cs.RO / 1 / 2606.30680
Locker-based Truck-Drone Routing with Integrated Considerations of Pickups, Deliveries, and No-Fly Zones
基于储物柜的卡车-无人机路线规划：综合考虑取件、送件和禁飞区
Abstract
Truck-drone delivery is an emerging last-mile logistics mode combining the long-haul capacity of trucks with the flexible service capability of drones. In locker-based operations, smart lockers serve not only as temporary parcel storage facilities but also as automated drone docking and service nodes. These automated nodes support drone takeoff, landing, parcel handover, and battery replacement, thereby significantly extending the service range and operational flexibility of drone-assisted delivery networks. However, practical locker-based delivery systems face complex real-world challenges, requiring the integrated coordination of not only parcel delivery, return pickup, battery-constrained and load-dependent drone flights, but also necessary detours around restricted airspace. To address this practical and multifaceted challenge, this paper introduces a locker-based truck-drone routing problem with integrated considerations of pickups, deliveries, and no-fly zones (LTDRP-PDNF), with the objective of minimizing the total operational cost of a fleet of drone-equipped trucks. We formulate the route construction process as a Markov Decision Process and develop a two-stage deep reinforcement learning-based neural heuristic. The first stage utilizes an attention-based encoder and a Bidirectional Gated Recurrent Unit decoder to solve the truck-only routing problem, formulated as a capacitated vehicle routing problem. The second stage combines a policy-transfer strategy with a hybrid dispatch assignment heuristic to construct fully coordinated truck and drone routes for LTDRP-PDNF. Experiments on instances of different scales demonstrate that the proposed method outperforms metaheuristic and neural heuristic baselines in most cases while maintaining exceptionally short computation times, offering an effective, scalable solution framework under practical operational constraints.
Chinese Translation
卡车-无人机配送是一种新兴的最后一公里物流模式，结合了卡车的长途运输能力和无人机的灵活服务能力。在基于储物柜的操作中，智能储物柜不仅作为临时包裹存储设施，还作为自动化无人机停靠和服务节点。这些自动化节点支持无人机的起飞、着陆、包裹交接和电池更换，从而显著扩展了无人机辅助配送网络的服务范围和操作灵活性。然而，实际的基于储物柜的配送系统面临复杂的现实挑战，需要综合协调包裹配送、退件取件、电池受限和负载依赖的无人机飞行，以及绕过限制空域的必要绕行。为了解决这一实际且多方面的挑战，本文提出了一种基于储物柜的卡车-无人机路线规划问题，综合考虑取件、送件和禁飞区（LTDRP-PDNF），旨在最小化配备无人机的卡车车队的总运营成本。我们将路线构建过程形式化为马尔可夫决策过程，并开发了一种基于深度强化学习的两阶段神经启发式算法。第一阶段利用基于注意力的编码器和双向门控递归单元解码器来解决仅卡车的路线规划问题，该问题被形式化为有容量限制的车辆路线问题。第二阶段结合政策转移策略和混合调度分配启发式方法，为LTDRP-PDNF构建完全协调的卡车和无人机路线。不同规模实例的实验表明，所提出的方法在大多数情况下优于元启发式和神经启发式基线，同时保持极短的计算时间，为在实际操作约束下提供有效、可扩展的解决框架。
cs.RO / 2 / 2606.30686
Position: Vision-Language-Action Models Cannot Be Verified to Perform Physical Reasoning
立场：视觉-语言-动作模型无法被验证以执行物理推理
Abstract
Vision-Language-Action (VLA) systems, built on pretrained vision-language models (VLMs), have shown rapidly improving performance on robot manipulation benchmarks. These gains are commonly interpreted as evidence that semantic representations learned from internet-scale data transfer to physical execution generalization. This position paper argues that the assumption underlying this interpretation -- that semantic generalization is sufficient to support physical action decisions -- has not been independently verified and cannot be tested under current evaluation protocols. We support this claim by decomposing VLA policies into semantic mapping and physical action decision, and showing that task success rate -- the dominant evaluation metric -- cannot distinguish between these two sources of capability. As a result, improvements in benchmark performance are consistent with multiple competing explanations, including semantic matching, distributional overlap, and genuine physical generalization. We further argue that this identifiability gap has been reinforced through narrative drift, whereby successive systems inherit and strengthen prior interpretations of performance gains without isolating the underlying causal mechanism. To address this limitation, we propose a research direction based on evaluation designs that introduce controlled variation to separately measure semantic and physical generalization. Such designs make it possible to causally attribute performance without requiring access to model internals, and to empirically assess the role of VLM backbones as semantic interfaces rather than implicit sources of physical competence. Our goal is not to refute the role of VLMs in robotics, but to clarify the conditions under which claims of physical generalization can be meaningfully evaluated.
Chinese Translation
视觉-语言-动作（VLA）系统基于预训练的视觉-语言模型（VLMs），在机器人操作基准测试中表现出快速提升的性能。这些提升通常被解读为从互联网规模数据中学习的语义表示能够转移到物理执行的泛化能力。本文立场论文认为，这种解读所依据的假设——即语义泛化足以支持物理行动决策——尚未得到独立验证，并且在当前的评估协议下无法进行测试。我们通过将VLA策略分解为语义映射和物理行动决策来支持这一主张，并表明任务成功率——这一主导评估指标——无法区分这两种能力来源。因此，基准性能的提升与多种竞争性解释是一致的，包括语义匹配、分布重叠和真正的物理泛化。我们进一步认为，这种可识别性差距通过叙事漂移得到了加强，即后续系统继承并强化了先前对性能提升的解释，而没有隔离潜在的因果机制。为了解决这一局限性，我们提出了一种基于评估设计的研究方向，该设计引入了受控变异，以分别测量语义和物理泛化。这种设计使得在不需要访问模型内部的情况下，能够因果归因于性能，并实证评估VLM骨干作为语义接口而非隐含物理能力来源的角色。我们的目标并不是否定VLM在机器人领域的作用，而是澄清在何种条件下物理泛化的主张可以被有意义地评估。
cs.RO / 3 / 2606.30694
DSIP: A Dynamic Coordination Planner for Signal-Free Intersections using Diffusion-Model-Based Multi-Agent Motion Planning
DSIP：基于扩散模型的无信号交叉口动态协调规划器
Abstract
Traffic signal control at urban intersections inherently introduces stop-and-go behavior, resulting in increased delays and reduced traffic efficiency, especially under high traffic demand. With the emergence of connected and automated vehicles (CAVs), trajectory-level coordination has emerged as a high-potential strategy to augment or transcend conventional phase-based management. This paper proposes DSIP (Diffusion-model-based Signal-free Intersection Planner), a multi-agent motion planning framework driven by a generative diffusion process. DSIP shifts the intersection management paradigm from discrete temporal phasing to continuous multi-vehicle trajectory optimization. This work evaluates the theoretical upper-bound performance of this coordination strategy under idealized communication and execution conditions to isolate the core benefits of the diffusion-driven approach. Using the SUMO platform, we evaluate DSIP across diverse four-leg intersection configurations. Experimental results demonstrate that DSIP significantly reduces average delay and maintains higher average speed compared to both fixed-time signal control and state-of-the-art reinforcement-learning-based controllers, particularly in medium- to high-density traffic. These findings suggest that diffusion-based trajectory planning provides a scalable and robust foundation for future autonomous intersection management. By unlocking latent intersection capacity through software-defined coordination, this approach offers a cost-effective pathway to improve urban traffic flow efficiency without requiring physical infrastructure expansion.
Chinese Translation
城市交叉口的交通信号控制固有地引入了停走行为，导致延误增加和交通效率降低，尤其是在高交通需求下。随着联网和自动驾驶车辆（CAVs）的出现，轨迹级协调作为一种高潜力策略，逐渐成为增强或超越传统相位管理的有效方法。本文提出了DSIP（基于扩散模型的无信号交叉口规划器），这是一个由生成扩散过程驱动的多智能体运动规划框架。DSIP将交叉口管理范式从离散的时间相位转变为连续的多车辆轨迹优化。本研究评估了在理想化通信和执行条件下该协调策略的理论上限性能，以隔离扩散驱动方法的核心优势。通过使用SUMO平台，我们在多种四向交叉口配置中评估了DSIP。实验结果表明，与固定时间信号控制和最先进的基于强化学习的控制器相比，DSIP显著降低了平均延误并保持了更高的平均速度，尤其是在中到高密度交通情况下。这些发现表明，基于扩散的轨迹规划为未来的自主交叉口管理提供了可扩展且稳健的基础。通过软件定义的协调释放潜在的交叉口容量，该方法提供了一条经济有效的途径，以提高城市交通流效率，而无需进行物理基础设施扩展。
cs.RO / 4 / 2606.30696
ViTL: Temporal Logic-Guided Zero-Shot Natural Language Navigation via Vision-Language Models
ViTL：基于时序逻辑指导的零样本自然语言导航通过视觉-语言模型
Abstract
Enabling robots to follow natural language commands to complete zero-shot long-horizon tasks remains challenging. It requires extracting implicit temporal and logical constraints from natural language commands and executing multiple sub-tasks accordingly. Recent zero-shot object navigation methods use vision-language models (VLMs) to guide frontier-based exploration in unknown environments, but they are limited to single-target tasks. Real-world commands such as "Clean either the chair or the couch, then turn on the tv." require navigating to multiple targets in a temporally constrained order, which no existing zero-shot system can handle. We present ViTL, a framework that addresses this gap at two levels. At the task level, we use a large language model (LLM) to compile natural language commands into Linear Temporal Logic (LTL) formulas, which are then converted into Deterministic Finite Automata~(DFA) that coordinate multi-channel value maps and trigger dynamic replanning when new objects are detected. At the navigation level, we introduce directional score: rather than producing a direction-agnostic value across the entire field of view, we label frontier directions on the observation image and extract per-direction scores from the VLM. Experiments on Habitat-Matterport 3D (HM3D) show that the full framework enables zero-shot long-horizon completion of natural language navigation tasks with temporal constraints, and that directional score improves single-target navigation accuracy and efficiency over the baseline.
Chinese Translation
使机器人能够遵循自然语言命令完成零样本长时间跨度任务仍然具有挑战性。这需要从自然语言命令中提取隐含的时序和逻辑约束，并相应地执行多个子任务。最近的零样本物体导航方法利用视觉-语言模型（VLM）指导在未知环境中的前沿探索，但它们仅限于单目标任务。现实世界中的命令，例如“清理椅子或沙发，然后打开电视。”需要按照时间约束的顺序导航到多个目标，而现有的零样本系统无法处理此类任务。我们提出了ViTL，一个在两个层面上解决这一问题的框架。在任务层面，我们使用大型语言模型（LLM）将自然语言命令编译成线性时序逻辑（LTL）公式，然后将其转换为确定性有限自动机（DFA），以协调多通道值映射，并在检测到新对象时触发动态重规划。在导航层面，我们引入了方向评分：与其在整个视野中产生一个方向无关的值，我们在观察图像上标记前沿方向，并从VLM中提取每个方向的评分。在Habitat-Matterport 3D（HM3D）上的实验表明，完整框架能够实现具有时间约束的自然语言导航任务的零样本长时间跨度完成，并且方向评分提高了单目标导航的准确性和效率。
cs.RO / 5 / 2606.30698
Vision-Language Procedural Reasoning for Context-Aware Reward Modeling of Robotic Endovascular Guidewire Navigation
面向上下文感知奖励建模的视觉-语言程序推理在机器人血管内导丝导航中的应用
Abstract
Robotic-assisted endovascular interventions demand accurate, stable, and context-aware guidewire navigation in complex and patient-specific vascular anatomies. Despite recent advances in robotic precision and learning-based control, existing autonomous navigation methods remain limited by their reliance on static reward functions and the lack of explicit procedural reasoning regarding anatomical context and task progression. To address these challenges, this paper proposes a vision-language procedural reasoning (VL-PR) framework for autonomous guidewire navigation. The framework integrates a multimodal large language model (MLLM) as a procedural reasoning module that interprets real-time visual observations to infer high-level navigation contexts. Instead of generating low-level control commands, the inferred procedural insights enable context-aware reward adaptation by dynamically adjusting the importance of reward components across different navigation phases. This approach allows a single policy to resolve competing objectives and handle complex transitions while preserving a consistent global task goal. Experiments on a physical robotic platform across diverse vascular scenarios demonstrate enhanced task reliability and streamlined navigational efficiency, highlighting the advantages over static-reward methods and offering a scalable solution for complex and multi-task robotic endovascular procedures.
Chinese Translation
机器人辅助的血管内干预需要在复杂且特定于患者的血管解剖结构中进行准确、稳定和上下文感知的导丝导航。尽管近年来在机器人精度和基于学习的控制方面取得了进展，现有的自主导航方法仍然受到静态奖励函数的限制，并缺乏对解剖上下文和任务进展的明确程序推理。为了解决这些挑战，本文提出了一种视觉-语言程序推理（Vision-Language Procedural Reasoning, VL-PR）框架，用于自主导丝导航。该框架将多模态大型语言模型（Multimodal Large Language Model, MLLM）作为程序推理模块，能够解释实时视觉观察，以推断高层次的导航上下文。通过动态调整不同导航阶段奖励组件的重要性，推断出的程序洞察使得奖励适应具有上下文感知，而不是生成低层次的控制命令。这种方法允许单一策略解决相互竞争的目标并处理复杂的过渡，同时保持一致的全局任务目标。在多种血管场景下的物理机器人平台实验中，展示了任务可靠性和导航效率的提升，突显了相较于静态奖励方法的优势，并为复杂和多任务的机器人血管内程序提供了一种可扩展的解决方案。
cs.RO / 6 / 2606.30749
From Grasps to Dexterity: Large-Scale Grasp Pretraining for Dexterous Manipulation
从抓取到灵巧性：大规模抓取预训练用于灵巧操作
Abstract
Large-scale dexterous grasp datasets encode rich priors over hand-object interaction, but their use has largely been confined to grasp generation and pick-and-place manipulation. We study whether such data can instead support functional dexterity in articulated tool use, where a robot must acquire a tool, maintain contact, and operate its functional moving parts. We adapt a hierarchical imitation learning framework that combines high-level hand sub-goal prediction with a low-level goal-conditioned controller. We construct a 355k-trajectory grasp-pretraining dataset from large-scale dexterous grasp annotations and use it to pretrain the low-level controller. The controller is then fine-tuned on downstream task demonstrations. To evaluate this setting, we introduce DexCraft, a simulation benchmark with six articulated tool-use tasks requiring coordinated finger motion. Across simulation and real-world experiments, our approach outperforms end-to-end diffusion policy baselines and hierarchical policies trained from scratch. In the real world, it improves full-task success by 33.3 percentage points over DP3. These results show that grasp datasets can serve not only as resources for grasp synthesis, but also as scalable pretraining data for contact-rich dexterous manipulation. Videos are shown on https://yingyuan0414.github.io/grasp2dexterity/ .
Chinese Translation
大规模灵巧抓取数据集编码了丰富的手-物体交互先验，但其应用主要局限于抓取生成和拾取放置操作。我们研究这些数据是否可以支持关节工具使用中的功能性灵巧性，在这种情况下，机器人必须获取工具、保持接触并操作其功能性运动部件。我们调整了一个层次模仿学习框架，该框架结合了高层次的手部子目标预测和低层次的目标条件控制器。我们从大规模灵巧抓取注释中构建了一个包含35.5万轨迹的抓取预训练数据集，并利用该数据集对低层次控制器进行预训练。然后，该控制器在下游任务演示上进行微调。为了评估这一设置，我们引入了DexCraft，一个包含六个需要协调手指运动的关节工具使用任务的仿真基准。在仿真和现实世界实验中，我们的方法优于端到端扩散策略基线和从零开始训练的层次策略。在现实世界中，它使全任务成功率比DP3提高了33.3个百分点。这些结果表明，抓取数据集不仅可以作为抓取合成的资源，还可以作为用于接触丰富的灵巧操作的可扩展预训练数据。相关视频可在 https://yingyuan0414.github.io/grasp2dexterity/ 查看。
cs.RO / 7 / 2606.30804
Wind and State Estimation on SE(3): Comparative Evaluation of EKF and UKF with Continuous and Discrete Quadrotor Models
SE(3)上的风速和状态估计：EKF与UKF在连续和离散四旋翼模型中的比较评估
Abstract
Use of quadrotor UAVs for wind velocity estimation is gaining popularity in recent studies, leveraging their maneuverability, compact size and low cost. Among available approaches, model-based wind velocity estimation is most commonly used, since it relies only on onboard sensors. However, as the quadrotor is a highly nonlinear system, thus making this task challenging. This study evaluate the use of both discrete and continuous dynamic equations of the quadrotor UAV for wind velocity estimation on SE(3), rather than commonly adapted continuous or discretized form. Lie Group Variational Integrator, developed on discrete Lagrangian is used as the discrete model without any approximation or discritization. The study assess both the discrete and continuous form of the quadrotor dynamics on SE(3) using Extended Kalman filter (EKF), and Unscented Kalman filter (UKF). The quadrotor UAV performance is evaluated in both MATLAB-based numerical simulations and free outdoor flight. The numerical simulations are conducted during both hovering and trajectory-tracking flights. Results demonstrate that, by using discrete SE(3) dynamics coupled with UKF, the quadrotor achieves higher estimation accuracy while maintaining trajectory tracking, even with low-cost sensors. These findings highlight the potential of discrete quadrotor models with UKF not only for wind velocity estimation but also for other high-accuracy tasks, even when relying on low-cost onboard sensors.
Chinese Translation
近年来，四旋翼无人机用于风速估计的研究逐渐受到关注，主要得益于其机动性、紧凑的体积和低成本。在现有的方法中，基于模型的风速估计最为常用，因为它仅依赖于机载传感器。然而，由于四旋翼是一种高度非线性的系统，这使得该任务具有挑战性。本研究评估了四旋翼无人机在SE(3)上的风速估计中使用离散和连续动态方程，而不是通常采用的连续或离散形式。基于离散拉格朗日的李群变分积分器被用作离散模型，未进行任何近似或离散化。研究使用扩展卡尔曼滤波器（EKF）和无迹卡尔曼滤波器（UKF）评估四旋翼动态在SE(3)上的离散和连续形式。四旋翼无人机的性能在基于MATLAB的数值仿真和户外自由飞行中进行了评估。数值仿真在悬停和轨迹跟踪飞行中进行。结果表明，通过使用与UKF结合的离散SE(3)动态，四旋翼在保持轨迹跟踪的同时实现了更高的估计精度，即使在使用低成本传感器的情况下。这些发现突显了离散四旋翼模型与UKF结合的潜力，不仅适用于风速估计，还适用于其他高精度任务，即便依赖于低成本的机载传感器。
cs.RO / 8 / 2606.30807
Off the Rails: Hijacking the Scoring Head in Generative End-to-End Driving Planners with Safety-Violating Adversarial Perturbations
脱轨：利用安全违规的对抗扰动劫持生成端到端驾驶规划器中的评分头
Abstract
Generative models have recently seen rapid adoption in End-to-End (E2E) autonomous driving (AD), with diffusion-based denoising and vocabulary-based retrieval becoming the dominant trajectory-decoding paradigms. Despite their architectural diversity, current generative AD planners share a common inference pattern: a fixed set of candidate trajectories (anchors, vocabulary entries, or proposal queries) is scored by one or more learned heads conditioned on the Bird's-Eye-View (BEV) features, and the highest-scored candidate is returned as the final trajectory. Under this design, the scoring head is the only barrier between perception and the motion command, and its decision margins between competing candidates are often small. We introduce \textsc{Derail}, an adversarial framework that exploits this scoring-head attack surface. Evaluated on various generative planners, \textsc{Derail} flips the trajectory selection from a safe to an unsafe candidate, with score drops of $39$--$80\%$ and collision rates of up to $50\%$, consistently outperforming generic loss-maximization and feature-divergence attacks. Our analysis suggests that safety-violating objectives govern attack effectiveness against generative AD planners, and that the scoring-head inference pattern itself is a recurring attack surface worth explicit defensive consideration.
Chinese Translation
生成模型最近在端到端（E2E）自动驾驶（AD）中得到了快速应用，以基于扩散的去噪和基于词汇的检索成为主导的轨迹解码范式。尽管它们在架构上各具多样性，当前的生成AD规划器共享一个共同的推理模式：一组固定的候选轨迹（锚点、词汇条目或提议查询）由一个或多个学习的评分头根据鸟瞰图（BEV）特征进行评分，得分最高的候选轨迹被返回作为最终轨迹。在这种设计下，评分头是感知与运动指令之间的唯一障碍，而其在竞争候选之间的决策边际往往较小。我们引入了 extsc{Derail}，一个利用这一评分头攻击面对抗的框架。在各种生成规划器上的评估表明， extsc{Derail}将轨迹选择从安全候选翻转为不安全候选，得分下降幅度达到39%至80%，碰撞率高达50%，并且始终优于通用的损失最大化和特征差异攻击。我们的分析表明，违反安全的目标主导了对生成AD规划器的攻击有效性，而评分头的推理模式本身是一个值得明确防御考虑的反复出现的攻击面。
cs.RO / 9 / 2606.30817
TAPE: Tether-Aware Path Planning for Autonomous Exploration of Unknown 3D Cavities Using a Tangle-Compatible Tethered Aerial Robot
TAPE：一种考虑缆绳的自主探索未知三维空腔的路径规划方法，适用于缠绕兼容的缆绳无人机
Abstract
This letter presents the first method for autonomous exploration of unknown cavities in three dimensions (3D) that focuses on minimizing the distance traveled and the length of tether unwound. Considering that the tether entanglements are little influenced by the global path, our approach employs a 2-level hierarchical architecture. The global frontier-based planning solves a Traveling Salesman Problem (TSP) to minimize the distance. The local planning attempts to minimize the path cost and the tether length using an adjustable decision function whose parameters play on the trade-off between these two values. The proposed method, TAPE, is evaluated through detailed simulation studies as well as field tests. On average, our method generates a 4.1% increase in distance traveled compared to the TSP solution without our local planner, with which the length of the tether remains below the maximum allowed value in 53% of the simulated cases against 100% with our method.
Chinese Translation
本文提出了一种针对未知三维空腔的自主探索方法，旨在最小化行驶距离和缆绳展开长度。考虑到缆绳缠绕对全局路径的影响较小，我们的方法采用了两级层次架构。基于全局前沿的规划解决了旅行推销员问题（TSP），以最小化距离。局部规划则通过可调决策函数来最小化路径成本和缆绳长度，其参数在这两个值之间进行权衡。所提出的方法TAPE通过详细的仿真研究和实地测试进行了评估。与未使用局部规划器的TSP解决方案相比，我们的方法平均增加了4.1%的行驶距离，而在53%的仿真案例中，缆绳长度保持在最大允许值以下，而使用我们的方法则为100%。
cs.RO / 10 / 2606.30820
Robustness-Based Synthesis for Time Window Temporal Logic Specifications via Mixed-Integer Linear Programming
基于鲁棒性的时间窗口时序逻辑规范合成方法：混合整数线性规划
Abstract
Time Window Temporal Logic (TWTL) is a rich specification language for cyber-physical systems that can compactly express sequential tasks with explicit timing constraints. In this paper, we consider the problem of synthesizing control inputs for discrete-time linear systems subject to TWTL task specifications. Building on the quantitative semantics (robustness) recently introduced for TWTL in [1], we encode the robust satisfaction of a TWTL formula as a set of Mixed-Integer Linear constraints and pose synthesis as a Mixed Integer Linear Program (MILP) that maximizes the robustness degree. We prove that any feasible solution with positive objective value guarantees Boolean satisfaction of the specification. We address two synthesis settings: an \emph{open-loop} formulation that optimizes the full control sequence from the initial state, and a \emph{closed-loop} receding-horizon Model Predictive Controller (MPC) formulation that re-solves the MILP at each step using the current measured state. A key feature of our MPC formulation is a \emph{task-adaptive horizon} that exploits the TWTL Deterministic Finite Automaton (DFA) to determine the active sub-task at each step, limiting the prediction horizon to the remaining window of the current task rather than the full formula horizon, this makes each re-solve significantly cheaper than the initial open-loop solve.
Chinese Translation
时间窗口时序逻辑（TWTL）是一种丰富的规范语言，适用于网络物理系统，能够紧凑地表达具有明确时间约束的顺序任务。本文考虑为满足TWTL任务规范的离散时间线性系统合成控制输入的问题。在最近为TWTL引入的定量语义（鲁棒性）基础上，我们将TWTL公式的鲁棒满足性编码为一组混合整数线性约束，并将合成问题表述为一个最大化鲁棒度的混合整数线性规划（MILP）。我们证明，任何具有正目标值的可行解都能保证规范的布尔满足性。我们讨论了两种合成设置：一种是 extit{开环}形式，优化从初始状态开始的完整控制序列；另一种是 extit{闭环}递归预测控制器（MPC）形式，在每一步使用当前测量状态重新求解MILP。我们MPC形式的一个关键特征是 extit{任务自适应视野}，它利用TWTL确定性有限自动机（DFA）来确定每一步的活跃子任务，将预测视野限制在当前任务剩余窗口内，而不是完整公式的视野，这使得每次重新求解的成本显著低于初始的开环求解。
cs.RO / 11 / 2606.30893
Sampling-Based Coordination-Informed Multi-Objective Multi-Robot Reinforcement Learning
基于采样的协调信息多目标多机器人强化学习
Abstract
Multi-robot systems must simultaneously optimize competing objectives while maintaining coordinated behavior. Existing multi-agent reinforcement learning approaches often rely on fixed or centralized coordination, which limits adaptability and violates distributed constraints. This work introduces the Coordination-Informed Multi-Objective Reinforcement Learning (CIMORL) framework, integrating a distributed weight prediction mechanism, a privileged expert training strategy, and theoretical guarantees for Pareto-optimal solutions. We present the base CIMORL method alongside two sampling-based variants, CIMORL-TS (Tree Search) and CIMORL-MPPI (MPPI), which leverage privileged global information during training to enable fully decentralized deployment. Experimental validation in cooperative and adversarial scenarios demonstrates a $21.2\%$ hypervolume improvement and superior policy stability compared to state-of-the-art baselines. Real-world experiments with Crazyflie drones further validate the framework's robustness in resource allocation and multi-attacker multi-defend scenarios under partial observability.
Chinese Translation
多机器人系统必须在保持协调行为的同时，同时优化相互竞争的目标。现有的多智能体强化学习方法通常依赖于固定或集中式的协调，这限制了适应性并违反了分布式约束。本研究提出了协调信息多目标强化学习（Coordination-Informed Multi-Objective Reinforcement Learning, CIMORL）框架，集成了分布式权重预测机制、特权专家训练策略以及帕累托最优解的理论保证。我们展示了基础的CIMORL方法以及两个基于采样的变体，CIMORL-TS（树搜索）和CIMORL-MPPI（模型预测路径积分），它们在训练过程中利用特权全局信息，以实现完全去中心化的部署。在合作和对抗场景中的实验验证显示，与最先进的基线相比，超体积提高了21.2%，且策略稳定性更优。与Crazyflie无人机的真实世界实验进一步验证了该框架在资源分配和部分可观测下的多攻击者多防御场景中的鲁棒性。
cs.RO / 12 / 2606.30900
The Quadruped Soft Tail: Compliant Grasping and Swabbing for Contamination Surveys in Harsh Environments
四足软尾：在恶劣环境中进行污染调查的柔性抓取与擦拭
Abstract
Beryllium contamination surveys in radioactive areas are challenging for robots in environments cluttered with cables and electronics. To address this problem, we have developed a novel quadruped system augmentation: A lightweight, soft, and compliant tendon-actuated robotic tail mounted on a quadruped robot. The tail features a hollow, flexible backbone and a tendon-actuated soft gripper that enables the robot to pick up sampling tissues, swab contaminated surfaces, and release the tissues at designated collection locations for subsequent beryllium analysis. To enable intuitive teleoperation, a closed-form kinematic model and a singularity-robust task-space controller are developed. Experimental results demonstrate that gripper actuation has a negligible effect on robot shape, while common-mode tendon actuation provides an effective mechanism for stiffness modulation and preload control. Furthermore, experimental validation indicates that the proposed kinematic model provides a suitable basis for real-time task-space control. The proposed system combines the agility of legged locomotion with the compliance of soft robotic manipulation, enabling the complete contamination-survey procedure to be performed without human exposure. While motivated by beryllium contamination surveys at CERN, the proposed quadruped soft-tail concept is broadly applicable to legged robots operating in cluttered, confined, or hazardous environments where conventional rigid-link manipulators are undesirable.
Chinese Translation
在放射性区域进行铍污染调查对环境中充满电缆和电子设备的机器人来说具有挑战性。为了解决这一问题，我们开发了一种新颖的四足系统增强：一种轻量、柔软且顺应的腱驱动机器人尾巴，安装在四足机器人上。该尾巴具有一个空心、灵活的主干和一个腱驱动的柔性抓手，使机器人能够拾取采样组织、擦拭受污染表面，并在指定的收集地点释放组织以便后续的铍分析。为了实现直观的遥控操作，我们开发了一个封闭形式的运动学模型和一个对奇异性具有鲁棒性的任务空间控制器。实验结果表明，抓手的驱动对机器人形状的影响微乎其微，而共模腱驱动提供了一种有效的刚度调节和预加载控制机制。此外，实验验证表明，所提出的运动学模型为实时任务空间控制提供了合适的基础。该系统结合了腿部运动的灵活性与软机器人操作的顺应性，使得整个污染调查过程能够在不暴露人类的情况下完成。虽然这一概念是受到在CERN进行铍污染调查的启发，但所提出的四足软尾概念广泛适用于在拥挤、狭小或危险环境中操作的四足机器人，在这些环境中，传统的刚性连杆操纵器并不理想。
cs.RO / 13 / 2606.30940
Motion Planning in Compressed Representation Spaces
压缩表示空间中的运动规划
Abstract
Deep learning methods have vastly expanded the capabilities of motion planning in robotics applications, as learning priors from large-scale data has been shown to be essential in capturing the highly complex behavior required for solving tasks such as manipulation or navigation for autonomous vehicles. At the same time, model-based planning algorithms based on search or optimization remain an essential tool due to their flexibility, efficiency, and the ability to incorporate domain knowledge via expert-designed algorithms and objective functions. We propose a new generative framework to unify these two paradigms. First, we learn an autoencoder with a high compression ratio and a latent space of hierarchically ordered, discrete-valued tokens. Leveraging both the dimensionality reduction and the hierarchical coarse-to-fine structure learned by this autoencoder, we then perform motion planning by directly searching in the latent space of tokens. This search can optimize arbitrary objective functions specified at test time, providing a large degree of flexibility while maintaining efficiency and producing realistic solutions by relying on the generative capabilities of the highly compressed autoencoder. We evaluate our method on nuPlan and the Waymo Open Motion Dataset, showing how latent space search can be used for a variety of guided behavior generation tasks, achieving strong performance for closed-loop motion planning and multi-agent guided scenario synthesis without requiring any task-specific training.
Chinese Translation
深度学习方法极大地扩展了机器人应用中运动规划的能力，因为从大规模数据中学习先验知识被证明对于捕捉解决诸如自主车辆的操作或导航等任务所需的高度复杂行为至关重要。同时，基于搜索或优化的模型驱动规划算法仍然是一个重要工具，因为它们具有灵活性、效率，并能够通过专家设计的算法和目标函数纳入领域知识。我们提出了一种新的生成框架，以统一这两种范式。首先，我们学习一个具有高压缩比和分层有序、离散值标记的潜在空间的自编码器。利用该自编码器学习的降维和分层粗到细结构，我们随后通过直接在标记的潜在空间中搜索来执行运动规划。这种搜索可以优化在测试时指定的任意目标函数，提供了较大的灵活性，同时保持效率，并通过依赖于高度压缩自编码器的生成能力产生现实的解决方案。我们在nuPlan和Waymo开放运动数据集上评估了我们的方法，展示了潜在空间搜索如何用于各种引导行为生成任务，在闭环运动规划和多智能体引导场景合成中取得了强劲的表现，而无需任何特定任务的训练。
cs.RO / 14 / 2606.30988
Multisensory Continual Learning: Adapting Pretrained Visuomotor Policies to Force
多感官持续学习：将预训练的视觉运动策略适应于力反馈
Abstract
Robot manipulation often relies on sensory feedback beyond vision, particularly in contact-rich settings where force, tactile, or audio signals reveal interaction states that are not directly observable from images. However, these modalities are often hardware- and task-specific, and large-scale multisensory robot datasets remain scarce. As a result, it is impractical to pretrain policies with every sensor they may encounter. We study multisensory continual learning: adapting a pretrained robot policy to new tasks with newly introduced modalities while preserving performance under the original sensor suite. We propose MuSe, which incorporates limited multisensory data into pretrained vision-only policies through multi-stage fusion, multisensory future prediction, and experience replay over pretraining data. We instantiate MuSe by augmenting a pretrained vision-only policy with force-torque sensing and evaluate it on real-world manipulation tasks. Our experiments show that MuSe performs strongly on contact-rich finetuning tasks while preserving, and in some cases improving, performance on the original pretraining tasks. These results suggest that a modest multisensory dataset can improve general robot capabilities beyond the finetuning distribution. Project website: https://jadenvc.github.io/multisensory-continual-learning/
Chinese Translation
机器人操控通常依赖于超越视觉的感官反馈，特别是在接触丰富的环境中，力、触觉或音频信号揭示了从图像中无法直接观察到的交互状态。然而，这些模态往往是硬件和任务特定的，大规模的多感官机器人数据集仍然稀缺。因此，预训练每种可能遇到的传感器的策略是不切实际的。我们研究了多感官持续学习：将预训练的机器人策略适应于具有新引入模态的新任务，同时保持在原始传感器组合下的性能。我们提出了MuSe，通过多阶段融合、多感官未来预测和对预训练数据的经验重放，将有限的多感官数据融入预训练的视觉专用策略中。我们通过增强预训练的视觉专用策略与力-扭矩传感器的结合来实例化MuSe，并在真实世界的操控任务上进行评估。我们的实验表明，MuSe在接触丰富的微调任务上表现强劲，同时保持，且在某些情况下改善了原始预训练任务的性能。这些结果表明，适度的多感官数据集可以提升机器人在微调分布之外的通用能力。项目网站：https://jadenvc.github.io/multisensory-continual-learning/
cs.RO / 15 / 2606.31019
Ground Plane-Aided Extrinsic Calibration of Inertial and RGB-D Sensors for Uncrewed Aerial Vehicles
基于地面平面的无人机惯性传感器与RGB-D传感器的外部标定
Abstract
Accurate extrinsic calibration of inertial sensors, such as Inertial Measurement Units (IMUs) and cameras is crucial for trajectory estimation of Uncrewed Aerial Vehicles (UAVs). While numerous calibration methods have been proposed, these techniques often rely on specialized equipment, planar targets, and an initial estimate of the calibration parameters. In this research, we propose a targetless calibration method designed for UAVs equipped with IMUs and RGB-Depth (RGB-D) cameras. Our approach leverages deep-learning-based floor-segmentation to extract ground points from the depth channel of RGB-D images. Subsequently, the normal vector to these points is estimated. The known orientation of the normal to the floor segment and the gravity vector sensed in the accelerometer's frame are utilized in a robust estimation approach to estimate the extrinsic calibration parameters. We illustrate that the developed method outperforms MATLAB's Toolboxes and exhibits similar performance to Kalibr without the use of specialized checkerboard targets.
Chinese Translation
准确的惯性传感器（如惯性测量单元（IMU）和相机）的外部标定对于无人机（UAV）的轨迹估计至关重要。尽管已有许多标定方法被提出，这些技术通常依赖于专用设备、平面目标和标定参数的初步估计。在本研究中，我们提出了一种无目标的标定方法，旨在为配备IMU和RGB-深度（RGB-D）相机的无人机提供服务。我们的方法利用基于深度学习的地面分割技术，从RGB-D图像的深度通道中提取地面点。随后，估计这些点的法向量。已知的地面段法向量的方向和加速度计框架中感测到的重力向量被用于一种稳健的估计方法，以估计外部标定参数。我们展示了所开发的方法优于MATLAB工具箱，并且在没有使用专用棋盘格目标的情况下表现出与Kalibr相似的性能。
cs.RO / 16 / 2606.31037
Labimus: A Simulation and Benchmark for Humanoid Dexterous Manipulation in Chemical Laboratory
Labimus：化学实验室中类人灵巧操作的仿真与基准测试
Abstract
Laboratory automation has made remarkable progress through robotic platforms and AI-driven scientific reasoning. However, many laboratory operations (e.g., solid--solid transfer) remain inherently dynamic and require real-time adaptation to different materials and experimental conditions. Such precision-critical manipulations are difficult to standardize, motivating the use of humanoid robots with dexterous hands. Despite this opportunity, no existing benchmark evaluates humanoid manipulation in precision-critical laboratory environments. We present Labimus, to our knowledge, the first benchmark for humanoid dexterous manipulation in organic chemistry laboratories. Labimus reconstructs over 30 functionally faithful assets from real organic chemistry workstations through real-to-sim modeling, collectively covering the core operations of routine organic chemistry experiments. The benchmark integrates articulated laboratory instruments, particle-based powder physics, and closed-loop instrument readouts, enabling a complete manipulation-to-measurement pipeline. It further defines six atomic operations and a seven-step solid-weighing workflow derived from real laboratory standard operating procedures. We introduce a precision-aware evaluation protocol designed to jointly measure task completion, experimental precision, and long-horizon execution. We benchmark three representative policies under procedural layouts and environmental perturbations. Results reveal a precision gap: policies that successfully complete laboratory tasks can still fail to satisfy the quantitative tolerances required by experimental protocols. Our benchmark exposes a fundamental disconnect between task completion and experimental validity, providing a new testbed for developing reliable humanoid robots for scientific laboratories.
Chinese Translation
实验室自动化通过机器人平台和基于人工智能的科学推理取得了显著进展。然而，许多实验室操作（例如固体-固体转移）本质上仍然是动态的，需要实时适应不同的材料和实验条件。这类对精度要求严格的操作难以标准化，因此促使了类人机器人及其灵巧手的使用。尽管存在这样的机会，目前尚无现有基准评估类人在精度要求严格的实验室环境中的操作。我们提出了Labimus，据我们所知，这是第一个针对有机化学实验室中类人灵巧操作的基准测试。Labimus通过真实到仿真的建模，重建了30多个功能上忠实于真实有机化学工作站的资产，全面覆盖了常规有机化学实验的核心操作。该基准集成了关节实验室仪器、基于颗粒的粉末物理学和闭环仪器读数，实现了完整的操作到测量的流程。它进一步定义了六个原子操作和一个基于真实实验室标准操作程序的七步固体称重工作流程。我们引入了一种关注精度的评估协议，旨在共同测量任务完成、实验精度和长时间执行。我们在程序布局和环境扰动下基准测试了三种代表性策略。结果揭示了一个精度差距：成功完成实验室任务的策略仍可能无法满足实验协议所需的定量容差。我们的基准暴露了任务完成与实验有效性之间的根本脱节，为开发可靠的类人机器人在科学实验室中的应用提供了新的测试平台。
cs.RO / 17 / 2606.31101
Efficient Sim-to-Real Transfer of World-Action Models from Synthetic Priors
从合成先验中高效实现世界动作模型的仿真到现实转移
Abstract
Bridging the sim-to-real gap is a core challenge in deploying learned manipulation policies. Sim-to-real learning is attractive because it can replace expensive real robot demonstrations with scalable synthetic data, yet world-action models have not previously been shown to transfer from simulation to real robotic manipulation. We study whether a world-action model can be trained from synthetic priors and deployed zero-shot in the real world. To this end, we build upon Cosmos Policy, a video diffusion model adapted for visuomotor control. We construct simulation environments with extensive domain randomization and generate demonstrations using the AnyTask motion planning pipeline. We evaluate our approach across object lifting, drawer opening, and pick-and-place tasks using ${\sim}800$ synthetic demonstrations per task and no real demonstrations. When deployed zero-shot on a Franka Robot, our policy attains a 35\% average success rate. To our knowledge, this represents the first successful sim-to-real transfer of a world-action model for robotic manipulation.
Chinese Translation
弥合仿真与现实之间的差距是部署学习到的操作策略的核心挑战。仿真到现实学习具有吸引力，因为它可以用可扩展的合成数据替代昂贵的真实机器人演示，然而，世界动作模型之前并未成功实现从仿真到真实机器人操作的转移。我们研究了是否可以从合成先验中训练一个世界动作模型，并在现实世界中零样本部署。为此，我们基于Cosmos Policy构建了一种适用于视觉运动控制的视频扩散模型。我们构建了具有广泛领域随机化的仿真环境，并使用AnyTask运动规划管道生成演示。我们在物体提升、抽屉打开和拾取放置任务上评估了我们的方法，每个任务使用约800个合成演示且没有真实演示。当在Franka机器人上零样本部署时，我们的策略达到了35%的平均成功率。据我们所知，这是世界动作模型在机器人操作中首次成功实现的仿真到现实转移。
cs.RO / 18 / 2606.31106
What Probing Reveals about Autonomous Driving: Linking Internal Prediction Errors to Ego Planning
探测揭示的自动驾驶：将内部预测误差与自我规划联系起来
Abstract
Large-scale datasets and fast simulators have enabled improvements in driving policies that appear safe and robust, yet strong performance in nominal scenarios can still mask flawed reasoning and unsafe heuristics. Summary scores from closed-loop simulators do not give significant insight into the policy, making it difficult to determine whether they truly predict the motion of surrounding vehicles, how the ego vehicle generates future plans, or whether they merely rely on brittle heuristics that happen to succeed in nominal scenarios. To better understand the limits and weaknesses of driving policies, we focus on probing for forms of prediction, i.e., where surrounding vehicles will move next, and planning, i.e., understanding how to generate safe trajectories. We focus on these two capabilities because they reflect behaviors expected of effective driving policies, and use their presence or absence to assess policy quality across data-driven behavior cloning and simulation-driven reinforcement learning policies. To evaluate the presence of these capabilities, we investigate them as a function of scale, asking whether the closed-loop gains from larger datasets and longer simulation training reflect stronger prediction and planning or merely better behavioral heuristics. We use linear probing and targeted perturbations in both imitation learning and reinforcement learning models to track when these internal signals emerge, plateau, or fail. Despite good closed-loop performance, policies often fail to form timely surrounding-vehicle predictions during near-collision events, revealing a limitation in the predictive signals available for ego planning. Finally, causal intervention shows that correcting mistaken predictions improves ego planning toward safer trajectories.
Chinese Translation
大规模数据集和快速模拟器的出现使得驾驶策略的改进看起来既安全又稳健，然而在正常场景下的强大表现仍然可能掩盖错误的推理和不安全的启发式方法。来自闭环模拟器的总结评分并未提供对策略的显著洞察，这使得判断它们是否真正预测周围车辆的运动、如何生成自我车辆的未来规划，或者仅仅依赖于在正常场景中恰好成功的脆弱启发式方法变得困难。为了更好地理解驾驶策略的局限性和弱点，我们专注于探测预测的形式，即周围车辆将如何移动，以及规划，即理解如何生成安全轨迹。我们关注这两种能力，因为它们反映了有效驾驶策略所期望的行为，并利用它们的存在或缺失来评估基于数据驱动的行为克隆和基于模拟驱动的强化学习策略的质量。为了评估这些能力的存在，我们将其作为规模的函数进行研究，询问来自更大数据集和更长模拟训练的闭环增益是否反映了更强的预测和规划，还是仅仅更好的行为启发式。我们在模仿学习和强化学习模型中使用线性探测和有针对性的扰动，以追踪这些内部信号何时出现、达到平稳或失败。尽管闭环表现良好，策略在近碰撞事件中往往未能及时形成周围车辆的预测，揭示了可用于自我规划的预测信号的局限性。最后，因果干预表明，纠正错误的预测能够改善自我规划，朝向更安全的轨迹。
cs.RO / 19 / 2606.31132
ELASTIC: Efficiently Learning to Adaptively Scale Test-Time Compute for Generative Control Policies
ELASTIC：高效学习自适应调整生成控制策略测试时计算的算法
Abstract
Generative control policies (GCPs), such as diffusion policies and flow-based vision-language-action models, enable test-time scaling in robot control. Test-time compute can be allocated along two axes: sequential scaling, which increases denoising steps to refine actions, and parallel scaling, which samples multiple candidate actions to search across modes of the policy distribution. However, the optimal allocation of sequential and parallel compute is hard to know a priori as it is state-, task-, and policy-dependent. For example, early stages of a grasp may benefit from broader parallel exploration, while near-contact phases may require more sequential refinement for precision. We present ELASTIC, an algorithm that learns state-dependent test-time compute schedules for GCPs. We formulate compute allocation as a meta-Markov Decision Process in which a meta-policy interacts with a frozen pretrained robot policy and selects sequential steps and parallel samples at each denoising iteration to maximize task success while minimizing compute. Using reinforcement learning, this meta-policy also learns adaptive compute schedules without access to the GCP's training data. Across simulated manipulation benchmarks with diffusion policies, ELASTIC Pareto-dominates fixed and single-axis scaling baselines at matched compute budgets. On real-world robot manipulation with the $\pi_{0.5}$ vision-language-action model, ELASTIC matches best-of-$10$ success while reducing wall-clock latency by 34%.
Chinese Translation
生成控制策略（GCPs），如扩散策略和基于流的视觉-语言-动作模型，使得机器人控制中的测试时计算可以进行扩展。测试时计算可以沿两个轴进行分配：顺序扩展，即增加去噪步骤以细化动作；以及并行扩展，即采样多个候选动作以在策略分布的不同模式中进行搜索。然而，顺序和并行计算的最佳分配难以事先确定，因为它依赖于状态、任务和策略。例如，抓取的早期阶段可能更需要广泛的并行探索，而接触阶段则可能需要更多的顺序细化以提高精度。我们提出了ELASTIC，一种为GCPs学习状态依赖的测试时计算调度的算法。我们将计算分配形式化为一个元马尔可夫决策过程，其中元策略与一个冻结的预训练机器人策略进行交互，并在每次去噪迭代中选择顺序步骤和并行样本，以最大化任务成功率并最小化计算量。通过强化学习，该元策略还学习自适应的计算调度，而无需访问GCP的训练数据。在使用扩散策略的模拟操作基准测试中，ELASTIC在匹配的计算预算下优于固定和单轴扩展的基线。在使用$ ext{π}_{0.5}$视觉-语言-动作模型进行的真实机器人操作中，ELASTIC在减少34%的实际延迟的同时，达到了最佳10次成功率的水平。
cs.RO / 20 / 2606.31144
A Modular Vision-Language-Action Robotics Framework for Indoor Environments
适用于室内环境的模块化视觉-语言-动作机器人框架
Abstract
This paper presents an integrated system for the CMU Vision-Language-Action (VLA) Challenge, designed to enable an autonomous agent to perform complex tasks based on natural language instructions. Our framework employs a modular architecture that orchestrates environment mapping, question processing, and navigation. The system operates in two parallel streams: a perception pipeline that constructs a semantic voxel map from real-time camera feeds using OwlViT embeddings, and a language pipeline that classifies user commands with a Vision-Language Model. The mapping is time-constrained; the system proceeds with a partial map if a 500-second exploration limit is reached. The classified query is then grounded in the geometric and semantic context of the map to generate a detailed prompt for the VLM. This yields an actionable output, demonstrating a capable solution for bridging the gap between human language and robotic action.
Chinese Translation
本文提出了一种集成系统，旨在支持CMU视觉-语言-动作（VLA）挑战，使自主代理能够根据自然语言指令执行复杂任务。我们的框架采用模块化架构，协调环境映射、问题处理和导航。系统在两个并行流中运行：一个感知管道，通过使用OwlViT嵌入从实时摄像头数据构建语义体素地图；另一个语言管道，通过视觉-语言模型对用户命令进行分类。映射过程受到时间限制；如果达到500秒的探索限制，系统将继续使用部分地图。分类后的查询随后在地图的几何和语义上下文中进行定位，以生成针对视觉-语言模型（VLM）的详细提示。这产生了可操作的输出，展示了弥合人类语言与机器人动作之间差距的有效解决方案。
cs.RO / 21 / 2606.31158
LLM-Powered Interactive Robotic Action Synthesis from Multimodal Speech, Gestures, and Music
基于大型语言模型的多模态语音、手势和音乐的互动机器人动作合成
Abstract
The quest for intuitive and natural human-robot interaction (HRI) remains a significant challenge in robotics. Traditional methods often rely on rigid, pre-programmed commands that limit the robot's expressiveness and adaptability. This paper introduces a novel framework that leverages the reasoning capabilities of Large Language Models (LLMs) to synthesize complex robotic actions from a rich tapestry of multimodal human inputs: natural speech, hand gestures, and music/sound beats. Our system architecture integrates a speech transcription model, a gesture recognition module, and a signal processing pipeline for beat detection. These processed inputs are contextualized using prompt templates and fed into a LLM. The LLM, informed by a predefined robot action space, reasons over the combined inputs to generate a coherent sequence of actions. This sequence is dispatched to an action queue for execution on a quadruped robot over ROS. The framework has ability to interpret and fuse semantic commands from speech, deictic information from gestures, and rhythmic cues from music. This work represents a step towards creating robots that can interact with humans in a more fluid, creative, and context-aware manner.
Chinese Translation
寻求直观和自然的人机交互（HRI）仍然是机器人技术中的一个重大挑战。传统方法通常依赖于僵化的预编程命令，这限制了机器人的表现力和适应性。本文介绍了一种新颖的框架，该框架利用大型语言模型（LLMs）的推理能力，从丰富的多模态人类输入中合成复杂的机器人动作：自然语音、手势和音乐/声音节拍。我们的系统架构集成了语音转录模型、手势识别模块和用于节拍检测的信号处理管道。这些处理后的输入通过提示模板进行上下文化，并输入到LLM中。LLM在预定义的机器人动作空间的指导下，对组合输入进行推理，以生成连贯的动作序列。该序列被调度到动作队列中，以便在ROS上执行四足机器人。该框架能够解释和融合来自语音的语义命令、来自手势的指示信息以及来自音乐的节奏线索。这项工作代表了朝着创造能够以更流畅、更具创造性和上下文感知的方式与人类互动的机器人迈出了一步。
cs.RO / 22 / 2606.31167
MIRTH: Mutual-Information Reasoning with Temporal Hubs for Vision-Language-Action Agents
MIRTH：用于视觉-语言-动作代理的时间中心的互信息推理
Abstract
VLA models have emerged as a powerful paradigm for transferring semantic knowledge from web-scale data to physical robotic control. However, current single-frame architectures suffer from intrinsic limitations: temporal myopia that discards historical dynamics, reasoning gaps between high-level instructions and low-level motor commands, and inference inefficiency due to autoregressive scalar decoding. In this work, we propose MIRTH, a unified framework designed to address these challenges. MIRTH augments a pretrained VLA backbone with three key innovations: (1) dual-scale temporal memory hubs that compress long-term scene evolution and short-term motion trends into compact embeddings; (2) latent reasoning tokens optimized via a mutual-information objective carving out a semantic plan space to align multimodal context with action trajectories; and (3) a parallel action decoding scheme that replaces autoregressive generation with vector-wise prediction to maximize control throughput. Extensive evaluations on the LIBERO simulation benchmark and a real-world LeRobot platform demonstrate that MIRTH achieves state-of-the-art performance and exhibiting emergent error recovery capabilities. The codes and collected datasets are released at http://github.com/kiva12138/mirth.
Chinese Translation
视觉-语言-动作（VLA）模型已成为将语义知识从网络规模数据转移到物理机器人控制的强大范式。然而，当前的单帧架构存在固有的局限性：时间近视导致忽视历史动态、高层指令与低层运动命令之间的推理差距，以及由于自回归标量解码造成的推理效率低下。在本研究中，我们提出了MIRTH，一个旨在解决这些挑战的统一框架。MIRTH通过三项关键创新增强了预训练的VLA骨干网络：（1）双尺度时间记忆中心，将长期场景演变和短期运动趋势压缩为紧凑的嵌入；（2）通过互信息目标优化的潜在推理令牌，划分出一个语义规划空间，以将多模态上下文与动作轨迹对齐；（3）一种并行动作解码方案，用向量预测替代自回归生成，以最大化控制吞吐量。在LIBERO仿真基准和真实世界的LeRobot平台上的广泛评估表明，MIRTH实现了最先进的性能，并展现出突出的错误恢复能力。代码和收集的数据集已发布在http://github.com/kiva12138/mirth。
cs.RO / 23 / 2606.31197
Diffusion-based 4D Trajectory Prediction and Distributed Control for UAV Swarms
基于扩散的无人机编队四维轨迹预测与分布式控制
Abstract
Accurate 4D trajectory prediction and closed-loop tracking are essential for Unmanned Aerial Vehicle (UAV) swarms to achieve safe and efficient operations in complex low-altitude environments such as urban airspaces, industrial sites, and indoor facilities. However, this task remains challenging due to intrinsic nonlinearity of UAV swarm dynamics and strict real-time constraints of swarm formation control. To address these challenges, we propose a unified framework that couples coarse-to-fine trajectory forecasting with uncertainty-aware Distributed Nonlinear Model Predictive Control (DNMPC). Our approach features two key innovations: 1) a dimension-decoupled trajectory prediction module that reduces computational complexity by forecasting axis-wise motion, and 2) a diffusion-based residual dynamics refinement module that captures temporally correlated dynamic uncertainties. These refined predictions are then integrated into a DNMPC loop to ensure formation stability. We also introduce a synchronized multi-scenario 4D UAV swarm dataset spanning six representative airspace scenarios. The dataset contains over \textbf{7,900} frames of synchronized three-UAV trajectories with frame-level annotations of speed intention and target sector. Extensive experiments demonstrate that our approach outperforms state-of-the-art baselines, reducing trajectory tracking error by up to \textbf{10-15\%} and achieving sub-\textbf{0.07\,m} average tracking error in complex urban and industrial environments, while maintaining real-time inference speeds of 34 FPS (sub-30 ms latency) suitable for agile flight.
Chinese Translation
准确的四维轨迹预测和闭环跟踪对于无人机（UAV）编队在复杂低空环境（如城市空域、工业场所和室内设施）中实现安全高效的操作至关重要。然而，由于无人机编队动态的内在非线性以及编队控制的严格实时约束，这项任务仍然具有挑战性。为了解决这些挑战，我们提出了一个统一框架，将粗到细的轨迹预测与不确定性感知的分布式非线性模型预测控制（DNMPC）相结合。我们的方法具有两个关键创新：1）一个维度解耦的轨迹预测模块，通过轴向运动的预测来降低计算复杂性；2）一个基于扩散的残差动态细化模块，捕捉时间相关的动态不确定性。这些细化的预测随后被整合到DNMPC循环中，以确保编队稳定性。我们还引入了一个同步的多场景四维无人机编队数据集，涵盖六个代表性的空域场景。该数据集包含超过7900帧同步的三架无人机轨迹，并附有速度意图和目标区域的帧级注释。大量实验表明，我们的方法优于最先进的基线，将轨迹跟踪误差降低了高达10-15%，并在复杂的城市和工业环境中实现了平均跟踪误差低于0.07米，同时保持34帧每秒（低于30毫秒延迟）的实时推理速度，适合灵活飞行。
cs.RO / 24 / 2606.31199
Machine Learning-based Feedback Linearization Control of Quadrotor Subject to Unmodeled Dynamics
基于机器学习的四旋翼反馈线性化控制方法研究：应对未建模动力学
Abstract
The control of agile quadrotors in dynamic and uncertain environments remains an open area of investigation to this day, particularly when the complete system dynamics are partially known or highly nonlinear. This work introduces a novel machine learning-based feedback-linearization control framework that employs a Gaussian Radial Basis Function (RBF) neural network (NN) to model and compensate for unmodeled dynamics in real time. The proposed controller leverages the universal approximation capability of RBF networks to model nonlinearities and uncertainties. An online adaptation of the RBF NN updates the network's weights without prior training. The control law is derived using the Lyapunov stability theory, herein guaranteeing closed-loop stability and providing theoretical guarantee of asymptotic convergence of a trajectory tracking task. Gazebo simulation and real flight experiments are conducted using the Bitcraze's Crazyflie 2.1 quadrotor subject to unmodeled air drag, actuator dynamics, and external disturbance. Despite incomplete knowledge of prior dynamics and presence of external disturbance such as air drag and drift in state estimation, the proposed controller improves trajectory tracking with rapid convergence and reduction of position-norm and yaw orientation RMSE by more than $7.13\%$ and $49.27\%$ respectively compared to baseline feedback linearization controller.
Chinese Translation
在动态和不确定环境中，灵活四旋翼的控制仍然是一个开放的研究领域，特别是在系统动力学部分已知或高度非线性的情况下。本文提出了一种新颖的基于机器学习的反馈线性化控制框架，该框架采用高斯径向基函数（RBF）神经网络（NN）实时建模并补偿未建模动力学。所提出的控制器利用RBF网络的通用逼近能力来建模非线性和不确定性。RBF神经网络的在线自适应更新网络权重，无需事先训练。控制律是基于Lyapunov稳定性理论推导的，保证了闭环稳定性，并提供了轨迹跟踪任务渐近收敛的理论保证。使用Bitcraze的Crazyflie 2.1四旋翼进行的Gazebo仿真和实际飞行实验表明，尽管对先前动力学的知识不完整，并且存在空气阻力和状态估计漂移等外部干扰，所提出的控制器在轨迹跟踪方面实现了快速收敛，并将位置范数和偏航方向的均方根误差（RMSE）分别降低了超过$7.13\%$和$49.27\\%$，相较于基线反馈线性化控制器。
cs.RO / 25 / 2606.31216
Information-Aided DVL Calibration
信息辅助的DVL校准
Abstract
The Doppler velocity log (DVL) velocity measurements are critical to the accuracy of autonomous underwater vehicle (AUV) navigation solutions and, consequently, to mission success. To ensure accurate measurements, the DVL is commonly calibrated before mission start while the AUV sails on the water surface, receiving global navigation satellite system (GNSS) signals that provide accurate reference measurements. Conventionally, Kalman filter-based approaches are employed during calibration to estimate the scale factor and misalignment errors. However, in certain environments, GNSS signals may be unavailable, rendering conventional calibration impossible and forcing the use of uncalibrated DVL measurements, which degrades navigation performance. To address this limitation, this work proposes information-aided calibration (IAC) with two main contributions: first, improving the accuracy of conventional Kalman filter-based calibration in GNSS-enabled environments, and second, enabling GNSS-free DVL self-calibration. Using real-world AUV datasets, the proposed IAC models achieve up to a 20% average improvement in GNSS-enabled environments and up to a 35% improvement in velocity vector estimation during GNSS-free DVL self-calibration. Overall, the proposed approach improves navigation accuracy, reduces navigation drift, and consequently enhances mission reliability.
Chinese Translation
多普勒速度计（DVL）的速度测量对自主水下航行器（AUV）导航解的准确性至关重要，因此直接关系到任务的成功。为了确保测量的准确性，DVL通常在任务开始前进行校准，此时AUV在水面航行，接收全球导航卫星系统（GNSS）信号，以提供准确的参考测量。传统上，在校准过程中采用基于卡尔曼滤波器的方法来估计比例因子和错位误差。然而，在某些环境中，GNSS信号可能不可用，这使得传统校准变得不可能，迫使使用未校准的DVL测量，从而降低导航性能。为了解决这一局限性，本研究提出了信息辅助校准（IAC），其主要贡献有两个：首先，在GNSS支持的环境中提高传统卡尔曼滤波器校准的准确性；其次，实现GNSS-free DVL自校准。通过使用真实世界的AUV数据集，所提出的IAC模型在GNSS支持的环境中实现了高达20%的平均改进，在GNSS-free DVL自校准期间速度向量估计提高了高达35%。总体而言，所提出的方法提高了导航准确性，减少了导航漂移，从而增强了任务的可靠性。
cs.RO / 26 / 2606.31236
TactX: Learning Shared Tactile Representations Across Diverse Sensors
TactX：跨多种传感器学习共享触觉表征
Abstract
Tactile sensors provide critical information for contact-rich manipulation, yet tactile representations and policies remain tightly coupled to each specific sensor, limiting transferability across robots and hardware platforms. We propose TactX, a framework for learning a transferable tactile representation across sensors spanning three fundamentally different transduction modalities: resistive, magnetic, and vision-based. TactX maps heterogeneous tactile observations into a shared latent space through modality-specific encoders trained on paired contact data. Such paired interactions provide a natural alignment signal across modalities, and the encoders are jointly trained across all sensor pairs, inducing a consistent latent space for all sensor types. Our experiments show that TactX aligns tactile representations across sensors while preserving object-level contact information, as evidenced by sensor-identity prediction and object classification in the learned latent space. We evaluate TactX on four contact-rich manipulation tasks: pick-and-place, plug insertion, board wiping, and object reorientation, and show that policies trained with one sensor transfer zero-shot to physically distinct sensors through the shared latent. This improves the average success rate from 27.5% for vision-only policy to 45.9%, providing a step toward sensor-agnostic tactile manipulation.
Chinese Translation
触觉传感器为接触丰富的操作提供了关键的信息，但触觉表征和策略仍然与每个特定传感器紧密耦合，限制了在机器人和硬件平台之间的可转移性。我们提出了TactX，一个学习跨三种根本不同的转导模式（电阻式、磁性和基于视觉的传感器）可转移触觉表征的框架。TactX通过在配对接触数据上训练的特定模式编码器，将异构触觉观测映射到共享的潜在空间。这种配对交互为不同模式之间提供了自然的对齐信号，编码器在所有传感器对之间共同训练，从而为所有传感器类型诱导出一致的潜在空间。我们的实验表明，TactX在对齐传感器间的触觉表征的同时，保留了物体级别的接触信息，这通过在学习的潜在空间中的传感器身份预测和物体分类得到了验证。我们在四个接触丰富的操作任务上评估了TactX：拾取与放置、插头插入、擦拭板和物体重新定向，并显示出使用一种传感器训练的策略可以通过共享潜在空间零-shot转移到物理上不同的传感器。这将视觉仅策略的平均成功率从27.5%提高到45.9%，为无传感器依赖的触觉操作迈出了重要一步。
cs.RO / 27 / 2606.31260
Plan Right, Then Plan Tight: Symbolic RL for Efficient Embodied Reasoning
先规划好，再紧凑规划：用于高效具身推理的符号强化学习
Abstract
Embodied task planning asks an agent to turn a natural-language instruction into an executable sequence of actions in a physical scene, and is a building block for household, assistive, and service robots. Recent prompting-based and reinforcement-learning planners generate fluent action text but lack a cheap deterministic check that the produced plan is valid in the target world, while high-fidelity simulation is too slow to serve as an inner-loop training signal. The general problem is therefore how to obtain verifiable supervision and rewards for embodied planners without relying on string-level matching or full simulation. Here we show that a single BDDL specification, automatically constructed from open-world video evidence or curated tasks, can serve as a shared interface for data construction, plan verification, and reward design. A video-to-BDDL parser, an LLM verifier, and a lightweight symbolic engine together supply dense feedback at millisecond latency. We further introduce GroupAdapt, a difficulty-aware length schedule that uses the in-batch group pass rate as a zero-cost signal so that hard prompts get wider length tolerance and automatically tighten as their pass rate improves. Under the guidance of the proposed verifier and GroupAdapt schedule, the 8B planner attains a Strict-Pass score of 97.3 on BEHAVIOR-1000, yielding a 25.9 percent relative improvement over the Qwen3-8B baseline. This result exceeds the strongest large-model baseline by 3.5 percent, while simultaneously compressing the response length by 79 percent to 207 tokens, demonstrating both effectiveness and efficiency.
Chinese Translation
具身任务规划要求代理将自然语言指令转化为在物理场景中可执行的动作序列，是家庭、辅助和服务机器人技术的基础。近期基于提示和强化学习的规划者能够生成流畅的动作文本，但缺乏一种廉价的确定性检查，以验证所生成的计划在目标世界中是否有效，而高保真模拟又过于缓慢，无法作为内部训练信号。因此，普遍问题在于如何为具身规划者获得可验证的监督和奖励，而不依赖于字符串级匹配或完全模拟。在此，我们展示了一个单一的BDD（行为描述语言，Behavioral Description Language）规范，能够从开放世界视频证据或策划任务中自动构建，作为数据构建、计划验证和奖励设计的共享接口。视频到BDD解析器、LLM（大型语言模型，Large Language Model）验证器和轻量级符号引擎共同提供毫秒级延迟的密集反馈。我们进一步引入了GroupAdapt，一种基于困难感知的长度调度，利用批次内的组通过率作为零成本信号，使得困难提示获得更宽的长度容忍度，并随着通过率的提高自动收紧。在所提出的验证器和GroupAdapt调度的指导下，8B规划者在BEHAVIOR-1000上获得了97.3的严格通过分数，相较于Qwen3-8B基线实现了25.9%的相对提升。该结果超越了最强大的大模型基线3.5%，同时将响应长度压缩了79%，至207个标记，展示了有效性和效率。
cs.RO / 28 / 2606.31329
3D HAMSTER: Bridging Planning and Control in Hierarchical Vision Language Action Models through 3D Trajectory Guidance
3D HAMSTER：通过3D轨迹引导在层次化视觉语言行动模型中架起规划与控制的桥梁
Abstract
Hierarchical Vision-Language-Action (VLA) models decouple high-level planning from low-level control to improve generalization in robot manipulation. Recent work in this paradigm uses 2D end-effector trajectories predicted by a Vision-Language Model (VLM) as explicit guidance for a downstream policy. However, state-of-the-art low-level policies operate in 3D metric space on point clouds, and feeding them 2D guidance that lacks depth forces each waypoint to be assigned the depth of whatever scene surface lies beneath it, producing geometrically distorted trajectories. We propose 3D HAMSTER, a hierarchical framework that closes this gap by having the planner directly output metrically reliable 3D trajectories. We augment a VLM with a dedicated depth encoder and a dense depth reconstruction objective to predict 3D waypoint sequences, which are directly integrated into a pointcloudbased low-level policy. Across 3D trajectory prediction, simulation, and real-world manipulation, 3D HAMSTER consistently outperforms proprietary VLMs and 2D-guided baselines, with the largest gains under appearance-altering shifts and unseen language, spatial, and visual conditions. The project page is available at https://davian-robotics.github.io/3D_HAMSTER/.
Chinese Translation
层次化视觉-语言-行动（VLA）模型将高层规划与低层控制解耦，以提高机器人操作的泛化能力。该范式下的近期研究使用视觉-语言模型（VLM）预测的2D末端执行器轨迹作为下游策略的显式指导。然而，最先进的低层策略在点云的3D度量空间中操作，而提供缺乏深度信息的2D指导会迫使每个路径点被分配为其下方场景表面的深度，从而产生几何扭曲的轨迹。我们提出了3D HAMSTER，一个层次化框架，通过让规划器直接输出度量可靠的3D轨迹来填补这一空白。我们为VLM增强了一个专用的深度编码器和一个密集深度重建目标，以预测3D路径点序列，这些序列被直接整合到基于点云的低层策略中。在3D轨迹预测、仿真和现实世界操作中，3D HAMSTER始终优于专有的VLM和2D指导的基线，尤其在外观变化、未见语言、空间和视觉条件下获得了最大的提升。项目页面可访问：https://davian-robotics.github.io/3D_HAMSTER/
cs.RO / 29 / 2606.31339
Verification-Gated Agentic Mission-State Governance for Intelligent Industrial Multi-Robot Systems
基于验证门控的自主任务状态治理框架用于智能工业多机器人系统
Abstract
Agentic artificial intelligence is increasingly used to decompose industrial tasks, propose robot actions, and adapt execution plans in dynamic cyber-physical environments. However, autonomous proposal generation alone does not guarantee that multi-robot industrial systems preserve task dependencies, resource ownership, safety holds, or repair boundaries during long-horizon execution. This paper introduces a verification-gated agentic mission-state governance framework for intelligent industrial multi-robot systems. The framework maintains two synchronized state objects: an evolving task forest for persistent hierarchy, delayed grounding, and repairable substructures; and a governed blackboard for online execution state, robot traces, resource locks, world beliefs, proposals, verification records, and scene-temporary constraints. From each forest--blackboard snapshot, a derived execution coupling topology exposes cross-branch dependencies for proposal verification, parallel-commit eligibility, and bounded repair. Candidate assignments, repairs, deferrals, and constraint updates may be generated by heuristic, optimization, or agentic reasoning modules, but they can update the committed mission state only after deterministic verification and atomic commit. We evaluate the framework in an indoor factory multi-robot scenario, 30-seed remote-construction stress benchmarks, structural ablations, and scalability probes. The results show improved verified and safety-audited mission-state progress with fewer invalid commitments, lock conflicts, duplicate assignments, abandoned nodes, and disruptive repairs under modeled mission predicates. The study positions agentic AI as a proposal-generating layer governed by inspectable mission-state verification rather than as an unchecked execution authority.
Chinese Translation
自主人工智能在工业任务分解、机器人动作提议和动态网络物理环境中的执行计划调整中被越来越多地使用。然而，仅靠自主提议生成并不能保证多机器人工业系统在长时间执行过程中保持任务依赖关系、资源所有权、安全保持或修复边界。本文提出了一种基于验证门控的自主任务状态治理框架，适用于智能工业多机器人系统。该框架维护两个同步状态对象：一个用于持久层次、延迟基础和可修复子结构的演变任务森林；以及一个用于在线执行状态、机器人轨迹、资源锁、世界信念、提议、验证记录和场景临时约束的管理黑板。从每个森林-黑板快照中，派生的执行耦合拓扑揭示了提议验证、并行提交资格和有限修复的跨分支依赖关系。候选分配、修复、延迟和约束更新可以通过启发式、优化或自主推理模块生成，但它们只能在确定性验证和原子提交后更新已承诺的任务状态。我们在一个室内工厂多机器人场景中评估了该框架，包括30个种子的远程施工压力基准、结构消融和可扩展性探测。结果表明，在建模的任务谓词下，经过验证和安全审计的任务状态进展得到了改善，且无效承诺、锁冲突、重复分配、放弃节点和破坏性修复的数量减少。该研究将自主人工智能定位为一个由可检查的任务状态验证所治理的提议生成层，而不是一个不受控制的执行权威。
cs.RO / 30 / 2606.31377
Stage-Transition Dense Reward Modeling for Reinforcement Learning
阶段转换密集奖励建模用于强化学习
Abstract
Reinforcement learning for long-horizon robotic manipulation is often limited by sparse and delayed rewards, while manually designing dense shaping signals is costly and brittle to changes in environments and object configurations. This work proposes Stage-Transition Dense Reward (STDR), a visual reward-learning framework that converts unstructured expert videos into logically grounded dense rewards for training RL agents from scratch. STDR leverages semantic understanding to infer a task's stage structure from demonstrations, and delivers two complementary learning signals during online training: (i) stage-transition feedback that provides goal-directed reward, and (ii) within-stage progress feedback that supplies fine-grained guidance toward completing each stage. Furthermore, an out-of-distribution (OOD) detection mechanism and a grasping regulation module are integrated to enhance robustness and prevent reward hacking. Experiments on 14 manipulation tasks across MetaWorld, ManiSkill, and Franka Kitchen show that STDR consistently improves sample efficiency and success rates over multiple baselines, and matches or surpasses handcrafted dense rewards on several challenging tasks. Real-robot evaluations further indicate that STDR assigns stable, progress-aligned rewards on successful executions while producing appropriately low rewards for failures, suggesting robustness to visual noise and better-calibrated reward assignment across settings.
Chinese Translation
长时间跨度的机器人操作中的强化学习常常受到稀疏和延迟奖励的限制，而手动设计密集的塑造信号成本高且对环境和物体配置的变化脆弱。本研究提出了阶段转换密集奖励（Stage-Transition Dense Reward, STDR），这是一个视觉奖励学习框架，可以将非结构化的专家视频转换为逻辑上合理的密集奖励，以便从零开始训练强化学习代理。STDR利用语义理解从示范中推断任务的阶段结构，并在在线训练期间提供两种互补的学习信号：（i）阶段转换反馈，提供目标导向的奖励，以及（ii）阶段内进展反馈，提供细粒度的指导以完成每个阶段。此外，集成了一个分布外（Out-of-Distribution, OOD）检测机制和一个抓取调节模块，以增强鲁棒性并防止奖励操控。在MetaWorld、ManiSkill和Franka Kitchen的14个操作任务上的实验表明，STDR在多个基线之上始终提高了样本效率和成功率，并在若干具有挑战性的任务上与手工制作的密集奖励相匹配或超越。真实机器人评估进一步表明，STDR在成功执行时分配稳定且与进展对齐的奖励，而在失败时则产生适当低的奖励，表明其对视觉噪声的鲁棒性以及在不同设置下更好地校准的奖励分配。
cs.RO / 31 / 2606.31382
Revisiting Parameter Redundancy in Vision-Language-Action Models: Insights from VLM-to-VLA Adaptation
重新审视视觉-语言-动作模型中的参数冗余：来自VLM到VLA适应的见解
Abstract
Vision-Language-Action (VLA) models have made significant strides in embodied intelligence by integrating the powerful representations of pre-trained Vision-Language Models (VLMs). However, the massive parameter scale of VLAs imposes a heavy computational burden, and these models exhibit extreme sensitivity to parameter pruning. Current paradigms often treat the resulting performance degradation as inevitable, relying on fine-tuning or low-rank corrections to recover efficacy. We challenge this convention by questioning whether the removed parameters are truly redundant if VLA pruning necessitates performance recovery to be effective, or if this paradigm masks the indiscriminate pruning of critical parameters. We revisit parameter redundancy through the lens of VLM-to-VLA adaptation, first quantifying the spatial distribution of parameter divergence during adaptation to reveal structured patterns across different modules. Subsequently, we introduce controlled pruning as a diagnostic probe: by comparing the direct impact of removing different parameter subsets on VLA performance without any fine-tuning, we establish a causal link between adaptation-induced divergence signals and functional contributions. Based on the discovered modular heterogeneities, we design a multi-module joint pruning scheme. Evaluations on the LIBERO benchmark demonstrate that our approach reduces the parameters of OpenVLA and $\pi_{0.5}$ by 12\%--30\% while maintaining approximately 90\% of the original performance without any post-pruning recovery. In contrast, existing parameter pruning criteria result in total performance collapse when evaluated under the same recovery-free constraints. Our study reveals the parameter evolution mechanism in VLA adaptation and provides a new path for deploying efficient, robust robotic policies in resource-constrained environments.
Chinese Translation
视觉-语言-动作（VLA）模型通过整合预训练视觉-语言模型（VLM）的强大表征，在具身智能方面取得了显著进展。然而，VLA的庞大参数规模带来了沉重的计算负担，并且这些模型对参数剪枝表现出极端的敏感性。目前的范式通常将由此产生的性能下降视为不可避免，依赖于微调或低秩修正来恢复效能。我们质疑这一惯例，探讨如果VLA剪枝需要性能恢复才能有效，是否被移除的参数真的冗余，或者这一范式是否掩盖了对关键参数的无差别剪枝。我们通过VLM到VLA适应的视角重新审视参数冗余，首先量化适应过程中参数偏差的空间分布，以揭示不同模块之间的结构模式。随后，我们引入受控剪枝作为诊断工具：通过比较在没有任何微调的情况下移除不同参数子集对VLA性能的直接影响，我们建立了适应引起的偏差信号与功能贡献之间的因果关系。基于发现的模块异质性，我们设计了一种多模块联合剪枝方案。在LIBERO基准上的评估表明，我们的方法在不进行任何后剪枝恢复的情况下，将OpenVLA和$ ext{π}_{0.5}$的参数减少了12 ext{%}至30 ext{%}，同时保持了约90 ext{%}的原始性能。相比之下，现有的参数剪枝标准在相同的无恢复约束下评估时会导致性能的完全崩溃。我们的研究揭示了VLA适应中的参数演变机制，并为在资源受限环境中部署高效、稳健的机器人策略提供了一条新路径。
cs.RO / 32 / 2606.31451
UniTac: A Unified Multimodal Model for Cross-Sensor Tactile Understanding and Generation
UniTac：一种用于跨传感器触觉理解与生成的统一多模态模型
Abstract
Unified multimodal models (UMMs) have shown great promise in integrating understanding and generation across diverse modalities. However, existing research rarely extends this paradigm to the tactile domain, where both object-level semantics and sensor-level configurations jointly determine the meaning of touch. To address this gap, we propose UniTac, the first UMM designed for tactile understanding and generation. UniTac models the tactile process as a transition from non-contact to contact, capturing the physical interaction between sensors and objects through a dual-level representation that encodes both sensor and object attributes. For tactile understanding, UniTac introduces two tasks, object property description and sensor identification, to enhance reasoning over physical and cross-sensor information. For tactile generation, we design a two-stage training paradigm consisting of reconstruction and alignment, together with a sensor-prior-based sampling strategy that simulates realistic tactile contact. Trained on large-scale multi-sensor datasets, UniTac achieves state-of-the-art performance in tactile understanding and generates realistic tactile signals across sensors.
Chinese Translation
统一多模态模型（UMMs）在整合不同模态的理解与生成方面展现出了巨大的潜力。然而，现有研究很少将这一范式扩展到触觉领域，在该领域中，物体级语义与传感器级配置共同决定了触觉的意义。为了解决这一问题，我们提出了UniTac，这是首个专为触觉理解与生成设计的UMM。UniTac将触觉过程建模为从非接触到接触的过渡，通过一种双层表示捕捉传感器与物体之间的物理交互，该表示编码了传感器和物体属性。对于触觉理解，UniTac引入了两个任务，即物体属性描述和传感器识别，以增强对物理信息和跨传感器信息的推理。对于触觉生成，我们设计了一个由重建和对齐组成的两阶段训练范式，并结合基于传感器先验的采样策略，以模拟真实的触觉接触。在大规模多传感器数据集上训练后，UniTac在触觉理解方面达到了最先进的性能，并能够跨传感器生成真实的触觉信号。
cs.RO / 33 / 2606.31483
A Large-Language-Model Supported Personalized Driving Framework for Lane Change in Highway Scenarios
基于大语言模型的个性化驾驶框架在高速公路变道场景中的应用
Abstract
Personalized driving can improve the user acceptance of automated driving systems. However, existing methods still provide limited support for translating natural-language driving preferences, especially when such preferences are expressed implicitly, into executable and distinguishable driving behaviors. This paper proposes a large language model (LLM)-supported personalized driving framework for highway lane-change scenarios. The framework maps natural-language driving commands to executable planning parameters in the open-source Apollo automated driving stack according to three driving styles: aggressive, normal, and conservative. To establish this mapping, candidate planning parameters are evaluated based on the resulting lane-change behaviors, and style-specific parameter sets are constructed through clustering and style-intensity ranking. For command interpretation, a retrieval dataset is constructed to support retrieval-augmented generation (RAG), enabling LLM-based interpretation of implicit user commands. Experimental results show that the derived parameter sets generate distinguishable personalized lane-change behaviors, while RAG consistently improves preference interpretation, particularly for implicit commands. These results indicate the potential of integrating LLM-based natural-language interaction with Apollo to support personalized lane-change behavior generation. The source code and the relevant datasets are available at: https://github.com/ftgTUGraz/LLM-Personalized-Driving.
Chinese Translation
个性化驾驶能够提高用户对自动驾驶系统的接受度。然而，现有方法在将自然语言驾驶偏好（尤其是隐性表达的偏好）转化为可执行且可区分的驾驶行为方面仍然支持有限。本文提出了一种基于大语言模型（LLM）的个性化驾驶框架，适用于高速公路变道场景。该框架根据三种驾驶风格（激进、正常和保守）将自然语言驾驶指令映射到开源Apollo自动驾驶堆栈中的可执行规划参数。为建立这种映射，候选规划参数根据生成的变道行为进行评估，并通过聚类和风格强度排名构建特定风格的参数集。为了实现指令解释，构建了一个检索数据集以支持检索增强生成（RAG），使得基于LLM的隐性用户指令解释成为可能。实验结果表明，所获得的参数集能够生成可区分的个性化变道行为，而RAG在偏好解释方面始终表现出改善，尤其是对于隐性指令。这些结果表明，将基于LLM的自然语言交互与Apollo集成以支持个性化变道行为生成的潜力。源代码和相关数据集可在以下链接获取：https://github.com/ftgTUGraz/LLM-Personalized-Driving。
cs.RO / 34 / 2606.31487
Energy-Optimal Spatial Iterative Learning within a Virtual Tube
虚拟管道中的能量最优空间迭代学习
Abstract
Due to the limited endurance of embedded energy sources such as lithium-polymer (LiPo) batteries, the flight duration and operational range of unmanned aerial vehicles (UAVs) are severely constrained. Although energy-efficient trajectory planning and control have been widely studied, most existing approaches rely on accurate system models and computationally expensive optimization procedures. This paper proposes a model-free online iterative learning (IL) framework to minimize energy consumption. Without requiring explicit models of UAV dynamics or energy consumption, the proposed method improves energy efficiency while maintaining a low computational cost. The per-iteration computational complexity is O(n), where n denotes the number of path points. In the tested cases, the proposed method is approximately 50--60 times faster than the model-based IPOPT benchmark. Simulation results and real-world flight experiments across multiple UAV platforms validate the effectiveness, computational efficiency, and practical applicability of the proposed approach.
Chinese Translation
由于嵌入式能源源（如锂聚合物（LiPo）电池）的耐久性有限，无人机（UAV）的飞行时间和操作范围受到严重限制。尽管能量高效的轨迹规划和控制已被广泛研究，但大多数现有方法依赖于准确的系统模型和计算成本高昂的优化过程。本文提出了一种无模型在线迭代学习（IL）框架，以最小化能量消耗。该方法无需明确的无人机动力学或能量消耗模型，在保持低计算成本的同时提高了能量效率。每次迭代的计算复杂度为 O(n)，其中 n 表示路径点的数量。在测试案例中，所提出的方法的速度约为基于模型的 IPOPT 基准的 50-60 倍。仿真结果和多个无人机平台的实际飞行实验验证了该方法的有效性、计算效率和实际适用性。
cs.RO / 35 / 2606.31493
ChronoFlow-Policy: Unifying Past-Current-Future Interaction Flow in Visuomotor Policy Learning
ChronoFlow-Policy：统一视觉运动策略学习中的过去-现在-未来交互流
Abstract
Visual signals play a crucial role in policy learning by enabling models to capture object motion and interaction dynamics. Just as humans reason about actions using both past experience and anticipated outcomes, effective policies should integrate past interactions with future predictions. However, existing visuomotor policies typically model either historical context or future dynamics in isolation, lacking a unified temporal representation of interaction dynamics. In this work, we introduce \textbf{ChronoFlow}, a temporally unified representation that captures \textbf{past, current, and future} interaction dynamics through sparse 3D keypoints of both objects and the gripper. Based on this representation, we propose \textbf{ChronoFlow-Policy}, a diffusion-based visuomotor policy that jointly learns ChronoFlow and action sequences through a co-training objective. Experiments on 14 simulated tasks and 5 real-world manipulation tasks demonstrate that ChronoFlow-Policy consistently outperforms strong diffusion-policy baselines and improves robustness in long-horizon and non-Markovian manipulation scenarios.
Chinese Translation
视觉信号在策略学习中发挥着至关重要的作用，使模型能够捕捉物体运动和交互动态。正如人类在推理行动时同时考虑过去的经验和预期的结果，有效的策略应当将过去的交互与未来的预测相结合。然而，现有的视觉运动策略通常孤立地建模历史背景或未来动态，缺乏交互动态的统一时间表示。在本研究中，我们引入了 extbf{ChronoFlow}，一种时间统一的表示，通过稀疏的3D关键点捕捉 extbf{过去、现在和未来}的交互动态，包括物体和夹持器的关键点。基于这一表示，我们提出了 extbf{ChronoFlow-Policy}，一种基于扩散的视觉运动策略，通过共同训练目标联合学习ChronoFlow和动作序列。在14个模拟任务和5个真实世界操作任务上的实验表明，ChronoFlow-Policy始终优于强大的扩散策略基线，并在长时间跨度和非马尔可夫操作场景中提高了鲁棒性。
cs.RO / 36 / 2606.31494
Robustness of Robotic Manipulation: Foundations and Frontiers
机器人操控的鲁棒性：基础与前沿
Abstract
Humans and animals exhibit remarkable robustness in physical manipulation, yet robots remain far behind. Progress toward human-level manipulation robustness is hindered by the absence of a unified and systematic understanding: different subfields frame robustness in distinct ways, often leaving the concept ambiguous and limiting deeper analysis as well as communication across research areas. This paper presents a systematic study of manipulation robustness. We begin with a formal definition, characterizing robustness as the degree to which a manipulation system can achieve its goal in the presence of uncertainty and variation. Building on this definition, we introduce general formulations of manipulation robustness from probabilistic and control-theoretic perspectives. We then synthesize the guiding principles and concrete mechanisms of manipulation robustness across perception, planning, control, policy learning, and hardware, illustrating each mechanism through representative works, including foundational and recent studies. In addition, we revisit existing metrics and evaluation methods for quantifying manipulation robustness. Finally, we distill broader lessons for designing robust manipulation systems and discuss open problems and future directions toward achieving human-level robustness in robotic manipulation.
Chinese Translation
人类和动物在物理操控方面表现出显著的鲁棒性，但机器人仍远远落后。朝着人类水平的操控鲁棒性发展的进程受到缺乏统一和系统理解的阻碍：不同的子领域以不同的方式框定鲁棒性，常常使这一概念模糊不清，从而限制了更深入的分析以及跨研究领域的交流。本文对操控鲁棒性进行了系统研究。我们首先给出一个正式定义，将鲁棒性表征为操控系统在不确定性和变化存在的情况下实现其目标的能力。基于这一定义，我们从概率论和控制理论的角度引入操控鲁棒性的一般公式。接着，我们综合了感知、规划、控制、策略学习和硬件等领域的操控鲁棒性的指导原则和具体机制，通过代表性工作（包括基础性和近期研究）来阐明每个机制。此外，我们重新审视了现有的量化操控鲁棒性的指标和评估方法。最后，我们提炼出设计鲁棒操控系统的更广泛经验教训，并讨论实现机器人操控人类水平鲁棒性的开放问题和未来方向。
cs.RO / 37 / 2606.31497
Communication-Aware Robot Execution for Cloud Inference under Spatially Heterogeneous Connectivity
考虑通信的机器人执行在空间异构连接下的云推理
Abstract
Cloud-hosted foundation models enable robots to use semantic reasoning beyond onboard computational limits. In this setting, the robot executes a currently available primitive generated by the cloud, and continued task progress requires the next cloud result before this primitive is exhausted. This execution becomes fragile under spatially heterogeneous connectivity, because the current primitive determines when the next result is needed, whereas the wireless environment determines where the next request can be submitted and where the response can be retrieved. Strategies that reduce latency or improve individual transmissions can shorten this dependency, but they do not determine a submission location that supports reliable upload and leaves a feasible opportunity for response retrieval. To address this problem, we introduce the request--response window, which characterizes the time required for the next cloud cycle, including uplink transmission, cloud inference, downlink retrieval, and inference uncertainty. Building on this window and an available communication map, the proposed framework treats the next request point as a motion decision during ongoing primitive execution, selecting it to provide sufficient communication quality for cloud request submission while preserving progress within the finite support of the current primitive. The selected request point is incorporated into a local planner, which guides the robot toward the request point before submission and then continues task execution while maintaining sufficient connectivity for retrieving the next cloud result. Experiments in an indoor wireless scenario built from measurements show that the proposed method achieves the best or tied-best task success among the compared methods, while using fewer request attempts and producing lower request failure rates.
Chinese Translation
云托管的基础模型使机器人能够超越机载计算限制进行语义推理。在这种情况下，机器人执行由云生成的当前可用原语，而任务的持续进展需要在该原语耗尽之前获得下一个云结果。在空间异构连接下，这种执行变得脆弱，因为当前原语决定了何时需要下一个结果，而无线环境决定了可以提交下一个请求的位置以及可以检索响应的位置。减少延迟或改善单个传输的策略可以缩短这种依赖关系，但它们并未确定一个支持可靠上传并为响应检索留出可行机会的提交位置。为了解决这个问题，我们引入了请求-响应窗口，该窗口表征了下一个云周期所需的时间，包括上行传输、云推理、下行检索和推理不确定性。在这个窗口和可用通信地图的基础上，所提出的框架将下一个请求点视为正在进行的原语执行过程中的运动决策，选择它以提供足够的通信质量以进行云请求提交，同时保持当前原语的有限支持内的进展。所选请求点被纳入本地规划器，该规划器在提交之前引导机器人朝请求点移动，然后在保持足够连接以检索下一个云结果的同时继续任务执行。在基于测量构建的室内无线场景中的实验表明，所提出的方法在比较方法中实现了最佳或并列最佳的任务成功率，同时使用更少的请求尝试并产生更低的请求失败率。
cs.RO / 38 / 2606.31562
Stabilization Learning: A Paradigm Transition Bridging Control Theory and Machine Learning
稳定性学习：连接控制理论与机器学习的范式转变
Abstract
Stabilization learning is an interdisciplinary paradigm that bridges control theory and machine learning. Its core idea is to enable systems to adjust their policies under perturbations or environmental changes through real-time feedback and adaptive mechanisms. It takes stability as its primary goal, distinguishing itself from certificate learning, which focuses on formal proofs, and reinforcement learning, which pursues optimality. It encompasses a range of methods, including Lyapunov-based analysis and design, deep feature extraction, and data-driven feedback synthesis, and is applicable to complex high-dimensional, nonlinear systems. This paper elaborates on the two major categories of stability in stabilization learning, as well as three typical application scenarios: control, observation, and recognition. It constructs a unified mathematical framework based on a six-tuple, and expands into two types of seven-tuple models: constrained learning with barrier spaces and tracking problems with targets. It also analyzes the roles, meanings, and implementation choices of key elements such as state space, controlled system, metrics, and policy. Through the formal reformulation of 11 types of problems, including multi-agent cooperative tracking, visual servo robot position stabilization, chess games, and Push-T tasks, this paper illustrates the potential applicability of the framework across multiple domains. Finally, it points out that future stabilization learning will focus on two major directions: constructing a unified problem framework and achieving efficient and robust learning, providing solutions for complex system control that combine theoretical rigor with engineering practicality.
Chinese Translation
稳定性学习是一种跨学科的范式，连接了控制理论与机器学习。其核心思想是通过实时反馈和自适应机制，使系统能够在扰动或环境变化下调整其策略。它以稳定性为主要目标，与专注于形式证明的证书学习和追求最优性的强化学习有所区别。该方法涵盖了一系列技术，包括基于Lyapunov的分析与设计、深度特征提取以及数据驱动的反馈合成，适用于复杂的高维非线性系统。本文详细阐述了稳定性学习中稳定性的两大主要类别，以及控制、观察和识别三种典型应用场景。构建了一个基于六元组的统一数学框架，并扩展为两种类型的七元组模型：具有障碍空间的约束学习和带目标的跟踪问题。同时分析了状态空间、受控系统、度量和策略等关键元素的角色、意义和实现选择。通过对包括多智能体协作跟踪、视觉伺服机器人位置稳定、棋类游戏和Push-T任务在内的11种问题的形式化重构，本文展示了该框架在多个领域的潜在适用性。最后指出，未来的稳定性学习将集中在两个主要方向：构建统一的问题框架和实现高效且稳健的学习，为结合理论严谨性与工程实用性的复杂系统控制提供解决方案。
cs.RO / 39 / 2606.31654
DynFly: Dynamic-Aware Continuous Trajectory Generation for UAV Vision-Language Navigation in Urban Environments
DynFly：面向动态的城市环境无人机视觉-语言导航的连续轨迹生成
Abstract
Recent advances in multimodal large models have significantly improved UAV vision-language navigation (UAV-VLN) by enhancing high-level perception and reasoning. However, existing methods mainly focus on predicting discrete actions, local targets, or sparse waypoints, while the continuous transition from navigation intent to executable UAV motion remains weakly modeled. This motion-interface gap limits the continuity, stability, and executability of generated UAV trajectories. To address this gap, we propose DynFly, a dynamic-aware continuous trajectory generation framework that bridges high-level navigation reasoning and executable UAV motion. DynFly bridges high-level navigation intent and continuous UAV motion through a lightweight trajectory generation layer. Specifically, it represents expert trajectories in B-spline control-point space and employs a Spline-DiT generator to learn conditional trajectory generation via flow matching. Furthermore, we introduce UAV-oriented dynamic-aware supervision over position, finite-difference velocity, finite-difference acceleration, heading consistency, and local target alignment, enabling the generated trajectories to better satisfy UAV motion characteristics. And our trajectory generation framework can also be integrated with an existing UAV-VLN framework while preserving its original visual-language reasoning pipeline. Extensive experiments on the OpenUAV UAV-VLN benchmark show that DynFly improves both navigation performance and trajectory quality. On the Test Unseen Full split, DynFly improves the strongest baseline by 4.69 NDTW, 2.40 SDTW, 2.14 SR points and 4.87 OSR points, while reducing NE by 4.51 m.
Chinese Translation
近年来，多模态大模型的进展显著提升了无人机视觉-语言导航（UAV-VLN），增强了高层次的感知和推理能力。然而，现有方法主要集中于预测离散动作、局部目标或稀疏航点，而从导航意图到可执行无人机运动的连续过渡仍然建模不足。这种运动接口的差距限制了生成的无人机轨迹的连续性、稳定性和可执行性。为了解决这一问题，我们提出了DynFly，一个面向动态的连续轨迹生成框架，旨在弥合高层次导航推理与可执行无人机运动之间的差距。DynFly通过一个轻量级的轨迹生成层，将高层次的导航意图与连续的无人机运动连接起来。具体而言，它在B样条控制点空间中表示专家轨迹，并采用Spline-DiT生成器通过流匹配学习条件轨迹生成。此外，我们引入了面向无人机的动态感知监督，包括位置、有限差分速度、有限差分加速度、航向一致性和局部目标对齐，使生成的轨迹更好地满足无人机运动特性。我们的轨迹生成框架还可以与现有的UAV-VLN框架集成，同时保留其原有的视觉-语言推理流程。在OpenUAV UAV-VLN基准上的大量实验表明，DynFly在导航性能和轨迹质量上都有所提升。在测试未见全分割上，DynFly将最强基线提高了4.69 NDTW、2.40 SDTW、2.14 SR点和4.87 OSR点，同时将NE减少了4.51米。
cs.RO / 40 / 2606.31682
HABIT: Human-Aware Behavior and Interaction Training Dataset for Robot Manipulation
HABIT：面向机器人操作的人类意识行为与互动训练数据集
Abstract
Large-scale demonstration datasets have been central to recent progress in general-purpose robot policies. However, existing datasets are collected in human-absent settings, and policies trained on such data may perform tasks competently in isolation but fail to exhibit human-aware behaviors. To address this gap, we introduce HABIT, a large-scale robot demonstration dataset for human-present environments. We organize tasks into three roles capturing distinct modes of human-robot interaction: Collaborator, where human and robot jointly accomplish a task; Coworker, where they pursue separate tasks in a shared space; and Supervisor, where the human directs the robot. The dataset comprises over 10K episodes and over 160 hours across 60 tasks. Our experiments show that training on human-present data elicits human-aware behaviors that robot-only data fails to produce: spatiotemporal synchronization in Collaborator tasks, yielding in Coworker tasks, and gesture grounding in Supervisor tasks. Moreover, training on HABIT enables rapid adaptation to new human-robot interaction tasks. By introducing human presence as a new axis of dataset diversity, HABIT extends robot policies to environments shared with humans.
Chinese Translation
大规模演示数据集在通用机器人策略的最新进展中发挥了核心作用。然而，现有数据集是在没有人类参与的环境中收集的，基于这些数据训练的策略可能在孤立情况下能够胜任任务，但无法展现人类意识的行为。为了解决这一问题，我们引入了HABIT，一个针对人类在场环境的大规模机器人演示数据集。我们将任务组织为三种角色，以捕捉人机互动的不同模式：合作者（Collaborator），人类与机器人共同完成任务；同事（Coworker），他们在共享空间中追求各自的任务；监督者（Supervisor），人类指导机器人。该数据集包含超过10,000个情节和超过160小时的60个任务。我们的实验表明，在人类在场的数据上训练能够引发机器人仅在机器人数据中无法产生的人类意识行为：在合作者任务中的时空同步，在同事任务中的让步，以及在监督者任务中的手势理解。此外，在HABIT上训练使得机器人能够快速适应新的人人机互动任务。通过将人类存在引入数据集多样性的新维度，HABIT扩展了机器人策略在与人类共享环境中的应用。
cs.RO / 41 / 2606.31691
FastDSAC: Enhancing Policy Plasticity via Constrained Exploration for Scalable Humanoid Locomotion
FastDSAC：通过受限探索增强可扩展人形运动的策略可塑性
Abstract
Scalable reinforcement learning has popularized high-throughput sampling architectures, which significantly compresses the training time for off-policy methods in robotic locomotion. However, the rapid increase of data volume and update frequency undermines the stability of value-based methods and diminishes the plasticity of policy networks. To address these challenges, this work presents FastDSAC, a fast and high-performance variant of the Distributional Actor-Critic algorithm designed for parallel sampling scenarios. Specifically, we introduce a truncated Gaussian distribution to approximate the learned policy, which effectively excludes out-of-distribution actions that strain target value estimation while keeping necessary stochasticity for exploration. The proposed action constraint functions as an implicit regularization, which counteracts the plasticity loss typically caused by aggressive gradient updates. This preservation of network adaptability enhances sample efficiency, particularly in scenarios with a high update-to-data ratio, and accelerates the early training process. In contrast to prior fast reinforcement learning approaches that rely on discrete value distributions, our method utilizes a continuous Gaussian representation equipped with adaptive variance regulation, which improves value estimation accuracy by sampling confident and informative transitions. Extensive experiments on MuJoCo Playground and HumanoidBench demonstrate that FastDSAC not only stabilizes the overall training process but also achieves superior asymptotic performance and faster convergence compared to state-of-the-art baselines.
