其精度达到100%；这只是一个领域内自动标签的上限，而不是临床安全声明，仍需人工验证。RAG中的信心是特定于对象的：当答案一致时，值得关注的问题是管道的哪个部分值得怀疑。
cs.CL / 116 / 2606.22745
Language-Specific Sentiment Polarity Biases in Encoder and Large Language Model Classification of Product Reviews
编码器和大型语言模型在产品评论分类中的语言特定情感极性偏差
Abstract
This study investigates sentiment polarity biases, specifically, differences in how accurately AI models classify positive versus negative reviews across languages and model architectures. Large language models show a negative bias in French and are more accurate on negative reviews, while encoder models exhibit positive bias in Japanese, missing negative reviews that use indirect criticism. These language-specific polarity biases have implications in both social and business domains deploying multilingual sentiment analysis systems.
Chinese Translation
本研究探讨了情感极性偏差，特别是人工智能模型在不同语言和模型架构中对正面与负面评论分类的准确性差异。大型语言模型在法语中表现出负向偏差，对负面评论的分类更为准确，而编码器模型在日语中则表现出正向偏差，未能准确识别使用间接批评的负面评论。这些语言特定的极性偏差对在社会和商业领域部署多语言情感分析系统具有重要影响。
cs.CL / 117 / 2606.22748
AI Fiction in the Wild
野外的人工智能小说
Abstract
Some professional authors are beginning to use AI tools to help produce their fiction writing. Are readers using AI to generate fiction, too? This paper examines how large language models are reshaping the production and consumption of fiction by enabling new forms of participation in narrative generation. Drawing on over 500,000 anonymized, English-language ChatGPT-user conversations (arXiv:2405.01470), we find that more than one third of the conversations involve some form of fiction generation -- including original stories, roleplay, fanfiction, and erotica. This AI-generated fiction is notably dominated by power users. We identify common fiction generation patterns and profiles among these users, including what we call "infinite story demanders," who repeatedly request and revise variations of the same or similar narratives over extended periods of time. We show that users especially gravitate toward fanfiction and erotica, and that they are broadly drawn to generic forms, repetition, immediacy, and niche combinations of story elements. Our findings motivate two theoretical provocations. First, we argue that AI technologies may lead to a shift in the conventional relationship between the author and reader, potentially producing what we call a "solipsistic reader-writer," who both generates and consumes fiction within a closed conversational loop, interacting with a machine rather than a human other. Second, we note that LLMs enable interactivity, play, and permutation in ways that are seemingly pleasurable for users, raising questions about where AI will fit into contemporary storytelling and entertainment ecosystems. We situate these developments within broader transformations in literature and media, including self-publishing, fanfiction, and pornography, and suggest that AI-generated fiction shares structural affinities with on-demand, personalized, and repetitive cultural forms.
Chinese Translation
一些专业作者开始使用人工智能工具来帮助创作小说。那么，读者是否也在使用人工智能生成小说呢？本文探讨了大型语言模型如何通过促进叙事生成的新形式参与，重塑小说的创作和消费。基于超过500,000个匿名的英语ChatGPT用户对话（arXiv:2405.01470），我们发现超过三分之一的对话涉及某种形式的小说生成——包括原创故事、角色扮演、同人小说和情色文学。这种人工智能生成的小说显著地由高频用户主导。我们识别出这些用户中常见的小说生成模式和特征，包括我们所称的“无限故事需求者”，他们在较长时间内反复请求和修订相同或相似叙事的变体。我们展示了用户特别倾向于同人小说和情色文学，并且他们普遍被通用形式、重复性、即时性和故事元素的小众组合所吸引。我们的研究结果引发了两个理论上的思考。首先，我们认为人工智能技术可能导致作者与读者之间传统关系的转变，可能产生我们所称的“自我中心的读者-作者”，他们在一个封闭的对话循环中既生成又消费小说，与机器互动而非与人类互动。其次，我们注意到大型语言模型（LLMs）以看似愉悦的方式促进了互动、游戏和变换，这引发了关于人工智能在当代叙事和娱乐生态系统中将如何定位的问题。我们将这些发展置于更广泛的文学和媒体转型背景中，包括自出版、同人小说和色情文学，并建议人工智能生成的小说在结构上与按需、个性化和重复的文化形式具有相似性。
cs.CL / 118 / 2606.22771
Learning Moral Diversity: Modelling Individual Perspectives in Moral Classification of Texts
学习道德多样性：文本道德分类中的个体视角建模
Abstract
Understanding moral values in social media text offers insight into moral judgement formation, and supervised NLP models trained on crowdsourced data have achieved strong classification performance. However, most approaches simplify the problem by aggregating multiple annotators' labels into a single "ground truth", overlooking the inherent subjectivity of the task. In practice, there are disagreements between annotators caused by personal viewpoint or inherent ambiguities, particularly for short tweets. Here, we extend a pretrained language model with a layer that learns annotator-specific features. Our model improves predictions of individual annotations and yields representations that reveal meaningful insights into annotators' moral perspectives. We show that models trained on aggregated labels may hide variation and give a misleading impression of performance. Overall, we demonstrate that disagreement reflects the inherent subjectivity of the task and that modelling individual perspectives creates benefits for moral classification of texts.
Chinese Translation
理解社交媒体文本中的道德价值观有助于洞察道德判断的形成，而基于众包数据训练的监督自然语言处理模型已取得了良好的分类性能。然而，大多数方法通过将多个标注者的标签聚合为单一的“真实标签”来简化问题，忽视了任务固有的主观性。在实际操作中，由于个人观点或固有的模糊性，标注者之间存在分歧，尤其是在短推文中。在此，我们扩展了一个预训练语言模型，增加了一层以学习标注者特定特征。我们的模型改善了对个体标注的预测，并生成了揭示标注者道德视角的有意义见解的表示。我们展示了基于聚合标签训练的模型可能隐藏变异，并给出误导性的性能印象。总体而言，我们证明了分歧反映了任务的固有主观性，并且建模个体视角为文本的道德分类带来了好处。
cs.CL / 119 / 2606.22798
Does the Same Token Mean the Same State? MoE Routing as Signal for Reasoning Control
相同的标记是否意味着相同的状态？MoE 路由作为推理控制的信号
Abstract
In sparse Mixture-of-Experts language models, does the same token id imply the same router state and the same experts producing it? Holding the emitted token id fixed at repeated anchors, we find it does not: the experts that produce it still separate task context, trajectory history, and reasoning-effort mode. This residual structure supports test-time control: near \emph{boundary} anchors (the final-response transition) and \emph{delimiter} anchors (which open the answer, e.g.\ \texttt{\textbackslash boxed\{} or code fences), routing neighborhoods already align with final-answer basins at a marker-only readout and strongest when the routing is read at the answer opening. We operationalize this as \textbf{RAD} (Routing Agreement Decoding), an answer-string-free multi-rollout selector: it locates a fixed anchor, represents each rollout by its anchor-window MoE routing states, and returns the densest Weighted-Jaccard $K$-NN route-basin center, without parsing, normalizing, executing, or voting over answer strings. Across 10 sparse-MoE configurations (gpt-oss, Qwen3-MoE) and 6 datasets spanning math, GPQA, and code, RAD is on par with Majority where string voting is well-posed, with small positive paired deltas (RAD $73.9$ / RAD+DC $74.2$ vs.\ Majority $73.6$). Like majority voting, RAD is not a verifier: a dense \emph{wrong} basin can still win. Its value is the interface: the same selector gives direct pass@1 on code, where exact-string voting is ill-defined, and the same routing-density principle, re-anchored to the agentic boundary, improves best-of-16 patch selection on SWE-bench Verified over random, where patches have no answer string to vote on.
Chinese Translation
在稀疏混合专家语言模型中，相同的标记 ID 是否意味着相同的路由器状态和相同的专家生成它？在重复锚点上固定发出的标记 ID，我们发现并非如此：生成它的专家仍然区分任务上下文、轨迹历史和推理努力模式。这种残余结构支持测试时控制：在接近 extit{边界} 锚点（最终响应过渡）和 extit{分隔符} 锚点（打开答案的标记，例如 exttt{ extbackslash boxed extbraceleft} 或代码围栏）时，路由邻域已经与最终答案的基线对齐，并且在路由在答案开启时读取时最强。我们将其操作化为 extbf{RAD}（路由一致解码），一种无答案字符串的多次回滚选择器：它定位一个固定的锚点，通过其锚窗口的 MoE 路由状态表示每次回滚，并返回最密集的加权 Jaccard $K$-最近邻路由基线中心，而无需解析、规范化、执行或对答案字符串进行投票。在 10 种稀疏 MoE 配置（gpt-oss、Qwen3-MoE）和涵盖数学、GPQA 和代码的 6 个数据集上，RAD 的表现与多数投票相当，在字符串投票良好定义的情况下，具有小的正配对增量（RAD $73.9$ / RAD+DC $74.2$ 对比 Majority $73.6$）。与多数投票一样，RAD 不是验证器：一个密集的 extit{错误} 基线仍然可以获胜。它的价值在于接口：相同的选择器在代码上提供直接的 pass@1，而精确字符串投票则定义不明确，并且相同的路由密度原则重新锚定到代理边界，提高了在 SWE-bench Verified 上的最佳 16 个补丁选择，相比随机选择，补丁没有答案字符串可供投票。
cs.CL / 120 / 2606.22807
KaLM-Reranker-V1: Fast but Not Late Interaction for Compressed Document Reranking
KaLM-Reranker-V1：快速但非迟到交互的压缩文档重排序
Abstract
As retrieval systems scale, high-quality reranking becomes increasingly important. However, most existing rerankers, whether encoder-based or decoder-based, jointly encode the query and passage, tightly coupling their computation and limiting deployment efficiency as well as flexibility. We present KaLM-Reranker-V1, a fast but not late-interaction (FBNL) reranker that decouples query and passage computation while retaining expressive relevance modeling. Built on an encoder-decoder architecture, KaLM-Reranker-V1 uses the encoder to pre-encode passages with Matryoshka embedding pooling, while the decoder models the system instruction, user instruction, and query intent; cross-attention then captures relevance between the query context and passage representations. This design makes KaLM-Reranker-V1 efficient through decoupled passage encoding, yet not late interaction, by preserving rich relevance modeling through cross-attention. We instantiate KaLM-Reranker-V1 in three sizes, Nano, Small, and Large, with 0.27B, 1B, and 4B activated parameters, respectively. Extensive experiments on BEIR, MIRACL, and LMEB demonstrate that KaLM-Reranker-V1 achieves strong reranking performance with superior efficiency. On BEIR, KaLM-Reranker-V1 achieves state-of-the-art performance, on par with strong industrial models such as the Qwen3-Reranker series; on MIRACL, despite not being extensively trained on multilingual data, KaLM-Reranker-V1 still shows excellent reranking performance. Moreover, on LMEB, reranking models demonstrate a clear advantage, with even the 0.27B Nano model remaining competitive with 7-12B embedding models.
Chinese Translation
随着检索系统的规模扩大，高质量的重排序变得越来越重要。然而，大多数现有的重排序模型，无论是基于编码器还是解码器，都将查询和文段的编码紧密耦合，限制了计算效率和灵活性。我们提出了KaLM-Reranker-V1，这是一种快速但非迟到交互（FBNL）的重排序模型，它在保留表达性相关性建模的同时，解耦了查询和文段的计算。KaLM-Reranker-V1基于编码器-解码器架构，使用编码器通过Matryoshka嵌入池化对文段进行预编码，而解码器则建模系统指令、用户指令和查询意图；交叉注意力机制捕捉查询上下文与文段表示之间的相关性。这一设计使得KaLM-Reranker-V1通过解耦的文段编码实现高效，同时通过交叉注意力保留丰富的相关性建模，避免了迟到交互。我们将KaLM-Reranker-V1实例化为三种尺寸：Nano、Small和Large，分别具有0.27B、1B和4B的激活参数。在BEIR、MIRACL和LMEB上的大量实验表明，KaLM-Reranker-V1在保持卓越效率的同时，取得了强大的重排序性能。在BEIR上，KaLM-Reranker-V1达到了与强大的工业模型如Qwen3-Reranker系列相当的最先进性能；在MIRACL上，尽管没有在多语言数据上进行广泛训练，KaLM-Reranker-V1仍然展现出优秀的重排序性能。此外，在LMEB上，重排序模型展现出明显的优势，即使是0.27B的Nano模型也与7-12B的嵌入模型保持竞争力。
cs.CL / 121 / 2606.22811
Bagpiper-TTS: Natural Language Guided Universal Speech Synthesis
Bagpiper-TTS：自然语言引导的通用语音合成
Abstract
Classical TTS systems typically rely on rigid input formats and predefined metadata slots, limiting their ability to fulfill flexible user requirements. This paper introduces Bagpiper-TTS, a universal speech synthesis system that deals with diverse natural language user requests. Given a natural language prompt, Bagpiper-TTS first reasons over the users' intent to derive a rich caption, i.e., a comprehensive textual blueprint encompassing both transcription and nuanced metadata. Subsequently, this caption guides the synthesis of the target speech. Our model inherently supports a broad spectrum of tasks besides classical TTS applications, including multi-talker, intent-to-speech, role-play synthesis, singing voice synthesis, and more. Experimental results demonstrate that Bagpiper-TTS achieves an 1.7% Word Error Rate (WER) on the Seed-TTS-Eval benchmark and match the performance of dedicated models in both LLM-as-a-judge and human subjective evaluations across multiple applications.