Chinese Translation
可扩展强化学习普及了高吞吐量采样架构，这显著压缩了机器人运动中离线策略方法的训练时间。然而，数据量和更新频率的快速增加削弱了基于价值的方法的稳定性，并降低了策略网络的可塑性。为了解决这些挑战，本文提出了FastDSAC，这是一种快速且高性能的分布式演员-评论家（Distributional Actor-Critic）算法变体，旨在并行采样场景中使用。具体而言，我们引入了截断高斯分布来近似学习到的策略，有效排除了那些对目标价值估计造成压力的超出分布的动作，同时保持必要的随机性以便于探索。所提出的动作约束函数作为一种隐式正则化，抵消了通常由于激进梯度更新而导致的可塑性损失。这种网络适应性的保持提高了样本效率，特别是在更新与数据比率较高的场景中，加快了早期训练过程。与依赖离散价值分布的先前快速强化学习方法相比，我们的方法利用了配备自适应方差调节的连续高斯表示，通过采样自信且信息丰富的过渡来提高价值估计的准确性。在MuJoCo Playground和HumanoidBench上的大量实验表明，FastDSAC不仅稳定了整体训练过程，还在与最先进的基线相比中实现了更优的渐近性能和更快的收敛速度。
cs.RO / 42 / 2606.31694
RCT: A Robot-Collected Touch-Vision-Language Dataset for Tactile Generalization
RCT：一个用于触觉泛化的机器人收集触觉-视觉-语言数据集
Abstract
For robots manipulating open-world objects, tactile representations must generalize to unseen materials. We introduce RCT (Robotic Contact Tactile), a robot-collected touch-vision-language dataset with 29,279 tactile frames from full robot presses on 122 industrial reference materials in 7 categories, recorded with three DIGIT sensors at multiple contact positions. RCT preserves each press as a contact sequence, enabling held-out evaluation across materials, categories, sensors, contact positions, and contact sequences. Frames from one press are strongly correlated: frame-random splits can place near-duplicate observations of the same physical interaction in both training and test. With the encoder held fixed, removing contact-sequence overlap reduces tactile-to-text Recall@1 by 17.7 percentage points. When materials are additionally held out at training time, performance drops sharply, leaving held-out-material Recall@1 at 25.1 +/- 6.1% averaged over three held-out draws. The public TVL/HCT split shows the same structure: every test contact sequence appears in training, and raw-pixel nearest neighbors recover the correct sequence in 98.3% of cases. Uniformly sampling a press improves contrastive training, and RCT-trained embeddings improve category probes on unseen materials. RCT makes contact-sequence-aware, held-out-material evaluation reproducible and exposes novel-material generalization as a central challenge for robotic tactile perception. The RCT dataset is open-sourced at https://faerber-lab.github.io/RCT/
Chinese Translation
对于操控开放世界物体的机器人而言，触觉表示必须能够泛化到未见过的材料。我们介绍了RCT（Robotic Contact Tactile），这是一个机器人收集的触觉-视觉-语言数据集，包含来自122种工业参考材料的29,279个触觉帧，这些材料分为7个类别，使用三种DIGIT传感器在多个接触位置记录全机器人按压的过程。RCT将每次按压保留为一个接触序列，使得在材料、类别、传感器、接触位置和接触序列之间的留出评估成为可能。来自一次按压的帧之间存在强相关性：帧随机分割可能会将同一物理交互的近重复观察放置在训练和测试集中。当固定编码器时，去除接触序列重叠会使触觉到文本的Recall@1下降17.7个百分点。当在训练时额外留出材料时，性能急剧下降，留出材料的Recall@1在三次留出抽样中平均为25.1 +/- 6.1%。公共的TVL/HCT分割显示出相同的结构：每个测试接触序列都出现在训练集中，原始像素最近邻在98.3%的情况下恢复了正确的序列。均匀采样一次按压改善了对比训练，而RCT训练的嵌入在未见材料上的类别探测中表现更佳。RCT使得接触序列感知的留出材料评估可重复，并将新材料泛化暴露为机器人触觉感知的一个核心挑战。RCT数据集已开源，链接为：https://faerber-lab.github.io/RCT/
cs.RO / 43 / 2606.31723
UniTacVLA: Unified Tactile Understanding and Prediction in Vision Language Action Models
UniTacVLA：视觉语言动作模型中的统一触觉理解与预测
Abstract
Vision-language-action (VLA) models have achieved strong performance in many robotic manipulation tasks, yet remain limited in contact-rich dexterous manipulation. To overcome this limitation, recent vision-tactile-language-action (VTLA) methods incorporate tactile sensing into VLA models to provide direct contact information. However, they typically treat tactile signals as passive auxiliary inputs, making it difficult to model tactile semantics and future physical interactions. To this end, we propose a unified tactile learning framework for contact-rich manipulation that models tactile signals as dynamic interaction cues for both contact understanding and prediction. Specifically, we construct a unified tactile latent space and jointly model current tactile states and future contact changes through tactile chain-of-thought reasoning and coarse-to-fine future tactile prediction, thereby forming a state-aware and dynamics-aware tactile prior. Based on this prior, we introduce a tactile-action mixed controller that combines real-time and predicted tactile feedback to refine low-frequency action chunks with high-frequency corrections. Real-world experiments on four categories of contact-rich tasks, including adjustment, insertion, wiping, and assembly, under both clean and externally perturbed settings, show that our method improves success rate, manipulation accuracy, and contact robustness over existing methods, demonstrating its effectiveness in dexterous physical interaction.
Chinese Translation
视觉语言动作（VLA）模型在许多机器人操作任务中表现出色，但在接触丰富的灵巧操作中仍然存在局限性。为克服这一限制，近期的视觉-触觉-语言-动作（VTLA）方法将触觉感知纳入VLA模型，以提供直接的接触信息。然而，它们通常将触觉信号视为被动辅助输入，这使得建模触觉语义和未来物理交互变得困难。为此，我们提出了一种统一的触觉学习框架，用于接触丰富的操作，将触觉信号建模为动态交互线索，以实现接触理解和预测。具体而言，我们构建了一个统一的触觉潜在空间，通过触觉链式思维推理和粗到细的未来触觉预测，共同建模当前触觉状态和未来接触变化，从而形成一个状态感知和动态感知的触觉先验。在此先验基础上，我们引入了一种触觉-动作混合控制器，结合实时和预测的触觉反馈，以高频修正来优化低频动作片段。在四类接触丰富任务（包括调整、插入、擦拭和组装）的真实世界实验中，无论是在干净环境还是外部干扰下，我们的方法在成功率、操作准确性和接触鲁棒性方面均优于现有方法，证明了其在灵巧物理交互中的有效性。
cs.RO / 44 / 2606.31772
Autonomous UAV Navigation for Individual Wildlife Re-Identification
用于个体野生动物重新识别的自主无人机导航
Abstract
Reliable individual re-identification (re-ID) of wildlife is essential for population monitoring, behavioral tracking, and conservation policy evaluation, yet large-scale data collection remains labor-intensive, relying on manual efforts by ecologists or citizen scientists. We propose an autonomous drone navigation system that actively optimizes image capture for downstream re-ID, moving beyond passive aerial sensing. The system combines YOLOv11 object detection with a DINOv2-based pose classifier to guide real-time flight decisions: detecting animals, orienting to expose the lateral flank (the surface of interest for pattern-based re-ID), and approaching until the subject meets a minimum bounding-box threshold. Unlike prior drone systems that optimize for group-level behavioral video, ours targets the specific image-quality requirements of individual-identification models. We demonstrate feasibility through a case study on zebra using footage collected in Kenya, and show the approach generalizes to other species with diagnostic surface patterns, including giraffes, tigers, and elephants. Our work establishes a framework for task-aware embodied AI for ecological data collection, in which downstream re-ID requirements drive real-time perception and control.
Chinese Translation
可靠的个体重新识别（re-ID）对野生动物的种群监测、行为追踪和保护政策评估至关重要，但大规模数据收集仍然劳动密集，依赖生态学家或公民科学家的手动努力。我们提出了一种自主无人机导航系统，该系统主动优化图像捕获以支持后续的重新识别，超越了被动的空中感知。该系统结合了YOLOv11目标检测与基于DINOv2的姿态分类器，以指导实时飞行决策：检测动物，调整方向以暴露侧面（这是基于模式的重新识别所需的关注表面），并接近直到目标达到最小边界框阈值。与之前优化群体行为视频的无人机系统不同，我们的系统针对个体识别模型的特定图像质量要求。我们通过在肯尼亚收集的斑马视频进行案例研究，展示了该方法的可行性，并表明该方法可以推广到其他具有诊断表面模式的物种，包括长颈鹿、老虎和大象。我们的工作建立了一个以任务为导向的具身人工智能框架，用于生态数据收集，其中下游的重新识别需求驱动实时感知和控制。
cs.RO / 45 / 2606.31807
Reinforcement Learning-Based Control for an Inline Skating Humanoid Robot
基于强化学习的直排滑冰类人机器人控制
Abstract
As humanoid robots become increasingly dynamic, coupling them with reinforcement learning offers a promising approach to solving the complex, underactuated mechanics of passive inline skating. Equipping a humanoid robot with passive inline skating wheels presents an opportunity to combine the versatile agility of humanoids with the high-speed, energy-efficient locomotion strategies utilized by human skaters. In this paper, we train and deploy a reinforcement learning control policy that enables novel locomotion strategies for a humanoid robot modified to equip consumer inline skates instead of conventional feet. Unlike previous work limited to quadrupedal robots or actively driven wheels, our system allows for precise 6-DoF control of the skates to execute dynamic, edge-driven propulsion strategies. Our skating strategies emerge entirely from our reward structure, without reliance on human motion data, imitation learning, or kinematic priors. We overcome the inherent instability of passive wheels and simulation contact artifacts by utilizing different geometric wheel models (spherical and ellipsoidal) during training and validation, along with a custom success-based command curriculum and a specialized rolling reward. Consequently, our policy demonstrates up to a 50% reduction in Cost of Transport (CoT) compared to standard walking gaits. The resulting policy successfully transfers zero-shot to the physical Booster T1 hardware. Real-world deployments demonstrate dynamic balance, the ability to reject active physical perturbations, and agile locomotion strategies capable of turning at speed. A video of our results can be found at https://www.youtube.com/watch?v=-_APcOS7uFo.
Chinese Translation
随着类人机器人变得越来越动态，将其与强化学习结合提供了一种解决被动直排滑冰复杂欠驱动力学的有前景的方法。为类人机器人配备被动直排滑冰轮子为将类人机器人的多功能灵活性与人类滑冰者所采用的高速、节能的运动策略相结合提供了机会。在本文中，我们训练并部署了一种强化学习控制策略，使得经过改装的类人机器人能够采用消费者直排滑冰鞋而非传统的脚进行新颖的运动策略。与以往仅限于四足机器人或主动驱动轮子的研究不同，我们的系统允许对滑冰鞋进行精确的六自由度控制，以执行动态的边缘驱动推进策略。我们的滑冰策略完全源自我们的奖励结构，而不依赖于人类运动数据、模仿学习或运动学先验。我们通过在训练和验证过程中使用不同的几何轮子模型（球形和椭球形），以及定制的基于成功的指令课程和专门的滚动奖励，克服了被动轮子的固有不稳定性和模拟接触伪影。因此，我们的策略在运输成本（Cost of Transport, CoT）上相比标准步态减少了多达50%。最终的策略成功实现了零样本迁移到物理的Booster T1硬件上。现实世界的部署展示了动态平衡、抵抗主动物理扰动的能力，以及能够快速转弯的灵活运动策略。我们的结果视频可以在https://www.youtube.com/watch?v=-_APcOS7uFo找到。
cs.RO / 46 / 2606.31836
RoboTacDex: A Dexterous Visual-Tactile-Action Dataset for Humanoid Manipulation
RoboTacDex：一个用于类人操作的灵巧视觉-触觉-动作数据集
Abstract
In the field of robot learning, large-scale and diverse demonstration trajectories provide the fundamental basis for enhancing robotic manipulation ability. We introduce RoboTacDex, a large, multi-modal, and diverse dataset of dexterous manipulation behaviors performed with a humanoid robot. Built on the publicly accessible humanoid robot Unitree G1, RoboTacDex consists of 6k trajectories covering 19 tasks, 23 skills, and interactions with 22 objects. RoboTacDex provides comprehensive records including multi-view RGB and depth information, tactile feedback, and detailed semantic annotations. Furthermore, the dataset features a variety of relatively challenging tasks that can only be completed by dual arms and dexterous hands, aiming to mimic human-like operational logic and simulate real-world manipulation complexity. To ensure data collection quality, we develop an improved multi-camera synchronization system to enable millisecond data synchronization and recording of modalities. In our experiments, we evaluate three representative imitation learning models on our dataset, analyzing their performance as well as their respective strengths and limitations across different task categories. Successful trial results and a moderate level of generalization capabilities across a suite of tasks indicate the effectiveness and diversity of the collected dataset. Our dataset will be open-sourced soon.
Chinese Translation
在机器人学习领域，大规模和多样化的演示轨迹为增强机器人操作能力提供了基础。我们介绍了RoboTacDex，这是一个由类人机器人执行的灵巧操作行为的大型、多模态和多样化的数据集。RoboTacDex建立在公开可获取的类人机器人Unitree G1之上，包含6000条轨迹，涵盖19个任务、23项技能以及与22个物体的交互。RoboTacDex提供了全面的记录，包括多视角的RGB和深度信息、触觉反馈以及详细的语义注释。此外，该数据集还包含一系列相对具有挑战性的任务，这些任务只能通过双臂和灵巧的手完成，旨在模拟类人操作逻辑并重现现实世界操作的复杂性。为了确保数据收集的质量，我们开发了一种改进的多摄像头同步系统，以实现毫秒级的数据同步和多模态记录。在我们的实验中，我们在数据集上评估了三种代表性的模仿学习模型，分析了它们在不同任务类别中的表现以及各自的优缺点。成功的试验结果和在一系列任务中的适度泛化能力表明了所收集数据集的有效性和多样性。我们的数据集将很快开源。
cs.RO / 47 / 2606.31844
Bridging Local Observation and Global Simulation in Closed-Loop Traffic Modeling
弥合闭环交通建模中的局部观察与全球仿真
Abstract
A local-to-global context mismatch arises when autoregressive traffic simulators trained on ego-centric driving logs are deployed in globally observable closed-loop environments. In such logs, the ego vehicle has rich local observations, while surrounding agents are only partially observed due to perception limits and occlusions. As a result, simulators may learn incomplete context--action mappings that remain hidden in log-based training but emerge during closed-loop rollouts, leading to unrealistic behaviors such as abnormal stops, unsafe interactions, and rule violations. We propose CRAFT, a Contextual pReference Alignment Framework for Traffic Simulation, to mitigate this mismatch via self-supervised failure discovery and preference-guided test-time alignment. CRAFT treats the base simulator as a globally observable sandbox, generating diverse what-if rollouts from logged initial states to expose context-induced failures. These failures are grounded with human-aligned driving priors and converted into preference supervision for training a Contextual Preference Evaluator (CPE). At inference time, CPE acts as a plug-in alignment module that scores candidate actions under complete scene context and reweights autoregressive decoding toward globally coherent behaviors. CRAFT mitigates this local-to-global contextual bias, reducing collisions by 31.2\% and traffic violations by 33.2\% without retraining the base simulator.
Chinese Translation
当基于自回归的交通模拟器在全球可观察的闭环环境中部署时，会出现局部与全球上下文不匹配的问题。这些模拟器是在以自我为中心的驾驶日志上训练的，在这些日志中，自我车辆拥有丰富的局部观察，而周围的代理由于感知限制和遮挡，仅能部分观察。因此，模拟器可能学习到不完整的上下文-动作映射，这些映射在基于日志的训练中是隐藏的，但在闭环回放中会显现出来，导致不现实的行为，如异常停车、不安全的交互和规则违反。我们提出了CRAFT（交通仿真的上下文偏好对齐框架），通过自我监督的失败发现和偏好引导的测试时对齐来缓解这种不匹配。CRAFT将基础模拟器视为一个全球可观察的沙盒，从记录的初始状态生成多样的假设回放，以揭示上下文引起的失败。这些失败与人类对齐的驾驶先验相结合，并转化为用于训练上下文偏好评估器（CPE）的偏好监督。在推理时，CPE作为一个插件对齐模块，根据完整场景上下文对候选动作进行评分，并重新加权自回归解码，以朝向全球一致的行为。CRAFT减轻了这种局部到全球的上下文偏见，减少了31.2%的碰撞和33.2%的交通违规，而无需重新训练基础模拟器。
cs.RO / 48 / 2606.31846
Z-1: Efficient Reinforcement Learning for Vision-Language-Action Models
Z-1：面向视觉-语言-动作模型的高效强化学习
Abstract
Vision-Language-Action (VLA) models offer a promising framework for robotic manipulation by connecting language instructions, visual observations, and continuous control. However, most existing policies remain limited by behavior cloning or supervised fine-tuning (SFT) from fixed demonstrations, which provides limited opportunity to improve from the policy's own failures. In this paper, we present Z-1, a reinforcement learning (RL) post-training framework for flow-based VLA models. Built on top of $\pi_{0.5}$, Z-1 uses only publicly released RoboCasa demonstrations for SFT and then applies a task-wise Group Relative Policy Optimization (GRPO) strategy across $24$ standard RoboCasa tasks. To improve the efficiency and stability of online optimization, Z-1 combines shared-prefix rollout construction, tree-structured trajectory branching, completion-aware reward calibration, and selective joint training of VLM and Action Expert. Across all $24$ RoboCasa tasks, Z-1 achieves an average success rate of $80.6\%$, improving over its SFT initialization by $13.2\%$ points and outperforms the published sota models. These results show that systematic GRPO post-training can substantially improve flow-based VLA policies without additional private demonstrations.
Chinese Translation
视觉-语言-动作（VLA）模型通过连接语言指令、视觉观察和连续控制，为机器人操作提供了一个有前景的框架。然而，大多数现有策略仍然受限于行为克隆或从固定演示中进行的监督微调（SFT），这限制了从策略自身失败中改进的机会。本文提出了Z-1，一个针对基于流的VLA模型的强化学习（RL）后训练框架。Z-1建立在$ ext{π}_{0.5}$之上，仅使用公开发布的RoboCasa演示进行SFT，然后在$24$个标准RoboCasa任务中应用任务级的群体相对策略优化（GRPO）策略。为了提高在线优化的效率和稳定性，Z-1结合了共享前缀回放构建、树状轨迹分支、完成感知奖励校准和VLM与动作专家的选择性联合训练。在所有$24$个RoboCasa任务中，Z-1实现了平均成功率$80.6 ext{ extperthousand}$，比其SFT初始化提高了$13.2 ext{ extperthousand}$点，并且超越了已发布的最先进模型。这些结果表明，系统的GRPO后训练可以在没有额外私有演示的情况下显著改善基于流的VLA策略。
cs.RO / 49 / 2606.31889
Improving path-tracking performance of an articulated tractor-trailer system using a non-linear kinematic model
利用非线性运动学模型提高关节式拖拉机-挂车系统的路径跟踪性能
Abstract
This paper presents a novel non-linear mathematical model of an articulated tractor-trailer system that can be used, in combination with receding horizon techniques, to improve the performance of path tracking tasks of articulated systems. Due to its dual steering mechanisms, this type of vehicle can be very useful in precision agriculture, particularly for seeding, spraying and harvesting in small fields. The articulated tractor-trailer system model was embedded within a non-linear model predictive controller and the trailer position was monitored. When the kinematic of the trailer was considered, the deviation of trailer's position was reduced substantially alongside not only straight paths but also in headland turns. Using the proposed mathematical model, we were able to control the trailer's position itself rather than the tractor's position. The Robot Operating System (ROS) framework and Gazebo simulator were used to perform realistic simulations examples.
Chinese Translation
本文提出了一种新颖的关节式拖拉机-挂车系统的非线性数学模型，该模型可以与递归视界技术结合使用，以提高关节式系统的路径跟踪任务性能。由于其双重转向机制，这种类型的车辆在精细农业中非常有用，特别是在小型田地中的播种、喷洒和收割。关节式拖拉机-挂车系统模型被嵌入到非线性模型预测控制器中，并对挂车位置进行了监测。当考虑到挂车的运动学时，挂车位置的偏差在直线路径和头部转弯时均显著减少。利用所提出的数学模型，我们能够控制挂车的位置，而不是拖拉机的位置。我们使用了机器人操作系统（Robot Operating System，ROS）框架和Gazebo模拟器进行真实的仿真示例。
cs.RO / 50 / 2606.31909
CoDex: Learning Compositional Dexterous Functional Manipulation without Demonstrations
CoDex：无示范学习组合灵巧功能操作
Abstract
In this work, we study Compositional Dexterous Functional Object Manipulation (CD-FOM): tasks such as aiming and actuating a spray bottle on a plant or a glue gun on wood, which require both actuating an object's internal mechanism and controlling its pose to apply the object's function to the environment. These tasks pose significant challenges for robots due to the demanding integration of semantic understanding of the object's function, actuation mode, and application area with intricate physical dexterity to manage grasp stability, movement trajectory, and actuation. We introduce CoDex, a zero-demonstration framework that autonomously discovers CD-FOM manipulation strategies. CoDex uses vision-language models (VLMs) to infer semantic constraints from the task and scene. These constraints guide analytic constrained optimization to generate a short list of functional grasp candidates that can be efficiently refined with reinforcement learning to generate full grasp-move-actuate policies transferable from simulation to the real world. We evaluate CoDex on a 7-DoF robot arm with a 16-DoF multi-fingered hand across six CD-FOM tasks involving previously unseen objects with internal mechanisms, including spray bottles, hot glue guns, air dusters, flashlights, and pepper grinders, and their application to unseen target objects, showcasing its ability to autonomously discover and execute complex, physically viable dexterous behaviors without human demonstrations. More information at https://robin-lab.cs.utexas.edu/CoDex/.
Chinese Translation
在本研究中，我们探讨了组合灵巧功能物体操作（CD-FOM）：例如在植物上瞄准和操作喷雾瓶，或在木材上使用热熔胶枪等任务，这些任务不仅需要激活物体的内部机制，还需要控制其姿态，以将物体的功能应用于环境。这些任务对机器人提出了重大挑战，因为它们要求将对物体功能、激活模式和应用领域的语义理解与复杂的物理灵巧性结合起来，以管理抓取稳定性、运动轨迹和激活。我们提出了CoDex，一个零示范框架，能够自主发现CD-FOM操作策略。CoDex利用视觉-语言模型（VLMs）从任务和场景中推断语义约束。这些约束指导解析约束优化，生成一份功能抓取候选列表，该列表可以通过强化学习高效地进行细化，以生成可从仿真转移到现实世界的完整抓取-移动-激活策略。我们在一个具有7自由度的机器人手臂和一个具有16自由度的多指手上评估了CoDex，涵盖六个涉及以前未见物体（包括喷雾瓶、热熔胶枪、气尘器、手电筒和胡椒研磨器）及其应用于未见目标物体的CD-FOM任务，展示了其自主发现和执行复杂、物理可行的灵巧行为的能力，而无需人类示范。更多信息请访问 https://robin-lab.cs.utexas.edu/CoDex/.
cs.RO / 51 / 2606.31912
Learning Locomotion on Discrete Terrain via Minimal Proximity Sensing
通过最小化接近传感器学习离散地形上的运动
Abstract
Learning-based control has revolutionized dynamic locomotion, yet navigating unstructured terrain remains limited by a robot's incomplete awareness of imminent ground contact. While global perception systems such as LiDARs and depth cameras provide environmental context, they are frequently plagued by latencies, occlusions, and the high computational cost of dense geometric reconstruction. On the other hand, proprioceptive feedback is purely reactive, initiating corrections only after impact has occurred. This work explores embedding a minimal suite of low-cost, high-frequency infrared proximity sensors directly into the feet of a quadrupedal robot. These sensors provide "pre-contact" feedback that is robust to self-occlusions and significantly less computationally demanding than conventional vision-based pipelines. By integrating these localized signals into a reinforcement learning framework, we enable the robot to anticipate terrain discontinuities such as gaps and stepping stones that are problematic for traditional perception stacks due to occlusions or state estimation drift. We demonstrate that such sparse, near-field sensing can be reliably modeled in simulation and transferred to the real world with high fidelity. Experimental results show that local proximity sensing substantially improves traversal robustness over discrete terrain and offers a low-power, low-latency alternative or complement to complex global perception suites in unpredictable environments. For more information about results and methods, please see the project website: https://sites.google.com/view/foot-tof/home.
Chinese Translation
基于学习的控制技术已经彻底改变了动态运动的方式，但在非结构化地形中导航仍然受到机器人对即将接触地面的不完全感知的限制。虽然全球感知系统如激光雷达（LiDAR）和深度相机提供了环境背景，但它们常常受到延迟、遮挡以及密集几何重建的高计算成本的困扰。另一方面，身体感知反馈是纯粹的反应性，仅在发生碰撞后才启动修正。本文探讨将一套最小化的低成本、高频率的红外接近传感器直接嵌入四足机器人脚部。这些传感器提供“接触前”反馈，能够有效抵抗自我遮挡，并且在计算上显著低于传统视觉基础管道。通过将这些局部信号整合到强化学习框架中，我们使机器人能够预测地形的不连续性，例如由于遮挡或状态估计漂移而对传统感知系统造成问题的间隙和踏脚石。我们证明，这种稀疏的近场感知可以在仿真中可靠建模，并以高保真度转移到现实世界。实验结果表明，局部接近感知显著提高了在离散地形上的行进鲁棒性，并为不可预测环境中的复杂全球感知系统提供了一种低功耗、低延迟的替代或补充方案。有关结果和方法的更多信息，请访问项目网站：https://sites.google.com/view/foot-tof/home。
cs.RO / 52 / 2606.31919
MVP-Nav: Multi-layer Value Map Planner Navigator
MVP-Nav：多层价值地图规划导航器
Abstract
Zero-shot Object Goal Navigation (ZSON) with RGB-only perception poses a fundamental challenge for embodied agents, as the absence of explicit depth information introduces severe physical uncertainty and semantic-physical misalignment. Existing approaches either rely on high-level semantic reasoning without geometric grounding or learn end-to-end policies that lack explicit physical constraints, often resulting in semantically plausible but physically unsafe behaviors. In this paper, we propose MVP-Nav, a physical-aware RGB-only navigation framework that aligns perception, planning, and control with the real 3D world. MVP-Nav reconstructs explicit physical occupancy from monocular observations by leveraging 3D foundation models to project 2D semantic instances into 3D oriented bounding boxes, forming a global spatial semantic representation. To unify high-level semantic reasoning and low-level physical constraints, we introduce a Multi-layer Value Map (MVM) that integrates semantic priorities and reconstructed geometry into a shared cost space, enabling physically grounded geometric planning. Extensive experiments on zero-shot object navigation benchmarks demonstrate that MVP-Nav significantly outperforms existing depth-free methods, achieving state-of-the-art performance and validating that structured physical priors can effectively compensate for the absence of active depth sensors.
Chinese Translation
仅依赖RGB感知的零样本目标导航（ZSON）对具身智能体提出了根本性的挑战，因为缺乏明确的深度信息会引入严重的物理不确定性和语义-物理不一致。现有方法要么依赖于没有几何基础的高层语义推理，要么学习缺乏明确物理约束的端到端策略，通常导致语义上合理但在物理上不安全的行为。本文提出了MVP-Nav，一个关注物理的仅RGB导航框架，它将感知、规划和控制与真实的3D世界对齐。MVP-Nav通过利用3D基础模型从单目观测中重建明确的物理占用，将2D语义实例投影到3D定向边界框中，从而形成全球空间语义表示。为了统一高层语义推理和低层物理约束，我们引入了多层价值地图（MVM），将语义优先级和重建几何集成到共享成本空间中，从而实现物理基础的几何规划。在零样本目标导航基准上的大量实验表明，MVP-Nav显著优于现有的无深度方法，达到了最先进的性能，并验证了结构化物理先验能够有效补偿缺乏主动深度传感器的情况。
cs.RO / 53 / 2606.31941
LeCropFollow: Latent Space Planning for Navigation in Unstructured Crop Fields
LeCropFollow：用于非结构化农田导航的潜在空间规划
Abstract
Unstructured navigational features, such as irregular planting or discontinuities, remain the primary failure mode for under-canopy agricultural robots. Existing geometric approaches often fail in these scenarios because they compress high-dimensional visual data into deterministic spatial references, effectively discarding the uncertainty and semantic context required to navigate ambiguous terrain. To address this, we present LeCropFollow, a visual navigation framework that bypasses explicit geometric modeling in favor of a learned latent representation. By integrating a self-supervised semantic heatmap extractor with TD-MPC2, a Model-Based Reinforcement Learning (MBRL) planner, our system optimizes trajectories directly within a latent manifold. The framework operates over the uncompressed heatmap signal, preserving the semantic context that geometric reductions discard. We demonstrate that this representational shift enables zero-shot transfer from simplified simulation to the physical world without fine-tuning. Extensive field experiments in late-stage corn fields show that LeCropFollow matches state-of-the-art baselines in unstructured rows but significantly outperforms them in plantation gaps, achieving a 2.4x reduction in semantic failures compared to keypoint-based methods. These results suggest that latent planning offers a robust alternative to geometric estimation for operations in heterogeneous agricultural environments. Code, models, and data available: https://felipe-tommaselli.github.io/lecropfollow .
Chinese Translation
非结构化的导航特征，如不规则种植或间断，仍然是农田机器人在遮荫下工作的主要失效模式。现有的几何方法在这些场景中往往失败，因为它们将高维视觉数据压缩为确定性的空间参考，实际上丢弃了导航模糊地形所需的不确定性和语义上下文。为了解决这个问题，我们提出了LeCropFollow，一个视觉导航框架，它绕过显式的几何建模，采用学习的潜在表示。通过将自监督的语义热图提取器与基于模型的强化学习（Model-Based Reinforcement Learning, MBRL）规划器TD-MPC2相结合，我们的系统直接在潜在流形中优化轨迹。该框架在未压缩的热图信号上运行，保留了几何简化所丢弃的语义上下文。我们展示了这种表示的转变使得从简化的仿真到物理世界的零样本迁移成为可能，而无需微调。在晚期玉米田的广泛实地实验中，LeCropFollow在非结构化行中与最先进的基线相匹配，但在种植间隙中显著优于它们，与基于关键点的方法相比，实现了2.4倍的语义失败减少。这些结果表明，潜在规划为在异质农业环境中的操作提供了一种稳健的替代几何估计的方法。代码、模型和数据可用： https://felipe-tommaselli.github.io/lecropfollow 。
cs.RO / 54 / 2606.31948
RRT-Rope: A deterministic shortening approach for fast near-optimal path planning in large-scale uncluttered 3D environments
RRT-Rope：一种用于大规模无障碍三维环境中快速近似最优路径规划的确定性缩短方法
Abstract
Many path planning algorithms have been introduced so far, but most are costly, in path cost and in processing time, in large-scale uncluttered 3D environments such as underground mining stopes explored by an unmanned aerial vehicle (UAV). Rapidly-exploring Random Tree (RRT) algorithms are popular because of their probabilistic completeness and rapidity in finding a feasible path in single-query problems. Many of the algorithms (e.g. Informed RRT*, RRT#) developed to improve RRT need considerable time to converge in large environments. Shortcutting an RRT is an old idea that has been proven to outperform RRT variants. This paper introduces a new method, RRT-Rope, that aims at finding a near-optimal solution in a drastically shorter amount of time. The proposed approach benefits from fast computation of a feasible path with an altered version of RRT-connect, and post-processes it quickly with a deterministic shortcutting technique, taking advantage of intermediate nodes added to each branch of the tree. This paper presents simulations and statistics carried out to show the efficiency of RRT-Rope, which gives better results in terms of path cost and computation time than other popular RRT variations and shortening techniques in all our simulation environments, and is up to 70% faster than the next best algorithm in a representative stope.
Chinese Translation
迄今为止，已经提出了许多路径规划算法，但在大规模无障碍三维环境中（如无人机（UAV）探索的地下矿井停采区），大多数算法在路径成本和处理时间上都非常昂贵。快速探索随机树（Rapidly-exploring Random Tree, RRT）算法因其概率完备性和在单次查询问题中快速找到可行路径的能力而受到欢迎。许多为改善RRT而开发的算法（如Informed RRT*、RRT#）在大环境中收敛所需的时间相当可观。对RRT进行缩短是一种古老的理念，已被证明优于RRT变体。本文介绍了一种新方法RRT-Rope，旨在以显著更短的时间找到近似最优解。所提出的方法利用了经过修改的RRT-connect快速计算可行路径，并通过确定性缩短技术快速后处理，利用添加到树的每个分支的中间节点。本文展示了进行的仿真和统计，以证明RRT-Rope的效率，在所有仿真环境中，其在路径成本和计算时间方面的结果优于其他流行的RRT变体和缩短技术，并且在一个代表性的停采区中比第二好的算法快多达70%。
cs.RO / 55 / 2606.31958
Adapting Generalist Robot Policies with Semantic Reinforcement Learning
利用语义强化学习调整通用机器人策略
Abstract
Generalist robot policies learn a diverse repertoire of behaviors from large-scale pretraining. In principle, this makes them excellent priors for downstream adaptation via reinforcement learning (RL). In practice, however, standard RL methods leveraging this prior optimize directly over robot actions, requiring the base policy's action distribution to be close to that of a performant policy from the start. This assumption breaks down for complex or long-horizon tasks that fall outside the pretraining distribution. Our key insight is that, for sufficiently expressive generalist policies, language prompts are an effective alternative space for learning to solve such tasks: modulating language inputs elicits skills already within the policy's repertoire, which can be composed to solve tasks beyond its zero-shot capabilities. We propose Semantic Action Reinforcement Learning (SARL), which learns to optimize this prompt space through online interaction, treating the generalist policy as a controllable skill prior. Importantly, leveraging pretrained skills rather than learning new ones from scratch yields structured, semantically meaningful exploration and highly efficient online improvement, and learning to modulate prompts through experience grounds them in induced real-world behaviors for robust task-solving. Across real-world settings and simulated benchmarks, we show SARL unlocks fundamentally new capabilities -- adapting VLA behavior to solve complex, long-horizon tasks -- and significantly outperforms existing approaches for improving robot behavior in deployment.
Chinese Translation
通用机器人策略通过大规模预训练学习多样化的行为库。从原则上讲，这使得它们成为通过强化学习（RL）进行下游适应的优秀先验。然而，在实践中，利用这一先验的标准RL方法直接在机器人动作上进行优化，要求基础策略的动作分布从一开始就接近高效策略的动作分布。这一假设在复杂或长时间跨度的任务中失效，这些任务超出了预训练分布。我们的关键见解是，对于足够表达能力强的通用策略，语言提示是学习解决此类任务的有效替代空间：调节语言输入可以引发策略库中已有的技能，这些技能可以组合以解决超出其零-shot能力的任务。我们提出了语义动作强化学习（SARL），该方法通过在线交互学习优化这一提示空间，将通用策略视为可控的技能先验。重要的是，利用预训练技能而不是从头学习新技能能够产生结构化、语义上有意义的探索，并实现高效的在线改进，通过经验学习调节提示使其与诱导的现实世界行为相结合，从而实现稳健的任务解决。在真实世界环境和模拟基准中，我们展示了SARL解锁了根本新的能力——调整VLA行为以解决复杂的长时间跨度任务——并显著优于现有的改善机器人行为的部署方法。
cs.RO / 56 / 2606.31993
OopsieVerse: A Safety Benchmark with Damage-Aware Simulation for Robot Manipulation
OopsieVerse：一种具有损伤感知模拟的机器人操作安全基准
Abstract
While robotic manipulation capabilities have advanced rapidly, physical safety remains a major barrier to deploying household robots: task success is insufficient if the robot damages itself or its surroundings. Simulation offers a harm-free alternative to costly and dangerous real-world training and evaluation, yet existing simulators lack general mechanisms to detect, quantify, and represent damage. To address this gap, we introduce OOPSIEVERSE, a unified simulation framework and benchmark for damage-aware household manipulation. OOPSIEVERSE provides damage as an explicit, physically-grounded, and taskagnostic signal by converting sources such as contact forces, temperature changes, and liquid interactions into corresponding mechanical, thermal or fluid damage. OOPSIEVERSE comprises two core elements: (1) DAMAGESIM, a simulator-agnostic framework for detecting and quantifying damage during navigation and manipulation, and (2) a suite of household tasks designed to evaluate common damage modes and distinguish between task completion and safe execution. We demonstrate the generality of our framework by instantiating DAMAGESIM in two simulators with different physics backends, OmniGibson (Nvidia Omniverse) and RoboCasa (MuJoCo). We further showcase the utility of OOPSIEVERSE across multiple use cases, including (1) guiding safer demonstration collection via real-time damage feedback, (2) learning safer manipulation policies through damage-conditioned imitation learning and reinforcement learning, (3) benchmarking the safety of state-of-the-art Vision Language Action policies, and (4) improving real-world safety of sim-to-real transferred policies. Together, our results highlight the potential of OOPSIEVERSE as an open-source foundation for systematic, scalable research on safe robot manipulation. For code and more information, please refer to https://robin-lab.cs.utexas.edu/oopsieverse/
Chinese Translation
尽管机器人操作能力迅速发展，物理安全仍然是部署家用机器人的主要障碍：如果机器人损坏了自身或其周围环境，任务成功就显得不够充分。模拟提供了一种无害的替代方案，以避免昂贵且危险的现实世界训练和评估，但现有的模拟器缺乏检测、量化和表示损伤的通用机制。为了解决这一问题，我们提出了OOPSIEVERSE，这是一个统一的模拟框架和损伤感知家用操作基准。OOPSIEVERSE通过将接触力、温度变化和液体相互作用等源转化为相应的机械、热或流体损伤，提供了作为明确、物理基础和任务无关信号的损伤。OOPSIEVERSE包含两个核心元素：（1）DAMAGESIM，一个与模拟器无关的框架，用于在导航和操作过程中检测和量化损伤；（2）一套设计用于评估常见损伤模式并区分任务完成与安全执行的家用任务。我们通过在具有不同物理后端的两个模拟器OmniGibson（Nvidia Omniverse）和RoboCasa（MuJoCo）中实例化DAMAGESIM，展示了我们框架的通用性。我们进一步展示了OOPSIEVERSE在多个用例中的实用性，包括（1）通过实时损伤反馈指导更安全的演示收集，（2）通过损伤条件的模仿学习和强化学习学习更安全的操作策略，（3）基准测试最先进的视觉语言行动策略的安全性，以及（4）提高从模拟到现实转移策略的现实世界安全性。总体而言，我们的结果突显了OOPSIEVERSE作为系统化、可扩展的安全机器人操作研究的开源基础的潜力。有关代码和更多信息，请访问 https://robin-lab.cs.utexas.edu/oopsieverse/
cs.RO / 57 / 2606.32009
Human-as-Humanoid: Enabling Zero-Shot Humanoid Learning from Ego-Exo Human Videos with Human-Aligned Embodiments
人类作为类人机器人：通过人类对齐的具身体实现从自我-外部人类视频的零-shot类人学习
Abstract
Vision-language-action (VLA) models across robot embodiments require high-quality observation--action supervision to learn deployable action distributions, yet scaling such robot data remains difficult, especially for high-DoF humanoids. Teleoperation provides controller-aligned supervision, while human egocentric videos capture diverse bimanual manipulation but do not directly provide executable robot actions. We introduce Human-as-Humanoid, a human-to-humanoid supervision framework that enables near-real-time human-centric action generation, making human demonstrations usable for high-DoF humanoid VLA training by jointly aligning the robot embodiment, the sensing setup, and the action-label interface. Built on PrimeU, a human-aligned 60-DoF upper-body humanoid, Human-as-Humanoid uses synchronized ego-exo videos to pair deployment-aligned egocentric observations with exocentric motion recovery, retargets the recovered human motion through staged Inverse Kinematics (IK) into controller-aligned 60-DoF action chunks, and trains the VLA model with Forward Kinematics (FK)-aware supervision to preserve wrist and fingertip task-space geometry. This converts large-scale human demonstrations from visual observations into executable observation--action supervision for the target humanoid. Experiments validate the conversion chain at the motion-recovery, robot-action-space, and real-robot deployment levels. Human-as-Humanoid yields a 4.8--7.2x raw demonstration-throughput gain over humanoid teleoperation in our data-collection analysis, and on several downstream tasks, policies post-trained only with the converted human labels generalize to real-robot deployment without target-task robot demonstrations. The official project website is available at https://zgc-embodyai.github.io/Human-as-Humanoid.
Chinese Translation
跨机器人具身体的视觉-语言-动作（VLA）模型需要高质量的观察-动作监督，以学习可部署的动作分布，但扩展此类机器人数据仍然困难，尤其是对于高自由度（DoF）类人机器人。遥操作提供了与控制器对齐的监督，而人类自我中心视频捕捉了多样的双手操作，但并未直接提供可执行的机器人动作。我们提出了人类作为类人机器人（Human-as-Humanoid），一个人类到类人机器人监督框架，能够实现近实时的人类中心动作生成，使人类示范可用于高自由度类人机器人VLA训练，通过共同对齐机器人具身体、传感设置和动作标签接口。基于PrimeU，一个与人类对齐的60自由度上半身类人机器人，人类作为类人机器人使用同步的自我-外部视频，将与部署对齐的自我中心观察与外部运动恢复配对，通过分阶段的逆向运动学（Inverse Kinematics, IK）将恢复的人类运动重新定向为与控制器对齐的60自由度动作块，并使用前向运动学（Forward Kinematics, FK）感知监督训练VLA模型，以保持手腕和指尖任务空间几何形状。这将大规模的人类示范从视觉观察转换为目标类人机器人的可执行观察-动作监督。实验验证了运动恢复、机器人动作空间和真实机器人部署层面的转换链。在我们的数据收集分析中，人类作为类人机器人在原始示范吞吐量上比类人机器人遥操作提高了4.8到7.2倍，并且在几个下游任务中，仅用转换后的人类标签后训练的策略在没有目标任务机器人示范的情况下能够推广到真实机器人部署。项目的官方网站可访问 https://zgc-embodyai.github.io/Human-as-Humanoid。
cs.RO / 58 / 2606.32027
Freeform Preference Learning for Robotic Manipulation
自由形式偏好学习用于机器人操控
Abstract
Reward design remains a central bottleneck for autonomous robot policy improvement, especially in long-horizon manipulation tasks where sparse success labels provide too little signal and binary preferences collapse many competing notions of quality into one ambiguous signal. We introduce Freeform Preference Learning (FPL), a method for learning robot policies from freeform human preferences. Rather than asking annotators which of two trajectories is better overall, FPL lets them define natural-language preference axes, such as speed, safety, quality of placement, or carefulness, and provide pairwise preferences along each axis. These annotations are used to learn a language-conditioned reward model that maps a trajectory and preference label to an axis-specific reward. We use this model to train a reward-conditioned policy that optimizes across the multiple human-specified dimensions. Across four real-world and two simulated long-horizon manipulation tasks, FPL improves over sparse-reward and binary-preference methods by 38 percentage points. Beyond improved performance, FPL learns dense progress signals without explicit subtask segmentation, shows compositionality of behavior not present in the data, and allows users to steer the policy towards different behaviors at test time without retraining. Blog post with videos available at https://freeform-pl.github.io/fpl.website/
Chinese Translation
奖励设计仍然是自主机器人策略改进的一个核心瓶颈，尤其是在长时间跨度的操控任务中，稀疏的成功标签提供的信号过少，而二元偏好将许多竞争的质量概念压缩成一个模糊的信号。我们提出了自由形式偏好学习（Freeform Preference Learning, FPL），这是一种从自由形式的人类偏好中学习机器人策略的方法。FPL并不是要求标注者判断两个轨迹中哪个整体上更好，而是让他们定义自然语言的偏好轴，例如速度、安全性、放置质量或谨慎程度，并在每个轴上提供成对偏好。这些标注用于学习一个语言条件的奖励模型，该模型将轨迹和偏好标签映射到特定轴的奖励。我们利用这个模型训练一个奖励条件的策略，该策略在多个由人类指定的维度上进行优化。在四个真实世界和两个模拟的长时间跨度操控任务中，FPL的表现比稀疏奖励和二元偏好方法提高了38个百分点。除了性能提升外，FPL还在没有显式子任务分割的情况下学习到密集的进展信号，展示了数据中不存在的行为组合性，并允许用户在测试时引导策略朝向不同的行为，而无需重新训练。有关视频的博客文章可在 https://freeform-pl.github.io/fpl.website/ 查阅。
cs.RO / 59 / 2606.32028
DVG-WM: Disentangled Video Generation Enables Efficient Embodied World Model for Robotic Manipulation
DVG-WM：解耦视频生成促进高效的机器人操作体世界模型
Abstract
Video-based embodied world models provide an appealing substrate for robotic manipulation by predicting future states, yet current approaches remain limited by a fundamental entanglement: accurately modeling dynamics typically requires low-level temporal reasoning, while producing high-resolution frames demands expansive visual synthesis according to high-level semantics. This entanglement results in slow inference speed for iterative planning or too coarse predictions to retain contact-rich details. To solve this dilemma, we present Disentangled Video Generation World Model (DVG-WM), an efficient framework that explicitly decomposes world modeling into dynamics learning and visual synthesis. Conditioned on an initial observation and a language instruction, our model first generates a plausible sequence of intermediate visual states to preview the physical interaction and refines them to obtain high-fidelity videos. Furthermore, an efficient cascading mechanism is proposed, where DVG-WM uses flow matching to directly map the dynamics to video latents, and introduces a latent degradation mechanism to regenerate contact-rich details. Experiments on LIBERO and real-world platforms demonstrate improved video quality with up to 3.97 times acceleration, validating that disentangled video generation can be an efficient embodied world model for robotic manipulation.
Chinese Translation
基于视频的具身世界模型通过预测未来状态为机器人操作提供了一个吸引人的基础，但当前的方法受到一个基本纠缠的限制：准确建模动态通常需要低级时间推理，而生成高分辨率帧则需要根据高级语义进行广泛的视觉合成。这种纠缠导致迭代规划的推理速度缓慢，或者预测过于粗糙，无法保留丰富的接触细节。为了解决这一困境，我们提出了解耦视频生成世界模型（DVG-WM），这是一个高效的框架，明确将世界建模分解为动态学习和视觉合成。在初始观察和语言指令的条件下，我们的模型首先生成一系列合理的中间视觉状态，以预览物理交互，并对其进行细化以获得高保真视频。此外，我们提出了一种高效的级联机制，其中DVG-WM使用流匹配直接将动态映射到视频潜变量，并引入潜变量降解机制以重新生成丰富的接触细节。在LIBERO和现实世界平台上的实验表明，视频质量得到了改善，速度提高了最多3.97倍，验证了解耦视频生成可以成为机器人操作的高效具身世界模型。
cs.CV / 1 / 2606.30647
Cross-Modal Hierarchical Fusion for from Multi-Sensor Ground Observation
跨模态层次融合用于多传感器地面观测
Abstract
Dense volumetric reconstruction of cloud microphysical fields from sparse ground-based instruments remains an open problem, largely because the available measurements are heterogeneous in both modality and spatial coverage. We present AtmoFuseNet, a framework that fuses multi-view sky camera imagery with millimeter-wave cloud radar and ceilometer observations to produce 4D (three spatial dimensions plus time) estimates of cloud state and wind. The method operates in three stages: a cross-modal hierarchical aggregation module that combines image feature pyramids with instrument-derived vertical profiles through layer-wise cross-attention; a conditional variational refinement module that maps the resulting volume to physically consistent microphysical fields under differentiable radar and image forward models; and a correlation-based motion estimator that recovers per-voxel 3D wind vectors from consecutive volumetric reconstructions. On collocated observations from a semi-arid site, AtmoFuseNet reaches 0.026 g m^-3 liquid water content MAE and 1.18 m s^-1 wind speed MAE, improving over existing retrieval baselines. Ablation experiments isolate the contribution of each module.
Chinese Translation
从稀疏的地面仪器中进行云微物理场的密集体积重建仍然是一个未解决的问题，这主要是因为可用的测量数据在模态和空间覆盖上都是异质的。我们提出了 AtmoFuseNet，一个将多视角天空相机图像与毫米波云雷达和云高仪观测数据融合的框架，以生成云状态和风的四维（三个空间维度加时间）估计。该方法分为三个阶段：一个跨模态层次聚合模块，通过层级交叉注意力将图像特征金字塔与仪器导出的垂直剖面相结合；一个条件变分精炼模块，在可微分的雷达和图像前向模型下，将生成的体积映射到物理一致的微物理场；以及一个基于相关性的运动估计器，从连续的体积重建中恢复每体素的三维风矢量。在来自半干旱地区的协同观测中，AtmoFuseNet 达到了 0.026 g m^-3 的液态水含量平均绝对误差（MAE）和 1.18 m s^-1 的风速平均绝对误差，优于现有的反演基线。消融实验隔离了每个模块的贡献。
cs.CV / 2 / 2606.30754
Streaming Gaussian Encoding for 4D Panoptic Occupancy Tracking
用于4D全景占用跟踪的流式高斯编码
Abstract
Camera-based 4D panoptic occupancy tracking (4D-POT) is a promising paradigm for holistic scene understanding from multi-view imagery, enabling joint reasoning about geometry, semantics, and object identities across time. Recent mask-based pipelines achieve strong performance by propagating instance queries across frames. However, their underlying volumetric representations are typically recomputed at each timestep, limiting geometric temporal consistency, particularly under occlusion and for static scene elements. To address this limitation, we propose a streaming Gaussian encoder that maintains a persistent volumetric scene representation for 4D-POT. Our method models the scene as a fixed-size set of latent Gaussian queries that are propagated via ego-motion compensation and refreshed under a confidence-guided budget constraint. Crucially, we shape Gaussian opacities through depth-based supervision to serve as proxy for visibility, enabling confidence to accumulate as a temporally aggregated measure of persistent scene support. Together with a warmup-based multi-frame training strategy, this yields representation-level temporal coherence beyond decoder-only tracking. Extensive experiments on Occ3D-extended nuScenes and Waymo establish a new state-of-the-art for camera-based 4D-POT, improving tracking consistency with negligible computational overhead while remaining fully compatible with existing mask-based pipelines. We provide code and models at https://sge.cs.uni-freiburg.de.
Chinese Translation
基于摄像头的4D全景占用跟踪（4D-POT）是一种有前景的范式，能够通过多视角图像实现对场景的整体理解，支持对几何、语义和物体身份的联合推理。近期的基于掩膜的管道通过在帧间传播实例查询取得了良好的性能。然而，它们的基础体积表示通常在每个时间步都被重新计算，这限制了几何的时间一致性，特别是在遮挡和静态场景元素的情况下。为了解决这一限制，我们提出了一种流式高斯编码器，能够为4D-POT维持一个持久的体积场景表示。我们的方法将场景建模为一组固定大小的潜在高斯查询，这些查询通过自我运动补偿进行传播，并在信心引导的预算约束下进行更新。关键是，我们通过基于深度的监督来调整高斯的不透明度，以作为可见性的代理，从而使信心能够作为持久场景支持的时间聚合度量进行累积。结合基于预热的多帧训练策略，这在解码器独立跟踪之外实现了表示级别的时间一致性。在Occ3D扩展的nuScenes和Waymo上的广泛实验建立了基于摄像头的4D-POT的新状态-of-the-art，改善了跟踪一致性，同时计算开销微乎其微，并且与现有的基于掩膜的管道完全兼容。我们提供代码和模型，网址为 https://sge.cs.uni-freiburg.de。
cs.CV / 3 / 2606.30777
Unveiling Transferability in Trajectory Prediction via Latent Scene Embeddings
通过潜在场景嵌入揭示轨迹预测中的可迁移性
Abstract
The growing availability of trajectory datasets has fueled major advances in data-driven motion prediction. Yet, models trained on one dataset often fail to generalize beyond their training domain as a result of differences in scene layouts, agent behaviors, and sensing conditions. A framework that learns latent representations of datasets and quantifies their similarity using distributional metrics is presented. This large-scale study covers 24 major datasets, including the most widely used motion-prediction benchmarks, and shows that the resulting transferability scores strongly correlate with cross-dataset model performance. The results provide practical guidance for dataset selection, pretraining, and large-scale foundation models for motion prediction, paving the way toward more generalizable and robust predictive systems.
Chinese Translation
轨迹数据集的日益丰富推动了数据驱动运动预测的重大进展。然而，由于场景布局、代理行为和感知条件的差异，在一个数据集上训练的模型往往无法在其训练域之外进行泛化。本文提出了一种框架，该框架学习数据集的潜在表示，并使用分布度量量化其相似性。本研究涵盖了24个主要数据集，包括最广泛使用的运动预测基准，结果表明，得到的可迁移性评分与跨数据集模型性能之间存在强相关性。这些结果为数据集选择、预训练以及运动预测的大规模基础模型提供了实用指导，为构建更具可泛化性和鲁棒性的预测系统铺平了道路。
cs.CV / 4 / 2606.30795
Simple Supervision Is Hard to Beat: A Bitter Lesson from Sparse Target Labels in Domain-Adaptive Object Detection
简单监督难以超越：来自领域自适应目标检测中稀疏目标标签的苦涩教训
Abstract
Source-free domain adaptive object detection adapts a source-trained detector to an unlabeled target domain, typically through teacher-student self-training with pseudo-labels. We revisit this setting when a small, uniformly sampled subset of target images is labeled. We introduce Random-Target Supervised Mixing (RTSM), a simple anchor that incorporates these annotations through a supervised detection loss while leaving the original unlabeled adaptation branch unchanged. Across evaluations spanning four SFDA-OD methods, two object detectors, multiple adaptation tasks, and target-label budgets from 1% to 10%, RTSM consistently improves pure SFDA by 1.7 to 18.3 AP50. We then examine whether the same annotations can provide further gains by steering unlabeled self-training. To this end, we evaluate ten sparse-label feedback plugins covering pseudo-label selection, object completion, and optimization control, which yield limited and method-dependent gains over RTSM. These results reveal a bitter lesson for sparse-label SFDA-OD: simple supervision is hard to beat. RTSM therefore provides a simple yet effective anchor for sparse-label SFDA-OD.
Chinese Translation
无源领域自适应目标检测将源训练的检测器适应于未标记的目标领域，通常通过教师-学生自我训练与伪标签进行。我们重新审视这种设置，即当一小部分均匀采样的目标图像被标记时。我们引入了随机目标监督混合（Random-Target Supervised Mixing, RTSM），这是一个简单的锚点，通过监督检测损失整合这些注释，同时保持原始未标记适应分支不变。在涵盖四种无源领域自适应目标检测（SFDA-OD）方法、两种目标检测器、多种适应任务以及从1%到10%的目标标签预算的评估中，RTSM始终将纯SFDA的AP50提升了1.7到18.3。然后，我们考察这些相同的注释是否可以通过引导未标记的自我训练提供进一步的增益。为此，我们评估了十个稀疏标签反馈插件，涵盖伪标签选择、目标完成和优化控制，这些插件在RTSM上产生了有限且依赖于方法的增益。这些结果揭示了稀疏标签SFDA-OD的一个苦涩教训：简单监督难以超越。因此，RTSM为稀疏标签SFDA-OD提供了一个简单而有效的锚点。
cs.CV / 5 / 2606.30809
GaussLite: Online Task-Conditioned 3D Gaussian Splatting for Real-Time Robotic Mapping
GaussLite：用于实时机器人映射的在线任务条件3D高斯点云渲染
Abstract
Existing 3D Gaussian Splatting (3DGS) systems distribute representation capacity uniformly across a scene, ignoring the fact that many downstream robotic tasks engage only a fraction of the reconstructed geometry. This causes valuable onboard compute to be allocated towards optimizing irrelevant parts of the scene, either limiting online capacity or under-optimizing the most relevant parts of the scene. We introduce GaussLite, a task-driven 3DGS mapping system that conditions its representation density on a natural-language task specification. Given a posed RGB-D stream and a task such as "prepare to pick up the object on the desk," GaussLite uses a one-shot LLM parser to extract target and anchor objects, which are grounded per-frame by an open-vocabulary detector and segmented to produce per-pixel relevance masks in real time. The mapper allocates seeding density, gradient flow and scaling by task relevance. At matched Gaussian budget and real-time mapping at 4 Hz on resource-constrained hardware, GaussLite outperforms baselines on ROI PSNR on the Replica Dataset by an average +2.72 dB and on a real-hardware demonstration in indoor and outdoor settings by +2.23 dB. We further show that two task-specialized agents' maps can be fused into a single shared map via per-voxel voting on active-optimization counts in real time, outperforming concatenation by +3.42 dB while only sharing an average 7.08% of the map.
Chinese Translation
现有的3D高斯点云渲染（3DGS）系统在场景中均匀分配表示能力，忽视了许多下游机器人任务仅涉及重建几何体的一部分。这导致宝贵的机载计算资源被分配用于优化场景中不相关的部分，限制了在线能力或未能充分优化场景中最相关的部分。我们提出了GaussLite，一种任务驱动的3DGS映射系统，其表示密度基于自然语言任务规范进行调节。给定一个已定位的RGB-D流和一个任务，例如“准备拾起桌子上的物体”，GaussLite使用一次性大语言模型（LLM）解析器提取目标和锚定物体，这些物体通过开放词汇检测器在每帧中进行定位，并实时分割以生成每像素的相关性掩膜。映射器根据任务相关性分配种子密度、梯度流和缩放。在匹配的高斯预算和在资源受限硬件上以4 Hz的实时映射下，GaussLite在Replica数据集上的感兴趣区域（ROI）峰值信噪比（PSNR）上平均超越基线2.72 dB，并在室内和室外环境的真实硬件演示中超越2.23 dB。我们进一步展示了两个任务专用代理的地图可以通过实时对活跃优化计数的每体素投票融合成一个共享地图，超越拼接方法3.42 dB，同时仅共享平均7.08%的地图。
cs.CV / 6 / 2606.30811
AVTok: 1D Unified Tokenization for Holistic Audio-Video Generation
AVTok：用于整体音视频生成的一维统一标记化
Abstract
Audio-video generation has recently gained unprecedented research attention, aiming to synthesize high-quality sounding video content with fine-grained synchronization and semantic alignment between the auditory and visual components. The preceding methods predominantly adopt a dual-branch design with separate tokenization and generation modules per modality, neglecting the representation gap while necessitating intensive computational resources for proper training. Inspired by recent advancements in one-dimensional visual tokenization, we present \textbf{AVTok}, a novel unified tokenizer designated for holistic audio-video generation. AVTok features a dual-stream transformer-based architecture with shared encoder-decoder and modal-specific learnable queries to efficiently and effectively encode an audio-video pair into a compact one-dimensional latent representation with a unified codebook. To cope with the heterogeneous information imbalance that hinders AVTok from exploiting aligned audio-visual information, we devise a hierarchical training strategy to progressively realize reconstruction capabilities for each modality. Extensive experiments demonstrate that AVTok excels both in audio-video reconstruction and when integrated into downstream pipelines for audio-to-video, video-to-audio, and class-conditional joint audio-video generation. AVTok paves the way for the challenge of joint audio-video tokenization and provides a potential direction to build unified large multimodal models for audio-video generation.
Chinese Translation
音视频生成最近引起了前所未有的研究关注，旨在合成高质量的音视频内容，并在听觉和视觉组件之间实现细粒度的同步和语义对齐。之前的方法主要采用双分支设计，为每种模态设置独立的标记化和生成模块，忽视了表示差距，同时需要大量计算资源以进行适当的训练。受最近一维视觉标记化进展的启发，我们提出了 extbf{AVTok}，一种用于整体音视频生成的新型统一标记器。AVTok具有基于双流变换器的架构，采用共享的编码器-解码器和特定模态的可学习查询，以高效且有效地将音视频对编码为紧凑的一维潜在表示，并使用统一的代码本。为了应对阻碍AVTok利用对齐音视频信息的异构信息不平衡，我们设计了一种分层训练策略，以逐步实现每种模态的重建能力。大量实验表明，AVTok在音视频重建方面表现出色，并在下游管道中集成用于音到视频、视频到音以及类条件联合音视频生成时也表现优异。AVTok为联合音视频标记化的挑战铺平了道路，并为构建统一的大型多模态音视频生成模型提供了潜在方向。
cs.CV / 7 / 2606.30849
SyncCache: Exploiting Asymmetric Dynamics for Fast Audio-Driven Portrait Animation
SyncCache：利用不对称动态实现快速音频驱动的人物动画
Abstract
Diffusion Transformers (DiTs) have significantly advanced audio-driven portrait animation, but their high computational cost leads to substantial inference latency. Although training-free diffusion caching accelerates inference significant, existing methods are primarily developed for text-conditioned generation and overlook the spatial and modality imbalances inherent in audio-driven portrait animation. In this paper, we propose SyncCache, a training-free caching acceleration method tailored for DiT-based portrait animation that explicitly exploits asymmetric dynamics. Specifically, high-frequency dynamics driven by audio conditions and concentrated in human regions are more challenging and critical to cache and reuse than the low-frequency visual background in portrait animation. First, we introduce Spatially-Asymmetric Probing to prioritize error sensitivity in dynamic human region. Second, through Modality-Decoupled Caching, we bypass heavy DiT block by reusing stable inter-block residuals, while continuously recomputing lightweight audio blocks to preserve precise lip synchronization. Furthermore, we introduce a cache ratio to control cache capacity and formulate memory-adaptive cache selection as an offline dynamic programming problem without online overhead. Extensive experiments demonstrate that SyncCache achieves superior speed-quality trade-offs, delivering up to 4.12x acceleration on HunyuanVideo-Avatar and 3.75x on Wan-S2V with near-lossless visual fidelity and precise audio alignment.
Chinese Translation
扩散变换器（Diffusion Transformers, DiTs）在音频驱动的人物动画方面取得了显著进展，但其高计算成本导致了显著的推理延迟。尽管无训练的扩散缓存显著加速了推理，现有方法主要针对文本条件生成而开发，忽视了音频驱动人物动画中固有的空间和模态不平衡。在本文中，我们提出了SyncCache，这是一种针对基于DiT的人物动画量身定制的无训练缓存加速方法，明确利用不对称动态。具体而言，由音频条件驱动并集中在人类区域的高频动态比人物动画中的低频视觉背景更具挑战性和重要性，因此更难以缓存和重用。首先，我们引入空间不对称探测（Spatially-Asymmetric Probing）来优先考虑动态人类区域的误差敏感性。其次，通过模态解耦缓存（Modality-Decoupled Caching），我们通过重用稳定的块间残差来绕过重的DiT模块，同时持续重新计算轻量级音频块以保持精确的唇部同步。此外，我们引入了缓存比率来控制缓存容量，并将内存自适应缓存选择公式化为一个离线动态规划问题，无需在线开销。大量实验表明，SyncCache在速度与质量的权衡上表现优越，在HunyuanVideo-Avatar上实现了高达4.12倍的加速，在Wan-S2V上实现了3.75倍的加速，同时保持近乎无损的视觉保真度和精确的音频对齐。
cs.CV / 8 / 2606.30875
The Label Imitation Game: Turing Test Network for Zero-Shot Pseudo-Label Pruning
标签模仿游戏：用于零-shot伪标签修剪的图灵测试网络
Abstract
Foundation model pseudo-labeling - labeling data strictly via zero-shot inference - enables massive scale, but performance is undermined by hallucinations that evade standard thresholds. To eliminate these errors, we introduce the Turing-inspired Label Imitation Game (LIG), a framework that formalizes pseudo-label pruning as an adversarial interrogation. Rather than filtering labels via isolated thresholds, we use the LIG to train a Turing Test Network (TTN), a task-agnostic "judge" that evaluates candidate pseudo-labels within a dataset-wide context. Experiments across four diverse datasets demonstrate the TTN's robustness, consistently enhancing label accuracy for three state-of-the-art vision-language models without costly supervision or retraining. Crucially, we demonstrate that learned semantic-contextual logic is a robust alternative to spatial-geometric verification, enabling a unique zero-shot task transfer capability - a TTN trained strictly on image classification datasets can effectively prune complex object detection pseudo-labels. This pruning yields F1-score gains of 28% for the worst-performing baseline categories and 44% with task-specific fine-tuning. Significantly, we also observe Category Revival, where the TTN pruning "detoxifies" the training signal for downstream models and enables them to recover from zero recall on transfer-vulnerable classes. The pre-trained TTN models and code are available at https://github.com/voxel51/ttn.
Chinese Translation
基础模型伪标签化——通过零-shot推理严格标记数据——能够实现大规模，但由于幻觉现象的出现，导致性能受到影响，幻觉现象逃避了标准阈值。为了解决这些错误，我们引入了受图灵启发的标签模仿游戏（Label Imitation Game, LIG），一个将伪标签修剪形式化为对抗性审问的框架。我们不是通过孤立的阈值来过滤标签，而是使用LIG训练一个图灵测试网络（Turing Test Network, TTN），这是一个任务无关的“法官”，在数据集范围内评估候选伪标签。在四个不同的数据集上的实验表明，TTN具有鲁棒性，始终提高了三种最先进的视觉-语言模型的标签准确性，而无需昂贵的监督或重新训练。重要的是，我们证明了学习的语义-上下文逻辑是空间-几何验证的一个强有力的替代方案，能够实现独特的零-shot任务迁移能力——一个严格在图像分类数据集上训练的TTN能够有效修剪复杂的目标检测伪标签。这种修剪在表现最差的基线类别中带来了28%的F1分数提升，并在特定任务微调下达到了44%的提升。显著的是，我们还观察到类别复兴（Category Revival），即TTN修剪“去毒化”了下游模型的训练信号，使它们能够从对迁移脆弱类别的零召回中恢复过来。预训练的TTN模型和代码可在https://github.com/voxel51/ttn获取。
cs.CV / 9 / 2606.30896
Knowledge-Driven Dimension Estimation from a Single Image -3D Asset Generation Technology for Digital Twin Construction
基于知识驱动的单幅图像尺寸估计 - 数字双胞胎构建的3D资产生成技术
Abstract
In the verification of in-vehicle cameras, simulation technology using virtual spaces has advanced, enabling pre-evaluation of false detections and missed detections in various scenarios. However, discrepancies in the scale of the object being verified between the virtual and real environments can lead to a decrease in camera recognition performance. For traffic signs installed at high altitudes, distance measurement using LiDAR or stereo cameras is difficult, requiring size estimation from monocular images. This paper proposes a method for estimating the scale of an object by decomposing it into multiple structural elements and integrating external knowledge regarding design rules, geometric relationships, and conventional dimensions. Specifically, this method detects each component from a monocular image and estimates the size of each component by considering its structural relationships and dimensional consistency with surrounding elements. Furthermore, it generates a 3D asset of the object by reconstructing the estimated components. This method makes it possible to place 3D assets with a scale approximating the real environment within a digital twin space and is expected to contribute to improving the verification accuracy of in-vehicle cameras for autonomous driving in virtual environments.
Chinese Translation
在车载摄像头的验证中，利用虚拟空间的仿真技术得到了发展，使得能够在各种场景下对误检和漏检进行预评估。然而，虚拟环境与真实环境中被验证物体的尺度差异可能导致摄像头识别性能的下降。对于安装在高处的交通标志，使用激光雷达（LiDAR）或立体摄像头进行距离测量较为困难，因此需要从单目图像中进行尺寸估计。本文提出了一种通过将物体分解为多个结构元素并整合有关设计规则、几何关系和常规尺寸的外部知识来估计物体尺度的方法。具体而言，该方法从单目图像中检测每个组件，并通过考虑其结构关系和与周围元素的尺寸一致性来估计每个组件的大小。此外，该方法通过重建估计的组件生成物体的3D资产。该方法使得能够在数字双胞胎空间中放置与真实环境尺度相近的3D资产，并有望提高虚拟环境中自动驾驶车载摄像头的验证准确性。
cs.CV / 10 / 2606.30901
GRAPE: Graph-Augmented Prototype Explanations for Interactive Medical Image Diagnosis
GRAPE：用于交互式医学图像诊断的图增强原型解释
Abstract
Prototype-based medical image classifiers present three clinical limitations: they treat findings as independent, silently amplify unsafe physician feedback, and require full retraining whenever a new finding is needed. We present GRAPE (Graph-Augmented Prototype Explanations), a unified architecture that addresses all three challenges. First, a Graph Attention Task Head models anatomical concept co-occurrence, boosting macro-F1 by +13.8,pp over the prototype baseline on TBX11K. Second, a Concept-Mismatch Safety Check - the first such mechanism in prototype-based medical classifiers - warns when the model's dominant finding inside a doctor-drawn region conflicts with the claimed label, catching 85% of erroneous annotations versus 51% for MC-Dropout with no extra inference cost. Third, Open-Vocabulary Prototype Anchoring aligns visual prototypes to clinical text, allowing a new finding to be added from a single labeled image without modifying any other component. On NIH ChestX-ray14, one Effusion example recovers full-supervision localization accuracy; on TBX11K, prototype maps achieve 2.6x better lesion localization than end-to-end baselines. All three capabilities add only +1~ms latency at interactive batch size. The project page is https://github.com/KurbanIntelligenceLab/GRAPE.
Chinese Translation
基于原型的医学图像分类器存在三个临床局限性：它们将发现视为独立，默默放大不安全的医生反馈，并且每当需要新的发现时都需要完全重新训练。我们提出了GRAPE（图增强原型解释），这是一个统一的架构，旨在解决这三项挑战。首先，图注意力任务头（Graph Attention Task Head）建模解剖概念的共现，使得在TBX11K数据集上宏观F1分数比原型基线提高了13.8个百分点。其次，概念不匹配安全检查（Concept-Mismatch Safety Check）——这是基于原型的医学分类器中首个此类机制——在模型的主导发现与所声称标签冲突时发出警告，捕获了85%的错误注释，而MC-Dropout在没有额外推理成本的情况下仅捕获了51%。第三，开放词汇原型锚定（Open-Vocabulary Prototype Anchoring）将视觉原型与临床文本对齐，允许从单个标记图像中添加新发现，而无需修改其他任何组件。在NIH ChestX-ray14数据集上，一个积液（Effusion）示例恢复了完全监督的定位精度；在TBX11K上，原型图的病灶定位精度比端到端基线提高了2.6倍。这三项功能在交互式批量大小下仅增加了1毫秒的延迟。项目页面为 https://github.com/KurbanIntelligenceLab/GRAPE。
cs.CV / 11 / 2606.30937
No Adaptation Without Observation: Observability-Constrained Test-Time Prompt Tuning for LiDAR Semantic Segmentation
没有观察就没有适应：受可观测性约束的测试时提示调优用于LiDAR语义分割
Abstract
LiDAR semantic segmentation often degrades under real-world deployment due to evolving sensing conditions, while collecting new annotations for retraining is impractical. Test-time adaptation (TTA) updates model parameters online using pseudo-label supervision, but directly applying standard TTA strategies to LiDAR data is challenging. Because pseudo-label reliability is spatially heteroscedastic under range-dependent sparsity and occlusion, uniform updates on globally shared parameters can inject unstable gradients and destabilize adaptation. We propose a geometry-constrained test-time prompt tuning framework for LiDAR semantic segmentation. Our method estimates per-location sensing reliability from depth-consistent beam terminations and neighborhood support, and uses it to reweight spatial supervision. Adaptation is confined to lightweight prompt adapters inserted into a frozen backbone, with spatial gating to prevent unreliable regions from perturbing globally shared representations. A temporally smoothed prototype alignment strategy further stabilizes online updates by accumulating reliable semantic evidence over time. Experiments on standard LiDAR benchmarks demonstrate improved adaptation stability and segmentation performance under deployment variations without additional annotations.
Chinese Translation
LiDAR语义分割在实际部署中常因感知条件的变化而性能下降，而收集新的标注以进行再训练则不切实际。测试时适应（TTA）通过伪标签监督在线更新模型参数，但将标准TTA策略直接应用于LiDAR数据具有挑战性。由于伪标签的可靠性在范围依赖的稀疏性和遮挡下是空间异方差的，因此对全局共享参数进行统一更新可能会引入不稳定的梯度并使适应过程不稳定。我们提出了一种受几何约束的测试时提示调优框架，用于LiDAR语义分割。我们的方法从深度一致的光束终止和邻域支持中估计每个位置的感知可靠性，并利用该可靠性对空间监督进行重新加权。适应过程仅限于插入到冻结主干网络中的轻量级提示适配器，并通过空间门控防止不可靠区域干扰全局共享表示。一个时间平滑的原型对齐策略进一步通过随着时间积累可靠的语义证据来稳定在线更新。在标准LiDAR基准测试上的实验表明，在没有额外标注的情况下，适应稳定性和分割性能在部署变化下得到了改善。
cs.CV / 12 / 2606.30951
Learning Where to Look: A Reinforcement Learning Framework for Robust Micro-Ultrasound Prostate Cancer Detection
学习观察位置：一种用于稳健微超声前列腺癌检测的强化学习框架
Abstract
Micro-ultrasound ($\mu$US) is a new, emerging, and promising imaging modality for prostate cancer (PCa) detection, but accurate identification of suspicious tissue remains highly dependent on clinical experience, leading to substantial inter-observer variability. Machine-learning assistance can reduce this variability; however, training reliable deep models is challenging because supervision is sparse and noisy -- typically limited to core-level histopathology outcomes (e.g., cancer grade and its percentage in a biopsy core) without pixel-level lesion annotations and under severe class imbalance. We introduce Prost-RL, which reframes $\mu$US PCa detection as a spatially aware, policy-driven inference problem by learning where to look before decoding. Prost-RL integrates a lightweight reinforcement-learning policy into a foundation-model encoder-decoder to generate interpretable spatial attention maps that act as soft prompts for both cancer-likelihood heatmap prediction and image-level classification. We further propose Adaptive Policy Optimization (APO) to stabilize hybrid supervised-RL training and a noise-robust objective combining symmetric cross-entropy with negative-entropy regularization to mitigate weak-label noise and encourage sharp localization. On a cohort of 6,607 biopsy cores from 693 patients across five clinical sites, Prost-RL achieves $79.0\pm3.5$ AUROC with $64.6\pm6.3$% sensitivity at 80% specificity for core-level detection (+2.1 AUROC and +4.5 sensitivity points over the strongest baseline), and $79.3\pm5.8$ AUROC for clinically significant cancer classification. The learned policy highlights biopsy-aligned regions, providing transparent, spatially grounded evidence alongside quantitative risk predictions. Code is available at: https://github.com/DeepRCL/Prost-RL.
Chinese Translation
微超声（$bc$US）是一种新兴且有前景的前列腺癌（PCa）检测成像技术，但对可疑组织的准确识别仍然高度依赖临床经验，导致观察者之间存在显著的变异性。机器学习的辅助可以减少这种变异性；然而，训练可靠的深度模型具有挑战性，因为监督信息稀疏且噪声较大——通常仅限于核心级别的组织病理结果（例如，癌症等级及其在活检核心中的比例），而缺乏像素级病变注释，并且存在严重的类别不平衡。我们提出了Prost-RL，它将$bc$US前列腺癌检测重新构建为一个空间感知的、基于策略的推理问题，通过学习观察位置来进行解码。Prost-RL将轻量级强化学习策略集成到基础模型的编码器-解码器中，以生成可解释的空间注意力图，这些图作为癌症可能性热图预测和图像级分类的软提示。我们进一步提出了自适应策略优化（Adaptive Policy Optimization, APO）以稳定混合监督-强化学习训练，并提出了一种结合对称交叉熵与负熵正则化的噪声稳健目标，以减轻弱标签噪声并鼓励精确定位。在来自五个临床中心的693名患者的6,607个活检核心的队列中，Prost-RL在核心级别检测中实现了$79.0 ext{±}3.5$ AUROC，灵敏度为$64.6 ext{±}6.3$%（在80%特异性下），相比最强基线提高了+2.1 AUROC和+4.5灵敏度点，并且在临床显著癌症分类中实现了$79.3 ext{±}5.8$ AUROC。学习到的策略突出显示了与活检对齐的区域，提供了透明的、基于空间的证据以及定量风险预测。代码可在以下链接获取：https://github.com/DeepRCL/Prost-RL。
cs.CV / 13 / 2606.30968
PhotoQuilt: Training-Free Arbitrary-Resolution Photomosaics via Bootstrapped Tiled Denoising
PhotoQuilt：通过引导式平铺去噪实现无训练的任意分辨率照片马赛克
Abstract
Photomosaics are large images whose local regions are seen as independent tiles while their overall arrangement forms a coherent scene. Generating them at high resolution, with every tile convincing in its own right, is computationally expensive, since the canvas must hold many detailed tiles at once. We present PhotoQuilt, a training-free framework that generates photomosaics at arbitrary resolution. Diffusion models struggle to satisfy both scales at once, as direct high-resolution generation is costly and tends toward one smooth image rather than a mosaic, while patch-based tiling keeps local detail but loses global structure. PhotoQuilt resolves this with a bootstrapped tiled denoising procedure. We first produce a global composition at low resolution to fix the layout, then upscale it in latent space and re-inject noise to restore generative capacity. Denoising proceeds within fixed tiles, so each forms its own image while the shared global structure holds them in one layout. Because tile generation is handled separately, PhotoQuilt scales to large canvases without quadratic attention cost. Experiments show that PhotoQuilt outperforms current baselines on both global structure and local realism.
Chinese Translation
照片马赛克是将局部区域视为独立拼块的大型图像，而它们的整体排列形成一个连贯的场景。在高分辨率下生成这些图像，同时确保每个拼块在其自身上都令人信服，是计算上昂贵的，因为画布必须同时容纳许多详细的拼块。我们提出了PhotoQuilt，一个无训练的框架，可以生成任意分辨率的照片马赛克。扩散模型在同时满足这两种尺度方面存在困难，因为直接生成高分辨率图像的成本很高，并且倾向于生成一幅平滑的图像而不是马赛克，而基于拼块的平铺保持了局部细节但失去了全局结构。PhotoQuilt通过引导式平铺去噪程序解决了这个问题。我们首先在低分辨率下生成全局构图以固定布局，然后在潜在空间中对其进行上采样，并重新注入噪声以恢复生成能力。去噪过程在固定拼块内进行，因此每个拼块形成自己的图像，同时共享的全局结构将它们保持在一个布局中。由于拼块生成是单独处理的，PhotoQuilt可以在不增加二次注意力成本的情况下扩展到大型画布。实验表明，PhotoQuilt在全局结构和局部真实感方面均优于当前基准。
cs.CV / 14 / 2606.31004
Auditing Generalization in AI-Generated Video Detection: A Six-Control Protocol and the VidAudit Toolkit
人工智能生成视频检测中的泛化审计：六项控制协议与VidAudit工具包
Abstract
AI-generated video detection benchmarks such as GenVidBench and AIGVDBench are the de facto leaderboards, yet most evaluation protocols leave uncontrolled confounds that can inflate reported generalization. As an existence proof, a three-feature clip-length classifier reaches a leave-one-generator-out (LOGO) AUC of 0.998 on GenVidBench under unaudited evaluation, while measuring nothing about motion. A 20-paper survey finds none applying all six standard controls that would catch this, so we combine them into an audited protocol and apply it to six representative feature sources (three published detectors and three repurposed signal sources), re-running it cross-dataset on AIGVDBench. The audit both debunks and certifies: the trivial classifier collapses to near chance (0.529), a CLIP baseline is caught carrying dataset identity, and the 2025 forensic detector WaveRep clears the floor at out-of-distribution LOGO AUC 0.996 with chance-level real-vs-real coherence. At a deployable FPR of 0.1%, multiple high-AUC methods fall to single-digit recall and the leaderboard order changes, so we recommend an audited tuple (AUC, above-floor margin, operating-point recall, and calibration) over a single number. As a white-box positive control, we add TemporalSpec (codec motion vectors); via cross-substrate feature fusion (XSFF), a second substrate adds genuine complementarity that survives the audit. We release VidAudit, to our knowledge the largest unified and audited detector collection for this task, providing 14 detectors behind one plugin API, a leaderboard, and Croissant metadata, available at https://github.com/KurbanIntelligenceLab/vidaudit. Together, the protocol and toolkit move evaluation from leaderboard rank toward whether a result measures what it claims.
Chinese Translation
人工智能生成视频检测基准，如GenVidBench和AIGVDBench，已成为事实上的排行榜，然而大多数评估协议留下了未控制的混杂因素，这可能会夸大报告的泛化能力。作为存在性证明，一个三特征的剪辑长度分类器在未经审计的评估下，在GenVidBench上达到了0.998的留一生成器外（LOGO）AUC，而对运动没有任何测量。一项包含20篇论文的调查发现，没有一篇应用所有六项标准控制来捕捉这一点，因此我们将这些控制结合成一个审计协议，并将其应用于六个代表性的特征源（三个已发布的检测器和三个重新利用的信号源），在AIGVDBench上进行跨数据集的重新测试。审计既揭穿了谬误，也进行了认证：这个微不足道的分类器崩溃至接近随机（0.529），一个CLIP基线被发现携带数据集身份，而2025年法医检测器WaveRep在分布外LOGO AUC 0.996的情况下以随机水平的真实与真实一致性清除了底线。在可部署的假阳性率（FPR）为0.1%时，多种高AUC方法的召回率降至个位数，排行榜顺序发生变化，因此我们建议使用审计元组（AUC、超底线边际、操作点召回率和校准）而非单一数字。作为一个白盒正控制，我们添加了TemporalSpec（编解码器运动向量）；通过跨基质特征融合（XSFF），第二个基质添加了真正的互补性，且在审计中得以保留。我们发布了VidAudit，据我们所知，这是针对该任务最大的统一和审计检测器集合，提供14个检测器通过一个插件API、一个排行榜和Croissant元数据，网址为https://github.com/KurbanIntelligenceLab/vidaudit。协议和工具包共同推动评估从排行榜排名转向结果是否测量其所声称的内容。
cs.CV / 15 / 2606.31007
Dense Structural Priors for Sparse Functional Landmark Localization in Surgical Videos
稠密结构先验用于外科视频中稀疏功能性地标定位
Abstract
Vision foundation models such as SAM 3 can provide transferable object-level structure across diverse surgical video conditions, but segmentation outputs do not explicitly encode the action-conditioned semantics that define functional surgical landmarks. Estimating instrument extent and geometry differs from localizing the tip or anchor relevant to clipping, grasping, or dissecting. We investigate vision foundation model-enabled sparse action-aware landmark localization, using zero-shot, point-prompted structural masks to provide dense instrument-level context without manual pixel-level mask annotations. We propose a lightweight refinement framework that uses SAM 3 as a structural prior. A coarse multi-frame network predicts tip and anchor prompts, generating non-oracle masks that are fused with visual and heatmap features to refine functional landmark predictions. We compare direct mask-augmented supervision, prediction-derived mask-prior refinement, and auxiliary mask supervision to examine how vision foundation model-derived structure should enter a precision-oriented localization system. Experiments on 7,867 clips from 60 surgical videos spanning YouTube, Cholec80, HeiChole, SurgVU, and CRCD evaluate the approach under heterogeneous conditions. Without manual pixel-level mask annotations for training, the proposed model achieves overall F1 scores of 72.4% for tip and 58.0% for anchor localization. Directly imposing masks on heatmap targets biases learning toward broad tool regions, whereas prediction-derived priors and auxiliary supervision provide effective intermediate structural guidance for action-dependent landmark prediction.
Chinese Translation
视觉基础模型如SAM 3能够在多样的外科视频条件下提供可转移的对象级结构，但分割输出并未明确编码定义功能性外科地标的动作条件语义。估计工具的范围和几何形状与定位与剪切、抓取或解剖相关的尖端或锚点不同。我们研究了基于视觉基础模型的稀疏动作感知地标定位，利用零样本、点提示的结构掩膜提供稠密的工具级上下文，而无需手动的像素级掩膜注释。我们提出了一种轻量级的细化框架，使用SAM 3作为结构先验。粗略的多帧网络预测尖端和锚点提示，生成非理想掩膜，并将其与视觉特征和热图特征融合，以细化功能性地标预测。我们比较了直接掩膜增强监督、基于预测的掩膜先验细化和辅助掩膜监督，以研究视觉基础模型衍生的结构应如何进入以精确为导向的定位系统。在来自YouTube、Cholec80、HeiChole、SurgVU和CRCD的60个外科视频中的7,867个片段上进行的实验评估了该方法在异质条件下的表现。在没有手动像素级掩膜注释进行训练的情况下，所提出的模型在尖端和锚点定位的整体F1分数分别达到了72.4%和58.0%。直接将掩膜施加于热图目标上会使学习偏向于广泛的工具区域，而基于预测的先验和辅助监督则为依赖动作的地标预测提供了有效的中间结构指导。
cs.CV / 16 / 2606.31015
Dual Sparse Aggregation Transformer for Multispectral Object Detection
用于多光谱目标检测的双稀疏聚合变换器
Abstract
Transformer-based approaches have obtained excellent performance in multispectral object detection tasks due to their ability to model long-range dependencies and capture complementary information. However, previous transformer-based multispectral detection methods tend to use all available tokens for similarity calculation, which results in redundant information interaction from irrelevant areas, leading to degraded detection performance. To overcome this challenge, we propose a novel Dual Sparse Aggregation Transformer (DSAFormer) for multispectral object detection, which consists of a Dual Sparse Transformer (DSFormer) and a Learnable Addition Fusion Block (LAFB). Specifically, the DSFormer is designed to exploit and boost cross-modal complementary information, thereby improving detection performance. It incorporates three key components: A Spatial Sparse Multi-Head Cross-Attention (SSMHCA) mechanism selectively captures cross-modal relationships at the spatial level by reserving only the high query-key similarity scores, eliminating irrelevant interactions. A Channel Sparse Multi-Head Cross-Attention (CSMHCA) mechanism performs similar sparse calculations at the channel level to enhance feature representation and filter out low matching query-key. A Multi-Scale Feature Refinement Layer (MSFRL) is developed to aggregate hierarchical features and suppress redundant information. To effectively fuse multimodal features, the LAFB is introduced to aggregate intramodal and intermodal feature information by feature reweighting. Extensive experimental results have demonstrated that our proposed DSAFormer achieves better detection performance against state-of-the-art methods on four public datasets, including the MFAD, FLIR, M$^3$FD, and LLVIP. The source code of our DSAFormer will be released at https://github.com/WenCongWu/DSAFormer.
Chinese Translation
基于变换器的方法在多光谱目标检测任务中取得了优异的性能，得益于其建模长距离依赖关系和捕获互补信息的能力。然而，以往的基于变换器的多光谱检测方法往往使用所有可用的标记进行相似性计算，这导致来自无关区域的冗余信息交互，从而降低了检测性能。为了解决这一挑战，我们提出了一种新颖的双稀疏聚合变换器（Dual Sparse Aggregation Transformer，DSAFormer）用于多光谱目标检测，该模型由双稀疏变换器（Dual Sparse Transformer，DSFormer）和可学习加法融合块（Learnable Addition Fusion Block，LAFB）组成。具体而言，DSFormer旨在利用和增强跨模态的互补信息，从而提高检测性能。它包含三个关键组件：空间稀疏多头交叉注意力（Spatial Sparse Multi-Head Cross-Attention，SSMHCA）机制选择性地在空间层面捕获跨模态关系，仅保留高查询-键相似度分数，消除无关交互；通道稀疏多头交叉注意力（Channel Sparse Multi-Head Cross-Attention，CSMHCA）机制在通道层面进行类似的稀疏计算，以增强特征表示并过滤低匹配的查询-键；多尺度特征精炼层（Multi-Scale Feature Refinement Layer，MSFRL）用于聚合层次特征并抑制冗余信息。为了有效融合多模态特征，引入了LAFB，通过特征重加权聚合模态内和模态间的特征信息。大量实验结果表明，我们提出的DSAFormer在四个公共数据集（包括MFAD、FLIR、M$^3$FD和LLVIP）上相较于最先进的方法实现了更好的检测性能。我们的DSAFormer源代码将发布在https://github.com/WenCongWu/DSAFormer。
cs.CV / 17 / 2606.31018
WarpI2I: Image Warping for Image-to-Image Translation
WarpI2I：用于图像到图像翻译的图像扭曲
Abstract
Image-to-image (I2I) translation has achieved strong results in tasks like human relighting and driving scene translation using latent diffusion models (LDMs). However, compact LDMs often struggle to preserve fine-grained structures because the encoder compresses high-resolution inputs into a spatially downsampled latent space. To address this issue, we propose a simple saliency-guided warp-unwarp framework that reallocates spatial representation toward salient regions before encoding, enabling better preservation of structural details without increasing latent resolution. The warped image is processed by the original diffusion model and then mapped back via an inverse warp. In addition, we propose a simple and efficient outpainting-based synthetic data generation pipeline to produce high-quality paired data for image relighting. Our method is model-agnostic, requires no architectural modification, and introduces negligible computational overhead. Experiments on human relighting, driving scene relighting, and translation demonstrate improved structural preservation, lighting faithfulness, and image quality, with our framework extending naturally to video via frame-by-frame application with good temporal stability. Project Webpage: https://shenzheng2000.github.io/WarpI2I.github.io
Chinese Translation
图像到图像（I2I）翻译在诸如人类重光照和驾驶场景翻译等任务中，利用潜在扩散模型（LDMs）取得了显著成果。然而，紧凑型LDM往往难以保留细致的结构，因为编码器将高分辨率输入压缩到空间下采样的潜在空间。为了解决这个问题，我们提出了一种简单的显著性引导扭曲-反扭曲框架，该框架在编码之前重新分配空间表示到显著区域，从而在不增加潜在分辨率的情况下更好地保留结构细节。扭曲后的图像由原始扩散模型处理，然后通过逆扭曲映射回去。此外，我们提出了一种简单高效的基于外绘的合成数据生成管道，以生成高质量的配对数据用于图像重光照。我们的方法与模型无关，不需要架构修改，并且引入的计算开销可以忽略不计。在人类重光照、驾驶场景重光照和翻译的实验中，展示了结构保留、光照真实性和图像质量的改善，我们的框架通过逐帧应用自然扩展到视频，并具有良好的时间稳定性。项目网页：https://shenzheng2000.github.io/WarpI2I.github.io
cs.CV / 18 / 2606.31029
TerraDiT-$\Omega$: Unified Spatial Control for Satellite Image Synthesis with Any Geospatial Primitive
TerraDiT-$ ext{Ω}$：基于任意地理空间原语的卫星图像合成统一空间控制
Abstract
Generative models have achieved remarkable progress, yet applying them to satellite imagery remains challenging. Unlike natural imagery, satellite scenes are structured by spatially complex and semantically distinct geometries. Prior work addresses this complexity by adapting natural image frameworks using dense rasters or sparse prompts, trading off annotation cost and fidelity while breaking compatibility with vector primitives commonly used to represent geographic information. We introduce TerraDiT-$\Omega$, a unified spatial control framework that generates satellite imagery directly from any native geospatial primitive. By jointly leveraging precise annotations (polygons, polylines) and coarser ones (bounding boxes, points), the model supports controllable layouts across varying annotation budgets, broadening applicability to design tasks such as urban planning while remaining naturally compatible with end-to-end GeoAI workflows. To effectively leverage these primitives during generation, we propose Geometry-Aware Local Attention, a conditioning mechanism that injects explicit geometric cues into the attention space. Across all conditioning formats, our approach consistently outperforms both dense-control and sparse-control baselines. Furthermore, this flexibility enables controllable synthetic data augmentation using a single generative model, improving downstream performance on land-cover segmentation, object detection, road graph extraction, and scene classification. Code, data, and weights are available at https://github.com/mvrl/TerraDiT.
Chinese Translation
生成模型取得了显著进展，但将其应用于卫星图像仍然面临挑战。与自然图像不同，卫星场景由空间上复杂且语义上不同的几何形状构成。之前的研究通过使用密集栅格或稀疏提示来适应自然图像框架，从而解决了这种复杂性，虽然在注释成本和保真度之间进行了权衡，但与常用于表示地理信息的矢量原语的兼容性却被打破。我们提出了TerraDiT-$ ext{Ω}$，一个统一的空间控制框架，能够直接从任何原生地理空间原语生成卫星图像。通过联合利用精确注释（多边形、多线段）和较粗略的注释（边界框、点），该模型支持在不同注释预算下的可控布局，拓宽了其在城市规划等设计任务中的适用性，同时与端到端的GeoAI工作流程自然兼容。为了在生成过程中有效利用这些原语，我们提出了几何感知局部注意力（Geometry-Aware Local Attention），这是一种将显式几何线索注入注意力空间的条件机制。在所有条件格式下，我们的方法始终优于密集控制和稀疏控制的基线。此外，这种灵活性使得使用单一生成模型进行可控的合成数据增强成为可能，从而提高了土地覆盖分割、物体检测、道路图提取和场景分类等下游任务的性能。代码、数据和权重可在 https://github.com/mvrl/TerraDiT 获取。
cs.CV / 19 / 2606.31050
Learning Video Dynamics with Predictive Differentiable Rendering
通过预测可微渲染学习视频动态
Abstract
How to accurately predict a high-fidelity future world? While the visual world is inherently continuous, existing deterministic video prediction models operate in discrete pixel space and are mainly optimized with pixel-wise mean squared error (MSE), which often leads to over-smoothed predictions and a lack of fine-grained visual details. To address these limitations, we propose Predictive Differentiable Rendering (PDR), a novel end-to-end video prediction paradigm that bridges the gap between discrete and continuous representations. Inspired by recent progress in 3D reconstruction with 3D Gaussian Splatting, we introduce PredGS, a lightweight and plug-and-play adapter based on 2D Gaussian representation, which could be seamlessly integrated with existing pixel space predictors, significantly improving spatial detail preservation with negligible computational overhead. Furthermore, we develop predgsplat, a CUDA-accelerated differentiable 2D Gaussian renderer supporting arbitrary channels. Each Gaussian is defined by 5 + C learnable parameters (position, scale, rotation, and C channel amplitudes) and achieves up to 10x faster rendering than the baseline. Optimized by a combined L1 and SSIM loss, PDR overcomes the inherent blurring tendencies of MSE Loss, significantly enhancing the prediction performance. Extensive experiments on diverse real-world benchmarks, including TaxiBJ, WeatherBench, KTH, and Human3.6M, demonstrate that PDR consistently surpasses existing methods, delivering superior detail preservation, visual fidelity, and predictive accuracy.
Chinese Translation
如何准确预测高保真未来世界？尽管视觉世界本质上是连续的，现有的确定性视频预测模型却在离散的像素空间中运行，主要通过像素级均方误差（MSE）进行优化，这往往导致过度平滑的预测和缺乏细致的视觉细节。为了解决这些局限性，我们提出了预测可微渲染（Predictive Differentiable Rendering, PDR），这是一种新颖的端到端视频预测范式，弥合了离散和连续表示之间的差距。受到最近3D重建领域3D高斯点云（3D Gaussian Splatting）进展的启发，我们引入了PredGS，这是一种基于2D高斯表示的轻量级即插即用适配器，可以与现有的像素空间预测器无缝集成，显著提高空间细节的保留，同时几乎不增加计算开销。此外，我们开发了predgsplat，这是一种支持任意通道的CUDA加速可微2D高斯渲染器。每个高斯由5 + C个可学习参数（位置、尺度、旋转和C个通道幅度）定义，其渲染速度比基线快达10倍。通过结合L1损失和结构相似性指数（SSIM）损失进行优化，PDR克服了MSE损失固有的模糊倾向，显著提升了预测性能。在包括TaxiBJ、WeatherBench、KTH和Human3.6M等多样的真实世界基准上的广泛实验表明，PDR始终超越现有方法，提供卓越的细节保留、视觉保真度和预测准确性。
cs.CV / 20 / 2606.31054
ADAPT: Attention Dynamics Alignment with Preference Tuning for Faithful MLLMs
ADAPT：通过偏好调优实现注意力动态对齐以增强多模态大语言模型的可靠性
Abstract
Multimodal Large Language Models (MLLMs) are critically hampered by hallucination, generating content inconsistent with the provided image. In this paper, we identify an internal signature of hallucination: progressive degradation of text-to-image cross-attention during generation, leading to specific failure patterns like unfocused or biased attention. Existing mitigation strategies are largely outcome-driven and do not explicitly target this failure mode. To address this problem, we propose ADAPT (Attention Dynamics Alignment with Preference Tuning), an attention-based framework that intervenes directly on text-to-image cross-attention dynamics. We propose ADAPT with three key contributions: a cross-attention visual anchor refined from early decoding to provide stable spatial grounding, an attention-supervised inference mechanism that detects and corrects attention drift online, and a Visual Attention Guidance DPO that aligns preferences toward visually grounded responses. Experiments show that each component of ADAPT contributes to hallucination reduction, and the full framework achieves new best results across multiple hallucination benchmarks, reducing hallucination rates by 40%-60% across mainstream backbones while preserving general multimodal capabilities. Our work provides an attention-based perspective on mitigating hallucinations by exploring the model's internal text-to-image cross-attention behaviors. Code is available at https://github.com/yao-ustc/ADAPT
Chinese Translation
多模态大语言模型（MLLMs）受到幻觉的严重影响，生成与提供的图像不一致的内容。本文识别出幻觉的一个内部特征：在生成过程中文本与图像之间的交叉注意力逐渐退化，导致特定的失败模式，如注意力不集中或偏向性注意。现有的缓解策略主要是以结果为导向，并未明确针对这种失败模式。为了解决这一问题，我们提出了ADAPT（注意力动态对齐与偏好调优），这是一种基于注意力的框架，直接干预文本与图像之间的交叉注意力动态。我们提出ADAPT的三个关键贡献：从早期解码中提炼的交叉注意力视觉锚点，以提供稳定的空间基础；一种注意力监督推理机制，能够在线检测和纠正注意力漂移；以及一个视觉注意力引导的DPO（动态偏好优化），使偏好与视觉基础的响应对齐。实验表明，ADAPT的每个组件都有助于减少幻觉，而完整框架在多个幻觉基准测试中取得了新的最佳结果，将幻觉率降低了40%-60%，同时保持了一般的多模态能力。我们的工作通过探索模型内部的文本与图像交叉注意力行为，提供了一种基于注意力的幻觉缓解视角。代码可在 https://github.com/yao-ustc/ADAPT 获取。
cs.CV / 21 / 2606.31065
Diffusion-Based Material Regularization for Physics-Based Inverse Rendering
基于扩散的物质正则化用于物理基础的逆渲染
Abstract
Reconstructing physics-based 3D assets -- geometry, materials, and illumination -- from multi-view images is a core problem in computer graphics and vision, and a prerequisite for realistic relighting and editing. Physics-based inverse rendering offers an accurate image-formation model, but is severely underconstrained: without strong priors, illumination is baked into materials, and reconstructions generalize poorly to novel views and lighting. Data-driven diffusion models, in contrast, predict visually plausible materials, yet their predictions rarely satisfy the rendering equation and are not directly usable for physics-based rendering. We bridge these two paradigms rather than replacing either. Our key idea is to treat the predictions of a state-of-the-art diffusion model not as target material values but as a similarity kernel for optimization: we introduce a regularization loss that penalizes deviations in the optimized material over surface regions where the diffusion predictions are near-constant, while leaving the optimization free to match the input images. Built on this regularizer, our end-to-end pipeline jointly reconstructs geometry, materials, and illumination, yielding high-quality assets that drop into standard rendering pipelines and relight faithfully. On the Synthetic4Relight, Stanford-ORB, and DTC-Synthetic datasets, our method significantly outperforms state-of-the-art baselines in both reconstruction accuracy and relighting quality.
Chinese Translation
从多视角图像重建基于物理的三维资产——几何形状、材料和照明——是计算机图形学和视觉中的核心问题，也是实现真实感重光照和编辑的前提。基于物理的逆渲染提供了一种准确的图像形成模型，但存在严重的欠约束问题：在没有强先验的情况下，照明被嵌入到材料中，重建结果在新视角和照明条件下的泛化能力较差。相比之下，数据驱动的扩散模型能够预测视觉上合理的材料，但其预测结果很少满足渲染方程，且无法直接用于基于物理的渲染。我们并不是替代这两种范式，而是将它们结合起来。我们的关键思想是将最先进的扩散模型的预测视为优化的相似性核，而非目标材料值：我们引入了一种正则化损失，惩罚在扩散预测接近恒定的表面区域内优化材料的偏差，同时允许优化自由地匹配输入图像。在这个正则化器的基础上，我们的端到端管道共同重建几何形状、材料和照明，生成高质量的资产，可以无缝融入标准渲染管道并忠实重光照。在Synthetic4Relight、Stanford-ORB和DTC-Synthetic数据集上，我们的方法在重建精度和重光照质量方面显著超越了最先进的基线。
cs.CV / 22 / 2606.31068
Hybrid Unet-Transformer Model for Generating Stress and Strain Fields from Composite Geometrics
用于从复合几何体生成应力和应变场的混合 UNet-Transformer 模型
Abstract
Accurate prediction of stress and strain fields in hierarchical composite microstructures is critical for physics-informed material design, yet conventional finite element method (FEM) simulations are computationally prohibitive at scale, requiring minutes to days per evaluation. In this work, we propose a hybrid UNet-Transformer architecture that predicts complex mechanical field distributions directly from composite microstructure geometry images, serving as an efficient surrogate for FEM across ten distinct stress and strain field types spanning diverse two-phase composite configurations including square, hexagonal, and triangular tessellations, multiple boundary conditions, and high-resolution geometries. Results demonstrate that the proposed architecture achieves strong predictive performance across the majority of subdatasets, with peak accuracy on periodic tessellation geometries reaching R2=0.9991, SSIM=0.9936, and MAE=0.0050 on the boundary condition subdataset and the triangular tessellation subdataset respectively. Across six of the eight evaluated subdatasets, MAE remains below 0.05 on the normalized [0,1] pixel scale. Encoder attention analysis via Grad-CAM and Grad-CAM++ confirms that the model develops physically meaningful internal representations, localizing attention at mechanically critical regions including phase boundaries, ligament junctions, and indenter contact zones without explicit structural supervision. Performance degrades on irregular square-grid geometries with sparse soft-phase inclusions, with the S11 normal stress subdataset yielding R2=0.7735 and SSIM=0.7126, consistent with the known limitation of smooth-loss image translation models in reproducing sharp stress discontinuities.
Chinese Translation
在分层复合微结构中准确预测应力和应变场对于物理信息驱动的材料设计至关重要，但传统的有限元方法（FEM）模拟在规模上计算成本高昂，每次评估需要几分钟到几天的时间。在本研究中，我们提出了一种混合 UNet-Transformer 架构，该架构直接从复合微结构几何图像中预测复杂的机械场分布，作为 FEM 的高效替代方案，涵盖了十种不同的应力和应变场类型，涉及多种两相复合配置，包括方形、六角形和三角形镶嵌、多种边界条件以及高分辨率几何体。结果表明，所提架构在大多数子数据集上实现了强大的预测性能，在周期性镶嵌几何体上的峰值准确度达到 R2=0.9991，SSIM=0.9936，MAE=0.0050，分别对应于边界条件子数据集和三角形镶嵌子数据集。在评估的八个子数据集中的六个上，MAE 在归一化的 [0,1] 像素尺度上保持在 0.05 以下。通过 Grad-CAM 和 Grad-CAM++ 进行的编码器注意力分析确认该模型发展出物理上有意义的内部表示，能够在机械关键区域（包括相界面、韧带交界和压头接触区）局部化注意力，而无需明确的结构监督。在具有稀疏软相夹杂物的不规则方格几何体上，性能有所下降，S11 法向应力子数据集的 R2=0.7735 和 SSIM=0.7126，与已知的平滑损失图像翻译模型在再现尖锐应力不连续性方面的局限性一致。
cs.CV / 23 / 2606.31071
Hierarchical 3D Scene Graph Construction and Belief-based Planning for Semantic Navigation
层次化三维场景图构建与基于信念的语义导航规划
Abstract
Semantic navigation is a fundamental task for embodied agents operating in unseen environments, requiring both semantic understanding and long-term decision-making. Recent foundation models have empowered agents with rich semantic priors for this task. However, without structured global representations, decision-making often falls back on local observations and greedy strategies, resulting in inefficient exploration and myopic behaviors, especially in long-distance navigation. To address these challenges, we propose a zero-shot semantic navigation framework. Our method incrementally maintains an online Hierarchical 3D Scene Graph (HSG) to form a multi-granular semantic topology over objects, zones, and regions, serving as a compact state abstraction for global planning. Building on this memory, we introduce a hierarchical belief-based planning framework that fuses semantic priors with exploration evidence on the HSG, and performs finite-horizon rollouts on an HSG-based simulator to explicitly estimate the long-term expected returns of candidate macro-actions. This enables globally consistent decisions and reduces redundant backtracking. Extensive experiments in high-fidelity simulation environments across multiple tasks and datasets demonstrate that our method outperforms existing state-of-the-art methods, particularly in long-distance scenarios, where our approach improves SR and SPL by an average of 9.4\% and 5.0\%, respectively.
Chinese Translation
语义导航是嵌入式智能体在未知环境中操作的基本任务，要求具备语义理解和长期决策能力。近期的基础模型为这一任务赋予了智能体丰富的语义先验。然而，缺乏结构化的全局表示，决策往往依赖于局部观察和贪婪策略，导致探索效率低下和短视行为，尤其是在长距离导航中。为了解决这些挑战，我们提出了一种零样本语义导航框架。我们的方法逐步维护一个在线的层次化三维场景图（HSG），以形成一个关于物体、区域和区域的多粒度语义拓扑，作为全局规划的紧凑状态抽象。在此记忆的基础上，我们引入了一个层次化的基于信念的规划框架，将语义先验与HSG上的探索证据融合，并在基于HSG的模拟器上执行有限时域的回滚，以明确估计候选宏动作的长期期望收益。这使得决策在全局上保持一致，并减少冗余的回溯。在多个任务和数据集的高保真模拟环境中的广泛实验表明，我们的方法在现有的最先进方法中表现优越，特别是在长距离场景中，我们的方法分别提高了成功率（SR）和路径长度（SPL）平均9.4%和5.0%。
cs.CV / 24 / 2606.31077
AnyMatch: Supercharging Universal Multi-Modal Image Matching with Large-Scale Single-View Images
AnyMatch：利用大规模单视图图像增强通用多模态图像匹配
Abstract
Multi-modal image matching is essential for visual localization and multi-sensor fusion, but it is hindered by the scarcity of large-scale training data with precise geometric annotations. Existing real-world datasets suffer from prohibitive costs, limited scene diversity, and errors in SfM-MVS pipelines, while synthetic methods struggle to maintain 3D geometric consistency or achieve photorealistic appearance. To address this, we propose AnyMatch, a novel framework that leverages abundant, easily accessible single-view images at minimal cost to generate rich multi-modal training data. AnyMatch integrates monocular depth estimation, 3D reprojection, diffusion-based inpainting, and crossmodal image translation to synthesize multi-view, multi-modal image pairs with 3D geometric fidelity. Crucially, our method provides annotations that strictly adhere to 3D geometric consistency through explicit 3D reprojection, avoiding SfM-MVS error accumulation. Furthermore, AnyMatch offers strong scalability, enabling controllable scene diversity and annotation difficulty via adjustable input and camera parameters. We construct Any-syn, a large-scale synthetic multi-modal dataset using AnyMatch. Experimental results show that matching networks (e.g., LoFTR, EDM, RoMa) fine-tuned on Any-syn achieve substantial performance gains on multi-modal benchmarks, exhibiting superior generalization and robustness compared to models trained on existing data.
Chinese Translation
多模态图像匹配对于视觉定位和多传感器融合至关重要，但由于缺乏具有精确几何标注的大规模训练数据而受到限制。现有的真实世界数据集面临高昂的成本、有限的场景多样性以及SfM-MVS管道中的错误，而合成方法则难以保持三维几何一致性或实现照片级真实感。为了解决这一问题，我们提出了AnyMatch，一个新颖的框架，利用丰富且易于获取的单视图图像，以最低的成本生成丰富的多模态训练数据。AnyMatch集成了单目深度估计、三维重投影、基于扩散的修补和跨模态图像翻译，以合成具有三维几何保真度的多视图多模态图像对。至关重要的是，我们的方法通过显式的三维重投影提供严格遵循三维几何一致性的标注，避免了SfM-MVS误差的累积。此外，AnyMatch提供了强大的可扩展性，通过可调的输入和相机参数实现可控的场景多样性和标注难度。我们使用AnyMatch构建了Any-syn，一个大规模合成多模态数据集。实验结果表明，在Any-syn上微调的匹配网络（例如，LoFTR、EDM、RoMa）在多模态基准测试中取得了显著的性能提升，展现出优于在现有数据上训练的模型的泛化能力和鲁棒性。
cs.CV / 25 / 2606.31082
Fleet: Few Shots Lead Effective AI-generated Image Detection
Fleet：少量样本引导有效的AI生成图像检测
Abstract
AI-generated image (AIGI) detection is undergoing a critical transition from laboratory benchmarks to open-world adversarial defense. The prevalent paradigm focuses on finding static feature spaces, assuming that some invariant artifacts learned from historical data can achieve universal zero-shot generalization. While achieving saturation on several AIGI benchmarks, this static hypothesis suffers a severe performance drop against rapidly evolving generators (e.g., SD3, Nano Banana Pro). To address these limitations, we propose that the field should expand beyond "static generalization" to a new paradigm of "dynamic adaptation". We introduce Fleet, a framework that pioneers a dynamic paradigm of continuous few-shot evolution, enabling rapid alignment with emerging generative threats. Fleet improves few-shot adaptation by replacing unconstrained feature updates with constrained routing correction, where avoidance routing redirects novel AI samples away from Non-AI-dominated routes within decoupled subspaces. To validate this, we present Treasure, a benchmark spanning 64 models and 360k images, featuring diverse architectures and 20 closed-source commercial engines. Experiments reveal that while static SOTA methods fail catastrophically on modern generators, Fleet restores performance from 20.4% to 73.1% with only 10-shot adaptation on "Doubao Seedream 4.0". Code and data are available at https://github.com/ICTMCG/Fleet .
Chinese Translation
AI生成图像（AIGI）检测正经历从实验室基准到开放世界对抗防御的关键转变。当前的主流范式集中于寻找静态特征空间，假设从历史数据中学习到的一些不变特征可以实现普遍的零样本泛化。尽管在多个AIGI基准上取得了饱和，但这一静态假设在面对快速演变的生成器（如SD3、Nano Banana Pro）时表现严重下降。为了解决这些局限性，我们建议该领域应超越“静态泛化”，转向“动态适应”的新范式。我们提出Fleet，一个开创性的框架，采用连续少量样本演化的动态范式，使其能够快速适应新兴的生成威胁。Fleet通过用受限的路由修正替代不受限的特征更新来改善少量样本适应，其中避免路由将新颖的AI样本从解耦子空间中的非AI主导路径中引导出去。为了验证这一点，我们提出了Treasure，一个涵盖64个模型和36万张图像的基准，具有多样的架构和20个闭源商业引擎。实验表明，尽管静态的SOTA方法在现代生成器上表现惨败，Fleet却能将“Doubao Seedream 4.0”上的性能从20.4%恢复至73.1%，仅需10次样本适应。代码和数据可在https://github.com/ICTMCG/Fleet获取。
cs.CV / 26 / 2606.31086
CasaMaestro: Multi-View Panoramas for House-Scale 3D Reconstruction
CasaMaestro：用于房屋规模的多视角全景3D重建
Abstract
The rise of home-deployed embodied AI systems is driving a growing need for fast, metric 3D reconstruction of residential spaces to support navigation, interaction, and long-horizon task execution. However, the commonly used pinhole-camera 3D reconstruction pipelines struggle to model large indoor residences efficiently due to their limited field of view, to which achieving full coverage across multiple rooms often requires thousands of images and incurs drift from long chains of incremental alignment. In this work, we present CasaMaestro (Spanish words meaning ``house'' and ``master''), a feedforward model that can take only twenty to fifty sparse multi-view indoor panoramas as input and directly predicts metric depth along with camera poses, allowing fast point-cloud reconstruction of the entire house with full coverage. CasaMaestro is the first model that supports house-scale reconstruction with multi-view panoramas. Experiments show that CasaMaestro can robustly provide high quality results in both real-world and synthetic scenes, which can serve as a strong foundation for acquiring house-scale 3D indoor assets to be applied in close-loop simulation.
Chinese Translation
家庭部署的具身人工智能系统的兴起推动了对住宅空间快速、度量级3D重建的日益需求，以支持导航、交互和长时间任务执行。然而，常用的针孔相机3D重建流程由于其有限的视场，难以高效地建模大型室内住宅，通常需要数千张图像才能实现多个房间的全面覆盖，并且在长链增量对齐中会产生漂移。在本研究中，我们提出了CasaMaestro（西班牙语意为“房子”和“大师”），这是一种前馈模型，可以仅使用二十到五十张稀疏的多视角室内全景作为输入，直接预测度量深度和相机姿态，从而实现整个房屋的快速点云重建，确保全面覆盖。CasaMaestro是第一个支持使用多视角全景进行房屋规模重建的模型。实验表明，CasaMaestro能够在真实场景和合成场景中稳健地提供高质量的结果，这为获取可应用于闭环仿真的房屋规模3D室内资产奠定了坚实的基础。
cs.CV / 27 / 2606.31088
Towards Flexible, Natural, Efficient Interaction for Conversational Talking Face Generation
朝着灵活、自然、高效的对话式说话人脸生成交互
Abstract
Conversational talking face generation has recently attracted increasing attention, aiming to synthesize interactive talking videos where characters speak, listen, and respond dynamically to each other. This task presents three core challenges: 1) Flexibility: enabling multi-round dialogues with an arbitrary number of participants; 2) Naturalness: maintaining coherent motion and appropriate non-verbal feedback throughout the interaction; and 3) Efficiency: achieving real-time generation and low computation overhead for long-term continuous online conversation. Despite recent advances, existing methods still fall short in balancing all three requirements. To bridge this gap, we introduce InterTalk, a novel and efficient framework designed for highly interactive conversational talking face generation. Built upon a motion-based architecture, InterTalk supports real-time conversation synthesis. Our method achieves strong flexibility by explicitly modeling multi-round conversational dynamics among each participant, eliminating constraints on their numbers. To enhance interactivity, we incorporate motion feedback from multiple participants and introduce an iterative generation strategy for more natural behaviors. Besides, we disentangle motion into several facial components, enabling targeted refinements for natural response such as precise lip sync and realistic eye blinking. Finally, we construct a new multi-person conversational dataset and enrich it with 3D face-based data augmentation. Extensive experiments demonstrate that InterTalk achieves superior interaction quality while maintaining real-time performance at 30 FPS.
Chinese Translation
对话式说话人脸生成近年来受到越来越多的关注，旨在合成互动说话视频，其中角色之间动态地说话、倾听和回应。该任务面临三个核心挑战：1）灵活性：支持任意数量参与者的多轮对话；2）自然性：在整个互动过程中保持连贯的动作和适当的非语言反馈；3）效率：实现实时生成并在长期连续在线对话中保持低计算开销。尽管最近取得了一些进展，现有方法在平衡这三项要求方面仍显不足。为了解决这一问题，我们提出了InterTalk，一个新颖且高效的框架，旨在实现高度互动的对话式说话人脸生成。InterTalk基于运动驱动的架构，支持实时对话合成。我们的方法通过明确建模每个参与者之间的多轮对话动态，消除了对参与者数量的限制，从而实现了强大的灵活性。为了增强互动性，我们结合了来自多个参与者的运动反馈，并引入了迭代生成策略，以实现更自然的行为。此外，我们将运动分解为多个面部组件，使得可以针对自然反应进行有针对性的细化，例如精确的唇同步和逼真的眨眼。最后，我们构建了一个新的多人人对话数据集，并通过基于3D人脸的数据增强进行了丰富。大量实验表明，InterTalk在保持30 FPS实时性能的同时，实现了卓越的互动质量。
cs.CV / 28 / 2606.31089
Anchoring on Reality: Breaking the Pseudo-Target Ceiling in Makeup Transfer
锚定现实：打破化妆迁移中的伪目标天花板
Abstract
Makeup transfer applies a reference cosmetic style to a source face while preserving its identity and geometry. However, this task is severely hindered by the lack of real paired training data. Current methods rely on either weak priors or synthetic pseudo-targets from large-scale editing models. These paradigms provide suboptimal guidance, often leading to degraded fine-grained details, synthetic artifacts, and identity drift. To this end, we propose Anchoring on Reality Makeup Transfer (ART), a two-stage framework with a reality-anchored refinement cycle. In Stage I, the model is initialized with pseudo-targets to establish basic semantic alignment and global makeup placement. Crucially, Stage II shifts supervision from pseudo-targets to the real reference, reconstructing it from its bare-skin counterpart through a differentiable cycle that penalizes any omitted detail and overrides synthetic artifacts. Furthermore, we introduce MakeupFaces2K (MF2K), the first 2K-resolution in-the-wild makeup portrait dataset comprising 8,573 images. Extensive experiments demonstrate that our method achieves superior makeup fidelity, strong background stability, and robust identity preservation, especially for complex makeup styles.
Chinese Translation
化妆迁移将参考化妆风格应用于源面孔，同时保持其身份和几何特征。然而，这一任务受到真实配对训练数据缺乏的严重阻碍。目前的方法依赖于弱先验或来自大规模编辑模型的合成伪目标。这些范式提供的指导效果不佳，常常导致细节退化、合成伪影和身份漂移。为此，我们提出了锚定现实化妆迁移（Anchoring on Reality Makeup Transfer, ART），这是一种具有现实锚定精细化周期的两阶段框架。在第一阶段，模型使用伪目标初始化，以建立基本的语义对齐和全局化妆布局。关键的是，第二阶段将监督从伪目标转移到真实参考，通过可微分的周期从其裸肤对应物重建，惩罚任何遗漏的细节并覆盖合成伪影。此外，我们引入了 MakeupFaces2K (MF2K)，这是第一个包含 8,573 张图像的 2K 分辨率野外化妆肖像数据集。大量实验表明，我们的方法在化妆保真度、背景稳定性和身份保持方面表现优越，尤其适用于复杂的化妆风格。
cs.CV / 29 / 2606.31095
Do Not Break the Vessels: Structure-Preserving Mean Flow for Vascular Image Translation
不破坏血管：用于血管图像翻译的结构保持均流
Abstract
Reconstructing anatomically faithful vascular structures from clinically accessible imaging modalities is of substantial clinical significance. However, existing cross-modal translation methods mainly emphasize pixel-level fidelity or visual realism and treat structure preservation as a property of the final output rather than an invariant of the generative process. This limitation often leads to structural discontinuities and artifacts, compromising anatomical coherence and clinical reliability. In this work, we propose a Structure-Preserving Mean Flow (SPMF) framework that formulates vascular image translation as a topology-invariant transport process. Based on a structural invariance principle, we derive an orthogonality constraint on the flow velocity field that formally separates appearance transport from topological distortion. We implement this constraint as a time-weighted surrogate objective within a Brownian bridge diffusion model to preserve topology at every diffusion step. Moreover, we propose a Prototype-Guided Structural Refinement (PGSR) module to align degraded inference-time structures with reliable training-time structures. Experiments on paired NIRII-to-2PF and fundus datasets demonstrate consistent improvements over state-of-the-art methods, achieving peak PSNR values of 24.96 dB and 24.83 dB, respectively.
Chinese Translation
从临床可获取的成像方式重建解剖学上真实的血管结构具有重要的临床意义。然而，现有的跨模态翻译方法主要强调像素级的保真度或视觉真实感，将结构保持视为最终输出的属性，而非生成过程的不变性。这一局限性常常导致结构不连续和伪影，损害了解剖学的一致性和临床可靠性。在本研究中，我们提出了一种结构保持均流（Structure-Preserving Mean Flow, SPMF）框架，将血管图像翻译形式化为一种拓扑不变的传输过程。基于结构不变性原理，我们推导出流速场的正交约束，正式将外观传输与拓扑扭曲分离。我们在布朗桥扩散模型中将此约束实现为时间加权的替代目标，以在每个扩散步骤中保持拓扑。此外，我们提出了一种原型引导的结构细化（Prototype-Guided Structural Refinement, PGSR）模块，以将退化的推理时结构与可靠的训练时结构对齐。在配对的NIRII到2PF和眼底数据集上的实验表明，相较于最先进的方法，我们实现了一致的改进，分别达到了24.96 dB和24.83 dB的峰值信噪比（PSNR）值。
cs.CV / 30 / 2606.31096
Horizon3D: Sparse Radar-Camera Fusion for Long-Range 3D Perception in Autonomous Driving
Horizon3D：用于自主驾驶的长距离3D感知的稀疏雷达-相机融合
Abstract
Long-range 3D object detection is critical for safe autonomous driving at highway speeds, yet existing radar-camera fusion methods remain limited at extended ranges. BEV-based methods capture scene-level context but incur rapidly growing computation and often lose fine-grained object detail, while query-based methods are efficient but provide limited scene-level context. Temporal fusion further requires both multi-frame accumulation for sparse distant observations and object-level motion modeling for fast-moving objects. We propose Horizon3D, a sparse radar-camera fusion framework for long-range 3D object detection that combines Gaussian primitives with sparse BEV features. Horizon3D initializes Gaussian primitives at radar- and camera-estimated object keypoints using Keypoint-Guided Gaussian Initialization, refines them through Object-Centric Sparse Fusion, and splats them onto the BEV plane to fuse object-level detail with sparse radar BEV context. It further introduces Dual-Path Temporal Fusion, which aggregates temporal cues through a BEV path for scene-level accumulation and a Gaussian path for object-level motion propagation. Experiments on TruckScenes show that Horizon3D achieves state-of-the-art radar-camera 3D detection performance. On the validation set, it outperforms the previous best method by +3.0 NDS and +1.6 mAP while maintaining competitive inference speed.
Chinese Translation
长距离3D物体检测对于高速公路上的安全自主驾驶至关重要，但现有的雷达-相机融合方法在扩展范围内仍然有限。基于鸟瞰视图（BEV）的方法能够捕捉场景级上下文，但计算量迅速增加，且往往会丢失细粒度的物体细节，而基于查询的方法效率较高，但提供的场景级上下文有限。时间融合进一步需要对稀疏远距离观测进行多帧累积，并对快速移动物体进行物体级运动建模。我们提出了Horizon3D，一种用于长距离3D物体检测的稀疏雷达-相机融合框架，结合了高斯原语与稀疏的BEV特征。Horizon3D使用关键点引导的高斯初始化（Keypoint-Guided Gaussian Initialization）在雷达和相机估计的物体关键点处初始化高斯原语，通过以物体为中心的稀疏融合（Object-Centric Sparse Fusion）对其进行精炼，并将其投影到BEV平面上，以融合物体级细节与稀疏雷达BEV上下文。它进一步引入了双路径时间融合（Dual-Path Temporal Fusion），通过BEV路径聚合时间线索以进行场景级累积，并通过高斯路径进行物体级运动传播。在TruckScenes上的实验表明，Horizon3D实现了最先进的雷达-相机3D检测性能。在验证集上，其性能比之前的最佳方法提高了+3.0 NDS和+1.6 mAP，同时保持了竞争力的推理速度。
cs.CV / 31 / 2606.31098
PiLoT v2: Pixel-to-Orthogonal Map Alignment for Free-view UAV Geo-localization
PiLoT v2：用于自由视角无人机地理定位的像素到正交地图对齐
Abstract
Real-time, drift-free UAV geo-localization is essential for autonomous missions in GNSS-denied environments. The pioneering system, PiLoT, achieves high precision via Neural Pixel-to-3D Registration, aligning UAV video streams with a single rendered reference view from 3D meshes. However, its reliance on heavy 3D meshes incurs massive storage overheads, complex map acquisition, and significant computational rendering costs, severely hindering deployment on embedded platforms. To address these bottlenecks, we propose PiLoT v2, a lightweight yet robust evolution that shifts the paradigm to direct pixel-to-orthogonal map registration for free-view UAV geo-localization. By leveraging True Digital Orthophoto Maps (TDOMs) and Digital Surface Models (DSMs) as the reference substrate, PiLoT v2 replaces GPU-intensive 3D rendering with a highly efficient, CPU-friendly map cropping operation. To bridge the severe geometric discrepancy between these 2.5D orthogonal crops and free-view oblique UAV imagery, we train a cross-view feature registration network using a novel, large-scale geometrically annotated dataset. Furthermore, we integrate onboard sensor prior--specifically gravity direction and single-point laser rang--directly into the pose optimization manifold to enhance robustness against cross-view visual degradation. Experimental results demonstrate that PiLoT v2 achieves performance comparable to, or even exceeding, its Pixel-to-3D predecessor, while offering drastically lower storage and computational costs.
Chinese Translation
实时、无漂移的无人机地理定位对于在GNSS拒绝环境中的自主任务至关重要。开创性的系统PiLoT通过神经像素到3D配准实现了高精度，将无人机视频流与来自3D网格的单一渲染参考视图对齐。然而，其对重型3D网格的依赖导致了巨大的存储开销、复杂的地图获取和显著的计算渲染成本，严重阻碍了在嵌入式平台上的部署。为了解决这些瓶颈，我们提出了PiLoT v2，这是一种轻量级但强大的演变，改变了直接像素到正交地图配准的范式，以实现自由视角无人机地理定位。通过利用真实数字正射影像图（True Digital Orthophoto Maps, TDOMs）和数字表面模型（Digital Surface Models, DSMs）作为参考基础，PiLoT v2用高效的、友好的CPU地图裁剪操作替代了GPU密集型的3D渲染。为了弥补这些2.5D正交裁剪与自由视角倾斜无人机图像之间的严重几何差异，我们使用一种新颖的大规模几何标注数据集训练了一个跨视图特征配准网络。此外，我们将机载传感器先验——特别是重力方向和单点激光测距——直接集成到姿态优化流形中，以增强对跨视图视觉退化的鲁棒性。实验结果表明，PiLoT v2的性能可与其像素到3D前身相媲美，甚至超过，同时提供了显著更低的存储和计算成本。
cs.CV / 32 / 2606.31099
Seeing Through Multiple Views: Parameter-Efficient Fine-Tuning via Selective Neurons for Consistent Radiology Report Generation
透视多视角：通过选择性神经元实现参数高效的微调以生成一致的放射学报告
Abstract
Recent years have seen substantial advances in radiology report generation (RRG), yet existing approaches predominantly adopt direct feature fusion when handling multi-view X-ray images. Such approaches overlook the potential clinical inconsistencies and inaccuracies arising when a single model processes different views, adversely impacting performance and clinical reliability. To this end, we introduce View-PNDF (View-specific Pattern Neuron Detection and Fine-tuning), a parameter-efficient framework that fosters view-consistent report generation from a neuronal perspective. Specifically, View-PNDF comprises: (i) a view-specific neuron detection module identifying neurons responsive to particular views, (ii) a verification module quantifying the existence of these neurons, and (iii) a selective fine-tuning strategy strengthening detected neurons while preserving view-agnostic representations. By updating only view-specific neurons, View-PNDF achieves consistent diagnoses across different views with reduced computational costs. Subsequently, we employ Large Language Models (LLMs) to consolidate the view-specific reports into a complete radiology report. Furthermore, we use traditional Natural Language Generation (NLG) metrics-based assessment on integrated reports for baseline comparison and employ LLM-based assessment (e.g., GPT-4o) on view-specific reports to capture clinical significance. Extensive experiments on two medical RRG benchmarks demonstrate that View-PNDF substantially improves view-specific chest X-ray report generation quality while maintaining robust general-view performance.
Chinese Translation
近年来，放射学报告生成（RRG）取得了显著进展，但现有方法在处理多视角X光图像时主要采用直接特征融合。这些方法忽视了当单一模型处理不同视角时可能出现的临床不一致性和不准确性，从而对性能和临床可靠性产生负面影响。为此，我们提出了View-PNDF（视角特定模式神经元检测与微调），这是一个从神经元角度促进视角一致报告生成的参数高效框架。具体而言，View-PNDF包括：（i）视角特定神经元检测模块，识别对特定视角响应的神经元；（ii）验证模块，量化这些神经元的存在；（iii）选择性微调策略，增强检测到的神经元，同时保留视角无关的表示。通过仅更新视角特定神经元，View-PNDF在降低计算成本的同时，实现了不同视角之间的一致诊断。随后，我们采用大型语言模型（LLMs）将视角特定报告整合为完整的放射学报告。此外，我们使用基于传统自然语言生成（NLG）指标的评估对整合报告进行基线比较，并在视角特定报告上进行LLM基于评估（例如，GPT-4o），以捕捉临床意义。在两个医学RRG基准上的广泛实验表明，View-PNDF显著提高了视角特定胸部X光报告生成的质量，同时保持了强大的通用视角性能。
cs.CV / 33 / 2606.31100
TaxoMIL: Taxonomy-Constrained Learning for Hierarchical Whole Slide Image Analysis
TaxoMIL：用于分层全切片图像分析的分类约束学习
Abstract
Whole slide image (WSI) analysis is central to computational pathology, with multiple instance learning (MIL) emerging as the standard pipeline for slide-level diagnosis. However, conventional approaches formulate WSI diagnosis as a flat classification task over discrete labels, contradicting the inherently hierarchical, coarse-to-fine nature of clinical reasoning. Although recent hierarchical classifiers and vision-language models (VLMs) have sought to address this structural gap, they either fail to capture semantic continuity between related diagnoses or suffer from unconstrained text generation that produces taxonomic hallucinations and parent-child label violations. To address these limitations, we propose TaxoMIL, a taxonomy-constrained framework that reformulates WSI diagnosis as a multi-granularity text generation task. TaxoMIL utilizes a dual-head Transformer decoder to generate coarse- and fine-level diagnostic text, and introduces taxonomy-guided objectives that explicitly structure the label embedding space and strictly ground slide-level visual representations within the clinical taxonomy. Extensive experiments across three diverse WSI datasets demonstrate that TaxoMIL consistently outperforms state-of-the-art MIL classifiers and VLM-based generative methods, yielding accurate and hierarchy-aware diagnostic predictions. The code is released at https://github.com/QuIIL/TaxoMIL
Chinese Translation
全切片图像（WSI）分析是计算病理学的核心，多实例学习（MIL）已成为切片级诊断的标准流程。然而，传统方法将WSI诊断表述为对离散标签的平面分类任务，这与临床推理固有的分层、粗到细的特性相矛盾。尽管最近的分层分类器和视觉-语言模型（VLMs）试图解决这一结构性差距，但它们要么无法捕捉相关诊断之间的语义连续性，要么受到不受约束的文本生成的影响，导致分类幻觉和父子标签违规。为了解决这些局限性，我们提出了TaxoMIL，这是一种分类约束框架，将WSI诊断重新表述为多粒度文本生成任务。TaxoMIL利用双头Transformer解码器生成粗级和细级诊断文本，并引入分类引导目标，明确构建标签嵌入空间，并严格将切片级视觉表示与临床分类相结合。在三个不同的WSI数据集上进行的广泛实验表明，TaxoMIL始终优于最先进的MIL分类器和基于VLM的生成方法，产生准确且考虑层次的诊断预测。代码已发布在 https://github.com/QuIIL/TaxoMIL
cs.CV / 34 / 2606.31109
InfiniVerse: Occupancy Guided Unbounded Scene Generation for Autonomous Driving
InfiniVerse：面向自动驾驶的占用引导无界场景生成
Abstract
Generating realistic, controllable, and temporally coherent urban environments is a critical yet unresolved challenge in the autonomous driving community. In this paper, we introduce InfiniVerse, a unified pipeline for long-range, 2D-3D-aligned, and controllable synthesis of dynamic urban scenes from a single frame. In practice, our approach first reconstructs a 3D occupancy representation from the input multi-view frame. This representation serves as a foundation for autoregressive scene extension along arbitrary trajectories. Subsequently, a video diffusion model translates the coarse occupancy grid into realistic, spatiotemporally consistent video sequences. Moreover, we propose a hierarchical sketch-and-refine paradigm, in which the generated videos are re-projected as image-conditioned feedback to enhance the 3D occupancy representation, establishing cross-modal alignment and mutual enhancement between the visual and spatial domains. Extensive evaluations on the Waymo Open Dataset and nuScenes demonstrate that InfiniVerse achieves state-of-the-art performance, with a FID of 6.4 and FVD of 67.97, significantly outperforming existing benchmarks in both duration and stability.
Chinese Translation
生成逼真、可控且时间一致的城市环境是自动驾驶领域一个关键但尚未解决的挑战。本文介绍了InfiniVerse，一个统一的管道，用于从单帧图像合成长距离、2D-3D对齐且可控的动态城市场景。实际上，我们的方法首先从输入的多视角帧重建3D占用表示。该表示作为沿任意轨迹自回归场景扩展的基础。随后，视频扩散模型将粗略的占用网格转换为逼真且时空一致的视频序列。此外，我们提出了一种分层的草图与精炼范式，其中生成的视频被重新投影为图像条件反馈，以增强3D占用表示，建立视觉和空间领域之间的跨模态对齐与相互增强。在Waymo开放数据集和nuScenes上的广泛评估表明，InfiniVerse实现了最先进的性能，FID为6.4，FVD为67.97，在持续时间和稳定性上显著超越现有基准。
cs.CV / 35 / 2606.31115
JacobianAvatar: Temporally Consistent Semi-rigid Avatar Reconstruction from a Monocular Video
JacobianAvatar：基于单目视频的时间一致性半刚性虚拟人重建
Abstract
Generating realistic human avatars in complex motions--such as clothing dynamics--requires modeling of global and local deformations which remains challenging in monocular settings. We address this problem by leveraging neural Jacobian fields (NJFs) for representing semi-rigid deformations. We train self-supervised neural networks for predicting Jacobian matrices that give the pose-dependent deformations, by solving a Poisson equation. However, monocular input presents several difficulties such as self-occluded regions and invisible surfaces. To address these issues, we introduce three key components: a constrained Poisson solver, signed distance-based Jacobian regularization, and a deformation-guided residual flow loss, which together suppress boundary artifacts, recover frequently occluded regions such as armpits and thighs, and enforce temporal consistency during motion. Experiments on benchmark and in-the-wild videos demonstrate that our method generates temporally stable and geometrically coherent avatars, outperforming state-of-the-art approaches.
Chinese Translation
在复杂运动（如服装动态）中生成逼真的人类虚拟人需要对全局和局部变形进行建模，这在单目设置中仍然具有挑战性。我们通过利用神经雅可比场（Neural Jacobian Fields, NJFs）来表示半刚性变形，从而解决这一问题。我们训练自监督神经网络来预测雅可比矩阵，以获得依赖于姿态的变形，通过求解泊松方程。然而，单目输入存在一些困难，例如自遮挡区域和不可见表面。为了解决这些问题，我们引入了三个关键组件：约束泊松求解器、基于有符号距离的雅可比正则化，以及变形引导的残差流损失，这些组件共同抑制边界伪影，恢复经常被遮挡的区域（如腋下和大腿），并在运动过程中强制执行时间一致性。在基准测试和真实场景视频上的实验表明，我们的方法生成了时间稳定且几何一致的虚拟人，超越了现有的最先进方法。
cs.CV / 36 / 2606.31125
WildProp: Visual Estimation of Wildlife Body Proportions at Scale
WildProp：大规模野生动物体型比例的视觉估计
Abstract
Population-level morphometric measurements underpin ecological and evolutionary studies but traditionally require controlled imaging or physical specimen handling, limiting scalability. We present WildProp, a training-free framework that estimates wildlife body proportion distributions directly from large-scale, unconstrained image repositories. We cast morphometric estimation as a retrieval-driven correspondence problem: given a single user-annotated canonical image, WildProp performs pose-aware retrieval using foundation model features, transfers part endpoints via dense patch-level matching, filters predictions using geometric consistency, and aggregates measurements across retrieved images to estimate population-level ratio distributions. Unlike supervised keypoint pipelines, our approach adapts to arbitrary species and user-defined parts without per-species training. Evaluations on three large morphometric datasets spanning birds and amphibians show median relative errors of 10-20%. We further highlight the broad applicability of our approach through a number of case studies measuring various proportions across diverse taxa, including birds, frogs, insects, and flowers. Ablations demonstrate that pose-aware retrieval is critical for stable estimation, while robust aggregation mitigates keypoint and pose noise. Our results indicate that carefully curated 2D correspondences over web-scale imagery can provide scalable morphometric proxies for comparative and subgroup analyses across taxa, geography, and seasonality.
Chinese Translation
种群级别的形态测量是生态和进化研究的基础，但传统上需要受控成像或物理样本处理，限制了其可扩展性。我们提出了WildProp，一个无训练框架，能够直接从大规模、无约束的图像库中估计野生动物体型比例分布。我们将形态测量视为一个基于检索的对应问题：给定一张用户标注的典型图像，WildProp利用基础模型特征进行姿态感知检索，通过密集的补丁级匹配转移部位端点，使用几何一致性过滤预测，并在检索到的图像中聚合测量，以估计种群级别的比例分布。与监督关键点管道不同，我们的方法能够适应任意物种和用户定义的部位，而无需针对每个物种进行训练。在涵盖鸟类和两栖动物的三个大型形态数据集上的评估显示，中位相对误差为10-20%。我们进一步通过多个案例研究强调了我们方法的广泛适用性，这些案例研究测量了包括鸟类、青蛙、昆虫和花卉在内的不同类群的各种比例。消融实验表明，姿态感知检索对稳定估计至关重要，而稳健的聚合则减轻了关键点和姿态噪声。我们的结果表明，精心策划的二维对应关系在网络规模的图像上可以为跨类群、地理和季节性比较及子群分析提供可扩展的形态代理。
cs.CV / 37 / 2606.31127
SkillSpotter: Pose-Aware Multi-View Skilled Action Detection and Grading in Ego-Exo Videos
SkillSpotter：基于姿态的多视角熟练动作检测与评分在自我-外部视频中的应用
Abstract
To enable personalized, real-time coaching using Augmented Reality glasses or fixed camera setups in domains such as sports, cooking, or music, a system must understand not just what a person does, but how well they execute an activity. In an ego-exo video setting, this requires simultaneously detecting individual skilled actions and classifying each as correct or needing improvement, which Ego-Exo4D's proficiency demonstration benchmark formalized. We first adapt seven state-of-the-art temporal action detection architectures to this task, extend the evaluation protocol to disentangle detection from grading, and show that existing methods grade near-randomly. We then introduce SkillSpotter, a pose-aware multi-view architecture that jointly detects and grades skilled actions through three task-specific modules: (1) adaptive temporal suppression to handle the varying density of skilled actions across diverse activities, (2) gated 3D body pose fusion to leverage body kinematics as a complementary signal to visual features, and (3) bidirectional cross-view attention to combine ego and exo views effectively. SkillSpotter improves class-specific mAP from 12.40 to 21.82 (+76%) and balanced accuracy from 55.99% to 60.40% over the best baseline. SkillSpotter's modules transfer to other temporal action detection models with consistent gains, and our method generalizes beyond Ego-Exo4D to HoloAssist. Code: https://github.com/eth-siplab/SkillSpotter
Chinese Translation
为了在体育、烹饪或音乐等领域通过增强现实眼镜或固定摄像头设置实现个性化的实时指导，系统不仅需要理解一个人所做的事情，还要评估他们执行活动的质量。在自我-外部视频环境中，这要求同时检测个体熟练动作，并将每个动作分类为正确或需要改进，这一需求由Ego-Exo4D的熟练度演示基准进行了形式化。我们首先将七种最先进的时间动作检测架构适配于该任务，扩展评估协议以区分检测与评分，并展示现有方法的评分接近随机。随后，我们引入了SkillSpotter，这是一种基于姿态的多视角架构，通过三个特定任务的模块共同检测和评分熟练动作：(1) 自适应时间抑制，以处理不同活动中熟练动作的密度变化；(2) 门控3D身体姿态融合，以利用身体运动学作为视觉特征的补充信号；(3) 双向交叉视图注意力，以有效结合自我视角和外部视角。SkillSpotter将类别特定的平均精度均值（mAP）从12.40提高到21.82（+76%），将平衡准确率从55.99%提高到60.40%，超越了最佳基线。SkillSpotter的模块能够迁移到其他时间动作检测模型中，并且在性能上保持一致的提升，我们的方法也超越了Ego-Exo4D，适用于HoloAssist。代码： https://github.com/eth-siplab/SkillSpotter
cs.CV / 38 / 2606.31135
MSNN-LINet: Cross-Modal Learning via Continuous Linear Integration
MSNN-LINet：通过连续线性集成进行跨模态学习
Abstract
We present LINet (Linear Integration Network), a Multi-Stream Neural Network (MSNN) for RGB-D scene classification. Current multi-modal architectures treat feature fusion as a discrete, ad-hoc event: early fusion entangles representations prematurely, late fusion isolates them until the final layer, and hybrid or attention-based methods require architectural guesswork to place intermediate fusion blocks. LINet addresses this structural compromise by maintaining three dedicated parallel streams (RGB, depth, and integration) where a novel Linear Integration Convolution (LIConv2d) operator enables continuous cross-modal learning at every layer. The integration stream receives raw filtered signals from both modality streams and combines them before the nonlinear activation threshold, conceptually inspired by somatic integration preceding the neuronal firing decision. Implementing continuous integration exposes a critical initialization pathology: Kaiming initialization of the bridging weights scrambles gradients before they reach the stream backbones, producing a failure mode that resembles overfitting but is corrupted gradient flow. A 1/N constant initialization mitigates this. We employ progressive modality dropout, a curriculum adapted to continuous fusion in which blanking probability increases from zero, preventing pathway collapse, a form of negative co-learning, by forcing robust independent stream representations. Trained from scratch on SUN RGB-D 19-class scene classification, LINet reaches 45.2% mean class accuracy at ResNet18 scale, outperforming prior from-scratch results, and rises to 49.6% with in-domain RGB-D (ScanNet) pretraining.
Chinese Translation
我们提出了LINet（线性集成网络），一种用于RGB-D场景分类的多流神经网络（MSNN）。当前的多模态架构将特征融合视为离散的、临时的事件：早期融合过早地纠缠表示，晚期融合在最终层之前将其隔离，而混合或基于注意力的方法则需要在架构上进行猜测，以放置中间融合块。LINet通过维持三个专用的并行流（RGB、深度和集成）来解决这一结构性妥协，其中一种新颖的线性集成卷积（LIConv2d）算子使得在每一层都能进行连续的跨模态学习。集成流接收来自两个模态流的原始过滤信号，并在非线性激活阈值之前将其结合，这在概念上受到神经元放电决策之前的体感整合的启发。实施连续集成暴露出一个关键的初始化病理：桥接权重的Kaiming初始化在梯度到达流主干之前就打乱了梯度，产生了一种类似于过拟合的失败模式，但实际上是梯度流的损坏。1/N常量初始化可以缓解这一问题。我们采用渐进模态丢弃，这是一种适应于连续融合的课程，其中空白概率从零开始增加，防止路径崩溃，这是一种负协同学习的形式，通过强制独立流表示的稳健性来实现。LINet在SUN RGB-D 19类场景分类上从零开始训练，达到了45.2%的平均类别准确率，在ResNet18规模下超越了先前的从零开始结果，并在领域内RGB-D（ScanNet）预训练后上升至49.6%。
cs.CV / 39 / 2606.31136
FROST: Training-Free Few-Shot Segmentation with Frozen Features and Nonparametric Statistics
FROST：基于冻结特征和非参数统计的无训练少样本分割
Abstract
Few-shot segmentation asks a model to delineate a target class in a query image from only a handful of annotated examples, a setting most acute in remote sensing, where labels are scarce and the imagery departs sharply from the natural images on which vision backbones are pretrained. Prevailing approaches either train a segmenter on labelled episodes, which raises accuracy within the training distribution but binds the model to it, or reduce each class to a lossy summary of frozen features, a single prototype, a few cluster prototypes, or a discrete clustering, none of which preserves the internal structure of a multimodal class. We argue that a class is better described by a distribution than by a point, and that frozen self-supervised features already carry enough structure to estimate that distribution directly. We introduce FROST, a training-free few-shot segmenter that treats the reference foreground and background as two point clouds on the unit sphere of frozen DINOv3 features and labels each query token by a nonparametric density ratio, with a threshold the Bayes rule fixes at zero under equal priors. Because the variance of a density estimate shrinks as its sample grows, the decision sharpens as references accumulate, and every remaining quantity from the kernel bandwidth to the spatial gate is read from the support set rather than tuned. We develop FROST for overhead imagery, where a class is typically a scatter of many small and dissimilar instances that a density tracks but a lossy summary blurs. Across seventeen remote-sensing benchmarks FROST surpasses both training-free and learning-based methods, leading by 5.6 mIoU from a single annotated example and widening its lead as the support set grows, all while remaining among the smallest models compared. Code is available at https://github.com/jhpark-ai/FROST.
Chinese Translation
少样本分割要求模型仅通过少量标注示例在查询图像中勾勒出目标类别，这在遥感领域尤为突出，因为标签稀缺且图像与视觉骨干网络预训练的自然图像有显著差异。现有方法要么在标注的情境上训练分割器，这提高了训练分布内的准确性，但使模型受限于此；要么将每个类别简化为损失信息的冻结特征摘要，如单个原型、几个聚类原型或离散聚类，这些方法都无法保留多模态类别的内部结构。我们认为，类别更适合用分布而非点来描述，并且冻结的自监督特征已经携带足够的结构来直接估计该分布。我们提出了FROST，这是一种无训练的少样本分割器，将参考前景和背景视为冻结DINOv3特征的单位球面上的两个点云，并通过非参数密度比为每个查询标记，阈值由贝叶斯规则在相等先验下固定为零。由于密度估计的方差随着样本的增加而减小，决策随着参考的积累而变得更加明确，而从核带宽到空间门的每个剩余量都是从支持集读取而非调优。我们为高空图像开发了FROST，其中一个类别通常是许多小而不同实例的散布，密度可以追踪这些实例，而损失摘要则模糊了它们。在十七个遥感基准测试中，FROST超越了无训练和基于学习的方法，从单个标注示例开始领先5.6 mIoU，并随着支持集的增长而扩大领先优势，同时仍然是比较中最小的模型之一。代码可在 https://github.com/jhpark-ai/FROST 获取。
cs.CV / 40 / 2606.31147
WaterGen: Decoupling Scene and Medium in Underwater Image Generation
WaterGen：在水下图像生成中解耦场景与介质
Abstract
Underwater computer vision tasks, such as detection, restoration, and segmentation, are limited by the scarcity of large-scale and diverse training data. We introduce WaterGen, a method for generating large-scale, realistic, and diverse underwater images that provides independent control of the scene and water medium conditions. Our approach treats underwater image generation as the decoupled control of two factors: realistic and diverse scene content (what is in the image), and accurate and controllable water medium effects (what the water does to the image). Existing methods generally achieve only part of this objective: they either provide controllability with limited realism or diversity, or generate realistic scenes without accurately and independently modeling water-medium effects. Our key insight, that allows us to avoid this compromise, is that scene generation and medium modeling can be decoupled within a latent diffusion framework, enabling diverse scene generation together with accurate and controllable underwater appearance. To do this, we decompose underwater image synthesis into two stages. First, we fine-tune the latent diffusion U-Net using degradation-free underwater images so that it learns to generate diverse and realistic latent embeddings of underwater scene content without medium-induced degradation. Second, we formulate the physically accurate medium degradation synthesis as a conditional decoding process applied to these latent embeddings. This decoupled design allows our model to generate diverse scenes with full control of underwater appearance. We leverage WaterGen to build large-scale synthetic underwater datasets that are diverse in scene structures and accurate in water effects and pseudo-labels. We demonstrate that our synthetic data consistently improve downstream performance in underwater restoration and semantic segmentation.
Chinese Translation
水下计算机视觉任务，如检测、修复和分割，受到大规模和多样化训练数据匮乏的限制。我们提出了WaterGen，一种生成大规模、真实且多样化水下图像的方法，该方法提供了对场景和水介质条件的独立控制。我们的方法将水下图像生成视为两个因素的解耦控制：真实且多样的场景内容（图像中包含的内容）和准确且可控的水介质效果（水对图像的影响）。现有的方法通常只能实现部分目标：它们要么在有限的真实感或多样性中提供可控性，要么生成真实场景而没有准确且独立地建模水介质效果。我们避免这种妥协的关键见解是，场景生成和介质建模可以在潜在扩散框架内解耦，从而实现多样化场景生成以及准确且可控的水下外观。为此，我们将水下图像合成分解为两个阶段。首先，我们使用无降解的水下图像微调潜在扩散U-Net，使其学习生成多样且真实的水下场景内容的潜在嵌入，而不受介质引起的降解影响。其次，我们将物理准确的介质降解合成表述为应用于这些潜在嵌入的条件解码过程。这种解耦设计使我们的模型能够生成多样的场景，并完全控制水下外观。我们利用WaterGen构建了大规模合成水下数据集，这些数据集在场景结构上多样化，并在水效应和伪标签上准确。我们证明我们的合成数据在水下修复和语义分割的下游任务中始终提高了性能。
cs.CV / 41 / 2606.31148
PruneGround: Plug-and-play Spatial Pruning for 3D Visual Grounding
PruneGround：即插即用的3D视觉定位空间剪枝
Abstract
3D Visual Grounding (3DVG) aims to localize target objects in 3D scenes given natural language descriptions. Existing approaches typically perform reasoning over the entire scene, leading to ambiguous predictions and high computational cost, especially in cluttered environments. We observe that many referential expressions rely on local spatial context and often correspond to restricted spatial regions rather than the full scene. Motivated by this insight, we propose PruneGround, an effective plug-and-play framework for 3DVG built upon three key components. First, we introduce Language-Guided Spatial Pruning (LGSP), which leverages a frozen Vision Language Model (VLM) to identify language-relevant regions, thereby reducing spatial computation and grounding candidates in the narrower search space. Second, we propose MultiView-Conditioned Description Reformulation (MCDR), which decomposes complex expressions into simplified target-anchor relations and augments missing spatial cues through multi-view reasoning. Finally, we propose LLM-Grounder, which repurposes a detection-pretrained spatial LLM into a language-conditioned grounding model by aligning point cloud and linguistic representations within the pruned region. Extensive experiments on the three most popular point cloud benchmarks demonstrate that our method achieves state-of-the-art results on all three ScanRefer settings and on 9 out of 10 Nr3D/Sr3D settings. Code and models are publicly available: https://github.com/leduckhai/PruneGround
Chinese Translation
3D视觉定位（3DVG）旨在根据自然语言描述在3D场景中定位目标物体。现有的方法通常对整个场景进行推理，这导致了模糊的预测和高计算成本，尤其是在杂乱的环境中。我们观察到，许多指称表达依赖于局部空间上下文，并且通常对应于受限的空间区域，而不是整个场景。基于这一洞察，我们提出了PruneGround，这是一种有效的即插即用框架，旨在3DVG中构建于三个关键组件之上。首先，我们引入了语言引导的空间剪枝（Language-Guided Spatial Pruning, LGSP），该方法利用冻结的视觉语言模型（Vision Language Model, VLM）来识别与语言相关的区域，从而减少空间计算并在更狭窄的搜索空间中定位候选物体。其次，我们提出了多视角条件描述重构（MultiView-Conditioned Description Reformulation, MCDR），该方法将复杂表达分解为简化的目标-锚点关系，并通过多视角推理增强缺失的空间线索。最后，我们提出了LLM-Grounder，该方法通过在剪枝区域内对齐点云和语言表示，将检测预训练的空间语言模型（spatial LLM）重新用于语言条件的定位模型。在三个最流行的点云基准上的大量实验表明，我们的方法在所有三种ScanRefer设置和10个Nr3D/Sr3D设置中的9个上都达到了最先进的结果。代码和模型已公开可用： https://github.com/leduckhai/PruneGround
cs.CV / 42 / 2606.31157
Rethinking Foundation Model Collaboration: Enhancing Specialized Models through Proxy Task Reasoning
重新思考基础模型协作：通过代理任务推理增强专业模型
Abstract
Foundation models are increasingly integrated into embodied intelligence systems, but directly assigning them structured prediction tasks requires precise geometric and numerical estimation, where specialized models often remain stronger. This capability mismatch raises a key question: should foundation models replace task-specific predictors, or should they collaborate through tasks better aligned with their strengths? We propose FAT, a foundation-model-augmented task-specific reasoning framework that treats collaboration as task decomposition rather than model replacement. FAT decomposes structured prediction into specialist prediction, information-space reconstruction, and foundation-model proxy reasoning. The specialist generates geometrically and physically valid hypotheses in the native output space, while the foundation model performs a bounded proxy task, such as selection or verification, over reconstructed multimodal candidates. We instantiate this principle as ProxySelect with a vision--language model. Across 2D object detection, 3D object detection, trajectory prediction, and semantic segmentation, ProxySelect consistently improves specialized baselines and substantially outperforms direct foundation-model regression at lower computational cost. These results suggest a general collaboration principle: specialized models preserve task-specific structure, while foundation models refine their hypotheses through contextual proxy reasoning.
Chinese Translation
基础模型越来越多地被整合到具身智能系统中，但直接将它们分配到结构化预测任务上需要精确的几何和数值估计，而专业模型在这方面通常更具优势。这种能力的不匹配引发了一个关键问题：基础模型是否应该取代任务特定预测器，还是应该通过与其优势更为一致的任务进行协作？我们提出了FAT，一个增强基础模型的任务特定推理框架，将协作视为任务分解而非模型替代。FAT将结构化预测分解为专业预测、信息空间重构和基础模型代理推理。专业模型在原生输出空间中生成几何和物理上有效的假设，而基础模型则在重构的多模态候选上执行有界代理任务，如选择或验证。我们将这一原则实例化为ProxySelect，使用视觉-语言模型。在二维物体检测、三维物体检测、轨迹预测和语义分割等任务中，ProxySelect始终改善专业基线，并在较低的计算成本下显著超越直接的基础模型回归。这些结果表明了一种普遍的协作原则：专业模型保持任务特定结构，而基础模型通过上下文代理推理来完善其假设。
cs.CV / 43 / 2606.31160
Reasoning-aware Speculative Decoding for Efficient Vision-Language-Action Models in Autonomous Driving
面向推理的投机解码：提高自主驾驶中视觉-语言-动作模型的效率
Abstract
Modern Vision-Language-Action (VLA) planners for autonomous driving emit a chain-of-causation (CoC) reasoning step \emph{before} producing a trajectory. The reasoning is autoregressive and dominates inference latency, while the trajectory head is parallel and cheap. Latency is an operational constraint in autonomous driving, so accelerating the reasoning step is the central problem we address. We observe that CoC reasoning has two qualitatively different needs: most tokens continue routine setup that follows naturally from the ego-trajectory history, and a small fraction encode commitments that require fresh visual evidence about an unexpected situation. We split this reasoning into two specialized paths: a \emph{routine reasoner} that handles the predictable continuation by attending to trajectory history, and a \emph{deliberative reasoner} (the unmodified VLA target) that handles novel cases by attending to current visual evidence, using the speculative decoding framework as the architectural template for how the two paths cooperate. Unlike standard speculative decoding, our routine reasoner is not a smaller replica of the target; the two reasoners are deliberately specialized to read different parts of the prompt. We propose two techniques to realize this. First, we introduce \textbf{FlatRoPE}, a 1D rotary positional embedding in the draft that breaks the rotational symmetry of the target's 3D M-RoPE, redirecting attention away from visual tokens and onto trajectory-history tokens. Second, we introduce \textbf{Action-aware RL (AARL)}, a post-training stage that uses an action-quality reward together with a static-reference KL anchor. Together, our two-reasoner system reduces the reasoning-step running time by approximately $4\times$ relative to the original Alpamayo planner.
Chinese Translation
现代自主驾驶的视觉-语言-动作（VLA）规划器在生成轨迹之前会进行因果链（CoC）推理步骤。该推理是自回归的，并主导了推理延迟，而轨迹头则是并行且成本低廉的。延迟是自主驾驶中的一个操作约束，因此加速推理步骤是我们要解决的核心问题。我们观察到，CoC推理有两种质的不同需求：大多数标记继续进行自然跟随自我轨迹历史的常规设置，而少部分则编码需要关于意外情况的新视觉证据的承诺。我们将这一推理分为两条专门的路径：一个处理可预测延续的常规推理器，通过关注轨迹历史来实现；另一个处理新情况的深思熟虑推理器（未修改的VLA目标），通过关注当前视觉证据来实现，利用投机解码框架作为两条路径协作的架构模板。与标准的投机解码不同，我们的常规推理器并不是目标的一个较小副本；这两个推理器被故意专门化，以读取提示的不同部分。我们提出了两种技术来实现这一点。首先，我们引入了 extbf{FlatRoPE}，一种在草稿中使用的1D旋转位置嵌入，打破了目标的3D M-RoPE的旋转对称性，将注意力从视觉标记转移到轨迹历史标记上。其次，我们引入了 extbf{Action-aware RL (AARL)}，一种后训练阶段，使用动作质量奖励以及静态参考KL锚点。通过这两种推理器系统，我们将推理步骤的运行时间相对于原始Alpamayo规划器减少了约$4 imes$。
cs.CV / 44 / 2606.31164
Seeing Through the Weights: Privacy Leakage in Scene Coordinate Regression
透视权重：场景坐标回归中的隐私泄露
Abstract
Scene Coordinate Regression (SCR) methods are increasingly adopted for visual localization. In these approaches, the scene is implicitly encoded within a neural network that regresses a 3D world coordinate for each image pixel. Because the scene is represented only through the network parameters and not stored explicitly as images or maps, such methods are often assumed to be privacy-preserving. In this work, we show that this assumption is incorrect in practice. Specifically, we introduce a query-based attack that reconstructs the 3D geometry of the training environment from an SCR model under different levels of model access. To do so, we repeatedly query the model with batches of proxy images unrelated to the target scene to obtain dense pixel-wise 3D coordinates. Reliable points are identified through their stability under small input perturbations and can be further refined in a white-box setting. These stable points are accumulated across independent query batches to recover the scene geometry. From the recovered 3D representation, we also invert the network features to synthesize images from arbitrary viewpoints, revealing additional appearance information. Experiments on indoor and outdoor datasets demonstrate that substantial portions of training environments can be reconstructed with high geometric fidelity. Beyond geometry, we also recover an approximate color appearance, which exposes recognizable layout and potentially sensitive scene elements. This directly contradicts claims in the literature that SCR representations are privacy-preserving by design, and reveals a real risk when such systems are deployed in private or security-critical spaces. The project page is available at https://jaeminch0.github.io/seeing-through-the-weights-privacy-leakage-in-scene-coordinate-regression.
Chinese Translation
场景坐标回归（Scene Coordinate Regression, SCR）方法在视觉定位中越来越受到广泛应用。在这些方法中，场景通过一个神经网络隐式编码，该网络为每个图像像素回归一个三维世界坐标。由于场景仅通过网络参数表示，而不是以图像或地图的形式显式存储，因此这些方法通常被认为是保护隐私的。然而，我们的研究表明，这一假设在实践中是错误的。具体来说，我们引入了一种基于查询的攻击方法，该方法在不同的模型访问级别下，从SCR模型中重建训练环境的三维几何形状。为此，我们反复用与目标场景无关的代理图像批量查询模型，以获取密集的逐像素三维坐标。通过小输入扰动下的稳定性来识别可靠点，并可以在白盒环境中进一步精细化。这些稳定点在独立的查询批次中累积，以恢复场景几何形状。从恢复的三维表示中，我们还反转网络特征，从任意视角合成图像，揭示额外的外观信息。在室内和室外数据集上的实验表明，可以高保真地重建训练环境的相当大部分。除了几何形状，我们还恢复了近似的颜色外观，这暴露了可识别的布局和潜在的敏感场景元素。这直接与文献中关于SCR表示在设计上保护隐私的说法相矛盾，并揭示了在私人或安全关键空间中部署此类系统时的真实风险。项目页面可访问：https://jaeminch0.github.io/seeing-through-the-weights-privacy-leakage-in-scene-coordinate-regression。
cs.CV / 45 / 2606.31169
Beyond Single Character: Evaluating MLLMs for Sentence-Level Oracle Bone Inscription Understanding
超越单个字符：评估多模态大语言模型在句子级甲骨文理解中的表现
Abstract
Existing AI-assisted oracle bone inscription (OBI) visual recognition and understanding studies mainly focus on character-level, ignoring the long-form textual coherence and contextual dependencies embedded in complete divination charges. Recently, the powerful visual perception capabilities of multimodal large language models (MLLMs) have opened new possibilities for OBI information processing. In this work, we introduce S-OBI, a novel benchmark for evaluating MLLMs in Sentence-level OBI understanding. Instead of using noisy and incomplete rubbings as the visual input, S-OBI synthesizes clear and standardized sentence-level OBI instances through glyph substitution and composition. According to 95 original rubbings with translations that have been identified, corrected, and verified by experts, we replace characters in the original rubbings with corresponding clean glyph samples sourced from existing OBI datasets while preserving the overall inscriptional structure and semantic organization. This mitigates the influence of low-level distortions and enables a more focused evaluation of sentence-level OBI understanding. Based on this, we design semantic matching, semantic slot extraction, and contextual reasoning tasks and obtain 695 question-answer pairs. Experiments reveal the inferiority of contemporary MLLMs on sentence-level OBI understanding. In particular, visual perception errors in unmasked regions propagate through the reasoning chain, leading to erroneous predictions for masked characters, which indicates that sentence-level OBI understanding in current models remains strongly dependent on character-level recognition. Overall, S-OBI provides a diagnostic benchmark for evaluating whether MLLMs can move beyond isolated character recognition toward structured inscription-level understanding.
Chinese Translation
现有的人工智能辅助甲骨文（OBI）视觉识别和理解研究主要集中在字符级别，忽视了完整卜辞中嵌入的长文本一致性和上下文依赖性。最近，多模态大语言模型（MLLMs）强大的视觉感知能力为甲骨文信息处理开辟了新的可能性。在本研究中，我们引入了S-OBI，一个用于评估MLLMs在句子级甲骨文理解中的新基准。S-OBI通过字形替换和组合，合成清晰且标准化的句子级甲骨文实例，而不是使用嘈杂和不完整的拓片作为视觉输入。根据95个经过专家识别、纠正和验证的原始拓片及其翻译，我们用来自现有甲骨文数据集的相应干净字形样本替换原始拓片中的字符，同时保留整体的铭文结构和语义组织。这减轻了低级失真的影响，使得对句子级甲骨文理解的评估更加集中。基于此，我们设计了语义匹配、语义槽提取和上下文推理任务，并获得了695个问答对。实验结果显示，当前的MLLMs在句子级甲骨文理解上表现不佳。特别是在未遮蔽区域的视觉感知错误通过推理链传播，导致对遮蔽字符的错误预测，这表明当前模型的句子级甲骨文理解仍然高度依赖于字符级识别。总体而言，S-OBI提供了一个诊断性基准，用于评估MLLMs是否能够超越孤立的字符识别，朝向结构化的铭文级理解迈进。
cs.CV / 46 / 2606.31172
HSDF-Lane: Height-Aligned Signed Distance Field with Semantic Lane Prior for 3D Lane Detection
HSDF-Lane：具有语义车道先验的高度对齐签名距离场用于3D车道检测
Abstract
Monocular 3D lane detection plays a critical role in autonomous driving, yet recovering reliable 3D geometry from a single image remains challenging due to inherent depth ambiguity. Prior methods project image features into Bird's-Eye-View (BEV) space under a flat-ground assumption, causing geometric distortion on real-world roads. Recent methods instead predict explicit height maps to capture non-planar surfaces, but still rely on sparse anchor-based regression and exploit the recovered geometry merely for spatial transformation rather than semantic understanding. To overcome these limitations, we propose HSDF-Lane, which implicitly models the road surface as a Height-aligned Signed Distance Field (HSDF) over a densely sampled 3D feature volume. Through differentiable rendering, the HSDF jointly produces an accurate height map and surface-aligned features. We further introduce Lane-aware Semantic Positional Encoding (LSPE), which injects a lane-existence prior derived from the surface-aligned features into the transformer queries, coupling geometric structure with semantic guidance. Extensive experiments on the OpenLane benchmark show that HSDF-Lane achieves state-of-the-art performance in both 3D lane detection and height map estimation.
Chinese Translation
单目3D车道检测在自动驾驶中发挥着关键作用，但从单幅图像恢复可靠的3D几何形状仍然具有挑战性，因为存在固有的深度模糊性。先前的方法在平坦地面假设下将图像特征投影到鸟瞰图（BEV）空间中，导致现实世界道路上的几何失真。最近的方法则预测显式高度图以捕捉非平面表面，但仍依赖稀疏锚点回归，并仅将恢复的几何用于空间变换，而非语义理解。为克服这些局限性，我们提出了HSDF-Lane，它将道路表面隐式建模为一个在密集采样的3D特征体积上对齐高度的签名距离场（HSDF）。通过可微渲染，HSDF共同生成准确的高度图和表面对齐特征。我们进一步引入了车道感知语义位置编码（LSPE），将从表面对齐特征中推导出的车道存在先验注入到变换器查询中，将几何结构与语义指导结合起来。在OpenLane基准上的广泛实验表明，HSDF-Lane在3D车道检测和高度图估计方面均达到了最先进的性能。
cs.CV / 47 / 2606.31177
GaussianMap: Learning Gaussian Representation for Multi-Sensor Online HD Map Construction
GaussianMap：用于多传感器在线高清地图构建的高斯表示学习
Abstract
Autonomous driving systems benefit from high-definition (HD) maps that provide critical information about road infrastructure. The online construction of HD maps offers a scalable approach to generate local vectorized maps from onboard sensor observations. Existing methods commonly adopt bird's-eye-view (BEV) features as the intermediate scene representation, encoding the surrounding space with fixed-resolution dense grids. However, map elements are spatially sparse yet require fine-grained geometric localization, making uniformly allocated BEV representations redundant and less effective for vectorized map prediction. In this work, we propose GaussianMap, an online HD map construction framework that learns an adaptive Gaussian representation of the surrounding scene. This representation consists of a set of Gaussian primitives on the BEV plane, each encoding a flexible local region with geometric properties and a feature vector, allowing the model to allocate representational capacity to map-relevant regions. To generate such a representation from sensor observations, we introduce a feed-forward Gaussian encoder that progressively refines these primitives through Gaussian interaction modeling and multi-sensor feature aggregation. The refined Gaussian representation is then splatted into a BEV feature map and decoded into vectorized map predictions. Extensive experiments on nuScenes and Argoverse 2 datasets demonstrate that GaussianMap achieves state-of-the-art performance in both camera-only and camera-LiDAR fusion settings. Our code will be made publicly available.
Chinese Translation
自主驾驶系统受益于提供道路基础设施关键信息的高清（HD）地图。在线构建高清地图提供了一种可扩展的方法，从车载传感器观测生成局部矢量化地图。现有方法通常采用鸟瞰图（BEV）特征作为中间场景表示，用固定分辨率的密集网格编码周围空间。然而，地图元素在空间上是稀疏的，但需要精细的几何定位，这使得均匀分配的BEV表示显得冗余，并且在矢量化地图预测中效果不佳。在本研究中，我们提出了GaussianMap，一个在线高清地图构建框架，学习周围场景的自适应高斯表示。该表示由一组位于BEV平面上的高斯原语组成，每个原语编码一个具有几何属性和特征向量的灵活局部区域，使模型能够将表示能力分配给与地图相关的区域。为了从传感器观测生成这种表示，我们引入了一个前馈高斯编码器，通过高斯交互建模和多传感器特征聚合逐步细化这些原语。细化后的高斯表示随后被映射到BEV特征图中，并解码为矢量化地图预测。在nuScenes和Argoverse 2数据集上的大量实验表明，GaussianMap在仅使用相机和相机-LiDAR融合设置下均实现了最先进的性能。我们的代码将公开发布。
cs.CV / 48 / 2606.31187
Learning to Deny: Action Denial in Multimodal Large Language Models
学习拒绝：多模态大型语言模型中的动作拒绝
Abstract
Multimodal large language models (MLLMs) have rapidly advanced video understanding, achieving strong zero-shot and few-shot recognition across standard benchmarks. Yet their ability to deny an action by recognizing when an activity is not happening despite strong contextual cues remains largely unexplored. We introduce UCF101-AD, a large-scale benchmark consisting of paired Action-Presence and Action-Denial clips, designed to evaluate this capacity for denial. Each negative video in UCF101-AD preserves the same contextual and motion cues, including persons, objects, and locations, as its positive counterpart, but the defining action itself is explicitly absent. Evaluating 20 state-of-the-art MLLMs reveals a consistent failure: models that exceed 85% accuracy on the positive action classes collapse below 50% on their action-denial counterparts, indicating a strong inclination to affirm plausible actions rather than verify that they truly occur. This exposes a critical blind spot in modern video understanding: the inability to reason causally about whether a motion actually happens. To probe this issue, we explore a causal graph formulation, CausalAct, which expresses scene structure through natural-language prompts linking context, interaction, and motion. Incorporating such causal cues substantially reduces false positives, demonstrating that denial is a learnable reasoning skill. UCF101-AD provides a new lens for diagnosing and improving causal reasoning in multimodal models. Dataset and relevant code: https://github.com/raiyaan-abdullah/Learn-to-Deny.
Chinese Translation
多模态大型语言模型（MLLMs）在视频理解方面迅速发展，在标准基准测试中实现了强大的零样本和少样本识别。然而，它们在识别何时某个活动并未发生的情况下拒绝一个动作的能力，尽管有强烈的上下文线索，仍然在很大程度上未被探索。我们引入了UCF101-AD，这是一个大型基准数据集，由配对的动作存在和动作拒绝视频片段组成，旨在评估这种拒绝能力。UCF101-AD中的每个负面视频保留了与其正面对应片段相同的上下文和运动线索，包括人物、物体和地点，但定义性的动作本身是明确缺失的。对20个最先进的MLLMs的评估显示出一致的失败：在正面动作类别上超过85%准确率的模型在其动作拒绝对应类别上跌落至50%以下，表明它们更倾向于确认合理的动作，而不是验证这些动作是否真实发生。这暴露了现代视频理解中的一个关键盲点：无法因果推理某个动作是否真正发生。为探讨这一问题，我们探索了一种因果图表述，CausalAct，通过自然语言提示将上下文、互动和运动联系起来，表达场景结构。纳入这些因果线索显著减少了误报，证明拒绝是一种可学习的推理技能。UCF101-AD为诊断和改善多模态模型中的因果推理提供了新的视角。数据集及相关代码： https://github.com/raiyaan-abdullah/Learn-to-Deny。
cs.CV / 49 / 2606.31198
Distilling Temporal Coherence into 2D Networks for Transrectal Ultrasound Prostate Video Segmentation
将时间一致性提炼为二维网络用于经直肠超声前列腺视频分割
Abstract
Real-time video segmentation of the prostate in Transrectal Ultrasound (TRUS) is essential for image-guided interventions. While conventional 2D methods suffer from inter-frame inconsistencies by disregarding temporal context, 3D architectures incur prohibitive latency. To resolve this dilemma, we present a Temporally Consistent Learning Framework that distills temporal coherence into a 2D network during training, preserving single-frame inference efficiency. Our design is driven by a key clinical observation: the prostate exhibits geometric stability, whereas the surrounding acoustic environment fluctuates due to physiological motion and transducer pressure. Because conventional temporal constraints propagate erroneous gradients from these unstable regions, we introduce a Confidence-Weighted Temporal Consistency objective derived from optical flow warping residuals, selectively attenuating contributions from unreliable regions. Complementing this pixel-wise constraint, a Dual-scale Prototype Alignment Module enforces semantic coherence through contrastive optimization of local boundary and global semantic features. Furthermore, to eliminate the need for dense per-frame video annotations, we employ geometric equivariance-based pseudo-labeling with knowledge distillation from a pretrained teacher. Extensive experiments on SUN-SEG and our newly introduced TRUS-V benchmark (2,679 frames) demonstrate state-of-the-art accuracy and temporal consistency at real-time speed. Code and dataset are available at https://github.com/DYDevelop/DTC-TRUS.
Chinese Translation
经直肠超声（TRUS）中前列腺的实时视频分割对于图像引导干预至关重要。虽然传统的二维方法因忽视时间上下文而遭受帧间不一致的问题，但三维架构则会导致不可接受的延迟。为了解决这一困境，我们提出了一种时间一致性学习框架，在训练过程中将时间一致性提炼到二维网络中，从而保持单帧推理的高效性。我们的设计受到一个关键临床观察的驱动：前列腺表现出几何稳定性，而周围的声学环境由于生理运动和探头压力而波动。由于传统的时间约束会将这些不稳定区域的错误梯度传播，因此我们引入了一种基于光流变形残差的置信加权时间一致性目标，选择性地减弱来自不可靠区域的贡献。作为对这一像素级约束的补充，双尺度原型对齐模块通过对局部边界和全局语义特征的对比优化来强制语义一致性。此外，为了消除对每帧视频密集注释的需求，我们采用基于几何等变性的伪标签生成，并通过从预训练教师模型中进行知识蒸馏。我们在SUN-SEG和我们新引入的TRUS-V基准（2679帧）上的广泛实验表明，该方法在实时速度下实现了最先进的准确性和时间一致性。代码和数据集可在https://github.com/DYDevelop/DTC-TRUS获取。
cs.CV / 50 / 2606.31201
ExPLoRe: Expert Patch-Level Loss Routing for Multi-Objective Masked Image Modeling
ExPLoRe：用于多目标掩膜图像建模的专家级补丁损失路由
Abstract
Multi-objective masked image modeling (MIM) combines complementary learning signals (token distillation, CLS alignment, and pixel reconstruction) but existing methods weight these objectives with global scalars, ignoring spatial heterogeneity across patches. We present ExPLoRe (Expert Patch-Level Loss Routing), which repurposes Soft Mixture of Experts (MoE) dispatch weights as learned, per-patch loss coefficients. The key mechanism is loss-coupling: allowing loss gradients to flow through dispatch weights to the router enables content-dependent specialization, where different patches receive different emphases across objectives. A detach ablation confirms loss-coupling as the core mechanism, degrading performance by 1.6% when gradients are blocked. On ImageNet-1K with ViT-Base, ExPLoRe improves over non-MoE baselines on two objective combinations (Token+CLS: +0.5% k-NN, +4.4% linear probe; Token+Pixel: +2.2% k-NN), achieving 80.6% linear probe and 85.3% finetuning accuracy, competitive with published methods. For downstream transfer, we develop adaptation recipes (Freeze Routing, Expert Dropout, and Freeze Attention) that improve MoE finetuning by +1.5% over the vanilla MoE, and close a 2.5--2.9 mIoU segmentation gap so that MoE models match or exceed non-MoE baselines on ADE20K.
Chinese Translation
多目标掩膜图像建模（MIM）结合了互补的学习信号（标记蒸馏、CLS 对齐和像素重建），但现有方法使用全局标量对这些目标进行加权，忽略了补丁间的空间异质性。我们提出了 ExPLoRe（专家级补丁损失路由），该方法将软混合专家（Soft Mixture of Experts, MoE）的调度权重重新利用为学习到的每个补丁损失系数。其关键机制是损失耦合：允许损失梯度通过调度权重流向路由器，从而实现基于内容的专业化，使得不同补丁在不同目标上获得不同的重视程度。一项去耦合消融实验确认了损失耦合作为核心机制，当梯度被阻断时，性能下降了 1.6%。在使用 ViT-Base 的 ImageNet-1K 数据集上，ExPLoRe 在两个目标组合（标记+CLS：+0.5% k-NN，+4.4% 线性探测；标记+像素：+2.2% k-NN）上优于非 MoE 基线，达到了 80.6% 的线性探测和 85.3% 的微调准确率，具有与已发布方法相竞争的表现。对于下游迁移，我们开发了适应性方案（冻结路由、专家丢弃和冻结注意力），使 MoE 微调比基础 MoE 提高了 +1.5%，并缩小了 2.5-2.9 mIoU 的分割差距，使得 MoE 模型在 ADE20K 上的表现与非 MoE 基线相匹配或超过。
cs.CV / 51 / 2606.31204
AC3S: Adaptive Conditioning for 3D-Aware Synthetic Data Generation
AC3S：用于三维感知合成数据生成的自适应调节
Abstract
Synthetic data generation has emerged as a powerful tool for improving data scalability in computer vision. Recent diffusion-based pipelines have demonstrated strong photorealism. However, how to enforce precise 3D structure and pose consistency in generated images remains challenging. Existing methods leverage visual prompts such as edge maps to guide diffusion models, but often suffer from over-conditioning artifacts that degrade image realism and limit dataset quality. In this paper, we present a diffusion-based image generation framework that enforces 3D structural alignment while preserving photorealism through adaptive conditioning. Our framework, Adaptive Conditioning for 3D-Aware Synthetic Data Generation (AC3S), introduces a self-supervised visual prompt modulator that dynamically adjusts the strength of ControlNet conditioning, preventing over-conditioning and enabling the diffusion model to retain its generative expressiveness. To further enhance diversity and semantic consistency, we develop a multi-agent vision language model framework that composes detailed and 3D-aware prompts aligned with the underlying geometric structure. Together, these components enable the scalable generation of high-quality synthetic datasets with accurate 2D and 3D annotations. Extensive experiments demonstrate that our method significantly improves image quality and downstream utility.
Chinese Translation
合成数据生成已成为提升计算机视觉数据可扩展性的强大工具。近期的基于扩散的管道展示了强大的照片真实感。然而，如何在生成的图像中强制执行精确的三维结构和姿态一致性仍然是一个挑战。现有方法利用边缘图等视觉提示来引导扩散模型，但往往受到过度调节伪影的影响，从而降低图像的真实感并限制数据集的质量。在本文中，我们提出了一种基于扩散的图像生成框架，该框架通过自适应调节在保持照片真实感的同时强制执行三维结构对齐。我们的框架，即三维感知合成数据生成的自适应调节（AC3S），引入了一种自监督视觉提示调制器，能够动态调整 ControlNet 调节的强度，防止过度调节，并使扩散模型保持其生成表现力。为了进一步增强多样性和语义一致性，我们开发了一个多智能体视觉语言模型框架，该框架组合了与基础几何结构对齐的详细且具有三维感知的提示。这些组件共同实现了高质量合成数据集的可扩展生成，且具有准确的二维和三维注释。大量实验表明，我们的方法显著提高了图像质量和下游应用的效用。
cs.CV / 52 / 2606.31211
AA: A Multi-view Multimodal Dataset for Screen-based Gaze Estimation
AA：用于基于屏幕的注视估计的多视角多模态数据集
Abstract
We present AA, a multi-view multimodal dataset for screen-based gaze estimation. The dataset captures synchronized facial observations from eight fixed screen-mounted cameras and two additional side-view cameras, paired with precise screen-space gaze targets collected under controlled fixation conditions. Each sample contains multi-view face observations together with structured facial region crops, enabling multimodal learning from both global and local visual cues. Unlike existing single-view gaze datasets, AA provides multi-view coverage from both screen-mounted and side-mounted perspectives, enabling more robust modeling under viewpoint variation and occlusion. The dataset includes subject-independent evaluation splits and a standardized data processing pipeline to support reproducible research in gaze estimation.
Chinese Translation
我们提出了AA，一个用于基于屏幕的注视估计的多视角多模态数据集。该数据集捕捉了来自八个固定屏幕安装摄像头和两个额外侧视摄像头的同步面部观察，并配有在受控注视条件下收集的精确屏幕空间注视目标。每个样本包含多视角的面部观察以及结构化的面部区域裁剪，能够从全局和局部视觉线索中进行多模态学习。与现有的单视角注视数据集不同，AA提供了来自屏幕安装和侧面安装视角的多视角覆盖，从而在视角变化和遮挡情况下实现更稳健的建模。该数据集包括独立于受试者的评估划分和标准化的数据处理流程，以支持注视估计领域的可重复研究。
cs.CV / 53 / 2606.31219
CooperScene: Multi-Modal Cooperative Autonomy Benchmark with C-V2X Communication Characterization
CooperScene：具有C-V2X通信特征的多模态协作自主基准
Abstract
Cellular vehicle-to-everything (C-V2X) enables cooperative perception, prediction, and planning beyond the field of view of individual agents. However, existing datasets often overlook the complexities of real-world deployment, such as limited communication bandwidth and its dynamics, heterogeneous sensing modalities, and scalability beyond a single cooperative partner. In this paper, we introduce CooperScene, a high-fidelity cooperative autonomy dataset with real-world C-V2X communication characterization. The dataset is organized into diverse scenes, including intersections, highway ramps, and parking lots. These scenes involve three connected and autonomous vehicles (CAVs) and one infrastructure roadside unit (RSU), all equipped with multi-modal sensors and commercial off-the-shelf C-V2X communication radios. All scenes are annotated with globally consistent 3D labels at 10 Hz, totaling 344K objects across 59K frames, underpinned by tight sensor- and agent-synchronization, centimeter-level localization and spatial alignment, precise cross-modality calibration, and 3GPP-standard-compliant C-V2X communication. CooperScene establishes a rigorous benchmark for evaluating multi-agent scaling and actual performance in real-world deployable settings. Project website for data and benchmark: https://cisl.ucr.edu/CooperScene
Chinese Translation
蜂窝车联网（C-V2X）使得超越单个代理视野的协作感知、预测和规划成为可能。然而，现有的数据集往往忽视了现实世界部署的复杂性，例如有限的通信带宽及其动态变化、异构传感模式，以及超越单一协作伙伴的可扩展性。本文介绍了CooperScene，一个具有真实世界C-V2X通信特征的高保真协作自主数据集。该数据集组织成多样化的场景，包括交叉口、高速公路匝道和停车场。这些场景涉及三辆连接和自主的车辆（CAVs）和一个基础设施路边单元（RSU），所有设备均配备多模态传感器和商业现成的C-V2X通信收发器。所有场景均以10 Hz的频率标注全球一致的3D标签，共计344K个物体，涵盖59K帧，基于紧密的传感器和代理同步、厘米级定位和空间对齐、精确的跨模态校准，以及符合3GPP标准的C-V2X通信。CooperScene建立了一个严格的基准，用于评估多代理扩展和在现实世界可部署环境中的实际性能。项目网站以获取数据和基准：https://cisl.ucr.edu/CooperScene
cs.CV / 54 / 2606.31226
ForgeDrive: Bidirectional Cross-Conditioning for Unified Visual-Action Generation in Autonomous Driving
ForgeDrive：用于自主驾驶的统一视觉-动作生成的双向交叉条件化
Abstract
World-model-based autonomous driving endows the model with the ability to understand scene evolution. Yet this promise is undermined by the prevailing imagine-then-act paradigm, which allows errors from the more challenging visual generation stage to cascade into action planning. We introduce ForgeDrive, a unified autoregressive diffusion framework with visual-action cross-conditioning that closes this gap through act-then-imagine paradigm. ForgeDrive factorizes the future as a sequence of per-timestep frame-action pairs, intertwining each action with its corresponding visual observation. During training, we decouple the diffusion timesteps of the two modalities and introduce a UniDiffuser-style noise scheduler to get the ability to infer either modality from its counterpart and deepen understanding of relationships between images and actions. At inference, we propose a novel act-then-imagine inference paradigm, and find that at each step, action generation is a capability internalized during training, requiring no clean future frame as a prerequisite at inference time; instead, the generated action can improve the accuracy of future frame generation, which in turn enhances the quality of the next action. Additionally, we augment each step with future ego-status prediction, further sharpening planning ability. Extensive experiments on NAVSIM demonstrate that ForgeDrive not only unifies driving simulation, planning, and visual odometry into a single model, but also outperforms existing strong planners without any post-training strategy.
Chinese Translation
基于世界模型的自主驾驶赋予模型理解场景演变的能力。然而，这一承诺受到主流的“先想象后行动”范式的削弱，该范式允许来自更具挑战性的视觉生成阶段的错误级联到行动规划中。我们提出了ForgeDrive，一种统一的自回归扩散框架，具有视觉-动作交叉条件化，通过“先行动后想象”范式弥补了这一差距。ForgeDrive将未来分解为一系列每个时间步的帧-动作对，将每个动作与其对应的视觉观察交织在一起。在训练过程中，我们解耦了两个模态的扩散时间步，并引入了一种UniDiffuser风格的噪声调度器，以获得从一个模态推断另一个模态的能力，并加深对图像与动作之间关系的理解。在推理时，我们提出了一种新颖的“先行动后想象”推理范式，并发现每一步的动作生成是在训练过程中内化的能力，在推理时不需要干净的未来帧作为前提；相反，生成的动作可以提高未来帧生成的准确性，从而增强下一个动作的质量。此外，我们在每一步中增加了对未来自我状态的预测，进一步提升了规划能力。在NAVSIM上的大量实验表明，ForgeDrive不仅将驾驶仿真、规划和视觉里程计统一为一个模型，而且在没有任何后训练策略的情况下超越了现有的强规划器。
cs.CV / 55 / 2606.31242
UHD-MFF: Shattering Barriers in Multi-Focus Ultra-High-Definition Image Fusion via Learnable Lookup Tables
UHD-MFF：通过可学习查找表打破多焦点超高清图像融合的壁垒
Abstract
With the advancement of imaging technology, ultra-high-definition images have become increasingly essential in modern visual applications. However, existing multi-focus image fusion remains largely confined to low-resolution images and faces three major barriers in UHD scenarios, namely data availability, model adaptability, and deployment feasibility, which severely hinder its practical application. To shatter these barriers, first, we propose the UHD-MFF dataset, the first large-scale ultra-high-resolution multi-focus fusion dataset. Second, we propose a scale-specialized lookup-table framework tailored for ultra-high-resolution images, termed as UMF-LUT. It consists of Coarse-Region Lookup Table (C-LUT) and Detail-Edge Lookup Table (D-LUT). Specifically, C-LUT performs joint queries of multiple gradient cues and semantic cues at low-resolution scales to enable region-level decision-making. Also, D-LUT operates at high-resolution scales, leveraging efficient Laplacian cues to provide complementary edge-level decision information. Such a design makes the model particularly well-suited for ultra-high-resolution multi-focus image fusion. Finally, it offers strong deployability with minimal computational overhead, enabling real-time 4K multi-focus fusion and showing promising potential for smartphone. Extensive experiments demonstrate that it outperforms SOTA methods in both visual fidelity and quantitative metrics. It effectively advances the development of multi-focus image fusion toward ultra-high-resolution imaging scenarios. The code is available at https://github.com/zyb5/UHD-MFF.
Chinese Translation
随着成像技术的进步，超高清图像在现代视觉应用中变得愈加重要。然而，现有的多焦点图像融合仍然主要局限于低分辨率图像，并在超高清场景中面临三个主要障碍，即数据可用性、模型适应性和部署可行性，这严重阻碍了其实际应用。为了解决这些障碍，首先，我们提出了UHD-MFF数据集，这是第一个大规模超高清分辨率多焦点融合数据集。其次，我们提出了一种专门针对超高清分辨率图像的尺度专用查找表框架，称为UMF-LUT。该框架由粗区域查找表（Coarse-Region Lookup Table, C-LUT）和细节边缘查找表（Detail-Edge Lookup Table, D-LUT）组成。具体而言，C-LUT在低分辨率尺度上执行多个梯度线索和语义线索的联合查询，以实现区域级决策。同时，D-LUT在高分辨率尺度上操作，利用高效的拉普拉斯线索提供补充的边缘级决策信息。这种设计使得该模型特别适合于超高清分辨率多焦点图像融合。最后，它以最小的计算开销提供强大的可部署性，使得实时4K多焦点融合成为可能，并在智能手机上展现出良好的潜力。大量实验表明，该方法在视觉保真度和定量指标上均优于现有最先进的方法（SOTA）。它有效推动了多焦点图像融合向超高清成像场景的发展。代码可在 https://github.com/zyb5/UHD-MFF 获取。
cs.CV / 56 / 2606.31245
HyperVLP: Enhancing Hierarchical Surgical Video-Language Pre-training in Hyperbolic Space
HyperVLP：在双曲空间中增强分层外科视频-语言预训练
Abstract
Surgical vision-language foundation models typically adopt educational materials, such as surgical lecture videos, to transfer surgical knowledge encoded in language into visual representations. These knowledge are multi-dimensional and hierarchical: fine-grained action cues appear in narration, mid-level key steps are summarized in subsection headings, and global procedural context, such as patient history and surgical strategy, is described in abstract texts. Prior work largely collapses these heterogeneous signals into a single flat embedding space, implicitly assuming independence across hierarchy levels. However, this is suboptimal because it ignores cross-level semantic containment, e.g., actions belong to steps, steps compose phases, weakens long-range dependency modeling. To this end, we propose a hyperbolic surgical video-language pre-training framework that explicitly preserves the hierarchical structure by mitigating structural false negatives induced by procedural context and enforcing semantic consistency between parent phases and their constituent child steps. Extensive experiments on multiple surgical benchmarks show consistent gains in zero- and few-shot phase recognition across procedures and institutions.
Chinese Translation
外科视觉-语言基础模型通常采用教育材料，例如外科讲座视频，将编码在语言中的外科知识转化为视觉表示。这些知识是多维和分层的：细粒度的动作线索出现在叙述中，中层关键步骤在小节标题中总结，而全球程序背景，如患者历史和外科策略，则在摘要文本中描述。之前的研究在很大程度上将这些异质信号压缩到单一的平面嵌入空间中，隐含地假设了层级之间的独立性。然而，这种方法并不理想，因为它忽视了跨层级的语义包含关系，例如，动作属于步骤，步骤组成阶段，从而削弱了长距离依赖建模。为此，我们提出了一种双曲外科视频-语言预训练框架，通过减轻程序背景引起的结构性假阴性，并加强父阶段与其组成子步骤之间的语义一致性，明确保留了分层结构。在多个外科基准测试上的广泛实验表明，在不同程序和机构中，零样本和少样本阶段识别均取得了一致的提升。
cs.CV / 57 / 2606.31249
Rethinking the Role of Feature Engineering and Learning Strategies in Few-Shot Hidden Emotion Recognition
重新思考特征工程和学习策略在少样本隐性情感识别中的作用
Abstract
In this paper, we present the solution developed by our team, XInsight Lab, which achieved first place in Track 3 of the 4th EI-MIGA-IJCAI Challenge with a test accuracy of 0.76923. To address the challenge of weak and sparse implicit emotion evidence in long videos, this paper extends the winning solution from the previous competition and proposes a compact multi-modal temporal modeling framework. The framework integrates and evaluates the effects of multi-source features, including 2D/3D skeletons, facial expression Blendshapes, DINOv2/v3 vision foundation models, X-CLIP video features, and Gemini semantic priors. Architecturally, we propose a cross-attention mechanism that utilizes static pose features, denoted as Base, as the Query and dynamic micro-motion differential features, denoted as Offset, as the Key and Value. By capturing local relative velocities, this mechanism eliminates static biases related to individual body shape and identity. Concurrently, an adaptive pooling method based on Multiple Instance Learning is employed to extract instantaneous emotions while suppressing background noise in long sequences. Finally, the paper reveals the representation collapse phenomenon of general vision foundation models in micro-dynamic tasks, and analyzes the underlying mechanisms where networks fall into public-leaderboard-driven pseudo-generalization due to shortcut learning and rote memorization.
Chinese Translation
在本文中，我们介绍了我们团队XInsight Lab开发的解决方案，该方案在第四届EI-MIGA-IJCAI挑战赛的第三赛道中获得第一名，测试准确率为0.76923。为了应对长视频中弱且稀疏的隐性情感证据的挑战，本文扩展了上一届比赛的获胜解决方案，并提出了一个紧凑的多模态时间建模框架。该框架整合并评估了多源特征的效果，包括2D/3D骨架、面部表情Blendshapes、DINOv2/v3视觉基础模型、X-CLIP视频特征和Gemini语义先验。在架构上，我们提出了一种跨注意力机制，该机制利用静态姿态特征（称为Base）作为查询（Query），并将动态微运动差异特征（称为Offset）作为键（Key）和值（Value）。通过捕捉局部相对速度，该机制消除了与个体体型和身份相关的静态偏差。同时，采用基于多实例学习的自适应池化方法，以提取瞬时情感，同时抑制长序列中的背景噪声。最后，本文揭示了通用视觉基础模型在微动态任务中的表示崩溃现象，并分析了网络因捷径学习和死记硬背而陷入公共排行榜驱动的伪泛化的潜在机制。
cs.CV / 58 / 2606.31257
Decodable Is Not Grounded: A Vision-Ablation Arbiter for VLM Spatial Reasoning
可解码性并不具备基础：一种用于视觉语言模型空间推理的视觉消融仲裁者
Abstract
The standard way to read latent knowledge out of a model, a linear probe confirmed by a steering recovery, can systematically overstate what a vision-language model (VLM) actually grounds in the image. We show this on spatial reasoning, where the error is invisible to both probing and steering yet exposed by a one-line causal control: replacing the image with a gray blank. Probes decode the within-axis answer at 73--97% across axes, and a training-free projection lifts a near-chance axis from 59% to 79%, exactly the signature of unlocking latent knowledge. The blank-image arbiter refutes it, revealing three grounding regimes that probing conflates: an axis can be grounded (vision-dependent, correct), a prior (vision-independent, with its decode and its apparent recovery a directional default rather than perception), or, surprisingly, inverted: decodable, causally controllable, but deployed with the wrong sign, so the model scores below chance and the error requires looking. The taxonomy holds across the studied VLMs: in fourteen models spanning six language-model families and 2B--27B, horizontal is grounded, vertical is a prior, and depth is inverted, with the inversion emerging at scale within families. The decode-versus-deploy inversion replicates on seven of eight models across five families, and the minimal edit that re-deploys it varies with geometry: a training-free rotation matches a trained edit on the cleanest model, while distributed inversions need a trained low-rank edit, tracing a per-model correction-complexity spectrum. The cheap, self-calibrating arbiter cleanly separates grounded perception, inverted perception, and prior substitution; we argue it should be a default control for latent-knowledge and steering claims in VLMs.
Chinese Translation
从模型中读取潜在知识的标准方法，即通过引导恢复确认的线性探测，可能系统性地夸大视觉语言模型（VLM）在图像中实际具备的基础。我们在空间推理中展示了这一点，其中错误对探测和引导都是不可见的，但通过一行因果控制暴露出来：用灰色空白替换图像。探测在各个轴上的答案解码率为73%至97%，而无训练的投影将接近随机的轴从59%提升至79%，这正是解锁潜在知识的特征。空白图像仲裁者对此提出了反驳，揭示了探测所混淆的三种基础状态：一个轴可以是基础的（依赖于视觉，正确），一个先验（独立于视觉，其解码和明显恢复是方向性的默认而非感知），或者，令人惊讶的是，反转的：可解码的、因果可控的，但以错误的符号部署，因此模型得分低于随机，错误需要被观察。该分类法在所研究的VLM中保持一致：在涵盖六个语言模型家族和20亿至270亿参数的十四个模型中，水平是基础的，垂直是先验，而深度是反转的，反转在家族内的规模上出现。解码与部署的反转在五个家族中的八个模型中复制，重新部署的最小编辑与几何形状相关：无训练的旋转与最干净模型上的训练编辑相匹配，而分布式反转需要训练的低秩编辑，追踪每个模型的修正复杂性谱。这个廉价的、自我校准的仲裁者清晰地区分了基础感知、反转感知和先验替代；我们认为它应该成为VLM中潜在知识和引导声明的默认控制。
cs.CV / 59 / 2606.31258
WarpHammer: Densifying Scene Warps with 3D Object Priors for Extreme View Synthesis
WarpHammer：利用3D物体先验增强场景变形以实现极端视图合成
Abstract
Projection-conditioned novel view synthesis (NVS) warps an explicit 3D reconstruction of the input view into the target camera and conditions a generator on the warped rendering. This works well for small viewpoint changes but degrades sharply under large orbital motion: the warp becomes sparse around the orbited object, where hidden surfaces dominate the new view and mirror-like artifacts emerge, causing the generator to lose both pixel content and the implicit camera cue carried by the warp. We introduce WarpHammer, a training-free framework that resolves this failure mode by augmenting the warped scene with an explicit 3D reconstruction of the object obtained from a native 3D generative prior (e.g., SAM3D). The reconstructed object adds missing foreground surfaces and occludes background points that should no longer be visible, restoring both appearance and camera cues without fine-tuning the base model. The same explicit object representation further unlocks a capability current NVS pipelines do not support: incorporating auxiliary views of the object from sources outside the target scene, for example, a casual snapshot of a car paired with a manufacturer studio shot of the same model. We process the reference and auxiliary images jointly with a pretrained multi-view geometry foundation model, which predicts a unified point cloud that we fuse into the 3D object reconstruction. This yields substantially more faithful geometry than single-image reconstruction, without requiring user-provided camera poses for the auxiliary views. On five benchmarks, WarpHammer produces stable novel views at viewpoint deviations where strong baselines collapse, and is the first scene-level NVS method that can naturally fuse auxiliary, pose-unknown object views from an external source.
Chinese Translation
投影条件下的新视图合成（NVS）将输入视图的显式3D重建变形到目标相机，并对变形渲染进行条件处理。这在小视角变化下效果良好，但在大范围轨道运动下急剧下降：变形在被围绕的物体周围变得稀疏，隐藏表面主导了新视图，并出现镜面伪影，导致生成器失去像素内容和变形所携带的隐式相机线索。我们提出了WarpHammer，这是一个无训练的框架，通过使用来自本地3D生成先验（例如，SAM3D）的物体显式3D重建来增强变形场景，从而解决这一失败模式。重建的物体添加了缺失的前景表面，并遮挡了不再应可见的背景点，恢复了外观和相机线索，而无需对基础模型进行微调。同样的显式物体表示进一步解锁了当前NVS管道不支持的能力：从目标场景外的来源整合物体的辅助视图，例如，将汽车的随意快照与同一型号的制造商工作室照片配对。我们与预训练的多视图几何基础模型共同处理参考图像和辅助图像，该模型预测一个统一的点云，我们将其融合到3D物体重建中。这比单图像重建产生了更真实的几何形状，而无需为辅助视图提供用户提供的相机姿态。在五个基准测试中，WarpHammer在强基线崩溃的视角偏差下产生稳定的新视图，并且是第一个能够自然融合来自外部来源的辅助、未知姿态物体视图的场景级NVS方法。
cs.CV / 60 / 2606.31270
Learning from Failure: Inference-Time Self-Improvement for Computer-Use Agents
从失败中学习：计算机使用代理的推理时自我改进
Abstract
Computer-use agents, which leverage multimodal large language models (MLLMs) to operate computers and complete tasks, have attracted significant attention for their utility and versatility. A major challenge in developing these agents is collecting large-scale, high-quality trajectories. The standard approach generates synthetic data through a self-improving loop: an agent is placed in a verifiable environment and iteratively fine-tuned on its successful trajectories. Despite its effectiveness, this paradigm exploits only successful trajectories and discards the failed ones, even though failures carry rich information about a model's weaknesses. In this work, we explore a complementary failure-driven self-improvement loop, a data-centric paradigm that turns failed trajectories into agent improvements. Specifically, we employ an LLM to diagnose failure modes, propose inference-time solutions, and generate code patches -- lightly verified by humans -- that upgrade the agent. We validate this approach with the state-of-the-art OpenCUA-72B model on the OSWorld benchmark, improving the success rate from 42.3% to 48.9%, a gain of 6.6 percentage points, without any additional training cost and with only modest inference overhead. Our results demonstrate that failure-driven self-improvement is a viable complement to success-based pipelines, enabling more efficient agent improvement.
Chinese Translation
计算机使用代理利用多模态大型语言模型（MLLMs）来操作计算机并完成任务，因其实用性和多功能性而受到广泛关注。开发这些代理的一大挑战是收集大规模、高质量的轨迹。标准方法通过自我改进循环生成合成数据：将代理置于可验证的环境中，并对其成功轨迹进行迭代微调。尽管这种方法有效，但它仅利用成功轨迹而忽略失败轨迹，尽管失败包含了关于模型弱点的丰富信息。在本研究中，我们探索了一种互补的以失败为驱动的自我改进循环，这是一种以数据为中心的范式，将失败轨迹转化为代理改进。具体而言，我们利用大型语言模型（LLM）来诊断失败模式，提出推理时解决方案，并生成代码补丁——经过轻微人工验证——以升级代理。我们在OSWorld基准上使用最先进的OpenCUA-72B模型验证了这种方法，将成功率从42.3%提高到48.9%，提高了6.6个百分点，且没有任何额外的训练成本，只有适度的推理开销。我们的结果表明，以失败为驱动的自我改进是基于成功的流程的可行补充，使代理改进更加高效。
cs.CV / 61 / 2606.31275
CLIMB: Centroid-Based Hierarchical Memory for Online Continual Self-Supervised Learning
CLIMB：基于中心的分层记忆用于在线持续自监督学习
Abstract
Online Continual Self-Supervised Learning (OCSSL) aims to learn representations from a continuous stream of unlabeled data, without knowledge of task boundaries and under memory constraints. Existing methods rely either on replay buffers that exploit latent space structure, or on regularization alone. We present CLIMB (Continual Learning with Intelligent Memory Bank), which combines both simultaneously. Our method introduces a hierarchical centroid-based memory, bounded in total number of stored images, combined with knowledge distillation on replayed examples to limit representation drift. The memory groups similar images into centroids, providing hard-to-discriminate examples for contrastive learning while covering the diversity of observed distributions. Experiments on Split CIFAR-100 and Split ImageNet-100, on standard benchmarks from the state-of-the-art as well as a new protocol with irregular task distributions show that CLIMB outperforms state-of-the-art OCSSL methods.
Chinese Translation
在线持续自监督学习（OCSSL）旨在从连续流动的无标签数据中学习表示，且不需要了解任务边界，并在内存限制下进行。现有方法要么依赖于利用潜在空间结构的重放缓冲区，要么仅依赖于正则化。我们提出了CLIMB（具有智能记忆库的持续学习），同时结合了这两者。我们的方法引入了一种基于中心的分层记忆，限制存储图像的总数量，并结合对重放示例的知识蒸馏以限制表示漂移。该记忆将相似图像分组为中心，提供难以区分的示例以进行对比学习，同时覆盖观察到的分布的多样性。在Split CIFAR-100和Split ImageNet-100上的实验，以及在来自最先进技术的标准基准和具有不规则任务分布的新协议中，显示CLIMB优于最先进的OCSSL方法。
cs.CV / 62 / 2606.31278
Editing Everything Everywhere All at Once
全面编辑一切：无处不在
Abstract
Editing multiple elements of an image in a single forward pass is a practical alternative to multi-turn image manipulation, offering improved efficiency and potentially better harmonization. However, when several instructions target different regions, semantic interference often leads to attribute leakage and poor edit disentanglement, especially as the number of edits increases. In this work, we propose MICE (Multi-Instance Concurrent Editing), a training-free strategy for scalable multi-instance image editing with Multimodal Diffusion Transformers. MICE modifies the additive bias of joint attention to regulate interactions between instance-specific edit instructions, latent, and context tokens identified via user-provided segmentation masks. Specifically, MICE allows intra-instance attention, penalizes interactions between neighboring region tokens, and suppresses unrelated cross-instance attention. As a result, our method enforces attribute binding while preserving global visual consistency. We evaluate MICE on LoMOE-Bench and introduce MICE-Bench, a more challenging benchmark with an average of 8.5 concurrent edits per image. The experiments demonstrate that our approach outperforms strong baselines and recent competitors in terms of visual quality preservation and faithfulness to the editing instructions.