Chinese Translation
传统的文本到语音（TTS）系统通常依赖于固定的输入格式和预定义的元数据槽，这限制了它们满足灵活用户需求的能力。本文介绍了Bagpiper-TTS，一种通用语音合成系统，能够处理多样的自然语言用户请求。在接收到自然语言提示后，Bagpiper-TTS首先推理用户的意图，以生成丰富的说明，即一个全面的文本蓝图，涵盖转录和细致的元数据。随后，这个说明指导目标语音的合成。我们的模型本质上支持广泛的任务，除了传统的TTS应用外，还包括多说话者、意图到语音、角色扮演合成、歌声合成等。实验结果表明，Bagpiper-TTS在Seed-TTS-Eval基准测试中实现了1.7%的词错误率（WER），并在多个应用中与专用模型的性能相匹配，包括LLM作为评判者和人类主观评估。
cs.CL / 122 / 2606.22841
IndicGuard: A Multilingual Safety Guard Model and Dataset for Indic Languages
IndicGuard：一个针对印度语言的多语言安全防护模型和数据集
Abstract
As Large Language Models (LLMs) achieve widespread integration across diverse linguistic landscapes, ensuring their safety and alignment with regional normative values remains a critical challenge. Current safety mechanisms are predominantly optimized for English-centric frameworks, often failing to capture the unique socio-cultural sensitivities and localized categories of harm inherent to the Indic region. To address this gap, we introduce IndicGuard, a multilingual safety guard model and dataset for Indic languages. We construct a high-volume, culturally nuanced safety dataset encompassing ten major Indic languages, systematically curated to capture regional harms, sensitive socio-political contexts, and adversarial jailbreaks. Leveraging this corpus, we fine-tune a 4B-parameter instruction-tuned model based on Gemma-3-4B-IT to serve as a multilingual safety guardrail for real-time content moderation and policy compliance checking. Our empirical evaluations demonstrate that IndicGuard significantly enhances LLM robustness against localized vulnerabilities, achieving high moderation consistency across different conversational turns. Crucially, IndicGuard consistently outperforms the existing baseline model, CultureGuard, across evaluated languages. Finally, we demonstrate that our model effectively generalizes to low-resource Indic languages excluded from training, substantiating the structural robustness and cross-lingual transfer capabilities of the framework.
Chinese Translation
随着大型语言模型（LLMs）在多样化语言环境中的广泛应用，确保它们的安全性和与地区规范价值观的一致性仍然是一个关键挑战。目前的安全机制主要针对以英语为中心的框架，往往无法捕捉到印度地区独特的社会文化敏感性和本地化的伤害类别。为了解决这一问题，我们提出了IndicGuard，一个针对印度语言的多语言安全防护模型和数据集。我们构建了一个高容量、文化细致的安全数据集，涵盖十种主要的印度语言，系统性地策划以捕捉地区性伤害、敏感的社会政治背景和对抗性越狱。利用这一语料库，我们对基于Gemma-3-4B-IT的4B参数指令调优模型进行了微调，以作为实时内容审核和政策合规检查的多语言安全防护措施。我们的实证评估表明，IndicGuard显著增强了LLM对本地化脆弱性的鲁棒性，在不同对话轮次中实现了高一致性的审核。重要的是，IndicGuard在评估的语言中始终优于现有的基线模型CultureGuard。最后，我们证明了我们的模型能够有效地推广到训练中未包含的低资源印度语言，证实了该框架的结构鲁棒性和跨语言迁移能力。
cs.CL / 123 / 2606.22877
DynamicMem: A Long-Horizon Memory Benchmark in Real-World Settings
DynamicMem：真实环境中的长时间记忆基准测试
Abstract
LLM agents increasingly act as personal assistants that must remember a user's profile over months: who they are (attributes), what they routinely do (habits), and what they prefer (preferences), and keep it updated as jobs, routines, and tastes drift. Existing benchmarks evaluate this "memory" ability through short, simplified interactions, missing three core properties of real behavior: the profile is heterogeneous, with attributes, habits, and preferences evolving on different timelines; changes are driven by external context such as seasons and life events; and evidence is rarely stated explicitly, instead scattered across many small actions in different apps that a memory system must infer from. We introduce DynamicMem, a synthetic benchmark that constructs 15 months of activity per user, providing long-term multi-app data that real users' privacy keeps out of reach. It provides user-consistent trajectories averaging 2.2M tokens and 1,772 grounded events per user across 16 applications such as e-commerce, fitness, and social platforms. The profile evolves over this period and is never given explicitly: each attribute, habit, or preference must be inferred from small signals scattered across apps. We evaluate at five quarterly checkpoints to track how systems scale as history grows. Benchmarking five representative systems exposes problems a single accuracy score hides: (i) profile reconstruction degrades with history length while service-task accuracy stays flat, despite both drawing on the same memory; (ii) no system both keeps facts that stay true and replaces facts that change, with errors clustering on preferences and on naming the exact referent; and (iii) over 93% of failures trace to what the memory retrieves, not to the model writing the answer, so the largest room for improvement lies in memory itself. Code: https://wenyaxie023.github.io/DynamicMem/
Chinese Translation
大型语言模型（LLM）代理越来越多地作为个人助手，必须在数月内记住用户的个人资料：他们是谁（属性）、他们的日常活动（习惯）以及他们的偏好（偏好），并在工作、日常和口味变化时保持更新。现有基准通过短期、简化的互动来评估这种“记忆”能力，忽视了真实行为的三个核心特性：个人资料是异质的，属性、习惯和偏好在不同的时间线上演变；变化受到外部环境的驱动，例如季节和生活事件；证据通常不会明确陈述，而是散布在不同应用中的许多小动作中，记忆系统必须从中推断。我们引入了DynamicMem，这是一个合成基准，为每个用户构建15个月的活动数据，提供长期的多应用数据，这些数据因真实用户的隐私而无法获取。它提供了一致的用户轨迹，平均每个用户有220万个标记和1,772个实际事件，涵盖电子商务、健身和社交平台等16个应用。个人资料在此期间不断演变，且从未明确给出：每个属性、习惯或偏好必须从散布在应用中的小信号中推断。我们在五个季度检查点进行评估，以跟踪系统在历史增长时的扩展情况。对五个代表性系统的基准测试揭示了单一准确性评分所隐藏的问题：（i）个人资料重建随着历史长度的增加而退化，而服务任务的准确性保持平稳，尽管两者都依赖于相同的记忆；（ii）没有任何系统能够同时保持真实的事实并替换变化的事实，错误主要集中在偏好和确切指称的命名上；（iii）超过93%的失败源于记忆检索，而不是模型写出答案，因此最大的改进空间在于记忆本身。代码链接： https://wenyaxie023.github.io/DynamicMem/
cs.CL / 124 / 2606.22886
Explanation-Guided Medical Named Entity Recognition with Stability and Boundary Awareness for Atopic Dermatitis
基于解释引导的特应性皮炎医学命名实体识别的稳定性与边界意识
Abstract
Objective: This study aims to improve the reliability and robustness of medical named entity recognition (NER) in Chinese atopic dermatitis (AD) clinical texts through explanation-guided learning. Methods: We propose a stability and boundary-aware explanation-guided NER framework. Perturbation-based analysis is used to evaluate explanation stability and entity boundary sensitivity. An adaptive fusion strategy dynamically combines local and global explanation to generate more reliable token-level explanations. The fused explanation signals are further incorporated into model training through stability, boundary-aware, and consistency constraints. Results: Experiments on Chinese AD NER datasets show that the proposed framework improves explanation robustness and achieves consistent performance gains across multiple NER models. The adaptive fusion strategy also provides more stable explanations and stronger boundary perception than individual explanation methods. Conclusion: The proposed method effectively integrates reliable explanation signals into medical NER training, improving both recognition performance and explanation reliability. The framework provides a practical and generalizable solution for explainable medical NER and offers reliable support for downstream clinical decision-making and medical knowledge applications.
Chinese Translation
目的：本研究旨在通过解释引导学习提高中文特应性皮炎（AD）临床文本中医学命名实体识别（NER）的可靠性和鲁棒性。方法：我们提出了一种稳定性与边界意识的解释引导NER框架。采用基于扰动的分析方法评估解释的稳定性和实体边界的敏感性。自适应融合策略动态结合局部和全局解释，以生成更可靠的标记级解释。融合的解释信号进一步通过稳定性、边界意识和一致性约束纳入模型训练。结果：在中文AD NER数据集上的实验表明，所提出的框架提高了解释的鲁棒性，并在多个NER模型中实现了一致的性能提升。自适应融合策略还提供了比单一解释方法更稳定的解释和更强的边界感知。结论：所提方法有效地将可靠的解释信号整合到医学NER训练中，提升了识别性能和解释可靠性。该框架为可解释的医学NER提供了一个实用且具有普适性的解决方案，并为下游临床决策和医学知识应用提供了可靠支持。
cs.CL / 125 / 2606.22942
Understanding Knowledge Distillation in Post-Training: When It Helps and When It Fails
理解后训练中的知识蒸馏：何时有效，何时失效
Abstract
Large language models (LLMs) achieve strong performance across many tasks, but their high computational cost limits deployment in resource-constrained environments. Knowledge Distillation (KD) offers a practical solution by transferring knowledge from a teacher model of a larger size to a smaller student model. While prior work has mainly examined task-specific or small-scale settings, the post-training stage for building general instruction-following models has received limited attention. In this paper, we conduct a systematic study of KD in post-training using the large-scale Tulu 3 dataset. We find that KD outperforms supervised fine-tuning (SFT) in low-data regimes, but its advantage diminishes as more training data is added. Distilling from a stronger instruction-tuned teacher restores substantial gains even with abundant data, indicating that KD remains effective when the teacher provides knowledge that the student cannot easily acquire from the training data alone. We further study domain-specific, low-resource scenarios and propose a two-stage KD strategy that leverages synthetic teacher-labeled data followed by refinement on human annotations. This method consistently improves student performance, providing practical guidance for building compact models in data-scarce environments.
Chinese Translation
大型语言模型（LLMs）在许多任务中表现出色，但其高计算成本限制了在资源受限环境中的部署。知识蒸馏（Knowledge Distillation, KD）通过将来自更大规模教师模型的知识转移到较小的学生模型中，提供了一种实用的解决方案。尽管之前的研究主要集中在任务特定或小规模设置上，但用于构建通用指令遵循模型的后训练阶段却受到的关注有限。在本文中，我们使用大规模的 Tulu 3 数据集对后训练中的知识蒸馏进行了系统研究。我们发现，在低数据环境下，KD 的表现优于监督微调（Supervised Fine-Tuning, SFT），但随着训练数据的增加，其优势逐渐减弱。从更强的指令调优教师模型进行蒸馏，即使在数据丰富的情况下也能恢复显著的收益，这表明当教师提供学生无法仅从训练数据中轻易获得的知识时，KD 仍然有效。我们进一步研究了特定领域的低资源场景，并提出了一种两阶段的 KD 策略，该策略利用合成教师标注数据，随后在人工注释上进行精炼。这种方法持续提高学生模型的性能，为在数据稀缺环境中构建紧凑模型提供了实用指导。
cs.CL / 126 / 2606.22977
StatABench: Dataset and Framework for Evaluating Statistical Analysis Capabilities of LLMs
StatABench：评估大型语言模型统计分析能力的数据集与框架
Abstract
Statistical analysis is a broad, complex field requiring both domain knowledge and tool proficiency. While prior work has evaluated large language models (LLMs) in this domain, existing benchmarks remain limited in scope and format. To bridge this gap, we introduce StatABench (Statistical AnalysisBenchmark), a benchmark designed to systematically assess LLMs' statistical analysis capabilities. StatABench comprises two complementary components: Stat-Closed, containing 404 questions across 18 statistical topics in multiple formats (multiple-choice, fill-in-the-blank, decision-making, and practical application), and Stat-Open, featuring 30 complex open-ended modeling tasks adapted from professional competitions. We evaluate diverse LLMs using the LangChain MCP framework and multiple data science agents, and assess Stat-Open solutions via a validated LLM-as-Judge protocol. Experiments show that even GPT-5.1 achieves only 68.6% on Stat-Closed, while the best open-source model reaches 60.6%. On Stat-Open, the top agent framework scores 61.86 on average. These results reveal the gap between current LLMs and reliable statistical analysis, highlighting persistent challenges in tool-grounded reasoning, methodological decision-making, and end-to-end statistical modeling.
Chinese Translation
统计分析是一个广泛而复杂的领域，既需要领域知识又需要工具熟练度。尽管之前的研究已经评估了大型语言模型（LLMs）在这一领域的表现，但现有的基准测试在范围和格式上仍然有限。为了解决这一问题，我们提出了StatABench（统计分析基准），这是一个旨在系统评估LLMs统计分析能力的基准。StatABench包含两个互补的组成部分：Stat-Closed，包含18个统计主题下的404个问题，涵盖多种格式（选择题、填空题、决策制定和实际应用）；以及Stat-Open，包含30个复杂的开放式建模任务，这些任务改编自专业竞赛。我们使用LangChain MCP框架和多个数据科学代理评估不同的LLMs，并通过经过验证的LLM-as-Judge协议评估Stat-Open的解决方案。实验结果表明，即使是GPT-5.1在Stat-Closed上也仅获得68.6%的分数，而最佳的开源模型得分为60.6%。在Stat-Open上，顶级代理框架的平均得分为61.86。这些结果揭示了当前LLMs与可靠统计分析之间的差距，突显了在工具基础推理、方法决策和端到端统计建模方面持续存在的挑战。
cs.CL / 127 / 2606.22992
Predicate Importance Estimation and Decoupled Rationale-Score Distillation for Entity Alignment
谓词重要性估计与解耦的理由分数蒸馏用于实体对齐
Abstract
Knowledge graphs (KGs) are increasingly used as structured context for Large Language Models (LLMs), but industrial KG-RAG systems often need to integrate public and domain-specific KGs constructed from heterogeneous databases. This integration relies on Entity Alignment (EA), where lexical matching alone is insufficient under predicate-name variation and incomplete local neighborhoods. We address EA for KG integration by constructing a pairwise EA dataset and proposing two complementary modules: Predicate Importance Estimation (PIE) and Decoupled Rationale-Score Distillation (DRSD). PIE is a compact embedding-based approach that removes the subject information from each 1-hop triple, encodes the resulting subjectless triples, and aggregates them with learnable predicate-importance weights to build predicate-aware entity embeddings. DRSD trains a distilled small language model (SLM) with pseudo-answers produced by a teacher LLM through distinct prompts. By converting binary EA labels into text-based supervision and decoupling confidence-score estimation from label-consistent rationales, DRSD enables the SLM to learn task-specific reasoning while retaining a less label-biased confidence signal. Experiments show that PIE and DRSD improve EA classification. Moreover, because DRSD decouples confidence-score estimation from the decision, a discrepancy between the two flags an uncertain prediction for human review, thereby enabling a practical discrepancy between automatic acceptance and human-in-the-loop verification.
Chinese Translation
知识图谱（KGs）越来越多地被用作大型语言模型（LLMs）的结构化上下文，但工业KG-RAG系统通常需要整合来自异构数据库的公共和特定领域的知识图谱。这种整合依赖于实体对齐（EA），而单靠词汇匹配在谓词名称变异和局部邻域不完整的情况下是不够的。我们通过构建一个成对的EA数据集并提出两个互补模块来解决KG整合中的EA问题：谓词重要性估计（PIE）和解耦的理由分数蒸馏（DRSD）。PIE是一种基于紧凑嵌入的方法，它从每个1-hop三元组中去除主体信息，编码得到的无主体三元组，并通过可学习的谓词重要性权重对其进行聚合，以构建谓词感知的实体嵌入。DRSD通过不同的提示训练一个蒸馏的小型语言模型（SLM），使用由教师LLM生成的伪答案。通过将二元EA标签转换为基于文本的监督，并将置信度分数估计与标签一致的理由解耦，DRSD使SLM能够学习特定任务的推理，同时保留较少标签偏见的置信信号。实验表明，PIE和DRSD提高了EA分类的效果。此外，由于DRSD将置信度分数估计与决策解耦，二者之间的差异标志着不确定的预测，供人工审核，从而实现自动接受与人类审核之间的实际差异。
cs.CL / 128 / 2606.23002
Machine Translation and Post-Editing: Comparative Evaluation of Different MT Systems and Post-Editor Groups in Specialised Translation
机器翻译与后编辑：专业翻译中不同机器翻译系统与后编辑者群体的比较评估
Abstract
This article aims to evaluate the quality of machine translation (MT) and post-editing (PE) in the context of specialised translation from English into French. Three MT systems (DeepL, eTranslation and Systran) were compared, and two groups of post-editors -linguists/translators and NLP experts -were asked to perform post-editing. Translation assessment is based on error annotation using an error typology adapted to MT and PE evaluation. The results reveal significant differences between the three MT systems and the two groups of post-editors, particularly in terms of terminological accuracy and fluency. This study highlights the importance of domain knowledge in specialised translation, as well as the limitations and variable performance of MT systems in language for specific purposes (LSP).
Chinese Translation
本文旨在评估机器翻译（MT）和后编辑（PE）在英语到法语的专业翻译中的质量。比较了三种机器翻译系统（DeepL、eTranslation 和 Systran），并要求两组后编辑者——语言学家/翻译者和自然语言处理（NLP）专家——进行后编辑。翻译评估基于使用适应于机器翻译和后编辑评估的错误类型学进行的错误标注。结果显示，三种机器翻译系统和两组后编辑者之间存在显著差异，特别是在术语准确性和流畅性方面。本研究强调了领域知识在专业翻译中的重要性，以及机器翻译系统在特定目的语言（LSP）中的局限性和可变表现。
cs.CL / 129 / 2606.23030
Have You Ever Seen Them? Entity-level Membership Inference through Interrogating Large Language Models
你见过它们吗？通过询问大型语言模型进行实体级成员推断
Abstract
Large Language Models (LLMs) raise growing concerns about privacy leakage and copyright compliance. Membership inference is a key tool for assessing such risks, but existing studies mainly focus on whether specific samples or sample-based data units are used for training. We argue that LLMs exhibit a human-memory-like behavior: an LLM may not memorize a specific sample verbatim, yet it can accumulate and reveal knowledge about a real-world entity from scattered mentions. This analogy motivates us to examine whether an LLM can be interrogated like a human interviewee to reveal its exposure to entity-related information. Motivated by this question, we propose entity-level membership inference, which determines whether information related to a target entity is used in LLM training. We study this task in the practical label-only black-box setting, where only generated texts are observable. We formalize the task under clue, input, and model constraints, establish the necessary and sufficient conditions for its feasibility, and instantiate five interrogation strategies based on this formalization. The strategies use limited entity clues to construct prompts, elicit entity-related responses, and infer membership from semantic features among the generated texts. We construct entity-level datasets and adapt state-of-the-art sample-level label-only methods to the entity-level setting as baselines. Experiments on person entities show that our methods achieve AUC up to 0.97 and bring gains of 6.0%--17.5% in Balanced Accuracy over the best adapted baseline.
Chinese Translation
大型语言模型（LLMs）引发了对隐私泄露和版权合规性日益增长的关注。成员推断是评估此类风险的关键工具，但现有研究主要集中在特定样本或基于样本的数据单元是否被用于训练。我们认为，LLMs表现出类似人类记忆的行为：一个LLM可能不会逐字记忆特定样本，但它可以从零散的提及中积累并揭示关于现实世界实体的知识。这一类比促使我们探讨是否可以像对待人类受访者一样询问LLM，以揭示其对实体相关信息的暴露。受到这一问题的启发，我们提出了实体级成员推断，旨在确定与目标实体相关的信息是否被用于LLM的训练。我们在实际的仅标签黑箱设置中研究这一任务，其中仅可观察生成的文本。我们在线索、输入和模型约束下对任务进行了形式化，建立了其可行性的必要和充分条件，并基于这一形式化构建了五种询问策略。这些策略利用有限的实体线索构建提示，诱导与实体相关的响应，并从生成文本中的语义特征推断成员资格。我们构建了实体级数据集，并将最先进的样本级仅标签方法调整为实体级设置作为基线。对人实体的实验表明，我们的方法在AUC上达到0.97，并在最佳适应基线的平衡准确率上带来了6.0%至17.5%的提升。
cs.CL / 130 / 2606.23049
Training Open Models for Agentic Phone Use
为自主手机使用训练开放模型
Tang, Zhengyang, Lai, Xin, Lyu, Pengyuan, Wang, Xinyuan, Bai, Tianyi, Li, Chenxin, Guo, Yiduo, Shen, Huawen, Liu, Yuxuan, Li, Junyi, Fang, Zhengyao, Ding, Yang, Zhang, Yi, Wang, Weinong, Zhou, Xingran, Wu, Liang, Tang, Fei, Fan, Sunqi, Peng, Shangpin, Ruan, Zheng, Zhang, Anran, Wang, Benyou, Wen, Ji-Rong, Yan, Rui, Zhang, Chengquan, Hu, Han
Abstract
Phones are becoming an important execution surface for general-purpose agents, but training open models for reliable phone use remains difficult because the environment that matters at deployment, real devices running real apps, is slow, stateful, side-effectful, and hard to reset or verify, while scalable mock environments only approximate real behavior. We present PhoneBuddy, a training recipe and open-model line for agentic phone use that combines a real-app environment with a mock-app environment, PhoneWorld, which reconstructs runnable mock apps from real GUI usage structure. PhoneBuddy first builds a shared supervised fine-tuning stage from trajectories collected in both environments, then compares real-app RL against mixed RL across both environments. Across a 150-task human evaluation on real phones spanning apps, mini-apps, and cross-app workflows, task success rate improves from 36.67\% after supervised fine-tuning to 40.67\% after real-app RL and 45.33\% after mixed RL. On AndroidWorld, the same progression rises from 60.3\% to 77.2\% to 83.2\%. These results show that mock-app training is not a replacement for real-app RL, but a complementary source of scalable, resettable, and automatically checked interaction. The gains are strongest on app and mini-app tasks, while long-horizontal cross-app workflows remain an important open challenge.
Chinese Translation
手机正成为通用智能体的重要执行平台，但为可靠的手机使用训练开放模型仍然困难，因为在部署时重要的环境——真实设备运行真实应用——是缓慢的、有状态的、具有副作用的，并且难以重置或验证，而可扩展的模拟环境仅能近似真实行为。我们提出了PhoneBuddy，这是一种用于自主手机使用的训练方案和开放模型系列，它结合了真实应用环境和模拟应用环境PhoneWorld，该环境从真实的图形用户界面（GUI）使用结构重建可运行的模拟应用。PhoneBuddy首先从两个环境中收集的轨迹构建一个共享的监督微调阶段，然后比较真实应用的强化学习（RL）与两个环境中的混合RL。在涵盖应用、迷你应用和跨应用工作流的150项任务的人类评估中，任务成功率从监督微调后的36.67\%提高到真实应用RL后的40.67\%，再到混合RL后的45.33\%。在AndroidWorld中，相同的进展从60.3\%上升到77.2\%再到83.2\%。这些结果表明，模拟应用训练并不能替代真实应用的RL，而是可扩展的、可重置的和自动检查的交互的补充来源。增益在应用和迷你应用任务上最为显著，而长水平的跨应用工作流仍然是一个重要的开放挑战。
cs.CL / 131 / 2606.23092
PIVOTSBench: Evaluating Fine-Grained Interpersonal Relationship Reasoning in Multimodal Large Language Models
PIVOTSBench：评估多模态大型语言模型中的细粒度人际关系推理
Abstract
Humans possess an innate ability to understand fine-grained interpersonal relationships, which is central to everyday social interactions. Although such reasoning is inherently multimodal, it remains largely unexplored by existing multimodal large language models (MLLMs). To address this gap, we introduce PIVOTS, the first benchmark built from Social-IQ 2.0 and YouTube data to evaluate MLLMs' ability to predict bidirectional interpersonal relationship dimensions grounded in established psychology research. In addition, PIVOTS includes auxiliary tasks that assess models' ability to identify and leverage the critical visual cues underlying such predictions. We evaluate both proprietary and open-source MLLMs and conduct detailed ablation studies to analyze the effects of visual modalities and explicit social role information in conversational utterances. We further examine how joint and pairwise prediction settings benefit MLLMs in scoring bidirectional PIVOTS dimensions. Project page and resources: https://flynnzhangsx.github.io/PIVOTSBench/ .