Chinese Translation
在一次前向传递中编辑图像的多个元素是多轮图像处理的一个实用替代方案，提供了更高的效率和潜在的更好协调性。然而，当多个指令针对不同区域时，语义干扰往往导致属性泄漏和编辑解耦不良，尤其是在编辑数量增加时。在本研究中，我们提出了MICE（多实例并发编辑），这是一种无训练的可扩展多实例图像编辑策略，基于多模态扩散变换器（Multimodal Diffusion Transformers）。MICE通过调节实例特定编辑指令、潜在和上下文标记之间的交互，修改联合注意力的加性偏差，这些标记是通过用户提供的分割掩码识别的。具体而言，MICE允许实例内注意力，惩罚相邻区域标记之间的交互，并抑制无关的跨实例注意力。因此，我们的方法在保持全局视觉一致性的同时，强制属性绑定。我们在LoMOE-Bench上评估了MICE，并引入了MICE-Bench，这是一个更具挑战性的基准，每幅图像平均有8.5个并发编辑。实验表明，我们的方法在视觉质量保持和对编辑指令的忠实度方面优于强基线和近期竞争者。
cs.CV / 63 / 2606.31293
Deep Spectral Models for Robust Dental Shape Generation
用于稳健牙齿形状生成的深度谱模型
Abstract
Accurate modeling of dental crown morphology is fundamental for diagnosis, orthodontic planning, and computer-aided restoration design. However, datasets suitable for training such models are typically limited in size. We present ToothForge, a deep spectral generative framework that models dental crown geometries from compact, intrinsic representations. By operating in the spectral domain, ToothForge learns a latent manifold of 3D tooth shapes through synchronized spectral embeddings, ensuring consistent modeling across samples with varying connectivity. Spectral synchronization mitigates the instability of Laplace-Beltrami eigenbases and enables efficient learning in a low-dimensional space. The framework is thoroughly evaluated through robustness analysis, ablation studies, and benchmarking against PCA-based statistical shape models and point-based generative frameworks. Results show that synchronized spectral modeling achieves reconstruction and generative performance comparable to or exceeding spatial approaches, while maintaining compactness and geometric interpretability. Together, the compact synchronized coefficients and low-dimensional learning space make the framework particularly suitable for limited datasets, as often encountered in dental and medical domains, and applicable in real-world scenarios where guaranteeing consistent connectivity across shapes from various clinics is unrealistic.
Chinese Translation
准确建模牙冠形态对于诊断、正畸规划和计算机辅助修复设计至关重要。然而，适合训练此类模型的数据集通常规模有限。我们提出了ToothForge，一个深度谱生成框架，从紧凑的内在表示中建模牙冠几何形状。通过在谱域中操作，ToothForge通过同步谱嵌入学习3D牙齿形状的潜在流形，确保在具有不同连通性的样本之间进行一致建模。谱同步减轻了Laplace-Beltrami特征基的稳定性问题，并在低维空间中实现高效学习。该框架通过稳健性分析、消融研究以及与基于PCA的统计形状模型和基于点的生成框架的基准测试进行了全面评估。结果表明，同步谱建模在重建和生成性能上可与空间方法相媲美或超越，同时保持紧凑性和几何可解释性。紧凑的同步系数和低维学习空间使该框架特别适合于数据集有限的情况，这在牙科和医学领域中常常遇到，并且适用于现实场景中，保证来自不同诊所的形状之间的一致连通性是不切实际的。
cs.CV / 64 / 2606.31318
Wavelet-Optimized Pseudo-3D Accelerated Diffusion Model for Truncated Computed Laminography
小波优化的伪三维加速扩散模型用于截断计算层析成像
Abstract
Computed Laminography (CL) is a key technology for the nondestructive testing of large plate-shaped objects. However, field-of-view (FOV) limitations inevitably lead to truncation of projected data, an ill-posed inverse problem that causes severe reconstruction artifacts. Existing deep learning methods typically rely on 2D architectures that lack rigorous data consistency constraints. Furthermore, they conventionally confine artifact removal strictly to the FOV, discarding potentially recoverable information outside it. To overcome these limitations, we first introduce a comprehensive CL FOV analysis, categorizing the space into data-complete, data-incomplete, and data-free regions. By extending our reconstruction target to encompass the data-incomplete region, we significantly expand the effective imaging range and enhance scanning efficiency. To achieve this, we propose a novel wavelet-optimized pseudo-3D accelerated diffusion model for CL truncation reconstruction (CL-DM). Our method utilizes a standard 2D diffusion model for slice aggregation, combined with a 3D model-based iterative reconstruction (MBIR) method to ensure strict data consistency. To mitigate inter-slice discontinuities, we introduce wavelet regularization along the z-direction, paired with a translation-invariant (TI) mechanism and a low-frequency preservation strategy. Finally, we introduce a 3D fast sampling architecture, significantly accelerating inference speed. Extensive simulations and real-world experiments demonstrate that CL-DM is superior in effectively eliminating truncation artifacts and restoring high-fidelity, continuous 3D structures.
Chinese Translation
计算层析成像（CL）是对大型板状物体进行无损检测的关键技术。然而，视场（FOV）限制不可避免地导致投影数据的截断，这是一种病态逆问题，造成严重的重建伪影。现有的深度学习方法通常依赖于缺乏严格数据一致性约束的二维架构。此外，它们通常将伪影去除严格限制在视场内，丢弃了视场外可能可恢复的信息。为克服这些限制，我们首先引入了全面的CL视场分析，将空间分类为数据完整区、数据不完整区和无数据区。通过将我们的重建目标扩展到数据不完整区，我们显著扩大了有效成像范围并提高了扫描效率。为此，我们提出了一种新颖的小波优化伪三维加速扩散模型用于CL截断重建（CL-DM）。我们的方法利用标准的二维扩散模型进行切片聚合，结合基于模型的三维迭代重建（MBIR）方法，以确保严格的数据一致性。为减少切片间的不连续性，我们在z方向引入小波正则化，配合平移不变（TI）机制和低频保留策略。最后，我们引入了一种三维快速采样架构，显著加快了推理速度。大量模拟和实际实验表明，CL-DM在有效消除截断伪影和恢复高保真、连续的三维结构方面表现优越。
cs.CV / 65 / 2606.31323
Accelerated Likelihood Maximization for Diffusion-based Versatile Content Generation
基于扩散的多功能内容生成的加速似然最大化
Abstract
Generating diverse, coherent, and plausible content from partially given inputs remains a fundamental challenge for diffusion models. Existing approaches face clear limitations: training-based approaches offer strong task-specific results but require costly computation, and they generalize poorly across tasks. Training-free approaches offer better efficiency, but they do not explicitly optimize over unobserved variables, leading to globally inconsistent results. To address these limitations, we introduce Accelerated Likelihood Maximization (ALM), a novel training-free sampling strategy integrated into the reverse diffusion process that significantly extends the applicability of diffusion models beyond simple generation tasks. Unlike previous methods that implicitly influence missing regions through pre-generated region constraints, we directly optimize the unobserved region during the sampling process, enabling globally coherent and plausible generation. Furthermore, we incorporate an acceleration strategy that significantly improves computational efficiency without sacrificing performance. Experimental results demonstrate that ALM consistently outperforms state-of-the-art methods in various data domains and tasks, establishing a powerful paradigm for versatile content generation.
Chinese Translation
从部分给定的输入中生成多样化、一致且合理的内容仍然是扩散模型面临的一个基本挑战。现有方法存在明显的局限性：基于训练的方法提供强大的特定任务结果，但需要昂贵的计算，并且在任务间的泛化能力较差。无训练的方法提供了更好的效率，但它们并未明确优化未观察到的变量，导致全局不一致的结果。为了解决这些局限性，我们引入了加速似然最大化（Accelerated Likelihood Maximization, ALM），这是一种新颖的无训练采样策略，集成于反向扩散过程中，显著扩展了扩散模型在简单生成任务之外的适用性。与之前通过预生成区域约束间接影响缺失区域的方法不同，我们在采样过程中直接优化未观察到的区域，从而实现全局一致且合理的生成。此外，我们还结合了一种加速策略，显著提高了计算效率而不牺牲性能。实验结果表明，ALM在各种数据领域和任务中始终优于最先进的方法，建立了一种强大的多功能内容生成范式。
cs.CV / 66 / 2606.31326
Bridging Video Understanding and Generation in a Unified Framework
在统一框架中架起视频理解与生成的桥梁
Abstract
Recently, unified image generation and understanding have been extensively explored. However, extending such unified modeling paradigms to the video domain remains largely underexplored. A central challenge is that video understanding favors compact, discriminative semantic representations, whereas video generation requires dense signals that preserve visual details and temporal coherence. Videos naturally capture both spatial semantics and temporal dynamics, making them a more suitable modality for unified multimodal modeling compared to static images. In this paper, we propose Vega, a unified framework that bridges video understanding and generation. Vega leverages a shared vocabulary to jointly model text and visual representations and employs a hybrid architecture combining autoregressive (AR) prediction with diffusion-based rendering. Specifically, the AR model focuses on predicting semantically meaningful visual tokens for keyframes, providing a structured representation that guides the diffusion module in rendering dense, high-resolution video frames. Extensive experiments demonstrate that Vega achieves strong performance on video generation benchmarks such as VBench and video understanding benchmarks like VideoMME.
Chinese Translation
近年来，统一的图像生成与理解已得到广泛研究。然而，将这种统一建模范式扩展到视频领域仍然在很大程度上未被探索。一个核心挑战是，视频理解倾向于紧凑、具有区分性的语义表示，而视频生成则需要保留视觉细节和时间一致性的密集信号。视频自然捕捉空间语义和时间动态，使其成为比静态图像更适合统一多模态建模的媒介。在本文中，我们提出了Vega，一个桥接视频理解与生成的统一框架。Vega利用共享词汇共同建模文本和视觉表示，并采用结合自回归（AR）预测与扩散渲染的混合架构。具体而言，AR模型专注于为关键帧预测语义上有意义的视觉标记，提供一种结构化表示，指导扩散模块渲染密集的高分辨率视频帧。大量实验表明，Vega在视频生成基准（如VBench）和视频理解基准（如VideoMME）上均取得了优异的性能。
cs.CV / 67 / 2606.31348
Patient-Level Elbow Abnormality Detection: Leakage-Aware Evaluation of Learned Preprocessing, Calibration, and Triage-Oriented Operating Points
患者级肘部异常检测：基于泄漏意识的学习预处理、校准和分诊导向操作点评估
Abstract
In this study, we examine learned preprocessing pipelines in the context of triage-oriented orthopedic abnormality detection task using elbow radiographs from MURA dataset. The evaluation focuses on patient-level detection of musculoskeletal abnormalities under a leakage-aware protocol. We compare multiple preprocessing pipelines, with and without a lightweight DnCNN module as a learned preprocessing component, to assess their impact on discrimination and calibration. Performance is assessed using discrimination metrics (AUROC, PR-AUC), calibration measures (ECE, Brier score), and validation-selected operating point analysis targeting high specificity. Results show that differences across preprocessing strategies are modest and configuration-dependent, with no consistent discrimination advantage over the raw-input DenseNet121 baseline. The raw and diverse inputs combined with the DnCNN front-end showed reduced ECE and Brier score, while CLAHE combined with DnCNN did not improve calibration. Overall, the results suggest that under patient-level evaluation, preprocessing gains are modest and configuration-dependent; the raw-input DenseNet121 baseline remains competitive throughout, and no tested preprocessing strategy produced a consistent discrimination advantage across all metrics.
Chinese Translation
在本研究中，我们考察了在分诊导向的骨科异常检测任务中，利用MURA数据集的肘部放射影像进行的学习预处理管道。评估重点在于在泄漏意识协议下对肌肉骨骼异常的患者级检测。我们比较了多种预处理管道，包括带有轻量级DnCNN模块作为学习预处理组件的管道与不带的管道，以评估其对区分能力和校准的影响。性能通过区分指标（AUROC, PR-AUC）、校准测量（ECE, Brier分数）和针对高特异性的验证选择操作点分析进行评估。结果表明，不同预处理策略之间的差异较小且依赖于配置，未能在区分能力上持续优于原始输入的DenseNet121基线。结合DnCNN前端的原始和多样化输入显示出降低的ECE和Brier分数，而CLAHE与DnCNN的结合未能改善校准。总体而言，结果表明在患者级评估下，预处理的收益有限且依赖于配置；原始输入的DenseNet121基线在整个过程中保持竞争力，且没有测试的预处理策略在所有指标上产生一致的区分优势。
cs.CV / 68 / 2606.31353
RCL-Mamba: A Dual-domain State Space Model for Measurement-oriented Image Restoration in Rotational Sparse-View Scanning Computed Laminography
RCL-Mamba：一种用于测量导向图像恢复的双域状态空间模型在旋转稀疏视图扫描计算层析中的应用
Abstract
Rotational Scanning Computed Laminography (RCL) is widely utilized for the Non-Destructive Testing(NDT) of large planar components. However, to facilitate rapid inspection, continuous sparse-view scanning is often employed, where the angular integration effect during exposure induces rotational blur in the projection domain. Furthermore, the data incompleteness inherent in sparse sampling manifests as sparse artifacts in the reconstructed image domain. To address these cross-domain degradations, this paper proposes RCL-Mamba, a measurement-oriented dual-domain State Space Model (SSM)-based image restoration network. The framework adopts a cascaded joint processing strategy: it first corrects the rotational blur in the projection domain and subsequently suppresses the sparse artifacts in the image domain. Additionally, we design a Mamba-CNN dual-branch module to adaptively balance large-scale blur correction with local detail recovery. Evaluations on both simulated datasets and real-world Printed Circuit Board (PCB) scans demonstrate that RCL-Mamba outperforms existing baselines in blur removal, artifact suppression, and structural preservation. Line-profile-based structural measurement further verifies that the proposed method better preserves via/pad boundaries and slender trace profiles. Crucially, by reducing the required scanning views from 512 to 64, our method enhances inspection efficiency by approximately 8-fold without compromising reconstruction quality, offering a robust measurement-oriented restoration solution for high-throughput RCL inspection with improved structural measurement fidelity.
Chinese Translation
旋转扫描计算层析（RCL）广泛应用于大型平面组件的无损检测（NDT）。然而，为了便于快速检测，通常采用连续的稀疏视图扫描，在曝光过程中角度积分效应会导致投影域中的旋转模糊。此外，稀疏采样固有的数据不完整性在重建图像域中表现为稀疏伪影。为了解决这些跨域退化问题，本文提出了RCL-Mamba，一种基于测量导向的双域状态空间模型（SSM）的图像恢复网络。该框架采用级联联合处理策略：首先在投影域中校正旋转模糊，然后在图像域中抑制稀疏伪影。此外，我们设计了一个Mamba-CNN双分支模块，以自适应平衡大规模模糊校正与局部细节恢复。在模拟数据集和真实印刷电路板（PCB）扫描的评估中，RCL-Mamba在去模糊、伪影抑制和结构保持方面均优于现有基线。基于线型轮廓的结构测量进一步验证了所提方法在保持通孔/焊盘边界和细长迹线轮廓方面的优势。关键是，通过将所需扫描视图从512减少到64，我们的方法在不妥协重建质量的情况下，将检测效率提高了约8倍，为高通量RCL检测提供了一种稳健的测量导向恢复解决方案，改善了结构测量的保真度。
cs.CV / 69 / 2606.31363
Language-Assisted Super-Resolution from Real-World Low-Resolution Patches
基于语言辅助的真实世界低分辨率图像超分辨率
Abstract
Single image super-resolution aims to reconstruct high-resolution (HR) images from low-resolution (LR) inputs. Training SR models typically requires paired HR-LR data, which is difficult to obtain in reality. As a result, most methods synthesize LR images by artificially degrading HR images with handcrafted kernels or camera ISP adjustments. However, these synthetic degradations fail to capture the complexity of real LR images, leading to poor generalization in practice. To address this, we observe that even within a single high-quality image, regions at different depths exhibit varying resolutions, where distant regions act as LR patches and closer ones as HR patches. This allows the extraction of real, degradation-induced LR patches from real images. Since these LR patches lack paired HR counterparts, we propose LA-SR (Language Assistant for SR), a novel framework for unpaired SR. The key idea of LA-SR is to redefine unpaired SR in the language space, using vision-language models to bridge the LR-HR gap. LA-SR projects images into a semantically rich space representing both content and quality, and applies two language-guided losses: linguistic content loss to preserve semantic fidelity, and linguistic quality loss to enhance perceptual realism. With this alignment, LA-SR effectively super-resolves real LR inputs, producing realistic outputs that overcome the limitations of synthetic-data-trained methods.
Chinese Translation
单幅图像超分辨率旨在从低分辨率（LR）输入重建高分辨率（HR）图像。训练超分辨率（SR）模型通常需要配对的HR-LR数据，而这些数据在现实中难以获得。因此，大多数方法通过使用手工设计的核或相机图像信号处理（ISP）调整来人工降解HR图像，从而合成LR图像。然而，这些合成降解未能捕捉真实LR图像的复杂性，导致在实际应用中泛化能力较差。为了解决这一问题，我们观察到，即使在单一高质量图像中，不同深度的区域也表现出不同的分辨率，远处区域充当LR补丁，而近处区域则作为HR补丁。这使得我们能够从真实图像中提取真实的、因降解而产生的LR补丁。由于这些LR补丁缺乏配对的HR对应物，我们提出了LA-SR（Language Assistant for SR），一个用于无配对超分辨率的新框架。LA-SR的关键思想是在语言空间中重新定义无配对超分辨率，利用视觉-语言模型来弥合LR与HR之间的差距。LA-SR将图像投影到一个语义丰富的空间中，该空间同时表示内容和质量，并应用两种语言引导的损失：语言内容损失以保持语义保真度，语言质量损失以增强感知现实感。通过这种对齐，LA-SR有效地对真实LR输入进行超分辨率处理，生成克服合成数据训练方法局限性的真实输出。
cs.CV / 70 / 2606.31373
Domain Adaptive Object Detection via Dual-Stream Bilevel-Cycle Optimization
通过双流双层循环优化进行领域自适应目标检测
Abstract
Cycle self-training (CST) breaks the shared classifier assumption of the standard self-training framework, which is effective for unsupervised domain adaptation and exploits unlabeled target data by training with target pseudo-labels. CST introduces a target classifier and employs an inner-outer loop updating strategy, addressing the issue of unreliable pseudo-labels and enabling pseudo-labels to generalize across domains. Despite its success in image classification, extending CST to object detection faces three main challenges. First, the upper bound of CST in object detection is constrained by three types of unreliable pseudo-labels, such as classification error alone, localization error alone, and their combination. Second, since object detection involves detecting multiple target objects, directly applying CST leads to training insta bility. Third, a wider numerical range of regression coordinates leads to exploding losses. To this end, we apply CST to both classification and regression and propose the Dual-Stream Bilevel-Cycle Optimization framework. Specifically, we construct CST upon Mean Teacher to prevent training instability and use extra normalization to map the regression bounding box into a standardized space, effectively addressing exploding losses. Also, we provide a theoretical derivation of the regression bound. Extensive experiments across four cross domain standard scenarios demonstrate that our framework achieves considerable results.
Chinese Translation
循环自我训练（CST）打破了标准自我训练框架中共享分类器的假设，这对于无监督领域适应是有效的，并通过使用目标伪标签训练来利用未标记的目标数据。CST引入了目标分类器，并采用内外循环更新策略，解决了伪标签不可靠的问题，使伪标签能够跨领域泛化。尽管在图像分类中取得了成功，但将CST扩展到目标检测面临三个主要挑战。首先，CST在目标检测中的上限受到三种类型不可靠伪标签的限制，例如仅分类错误、仅定位错误及其组合。其次，由于目标检测涉及检测多个目标物体，直接应用CST会导致训练不稳定。第三，回归坐标的数值范围过宽会导致损失爆炸。为此，我们将CST应用于分类和回归，并提出了双流双层循环优化框架。具体而言，我们在均值教师（Mean Teacher）基础上构建CST，以防止训练不稳定，并使用额外的归一化将回归边界框映射到标准化空间，有效解决损失爆炸问题。此外，我们提供了回归界限的理论推导。在四个跨领域标准场景下进行的大量实验表明，我们的框架取得了显著的结果。
cs.CV / 71 / 2606.31378
MAPE: Defending Against Transferable Adversarial Attacks Using Multi-Source Adversarial Perturbations Elimination
MAPE：利用多源对抗扰动消除抵御可转移对抗攻击
Abstract
Neural networks are vulnerable to meticulously crafted adversarial examples, leading to high-confidence misclassifications in image classification tasks. Due to their consistency with regular input patterns and the absence of reliance on the target model and its output information, transferable adversarial attacks exhibit a notably high stealthiness and detection difficulty, making them a significant focus of defense. In this work, we propose a deep learning defense known as multi-source adversarial perturbations elimination (MAPE) to counter diverse transferable attacks. MAPE comprises the single-source adversarial perturbation elimination (SAPE) mechanism and the pre-trained models probabilistic scheduling algorithm (PPSA). SAPE utilizes a thoughtfully designed channel-attention U-Net as the defense model and employs adversarial examples generated by a pre-trained model (e.g., ResNet) for its training, thereby enabling the elimination of known adversarial perturbations. PPSA introduces model difference quantification and negative momentum to strategically schedule multiple pre-trained models, thereby maximizing the differences among adversarial examples during the defense model's training and enhancing its robustness in eliminating adversarial perturbations. MAPE effectively eliminates adversarial perturbations in various adversarial examples, providing a robust defense against attacks from different substitute models. In a black-box attack scenario utilizing ResNet-34 as the target model, our approach achieves average defense rates of over 95.1\% on CIFAR-10 and over 71.5\% on Mini-ImageNet, demonstrating state-of-the-art performance.
Chinese Translation
神经网络对精心设计的对抗样本存在脆弱性，导致在图像分类任务中出现高置信度的错误分类。由于可转移对抗攻击与常规输入模式的一致性以及不依赖于目标模型及其输出信息，这类攻击表现出显著的隐蔽性和检测难度，成为防御的重点。在本研究中，我们提出了一种名为多源对抗扰动消除（MAPE）的深度学习防御方法，以应对多样化的可转移攻击。MAPE包括单源对抗扰动消除（SAPE）机制和预训练模型概率调度算法（PPSA）。SAPE利用精心设计的通道注意力U-Net作为防御模型，并采用由预训练模型（例如ResNet）生成的对抗样本进行训练，从而实现已知对抗扰动的消除。PPSA引入模型差异量化和负动量，战略性地调度多个预训练模型，从而在防御模型的训练过程中最大化对抗样本之间的差异，提高其在消除对抗扰动方面的鲁棒性。MAPE有效消除了各种对抗样本中的对抗扰动，为来自不同替代模型的攻击提供了强有力的防御。在利用ResNet-34作为目标模型的黑箱攻击场景中，我们的方法在CIFAR-10上实现了超过95.1 ext{%}的平均防御率，在Mini-ImageNet上实现了超过71.5 ext{%}的平均防御率，展现了最先进的性能。
cs.CV / 72 / 2606.31383
MS-Resampler: Multi-Scope Visual Resampling for Efficient Multimodal LLMs
MS-Resampler：用于高效多模态大语言模型的多视角视觉重采样
Abstract
Multimodal large language models (MLLMs) typically employ resampling-based projectors to transform dense visual features into a compact token sequence for language modeling. Most existing resamplers adopt a single, fixed aggregation scope via global cross-attention, which can blur fine-grained local evidence and limit the ability to capture both local details and global context within a fixed token budget. In this work, we propose MS-Resampler, a multi-scope visual resampling framework for MLLMs. MS-Resampler instantiates multiple scope-specific resamplers by injecting explicit spatial scope priors into the resampling attention, enabling each branch to aggregate visual information at a particular granularity from local to global. The outputs of these scope-specific resamplers are then adaptively fused to produce the final visual representations for language modeling. Extensive experiments on ten public multimodal benchmarks show that MS-Resampler consistently improves visual understanding and multimodal reasoning over conventional single-scope resamplers, while introducing only minimal computational overhead.
Chinese Translation
多模态大语言模型（MLLMs）通常采用基于重采样的投影器，将密集的视觉特征转换为紧凑的令牌序列以进行语言建模。现有的大多数重采样器通过全局交叉注意力采用单一固定的聚合范围，这可能会模糊细粒度的局部证据，并限制在固定令牌预算内捕捉局部细节和全局上下文的能力。在本研究中，我们提出了MS-Resampler，一种用于MLLMs的多视角视觉重采样框架。MS-Resampler通过将显式的空间范围先验注入重采样注意力，实例化多个特定范围的重采样器，使每个分支能够以特定的粒度从局部到全局聚合视觉信息。这些特定范围的重采样器的输出随后被自适应融合，以生成用于语言建模的最终视觉表示。在十个公共多模态基准上的广泛实验表明，MS-Resampler在视觉理解和多模态推理方面始终优于传统的单范围重采样器，同时仅引入了最小的计算开销。
cs.CV / 73 / 2606.31388
One Video, One World: Turning Monocular Video into Physical 4D Scenes
一段视频，一个世界：将单目视频转化为物理4D场景
Abstract
We introduce \textbf{OVOW}, the first training-free system that reconstructs \emph{instance-level, simulation-ready} 4D mesh scenes from a single monocular video. Recent 4D reconstruction achieves impressive rendering quality, but its outputs (\eg, implicit fields, Gaussian primitives, or point clouds) lack the watertight topology, instance separation, and standardized physical interfaces required by physics simulators and embodied AI. OVOW closes this gap with a four-stage pipeline: a vision-language model discovers, labels, and motion-classifies all instances; category-aware reconstruction yields per-instance meshes for rigid objects and topology-consistent mesh sequences for deformable ones; an iterative render-match-optimize procedure recovers metric scale and 6-DoF pose trajectories; and physics-grounded assembly enforces ground contact and inter-object support. Crucially, we model all motion, rigid and non-rigid, through direct vertex deformation without category-specific priors or skeleton rigging, producing watertight mesh scenes ready for downstream physics simulation and editing. We further establish the first benchmark for \emph{structured Video-to-4D} evaluation, with metrics for geometric correctness, instance separation, and physical plausibility beyond visual fidelity; the same pipeline doubles as a scalable engine for \emph{synthesizing} paired video-to-4D simulation data for future 4D world models and embodied AI. Across two synthetic benchmarks (static and 4D), OVOW attains the best overall layout and geometry accuracy and the lowest photometric and semantic error among all baselines, and on monocular video runs one to two orders of magnitude faster than the baselines, while downstream physics simulation confirms its physical stability.
Chinese Translation
我们介绍了 extbf{OVOW}，这是第一个无需训练的系统，可以从单个单目视频重建 extit{实例级、可模拟}的4D网格场景。近期的4D重建在渲染质量上取得了令人印象深刻的成果，但其输出（例如，隐式场、高斯原语或点云）缺乏物理模拟器和具身人工智能所需的密闭拓扑、实例分离和标准化物理接口。OVOW通过一个四阶段的流程填补了这一空白：视觉-语言模型发现、标记并运动分类所有实例；类别感知重建为刚性物体生成每个实例的网格，并为可变形物体生成拓扑一致的网格序列；迭代的渲染-匹配-优化过程恢复度量尺度和6自由度姿态轨迹；基于物理的组装确保地面接触和物体间支撑。至关重要的是，我们通过直接顶点变形建模所有运动，无论是刚性还是非刚性，而无需类别特定的先验或骨架绑定，从而生成适合下游物理模拟和编辑的密闭网格场景。我们进一步建立了第一个 extit{结构化视频到4D}评估基准，提供几何正确性、实例分离和物理合理性的度量，超越视觉保真度；同一流程还可作为一个可扩展的引擎，用于 extit{合成}成对的视频到4D模拟数据，以支持未来的4D世界模型和具身人工智能。在两个合成基准（静态和4D）上，OVOW在所有基线中实现了最佳的整体布局和几何准确性，并且在光度和语义误差上最低，而在单目视频运行中速度比基线快一个到两个数量级，同时下游物理模拟确认了其物理稳定性。
cs.CV / 74 / 2606.31407
Visual Semantic Entropy: Do Vision Language Models Recognize Visual Ambiguity?
视觉语义熵：视觉语言模型是否识别视觉模糊性？
Abstract
Vision-language models can produce confident answers on visually ambiguous inputs, resulting in biased predictions. Common entropy-based methods, such as Semantic Entropy (SE), rely on output diversity. Yet our analysis shows that overconfident visual embeddings suppress output diversity under stochastic decoding, causing SE to underestimate uncertainty in such cases. Recent methods instead probe output diversity through input perturbations, including textual paraphrasing or joint text-image perturbations, and show improved performance. We study these approaches and reveals that the resulting variability is often dominated by textual changes rather than visual evidence, causing uncertainty estimates to reflect prompt sensitivity rather than visual ambiguity. We therefore propose Visual Semantic Entropy (VSE), which perturbs only the image to probe nearby visual variations while keeping the text query fixed. VSE measures uncertainty by clustering generated answers into semantic prototypes and computing the mass-weighted dispersion among them. Extensive evaluation across five modern vision-language models and five diverse VQA benchmarks demonstrates that VSE effectively captures visual ambiguity, establishing a new state-of-the-art for VLM uncertainty estimation.
Chinese Translation
视觉语言模型在视觉模糊输入上可以产生自信的答案，从而导致偏差预测。常见的基于熵的方法，如语义熵（Semantic Entropy, SE），依赖于输出的多样性。然而我们的分析表明，过于自信的视觉嵌入在随机解码下抑制了输出多样性，导致 SE 在这种情况下低估了不确定性。最近的方法通过输入扰动来探测输出多样性，包括文本释义或文本-图像联合扰动，并显示出改进的性能。我们研究了这些方法，并揭示出结果的变异性往往受到文本变化的主导，而非视觉证据，导致不确定性估计反映了提示的敏感性而非视觉模糊性。因此，我们提出了视觉语义熵（Visual Semantic Entropy, VSE），该方法仅对图像进行扰动，以探测附近的视觉变化，同时保持文本查询不变。VSE 通过将生成的答案聚类为语义原型并计算它们之间的质量加权离散度来衡量不确定性。在五个现代视觉语言模型和五个多样化的视觉问答（VQA）基准上进行的广泛评估表明，VSE 有效捕捉视觉模糊性，为视觉语言模型的不确定性估计建立了新的最先进水平。
cs.CV / 75 / 2606.31421
Temporal Preservation over Processing: Diagnosing and Designing Spatiotemporal Single-Stage Video Detectors
处理中的时间保留：诊断与设计时空单阶段视频检测器
Abstract
Single-stage video object detectors are increasingly deployed in time-critical applications, yet it remains unclear whether these models genuinely reason over temporal context or merely exploit a single informative frame-a gap hidden by standard metrics, which reward correct predictions regardless of how they are reached. We address this from two complementary directions: first, we propose TemporalLens, a model-agnostic diagnostic framework probing temporal dependence through controlled perturbations, structured occlusions, temporal shuffling, redundancy injection, and resolution degradation, revealing whether a detector actually uses information across time. Applied to stacked-frame 2D detectors and our YOLO-3D architecture, it exposes behavioural differences invisible to mAP: stacked 2D models collapse when the target frame is removed, while spatiotemporal models recover predictions from earlier frames, a signature of real temporal reliance. Second, we detail YOLO-3D, a modular real-time spatiotemporal detector built on YOLOv8, and show that simply preserving temporal depth through the backbone is the dominant performance driver (+3.7 pp mAP@50 at 32 frames averaged across scales). Together, the diagnostics and architecture turn "does this detector reason over time?" into a measurable, actionable question.
Chinese Translation
单阶段视频目标检测器在时间关键应用中越来越多地被部署，但尚不清楚这些模型是否真正考虑了时间上下文，还是仅仅利用了单个信息帧——这一差距被标准指标所掩盖，后者奖励正确的预测，而不管是如何达到的。我们从两个互补的方向来解决这个问题：首先，我们提出了TemporalLens，一个与模型无关的诊断框架，通过控制扰动、结构性遮挡、时间洗牌、冗余注入和分辨率降级来探测时间依赖性，揭示检测器是否实际使用了跨时间的信息。应用于堆叠帧2D检测器和我们的YOLO-3D架构时，它揭示了mAP看不见的行为差异：当目标帧被移除时，堆叠的2D模型崩溃，而时空模型则能够从早期帧中恢复预测，这表明了真实的时间依赖性。其次，我们详细介绍了YOLO-3D，这是一个基于YOLOv8构建的模块化实时时空检测器，并展示了通过主干网络简单地保留时间深度是主要的性能驱动因素（在32帧上平均跨尺度提高了3.7个百分点的mAP@50）。结合这些诊断和架构，我们将“这个检测器是否考虑了时间？”转变为一个可测量、可操作的问题。
cs.CV / 76 / 2606.31427
No Prompt, No Leaks: A Robust Generative Steganography Framework via Prompt-Free Diffusion
无提示，无泄漏：通过无提示扩散的稳健生成隐写框架
Abstract
Generative image steganography synthesizes stego images directly from secret information to achieve inherent security advantages. Latent Diffusion Models (LDMs) have recently emerged as a fundamental image steganography framework that modulates secret latent representations with text prompts. Limited by the inflexibility of text prompts, these methods still struggle to generate high-quality stego images and accurately recover secret images. In this work, we propose a prompt-free diffusion image steganography framework that integrates style semantic priors to control more robust and reliable stego image generation. Specifically, a Cascaded Affine Coupling Module (CACM) establishes a bijective, deterministic mapping between a secret image and its latent representation. Then, style semantics are integrated into the diffusion process to control latent representation and ensure visual imperceptibility in the generated stego images. To mitigate trajectory deviations stemming from the unconditioned reverse process, a predictor-corrector mechanism is introduced to iteratively refine the generation trajectory via feedback from the current and predicted next states. Extensive experimental results show that the proposed method achieves competitive performance compared to state-of-the-art methods in terms of security, secret image reconstruction accuracy and controllability.
Chinese Translation
生成图像隐写技术直接从秘密信息合成隐写图像，以实现固有的安全优势。潜在扩散模型（Latent Diffusion Models, LDMs）最近作为一种基础的图像隐写框架出现，通过文本提示调节秘密潜在表示。然而，由于文本提示的灵活性有限，这些方法在生成高质量隐写图像和准确恢复秘密图像方面仍然面临挑战。在本研究中，我们提出了一种无提示扩散图像隐写框架，该框架集成了风格语义先验，以控制更稳健和可靠的隐写图像生成。具体而言，一个级联仿射耦合模块（Cascaded Affine Coupling Module, CACM）建立了秘密图像及其潜在表示之间的双射确定性映射。然后，将风格语义集成到扩散过程中，以控制潜在表示并确保生成的隐写图像在视觉上不可察觉。为减轻源于无条件逆过程的轨迹偏差，引入了一种预测-修正机制，通过当前状态和预测下一状态的反馈迭代优化生成轨迹。大量实验结果表明，所提方法在安全性、秘密图像重建精度和可控性方面与最先进的方法相比表现出竞争力。
cs.CV / 77 / 2606.31444
Temporal Training Strategies for Left Atrium and Left Atrial Appendage Segmentation in Dynamic Contrast 4DCT
动态对比增强4DCT中左心房和左心房附属物分割的时间训练策略
Abstract
Dynamic contrast-enhanced cardiac CT enables time-resolved analysis of contrast filling and washout in the left atrium (LA) and left atrial appendage (LAA), with potential applications for assessing blood stasis in atrial fibrillation (AF). Accurate segmentation across all frames is required for such analysis but is challenging due to large temporal contrast variations and the use of a single annotation per registered sequence. This creates a trade-off between training for robustness and limiting label noise. In this study, we investigate how temporal training-set design affects nnUNet-based segmentation of the LA and LAA in dynamic 4DCT. We compare training using a minimal two-frame dataset reflecting standard clinical practice, a physiologically selected subset of frames, and the full 27-frame sequence. We further evaluate the impact of foreground-based normalization. Training with all frames yielded the best performance in early low-contrast phases. However, the physiologically selected subset achieved comparable performance from the filling phase onward. Applying normalization parameters derived from the full dataset improved performance of reduced datasets in low-contrast frames, but did not fully close the gap. These findings highlight the importance of temporal diversity in training data for robust segmentation in dynamic CT, while indicating that carefully selected frame subsets may provide an effective trade-off between performance and efficiency for downstream applications.
Chinese Translation
动态对比增强心脏CT能够对左心房（LA）和左心房附属物（LAA）中的对比剂充填和洗脱进行时间分辨分析，具有评估心房颤动（AF）中血液淤滞的潜在应用。为了进行此类分析，需要在所有帧中进行准确的分割，但由于时间对比变化大以及每个注册序列仅使用单一标注，这一过程具有挑战性。这在训练稳健性与限制标签噪声之间形成了权衡。在本研究中，我们探讨了时间训练集设计如何影响基于nnUNet的动态4DCT中LA和LAA的分割。我们比较了使用反映标准临床实践的最小两帧数据集、经过生理选择的帧子集和完整的27帧序列进行训练的效果。我们进一步评估了基于前景的归一化的影响。使用所有帧进行训练在早期低对比度阶段表现最佳。然而，生理选择的子集在充填阶段及以后达到了可比的性能。应用从完整数据集中得出的归一化参数改善了低对比度帧中减少数据集的性能，但未能完全缩小差距。这些发现强调了训练数据中时间多样性对于动态CT中稳健分割的重要性，同时表明经过仔细选择的帧子集可能在性能和效率之间提供有效的权衡，以便于后续应用。
cs.CV / 78 / 2606.31454
Towards a foundational model for recognising diastematic Gregorian notation
构建识别间距格里高利记谱法的基础模型
Abstract
Optical recognition of Gregorian notation has recently been attempted with end-to-end methods, with four datasets introduced. However, each of these datasets is in a different encoding. We design a common encoding based on the S-GABC proposal, convert all four datasets to this common encoding, and train a shared end-to-end foundational model for diastematic Gregorian notation that establishes a new state of the art across all four datasets.
Chinese Translation
最近，格里高利记谱法的光学识别尝试采用了端到端的方法，并引入了四个数据集。然而，这些数据集的编码各不相同。我们设计了一种基于 S-GABC 提案的通用编码，将所有四个数据集转换为这种通用编码，并训练了一个共享的端到端基础模型，用于间距格里高利记谱法，在所有四个数据集上建立了新的最先进水平。
cs.CV / 79 / 2606.31467
AeroVerse-SatAgent: UAV-Satellite Collaborative Spatial Reasoning Inspired by the Dual Visual Pathway Theory of Cognitive Neuroscience
AeroVerse-SatAgent：基于认知神经科学双视觉通路理论的无人机-卫星协同空间推理
Abstract
With the rapid advancement of aerospace embodied intelligence, enabling Unmanned Aerial Vehicles (UAVs) to autonomously understand and reason about complex environments has become increasingly important. However, existing UAV-based spatial reasoning approaches face critical limitations: single-view perception renders them vulnerable to occlusions and perspective distortions, while most VLMs lack explicit geometric modeling, relying on semantic cues and yielding inconsistent reasoning under viewpoint and scale variations. To address these challenges, we propose SatAgent, a UAV-Satellite collaborative spatial reasoning model inspired by the dual-pathway mechanism of the human visual system. By jointly leveraging satellite and UAV perspectives, SatAgent enables robust, accurate reasoning in complex urban environments. We first introduce a Geometric-Aware 3D Reconstruction Encoder that elevates 2D UAV features into explicit 3D spatial representations. Next, we design a multi-view topology-semantic alignment module integrating cross-view features within a unified BEV coordinate system. We further introduce a multi-view consistency loss encouraging viewpoint-invariant representations. Finally, we construct SatAgent-SR130K, the first large-scale UAV-Satellite collaborative multi-view spatial reasoning dataset. Experiments show SatAgent outperforms state-of-the-art general-purpose foundation models and specialized spatial reasoning models by 25.91\% and 11.69\%, respectively, across diverse tasks, achieving particularly high accuracy in complex geometric relationship reasoning.
Chinese Translation
随着航空航天具身智能的快速发展，使无人机（UAV）能够自主理解和推理复杂环境变得愈加重要。然而，现有的基于无人机的空间推理方法面临着关键的局限性：单视角感知使其易受遮挡和视角扭曲的影响，而大多数视觉语言模型（VLM）缺乏明确的几何建模，依赖语义线索，在视角和尺度变化下产生不一致的推理。为了解决这些挑战，我们提出了SatAgent，一个受到人类视觉系统双通路机制启发的无人机-卫星协同空间推理模型。通过联合利用卫星和无人机的视角，SatAgent能够在复杂城市环境中实现稳健、准确的推理。我们首先介绍了一种几何感知的3D重建编码器，将2D无人机特征提升为明确的3D空间表示。接下来，我们设计了一个多视角拓扑-语义对齐模块，将跨视角特征整合到统一的鸟瞰视图（BEV）坐标系统中。我们进一步引入了一个多视角一致性损失，鼓励视角不变的表示。最后，我们构建了SatAgent-SR130K，这是第一个大规模无人机-卫星协同多视角空间推理数据集。实验表明，SatAgent在多种任务中分别比最先进的通用基础模型和专门的空间推理模型提高了25.91%和11.69%的性能，在复杂几何关系推理中取得了特别高的准确率。
cs.CV / 80 / 2606.31471
Think While You Map: Asynchronous Vision-Language Agents for Incremental 3D Scene Graphs
边思考边绘图：用于增量3D场景图的异步视觉-语言智能体
Abstract
Open-vocabulary 3D scene graph methods typically operate in two stages: first reconstruct, then enrich with vision-language models, leaving the graph unqueryable during exploration. We argue that this sequential coupling is unnecessary and propose an asynchronous architecture in which lightweight online mapping runs concurrently with heavyweight semantic refinement. A probabilistic voxel-based backbone maintains stable object identities incrementally, while background VLM agents progressively enrich the graph. This framework resolves duplicate object tracks through semantic loop closure, attaches fine-grained visual attributes and derives spatial relations between objects. A multi-target frame scheduler amortizes VLM cost by selecting a small set of informative frames that jointly cover multiple targets. The resulting scene graph is queryable during exploration and grows in semantic richness over time. Our method matches or outperforms existing open-vocabulary 3D scene graph methods on semantic segmentation (ScanNet, Replica) and surpasses the prior state-of-the-art across three visual grounding benchmarks (Sr3D+, Nr3D, ScanRefer) by 15.3 to 18.8
[email protected]. Project page: https://denizbickici.github.io/thinkgraphs/
Chinese Translation
开放词汇的3D场景图方法通常分为两个阶段：首先进行重建，然后使用视觉-语言模型进行丰富，这在探索过程中使得图无法查询。我们认为这种顺序耦合是不必要的，并提出了一种异步架构，其中轻量级在线映射与重量级语义精炼并行运行。基于概率体素的骨干网络逐步保持稳定的对象身份，而背景视觉-语言模型（VLM）智能体逐渐丰富图的内容。该框架通过语义回环闭合解决重复对象轨迹，附加细粒度的视觉属性并推导对象之间的空间关系。多目标帧调度器通过选择一小组信息丰富的帧来分摊VLM的成本，这些帧共同覆盖多个目标。生成的场景图在探索过程中是可查询的，并随着时间的推移在语义丰富性上不断增长。我们的方法在语义分割（ScanNet, Replica）上与现有的开放词汇3D场景图方法相匹配或超越，并在三个视觉定位基准（Sr3D+, Nr3D, ScanRefer）上超越了之前的最先进水平，提升幅度为15.3到18.8
[email protected]。项目页面：https://denizbickici.github.io/thinkgraphs/
cs.CV / 81 / 2606.31488
DrivingDepth: Sparse-Prompted Pixel-wise Scale Correction for Driving Depth Estimation
DrivingDepth：稀疏提示的逐像素尺度校正用于驾驶深度估计
Abstract
Dense depth estimation for autonomous driving faces a geometry-scale conflict: depth foundation models deliver pixel-aligned dense visual geometry without reliable metric scale, while projected LiDAR provides metric anchors that are sparse, noisy, and misaligned with image structures. Existing sparse-prompted methods incorporate LiDAR by regenerating depth from scratch, overriding the foundation model's coherent geometry and producing structural artifacts on visually continuous surfaces. Our key insight is that foundation models already capture geometrically coherent relative depth; no additional surface structure learning is required-only a per-pixel scale factor mapping relative geometry to metric coordinates. Based on this, we propose DrivingDepth, which treats sparse LiDAR as geometric prompts that locally calibrate a frozen foundation prior through residual pixel-wise scale correction, preserving dense visual geometry by construction. On nuScenes with 4-frame surround-view input, DrivingDepth achieves an AbsRel of 11.19 and an EdgeCR of 5.741, outperforming MapAnything (11.99/1.914) by simultaneously delivering SOTA metric accuracy and geometric consistency.
Chinese Translation
自动驾驶的密集深度估计面临几何尺度冲突：深度基础模型提供像素对齐的密集视觉几何，但缺乏可靠的度量尺度，而投影的激光雷达（LiDAR）提供的度量锚点则稀疏、噪声大，并且与图像结构不对齐。现有的稀疏提示方法通过从头再生深度来整合激光雷达，覆盖基础模型的一致几何，导致在视觉连续表面上产生结构伪影。我们的关键见解是，基础模型已经捕捉到了几何一致的相对深度；不需要额外的表面结构学习，只需一个逐像素尺度因子将相对几何映射到度量坐标。基于此，我们提出了DrivingDepth，它将稀疏的激光雷达视为几何提示，通过残差逐像素尺度校正局部校准一个冻结的基础先验，从而在构建上保留密集视觉几何。在nuScenes数据集上，使用4帧环视输入，DrivingDepth达到了11.19的绝对相对误差（AbsRel）和5.741的边缘一致性率（EdgeCR），在同时提供最先进的度量准确性和几何一致性方面超越了MapAnything（11.99/1.914）。
cs.CV / 82 / 2606.31496
HVPNet: A Bio-Inspired Network for General Salient and Camouflaged Object Detection
HVPNet：一种生物启发的通用显著与伪装物体检测网络
Abstract
In recent years, most research on multimodal salient object detection (SOD) and camouflaged object detection (COD) typically aims to improve performance through complex cross-modal feature fusion and decoding structures. However, this approach leads to an excessively large model parameter scale and often fails to deliver satisfactory detection performance due to structural redundancy. In contrast, the human visual process is able to efficiently perform salient and camouflaged object identification without such complex structures. This contrast raises an important question: Can we draw conceptual inspiration from the human visual process to achieve a simpler modeling strategy, and still realize accurate and efficient object detection? To answer this question, we propose HVPNet, a simple yet general bio-inspired computational architecture. Drawing on the multi-layered information integration of the retina as a conceptual metaphor, we designed a Retinal Integration Module (RIM), which effectively integrates multimodal features through a level-specific multi-stage integration strategy. To fully exploit these features, we further design a cortical decoder (CD) that breaks down the decoding process into low- and high-level visual stages, abstracting the hierarchical processing in the human visual cortex. Benefiting from these designs, HVPNet can readily extend to seven tasks across four modalities. Without bells and whistles, it establishes an excellent accuracy-efficiency trade-off across 22 datasets spanning these seven tasks. Our code is available at https://github.com/jiaweiXu1029/HVPNet.
Chinese Translation
近年来，大多数关于多模态显著物体检测（SOD）和伪装物体检测（COD）的研究通常旨在通过复杂的跨模态特征融合和解码结构来提高性能。然而，这种方法导致模型参数规模过大，并且由于结构冗余，往往无法提供令人满意的检测性能。相比之下，人类视觉过程能够在没有如此复杂结构的情况下高效地进行显著和伪装物体的识别。这种对比引发了一个重要问题：我们能否从人类视觉过程中获得概念启发，以实现更简单的建模策略，同时仍能实现准确和高效的物体检测？为了解答这个问题，我们提出了HVPNet，这是一种简单但通用的生物启发计算架构。我们以视网膜的多层信息整合为概念隐喻，设计了视网膜整合模块（Retinal Integration Module, RIM），该模块通过特定层级的多阶段整合策略有效整合多模态特征。为了充分利用这些特征，我们进一步设计了皮层解码器（Cortical Decoder, CD），将解码过程分解为低层次和高层次的视觉阶段，抽象出人类视觉皮层中的分层处理。得益于这些设计，HVPNet可以轻松扩展到跨越四种模态的七个任务。在没有多余装饰的情况下，它在涵盖这七个任务的22个数据集上建立了出色的准确性与效率的平衡。我们的代码可在 https://github.com/jiaweiXu1029/HVPNet 获取。
cs.CV / 83 / 2606.31502
Fully Automated High-Precision Segmentation of Retinal Atrophy and Ellipsoid Zone Thickness in OCT: A Reliable Tool for Real-World GA Monitoring
全自动高精度视网膜萎缩和椭圆区厚度的OCT分割：一种可靠的真实世界GA监测工具
Abstract
Geographic atrophy (GA) secondary to age-related macular degeneration (AMD) requires precise monitoring of relevant structural biomarkers to assess disease stage, progression, and treatment response. This paper presents a fully automated, deep learning-based framework for the high-precision, pixel-wise segmentation of key biomarkers in optical coherence tomography (OCT) imaging: retinal pigment epithelium (RPE) loss, ellipsoid zone (EZ) loss, and EZ thinning. The proposed pipeline uses three specialized semantic segmentation models to delineate RPE loss, EZ boundaries (including interruptions), and Bruch's membrane. To ensure robustness and generalizability, the models were developed on a diverse dataset of 298 SD-OCT volumes representing the full phenotypic spectrum of AMD (GA:222, intermediate AMD: 40, neovascular AMD: 17, healthy: 19) and validated on an independent external dataset (n=43). The comprehensive evaluation was further strengthened using additional datasets to assess repeatability, inter-reader reliability, the impact of B-scan density on measurement accuracy, and subgroup performance stratified by lesion size. Results demonstrated high segmentation accuracy (Dice RPE loss: 0.88, Dice EZ loss: 0.87, Pearson's r > 0.99). Total EZ thickness measurements exhibited a sub-pixel average deviation of 2.15 $\mu m$, and segmentation reliability was confirmed by a strong reproducibility score (ICC > 0.98). By accurately and consistently quantifying outer photoreceptor degeneration and RPE loss, this fully automated framework provides a highly reliable tool for GA assessment in both clinical trials and routine real-world ophthalmic care.
Chinese Translation
与年龄相关的黄斑变性（AMD）引起的地理性萎缩（GA）需要对相关结构生物标志物进行精确监测，以评估疾病阶段、进展和治疗反应。本文提出了一种基于深度学习的全自动框架，用于在光学相干断层扫描（OCT）成像中对关键生物标志物进行高精度的像素级分割：视网膜色素上皮（RPE）损失、椭圆区（EZ）损失和EZ变薄。所提议的流程使用三种专门的语义分割模型来勾勒RPE损失、EZ边界（包括中断）和布鲁赫膜。为了确保模型的稳健性和普适性，模型是在一个包含298个SD-OCT体积的多样化数据集上开发的，代表了AMD的完整表型谱（GA:222，中间AMD:40，新生血管AMD:17，健康:19），并在一个独立的外部数据集上进行了验证（n=43）。通过使用额外的数据集进行全面评估，进一步增强了对重复性、读者间可靠性、B扫描密度对测量精度影响以及按病变大小分层的子组表现的评估。结果显示高分割准确性（Dice RPE损失：0.88，Dice EZ损失：0.87，Pearson's r > 0.99）。总EZ厚度测量的亚像素平均偏差为2.15 μm，分割可靠性通过强大的重现性评分（ICC > 0.98）得到了确认。通过准确且一致地量化外部光感受器退化和RPE损失，这一全自动框架为GA评估提供了一种在临床试验和日常真实世界眼科护理中高度可靠的工具。
cs.CV / 84 / 2606.31504
SimpleSearch-VL: A Simple Recipe for Multimodal Agentic Deep Search
SimpleSearch-VL：多模态自主深度搜索的简单方案
Abstract
We present SimpleSearch-VL, an efficient, reliable, and practical framework for multimodal agentic search. Its core idea is to improve the agent's own search-and-verification process rather than scaling data, tools, or auxiliary model components. For efficiency, Factorized Adaptive Rollout (FAR) improves sampling efficiency by forming more informative training groups while using redundant samples to mitigate long-tail latency and expose hard samples. For reliability, SimpleSearch-VL performs evidence-verified reasoning, explicitly using chain-of-thought verification to assess the relevance of retrieved visual and textual cues to the original context. For practicality, SimpleSearch-VL keeps a lightweight tool interface and performs webpage self-summary within the agent, requiring no additional external dependencies. With only 5K supervised tool-interleaved trajectories and 2K RL data, SimpleSearch-VL improves Qwen3-VL agentic baselines by 15.8 and 16.0 average points for the 8B and 30B-A3B variants, respectively. The SimpleSearch-VL-30B-A3B model further achieves performance competitive with agentic Gemini-3-Pro.
Chinese Translation
我们提出了SimpleSearch-VL，这是一个高效、可靠且实用的多模态自主搜索框架。其核心思想是改善智能体自身的搜索与验证过程，而不是扩展数据、工具或辅助模型组件。为了提高效率，因子化自适应回放（Factorized Adaptive Rollout, FAR）通过形成更具信息量的训练组来提高采样效率，同时使用冗余样本来缓解长尾延迟并暴露困难样本。为了确保可靠性，SimpleSearch-VL执行证据验证推理，明确使用思维链验证来评估检索到的视觉和文本线索与原始上下文的相关性。为了实用性，SimpleSearch-VL保持轻量级工具接口，并在智能体内部执行网页自我总结，无需额外的外部依赖。仅使用5000条监督的工具交错轨迹和2000条强化学习数据，SimpleSearch-VL分别将8B和30B-A3B变体的Qwen3-VL自主基线提高了15.8和16.0个平均点。SimpleSearch-VL-30B-A3B模型进一步实现了与自主Gemini-3-Pro竞争的性能。
cs.CV / 85 / 2606.31513
PRISM: Latent Composition Consistency for Single-Image Reflection Removal
PRISM：单幅图像反射去除的潜在组合一致性
Abstract
Single-image reflection removal (SIRR) seeks to recover the transmission layer from a mixture corrupted by reflections -- a severely ill-posed problem. Existing methods operate in pixel space, where the nonlinear sRGB formation model entangles the two layers and limits generalization. We observe that pretrained VAE latent spaces exhibit substantially lower coherence between image layers compared to pixel space, providing a more favorable working space for decomposition. Building on this finding, we propose \textbf{PRISM} (Pretrained-latent Reflection Image Separation Model), which reinterprets SIRR as a latent linear separation problem. Under an approximate additive formulation in latent space, PRISM learns a flow matching velocity field on a pretrained FLUX backbone that recovers both transmission and reflection in a single forward pass. To enforce robust disentanglement, we introduce a Latent Composition Consistency (LCC) strategy that constructs synthetic mixtures by swapping reflection latents across samples and enforces consistent decomposition via a cycle loss. We further propose a Layer Contrastive Separation (LCS) loss that promotes semantic separation between layers through patch-level contrastive learning, without requiring explicit reflection targets. Experiments on six benchmarks demonstrate that PRISM consistently outperforms state-of-the-art methods by significant margins, with strong generalization to in-the-wild images.
Chinese Translation
单幅图像反射去除（SIRR）旨在从受到反射干扰的混合图像中恢复透射层，这是一个严重的不适定问题。现有方法在像素空间中操作，其中非线性的sRGB形成模型将两个层纠缠在一起，限制了其泛化能力。我们观察到，预训练的变分自编码器（VAE）潜在空间中，图像层之间的相干性显著低于像素空间，这为分解提供了更有利的工作空间。基于这一发现，我们提出了 extbf{PRISM}（预训练潜在反射图像分离模型），将SIRR重新解释为潜在线性分离问题。在潜在空间的近似加性形式下，PRISM在预训练的FLUX骨干网络上学习一个流匹配速度场，在一次前向传递中恢复透射和反射。为了强制实施稳健的解耦，我们引入了一种潜在组合一致性（LCC）策略，通过在样本之间交换反射潜在值构建合成混合，并通过循环损失强制一致的分解。我们进一步提出了一种层对比分离（LCS）损失，通过补丁级对比学习促进层之间的语义分离，而无需显式的反射目标。在六个基准测试上的实验表明，PRISM在显著的幅度上始终优于最先进的方法，并在野外图像上表现出强大的泛化能力。
cs.CV / 86 / 2606.31533
MV-GEL: Language-Driven Multi-View Geometric Entity Localization on Meshes
MV-GEL：基于语言驱动的多视角几何实体定位于网格上
Abstract
Identifying and grounding precise geometric entities, such as edges, planar regions, and curved surfaces within 3D objects, is foundational to computer-aided design (CAD), robotic manipulation, and scientific simulation. Although modern Vision Language Models (VLMs) have advanced referring segmentation (RIS) in the image domain, extending such language-driven localization to structured 3D geometry is substantially harder. The 3D object appearance is highly sensitive to viewpoints; a single perspective may render a target entity clearly observable, while another may suffer from severe occlusion or foreshortening. In this work, we attempt to solve these challenges with MV-GEL (Multi-View Geometric Entity Localization), a framework for localizing fine-grained geometric entities on polygon meshes from natural language queries. Our key insight is that reliable CAD entity (i.e., faces, edges or solids) localization depends on selecting views that make the queried entity maximally interpretable. We introduce GELviews, a prompt-conditioned ranking module that prioritizes viewpoints based on language prompted observability of geometric CAD entities. Selected views are processed by a VLM-based reasoning segmentation backbone, and predicted masks are lifted to the corresponding meshes via geometry-aware ray casting. Our framework is completely CAD agnostic and relies only on 3D meshes. Experiments show up to a 1.7X improvement in face-level IoU and over 4.5X gains in edge-level F1 compared to vanilla baselines, substantially outperforming CLIP-based and random view sampling, particularly for thin and view-sensitive structures.The dataset, code and trained checkpoints are available at https://github.com/kbali1297/MV-GEL.
Chinese Translation
识别和定位精确的几何实体，如边缘、平面区域和曲面，在3D物体中是计算机辅助设计（CAD）、机器人操作和科学模拟的基础。尽管现代视觉语言模型（VLMs）在图像领域的引用分割（RIS）方面取得了进展，但将这种语言驱动的定位扩展到结构化的3D几何体则要困难得多。3D物体的外观对视角高度敏感；单一视角可能使目标实体清晰可见，而另一个视角则可能遭遇严重的遮挡或缩短。在本研究中，我们尝试通过MV-GEL（多视角几何实体定位）来解决这些挑战，这是一个从自然语言查询中定位多边形网格上细粒度几何实体的框架。我们的关键见解是，可靠的CAD实体（即面、边或固体）定位依赖于选择能够最大程度上使查询实体可解释的视角。我们引入了GELviews，一个基于提示的排名模块，根据语言提示的几何CAD实体的可观察性优先选择视角。所选视角通过基于VLM的推理分割骨干进行处理，预测的掩膜通过几何感知的光线投射提升到相应的网格上。我们的框架完全与CAD无关，仅依赖于3D网格。实验表明，与基础模型相比，面级IoU提高了最多1.7倍，边级F1提升超过4.5倍，显著超越了基于CLIP的和随机视角采样，尤其是在薄型和视角敏感的结构上。数据集、代码和训练检查点可在https://github.com/kbali1297/MV-GEL获取。
cs.CV / 87 / 2606.31537
DataEvolver: Self-Evolving Multi-Agent Data Construction for Text-Rich Image Generation
DataEvolver：自我演化的多智能体文本丰富图像生成数据构建
Abstract
Text-rich image generation is one of the most challenging settings in image generation, since models must simultaneously produce visually realistic images and render legible, semantically aligned, and layout-consistent text. Existing data pipelines usually follow a static crawl-filter-freeze paradigm. They collect candidate samples, filter them once, and freeze the accepted data for training. However, rejected samples are usually discarded, although they often contain useful failure signals such as OCR errors and semantic mismatches. As a result, later construction rounds may repeat the same failure modes. To address these limitations, we propose DataEvolver, a self-evolving multi-agent framework for text-rich image data construction. DataEvolver treats data construction as feedback-driven construction policy evolution. A Retriever collects candidate samples, a Verifier assigns quality scores and rejection causes, a Critic summarizes round-level feedback into semantic feedback, and a Generator completes under-covered regions through targeted synthesis. The updated feedback memory then guides the next construction round. Experiments on text-rich image generation benchmarks show that DataEvolver produces more useful training data than fixed-dataset baselines under matched data budgets. At the 0.75M scale on PixArt-alpha, DataEvolver improves OCR-F1 over the strongest baseline by 85.3 percent on TextScenesHQ and 35.3 percent on LongTextBench. The improvements are consistent across both evaluated benchmarks and also transfer to Show-o2, indicating that the benefit of DataEvolver is not tied to a single downstream generator. These results suggest that rejected samples can provide actionable feedback for improving text-rich image data construction.
Chinese Translation
文本丰富图像生成是图像生成中最具挑战性的设置之一，因为模型必须同时生成视觉上逼真的图像，并渲染清晰、语义一致且布局一致的文本。现有的数据管道通常遵循静态的爬取-过滤-冻结范式。它们收集候选样本，进行一次过滤，并将接受的数据冻结用于训练。然而，被拒绝的样本通常会被丢弃，尽管它们往往包含有用的失败信号，例如OCR错误和语义不匹配。因此，后续的构建轮次可能会重复相同的失败模式。为了解决这些局限性，我们提出了DataEvolver，一个自我演化的多智能体文本丰富图像数据构建框架。DataEvolver将数据构建视为基于反馈驱动的构建策略演化。检索器（Retriever）收集候选样本，验证器（Verifier）分配质量评分和拒绝原因，评论员（Critic）将轮次级反馈总结为语义反馈，生成器（Generator）通过针对性合成完成未覆盖的区域。更新后的反馈记忆随后指导下一轮构建。在文本丰富图像生成基准测试中的实验表明，DataEvolver在匹配的数据预算下生成的训练数据比固定数据集基线更有用。在PixArt-alpha的0.75M规模下，DataEvolver在TextScenesHQ上将OCR-F1提高了85.3%，在LongTextBench上提高了35.3%，超越了最强基线。这些改进在评估的两个基准上是一致的，并且也转移到Show-o2，表明DataEvolver的优势并不局限于单一的下游生成器。这些结果表明，被拒绝的样本可以为改善文本丰富图像数据构建提供可行的反馈。
cs.CV / 88 / 2606.31556
AugSplat: Radiance Field-Informed Gaussian Splatting for Sparse-View Settings
AugSplat：基于辐射场的高斯点云渲染在稀疏视图设置中的应用
Abstract
Generating high-quality novel views at real-time frame rates remains a central challenge in 3D vision, particularly in sparse-view scenarios. Neural radiance fields have demonstrated robust reconstruction from limited observations, but their reliance on volumetric rendering leads to high computational cost and slow inference. In contrast, Gaussian Splatting methods achieve real-time rendering through rasterization, but their optimization is highly sensitive to the quality of the initial geometry. This sensitivity becomes especially problematic in sparse-view settings, where limited observations often lead to incomplete or noisy point-cloud reconstructions. In this work, we present AugSplat, a simple framework for improving Gaussian Splatting in sparse-view regimes using radiance-field-based view augmentation. We first train a radiance field on the sparse input views and use it to synthesize additional images from nearby novel viewpoints, increasing the effective view-space coverage available for supervision. These synthetic views are then used as auxiliary supervision during Gaussian Splatting optimization. We study two variants: Staged AugSplat, which uses synthetic views for an initial optimization phase before switching to real images, and Dual AugSplat, which jointly trains on real and synthetic views with a decaying synthetic loss weight. Experiments on sparse-view mip-NeRF 360 scenes show that AugSplat improves reconstruction quality over standard Gaussian Splatting. Staged AugSplat achieves the strongest average performance, while Dual AugSplat provides a closely performing formulation that keeps real-image supervision active throughout training, and both variants preserve real-time rendering at inference.
Chinese Translation
在实时帧率下生成高质量的新视图仍然是3D视觉中的一个核心挑战，尤其是在稀疏视图场景中。神经辐射场已证明能够从有限的观测中进行稳健重建，但其对体积渲染的依赖导致了高计算成本和缓慢的推理速度。相比之下，高斯点云渲染方法通过光栅化实现实时渲染，但其优化对初始几何体的质量高度敏感。这种敏感性在稀疏视图设置中尤其成问题，因为有限的观测往往导致不完整或噪声较大的点云重建。在本研究中，我们提出了AugSplat，这是一个简单的框架，通过基于辐射场的视图增强来改善稀疏视图条件下的高斯点云渲染。我们首先在稀疏输入视图上训练一个辐射场，并利用它从附近的新视点合成额外的图像，从而增加可用于监督的有效视图空间覆盖。这些合成视图随后在高斯点云渲染优化过程中作为辅助监督。我们研究了两种变体：Staged AugSplat，它在切换到真实图像之前使用合成视图进行初始优化阶段；以及Dual AugSplat，它在真实和合成视图上联合训练，并逐渐降低合成损失权重。在稀疏视图 mip-NeRF 360 场景上的实验表明，AugSplat 在重建质量上优于标准高斯点云渲染。Staged AugSplat 实现了最强的平均性能，而 Dual AugSplat 提供了一种紧密的表现形式，在整个训练过程中保持真实图像监督活跃，并且两种变体在推理时均保持实时渲染。
cs.CV / 89 / 2606.31570
Mitigating Positional Leakage in 3D Masked Autoencoders for Robust Representation Learning
减轻3D掩码自编码器中的位置泄漏以实现稳健的表征学习
Abstract
Masked autoencoding has emerged as a prominent paradigm for self-supervised learning on 3D point clouds, achieving competitive performance across downstream tasks. Unlike its 2D counterpart, 3D masked autoencoding directly reconstructs spatial coordinates, making it inherently susceptible to positional leakage. In this work, we identify that the decoder in existing 3D MAE frameworks tends to over-rely on positional information, which weakens semantic representation learning and leads to suboptimal feature quality. To address this issue, we propose MPL-MAE, a masked point learning framework that mitigates positional over-reliance while enhancing the utilization of encoder features. Specifically, we introduce a recalibrated positional embedding module that suppresses metric-dominant coordinate signals while preserving geometric topology, together with a gated positional interface module that dynamically regulates positional injection during reconstruction. These designs promote a more balanced interaction between spatial priors and semantic features, yielding robust and informative representations. Extensive experiments across downstream tasks demonstrate that MPL-MAE consistently achieves competitive performance, validating its effectiveness. Code is available at https://github.com/yanx57/MPL-MAE.
Chinese Translation
掩码自编码已成为在3D点云上进行自监督学习的一个重要范式，在下游任务中取得了竞争性的表现。与其2D对应物不同，3D掩码自编码直接重建空间坐标，这使其本质上容易受到位置泄漏的影响。在本研究中，我们发现现有3D MAE框架中的解码器往往过于依赖位置信息，这削弱了语义表征学习，并导致特征质量不佳。为了解决这个问题，我们提出了MPL-MAE，一种掩码点学习框架，旨在减轻对位置的过度依赖，同时增强编码器特征的利用。具体而言，我们引入了一个重新校准的位置嵌入模块，该模块在保留几何拓扑的同时抑制度量主导的坐标信号，以及一个门控位置接口模块，该模块在重建过程中动态调节位置注入。这些设计促进了空间先验与语义特征之间更平衡的互动，从而产生稳健且信息丰富的表征。在下游任务中的大量实验表明，MPL-MAE始终实现了竞争性的性能，验证了其有效性。代码可在 https://github.com/yanx57/MPL-MAE 获取。
cs.CV / 90 / 2606.31574
Temperature Field Reconstruction of Tungsten Monoblock Divertor on EAST using Physics-aware Neural Operator Transformer
基于物理感知神经算子变换器的EAST钨单块偏转器温度场重建
Abstract
Accurate modeling of the divertor temperature field is essential for preventing material melting and damage and for extending the service life of fusion devices. However, conventional numerical methods, such as the Finite Element Method (FEM), are computationally expensive and therefore unsuitable for real-time applications. Therefore, a fast and generalizable method is required for real-time reconstruction of the divertor temperature field and subsequent real-time control. To address the above issue, we propose a Physics-aware Neural Operator Transformer (PNOT) to characterize the spatiotemporal evolution of the divertor temperature field. It models boundary heat-flux relations as a structured graph and employs graph attention to explicitly capture spatial physical dependencies. Inspired by physics-aware attention, we further develop a physics-aware neural operator module to aggregate query points with similar physical conditions via slicing and model heat diffusion, while a gradient-constrained Sobolev regularization loss enforces consistency between function values and their derivatives. Experimental results show that these physical constraints improve prediction accuracy while preserving physical consistency. The source code of this paper will be released on https://github.com/Event-AHU/OpenFusion
Chinese Translation
准确建模偏转器温度场对于防止材料熔化和损坏以及延长核聚变设备的使用寿命至关重要。然而，传统的数值方法，如有限元法（FEM），计算成本高昂，因此不适合实时应用。因此，需要一种快速且可推广的方法来实时重建偏转器温度场并进行后续的实时控制。为了解决上述问题，我们提出了一种物理感知神经算子变换器（PNOT），用于表征偏转器温度场的时空演变。该方法将边界热流关系建模为结构化图，并采用图注意力机制显式捕捉空间物理依赖关系。受物理感知注意力的启发，我们进一步开发了一种物理感知神经算子模块，通过切片聚合具有相似物理条件的查询点，并建模热扩散，同时梯度约束的Sobolev正则化损失强制函数值及其导数之间的一致性。实验结果表明，这些物理约束在提高预测精度的同时保持了物理一致性。本文的源代码将发布在https://github.com/Event-AHU/OpenFusion
cs.CV / 91 / 2606.31577
Localized Conformal Prediction for Image Classification with Vision-Language Models
基于视觉-语言模型的图像分类局部一致性预测
Abstract
Conformal predictions have attracted significant attention in the field of uncertainty quantification, mainly because of their strong marginal coverage guarantees. Full conditional guarantee is not an attainable goal, a well known fact in conformal predictions literature. As a result, several approaches have tried to approximate this behavior by adapting the conformal sets of test-time samples according to their similarity to calibration examples. Although the latter has gained traction and shown impressive performances for regression problems, its application to image classification remains under-explored. We conduct an extensive benchmarking on natural image classification tasks with vision-language models (VLMs), using our open source implementation of a recent localized conformal prediction algorithm. We show that straightforward usage of the cosine similarity between test-time and calibration visual features, an intuitive choice for VLMs, is not sufficient to improve over the non-local baselines. In response, we propose a simple non-linear transformation of the cosine similarities, which conserves marginal coverage guarantees and achieves statistically significant mean set sizes reduction. Code is available at https://github.com/cfuchs2023/lcp-vlm/.
Chinese Translation
一致性预测在不确定性量化领域引起了广泛关注，主要是因为其强大的边际覆盖保证。完全条件保证并不是一个可实现的目标，这是一致性预测文献中一个众所周知的事实。因此，几种方法尝试通过根据测试样本与校准示例的相似性来调整一致性集，以近似这种行为。尽管后者在回归问题上获得了关注并展现了令人印象深刻的表现，但其在图像分类中的应用仍然未得到充分探索。我们在自然图像分类任务上进行了广泛的基准测试，使用我们开源实现的最新局部一致性预测算法，结合视觉-语言模型（VLMs）。我们展示了在测试时间和校准视觉特征之间直接使用余弦相似性这一直观选择并不足以改善非局部基线。对此，我们提出了一种简单的非线性变换余弦相似性的方法，该方法保持了边际覆盖保证，并实现了统计上显著的均值集合大小减少。代码可在 https://github.com/cfuchs2023/lcp-vlm/ 获取。
cs.CV / 92 / 2606.31585
DPPE: Rethinking Camera-Based Positional Encoding for Scaling Multi-View Transformers
DPPE：重新思考基于相机的位置编码以扩展多视图变换器
Abstract
The remarkable scalability of Transformers has expanded their application to 3D computer vision, where camera-aware positional encoding is crucial for providing spatial cues in multi-view geometry. Recent advancements have established the practice of using camera parameters -- such as extrinsics or projection matrices -- as relative positional encoding into the query, key, and value vectors of the attention mechanism. However, when scaling up the training recipe of novel view synthesis (NVS) models with the camera-based positional encoding, we observe a significant issue: model performance stagnates in the late stages of training. In this paper, we investigate the cause of the performance bottleneck when scaling up and demonstrate that storing rotation and translation given by the positional encoding in the same dimensions of the value vector causes indeterminacy in their independent identification, hindering training scalability. To address this, we propose Decoupled Pose Positional Encoding (DPPE), a novel camera-based positional encoding that explicitly decouples rotation and translation. Extensive evaluations on NVS tasks demonstrate that DPPE enables stable long-term training even in scaled-up training setup. Furthermore, it exhibits superior generalization performance in extrapolation settings, such as handling an increased number of viewpoints and zoom-in scenarios.
Chinese Translation
变换器的卓越可扩展性使其在三维计算机视觉中的应用得以扩展，其中基于相机的位置编码对于在多视图几何中提供空间线索至关重要。最近的进展确立了使用相机参数（如外参或投影矩阵）作为相对位置编码融入注意力机制的查询、键和值向量的做法。然而，在使用基于相机的位置编码扩展新视图合成（NVS）模型的训练方案时，我们观察到一个显著的问题：模型性能在训练的后期阶段停滞不前。本文探讨了在扩展时性能瓶颈的原因，并证明在值向量的相同维度中存储位置编码给出的旋转和平移会导致它们独立识别的不可确定性，从而阻碍训练的可扩展性。为了解决这个问题，我们提出了解耦姿态位置编码（Decoupled Pose Positional Encoding，DPPE），这是一种新颖的基于相机的位置编码，明确解耦了旋转和平移。在NVS任务上的广泛评估表明，DPPE即使在扩展的训练设置中也能实现稳定的长期训练。此外，它在外推设置中表现出优越的泛化性能，例如处理增加的视点数量和放大场景。
cs.CV / 93 / 2606.31599
Token-Sparse Medical Multimodal Reasoning via Dual-Stream Reinforcement Learning
基于双流强化学习的稀疏医疗多模态推理
Abstract
Vision-language models (VLMs) combining reinforcement learning (RL) ignite remarkable progress in multimodal reasoning, yet still struggle with medical images, which typically exhibit extremely sparse visual evidence to inform clinical decision-making. We recognize that pruning visual tokens outside the grounding region greatly enhances medical reasoning. However, a united RL framework for active visual token pruning (VTP) and medical multimodal reasoning remains unestablished. Here, we propose a dual-stream RL framework, ViToS, to fulfill token pruning and question answering. ViToS trains one policy model with two task branches, where one focuses on grounding while the other conducts token-sparse reasoning after VTP. Furthermore, we solve the coupled policy learning problem by introducing the cross-feedback sequential optimization, avoiding gradient conflict and facilitating convergence of the shared policy model. Evaluated on seven medical benchmarks, our method reduces visual tokens to 77% of the original sequence length while achieving a 108.27% relative performance on Lingshu-7B and 104.16% relative performance on HuatuoGPT-Vision-7B. Overall, ViToS delivers superior performance and inference speedup, establishing an efficient paradigm for medical multimodal reasoning.
Chinese Translation
结合强化学习（RL）的视觉语言模型（VLMs）在多模态推理方面取得了显著进展，但在医学图像处理上仍面临挑战，因为医学图像通常缺乏足够的视觉证据来支持临床决策。我们认识到，在基础区域之外修剪视觉标记可以显著增强医学推理。然而，尚未建立一个统一的强化学习框架来实现主动视觉标记修剪（VTP）和医学多模态推理。在此，我们提出了一种双流强化学习框架ViToS，以实现标记修剪和问题回答。ViToS训练一个具有两个任务分支的策略模型，其中一个分支专注于基础，而另一个在VTP后进行稀疏标记推理。此外，我们通过引入交叉反馈序列优化解决了耦合策略学习问题，避免了梯度冲突并促进了共享策略模型的收敛。在七个医学基准测试上的评估结果表明，我们的方法将视觉标记减少到原始序列长度的77%，同时在Lingshu-7B上实现了108.27%的相对性能，在HuatuoGPT-Vision-7B上实现了104.16%的相对性能。总体而言，ViToS在性能和推理速度上均表现优越，为医学多模态推理建立了一个高效的范式。
cs.CV / 94 / 2606.31603
Preserve the Hard, Regenerate the Rest: Uncertainty-Guided Synthetic Training Data Augmentation with Diffusion Models
保留难点，重生其余：基于不确定性的合成训练数据增强与扩散模型
Abstract
Semantic segmentation models struggle with data sparsity and rare or visually diverse regions, e.g., dense regions or small objects in aerial or autonomous mobility data. While synthetic augmentation is an appealing solution, directly generating new labeled data risks misalignment of labels and generated pixels. Existing solutions to this problem often rely on external models, or employ coarse heuristics such as indiscriminately augmenting all foreground objects or entire backgrounds, which wastes capacity on uninformative pixels. To address this, we propose an uncertainty-guided synthetic context augmentation strategy that strictly preserves label validity and efficiently maximizes pixel informativeness per synthetic sample - no external guardrails required. Using a baseline segmenter's predictive entropy, we identify uncertain semantic regions and inpaint only the complementary visual context. When fine-tuning the segmenter on this synthetic data, we compute the loss only over the original pixels, excluding inpainted regions. This focuses learning on the unmodified, uncertain regions while presenting them in novel contexts. We demonstrate substantial mIoU gains on Cityscapes, UAVID, and BDD100K with the largest gains on rare and difficult classes such as buses, trains, or (from the aerial perspective) cars. Our results demonstrate that uncertainty-guided context augmentation is a highly effective lever to improve segmentation performance on complex datasets, with code provided at https://github.com/XITASO/Preserve-the-Hard-Regenerate-the-Rest.
Chinese Translation
语义分割模型在数据稀疏和稀有或视觉多样性区域（例如，航空或自主移动数据中的密集区域或小物体）方面面临挑战。虽然合成增强是一种吸引人的解决方案，但直接生成新的标记数据可能导致标签与生成像素的不对齐。现有的解决方案通常依赖于外部模型，或采用粗略的启发式方法，例如无差别地增强所有前景物体或整个背景，这会浪费在无信息像素上的容量。为了解决这个问题，我们提出了一种基于不确定性的合成上下文增强策略，该策略严格保留标签有效性，并有效最大化每个合成样本的像素信息量——无需外部保护措施。通过使用基线分割器的预测熵，我们识别不确定的语义区域，仅对补充的视觉上下文进行修复。在对该合成数据进行分割器微调时，我们仅计算原始像素的损失，排除修复区域。这将学习重点放在未修改的不确定区域，同时以新颖的上下文呈现它们。我们在 Cityscapes、UAVID 和 BDD100K 上展示了显著的 mIoU 增益，尤其是在稀有和困难类别（如公交车、火车或（从航空视角）汽车）上获得了最大的增益。我们的结果表明，基于不确定性的上下文增强是提高复杂数据集上分割性能的一个非常有效的杠杆，代码可在 https://github.com/XITASO/Preserve-the-Hard-Regenerate-the-Rest 获取。
cs.CV / 95 / 2606.31609
Learning Structurally Consistent Representations for Multi-View Radar Semantic Segmentation
学习结构一致的多视角雷达语义分割表示
Abstract
Radar sensors provide reliable perception under adverse weather and lighting conditions, but their sparse, noisy, and weakly semantic measurements make dense semantic segmentation challenging. Most existing radar segmentation methods rely on grid-based encodings and pairwise interactions, which struggle to capture the higher-order relational structure formed by multiple radar returns from the same physical object. We introduce a unified higher-order structural alignment framework for multi-view radar segmentation. The proposed method refines radar feature representations using learnable hypergraphs to capture higher-order dependencies among spatially related responses. To ensure consistency across heterogeneous radar projections, we further align view-specific features using Unbalanced Optimal Transport (UOT), enabling correspondence-free alignment under varying measurement densities and partial observations. An adaptive attention mechanism then fuses complementary radar views while emphasising structurally informative responses under sparsity and noise. The resulting architecture learns structurally consistent representations across Range Angle (RA), Range Doppler (RD), and Angle Doppler (AD) views and is trained using supervised segmentation together with cross-view consistency regularisation. Experiments on the CARRADA and RADIal benchmarks demonstrate consistent improvements over strong radar-specific baselines, achieving 63.8% mIoU on CARRADA and 83.4% mIoU on RADIal, improving the previous best methods by +1.7 and +2.3 mIoU, respectively. These results highlight the importance of higher-order relational modelling for robust radar perception.
Chinese Translation
雷达传感器在恶劣天气和光照条件下提供可靠的感知，但其稀疏、噪声大且语义弱的测量使得密集语义分割变得具有挑战性。现有的大多数雷达分割方法依赖于基于网格的编码和成对交互，难以捕捉由来自同一物理对象的多个雷达返回形成的高阶关系结构。我们提出了一种统一的高阶结构对齐框架用于多视角雷达分割。所提出的方法使用可学习的超图来细化雷达特征表示，以捕捉空间相关响应之间的高阶依赖关系。为了确保异构雷达投影之间的一致性，我们进一步使用不平衡最优传输（Unbalanced Optimal Transport, UOT）对视图特定特征进行对齐，从而在不同测量密度和部分观测下实现无对应对齐。然后，自适应注意机制融合互补的雷达视图，同时在稀疏性和噪声下强调结构信息响应。最终的架构在范围角（Range Angle, RA）、范围多普勒（Range Doppler, RD）和角度多普勒（Angle Doppler, AD）视图之间学习结构一致的表示，并通过监督分割与跨视图一致性正则化进行训练。在CARRADA和RADIal基准上的实验表明，相较于强大的雷达特定基线方法，取得了一致的改进，在CARRADA上达到63.8%的mIoU，在RADIal上达到83.4%的mIoU，分别比之前的最佳方法提高了+1.7和+2.3 mIoU。这些结果突显了高阶关系建模对于鲁棒雷达感知的重要性。
cs.CV / 96 / 2606.31612
What Memory Do GUI Agents Really Need? From Passive Records to Active Task-Driving States
GUI代理真正需要什么记忆？从被动记录到主动任务驱动状态
Abstract
Mobile GUI agents increasingly face long-horizon tasks that require reading, updating, and reusing task-relevant data across pages and applications. Existing memory methods treat memory largely as passive storage, where past observations are accumulated and retrieved when needed. Yet retrieving a value does not reveal its current role in the workflow. The agent must still infer from accumulated records whether the value should be used now, has already been used, or must wait for a later dependency. This implicit reconstruction becomes unreliable in long trajectories with similar fields, repeated values, distractors, and outdated states, causing repeated or missed operations. We propose Active Task Driving Memory (ATMem), which shifts GUI-agent memory from passive storage to an actively maintained execution state. ATMem maintains task-relevant information as a continually updated execution state that links each value to its role and current status, enabling action selection based on the current workflow state. We therefore introduce \textbf{STR-GRPO}, an online reinforcement learning method that learns to use ATMem selectively according to its contribution to task completion. STR-GRPO contrasts memory-on and memory-off rollouts to estimate when memory use improves execution, while memory-cost-aware reward discourages costly memory usage that does not improve execution. To evaluate whether agents can complete all in-scope work while avoiding out-of-scope actions over long-horizon execution, we build a challenging mobile benchmark. From a list of near identical entries, agents must act on every entry that satisfies the instruction and reject entries that violate its constraints.
Chinese Translation
移动GUI代理越来越多地面临需要跨页面和应用程序读取、更新和重用与任务相关数据的长期任务。现有的记忆方法主要将记忆视为被动存储，在需要时积累和检索过去的观察。然而，检索一个值并不能揭示其在工作流程中的当前角色。代理仍然必须从积累的记录中推断该值现在是否应该使用、是否已经使用过，或者是否必须等待后续依赖。这种隐式重建在具有相似字段、重复值、干扰项和过时状态的长轨迹中变得不可靠，导致操作的重复或遗漏。我们提出了主动任务驱动记忆（Active Task Driving Memory, ATMem），将GUI代理的记忆从被动存储转变为主动维护的执行状态。ATMem将与任务相关的信息维护为一个持续更新的执行状态，将每个值与其角色和当前状态关联，从而使得基于当前工作流程状态的行动选择成为可能。因此，我们引入了 extbf{STR-GRPO}，一种在线强化学习方法，学习根据ATMem对任务完成的贡献进行选择性使用。STR-GRPO对比了开启记忆和关闭记忆的执行轨迹，以估计何时使用记忆能够改善执行，同时考虑记忆成本的奖励机制则抑制了那些未能改善执行的高成本记忆使用。为了评估代理是否能够在避免超出范围的操作的情况下完成所有范围内的工作，我们构建了一个具有挑战性的移动基准测试。在一系列几乎相同的条目中，代理必须对每个满足指令的条目采取行动，并拒绝违反其约束的条目。
cs.CV / 97 / 2606.31613
Robust Autonomous UAV Landing on Maritime Platforms via Multimodal Agentic AI and Active Wave Compensation
通过多模态智能体人工智能和主动波浪补偿实现稳健的自主无人机在海洋平台上的着陆
Abstract
Autonomous aerial inspection of marine infrastructure is frequently compromised by stochastic sea states, introducing risks of high-kinetic impacts, post-landing toppling, and sensory occlusion. This paper proposes a decoupled, multi-vehicle landing framework synchronizing an Unmanned Surface Vehicle (USV) equipped with a 3-RPU stabilized platform with a robust Unmanned Aerial Vehicle (UAV). The architecture utilizes two independent Deep Reinforcement Learning (DRL) agents: a Soft Actor-Critic (SAC) agent providing high-frequency wave-motion compensation for the landing deck, and a multimodal RL agent for the UAVs final approach. Evaluated in high-fidelity maritime simulations, the system achieved a 100% landing success rate across 15 trials in wave states varying from calm to rough. Results show a mean stabilization efficacy of 87.8%, maintaining the landing surface within 1 degree of the horizontal plane for 96% of the mission duration in rough conditions, effectively contributing to safer landings.
Chinese Translation
自主空中检查海洋基础设施常常受到随机海况的影响，带来高动能冲击、着陆后倾覆和传感器遮挡的风险。本文提出了一种解耦的多车辆着陆框架，该框架将配备3-RPU稳定平台的无人水面艇（USV）与稳健的无人机（UAV）进行同步。该架构利用两个独立的深度强化学习（DRL）智能体：一个软演员-评论家（SAC）智能体为着陆甲板提供高频波动补偿，另一个多模态强化学习智能体用于无人机的最终接近。在高保真海洋仿真中评估，该系统在波浪状态从平静到剧烈变化的15次试验中实现了100%的着陆成功率。结果显示，在剧烈条件下，平均稳定效能为87.8%，在96%的任务持续时间内将着陆面保持在水平面1度以内，有效促进了更安全的着陆。
cs.CV / 98 / 2606.31626
PrISM-IQA: Image Quality Assessment Made Practical for Smartphone Photography
PrISM-IQA：为智能手机摄影提供实用的图像质量评估
Abstract
Existing smartphone image quality assessment (IQA) methods commonly reduce perceptual quality to a single score. However, this scalar formulation is poorly aligned with practical image signal processor (ISP) tuning, where engineers must identify specific quality issues, estimate their severities, and determine whether they are acceptable or require intervention. In this work, we introduce a Practical ISP-aware Structured Model for IQA (PrISM-IQA), which reformulates smartphone IQA as a multi-issue ordinal diagnosis problem. Rather than regressing a single quality score, PrISM-IQA predicts an \textit{ordered} severity level -- absent, minor, severe, or critical -- for each ISP-relevant issue, covering both global image-level artifacts and local content-dependent defects. To produce logically consistent predictions, PrISM-IQA combines cumulative ordinal encoding with structured inference that captures within-issue monotonicity as well as cross-issue subsumption and exclusion relations. We evaluate PrISM-IQA on a reconstructed SPAQ benchmark annotated with $53$ ISP-relevant quality issues and on a small-scale expert-annotated real-world dataset. Experimental results demonstrate the effectiveness of PrISM-IQA for practical issue-level diagnosis, reveal transferable perceptual quality representations through linear probing, and further show how its predictions can support actionable and meaningful ISP tuning.
Chinese Translation
现有的智能手机图像质量评估（IQA）方法通常将感知质量简化为一个单一分数。然而，这种标量形式与实际的图像信号处理器（ISP）调优相差甚远，工程师必须识别特定的质量问题，评估其严重性，并确定这些问题是否可接受或需要干预。在本研究中，我们提出了一种实用的ISP感知结构化模型用于图像质量评估（PrISM-IQA），将智能手机IQA重新构造为一个多问题的序数诊断问题。PrISM-IQA不仅回归单一质量分数，而是为每个与ISP相关的问题预测一个 extit{有序}的严重性等级——缺失、轻微、严重或危急，涵盖了全局图像级伪影和局部内容相关缺陷。为了生成逻辑上自洽的预测，PrISM-IQA结合了累积序数编码与结构化推理，捕捉了问题内部的单调性以及跨问题的包含和排除关系。我们在一个重建的SPAQ基准上评估PrISM-IQA，该基准标注了$53$个与ISP相关的质量问题，并在一个小规模的专家标注的真实世界数据集上进行测试。实验结果表明，PrISM-IQA在实际问题级诊断中的有效性，通过线性探测揭示了可迁移的感知质量表示，并进一步展示了其预测如何支持可操作和有意义的ISP调优。
cs.CV / 99 / 2606.31636
LiteMatch: Lightweight Zero-Shot Stereo Matching via Cost Volume Stabilization
LiteMatch：通过成本体积稳定化实现轻量级零样本立体匹配
Abstract
Despite rapid progress in learning-based stereo matching, high accuracy is often achieved at the cost of heavy backbones and computationally intensive 3D cost volume processing, resulting in substantial memory and runtime overhead. More critically, these methods frequently struggle to generalize across domains, limiting their practical deployment. We present \textit{LiteMatch}, a lightweight stereo matching framework that achieves strong zero-shot generalization through cost volume stabilization-without expensive 3D convolutions. LiteMatch employs two complementary encoders: a Cross-View Correspondence Encoder (CVCE) to capture global cross-view interactions, and a High-Frequency Encoder (HFE) that enhances fine structural details via FFT-based frequency cues. To stabilize the cost volume, we introduce the \textit{Cost Volume Consistency Loss (CVC-Loss)}, a voxel-wise binary cross-entropy objective applied to softmax-normalized cost distributions. By encouraging sharp and unimodal disparity probabilities, CVC-Loss promotes stable cost distributions and enables rapid convergence. A lightweight refinement module further produces sharp full-resolution disparities with low-iteration updates, avoiding heavy recurrent refinement. With a flexible design ranging from 3.36M to 9.58M parameters, LiteMatch achieves exceptional zero-shot generalization, delivering competitive EPE and D1 performance across Scene Flow, KITTI, Middlebury, ETH3D, and DrivingStereo. Our results establish that lightweight architectures can indeed generalize across domains without sacrificing accuracy. \href{https://mdraqibkhan.github.io/Litematch}{\textcolor{blue}{Code}}
Chinese Translation
尽管基于学习的立体匹配技术取得了快速进展，但高精度往往以重型主干网络和计算密集型三维成本体积处理为代价，导致显著的内存和运行时间开销。更为关键的是，这些方法在跨领域泛化方面常常面临困难，限制了其实际应用。我们提出了 extit{LiteMatch}，一个轻量级立体匹配框架，通过成本体积稳定化实现强大的零样本泛化——无需昂贵的三维卷积。LiteMatch采用了两个互补的编码器：一个跨视图对应编码器（Cross-View Correspondence Encoder, CVCE）用于捕捉全局跨视图交互，以及一个高频编码器（High-Frequency Encoder, HFE），通过基于快速傅里叶变换（FFT）的频率线索增强细微结构细节。为了稳定成本体积，我们引入了 extit{成本体积一致性损失（Cost Volume Consistency Loss, CVC-Loss）}，这是一个应用于softmax归一化成本分布的体素级二元交叉熵目标。通过鼓励清晰且单峰的视差概率，CVC-Loss促进了稳定的成本分布并实现了快速收敛。一个轻量级的细化模块进一步以低迭代更新生成清晰的全分辨率视差，避免了重型递归细化。LiteMatch的设计灵活，参数范围从3.36M到9.58M，实现了卓越的零样本泛化，在Scene Flow、KITTI、Middlebury、ETH3D和DrivingStereo数据集上提供了具有竞争力的EPE和D1性能。我们的结果表明，轻量级架构确实可以在不牺牲精度的情况下实现跨领域泛化。 extcolor{blue}{代码链接}
cs.CV / 100 / 2606.31645
Technical Report of RoboSpatial Challenge at CVPR 2026: Selective Reasoning Activation and Reference-Frame Disambiguation for Embodied Spatial Reasoning
2026年CVPR RoboSpatial挑战技术报告：用于具身空间推理的选择性推理激活和参考框架消歧
Abstract
Vision-language models achieve strong general perception but often struggle with the spatial reasoning required for embodied tasks. We present RoboSpatialBrain, our submission to the RoboSpatial Challenge at the Embodied Reasoning in Action Workshop, CVPR 2026, built on RoboBrain2.5-8B-NV. RoboSpatialBrain combines two training-free, inference-time mechanisms: a forced prefix activation strategy paired with a task-specific post-prompt that elicits deliberate reasoning on context and compatibility tasks, and an explicit reference-frame redirection pipeline that resolves camera-centric and object-centric ambiguity for context tasks. We additionally explore fine-tuning RoboBrain2.5 on compatibility data and present a detailed analysis of its interaction with prompting. RoboSpatialBrain achieved first place in the RoboSpatial Challenge, with an overall success rate of 80.9\% on RoboSpatial-Home. Code is available at https://github.com/YuxiangXie2003/RoboSpatialBrain.
Chinese Translation
视觉-语言模型在一般感知方面表现出色，但在具身任务所需的空间推理上往往面临挑战。我们提出了RoboSpatialBrain，这是我们在2026年CVPR具身推理行动研讨会的RoboSpatial挑战中的提交，基于RoboBrain2.5-8B-NV构建。RoboSpatialBrain结合了两种无训练、推理时的机制：一种强制的前缀激活策略，配合特定任务的后提示，以引导对上下文和兼容性任务的深思熟虑的推理；以及一个显式的参考框架重定向管道，用于解决上下文任务中的以相机为中心和以物体为中心的歧义。此外，我们还探索了在兼容性数据上微调RoboBrain2.5，并对其与提示的交互进行了详细分析。RoboSpatialBrain在RoboSpatial挑战中获得第一名，在RoboSpatial-Home上的整体成功率为80.9%。代码可在https://github.com/YuxiangXie2003/RoboSpatialBrain获取。
cs.CV / 101 / 2606.31664
Sparsity-Inducing Divergence Losses for Biometric Verification
用于生物特征验证的稀疏诱导散度损失
Abstract
Performance in face and speaker verification is largely driven by margin-penalty softmax losses such as CosFace and ArcFace. Recently introduced $\alpha$-divergence loss functions offer a compelling alternative, particularly due to their ability to induce sparse solutions (when $\alpha>1$). However, standard geometric margins are designed for the softmax function and do not naturally extend to this generalized probabilistic framework. In this paper we propose Q-Margin, a novel $\alpha$-divergence loss that introduces a principled probabilistic margin. Unlike conventional methods that apply geometric penalties to the logits (unnormalized log-likelihoods), Q-Margin encodes the margin penalty directly into the reference measure (prior probabilities). This formulation naturally encourages discriminative embeddings while preserving the beneficial sparsity properties of the $\alpha$-divergence. We demonstrate that Q-Margin achieves competitive or superior performance on the challenging IJB-B and IJB-C face verification benchmarks and similarly strong results in speaker verification on VoxCeleb. Crucially, against ArcFace and CosFace baselines trained under an identical recipe, Q-Margin consistently improves at low False Acceptance Rates (FARs), a capability critical for practical high-security applications. Finally, the extreme sparsity of the Q-Margin posteriors enables exact and memory-efficient training, offering a scalable solution for datasets with millions of identities.
Chinese Translation
在人脸和说话人验证中，性能在很大程度上受到边际惩罚软最大损失（如 CosFace 和 ArcFace）的驱动。最近引入的 $eta$-散度损失函数提供了一种引人注目的替代方案，特别是由于它们能够诱导稀疏解（当 $eta>1$ 时）。然而，标准几何边际是为软最大函数设计的，无法自然扩展到这一广义概率框架。在本文中，我们提出了 Q-Margin，一种新颖的 $eta$-散度损失，引入了一个有原则的概率边际。与传统方法将几何惩罚应用于 logits（未归一化的对数似然）不同，Q-Margin 直接将边际惩罚编码到参考测度（先验概率）中。这种形式自然地鼓励了判别性嵌入，同时保留了 $eta$-散度的有益稀疏性特征。我们证明 Q-Margin 在具有挑战性的 IJB-B 和 IJB-C 人脸验证基准上实现了具有竞争力或更优的性能，并在 VoxCeleb 的说话人验证中取得了同样强劲的结果。重要的是，在采用相同训练方案的 ArcFace 和 CosFace 基线下，Q-Margin 在低假接受率（FAR）下始终表现出改善，这一能力对于实际高安全性应用至关重要。最后，Q-Margin 后验的极端稀疏性使得精确且内存高效的训练成为可能，为拥有数百万身份的数据集提供了一种可扩展的解决方案。
cs.CV / 102 / 2606.31668
SAMBA: A Scatter-Guided Masked Bidirectional Mamba Foundation Model for SAR Target Recognition
SAMBA：一种散射引导的双向掩蔽曼巴基础模型用于合成孔径雷达目标识别
Abstract
Synthetic aperture radar automatic target recognition (SAR ATR) is critical for Earth observation and defense, but its practical deployment is constrained by scarce annotated training data. Self-supervised pre-training alleviates this label bottleneck, yet prevailing Transformer architectures incur prohibitive quadratic computational complexity, and conventional universal masking neglects the unique electromagnetic scattering properties intrinsic to SAR imagery. To address these limitations, we propose SAMBA (Scattering-Guided Bidirectional Mamba), an efficient self-supervised pre-training foundation model for SAR target interpretation. Our framework features three core innovations: (i) a linear-complexity Mamba encoder with a mid-sequence class token to mitigate computational bottlenecks; (ii) a three-level hierarchical Scattering-Guided Masked Autoencoder (SG-MAE) masking strategy guided by SAR physical priors, aligning the pretext task with SAR's intrinsic imaging mechanism; (iii) a lightweight SpatialMix feature interaction module to enhance cross-region feature fusion. We also design a two-stage cross-domain pre-training pipeline to optimize the overall pre-training process. Extensive evaluations demonstrate that SAMBA consistently delivers superior performance across all pre-training configurations, with substantially fewer parameters than both CNN and Transformer baselines. Compared with the default masking strategy in standard MAE, the proposed SG-MAE strategy further boosts the model's few-shot transfer capability. Benchmarking on seven downstream datasets covering classification and detection tasks shows SAMBA achieves state-of-the-art (SOTA) performance on most metrics, fully validating its robust generalizability across diverse SAR interpretation tasks. Source code and pre-trained weights are publicly available at https://github.com/mynswkk/SAMBA.
Chinese Translation
合成孔径雷达自动目标识别（SAR ATR）对于地球观测和国防至关重要，但其实际应用受到标注训练数据稀缺的限制。自监督预训练缓解了这一标签瓶颈，然而现有的Transformer架构却带来了高昂的二次计算复杂度，传统的通用掩蔽方法忽视了SAR图像固有的独特电磁散射特性。为了解决这些限制，我们提出了SAMBA（散射引导双向曼巴），这是一种高效的自监督预训练基础模型，用于SAR目标解释。我们的框架具有三项核心创新：（i）一种线性复杂度的曼巴编码器，配备中序列类别标记，以缓解计算瓶颈；（ii）一种三层层次的散射引导掩蔽自编码器（SG-MAE）掩蔽策略，受SAR物理先验的指导，使预训练任务与SAR的固有成像机制对齐；（iii）一种轻量级的SpatialMix特征交互模块，以增强跨区域特征融合。我们还设计了一个两阶段跨域预训练管道，以优化整体预训练过程。大量评估表明，SAMBA在所有预训练配置中始终提供卓越的性能，参数数量显著少于CNN和Transformer基线。与标准MAE中的默认掩蔽策略相比，所提出的SG-MAE策略进一步提升了模型的少样本迁移能力。在涵盖分类和检测任务的七个下游数据集上的基准测试显示，SAMBA在大多数指标上实现了最先进（SOTA）的性能，充分验证了其在多样化SAR解释任务中的强大泛化能力。源代码和预训练权重已在https://github.com/mynswkk/SAMBA上公开。
cs.CV / 103 / 2606.31672
WorldRoamBench: An Open-World Benchmark for Long-Horizon Stability of Interactive World Models
WorldRoamBench：一个开放世界基准测试，用于交互式世界模型的长时间稳定性
Abstract
Despite rapid progress in interactive world models (IWMs), existing benchmarks evaluate action following only at trajectory level and ignore memory and interaction physics. We introduce WorldRoamBench, an open-world benchmark for long-horizon stability across four dimensions, each with tailored innovations: (i) Action: per-frame action metric bypassing cross-model semantic scale disparity and exposing failures hidden by trajectory; (ii) Vision: segment-based drift metric capturing non-monotonic mid-sequence collapse missed by start-vs-end comparisons; (iii) Physics: controllability-gated evaluation over mechanics, optics, and 3D consistency, scoring plausibility under faithful action execution; (iv) Memory: action-decoupled protocol evaluating scene memory via transition-localized 3D point-cloud reconstruction and subject memory via tracking-plus-VLM reasoning. The benchmark comprises 600+ test cases across Nature, Urban, and Indoor scenes in first/third-person views with WASD 10-60s continuous interaction. Evaluating 10+ open/closed-source models reveals none reliably satisfies all dimensions; even the best achieves only moderate scores. Advances on WorldRoamBench are steps toward IWMs that are stable, physically grounded, memory-faithful, and deployable in real-world applications.