Chinese Translation
人类具备理解细粒度人际关系的天赋，这对于日常社交互动至关重要。尽管这种推理本质上是多模态的，但现有的多模态大型语言模型（MLLMs）对此仍然探索不足。为了解决这一问题，我们引入了PIVOTS，这是第一个基于Social-IQ 2.0和YouTube数据构建的基准，用于评估MLLMs预测基于已建立心理学研究的双向人际关系维度的能力。此外，PIVOTS还包括辅助任务，以评估模型识别和利用这些预测背后关键视觉线索的能力。我们评估了专有和开源的MLLMs，并进行了详细的消融研究，以分析视觉模态和对话语句中显式社会角色信息的影响。我们进一步考察了联合和成对预测设置如何帮助MLLMs在评分双向PIVOTS维度时获益。项目页面和资源： https://flynnzhangsx.github.io/PIVOTSBench/ 。
cs.CL / 132 / 2606.23107
A Dual-Track Framework for Template-Constrained LaTeX Conversion
一种双轨框架用于模板约束的LaTeX转换
Abstract
With the increasing demands for advanced document conversion, mapping structured Markdown drafts into template-compliant formats like LaTeX remains a challenge. Existing approaches largely depend on either deterministic rule-based converters or pure end-to-end Large Language Model (LLM) generation. The former fails to correctly handle asset insertions and template-specific constraints, while the latter tends to induce semantic drift, leading to hallucinations that are difficult to debug. To address these limitations, we introduce a robust Dual-Track Framework that systematically decouples template formatting from document processing: an offline track extracts template constraints into a reusable manifest, while an online track implements a hybrid execution pipeline. This pipeline confines LLM usage exclusively to reasoning-intensive components (e.g., semantic metadata, bibliographic references, and complex visual/tabular layouts) while delegating rule-based engines for deterministic processing. Empirical evaluation across 7 LaTeX templates and 56 published research papers demonstrates that our method preserves better structural fidelity, satisfies diverse layout constraints, and achieves a higher compilation success rate compared to the previous baselines.
Chinese Translation
随着对高级文档转换需求的增加，将结构化的Markdown草稿映射到符合模板的格式（如LaTeX）仍然是一项挑战。现有的方法主要依赖于确定性的基于规则的转换器或纯端到端的大型语言模型（LLM）生成。前者无法正确处理资产插入和特定模板的约束，而后者则容易导致语义漂移，产生难以调试的幻觉。为了解决这些局限性，我们提出了一种强大的双轨框架，该框架系统地将模板格式化与文档处理解耦：离线轨道提取模板约束并生成可重用的清单，而在线轨道则实现混合执行管道。该管道将LLM的使用严格限制在推理密集型组件（例如，语义元数据、参考文献和复杂的视觉/表格布局）上，同时将确定性处理委托给基于规则的引擎。对7个LaTeX模板和56篇已发表研究论文的实证评估表明，我们的方法在结构保真度、满足多样化布局约束以及实现更高的编译成功率方面优于之前的基线。
cs.CL / 133 / 2606.23124
PRIDE: Privileged Information-enhanced Distillation for Empathetic Dialogue Generation
PRIDE：增强特权信息的知识蒸馏用于同理心对话生成
Abstract
Large language models have demonstrated significant capabilities in generating diverse and context-aware responses for empathetic dialogue. However, their computational demands severely limit their deployment in resource-constrained environments. While knowledge distillation offers a promising compression solution, it often fails to transfer the nuanced understanding essential for empathy, as it overlooks the implicit contextual cues that guide human connection. To bridge this gap, we propose a \textbf{pr}ivileged \textbf{i}nformation-enhanced knowledge \textbf{d}istillation method for \textbf{e}mpathetic dialogue generation (PRIDE). Our method leverages privileged information, such as expert psychological annotations or future event summaries, which is available exclusively during training but unavailable at inference time. This allows us to transfer the teacher model's empathetic reasoning to smaller models without relying on extra inputs during deployment. Specifically, PRIDE has three key components: (1) An empathy-reasoning prompt that guides the teacher to explicitly decompose the empathetic process into understanding feelings and analyzing situations step-by-step; (2) A multi-source attention mechanism that directs the student to effectively integrate privileged information; (3) A dual-alignment loss that combines reversed Kullback-Leibler divergence and maximum mean discrepancy to ensure robust knowledge transfer at both logit and feature levels. Experiments on multi-modal and text-only datasets demonstrate that our method achieves competitive performance, and in some cases matches or even surpasses larger teacher models in terms of accuracy and semantic relevance.
Chinese Translation
大型语言模型在生成多样化和上下文感知的同理心对话响应方面展现了显著的能力。然而，它们的计算需求严重限制了在资源受限环境中的部署。尽管知识蒸馏提供了一种有前景的压缩解决方案，但它常常无法传递对同理心至关重要的细微理解，因为它忽视了指导人际连接的隐含上下文线索。为了解决这一问题，我们提出了一种增强特权信息的知识蒸馏方法，用于同理心对话生成（PRIDE）。我们的方法利用特权信息，如专家心理注释或未来事件摘要，这些信息在训练期间独有，但在推理时不可用。这使我们能够在不依赖额外输入的情况下，将教师模型的同理心推理转移到较小的模型上。具体而言，PRIDE包含三个关键组件：（1）一个同理心推理提示，引导教师明确将同理心过程分解为逐步理解情感和分析情境；（2）一个多源注意机制，指导学生有效整合特权信息；（3）一个双重对齐损失，结合反向Kullback-Leibler散度和最大均值差异，以确保在logit和特征层面上稳健的知识转移。在多模态和仅文本数据集上的实验表明，我们的方法实现了具有竞争力的性能，在某些情况下在准确性和语义相关性方面与更大的教师模型相匹配甚至超越。
cs.CL / 134 / 2606.23164
Same question, different history: language, national identity, and credit in large language models
相同的问题，不同的历史：语言、国家认同与大型语言模型中的信用
Abstract
Who invented the radio, Russia's Alexander Popov or Italy's Guglielmo Marconi? Was the telephone the achievement of Bell in the United States or Meucci in Italy? Does printing belong to China's Bi Sheng or Germany's Gutenberg? The answer depends not only on historical record but also on language and perspective. We analyse eleven widely used large language models across 21 disputed inventions and discoveries, evaluated in twelve languages and 75,896 responses. While models generally acknowledge that credit is contested, query language systematically affects which claimant is surfaced. Lower-status claimants are more likely to appear when questions are asked in their associated language, whereas dominant Anglophone figures remain stable across languages. These patterns persist after controlling for response length, model differences, historical prominence, and levels of national commemoration. Language thus acts as a switch that activates different national versions of the same history, producing systematically different national memories from the same question. We interpret this as evidence that large language models function as distributed systems of cultural memory, where language conditions which histories become visible, contributing to a computational form of banal nationalism.
Chinese Translation
谁发明了无线电，是俄罗斯的亚历山大·波波夫（Alexander Popov）还是意大利的古列尔莫·马可尼（Guglielmo Marconi）？电话是美国的贝尔（Bell）还是意大利的梅乌奇（Meucci）的成就？印刷术属于中国的毕昇（Bi Sheng）还是德国的古滕贝格（Gutenberg）？答案不仅取决于历史记录，还与语言和视角有关。我们分析了十一种广泛使用的大型语言模型在21项有争议的发明和发现中的表现，评估了12种语言和75,896个响应。尽管模型通常承认信用存在争议，但查询语言系统性地影响了哪些索赔者被突出。当问题用其相关语言提出时，低地位的索赔者更可能出现，而主导的英语国家人物在不同语言中保持稳定。这些模式在控制响应长度、模型差异、历史显著性和国家纪念水平后依然存在。因此，语言充当了一个开关，激活了相同历史的不同国家版本，从而从相同的问题中产生系统性不同的国家记忆。我们将此解读为大型语言模型作为文化记忆的分布式系统的证据，其中语言决定了哪些历史变得可见，促进了一种计算形式的平凡民族主义。
cs.CL / 135 / 2606.23196
When Does Intrinsic Self-Correction Help? A Task-Sensitive Analysis
内在自我修正何时有效？一种任务敏感的分析
Abstract
Intrinsic self-correction (SC) aims to improve large language model outputs by prompting a model to revisit its own initial answer without external feedback. Recent studies have questioned the reliability of this approach, showing that models often struggle to judge whether their initial responses are correct. In this work, we take a task-sensitive view of SC. Rather than asking whether it works in general, we examine settings where SC may operate through different mechanisms: verifying explicit constraints, revisiting a complex reasoning process, or providing a second opinion over competing strategies in word-game tasks. Across multiple benchmarks and models, we find that SC can yield consistent performance gains when the underlying task structure facilitates these modes of revision. These results suggest that SC is best understood as a task-dependent inference-time strategy whose usefulness depends on the role the revision stage can play in a given task, rather than as a uniformly reliable method for improving initial model outputs.
Chinese Translation
内在自我修正（SC）旨在通过促使模型重新审视其初始答案而不依赖外部反馈来改善大型语言模型的输出。近期研究质疑了这种方法的可靠性，显示模型往往难以判断其初始响应是否正确。在本研究中，我们采取了任务敏感的视角来分析SC。我们并不是简单地询问其是否在一般情况下有效，而是考察SC可能通过不同机制运作的情境：验证显式约束、重新审视复杂推理过程，或在文字游戏任务中对竞争策略提供第二意见。在多个基准和模型中，我们发现当基础任务结构促进这些修订模式时，SC能够带来一致的性能提升。这些结果表明，SC最好被理解为一种依赖于任务的推理时策略，其有效性取决于修订阶段在特定任务中所能发挥的作用，而不是作为一种普遍可靠的方法来改善初始模型输出。
cs.CL / 136 / 2606.23217
MuPPET: A Benchmark for Contextual Privacy of LLM Assistants in Multi-Party Conversations
MuPPET：多方对话中大型语言模型助手的上下文隐私基准
Abstract
LLM agents are increasingly deployed in multi-party environments, handling sensitive personal data on behalf of individual users, for instance in group chats. When such an agent discloses private information, it reaches every group member at once. This risk is structurally harder to control than in one-to-one settings, as every piece of private information must be appropriate for every recipient in the group. Yet all existing contextual privacy benchmarks consider only single-interlocutor settings, leaving multi-party privacy risks unmeasured. We introduce MuPPET (Multi-Party Privacy Exposure Testing), a benchmark for contextual privacy in multi-party conversations. Our experiments show that models leak substantially more in multi-party settings than one-to-one evaluations suggest. Frontier models are vulnerable, and smaller open-weights models, often preferred for local deployment with sensitive data, even more so. Existing contextual privacy defences offer only partial protection, degrade utility, and do not resolve the underlying party-tracking problem.
Chinese Translation
大型语言模型（LLM）代理越来越多地在多方环境中部署，代表个别用户处理敏感个人数据，例如在群聊中。当这样的代理披露私人信息时，它会同时影响到每个群组成员。这种风险在结构上比一对一的环境更难以控制，因为每一条私人信息都必须适合群组中的每一个接收者。然而，现有的上下文隐私基准仅考虑单一对话者的设置，导致多方隐私风险未被衡量。我们引入了MuPPET（多方隐私曝光测试），这是一个针对多方对话的上下文隐私基准。我们的实验表明，在多方设置中，模型的泄露程度显著高于一对一评估所暗示的程度。前沿模型存在脆弱性，而较小的开放权重模型，通常因处理敏感数据而被优先用于本地部署，泄露风险更高。现有的上下文隐私防御措施仅提供部分保护，降低了实用性，并未解决根本的参与者追踪问题。
cs.CL / 137 / 2606.23233
Judgment-Grounded Expansion for Peer Review Generation
基于判断的扩展用于同行评审生成
Abstract
Automatic review generation is a promising direction for accelerating scientific progress. While most work adopts an end-to-end setup, its fully automated nature makes it less suitable for settings that demand accountability. To better balance automation and accountability, we formalize judgment-grounded expansion, a human-AI collaboration mode where a reviewer provides an evaluative claim and the system expands it into review comment candidate(s). We model it as a structured generate-check-refine process and conduct a user study to collect human-model interaction data. We study two practical challenges for judgment-grounded expansion: scalable evaluation and candidate set curation. We develop methods to simulate the process for large-scale evaluation, and show that conformal prediction is well suited to balancing candidate set size and target coverage. Our work establishes judgment-grounded expansion as a concrete task and provides empirical and methodological foundations for the design of future collaborative review generation systems.
Chinese Translation
自动评审生成是加速科学进步的一个有前景的方向。尽管大多数工作采用端到端的设置，但其完全自动化的特性使其不太适合需要问责的环境。为了更好地平衡自动化与问责，我们正式化了基于判断的扩展，这是一种人机协作模式，其中评审者提供评估性声明，系统将其扩展为评审评论候选项。我们将其建模为一个结构化的生成-检查-精炼过程，并进行用户研究以收集人机交互数据。我们研究了基于判断的扩展的两个实际挑战：可扩展评估和候选集策划。我们开发了模拟该过程的大规模评估方法，并展示了保守预测（conformal prediction）在平衡候选集大小和目标覆盖方面的适用性。我们的工作确立了基于判断的扩展作为一个具体任务，并为未来协作评审生成系统的设计提供了实证和方法论基础。
cs.CL / 138 / 2606.23271
Scaling LLM Knowledge Boundaries via Distribution-Optimized Synthesis
通过分布优化合成扩展大型语言模型的知识边界
Abstract
Knowledge injection via synthetic data is crucial for enhancing Large Language Models (LLMs). However, current synthesis methods simply stop at preset token counts or fixed data ratios, lacking awareness of knowledge distribution. This results in some domains being sparse while others are redundant, limiting LLM knowledge boundaries. We revisit knowledge injection from a distribution perspective and hypothesize that an optimal knowledge distribution exists to maximize knowledge boundary expansion. We propose KDoS (Knowledge Distribution-optimized Synthesis), a framework that introduces knowledge density to drive synthesis through a three-stage feedback mechanism, shifting from blind generation to distribution-optimized synthesis. We construct Wikipedia-based synthetic data with varying knowledge distributions and conduct experiments on models from 0.6B to 16B (Qwen, Ling, LLaMA) and data scales from 1B to 5B tokens. Our key findings are: (1) an optimal knowledge distribution consistently maximizes boundary expansion; (2) this distribution is stable across backbones and scales; (3) KDoS outperforms baselines across six knowledge benchmarks. Our work offers a new perspective and practical framework for synthetic data-driven knowledge injection.
Chinese Translation
通过合成数据进行知识注入对于增强大型语言模型（LLMs）至关重要。然而，当前的合成方法仅停留在预设的标记数量或固定的数据比例上，缺乏对知识分布的关注。这导致某些领域知识稀疏，而其他领域则冗余，从而限制了LLM的知识边界。我们从分布的角度重新审视知识注入，并假设存在一种最佳知识分布，以最大化知识边界的扩展。我们提出了KDoS（知识分布优化合成）框架，引入知识密度，通过三阶段反馈机制驱动合成，从盲目生成转向分布优化合成。我们构建了基于维基百科的合成数据，具有不同的知识分布，并对0.6B到16B（Qwen、Ling、LLaMA）模型及1B到5B标记的数据规模进行了实验。我们的主要发现是：（1）最佳知识分布始终最大化边界扩展；（2）这种分布在不同的基础模型和规模中是稳定的；（3）KDoS在六个知识基准测试中优于基线。我们的工作为基于合成数据的知识注入提供了新的视角和实用框架。
cs.CL / 139 / 2606.23283
Towards Root Memories: Benchmarking and Enhancing Implicit Logical Memory Retrieval for Personalized LLMs
迈向根记忆：基准测试与增强个性化大型语言模型的隐式逻辑记忆检索
Abstract
Memory systems are essential for personalized Large Language Models (LLMs). However, existing retrieval methods in these systems primarily rely on semantic similarity, potentially missing logically critical memories with limited semantic overlap. Current benchmarks remain inadequate for evaluating this problem. To address this gap, we construct IMLogic, the first high-quality benchmark targeting implicit logical memory retrieval in long-dialogue scenarios. Motivated by this challenge, we introduce root memory, a structured, decision-preserving representation that distills reusable personalized logic from long-term user histories. We then propose RootMem, a plug-and-play framework that first distills raw histories into structured root memories and then uses an LLM-based router to activate logically relevant ones, complementing semantic retrieval with personalized decision logic. Extensive experiments demonstrate that RootMem significantly outperforms the strongest retrieval baselines and consistently boosts the accuracy of existing memory agents. Our benchmark and codes will be available at https://anonymous.4open.science/r/IMLogic-DBB3.
Chinese Translation
记忆系统对于个性化大型语言模型（LLMs）至关重要。然而，这些系统中现有的检索方法主要依赖于语义相似性，可能会错过逻辑上关键的记忆，尽管它们的语义重叠有限。目前的基准测试对于评估这一问题仍显不足。为了解决这一空白，我们构建了IMLogic，这是第一个针对长对话场景中隐式逻辑记忆检索的高质量基准。受到这一挑战的启发，我们引入了根记忆（root memory），这是一种结构化的、保持决策的表示，提炼了来自长期用户历史的可重用个性化逻辑。然后，我们提出了RootMem，这是一个即插即用的框架，首先将原始历史提炼为结构化的根记忆，然后使用基于LLM的路由器激活逻辑相关的记忆，从而用个性化决策逻辑补充语义检索。大量实验表明，RootMem显著优于最强的检索基线，并持续提高现有记忆代理的准确性。我们的基准和代码将会在 https://anonymous.4open.science/r/IMLogic-DBB3 上提供。
cs.CL / 140 / 2606.23285
On the Effect of Segmentation Width and Cluster Size on Speech Resynthesis and Continuation in Generative Spoken Language Models
分割宽度和聚类大小对生成语音模型中的语音重合成和延续的影响
Abstract
Generative Spoken Language Modeling (GSLM) enables text-free speech modeling by training language models (LMs) using discrete speech representations instead of textual transcription. In this paper, we investigate the performance of GSLM on speech synthesis and continuation using discrete speech representations with varying bitrates. We segment speech representations with fixed widths and train K-means models in multiple cluster sizes, resulting in various bitrate settings. We demonstrate that intelligible and natural speech can be synthesized at lower bitrate settings than the baseline. Furthermore, speech continuation quality remains stable at lower bitrates across multiple metrics, suggesting that the conventional GSLM setting may be redundant for effective speech generation. Although LLM-based metrics show higher correlation with human subjective score than conventional metrics, it remains low, highlighting the need for more stable automatic evaluation methods.
Chinese Translation
生成语音语言建模（GSLM）通过使用离散语音表示而非文本转录来实现无文本的语音建模。在本文中，我们研究了GSLM在使用不同比特率的离散语音表示进行语音合成和延续时的性能。我们以固定宽度对语音表示进行分割，并在多个聚类大小中训练K均值模型，从而产生各种比特率设置。我们证明，在低于基线的比特率设置下，可以合成出可理解且自然的语音。此外，语音延续质量在多个指标上在较低比特率下保持稳定，这表明传统的GSLM设置可能对有效的语音生成是多余的。尽管基于大语言模型（LLM）的指标与人类主观评分的相关性高于传统指标，但相关性仍然较低，突显了对更稳定的自动评估方法的需求。
cs.CL / 141 / 2606.23306
The Anatomy of the CTC Oracle Gap: Acoustic Exhaustion and Linguistic Recovery
CTC Oracle Gap的解剖：声学疲劳与语言恢复
Abstract
We study the limits of CTC-internal scoring for N-best hypothesis selection and locate the information bottleneck separating acoustic confidence from linguistic plausibility. Eleven CTC-internal and acoustic-feature scoring strategies produce no statistically significant WER improvement over greedy decoding on LibriSpeech dev-other at G=16 (all p > 0.05). The exhaustion is systematic: CTC's Spearman $\rho$ between hypothesis score and per-utterance WER degrades from -0.574 at G=4 to -0.270 at G=128, a 53% loss driven by blank-path proliferation. This establishes that the discriminative capacity of CTC-internal representations is saturated: no recombination of acoustic signals can close the oracle gap. Confirming that the bottleneck is linguistic, not acoustic, external linguistic information introduced via MBR decoding breaks through it. MBR-CER decoding with a RoBERTa pseudo-log-likelihood (PLL) posterior ($\tau$=10, G=128) achieves 5.42% WER on held-out LibriSpeech test-other (greedy 5.96%, $\Delta$=-0.535 pp, p<0.0001, 9.0% relative). RoBERTa PLL $\rho$ degrades only 21% over the same range, retaining discriminating power where CTC loses it. Applied without retuning across two Zipformer architectures, three domains (LibriSpeech, TED-LIUM 3, VoxPopuli), and four MUSAN noise levels, the recipe gives significant gains in 11 of 13 conditions. On the training side, standard MWER training via the CTC forward-backward algorithm implements Rao-Blackwellized REINFORCE at the output projection (variance about 3x below Viterbi). Yet sequence-level fine-tuning fails at near-converged checkpoints: all four MWER configurations on CR-CTC collapse (+6.18 to +8.90 pp WER), as a training oracle gap of 0.007 pp provides no usable reward signal.
Chinese Translation
我们研究了CTC内部评分在N-best假设选择中的限制，并找到了将声学置信度与语言合理性分开的信息瓶颈。十一种CTC内部和声学特征评分策略在LibriSpeech dev-other上以G=16进行贪婪解码时未能显著改善字错误率（WER）（所有p > 0.05）。这种疲劳是系统性的：CTC的Spearman $
ho$在假设评分与每次发声的WER之间，从G=4的-0.574降至G=128的-0.270，损失幅度达到53%，主要由空路径的增殖驱动。这表明CTC内部表示的判别能力已达到饱和状态：没有声学信号的重组能够弥补oracle gap。确认瓶颈是语言性的，而非声学性的，通过MBR解码引入的外部语言信息突破了这一瓶颈。使用RoBERTa伪对数似然（PLL）后验的MBR-CER解码（$ au$=10, G=128）在保留的LibriSpeech test-other上达到了5.42%的WER（贪婪解码为5.96%，$ riangle$=-0.535 pp, p<0.0001，相对改善9.0%）。在相同范围内，RoBERTa PLL的$
ho$仅下降21%，在CTC失去判别能力的地方保持了判别能力。在两个Zipformer架构、三个领域（LibriSpeech、TED-LIUM 3、VoxPopuli）和四个MUSAN噪声级别下，未进行重新调优的应用在13种条件中的11种中取得了显著提升。在训练方面，通过CTC前向-后向算法的标准MWER训练在输出投影中实现了Rao-Blackwell化的REINFORCE（方差约为Viterbi的3倍以下）。然而，序列级微调在接近收敛的检查点处失败：CR-CTC上的所有四种MWER配置崩溃（+6.18至+8.90 pp WER），因为0.007 pp的训练oracle gap未能提供可用的奖励信号。
cs.CL / 142 / 2606.23321
Tmax: A simple recipe for terminal agents
Tmax：终端智能体的简单配方
Abstract
Terminal-using agents have quickly become the most popular downstream application of language models (LMs). Despite their prevalence, relatively little academic work has examined RL-based training of these models, likely due to difficult benchmarks, a lack of data, and a lack of simple baseline recipes. We present Tmax, the strongest open RL recipe for terminal agents to date, bringing open data recipes closer to the frontier. While simple, our recipe achieves 27\% on Terminal-Bench 2.0 with only 9B parameters, outperforming much larger models from prior work. Concretely, we generate data using a novel taxonomy, combining difficulty control, personas, and verifier diversification, which allows us to cheaply generate large amounts of terminal environments for RL and SFT training. We open-source our terminal dataset, which is over 2.5x larger than previously released terminal-agent datasets. We then train open-weight models using RL with our data, using a simple, outcome-only recipe. We release our data, models, and code as a strong baseline for future open academic work on terminal agents at https://github.com/hamishivi/tmax.