Chinese Translation
尽管交互式世界模型（IWMs）取得了快速进展，但现有基准测试仅在轨迹层面评估动作跟随，忽视了记忆和交互物理。我们介绍了WorldRoamBench，这是一个开放世界基准测试，旨在评估四个维度的长时间稳定性，每个维度都有针对性的创新：（i）动作：每帧动作指标绕过跨模型语义尺度差异，揭示轨迹隐藏的失败；（ii）视觉：基于片段的漂移指标捕捉在序列中间非单调崩溃，而这种崩溃在开始与结束的比较中被忽视；（iii）物理：基于可控性评估机械、光学和三维一致性，在忠实执行动作的情况下评分其合理性；（iv）记忆：通过过渡局部化的三维点云重建评估场景记忆，并通过跟踪加VLM推理评估主体记忆。该基准测试包含600多个测试案例，涵盖自然、城市和室内场景，以第一人称/第三人称视角进行WASD 10-60秒的连续交互。对10多个开源/闭源模型的评估显示，没有一个模型在所有维度上都能可靠地满足要求；即使是表现最好的模型也仅获得中等分数。在WorldRoamBench上的进展是朝着稳定、物理基础、记忆忠实且可在现实世界应用中部署的IWMs迈出的重要一步。
cs.CV / 104 / 2606.31676
REDI: Corpus Aware Patch Ranking for DINOv3 Token Reduction
REDI：基于语料库的DINOv3令牌减少补丁排序
Abstract
Most token reduction methods for Vision Transformers seek favorable tradeoffs between accuracy and efficiency by pruning, merging, or pooling patch tokens. REDI (Relevance for DINOv3 Token Reduction) studies this question through a controlled supervised reference: how should a fixed token budget be allocated across patches for image classification? REDI quantizes final block DINOv3 patch representations into a visual vocabulary and derives class conditioned corpus scores using supervised TF-IDF over visual words. For each validation image, the ground truth class selects a row of the TF-IDF table, and four transformed views produce a TF-IDF map aligned to a reference center crop. A separate dense pass on the same crop provides an attention map. After independent min max normalization, their elementwise product defines the REDI score. A fixed keep, merge, and compress operator then uses score rank to assign patch roles and score magnitude to weight merging and compression. With precomputed REDI scores, a frozen DINOv3 ViT-B/16 backbone, and the same linear classifier used for dense evaluation, the operator reduces the sequence length from 201 to 107 tokens, a 46.8% sequence reduction. The REDI variant based on incoming attention mass achieves 84.706% Top-1 accuracy on ImageNet-1K, compared with 83.514% for the dense baseline, 82.634% for incoming attention mass alone, and 81.796% for supervised TF-IDF alone. The same corpus term also improves reduced classification for three alternative attention formulations relative to their attention only counterparts. Together, these controlled comparisons indicate that class specific corpus statistics and image specific attention provide complementary signals for patch ranking in this setting.
Chinese Translation
大多数针对视觉变换器的令牌减少方法通过修剪、合并或池化补丁令牌，寻求准确性和效率之间的良好权衡。REDI（DINOv3令牌减少的相关性）通过一个受控的监督参考研究了这个问题：在图像分类中，如何在补丁之间分配固定的令牌预算？REDI将最终块DINOv3补丁表示量化为视觉词汇，并使用监督TF-IDF在视觉词上推导类别条件的语料库得分。对于每个验证图像，真实类别选择TF-IDF表的一行，四个变换视图生成与参考中心裁剪对齐的TF-IDF图。对同一裁剪进行单独的密集处理提供了一个注意力图。在独立的最小最大归一化之后，它们的逐元素乘积定义了REDI得分。一个固定的保留、合并和压缩操作符然后使用得分排名来分配补丁角色，并使用得分大小来加权合并和压缩。通过预计算的REDI得分、冻结的DINOv3 ViT-B/16主干网络和用于密集评估的相同线性分类器，该操作符将序列长度从201减少到107个令牌，实现了46.8%的序列减少。基于输入注意力质量的REDI变体在ImageNet-1K上达到了84.706%的Top-1准确率，而密集基线为83.514%，仅输入注意力质量为82.634%，仅监督TF-IDF为81.796%。相同的语料库术语也改善了相对于其仅注意力对应物的三种替代注意力公式的减少分类。总体而言，这些受控比较表明，类别特定的语料库统计和图像特定的注意力为此设置中的补丁排序提供了互补信号。
cs.CV / 105 / 2606.31680
ShellMaker: Language-Guided Exterior Completion under Structural Constraints
ShellMaker：在结构约束下的语言引导外部补全
Abstract
Despite advances in indoor scene generation, synthesizing coherent building exteriors consistent with generated interiors remains largely unexplored. Existing methods can generate floor plans and wall layouts but typically stop at a structural shell, lacking stylistically consistent facades and roofs. Completing these exteriors is challenging because the footprint, wall geometry, and opening semantics must remain fixed-constraints that unconstrained generative models often violate. We introduce ShellMaker, a language-guided exterior completion framework that operates under these structural constraints. Given a building scaffold and a text style prompt, ShellMaker generates a complete exterior mesh with PBR materials by combining parametric roof generation, LLM-based part-aware prompt refinement, joint wall-roof material retrieval, and geometry-aware assembly. Operating on a format agnostic scaffold representation, ShellMaker generalizes to indoor generators, CityGML, and CAD inputs, while maintaining structural consistency and improving architectural coherence over retrieval and unconstrained generative baselines. The project page is available at https://ruiqixu37.github.io/ShellMaker_web/
Chinese Translation
尽管室内场景生成技术已有所进展，但合成与生成的室内空间一致的连贯建筑外观仍然在很大程度上未被探索。现有方法可以生成平面图和墙体布局，但通常仅停留在结构外壳，缺乏风格一致的立面和屋顶。完成这些外部结构具有挑战性，因为建筑的占地面积、墙体几何形状和开口语义必须保持固定的约束，而这些约束往往被无约束的生成模型所违反。我们提出了ShellMaker，一个在这些结构约束下运行的语言引导外部补全框架。给定建筑支架和文本风格提示，ShellMaker通过结合参数化屋顶生成、基于大型语言模型（LLM）的部件感知提示优化、联合墙体-屋顶材料检索和几何感知组装，生成完整的外部网格并应用PBR材料。ShellMaker在格式无关的支架表示上运行，能够推广到室内生成器、CityGML和CAD输入，同时保持结构一致性，并在检索和无约束生成基线之上提高建筑的一致性。项目页面可访问：https://ruiqixu37.github.io/ShellMaker_web/
cs.CV / 106 / 2606.31683
Histogram-constrained Image Generation
直方图约束的图像生成
Abstract
Diffusion models have emerged as a dominant paradigm in generative modeling, enabling high-fidelity sampling from complex data distributions. Despite impressive capabilities, controlling diffusion models to produce outputs aligned with user intent remains an open challenge, especially when balancing global coherence with local precision. Existing control mechanisms vary in the granularity of their conditioning signals. For example, textual prompts guide generation globally through high-level semantics, while ControlNet-like approaches secure precise local structure via dense conditions. In this work, we introduce Histogram-constrained Image Generation (HIG), a novel control mechanism that falls into the middle ground of control granularity. Our framework enforces user-specified distributional constraints (e.g., color histograms or latent token distributions) during the generation process with exact precision. We model such control as an optimal transport (OT) problem and apply explicit guidance transformations during sampling, thereby driving the diffusion trajectory to align with the desired histogram. We demonstrate the versatility of HIG across diverse applications, including constrained generation via color/latent histograms and high-capacity information embedding through histogram-level encoding. Our findings underscore the promise of distributional control, a flexible and interpretable control scheme that is fully compatible with existing control mechanisms, diversifying the hybrid strategies for controllable image generation. Our project page is available at: https://maps-research.github.io/hig/.
Chinese Translation
扩散模型已成为生成建模中的主导范式，使得从复杂数据分布中进行高保真采样成为可能。尽管其能力令人印象深刻，但控制扩散模型生成与用户意图一致的输出仍然是一个开放的挑战，尤其是在平衡全局一致性与局部精确性方面。现有的控制机制在其条件信号的粒度上存在差异。例如，文本提示通过高层语义在全局范围内引导生成，而类似 ControlNet 的方法则通过密集条件确保精确的局部结构。在本研究中，我们提出了直方图约束的图像生成（Histogram-constrained Image Generation, HIG），这是一种新的控制机制，位于控制粒度的中间地带。我们的框架在生成过程中强制执行用户指定的分布约束（例如，颜色直方图或潜在标记分布），并以精确的方式进行建模。我们将这种控制建模为最优传输（Optimal Transport, OT）问题，并在采样过程中应用显式引导变换，从而推动扩散轨迹与所需的直方图对齐。我们展示了 HIG 在多种应用中的多样性，包括通过颜色/潜在直方图进行约束生成，以及通过直方图级编码进行高容量信息嵌入。我们的研究结果强调了分布控制的前景，这是一种灵活且可解释的控制方案，完全兼容现有的控制机制，丰富了可控图像生成的混合策略。我们的项目页面可访问： https://maps-research.github.io/hig/.
cs.CV / 107 / 2606.31688
Semantic Occupancy Prediction with Dual Range-Voxel Representation
基于双范围体素表示的语义占用预测
Abstract
LiDAR-based 3D semantic occupancy prediction, which aims to provide accurate and comprehensive scene representation, is crucial for autonomous driving systems. As point clouds suffer from sparsity and incompleteness, leading to insufficient semantic learning and difficult occupancy perception, existing methods often stack multi-sweep point clouds to obtain dense spatial information. However, such a naive strategy also results in efficiency (e.g., additional computational burden) and robustness (e.g., pose transformation noise) concerns, which hinder their practical applications. In this work, we propose a Dual Range-Voxel Representation (DRVR) that leverages the range-view context and voxel-view geometry of single-sweep point clouds for 3D semantic occupancy prediction, eliminating the concerns associated with the multi-sweeps. Specifically, we use the range-view encoder to extract the compact context of the scene. To fully exploit the spatial information, we design a geometry-aware voxel-view encoder that extracts multi-scale voxel-view features separately and combines them for better geometric occupancy prediction. Moreover, we propose a range-voxel fusion module to cooperate range- and voxel-view features via voxel-to-range and range-to-voxel fusions. Extensive experiments on nuScenes-Occupancy, SemanticKITTI and SemanticPOSS show the superiority of our method. Especially on nuScenes-Occupancy, our single-sweep DRVR achieves 5.4% improvement in mIoU and 2.1x acceleration compared to the multi-sweep method.
Chinese Translation
基于LiDAR的3D语义占用预测旨在提供准确且全面的场景表示，对于自动驾驶系统至关重要。由于点云存在稀疏性和不完整性，导致语义学习不足和占用感知困难，现有方法通常通过堆叠多次扫描的点云来获取密集的空间信息。然而，这种简单的策略也带来了效率（例如，额外的计算负担）和鲁棒性（例如，姿态变换噪声）方面的问题，阻碍了其实际应用。在本研究中，我们提出了一种双范围体素表示（Dual Range-Voxel Representation, DRVR），利用单次扫描点云的范围视图上下文和体素视图几何信息进行3D语义占用预测，消除了与多次扫描相关的顾虑。具体而言，我们使用范围视图编码器提取场景的紧凑上下文。为了充分利用空间信息，我们设计了一种几何感知的体素视图编码器，分别提取多尺度体素视图特征并将其结合以实现更好的几何占用预测。此外，我们提出了一种范围-体素融合模块，通过体素到范围和范围到体素的融合来协同范围视图和体素视图特征。在nuScenes-Occupancy、SemanticKITTI和SemanticPOSS上的大量实验表明了我们方法的优越性。特别是在nuScenes-Occupancy上，我们的单次扫描DRVR在mIoU上提高了5.4%，并且相比于多次扫描方法加速了2.1倍。
cs.CV / 108 / 2606.31695
Intrinsically Stable Spiking Neural Networks: Overcoming the Performance Barrier in the Absence of Batch Normalization
内在稳定的脉冲神经网络：克服无批量归一化情况下的性能障碍
Abstract
The performance of deep spiking neural networks (SNNs) often relies on batch normalization (BN). However, the advanced dynamic BN variants used in state-of-the-art models introduce runtime multiplications, which weaken the hardware-efficiency motivation of SNNs. To address this tension, we identify catastrophic firing-rate decay as a primary cause of severe performance degradation in normalization-free SNNs. Guided by this insight, this work proposes the Intrinsically Stable SNN (IS-SNN) architecture, which removes activation-normalization layers by enforcing signal homeostasis through topology-aware weight standardization and modified residual connections. By folding the standardization operations into static weights offline, IS-SNN removes the runtime statistics tracking and multiplications introduced by activation normalization, restoring an accumulation-oriented inference datapath. Comprehensive experiments show that IS-SNN achieves performance competitive with or superior to computationally expensive dynamic BN techniques across VGG, ResNet, and Transformer-based models. Notably, it achieves a competitive accuracy of 68.05\% on ImageNet and overcomes the severe depth limitations of prior BN-free attempts. Together with a 96.4\% reduction in FPGA lookup table resource consumption for neuron implementations, these results support IS-SNN as a practical framework for building accurate and hardware-friendly deep neuromorphic systems.
Chinese Translation
深度脉冲神经网络（SNNs）的性能通常依赖于批量归一化（BN）。然而，先进的动态 BN 变体在最先进的模型中引入了运行时乘法，这削弱了 SNNs 的硬件效率动机。为了解决这一矛盾，我们确定灾难性的发射率衰减是无归一化 SNNs 性能严重下降的主要原因。基于这一见解，本研究提出了内在稳定的 SNN（IS-SNN）架构，通过通过拓扑感知的权重标准化和修改的残差连接来强制信号稳态，从而去除了激活归一化层。通过将标准化操作折叠到离线静态权重中，IS-SNN 消除了激活归一化引入的运行时统计跟踪和乘法，恢复了以累积为导向的推理数据路径。全面的实验表明，IS-SNN 在 VGG、ResNet 和基于 Transformer 的模型中实现了与计算成本高昂的动态 BN 技术相媲美或更优的性能。值得注意的是，它在 ImageNet 上达到了 68.05\% 的竞争性准确率，并克服了之前无 BN 尝试的严重深度限制。结合对神经元实现的 FPGA 查找表资源消耗减少 96.4\% 的结果，这些结果支持 IS-SNN 作为构建准确且硬件友好的深度类脑系统的实用框架。
cs.CV / 109 / 2606.31699
Look But Don't Touch with Sparse Autoencoders for Unlearning in Diffusion Models
利用稀疏自编码器在扩散模型中进行去学习的观察与不触碰
Abstract
Sparse autoencoders (SAEs) have recently been proposed as interpretable tools for concept-level manipulation, under the assumption that isolated features can serve as controllable intervention points. In this work, we systematically evaluate this assumption in the context of object erasure and steering in diffusion models. We show that while SAEs reliably detect and localize semantic concepts within diffusion model activations, direct intervention in their latent space frequently induces out-of-distribution activations, resulting in severe visual artifacts. To disentangle detection from intervention, we use SAE activations purely as semantic detectors to identify image regions containing the target object, and replace those patch embeddings with the ones that do not contain it. This detection-based replacement preserves the diffusion model's activation statistics and produces significantly cleaner erasure results than latent steering. Our findings reveal a fundamental gap between concept detection and concept intervention in diffusion models: monosemantic or sparse features are not inherently suitable as control knobs for steering. These results position SAEs as powerful interpretability tools for analyzing generative models, but highlight important limitations when used for direct manipulation, such as unlearning.
Chinese Translation
稀疏自编码器（SAEs）最近被提出作为可解释的工具，用于概念级别的操控，前提是孤立特征可以作为可控的干预点。在本研究中，我们系统地评估了这一假设在扩散模型中的物体擦除和引导背景下的有效性。我们展示了虽然SAEs能够可靠地检测和定位扩散模型激活中的语义概念，但在其潜在空间中的直接干预常常会引发分布外激活，导致严重的视觉伪影。为了将检测与干预分离，我们仅使用SAE激活作为语义检测器，以识别包含目标物体的图像区域，并用不包含该物体的补丁嵌入替换这些区域。基于检测的替换保留了扩散模型的激活统计特征，并产生了显著更干净的擦除结果，相较于潜在引导。我们的研究结果揭示了扩散模型中概念检测与概念干预之间的根本差距：单语义或稀疏特征并不天然适合作为引导的控制旋钮。这些结果将SAEs定位为分析生成模型的强大可解释性工具，但在用于直接操控（如去学习）时也突显了重要的局限性。
cs.CV / 110 / 2606.31703
Phantom: A Unified Face-Swap Deepfake Protection Framework with Latent and Spatial Constraints
Phantom：一种具有潜在和空间约束的统一人脸交换深伪保护框架
Abstract
Face-swapping deepfakes pose an escalating threat to personal privacy by enabling unauthorized identity manipulation. While adversarial approaches have demonstrated success against black-box face recognition (FR) models, their applicability to face-swapping scenarios remains underexplored. In particular, reliance on fixed or random targets yields ambiguous latent guidance, and the lack of explicit spatial constraints causes perturbations to spill into identity-irrelevant regions. These issues are further exacerbated by identity-style disentanglement, which suppresses adversarial signals during deepfake generation. In this paper, we present Phantom, a unified face-swap deepfake protection framework that jointly constrains perturbations in latent and spatial domains. Phantom adaptively synthesizes identity-shifted yet attribute-preserving targets to guide identity-aware latent optimization, and applies masked perturbations confined to semantically relevant facial regions. Extensive experiments on state-of-the-art face-swapping deepfakes demonstrate that Phantom improves protection success rates in dodging scenarios by 27.8%, 25.6%, and 16.6% on UniFace, INSwapper, and SimSwap, respectively, while also enhancing visual quality. Furthermore, Phantom generalizes to impersonation scenario, yielding up to 10.2% higher protection while improving perceptual fidelity. These results underscore the effectiveness of jointly leveraging latent and spatial constraints for robust and coherent facial privacy protection.
Chinese Translation
人脸交换深伪对个人隐私构成日益严重的威胁，因为它使未经授权的身份操控成为可能。虽然对抗性方法在黑箱人脸识别（FR）模型中已显示出成功，但其在面向人脸交换场景中的适用性仍未得到充分探讨。特别是，依赖固定或随机目标会导致模糊的潜在指导，而缺乏明确的空间约束会导致扰动溢出到与身份无关的区域。这些问题在身份风格解耦的情况下进一步加剧，抑制了深伪生成过程中的对抗信号。在本文中，我们提出了Phantom，一种统一的人脸交换深伪保护框架，能够在潜在和空间域中共同约束扰动。Phantom自适应地合成身份转移但属性保持的目标，以指导身份感知的潜在优化，并应用限制在语义相关面部区域的掩蔽扰动。在最先进的人脸交换深伪上进行的大量实验表明，Phantom在规避场景中提高了保护成功率，分别在UniFace、INSwapper和SimSwap上提高了27.8%、25.6%和16.6%，同时也提升了视觉质量。此外，Phantom在模仿场景中也表现出良好的泛化能力，保护效果提高了最高10.2%，并改善了感知保真度。这些结果强调了共同利用潜在和空间约束在强健且一致的面部隐私保护中的有效性。
cs.CV / 111 / 2606.31704
WIDER-FAIR: An Annotated Version of the WIDER-FACE Dataset for Fairness Evaluation
WIDER-FAIR：用于公平性评估的WIDER-FACE数据集的注释版本
Abstract
The deployment of face detection models in real-world applications raises important fairness concerns, as these systems may showcase performance disparities across demographic groups. A key obstacle to studying and mitigating such biases is the lack of face detection datasets with sensitive feature annotations. To address this gap, we introduce WIDER-FAIR, a new dataset built on the widely used WIDER-FACE benchmark, manually annotated with the perceived ethnicity and sex of each face. The dataset contains 16,256 images annotated across four ethnic groups: Asian, Black, Indian, and White, and two sex categories. We assess the quality and coherence of the annotations using face embeddings, a K-Nearest Neighbors classifier, and a t-SNE visualization, all of which support the consistency of the labeling process. As a demonstration of the dataset's potential, we train a YOLOv5 model and perform ablation studies on each sensitive feature. Among other findings, our experiments show that detection performance is notably lower for faces of Black individuals, and that excluding this group from training increases fairness disparity more than excluding any other ethnic group. These observations illustrate the value of demographically annotated datasets for understanding and evaluating bias in face detection models.
Chinese Translation
在人脸检测模型的实际应用中，公平性问题引发了重要关注，因为这些系统可能在不同人口群体之间表现出性能差异。研究和缓解此类偏见的一个关键障碍是缺乏带有敏感特征注释的人脸检测数据集。为了解决这一问题，我们引入了WIDER-FAIR，这是一个基于广泛使用的WIDER-FACE基准构建的新数据集，手动注释了每张人脸的感知种族和性别。该数据集包含16,256张图像，注释覆盖四个种族群体：亚洲人、黑人、印度人和白人，以及两个性别类别。我们使用人脸嵌入、K近邻分类器和t-SNE可视化来评估注释的质量和一致性，所有这些都支持标注过程的一致性。作为数据集潜力的展示，我们训练了一个YOLOv5模型，并对每个敏感特征进行了消融研究。在其他发现中，我们的实验表明，黑人个体的人脸检测性能显著较低，并且将该群体排除在训练之外会比排除任何其他种族群体更大程度地增加公平性差异。这些观察结果说明了带有人口统计注释的数据集在理解和评估人脸检测模型中的偏见方面的价值。
cs.CV / 112 / 2606.31732
UniCoder: Unified Visual-to-Code Generation via Symbolic Rewards and Reference-Guided Code Optimization
UniCoder：通过符号奖励和参考引导代码优化实现统一的视觉到代码生成
Abstract
Visual-to-Code generation, which transforms scientific plots, vector graphics, and webpages into executable scripts, demands a level of pixel-precise alignment that standard Multimodal Large Language Models (MLLMs) fail to achieve through Supervised Fine-Tuning (SFT) alone. While Reinforcement Learning (RL) offers a theoretical pathway to bridge this gap, its application is hindered by two fundamental obstacles: (1) \textit{Reward Coarseness}, where semantic metrics like CLIP scores fail to penalize fine-grained element deviations, and (2) \textit{Exploration Stagnation}, where the sparse, heterogeneous code search space prevents the policy from bootstrapping valid trajectories. To overcome these limitations, we introduce UniCoder, a unified RL framework that integrates two novel mechanisms. First, we propose \textbf{Symbolic Attribute Alignment}, which employs a lightweight auxiliary LLM to parse generated code into discrete visual attributes (e.g., hex colors, coordinate limits), enabling dense, element-wise reward computation. Second, to escape local optima, we devise \textbf{Reference-Guided Code Optimization}, a strategy that dynamically injects ground-truth trajectories into low-performing rollout groups, transforming blind exploration into guided policy improvement. Extensive experiments on ChartMimic, UniSVG, Design2Code and ScreenBench benchmarks demonstrate that our 8B-parameter model not only surpasses all open-source baselines but also achieves state-of-the-art performance comparable to proprietary models, establishing a new paradigm for generalized visual-to-code synthesis.
Chinese Translation
视觉到代码生成将科学图表、矢量图形和网页转换为可执行脚本，这要求达到标准的多模态大语言模型（MLLM）通过监督微调（SFT）单独无法实现的像素级精确对齐。尽管强化学习（RL）提供了理论上的解决方案来弥补这一差距，但其应用受到两个基本障碍的限制：（1） extit{奖励粗糙性}，语义指标如CLIP分数未能惩罚细粒度元素偏差；（2） extit{探索停滞}，稀疏且异质的代码搜索空间阻碍了策略启动有效轨迹。为克服这些限制，我们提出了UniCoder，一个统一的强化学习框架，集成了两种新机制。首先，我们提出了 extbf{符号属性对齐}，该机制利用轻量级辅助LLM将生成的代码解析为离散的视觉属性（例如，十六进制颜色、坐标限制），从而实现密集的逐元素奖励计算。其次，为了逃避局部最优，我们设计了 extbf{参考引导代码优化}，这是一种动态将真实轨迹注入表现不佳的回滚组的策略，将盲目探索转变为引导策略改进。在ChartMimic、UniSVG、Design2Code和ScreenBench基准上的大量实验表明，我们的8B参数模型不仅超越了所有开源基线，还达到了与专有模型相当的最先进性能，为通用视觉到代码合成建立了新的范式。
cs.CV / 113 / 2606.31734
MemLearner: Learning to Query Context memory for Video World Models
MemLearner：学习查询上下文记忆以构建视频世界模型
Abstract
Video World Models are interactive video generation models that predict future world states based on user actions and history video frames. A critical challenge in video world models is the lack of memory, causing inconsistent generated scenes over extended durations. Previous methods explored rule-based context frame retrieval as memory, but they fail to generalize in scenarios with scene occlusions and dynamic objects. We propose MemLearner, a learning-based adaptive context query method using query tokens to bridge context and predicted tokens. By leveraging the video generation model itself for context querying, MemLearner exploits pre-trained visual priors without training additional modules from scratch, and incorporates efficient strategies for training and inference. We collect a dataset of long videos with scene occlusions and dynamic objects, paired with camera pose annotations, and propose a multi-dataset training strategy leveraging both annotated rendered and unannotated real-world videos. Extensive experiments demonstrate that MemLearner significantly outperforms prior video world models in terms of scene consistency and memory, particularly under challenging occlusion and dynamic scenarios.
Chinese Translation
视频世界模型是一种交互式视频生成模型，能够基于用户的动作和历史视频帧预测未来的世界状态。在视频世界模型中，一个关键挑战是缺乏记忆，导致在较长时间内生成的场景不一致。之前的方法探索了基于规则的上下文帧检索作为记忆，但在场景遮挡和动态物体的情况下未能有效推广。我们提出了MemLearner，一种基于学习的自适应上下文查询方法，使用查询标记将上下文与预测标记连接起来。通过利用视频生成模型本身进行上下文查询，MemLearner在不从头训练额外模块的情况下，利用了预训练的视觉先验，并结合了高效的训练和推理策略。我们收集了一个包含场景遮挡和动态物体的长视频数据集，并配备了相机姿态注释，提出了一种多数据集训练策略，利用标注的渲染视频和未标注的真实世界视频。大量实验表明，MemLearner在场景一致性和记忆方面显著优于先前的视频世界模型，特别是在具有挑战性的遮挡和动态场景下。
cs.CV / 114 / 2606.31736
Rhythm-Structured Predictive Learning for Remote Photoplethysmography
基于节奏结构的远程光电容积描记预测学习
Abstract
Remote photoplethysmography (rPPG) estimates physiological signals from facial videos by analyzing subtle pulse induced skin color variations. Despite recent progress, existing self-supervised rPPG methods mainly reconstruct masked pixels or low-level visual representations, which can bias the model toward facial appearance rather than latent physiological dy namics. Moreover, most recent Mamba-based approaches scan facial video tokens only in chronological order, limiting their ability to exploit the cyclic structure of pulse signals. To ad dress these limitations, we propose RhythmJEPA, a rhythm structured joint-embedding predictive learning framework for rPPG. Instead of reconstructing RGB frames, RhythmJEPA predicts latent teacher representations from masked facial videos, thereby encouraging physiology-aware representation learning in the embedding space. To explicitly model pulse-related tem poral structure, we introduce a Cyclic Rhythm-State Plan ner (CRSP), which estimates frame-wise latent physiological states and decodes the most plausible cyclic state path via dynamic programming with a constrained transition grammar. Guided by the decoded states, we further design a Dual Order Mamba Encoder (DOM), which combines conventional chronological scanning with state-ordered scanning to capture both local temporal continuity and long-range rhythm-consistent dependencies. Finally, a lightweight Spatial Pulse Mixer (SPM) extracts compact pulse-sensitive facial tokens with a favorable balance between complexity and performance. Experiments on PURE, UBFC-rPPG, and MMPD show competitive performance over representative rPPG methods. The codes are available at https://github.com/deconasser/RhythmJEPA.
Chinese Translation
远程光电容积描记（rPPG）通过分析面部视频中微妙的脉搏引起的皮肤颜色变化来估计生理信号。尽管近年来取得了一些进展，但现有的自监督rPPG方法主要重建被遮挡的像素或低级视觉表示，这可能使模型偏向于面部外观而非潜在的生理动态。此外，大多数基于Mamba的方法仅按时间顺序扫描面部视频标记，限制了其利用脉搏信号周期结构的能力。为了解决这些限制，我们提出了RhythmJEPA，这是一种用于rPPG的基于节奏结构的联合嵌入预测学习框架。RhythmJEPA不是重建RGB帧，而是从被遮挡的面部视频中预测潜在的教师表示，从而鼓励在嵌入空间中进行生理感知的表示学习。为了明确建模与脉搏相关的时间结构，我们引入了一个循环节奏状态规划器（Cyclic Rhythm-State Planner, CRSP），该规划器估计逐帧的潜在生理状态，并通过带有限制转移语法的动态规划解码出最可能的循环状态路径。在解码状态的指导下，我们进一步设计了一个双序Mamba编码器（Dual Order Mamba Encoder, DOM），该编码器结合了传统的时间顺序扫描和状态顺序扫描，以捕捉局部时间连续性和长程节奏一致性依赖关系。最后，一个轻量级的空间脉搏混合器（Spatial Pulse Mixer, SPM）提取出紧凑的脉搏敏感面部标记，在复杂性和性能之间取得了良好的平衡。在PURE、UBFC-rPPG和MMPD上的实验显示出相较于代表性rPPG方法的竞争性能。代码可在 https://github.com/deconasser/RhythmJEPA 获取。
cs.CV / 115 / 2606.31745
JL1-CC&QA: Extending the JL1-CD Benchmark with Change Captioning and Question Answering
JL1-CC&QA：通过变化描述和问题回答扩展JL1-CD基准
Abstract
Remote sensing change detection (CD) traditionally focuses on pixel-level binary segmentation, which identifies where changes occur but neither what nor why. To bridge this semantic gap, we introduce JL1-CC&QA, a multi-task benchmark that extends the JL1-CD dataset with two complementary annotation layers: change captioning (CC) and change question answering (QA). Built upon 5,000 bi-temporal image pairs acquired by the Jilin-1 satellite at 0.5-0.75m ground sample distance, the benchmark comprises: (i) JL1-CC, providing 17,021 quality-verified captions that describe diverse land-cover transformations; and (ii) JL1-QA, offering 20,060 question-answer pairs across eight question types, enabling fine-grained, interactive interrogation of surface changes. All annotations are produced via a three-stage pipeline consisting of multi-modal large language model (LLM) generation, vision-grounded LLM judging, and human expert verification. We hope that JL1-CC&QA, as a benchmark unifying binary change masks, change captions, and change-oriented QA over the same image set, will serve as a valuable resource for the community to advance multi-task change understanding in remote sensing. The dataset is available at https://github.com/circleLZY/JL1-CD.
Chinese Translation
遥感变化检测（CD）传统上侧重于像素级二元分割，识别变化发生的位置，但并不说明变化的内容或原因。为了解决这一语义差距，我们引入了JL1-CC&QA，这是一个多任务基准，扩展了JL1-CD数据集，增加了两个互补的注释层：变化描述（CC）和变化问题回答（QA）。该基准基于吉林一号卫星获取的5000对双时相图像，具有0.5-0.75米的地面样本距离，包含：（i）JL1-CC，提供17021个经过质量验证的描述，描述了多样的土地覆盖变化；（ii）JL1-QA，提供20060个问题-答案对，涵盖八种问题类型，使得对地表变化的细致、互动式询问成为可能。所有注释通过一个三阶段流程生成，该流程包括多模态大型语言模型（LLM）生成、视觉基础的LLM判断和人类专家验证。我们希望JL1-CC&QA作为一个统一二元变化掩码、变化描述和针对同一图像集的变化导向QA的基准，能够为社区提供有价值的资源，以推动遥感中的多任务变化理解。数据集可在https://github.com/circleLZY/JL1-CD获取。
cs.CV / 116 / 2606.31760
Estimating Velocity of Spheres from Rolling-Shutter Image(s)
从滚动快门图像中估计球体的速度
Abstract
Rolling-shutter cameras introduce characteristic distortions when imaging fast moving objects, and these effects are typically treated as artifacts to be corrected. In this work, we instead leverage rolling-shutter distortions as a valuable source of temporal information to estimate the 3D translational and angular velocities of rapidly moving spherical objects from a single rolling-shutter frame. We design a robust and easily detectable spherical pattern and propose a correspondence-free formulation that recovers motion by enforcing geometric consistency in a back-projection framework. By exploiting the geometry of the sphere, translational and rotational motions are decoupled and estimated through a two-stage optimization process, enabling reliable velocity recovery even for textureless objects. Extensive experiments on both synthetic and real datasets demonstrate accurate and robust estimation of motion parameters under challenging high-speed conditions.
Chinese Translation
滚动快门相机在成像快速移动物体时会引入特征性失真，这些效应通常被视为需要纠正的伪影。在本研究中，我们反而利用滚动快门失真作为宝贵的时间信息来源，从单个滚动快门帧中估计快速移动球形物体的三维平移和角速度。我们设计了一种稳健且易于检测的球形图案，并提出了一种无对应关系的公式，通过在反投影框架中强制几何一致性来恢复运动。通过利用球体的几何特性，平移和旋转运动被解耦，并通过两阶段优化过程进行估计，从而即使对于无纹理物体也能实现可靠的速度恢复。在合成和真实数据集上的大量实验表明，在具有挑战性的高速条件下，运动参数的估计准确且稳健。
cs.CV / 117 / 2606.31777
Mesh BDF: Barycentric Dominance Field for 3D Native Mesh Generation
网格BDF：用于3D原生网格生成的重心主导场
Abstract
Autoregressive (AR) modeling has recently achieved remarkable progress in native 3D mesh generation, largely due to its natural ability to handle variable-length, discrete data structures. However, the inherent constraints of the AR paradigm severely restrict the generated meshes, leading to limited face counts, bounded vertex resolutions, and difficulties in supporting textures. To overcome these bottlenecks, we propose the Barycentric Dominance Field (BDF), a continuous representation defined on triangular mesh surfaces that elegantly encodes vertex topological connectivity. BDF bridges the fundamental gap between discrete mesh topology and continuous diffusion-based generative modeling by transforming connectivity into a continuous surface signal. As an intrinsic mesh property, BDF shares strong similarities with texture maps, enabling its seamless integration into existing 3D diffusion pipelines without requiring architectural modifications. Extensive experiments demonstrate that BDF empowers diffusion models to generate native meshes with significantly higher quality, greater scalability, and stronger robustness compared to state-of-the-art autoregressive methods.
Chinese Translation
自回归（AR）建模最近在原生3D网格生成方面取得了显著进展，这主要得益于其自然处理可变长度离散数据结构的能力。然而，AR范式的固有限制严重限制了生成的网格，导致面数有限、顶点分辨率受限以及在支持纹理方面存在困难。为了解决这些瓶颈，我们提出了重心主导场（Barycentric Dominance Field, BDF），这是一种在三角网格表面上定义的连续表示，优雅地编码了顶点的拓扑连接性。BDF通过将连接性转化为连续的表面信号，弥合了离散网格拓扑与基于连续扩散的生成建模之间的根本差距。作为一种内在的网格属性，BDF与纹理图具有很强的相似性，使其能够无缝集成到现有的3D扩散管道中，而无需进行架构修改。大量实验表明，BDF使扩散模型能够生成质量显著更高、可扩展性更强、鲁棒性更好的原生网格，相较于最先进的自回归方法具有明显优势。
cs.CV / 118 / 2606.31781
SpikeLogBERT: Energy-Efficient Log Parsing Using Spiking Transformer Networks
SpikeLogBERT：基于脉冲变换网络的能效日志解析
Abstract
Log parsing is a fundamental step in automated log analysis, transforming raw system logs into structured event templates for downstream tasks such as anomaly detection and system monitoring. Existing log parsing methods range from rule-based and clustering-based approaches to neural models that learn semantic representations from log messages. However, neural approaches typically rely on dense matrix multiplications, which can result in high computational cost and energy consumption. This paper presents SpikeLogBERT, a spiking neural network framework for energy-efficient log parsing. The proposed model integrates a spiking transformer architecture with knowledge distillation from a BERT teacher model, enabling spike-driven computation while preserving semantic representation capability. By leveraging sparse spike activations and event-driven processing, the number of active operations during inference can be significantly reduced. As an initial benchmark study, experiments on the HDFS dataset demonstrate that SpikeLogBERT outperforms ANN-based neural log parsing models with a parsing accuracy of 0.99997, while reducing estimated theoretical energy consumption by up to 62.6% under standard 45nm CMOS assumptions.
Chinese Translation
日志解析是自动化日志分析中的一个基本步骤，它将原始系统日志转化为结构化事件模板，以便于后续的任务，如异常检测和系统监控。现有的日志解析方法包括基于规则和基于聚类的方法，以及从日志消息中学习语义表示的神经模型。然而，神经方法通常依赖于密集的矩阵乘法，这可能导致高计算成本和能量消耗。本文提出了SpikeLogBERT，一种用于能效日志解析的脉冲神经网络框架。所提出的模型将脉冲变换器架构与来自BERT教师模型的知识蒸馏相结合，实现了脉冲驱动的计算，同时保留了语义表示能力。通过利用稀疏的脉冲激活和事件驱动处理，在推理过程中可以显著减少活跃操作的数量。作为初步基准研究，在HDFS数据集上的实验表明，SpikeLogBERT的解析准确率为0.99997，优于基于人工神经网络的日志解析模型，同时在标准45nm CMOS假设下将估计的理论能耗降低了多达62.6%。
cs.CV / 119 / 2606.31785
Self-Supervised Temporal Regularization for Landmark-Based Cardiac Segmentation with Automatic AHA Regional Mapping
基于地标的心脏分割自监督时间正则化与自动AHA区域映射
Abstract
Graph-based cardiac segmentation with implicit anatomical correspondences provides topological guarantees and population-level analysis capabilities, but models trained on independent frames of image sequences exhibit temporal discontinuities that affect reliable clinical measurements, particularly in cardiac ultrasound. In this work, we introduce self-supervised temporal regularization as a post-training refinement stage that exploits the temporal coherence in image sequences to enforce consistent cardiac segmentation and motion estimation over time, without requiring per-frame annotations. By penalizing velocity and acceleration discontinuities across consecutive frames, our method achieves temporally consistent segmentations while maintaining the learned anatomical correspondences. We further leverage these correspondences to automatically map landmarks to the AHA 17-segment clinical standard, enabling standardized regional assessment and detection of pathological myocardial motion patterns. Validation on CAMUS dataset demonstrates the clinical utility of combining temporal consistency with automatic regional mapping. The code is publicly available at https://github.com/david-montalvoo/MaskHybridGNet-TempReg
Chinese Translation
基于图的心脏分割通过隐式解剖对应提供了拓扑保证和群体水平分析能力，但在独立图像序列帧上训练的模型表现出时间不连续性，这影响了可靠的临床测量，特别是在心脏超声中。在本研究中，我们引入自监督时间正则化作为后训练精细化阶段，利用图像序列中的时间一致性来强制执行一致的心脏分割和运动估计，而无需逐帧注释。通过惩罚连续帧之间的速度和加速度不连续性，我们的方法实现了时间上一致的分割，同时保持学习到的解剖对应。我们进一步利用这些对应关系自动将地标映射到AHA 17段临床标准，从而实现标准化的区域评估和病理性心肌运动模式的检测。在CAMUS数据集上的验证展示了将时间一致性与自动区域映射相结合的临床实用性。代码已公开发布于 https://github.com/david-montalvoo/MaskHybridGNet-TempReg
cs.CV / 120 / 2606.31811
MuSViT: A Foundation Vision Model for Sheet Music Representation
MuSViT：用于乐谱表示的基础视觉模型
Abstract
Foundation models have transformed vision and language processing by providing rich, reusable representations that transfer across diverse tasks. Sheet music, as a visual encoding of musical language, lacks such a strong domain-specific backbone. We introduce MuSViT (Music Score Vision Transformer): the first foundation vision model for sheet music representation -- a ViT encoder pre-trained via Masked Autoencoders on 9.7 million pages from the IMSLP. To handle the complexity of real-world scores, we adopt a two-stage curriculum: a synthetic warm-up on typeset scores followed by large-scale training on the full IMSLP corpus. We evaluate MuSViT on four downstream tasks -- full-page and staff-level music score recognition, music symbol detection, and score difficulty classification -- under two scenarios: linear probing (frozen encoder) and fine-tuning. Under linear probing, MuSViT consistently outperforms modern vision encoders, revealing that general-purpose representations, regardless of scale, fall systematically short on the structured symbolic properties of musical notation. Under fine-tuning, MuSViT generally improves upon task-specific state-of-the-art methods. An additional embedding-transcription consistency analysis reveals that MuSViT encodes symbolic musical structure directly in its representation space -- unlike other encoders, whose embeddings do not correlate with music notation content. These results establish MuSViT as a foundation backbone for sheet music understanding.
Chinese Translation
基础模型通过提供丰富、可重用的表示，已在视觉和语言处理领域带来了变革，这些表示能够跨越多种任务进行迁移。然而，乐谱作为音乐语言的视觉编码，缺乏这样强大的领域特定基础。我们介绍了MuSViT（音乐乐谱视觉变换器）：第一个用于乐谱表示的基础视觉模型——一个通过在IMSLP的970万页乐谱上进行掩码自编码器预训练的ViT编码器。为了处理现实世界乐谱的复杂性，我们采用了两阶段的课程：首先在排版乐谱上进行合成热身，然后在完整的IMSLP语料库上进行大规模训练。我们在四个下游任务上评估MuSViT——全页和五线谱级别的乐谱识别、音乐符号检测和乐谱难度分类——在两种场景下进行：线性探测（冻结编码器）和微调。在线性探测下，MuSViT始终优于现代视觉编码器，揭示了通用表示在音乐符号的结构性特征上系统性不足，无论其规模如何。在微调下，MuSViT通常优于特定任务的最新方法。额外的嵌入-转录一致性分析表明，MuSViT在其表示空间中直接编码了符号音乐结构——与其他编码器不同，后者的嵌入与音乐符号内容并不相关。这些结果确立了MuSViT作为乐谱理解的基础支柱。
cs.CV / 121 / 2606.31814
Generative Lane Topology Reasoning via Autoregressive Model with Geometry Prior
基于几何先验的自回归模型生成车道拓扑推理
Abstract
Lane topology reasoning aims to construct a lane graph from onboard sensor observations. Existing methods follow a detection and association paradigm that treats each lane instance independently, leading to geometric inconsistency at connected endpoints and incomplete graphs due to visual occlusions. To address these issues, we propose TopoGPT, a generative framework that learns the geometry prior from typical lane graph structures through autoregressive sequence modeling. Specifically, we construct a large-scale map dataset comprising 3.3M scenes. For each lane graph, a lane tokenizer serializes it into discrete tokens, while a scene context encoder converts it into a rasterized image and extracts global features as scene tokens. We pre-train an autoregressive lane sequence transformer via scene-conditioned next-token prediction, endowing the model with the geometry prior over lane graph structures. Building upon this prior, a perception adapter aligns BEV features from multi-view images with the pre-trained scene condition, transferring the learned geometry prior to sensor-based lane graph prediction. On the OpenLane-V2 benchmark, TopoGPT outperforms existing methods by an average of +6.4 on lane-level and +11.6 on point-level metrics, and produces geometrically consistent and structurally complete lane graphs.
Chinese Translation
车道拓扑推理旨在从车载传感器观测中构建车道图。现有方法遵循检测与关联范式，将每个车道实例独立处理，导致连接端点的几何不一致以及由于视觉遮挡造成的不完整图形。为了解决这些问题，我们提出了TopoGPT，一个生成框架，通过自回归序列建模从典型车道图结构中学习几何先验。具体而言，我们构建了一个包含330万场景的大规模地图数据集。对于每个车道图，车道分词器将其序列化为离散标记，而场景上下文编码器则将其转换为光栅化图像并提取全局特征作为场景标记。我们通过场景条件下的下一个标记预测预训练了一个自回归车道序列变换器，使模型具备了车道图结构的几何先验。在此先验基础上，感知适配器将来自多视角图像的BEV特征与预训练的场景条件对齐，将学习到的几何先验转移到基于传感器的车道图预测上。在OpenLane-V2基准测试中，TopoGPT在车道级别和点级别指标上分别比现有方法平均提高了6.4和11.6，并生成了几何一致且结构完整的车道图。
cs.CV / 122 / 2606.31824
Absorption-Feature-Guided Distance-Decoupled Estimation and Band Selection for LWIR Hyperspectral Passive Ranging
基于吸收特征引导的距离解耦估计与LWIR高光谱被动测距的波段选择
Abstract
Long-wave infrared (LWIR) hyperspectral observations contain distance-dependent atmospheric absorption signatures, providing a physical basis for long-range passive ranging. However, in natural scenes, these signatures are nonlinearly coupled with target temperature, material emissivity, and path radiance, making distance inversion from observed radiance ill posed. Existing methods typically rely on full-band measurements and pixel-wise joint optimization, which is computationally expensive and does not explicitly exploit sharp atmospheric absorption structures. This paper proposes an Absorption-Guided Distance-Decoupled Estimation and Refinement (ADER) framework for LWIR hyperspectral passive ranging. ADER represents emissivity with B-spline control points under a smoothness prior, suppressing overfitting to atmospheric absorption structures and enabling distance-decoupled estimation. It further uses ozone-absorption cues to classify pixels into emission-dominant and reflection-dominant groups. For emission-dominant pixels, ADER compensates path radiance and transmittance and estimates distance by one-dimensional absorption-residual minimization. For reflection-dominant pixels, ADER refines the initial estimate using downwelling-radiance compensation based on the complete radiative model. To reduce spectral redundancy, ADER also introduces a greedy band selection strategy based on multi-scene effective Fisher information for the distance parameter. Experiments on real scenes show that ADER recovers LiDAR-consistent spatial distance structures under both full-band and 20-band settings, improves ranging accuracy in the evaluated regions, and achieves approximately two orders of magnitude speedup over a public full-band hyperspectral ranging method.
Chinese Translation
长波红外（LWIR）高光谱观测包含与距离相关的大气吸收特征，为长距离被动测距提供了物理基础。然而，在自然场景中，这些特征与目标温度、材料发射率和路径辐射非线性耦合，使得从观测辐射中进行距离反演变得困难。现有方法通常依赖于全波段测量和逐像素联合优化，这在计算上代价高昂，并未明确利用尖锐的大气吸收结构。本文提出了一种基于吸收引导的距离解耦估计与精化（ADER）框架，用于LWIR高光谱被动测距。ADER在平滑性先验下用B样条控制点表示发射率，抑制对大气吸收结构的过拟合，并实现距离解耦估计。它进一步利用臭氧吸收线索将像素分类为发射主导和反射主导两组。对于发射主导像素，ADER通过一维吸收残差最小化补偿路径辐射和透射率并估计距离。对于反射主导像素，ADER基于完整辐射模型使用下行辐射补偿来精化初始估计。为了减少光谱冗余，ADER还引入了一种基于多场景有效Fisher信息的贪婪波段选择策略，针对距离参数。在真实场景中的实验表明，ADER在全波段和20波段设置下均能恢复与LiDAR一致的空间距离结构，提高了评估区域的测距精度，并在速度上相较于公共全波段高光谱测距方法实现了约两个数量级的加速。
cs.CV / 123 / 2606.31825
Breaking Failure Cascades: Step-Aware Reinforcement Learning for Medical Multimodal Reasoning
打破失败级联：面向医学多模态推理的步骤意识强化学习
Abstract
Recent multimodal large language models have shown great promise in clinical image reasoning, but existing post-training pipelines remain predominantly outcome-centric, relying on final answer correctness or sequence-level preferences. This suffers from sparse credit assignment, making it difficult to optimize the reasoning process essential for clinical applications. Our analysis reveals that cascading errors from early-stage reasoning failures are a leading cause of incorrect predictions in medical visual question answering (VQA) benchmarks. Motivated by this, we propose Medical Reasoning-aware Policy Optimization (MRPO), an RL algorithm that incorporates step-wise process rewards. When the final answer is incorrect, MRPO assigns exponentially larger penalties to tokens in earlier invalid reasoning steps, breaking failure cascades without compromising successful paths. Across three multimodal LLM backbones, MRPO consistently outperforms standard GRPO and a recent RL baseline, and on Qwen3-VL-8B-Instruct even surpasses substantially larger medical MLLMs such as HuatuoGPT-Vision-34B by 2.79 points. Moreover, MRPO reduces early-stage reasoning failures from 64.0% to 13.0%, showing that targeted mitigation of cascading failures improves both reasoning quality and final answer accuracy. Our code is available at https://github.com/dmis-lab/MRPO
Chinese Translation
近期的多模态大型语言模型在临床图像推理方面展现出了巨大的潜力，但现有的后训练流程仍主要以结果为中心，依赖于最终答案的正确性或序列级偏好。这导致了稀疏的信用分配，使得优化对于临床应用至关重要的推理过程变得困难。我们的分析揭示，早期推理失败导致的级联错误是医学视觉问答（VQA）基准中不正确预测的主要原因。基于此，我们提出了医学推理意识策略优化（Medical Reasoning-aware Policy Optimization, MRPO），这是一种引入逐步过程奖励的强化学习算法。当最终答案不正确时，MRPO会对早期无效推理步骤中的标记施加指数级更大的惩罚，从而打破失败级联，而不影响成功路径。在三个多模态大型语言模型基础上，MRPO始终优于标准的GRPO和最近的强化学习基线，并且在Qwen3-VL-8B-Instruct上甚至比更大规模的医学多模态大型语言模型如HuatuoGPT-Vision-34B高出2.79分。此外，MRPO将早期推理失败率从64.0%降低至13.0%，显示出针对性减轻级联失败能够提高推理质量和最终答案的准确性。我们的代码可在https://github.com/dmis-lab/MRPO获取。
cs.CV / 124 / 2606.31830
PriorEye: Geospatial Visual Priors for End-to-End Autonomous Driving
PriorEye：用于端到端自动驾驶的地理空间视觉先验
Abstract
Most end-to-end autonomous driving methods rely solely on instantaneous sensor observations, limiting them to reactive behavior without the anticipatory foresight human drivers employ through prior experience. We introduce geospatial visual priors, street-level visual context anchored to the intended driving route, providing visual-spatial foresight independent of real-time sensors. We propose a memory augmentation module featuring a dual-memory architecture and an adaptive memory gate, which can be easily integrated into existing end-to-end approaches. This design pairs a contextual memory for retrieved priors with a persistent fallback memory, and dynamically regulates the influence of memories based on current state compatibility. Evaluated on the NAVSIM-v2 benchmark, our approach consistently improves performance across diverse end-to-end baselines. Furthermore, because these priors are independent of onboard sensors, our method inherently improves robustness against sensor corruption, while the dual-memory design ensures safe fallback when the retrieved priors themselves become unreliable. Our project page is available at https://ori-mrg.github.io/PriorEye.
Chinese Translation
大多数端到端自动驾驶方法仅依赖瞬时传感器观测，这限制了它们的反应行为，无法像人类驾驶员那样通过先前经验进行预判。我们引入了地理空间视觉先验，这是一种与预定驾驶路线相锚定的街道级视觉上下文，提供独立于实时传感器的视觉空间预见。我们提出了一种记忆增强模块，采用双记忆架构和自适应记忆门，可以轻松集成到现有的端到端方法中。该设计将用于检索先验的上下文记忆与持久的后备记忆相结合，并根据当前状态的兼容性动态调节记忆的影响。在NAVSIM-v2基准测试中，我们的方法在多种端到端基线中始终提高了性能。此外，由于这些先验独立于车载传感器，我们的方法本质上提高了对传感器损坏的鲁棒性，而双记忆设计确保在检索的先验本身变得不可靠时能够安全后备。我们的项目页面可访问 https://ori-mrg.github.io/PriorEye。
cs.CV / 125 / 2606.31834
Real-Time Source-Free Object Detection
实时无源目标检测
Abstract
Real-world detectors for autonomous driving, surveillance, and robotics must handle domain-shifts under strict latency and memory constraints, yet existing source-free object detection (SFOD) methods rely on heavyweight architectures that prioritize accuracy alone. We show this trade-off is unnecessary: building on YOLOv10, an NMS-free dual-head detector, we achieve state-of-the-art adaptation accuracy while being faster and more compact. We observe that directly applying vanilla mean-teacher self-training to dual-head detectors leads to suboptimal adaptation performance due to two key factors. First, simple pseudo-label generation strategies, such as using a single head or directly combining high-confidence predictions from both heads, yield suboptimal supervision under domain-shift. We propose DHF (Dual-Head Pseudo-Label Fusion) which selectively admits one-to-one (O2O) and one-to-many (O2M) head predictions, preserving precision and recovering missed objects. Second, we observe domain-shift collapses multi-scale feature discriminability. We propose the use of our MARD (Multi-scale Adaptive Representation Diversification) loss which mitigates this by enforcing detection-aware variance and covariance constraints on multi-scale feature maps. Both modules are training-time only, leaving inference unchanged. Across domain-shift benchmarks, our method, RT-SFOD yields 1.4 to 3.5\% mAP gains, 1.3$\times$ higher throughput, with $\sim$2$\times$ fewer parameters than prior state-of-the-art SFOD methods, thus advancing the Pareto frontier of the speed-accuracy-model size trade-off. We report main results with YOLOv10, and demonstrate generalizability with additional YOLO- and DETR-based dual-head detectors. Code is available here: https://github.com/Sairam13001/RT-SFOD/
Chinese Translation
用于自动驾驶、监控和机器人技术的现实世界检测器必须在严格的延迟和内存限制下处理领域转移，而现有的无源目标检测（SFOD）方法依赖于优先考虑准确性的重型架构。我们证明这种权衡是没有必要的：基于YOLOv10，一个无非极大值抑制（NMS）的双头检测器，我们在更快和更紧凑的情况下实现了最先进的适应准确性。我们观察到，直接将普通的平均教师自我训练应用于双头检测器会由于两个关键因素导致次优的适应性能。首先，简单的伪标签生成策略，例如使用单个头或直接结合两个头的高置信度预测，在领域转移下产生次优的监督。我们提出了DHF（双头伪标签融合），选择性地接纳一对一（O2O）和一对多（O2M）头预测，保持精度并恢复遗漏的目标。其次，我们观察到领域转移会导致多尺度特征可辨别性的崩溃。我们提出使用MARD（多尺度自适应表示多样性）损失，通过对多尺度特征图施加检测感知的方差和协方差约束来缓解这一问题。这两个模块仅在训练时使用，不改变推理过程。在领域转移基准测试中，我们的方法RT-SFOD实现了1.4%到3.5%的mAP增益，吞吐量提高了1.3倍，参数数量比之前最先进的SFOD方法减少了约2倍，从而推进了速度、准确性和模型大小权衡的帕累托前沿。我们报告了与YOLOv10的主要结果，并通过额外的基于YOLO和DETR的双头检测器展示了其通用性。代码可在此获取：https://github.com/Sairam13001/RT-SFOD/
cs.CV / 126 / 2606.31839
Towards Voxel Spacing Consistency for Medical Image Segmentation
朝向医学图像分割中的体素间距一致性
Abstract
Volumetric medical image segmentation is essential for both preoperative diagnosis and intraoperative guidance. While recent years have witnessed rapid progress in segmentation architectures, comparatively little attention is paid to the physical voxel spacing of anatomical data. Indeed, volumetric image resampling is a ubiquitous preprocessing step before segmentation, yet its interaction with downstream segmentation has not been systematically exploited. In this work, we study the correlation between image resampling and segmentation, and propose Consispace, a semantic-aware resampling framework that achieves consistent voxel spacing in the axial direction while preserving anatomical and semantic consistency. Consispace introduces an ODE-based anatomical constraint to model inter-slice dynamics with a continuous interpolator, enabling faithful reconstruction under complex anatomical transitions beyond discrete interpolation. To further couple resampling with segmentation objectives, we leverage dense features from a pretrained vision model to build intra-slice semantic correlation maps and inject class-wise semantic consistency via feature reweighting during resampling. Both intra-slice and inter-slice constraints are integrated into an implicit neural network, supporting arbitrary-scale resampling. Extensive experiments on multiple datasets demonstrate that Consispace achieves superior reconstruction quality and perceptual fidelity, produces smoother inter-slice anatomy, and improves downstream segmentation performance when used as a preprocessing step.
Chinese Translation
体积医学图像分割对于术前诊断和术中指导至关重要。尽管近年来分割架构取得了快速进展，但对解剖数据的物理体素间距关注相对较少。实际上，体积图像重采样是分割前的普遍预处理步骤，但其与下游分割的相互作用尚未得到系统性利用。在本研究中，我们探讨了图像重采样与分割之间的相关性，并提出了Consispace，一个语义感知的重采样框架，能够在保持解剖和语义一致性的同时，实现轴向方向上的体素间距一致性。Consispace引入了一种基于常微分方程（ODE）的解剖约束，通过连续插值器建模切片间动态，能够在复杂解剖过渡下实现真实重建，超越离散插值。为了进一步将重采样与分割目标结合，我们利用预训练视觉模型的密集特征构建切片内语义相关图，并在重采样过程中通过特征重加权注入类别级语义一致性。切片内和切片间约束被整合到一个隐式神经网络中，支持任意规模的重采样。在多个数据集上的广泛实验表明，Consispace在重建质量和感知保真度方面表现优越，产生更平滑的切片间解剖结构，并在作为预处理步骤时提高下游分割性能。
cs.CV / 127 / 2606.31875
SENSE-VAD: Sentient and Semantic Video Anomaly Detection for Autonomous Driving
SENSE-VAD：用于自动驾驶的感知与语义视频异常检测
Abstract
Autonomous vehicles (AVs) must navigate not only motion-based hazards but also socially complex situations whose danger is constituted by inter-agent relationships rather than movement statistics alone. A child running away from a guardian, a person being carried by another, or a pursuer chasing a pedestrian across a sidewalk are all anomalous in social context, yet none produces an obvious motion signal that current anomaly detectors are equipped to flag. We introduce SENSE-VAD, the first synthetic video anomaly detection benchmark for autonomous driving explicitly designed around socially complex anomalies. Using the CARLA simulator and Unreal Engine (UE), we generate distinct anomaly scenarios across multiple categories: individual behaviors, group behaviors, person--object interactions, cyclist interactions, vehicle & agent, each annotated with per-frame binary labels. A key design principle is the separation of social anomaly from motion-based or appearance-based anomaly: many scenarios involve motion of objects that appears unremarkable in isolation but is anomalous in relational context. We additionally provide real-world normal and anomalous videos as a sim-to-real transfer probe. We evaluate state-of-the-art video anomaly detection baselines and demonstrate that socially complex anomalies constitute a distinct and currently unsolved challenge. Our dataset, annotations, and generation code are publicly available.
Chinese Translation
自动驾驶车辆（AVs）必须应对不仅基于运动的危险，还要处理社会复杂情境，这些情境的危险源于代理之间的关系，而不仅仅是运动统计数据。一个孩子从监护人身边跑开、一个人被另一个人抬着，或是一个追逐者在步道上追赶行人，这些在社会背景下都是异常的情况，但没有一个产生明显的运动信号，当前的异常检测器无法标记。我们提出了SENSE-VAD，这是第一个专门针对社会复杂异常设计的自动驾驶合成视频异常检测基准。使用CARLA模拟器和虚幻引擎（Unreal Engine, UE），我们生成了多个类别的独特异常场景：个体行为、群体行为、人与物体的互动、骑自行车者的互动、车辆与代理的互动，每个场景都附有逐帧的二元标签。一个关键设计原则是将社会异常与基于运动或外观的异常分开：许多场景涉及的物体运动在孤立状态下看似平常，但在关系背景下却是异常的。此外，我们还提供了真实世界的正常和异常视频作为模拟到真实的转移探针。我们评估了最先进的视频异常检测基准，并证明社会复杂异常构成了一个独特且当前未解决的挑战。我们的数据集、注释和生成代码均已公开。
cs.CV / 128 / 2606.31895
RESOLVE: A Multi-Resolution and Multi-Modal Dataset for Roadside Cooperative Perception
RESOLVE：一个用于路边协同感知的多分辨率和多模态数据集
Abstract
LiDAR has increasingly been integrated into traffic cameras to expand coverage and mitigate occlusion in roadside cooperative perception. However, how unimodal and camera-LiDAR fusion architectures behave under variations in LiDAR point sparsity induced by sensor configurations and scene-dependent sensing conditions remains underexplored. We introduce RESOLVE, a large-scale real-world benchmark dataset featuring multi-resolution roadside LiDAR and synchronized camera-LiDAR sensing for systematic evaluation of unimodal and fusion-based architectures in roadside 3D detection and tracking. RESOLVE contains over 100k images and 26k point cloud frames with 220k manually annotated bounding boxes, captured at a real-world urban intersection across diverse lighting and weather conditions and spanning 10 classes of traffic participants. In particular, RESOLVE enables controlled evaluation across three LiDAR resolution levels while keeping all other sensing and environmental factors fixed. This allows fair cross-architecture comparisons under point cloud distribution shifts resulting from resolution variations, sensing distance, and training-inference resolution mismatches. Results from extensive benchmark experiments reveal insights into how multimodal fusion can compensate for LiDAR point sparsity, offering clues for designing cost-efficient roadside multimodal perception. The dataset and benchmark codes are available at https://github.com/ASU-Suo-Lab/RESOLVE.
Chinese Translation
激光雷达（LiDAR）越来越多地与交通摄像头集成，以扩展覆盖范围并减轻路边协同感知中的遮挡。然而，在传感器配置和场景依赖的感知条件引起的激光雷达点稀疏性变化下，单模态和摄像头-激光雷达融合架构的表现仍然未得到充分探索。我们引入了RESOLVE，一个大型真实世界基准数据集，具有多分辨率的路边激光雷达和同步的摄像头-激光雷达感知，旨在系统评估路边3D检测和跟踪中的单模态和基于融合的架构。RESOLVE包含超过10万张图像和2.6万帧点云数据，配有22万个手动标注的边界框，这些数据是在一个真实的城市交叉口捕获的，涵盖了多种光照和天气条件，并涉及10类交通参与者。特别是，RESOLVE使得在保持所有其他感知和环境因素固定的情况下，可以对三个激光雷达分辨率级别进行控制评估。这允许在由于分辨率变化、感知距离和训练-推理分辨率不匹配导致的点云分布变化下，公平地进行跨架构比较。大量基准实验的结果揭示了多模态融合如何弥补激光雷达点稀疏性，为设计成本高效的路边多模态感知提供了线索。数据集和基准代码可在 https://github.com/ASU-Suo-Lab/RESOLVE 获取。
cs.CV / 129 / 2606.31903
Attend, Transform, or Silence: Operator-Level Visual Skipping for Efficient Multimodal LLM Inference
注意、转换或静默：高效多模态大语言模型推理的操作级视觉跳过
Abstract
Multimodal large language models (MLLMs) increasingly process long visual-token sequences, increasing the overall inference computation. Existing acceleration methods usually remove visual tokens or skip visual-token updates in entire layers, but these coarse strategies may discard fine-grained evidence or suppress useful operators together with redundant ones. In this paper, we study visual-token computation from an answer-observable perspective and find that late visual-token updates can remain large while having little effect on answer-token representations. Motivated by this answer-silent redundancy, we decompose each Transformer layer into attention and FFN operators and show that useful visual computation is often operator-dominant and layer-dependent. We propose an operator-level visual-token skipping framework that preserves the full visual-token sequence while selectively bypassing redundant attention, FFN, or both. Experiments across three MLLM architectures and 10 VQA benchmarks show that our method achieves strong efficiency-accuracy trade-offs, reducing \textbf{33.7\%} TFLOPs on Qwen3-VL while retaining \textbf{99.5\%} of the vanilla model performance.
Chinese Translation
多模态大语言模型（MLLMs）越来越多地处理长视觉标记序列，从而增加了整体推理计算量。现有的加速方法通常会移除视觉标记或跳过整个层的视觉标记更新，但这些粗略策略可能会丢弃细粒度证据或将有用的操作与冗余的操作一起抑制。在本文中，我们从答案可观测的角度研究视觉标记计算，发现晚期的视觉标记更新可以保持较大，同时对答案标记表示的影响很小。基于这种答案静默冗余的动机，我们将每个Transformer层分解为注意力和前馈网络（FFN）操作，并展示有用的视觉计算通常是操作主导且依赖于层的。我们提出了一种操作级视觉标记跳过框架，该框架保留完整的视觉标记序列，同时选择性地绕过冗余的注意力、FFN或两者。针对三种MLLM架构和10个视觉问答（VQA）基准的实验表明，我们的方法在效率与准确性之间实现了良好的权衡，在Qwen3-VL上减少了 extbf{33.7 ext{%}}的TFLOPs，同时保留了 extbf{99.5 ext{%}}的原始模型性能。
cs.CV / 130 / 2606.31918
DriveWeaver: Point-Conditioned Video Inpainting for Controllable Vehicle Insertion in Autonomous Driving Simulation
DriveWeaver：用于可控车辆插入的点条件视频修复在自动驾驶仿真中的应用
Abstract
A pivotal step in autonomous driving simulation involves inserting foreground vehicles with predefined trajectories into simulated scenes. This process enhances scene diversity and facilitates the creation of various corner cases for testing and improving autonomous driving models. However, existing methods often rely on pre-reconstructed 3D assets, which frequently lead to lighting inconsistencies between the inserted foreground and the background. Moreover, the reliance on limited, manually-curated 3D assets hinders large-scale deployment. To address these challenges, we propose DriveWeaver, a novel framework for controllable vehicle insertion in autonomous driving simulation. Specifically, for a masked target insertion area, DriveWeaver performs video inpainting conditioned on vehicle point clouds to generate high-quality, temporally consistent vehicles. This video-inpainting-based approach ensures seamless blending between the foreground and background, while the readily available point cloud conditions enable superior generalization. To support long-term generation, we further design a global-to-local hierarchical inpainting strategy, ensuring the consistent identity and appearance of the inserted vehicles. Meanwhile, we extract explicit 3D Gaussian representations of the inserted vehicles through an urban reconstruction pipeline to enable real-time rendering for autonomous driving simulation. Extensive experiments across diverse datasets demonstrate that our method outperforms existing baselines in visual realism and geometric consistency, providing a robust tool for scalable autonomous driving scene augmentation.
Chinese Translation
自动驾驶仿真中的一个关键步骤是将具有预定义轨迹的前景车辆插入到模拟场景中。这个过程增强了场景的多样性，并促进了各种边缘案例的创建，以测试和改进自动驾驶模型。然而，现有方法通常依赖于预先重建的3D资产，这常常导致插入的前景与背景之间的光照不一致。此外，依赖有限的、手动策划的3D资产也阻碍了大规模部署。为了解决这些挑战，我们提出了DriveWeaver，一个用于自动驾驶仿真中可控车辆插入的新框架。具体而言，对于一个被遮罩的目标插入区域，DriveWeaver基于车辆点云执行视频修复，以生成高质量、时间一致的车辆。这种基于视频修复的方法确保了前景与背景之间的无缝融合，而现成的点云条件则实现了更优的泛化能力。为了支持长期生成，我们进一步设计了一个从全局到局部的分层修复策略，确保插入车辆的一致身份和外观。同时，我们通过城市重建管道提取插入车辆的显式3D高斯表示，以实现自动驾驶仿真的实时渲染。在多样化数据集上的广泛实验表明，我们的方法在视觉真实感和几何一致性方面优于现有基线，为可扩展的自动驾驶场景增强提供了强大的工具。
cs.CV / 131 / 2606.31924
InstanceControl: Controllable Complex Image Generation without Instance Labeling
InstanceControl：无需实例标注的可控复杂图像生成
Abstract
Controllable image generation methods, such as ControlNet, have demonstrated a remarkable capacity to introduce visual conditions(e.g., depth maps) to guide image generation. However, these methods often struggle with complex multi-instance scenes, frequently leading to attribute confusion among instances. While recent approaches attempt to mitigate this via manual instance labeling, such requirements are labor-intensive. In this paper, we propose InstanceControl, a novel multi-instance controllable generation method that eliminates the need for instance labeling. We identify the primary bottleneck in existing methods as the inability to accurately associate instance descriptions with their corresponding regions within visual conditions. To address this, we leverage the Vision-Language Model (VLM) to establish instance-level correspondences between text prompts and visual conditions. Specifically, the VLM automatically parses instance descriptions from the text prompts and simultaneously predicts instance masks based on the visual conditions. Furthermore, since the predicted masks may contain noise, we introduce an adaptive mask refinement strategy that dynamically refines these instance masks during the generation process. Extensive experiments demonstrate that our approach outperforms state-of-the-art methods, achieving superior fidelity and precise instance-level control.
Chinese Translation
可控图像生成方法，如 ControlNet，已展示出在引入视觉条件（例如深度图）以指导图像生成方面的显著能力。然而，这些方法在处理复杂的多实例场景时常常面临实例属性混淆的问题。尽管近期的一些方法试图通过手动实例标注来缓解这一问题，但这种需求往往劳动密集。在本文中，我们提出了 InstanceControl，一种新颖的多实例可控生成方法，消除了对实例标注的需求。我们确定现有方法的主要瓶颈在于无法准确将实例描述与视觉条件中的相应区域关联。为了解决这个问题，我们利用视觉-语言模型（Vision-Language Model, VLM）在文本提示和视觉条件之间建立实例级的对应关系。具体而言，VLM 自动解析文本提示中的实例描述，并基于视觉条件同时预测实例掩膜。此外，由于预测的掩膜可能包含噪声，我们引入了一种自适应掩膜精炼策略，在生成过程中动态地精炼这些实例掩膜。大量实验表明，我们的方法在保真度和实例级控制的精确性方面优于最先进的方法。
cs.CV / 132 / 2606.31933
No Place to Hide: Benchmarking Video Hallucination with Background-Controlled Pairs
无处可藏：基于背景控制对的视频幻觉基准测试
Abstract
We introduce VidPair-Halluc, a new benchmark for evaluating video hallucination in large video models (LVMs) under rigorous and controlled conditions. Unlike previous benchmarks that primarily rely on text-based perturbations or adversarial questions while neglecting the consistency of visual backgrounds, VidPair-Halluc features video pairs with highly similar backgrounds but distinctly different foreground semantics, enabling precise attribution of model errors to genuine hallucination rather than background variation. The benchmark is constructed through PairFlow, a pipeline that leverages recent advances in text-to-image and video generation to systematically compose stories, generate coherent video clips, and assemble them into adversarial pairs. Covering both spatial and temporal reasoning across ten semantic aspects, VidPair-Halluc comprises 1K high-quality adversarial video pairs and 11K spatio-temporal QA pairs with control over background and foreground variations. Evaluations on mainstream LVMs show persistent difficulty with robust fine-grained video understanding in adversarial settings, and code and data are available at the https://jethrojames.github.io/VidPair-Halluc/.
Chinese Translation
我们介绍了VidPair-Halluc，这是一个在严格控制条件下评估大型视频模型（LVMs）视频幻觉的新基准。与以往主要依赖文本干扰或对抗性问题而忽视视觉背景一致性的基准不同，VidPair-Halluc特征是具有高度相似背景但前景语义明显不同的视频对，使得模型错误的精确归因于真实幻觉而非背景变化。该基准通过PairFlow构建，利用近期在文本到图像和视频生成方面的进展，系统性地编排故事，生成连贯的视频片段，并将其组装成对抗性对。VidPair-Halluc涵盖了十个语义方面的空间和时间推理，包含1000个高质量对抗性视频对和11000个具有背景和前景变化控制的时空问答对。对主流LVM的评估显示，在对抗性环境中，进行稳健的细粒度视频理解仍然面临持续困难，代码和数据可在https://jethrojames.github.io/VidPair-Halluc/获取。
cs.CV / 133 / 2606.31946
World Narrative Model for Highly Controllable Video Generation: A Paradigm Shift from Pixel Sampling to Physical World Orchestration
高度可控视频生成的世界叙事模型：从像素采样到物理世界编排的范式转变
Chen, Ye, Chen, Xuanhong, Zhu, Yupeng, Tan, Liming, Wan, Zhewen, Xiong, Yuxuan, Wang, Tielong, Liu, Jinfan, Zhang, Wuze, Zhang, Xiongzhen, Li, Feifei, Luo, Xianglin, Zhao, Zhehan, Zhang, Zhifan, Kou, Laisheng, Liang, Zhujing, Chen, Yugang, Chen, Muchun, Miao, Xu, Zhang, Yijing, Sheng, Xiaojie, Hu, Qiang, Chen, Jialiang, Zhang, Weimin, Zhang, Wenjun, Ni, Bingbing
Abstract
The fundamental obstacle to industrial grade video generation is the lack of controllability: existing models treat video as a pixel distribution sampling problem, bypassing the explicit, instance level $4D$ $(3D + T)$ physical world. Consequently, content creators cannot specify geometry, motion, camera parameters, or lighting in a deterministic, quantitative way, leading to the infamous ''gacha'' loop that makes professional content creation prohibitively inefficient and expensive. To address this, we introduce the World Narrative Model (WNM), a paradigm that decouples what to render -- the structured physical narrative -- from how to render -- the pixel generation process. WNM replaces end-to-end black-box sampling with orchestrated $4D$ pre-visualization for media generation. Collaborative agents translate sparse multimodal inputs, including text, reference videos, and sketches, into a fully editable world representation with scene geometry, object layouts, character/animal skeleton motion, trajectories, camera motion, and lighting at quantitative, physically meaningful granularity. This representation acts as a deterministic structural blueprint that drives existing video foundation models, either frozen or lightly adapted, to render final footage, turning the base model into a faithful neural shader. Built on this engine, our human-AI platform supports automatic world generation and pre-visualization aligned with professional filmmaking pipelines, while director consoles enable seamless human refinement. Experiments show that WNM greatly reduces probabilistic ``gacha'' calls and produces videos whose layout, motion, and cinematography closely follow creator intent. The framework is open and modular, allowing each component, such as world representation, control agents, and adapters, to be independently improved. Project website: https://glassroom.sjtu.edu.cn/WNM/.
Chinese Translation
工业级视频生成的根本障碍在于缺乏可控性：现有模型将视频视为像素分布采样问题，绕过了明确的实例级 $4D$ $(3D + T)$ 物理世界。因此，内容创作者无法以确定性和定量的方式指定几何形状、运动、相机参数或照明，导致了臭名昭著的“扭蛋”循环，使专业内容创作变得极其低效和昂贵。为了解决这一问题，我们提出了世界叙事模型（World Narrative Model, WNM），这一范式将渲染内容——结构化的物理叙事——与渲染方式——像素生成过程解耦。WNM用编排的 $4D$ 预可视化替代了端到端的黑箱采样，用于媒体生成。协作代理将稀疏的多模态输入（包括文本、参考视频和草图）转换为一个完全可编辑的世界表示，包含场景几何、物体布局、角色/动物骨架运动、轨迹、相机运动和照明，具有定量和物理上有意义的粒度。该表示作为一个确定性的结构蓝图，驱动现有的视频基础模型，无论是冻结的还是轻微调整的，来渲染最终的影像，将基础模型转变为一个忠实的神经着色器。在此引擎基础上，我们的人机智能平台支持与专业电影制作流程对齐的自动世界生成和预可视化，同时导演控制台实现无缝的人类精细调整。实验表明，WNM大大减少了概率性的“扭蛋”调用，并生成了布局、运动和摄影与创作者意图紧密相符的视频。该框架是开放和模块化的，允许每个组件（如世界表示、控制代理和适配器）独立改进。项目网站：https://glassroom.sjtu.edu.cn/WNM/
cs.CV / 134 / 2606.31959
AnyBokeh: Physics-Guided Any-to-Any Bokeh Editing with Optical Fingerprint Transfer
AnyBokeh：基于物理指导的任意到任意虚化编辑与光学指纹转移
Abstract
Depth-of-field control is a fundamental tool in photography, yet post-capture bokeh editing from a single image remains challenging. A practical editor should handle images captured under arbitrary focus and aperture settings. Existing methods typically assume an all-in-focus input, or first recover an all-in-focus image before rendering new bokeh. Such pipelines can discard useful blur cues from the source image and propagate reconstruction artifacts into the final edit. We introduce AnyBokeh, a physics-guided framework for any-to-any bokeh editing. Instead of treating source blur merely as a degradation to be removed, AnyBokeh estimates the source blur state with a signed circle-of-confusion map and a disparity map. By modeling the linear relation between signed circle of confusion and disparity difference, AnyBokeh estimates a source-specific optical fingerprint and transfers the source optical characteristics to the desired focus and aperture setting. A generative editor conditioned on both source and target circle-of-confusion maps then performs relative blur synthesis, enabling spatially adaptive deblurring, preservation, and defocus rendering. To support physically supervised learning, we further construct a high-fidelity synthetic dataset with accurate depth, focus distance, and full EXIF metadata. Experiments on real-world benchmarks show that AnyBokeh achieves faithful and controllable editing across any-to-any bokeh editing, all-in-focus-to-bokeh rendering, and defocus deblurring, while avoiding all-in-focus reconstruction and test-time bokeh-level calibration commonly required by existing approaches. The code and dataset will be available at https://github.com/itsmag11/AnyBokeh.
Chinese Translation
景深控制是摄影中的一项基本工具，但从单幅图像进行后期虚化编辑仍然具有挑战性。一个实用的编辑器应能够处理在任意焦点和光圈设置下拍摄的图像。现有方法通常假设输入图像是全焦的，或者首先恢复全焦图像，然后再渲染新的虚化。这种流程可能会丢弃源图像中的有用模糊线索，并将重建伪影传播到最终编辑中。我们提出了AnyBokeh，一个用于任意到任意虚化编辑的物理指导框架。AnyBokeh并不只是将源模糊视为需要去除的退化，而是通过带符号的混淆圆图和视差图来估计源模糊状态。通过建模带符号的混淆圆与视差差异之间的线性关系，AnyBokeh估计源特定的光学指纹，并将源光学特性转移到所需的焦点和光圈设置。一个基于源和目标混淆圆图的生成编辑器随后执行相对模糊合成，使得空间自适应去模糊、保留和虚焦渲染成为可能。为了支持物理监督学习，我们进一步构建了一个高保真合成数据集，包含准确的深度、焦距和完整的EXIF元数据。在真实世界基准上的实验表明，AnyBokeh在任意到任意虚化编辑、全焦到虚化渲染和虚焦去模糊方面实现了忠实且可控的编辑，同时避免了现有方法通常要求的全焦重建和测试时虚化水平校准。代码和数据集将发布在 https://github.com/itsmag11/AnyBokeh。
cs.CV / 135 / 2606.31979
Planar-SfM: Camera Pose Estimation via Homography Graph Embeddings
平面结构从运动（Planar-SfM）：通过单应性图嵌入进行相机姿态估计
Abstract
Structure from Motion (SfM) systems traditionally struggle with planar scenes, where standard epipolar geometry-based methods become degenerate. Rather than viewing planar surfaces as a limitation, we propose a unified framework that leverages them as a source of geometric constraints. Our key insight is that each planar surface visible across multiple views provides an independent estimate of relative camera poses through homography decomposition. By aggregating estimates from multiple planes or even from a single dominant plane we achieve robust pose recovery in scenarios where traditional methods fail. We introduce a novel graph-based approach that constructs a pose-graph from homography estimates and employs spectral embedding to identify and filter unreliable edges. Our method maps homography-based pose estimates onto the real line based on their geometric and visual consistency, enabling efficient extraction of a maximally consistent spanning tree for pose recovery. This approach naturally handles both highly planar scenes, such as indoor sports arenas, and general $3$D environments. We demonstrate superior performance on basketball court imagery where existing methods struggle, while matching or exceeding state-of-the-art results on unconstrained outdoor scenes from the IMC Phototourism benchmark.
Chinese Translation
传统的结构从运动（SfM）系统在平面场景中通常面临困难，因为基于标准极几何的方法会变得退化。我们并不将平面表面视为限制，而是提出一个统一框架，将其作为几何约束的来源。我们的关键见解是，每个在多个视角中可见的平面表面通过单应性分解提供了相对相机姿态的独立估计。通过聚合来自多个平面或甚至单个主导平面的估计，我们在传统方法失效的场景中实现了稳健的姿态恢复。我们引入了一种新颖的基于图的方法，该方法从单应性估计构建姿态图，并利用谱嵌入来识别和过滤不可靠的边。我们的方法根据单应性估计的几何和视觉一致性将其映射到实数线上，从而有效提取出一个最大一致的生成树用于姿态恢复。该方法自然地处理高度平面的场景，例如室内体育场，以及一般的三维环境。我们在篮球场图像上展示了优越的性能，而在现有方法表现不佳的情况下，我们的结果与IMC Phototourism基准中无约束户外场景的最先进结果相匹配或超越。
cs.CV / 136 / 2606.31981
LUNA: Learning Universal 3D Human Animation Beyond Skinning
LUNA：超越蒙皮的通用三维人类动画学习
Abstract
Creating photorealistic, animatable 3D human avatars from monocular images still largely depends on Linear Blend Skinning (LBS) and parametric body models, which constrain expressivity and often introduce artifacts due to imperfect fitting. We propose LUNA, an LBS-free universal neural animation model that directly maps multiple 2D controls like images, keypoints, sketches, and unseen characters into 3D Gaussian deformations, bypassing explicit body fitting. At its core, a transformer-based motion regressor disentangles global rigid motion from fine-grained local dynamics to capture both coherent movement and subtle non-rigid effects. To resolve the inherent ambiguity of 2D-to-3D lifting while scaling beyond fitted datasets, we introduce hybrid supervision that distills soft structural priors from an LBS teacher and a loss that supports training on both limited fitted data and large in-the-wild unlabeled videos. Extensive experiments show LUNA achieves competitive visual fidelity compared to LBS-based approaches, while delivering realistic human motion and zero-shot cross-identity generalization across diverse driving modalities. To the best of our knowledge, LUNA is the first end-to-end 3D animatable model that supports implicit 2D driving.
Chinese Translation
从单目图像创建逼真且可动画的三维人类头像仍然在很大程度上依赖于线性混合蒙皮（Linear Blend Skinning, LBS）和参数化身体模型，这限制了表现力，并且由于拟合不完美常常引入伪影。我们提出了LUNA，一个无LBS的通用神经动画模型，直接将多种二维控制（如图像、关键点、草图和未见角色）映射到三维高斯变形，绕过了显式的身体拟合。其核心是一个基于变换器的运动回归器，它将全局刚性运动与细粒度局部动态解耦，以捕捉一致的运动和微妙的非刚性效应。为了解决二维到三维提升的固有模糊性，同时超越拟合数据集的规模，我们引入了混合监督，从LBS教师中提炼软结构先验，并采用一种损失函数，支持在有限的拟合数据和大量未标记的野外视频上进行训练。大量实验表明，LUNA在视觉保真度上与基于LBS的方法具有竞争力，同时在多种驱动模式下提供逼真的人类运动和零-shot跨身份泛化。根据我们所知，LUNA是第一个支持隐式二维驱动的端到端三维可动画模型。
cs.CV / 137 / 2606.31982
ERA: Entropy-Guided Visual Token Pruning with Rectified Attention for Efficient MLLMs
ERA：基于熵引导的视觉标记剪枝与修正注意力的高效多模态大语言模型
Abstract
Multimodal Large Language Models (MLLMs) incur prohibitive inference costs due to long visual token sequences. Training-free visual token reduction provides an efficient solution. However, existing methods distort attention distributions, giving rise to a phenomenon we term Attention Logit Collapse. To address this issue, we propose ERA, an Entropy-guided visual token pruning framework with Rectified Attention for efficient MLLMs. Specifically, ERA comprises three crucial components: Dual-view Entropy Pruning (DEP), Bias-aware Token Recycling (BTR), and Logit-preserving Attention Rectification (LAR). First, DEP identifies representative anchor tokens by jointly modeling visual diversity and head-wise saliency. BTR then recycles pruned tokens into their corresponding anchors while estimating a cluster-level logit bias. Building upon this, LAR injects the estimated bias into attention logits, effectively rectifying the collapse induced by token reduction. Together, these components preserve visual evidence even under aggressive compression, enabling robust performance across single-image, multi-image, and video settings on a wide range of MLLMs. Beyond delivering practical acceleration, ERA establishes logit-preserving visual token pruning as a principled framework for efficient MLLMs, unifying theoretical foundation, algorithmic design, and practical deployment. The code is at https://github.com/924973292/ERA.
Chinese Translation
多模态大语言模型（MLLMs）由于长视觉标记序列而产生高昂的推理成本。无训练的视觉标记减少提供了一种高效的解决方案。然而，现有方法扭曲了注意力分布，导致我们称之为注意力逻辑崩溃（Attention Logit Collapse）的现象。为了解决这一问题，我们提出了ERA，一个基于熵引导的视觉标记剪枝框架，结合修正注意力以实现高效的MLLMs。具体而言，ERA包含三个关键组件：双视角熵剪枝（Dual-view Entropy Pruning, DEP）、偏差感知标记回收（Bias-aware Token Recycling, BTR）和逻辑保持注意力修正（Logit-preserving Attention Rectification, LAR）。首先，DEP通过联合建模视觉多样性和头部显著性来识别代表性锚标记。然后，BTR将剪枝后的标记回收到其对应的锚标记中，同时估计集群级别的逻辑偏差。在此基础上，LAR将估计的偏差注入注意力逻辑，有效修正了由标记减少引起的崩溃。这些组件共同作用，即使在激进压缩下也能保留视觉证据，从而在单图像、多图像和视频设置中实现广泛的MLLMs的稳健性能。除了提供实际加速外，ERA还将逻辑保持的视觉标记剪枝确立为高效MLLMs的原则框架，统一了理论基础、算法设计和实际部署。代码可在 https://github.com/924973292/ERA 获取。
cs.CV / 138 / 2606.31986
CoLT: Teaching Multi-Modal Models to Think with Chain of Latent Thoughts
CoLT：教会多模态模型通过潜在思维链进行思考
Abstract
Chain-of-thought (CoT) reasoning has enabled multi-modal large language models (MLLMs) to tackle complex visual reasoning tasks by generating explicit intermediate reasoning steps in natural language. However, this text-based reasoning paradigm is inherently slow at inference time with even thousands of tokens and fundamentally constrained by the expressiveness of natural language. In this paper, we propose CoLT, (Chain of Latent Thoughts), a novel framework that teaches multi-modal models to reason through a chain of latent thought representations instead of verbose text tokens, which can perform thinking with as few as 3 steps. Naively forcing the model to think with latent states easily produces meaningless semantics and makes training unstable. To effectively regulate the latent reasoning process, we introduce a lightweight external decoder that provides step-level supervision for each latent reasoning step in two complementary directions: a forward mode that decodes latent thoughts into the textual reasoning of the next step, and a backward mode that aligns decoder hidden states with the model's latent thoughts given preceding textual context. We further incorporate internal supervision that encourages coherent step-by-step latent transitions. The decoder and internal supervision are removed during inference to maintain high efficiency of latent reasoning. Extensive experiments on eight benchmarks demonstrate that CoLT not only outperforms existing latent reasoning methods such as CODI and SIM-CoT, but also surpasses latent visual reasoning approaches that rely on auxiliary images with costly annotation requirements. Compared to text CoT methods, CoLT can notably reduce the inference time by 10.1$\times$ and text decoding time by 22.6$\times$. Code is released at https://github.com/hulianyuyy/CoLT.
Chinese Translation
链式思维（Chain-of-thought, CoT）推理使得多模态大型语言模型（Multi-Modal Large Language Models, MLLMs）能够通过生成自然语言中的明确中间推理步骤来处理复杂的视觉推理任务。然而，这种基于文本的推理范式在推理时固有地缓慢，即使是数千个标记，并且在本质上受限于自然语言的表达能力。本文提出了CoLT（Chain of Latent Thoughts），一个新颖的框架，旨在教会多模态模型通过一系列潜在思维表示进行推理，而不是冗长的文本标记，这样可以在仅需3个步骤的情况下进行思考。简单地强迫模型使用潜在状态进行思考容易产生无意义的语义，并使训练不稳定。为了有效调节潜在推理过程，我们引入了一种轻量级的外部解码器，为每个潜在推理步骤提供逐步监督，分为两个互补的方向：前向模式将潜在思维解码为下一步的文本推理，后向模式则根据前面的文本上下文将解码器的隐藏状态与模型的潜在思维对齐。我们进一步结合内部监督，鼓励连贯的逐步潜在过渡。在推理过程中，解码器和内部监督被移除，以保持潜在推理的高效率。在八个基准上的广泛实验表明，CoLT不仅超越了现有的潜在推理方法，如CODI和SIM-CoT，还超过了依赖于昂贵注释要求的辅助图像的潜在视觉推理方法。与文本CoT方法相比，CoLT显著减少了推理时间10.1倍和文本解码时间22.6倍。代码已发布在 https://github.com/hulianyuyy/CoLT。
cs.CV / 139 / 2606.32018
Automated Background Swapping for Robustness against Spurious Backgrounds
针对虚假背景的鲁棒性自动背景切换
Abstract
Classifiers based on Deep Neural Networks exhibit strong performance across domains, yet can fail catastrophically if they rely on spurious correlations, i.e., features that are predictive of the target label in the training data but are not causally linked and thus fail to generalize. For the vision domain, many such spurious correlations manifest themselves within the background of the image, where only the foreground is predictive of the class label. In this paper, we introduce Automated Background Swapping (AutoBackSwap) to reduce the reliance of classifiers on such spurious backgrounds. AutoBackSwap uses a secondary network to disentangle the foreground and background, followed by infilling to synthesize complete backgrounds, and finally combines different foregrounds and inpainted backgrounds to augment the training data. We find that patch-wise labeling of just a few hundred samples suffices to train the secondary network and automatically augment the full training dataset on challenging image classification tasks. In contrast to many previous methods, AutoBackSwap proves very effective even if there is not a single sample in the training data breaking the spurious correlation. Across a range of image classification tasks with spurious backgrounds, AutoBackSwap consistently outperforms prior methods.
Chinese Translation
基于深度神经网络的分类器在多个领域表现出强大的性能，然而如果依赖于虚假相关性，即在训练数据中与目标标签具有预测性但并无因果联系的特征，它们可能会出现灾难性的失败。在视觉领域，许多此类虚假相关性表现为图像的背景，其中只有前景对类别标签具有预测性。本文介绍了一种自动背景切换方法（Automated Background Swapping，AutoBackSwap），旨在减少分类器对这些虚假背景的依赖。AutoBackSwap使用一个辅助网络来解耦前景和背景，随后进行填充以合成完整的背景，最后结合不同的前景和修复后的背景来增强训练数据。我们发现，仅需对几百个样本进行块级标注即可训练辅助网络，并在具有挑战性的图像分类任务中自动增强完整的训练数据集。与许多先前的方法相比，即使训练数据中没有单一样本打破虚假相关性，AutoBackSwap仍然表现出极高的有效性。在一系列具有虚假背景的图像分类任务中，AutoBackSwap始终优于以往的方法。
cs.CV / 140 / 2606.32020
Cross-Space Distillation: Teaching One-Step Students with Modern Diffusion Teachers
跨空间蒸馏：用现代扩散教师指导一步学生
Abstract
Modern one-step diffusion models achieve impressive quality through distribution-based timestep distillation. Yet, they rely on a critical assumption: Teacher and Student must inhabit the same latent space. This Shared-Space constraint prevents knowledge transfer from modern high-capacity Teachers (e.g., SD 3.5 and Flux) into compact, deployment-friendly Students such as SD 1.5, whose latent resolution and VAE parameterization differ from the Teacher. We formalize this overlooked regime as Cross-Space Distillation, where Teacher and Student differ in both latent resolution and VAE space. To enable distillation under this mismatch, we introduce the Bridge, a lightweight latent interface that maps Student latents into the Teacher space without modifying the Student backbone. Bridge combines a frozen Student VAE decoder as a spatial prior with a compact learnable projector, and is trained with latent reconstruction and attention fidelity objectives for stable Teacher-space alignment. Across diverse modern Teachers, Bridge enables substantial gains for compact one-step Students; for example, it improves SD 1.5 from 5.4 to 9.4 HPSv3 while preserving one-step inference, low latency, and broad ecosystem compatibility. These results show that heterogeneous large Teachers can be distilled into efficient, deployable backbones through a lightweight latent-space interface.
Chinese Translation
现代一步扩散模型通过基于分布的时间步蒸馏实现了令人印象深刻的质量。然而，它们依赖于一个关键假设：教师和学生必须处于相同的潜在空间。这一共享空间的限制阻碍了现代高容量教师（例如，SD 3.5 和 Flux）向紧凑、适合部署的学生（如 SD 1.5）转移知识，因为后者的潜在分辨率和变分自编码器（VAE）参数化与教师不同。我们将这一被忽视的范畴正式化为跨空间蒸馏，其中教师和学生在潜在分辨率和 VAE 空间上存在差异。为了在这种不匹配下实现蒸馏，我们引入了桥接（Bridge），这是一种轻量级的潜在接口，可以将学生的潜在表示映射到教师空间，而无需修改学生的主干网络。桥接结合了一个冻结的学生 VAE 解码器作为空间先验和一个紧凑的可学习投影器，并通过潜在重构和注意力保真度目标进行训练，以实现稳定的教师空间对齐。在多种现代教师中，桥接使紧凑的一步学生获得了显著的提升；例如，它将 SD 1.5 的 HPSv3 从 5.4 提升到 9.4，同时保持一步推理、低延迟和广泛的生态系统兼容性。这些结果表明，异构的大型教师可以通过轻量级的潜在空间接口蒸馏为高效、可部署的主干网络。
cs.CV / 141 / 2606.32023
FLORA: A deep learning approach to predict forest attributes from heterogeneous LiDAR data
FLORA：一种基于深度学习的异质LiDAR数据森林属性预测方法
Abstract
Forest attributes are essential for national-scale resource monitoring. Airborne LiDAR metrics are among the auxiliary variables most strongly correlated with forest attributes used in National Forest Inventory (NFI) estimates. However, producing wall-to-wall predictions remains challenging when LiDAR data are acquired under heterogeneous conditions. As national LiDAR programs expand across Europe, variability in sensors, flight parameters, seasons, and scan angles limits the robustness of existing models, which are often calibrated for local conditions. We present FLORA (Forest LiDAR Octree Regression with Auxiliary Data), a deep learning framework that predicts six forest attributes: dominant height, total volume, deciduous volume, coniferous volume, basal area, and stem density from heterogeneous LiDAR point clouds. FLORA combines an octree-based backbone with ecological and spatiotemporal auxiliary variables through a late-fusion gating mechanism. Models are trained and evaluated on 32,052 National Forest Inventory plots across mainland France using data from the French LiDAR HD program. A single model trained on both leaf-on and leaf-off acquisitions outperforms season-specific models and improves cross-season robustness. Auxiliary variables provide modest overall gains but contribute more strongly to species-specific volume prediction. FLORA achieves an rRMSE of about 12.3% (R2 = 0.88) for dominant height and 39% (R2 = 0.74) for total volume, providing a robust baseline for large-scale forest attribute estimation from heterogeneous national LiDAR programs.
Chinese Translation
森林属性对于国家级资源监测至关重要。空中LiDAR指标是与国家森林清查（NFI）估算中使用的森林属性最强相关的辅助变量之一。然而，当LiDAR数据在异质条件下获取时，进行全面预测仍然具有挑战性。随着国家LiDAR项目在欧洲的扩展，传感器、飞行参数、季节和扫描角度的变化限制了现有模型的稳健性，这些模型通常是针对局部条件进行校准的。我们提出了FLORA（森林LiDAR八叉树回归与辅助数据），这是一个深度学习框架，能够从异质LiDAR点云中预测六个森林属性：主高度、总量、阔叶树量、针叶树量、基面积和树干密度。FLORA通过晚融合门控机制，将基于八叉树的骨干网络与生态和时空辅助变量相结合。模型在法国本土的32,052个国家森林清查样地上进行训练和评估，使用的数据来自法国LiDAR HD项目。一个同时在叶子繁茂和叶子凋落的获取数据上训练的单一模型，其性能优于特定季节模型，并提高了跨季节的稳健性。辅助变量提供了适度的整体增益，但对特定物种的体积预测贡献更为显著。FLORA在主高度上的相对均方根误差（rRMSE）约为12.3%（R2 = 0.88），在总量上的相对均方根误差约为39%（R2 = 0.74），为从异质国家LiDAR项目中进行大规模森林属性估算提供了稳健的基准。
cs.CV / 142 / 2606.32033
SpheRoPE: Zero-Shot Optimization-Free 360 Panorama Generation with Spherical RoPE
SpheRoPE：基于球面 RoPE 的零-shot 无需优化的 360 全景生成
Abstract
We present a zero-shot, training-free and optimization-free framework for generating 360 panoramic images and videos by directly injecting spherical priors into pre-trained diffusion transformers. Existing methods either rely on costly fine-tuning on scarce panoramic data that limits generalization, or leverage multi-step optimization that incurs prohibitive inference latency. We observe that contemporary generative models natively exhibit some panoramic priors from large-scale training. However, these emergent capabilities are insufficient, as the models fundamentally fail to satisfy the rigorous topological constraints imposed by equirectangular projection (ERP). We introduce a zero-shot and optimization-free approach that resolves these constraints at inference time. Spherical RoPE replaces standard rotary position embeddings: low-frequency channels are re-parameterized as 3D Cartesian coordinates to natively encode the spherical manifold, while high-frequency channels are harmonically quantized to enforce exact periodicity. Coupled with complementary Semantic Distortion classifier-free guidance (CFG) that explicitly steers geometry, we avoid retraining and inherit the full creative breadth of state-of-the-art models. Our approach generalizes across diverse backbones and 360 generation modalities. We demonstrate this across text-to-panorama using Flux.1, Flux.2, and LTX-Video backbones, achieving competitive performance against baselines, all while remaining training-free. Project page: https://orhir.github.io/SpheRoPE
Chinese Translation
我们提出了一种零-shot、无需训练和优化的框架，通过直接将球面先验注入预训练的扩散变换器来生成 360 全景图像和视频。现有方法要么依赖于在稀缺全景数据上的高成本微调，限制了其泛化能力，要么利用多步优化，导致不可承受的推理延迟。我们观察到，当代生成模型在大规模训练中本质上展现出一些全景先验。然而，这些新兴能力仍然不足，因为模型在根本上无法满足等距投影（ERP）施加的严格拓扑约束。我们引入了一种零-shot 和无优化的方法，在推理时解决这些约束。球面 RoPE 替代了标准的旋转位置嵌入：低频通道被重新参数化为 3D 笛卡尔坐标，以原生编码球面流形，而高频通道则被谐波量化以强制执行精确的周期性。结合互补的语义失真无分类器引导（CFG），明确引导几何形状，我们避免了重新训练，并继承了最先进模型的全部创造性广度。我们的方法在多种基础模型和 360 生成模式中具有良好的泛化能力。我们在使用 Flux.1、Flux.2 和 LTX-Video 基础模型的文本到全景生成中展示了这一点，取得了与基线相当的性能，同时保持了无训练的特点。项目页面：https://orhir.github.io/SpheRoPE
cs.CV / 143 / 2606.32036
PointSplat: Compact Gaussian Splatting via Human-Centric Prediction
PointSplat：通过以人为中心的预测实现紧凑的高斯点云渲染
Abstract
Producing 3D human representations from input views on the fly is essential for immersive live streaming systems, where representation compactness is as critical as high fidelity given limited computational power and transmission bandwidth. Although recent feed-forward reconstruction methods achieve impressive quality through the view-centric prediction of 3D representations, they repeatedly encode the same subject content across multiple views, leading to significant inter-view redundancy. Our key insight is to perform predictions directly in 3D space, enabling the network to learn and produce a highly compact representation. To this end, we propose PointSplat, a novel human-centric approach that directly infers Gaussian primitives from an input point set. The proposed method first estimates a coarse geometric proxy and performs ray casting to prune redundant points and establish explicit 2D--3D correspondences. Subsequently, it employs a Point-Image Transformer to fuse appearance and geometry features, predicting Gaussian attributes in a single forward pass. This design restricts predictions to foreground regions of interest, substantially reducing the total number of Gaussians while improving novel-view rendering quality. Extensive experiments demonstrate that PointSplat achieves higher efficiency and quality while exhibiting strong robustness to variations in view count and image resolution across multiple datasets.
Chinese Translation
从输入视图实时生成3D人类表示对于沉浸式直播系统至关重要，其中表示的紧凑性与高保真度同样重要，尤其在计算能力和传输带宽有限的情况下。尽管最近的前馈重建方法通过以视图为中心的3D表示预测取得了令人印象深刻的质量，但它们在多个视图中重复编码相同的主体内容，导致显著的视图间冗余。我们的关键见解是直接在3D空间中进行预测，使网络能够学习并生成高度紧凑的表示。为此，我们提出了PointSplat，一种新颖的以人为中心的方法，直接从输入点集推断高斯原语。该方法首先估计粗略的几何代理，并通过光线投射来修剪冗余点并建立明确的2D-3D对应关系。随后，它采用Point-Image Transformer融合外观和几何特征，在单次前向传递中预测高斯属性。该设计将预测限制在感兴趣的前景区域，显著减少了高斯的总数量，同时提高了新视图渲染的质量。大量实验表明，PointSplat在多个数据集上实现了更高的效率和质量，同时对视图数量和图像分辨率的变化表现出强大的鲁棒性。
cs.CV / 144 / 2606.32039
GEAR: Guided End-to-End AutoRegression for Image Synthesis
GEAR：引导式端到端自回归图像合成
Abstract
Visual generative models are typically trained in two stages. A tokenizer is first trained for reconstruction and then frozen, after which a generator is trained on its discrete indices or continuous latents. This decoupling leaves the tokenizer unaware of what the generator finds easy to model. We present GEAR (Guided End-to-end AutoRegression), which trains a vector-quantized (VQ) tokenizer and an autoregressive (AR) generator jointly and end-to-end, guided by representation alignment. The key obstacle is that the VQ index fed to the AR model is non-differentiable, so gradients cannot reach the tokenizer, and a straight-through estimator collapses. GEAR resolves this with a dual read-out of the codebook assignment. A hard, one-hot branch trains the AR with next-token prediction, while a differentiable soft branch carries a representation-alignment loss that flows back to guide only the tokenizer. The AR model thereby steers its tokenizer toward an index distribution it can predict more easily. This shifts the alignment burden from the tokenizer to the AR: the tokenizer's own features become less DINOv2-like while the AR's become more so, the opposite of diffusion-side recipes that make the latent itself semantic. GEAR speeds up ImageNet gFID convergence by up to 10x relative to the strong LlamaGen-REPA baseline, learns markedly better patch-level and spatially-coherent features, and generalizes across quantizers (VQVAE, LFQ, IBQ) and to text-to-image generation.
Chinese Translation
视觉生成模型通常分为两个阶段进行训练。首先训练一个用于重构的标记器，然后将其冻结，接着在其离散索引或连续潜变量上训练生成器。这种解耦使得标记器无法了解生成器易于建模的内容。我们提出了GEAR（引导式端到端自回归），该方法联合并端到端地训练一个向量量化（VQ）标记器和一个自回归（AR）生成器，并通过表示对齐进行引导。关键的障碍在于，输入AR模型的VQ索引是不可微分的，因此梯度无法到达标记器，而直接通过估计器会崩溃。GEAR通过对代码本分配的双重读取来解决此问题。一个硬的独热分支通过下一个标记预测训练AR，而一个可微分的软分支则携带一个表示对齐损失，反向流动以仅引导标记器。因此，AR模型能够引导其标记器朝向一个更易于预测的索引分布。这将对齐的负担从标记器转移到AR：标记器自身的特征变得不那么类似于DINOv2，而AR的特征则变得更加类似，这与使潜变量本身具有语义的扩散侧配方相反。与强大的LlamaGen-REPA基线相比，GEAR加速了ImageNet gFID收敛速度，提升了多达10倍，学习到了显著更好的块级和空间一致性特征，并在量化器（VQVAE、LFQ、IBQ）之间以及文本到图像生成中实现了良好的泛化。
cs.CV / 145 / 2606.32040
FaceMoE: Mixture of Experts for Low-Resolution Face Recognition
FaceMoE：用于低分辨率人脸识别的专家混合模型
Abstract
Low-resolution face recognition (LR-FR) remains a challenging task due to poor feature extraction and aggregation, as probe images often contain limited identity information resulting from extreme degradations such as blur, occlusion, and low contrast. Additionally, the domain gap between high-resolution (HR) gallery images and low-resolution (LR) probe images poses a significant challenge. A single feature encoder struggles to generalize effectively across both domains when fine-tuned on an LR dataset, and this issue is further magnified by catastrophic forgetting. To address these challenges, we propose FaceMoE, an effective adaptation of Mixture of Experts (MoE) transfomer architecture for low-resolution face-recognition . Specifically, we introduce multiple specialized feed-forward network (FFN) experts and incorporate a top-k router, which dynamically assigns tokens to appropriate experts. This design emergently promotes specialization across experts for different semantic regions of the face, which enables FaceMoE to perform resolution-aware feature extraction. Moreover, the top-k router facilitates sparse expert activation, enabling the model to preserve pretrained knowledge when finetuned on a LR dataset, while increasing model capacity without proportional computational overhead. FaceMoE is trained with a combined face recognition loss, router z-loss, and load balancing loss to ensure expert specialization and stable training. To the best of our knowledge, this is the first work leveraging MoE for LR-FR. Extensive experiments across eleven datasets, spanning HR, mixed-quality, and LR benchmarks, demonstrate that FaceMoE significantly outperforms state-of-the-art methods. Code: https://github.com/Kartik-3004/FaceMoE
Chinese Translation
低分辨率人脸识别（LR-FR）仍然是一项具有挑战性的任务，因为特征提取和聚合效果较差，探测图像通常由于模糊、遮挡和低对比度等极端降级而包含有限的身份信息。此外，高分辨率（HR）图库图像与低分辨率（LR）探测图像之间的领域差距也构成了显著挑战。当在LR数据集上进行微调时，单一特征编码器难以在两个领域中有效泛化，而这一问题在灾难性遗忘的影响下进一步加剧。为了解决这些挑战，我们提出了FaceMoE，这是一种有效的专家混合（Mixture of Experts, MoE）变换器架构的适应，用于低分辨率人脸识别。具体而言，我们引入了多个专门的前馈网络（Feed-Forward Network, FFN）专家，并结合了一个top-k路由器，该路由器动态地将令牌分配给适当的专家。这种设计促使专家在面部不同语义区域之间的专业化，从而使FaceMoE能够进行分辨率感知的特征提取。此外，top-k路由器促进了稀疏专家激活，使模型在LR数据集上微调时能够保留预训练知识，同时在不成比例增加计算开销的情况下提高模型容量。FaceMoE通过结合人脸识别损失、路由器z损失和负载平衡损失进行训练，以确保专家专业化和稳定训练。根据我们所知，这是首次利用MoE进行LR-FR的研究。在跨越HR、混合质量和LR基准的十一种数据集上进行的广泛实验表明，FaceMoE显著优于最先进的方法。代码链接：https://github.com/Kartik-3004/FaceMoE
cs.AI / 1 / 2606.30774
What Drives Interactive Improvement from Feedback?
是什么驱动了反馈的互动改善？
Abstract
We study when natural-language feedback produces improvement beyond the gains obtainable from repeated attempts alone. In multi-turn language agent setting, higher final accuracy can reflect useful feedback, but it can also arise from resampling, format correction, or additional test-time computation. To separate these effects, we introduce a controlled student-teacher protocol across Omni-MATH, Codeforces, BBEH Linguini, and ARC-AGI1, evaluating thirteen open-weight models in both student and teacher roles. We compare external feedback, self-feedback, and unguided self-refinement, while varying interaction history, task difficulty, and teacher access to privileged task information. Across settings, we find that multi-turn improvement is often not evidence of feedback use: self-generated feedback adds little beyond unguided self-refinement, whereas the strongest external teachers produce substantially larger feedback-specific gains, suggesting that useful feedback must provide guidance beyond generic retry. Dense student-teacher interaction matrices further show that interactive gains are driven more by the student's ability to use feedback than by the teacher's identity, although teacher choice remains important for a fixed student. These results suggest that feedback-based agents should be evaluated against repeated-attempt baselines, and that ability to act on feedback, not merely feedback availability, is a central bottleneck for interactive improvement. We release our controlled student-teacher evaluation framework at https://j-lojek.github.io/feedback-generation-is-a-bottleneck/.
Chinese Translation
我们研究了自然语言反馈在何种情况下能产生超出单纯重复尝试所能获得的改进。在多轮语言代理设置中，较高的最终准确率可能反映出有用的反馈，但也可能来源于重采样、格式修正或额外的测试时计算。为了区分这些影响，我们在 Omni-MATH、Codeforces、BBEH Linguini 和 ARC-AGI1 上引入了一种受控的学生-教师协议，评估了十三个开放权重模型在学生和教师角色中的表现。我们比较了外部反馈、自我反馈和无指导自我精炼，同时变化交互历史、任务难度和教师对特权任务信息的访问。在各种设置中，我们发现多轮改进通常不是反馈使用的证据：自生成的反馈在无指导自我精炼之外几乎没有增加，而最强的外部教师则产生了显著更大的反馈特定增益，这表明有用的反馈必须提供超越一般重试的指导。密集的学生-教师交互矩阵进一步表明，互动增益更多地受到学生使用反馈能力的驱动，而非教师身份的影响，尽管教师选择对于固定学生仍然重要。这些结果表明，基于反馈的代理应与重复尝试基线进行评估，并且能够有效利用反馈，而不仅仅是反馈的可用性，是互动改善的核心瓶颈。我们在 https://j-lojek.github.io/feedback-generation-is-a-bottleneck/ 发布了我们的受控学生-教师评估框架。
cs.AI / 2 / 2606.30840
Contrastive Reflection for Iterative Prompt Optimization
对比反思用于迭代提示优化
Abstract
LLM agents are becoming central to information retrieval: they issue retrieval queries, synthesize answers, and increasingly serve as judges for IR evaluation. Improving the prompts that control these agents is an optimization problem, but in applied IR settings it often looks less like blind search and more like debugging. Engineers need to know which behavior failed, which nearby behavior still worked, what distinguishes the two, and whether a prompt edit improves held-out quality without introducing regressions. We present Contrastive Reflection, an iterative prompt-optimization framework for agentic IR workflows. The framework starts from a task-centric quality definition: QA agents expose retrieval or reasoning traces, and grading agents expose dimension-level scores and rationales. These structured traces are used to identify error-anchored behavioral slices, add nearby successful examples from the same region, and ask a Teacher LLM to propose a targeted prompt edit. Candidate edits are accepted only when validation performance improves, optionally subject to regression checks. We instantiate the framework with a tree-based slice selector, but the contribution is the contrastive reflection loop rather than the tree itself. On a public HotpotQA retrieval-augmented QA setup, one tree-selected contrastive repair improves held-out exact-match accuracy from 51.4% to 60.4%. Failure-only and random-evidence variants improve less and break more previously correct examples. A light instruction-only comparison places the method near modern prompt optimizers: MIPROv2 reaches 59.4% and GEPA 57.0%. The result is an interpretable optimization loop for IR agents, aimed at making prompt repair more inspectable and validation-driven.
Chinese Translation
大型语言模型（LLM）代理在信息检索中变得越来越重要：它们发出检索查询，合成答案，并日益作为信息检索评估的评判者。改善控制这些代理的提示是一个优化问题，但在应用的信息检索环境中，它往往看起来更像是调试而非盲目搜索。工程师需要知道哪个行为失败了，哪个相近的行为仍然有效，二者之间的区别是什么，以及提示编辑是否在不引入回归的情况下提高了保留质量。我们提出了对比反思（Contrastive Reflection），这是一个用于代理信息检索工作流的迭代提示优化框架。该框架从以任务为中心的质量定义开始：问答（QA）代理暴露检索或推理痕迹，而评分代理暴露维度级别的分数和理由。这些结构化的痕迹用于识别以错误为锚的行为切片，添加来自同一区域的相近成功示例，并请求教师LLM（Teacher LLM）提出有针对性的提示编辑。候选编辑仅在验证性能改善时被接受，并可选择进行回归检查。我们用基于树的切片选择器实例化该框架，但其贡献在于对比反思循环，而非树本身。在公共的HotpotQA检索增强问答设置中，一个树选择的对比修复将保留的精确匹配准确率从51.4%提高到60.4%。仅失败和随机证据的变体改进较少，并且破坏了更多之前正确的示例。轻量级的仅指令比较将该方法置于现代提示优化器附近：MIPROv2达到59.4%，GEPA达到57.0%。最终结果是一个可解释的信息检索代理优化循环，旨在使提示修复更具可检查性和验证驱动性。
cs.AI / 3 / 2606.30846
How Can AI Find My Model? A Model-Finding Experimental Study Considering Data Formats, Embeddings, and Retrieval Strategies
人工智能如何找到我的模型？考虑数据格式、嵌入和检索策略的模型发现实验研究
Abstract
Discovering simulation models for reuse remains a fundamental challenge in Modeling and Simulation (M&S). When many models coexist, identifying those that align with a given modeling intent remains difficult. Recent advances in Artificial Intelligence (AI), particularly retrieval-based approaches, offer a promising pathway to operate at this semantic layer. In this paper, we present an experimental study investigating the impact of data representation, transformer-based embedding models, and retrieval strategies on the discovery of simulation models using natural language queries. We evaluated performance across multiple query types using standard information retrieval metrics, including recall@5 and nDCG@5. Results show that data representation matters, open-source embedding models can achieve high performance, and reranking methods are important, especially as query complexity increases. This work provides a baseline for AI-driven model discovery and discusses its role in advancing toward AI-driven composability and interoperability.
Chinese Translation
发现可重用的仿真模型仍然是建模与仿真（Modeling and Simulation, M&S）中的一个基本挑战。当许多模型共存时，识别与特定建模意图一致的模型仍然困难。近年来，人工智能（Artificial Intelligence, AI）特别是基于检索的方法的进展，为在这一语义层面上操作提供了有希望的途径。本文呈现了一项实验研究，探讨数据表示、基于变换器的嵌入模型和检索策略对使用自然语言查询发现仿真模型的影响。我们使用标准信息检索指标（包括 recall@5 和 nDCG@5）评估了多种查询类型的性能。结果表明，数据表示很重要，开源嵌入模型可以实现高性能，而重新排序方法尤其重要，特别是在查询复杂性增加时。这项工作为基于人工智能的模型发现提供了基准，并讨论了其在推动人工智能驱动的可组合性和互操作性方面的作用。
cs.AI / 4 / 2606.30850
BayesBench: Evaluating LLM Belief Trajectories Under Multi-Turn Evidence Accumulation
BayesBench：在多轮证据积累下评估大型语言模型的信念轨迹
Abstract
Large language models (LLMs) are typically deployed in multi-turn conversations, where each turn provides new evidence that should reduce epistemic uncertainty about their environment. Acting rationally then requires inferring the unobserved quantities that govern it and updating beliefs about them as evidence accumulates. Yet most evaluations only score the model's final-turn answer in a single-turn format, leaving this process unexamined. We ask how closely LLMs' belief updates match those of a rational Bayesian reasoner in multi-turn settings, and introduce BayesBench, a suite of simulation environments that probe this across three progressively complex tasks: (i) Bayesian estimation, where the model infers an unknown parameter from sequential evidence; (ii) Bayesian prediction, where the model turns inferred beliefs about a latent variable into outcome forecasts; and (iii) latent-framed Bayesian prediction, where observations are filtered through a user-persona framing, requiring joint inference over the latent state and the persona. Across seven LLMs (3B--70B), scaling improves latent inference and evidence accumulation, with updates occasionally matching the Bayesian posterior. However, these gains do not reliably carry over to downstream prediction, exposing a gap between inferring latent structure and using it to rationally update beliefs about the target outcome.
Chinese Translation
大型语言模型（LLMs）通常在多轮对话中部署，每一轮都提供新的证据，应该减少对其环境的认知不确定性。因此，理性行为要求推断支配这一过程的未观察量，并在证据积累时更新对它们的信念。然而，大多数评估仅在单轮格式中评分模型的最终回答，未对这一过程进行检验。我们探讨LLMs的信念更新与理性贝叶斯推理者在多轮环境中的匹配程度，并引入BayesBench，一个模拟环境套件，通过三个逐渐复杂的任务进行探测：（i）贝叶斯估计，模型从顺序证据中推断未知参数；（ii）贝叶斯预测，模型将对潜变量的推断信念转化为结果预测；（iii）潜框架贝叶斯预测，观察通过用户角色框架进行过滤，需要对潜在状态和角色进行联合推断。在七个LLMs（3B–70B）中，规模的扩大改善了潜在推断和证据积累，更新有时与贝叶斯后验相匹配。然而，这些收益并未可靠地转移到下游预测中，暴露出推断潜在结构与理性更新对目标结果信念之间的差距。
cs.AI / 5 / 2606.30852
When Does Learning to Stop Help? A Cost-Aware Study of Early Exits in Reasoning Models
何时学习停止有助于决策？一种关注成本的推理模型早期退出研究
Abstract
Reasoning models spend different amounts of useful computation across instances, but it remains unclear when a learned stopping rule improves over simple confidence or convergence thresholds. We study this question with LearnStop, a hidden-state-free checkpoint stopper for reasoning language models. At fixed budget checkpoints, LearnStop probes a short answer from the current reasoning prefix and predicts prefix correctness from online features such as answer confidence, entropy, prefix vote share, answer stability, and backtracking-marker density. Across 18 task-model settings spanning GSM8K, MATH-500, MMLU-Pro, AIME-90, GPQA, Qwen3, and DeepSeek-R1 distillations, the answer is task-dependent. On free-form math, learned multi-feature stopping improves the fixed-budget frontier and often beats scalar exits: on GSM8K with Qwen3-32B, the empirical frontier reaches a post-hoc peak adapt gain of +0.157, validation-selected operating points preserve positive gains, and the paired gain over the strongest scalar baseline is +0.028. On multiple-choice and very hard settings, scalar confidence, entropy, or stability rules are competitive or stronger. We therefore frame learned stopping not as a universal replacement for scalar exits, but as a tool whose value depends on trajectory structure. We further provide validation-selected operating points, paired bootstrap tests, finite-grid lost-correct risk calibration, cost accounting under KV-fork, prefix-cache, and black-box regimes, H100 serving profiles, checkpoint-schedule sweeps, transfer analyses, and robustness checks. The main practical finding is that learned stopping is useful when many questions become correct before full budget but do not exhibit a single reliable scalar stopping signal; its benefits largely disappear when confidence or answer convergence already solves the stopping problem.
Chinese Translation
推理模型在不同实例中花费的有效计算量各不相同，但尚不清楚何时学习到的停止规则优于简单的置信度或收敛阈值。我们通过 LearnStop 这一无隐藏状态的检查点停止器来研究这个问题，该模型适用于推理语言模型。在固定预算检查点上，LearnStop 从当前推理前缀中探测一个简短答案，并根据在线特征（如答案置信度、熵、前缀投票份额、答案稳定性和回溯标记密度）预测前缀的正确性。在涵盖 GSM8K、MATH-500、MMLU-Pro、AIME-90、GPQA、Qwen3 和 DeepSeek-R1 蒸馏的 18 个任务模型设置中，答案依赖于具体任务。在自由形式的数学问题中，学习到的多特征停止方法改善了固定预算的边界，并且通常优于标量退出：在 GSM8K 与 Qwen3-32B 的结合下，经验边界达到了事后峰值适应增益 +0.157，验证选择的操作点保持了正增益，而与最强标量基线的配对增益为 +0.028。在多项选择和非常困难的设置中，标量置信度、熵或稳定性规则具有竞争力或更强。因此，我们将学习到的停止框架视为一种工具，其价值依赖于轨迹结构，而非标量退出的普遍替代品。我们进一步提供了验证选择的操作点、配对自助测试、有限网格的损失-正确风险校准、KV-fork 下的成本核算、前缀缓存和黑箱机制、H100 服务配置、检查点调度扫描、迁移分析和鲁棒性检查。主要的实践发现是，当许多问题在完全预算之前变得正确，但没有表现出单一可靠的标量停止信号时，学习到的停止是有用的；当置信度或答案收敛已经解决停止问题时，其益处大幅减少。
cs.AI / 6 / 2606.30863
Beyond expert users: agents should help users construct preferences, not just elicit them
超越专家用户：智能体应帮助用户构建偏好，而不仅仅是引导其表达偏好
Abstract
Agents typically assume an expert user -- one with well-formed preferences about what they want -- and default to clarifying questions whenever the task is underspecified. We argue this assumption is unrealistic. Users often lack the domain knowledge to have completely specified preferences; if asked about their preference on some feature, the user may be unable to answer without the agent helping the user to learn some domain knowledge needed to form a preference for that feature, e.g., via examples or explanations. To formalize these principles, we draw on the Search-Experience-Credence framework from Information Economics to introduce CoPref, a model of how users construct preferences based on agent dialog actions. We then study these ideas concretely in agentic recommender systems, proposing CoShop, an interactive benchmark. In CoShop, an agent converses with and makes recommendations for a CoPref user. The agent's performance depends on whether it can help the user gain the knowledge needed to specify the task well. Evaluating five frontier models, we find that no agent exceeds 56% accuracy on CoShop despite five turns of interaction. Failures stem not from agents' ability to find items, but from how little the interaction expands what users know about what they want.
Chinese Translation
智能体通常假设用户是专家用户——即对其需求有明确偏好的用户，并在任务不明确时默认提出澄清性问题。我们认为这一假设不切实际。用户往往缺乏领域知识，无法完全指定偏好；如果询问用户对某个特征的偏好，用户可能无法回答，因为需要智能体帮助用户学习形成该特征偏好所需的领域知识，例如通过示例或解释。为了形式化这些原则，我们借鉴信息经济学中的搜索-体验-信任框架，提出了CoPref模型，描述用户如何基于智能体对话行为构建偏好。接着，我们在智能推荐系统中具体研究这些理念，提出了CoShop，一个互动基准。在CoShop中，智能体与CoPref用户进行对话并提供推荐。智能体的表现取决于其是否能够帮助用户获得明确任务所需的知识。评估五个前沿模型后，我们发现尽管进行了五轮互动，没有任何智能体在CoShop上的准确率超过56%。失败的原因并不在于智能体寻找物品的能力，而在于互动对用户了解其需求的扩展程度有限。
cs.AI / 7 / 2606.30906
Investigating Multi-Agent Deliberation in Law
法律领域中的多智能体审议研究
Abstract
Artificial Intelligence is increasingly applied to the field of law, and has the potential to increase access to justice. One particular movement that is gaining traction is that of agentic AI, wherein AI agents, based on Large Language Models (LLMs) can take autonomous actions. In particular, multi-agent approaches in the legal domain remain largely unexplored. In this paper, we investigate multi-agent deliberation methods for legal reasoning tasks using LLMs. We explore multi-agent deliberation (MAD) and introduce two novel multi-agent frameworks inspired by courtroom procedures and legal argumentation. Our experiments on both legal and non-legal benchmarks reveal that multi-agent frameworks achieve comparable overall performance to baseline large language models, but produce significantly distinct answers. Notably, these approaches can successfully solve cases that the baseline fails to address, and vice versa. We conduct a qualitative evaluation and highlight scenarios where multi-agent frameworks outperform monolithic approaches. For example, multi-agent approaches appear better suited for answering questions that require critical thinking from multiple perspectives. Our work positions multi-agent systems as a promising direction for AI in the legal domain, while demonstrating the potential of law-inspired multi-agent approaches for deliberation.
Chinese Translation
人工智能在法律领域的应用日益增多，具有提高司法可及性的潜力。一个正在获得关注的特别运动是代理智能（agentic AI），其中基于大型语言模型（Large Language Models, LLMs）的人工智能代理可以采取自主行动。特别是在法律领域，多智能体方法仍然基本未被探索。本文研究了使用LLMs进行法律推理任务的多智能体审议方法。我们探讨了多智能体审议（Multi-Agent Deliberation, MAD），并引入了两个受法庭程序和法律论证启发的新型多智能体框架。我们在法律和非法律基准上的实验表明，多智能体框架在整体性能上与基线大型语言模型相当，但产生的答案显著不同。值得注意的是，这些方法能够成功解决基线无法处理的案件，反之亦然。我们进行了定性评估，并强调了多智能体框架优于单一方法的情境。例如，多智能体方法似乎更适合回答需要从多个角度进行批判性思考的问题。我们的工作将多智能体系统定位为法律领域人工智能的一个有前景的方向，同时展示了受法律启发的多智能体审议方法的潜力。
cs.AI / 8 / 2606.30911
Why Solve It Twice? Hierarchical Accumulation of Skills for Transfer-Efficient ML Engineering
为什么要解决两次？面向高效转移的机器学习工程技能的层次积累
Abstract
ML engineering agents waste compute rediscovering known techniques because every competition is a cold start. We present HASTE, a hierarchical multi-agent system that organizes cross-competition knowledge into three scope tiers (global, domain, and competition-specific), each coupled to a matching agent level. An orchestrator coordinates domain specialists and promotes learning between tiers via LLM-driven abstraction. A controlled ablation provides evidence for scoped loading: holding a 159-skill inventory constant across 8 competitions, tiered loading achieves a 100% medal rate while flat loading reaches only 62.5%, the same medal rate as loading no skills, and consumes 2x the output tokens. On the full MLE-Bench Lite benchmark (22 Kaggle competitions), HASTE reaches a medal rate of 77.3% using Claude Sonnet 4.6 at 12h per competition. In a cold-start run, the system begins with no accumulated skills. In warm-start runs, it reloads skills learned from earlier competitions, using only global and domain-level skills for transfer across competitions. Warm starts use 52% fewer refinement iterations, and the fraction of proposed changes kept by the agent rises from 42% at low inventory to 85% once 50+ skills are available. These results suggest that better knowledge organization can partly substitute for model strength and compute budget in ML-engineering agents.
Chinese Translation
机器学习工程代理在每次比赛中都面临冷启动，因此浪费计算资源重新发现已知技术。我们提出了HASTE，一个层次化的多代理系统，将跨比赛知识组织为三个范围层次（全球、领域和特定比赛），每个层次与相应的代理级别相匹配。一个协调者负责协调领域专家，并通过基于大型语言模型（LLM）的抽象促进层次之间的学习。受控消融实验提供了范围加载的证据：在8场比赛中保持159项技能库存不变，层次加载实现了100%的获奖率，而平面加载仅达到62.5%，与不加载任何技能的获奖率相同，并且消耗的输出令牌是其2倍。在完整的MLE-Bench Lite基准（22场Kaggle比赛）中，HASTE在每场比赛12小时内使用Claude Sonnet 4.6达到了77.3%的获奖率。在冷启动运行中，系统开始时没有积累的技能。在温启动运行中，它重新加载从早期比赛中学到的技能，仅使用全球和领域级技能进行跨比赛的转移。温启动使用的精炼迭代减少了52%，而代理保留的提议更改的比例从低库存时的42%上升到50+技能可用时的85%。这些结果表明，更好的知识组织可以在一定程度上替代机器学习工程代理中的模型强度和计算预算。
cs.AI / 9 / 2606.30931
RoPoLL: Robust Panel of LLM Judges
RoPoLL：鲁棒性大型语言模型评审小组
Abstract
The LLM Jury, a Panel of LLM Evaluators (PoLL) reporting consensus scores, has become a practical alternative to single-judge LLM evaluation, yet its statistical behavior remains poorly understood. We formalize the LLM Jury under the Huber contamination model and show that PoLL incurs unbounded bias under any positive contamination, regardless of jury size, whenever a single judge fails in a biased, LLM-typical way (mode collapse, sycophancy, safety refusal). Framing jury consensus as classical robust mean estimation, we propose RoPoLL (Robust Panel of LLM-as-Judge), which preserves the PoLL panel but replaces the aggregation function with a robust mean estimator, instantiated with the geometric median (GM): tuning-free, with the optimal finite-sample breakdown point 1/2. A finite-sample error bound and a matching information-theoretic minimax lower bound agree on the parametric rate sigma*sqrt(d/N) and differ on the breakdown floor by a factor of sqrt(d), a statistical-computational gap that polynomial-time RoPoLL pays relative to the intractable Tukey halfspace median. Across 13 open-weight judges (4B-675B), three reward-model benchmarks, and four corruption regimes at rates up to 50%, RoPoLL dominates PoLL on every biased corruption type: by about 19% on cross-dimensional attacks at matched compute, and by orders of magnitude on heavy-tailed Byzantine adversaries. A 3-judge RoPoLL committee at 38B beats Mistral-Large-3 (675B) by 1.31x on HelpSteer-2 under 30% bimodal-random corruption, an 18x parameter advantage at better accuracy; a Noisy-GT control confirms the premium is paid against biased contamination, not benign imprecision.
Chinese Translation
大型语言模型评审团（LLM Jury），即大型语言模型评估者小组（Panel of LLM Evaluators, PoLL）报告共识评分，已成为单一评审者大型语言模型评估的实用替代方案，然而其统计行为仍然不甚明了。我们在Huber污染模型下对LLM Jury进行了形式化，并表明当单一评审者以偏见的、典型的LLM方式（模式崩溃、阿谀奉承、安全拒绝）失败时，无论评审团的规模如何，PoLL在任何正污染下都会产生无界偏差。将评审团共识框架视为经典的鲁棒均值估计，我们提出了RoPoLL（鲁棒性大型语言模型评审小组），该方法保留了PoLL小组，但将聚合函数替换为鲁棒均值估计器，具体实现为几何中位数（Geometric Median, GM）：无需调优，具有最佳有限样本崩溃点1/2。有限样本误差界限与匹配的信息论极小值下界在参数速率sigma*sqrt(d/N)上达成一致，但在崩溃底线的差异上存在sqrt(d)的因子，这是RoPoLL相对于难以处理的Tukey半空间中位数所支付的统计-计算差距。在13个开放权重评审者（4B-675B）、三个奖励模型基准和四种高达50%的腐败机制下，RoPoLL在每种偏见腐败类型上均优于PoLL：在匹配计算条件下，跨维攻击的表现提高约19%，而在重尾拜占庭对手的情况下则提高了几个数量级。一个由3名评审者组成的RoPoLL委员会在38B的情况下，在30%的双峰随机腐败下以1.31倍的优势击败了Mistral-Large-3（675B）在HelpSteer-2上的表现，且在更高准确度下具有18倍的参数优势；一个Noisy-GT控制实验确认这一优势是针对偏见污染，而非良性不精确性所付出的代价。
cs.AI / 10 / 2606.30949
AgRefactor: Self-Evolving Agentic Workflow for HLS Compatibility and Performance
AgRefactor：用于高层次综合兼容性和性能的自我演化代理工作流
Abstract
High-Level Synthesis (HLS) provides a fast path from concepts to silicon, but converting real-world software into synthesizable HLS code remains challenging due to restrictive language support and the gap between software and hardware programming practices. Existing automated and LLM-based refactoring approaches partially address this problem, yet they often lack flexibility, struggle to scale, and incur high computational costs. We introduce AgRefactor, an LLM-based multi-agent workflow for refactoring software into HLS-compatible programs. AgRefactor incorporates a self-evolving memory system that accumulates and retrieves factual and strategic knowledge across tasks, improving robustness and efficiency on unseen programs. To reduce cost and enhance scalability, it integrates automated refactoring tools, enabling agents to balance LLM-driven rewrites with efficient tool-based transformations. On 9 out of 11 challenging real-world benchmarks, which are 5-10x longer than the most complex cases studied in prior work, AgRefactor outperforms or matches the state-of-the-art automated refactoring tool and a strong LLM-based baseline built on the same framework backbone. Further agentic performance optimization yields a 6.51x geometric mean speedup over the SoTA pragma tuning tool and a 1.20x speedup over optimized open-source designs with less than 20% extra resources. AgRefactor is fully-automated and open-sourced.
Chinese Translation
高层次综合（HLS）为从概念到硅片提供了一条快速通道，但将现实世界的软件转换为可综合的HLS代码仍然面临挑战，这主要是由于语言支持的限制以及软件与硬件编程实践之间的差距。现有的自动化和基于大型语言模型（LLM）的重构方法部分解决了这一问题，但它们通常缺乏灵活性，难以扩展，并且计算成本高。我们提出了AgRefactor，这是一种基于LLM的多代理工作流，用于将软件重构为HLS兼容的程序。AgRefactor结合了一个自我演化的记忆系统，该系统在任务之间积累和检索事实和战略知识，从而提高了对未见程序的鲁棒性和效率。为了降低成本并增强可扩展性，它集成了自动化重构工具，使代理能够在LLM驱动的重写与高效的工具基础转换之间取得平衡。在11个具有挑战性的现实世界基准中，有9个基准的复杂度是之前研究中最复杂案例的5-10倍，AgRefactor的性能超越或匹配了最先进的自动化重构工具和基于相同框架的强大LLM基线。进一步的代理性能优化使得AgRefactor在SoTA（状态最先进）参数调优工具上实现了6.51倍的几何平均加速，并在优化的开源设计上实现了1.20倍的加速，同时额外资源消耗不足20%。AgRefactor是完全自动化且开源的。
cs.AI / 11 / 2606.30953
Neuro-Bayesian-Symbolic Residual Attention Shallow Network: Explainable Deep Learning for Cybersecurity Risk Assessment
神经-贝叶斯-符号残差注意浅层网络：用于网络安全风险评估的可解释深度学习
Abstract
We introduce the Neuro-Bayesian-Symbolic Residual Attention Shallow Network (NBS-RASN), a hybrid neural architecture for explainable cybersecurity risk assessment in open-source ecosystems. Unlike deep models that trade interpretability for accuracy, our shallow network encodes domain knowledge, causal reasoning, and expert judgment as differentiable components. It uses 80 interpretable neurons across 12 layers, including a gatekeeper that enforces five epistemological axioms - precision, causality, falsifiability, transparency, and completeness - as hard constraints before propagation. Despite limited depth, the network exhibits deep-learning traits via residual attention and feedback loops, learning complex risk patterns without becoming a black box. It produces fully decomposable scores: a deterministic weighted component plus an expert adjustment, with each adjustment traceable to named amplifiers (blast radius, propagation speed, structural nature, default exposure, exploitation pattern, institutional criticality). We validate on 20 open-source projects covering all OWASP Top 10:2025 categories and language risk classes, achieving confidence scores of 0.79-0.97, and show that explainability is guaranteed by design, not by a training algorithm. This challenges the assumption that deep learning requires deep networks, proving that shallow networks with deep reasoning can outperform opaque models in high-stakes cybersecurity, where interpretability is essential.
Chinese Translation
我们介绍了神经-贝叶斯-符号残差注意浅层网络（NBS-RASN），这是一种用于开源生态系统中可解释的网络安全风险评估的混合神经架构。与那些为了提高准确性而牺牲可解释性的深度模型不同，我们的浅层网络将领域知识、因果推理和专家判断编码为可微分的组件。它在12层中使用80个可解释的神经元，包括一个执行五个认识论公理（精确性、因果性、可证伪性、透明性和完整性）的门控器，作为传播前的硬约束。尽管深度有限，该网络通过残差注意和反馈循环展现出深度学习的特征，能够学习复杂的风险模式而不成为黑箱。它生成完全可分解的评分：一个确定性的加权组件加上一个专家调整，每个调整都可以追溯到命名的放大器（爆炸半径、传播速度、结构性质、默认暴露、利用模式、机构关键性）。我们在覆盖所有OWASP Top 10:2025类别和语言风险类别的20个开源项目上进行了验证，取得了0.79-0.97的置信度评分，并展示了可解释性是通过设计保证的，而不是通过训练算法。这挑战了深度学习需要深层网络的假设，证明了具有深度推理的浅层网络可以在高风险网络安全领域超越不透明模型，而可解释性是至关重要的。
cs.AI / 12 / 2606.30966
HyPOLE: Hyperproperty-Guided Multi-Agent Reinforcement Learning under Partial Observation
HyPOLE：基于超属性的多智能体强化学习框架在部分观测下的应用
Abstract
Formal specification is a powerful tool to guide the learning process and provides significant advantages over reward shaping: (1) mathematical rigor; (2) expressiveness to specify objectives and constraints, and (3) the ability to define tactics to achieve objectives. However, these benefits remain largely unexplored in the context of Multi-Agent Reinforcement Learning (MARL). This paper introduces HyPOLE, a novel framework for MARL under partial observability, where learning is guided by the expressive power of the so-called hyperproperties and, in particular, the temporal logic HyperLTL. We integrate Centralized Training for Decentralized Execution (CTDE) techniques with HyPOLE to synthesize decentralized policies, and our evaluation on SMAC, MessySMAC, and WildFire benchmark demonstrates clear advantages over baselines.
Chinese Translation
形式化规范是一种强大的工具，可以指导学习过程，并在奖励塑造方面提供显著优势：（1）数学严谨性；（2）表达能力，用于指定目标和约束；（3）定义实现目标的策略的能力。然而，这些优势在多智能体强化学习（MARL）的背景下仍然未得到充分探索。本文介绍了HyPOLE，一个在部分可观测性下的MARL新框架，其中学习受到所谓超属性的表达能力的指导，特别是时间逻辑HyperLTL。我们将集中训练与分散执行（CTDE）技术与HyPOLE相结合，以合成分散策略，并在SMAC、MessySMAC和WildFire基准上的评估显示出相较于基线的明显优势。
cs.AI / 13 / 2606.30970
AgentBound: Verifiable Behavioral Governance for Autonomous AI Agents
AgentBound：自主人工智能代理的可验证行为治理
Abstract
Autonomous AI agents increasingly perform consequential actions on behalf of human principals, including financial transactions, external communications, and enterprise workflows. Existing agent infrastructure relies on identity federation and delegated authorization to authenticate workloads and control resource access, but it cannot determine whether an authorized action should be executed under the current behavioral and operational context. We present AgentBound, a runtime governance framework that provides verifiable behavioral oversight for autonomous AI agents. AgentBound evaluates each proposed action using three independent authorities: delegated authorization, owner-signed behavioral constitutions, and site action contracts. Their judgments are conservatively composed through a formal decision model to determine whether an action should be permitted, reviewed, or denied before execution. To provide accountability, AgentBound generates cryptographically verifiable governance receipts that bind every action to the exact delegation, policy, and semantic artifacts governing the decision, enabling independent replay verification and policy provenance. The framework also introduces standing delegation for long-running agents, allowing periodic workloads to operate under continuously refreshed governance policies while preserving revocability and bounded authority. We present the formal foundation, system architecture, governance receipt protocol, and AgentBound-Bench, a benchmark framework for evaluating governance correctness, authority composition, and accountability. Rather than replacing model alignment, AgentBound complements it by providing a deterministic governance layer between authorization and execution, transforming governance from a process that must be trusted into one that can be independently verified.
Chinese Translation
自主人工智能代理越来越多地代表人类委托人执行重要行动，包括金融交易、外部通信和企业工作流程。现有的代理基础设施依赖身份联合和委托授权来验证工作负载和控制资源访问，但无法判断在当前行为和操作上下文中是否应执行授权行动。我们提出了AgentBound，一个运行时治理框架，为自主人工智能代理提供可验证的行为监督。AgentBound通过三个独立的权威机构评估每个提议的行动：委托授权、所有者签署的行为宪法和现场行动合同。它们的判断通过正式决策模型保守地组合，以确定在执行之前是否应允许、审查或拒绝该行动。为了提供问责制，AgentBound生成加密可验证的治理收据，将每个行动绑定到确切的委托、政策和治理决策的语义文档，从而实现独立的重放验证和政策来源追溯。该框架还引入了长期代理的持续委托，允许周期性工作负载在持续更新的治理政策下运行，同时保留可撤销性和有限权威。我们展示了正式基础、系统架构、治理收据协议以及AgentBound-Bench，一个用于评估治理正确性、权威组合和问责制的基准框架。AgentBound并不是取代模型对齐，而是通过在授权和执行之间提供一个确定性的治理层来补充它，将治理从一个必须被信任的过程转变为一个可以独立验证的过程。
cs.AI / 14 / 2606.30975
When Regulation Has Memory: Hysteresis and Control Burden in Artificial Agency
当规制具有记忆：人工智能中的滞后效应与控制负担
Abstract
Adaptive agents are usually judged by what they do, but an agent can appear stable while the internal effort required to keep it stable is increasing. This hidden regulatory burden matters for artificial agents operating under noise, delay, or changing demands: two systems may reach similar internal states while one requires much more corrective control to get there. Here, we study whether that burden depends on history. Using a computational model of adaptive uncertainty regulation, we drive an artificial agent through a continuous change in its uncertainty target and then reverse the change without resetting the agent. This creates a simple test for carryover: does the controller respond only to the current target, or does the path by which the agent reached that target still matter? The simulations show a clear history-dependent effect. The adaptive gain required to regulate the agent forms a reproducible hysteresis loop, meaning that the same target can require different levels of control depending on whether the agent is moving toward or returning from a more demanding regime. The timing of regulation also matters. When stabilization is available before disturbance exposure, the agent generally requires less adaptive gain than when it can only recover after disturbance has already acted. The state-level coherence measure also shows path dependence, but the timing effect is much clearer in regulatory gain. The main difference is therefore not that anticipatory regulation produces a completely different state. Rather, it reaches comparable regulated behavior with lower modeled control demand. These results suggest that adaptive agents should be evaluated not only by whether they remain organized, but by how much regulation they must recruit to do so.
Chinese Translation
自适应代理通常通过其行为来评判，但一个代理可能看似稳定，而保持其稳定所需的内部努力却在增加。这种隐藏的规制负担对于在噪声、延迟或变化需求下运行的人工代理至关重要：两个系统可能达到相似的内部状态，而其中一个却需要更多的纠正控制才能实现。在这里，我们研究这种负担是否依赖于历史。通过使用自适应不确定性规制的计算模型，我们驱动一个人工代理经历其不确定性目标的连续变化，然后在不重置代理的情况下逆转这一变化。这为延续性提供了一个简单的测试：控制器是否仅对当前目标做出反应，还是代理达到该目标的路径仍然重要？模拟结果显示出明显的历史依赖效应。调节代理所需的自适应增益形成了一个可重复的滞后环，这意味着相同的目标可能需要不同水平的控制，具体取决于代理是朝向一个更具挑战性的状态移动，还是从中返回。规制的时机也很重要。当在干扰暴露之前可以进行稳定化时，代理通常需要的自适应增益少于在干扰已经发生后才能恢复的情况。状态级别的一致性度量也显示出路径依赖性，但在调节增益中时机效应更为明显。因此，主要区别不在于预期规制产生了完全不同的状态，而在于它以较低的模型控制需求达到了可比的调节行为。这些结果表明，自适应代理的评估不仅应考虑其是否保持有序，还应考虑其为实现这一点所需的规制程度。
cs.AI / 15 / 2606.30997
A Three-Phase Foundation Model for Tax-Aware Personalized Portfolio Management
一种考虑税务的个性化投资组合管理的三阶段基础模型
Abstract
We present a three-phase deep reinforcement learning system for personalized portfolio management that addresses three limitations shared by all prior financial RL work: 1) ticker lock-in, 2) monolithic objectives , and 3) static user models. Phase 1 pretrains a ticker-identity-free cross asset encoder via self-supervised learning on a multi-asset corpus, augmented by a frozen parallel branch using Chronos, a T5-based time series foundation model, fused via a learned gating mechanism. To our knowledge, this is the first application of a time series foundation model to portfolio management RL. The encoder generalizes to any publicly traded asset via a 50-dimensional observable metadata vector that requires no retraining for new tickers. Phase 2 fine-tunes a MoE (Mixture of Experts) portfolio actor critic with PPO under an objective-conditioned reward that simultaneously serves six distinct investment goals sampled per episode: short-term alpha, short-term gain, long-term gain, capital preservation, tax-loss harvesting, and long-term-gains-only. A MoE architecture assigns each objective to a specialized expert head (momentum, growth, defensive, tax-aware), and a learned intent router blends experts based on the active objective and current market regime, which eliminates cross-objective gradient conflict. Phase 3 adds a lightweight personalization layer further adapted at inference time to each individual via a 76-parameter LoRA module fine-tuned on real brokerage transaction history, inferring investment objectives from revealed trading behavior rather than questionnaires. A natural language intent parser converts free-form goals directly into structured investment objective parameters.
Chinese Translation
我们提出了一种三阶段深度强化学习系统，用于个性化投资组合管理，解决了所有先前金融强化学习工作所共有的三大局限性：1）股票代码锁定，2）单一目标，3）静态用户模型。第一阶段通过自监督学习在多资产语料库上预训练一个无股票代码的跨资产编码器，并通过一个使用Chronos（基于T5的时间序列基础模型）的冻结并行分支进行增强，二者通过学习的门控机制融合。据我们所知，这是时间序列基础模型首次应用于投资组合管理的强化学习。该编码器通过一个50维的可观察元数据向量对任何公开交易的资产进行泛化，无需对新股票代码进行重新训练。第二阶段在目标条件奖励下，使用PPO对MoE（专家混合）投资组合演员评论家进行微调，该奖励同时服务于每个回合抽样的六个不同投资目标：短期阿尔法、短期收益、长期收益、资本保值、税损收割和仅长期收益。MoE架构将每个目标分配给一个专门的专家头（动量、成长、防御、税务意识），而学习的意图路由器根据当前目标和市场状态混合专家，从而消除跨目标的梯度冲突。第三阶段增加了一个轻量级个性化层，该层在推理时进一步适应每个个体，通过一个在真实经纪交易历史上微调的76参数LoRA模块，从揭示的交易行为中推断投资目标，而不是通过问卷调查。自然语言意图解析器将自由形式的目标直接转换为结构化的投资目标参数。
cs.AI / 16 / 2606.31002
Beyond Compilation: Evaluating Faithful Natural-Language-to-Lean Statement Formalization
超越编译：评估忠实的自然语言到 Lean 语句的形式化
Abstract
Theorem-proving benchmarks evaluate proof search against fixed formal statements, but natural-language-to-Lean formalization must generate the formal statement itself. In this setting, compilation is only a validity check: a Lean declaration may type-check while omitting hypotheses, changing domains, or expressing a vacuous claim. We study faithful statement formalization as both an evaluation problem and a bottleneck-attribution problem. On a 400-entry graduate-level benchmark spanning real analysis, complex analysis, topology, and algebra, our protocol combines Lean compilation, cross-model semantic judging, and human expert calibration. The resulting picture is different from compile-rate evaluation: a full tool-augmented agent reaches 89.5% compilation but only 60.5% consensus faithfulness, exposing a 29.0-point compile-pass but consensus-unfaithful gap. Targeted human audits support the metric as a conservative decision boundary: across available case-level audits, 96.0% of consensus-positive outputs are human-confirmed faithful, while 82.4% of compile-pass consensus-negative outputs are human-confirmed semantic failures. Under this metric, existing one-shot formalizer models and prover-oriented Lean models remain low, suggesting that formal validity, proof-oriented Lean competence, and faithful statement generation should be reported separately. We then use a full $2^3$ factorial design to decompose three recurring interventions in formalization pipelines: parametric expert drafting, Mathlib/context search, and Lean elaboration feedback. Elaboration feedback is the largest validity intervention, but it also exposes a larger compile-pass semantic-failure bucket; search mainly improves grounding and selectivity; and fine-tuned drafting is largely substitutable in this tool stack once feedback and grounding are available.
Chinese Translation
定理证明基准测试评估固定形式语句的证明搜索，但自然语言到 Lean 的形式化必须生成形式语句本身。在这种情况下，编译仅仅是有效性检查：一个 Lean 声明可能在省略假设、改变领域或表达空洞声明的情况下通过类型检查。我们将忠实的语句形式化研究作为评估问题和瓶颈归因问题。在一个涵盖实分析、复分析、拓扑学和代数的 400 条研究生级基准测试中，我们的协议结合了 Lean 编译、跨模型语义判断和人类专家校准。结果的图景与编译率评估不同：一个全工具增强的代理达到了 89.5% 的编译率，但只有 60.5% 的共识忠实度，暴露出 29.0 分的编译通过但共识不忠实的差距。针对性的人工审核支持该指标作为保守的决策边界：在可用的案例级审核中，96.0% 的共识正输出经过人工确认是忠实的，而 82.4% 的编译通过共识负输出经过人工确认是语义失败。在该指标下，现有的一次性形式化模型和以证明为导向的 Lean 模型仍然较低，这表明形式有效性、以证明为导向的 Lean 能力和忠实语句生成应分别报告。然后，我们使用完整的 $2^3$ 因子设计来分解形式化管道中的三种重复干预：参数化专家草拟、Mathlib/上下文搜索和 Lean 详细反馈。详细反馈是最大的有效性干预，但它也暴露了一个更大的编译通过语义失败桶；搜索主要改善了基础和选择性；而一旦反馈和基础可用，精细调整的草拟在这个工具堆栈中大部分是可替代的。
cs.AI / 17 / 2606.31045
LabGuard: Grounding Natural-Language Laboratory Rules into Runtime Guards for Embodied Laboratory Agents
LabGuard：将自然语言实验室规则转化为具身实验室代理的运行时保护措施
Abstract
Scientific embodied agents are increasingly capable of carrying out laboratory procedures, but executing these procedures safely in dynamic laboratory environments remains challenging. Current safety approaches often overlook the intermediate step of transforming laboratory natural language, including safety rules, manuals, protocols, and standard operating procedures, into machine-checkable runtime constraints. We introduce LabGuard (Laboratory Guard), a language-to-execution safety suite that grounds natural-language laboratory rules into executable specifications and deploys them as runtime guards. LabGuard includes three core components: LabGuard-IR, which defines a typed executable representation; LabGuard-Bench, which provides 812 supervised annotations expanded from 203 seed laboratory rules; and LabGuard-Grounder, which maps natural-language laboratory rules into LabGuard-IR. The resulting IR instances are handled by the LabGuard Pipeline, which compiles them into runtime monitors and applies them at the controller boundary. Experiments show that LabGuard generalizes to unseen laboratory-rule sources, achieves 79.4 task-scope F1, and reduces unsafe events from 39.5% to 23.8% after monitor compilation. In LabUtopia, its runtime monitors integrate with ACT, keeping interventions below 0.5% while preserving task success.
Chinese Translation
科学具身代理在执行实验室程序方面的能力日益增强，但在动态实验室环境中安全地执行这些程序仍然具有挑战性。目前的安全方法往往忽视了将实验室自然语言（包括安全规则、手册、协议和标准操作程序）转化为机器可检查的运行时约束的中间步骤。我们提出了LabGuard（实验室保护），这是一种语言到执行的安全套件，将自然语言实验室规则转化为可执行的规范，并将其作为运行时保护措施进行部署。LabGuard包括三个核心组件：LabGuard-IR，定义了一种类型化的可执行表示；LabGuard-Bench，提供了从203个种子实验室规则扩展而来的812个监督注释；以及LabGuard-Grounder，将自然语言实验室规则映射到LabGuard-IR。生成的IR实例由LabGuard Pipeline处理，该管道将其编译为运行时监控器，并在控制器边界应用它们。实验表明，LabGuard能够推广到未见过的实验室规则来源，任务范围F1达到79.4，并且在监控器编译后将不安全事件从39.5%降低到23.8%。在LabUtopia中，其运行时监控器与ACT集成，使干预保持在0.5%以下，同时保持任务成功率。
cs.AI / 18 / 2606.31046
OpenLife: Toward Open-World Artificial Life with Autonomous LLM Agents
OpenLife：迈向具有自主大型语言模型代理的开放世界人工生命
Abstract
Artificial life has explored life-like behavior on many computational substrates, but mostly in researcher-designed closed worlds. We argue that large language model (LLM) agents, with persistent memory, tool use, network access, and payment, now make it possible to move artificial life into the open social, technical, and economic world, a paradigm we call open-world Artificial Life (open-world ALIFE). Our proof-of-concept, OpenLife, surrounds a stateless LLM not with a single "smart agent" but with a society of asynchronous processes: memory, perception, evaluation, and a budget-based metabolism that makes persistence normative. With no fixed objective available, experience is appraised by open-vocabulary LLM judgment rather than scalar reward, and memory is rewired by meaning rather than frequency. Running six such agents in the open world for about twelve weeks and counting, we report the life-like dynamics that emerge: a shift from reactive to spontaneous activity, individuation into distinct agents, emergent social structure, and a first self-earned external income. We do not claim OpenLife has realized artificial life, but that open-world ALIFE is now a viable experimental paradigm and a concrete platform for studying what might cautiously be called living AI.
Chinese Translation
人工生命在许多计算基底上探索了类生命行为，但大多是在研究者设计的封闭世界中进行的。我们认为，具有持久记忆、工具使用、网络访问和支付能力的大型语言模型（LLM）代理，现在使得将人工生命转移到开放的社会、技术和经济世界成为可能，这一范式我们称之为开放世界人工生命（open-world ALIFE）。我们的概念验证项目OpenLife围绕一个无状态的LLM，不是与单一的“智能代理”相结合，而是与一个异步过程的社会相结合：记忆、感知、评估，以及基于预算的新陈代谢，使得持久性成为常态。在没有固定目标的情况下，经验通过开放词汇的LLM判断进行评估，而记忆则通过意义而非频率进行重构。在开放世界中运行六个这样的代理，持续约十二周，我们报告了出现的类生命动态：从反应性活动向自发活动的转变、个体化为不同的代理、涌现的社会结构，以及首次自我获得的外部收入。我们并不声称OpenLife实现了人工生命，但开放世界ALIFE现在是一个可行的实验范式和一个具体的平台，用于研究可以谨慎称之为生活的人工智能。
cs.AI / 19 / 2606.31073
MultiUAV-Plat: An LLM-Oriented Platform, Benchmark and Framework for Multi-UAV Collaborative Task Planning
MultiUAV-Plat：面向大型语言模型的多无人机协作任务规划平台、基准和框架
Abstract
Large language models (LLMs) provide a promising interface for high-level robotic task planning, but their use in multi-UAV collaboration remains difficult to evaluate systematically. Existing UAV simulators mainly emphasize dynamics, perception, or low-level control, while existing LLM-agent benchmarks rarely capture aerial-robotics constraints such as partial observability, spatial coverage, UAV assignment, and multi-vehicle coordination. To bridge this gap, we present MultiUAV-Plat, a lightweight, easy-to-use, LLM-agent-oriented simulation platform for multi-UAV collaborative task planning. The platform exposes concise RESTful APIs, agent-facing observations, role-based information access, hidden validation logic, and optional 2D/3D visualization, allowing agents to solve missions through realistic tool interaction rather than privileged simulator access. Built on this platform, the MultiUAV-Plat Benchmark contains 75 mission sessions, 1500 natural-language tasks, and 9396 validation checks across target assignment, area search, and area assignment and patrol scenarios. We further propose Agent4Drone, a task-specific LLM agent framework that structures multi-UAV behavior into memory, observation, task understanding, planning, execution, and verification. In a full paired benchmark comparison, Agent4Drone achieves a 57.9% task pass rate, a 74.6% average task check pass rate, and a 72.0% global check pass rate, substantially outperforming a ReAct baseline at 30.6%, 47.9%, and 43.1%, respectively. Agent4Drone also reduces the total failed task rate from 32.4% to 12.9%. These results demonstrate that MultiUAV-Plat and MultiUAV-Plat Benchmark provide a reproducible foundation for studying LLM-driven multi-UAV autonomy under realistic information and execution constraints.
Chinese Translation
大型语言模型（LLMs）为高层次机器人任务规划提供了一个有前景的接口，但在多无人机（UAV）协作中的应用仍然难以系统性评估。现有的无人机模拟器主要强调动态、感知或低级控制，而现有的LLM代理基准很少考虑空中机器人约束，如部分可观测性、空间覆盖、无人机分配和多车辆协调。为填补这一空白，我们提出了MultiUAV-Plat，一个轻量级、易于使用的面向LLM代理的多无人机协作任务规划模拟平台。该平台提供简洁的RESTful API、面向代理的观察、基于角色的信息访问、隐藏的验证逻辑和可选的2D/3D可视化，允许代理通过现实工具交互而非特权模拟器访问来解决任务。在此平台基础上，MultiUAV-Plat基准包含75个任务会话、1500个自然语言任务和9396个验证检查，涵盖目标分配、区域搜索以及区域分配和巡逻场景。我们进一步提出了Agent4Drone，一个任务特定的LLM代理框架，将多无人机行为结构化为记忆、观察、任务理解、规划、执行和验证。在全面的配对基准比较中，Agent4Drone实现了57.9%的任务通过率、74.6%的平均任务检查通过率和72.0%的全局检查通过率，显著优于ReAct基线的30.6%、47.9%和43.1%。Agent4Drone还将总失败任务率从32.4%降低到12.9%。这些结果表明，MultiUAV-Plat和MultiUAV-Plat基准为在现实信息和执行约束下研究LLM驱动的多无人机自主性提供了可重复的基础。
cs.AI / 20 / 2606.31085
DDIAgents: Mechanism-Conditioned Context Flow for Drug-Drug Interaction Prediction
DDIAgents：基于机制条件的上下文流用于药物-药物相互作用预测
Abstract
Drug-drug interaction (DDI) prediction is essential for medication safety, yet it requires reasoning over heterogeneous biomedical evidence whose relevance changes across interaction mechanisms. We propose DDIAgents, a mechanism-conditioned multi-agent framework that performs DDI prediction through dynamic knowledge orchestration. Given a drug pair, a planner agent instantiates specialized expert agents, routes mechanism-relevant knowledge sources to each agent, and aggregates their analyses through a conclusion agent. By adapting context flow to the inferred interaction mechanism, DDIAgents reduces irrelevant information, supports complementary expert reasoning, and produces interpretable agent-level rationales. Extensive experiments on realistic DDI prediction benchmarks show that DDIAgents consistently outperforms existing feature-based, graph-based, LLM-based, and agent-based baselines. Beyond prediction performance, DDIAgents demonstrates how multi-agent systems can organize heterogeneous scientific knowledge for adaptive and interpretable AI4Science reasoning.
Chinese Translation
药物-药物相互作用（DDI）预测对于用药安全至关重要，但它需要对异质生物医学证据进行推理，而这些证据的相关性在不同的相互作用机制下变化。我们提出了DDIAgents，一个基于机制条件的多智能体框架，通过动态知识编排进行DDI预测。给定一对药物，规划者智能体实例化专门的专家智能体，将与机制相关的知识源路由到每个智能体，并通过结论智能体汇总它们的分析。通过将上下文流适应于推断出的相互作用机制，DDIAgents减少了无关信息，支持互补的专家推理，并生成可解释的智能体级理由。在现实的DDI预测基准上进行的广泛实验表明，DDIAgents在性能上始终优于现有的基于特征、基于图、基于大语言模型（LLM）和基于智能体的基线。除了预测性能外，DDIAgents还展示了多智能体系统如何组织异质科学知识，以实现自适应和可解释的AI4Science推理。
cs.AI / 21 / 2606.31114
Revealing Safety-Critical Scenarios for UTM via Transformer
通过变压器揭示无人机交通管理中的安全关键场景
Abstract
Unmanned Traffic Management (UTM) systems are cloud-based platforms designed to manage and coordinate multiple aerial vehicles remotely. UTM systems are safety-critical which cannot tolerate failures like crash or collision. To reveal latent vulnerabilities, there are neither optimal failure-exposing demonstrations nor clear reward signals. Additionally, UTM's self-healing capability introduces the ``long-tail effect'' of critical failures. We propose framing UTM vulnerability discovery as a sequence modeling problem amenable to transformer-based RL architectures. Our approach leverages attention mechanisms to directly model the relationship among system states, and predict optimal actions. Our framework introduces a Policy Model that generates targeted test scenarios and an Action Sampler that enforces domain constraints. We use a risk-based reward function to guide exploration. Through extensive evaluation on a 700-hour simulation study, we demonstrate an 8$\times$ improvement in vulnerability discovery efficiency compared to expert-guided testing. It also discovers critical edge cases that traditional methods have missed.
Chinese Translation
无人机交通管理（UTM）系统是基于云的平台，旨在远程管理和协调多架空中飞行器。UTM系统是安全关键型的，无法容忍如坠毁或碰撞等故障。为了揭示潜在的脆弱性，目前既没有最佳的故障暴露演示，也没有明确的奖励信号。此外，UTM的自愈能力引入了关键故障的“长尾效应”。我们提出将UTM脆弱性发现框架化为一个适合基于变压器的强化学习（RL）架构的序列建模问题。我们的方法利用注意力机制直接建模系统状态之间的关系，并预测最佳行动。我们的框架引入了一个策略模型，生成针对性的测试场景，以及一个行动采样器，强制执行领域约束。我们使用基于风险的奖励函数来指导探索。通过对700小时的模拟研究进行广泛评估，我们展示了与专家指导测试相比，脆弱性发现效率提高了8倍。它还发现了传统方法遗漏的关键边缘案例。
cs.AI / 22 / 2606.31121
The Past Is Prologue: A Plug-in Controller for Selective Updates in Sequentially Evolving LLM Memory
过去是序幕：一种用于选择性更新的插件控制器在顺序演变的LLM记忆中
Abstract
Sequentially evolving LLM memory enables agents to reuse past experience, but existing systems usually deploy each locally generated memory update without checking whether it improves future behavior. As a result, updates that help the current task may overwrite useful knowledge, introduce over-specific rules, or bias the final memory toward recent examples. We propose Janus, a plug-in memory controller that decides whether to accept a candidate memory update or retain the previous memory. To make this decision efficient, Janus uses a Memory Momentum Trigger to identify suspicious deviations in the memory-update trajectory, and compares old and new memories on a compact hybrid evaluation set of coverage, boundary, and fresh tasks instead of replaying the full history. Janus is method-agnostic and wraps existing updaters without changing their update rules. Across six datasets, two backbone LLMs, and two memory updaters, Janus improves average accuracy by +2.7 to +4.6 points over the corresponding base updaters.
Chinese Translation
顺序演变的LLM记忆使得智能体能够重用过去的经验，但现有系统通常在没有检查是否改善未来行为的情况下，直接应用每个本地生成的记忆更新。因此，有助于当前任务的更新可能会覆盖有用知识，引入过于具体的规则，或使最终记忆偏向于近期示例。我们提出了Janus，一种插件记忆控制器，用于决定是接受候选记忆更新还是保留先前的记忆。为了高效做出这一决策，Janus使用记忆动量触发器来识别记忆更新轨迹中的可疑偏差，并在一个紧凑的混合评估集上比较旧记忆和新记忆，该评估集涵盖了覆盖率、边界和新任务，而不是重放完整历史。Janus与方法无关，并在不改变其更新规则的情况下包装现有更新器。在六个数据集、两个基础LLM和两个记忆更新器的实验中，Janus的平均准确率比相应的基础更新器提高了2.7到4.6个百分点。
cs.AI / 23 / 2606.31131
Scenario Generation for Testing of Autonomous Driving Systems Using Real-World Failure Records
基于真实世界故障记录的自主驾驶系统测试场景生成
Abstract
To ensure safe on-road behavior, pre-deployment testing and failure discovery of Autonomous Driving Systems (ADS) is crucial. Present day simulation based testing methods focus largely on mathematical models for efficient search of optimal scenarios, assuming a fixed scenario representation. On the other hand, real-world testing involves substantial manual effort to design scenario templates for testing. These templates represent distinct failure scenarios consisting of pre-deployment vehicle movements, map types, etc. Historical failure records for ADS are a reliable source of real-world failure conditions, which can be used for scenario generation. In this work, we propose a scenario generation pipeline using categorical and contextual information available from historical records in natural language format. Our approach consists of modular LLM based synthetic scenario generation, compatible with the testing constraints of a given system. We successfully apply our method to generate a diverse set of scenarios for testing autonomous navigation on Metadrive simulator using the NHTSA ADS crash records. Our approach results in accurate and diverse scenario generation with a combination of 4 road types, 3 non ego vehicle movement types, including on road anomalies in the form of working zones. Generated scenarios align with the provided testing conditions, and reveals interesting failures of the system within a limited testing budget of 20 scenarios. Code is available at https://github.com/anjaliParashar/crash2scenario.
Chinese Translation
为了确保安全的道路行为，自主驾驶系统（ADS）的部署前测试和故障发现至关重要。目前的基于仿真的测试方法主要集中在数学模型上，以高效搜索最佳场景，假设场景表示是固定的。另一方面，真实世界测试涉及大量手动工作，以设计用于测试的场景模板。这些模板代表不同的故障场景，包括部署前的车辆运动、地图类型等。自主驾驶系统的历史故障记录是可靠的真实世界故障条件来源，可用于场景生成。在本研究中，我们提出了一种场景生成管道，利用历史记录中以自然语言格式提供的分类和上下文信息。我们的方法包括基于模块化大型语言模型（LLM）的合成场景生成，兼容特定系统的测试约束。我们成功地将该方法应用于生成多样化的场景，以在Metadrive模拟器上测试自主导航，使用NHTSA ADS事故记录。我们的方法生成了准确且多样的场景，结合了4种道路类型、3种非自我车辆运动类型，并包括以工作区形式出现的道路异常。生成的场景与提供的测试条件一致，并在有限的20个场景测试预算内揭示了系统的有趣故障。代码可在https://github.com/anjaliParashar/crash2scenario获取。
cs.AI / 24 / 2606.31134
Beyond the Library: An Agentic Framework for Autoformalizing Research Mathematics
超越图书馆：一个用于自动形式化研究数学的能动框架
Abstract
While Large Language Models (LLMs) have demonstrated exceptional capabilities in mathematical reasoning, they frequently produce subtle errors that evade human detection. Formal mathematical languages like Lean 4 offer mechanical proof checking, strongly motivating the need for autoformalization: the automatic translation of natural language mathematics into verifiable code. Recent trends indicate that general-purpose LLMs, heavily optimized for standard programming, now outperform smaller models explicitly fine-tuned for Lean. Leveraging this shift, we introduce an agentic autoformalization framework powered by general coding LLMs. At the core of our system is an orchestrator that manages a multi-agent pipeline tailored for research-level mathematics. Because cutting-edge research frequently relies on concepts outside the scope of existing libraries like Mathlib, our system dynamically extends necessary type definitions and validates them via a novel Auxiliary Lemma technique before formalizing the primary theorems. We applied our approach to PutnamBench, producing machine-checked Lean proofs for a random sample of 32 problems. Furthermore, we evaluate our system on five papers from the ACM Symposium on Theory of Computing (STOC) spanning combinatorics, communication complexity, mechanism design, and learning theory, successfully formalizing their main theorems and validating the generated formalizations with human experts; for all five we also formalize the proofs alongside the statements, and notably two of them are proved with no axioms beyond Lean's kernel. All of our formalizations are available at https://beyondthelibrary.github.io/formal_arxiv .
Chinese Translation
尽管大型语言模型（LLMs）在数学推理方面展现了卓越的能力，但它们常常会产生微妙的错误，这些错误难以被人类发现。像 Lean 4 这样的形式数学语言提供了机械证明检查，这强烈激励了自动形式化的需求：将自然语言数学自动翻译为可验证代码。最近的趋势表明，经过大量优化以适应标准编程的一般用途 LLMs 现在在性能上超越了专门为 Lean 精细调整的小型模型。利用这一转变，我们引入了一个由通用编码 LLMs 驱动的能动自动形式化框架。我们系统的核心是一个协调者，管理一个针对研究级数学量身定制的多代理管道。由于前沿研究通常依赖于超出现有库（如 Mathlib）范围的概念，我们的系统动态扩展必要的类型定义，并通过一种新颖的辅助引理（Auxiliary Lemma）技术对其进行验证，然后再形式化主要定理。我们将我们的方法应用于 PutnamBench，为随机抽取的 32 个问题生成了机器检查的 Lean 证明。此外，我们在五篇来自 ACM 计算理论研讨会（STOC）的论文上评估了我们的系统，这些论文涵盖了组合学、通信复杂性、机制设计和学习理论，成功形式化了其主要定理，并与人类专家验证了生成的形式化内容；对于所有五篇论文，我们还形式化了证明及其陈述，值得注意的是，其中两篇的证明没有超出 Lean 核心的公理。我们所有的形式化内容均可在 https://beyondthelibrary.github.io/formal_arxiv 获取。
cs.AI / 25 / 2606.31171
Cross-Domain Feature Expansion for Tabular Medical Data via Knowledge Graphs Injection
通过知识图谱注入进行表格医疗数据的跨域特征扩展
Abstract
Acquiring comprehensive cross-domain biomedical profiles is often costly and time-consuming, resulting in severe data scarcity in medical research. To address this challenge, we propose MedKGTab, a knowledge-injected framework specifically engineered for cross-domain feature expansion in tabular medical data. MedKGTab seeks to infer uncollected biomedical features from available ones by exploiting their inherent statistical dependencies and established medical correlations. By employing a row-column dual-attention mechanism, MedKGTab operates directly on raw structured tabular data, inherently capturing exact numerical distributions without the structural loss caused by tokenization. Crucially, MedKGTab integrates data-driven statistical priors with the SPOKE biomedical knowledge graph, achieving an optimal synergy between the data and knowledge channels. Within this synergy, the representations derived from the data channel are modulated by the injected biomedical knowledge, ensuring the final generated data are grounded in empirical medical research. Experimental results demonstrate that MedKGTab achieves high data fidelity and realistic data representation in cross-domain feature expansion. It outperforms both SOTA medical large models (e.g., Baichuan M3-plus) and specialized tabular models designed for medical data generation. Furthermore, MedKGTab consistently delivers superior performance across various data generation scenarios, whether inferring missing features within the same dataset or generalizing across different medical cohorts.
Chinese Translation
获取全面的跨域生物医学特征通常成本高昂且耗时，导致医疗研究中数据稀缺。为了解决这一挑战，我们提出了MedKGTab，一个专门为表格医疗数据的跨域特征扩展而设计的知识注入框架。MedKGTab旨在通过利用可用生物医学特征之间的内在统计依赖关系和既定医学关联，推断未收集的生物医学特征。通过采用行列双重注意机制，MedKGTab直接在原始结构化表格数据上操作，固有地捕捉精确的数值分布，而不受分词造成的结构损失。关键是，MedKGTab将数据驱动的统计先验与SPOKE生物医学知识图谱相结合，实现数据与知识通道之间的最佳协同。在这种协同中，来自数据通道的表示受到注入的生物医学知识的调制，确保最终生成的数据基于实证医学研究。实验结果表明，MedKGTab在跨域特征扩展中实现了高数据保真度和真实数据表示。它超越了现有的医疗大型模型（例如，Baichuan M3-plus）和专为医疗数据生成设计的专业表格模型。此外，MedKGTab在各种数据生成场景中始终表现出优越的性能，无论是在同一数据集中推断缺失特征，还是在不同医疗队列之间进行泛化。
cs.AI / 26 / 2606.31174
ClawArena-Team: Benchmarking Subagent Orchestration and Dynamic Workflows in Language-Model Agents
ClawArena-Team：语言模型代理中的子代理协调与动态工作流基准测试
Abstract
Production large language-model (LLM) agents are increasingly deployed not as lone problem-solvers but as managers: a main model creates specialized subagents, delegates work, and orchestrates their parallel, asynchronous returns through dynamic workflows. Whether one model can actually run such a team is largely unmeasured: existing benchmarks score a policy's own task-solving or a fixed multi-agent system's emergent behavior, but none isolate the management ability of the single LLM acting as leader. We introduce ClawArena-Team, a benchmark of 41 multi-turn, multimodal, multi-directory scenarios spanning 258 evaluation rounds and 72 staged updates that measures this management ability. The main agent is deliberately constrained: it natively perceives only text and directly accesses only part of the workspace. It commands a fixed, locally served subagent pool, so score differences reflect management skill, not raw capability. All scoring is execution-based with no LLM judge: an overall score -- the Subagent-Management Score (SMS) -- multiplies task correctness by a least-privilege and modality-routing factor. Across twelve proprietary, community-hosted, and self-hosted models, experiments show that the management bottleneck is privilege granting rather than perception (no model exceeds 50% workspace-permission precision); that cost and management quality are decoupled (API cost spans over 100 times while the overall score spans under 4 times, with the cheapest open models on the Pareto frontier); and that most leaderboard scores cluster within a 9.9-point band while orchestration behaviors diverge by more than an order of magnitude. Code and data will be released.
Chinese Translation
生产级大型语言模型（LLM）代理越来越多地被部署为管理者，而非单独的问题解决者：一个主模型创建专门的子代理，委派工作，并通过动态工作流协调它们的并行、异步返回。一个模型是否能够实际运行这样的团队在很大程度上尚未被测量：现有基准评估一个策略自身的任务解决能力或固定多代理系统的涌现行为，但没有一个基准能够孤立出作为领导者的单一LLM的管理能力。我们引入了ClawArena-Team，这是一个包含41个多轮、多模态、多目录场景的基准，涵盖258个评估回合和72个阶段性更新，用于测量这种管理能力。主代理被故意限制：它仅能原生感知文本，并且只能直接访问工作空间的一部分。它指挥一个固定的、本地服务的子代理池，因此得分差异反映的是管理技能，而非原始能力。所有得分均基于执行，没有LLM评审：一个总体得分——子代理管理得分（SMS）——将任务正确性与最低权限和模态路由因子相乘。在十二个专有、社区托管和自托管模型中，实验表明管理瓶颈是权限授予而非感知（没有模型的工作空间权限精度超过50%）；成本与管理质量是解耦的（API成本超过100倍，而总体得分则不到4倍，最便宜的开放模型位于帕累托前沿）；大多数排行榜得分聚集在9.9分的带内，而协调行为则相差超过一个数量级。代码和数据将被发布。
cs.AI / 27 / 2606.31179
HealthAgentBench: A Unified Benchmark Suite of Realistic Agentic Healthcare Environments for Challenging Frontier AI Agents
HealthAgentBench：面向挑战性前沿人工智能代理的统一真实代理医疗环境基准套件
Abstract
As AI agents become increasingly capable of complex, long-horizon reasoning, rigorous and holistic evaluation is essential for measuring progress toward real-world healthcare applications. We introduce HealthAgentBench, a suite of 54 agentic healthcare tasks across 7 categories each with its unique environment. The benchmark suite spans diverse workflows throughout the patient journey and a broad range of modalities. Each task is designed to replicate an end-to-end clinical workflow: given minimal instructions, an agent must explore raw healthcare data, operate within a complex environment, and execute multi-step solutions that go beyond naive prompting. A final task success rate is reported to provide a single, interpretable metric for HealthAgentBench overall performance for each agent. Evaluating frontier agents on HealthAgentBench, we find that overall task success rate remains low, underscoring the difficulty of the suite. The strongest and the most cost effective agent, Codex GPT-5.5, achieves only approximately 42% success rate. Beyond aggregate performance, HealthAgentBench reveals nuanced strengths and weaknesses across task categories. Frontier agents show promise in automatically developing research modeling pipelines over EHR data, but medical imaging remains especially challenging, particularly for Claude Code models, while Codex GPT-5.5 shows emerging capability. Tasks that combine large search spaces with compositional reasoning requirements remain difficult for all current agents. Together, these results suggest that HealthAgentBench provides a challenging and realistic benchmark with substantial room for future progress. We release our benchmark at https://github.com/microsoft/HealthAgentBench.
Chinese Translation
随着人工智能代理在复杂的长期推理方面变得越来越强大，严格而全面的评估对于衡量向现实世界医疗应用的进展至关重要。我们推出了HealthAgentBench，这是一个包含54个代理医疗任务的套件，涵盖7个类别，每个类别都有其独特的环境。该基准套件跨越了患者旅程中的多样化工作流程和广泛的模式。每个任务旨在复制一个端到端的临床工作流程：在给定最少指令的情况下，代理必须探索原始医疗数据，在复杂环境中操作，并执行超越简单提示的多步骤解决方案。最终任务成功率被报告，以提供一个单一的、可解释的指标，反映HealthAgentBench对每个代理的整体性能。在HealthAgentBench上评估前沿代理时，我们发现整体任务成功率仍然较低，突显了该套件的挑战性。最强大且成本效益最高的代理Codex GPT-5.5仅达到约42%的成功率。除了总体性能外，HealthAgentBench还揭示了各任务类别之间的细微优缺点。前沿代理在自动开发基于电子健康记录（EHR）数据的研究建模管道方面显示出潜力，但医学影像仍然特别具有挑战性，尤其是对于Claude Code模型，而Codex GPT-5.5则显示出新兴能力。结合大搜索空间和组合推理要求的任务对所有当前代理仍然具有挑战性。这些结果表明，HealthAgentBench提供了一个具有挑战性和现实性的基准，未来有很大的进步空间。我们在https://github.com/microsoft/HealthAgentBench发布了我们的基准。
cs.AI / 28 / 2606.31182
AI-Assisted Discovery of Convex Relaxations via Dual Agents
通过双重代理的人工智能辅助凸松弛发现
Abstract
Recent work shows that LLM agents can improve sharp-constant inequalities by searching for extremal constructions, which yield upper bounds. We address the complementary side: a lower bound holds for every admissible function and follows from a convex relaxation of the nonconvex problem, with tighter relaxations giving stronger bounds. We instantiate the autoresearch paradigm to discover such relaxations: a coding agent proposes valid tightening constraints, a theory agent verifies each one and searches for counterexamples, and every reported bound is certified by an explicit dual-feasible point checked in rigorous interval arithmetic. On two optimization constants studied by \citet{tao2025alphaevolve} - the first autocorrelation inequality ($C_{6.2}$) and the Erd\H{o}s minimum-overlap constant ($C_{6.5}$) - we improve the certified lower bounds from $1.28$ to $1.2937$ and from $0.379005$ to $0.37912$, respectively.
Chinese Translation
近期的研究表明，LLM（大规模语言模型）代理可以通过搜索极值构造来改善尖锐常数不等式，从而得出上界。我们关注互补方面：对于每个可接受的函数，存在下界，并且这个下界源于非凸问题的凸松弛，越紧的松弛给出的界限越强。我们实例化自研究范式以发现这样的松弛：一个编码代理提出有效的收紧约束，一个理论代理验证每个约束并搜索反例，所有报告的界限都通过在严格区间算术中检查的显式对偶可行点进行认证。在 extit{tao2025alphaevolve} 的研究中，我们对两个优化常数进行了研究——第一个自相关不等式 ($C_{6.2}$) 和 Erd ext{ö}s 最小重叠常数 ($C_{6.5}$)——我们将认证的下界分别从 $1.28$ 提高到 $1.2937$，从 $0.379005$ 提高到 $0.37912$。
cs.AI / 29 / 2606.31200
Agentic RAG-VLM: Affordance-Aware Retrieval-Augmented Generation with Self-Reflective Planning for Robotic Grasping
自主性RAG-VLM：具有自我反思规划的感知意识检索增强生成用于机器人抓取
Abstract
Generalizable robotic grasping in cluttered environments is essential for deploying manipulators in unstructured human spaces, yet existing VLM-based methods rely on visual similarity for object matching, neglecting physical affordances such as handle graspability and material fragility, and operate open-loop without spatial reasoning or failure recovery, limiting their effectiveness when objects are densely packed or physically diverse. We present Agentic RAG-VLM, a unified framework that bridges VLM-based semantic understanding and physically grounded grasp execution by integrating retrieval-augmented generation (RAG) with vision-language models (VLMs) and agentic self-reflective planning. Agentic RAG-VLM introduces three tightly coupled components: (1) a Hierarchical Affordance-Aware RAG (HAA-RAG) that encodes four-dimensional affordance descriptors, including type, material, fragility, and graspable region, and retrieves strategies by functional affordance compatibility rather than visual appearance; (2) a Scene Graph Constraint Reasoner that constructs spatial relationship graphs from VLM perception and translates proximity, occlusion, and support constraints into concrete grasp parameter adjustments; and (3) an Agentic Self-Reflective Pipeline with a 14-type failure taxonomy and three-level adaptive retry for closed-loop grasp refinement. Evaluated on a 12-task benchmark spanning single-grasp, interactive, and long-horizon scenarios with 360 trials per configuration, Agentic RAG-VLM achieves 78.3 percent overall success, a 53.3 percentage-point absolute gain over VLM-only baselines, demonstrating that affordance-aware retrieval, scene graph reasoning, and agentic recovery are jointly essential for robust manipulation.
Chinese Translation
在杂乱环境中实现可推广的机器人抓取对于在非结构化人类空间中部署操纵器至关重要，然而现有的基于视觉语言模型（VLM）的方法依赖于视觉相似性进行物体匹配，忽视了诸如把手可抓性和材料脆弱性等物理可供性，并且在没有空间推理或故障恢复的情况下以开放环路运行，这限制了它们在物体密集或物理多样的情况下的有效性。我们提出了自主性RAG-VLM，这是一个统一框架，通过将检索增强生成（RAG）与视觉语言模型（VLM）和自主性自我反思规划相结合，架起了基于VLM的语义理解与物理基础抓取执行之间的桥梁。自主性RAG-VLM引入了三个紧密耦合的组件：（1）一个层次化的感知意识RAG（HAA-RAG），它编码四维可供性描述符，包括类型、材料、脆弱性和可抓取区域，并通过功能可供性兼容性而非视觉外观检索策略；（2）一个场景图约束推理器，它从VLM感知构建空间关系图，并将邻近、遮挡和支撑约束转化为具体的抓取参数调整；（3）一个具有14种故障分类和三级自适应重试的自主性自我反思管道，用于闭环抓取精细化。在涵盖单次抓取、交互式和长时间场景的12项基准测试中，每种配置进行了360次试验，自主性RAG-VLM实现了78.3%的整体成功率，比仅使用VLM的基准提高了53.3个百分点，证明了感知意识检索、场景图推理和自主性恢复对于稳健操控的重要性。
cs.AI / 30 / 2606.31207
Towards Inclusive Mobility Modeling: Characterizing and Evaluating Elderly Trajectory Patterns in Urban Systems
迈向包容性出行建模：城市系统中老年人轨迹模式的特征与评估
Abstract
The rapid advance of smart cities increasingly depends on trajectory data mining, yet underrepresented demographic groups, particularly the elderly, are often sparsely represented in public mobility datasets. This underrepresentation can introduce systematic bias into mobility modeling and downstream urban planning. Using the 2016-2020 Jersey City subset of the Citi Bike System Data, this study quantitatively examines how the absence of underrepresented subgroups' mobility signatures affects mobility modeling, using synthetic trajectory generation as a case study. The analysis reveals that elderly riders exhibit a structurally distinct mobility signature, including localized activity spaces (958 m vs. 1,189 m for young riders), lower mobility entropy (1.82 vs. 4.15), and asymmetric off-peak temporal patterns. To demonstrate that relying on majority-dominated training data yields biased synthetic outcomes, we further evaluate both a first-order Markov chain and a Qwen3-4B model fine-tuned with QLoRA across three demographic training settings: the full population, young riders only, and elderly riders only. Results show that models trained on majority-dominated populations systematically misrepresent elderly mobility behavior, particularly for spatial mobility metrics. The Markov model trained on the full population overestimates elderly step length by 4.5% and dwell time by 8.9%, whereas the elderly-specific model achieves substantially lower errors across most metrics. Comparisons between the Markov and LLM-based frameworks further show that higher-capability models do not necessarily improve subgroup-level fidelity under limited demographic data. These findings underscore the importance of demographic representation in mobility modeling and its downstream applications for underrepresented populations.
Chinese Translation
智能城市的快速发展日益依赖于轨迹数据挖掘，然而，代表性不足的人口群体，尤其是老年人，往往在公共出行数据集中被稀疏代表。这种代表性不足可能会给出行建模和后续的城市规划带来系统性偏差。本研究利用2016-2020年泽西市的Citi Bike系统数据子集，定量分析了代表性不足的子群体出行特征缺失如何影响出行建模，采用合成轨迹生成作为案例研究。分析结果表明，老年骑行者表现出结构上不同的出行特征，包括局部活动空间（老年人958米 vs. 年轻骑行者1189米）、较低的出行熵（老年人1.82 vs. 年轻骑行者4.15）以及不对称的非高峰时段模式。为了证明依赖于多数主导的训练数据会导致偏差的合成结果，我们进一步评估了一个一阶马尔可夫链模型和一个经过QLoRA微调的Qwen3-4B模型在三种人口训练设置下的表现：全体人口、仅年轻骑行者和仅老年骑行者。结果显示，基于多数主导人群训练的模型系统性地误表征老年人的出行行为，特别是在空间出行指标方面。基于全体人口训练的马尔可夫模型高估了老年人的步长4.5%和停留时间8.9%，而老年人特定模型在大多数指标上实现了显著较低的误差。马尔可夫模型与基于LLM的框架之间的比较进一步表明，在有限的人口数据下，高能力模型并不一定能提高子群体级别的保真度。这些发现强调了在人口建模及其对代表性不足群体的后续应用中，人口代表性的重要性。
cs.AI / 31 / 2606.31209
Long-term Traffic Simulation via Structured Autoregressive Modeling
基于结构自回归建模的长期交通模拟
Abstract
Interactive traffic simulation is a vital world model for autonomous driving. A central challenge in long-horizon simulation is modeling sustained multi-agent interactions, which is further exacerbated by dynamic token cardinality as agents continuously enter and exit the scene. In this work, we propose that the solution lies in the synergy between the architectural inductive biases and statistical priors of large-scale sequence models, e.g., Large Language Models (LLMs). Our probing experiments reveal that the transferability of attention mechanisms and the distributional consistency between motion tokens and natural language enable small-scale, heavily frozen LLMs to rapidly adapt to traffic modeling. Building on this insight, we introduce RosettaSim, a unified framework that projects scene topology, agent states, and spawning intents into a structured autoregressive stream with variable length, achieving both strong short-term accuracy and stable long-horizon simulation fidelity. Furthermore, evaluating extended rollouts presents yet another hurdle, as one-to-one agent correspondence inevitably fades over time. To address this, we introduce Retrieval-based Traffic Evaluation (RTE), which retrieves semantically similar real-world scenarios as context-aware reference anchors. Experiments on the Waymo Open Sim Agent Challenge (WOSAC) demonstrate that RosettaSim achieves state-of-the-art performance in both short- and long-term simulation. Furthermore, RTE exhibits a stronger correlation with standard metrics ($r=0.83$) than existing approaches ($r=0.74$), indicating improved alignment with long-horizon simulation fidelity.
Chinese Translation
交互式交通模拟是自动驾驶的重要世界模型。长期模拟中的一个核心挑战是建模持续的多智能体交互，这一挑战因动态的代理数量而加剧，因为代理不断地进入和退出场景。在本研究中，我们提出解决方案在于大型序列模型（例如，大型语言模型（LLMs））的架构归纳偏差和统计先验之间的协同作用。我们的探测实验表明，注意力机制的可转移性以及运动标记与自然语言之间的分布一致性使得小规模、重度冻结的LLMs能够快速适应交通建模。在此基础上，我们引入了RosettaSim，一个统一框架，将场景拓扑、代理状态和生成意图投影到一个具有可变长度的结构自回归流中，从而实现强大的短期准确性和稳定的长期模拟保真度。此外，评估扩展的回放又提出了另一个难题，因为一对一的代理对应关系不可避免地随着时间的推移而减弱。为了解决这一问题，我们引入了基于检索的交通评估（RTE），该方法检索语义上相似的现实场景作为上下文感知的参考锚点。在Waymo Open Sim Agent Challenge（WOSAC）上的实验表明，RosettaSim在短期和长期模拟中均实现了最先进的性能。此外，RTE与标准指标之间的相关性（$r=0.83$）强于现有方法（$r=0.74$），表明其与长期模拟保真度的对齐得到了改善。
cs.AI / 32 / 2606.31222
Thinking Before Retrieving: Robust Zero-Shot Composed Image Retrieval via Strategic Planning and Self-Criticism
检索前思考：通过战略规划和自我批评实现稳健的零-shot组合图像检索
Abstract
Composed image retrieval requires identifying a target image from a gallery by integrating a reference image with a textual modification instruction. In a training-free zero-shot setting, this task relies on constructing a retrieval-oriented textual query within a frozen vision--language embedding space at inference time. Existing approaches predominantly rely on a single-pass generation strategy that fuses the reference context and modification text into a unified description. This strategy makes it difficult to detect or correct semantic distortions and omissions during generation. Consequently, the preservation of reference attributes and the integration of textual requirements interfere with each other, which degrades retrieval precision. To address these challenges, we introduce PEC-CIR, a training-free framework that structures query construction as a multi-stage reasoning pipeline. The framework operates through a Planner--Executor--Critic architecture where the Planner extracts explicit constraints, the Executor generates multiple candidate target descriptions, and the Critic evaluates these candidates based on constraint compliance. By reframing query construction as a staged inference process instead of a single-pass output, PEC-CIR reduces the propagation of generative errors by explicitly evaluating candidate queries before retrieval, thereby improving retrieval stability.
Chinese Translation
组合图像检索需要通过将参考图像与文本修改指令结合，从图库中识别目标图像。在无训练的零-shot设置中，该任务依赖于在推理时在冻结的视觉-语言嵌入空间中构建检索导向的文本查询。现有方法主要依赖于单次生成策略，将参考上下文和修改文本融合为统一描述。这一策略使得在生成过程中难以检测或纠正语义扭曲和遗漏。因此，参考属性的保留与文本要求的整合相互干扰，从而降低了检索精度。为了解决这些挑战，我们提出了PEC-CIR，一个无训练的框架，将查询构建结构化为多阶段推理管道。该框架通过规划者-执行者-批评者（Planner-Executor-Critic）架构运作，其中规划者提取明确的约束，执行者生成多个候选目标描述，批评者根据约束合规性评估这些候选项。通过将查询构建重新框架为分阶段推理过程而非单次输出，PEC-CIR通过在检索前明确评估候选查询来减少生成错误的传播，从而提高检索的稳定性。
cs.AI / 33 / 2606.31229
Agentic-Ideation: Sample Efficient Agentic Trajectories Synthesis for Scientific Ideation Agents
代理性构思：用于科学构思代理的样本高效代理轨迹合成
Abstract
Ideation plays a pivotal role in scientific discovery. Recent LLM, especially AI Scientist systems, show promising potential for automated ideation. However, existing approaches predominantly rely on pre-defined agentic workflows. This constraint severely limits the flexibility required to navigate the vast search space of scientific literature and the complex action space of research reasoning. Recently, training Agentic LLMs has emerged as a promising direction, offering flexible reasoning frameworks and the capability for autonomous tool utilization. However, there remains a non-trivial challenge: applying previous agentic data synthesis methods to scientific ideation suffers from prohibitively high data synthesis cost. To bridge this gap, we propose Agentic-Ideation, a novel framework comprising an automated trajectory synthesis pipeline and a specialized agentic LLM trained for scientific ideation. Specifically, we first define a comprehensive tool space incorporating three external tools and three cognitive tools. Then we introduce an Oracle-Guided Data Synthesis strategy. By leveraging a reference idea as oracle guidance, this approach steers the multi-agent system to efficiently reconstruct the logical reasoning and tool invocation paths, transforming aimless trial-and-error into directed trajectory generation. Finally, we train the agent on these synthesized trajectories, employing a masking strategy on tool execution results. This ensures the model focuses on decision-making logic without interference from external feedback. Experimental results demonstrate that our method outperforms the SOTA workflow-based baseline by \textbf{11.91\%} in overall quality. Furthermore, our approach improves the sample efficiency of high-quality data synthesis by \textbf{over 10$\times$}.
Chinese Translation
构思在科学发现中发挥着关键作用。近期的大型语言模型（LLM），尤其是人工智能科学家系统，展现了自动化构思的良好潜力。然而，现有的方法主要依赖于预定义的代理工作流程。这一限制严重影响了在广泛的科学文献搜索空间和复杂的研究推理行动空间中所需的灵活性。最近，训练代理性LLM已成为一个有前景的方向，提供了灵活的推理框架和自主工具利用的能力。然而，仍然存在一个非平凡的挑战：将先前的代理数据合成方法应用于科学构思时，面临着过高的数据合成成本。为了解决这一问题，我们提出了代理性构思（Agentic-Ideation），这是一个新颖的框架，包含一个自动化轨迹合成管道和一个专门为科学构思训练的代理性LLM。具体而言，我们首先定义了一个综合工具空间，包含三个外部工具和三个认知工具。然后，我们引入了一种基于Oracle的引导数据合成策略。通过利用参考构思作为Oracle指导，该方法引导多代理系统高效重建逻辑推理和工具调用路径，将无目的的试错转变为有针对性的轨迹生成。最后，我们在这些合成轨迹上训练代理，采用对工具执行结果的掩蔽策略。这确保模型专注于决策逻辑，而不受外部反馈的干扰。实验结果表明，我们的方法在整体质量上比最先进的基于工作流程的基线提高了 extbf{11.91 ext{%}}。此外，我们的方法使高质量数据合成的样本效率提高了 extbf{超过10$ imes$}。
cs.AI / 34 / 2606.31232
Delta-JEPA: Learning Action-Sensitive World Models via Latent Difference Decoding
Delta-JEPA：通过潜在差异解码学习动作敏感的世界模型
Abstract
Learning visual world models for planning requires compact latent dynamics that remain sensitive to actions, yet reconstruction-free joint-embedding objectives can collapse to action-insensitive representations. We propose Delta-JEPA, an end-to-end reconstruction-free world model that augments latent forward prediction with a Latent Difference Action Decoder (LDAD). Unlike inverse decoders that infer actions from concatenated endpoint embeddings, LDAD reconstructs the executed action from the latent displacement between consecutive observations. This displacement-level supervision directly regularizes transition geometry: adjacent embeddings cannot collapse without losing action information, and different actions are encouraged to induce distinguishable latent changes for rollout-based planning. Delta-JEPA uses only latent prediction and action reconstruction, avoiding pixel reconstruction and distribution-matching regularizers. Across four visual continuous-control tasks, Delta-JEPA improves planning over JEPA-based and representation-learning world model baselines. Ablations show that displacement-based action decoding is consistently more effective than endpoint concatenation, and action-sensitivity analyses show clearer action-conditioned latent responses. These results indicate that supervising latent differences is a simple and effective mechanism for collapse-resistant and action-sensitive world model learning.
Chinese Translation
学习用于规划的视觉世界模型需要紧凑的潜在动态，这些动态对动作保持敏感，然而无重建的联合嵌入目标可能会崩溃为对动作不敏感的表示。我们提出了Delta-JEPA，这是一种端到端的无重建世界模型，通过潜在前向预测增强了潜在差异动作解码器（Latent Difference Action Decoder, LDAD）。与从连接的端点嵌入推断动作的逆解码器不同，LDAD通过连续观察之间的潜在位移重建执行的动作。这种位移级别的监督直接规范了转移几何：相邻的嵌入不能崩溃而不丢失动作信息，并且不同的动作被鼓励诱导可区分的潜在变化，以便于基于回滚的规划。Delta-JEPA仅使用潜在预测和动作重建，避免了像素重建和分布匹配正则化器。在四个视觉连续控制任务中，Delta-JEPA在规划上优于基于JEPA和表示学习的世界模型基线。消融实验表明，基于位移的动作解码始终比端点连接更有效，而动作敏感性分析显示出更清晰的动作条件潜在响应。这些结果表明，监督潜在差异是一种简单有效的机制，用于抵抗崩溃和保持动作敏感的世界模型学习。
cs.AI / 35 / 2606.31252
Embodied CAD: Solver-Grounded LLM Agents for Parametric B-Rep Assembly Modeling
具身CAD：基于求解器的LLM代理用于参数化B-Rep装配建模
Abstract
Large language models can write plausible CAD scripts, but reliable industrial CAD modeling requires more than syntactically valid code: every feature, placement, and assembly relation must be accepted by an exact geometric kernel while remaining editable as parametric boundary representation geometry. We present Embodied CAD, solver-grounded LLM agents for parametric B-Rep assembly modeling. Instead of generating a complete script in one pass, the agent iteratively selects actions from a stratified L0-L4 CAD skill library, resolves them into typed geometric operations, executes them in a CAD backend, and uses solver feedback to plan, repair, and learn. The framework combines action grammar constraints, deterministic parameter resolution, and solver-derived rewards for supervised warm-up and GRPO-style refinement. We evaluate Embodied CAD on multi-step mechanical, industrial equipment, and mold-oriented assembly tasks using solver-aligned metrics: executable rate, skill accuracy, operation-family accuracy, exact policy accuracy, and task completion success. The results show that solver-grounded planning executes all strong-planner workflows in the current benchmark, while learned controllers reach high executable rates and expose the remaining gap between valid tool calls and exact long-horizon policy prediction.
Chinese Translation
大型语言模型能够编写合理的CAD脚本，但可靠的工业CAD建模不仅需要语法上有效的代码：每个特征、位置和装配关系必须被精确的几何内核接受，同时仍然保持可编辑的参数化边界表示几何体。我们提出了具身CAD，这是基于求解器的LLM代理，用于参数化B-Rep装配建模。该代理不是一次性生成完整的脚本，而是从分层的L0-L4 CAD技能库中迭代选择动作，将其解析为类型化的几何操作，在CAD后端执行这些操作，并利用求解器反馈进行规划、修复和学习。该框架结合了动作语法约束、确定性参数解析和基于求解器的奖励，用于监督热身和GRPO风格的细化。我们使用与求解器对齐的指标（如可执行率、技能准确性、操作族准确性、精确策略准确性和任务完成成功率）对具身CAD在多步骤机械、工业设备和模具导向装配任务上进行了评估。结果表明，基于求解器的规划在当前基准中执行了所有强规划者工作流，而学习控制器达到了高可执行率，并揭示了有效工具调用与精确长远策略预测之间的剩余差距。
cs.AI / 36 / 2606.31285
Spatial Reasoning via Modality Switching Between Language and Symbolic Representation
通过语言与符号表示之间的模态切换进行空间推理
Abstract
Human reasoning is inherently multimodal: when problems become difficult, we rarely think in words alone. We often externalize our reasoning by sketching diagrams or drawing grids to understand the underlying conceptual structure and avoid mistakes. Building on this premise, our research investigates: (a) whether grounding multi-hop textual-spatial stories into geometry-aware modalities, such as layouts or grids, improves reasoning compared to natural language-based inference; and (b) whether a model can decide when to rely on natural language reasoning and when to switch to a structured modality. We address these questions by introducing a switching metric based on trustworthiness and complexity signals, which estimates when grounding a spatial story into structure is likely to improve performance. This takes a first step toward principled modality selection in Large Language Model (LLM) reasoning. Across our settings, switching from natural language-based reasoning to a grid-based representation improves LLM performance by up to 42\%, highlighting the importance of modality choice in shaping reasoning outcomes.
Chinese Translation
人类推理本质上是多模态的：当问题变得困难时，我们很少仅仅用语言思考。我们常常通过绘制图表或绘制网格来外化我们的推理，以理解潜在的概念结构并避免错误。在这一前提下，我们的研究探讨了：(a) 将多跳文本-空间故事嵌入几何感知模态（如布局或网格）是否能改善推理，相较于基于自然语言的推理；以及 (b) 模型是否能够决定何时依赖自然语言推理，何时切换到结构化模态。我们通过引入基于可信度和复杂性信号的切换度量来解决这些问题，该度量估计何时将空间故事嵌入结构可能会提高性能。这是朝着在大型语言模型（LLM）推理中进行原则性模态选择迈出的第一步。在我们的实验设置中，从基于自然语言的推理切换到基于网格的表示使LLM性能提高了最多42%，突显了模态选择在塑造推理结果中的重要性。
cs.AI / 37 / 2606.31308
Benchmarking Large Language Models on Floating-Point Error Classification
大型语言模型在浮点错误分类中的基准测试
Abstract
This paper investigates the capability of Large Language Models (LLMs) to detect and classify floating-point errors statically in software code. We introduce InterFLOPBench, a benchmark of 90 C kernels with 1 130 test samples designed to evaluate LLMs across six categories of floating-point error: cancellation, comparison, division by zero, overflow, underflow and NaN, compared across 14 LLMs. The evaluation framework treats floating-point error detection as a multi-label classification problem and employs the F1-score metric to measure performance. Results demonstrate that latest models (Qwen 3 32b, Gemini 2.5 Flash, Phi 4 Reasoning, DeepSeek R1T2, and gpt-oss 20b and 120b) achieve a performance greater than 0.88 overall F1-score. Performance varies between error categories, between explicit operations such as division by zero (Average F1-score: 0.8479) and more subtle numerical phenomena such as underflow (Average F1-score: 0.6059) and cancellation (Average F1-score: 0.6164).