Chinese Translation
使用终端的智能体迅速成为语言模型（LMs）最受欢迎的下游应用。尽管它们的普及，关于这些模型的基于强化学习（RL）的训练的学术研究相对较少，这可能是由于基准测试的困难、数据的缺乏以及缺乏简单的基线配方。我们提出了Tmax，这是迄今为止最强大的开放强化学习配方，针对终端智能体，使开放数据配方更接近前沿。尽管简单，我们的配方在Terminal-Bench 2.0上达到了27\%的成绩，仅使用9B参数，超越了之前工作中更大规模的模型。具体而言，我们使用一种新颖的分类法生成数据，结合了难度控制、角色设定和验证者多样化，使我们能够廉价地生成大量的终端环境用于强化学习和监督微调（SFT）训练。我们开源了我们的终端数据集，其规模超过之前发布的终端智能体数据集的2.5倍。然后，我们使用我们的数据通过强化学习训练开放权重模型，采用简单的仅基于结果的配方。我们将数据、模型和代码发布为未来开放学术工作在终端智能体领域的强基线，网址为 https://github.com/hamishivi/tmax。
cs.CL / 143 / 2606.23336
WaveDetect: Robust Framework for Machine-Generated Text Detection via Wavelet Transform
WaveDetect：基于小波变换的机器生成文本检测的鲁棒框架
Abstract
As Large Language Models asymptotically approach human-level fluency in natural language generation, solely relying on surface-level semantic artifacts for detecting LLM-generated texts has become increasingly precarious. Existing detectors often falter when facing three critical challenges: adversarial perturbations, cross-domain shifts, and the rapid temporal evolution of the foundation model. To address these issues, we propose \wavedetect, a novel framework that reformulates text detection as a signal processing task within the time-frequency domain. Unlike previous methods that analyze static token probability distributions, \wavedetect models the generated output as a probability signal, upon which a differentiable Continuous Wavelet Transform is applied to convert them into learnable spectral representations. This process reveals the intrinsic ``spectral fingerprints'' in machine-generated texts--patterns that remain invisible in time domain. Comprehensive evaluations on three well-curated datasets (RAID, EvoBench, and Domain-Shift) show that our method achieves a new state-of-the-art. It not only achieves superior accuracy but also exhibits remarkable robustness against sophisticated attacks, generalization across out-of-distribution topics and unseen evolving LLMs. Our results validate the efficacy of spectral analysis as a promising paradigm for LLM-generated texts detection.
Chinese Translation
随着大型语言模型在自然语言生成中逐渐接近人类水平的流畅性，仅依赖表层语义特征来检测LLM生成的文本变得越来越不可靠。现有的检测器在面对三个关键挑战时往往表现不佳：对抗性扰动、跨领域转移以及基础模型的快速时间演变。为了解决这些问题，我们提出了 extit{WaveDetect}，一个将文本检测重新表述为时频域内信号处理任务的新框架。与以往分析静态标记概率分布的方法不同， extit{WaveDetect}将生成的输出建模为概率信号，并在其上应用可微分的连续小波变换，将其转换为可学习的谱表示。这个过程揭示了机器生成文本中的内在“谱指纹”——在时间域中不可见的模式。对三个精心策划的数据集（RAID、EvoBench和Domain-Shift）的全面评估表明，我们的方法达到了新的最先进水平。它不仅实现了卓越的准确性，还在复杂攻击、跨分布主题的泛化以及未见演变的LLM方面表现出显著的鲁棒性。我们的结果验证了谱分析作为检测LLM生成文本的有前景范式的有效性。
cs.CL / 144 / 2606.23375
Measuring & Mitigating Over-Alignment for LLMs in Multilingual Criminal Law Courts
在多语言刑法法庭中测量与缓解大型语言模型的过度对齐
Abstract
While the wider applicability of LLMs in the legal field is currently debated due to their reliability and the gravity of any errors, narrow uses with well-understood and mitigated risks have emerged. Notably the Swiss Federal Supreme Court uses small on-premises models for tentative translations and short-passage summarization across the four official languages. However, such usage is challenging in the context of Criminal Law. Since rulings and cases employees work on routinely can contain detailed descriptions of violent and sexual offenses, their legitimate work is compromised by refusals and disclaimers due to the activation of model guardrails (over-alignment). To measure this phenomenon, we introduce TF-RefusalBench, a multilingual benchmark for criminal-law translation and summarization derived from public Swiss Supreme Court rulings. TF-RefusalBench contains 5,200 total prompts across French, German, Italian, and English, corresponding to common task prompts and passages likely to trigger refusal. We then use TF-RefusalBench to show that over-alignment is a multifaceted phenomenon, influenced by the model and the prompt and text languages being processed, and that its impact cannot be evaluated solely from an over-refusal perspective, given the disclaimer's impact on task faithfulness. Finally, we evaluate approaches to enable on-premises LLMs for Criminal Law Tasks, demonstrating that while prompting can be effective, abliteration (refusal directions ablation) eliminates refusal with minimal impact on task performance.
Chinese Translation
尽管由于大型语言模型（LLMs）的可靠性及其错误的严重性，其在法律领域的广泛适用性目前仍存在争议，但在风险明确且已得到缓解的狭窄应用中，已出现了一些实际案例。值得注意的是，瑞士联邦最高法院使用小型本地模型进行四种官方语言的初步翻译和短段落摘要。然而，在刑法的背景下，这种使用面临挑战。由于裁决和案件中员工日常处理的内容可能包含对暴力和性犯罪的详细描述，因此由于模型保护措施（过度对齐）的激活，员工的合法工作受到拒绝和免责声明的影响。为了测量这一现象，我们引入了TF-RefusalBench，这是一个基于瑞士最高法院公开裁决的多语言刑法翻译和摘要基准。TF-RefusalBench包含5200个总提示，涵盖法语、德语、意大利语和英语，针对常见任务提示和可能触发拒绝的段落。我们随后使用TF-RefusalBench展示过度对齐是一个多方面的现象，受模型及处理的提示和文本语言的影响，其影响不能仅从过度拒绝的角度进行评估，因为免责声明对任务忠实度的影响。最后，我们评估了使本地大型语言模型适用于刑法任务的方法，证明虽然提示可能有效，但消除拒绝方向（abliteration）可以在对任务性能影响最小的情况下消除拒绝。
cs.CL / 145 / 2606.23382
Energy-Based Transformers as Predictors of Reading Difficulty
基于能量的变换器作为阅读难度的预测工具
Abstract
Transformer language models have become established tools for modeling human sentence processing, with measures such as surprisal and attention entropy serving as effective predictors of reading difficulty that together capture complementary aspects of processing load. Here, we explore a related class of transformer models: energy-based transformers, which provide a principled formal link to associative memory models, bringing processing research into direct contact with the broader literature on Hopfield networks and dense associative memory. To our knowledge, this is the first exploration of an energy-based transformer measure in computational psycholinguistics. Across reading-time corpora (Natural Stories, UCL eye-tracking, UCL self-paced reading), the energy measure is a robust predictor of reading times, providing significant fit beyond surprisal in all three. In a controlled experiment on relative clause processing, energy at a single layer captures the well-known object/subject asymmetry. We find evidence that it subsumes effects attributable to both attention entropy and surprisal, suggesting that energy may serve as a single unified predictor where multiple complementary measures have previously been required.
Chinese Translation
变换器语言模型已成为建模人类句子处理的成熟工具，诸如惊讶度（surprisal）和注意力熵（attention entropy）等指标有效地预测了阅读难度，并共同捕捉了处理负荷的互补方面。在此，我们探讨了一类相关的变换器模型：基于能量的变换器（energy-based transformers），它们为联想记忆模型提供了一个原则性的形式联系，将处理研究与霍普菲尔德网络（Hopfield networks）和密集联想记忆（dense associative memory）等更广泛文献直接联系起来。根据我们的了解，这是在计算心理语言学中首次探讨基于能量的变换器测量。在阅读时间语料库（Natural Stories、UCL眼动追踪、UCL自定进度阅读）中，能量测量是阅读时间的稳健预测指标，在所有三者中提供了超越惊讶度的显著拟合。在对相对从句处理的控制实验中，单层的能量捕捉到了众所周知的宾语/主语不对称性。我们发现证据表明，它包含了可归因于注意力熵和惊讶度的效应，暗示能量可能作为一个统一的预测指标，而在之前需要多个互补的测量。
cs.CL / 146 / 2606.23387
Self-Stigma Is Not a Monolith, but Generic Empathy Is: Persona-Conditioned LLM Support for People Who Use Drugs
自我污名并非单一体，但通用同理心是：基于角色的 LLM 支持药物使用者
Abstract
Self-stigma predicts treatment avoidance and disengagement among people who use drugs (PWUD), yet conversational systems aiming to provide support typically treat self-stigma expression as a uniform signal. We present a three-phase, proof-of-concept study of a persona-aware approach to LLM support. Latent Profile Analysis (LPA) on indicator-level features from 1,174 self-stigma expressors on Reddit yields a four-persona typology validated against held-out behavioral and linguistic features. Sequential Bayesian and recurrent neural classifiers recover these personas from limited posting histories, substantially outperforming batch and few-shot LLM baselines (macro-F1 = 0.74 at 30 posts). Evaluation by eight clinical experts across three contemporary LLMs revealed a misalignment: persona-matched responses successfully achieved targeted behavioral shifts, yet raters holistically preferred the generic empathy of the persona-neutral baseline. Our findings suggest that holistic empathy judgments and clinically-aligned response design can pull in opposite directions, and that evaluating LLM-based stigma support requires rubrics capable of decomposing the two.
Chinese Translation
自我污名预测药物使用者（PWUD）的治疗回避和脱离，然而旨在提供支持的对话系统通常将自我污名的表达视为统一信号。我们提出了一项三阶段的概念验证研究，采用基于角色的 LLM 支持。对来自 1,174 名 Reddit 自我污名表达者的指标级特征进行潜在剖面分析（LPA），得出一种经过验证的四种角色类型学，基于保留的行为和语言特征。顺序贝叶斯和递归神经分类器从有限的发帖历史中恢复这些角色，显著优于批量和少量样本的 LLM 基线（宏观 F1 = 0.74，基于 30 条帖子）。三位当代 LLM 的八位临床专家的评估揭示了一种不一致性：与角色匹配的回应成功实现了目标行为转变，但评审者整体上更偏好于角色中立基线的通用同理心。我们的研究结果表明，整体同理心判断与临床对齐的回应设计可能朝着相反的方向发展，并且评估基于 LLM 的污名支持需要能够分解这两者的标准。
cs.CL / 147 / 2606.23394
Do LLM Embedding Spaces Recover Expert Structure?
大规模语言模型嵌入空间能否恢复专家结构？
Abstract
Pretrained text embeddings are increasingly used as representational maps, yet high category separability does not imply that their geometry recovers expert-defined structure. We study this problem in mental-health-related language, where symptom relations provide an external reference and online communities introduce strong domain, affective, stylistic, and discourse confounds. Using 28 Reddit communities, we compare pretrained and supervised fine-tuned Qwen3 embedding spaces at two scales (0.6B and 4B). We construct category prototypes, evaluate their representational dissimilarity matrices against an expert symptom matrix with representational similarity analysis, and complement this global test with prototype-based typicality and multi-baseline confound controls. Pretrained embeddings show measurable alignment with expert structure within the mental-health subset; fine-tuning strengthens this alignment most at the finest category level; and larger scale improves both zero-shot alignment and supervision-induced gains. Residual alignment remains substantial after controlling for VAD, LIWC, lexical style, and topic-distribution structure. These results suggest that LLM embeddings can recover expert-relevant category geometry, but this recovery is level-dependent and should be tested against explicit confounds rather than inferred from classification alone.
Chinese Translation
预训练文本嵌入越来越多地被用作表征映射，然而，高类别可分离性并不意味着它们的几何结构能够恢复专家定义的结构。我们在与心理健康相关的语言中研究这个问题，其中症状关系提供了外部参考，而在线社区则引入了强烈的领域、情感、风格和话语混淆。我们使用28个Reddit社区，比较了预训练和监督微调的Qwen3嵌入空间在两个规模（0.6B和4B）上的表现。我们构建了类别原型，利用表征相似性分析评估其表征不相似矩阵与专家症状矩阵的对比，并通过基于原型的典型性和多基线混淆控制补充这一全球测试。预训练嵌入在心理健康子集内显示出与专家结构的可测量对齐；微调在最细类别层面上增强了这种对齐；而更大的规模则改善了零样本对齐和监督引导的增益。在控制VAD、LIWC、词汇风格和主题分布结构后，残余对齐仍然显著。这些结果表明，大规模语言模型嵌入可以恢复与专家相关的类别几何结构，但这种恢复是层次依赖的，应针对显式混淆进行测试，而不是仅仅从分类中推断。
cs.CL / 148 / 2606.23404
ReasoningLens: Hierarchical Visualization and Diagnostic Auditing for Large Reasoning Models
ReasoningLens：大型推理模型的层次可视化与诊断审计
Abstract
The emergence of Large Reasoning Models has introduced exceptionally long Chain-of-Thought traces, creating a transparency burden where critical logic is often buried under massive procedural text. To address this, we present ReasoningLens, an open-source framework designed for the hierarchical visualization and diagnostic auditing of complex reasoning chains. ReasoningLens addresses information necropsy by: (1) structuring traces into interactive hierarchies that separate high-level strategy from low-level execution; (2) leveraging an agentic auditor for automated error detection and tool-augmented verification; and (3) synthesizing systemic reasoning profiles to reveal model-specific blind spots. By transforming unstructured walls of text into actionable insights, ReasoningLens provides a modular foundation for interpreting, debugging, and optimizing the next generation of reasoning-centric AI.
Chinese Translation
大型推理模型的出现引入了异常长的思维链追踪，造成了透明性负担，关键逻辑常常埋没在大量的程序性文本之下。为了解决这一问题，我们提出了ReasoningLens，这是一个旨在对复杂推理链进行层次可视化和诊断审计的开源框架。ReasoningLens通过以下方式解决信息解剖问题：(1) 将追踪结构化为交互式层次，区分高层策略与低层执行；(2) 利用代理审计员进行自动错误检测和工具增强验证；(3) 综合系统推理概况以揭示模型特定的盲点。通过将非结构化的文本墙转化为可操作的见解，ReasoningLens为解释、调试和优化下一代以推理为中心的人工智能提供了模块化基础。
cs.CL / 149 / 2606.23412
UnBias-Plus: Detect, Explain, and Rewrite Bias
UnBias-Plus：检测、解释和重写偏见
Abstract
Bias in natural language remains a persistent challenge in both human-written and AI-generated content, affecting domains such as journalism, education, and AI research. Most existing detection methods identify only the presence of bias, with limited support for granular detection, interpretable explanations, neutral rewriting, and openly available trained models. We present UnBias-Plus, an open-source toolkit unifying (1) segment-level multi-class bias classification, (2) biased span localization, (3) neutral text rewriting, and (4) reasoning for each decision. Available via Python, CLI, REST API, and web interfaces, UnBias-Plus supports accessible bias analysis. The toolkit, source code, models, datasets, and documentation are publicly available.
Chinese Translation
自然语言中的偏见在人工撰写和AI生成的内容中仍然是一个持续的挑战，影响着新闻、教育和AI研究等领域。现有的大多数检测方法仅识别偏见的存在，缺乏对细粒度检测、可解释性解释、中立重写以及公开可用的训练模型的支持。我们提出了UnBias-Plus，一个开源工具包，统一了（1）段落级多类偏见分类，（2）偏见跨度定位，（3）中立文本重写，以及（4）每个决策的推理。UnBias-Plus通过Python、CLI、REST API和Web界面提供，支持便捷的偏见分析。该工具包、源代码、模型、数据集和文档均可公开获取。
cs.CL / 150 / 2606.23459
TriggerBench: Investigating Prospective Memory for Large Language Models
TriggerBench：探讨大型语言模型的前瞻性记忆
Abstract
While Large Language Models (LLMs) are increasingly deployed in long interactions, existing evaluations focus predominantly on retrospective memory (RM) via explicit queries. Prospective memory (PM), the critical ability to spontaneously recall and act on latent constraints without direct prompts, remains largely unevaluated. We introduce TriggerBench, a comprehensive PM benchmark spanning five dimensions across both daily assistants and professional workflows. TriggerBench pairs scenarios with matched RM controls, contrastive positive/negative variants, and overloaded triggers, enabling fine-grained measurement of proactive recall, false-alarm rate, and attentional robustness under a single protocol. Our evaluation yields three key findings. (i) PM shows a precision-recall trade-off and attentional fragility. Though enhanced reasoning significantly improves proactive recall, models may overfit to an "always-remind" heuristic. Furthermore, PM accuracy degrades substantially under implicit constraints or triggers overloaded by concurrent user requests, indicating that robust PM remains an open challenge. (ii) PM is notably harder than RM: on identical contexts, RM near-saturates up to 100K tokens, while PM decays sharply as context length scales. (iii) PM may serve as a behavioral probe of spare reasoning capacity. Pairing PM scenarios with AIME-2025 math problems reveals that successful trajectories yield higher PM accuracy than failed ones at the same context length, showing PM tracks spare reasoning budget that token count obscures. Project page: https://github.com/KristenZHANG/TriggerBench-Official.
Chinese Translation
随着大型语言模型（LLMs）在长时间交互中的应用越来越广泛，现有评估主要集中在通过显式查询进行的回顾性记忆（RM）。而前瞻性记忆（PM），即在没有直接提示的情况下自发回忆和执行潜在约束的关键能力，仍然未得到充分评估。我们推出了TriggerBench，这是一个涵盖日常助手和专业工作流程的全面PM基准，跨越五个维度。TriggerBench将场景与匹配的RM控制、对比的正/负变体以及过载触发器配对，从而在单一协议下实现对主动回忆、误报率和注意力稳健性的精细测量。我们的评估得出了三个关键发现。（i）PM展现出精确度-召回率的权衡和注意力脆弱性。尽管增强的推理显著提高了主动回忆，但模型可能会过度拟合于“始终提醒”的启发式。此外，在隐性约束或被并发用户请求过载的触发器下，PM的准确性显著下降，这表明稳健的PM仍然是一个未解决的挑战。（ii）PM显著比RM更困难：在相同的上下文中，RM在达到100K个标记时几乎饱和，而PM在上下文长度增加时急剧衰减。（iii）PM可能作为稀余推理能力的行为探针。将PM场景与AIME-2025数学问题配对显示，成功的轨迹在相同上下文长度下产生的PM准确性高于失败的轨迹，表明PM跟踪了被标记数量掩盖的稀余推理预算。项目页面：https://github.com/KristenZHANG/TriggerBench-Official。
cs.CL / 151 / 2606.23462
War in the Abstract: The Rise and Consequences of Militarized Language in Scientific Communication
抽象中的战争：军事化语言在科学交流中的崛起及其后果
Abstract
Scientists do not, by profession, wage war. Yet warfare's vocabulary consistently appears in their abstracts. To quantify the extent to which warfare's vocabulary pervades scientific abstracts, we analyze 21.4 million papers (2010-2025; OpenAlex, PubMed). We additionally run a within-subject war-framing experiment (N = 801; 32,040 trials) designed to provide causal insight into the effects of militaristic language on persuasion. Between 2010 and 2025, the presence of militaristic terms in scientific abstracts rose 48% in OpenAlex and 32% in PubMed, with the rise accelerating sharply after 2019 (cross-database r = 0.96, p < 10^-8). The prevalence of militaristic language is conflict-aligned at both country and annual scales (Uppsala Conflict Data Program; r = 0.77-0.84), with the abstracts from the Global South displaying the fastest rise in militaristic language. Among disciplines, social sciences leads in level of such language while engineering and computer science lead in growth. The COVID and post-2022 large-language-model eras also saw the rise and narrowed the language gap between native-English and non-English authors. In our follow-up experiment, we found that war framing reduced credibility (mean shift -0.18 Likert units, 95% CI [-0.21, -0.14]; d_z = -0.28, p < 10^-20), funding willingness (d_z = -0.12) and policy support (d_z = -0.08), with a trend-level increase in sense of urgency (d_z = +0.07). Collectively, findings reveal that while scientific abstracts drift toward warfare, the use of militaristic language may erode credibility, funding willingness, and policy support.
Chinese Translation
科学家并不以战争为职业。然而，战争的词汇在他们的摘要中却不断出现。为了量化战争词汇在科学摘要中的普遍程度，我们分析了2140万篇论文（2010-2025；OpenAlex，PubMed）。此外，我们还进行了一个内部实验，设计了一个战争框架实验（N = 801；32,040次试验），旨在提供军事化语言对说服力影响的因果洞察。在2010年至2025年期间，科学摘要中军事化术语的出现率在OpenAlex中上升了48%，在PubMed中上升了32%，并且在2019年后增长速度急剧加快（跨数据库相关系数 r = 0.96，p < 10^-8）。军事化语言的普遍性在国家和年度层面上与冲突相关（乌普萨拉冲突数据计划；r = 0.77-0.84），来自全球南方的摘要在军事化语言的使用上显示出最快的增长。在各学科中，社会科学在此类语言的使用水平上领先，而工程学和计算机科学在增长速度上领先。COVID大流行及2022年后的大型语言模型时代也见证了军事化语言的上升，并缩小了母语英语与非英语作者之间的语言差距。在我们的后续实验中，我们发现战争框架降低了可信度（平均变化 -0.18 Likert单位，95%置信区间[-0.21, -0.14]；d_z = -0.28，p < 10^-20），资金意愿（d_z = -0.12）和政策支持（d_z = -0.08），同时在紧迫感上有趋势性增加（d_z = +0.07）。总体而言，研究结果表明，尽管科学摘要逐渐倾向于军事化，使用军事化语言可能会侵蚀可信度、资金意愿和政策支持。
cs.CL / 152 / 2606.23525
Self-Compacting Language Model Agents
自我压缩语言模型代理
Abstract
Long agent traces composed of chains of thought and tool calls accumulate stale content that anchor subsequent generations, and eventually outgrow the context window. Existing scaffolds mitigate it with fixed-interval compaction triggered at a token threshold. Such triggers pay no heed to trajectory structure, risking discard of partial results mid-derivation or mid-search. We propose SelfCompact, a scaffold that allows the model itself to decide when and how to compact. Specifically, it pairs two inference-time elements: (i) a compaction tool the model invokes to summarize the accumulated context, and (ii) a lightweight rubric specifying when to fire (a sub-task has resolved, or the trajectory is converging) and when to suppress (mid-derivation, or when stuck). Both are needed. The tool alone is unevenly used across open-weight models, often invoked at unhelpful moments or not at all; the rubric alone cannot act. Together, they elicit effective adaptive compaction without any fine-tuning or external supervision. We present empirical results on six benchmarks (competitive math and agentic search) and seven models. Our results show that SelfCompact matches or exceeds fixed-interval summarization at a fraction of the token cost, improving over a no-summarization baseline by up to 18.1 points on math and 5-9 points on agentic search at 30-70% lower per-question cost. Our results expose a meta-cognitive gap: although unprompted models cannot reliably tell when their own context is rotting, a lightweight rubric closes this gap, reframing when to compact as a capability that scaffolds can supply without training.