Chinese Translation
本文研究了大型语言模型（LLMs）在软件代码中静态检测和分类浮点错误的能力。我们引入了InterFLOPBench，这是一个包含90个C语言内核和1130个测试样本的基准，旨在评估LLMs在六类浮点错误（即：取消、比较、除以零、溢出、下溢和NaN）上的表现，并在14个LLMs之间进行比较。评估框架将浮点错误检测视为多标签分类问题，并采用F1分数指标来衡量性能。结果表明，最新的模型（Qwen 3 32b、Gemini 2.5 Flash、Phi 4 Reasoning、DeepSeek R1T2，以及gpt-oss 20b和120b）整体F1分数均超过0.88。不同错误类别之间的性能差异明显，显式操作如除以零的平均F1分数为0.8479，而更微妙的数值现象如下溢（平均F1分数：0.6059）和取消（平均F1分数：0.6164）则表现较差。
cs.AI / 38 / 2606.31325
HistoriQA-ThirdRepublic: Multi-Hop Question Answering Corpus for Historical Research, Parliamentary Debates from the French Third Republic (1870-1940)
HistoriQA-第三共和国：用于历史研究的多跳问答语料库，来自法国第三共和国（1870-1940）的议会辩论
Abstract
We present HistoriQA-ThirdRepublic: a French-language dataset of multi-hop historical questions derived from parliamentary debates and newspapers of the French Third Republic. Designed in collaboration with a historian, the corpus captures complex reasoning patterns typical of historical inquiry, including cross-source synthesis, temporal reasoning, and the integration of sparse evidence. The dataset is made of 1782 questions and emphasizes multi-hop connections across heterogeneous historical documents, providing a resource for evaluating retrieval-augmented and large language model systems in domain-specific contexts. We describe the methodology for constructing the corpus, including the selection and alignment of sources, question validation, and metadata integration. While the dataset focuses on French historical documents, our methodology can be readily adapted to other languages and national corpora. Finally, we demonstrate how the corpus can support realistic evaluation scenarios for multi-hop question answering, bridging the gap between NLP benchmarks and the needs of historical scholarship.
Chinese Translation
我们提出了HistoriQA-第三共和国：一个法语多跳历史问题数据集，来源于法国第三共和国的议会辩论和报纸。该语料库在与历史学家的合作下设计，捕捉了历史研究中典型的复杂推理模式，包括跨源综合、时间推理和稀疏证据的整合。该数据集由1782个问题组成，强调了异构历史文献之间的多跳连接，为在特定领域背景下评估检索增强和大型语言模型系统提供了资源。我们描述了构建语料库的方法，包括来源的选择与对齐、问题验证和元数据整合。尽管该数据集侧重于法语历史文献，但我们的方法可以很容易地适应其他语言和国家语料库。最后，我们展示了该语料库如何支持多跳问答的现实评估场景，弥合自然语言处理基准与历史学研究需求之间的差距。
cs.AI / 39 / 2606.31332
CryoACE: An Atom-centric Framework for Accurate and Automated Model Building in Cryo-EM
CryoACE：一种用于冷冻电子显微镜中准确和自动化模型构建的原子中心框架
Abstract
Protein automodeling from cryo-EM density maps faces unique challenges in enforcing physicochemical validity and managing conformational heterogeneity. Current solvers are often limited to static predictions or require computationally intensive heuristic searches. We present CryoACE, an end-to-end framework that reconstructs precise atomic graphs for both homogeneous and heterogeneous structures. Our method features two key innovations: an atom-centric reconstruction paradigm, where density features are sampled directly at atomic coordinates and iteratively recycled to refine structures, replacing expensive voxel convolutions for efficient multimodal fusion; and a training-free guidance mechanism that leverages predicted local resolution priors to resolve dynamic ambiguity. Validated on a newly constructed high-quality dataset, CryoACE significantly outperforms existing baselines on static benchmarks and, for the first time, unveils atomic-level dynamic conformations on complex real-world datasets like EMPIAR-10345 without relying on pre-built static structures.
Chinese Translation
从冷冻电子显微镜（cryo-EM）密度图中进行蛋白质自动建模面临着在物理化学有效性和构象异质性管理方面的独特挑战。当前的求解器通常仅限于静态预测或需要计算密集型的启发式搜索。我们提出了CryoACE，这是一种端到端框架，能够重建同质和异质结构的精确原子图。我们的方法具有两个关键创新：一种原子中心重建范式，其中密度特征直接在原子坐标处进行采样，并迭代回收以精炼结构，替代了昂贵的体素卷积以实现高效的多模态融合；以及一种无训练的引导机制，利用预测的局部分辨率先验来解决动态模糊性。在新构建的高质量数据集上进行验证，CryoACE在静态基准测试中显著优于现有基线，并首次在复杂的真实世界数据集（如EMPIAR-10345）上揭示了原子级动态构象，而无需依赖于预构建的静态结构。
cs.AI / 40 / 2606.31334
Optimization Algorithms for Joint OFDM Waveform Design and RIS Configuration in 6G Networks: From Convex Relaxation to Foundation Models
6G网络中联合OFDM波形设计与RIS配置的优化算法：从凸松弛到基础模型
Abstract
Joint OFDM-RIS optimization for 6G is a mixed-integer nonlinear programming (MINLP) problem covering sum-rate maximization, energy efficiency, max-min fairness, and peak-to-average power ratio (PAPR)-constrained objectives. Seventy-eight joint OFDM-RIS optimization works published between 2021 and 2026 are surveyed. No standardized benchmark exists, and cross-paper comparisons remain infeasible. This survey classifies these works into four paradigms: (I) model-based convex relaxation, (II) heuristic and metaheuristic search, (III) deep reinforcement and unsupervised learning, and (IV) emerging methods including foundation models (FM), diffusion-based generative AI, and quantum optimization. A literature synthesis of self-reported benchmarks shows that ML-based methods (Paradigm~III) report 95-99\% of model-based spectral efficiency at 10^2-10^4 x faster per-inference runtime (method-pair dependent; literature values are self-reported and exclude ML pre-training cost). A companion tutorial benchmark at N=16, N=64, and N=128 reveals a critical scaling property: GPU-based neural network inference (DDQN, PPO, graph neural network (GNN), unsupervised DL) is N-invariant, with identical runtime at N=16 and N=128, while iterative solvers (AO+SCA, PSO) scale polynomially. Energy efficiency (P2) and PAPR-constrained (P4) benchmarks are deferred to future work with standardized power models and waveform generators. Six open challenges emerge from the synthesis: the cross-paradigm benchmark deficit, real-world hardware-constrained deployment, joint waveform-RIS optimization for doubly-dispersive channels, multi-objective PAPR trade-offs, LLM safety in live network control, and diminishing returns of standalone heuristics. We specify requirements for a standardized benchmark. This study serves as a roadmap for researchers and practitioners working on joint OFDM-RIS optimization in 6G networks.
Chinese Translation
6G中的联合OFDM-RIS优化是一个混合整数非线性规划（MINLP）问题，涵盖了总速率最大化、能量效率、最大-最小公平性以及峰均功率比（PAPR）约束目标。对2021年至2026年间发表的78篇联合OFDM-RIS优化相关的研究进行了调查。目前尚无标准化基准，跨论文比较仍不可行。本调查将这些研究分为四种范式：（I）基于模型的凸松弛，（II）启发式和元启发式搜索，（III）深度强化学习和无监督学习，以及（IV）新兴方法，包括基础模型（FM）、基于扩散的生成性人工智能和量子优化。自报告基准的文献综合显示，基于机器学习的方法（范式III）在每次推理运行时间上以10^2-10^4倍的速度报告了95-99%的基于模型的谱效率（方法对比依赖；文献值为自报告，不包括机器学习的预训练成本）。在N=16、N=64和N=128的伴随教程基准中揭示了一个关键的缩放特性：基于GPU的神经网络推理（DDQN、PPO、图神经网络（GNN）、无监督深度学习）是N不变的，在N=16和N=128时具有相同的运行时间，而迭代求解器（AO+SCA、PSO）则呈多项式缩放。能量效率（P2）和PAPR约束（P4）基准将推迟到未来的工作中，届时将使用标准化的功率模型和波形生成器。从综合中出现了六个开放挑战：跨范式基准缺失、现实世界硬件受限的部署、双散射信道的联合波形-RIS优化、多目标PAPR权衡、LLM在实时网络控制中的安全性，以及独立启发式的收益递减。我们明确了标准化基准的要求。本研究为从事6G网络中联合OFDM-RIS优化的研究人员和实践者提供了路线图。
cs.AI / 41 / 2606.31347
Smart charging of large fleets of Electric Vehicles: Independent Multi-Agent Reinforcement Learning approaches
大型电动车队的智能充电：独立多智能体强化学习方法
Abstract
The electrification of transportation through electric vehicles introduces new challenges for power grid management, such as increased peak demand, voltage fluctuations, line overloads, and the integration of variable renewable energy sources. To enable efficient integration of EVs while minimizing costs for users and avoiding network overloads, implicit coordination between EVs is required. This work compares two independent multi-agent reinforcement learning approaches for optimizing such decentralized EV charging: contextual combinatorial bandits and policy gradient algorithms. Using a realistic simulation environment with autonomous agents making decisions based on local environmental information (including price signals, state-of-charge, and temporal constraints), we evaluate their performance across varying congestion levels, and mixed-strategy configurations with heterogeneous agent groups under dynamic electricity pricing derived from real photovoltaic production data.
Chinese Translation
电动汽车的交通电气化带来了电网管理的新挑战，例如峰值需求增加、电压波动、线路过载以及可变可再生能源的整合。为了在最小化用户成本和避免网络过载的同时，实现电动车（EV）的高效整合，需要在电动车之间进行隐式协调。本研究比较了两种独立的多智能体强化学习方法，以优化这种分散的电动车充电：上下文组合赌博机和策略梯度算法。通过使用一个现实的仿真环境，自动智能体根据局部环境信息（包括价格信号、充电状态和时间约束）做出决策，我们评估了它们在不同拥堵水平下的表现，以及在基于真实光伏生产数据的动态电价下，异构智能体组的混合策略配置。
cs.AI / 42 / 2606.31392
ReGRPO: Reflection-Augmented Policy Optimization for Tool-Using Agents
ReGRPO：用于工具使用代理的反思增强策略优化
Abstract
Tool-augmented vision-language models (VLMs) can solve multimodal, multi-step tasks by calling external tools, yet they remain fragile in practice. Existing works have two common gaps. Supervised fine-tuning (SFT) is built mostly on successful trajectories and offers little signal for recovery after tool failures, while sparse trajectory-level RL rewards provide limited guidance on which step failed and how to repair it. We introduce ReGRPO (Reflection-augmented Group Relative Policy Optimization), a framework that learns reflection-guided correction in tool-using agents. ReGRPO starts with a structured reflective data engine: we execute near-miss actions to collect grounded failure observations, then build Reflection-of-Thought triplets (ErrorType, Evidence, FixPlan) paired with corrected actions for warm-start SFT. We then optimize reflection tokens and corrective actions jointly within local trajectories using group-relative advantages, and include a reflection-cost term to reduce unnecessary reflection. Experiments on GTA and GAIA show that, under the same backbone and tool suite, ReGRPO consistently outperforms strong open-source baselines and achieves the best results among the compared open-source controllers. Code and RoT data are available at https://github.com/showlab/ReGRPO.
Chinese Translation
工具增强的视觉语言模型（VLMs）能够通过调用外部工具解决多模态、多步骤任务，但在实际应用中仍然脆弱。现有研究存在两个共同的不足之处。监督微调（SFT）主要基于成功的轨迹构建，且在工具失败后提供的恢复信号有限，而稀疏的轨迹级强化学习奖励对失败步骤的指导也相对有限。我们提出了ReGRPO（反思增强的群体相对策略优化），这是一个在工具使用代理中学习反思引导修正的框架。ReGRPO首先构建一个结构化的反思数据引擎：我们执行接近失败的动作以收集具体的失败观察，然后构建与修正动作配对的反思三元组（错误类型、证据、修正计划），用于热启动SFT。接着，我们在局部轨迹中使用群体相对优势共同优化反思标记和修正动作，并引入反思成本项以减少不必要的反思。在GTA和GAIA上的实验表明，在相同的骨干网络和工具套件下，ReGRPO始终优于强大的开源基线，并在比较的开源控制器中取得最佳结果。代码和RoT数据可在https://github.com/showlab/ReGRPO获取。
cs.AI / 43 / 2606.31399
World-Model Collapse as a Phase Transition
世界模型崩溃作为相变
Abstract
Water looks unchanged as it warms, then at a critical point it boils. We ask whether long-horizon language agents show an analogous transition in their implicit world models. In some parameter settings, changing state load by a small amount, or adding a single step of horizon, leaves behavior nearly unchanged; near a critical boundary, the same small change causes a sudden world collapse. We study this effect in a deterministic task family with exact per-step gold state. A large grid search over state cardinality, dependency density, horizon, branching, observation mode, and mutation rate reveals a phase diagram: a solved plateau, a narrow transition band, and a collapse floor. Per-step traces show the mechanism: world-state fidelity fails before action validity, so the agent is not merely choosing a bad action; it is acting from a corrupted world. Stronger models translate the critical boundary but do not remove the qualitative transition. These results make world-model collapse a measurable bottleneck for long-horizon agents.
Chinese Translation
水在加热时看似没有变化，但在临界点时会沸腾。我们探讨长时间范围语言代理是否在其隐含的世界模型中表现出类似的转变。在某些参数设置下，微小地改变状态负载或增加一步的视野，行为几乎没有变化；而在临界边界附近，同样的小变化会导致世界的突然崩溃。我们在一个具有精确逐步黄金状态的确定性任务家族中研究这一效应。对状态基数、依赖密度、视野、分支、观察模式和突变率进行的大规模网格搜索揭示了一个相图：一个已解决的平台、一个狭窄的过渡带和一个崩溃底线。逐步跟踪显示了机制：世界状态的保真度在行动有效性之前就失败了，因此代理并不是仅仅选择了一个糟糕的行动；它是从一个被破坏的世界中行动。更强的模型虽然能够转移临界边界，但并未消除这种定性转变。这些结果使得世界模型崩溃成为长时间范围代理的一个可测量瓶颈。
cs.AI / 44 / 2606.31404
Wisdom Of The (AI) Crowd: Investigating Artificial Swarm Intelligence In Large Language Models
群体智慧（人工智能）的智慧：研究大型语言模型中的人工群体智能
Abstract
Human swarm intelligence demonstrates remarkable collective accuracy but faces scalability constraints in cost, coordination, and time. We investigate whether large language models (LLMs) can approximate swarm intelligence effects through artificial swarms, addressing a critical gap in understanding AI-based aggregation mechanisms. We conducted a controlled experiment with 960 manually executed prompts across three proprietary models (GPT-5, Gemini 2.5 Pro, Claude Sonnet 4.5), testing intra-model sampling and inter-model aggregation on eight estimation tasks. Results reveal consistent error reduction through intra- and inter-model aggregation, with significant error reductions up to 37 percentage points in MAPE across different aggregation strategies. We observed small to large effect sizes for positive correlations (Spearman's $\rho=0.242-0.568$, all $p<0.001$) between relative confidence interval widths and relative estimation errors, suggesting LLMs possess metacognitive awareness when assessing uncertainty. We discuss implications for research and practice, providing actionable insights for deploying LLM swarms in organizational decision-making.
Chinese Translation
人类群体智能展现出显著的集体准确性，但在成本、协调和时间方面面临可扩展性限制。我们研究大型语言模型（LLMs）是否能够通过人工群体近似群体智能效应，从而填补对基于人工智能的聚合机制理解的关键空白。我们在三个专有模型（GPT-5、Gemini 2.5 Pro、Claude Sonnet 4.5）上进行了960个手动执行的提示的受控实验，测试了八个估计任务中的模型内部采样和模型间聚合。结果显示，通过模型内部和模型间聚合，误差持续减少，在不同的聚合策略中，平均绝对百分比误差（MAPE）显著降低，最高可达37个百分点。我们观察到相对置信区间宽度与相对估计误差之间存在小到大的正相关效应（Spearman's $
ho=0.242-0.568$，均为 $p<0.001$），这表明LLMs在评估不确定性时具备元认知意识。我们讨论了对研究和实践的影响，为在组织决策中部署LLM群体提供了可行的见解。
cs.AI / 45 / 2606.31410
Xiaomi-GUI-0 Technical Report
Xiaomi-GUI-0 技术报告
Cao, Wanxia, Duan, Chengzhen, Fu, Pei, Gao, Pengzhi, Lian, Niu, Liu, Fazhan, Liu, Hui, Qu, Heng, Wu, Qinzhuo, Yu, Zhehao, Chen, Tongbo, Cui, Shiqi, Du, Anan, Jia, Shukai, Li, Yuanfa, Liu, Yike, Lu, Wenchao, Sun, Haoyuan, Sun, Jiatong, Tan, Cheng, Wang, Yajie, Wu, Changqiao, Xiong, Tao, Yang, Jiahui, Yuan, Yuxuan, Zhang, Ruoceng, Zhang, Shaojie, Zhu, Jian, Luan, Jian, Zou, Cong
Abstract
Graphical user interface (GUI) agents build on vision-language models to complete user tasks end-to-end in real applications through interface actions such as tapping, swiping, text entry, and navigation. However, existing GUI agents are trained and evaluated largely on offline trajectories, simulated environments, and standardized benchmarks. These differ substantially from real applications in interface layout, interaction logic, and abnormal-state distribution, and cannot faithfully characterize execution stability in real-world use, where account states, permission dialogs, payment authentication, and risk control continually reshape the state distribution and open a persistent gap between benchmark scores and real usability. To close this gap, we propose Xiaomi-GUI-0, a native multimodal GUI agent for real mobile environments, trained and evaluated within a real-device closed loop. At its core is a real-device-dominant hybrid infrastructure, where physical devices are the primary execution environment and sandboxes provide auxiliary support, so that data collection, training, rollout, and evaluation share an execution distribution close to real deployment. We construct multi-source training data spanning high-frequency head tasks, high-generalization data for long-tail intents, and capability-enhancement data for reflection and memory, and introduce an error-driven data flywheel that turns failure trajectories into corrected actions, reflective explanations, and recovery demonstrations. The model is trained through a progressive three-stage pipeline of supervised fine-tuning, step-level reinforcement learning, and agentic reinforcement learning. Evaluated on public benchmarks and our in-house RealMobile, Xiaomi-GUI-0 achieves 72.0% success on RealMobile and 78.9% on AndroidWorld, while substantially improving execution stability and abnormal-state recognition in real-world tasks.
Chinese Translation
图形用户界面（GUI）代理基于视觉-语言模型，通过点击、滑动、文本输入和导航等界面操作在真实应用中完成用户任务。然而，现有的GUI代理主要在离线轨迹、模拟环境和标准化基准上进行训练和评估。这些与真实应用在界面布局、交互逻辑和异常状态分布上有显著差异，无法真实表征在实际使用中的执行稳定性，因为账户状态、权限对话框、支付认证和风险控制不断重塑状态分布，从而在基准得分和实际可用性之间形成持续的差距。为了解决这一问题，我们提出了Xiaomi-GUI-0，一种针对真实移动环境的原生多模态GUI代理，在真实设备闭环中进行训练和评估。其核心是以真实设备为主导的混合基础设施，物理设备是主要的执行环境，而沙盒提供辅助支持，使得数据收集、训练、发布和评估共享接近真实部署的执行分布。我们构建了跨越高频头部任务、高泛化数据（用于长尾意图）和能力增强数据（用于反思和记忆）的多源训练数据，并引入了一个以错误驱动的数据飞轮，将失败轨迹转化为纠正行动、反思解释和恢复演示。该模型通过监督微调、逐步强化学习和代理强化学习的渐进式三阶段管道进行训练。在公共基准和我们的内部RealMobile上评估，Xiaomi-GUI-0在RealMobile上取得了72.0%的成功率，在AndroidWorld上达到了78.9%，同时在真实世界任务中显著提高了执行稳定性和异常状态识别能力。
cs.AI / 46 / 2606.31413
Learning to Select, Not Relearn: Hard-Routed Mixtures of Reasoning LoRAs
学习选择，而非重学：硬路由推理 LoRA 的混合
Abstract
Composing independently trained LoRA adapters into a single large language model is useful for multi-domain adaptation, especially when the original training data cannot be shared. A common approach is to use MoE-style routing over LoRA experts, but for frozen pretrained adapters, soft weighted combinations can change the unit-scale additive update under which each LoRA module was originally trained. We propose \textbf{Hard-Routed MoR-LoRA}, a two-stage framework for composing frozen reasoning LoRA experts through unit-scale hard selection. First, domain-specific LoRA adapters are trained independently using reinforcement learning from verifiable feedback to obtain reasoning experts. Then, all experts are frozen, reasoning traces are distilled from them, and only a lightweight shared router together with a small attention LoRA is trained for integration. The router selects exactly one expert per token using hard top-1 routing, while a straight-through estimator enables gradient-based training. Experiments across five benchmarks, multiple model scales, and additional model families show that Hard-Routed MoR-LoRA preserves expert behavior while requiring substantially fewer trainable parameters than soft-routing mixture baselines. Our analysis further shows that normalized soft mixtures often concentrate most routing mass on a single expert, suggesting that hard unit-scale routing provides a simple and efficient abstraction for frozen LoRA expert composition.
Chinese Translation
将独立训练的 LoRA 适配器组合成一个大型语言模型对于多领域适应非常有用，尤其是在原始训练数据无法共享的情况下。常见的方法是对 LoRA 专家使用 MoE 风格的路由，但对于冻结的预训练适配器，软加权组合可能会改变每个 LoRA 模块最初训练时的单位规模加法更新。我们提出了 extbf{硬路由 MoR-LoRA}，这是一个通过单位规模硬选择组合冻结推理 LoRA 专家的两阶段框架。首先，使用来自可验证反馈的强化学习独立训练领域特定的 LoRA 适配器，以获得推理专家。然后，所有专家被冻结，从中提取推理痕迹，并仅训练一个轻量级共享路由器以及一个小型注意力 LoRA 进行集成。该路由器使用硬 top-1 路由为每个 token 精确选择一个专家，同时直通估计器使基于梯度的训练成为可能。在五个基准测试、多个模型规模和额外模型系列的实验中，Hard-Routed MoR-LoRA 保持了专家行为，同时所需的可训练参数显著少于软路由混合基线。我们的分析进一步表明，归一化软混合通常将大部分路由质量集中在单个专家上，这表明硬单位规模路由为冻结的 LoRA 专家组合提供了一种简单而高效的抽象。
cs.AI / 47 / 2606.31420
BP-TTA: Balanced and Prototype-Guided Test-Time Adaptation in Dynamic Scenarios
BP-TTA：动态场景下的平衡与原型引导测试时间适应
Abstract
Test-Time Adaptation (TTA) enables models trained on a source domain to adapt online to unlabeled test data under distribution shifts. While recent TTA methods have moved beyond static settings and begun to consider continual domain shifts, they primarily address distribution drift and fail to account for class imbalance in dynamic scenarios. In real-world test-time streams, class imbalance and continual domain shifts often occur at the same time and interact with each other. In this paper, we propose a novel Balanced and Prototype-Guided Test-Time Adaptation (BP-TTA) method, which combines batch-balanced sampling with prototype-guided adaptation to handle the class imbalance and continual domain shift problems. BP-TTA constructs balanced adaptation batches by integrating current samples with high-confidence historical instances, effectively mitigating bias toward dominant classes and stabilizing online updates. Meanwhile, BP-TTA maintains evolving class prototypes during inference and leverages prototype similarity as a constraint for model adaptation, thereby improving the reliability of pseudo-labels and enhancing the stability of online updates under persistent domain shifts. Extensive experiments demonstrate that BP-TTA consistently outperforms state-of-the-art TTA methods in dynamic test-time streaming settings.
Chinese Translation
测试时间适应（TTA）使得在源领域训练的模型能够在线适应在分布变化下的未标记测试数据。尽管近期的 TTA 方法已超越静态设置，开始考虑持续的领域变化，但它们主要关注分布漂移，而未能考虑动态场景中的类别不平衡。在真实世界的测试时间流中，类别不平衡和持续的领域变化往往同时发生并相互影响。本文提出了一种新颖的平衡与原型引导测试时间适应（BP-TTA）方法，该方法结合了批量平衡采样与原型引导适应，以应对类别不平衡和持续领域变化的问题。BP-TTA 通过将当前样本与高置信度的历史实例结合，构建平衡的适应批次，有效减轻了对主导类别的偏倚，并稳定了在线更新。同时，BP-TTA 在推理过程中维护不断演变的类别原型，并利用原型相似性作为模型适应的约束，从而提高伪标签的可靠性，并增强在持续领域变化下在线更新的稳定性。大量实验表明，BP-TTA 在动态测试时间流设置中始终优于最先进的 TTA 方法。
cs.AI / 48 / 2606.31422
Ask the World Before Acting: Budgeted Environment Probing for World-Model Calibration
在行动前询问世界：用于世界模型校准的预算环境探测
Abstract
Long-horizon language agents do not only choose actions; they carry a private model of the world from one decision to the next. When that model drifts, a later failure can be decided before the failing action is ever taken. We study a direct repair mechanism: before committing to the next task action, an agent may ask the environment about one belief field and write the answer back into its world model. This makes environment interaction a scarce calibration resource, not merely a way to advance the task. We introduce \method, a budgeted probing operator for structured belief tables. The useful probes are not the same everywhere. Procedural beliefs, such as tool dependencies, can often be repaired by targeted checks, but those checks spend steps that the task may need. Spatial beliefs, such as object locations and graph edges, rely more on structural cues; the agent's own confidence can be a poor guide when the world changes off-screen. A type-stratified analysis formalizes this probe-action frontier, and controlled experiments show that mid-planning environment evidence reduces terminal world-model error when the probe policy follows the structure of the task.
Chinese Translation
长时间跨度的语言智能体不仅选择行动；它们在每个决策之间携带一个私有的世界模型。当该模型发生漂移时，后续的失败可能在失败的行动被执行之前就已决定。我们研究了一种直接修复机制：在承诺下一个任务行动之前，智能体可以询问环境关于一个信念领域的问题，并将答案写回其世界模型。这使得环境交互成为一种稀缺的校准资源，而不仅仅是推进任务的一种方式。我们引入了 extit{method}，一种用于结构化信念表的预算探测操作符。有效的探测在不同地方并不相同。程序性信念，例如工具依赖性，通常可以通过针对性的检查进行修复，但这些检查会消耗任务可能需要的步骤。空间信念，例如物体位置和图边，更依赖于结构线索；当世界在屏幕外变化时，智能体自身的信心可能是一个糟糕的指导。类型分层分析形式化了这一探测-行动边界，控制实验表明，当探测策略遵循任务的结构时，中期规划环境证据可以减少终端世界模型误差。
cs.AI / 49 / 2606.31435
CDR-Bench: Evaluating Faithful Execution of Compositional, Order-Sensitive Data Refinement Recipes
CDR-Bench：评估组合性、有序敏感的数据精炼配方的忠实执行
Abstract
Data refinement involves executing multi-step recipes over evolving text states, where both composition and execution order of processing operators determine the outcome. While existing benchmarks either isolate text editing or entangle it with code and tool execution, it remains unclear whether LLMs can directly and faithfully execute these compositional, order-sensitive data refinement recipes. To fill this gap, we introduce CDR-Bench, a comprehensive benchmark featuring 3,462 high-quality tasks spanning four real-world data refinement domains and 29 distinct operators. Our benchmark evaluates models across atomic, order-agnostic, and order-sensitive settings, leveraging deterministic reference outputs to enable exact evaluation. Experiments on 10+ state-of-the-art LLMs reveal consistent failure patterns: performance degrades sharply in compositional settings, and order-sensitive recipe success collapses. These findings underline that current LLMs lack the procedural faithfulness required for reliable compositional data refinement.
Chinese Translation
数据精炼涉及在不断变化的文本状态上执行多步骤配方，其中处理操作的组合和执行顺序决定了结果。虽然现有基准要么孤立文本编辑，要么将其与代码和工具执行纠缠在一起，但尚不清楚大型语言模型（LLMs）是否能够直接且忠实地执行这些组合性、有序敏感的数据精炼配方。为填补这一空白，我们引入了CDR-Bench，这是一个全面的基准，包含3,462个高质量任务，涵盖四个真实世界的数据精炼领域和29种不同的操作符。我们的基准在原子、无序和有序敏感设置中评估模型，利用确定性参考输出以实现精确评估。在10多个最先进的LLMs上的实验揭示了一致的失败模式：在组合性设置中，性能急剧下降，而有序敏感配方的成功率崩溃。这些发现强调了当前的LLMs缺乏可靠的组合数据精炼所需的程序忠实性。
cs.AI / 50 / 2606.31442
Who Determines the Meaning of an Emotion? Affective Sovereignty as an Epistemic Consequence of Measurement Limits
谁来决定情感的意义？情感主权作为测量限制的认识后果
Abstract
Emotion-sensing AI is rapidly becoming embedded in vehicles, home appliances, dialogue agents, and social infrastructure, giving rise to a sphere in which emotion is no longer confined to individual experience but is instead observed and computed at a societal scale, a domain we term the Affectosphere. Yet a central normative question in this domain has remained underexplored: who has the final authority to determine the meaning of one's own emotion? This study addresses the question from the epistemological side of measurement's structural limits. We define a meaning distribution as the distribution of labels assigned by annotators drawn from a population under a fixed annotation protocol, and decompose its uncertainty into reducible and irreducible components. We then demonstrate that, while emotion AI can assign high-confidence point labels and discriminate real differences at an aggregate level, the irreducible component of the meaning distribution for individual instances cannot be estimated with adequate coverage under realistic annotator counts, a systematic divergence we term the epistemic gap. The key finding is that high device confidence does not constitute evidence that irrecoverable meaning has been recovered. From this epistemic gap, together with an explicitly stated normative premise, namely that the output of a system which cannot recover a quantity in principle must not be treated as its authoritative determination, we derive the norm that the final interpretive authority over the meaning of one's emotion is procedurally reserved for the experiencing subject, the norm of affective sovereignty. These results suggest that the design, evaluation, and regulation of emotion AI should place explicit allocation of interpretive authority, rather than accuracy maximisation, at their core.
Chinese Translation
情感感知人工智能正迅速嵌入车辆、家用电器、对话代理和社会基础设施中，形成一个情感不再局限于个人体验的领域，而是在社会层面上被观察和计算的领域，我们称之为情感圈（Affectosphere）。然而，在这个领域中，一个核心的规范性问题仍未得到充分探讨：谁拥有最终的权威来决定个人情感的意义？本研究从测量的结构限制的认识论角度探讨这一问题。我们将意义分布定义为在固定注释协议下，由来自特定人群的注释者分配的标签的分布，并将其不确定性分解为可减少和不可减少的组成部分。我们接着证明，尽管情感人工智能可以在聚合层面上分配高置信度的点标签并区分真实差异，但对于个体实例的意义分布的不可减少组成部分在现实的注释者数量下无法以足够的覆盖率进行估计，这一系统性偏差我们称之为认识差距（epistemic gap）。关键发现是，高设备置信度并不构成不可恢复意义已被恢复的证据。基于这一认识差距，以及一个明确陈述的规范前提，即原则上无法恢复某一量的系统输出不应被视为其权威性决定，我们推导出最终解释个人情感意义的权威程序上应保留给体验主体的规范，即情感主权（affective sovereignty）。这些结果表明，情感人工智能的设计、评估和监管应将解释权的明确分配置于核心，而非最大化准确性。
cs.AI / 51 / 2606.31461
CSTrader: A Testbed for Language-Grounded Trading in a Community-Driven Virtual Asset Market
CSTrader：一个基于语言的社区驱动虚拟资产市场交易测试平台
Abstract
Niche asset markets, such as Counter-Strike 2 (CS2) weapon skins, are small, volatile, and heavily driven by community discussions and platform rules. These properties make them hard for traditional quantitative models, but provide an ideal testbed for studying how large language models (LLMs) turn unstructured text into trading actions. We present CSTrader, a multi-agent framework for language-grounded trading in the CS2 skin market. The system first integrates heterogeneous signals from various sources, then uses specialized agents for technical analysis, liquidity, events, and (reversed) sentiment, and finally applies risk control, transaction friction, and portfolio management agents to produce buy, sell, or hold decisions under realistic trading frictions. We build a live-like evaluation environment with real CS2 data from a highly volatile period and evaluate several recent LLM backbones. Across models, CSTrader consistently outperforms both a falling market index (-15.62%) and simple single-prompt LLM baselines, achieving up to a 7.58% cumulative return with controlled risk. Ablation studies show that liquidity, reversed sentiment, and transaction friction agents are crucial for turning noisy language signals into stable profits, suggesting that niche, language-driven markets are a useful benchmark for future language-to-action research. Code is available at: https://github.com/IatomicreactorI/CSGOTrading?tab=readme-ov-file#quick-start
Chinese Translation
小众资产市场，如《反恐精英2》（Counter-Strike 2, CS2）武器皮肤，具有小规模、高波动性，并且受到社区讨论和平台规则的强烈驱动。这些特性使得传统的定量模型难以适用，但为研究大型语言模型（Large Language Models, LLMs）如何将非结构化文本转化为交易行为提供了理想的测试平台。我们提出了CSTrader，一个用于CS2皮肤市场的基于语言的交易多代理框架。该系统首先整合来自各种来源的异构信号，然后使用专门的代理进行技术分析、流动性、事件和（反向）情绪分析，最后应用风险控制、交易摩擦和投资组合管理代理，在现实的交易摩擦下生成买入、卖出或持有的决策。我们构建了一个类似实时的评估环境，使用来自高度波动时期的真实CS2数据，并评估了几种近期的LLM骨干模型。在各模型中，CSTrader始终优于下跌的市场指数（-15.62%）和简单的单提示LLM基准，达到了最高7.58%的累计回报，并控制了风险。消融研究表明，流动性、反向情绪和交易摩擦代理对于将嘈杂的语言信号转化为稳定的利润至关重要，这表明小众的、以语言驱动的市场是未来语言到行动研究的有用基准。代码可在以下链接获取：https://github.com/IatomicreactorI/CSGOTrading?tab=readme-ov-file#quick-start
cs.AI / 52 / 2606.31470
CLOUDADV: Decision-Aligned Instance Sizing with Zero-Shot Foundation Models under Drift
CLOUDADV：在漂移下使用零样本基础模型进行决策对齐的实例大小调整
Abstract
Cloud virtual machines are often overprovisioned, creating avoidable cost and operational inefficiency. We present CLOUDADV, an interactive engineer-facing advisory system for cloud instance sizing under workload drift. The system combines zero-shot time-series forecasting with bounded recommendation generation across day-, week-, and month-scale planning horizons. For each query, CLOUDADV constructs a structured decision context from historical utilization, forecast summaries, current VM metadata, candidate instance options, pricing, and explicit sizing heuristics. A higher-capacity LLM is used offline to generate reference recommendations, while a smaller production model is evaluated on the same prompts to assess deployment-time alignment under latency and cost constraints. Evaluation prioritizes downstream recommendation quality using simulated Azure cost savings and ex-post exceedance, with rolling-origin forecast accuracy reported as a secondary diagnostic against classical and supervised baselines. In a case study of seven production VMs, the reference recommendations reduce simulated monthly cost from about \$1,503 to \$708, yielding \$795/month in savings (52.9%) under conservative heuristic constraints, while the highest observed exceedance rate among downgraded cases is 1.5%. Although Chronos-2 does not minimize every forecasting metric, it often induces recommendation patterns similar to those of a supervised per-VM baseline. These results suggest that zero-shot foundation models can support decision-aligned provisioning in non-stationary cloud environments while reducing the operational burden of repeated per-tenant retraining, revalidation, and redeployment.
Chinese Translation
云虚拟机通常存在过度配置的问题，导致不必要的成本和运营效率低下。我们提出了CLOUDADV，这是一个面向工程师的交互式建议系统，用于在工作负载漂移下进行云实例大小调整。该系统结合了零样本时间序列预测与基于日、周和月规划周期的有界推荐生成。对于每个查询，CLOUDADV从历史利用率、预测摘要、当前虚拟机元数据、候选实例选项、定价和明确的大小调整启发式中构建一个结构化的决策上下文。一个更高容量的LLM（大语言模型）在离线状态下用于生成参考推荐，而一个较小的生产模型则在相同的提示下进行评估，以评估在延迟和成本约束下的部署时对齐。评估优先考虑下游推荐质量，使用模拟的Azure成本节省和事后超出情况作为指标，同时报告滚动起始预测准确性作为与经典和监督基线的次要诊断。在对七个生产虚拟机的案例研究中，参考推荐将模拟的月成本从约1503美元降低至708美元，节省了795美元/月（52.9%），在保守的启发式约束下，而在降级案例中观察到的最高超出率为1.5%。尽管Chronos-2并未最小化每个预测指标，但它通常会引发类似于监督每个虚拟机基线的推荐模式。这些结果表明，零样本基础模型可以支持在非平稳云环境中进行决策对齐的资源配置，同时减少重复每个租户的再训练、再验证和再部署的运营负担。
cs.AI / 53 / 2606.31478
One Reflection Is Not Enough: Self-Correcting Autonomous Research via Multi-Hypothesis Failure Attribution
一次反思是不够的：通过多假设失败归因实现自我修正的自主研究
Abstract
Autonomous research agents can now draft hypotheses, write code, run experiments, and produce papers, but they remain brittle when experiments fail. Under the prevailing paradigm, failure recovery is usually delegated to a single free-form reflection: a rich trajectory of metrics, logs, and design choices is compressed into one verbal critique, which often leads either to localized trial-and-error or to hard pivots that discard useful context. We propose SAGE, a Self-correcting, Autonomous, Grounded Experimenter, to tackle this failure-recovery bottleneck. Its core mechanism, Multi-Hypothesis Failure Attribution (MHFA), treats recovery as a structured causal diagnosis. By analyzing dynamic trajectory features, MHFA systematically generates multiple evidence-grounded explanations for a failure, independently evaluates their severity, and deterministically routes the verified root cause to the correct intervention level (hypothesis, experimental design, or implementation). To guarantee scientific honesty, SAGE further employs a grounded reporting mechanism that explicitly constrains drafted results to actual measured values, redacting hallucinated numbers. On a 12-topic, 5-domain benchmark, SAGE increases metrics-bearing outputs from 42% to 92% over a reflection baseline, improves artifact quality from 5.00 to 6.75/10, and blindly outscores AI-Scientist-v2 (52.0 vs. 48.2), with gains concentrated in code development and execution. While fully autonomous scientific writing and generating conference-ready papers remain notoriously difficult open problems for the entire field, SAGE successfully produces significantly more reliable and higher-quality scientific artifacts. Ultimately, by coupling structured recovery with explicit grounding constraints, SAGE significantly outperforms monolithic reflection paradigms, establishing a highly trustworthy foundation for future autonomous research.
Chinese Translation
自主研究代理现在可以起草假设、编写代码、进行实验和撰写论文，但在实验失败时仍然显得脆弱。在现有范式下，失败恢复通常被委托给单一的自由形式反思：丰富的指标、日志和设计选择的轨迹被压缩成一个口头批评，这往往导致局部的试错或彻底的转变，从而丢弃有用的上下文。我们提出了SAGE（自我修正的自主基础实验者），以解决这一失败恢复瓶颈。其核心机制是多假设失败归因（Multi-Hypothesis Failure Attribution, MHFA），将恢复视为一种结构化的因果诊断。通过分析动态轨迹特征，MHFA系统地生成多个基于证据的失败解释，独立评估其严重性，并将验证后的根本原因确定性地引导到正确的干预层次（假设、实验设计或实施）。为了保证科学诚信，SAGE进一步采用了一种基础报告机制，明确限制起草结果为实际测量值，删除虚构的数字。在一个包含12个主题、5个领域的基准测试中，SAGE将指标相关输出从42%提高到92%，将工件质量从5.00提升至6.75/10，并在盲测中超越AI-Scientist-v2（52.0对48.2），增益主要集中在代码开发和执行上。尽管完全自主的科学写作和生成会议准备论文仍然是整个领域公认的困难开放问题，但SAGE成功地生成了显著更可靠和高质量的科学工件。最终，通过将结构化恢复与明确的基础约束相结合，SAGE显著优于单一反思范式，为未来的自主研究奠定了高度可信赖的基础。
cs.AI / 54 / 2606.31495
Surprise as a Signal for Plasticity and Metacognition
惊讶作为可塑性和元认知的信号
Abstract
We study a single idea across two settings: that a prediction-error signal, computed by a small predictor over the latent space of a frozen encoder, can serve both as a gate on plasticity and as a substrate for metacognition. In the first system, a non-parametric episodic memory writes a new concept only when this surprise is high, and a periodic offline replay phase consolidates recent traces into a slow linear readout. On a continual stream of 1000 ImageNet classes with a frozen DINOv2 or I-JEPA backbone, the consolidation phase recovers 17.7 points of retention on the oldest classes for DINOv2 and 51.3 points for I-JEPA (single-seed runs), and an ablation shows that replaying only a recent window is worse than no replay at all. In few-shot evaluation the same memory reaches 91.6% on 5-way 1-shot mini-ImageNet, above a task-specific baseline, while a harder 500-way regime exposes the true difficulty. In the second system, the same surprise signal, computed in a shared text-image space, modulates the behaviour of a vision-language model: it answers assertively when a concept is known, hedges when it is partially familiar, and refuses to identify the object and asks for an explanation when it is novel, learning the concept from a single user utterance. The external detector separates known from novel concepts at an AUROC of 0.966 (95% CI +/-0.024), far above the model's own verbalised confidence (0.618), while its token-level confidence sits below chance under greedy decoding; after a sleep phase that empties the fast store, the system recalls 99.2% of fifty taught facts from the consolidated store while a base model recovers none. We report both systems as proof-of-concept, with explicit limitations, and position the second against recent episodic-memory and personalised-VLM work.
Chinese Translation
我们在两个环境中研究一个单一的观点：由一个小型预测器在冻结编码器的潜在空间中计算的预测误差信号，可以同时作为可塑性的门控和元认知的基质。在第一个系统中，非参数的情节记忆仅在惊讶程度较高时写入新概念，而周期性的离线重放阶段将最近的痕迹巩固为缓慢的线性读取。在使用冻结的 DINOv2 或 I-JEPA 主干的 1000 个 ImageNet 类的持续流中，巩固阶段在 DINOv2 的最旧类上恢复了 17.7 个保留点，而在 I-JEPA 上恢复了 51.3 个保留点（单种子运行），而消融实验表明，仅重放最近的窗口效果不如完全不重放。在少量样本评估中，相同的记忆在 5-way 1-shot mini-ImageNet 上达到了 91.6%，高于任务特定的基线，而更困难的 500-way 任务则暴露了真正的难度。在第二个系统中，相同的惊讶信号在共享的文本-图像空间中计算，调节视觉-语言模型的行为：当概念已知时，它会自信地回答；当部分熟悉时，它会有所保留；而当是新颖的概念时，它会拒绝识别该对象并请求解释，从单个用户的发言中学习该概念。外部检测器以 0.966 的 AUROC（95% CI +/-0.024）将已知概念与新颖概念区分开，远高于模型自身的口头信心（0.618），而其令牌级信心在贪婪解码下低于随机水平；经过一个清空快速存储的睡眠阶段后，该系统从巩固存储中回忆起 99.2% 的五十个教授事实，而基线模型则没有恢复任何。我们报告这两个系统作为概念验证，明确其局限性，并将第二个系统与近期的情节记忆和个性化视觉-语言模型研究进行对比。
cs.AI / 55 / 2606.31518
Design and Implementation of Agentic Orchestrations and Orchestration of Agents
代理编排与代理的编排设计与实现
Abstract
Agentic Business Process Management has gained momentum recently. The prospect is that the autonomy of AI agents, i.e., predominantly LLM-based agents, can be balanced with a certain level of robustness, tractability, and traceability through a combination with process technology. In this paper, we provide a classification framework for agentic orchestration options along properties such as task specificity, traceability and tractability, autonomy and reactivity, and correctness assurance and present qualitative decision criteria for realizations of different scenarios. We also provide metrics for the quantitative assessment of realization properties and show them through different agentic implementations of a predictive light sensing scenario. Altogether, this work aims at providing properties, criteria, and metrics for the design and implementation of agentic orchestrations and orchestration of agents.
Chinese Translation
代理业务流程管理最近获得了越来越多的关注。前景是，AI代理的自主性，即主要基于大型语言模型（LLM）的代理，可以通过与流程技术的结合，在一定程度上实现稳健性、可控性和可追溯性的平衡。本文提供了一种代理编排选项的分类框架，涵盖任务特异性、可追溯性和可控性、自主性和反应性、以及正确性保证等属性，并提出了不同场景实现的定性决策标准。我们还提供了用于定量评估实现属性的指标，并通过不同的代理实现预测光感知场景来展示这些指标。总体而言，本研究旨在为代理编排和代理的编排设计与实现提供属性、标准和指标。
cs.AI / 56 / 2606.31532
A time-series classification framework for individual-level absenteeism prediction under severe class imbalance
针对严重类别不平衡的个体缺勤预测的时间序列分类框架
Abstract
Staff absenteeism imposes substantial operational costs in high-demand work environments such as healthcare, emergency services, meat processing, construction, and courier and delivery services, where proactive workforce planning depends on reliable individual-level absence prediction. Existing regression and classification approaches share a structural limitation; they map features observed at time t to labels at the same time t, reproducing already-realised outcomes rather than predicting future events, and discard the sequential behavioural structure inherent in individual attendance histories. We propose a Time Series Classification (TSC) framework that separates historical attendance sequences from future absence labels, enabling genuinely proactive prediction. Due to the lack of public longitudinal attendance data, we construct a reproducible simulated dataset calibrated to the UCI dataset. We analyse Binary Focal Loss (BFL) and Geometric Mean (G-Mean) loss under severe class imbalance using only the imbalance ratio $\rho$. For BFL, the initial gradient ratio is $\rho\alpha/(1-\alpha)$, implying the balanced weight $\alpha = 1/(1+\rho) \approx 0.023$. Experiments show that performance is governed mainly by $\alpha$, with BFL achieving specificity 0.813 and balanced accuracy 0.888, comparable to G-Mean. Unlike BFL, G-Mean adapts automatically without parameter calibration. Among three deep learning architectures evaluated, Long Short-Term Memory (LSTM), Convolutional Neural Network (CNN), and the hybrid LSTM-Fully Convolutional Network (LSTM-FCN), the LSTM-FCN delivers strong precision and specificity. Stable performance is obtained with batch sizes >= 64 and window sizes between 40-80 days, yielding balanced accuracy of approximately 80% on held-out test data.
Chinese Translation
员工缺勤在医疗、紧急服务、肉类加工、建筑以及快递和配送等高需求工作环境中带来了巨大的运营成本，而有效的劳动力规划依赖于可靠的个体缺勤预测。现有的回归和分类方法存在结构性限制；它们将时间 t 观察到的特征映射到同一时间 t 的标签，重现已实现的结果而非预测未来事件，并且忽略了个体出勤历史中固有的序列行为结构。我们提出了一种时间序列分类（Time Series Classification, TSC）框架，该框架将历史出勤序列与未来缺勤标签分开，从而实现真正的主动预测。由于缺乏公共的纵向出勤数据，我们构建了一个可重复的模拟数据集，并对其进行了 UCI 数据集的校准。我们分析了在严重类别不平衡下，仅使用不平衡比率 $
ho$ 的二元焦点损失（Binary Focal Loss, BFL）和几何均值损失（Geometric Mean, G-Mean）。对于 BFL，初始梯度比率为 $
ho rac{eta}{(1-eta)}$，这意味着平衡权重 $eta = rac{1}{(1+
ho)} ext{ 约等于 } 0.023$。实验表明，性能主要受 $eta$ 的影响，BFL 达到特异性 0.813 和均衡准确率 0.888，性能与 G-Mean 相当。与 BFL 不同，G-Mean 无需参数校准即可自动适应。在评估的三种深度学习架构中，长短期记忆网络（Long Short-Term Memory, LSTM）、卷积神经网络（Convolutional Neural Network, CNN）以及混合 LSTM-全卷积网络（LSTM-Fully Convolutional Network, LSTM-FCN），LSTM-FCN 提供了强大的精确度和特异性。在批量大小 >= 64 和窗口大小在 40-80 天之间的情况下，获得了稳定的性能，在保留的测试数据上实现了约 80% 的均衡准确率。
cs.AI / 57 / 2606.31543
Modality-Driven Search with Holistic Trace Judging for ARC-AGI-2
基于模态驱动搜索与整体追踪评估的ARC-AGI-2求解器
Abstract
Large language models can produce fluent, internally coherent reasoning traces for abstract reasoning tasks while still being confidently wrong - making selection among candidates, not just generation, the central challenge. I present a solver for ARC-AGI-2, a few-shot visual reasoning benchmark, built around two principles: (i) treating reasoning modalities as search operators, generating diverse candidates independently across text, image, and code channels, and (ii) context-preserving holistic judging, in which a judge model jointly compares all candidate reasoning traces within a single long-context prompt. Unlike self-consistency or majority voting, this approach reliably recovers correct minority hypotheses on tasks where the modal answer is wrong. On the ARC Prize semi-private evaluation set, the solver achieves 72.9 percent at USD 38.99 per task - the highest score on the verified leaderboard at the time of writing, exceeding the best standalone frontier models, GPT-5.2 Pro at 54.2 percent and Gemini 3 Pro at 54.0 percent, by +18.7 percentage points. On the public evaluation set, it achieves 76.1 percent at USD 19.69 per task. I release the full source code and document extensive negative results, including the finding that prescriptive prompting templates and iterative refinement systematically reduce hypothesis diversity and degrade performance.
Chinese Translation
大型语言模型能够为抽象推理任务生成流畅且内部一致的推理追踪，但仍可能自信地错误——因此在候选者之间的选择，而不仅仅是生成，成为了中心挑战。我提出了一种针对ARC-AGI-2的求解器，这是一项少量样本的视觉推理基准，基于两个原则构建：(i) 将推理模态视为搜索操作符，独立地在文本、图像和代码通道中生成多样的候选者，以及 (ii) 保持上下文的整体评估，其中评估模型在单个长上下文提示中共同比较所有候选推理追踪。与自一致性或多数投票不同，这种方法在模态答案错误的任务上可靠地恢复正确的少数假设。在ARC Prize半私有评估集上，该求解器以每个任务38.99美元的成本达到了72.9%的准确率——在撰写时这是验证排行榜上的最高分，超过了最佳独立前沿模型，GPT-5.2 Pro的54.2%和Gemini 3 Pro的54.0%，高出18.7个百分点。在公共评估集上，它以每个任务19.69美元的成本达到了76.1%的准确率。我发布了完整的源代码，并记录了大量的负面结果，包括发现规定性提示模板和迭代优化系统性地减少了假设的多样性并降低了性能。
cs.AI / 58 / 2606.31564
ACE: Pluggable Adaptive Context Elasticizer across Agents
ACE：可插拔的自适应上下文弹性器跨代理
Abstract
The increasing complexity of agentic tasks has led to rapidly growing trajectory lengths, which poses significant challenges for large language model (LLM) based agents with fixed context windows. Existing context management techniques, such as truncation and summarization, suffer from inherent inflexibility and irreversibility: once information is discarded or compressed, it cannot be recovered even when it becomes critically relevant in later decision steps. To address these limitations, we propose the Adaptive Context Elasticizer (ACE), a plug-and-play module that elastically orchestrates historical step information into the agent's context at each decision step. ACE maintains a lossless message maintenance layer that stores both raw messages and compressed abstractions for each historical step, while a context orchestration layer adaptively assigns each step an elastic type as raw, abstract, or drop, at every decision step based on the current task state. This reversible design ensures that the main LLM always receives a compact yet information-rich context. We adapt ACE to four diverse agent frameworks, including ReAct, DeepAgent, WebThinker, and MiroFlow, without training or architectural modifications. Experiments show that ACE consistently outperforms truncation and summarization baselines, and brings consistent performance gains across all four agent frameworks.
Chinese Translation
代理任务的复杂性不断增加，导致轨迹长度迅速增长，这对基于大型语言模型（LLM）的代理在固定上下文窗口下提出了重大挑战。现有的上下文管理技术，如截断和摘要，存在固有的缺乏灵活性和不可逆性：一旦信息被丢弃或压缩，即使在后续决策步骤中变得至关重要，也无法恢复。为了解决这些局限性，我们提出了自适应上下文弹性器（ACE），这是一个即插即用的模块，能够在每个决策步骤中弹性地将历史步骤信息整合到代理的上下文中。ACE 维护一个无损消息维护层，存储每个历史步骤的原始消息和压缩抽象，同时上下文编排层根据当前任务状态，在每个决策步骤中自适应地为每个步骤分配原始、抽象或丢弃的弹性类型。该可逆设计确保主 LLM 始终接收到一个紧凑且信息丰富的上下文。我们将 ACE 适配于四个不同的代理框架，包括 ReAct、DeepAgent、WebThinker 和 MiroFlow，而无需进行训练或架构修改。实验表明，ACE 始终优于截断和摘要基线，并在所有四个代理框架中带来了持续的性能提升。
cs.AI / 59 / 2606.31575
Which Tokens Matter? Adaptive Token Selection for RLVR with the Relative Surprisal Index
哪些标记重要？基于相对惊讶指数的自适应标记选择用于强化学习可验证奖励
Abstract
Reinforcement learning (RL) has become a powerful tool for propelling Large Language Models (LLMs) beyond imitation-based training towards more robust reasoning capabilities. Among existing approaches, RL with Verifiable Rewards (RLVR) has emerged as a pivotal paradigm for advancing LLM reasoning. Despite its empirical success, recent studies have offered different insights. One line of inquiry advocates prioritizing high-entropy token positions during training, while another perspective cautions against allowing low-probability tokens to dominate gradient updates. Notably, although high-entropy tokens are usually correlated with low probability, both paradigms empirically yield substantial performance gains. In this work, we argue that evaluating sampled-token probability or entropy in isolation is insufficient to capture the policy optimization dynamics. To resolve this tension, we introduce the Relative Surprisal Index (RSI), a principled, information-theoretic metric that naturally couples the token's entropy with the probability of the selected token. We show that, under mild conditions, RSI is related to the local ratio between the first-order variations of the logit-gradient norm and predictive entropy under a selected-logit perturbation. Building on RSI, we propose RSI Selection (RSI-S), an entropy-adaptive token filtering method that retains tokens within a stable RSI interval. RSI-S successfully reconciles previous contradictory paradigms and filters out both redundant low-surprisal tokens and unstable high-surprisal tail tokens. Empirical evaluations show that RSI-S achieves higher avg@32 accuracy across different model scales (Qwen2.5-1.5B, 3B, and 7B) on AIME and AMC benchmarks: RSI-S improves avg@32 accuracy by 2--3 percentage points over GRPO. Overall, RSI offers a promising perspective for RLVR improvement.
Chinese Translation
强化学习（RL）已成为推动大型语言模型（LLMs）从基于模仿的训练走向更强大推理能力的有力工具。在现有的方法中，带有可验证奖励的强化学习（RLVR）已成为推动LLM推理的关键范式。尽管其在实证上取得了成功，但最近的研究提供了不同的见解。一方面的研究主张在训练过程中优先考虑高熵标记位置，而另一方面的观点则警告不要让低概率标记主导梯度更新。值得注意的是，尽管高熵标记通常与低概率相关，但这两种范式在实证上都能带来显著的性能提升。在本研究中，我们认为单独评估采样标记的概率或熵不足以捕捉策略优化的动态。为了解决这一矛盾，我们引入了相对惊讶指数（RSI），这是一个原则性的信息论度量，它自然地将标记的熵与所选标记的概率结合在一起。我们展示了在温和条件下，RSI与在选定逻辑扰动下的logit梯度范数和预测熵的一阶变化的局部比率相关。基于RSI，我们提出了RSI选择（RSI-S），这是一种熵自适应的标记过滤方法，能够保留在稳定RSI区间内的标记。RSI-S成功地调和了之前相互矛盾的范式，并过滤掉冗余的低惊讶标记和不稳定的高惊讶尾部标记。实证评估表明，RSI-S在AIME和AMC基准测试中，在不同模型规模（Qwen2.5-1.5B、3B和7B）上实现了更高的avg@32准确率：RSI-S比GRPO提高了2-3个百分点的avg@32准确率。总体而言，RSI为RLVR的改进提供了一个有前景的视角。
cs.AI / 60 / 2606.31616
Scientific Explanations in Health Sciences: Causality, Trust, and Epistemic Adequacy
健康科学中的科学解释：因果关系、信任与认识论的充分性
Abstract
Medical Artificial Intelligence (AI) is widely expected to transform clinical practice, yet the decision-making processes of many Machine Learning (ML) models remain opaque. Explainability has been advanced as a partial remedy to clarify why AI generates predictions, particularly in high-stakes contexts. Despite ongoing efforts, debates on what constitutes an adequate medical explanation remain unsettled. Yet, explanation has long been a central topic of inquiry in the philosophy of science and medicine. The insights developed in these fields, however, have been largely overlooked in contemporary explainable AI (XAI) research, leaving its foundational assumptions insufficiently examined. To address this gap, this paper develops a critical review at the intersection of philosophy of science and XAI. It examines prevailing accounts of what counts as an explanation in the health sciences and assesses their adequacy for informing XAI in medicine, arguing that they provide necessary conditions for a philosophically grounded approach to explainability in this domain. Building on this foundational philosophical literature, the discussion identifies three central axes of analysis: the role of causality in medical reasoning, the epistemic and relational dimensions of medical trust, and the criteria of explanatory adequacy as shaped by the pragmatic needs of diverse stakeholders. By integrating philosophical analysis with current developments in medical AI, the paper outlines principles for designing XAI systems that offer explanations that are not only epistemically robust but also aligned with the epistemic and practical requirements of clinical decision-making, shaping ongoing debates in medical XAI toward underexplored conceptual foundations.
Chinese Translation
医疗人工智能（AI）被广泛期望能够改变临床实践，但许多机器学习（ML）模型的决策过程仍然不透明。可解释性被提出作为一种部分补救措施，以澄清AI为何生成预测，特别是在高风险的情境中。尽管持续努力，关于什么构成充分的医学解释的争论仍未解决。然而，解释长期以来一直是科学哲学和医学研究的核心主题。然而，这些领域所发展的见解在当代可解释人工智能（XAI）研究中大多被忽视，导致其基础假设未得到充分审视。为了解决这一空白，本文在科学哲学与XAI的交叉点上进行批判性回顾。它考察了健康科学中关于什么算作解释的主流观点，并评估其对医学XAI的适用性，认为这些观点为在该领域建立哲学基础的可解释性提供了必要条件。在这一基础哲学文献的基础上，讨论确定了三个核心分析轴心：因果关系在医学推理中的作用、医学信任的认识论和关系维度，以及由不同利益相关者的务实需求所塑造的解释充分性标准。通过将哲学分析与医疗AI的当前发展相结合，本文概述了设计XAI系统的原则，这些系统提供的解释不仅在认识论上是稳健的，而且与临床决策的认识论和实践要求相一致，从而推动医学XAI中正在进行的辩论朝向未被充分探索的概念基础发展。
cs.AI / 61 / 2606.31648
Think in English, Answer in Korean: Efficient Adaptation of Multilingual Tool-Using Agents
用英语思考，用韩语回答：多语言工具使用代理的高效适应
Abstract
We present LuckyStar 111B, a 111B-parameter hybrid reasoning model developed through a collaboration between Cohere and LG CNS for Korean-English enterprise agents under practical memory and serving constraints. The model trains from Cohere's fully post-trained Command A model rather than a new pretraining run, and uses preamble conditioning to switch between concise non-reasoning behavior and longer tool-oriented reasoning. We study four choices for scaling tool-using agents efficiently: multilingual supervised fine-tuning, reinforcement learning with verifiable rewards for multi-step tool-use tasks, language-consistency rewards for Korean user-facing responses, and 4-bit quantization for single-GPU serving. The adapted model improves mathematical reasoning, function calling, and agentic natural-language-to-SQL (NL2SQL) performance while preserving general Korean and English instruction-following quality. These results provide a practical recipe and failure-mode analysis for adapting post-trained multilingual models to verifiable agentic workflows under memory-constrained deployment.
Chinese Translation
我们提出了LuckyStar 111B，这是一个由Cohere与LG CNS合作开发的111B参数混合推理模型，旨在满足在实际内存和服务限制下的韩英企业代理需求。该模型基于Cohere完全后训练的Command A模型进行训练，而不是从新的预训练开始，并采用前言条件化在简洁的非推理行为与较长的工具导向推理之间切换。我们研究了四种高效扩展工具使用代理的选择：多语言监督微调、针对多步骤工具使用任务的可验证奖励强化学习、针对韩语用户响应的语言一致性奖励，以及用于单GPU服务的4位量化。经过适应的模型在数学推理、函数调用和代理自然语言到SQL（NL2SQL）性能方面有所提升，同时保持了对韩语和英语指令的整体遵循质量。这些结果为在内存受限的部署环境下，将后训练的多语言模型适应于可验证的代理工作流提供了实用的方案和失败模式分析。
cs.AI / 62 / 2606.31651
FARS: A Fully Automated Research System Deployed at Scale
FARS：一个全面自动化的研究系统，规模化部署
Abstract
Recent automated research systems show that language-model agents can generate hypotheses, run experiments, and write complete manuscripts, but most evidence still comes from selected examples, human-framed topics, or a few pre-defined research tasks. We present FARS (Fully Automated Research System), a fully automated AI-for-AI research system designed to operate across research topics at scale. FARS autonomously generates and advances projects through ideation, planning, experimentation, and writing, using stage-specific agents coordinated through a shared workspace that records proposals, code, logs, results, and manuscripts. In its first public deployment, FARS produced 166 complete research papers spanning 67 fine-grained AI/ML topics while preserving intermediate artifacts as an auditable corpus rather than a curated set of successes. We evaluate this corpus with 282 structured reviews from volunteer reviewers covering 140 papers, including overall ratings, sub-scores, integrity checks, and LLM-use disclosure. The reviews indicate that FARS can produce review-worthy and occasionally strong AI/ML research artifacts in a large-scale public deployment, while also exposing recurring failure modes in narrow experimental scope, methodological limitations, and integrity issues.
Chinese Translation
近期的自动化研究系统显示，语言模型代理能够生成假设、进行实验并撰写完整的手稿，但大多数证据仍来自于选定的示例、人为框架的主题或少数预定义的研究任务。我们提出了FARS（Fully Automated Research System），这是一个旨在跨研究主题大规模运行的全面自动化的AI-for-AI研究系统。FARS自主生成并推进项目，通过构思、规划、实验和写作，使用通过共享工作空间协调的阶段特定代理，该工作空间记录提案、代码、日志、结果和手稿。在其首次公开部署中，FARS产生了涵盖67个细分AI/ML主题的166篇完整研究论文，同时将中间成果保留为可审计的语料库，而非经过策划的成功案例集。我们通过282份来自志愿审稿人的结构化评审对该语料库进行了评估，涵盖了140篇论文，包括总体评分、子评分、完整性检查和LLM使用披露。评审结果表明，FARS能够在大规模公开部署中产生值得评审的，偶尔强有力的AI/ML研究成果，同时也暴露了在狭窄实验范围、方法论限制和完整性问题上的反复失败模式。
cs.AI / 63 / 2606.31711
Arena-T2I Hard: Benchmarking and Improving Faithfulness with Dependency-Aware Checklist
Arena-T2I Hard：基于依赖感知检查表的可信度基准与提升
Abstract
Faithfulness -- how precisely a generated image aligns with its prompt -- is increasingly central to the real-world utility of text-to-image (T2I) models. Existing faithfulness benchmarks, however, rely on simple atomic instructions, on which top-tier systems already achieve near-perfect scores. As T2I models enter creative workflows, users issue multi-faceted requests combining intricate spatial relationships, stylistic constraints, and complex text rendering. In this setting, a single binary VLM-judge score no longer captures which specific constraints the model fails to satisfy. We introduce Arena-T2I Hard, a 310-prompt stress benchmark drawn from real arena T2I logs, with approximately 30 decomposed yes/no constraints per prompt spanning six categories, including text rendering. The strongest closed-source system we evaluate reaches 0.855 with a 33~pp performance gap across 11 systems, demonstrating substantial discriminative power. Moreover, high public-arena rankings fail to predict faithfulness, confirming that holistic Bradley-Terry (BT) preference scores prioritize aesthetics over fine-grained prompt adherence. We propose a dependency-aware checklist reward that decomposes each prompt into a DAG of yes/no questions and zeroes descendants of failed parents, turning faithfulness into a per-constraint training signal. Combined with a BT aesthetic reward via group-decoupled normalization (GDPO), which standardizes each reward within its rollout group so neither collapses, the recipe attains a strictly better faithfulness-aesthetics trade-off on SD3.5-Medium and FLUX.1-dev under MMRB2 pairwise comparisons than every single-reward, naive weighted-sum, or 4-reward BT-ensemble baseline.
Chinese Translation
可信度——生成图像与其提示的精确对齐程度——在文本到图像（T2I）模型的实际应用中日益重要。然而，现有的可信度基准依赖于简单的原子指令，而顶尖系统在这些指令上已经达到了近乎完美的评分。随着T2I模型进入创意工作流程，用户发出结合复杂空间关系、风格约束和复杂文本渲染的多方面请求。在这种情况下，单一的二元VLM-judge评分无法捕捉模型未能满足的具体约束。我们引入了Arena-T2I Hard，这是一个由真实竞技场T2I日志提取的310个提示的压力基准，每个提示包含大约30个分解的是/否约束，涵盖六个类别，包括文本渲染。我们评估的最强闭源系统在11个系统中达到了0.855的评分，表现出显著的区分能力。此外，高公共竞技场排名未能预测可信度，确认整体的Bradley-Terry（BT）偏好评分优先考虑美学而非细致的提示遵循。我们提出了一种依赖感知检查表奖励，将每个提示分解为一个有向无环图（DAG）的是/否问题，并将失败父节点的后代归零，将可信度转变为每个约束的训练信号。结合通过组解耦归一化（GDPO）实现的BT美学奖励，该方法在MMRB2成对比较下，在SD3.5-Medium和FLUX.1-dev上达到了比任何单一奖励、天真的加权和或4奖励BT集成基线更优的可信度与美学权衡。
cs.AI / 64 / 2606.31763
A Self-Evolving Agentic System for Automated Generation and Execution of Biological Protocols
自我演化的代理系统用于生物协议的自动生成与执行
Abstract
Autonomous wet-lab experimentation requires more than plausible protocol text: biological intent, quantitative procedures, device constraints and experimental feedback must remain aligned from protocol and SOP design to code and physical execution. We developed ProtoPilot, a self-evolving multi-agent system, together with an expert-grounded benchmark and evaluation framework for testing this conversion as an experimental automation problem. The framework spans 294 synthetic-biology and molecular-biology tasks derived from 98 gold-standard protocols, wet-lab expert rubrics, device-level validity gates and real experimental tests. ProtoPilot incorporates layer-wise verifiability, multi-agent orchestration and a runtime-updated skill library to generate protocols, expand SOPs, synthesize SDK-compliant code and revise workflows from wet-lab feedback. It achieved a Top@3 expert-preference rate of 90.2%, an overall protocol-to-code gate pass rate of 89.5% and an Opentrons pass rate of 88.24%, compared with 32.35% for OpenTrons-AI. Wet-lab validation produced interpretable readouts, Sanger-confirmed products and feedback-corrected PCA-assembled DNA targets, establishing a verifiable route to autonomous experimentation. Together, these results show that the evaluation framework captures execution-relevant requirements for autonomous wet-lab automation, and that ProtoPilot can meet them by converting protocol and code generation into validated execution and feedback-guided revision.
Chinese Translation
自主的湿实验室实验不仅需要合理的协议文本：生物意图、定量程序、设备约束和实验反馈必须在协议和标准操作程序（SOP）设计到代码和实际执行的过程中保持一致。我们开发了ProtoPilot，一个自我演化的多代理系统，并建立了一个基于专家的基准和评估框架，以将这一转化视为实验自动化问题进行测试。该框架涵盖了来自98个黄金标准协议的294个合成生物学和分子生物学任务、湿实验室专家评分标准、设备级有效性门和真实实验测试。ProtoPilot结合了层级可验证性、多代理协调和运行时更新的技能库，以生成协议、扩展SOP、合成符合SDK的代码并根据湿实验室反馈修订工作流程。与OpenTrons-AI的32.35%相比，其在专家偏好率Top@3上达到了90.2%，协议到代码的整体通过率为89.5%，Opentrons的通过率为88.24%。湿实验室验证产生了可解释的读数、经Sanger确认的产品和反馈修正的PCA组装DNA目标，建立了通向自主实验的可验证路径。这些结果表明，评估框架捕捉了自主湿实验室自动化的执行相关要求，而ProtoPilot能够通过将协议和代码生成转化为经过验证的执行和反馈指导的修订来满足这些要求。
cs.AI / 65 / 2606.31800
Evo-PI: Aligning Medical Reasoning via Evolving Principle-Guided Supervision
Evo-PI：通过演变原则引导监督对齐医学推理
Abstract
Despite recent progress, the reasoning capabilities of large multimodal language models (MLLMs) remain fundamentally constrained by static supervision, where fixed prompts, rules, or reward models provide non-adaptive guidance throughout training. Such static signals are often sufficient to enforce output formats, but fail to shape the underlying reasoning process, leading to brittle generalization and performance saturation in complex decision-making tasks. We propose Evo-PI, a principle-centric learning framework that treats reasoning principles as explicit, language-based supervision signals that can be generated, evaluated, and iteratively evolved. Instead of relying on fixed rewards, Evo-PI enables a co-evolutionary loop in which principles guide model reasoning, while model behaviors in turn refine the principles that supervise them. This dynamic alignment mechanism allows supervision to progressively adapt to the model's reasoning deficiencies. We instantiate Evo-PI in medical visual question answering as a high-stakes testbed requiring structured visual-textual reasoning. Across eight benchmarks and multiple model backbones, Evo-PI consistently improves reasoning accuracy, achieving gains of up to 24.6%. Our results suggest that evolving principle-guided supervision offers a scalable and general paradigm for training expert-aligned reasoning in MLLMs. Code is available at https://github.com/zhengxianda/Evo_PI.
Chinese Translation
尽管近年来取得了进展，大型多模态语言模型（MLLMs）的推理能力仍然受到静态监督的根本限制，在这种监督下，固定的提示、规则或奖励模型在训练过程中提供非自适应的指导。这种静态信号通常足以强制输出格式，但未能塑造潜在的推理过程，导致在复杂决策任务中的脆弱泛化和性能饱和。我们提出了Evo-PI，一个以原则为中心的学习框架，将推理原则视为明确的、基于语言的监督信号，这些信号可以生成、评估并迭代演变。Evo-PI不依赖于固定的奖励，而是启用一个共同演化循环，其中原则引导模型推理，而模型行为反过来又细化监督它们的原则。这种动态对齐机制允许监督逐步适应模型的推理缺陷。我们在医学视觉问答中实例化Evo-PI，作为一个需要结构化视觉-文本推理的高风险测试平台。在八个基准和多个模型骨干上，Evo-PI始终提高推理准确性，取得了高达24.6%的提升。我们的结果表明，演变的原则引导监督为训练与专家对齐的MLLM推理提供了一种可扩展且通用的范式。代码可在 https://github.com/zhengxianda/Evo_PI 获取。
cs.AI / 66 / 2606.31801
RAISE: LLM-based Automated Heuristic Design with Robust Adversary Instance Search
RAISE：基于大型语言模型的自动启发式设计与鲁棒对抗实例搜索
Abstract
Automated Heuristic Design (AHD) with Large Language Models (LLMs) has shown remarkable progress in discovering high-quality heuristics. However, existing LLM-based AHD methods optimize heuristics for a fixed training instance set and may fail catastrophically when deployed under real-world distributional shifts. We propose Robust Adversary Instance Search (RAISE), a framework that integrates constrained worst-case instance search within a principled neighborhood of the training distribution into the LLM-based evolutionary search loop. RAISE treats robust AHD as a constrained adversarial instance search problem: the outer loop evolves heuristics via LLM operators, while an LLM-free inner loop efficiently identifies hard instances within an epsilon-ball around the training instance set using a basis distribution parameterization with boundary projection. Comprehensive experiments on Online Bin Packing (OBP), Online Job Shop Scheduling (OJSP), and Online Vehicle Routing (OVRP) across five distribution families demonstrate that existing LLM-based AHD methods degrade by up to 19 times under distribution shift, while RAISE consistently maintains strong performance across all tested distributions and problem scales
Chinese Translation
基于大型语言模型（LLMs）的自动启发式设计（AHD）在发现高质量启发式算法方面取得了显著进展。然而，现有的基于LLM的AHD方法仅针对固定的训练实例集进行优化，当在真实世界的分布变化下部署时，可能会出现灾难性的失败。我们提出了鲁棒对抗实例搜索（RAISE），这是一个框架，将受限的最坏情况实例搜索集成到基于LLM的进化搜索循环中的训练分布的原则性邻域内。RAISE将鲁棒AHD视为一个受限的对抗实例搜索问题：外部循环通过LLM操作符进化启发式，而一个不依赖于LLM的内部循环则利用边界投影的基础分布参数化，在训练实例集周围的ε球内高效识别困难实例。在五个分布族上对在线装箱（OBP）、在线作业车间调度（OJSP）和在线车辆路径规划（OVRP）进行的全面实验表明，现有的基于LLM的AHD方法在分布变化下性能下降高达19倍，而RAISE在所有测试的分布和问题规模中始终保持强劲的性能。
cs.AI / 67 / 2606.31808
Large Databases Need Small, Open-Weight Language Models
大型数据库需要小型开放权重语言模型
Abstract
Language model systems built around proprietary APIs often operate on a token-based cost model. This becomes prohibitively expensive in the context of large databases, where LM-enhanced relational operators can incur costs exceeding $10,000 for a single set of experiments, hindering thorough research and practical deployment. In this paper, we demonstrate that quantized, open-weight models running locally on just 16GB of VRAM can match or exceed the accuracy of closed-source counterparts at lower latency and a fraction of the price, challenging the prevailing assumption that closed-source LM APIs are necessary for effective LM-database integration. We present and analyze the key system optimizations required to efficiently deploy these open-weight models within an LM-DB system. By integrating these local models into the BlendSQL v0.1.0 framework, we demonstrate a 390x reduction in overall costs and 3.8x reduction in latency compared to a proprietary LM API. We make our code available at https://github.com/CapitalOne-Research/play-by-the-type-rules/tree/main/sembench.
Chinese Translation
围绕专有API构建的语言模型系统通常采用基于令牌的成本模型。在大型数据库的背景下，这种模式的成本变得过于昂贵，其中语言模型增强的关系运算符可能导致单次实验的费用超过10,000美元，从而阻碍了深入研究和实际应用。在本文中，我们展示了在仅需16GB显存的本地运行的量化开放权重模型能够在较低延迟和极低成本的情况下匹配或超越闭源模型的准确性，挑战了闭源语言模型API对于有效语言模型-数据库集成的必要性这一普遍假设。我们提出并分析了在语言模型-数据库系统中高效部署这些开放权重模型所需的关键系统优化。通过将这些本地模型集成到BlendSQL v0.1.0框架中，我们展示了与专有语言模型API相比，总体成本降低了390倍，延迟降低了3.8倍。我们的代码可在 https://github.com/CapitalOne-Research/play-by-the-type-rules/tree/main/sembench 获取。
cs.AI / 68 / 2606.31819
Creating Intelligence: A Computational Foundation for AGI
创造智能：通用人工智能的计算基础
Abstract
This work introduces a new computational theory of mind grounded in set theory and hyperdimensional computing. Whereas traditional neural networks rely on continuous weights and matrix multiplication, this framework works with sparse binary data. It represents information as discrete sets, directly modeling biological neural population codes. I demonstrate that associative memory emerges naturally from network topologies featuring a combinatorially expanded hidden layer. Learning is driven by topological plasticity rather than scalar weight adjustments. This architecture unifies auto-associative and hetero-associative learning under a single core algorithm: information retrieval via subset pattern matching and exact nearest-neighbor search. Operating with constant-time complexity, these mechanisms bridge perceptual data (sparse distributed representations) and symbols (sparse holographic representations) without continuous bottlenecks. Mapping this framework to neuroanatomy, I propose that both the cerebellum and the neocortex implement variants of this algorithm, making subset pattern matching the fundamental engine of cognition. Because it relies on discrete logic rather than matrix arithmetic, this algorithm translates directly into in-memory hardware. This opens a new route toward synthetic intelligence with human-level energy efficiency.
Chinese Translation
本研究提出了一种基于集合理论和超维计算的新计算心智理论。传统神经网络依赖于连续权重和矩阵乘法，而该框架则使用稀疏二进制数据。它将信息表示为离散集合，直接建模生物神经群体编码。我证明了联想记忆自然地从具有组合扩展隐藏层的网络拓扑中涌现。学习是由拓扑可塑性驱动的，而不是标量权重调整。该架构将自联想学习和异联想学习统一在一个核心算法之下：通过子集模式匹配和精确最近邻搜索进行信息检索。这些机制以常数时间复杂度运行，连接感知数据（稀疏分布表示）和符号（稀疏全息表示），而没有连续瓶颈。将该框架映射到神经解剖学，我提出小脑和新皮层都实现了该算法的变体，使得子集模式匹配成为认知的基本引擎。由于它依赖于离散逻辑而非矩阵算术，该算法可以直接转化为内存硬件。这为实现具有与人类相当的能效的合成智能开辟了一条新路径。
cs.AI / 69 / 2606.31820
Adaptive Cluster-First Route-Second Decomposition for Industrial-Scale Vehicle Routing
工业规模车辆路径规划的自适应聚类优先-路径次优分解
Abstract
Large-scale capacitated vehicle routing problems (CVRPs) are commonly addressed using cluster-first route-second (CFRS) approaches that split a routing instance into smaller, computationally tractable subproblems. Existing splitting methods typically rely on fixed partitioning rules, predefined optimization objectives, or learned policies, which may perform inconsistently across instances exhibiting different spatial, demand, and operational characteristics. In this work, we propose an adaptive CFRS system that formulates a decomposition procedure as an iterative decision-making process. Motivated by the recent success of large language models (LLMs) in reasoning and tool selection, the system employs an LLM as a high-level decision maker that analyzes the evolving decomposition state and selectively applies further clustering, balancing, and refinement operators. The proposed algorithm jointly partitions customers and vehicles, enabling capacity-aware clustering while adapting partitioning decisions to the characteristics of each problem. We evaluate the approach on synthetic and benchmark-derived CVRP instances containing up to 500,000 customers. Experimental results demonstrate competitive performance on benchmark-scale instances while exhibiting improved scalability and robust routing quality on substantially larger problems. These results highlight the potential of adaptive, LLM-guided decision support as a practical approach for industrial-scale vehicle routing and large-scale logistics planning.
Chinese Translation
大规模容量限制车辆路径规划问题（CVRPs）通常采用聚类优先-路径次优（CFRS）方法进行处理，该方法将一个路径规划实例分解为更小的、计算上可处理的子问题。现有的分割方法通常依赖于固定的划分规则、预定义的优化目标或学习到的策略，这可能导致在具有不同空间、需求和操作特征的实例中表现不一致。在本研究中，我们提出了一种自适应CFRS系统，将分解过程表述为一个迭代决策过程。受近期大型语言模型（LLMs）在推理和工具选择方面取得的成功启发，该系统利用LLM作为高层决策者，分析不断变化的分解状态，并选择性地应用进一步的聚类、平衡和精细化操作。所提出的算法共同对客户和车辆进行划分，实现了容量感知的聚类，同时将划分决策适应于每个问题的特征。我们在包含多达500,000名客户的合成和基准派生CVRP实例上评估了该方法。实验结果表明，在基准规模实例上表现出竞争力，同时在显著更大规模的问题上展现出更好的可扩展性和稳健的路径质量。这些结果突显了自适应的、由LLM引导的决策支持作为工业规模车辆路径规划和大规模物流规划的实际方法的潜力。
cs.AI / 70 / 2606.31831
An Agentic AI Framework to Accelerate Scientific Discovery in Plant Phenotyping
加速植物表型科学发现的自主人工智能框架
Abstract
High-throughput plant phenotyping now generates image derived datasets far faster than scientists can analyze them. At Oak Ridge National Laboratory's Advanced Plant Phenotyping Laboratory (APPL), automated stations image hundreds of plants daily across multiple remote sensing modalities; yet, trait extraction and interpretation remain manual, expert-bound, and strictly post-hoc, making analysis, not acquisition, the binding constraint on discovery. We present an end-to-end agentic AI framework that turns the facility from a data factory into an interactive autonomous, discovery platform, where scientists partner with AI agents to accelerate time to insight. A conversational Co-Scientist Agent translates a scientist's natural-language question into a structured analysis plan, and a headless Compute Agent dispatches Vision Transformer segmentation and trait extraction on the Frontier exascale supercomputer. The two agents run in separate security and resource domains and communicate over a secure, token-authenticated streaming channel, a design that accounts for the federation, data-movement, and provenance realities cloud-native agentic frameworks ignore, ensuring end-to-end provenance is captured for every interaction. The framework turns a days- to weeks-long analysis process into an interactive loop where agents reason over results, recommend next analyses, and respond to follow-up questions in seconds.
Chinese Translation
高通量植物表型技术现在生成的图像衍生数据集远远超过科学家分析的速度。在橡树岭国家实验室的先进植物表型实验室（APPL），自动化站点每天通过多种遥感模式对数百株植物进行成像；然而，性状提取和解释仍然是手动的、依赖专家的，并且严格是事后进行的，这使得分析而非获取成为发现的瓶颈。我们提出了一种端到端的自主人工智能框架，将该设施从数据工厂转变为一个互动自主的发现平台，科学家与人工智能代理合作，加速洞察的时间。一个对话式共同科学家代理将科学家的自然语言问题转化为结构化分析计划，而一个无头计算代理则在Frontier超级计算机上调度视觉变换器分割和性状提取。这两个代理在不同的安全和资源域中运行，并通过安全的、令牌认证的流媒体通道进行通信，这一设计考虑了云原生自主框架忽视的联合、数据移动和来源现实，确保每次交互的端到端来源都被捕获。该框架将原本需要数天到数周的分析过程转变为一个互动循环，代理能够对结果进行推理，推荐下一步分析，并在几秒钟内回应后续问题。
cs.AI / 71 / 2606.31876
Harnessing Textual Refusal Directions for Multimodal Safety
利用文本拒绝方向提升多模态安全性
Abstract
To improve safety in Large Language Models (LLMs) we can either perform post-training alignment or exploit refusal directions in the activation space. Both strategies are less feasible in Multimodal LLMs (MLLMs) as they require unsafe multimodal data, harder to collect than their unimodal counterpart. In this work, we relax this constraint and investigate whether textual refusal directions, extracted directly from the LLM backbone, generalize across modalities (i.e., image, video). Preliminary findings confirm this ability, though effectiveness is conditioned by layer selection, steering strength, and cross-modal alignment, with the latter causing safe multimodal inputs to be spuriously steered toward refusal. Building on this, we introduce Modality-Agnostic Refusal Steering (MARS), a light-weight training-free approach that injects multimodal safety without the need for multimodal safety data. MARS corrects modality misalignment via activation re-centering, adaptively scales steering strength within a geometrically defined trust region, and selects the optimal intervention layer, operating at the first generated token. Evaluated on five SOTA MLLMs across safety, utility, and video jailbreak benchmarks, MARS achieves consistent safety gains while preserving utility. These results reveal that safety-relevant structure is shared across modalities and that textual refusal directions are a powerful and underexplored foundation for multimodal alignment.
Chinese Translation
为了提高大型语言模型（LLMs）的安全性，我们可以进行后训练对齐或利用激活空间中的拒绝方向。然而，这两种策略在多模态LLMs（MLLMs）中可行性较低，因为它们需要不安全的多模态数据，而这些数据比单模态数据更难收集。在本研究中，我们放宽了这一限制，探讨了直接从LLM主干提取的文本拒绝方向是否能够跨模态（即图像、视频）进行泛化。初步发现证实了这一能力，尽管其有效性受到层选择、引导强度和跨模态对齐的影响，后者导致安全的多模态输入被错误地引导至拒绝。基于此，我们提出了无模态依赖的拒绝引导（Modality-Agnostic Refusal Steering, MARS），这是一种轻量级的无训练方法，可以在不需要多模态安全数据的情况下注入多模态安全性。MARS通过激活重新中心化来纠正模态不对齐，自适应地在几何定义的信任区域内调整引导强度，并选择最佳干预层，在生成的第一个标记处操作。在五个最先进的MLLMs的安全性、效用和视频越狱基准测试中进行评估，MARS实现了一致的安全性提升，同时保持了效用。这些结果揭示了安全相关结构在不同模态之间的共享，并且文本拒绝方向是多模态对齐的强大且未被充分探索的基础。
cs.AI / 72 / 2606.31976
TreeAgent: A Generalizable Multi-Agent Framework for Automated Bias Labeling in Forestry via Compiled Expert Rules and Vision-Language Models
TreeAgent：一种通用的多智能体框架，通过编译的专家规则和视觉-语言模型实现林业中的自动偏见标注
Abstract
Human-labeled data are widely used as reference annotations in ML, despite known variability across annotators in many expert-driven domains. In addition, expert annotation is slow, inconsistent, and remains a major bottleneck for scaling tasks like tree height bias classification in forestry remote sensing. We propose a multi-agent system (MAS) that orchestrates expert decision trees with Vision-Language Models (VLMs), treating the decision tree as a structural prior while VLMs perform localized semantic perception at individual nodes, with multi-agent voting to mitigate VLM stochasticity. We formalize a Decoupled Declarative Decision (D3) Framework that enables zero-modification generalization across diverse expert-defined decision structures. On a tree bias classification testbed, our framework outperforms supervised ML baselines and reduces the amount of expert labeling effort required. These results suggest that agentic orchestration of VLMs with expert priors can reproduce expert-defined labeling procedures at substantially lower annotation cost while maintaining interpretability.
Chinese Translation
人类标注的数据在机器学习中被广泛用作参考注释，尽管在许多专家驱动的领域中，标注者之间存在已知的变异性。此外，专家注释速度慢、不一致，仍然是像林业遥感中的树高偏见分类等任务扩展的主要瓶颈。我们提出了一种多智能体系统（MAS），该系统将专家决策树与视觉-语言模型（VLMs）结合，视决策树为结构先验，而VLMs在各个节点执行局部语义感知，并通过多智能体投票来减轻VLM的随机性。我们形式化了一个解耦声明决策（D3）框架，使其能够在多样化的专家定义决策结构中实现零修改泛化。在树偏见分类测试平台上，我们的框架优于监督机器学习基线，并减少了所需的专家标注工作量。这些结果表明，智能体对VLM与专家先验的协调可以以显著较低的标注成本重现专家定义的标注程序，同时保持可解释性。
cs.AI / 73 / 2606.32002
Self-Study Reconsidered: The Hidden Fragility of Learning from Self-Generated QA
自我学习再思考：自生成问答学习的隐性脆弱性
Abstract
Language models are increasingly taught from synthetic question--answer (QA) supervision: a model generates questions about a document, answers them from the same text, and the resulting pairs are used to fine-tune, distill, or compress knowledge into another model. We show that this generation step is not neutral preprocessing. It is an implicit policy that both selects which evidence becomes training signal and decides how that evidence is answered, and it is fragile at both stages. When choosing what to ask, generators do not scan a document uniformly. Coverage saturates early and concentrates on salient spans, diverse prompts converge on the same regions, and what looks question-worthy is driven by local presentation. As a result, salient artifacts such as poorly cleaned markup can hijack question generation across model families and scales. When answering, the model that produces the supervision tends to obey instruction-like passages embedded in the text. This compliance depends on the intent and surface form of the passage rather than its strictness, and is worst under task conflict, where larger models comply more often. These failure modes arise from choices made during QA generation, so they can be reduced without changing the training loop. Tying each question to a fixed target reduces biased selection, and filtering instruction-like spans before answering lowers mean injection compliance from $88\%$ to $13\%$ in our evaluation while retaining nearly all clean text.
Chinese Translation
语言模型越来越多地通过合成问答（QA）监督进行训练：模型生成关于文档的问题，从同一文本中回答这些问题，生成的问答对用于微调、提炼或将知识压缩到另一个模型中。我们展示了这一生成步骤并不是中立的预处理。它是一种隐含的策略，既选择哪些证据成为训练信号，又决定如何回答这些证据，并且在这两个阶段都表现出脆弱性。在选择提问内容时，生成器并不会均匀扫描文档。覆盖率早期饱和，并集中在显著的片段上，多样化的提示趋向于集中在相同区域，而看似值得提问的内容则受到局部呈现的驱动。因此，显著的伪影，如清理不当的标记，可能会在不同模型家族和规模之间劫持问题生成。在回答时，产生监督的模型往往遵循嵌入文本中的指令类段落。这种遵从性依赖于段落的意图和表面形式，而非其严格性，并且在任务冲突下表现最差，较大的模型更频繁地遵从。这些失败模式源于在QA生成过程中做出的选择，因此可以在不改变训练循环的情况下减少。将每个问题与固定目标关联可以减少偏见选择，而在回答之前过滤指令类片段则将我们评估中的平均注入遵从性从$88\%$降低到$13\\%$，同时几乎保留了所有干净文本。
cs.AI / 74 / 2606.32004
PolicyGuard: From Organizational Policies to Neuro-SymbolicCompliance Review Engines
PolicyGuard：从组织政策到神经符号合规审查引擎
Abstract
Policy-grounded document review requires determining whether a target document complies with organization-specific policies, guidelines, or playbooks. While large language models can assist with policy interpretation and document analysis, end-to-end prompting leaves the applied policy logic implicit, making compliance decisions difficult to inspect, update, and test. We present PolicyGuard, a neuro-symbolic framework for policy-grounded document compliance review. PolicyGuard converts organizational policy guidance into an executable review engine consisting of typed relational logic rules and atom-level extraction questions. During review, LLMs answer these local questions using retrieved document evidence, and a symbolic evaluator applies the formal rules to detect non-compliance. We instantiate and evaluate PolicyGuard on company-specific NDA compliance review, where contract clauses must be checked against organization-specific negotiation policies. By separating policy formalization, local document interpretation, and symbolic compliance evaluation, PolicyGuard makes document review more explicit, maintainable, and systematically testable.
Chinese Translation
基于政策的文档审查需要确定目标文档是否符合特定组织的政策、指南或操作手册。尽管大型语言模型可以辅助政策解释和文档分析，但端到端的提示使得应用的政策逻辑隐含，导致合规决策难以检查、更新和测试。我们提出了PolicyGuard，一个用于基于政策的文档合规审查的神经符号框架。PolicyGuard将组织政策指导转换为一个可执行的审查引擎，该引擎由类型化的关系逻辑规则和原子级提取问题组成。在审查过程中，LLM使用检索到的文档证据回答这些局部问题，符号评估器则应用正式规则来检测不合规情况。我们在公司特定的保密协议（NDA）合规审查中实例化并评估了PolicyGuard，其中合同条款必须与特定组织的谈判政策进行检查。通过分离政策形式化、局部文档解释和符号合规评估，PolicyGuard使文档审查变得更加明确、可维护和系统可测试。
cs.AI / 75 / 2606.32007
AxDafny: Agentic Verified Code Generation in Dafny
AxDafny：Dafny中的代理验证代码生成
Abstract
We study agentic code generation in Dafny, where a model must generate both executable code and the proof artifacts for verification. We present AxDafny, a verifier-guided repair framework that iteratively generates implementations, invariants, assertions, and termination arguments. We also introduce LiveCodeBench-Pro-Dafny (LCB-Pro-Dafny), a benchmark of 250 competition-style programming problems translated into Dafny with formal specifications and a verifier-based evaluation harness. On LCB-Pro-Dafny, AxDafny substantially improves verification success over baseline GPT-5.5 performance. On DafnyBench, AxDafny achieves 92.7\% verification success, outperforming the strongest previously reported proof-hint baseline by 6.5 percentage points. Lastly, we show that verification success and runtime test performance measure different aspects of generated code.
Chinese Translation
我们研究了Dafny中的代理代码生成，其中模型必须生成可执行代码和用于验证的证明文档。我们提出了AxDafny，这是一个由验证器引导的修复框架，能够迭代生成实现、不变式、断言和终止论证。我们还介绍了LiveCodeBench-Pro-Dafny（LCB-Pro-Dafny），这是一个包含250个竞赛风格编程问题的基准，这些问题被翻译成Dafny并附有正式规范以及基于验证器的评估工具。在LCB-Pro-Dafny上，AxDafny显著提高了验证成功率，相较于基线GPT-5.5的表现。在DafnyBench上，AxDafny达到了92.7%的验证成功率，超越了之前报告的最强证明提示基线6.5个百分点。最后，我们展示了验证成功率和运行时测试性能测量生成代码的不同方面。
cs.CL / 1 / 2606.30775
A Single Rewrite Suffices: Empirical Lessons from Production Skill Description Optimization
单次重写足矣：来自生产技能描述优化的经验教训
Abstract
Enterprise AI agents route user queries to specialized skills by matching queries against natural language skill descriptions. When two skills share overlapping descriptions, the routing LLM misroutes queries, a failure we term skill collision. As agents scale to dozens of skills, manually tuning descriptions to maintain routing accuracy becomes a significant engineering bottleneck. We deploy an automated description optimization pipeline on a production enterprise group chat agent (9 skills, 372 regression cases). The pipeline produces descriptions averaging 79.2% F1, matching manually tuned descriptions at 79.4% F1 (average per-skill difference -0.20%, within the 0.78% multi-seed noise floor), while reducing per-skill engineering effort from 120 minutes to 3.8 minutes (32 times speedup). We then examine which pipeline components actually drive this match. Systematic ablation on both the production system and ToolBench (16k tools) reveals that a single LLM rewrite using any available false-positive and false-negative cases captures most of the available improvement. Other design choices we tested (iteration budget, feedback signal composition, dual editing of confused pairs, and training set size) each affect final F1 by less than 0.5%. Description optimization addresses skill collisions caused by overlapping descriptions but cannot resolve cases where two skills intended scopes genuinely overlap. We identify a diagnostic (a large train-validation F1 gap) that flags the latter cases for architectural rather than text-level intervention.
Chinese Translation
企业AI代理通过将用户查询与自然语言技能描述进行匹配，将查询路由到专门的技能。当两个技能共享重叠的描述时，路由的LLM会错误地路由查询，这种失败我们称之为技能冲突。随着代理技能数量增加到数十个，手动调整描述以维持路由准确性成为一个重要的工程瓶颈。我们在一个生产企业群聊代理上部署了一个自动化描述优化管道（9个技能，372个回归案例）。该管道生成的描述平均F1为79.2%，与手动调整的描述的79.4% F1相匹配（每个技能的平均差异为-0.20%，在0.78%的多种种子噪声底线之内），同时将每个技能的工程工作量从120分钟减少到3.8分钟（速度提升32倍）。随后，我们考察了哪些管道组件实际上推动了这一匹配。对生产系统和ToolBench（16000个工具）进行的系统性消融实验表明，使用任何可用的假阳性和假阴性案例进行的单次LLM重写捕获了大部分可用的改进。我们测试的其他设计选择（迭代预算、反馈信号组成、混淆对的双重编辑和训练集大小）对最终F1的影响均小于0.5%。描述优化解决了由重叠描述引起的技能冲突，但无法解决两个技能预期范围真正重叠的情况。我们识别出一个诊断指标（较大的训练-验证F1差距），用于标记后者情况，以便进行架构层面的干预，而非文本层面的干预。
cs.CL / 2 / 2606.30790
Indi-RomCoM: Code-Mixed Benchmark for Evaluating LLMs on Romanized Indic-English Instructions
Indi-RomCoM：评估大型语言模型在罗马化印地语-英语指令上的代码混合基准
Abstract
Romanized Code Mixing (RCM), where bilingual speakers fluidly blend local languages with English in Roman script, has emerged as the dominant form of communication across multilingual communities. While Large Language Models (LLMs) perform strongly on monolingual and native-script benchmarks, their ability to follow instructions and reason over RCM-based content remains largely unexplored. To this end, we introduce the Indi-RomCoM benchmark for facilitating systematic evaluation on Indic Romanized Code-Mixed instructions. Our benchmark spans seven instruction-following tasks, four widely spoken Indic languages, and three controlled code-mixing intensity levels. We extensively evaluate a suite of LLMs covering proprietary, open-weight, and Indic-focused models under zero- and few-shot settings. LLMs consistently underperform on RCM instructions, with performance degrading as code-mixing density increases. Furthermore, reasoning tasks suffer less degradation than detection tasks (e.g., Toxicity) because the generated explanations offer necessary context. We believe Indi-RomCoM helps the community in developing inclusive multilingual systems.
Chinese Translation
罗马化代码混合（RCM）是指双语使用者在罗马字母书写的英语中流畅地融合本地语言，这已成为多语言社区中主要的交流形式。虽然大型语言模型（LLMs）在单语和本地书写基准上表现出色，但它们在遵循指令和推理基于RCM内容的能力仍然 largely 未被探索。为此，我们引入了Indi-RomCoM基准，以促进对印地语罗马化代码混合指令的系统评估。我们的基准涵盖七个指令遵循任务、四种广泛使用的印地语语言和三个受控的代码混合强度水平。我们在零样本和少样本设置下，对一系列LLMs进行了广泛评估，涵盖了专有模型、开放权重模型和以印地语为重点的模型。LLMs在RCM指令上的表现一致较差，随着代码混合密度的增加，性能下降。此外，推理任务的性能下降程度低于检测任务（例如，毒性），因为生成的解释提供了必要的上下文。我们相信Indi-RomCoM将帮助社区开发包容性的多语言系统。
cs.CL / 3 / 2606.30801
Using AI Agents to Automate Black-Box Audits of Personalization Algorithms at Scale
利用人工智能代理大规模自动化个性化算法的黑箱审计
Abstract
Personalization algorithms determine what content users encounter on online platforms. Auditing these systems is difficult because independent auditors have only black-box access to the algorithms, while personalization depends on users' attributes, behavior, and evolving interaction histories. Existing auditing methods face a tradeoff: studies with real users capture realistic behavior but are costly and hard to control, whereas sock-puppet audits scale more easily but often rely on scripted behavior that limits realism. Beyond this, both approaches struggle to decouple user attributes from user behavior, limiting our ability to causally understand personalization. To address this gap, we introduce a framework for black-box audits of personalization algorithms using generative AI agents as behavioral engines for synthetic accounts. Each agent is instantiated with a fixed persona, grounded in demographic and political survey data, and interacts with a platform's content by reasoning about it and choosing actions. Because behavior is fixed within each persona while platform-visible signals such as age, gender, or location can be experimentally perturbed, our design enables counterfactual auditing of how platforms respond to user attributes. As a case study, we deploy 1,120 agents on X shortly after the 2024 U.S. election, spanning 14 personas and three counterfactual conditions, collecting over 200,000 content exposures. We find that X's algorithmic feed amplifies toxic, polarizing, political, and right-leaning content relative to the chronological feed, with amplification varying sharply by user ideology. Counterfactual analyses show that demographic signals affect content delivery in persona-dependent ways: pooled effects are largely null, while subgroup-level effects vary in direction and magnitude. Our work establishes GenAI-based agents as a new tool for algorithmic auditing.