Chinese Translation
由思维链和工具调用组成的长代理轨迹积累了过时的内容，这些内容会影响后续生成，并最终超出上下文窗口。现有的支架通过在令牌阈值触发的固定间隔压缩来缓解这一问题。这种触发机制并未考虑轨迹结构，可能会在推导或搜索过程中丢弃部分结果。我们提出了SelfCompact，这是一种允许模型自主决定何时以及如何进行压缩的支架。具体而言，它结合了两个推理时的元素：(i) 模型调用的压缩工具，用于总结累积的上下文；(ii) 一种轻量级的标准，指定何时触发（子任务已解决，或轨迹正在收敛）以及何时抑制（推导中，或陷入困境时）。这两者都是必需的。仅使用工具在开放权重模型中效果不均，常常在不合适的时刻被调用或根本不被调用；而仅有标准则无法起作用。两者结合能够有效引导自适应压缩，而无需任何微调或外部监督。我们在六个基准（竞争性数学和代理搜索）和七个模型上进行了实证结果展示。我们的结果表明，SelfCompact在令牌成本的极小部分下，与固定间隔总结相匹配或超越，在数学上提高了高达18.1分，在代理搜索上提高了5-9分，且每个问题的成本降低了30-70%。我们的结果揭示了一个元认知差距：尽管未提示的模型无法可靠判断自身上下文何时变得无效，但轻量级标准弥补了这一差距，将何时进行压缩重新定义为支架可以在不训练的情况下提供的能力。
cs.CL / 153 / 2606.23566
LangMAP: A Language-Adaptive Approach to Tokenization
LangMAP：一种语言自适应的分词方法
Abstract
Language-specific tokenizers improve tokenization quality and the downstream performance of models on those languages. However, using such a tokenizer comes at a cost: either a new model must be trained from scratch, or the vocabulary of an existing pretrained model must be adapted. We propose Language-adaptive Maximum a Posteriori (LangMAP) Tokenization, a tokenization scheme that extends the UnigramLM algorithm to the multilingual setting, producing language-specific tokenization from a single shared vocabulary. Notably, LangMAP can be used when training a multilingual language model from scratch or to adapt a pretrained model's tokenizer to individual languages without changing its vocabulary. While language labels are required at training time, a key feature of the algorithm is that it then performs language-specific tokenization at inference without knowledge of the input's language. Across 14 open-source tokenizers, 9 natural languages, and 9 programming languages, LangMAP improves morphological boundary alignment and, for all coding languages tested, alignment with abstract syntax tree (AST) leaf boundaries. In fine-tuning experiments, results are mixed: LangMAP improves target-language grammatical acceptability (MultiBLiMP) on the languages tested; its benefits are less consistent on knowledge-related tasks (Global-PIQA, Belebele).
Chinese Translation
特定语言的分词器可以提高分词质量以及模型在这些语言上的下游性能。然而，使用这样的分词器是有代价的：要么必须从头开始训练一个新模型，要么必须调整现有预训练模型的词汇表。我们提出了语言自适应最大后验（LangMAP）分词方案，该方案将UnigramLM算法扩展到多语言环境，从单一共享词汇表中生成特定语言的分词。值得注意的是，LangMAP可以在从头训练多语言模型时使用，或者在不更改词汇表的情况下，将预训练模型的分词器适应于个别语言。虽然在训练时需要语言标签，但该算法的一个关键特性是它在推理时能够在不知道输入语言的情况下执行特定语言的分词。在14个开源分词器、9种自然语言和9种编程语言的实验中，LangMAP改善了形态边界对齐，并且在所有测试的编程语言中，与抽象语法树（AST）叶子边界的对齐也得到了改善。在微调实验中，结果呈现出混合的情况：LangMAP提高了测试语言的目标语言语法可接受性（MultiBLiMP）；在知识相关任务（Global-PIQA, Belebele）上的益处则不那么一致。
cs.CL / 154 / 2606.23583
Evaluation Awareness Is Not One Capability: Evidence from Open Language Models
评估意识并非单一能力：来自开放语言模型的证据
Abstract
Safety benchmarks assume that test-condition behavior predicts deployment behavior, an assumption that fails if models detect evaluation cues and adapt. This opens a gap between benchmark performance and deployment behavior: compliance measured under test conditions becomes an optimistic upper bound that overstates how safely a model behaves once the evaluation harness is removed. We characterize this evaluation awareness through eight experiments across 37 open-weight models and seven families. (i)Detection is moderate and training-driven (24/37 models exceed chance, best AUROC 0.714 vs.0.819 human, with instruction tuning dominating over scale). (ii)Detection shifts safety behavior (hard refusal drops 5.8 percentage points under hypothetical framing, and 21/140 HarmBench framing effects are significant, with compliance rising up to +30 percentage points. (iii)Representations survive behavioral collapse (probes retain AUROC 0.98 under rewrites that drive behavior below chance, and multi-layer steering causally moves three downstream tasks while random controls do not). (iv)These axes are weakly coupled (only 1/15 correlations are significant, the sole robust link being behavioral detection versus framing resistance, $\rho=-0.79$, $p<0.001$). We call this gap the benchmark illusion: because detectability, behavioral manifestation, and controllability vary independently, it is multivariate rather than a single number, so no single awareness score is a reliable proxy for deployment safety.
Chinese Translation
安全基准假设测试条件下的行为可以预测部署行为，但如果模型能够检测评估线索并进行适应，这一假设就会失效。这在基准性能与部署行为之间打开了一个差距：在测试条件下测量的合规性成为一个乐观的上限，夸大了模型在移除评估环境后表现的安全性。我们通过对37个开放权重模型和七个模型家族进行的八个实验来表征这种评估意识。(i) 检测能力适中且受训练驱动（24/37个模型超过随机水平，最佳AUROC为0.714，而人类为0.819，指令调优优于规模）。(ii) 检测影响安全行为（在假设框架下，强拒绝率下降了5.8个百分点，140个HarmBench框架效应中有21个显著，合规性最高上升了30个百分点）。(iii) 表示在行为崩溃中依然存在（探测器在驱动行为低于随机水平的重写下保持AUROC为0.98，多层引导因果性地推动了三个下游任务，而随机控制则没有）。(iv) 这些维度之间的耦合较弱（15个相关性中仅有1个显著，唯一稳健的联系是行为检测与框架抵抗之间，$
ho=-0.79$, $p<0.001$）。我们将这一差距称为基准幻觉：由于可检测性、行为表现和可控性独立变化，它是多变量的，而不是单一数字，因此没有单一的意识评分可以可靠地作为部署安全性的代理。
cs.CL / 155 / 2606.23654
EnterpriseClawBench: Benchmarking Agents from Real Workplace Sessions
EnterpriseClawBench：来自真实工作场景的代理基准测试
Abstract
Enterprise agents increasingly operate inside workspaces: they read heterogeneous files, invoke tools, and deliver business artifacts. We introduce EnterpriseClawBench, an enterprise agent benchmark constructed from proprietary, real-world agent sessions. Starting from a large archive of workplace sessions, the EnterpriseClawBench produces 852 reproducible tasks, each paired with recovered fixtures, rewritten prompts, role classes, skill subclasses, hard rules, and semantic rubrics. Because the sessions contain internal enterprise content, we do not release the benchmark data; instead, our reusable contribution is the construction and evaluation protocol. On EnterpriseClawBench, the best configuration reaches only 0.663 (Codex with GPT-5.5). These results show that enterprise agent evaluation must report harness--model combinations, artifact delivery, visual quality, cost, runtime, and skill-transfer behavior, rather than collapsing performance into a single score. Code: https://github.com/FrontisAI/EnterpriseClawBench
Chinese Translation
企业代理越来越多地在工作空间内操作：它们读取异构文件，调用工具，并交付业务文档。我们介绍了EnterpriseClawBench，这是一个基于专有的真实世界代理会话构建的企业代理基准测试。EnterpriseClawBench从大量工作场景档案中生成852个可重复的任务，每个任务都配有恢复的固定装置、重写的提示、角色类别、技能子类别、硬性规则和语义评分标准。由于这些会话包含内部企业内容，我们不发布基准数据；相反，我们可重用的贡献是构建和评估协议。在EnterpriseClawBench上，最佳配置的得分仅为0.663（Codex与GPT-5.5）。这些结果表明，企业代理评估必须报告工具-模型组合、文档交付、视觉质量、成本、运行时间和技能转移行为，而不是将性能简化为单一得分。代码链接：https://github.com/FrontisAI/EnterpriseClawBench
cs.CL / 156 / 2606.23671
Can LLMs Reliably Self-Report Adversarial Prefills, and How?
大型语言模型能否可靠地自我报告对抗性预填充攻击，以及如何实现？
Abstract
Prior work shows that large language models (LLMs) exhibit introspective capability on benign tasks. We extend the question to safety contexts and examine how reliably a model can recognize that its own prior response was elicited by an adversarial prefill attack. Across ten open-weight instruction-tuned LLMs (3B to 70B) and four safety benchmarks, no model reliably recognizes its own compromised outputs, with models claiming intent on prefilled responses at an average rate of $27.3\%$. Introspective signal stems largely from safety- and refusal-related reasoning. Orthogonalizing models' weights against the refusal direction collapses the gap between claiming rates on prefilled and natural outputs to near zero, though the direction is not its unique mediator. The signal is also probe-dependent: framing the question as internal intention versus external tampering elicits qualitatively different responses on the same models. We test three LoRA finetuning methods (SFT, GRPO, DPO) on eight models from 3B to 27B; all three widen the intention-probe gap on every model from 8B to 27B, with method ranking varying by model. The intervention does not transfer to the tampering probe and counterintuitively raises attack success rate under adversarial prefill on most models, amounting to a partial mitigation. These findings outline mechanisms underpinning the observed introspective signals in safety contexts and highlight risks in the reliability of LLM self-reports.
Chinese Translation
先前的研究表明，大型语言模型（LLMs）在良性任务上表现出内省能力。我们将问题扩展到安全背景，考察模型在多大程度上能够可靠地识别其自身先前响应是由对抗性预填充攻击引发的。在十个开放权重的指令调优LLMs（3B到70B）和四个安全基准测试中，没有模型能够可靠地识别其自身被损害的输出，模型在预填充响应上声称意图的平均比例为27.3%。内省信号主要源于与安全和拒绝相关的推理。将模型的权重与拒绝方向正交化，使得在预填充和自然输出上的声称率差距几乎降至零，尽管该方向并不是其唯一的中介。该信号还依赖于探测方式：将问题框架化为内部意图与外部干扰会在相同模型上引发质的不同响应。我们在八个模型（从3B到27B）上测试了三种LoRA微调方法（SFT、GRPO、DPO）；所有三种方法在每个8B到27B的模型上都扩大了意图探测差距，且方法排名因模型而异。该干预未能转移到干扰探测上，并且在大多数模型上反直觉地提高了对抗性预填充下的攻击成功率，形成部分缓解。这些发现勾勒出在安全背景下观察到的内省信号的机制，并强调了LLM自我报告可靠性中的风险。
cs.CL / 157 / 2606.23687
Randomized YaRN Improves Length Generalization for Long-Context Reasoning
随机化 YaRN 改善长上下文推理的长度泛化能力
Abstract
Large language models (LLMs) are typically pretrained on short sequences and then extended to work on longer sequences with additional training. However, such LLMs still struggle to further generalize to very long sequences. We propose Randomized YaRN, a training method that improves length generalization by combining YaRN-based positional extrapolation with randomized positional encoding and a length curriculum. During training on short context data, tokens are assigned YaRN positional encodings sampled from a larger position range, exposing the model to out-of-distribution positional representations even on short-context inputs. We evaluate Randomized YaRN on two challenging long-context reasoning benchmarks, BABILong and Multi-Round Coreference Resolution (MRCR). When training on data with <8K context, Randomized YaRN consistently improves reasoning performance on context lengths from 16K to 128K and outperforms standard fine-tuning, with the largest gains appearing at far out-of-distribution lengths. Our results suggest that progressively exposing models to OOD positional distributions provides an effective recipe for generalizable long-context reasoning.
Chinese Translation
大型语言模型（LLMs）通常是在短序列上进行预训练，然后通过额外的训练扩展到处理更长的序列。然而，这些 LLMs 在进一步泛化到非常长的序列时仍然面临挑战。我们提出了随机化 YaRN（Randomized YaRN），这是一种通过将基于 YaRN 的位置外推与随机化位置编码和长度课程相结合来改善长度泛化的训练方法。在短上下文数据的训练过程中，令牌被分配从更大位置范围中采样的 YaRN 位置编码，使模型即使在短上下文输入上也能接触到分布外（out-of-distribution, OOD）的位置表示。我们在两个具有挑战性的长上下文推理基准上评估了随机化 YaRN，分别是 BABILong 和多轮共指解析（Multi-Round Coreference Resolution, MRCR）。当在上下文长度小于 8K 的数据上进行训练时，随机化 YaRN 一直在 16K 到 128K 的上下文长度上持续提高推理性能，并且优于标准微调，最大的提升出现在远离分布的长度上。我们的结果表明，逐步让模型接触 OOD 位置分布为可泛化的长上下文推理提供了一种有效的方法。