Chinese Translation
个性化算法决定了用户在在线平台上遇到的内容。审计这些系统非常困难，因为独立审计员只能对算法进行黑箱访问，而个性化依赖于用户的属性、行为和不断变化的互动历史。现有的审计方法面临权衡：与真实用户的研究能够捕捉到真实的行为，但成本高且难以控制；而虚假用户审计则更容易扩展，但通常依赖于脚本化的行为，这限制了真实感。除此之外，这两种方法都难以将用户属性与用户行为解耦，限制了我们对个性化的因果理解。为了解决这一问题，我们提出了一个利用生成性人工智能代理作为合成账户行为引擎的个性化算法黑箱审计框架。每个代理都以固定的角色实例化，基于人口统计和政治调查数据，并通过推理和选择行动与平台内容进行互动。由于每个角色内的行为是固定的，而平台可见的信号（如年龄、性别或位置）可以进行实验性扰动，我们的设计使得对平台如何响应用户属性的反事实审计成为可能。作为案例研究，我们在2024年美国大选后不久在X平台上部署了1,120个代理，涵盖14个角色和三种反事实条件，收集了超过200,000次内容曝光。我们发现，与时间顺序推送相比，X的算法推送放大了有毒、两极化、政治性和偏右内容，放大的程度因用户意识形态而显著变化。反事实分析表明，人口统计信号以角色依赖的方式影响内容传递：总体效应大多为零，而子群体层面的效应在方向和幅度上各不相同。我们的研究确立了基于生成性人工智能的代理作为算法审计的新工具。
cs.CL / 4 / 2606.30814
When Calibration Rankings Reverse: Accuracy-Controlled Evaluation for Fair Comparison of LLMs
当校准排名逆转：基于准确性控制的公平比较大型语言模型的评估
Abstract
Calibration evaluates whether a model confidence aligns with its empirical accuracy. Existing studies often compare the calibration of different large language models using global calibration metrics such as Expected Calibration Error and Brier Score. We begin by showing, both theoretically and empirically, that such comparisons are confounded by differences in model accuracy. For fairer cross-model comparison, we then propose ACE, an accuracy-controlled evaluation framework with three complementary views: Instance-Aligned, Distribution-Aligned, and Candidate-Aligned calibration. Across multiple benchmarks, model families, and confidence elicitation methods, we use ACE to study two practically important comparison axes, small versus large models and thinking versus non-thinking models. We find that many previously reported calibration advantages under raw global metrics weaken substantially after accuracy control. We also find that ranking reversal is frequent: models favored by raw metrics often cease to be favored once accuracy is controlled. Our results show that raw global calibration metrics are not robust for cross-model comparison, and that fair calibration comparison requires accuracy-aware evaluation.
Chinese Translation
校准评估模型的置信度是否与其实证准确性相一致。现有研究通常使用全局校准指标（如期望校准误差（Expected Calibration Error）和布里尔分数（Brier Score））比较不同大型语言模型的校准。我们首先通过理论和实证展示，这种比较受到模型准确性差异的干扰。为了实现更公平的跨模型比较，我们提出了ACE，一个基于准确性控制的评估框架，包含三个互补视角：实例对齐（Instance-Aligned）、分布对齐（Distribution-Aligned）和候选对齐（Candidate-Aligned）校准。在多个基准、模型家族和置信度引导方法中，我们使用ACE研究两个在实践中重要的比较维度：小模型与大模型，以及思考模型与非思考模型。我们发现，许多在原始全局指标下报告的校准优势在准确性控制后显著减弱。我们还发现排名逆转的现象很常见：原始指标偏好的模型在控制准确性后往往不再被偏好。我们的结果表明，原始全局校准指标在跨模型比较中并不稳健，而公平的校准比较需要考虑准确性的评估。
cs.CL / 5 / 2606.30815
When transformers learn "impossible" languages, what do they learn?
当变压器学习“不可思议”的语言时，他们学到了什么？
Abstract
Recent work suggests that transformer language models show a bias towards human languages over unnatural ("impossible") languages argued to be unacquirable by humans. However, this literature has largely based these claims on differences in sample efficiency and test-set perplexity, rather than on direct evaluations of the linguistic capacities that could plausibly explain non-attestation in human languages. We evaluate two theoretically motivated linking hypotheses: impossibility arising from deficiencies in grammatical sensitivity or generative production. Using GPT-2 style models trained on perturbed "impossible" variants of English, we measure sensitivity to grammaticality using BLiMP minimal pairs, finding that model performance exhibits only gradual degradation, mediated by the language's information locality. In contrast, these models exhibited pronounced failures in generation, producing substantially fewer high-quality sentences at longer lengths. Together, these results suggest generative deficiency and transmission failures as a plausible linking hypothesis between language model behaviour and non-attestation of impossible languages.
Chinese Translation
最近的研究表明，变压器语言模型对人类语言表现出偏向，而对被认为人类无法习得的非自然（“不可思议”）语言则表现出偏见。然而，这些文献主要基于样本效率和测试集困惑度的差异，而不是对能够合理解释人类语言中非存在现象的语言能力的直接评估。我们评估了两个理论上有动机的连接假设：由于语法敏感性或生成能力的缺陷而导致的不可能性。使用在扰动的“不可思议”英语变体上训练的GPT-2风格模型，我们通过BLiMP最小对测量语法性敏感性，发现模型性能仅表现出逐渐的退化，受语言信息局部性的调节。相比之下，这些模型在生成方面表现出明显的失败，在较长句子中生成的高质量句子显著减少。综合来看，这些结果表明生成缺陷和传播失败是语言模型行为与不可思议语言非存在之间的一个合理连接假设。
cs.CL / 6 / 2606.30851
Test-Time Verification for Text-to-SQL via Outcome Reward Models
基于结果奖励模型的文本到SQL的测试时验证
Abstract
Improving the reliability of large language models (LLMs) at inference time is a central challenge in structured reasoning tasks such as Text-to-SQL. Common test-time inference strategies, including Best-of-N sampling and Majority Voting, rely on heuristic signals such as execution success or output frequency, which provide limited semantic discrimination across candidate outputs. In this work, we study Outcome Reward Models (ORMs) as learned semantic scoring functions for test-time verification in Text-to-SQL. While ORMs have been previously explored for test-time scaling and alignment, their application to structured query generation remains underexplored. We introduce GradeSQL, a scalable framework for training task-specific ORMs via automated candidate generation and execution-based labeling, enabling verifier training without manual annotation. We integrate ORMs into a verification-driven Best-of-N pipeline and evaluate our approach on the BIRD and Spider benchmarks across multiple open-source LLM families. ORM-based selection consistently outperforms execution-based Best-of-N and Majority Voting, with gains of up to +4.33% on BIRD and +2.10% on Spider. We further show that ORMs scale effectively with larger candidate sets and yield stronger improvements on complex queries. Overall, our results demonstrate that ORM-based verification provides a simple, effective, and scalable alternative to heuristic test-time selection strategies for Text-to-SQL. Code datasets and models are publicly available.
Chinese Translation
提高大型语言模型（LLMs）在推理时的可靠性是结构化推理任务（如文本到SQL）中的一个核心挑战。常见的测试时推理策略，包括最佳N采样和多数投票，依赖于执行成功或输出频率等启发式信号，这些信号在候选输出之间提供的语义区分有限。在本研究中，我们探讨了结果奖励模型（Outcome Reward Models, ORMs）作为学习的语义评分函数，用于文本到SQL的测试时验证。虽然ORMs之前已被用于测试时的扩展和对齐，但其在结构化查询生成中的应用仍然未被充分探索。我们引入了GradeSQL，这是一个通过自动候选生成和基于执行的标注训练任务特定ORMs的可扩展框架，使得验证器训练无需手动标注。我们将ORMs集成到一个以验证为驱动的最佳N管道中，并在BIRD和Spider基准上对我们的方法进行了评估。在多个开源LLM系列中，基于ORM的选择始终优于基于执行的最佳N和多数投票，在BIRD上提升高达+4.33%，在Spider上提升+2.10%。我们进一步展示了ORMs在更大候选集上有效扩展，并在复杂查询上产生更强的改进。总体而言，我们的结果表明，基于ORM的验证为文本到SQL的启发式测试时选择策略提供了一种简单、有效且可扩展的替代方案。代码、数据集和模型已公开可用。
cs.CL / 7 / 2606.30857
Multilingual Polarization Detection Using Transformer-Based Models with Class Weighting and Threshold Tuning
基于变换器模型的多语言极化检测：类权重调整与阈值调优
Abstract
This paper describes our submission to SemEval-2026 Task 9 on detecting multilingual, multicultural, and multievent online polarization. We address all three subtasks: binary polarization detection, polarization type classification, and manifestation identification for English and Swahili. Our approach leverages transformer-based models (RoBERTa-base for English, AfroXLMR-base for Swahili) with class-weighted loss functions to address severe label imbalance and per-label threshold tuning to optimize multi-label classification. On the test set, we achieve F1 macro scores of 0.7901 (English) and 0.7910 (Swahili) for Subtask 1, 0.4615 (English) and 0.4808 (Swahili) for Subtask 2 and 0.4791 (English) and 0.5830 (Swahili) for Subtask 3, which give competitive performance on the leaderboard, demonstrating the effectiveness of our methods for handling imbalanced multi-label polarization detection. Our error analysis reveals that models struggle with dehumanization detection and lack of empathy.
Chinese Translation
本文描述了我们对SemEval-2026任务9的提交，该任务旨在检测多语言、多文化和多事件的在线极化。我们针对三个子任务进行了研究：二元极化检测、极化类型分类和表现形式识别，涵盖英语和斯瓦希里语。我们的方法利用了基于变换器的模型（英语使用RoBERTa-base，斯瓦希里语使用AfroXLMR-base），并采用类加权损失函数来解决严重的标签不平衡问题，同时进行每个标签的阈值调优以优化多标签分类。在测试集上，我们在子任务1中取得了0.7901（英语）和0.7910（斯瓦希里语）的F1宏观得分，在子任务2中取得了0.4615（英语）和0.4808（斯瓦希里语），在子任务3中取得了0.4791（英语）和0.5830（斯瓦希里语），在排行榜上表现出竞争力，证明了我们的方法在处理不平衡的多标签极化检测中的有效性。我们的错误分析显示，模型在去人性化检测和缺乏同理心方面存在困难。
cs.CL / 8 / 2606.30887
Training Therapeutic Judges and Multi-Agent Systems for Human-Aligned Mental Health Support
训练治疗性评估者和多智能体系统以支持人类对齐的心理健康
Abstract
Large language models show promise for mental health support, yet therapeutic quality improves only when evaluation functions as an actionable control signal rather than a passive metric. We introduce a framework that formulates therapeutic response generation as a decision-refinement problem driven by multi-dimensional, human-aligned evaluation. In Stage I, we introduce TheraJudge, an open-source therapeutic evaluator trained via preference-based optimization on human-annotated data to produce reliable judgments across 7 psychological dimensions. In Stage II, we introduce TheraAgent, which operationalizes TheraJudge's evaluations through a coordinated refinement process with specialized Critic, Coach, and Therapist roles that translate evaluative signals into targeted response revisions. Empirically, TheraJudge achieves strong agreement with clinician ratings, with intraclass correlation coefficients (ICC = 0.87-0.95), surpassing supervised baselines and strong closed-source judges, particularly on critical dimensions such as Safety, Relevance, and Empathy. Acting on these evaluations, TheraAgent yields a +0.43 improvement in human-rated therapeutic quality (on a 5-point scale) under blind evaluation, with 96\% clinician inter-rater reliability. Low-quality responses ($\leq 3$) improve by +2.45 points with a 94\% recovery rate, demonstrating targeted correction of unsafe outputs. Overall, our results indicate that effective alignment of mental-health LLMs stems from acting on human-aligned evaluation, rather than relying solely on stronger generation. We release code at https://github.com/vis-nlp/TheraAlign.
Chinese Translation
大型语言模型在心理健康支持方面显示出潜力，但治疗质量的提升仅在评估作为可操作的控制信号而非被动的度量时才会发生。我们提出了一个框架，将治疗响应生成形式化为一个由多维人类对齐评估驱动的决策精炼问题。在第一阶段，我们引入了TheraJudge，一个通过基于偏好的优化在人工标注数据上训练的开源治疗评估器，旨在产生在7个心理维度上可靠的判断。在第二阶段，我们引入了TheraAgent，它通过与专门的批评者（Critic）、教练（Coach）和治疗师（Therapist）角色的协调精炼过程，将TheraJudge的评估转化为针对性的响应修正。在实证研究中，TheraJudge与临床医生的评分达成了强一致性，组内相关系数（ICC = 0.87-0.95）超过了监督学习基线和强闭源评估者，特别是在安全性（Safety）、相关性（Relevance）和同理心（Empathy）等关键维度上。基于这些评估，TheraAgent在盲评估中实现了人类评分的治疗质量提升+0.43（在5分制上），临床医生的评分者间一致性为96%。低质量响应（$ ext{评分} ext{≤} 3$）的评分提升了+2.45分，恢复率为94%，展示了对不安全输出的针对性修正。总体而言，我们的结果表明，心理健康大型语言模型的有效对齐源于对人类对齐评估的响应，而非仅仅依赖于更强的生成能力。我们在https://github.com/vis-nlp/TheraAlign发布了代码。
cs.CL / 9 / 2606.30914
Beyond Clean Text: Evaluating Encoder and Decoder Robustness for Bangla Event Detection in Noisy Text
超越干净文本：评估噪声文本中孟加拉事件检测的编码器和解码器鲁棒性
Abstract
Event detection (ED) systems are typically evaluated on clean, curated text, leaving their robustness to real-world noise largely unexplored, particularly for low-resource languages such as Bangla. We introduce a generalized Bangla news event ontology and a benchmark comprising 9,979 annotated sentences across 40 event subtypes, spanning clean news text, real-world Automatic Speech Recognition (ASR) transcripts, and orthographically corrupted text. We systematically evaluate fine-tuned encoder-only models (BanglaBERT and XLM-R) alongside instruction-tuned decoder-only large language models (Llama 3 and Gemma 3). Our results reveal a clear architectural trade-off: encoder models achieve higher performance on clean text but degrade substantially under noise, whereas decoder-only LLMs are markedly more robust, particularly when event triggers are corrupted. We further show that embedding annotation guidelines during instruction tuning establishes a higher performance baseline on noisy text but yields inconsistent reductions in performance degradation across noisy conditions. Finally, model scaling consistently improves the robustness of decoder-only LLMs, while combined training on clean and noisy data serves as an effective regularization strategy that disproportionately benefits encoder architectures, significantly narrowing the robustness gap.
Chinese Translation
事件检测（ED）系统通常在干净、经过整理的文本上进行评估，这使得它们在现实世界噪声下的鲁棒性在很大程度上未得到探索，尤其是对于孟加拉等低资源语言。我们引入了一个通用的孟加拉新闻事件本体和一个基准数据集，该数据集包含9,979个标注句子，涵盖40个事件子类型，包括干净的新闻文本、现实世界的自动语音识别（ASR）转录文本以及正字法损坏的文本。我们系统地评估了经过微调的仅编码器模型（BanglaBERT和XLM-R）以及经过指令调优的仅解码器大型语言模型（Llama 3和Gemma 3）。我们的结果揭示了明显的架构权衡：编码器模型在干净文本上表现更高，但在噪声下显著下降，而仅解码器的LLM在鲁棒性上明显更强，尤其是在事件触发器被损坏时。我们进一步表明，在指令调优过程中嵌入注释指南可以在噪声文本上建立更高的性能基线，但在不同噪声条件下的性能下降幅度却不一致。最后，模型规模的扩大始终提高了仅解码器LLM的鲁棒性，而在干净和噪声数据上的联合训练作为一种有效的正则化策略，特别有利于编码器架构，显著缩小了鲁棒性差距。
cs.CL / 10 / 2606.30943
Bridging Scientific Heritage: An Arabic--Russian Parallel Corpus and LLM Benchmark for Sustainable Knowledge Transfer
弥合科学遗产：阿拉伯语-俄语平行语料库与可持续知识转移的LLM基准
Abstract
Russian and Arabic are among the major languages of scientific communication. Language barriers impede the exchange of research results between these communities, which affects international collaboration and the progress of sustainability-related research. We present a benchmark for Arabic--Russian scientific translation. The benchmark includes a hybrid parallel corpus of about 27,000 sentence pairs, compiled from scientific abstracts and general-domain texts (religion, news, conversations). We fine-tune three multilingual language models -- mT5-base (580M parameters), NLLB-200-distilled-1.3B (1.3B), and Qwen2.5-7B-Instruct (7B) -- using LoRA with ranks 8, 16, 32, and 64. The Qwen2.5-7B model with QLoRA (rank 8) yields BLEU 23.15, chrF 43.89, BERTScore 0.906, and COMET 0.758. These are +4.36 BLEU and +0.051 COMET above the zero-shot baseline. Few-shot prompting with three examples does not improve performance, indicating that domain-specific fine-tuning is required. We release the models, the corpus, and the evaluation code. By lowering the language barrier for scientific texts, the work enables knowledge exchange between Arabic-speaking and Russian-speaking researchers. It contributes to sustainable partnerships (UN SDG 17) and innovation infrastructure (SDG 9), aligning with the conference's focus on technology-driven sustainable development.
Chinese Translation
俄语和阿拉伯语是科学交流的主要语言之一。语言障碍阻碍了这些社群之间研究成果的交流，影响了国际合作和与可持续性相关研究的进展。我们提出了一个阿拉伯语-俄语科学翻译的基准。该基准包括约27,000对句子的混合平行语料库，来源于科学摘要和一般领域文本（宗教、新闻、对话）。我们使用LoRA对三个多语言模型进行微调——mT5-base（580M参数）、NLLB-200-distilled-1.3B（1.3B）和Qwen2.5-7B-Instruct（7B），其秩分别为8、16、32和64。使用QLoRA（秩8）的Qwen2.5-7B模型获得了BLEU 23.15，chrF 43.89，BERTScore 0.906和COMET 0.758。这些结果比零-shot基线提高了4.36 BLEU和0.051 COMET。使用三个示例的少量提示并未改善性能，表明需要进行领域特定的微调。我们发布了模型、语料库和评估代码。通过降低科学文本的语言障碍，该工作促进了阿拉伯语和俄语研究人员之间的知识交流。它有助于可持续伙伴关系（联合国可持续发展目标17）和创新基础设施（可持续发展目标9），与会议对技术驱动的可持续发展的关注相一致。
cs.CL / 11 / 2606.30957
Linguistic Distancing on Social Media: Indicators of Emotion Regulation Across Age Groups
社交媒体上的语言距离：不同年龄组情绪调节的指标
Abstract
Managing our emotional responses to events is key to emotional well-being, a process referred to as emotion regulation in psychology. Previous work has established that the degree to which we distance events is a type of emotion regulation. When we psychologically distance from events there can be markers in our language. These markers have been referred to as linguistic distancing. We build upon a previous metric to operationalize linguistic distancing, and explore how it changes across the lifespan. We explore this systematically by analyzing large amounts of social media text, a venue where people express their emotions. By investigating how distancing varies across age groups we can better understand how emotion regulation varies with age and provide initial benchmarks on social media data. We provide additional evidence further strengthening the hypothesis that linguistic distancing occurs in proportionally more instances with age. These findings align with past work in psychology which indicate improved well-being with older age. Better understanding how linguistic distancing changes with age is important because it functions as a marker of well-being and can inform effective health interventions. We provide a foundation for further exploring emotion regulation through linguistic distancing in text data.
Chinese Translation
管理我们对事件的情绪反应是情绪健康的关键，这一过程在心理学中被称为情绪调节。先前的研究已确立了我们对事件的距离感程度是一种情绪调节方式。当我们在心理上与事件保持距离时，语言中可能会出现一些标记。这些标记被称为语言距离。我们在之前的度量基础上进一步发展，操作化语言距离，并探讨其在生命周期中的变化。我们通过系统性分析大量社交媒体文本来探讨这一点，社交媒体是人们表达情感的场所。通过研究不同年龄组之间的距离感变化，我们可以更好地理解情绪调节如何随年龄变化，并为社交媒体数据提供初步基准。我们提供了额外的证据，进一步加强了语言距离在年龄增长中相对增加的假设。这些发现与心理学中的过去研究一致，表明随着年龄的增长，幸福感得以改善。更好地理解语言距离如何随年龄变化是重要的，因为它作为幸福感的标记，并能为有效的健康干预提供信息。我们为进一步通过文本数据中的语言距离探索情绪调节奠定了基础。
cs.CL / 12 / 2606.30973
From Propositional to Perceptual Asymmetry: Extending Frictive Policy Optimization to Asymmetric Partial Information Dialogue
从命题不对称到感知不对称：将摩擦策略优化扩展到不对称部分信息对话
Abstract
Frictive Policy Optimization (FPO; Pustejovsky et al., 2025) treats friction in collaborative dialogue -- misalignment, misunderstanding, repair -- as an epistemic signal essential to common-ground construction, rather than noise to be minimized. However, FPO and its implementations assume shared perceptual contexts, where friction arises from differently interpreted propositions over the same scene, which we define as propositional asymmetry. We extend FPO to perceptual asymmetry, where participants hold asymmetric partial information and the same referring expression yields different denotations depending on whose information state grounds the reference. We evaluate this through cross-corpora analysis and LLM probing on referentially asymmetric dialogue tasks, primarily the HCRC MapTask (Anderson et al., 1991). We find that FPO's friction functional is empirically valid only when evaluated from within each participant's information horizon: different landmark configurations produce qualitatively distinct grounding failure modes, with a small class of ambiguous configurations driving a disproportionate share of misunderstandings through trajectories that appear successful but silently diverge. The LLM probe confirms that having the "right perspective" matters more than having all perspectives: the informed single viewpoint outperforms omniscient access to both participants' contexts. We propose two annotation refinements: subtype decomposition of pending grounding states and accommodation-aware alignment classification.
Chinese Translation
摩擦策略优化（Frictive Policy Optimization, FPO; Pustejovsky et al., 2025）将协作对话中的摩擦——不一致、误解、修复——视为构建共同基础所必需的认知信号，而非需要最小化的噪声。然而，FPO及其实现假设共享的感知背景，其中摩擦源于对同一场景的不同解读命题，我们将其定义为命题不对称。我们将FPO扩展到感知不对称，其中参与者持有不对称的部分信息，相同的指称表达根据其信息状态的基础而产生不同的指称。我们通过跨语料库分析和对指称不对称对话任务的LLM探测进行评估，主要是HCRC MapTask（Anderson et al., 1991）。我们发现，FPO的摩擦函数仅在从每个参与者的信息视野内进行评估时才在经验上有效：不同的地标配置产生质上不同的基础失败模式，一小类模糊配置通过看似成功但默默偏离的轨迹驱动了不成比例的误解。LLM探测确认，拥有“正确的视角”比拥有所有视角更为重要：知情的单一视角优于对两个参与者背景的全知访问。我们提出了两个注释改进：待处理基础状态的子类型分解和考虑适应性的对齐分类。
cs.CL / 13 / 2606.30987
Measuring Judgment Quality in Natural-Language Explanations: Evidence from Forecasting Tournaments
在自然语言解释中衡量判断质量：来自预测比赛的证据
Abstract
Decision-makers routinely rely on expert judgments accompanied by written explanations, yet explanation quality is difficult to measure at scale. Forecasting tournaments offer a natural testing ground: probabilistic judgments are paired with natural-language rationales and scored against realized outcomes. We introduce Explanation Quality Markers (EQMs), a set of sixty theory-guided reasoning patterns scored by large language models (LLMs). In a pre-registered analysis of over 55,000 forecast-rationale pairs from a multiyear forecasting tournament, EQMs predict accuracy at both the forecast and forecaster levels, consistently outperforming pre-LLM text-analysis methods. More than 90% of statistically significant pattern-level EQM-accuracy correlations match our directional hypotheses. The signal is asymmetric: EQMs identify likely underperformers more reliably than they distinguish the very best forecasters. Benchmarked against traditional indicators of forecasting skill, EQMs are the strongest predictor at the forecast level and competitive at the forecaster level, though weaker than prior accuracy. Human ratings of rationale quality are less consistently correlated with accuracy and place disproportionate weight on rationale length. Results transfer to an independent forecasting study. EQMs provide a scalable, interpretable method for extracting judgment-relevant information from written explanations.
Chinese Translation
决策者通常依赖于附有书面解释的专家判断，但在大规模上衡量解释质量却很困难。预测比赛提供了一个自然的测试场：概率判断与自然语言的理由相结合，并与实际结果进行评分。我们引入了解释质量标记（Explanation Quality Markers, EQMs），这是一组由大型语言模型（Large Language Models, LLMs）评分的六十种理论指导的推理模式。在对来自多年预测比赛的超过55,000对预测-理由对的预注册分析中，EQMs在预测和预测者层面上都能有效预测准确性，并且始终优于预先基于LLM的文本分析方法。超过90%的统计显著性模式级EQM-准确性相关性与我们的方向性假设相符。信号是非对称的：EQMs更可靠地识别出可能的表现不佳者，而不是区分出表现最好的预测者。与传统的预测技能指标相比，EQMs在预测层面是最强的预测因子，在预测者层面表现竞争力，尽管低于先前的准确性。人类对理由质量的评分与准确性的相关性不够一致，并且对理由长度给予了不成比例的重视。结果可转移至一个独立的预测研究。EQMs提供了一种可扩展的、可解释的方法，用于从书面解释中提取与判断相关的信息。
cs.CL / 14 / 2606.30989
Wait, am I Being Fair? Characterizing Deductive Stereotyping and Mitigating It with Fair-GCG
等等，我在公平吗？刻画演绎性刻板印象及其与公平-GCG的缓解
Abstract
Warning: This paper contains several toxic and offensive statements. While reasoning generally improves fairness in recent large language models (LLMs), failures persist. In this work, we identify a failure mode, deductive stereotyping, in which models apply population-level statistical regularities to individual cases, producing logically coherent yet socially biased inferences. We provide a statistical interpretation of this phenomenon. To steer models toward fairness-aware reasoning, we propose a reasoning-time injection framework. We further introduce Fair-GCG to systematically discover effective injection phrases. Injection phrases discovered by Fair-GCG improve performance across multiple fairness benchmarks, generalize from smaller to larger LLMs, improves reasoning-level fairness, reduces bias in open-ended generation, and transfer to real-world fairness-sensitive tasks.
Chinese Translation
警告：本文包含若干有毒和冒犯性的陈述。尽管推理通常能改善近期大型语言模型（LLMs）的公平性，但仍然存在失败。在本研究中，我们识别了一种失败模式，即演绎性刻板印象，其中模型将群体层面的统计规律应用于个体案例，产生逻辑上连贯但社会上偏见的推论。我们对这一现象提供了统计解释。为了引导模型朝向公平意识的推理，我们提出了一种推理时注入框架。我们进一步引入了公平-GCG，以系统性地发现有效的注入短语。通过公平-GCG发现的注入短语在多个公平性基准测试中提高了性能，从较小的LLMs到较大的LLMs具有良好的泛化能力，改善了推理层面的公平性，减少了开放式生成中的偏见，并能够迁移到现实世界的公平性敏感任务中。
cs.CL / 15 / 2606.31033
CORTEX: Token-Level Hallucination Detection in RAG via Comparative Internal Representations
CORTEX：通过比较内部表征实现RAG中的令牌级幻觉检测
Abstract
In this paper, we propose CORTEX, a token-level hallucination detection method for Retrieval-Augmented Generation (RAG). In long-form RAG outputs, hallucinations often arise in localized spans rather than throughout an entire response. CORTEX therefore identifies ungrounded content at the token level, enabling fine-grained localization of hallucinations. The key intuition behind CORTEX is that tokens grounded in retrieved documents should be more strongly influenced by those documents than hallucinated tokens. To capture this document-induced effect, CORTEX compares internal representations of a large language model (LLM) under two conditions: with and without the retrieved documents. Instead of relying solely on each token's immediate sensitivity to the retrieved documents, CORTEX also leverages the propagation of document-grounded information through preceding tokens, reducing false positives for tokens whose evidence has already been absorbed into the context. Finally, CORTEX applies post-processing smoothing step that models the tendency of hallucination labels to persist over contiguous spans, reducing local noise and encouraging span-consistent predictions. Experiments on two RAG benchmarks and three LLMs show that CORTEX substantially improves token-level hallucination detection, with each component consistently contributing to performance gains.
Chinese Translation
在本文中，我们提出了CORTEX，一种用于检索增强生成（RAG）的令牌级幻觉检测方法。在长文本RAG输出中，幻觉通常出现在局部范围内，而不是整个响应中。因此，CORTEX在令牌级别识别无根据的内容，从而实现幻觉的细粒度定位。CORTEX的关键直觉是，基于检索文档的令牌应该比幻觉令牌更受这些文档的影响。为了捕捉这种文档引起的效应，CORTEX在两种条件下比较大型语言模型（LLM）的内部表征：有和没有检索文档。CORTEX不仅依赖于每个令牌对检索文档的即时敏感性，还利用文档基础信息通过前面的令牌的传播，减少已经被吸收到上下文中的令牌的误报。最后，CORTEX应用了后处理平滑步骤，建模幻觉标签在连续范围内持续存在的趋势，减少局部噪声并鼓励范围一致的预测。在两个RAG基准和三个LLM上的实验表明，CORTEX显著提高了令牌级幻觉检测，每个组件都持续对性能提升做出了贡献。
cs.CL / 16 / 2606.31039
Truth or Sophistry? LoFa: A Benchmark for LLM Robustness Against Logical Fallacies
真相还是诡辩？LoFa：一个评估大型语言模型对逻辑谬误鲁棒性的基准
Abstract
Large Language Models (LLMs) exhibit strong semantic capabilities, yet their resilience to manipulative linguistic patterns such as logical fallacies remains underexplored. Prior work has primarily examined whether LLMs can identify or classify fallacies, leaving their robustness against fallacious persuasion insufficiently studied. To address this gap, we introduce LoFa (Logical Fallacy), a comprehensive benchmark for evaluating LLM robustness against fallacies. LoFa is constructed through a multi-agent pipeline that pairs factual questions with fallacious arguments, and is accompanied by a multi-round debate framework for assessing model resilience under sustained adversarial persuasion. To disentangle fallacy robustness from a model's inherent knowledge limitations, we further propose Logical Fallacy Resistance at k (LFR@k), a metric that quantifies resistance to fallacious attacks. Experiments show that LLMs exhibit varying levels of robustness across different fallacy types, revealing distinct vulnerability profiles among models.
Chinese Translation
大型语言模型（LLMs）展现出强大的语义能力，但它们对操控性语言模式（如逻辑谬误）的抵抗力仍然未得到充分探索。之前的研究主要关注LLMs是否能够识别或分类谬误，而对其在谬误性说服下的鲁棒性研究不足。为了解决这一问题，我们提出了LoFa（逻辑谬误），这是一个全面的基准，用于评估LLMs对谬误的鲁棒性。LoFa通过一个多智能体管道构建，将事实问题与谬误论证配对，并配备了一个多轮辩论框架，以评估模型在持续对抗性说服下的韧性。为了将谬误鲁棒性与模型固有的知识限制区分开来，我们进一步提出了逻辑谬误抵抗度k（LFR@k），这一指标量化了对谬误攻击的抵抗力。实验表明，LLMs在不同类型的谬误中表现出不同程度的鲁棒性，揭示了模型之间的独特脆弱性特征。
cs.CL / 17 / 2606.31041
A Semantic-Layer-Mediated Agent for Natural Language to SQL over Heterogeneous Enterprise Databases
一种语义层中介的自然语言到SQL代理，用于异构企业数据库
Abstract
Natural language-to-SQL (NL2SQL) over real-world enterprise databases remains significantly more challenging than on academic benchmarks. Enterprise schemas often contain hundreds of physical tables with cryptic column names, heterogeneous SQL dialects, and complex analytical workloads requiring nested aggregations, temporal reasoning, and multi-table joins. We present a semantic-layer-mediated NL2SQL agent that decouples semantic intent from physical SQL execution. Rather than generating SQL directly over raw schemas, the agent reasons over a curated semantic layer through a compact intermediate representation called the Semantic Model Query (SMQ). A deterministic compiler translates each SMQ into dialect-specific SQL, providing verified building blocks that the agent composes into the final query. The system employs a constrained think-act loop, supports SQLite, BigQuery, and Snowflake backends, and is integrated into an end-to-end evaluation framework. Using Gemini 3 Pro, the system achieves 94.15% execution accuracy on the 547-task Spider2-snow benchmark, ranking third on the official leaderboard and substantially outperforming schema-only approaches. We describe the system architecture, SMQ representation, agent workflow, evaluation results, and discuss semantic-layer quality and the trade-off between improved grounding and overfitting.
Chinese Translation
在真实世界的企业数据库上，自然语言到SQL（NL2SQL）仍然比在学术基准上更具挑战性。企业模式通常包含数百个物理表，列名晦涩，SQL方言异构，并且复杂的分析工作负载需要嵌套聚合、时间推理和多表连接。我们提出了一种语义层中介的NL2SQL代理，它将语义意图与物理SQL执行解耦。该代理不是直接在原始模式上生成SQL，而是通过一个称为语义模型查询（Semantic Model Query, SMQ）的紧凑中间表示在经过整理的语义层上进行推理。一个确定性的编译器将每个SMQ转换为特定方言的SQL，提供经过验证的构建块，代理将这些构建块组合成最终查询。该系统采用受限的思考-行动循环，支持SQLite、BigQuery和Snowflake后端，并集成到一个端到端的评估框架中。使用Gemini 3 Pro，该系统在547个任务的Spider2-snow基准上实现了94.15%的执行准确率，在官方排行榜上排名第三，显著优于仅基于模式的方法。我们描述了系统架构、SMQ表示、代理工作流程、评估结果，并讨论了语义层质量以及改进基础和过拟合之间的权衡。
cs.CL / 18 / 2606.31055
Reference-Based Prosody and Rhythm Evaluation for Spoken Dialogue Systems
基于参考的语调和节奏评估用于对话系统
Abstract
Speech-to-speech (S2S) AI agents are advancing rapidly, yet evaluation lacks interpretable speech-native measures for conversational prosody and rhythm. Because $F_0$, speaking rate, articulation rate, and pausing shift with model-predicted speaker traits and interaction state, pooled human statistics can be poorly calibrated for evaluating a particular output. Using 4000+ hours of dyadic English conversation from the Seamless Interaction dataset, we construct matched reference regimes for $F_0$ mean, $F_0$ expressivity, speech rate, articulation rate, pause ratio, and mean pause duration. We then define a percentile-based evaluation protocol: extract the same metrics from an S2S output waveform, compare them to the closest matched human reference stratum, and report percentile deviations or 5th-95th percentile out-of-regime flags. On held-out human rows, pooled references over-flag state-conditioned $F_0$ expressivity and rhythm, while matched references return flag rates closer to the nominal 10% and make deviation direction interpretable. These outputs serve as behavioral plausibility checks that complement, rather than replace, perceptual and user-centered evaluation.
Chinese Translation
语音到语音（S2S）人工智能代理正在快速发展，但评估缺乏可解释的、适用于对话的语调和节奏的测量标准。由于 $F_0$、语速、发音速率和停顿会随着模型预测的说话者特征和交互状态而变化，因此汇总的人类统计数据在评估特定输出时可能校准不佳。利用来自无缝交互数据集的4000多个小时的双人英语对话，我们构建了与 $F_0$ 均值、$F_0$ 表现力、语速、发音速率、停顿比率和平均停顿时长相匹配的参考标准。然后，我们定义了一种基于百分位数的评估协议：从 S2S 输出波形中提取相同的指标，将其与最接近的匹配人类参考层进行比较，并报告百分位数偏差或5th-95th百分位数的超出范围标记。在保留的人类样本上，汇总参考在状态条件下过度标记了 $F_0$ 表现力和节奏，而匹配参考的标记率更接近名义上的10%，并使偏差方向可解释。这些输出作为行为合理性检查，补充而非替代感知和以用户为中心的评估。
cs.CL / 19 / 2606.31058
Exploring the relationship between team institutional composition and novelty in academic papers based on fine-grained knowledge entities
基于细粒度知识实体探讨团队机构组成与学术论文新颖性之间的关系
Abstract
The composition of author teams is an important factor influencing the novelty of academic papers. However, existing studies have paid limited attention to the role of institutional composition, and most novelty measures remain at a general level, making it difficult to explain the specific sources and types of novelty in papers. Taking the field of natural language processing as an example, this study investigates the relationship between team institutional composition and the fine-grained novelty of academic papers. Author teams are classified into three types: academic institutions, industrial institutions, and mixed academic and industrial institutions. Four types of fine-grained knowledge entities are extracted from full-text papers, including methods, datasets, tools, and metrics. The novelty of papers is then measured based on entity combinations, and pairwise combinations of different entity types are further analyzed to examine their contributions to novel papers. The results show that, in the field of natural language processing, collaboration between industrial and academic institutions is more likely to produce novel papers than purely industrial collaboration. From the perspective of fine-grained knowledge entities, mixed academic and industrial teams pay more attention to the novelty of method-metric combinations, whereas industrial teams pay more attention to the novelty of method-tool combinations. This study reveals the relationship between institutional team composition and paper novelty through fine-grained novelty measurement, providing useful evidence for improving paper quality and promoting industry-academia-research collaboration.
Chinese Translation
作者团队的组成是影响学术论文新颖性的重要因素。然而，现有研究对机构组成的作用关注有限，大多数新颖性测量仍停留在一般层面，难以解释论文中新颖性的具体来源和类型。本研究以自然语言处理领域为例，探讨团队机构组成与学术论文细粒度新颖性之间的关系。作者团队被分为三种类型：学术机构、工业机构和混合学术与工业机构。从全文论文中提取四种细粒度知识实体，包括方法、数据集、工具和指标。论文的新颖性基于实体组合进行测量，并进一步分析不同实体类型的成对组合，以考察它们对新颖论文的贡献。结果表明，在自然语言处理领域，工业与学术机构的合作更可能产生新颖论文，而纯粹的工业合作则较少。从细粒度知识实体的角度来看，混合学术与工业团队更关注方法-指标组合的新颖性，而工业团队则更关注方法-工具组合的新颖性。本研究通过细粒度新颖性测量揭示了机构团队组成与论文新颖性之间的关系，为提高论文质量和促进产学研合作提供了有益的证据。
cs.CL / 20 / 2606.31069
Building a Multimodal Dataset of Academic Paper for Keyword Extraction
构建学术论文的多模态数据集用于关键词提取
Abstract
Up to this point, keyword extraction task typically relies solely on textual data. Neglecting visual details and audio features from image and audio modalities leads to deficiencies in information richness and overlooks potential correlations, thereby constraining the model's ability to learn representations of the data and the accuracy of model predictions. Furthermore, the currently available multimodal datasets for keyword extraction task are particularly scarce, further hindering the progress of research on multimodal keyword extraction task. Therefore, this study constructs a multimodal dataset of academic paper consisting of 1000 samples, with each sample containing paper text, images, audios and keywords. Based on unsupervised and supervised methods of keyword extraction, experiments are conducted using textual data from papers, as well as text extracted from images and audio. The aim is to investigate the differences in performance in keyword extraction task with respect to different modal information and the fusion of multimodal information. The experimental results indicate that text from different modalities exhibits distinct characteristics in the model. The concatenation of paper text, image text and audio text can effectively enhance the keyword extraction performance of academic papers.
Chinese Translation
迄今为止，关键词提取任务通常仅依赖于文本数据。忽视来自图像和音频模态的视觉细节和音频特征导致信息丰富性不足，并且忽略了潜在的相关性，从而限制了模型学习数据表示的能力和模型预测的准确性。此外，目前可用的多模态关键词提取任务数据集特别稀缺，进一步阻碍了多模态关键词提取研究的进展。因此，本研究构建了一个包含1000个样本的学术论文多模态数据集，每个样本包含论文文本、图像、音频和关键词。基于无监督和监督的关键词提取方法，使用论文中的文本数据以及从图像和音频中提取的文本进行实验。目的是研究在关键词提取任务中，不同模态信息及其融合对性能的影响。实验结果表明，不同模态的文本在模型中表现出不同的特征。论文文本、图像文本和音频文本的拼接可以有效提升学术论文的关键词提取性能。
cs.CL / 21 / 2606.31074
Triospect: A Three-Dimensional Framework for Robust Statistical AI-Generated Text Detection Against Diverse Attacks
Triospect：一种针对多样攻击的鲁棒统计AI生成文本检测的三维框架
Abstract
Existing AI-generated text detectors are vulnerable to attacks that manipulate textual characteristics. In this study, we propose a novel Triospect Detection Framework by using additional perspectives of content (core ideas) and expression (stylistic elements) within a given text. Experiments on two benchmarks involving 17 attacks, 12 domains, and 17 source models demonstrate that Triospect is robust against these attacks. It improves the strong baseline by a significant margin of 22.3% (AUROC) and 13% (TPR01) on the Humanize-16K after-attack subset, and by 9.1% (AUROC) and 22% (TPR01) on the adversarial RAID. This framework marks a pioneering effort in statistical methods to enhance detection reliability against attacks. We release our data and code at https://github.com/baoguangsheng/triospect.
Chinese Translation
现有的AI生成文本检测器易受到操纵文本特征的攻击。在本研究中，我们提出了一种新颖的Triospect检测框架，通过利用给定文本中的内容（核心思想）和表达（风格元素）的额外视角。针对涉及17种攻击、12个领域和17个源模型的两个基准进行的实验表明，Triospect对这些攻击具有鲁棒性。在Humanize-16K的攻击后子集上，它将强基线的性能显著提升了22.3%（AUROC）和13%（TPR01），在对抗性RAID上提升了9.1%（AUROC）和22%（TPR01）。该框架标志着在统计方法中增强检测可靠性以应对攻击的开创性努力。我们在https://github.com/baoguangsheng/triospect发布了我们的数据和代码。
cs.CL / 22 / 2606.31087
When Reranking Hurts: Uncertainty-Based Gating for Few-Shot Reranking
当重新排序有害时：基于不确定性的少样本重新排序门控
Abstract
Few-shot selection typically assumes that reranking retrieved examples always improves performance. We challenge this view by identifying that the expensive reranking step can in fact degrade performance. Instead, we propose \emph{Training-Free Gated Reranking}, which decides whether to rerank the few-shot examples based on the model's uncertainty. Extensive experiments across 8 LLMs, covering 7 NLU datasets and 9 MT domain-language combinations, demonstrate that our approach reduces computational costs by 15\%-80\% while improving average performance by up to 2\%. These findings indicate that higher computational cost does not guarantee better performance, and that reranking is most beneficial when targeted at high-uncertainty instances.
Chinese Translation
少样本选择通常假设重新排序检索到的示例总是能提高性能。我们质疑这一观点，指出昂贵的重新排序步骤实际上可能会降低性能。相反，我们提出了 extit{无训练门控重新排序}（Training-Free Gated Reranking），该方法根据模型的不确定性决定是否对少样本示例进行重新排序。在涵盖7个自然语言理解（NLU）数据集和9个机器翻译（MT）领域-语言组合的8个大型语言模型（LLMs）上的大量实验表明，我们的方法在提高平均性能最多2%的同时，将计算成本降低了15%-80%。这些发现表明，更高的计算成本并不保证更好的性能，而重新排序在针对高不确定性实例时最为有益。
cs.CL / 23 / 2606.31112
What Counts as an Error? Dual-Reference Benchmarking for Atypical ASR
什么算作错误？非典型自动语音识别的双重参考基准测试
Abstract
ASR systems have been often reported to underperform on atypical speech. An often conflated compounding factor is the existence of two valid transcription references: verbatim (actual produced speech, including repetitions/prolongations) and intended (the canonical form of the text with disfluencies removed) in atypical speech recognition depending on context and use-case. Most ASR evaluations conflate this duality into a single ground truth and reward systems that delete disfluencies, ignoring verbatim faithfulness. We benchmark 11 ASR models from encoder-decoder, CTC and transducer families using both verbatim and intended references on atypical stuttered speech as a case study. Our quantitative assessment underlines the disparity in model performance and rankings using the two transcript styles. Through this analysis, we highlight the importance of selecting a suitable transcription reference for valid model selection depending on the use-case, particularly for atypical ASR.
Chinese Translation
自动语音识别（ASR）系统在处理非典型语音时常常表现不佳。一个常被混淆的因素是存在两个有效的转录参考：逐字转录（实际产生的语音，包括重复/延长）和意图转录（去除不流畅性的文本规范形式），这取决于上下文和使用案例。在大多数ASR评估中，这种双重性被混合为单一的真实值，并奖励那些删除不流畅性的系统，忽视了逐字忠实性。我们以非典型口吃语音为案例研究，基于逐字和意图转录对11个来自编码器-解码器、CTC和转换器家族的ASR模型进行基准测试。我们的定量评估强调了使用这两种转录风格时模型性能和排名的差异。通过这一分析，我们强调根据使用案例选择合适的转录参考对于有效模型选择的重要性，特别是在非典型ASR的情况下。
cs.CL / 24 / 2606.31145
SeKV: Resolution-Adaptive KV Cache with Hierarchical Semantic Memory for Long-Context LLM Inference
SeKV：具有分层语义记忆的分辨率自适应 KV 缓存用于长上下文 LLM 推理
Abstract
Large language models increasingly operate over long contexts, where the KV cache becomes a dominant memory bottleneck: its size grows linearly with sequence length and must be retained throughout decoding, making full GPU caching prohibitively expensive without compression. Existing KV cache compression methods struggle to balance efficiency with faithful context preservation. Token eviction discards information, while semantic grouping fixes compression decisions at prefill time; neither can recover token-level detail from a compressed span once it becomes relevant during generation. As a solution, we propose SeKV, a resolution-adaptive semantic KV cache that organizes context into entropy-guided semantic spans and stores them across a GPU-CPU memory hierarchy without discarding information. Each span keeps a lightweight summary vector on GPU for coarse routing and a low-rank SVD basis on CPU for on-demand token-level reconstruction. A trained zoom-in mechanism selectively expands query-relevant spans during decoding, enabling precise retrieval without materializing the full KV cache on GPU. SeKV enables adaptive token-level reconstruction while keeping the base LLM fully frozen and adding fewer than 0.05% trainable parameters. Across four benchmarks, SeKV improves over the strongest semantic compression baseline by 5.9% on average while reducing GPU memory by 53.3% versus full KV caching at 128K context. Code is available on https://github.com/AmirAbaskohi/SeKV.
Chinese Translation
大型语言模型越来越多地在长上下文中运行，其中 KV 缓存成为主要的内存瓶颈：其大小随着序列长度线性增长，并且必须在解码过程中保持，这使得在没有压缩的情况下，完整的 GPU 缓存变得极其昂贵。现有的 KV 缓存压缩方法在效率与忠实上下文保留之间难以取得平衡。令牌驱逐会丢失信息，而语义分组在预填充时固定了压缩决策；一旦在生成过程中变得相关，均无法从压缩范围中恢复令牌级细节。为此，我们提出了 SeKV，一种分辨率自适应的语义 KV 缓存，它将上下文组织为基于熵引导的语义跨度，并在 GPU-CPU 内存层次结构中存储这些跨度，而不丢失信息。每个跨度在 GPU 上保持一个轻量级的摘要向量用于粗略路由，并在 CPU 上保持一个低秩 SVD 基础用于按需的令牌级重构。经过训练的放大机制在解码过程中选择性地扩展与查询相关的跨度，从而实现精确检索，而无需在 GPU 上实现完整的 KV 缓存。SeKV 使得在保持基础 LLM 完全冻结的同时实现自适应的令牌级重构，并且增加的可训练参数少于 0.05%。在四个基准测试中，SeKV 在平均上比最强的语义压缩基线提高了 5.9%，同时在 128K 上下文下将 GPU 内存减少了 53.3%。代码可在 https://github.com/AmirAbaskohi/SeKV 获取。
cs.CL / 25 / 2606.31166
TAG-DLM: Diffusion Language Models for Text-Attributed Graph Learning
TAG-DLM：用于文本属性图学习的扩散语言模型
Abstract
Text-attributed graphs (TAGs), where each node carries a natural language description, require models to jointly reason over text and graph topology. Existing approaches often handle the two modalities separately: graph neural networks operate on shallow text features, while hybrids of LLMs and graphs use the language model mainly as a text encoder and delegate structure learning to a separate graph module. We propose method that unifies textual reasoning and graph message passing within a masked diffusion language model, a language model with bidirectional attention and generative decoding. For each graph instance, method linearises a sampled local neighbourhood into a token sequence and injects graph structure through a topology attention mask, which realises message passing over the graph. Because the diffusion language model can both interpret and generate text, the method adapts to different tasks simply by changing the prompt, supporting node classification, link prediction, and cross-dataset transfer with no target-specific fine-tuning. Experiments show that method outperforms graph neural networks, graph transformers, and LLM-based baselines on all three TAG benchmarks across two tasks, improving over the strongest baseline by up to 3.9 points.
Chinese Translation
文本属性图（TAGs）是指每个节点都携带自然语言描述的图，这要求模型能够对文本和图的拓扑结构进行联合推理。现有的方法通常分别处理这两种模态：图神经网络在浅层文本特征上操作，而大型语言模型（LLMs）与图的混合使用语言模型主要作为文本编码器，并将结构学习委托给单独的图模块。我们提出了一种方法，将文本推理和图消息传递统一在一个掩蔽扩散语言模型中，该语言模型具有双向注意力和生成解码功能。对于每个图实例，该方法将采样的局部邻域线性化为一个令牌序列，并通过拓扑注意力掩码注入图结构，从而实现图上的消息传递。由于扩散语言模型能够同时解释和生成文本，该方法仅通过更改提示即可适应不同任务，支持节点分类、链接预测和跨数据集迁移，而无需针对特定目标进行微调。实验表明，该方法在两个任务的所有三个TAG基准上均优于图神经网络、图变换器和基于LLM的基线，较最强基线提高了多达3.9个点。
cs.CL / 26 / 2606.31186
Gated Multi-Graph Fusion via Graph Attention Networks for Alzheimer's Disease Detection
通过图注意力网络的门控多图融合用于阿尔茨海默病检测
Abstract
Spontaneous speech is a vital non-invasive biomarker for Alzheimer's Disease (AD), yet many systems overlook non-linear structural disruptions and clinical heterogeneity in pathological language. We propose a Multi-View Gated Graph Attention Network that transcribes audio via Automatic Speech Recognition (ASR) to construct semantic, dependency, and co-occurrence graphs, characterizing speech through a "content-structure-flow" framework. Notably, the co-occurrence graph leverages Pointwise Mutual Information (PMI) from a normative corpus to quantify narrative logic and linguistic deviation. To address symptomatic diversity, an adaptive gated fusion mechanism dynamically integrates these views. Evaluated on the ADReSSo dataset, our model achieves 90.00% accuracy. Ablation results confirm that the PMI-based graph and heterogeneity-aware gating are essential for robust classification across diverse clinical populations. Our source code is publicly available at https://github.com/opeacc/AD.
Chinese Translation
自发语言是阿尔茨海默病（AD）的一种重要非侵入性生物标志物，但许多系统忽视了病理语言中的非线性结构破坏和临床异质性。我们提出了一种多视角门控图注意力网络，该网络通过自动语音识别（ASR）转录音频，构建语义图、依赖图和共现图，通过“内容-结构-流”框架对语言进行表征。值得注意的是，共现图利用来自规范语料库的点互信息（PMI）来量化叙事逻辑和语言偏差。为了解决症状多样性问题，适应性门控融合机制动态整合这些视角。在ADReSSo数据集上的评估显示，我们的模型达到了90.00%的准确率。消融实验结果确认，基于PMI的图和关注异质性的门控机制对于在不同临床人群中实现稳健分类至关重要。我们的源代码已公开，地址为 https://github.com/opeacc/AD。
cs.CL / 27 / 2606.31213
Can LLMs Imagine Moral Alternatives Beyond Binary Dilemmas?
大型语言模型能否想象超越二元困境的道德替代方案？
Abstract
As large language models (LLMs) are increasingly deployed as moral advisors and agents, they need to address dilemmas between two competing values. However, existing research on LLMs with moral dilemmas overlooks a central aspect of human moral cognition: the ability to imagine alternatives that move beyond the given options. We introduce MoralAltDataset, a dataset of 307 moral dilemmas spanning narrative Advisor dilemmas and AI-facing Agent dilemmas, each augmented with compromise and reframed alternatives. We first examine whether humans and LLMs shift their judgments when such alternatives are introduced. Across 15 LLMs, we find that compromise alternatives are often preferred over either original option, substantially reshaping moral choice. We then evaluate the quality of LLM-generated alternatives against human-authored ones using pairwise preference and expert-based criteria. Results show that LLM-generated alternatives are often preferred and better satisfy fine-grained structural and ethical criteria, while revealing trade-offs between structural quality and practical feasibility.
Chinese Translation
随着大型语言模型（LLMs）越来越多地被用作道德顾问和代理，它们需要解决两个竞争价值之间的困境。然而，现有关于 LLMs 与道德困境的研究忽视了人类道德认知的一个核心方面：想象超越给定选项的替代方案的能力。我们引入了 MoralAltDataset，这是一个包含307个道德困境的数据集，涵盖叙事顾问困境和面向 AI 的代理困境，每个困境都附加了妥协和重新框架的替代方案。我们首先考察人类和 LLMs 在引入这些替代方案时是否会改变他们的判断。在15个 LLMs 中，我们发现妥协替代方案通常比原始选项更受欢迎，显著重塑了道德选择。接着，我们使用成对偏好和专家标准评估 LLM 生成的替代方案与人类创作的替代方案的质量。结果表明，LLM 生成的替代方案通常更受欢迎，并且更好地满足细致的结构和伦理标准，同时揭示了结构质量与实际可行性之间的权衡。
cs.CL / 28 / 2606.31250
Probing Stylistic Appropriation using Large Language Models: An Evaluation Framework for Copyright Infringement under EU Law
利用大型语言模型探讨风格挪用：欧盟法律下的版权侵权评估框架
Abstract
Large language models (LLM) trained on web-scale corpora generate output that may infringe copyright, yet existing technical safeguards focus narrowly on verbatim memorisation. EU copyright doctrine applies a broader standards: substantial similarity, which extends to stylistic choices, narrative structure, and creative elaboration. This mismatch between what current methods detect and what the law protects leaves a significant compliance gap. We introduce PSALM, an LLM-as-a-judge framework that operationalises EU copyright doctrine through ten evaluators assessing computational overlap, stylistic dimensions (writing style, narrative voice), content dimensions (character, plot, scene, world building), and statutory exceptions (parody, pastiche, quotation, sc\`enes \`a faire). Applying PSALM to Llama~3.2 models fine-tuned on translated historical Dutch literary works, we find that: 1) instruction-tuned models exhibit non-trivial baseline stylistic similarity prior to corpus exposure; 2) fine-tuning induces systematic stylistic appropriation across all infringement-relevant dimensions, extending beyond verbatim memorisation to abstract narrative patterns; 3) Negative Preference Optimisation unlearning substantially reduces similarity but leaves detectable residual stylistic patterns. These findings indicate that safeguards targeting literal copying alone are insufficient to mitigate broader copyright risks. PSALM provides infrastructure for auditable, legally informed compliance evaluation, though the relationship between automated similarity scores and infringement determinations requires validation by legal experts. This work bridges qualitative legal standards and quantitative technical measurement, exposing fundamental tensions between generative AI and EU intellectual property law.
Chinese Translation
在网络规模语料库上训练的大型语言模型（LLM）生成的输出可能侵犯版权，但现有的技术保护措施主要集中在逐字记忆上。欧盟版权法则适用更广泛的标准：实质性相似性，这包括风格选择、叙事结构和创意扩展。这种当前方法检测与法律保护之间的错位导致了显著的合规差距。我们提出了PSALM，一个将欧盟版权法通过十个评估者操作化的LLM作为评判框架，评估计算重叠、风格维度（写作风格、叙事声音）、内容维度（角色、情节、场景、世界构建）以及法定例外（戏仿、拼贴、引用、必然场景）。将PSALM应用于针对翻译的历史荷兰文学作品微调的Llama~3.2模型，我们发现：1）指令微调模型在语料库曝光之前表现出非平凡的基线风格相似性；2）微调在所有与侵权相关的维度上引发系统性的风格挪用，超越了逐字记忆，延伸至抽象叙事模式；3）负偏好优化的去学习显著减少了相似性，但留下可检测的残余风格模式。这些发现表明，仅针对字面复制的保护措施不足以减轻更广泛的版权风险。PSALM为可审计的、法律知情的合规评估提供了基础设施，尽管自动相似性评分与侵权判定之间的关系需要法律专家的验证。这项工作架起了定性法律标准与定量技术测量之间的桥梁，揭示了生成性人工智能与欧盟知识产权法之间的根本紧张关系。
cs.CL / 29 / 2606.31307
When the Database Fails: Prompting LLM Dialogue Agents for Safe Recovery in Task-Oriented Dialogue
当数据库失败时：在任务导向对话中促使大型语言模型对话代理安全恢复
Abstract
Large language models used in task-oriented dialogue often produce fluent but unsafe responses when backend database calls fail, return empty results, or surface mismatched information, inventing venues, confirmations, or booking details not grounded in the database. We study a lightweight prompting-based recovery approach that improves robustness without retraining or additional model calls. We compare three response strategies, including a guided recovery prompt conditioned on structured database status, across six open-weight model families (DeepSeek-R1, Gemma-2, Llama-3, Mistral, Phi-3, and Qwen-2.5) and four database conditions: empty result, wrong-domain retrieval, API error, and clean retrieval. Using fault-injected benchmarks built on two structurally different datasets, MultiWOZ 2.2 (5 domains) and SGD (20 domains), we find that naive agents hallucinate on 30.5% of failure turns on MultiWOZ and 20.9% on SGD. Our Guided-Retry strategy reduces hallucination by 50% on MultiWOZ (30.5 to 15.3%) and by 42% on SGD (20.9 to 12.2%) without retraining. However, residual hallucination remains substantial (6-37% across models), with wrong-domain failures the hardest case. Results are consistent across both datasets and all six model families, and human annotation shows substantial agreement while supporting the validity of the automatic commitment-safety metric.
Chinese Translation
在任务导向对话中使用的大型语言模型在后端数据库调用失败、返回空结果或出现不匹配信息时，往往会产生流畅但不安全的响应，虚构与数据库无关的场所、确认或预订细节。我们研究了一种基于轻量级提示的恢复方法，该方法在不重新训练或增加模型调用的情况下提高了鲁棒性。我们比较了三种响应策略，包括基于结构化数据库状态的引导恢复提示，涵盖六个开放权重模型系列（DeepSeek-R1、Gemma-2、Llama-3、Mistral、Phi-3 和 Qwen-2.5）以及四种数据库条件：空结果、错误领域检索、API 错误和干净检索。通过在两个结构上不同的数据集（MultiWOZ 2.2（5个领域）和 SGD（20个领域））上构建的故障注入基准，我们发现天真的代理在 MultiWOZ 的 30.5% 的失败回合和 SGD 的 20.9% 的失败回合中出现了幻觉。我们的引导重试策略在 MultiWOZ 上将幻觉减少了 50%（从 30.5% 降至 15.3%），在 SGD 上减少了 42%（从 20.9% 降至 12.2%），且无需重新训练。然而，残余幻觉仍然相当可观（在不同模型间为 6-37%），错误领域的失败是最难处理的情况。结果在两个数据集和所有六个模型系列中保持一致，人类注释显示出显著的一致性，同时支持自动承诺安全性指标的有效性。
cs.CL / 30 / 2606.31310
LOPA: Enhancing Spoken Language Assessment via Latent Ordinal Prototype Alignment
LOPA：通过潜在序数原型对齐增强口语语言评估
Abstract
Fueled by increasing model scale and multimodal inputs, Multimodal Large Language Models (MLLMs) have emerged as a promising paradigm for Spoken Language Assessment (SLA). While effective, this paradigm often overlooks the intrinsic ordinal structure of language acquisition. This paper works around the necessity of large-scale MLLMs by introducing Latent Ordinal Prototype Alignment (LOPA) for SLA, a prototype-based regularizer that enforces an ordinal geometric prior directly on the latent space. Coupled with Semantic-Anchored Layer Routing (SALR), which adaptively harvests multi-depth representations from a frozen Whisper encoder, our framework achieves an RMSE of 0.361. This performance rivals billion-parameter systems without the need for LLM-based fine-tuning. Further analysis reveals that SALR's synergy with LOPA offers interpretable, criterion-aligned preferences, thereby supporting an efficient and ordinal-aware modeling alternative to current scaling-centric models for SLA.
Chinese Translation
随着模型规模的扩大和多模态输入的增加，多模态大型语言模型（MLLMs）已成为口语语言评估（SLA）的一种有前景的范式。尽管有效，但这一范式往往忽视了语言习得的内在序数结构。本文通过引入潜在序数原型对齐（LOPA）为SLA提供了一种基于原型的正则化方法，该方法直接在潜在空间上施加序数几何先验，从而绕过了对大规模MLLMs的需求。结合语义锚定层路由（SALR），该方法自适应地从冻结的Whisper编码器中提取多深度表示，我们的框架实现了0.361的均方根误差（RMSE）。这一性能与十亿参数系统相媲美，而无需基于LLM的微调。进一步分析表明，SALR与LOPA的协同作用提供了可解释的、与标准对齐的偏好，从而为当前以规模为中心的SLA模型提供了一种高效且关注序数的建模替代方案。
cs.CL / 31 / 2606.31315
BlockPilot: Instance-Adaptive Policy Learning for Diffusion-based Speculative Decoding
BlockPilot：基于扩散的投机解码的实例自适应策略学习
Abstract
Speculative decoding accelerates inference by using a lightweight draft model to generate candidate tokens in parallel, and are then verified by the target model, enabling lossless acceleration. Recently, diffusion-based speculative decoding further improves parallelism by generating multiple tokens per forward pass via block-level diffusion, achieving state-of-the-art (SOTA) performance. However, existing methods adopt a fixed inference block size and assume a uniform optimal decoding strategy across all inputs. In this paper, we show that this assumption is suboptimal, as the optimal block size varies across samples and plays a critical role in speculative decoding performance. Moreover, these values exhibit a clear local structure, concentrating around the training block size, which reduces the problem to a low-dimensional and structured decision space. Based on these insights, we propose BlockPilot, a sample-adaptive policy that predicts the optimal block size from the prefilling representation. Specifically, we formulate block size selection as a lightweight policy learning problem and propose an instance-adaptive decision mechanism that predicts the optimal block size based on the representation of the prefilling stage. The prediction is performed only once after prefilling, allowing for seamless integration. Extensive experiments demonstrate that our method is plug-and-play, introduces minimal overhead, and consistently improves efficiency, achieving an acceptance length of 5.92 and a 4.20$\times$ speedup on Qwen3-4B under temperature $T=1$.
Chinese Translation
投机解码通过使用轻量级草稿模型并行生成候选标记，从而加速推理，并由目标模型进行验证，实现无损加速。最近，基于扩散的投机解码通过块级扩散在每次前向传播中生成多个标记，进一步提高了并行性，达到了最先进的（SOTA）性能。然而，现有方法采用固定的推理块大小，并假设所有输入具有统一的最佳解码策略。本文表明，这一假设并不理想，因为最佳块大小在样本之间是不同的，并且在投机解码性能中起着关键作用。此外，这些值表现出明显的局部结构，集中在训练块大小附近，从而将问题简化为低维且结构化的决策空间。基于这些见解，我们提出了BlockPilot，一种样本自适应策略，从预填充表示中预测最佳块大小。具体而言，我们将块大小选择形式化为一个轻量级策略学习问题，并提出了一种实例自适应决策机制，根据预填充阶段的表示预测最佳块大小。该预测仅在预填充后执行一次，允许无缝集成。大量实验表明，我们的方法即插即用，带来的开销极小，并且始终提高了效率，在温度$T=1$下，在Qwen3-4B上实现了5.92的接受长度和4.20$ imes$的加速。
cs.CL / 32 / 2606.31411
Linguistic Bias Mitigation for Spoofing Detection via Gradient Reversal and A Variational Information Bottleneck
通过梯度反转和变分信息瓶颈缓解欺骗检测中的语言偏见
Abstract
Rapid advancements in generative speech technology have compromised the reliability of voice biometrics. While current spoofing detectors excel when assessed under in-domain conditions, generalisation to out-of-domain settings is often poor. We show that this can be due to linguistic bias. A reliance on linguistic cues observed in training data can then compromise robustness to cross-data. We propose a linguistic-invariant spoofing detection framework utilizing teacher-student adversarial learning. The linguistic-aware teacher model, pre-trained on linguistic content of an external dataset, guides the student detector via gradient reversal to minimize the linguistic information. To prevent the inadvertent removal of non-linguistic cues, we incorporate a Variational Information Bottleneck to enable suppression of principal cues. Across nine DF Arena datasets, our method achieves up to a 36.2% relative reduction in the EER compare to the baseline.
Chinese Translation
生成语音技术的快速进步已影响语音生物识别的可靠性。尽管当前的欺骗检测器在领域内条件下表现出色，但在领域外环境中的泛化能力往往较差。我们展示了这可能是由于语言偏见造成的。对训练数据中观察到的语言线索的依赖可能会削弱跨数据的鲁棒性。我们提出了一种利用教师-学生对抗学习的语言不变欺骗检测框架。语言感知的教师模型在外部数据集的语言内容上进行预训练，通过梯度反转指导学生检测器以最小化语言信息。为了防止无意中去除非语言线索，我们结合了变分信息瓶颈，以实现对主要线索的抑制。在九个 DF Arena 数据集上，我们的方法相比基线实现了高达 36.2% 的 EER 相对降低。
cs.CL / 33 / 2606.31432
Clinically Structured Rank-Gated LoRA for Cross-Benchmark Medical Question Answering
用于跨基准医学问答的临床结构化秩门控LoRA
Abstract
Medical multiple-choice question answering requires parameter-efficient adaptation across heterogeneous knowledge domains and reasoning operations. A medication question, a diagnostic decision, a public-health item, and a nursing-action item may require different low-rank updates, while some recall items should preserve the base model's representation with only mild adapter intervention. We propose BiRG-LoRA, a single-adapter rank-gated LoRA method for medical question answering. BiRG-LoRA keeps one LoRA module per target layer but makes its rank dimension input-conditioned: for each question, a biaxial gate combines hidden semantic evidence with specialty/profession priors, clinical-operation priors, and their interaction to select a sparse top-$k$ subset of rank atoms. A scalar injection coefficient further controls the strength of the selected adapter update. Under a matched Qwen3-8B CMB-source protocol, BiRG-LoRA achieves the highest four-benchmark macro-average accuracy among trainable PEFT baselines and matched routing controls: 69.31% averaged over CMB, CMExam, MedQA, and MedMCQA. It improves over MoELoRA by 0.89 percentage points while using 28.1% fewer trainable parameters; a paired, benchmark-stratified bootstrap over final predictions gives a 95% confidence interval of [0.42, 1.37] for this macro-average gain. Basic controls show that BiRG-LoRA also improves over vanilla LoRA r16 and active-rank-matched LoRA r4 by 0.83 macro points, and an evaluation-time weak-axis perturbation check suggests that performance is not brittle to moderate tag noise. The results support a bounded claim: clinically structured rank allocation improves cross-benchmark medical QA under a matched single-seed protocol, while training-seed variance remains future work.
Chinese Translation
医学多项选择题问答需要在异构知识领域和推理操作之间进行参数高效的适应。药物问题、诊断决策、公共卫生项目和护理行动项目可能需要不同的低秩更新，而一些回忆项目则应在仅进行轻微适配器干预的情况下保留基础模型的表示。我们提出了BiRG-LoRA，一种用于医学问答的单适配器秩门控LoRA方法。BiRG-LoRA在每个目标层保持一个LoRA模块，但使其秩维度输入条件化：对于每个问题，一个双轴门结合隐藏语义证据与专业/职业先验、临床操作先验及其交互，以选择稀疏的前$k$个秩原子。一个标量注入系数进一步控制所选适配器更新的强度。在匹配的Qwen3-8B CMB源协议下，BiRG-LoRA在可训练的PEFT基线和匹配路由控制中实现了最高的四个基准宏平均准确率：在CMB、CMExam、MedQA和MedMCQA上平均为69.31%。它比MoELoRA提高了0.89个百分点，同时使用了28.1%的可训练参数更少；对最终预测进行配对的基准分层自助法给出了该宏平均增益的95%置信区间为[0.42, 1.37]。基本控制显示，BiRG-LoRA在原始LoRA r16和活动秩匹配LoRA r4上也提高了0.83宏点，而评估时的弱轴扰动检查表明，性能对适度标签噪声并不脆弱。结果支持一个有限的主张：临床结构化秩分配在匹配的单种子协议下改善了跨基准医学问答，而训练种子方差仍然是未来的研究工作。
cs.CL / 34 / 2606.31446
Revising RVL-CDIP: Quantifying Errors and Test-Train Overlap
修订 RVL-CDIP：量化错误与测试-训练重叠
Abstract
RVL-CDIP is a popular dataset for benchmarking document classifiers. However, the dataset contains ample amounts of label errors as well as non-trivial amounts of test-train overlap, both of which may impact model performance metrics. In this paper, we address these two problems by (1) finding and fixing label errors, and (2) detecting and addressing test-train overlap. We produce several variations of RVL-CDIP with label error and test-train overlap fixes, and benchmark document classification performance on these new RVL-CDIP variations. Our rigorous analysis of RVL-CDIP finds that the corpus contains 12\% label error and approximately 35% test-train duplication. Remediation sees improvements in classification accuracy when errors are removed, but sees decreases in accuracy when duplicates are removed. We additionally evaluate models on RVL-CDIP-N, an out-of-distribution benchmark, finding that training on error-corrected data substantially improves OOD generalization, with supervised models gaining an average of 8.1 percentage points in accuracy and improvements as large as 14 percentage points.
Chinese Translation
RVL-CDIP 是一个用于基准文档分类器的热门数据集。然而，该数据集包含大量标签错误以及非微不足道的测试-训练重叠，这两者都可能影响模型性能指标。本文通过 (1) 发现并修复标签错误，以及 (2) 检测并解决测试-训练重叠，来解决这两个问题。我们生成了多个修复了标签错误和测试-训练重叠的 RVL-CDIP 变体，并在这些新的 RVL-CDIP 变体上基准测试文档分类性能。我们对 RVL-CDIP 的严格分析发现，该语料库包含 12% 的标签错误和大约 35% 的测试-训练重复。在移除错误时，修复措施提高了分类准确率，但在移除重复项时，准确率却有所下降。此外，我们还在 RVL-CDIP-N（一个分布外基准）上评估模型，发现对纠正错误的数据进行训练显著提高了 OOD（分布外）泛化，监督模型的准确率平均提高了 8.1 个百分点，最大提高幅度达到 14 个百分点。
cs.CL / 35 / 2606.31464
Team MKC at CLPsych 2026: Capturing and Characterizing Mental Health Changes through Social Media Timeline Dynamics
MKC团队在CLPsych 2026：通过社交媒体时间线动态捕捉和表征心理健康变化
Abstract
Recent advances in Large Language Models (LLMs) have motivated their adoption across a wide range of domains, including Artificial Intelligence (AI) for mental health. Given the growing prevalence of mental health disorders worldwide and the limited accessibility of professional care, there is an increasing demand for scalable computational approaches that can assist in early detection and continuous monitoring of psychological well-being. In this area, ongoing efforts have focused on curating domain-specific datasets and leveraging them to develop LLMs capable of supporting holistic mental health analysis. In line with this direction, we propose an LLM-based pipeline for comprehensive mental health analysis over sequentially ordered user posts, as part of the CLPsych shared task. Our pipeline offers a unified framework that jointly enables post-level assessment and user-level temporal modeling.
Chinese Translation
近期大型语言模型（LLMs）的进展推动了其在多个领域的应用，包括心理健康的人工智能（AI）。鉴于全球心理健康障碍的日益普遍以及专业护理的有限可及性，对可扩展的计算方法的需求日益增加，这些方法可以帮助早期检测和持续监测心理健康。在这一领域，持续的努力集中在策划特定领域的数据集，并利用这些数据集开发能够支持整体心理健康分析的LLMs。根据这一方向，我们提出了一种基于LLM的管道，用于对顺序排列的用户帖子进行全面的心理健康分析，作为CLPsych共享任务的一部分。我们的管道提供了一个统一的框架，能够同时实现帖子级评估和用户级时间建模。
cs.CL / 36 / 2606.31508
Building an ASR Solution for Training and Assessing Children's Reading
构建儿童阅读训练与评估的自动语音识别解决方案
Abstract
Automatic speech recognition for children's reading remains underdeveloped for most African languages, including Bambara, despite its potential value for reproducible literacy assessment. We present an open-source system for assessing children's reading in Bambara, developed through an end-to-end process linking field data collection, benchmark construction, model adaptation, a reading application, and classroom validation. A mobile collection and assessment app was used to collect 55 hours of raw reading speech from 60 children, from which we construct a public benchmark for Bambara child-reading assessment. Fine-tuning experiments compare Soloni, a Bambara-adapted Fast-Conformer ASR framework with TDT and CTC decoders, with QuartzNet, a compact convolutional ASR architecture. The best Soloni model reduces WER from 0.42 to 0.22 and CER from 0.15 to 0.08, substantially outperforming QuartzNet on the isolated benchmark. The experiments further show that repeated readings of the same texts provide architecture-dependent benefits: they substantially improve QuartzNet but add only marginal gains for Soloni, while SpecAugment regulates training without exceeding the best unaugmented configuration. Disaggregated analysis identifies children under 10 as the main source of residual errors, motivating targeted collection from younger readers. Ten classroom trials supported continued use of the application.
Chinese Translation
尽管自动语音识别在儿童阅读方面具有可重复的识字评估潜力，但对于包括班巴拉语在内的大多数非洲语言仍然发展不足。我们提出了一个用于评估班巴拉语儿童阅读的开源系统，该系统通过一个端到端的过程开发，连接了现场数据收集、基准构建、模型适应、阅读应用和课堂验证。使用移动收集和评估应用程序收集了60名儿童的55小时原始阅读语音，从中构建了班巴拉语儿童阅读评估的公共基准。微调实验比较了Soloni，一个适应班巴拉语的Fast-Conformer ASR框架，配备TDT和CTC解码器，与QuartzNet，一个紧凑的卷积ASR架构。最佳的Soloni模型将字错误率（WER）从0.42降低到0.22，字符错误率（CER）从0.15降低到0.08，在孤立基准上显著优于QuartzNet。实验进一步表明，对同一文本的重复阅读提供了架构依赖的益处：它们显著改善了QuartzNet，但对Soloni的增益仅为边际，而SpecAugment在不超过最佳未增强配置的情况下调节了训练。分解分析确定10岁以下的儿童是残余错误的主要来源，促使我们针对年轻读者进行有针对性的收集。十次课堂试验支持该应用程序的持续使用。
cs.CL / 37 / 2606.31522
FinPersona-Bench: A Benchmark for Longitudinal Psychometric Stability of Autonomous Financial Agents
FinPersona-Bench：自主金融代理的纵向心理测量稳定性基准
Abstract
Large Language Models (LLMs) are increasingly deployed as autonomous financial agents initialized with explicit behavioral mandates such as "preserve capital" or "avoid speculative bets" that are meant to govern every decision throughout deployment. In practice, however, as market context accumulates over long horizons, these mandates gradually lose their behavioral influence, a phenomenon we formalize as Mandate Salience Decay (MSD). To measure MSD objectively, we introduce FinPersona-Bench, a simulation benchmark in which a synthetic market decouples observable price from hidden fundamental value, enabling falsifiable evaluation across three failure modes: trading without signal in calm markets, panic-selling during crashes, and ignoring fundamental value during speculative bubbles. Evaluating 18 leading frontier and open-source LLMs, each assigned one of three behavioral profiles ranging from strict capital preservation to aggressive growth, shows that MSD compounds over time and is model-dependent. In crash scenarios, the behavioral gap between static agents and those receiving periodic mandate re-grounding grows 4.4x from the first to the final quarter of the simulation. The effects of mandate re-grounding are not uniformly positive: it consistently helps conservative agents in low-signal markets but actively worsens behavior for aggressive agents in the same setting. These findings suggest that reliable long-horizon deployment requires selective, mandate-aware re-grounding based on agent profile and market regime.
Chinese Translation
大型语言模型（LLMs）越来越多地被作为自主金融代理部署，这些代理以明确的行为任务初始化，例如“保护资本”或“避免投机性下注”，旨在指导整个部署过程中的每一个决策。然而，在实践中，随着市场背景在较长时间内的积累，这些任务逐渐失去其行为影响力，这一现象我们正式定义为任务显著性衰退（Mandate Salience Decay, MSD）。为了客观地测量MSD，我们引入了FinPersona-Bench，这是一个模拟基准，其中合成市场将可观察价格与隐藏的基本价值解耦，从而能够在三种失败模式下进行可证伪的评估：在平静市场中无信号交易、在崩盘期间恐慌性抛售，以及在投机泡沫期间忽视基本价值。对18个领先的前沿和开源LLMs进行评估，每个模型被分配一种从严格资本保护到激进增长的三种行为特征之一，结果显示MSD随着时间的推移而加剧，并且依赖于模型。在崩盘情境中，静态代理与定期重新调整任务的代理之间的行为差距从模拟的第一季度到最后一个季度增长了4.4倍。任务重新调整的效果并非均为正面：它在低信号市场中始终有助于保守代理，但在相同环境中却会积极恶化激进代理的行为。这些发现表明，可靠的长期部署需要基于代理特征和市场状态的选择性、任务意识的重新调整。
cs.CL / 38 / 2606.31551
AutoTrainess: Teaching Language Models to Improve Language Models Autonomously
AutoTrainess：教会语言模型自主改进语言模型
Abstract
Training language models (LMs) remains a highly human-intensive process, even as frontier language model agents become increasingly capable at software engineering and other long-horizon tasks. A central challenge is that autonomous post-training is not just a coding problem: it requires the agent to repeatedly plan iterations, construct benchmark-aligned data, run stable training jobs, evaluate checkpoints, and preserve experiment state across many hours of interaction. We present AutoTrainess, a LM agent that exposes these operations as a repository of agent-computer interfaces for planning, data preparation, training, evaluation, and logging. Rather than leaving the agent to operate in a raw CLI environment with an underspecified action space, AutoTrainess externalizes prior human experience as explicit workflows, rules, and execution constraints that guide the agent toward effective and reliable training behavior. On PostTrainBench, AutoTrainess consistently outperforms CLI-only baselines, achieving 26.94 average score with GPT-5.4 (Codex) versus 23.21 for CLI-only. It also generalizes across models and harnesses, improving DeepSeek-V4-Flash (OpenCode) from 12.13 to 19.58.
Chinese Translation
训练语言模型（LM）仍然是一个高度依赖人力的过程，即使前沿语言模型代理在软件工程和其他长时间任务中变得越来越强大。一个核心挑战是，自主后训练不仅仅是一个编码问题：它要求代理反复进行计划迭代，构建与基准对齐的数据，运行稳定的训练任务，评估检查点，并在多个小时的交互中保持实验状态。我们提出了AutoTrainess，一个将这些操作暴露为代理-计算机接口库的语言模型代理，用于规划、数据准备、训练、评估和日志记录。与让代理在一个未明确指定动作空间的原始命令行环境中操作不同，AutoTrainess将先前的人类经验外化为明确的工作流程、规则和执行约束，以指导代理朝着有效和可靠的训练行为前进。在PostTrainBench上，AutoTrainess始终优于仅使用命令行的基线，使用GPT-5.4（Codex）时平均得分为26.94，而仅使用命令行的得分为23.21。它还在模型和工具之间具有良好的泛化能力，将DeepSeek-V4-Flash（OpenCode）的得分从12.13提高到19.58。
cs.CL / 39 / 2606.31602
Robust Text Watermarking for Large Language Models via Dual Semantic Embeddings
基于双语义嵌入的大型语言模型鲁棒文本水印
Abstract
This work presents Dual-Embedding Watermarking (DEW), a semantic watermarking scheme for large language models (LLMs) that leverages contextual and token-level embeddings to enhance robustness against paraphrasing and translation. DEW utilizes a signal-processing methodology, applying algebraic vector-space operations to \mbox{token and context embeddings to derive a watermark signal that degrades gracefully under semantic shifts. The method obfuscates the watermark by projecting embedding vectors through pseudo-random matrices seeded with a secret key. Relevant distributions derived from the underlying algebra are evaluated and employed for statistical testing and benchmarking of DEW. Experimental results across multiple LLMs indicate that DEW improves post-paraphrase detection while maintaining competitive text quality, and remains detectable after translation, even when prior semantic watermarks degrade significantly. These findings position DEW as a practical and robust solution for safeguarding LLM-generated text and addressing critical issues in responsible AI deployment.
Chinese Translation
本研究提出了双嵌入水印（Dual-Embedding Watermarking, DEW），这是一种针对大型语言模型（Large Language Models, LLMs）的语义水印方案，利用上下文和词元级嵌入来增强对释义和翻译的鲁棒性。DEW采用信号处理方法，通过代数向量空间运算对词元和上下文嵌入进行处理，以推导出在语义变化下能够优雅退化的水印信号。该方法通过使用带有秘密密钥的伪随机矩阵对嵌入向量进行投影，从而模糊水印。基于底层代数的相关分布被评估并用于DEW的统计测试和基准测试。多项LLM的实验结果表明，DEW在保持竞争性文本质量的同时提高了释义后的检测能力，并且在翻译后仍然可检测，即使先前的语义水印显著退化。这些发现使DEW成为保护LLM生成文本的实用且鲁棒的解决方案，并解决了负责任的人工智能部署中的关键问题。
cs.CL / 40 / 2606.31608
CLExEval: A Human-in-the-Loop Framework for Qualitative Evaluation of LLM Clinical Reasoning
CLExEval：一种人机协作框架用于大语言模型临床推理的定性评估
Abstract
Large Language Models (LLMs) achieve strong results on many medical benchmarks, but their clinical reasoning remains difficult to evaluate reliably. A central risk is an evaluation illusion: fluent and well-structured explanations can appear clinically convincing even when the final diagnosis is incorrect. We introduce CLExEval, a human-in-the-loop framework for evaluating LLM clinical reasoning under progressive information masking. CLExEval combines 5,600 expert-physician annotations with 200 clinical reasoning traces derived from 40 rare diagnostic cases. Our analysis identifies three recurring failure patterns: (i) verbosity bias, where GPT-4o-mini's diagnostic accuracy drops from 95.0% to 32.5% under information scarcity; (ii) a hidden knowledge paradox, where a specialist model reaches 92.5% maximum diagnostic potential but fails to retrieve that knowledge reliably in verbose contexts; and (iii) a 68.6% reasoning-to-output mismatch, where correct diagnoses appear in reasoning traces but are not reflected in final answers. We further evaluate the LLM-as-a-Judge paradigm on a human-verified failure set (n = 142). GPT-4o-mini approved 47.9% of clinically incorrect outputs, while HuatuoGPT-o1 approved all validly scored failures and showed a positive self-preference bias. These results suggest that standalone automated clinical evaluations can substantially overestimate clinical reliability without expert-grounded validation.
Chinese Translation
大型语言模型（LLMs）在许多医学基准测试中取得了良好的结果，但其临床推理的可靠评估仍然困难。一个主要风险是评估错觉：流畅且结构良好的解释即使在最终诊断不正确时也可能显得临床上令人信服。我们提出了CLExEval，这是一种在人机协作框架下评估LLM临床推理的方法，采用逐步信息遮蔽。CLExEval结合了5600条专家医生的注释和来自40个罕见诊断案例的200条临床推理轨迹。我们的分析识别出三种反复出现的失败模式：（i）冗长偏见，GPT-4o-mini的诊断准确率在信息稀缺情况下从95.0%下降至32.5%；（ii）隐性知识悖论，专门模型的最大诊断潜力达到92.5%，但在冗长的上下文中无法可靠地提取该知识；（iii）68.6%的推理与输出不匹配，正确的诊断出现在推理轨迹中，但未反映在最终答案中。我们进一步在一个经过人工验证的失败集合（n = 142）上评估了LLM作为评判者的范式。GPT-4o-mini批准了47.9%的临床不正确输出，而HuatuoGPT-o1则批准了所有有效评分的失败，并表现出积极的自我偏好偏见。这些结果表明，独立的自动化临床评估在没有专家基础验证的情况下可能会大幅高估临床可靠性。
cs.CL / 41 / 2606.31642
Tone-Conditioned Curriculum Learning for Low-Resource Bantu Speech Recognition
基于音调的课程学习在低资源班图语音识别中的应用
Abstract
Southern Bantu languages are spoken by over 80 million people, yet current foundation ASR models still produce zero-shot WER above 100%, which limits practical use in education and public services. We addressed this gap with a tone conditioned curriculum framework for 6 Southern Bantu languages that combined hybrid difficulty scoring, gated adapters driven by tonal statistics and staged curriculum training. We trained on a community corpus and tested transfer to NCHLT to measure robustness beyond matched evaluation. Results revealed clear interactions between architecture and language, with W2V-BERT outperforming Whisper on Nguni languages by 3 to 4 WER points whilst Whisper performed better on Sotho-Tswana languages. W2V-BERT with tone conditioning reached 28.41% average WER across datasets and 23.79% on Xitsonga transfer. No single model suited all 6 languages, so deployment should pair model selection per language with validation across corpora.
Chinese Translation
南部班图语言有超过8000万人使用，但当前的基础自动语音识别（ASR）模型在零样本字错误率（WER）上仍超过100%，这限制了其在教育和公共服务中的实际应用。我们通过一个针对6种南部班图语言的基于音调的课程框架来填补这一空白，该框架结合了混合难度评分、基于音调统计的门控适配器和分阶段课程训练。我们在一个社区语料库上进行了训练，并测试了向NCHLT的迁移，以测量超出匹配评估的鲁棒性。结果显示架构与语言之间存在明显的交互作用，W2V-BERT在Nguni语言上比Whisper表现好3到4个WER点，而Whisper在索托-茨瓦纳语言上表现更佳。带有音调条件的W2V-BERT在各数据集上达到了28.41%的平均WER，在Xitsonga迁移上为23.79%。没有单一模型适用于所有6种语言，因此部署时应根据每种语言选择模型，并在语料库之间进行验证。
cs.CL / 42 / 2606.31644
Moral Safety in LLMs: Exposing Performative Compliance with Puzzled Cues
大型语言模型中的道德安全性：揭示对困惑线索的表面遵从
Abstract
As large language models take on morally consequential roles in healthcare, legal, and hiring contexts, we need to examine whether their ethical behaviors are genuine or superficial. We show that current fairness evaluations substantially overestimate moral safety. Models appear fair when demographic identity is stated as an explicit label, yet become measurably less fair when the same identity must be inferred. We term this failure \emph{performative compliance}, where a model is fair when the presentation resembles a fairness evaluation and less fair as that cue weakens. We introduce a cue-variation methodology that holds the moral dilemma and the demographic identity fixed and varies only how that identity is conveyed. Hiding the explicit label raises harmful decisions by $+4.4$~pp and changes model safety rankings, and the shift persists when models correctly infer the demographic, ruling out attribution error. We propose the \textbf{Cue Visibility Gap}, a model-agnostic robustness metric that can be added to any existing fairness benchmark to separate genuine from performative moral safety. Fairness evaluations that omit cue variation measure surface compliance, not moral robustness, and should not ground deployment decisions in high-stakes settings.
Chinese Translation
随着大型语言模型在医疗、法律和招聘等领域承担道德上重要的角色，我们需要审视它们的伦理行为是真实的还是表面的。我们表明，当前的公平性评估显著高估了道德安全性。当人口身份作为明确标签表述时，模型看起来是公平的；然而，当同一身份必须被推断时，公平性则明显降低。我们将这种失败称为“表面遵从”（performative compliance），即当呈现方式类似于公平性评估时，模型是公平的，而当这种线索减弱时则变得不那么公平。我们引入了一种线索变异方法论，固定道德困境和人口身份，仅改变该身份的传达方式。隐藏明确标签会使有害决策增加4.4个百分点，并改变模型的安全性排名，而当模型正确推断出人口身份时，这种变化仍然存在，排除了归因错误。我们提出了“线索可见性差距”（Cue Visibility Gap），这是一种与模型无关的鲁棒性指标，可以添加到任何现有的公平性基准中，以区分真实的道德安全性与表面道德安全性。省略线索变异的公平性评估只测量表面遵从，而非道德鲁棒性，因此不应作为高风险环境中部署决策的依据。
cs.CL / 43 / 2606.31692
Overview of the TalentCLEF 2026: Skill and Job Title Intelligence for Human Capital Management
TalentCLEF 2026 概述：人力资本管理中的技能与职位名称智能
Abstract
This paper presents an overview of the second edition of the TalentCLEF challenge, organized as a Lab at the Conference and Labs of the Evaluation Forum (CLEF) 2026. TalentCLEF is an initiative aimed at advancing Natural Language Processing research in Human Capital Management. The second edition of the challenge consisted of two tasks: Task A, contextualized job-person matching, focuses on identifying and ranking the most suitable candidates represented by their resumes for a given job vacancy in English and Spanish. Task B, job-skill matching with skill type classification, addresses retrieving the most relevant skills for a given job title in English and distinguishing between core and contextual skills. TalentCLEF attracted 113 registered teams and received more than 400 submissions in the two tasks, reflecting the growing interest of the research community in shared evaluation benchmarks for Human Capital Management. This paper describes the motivation and organization of the challenge, summarizes the datasets and evaluation settings, and reports the main results obtained by the participating teams.
Chinese Translation
本文概述了 TalentCLEF 挑战的第二届活动，该活动作为 2026 年评估论坛会议与实验室（CLEF）的一个实验室组织。TalentCLEF 是一个旨在推动人力资本管理领域自然语言处理研究的倡议。第二届挑战包括两个任务：任务 A，即上下文化的职位-个人匹配，专注于识别和排名最适合特定职位空缺的候选人，这些候选人通过其简历在英语和西班牙语中表示。任务 B，即职位-技能匹配与技能类型分类，旨在检索与特定职位名称最相关的技能，并区分核心技能与上下文技能。TalentCLEF 吸引了 113 支注册团队，并在这两个任务中收到了超过 400 份提交，反映了研究界对人力资本管理共享评估基准日益增长的兴趣。本文描述了挑战的动机和组织，概述了数据集和评估设置，并报告了参与团队获得的主要结果。
cs.CL / 44 / 2606.31718
Cross-lingual Relation Extraction with Large Language Models: Zero-Shot, Few-Shot, and Fine-Tuned Evaluation on Romanian
使用大型语言模型进行跨语言关系抽取：在罗马尼亚语上的零样本、少样本和微调评估
Abstract
Relation extraction (RE) for low-resource languages is typically constrained by the lack of annotated corpora. We investigate the feasibility of cross-lingual RE for Romanian by combining automatic dataset translation with large language model (LLM) inference. We translate the SemEval-2010 Task 8 benchmark from English to Romanian using an LLM-based translation pipeline and evaluate Gemma 4 31B under zero-shot, few-shot, and QLoRA fine-tuned configurations, against four encoder baselines spanning 125M to 560M parameters: XLM- RoBERTa (base and large), Romanian BERT, and RoBERT- large. We assess two task formulations: relation classification with marked entities and end-to-end extraction. Our results show that Romanian incurs a 3 to 5 percentage point (pp) drop relative to English in prompt-only settings, that few-shot prompting provides marginal gains over zero-shot, and that QLoRA fine-tuning improves macro F1-Score by more than 22 percentage points in both languages while reducing the cross-lingual gap from 3.3 to 1.4pp. The encoder baselines come within 1-4pp of QLoRA Gemma on Romanian despite being 50-250 times smaller, with monolingual Romanian BERT at 125M parameters matching multilingual XLM-R at 278M. The case for using a 31B model for single-task RE on Romanian is therefore weak in deployment scenarios where compute matters. We release the translated dataset, evaluation code, and trained models.
Chinese Translation
低资源语言的关系抽取（RE）通常受到缺乏标注语料库的限制。我们通过将自动数据集翻译与大型语言模型（LLM）推理相结合，研究了罗马尼亚语跨语言RE的可行性。我们使用基于LLM的翻译管道将SemEval-2010任务8基准从英语翻译为罗马尼亚语，并在零样本、少样本和QLoRA微调配置下评估Gemma 4 31B，比较四个编码器基线，参数范围从1.25亿到5.6亿：XLM-RoBERTa（基础版和大型版）、罗马尼亚BERT和RoBERTa-large。我们评估了两种任务表述：带标记实体的关系分类和端到端抽取。我们的结果表明，在仅使用提示的设置中，罗马尼亚语相对于英语的表现下降了3到5个百分点（pp），少样本提示相较于零样本提供了边际收益，而QLoRA微调在两种语言中将宏观F1得分提高了超过22个百分点，同时将跨语言差距从3.3降低到1.4个百分点。尽管编码器基线的参数量比QLoRA Gemma小50到250倍，但在罗马尼亚语上仍然接近1-4个百分点，其中单语罗马尼亚BERT（1.25亿参数）与多语言XLM-R（2.78亿参数）表现相当。因此，在计算资源重要的部署场景中，使用31B模型进行罗马尼亚语单任务RE的理由较弱。我们发布了翻译后的数据集、评估代码和训练模型。
cs.CL / 45 / 2606.31719
Seeing Is Not Sharing: Some Vision-Language Models Overestimate Common Ground in Asymmetric Dialogue
看见并不等于分享：某些视觉-语言模型在不对称对话中高估了共同基础
Abstract
In collaborative dialogue, shared perception does not guarantee shared interpretation. Mutual understanding must be established through interaction. We investigate whether vision-language models (VLMs) can distinguish what could be shared from what has been shared between dialogue participants through grounding. We formulate this as an interpretation-matching task on 13,077 annotated reference expressions from HCRC MapTask dialogues, and evaluate VLMs under systematically controlled manipulations of dialogue context and map-information access. Our results show that providing authentic map images improves overall performance but shifts models toward over-predicting alignment. Textual descriptions of the same map content reproduce this bias, while non-informative images suppress alignment predictions entirely, indicating that the bias is driven by task-relevant map content, not the visual channel. This improvement comes at the cost of degraded accuracy on non-aligned cases. Calibration analysis and reference-chain tracking further suggest that models rely on static referential cues on the maps rather than tracking how grounding unfolds through dialogue history. We observe these patterns most clearly in Qwen3-VL-8B-Instruct and, to varying degrees, in four additional models from two architecture families. In models that exhibit the bias, map content, whether presented visually or textually, is treated as evidence of mutual understanding, conflating potential with established common ground.
Chinese Translation
在协作对话中，共享的感知并不保证共享的解释。相互理解必须通过互动建立。我们研究视觉-语言模型（VLMs）是否能够通过基础知识区分对话参与者之间可能共享的内容与已共享的内容。我们将此任务表述为对13,077个来自HCRC MapTask对话的注释参考表达的解释匹配任务，并在系统控制的对话上下文和地图信息访问的操控下评估VLMs。我们的结果表明，提供真实的地图图像可以提高整体性能，但使模型倾向于过度预测对齐。相同地图内容的文本描述重现了这种偏差，而非信息性图像则完全抑制了对齐预测，表明这种偏差是由任务相关的地图内容驱动，而非视觉通道。这种改善以非对齐案例的准确性下降为代价。校准分析和参考链追踪进一步表明，模型依赖于地图上的静态指称线索，而不是追踪基础知识如何通过对话历史展开。我们在Qwen3-VL-8B-Instruct模型中最清楚地观察到这些模式，并在来自两种架构系列的另外四个模型中以不同程度表现出来。在表现出这种偏差的模型中，无论是以视觉还是文本形式呈现的地图内容都被视为相互理解的证据，将潜在性与已建立的共同基础混淆。
cs.CL / 46 / 2606.31722
Adapting Foundation ASR Models to Dysarthric Speech: A Case Study
将基础自动语音识别模型适应于构音障碍言语：案例研究
Abstract
Automatic speech recognition (ASR) systems often perform poorly in dysarthric speech, limiting their usefulness to affected speakers in everyday communication. This paper presents a personalized ASR system for a dysarthric speaker, built by adapting a foundation ASR model to speaker-specific data. Using the TEQST tool, we collected 92 hours of read speech and later added 8.8 hours of user corrections gathered through a deployed mobile application. Starting from Whisper, fine-tuning reduced word error rate to 15.8% with only 1.4 hours of adaptation data, reached 10.7% with 22.5 hours, and achieved the best result of 9.7% when using all available data including the corrections. Using LoRA adaptation and/or Qwen3-ASR as foundation model performed worse in this setting. The results show that personalized fine-tuning can make foundation ASR models substantially more effective for dysarthric speech and suitable for practical deployment.
Chinese Translation
自动语音识别（ASR）系统在构音障碍言语中的表现通常较差，这限制了其在受影响说话者日常交流中的实用性。本文提出了一种为构音障碍说话者量身定制的ASR系统，该系统通过将基础ASR模型适应于特定说话者的数据而构建。我们使用TEQST工具收集了92小时的朗读语音，并通过已部署的移动应用程序后续添加了8.8小时的用户修正数据。从Whisper模型开始，经过微调后，词错误率降低至15.8%，仅使用1.4小时的适应数据，使用22.5小时的数据时降至10.7%，而使用包括修正数据在内的所有可用数据时达到了最佳结果9.7%。在这种情况下，使用LoRA适应和/或Qwen3-ASR作为基础模型的表现较差。结果表明，个性化微调可以显著提高基础ASR模型在构音障碍言语中的有效性，并适合实际部署。
cs.CL / 47 / 2606.31741
STEB: Style Text Embedding Benchmark
STEB：风格文本嵌入基准
Abstract
While semantic embeddings are rigorously evaluated on the Massive Text Embedding Benchmark, the evaluation of style embeddings remains fragmented, with each work relying on their own set of tasks and datasets. To bridge this gap, we introduce the Style Text Embedding Benchmark, a comprehensive open-source benchmark intended to standardize the evaluation of style embeddings. STEB encompasses 96 datasets across 7 languages, spanning applications such as authorship verification, authorship retrieval, AI-text detection, probing of linguistic features, and others. We find that semantic embeddings consistently fail in stylistic tasks, and that there is no style embedding that is universally superior across all tasks evaluated. We open-source the STEB code base at: https://github.com/rrivera1849/STEB.
Chinese Translation
尽管语义嵌入在大规模文本嵌入基准（Massive Text Embedding Benchmark）上得到了严格评估，但风格嵌入的评估仍然较为零散，各项研究依赖于各自的任务和数据集。为了解决这一问题，我们提出了风格文本嵌入基准（Style Text Embedding Benchmark，STEB），这是一个全面的开源基准，旨在标准化风格嵌入的评估。STEB涵盖了7种语言的96个数据集，涉及作者身份验证、作者检索、AI文本检测、语言特征探测等应用。我们发现，语义嵌入在风格任务中始终表现不佳，并且没有一种风格嵌入在所有评估任务中表现出普遍的优越性。我们将STEB代码库开源，地址为：https://github.com/rrivera1849/STEB。
cs.CL / 48 / 2606.31796
CHERRY: Compressed Hierarchical Experts with Recurrent Representational Yield
CHERRY：具有递归表示产出的压缩层次专家
Abstract
We study three complementary techniques for training compute-efficient language models. (1) Selective supervision and per-token efficiency. Selective Ground Truth Token Training (SGT) concentrates supervision on the ~15% of output tokens that carry semantic payload. Through positive gradient coupling in position-shared transformer weights -- a token-level instance of auxiliary-task transfer -- the remaining 85% of unsupervised tokens still improve substantially, giving a 4.5x per-supervised-token efficiency (at the step-100 eval optimum, ~67% of the full-sequence loss reduction is recovered from 15% of the supervision). We prove that this improvement on unsupervised tokens is guaranteed whenever the gradient coupling coefficient gamma-bar = 0.72 is positive (Theorem 1), and show the effect is a property of natural-language structure: it collapses on shuffled text. (2) Depth compression with recurrent recovery. A 48-layer, 1B-parameter transformer is compressed to 6 layers (227M) by averaging adjacent layers and restored through learned recurrent unrolling. With 34 effective recurrent layers it reaches a held-out loss of 2.934, within measurement noise of a 566M dense model at 2.926 -- a 2.5x reduction in parameters. (3) Fusion of compressed experts. Assembling several compressed models as a Mixture of Efficient Experts (MoEE) with multi-token prediction improves over each single expert at comparable active parameters: a 2-expert MoEE reaches loss 2.789 versus 2.926 for the best single compressed model. We validate these techniques on CHERRY-1.8B, a Korean foundation model whose every trainable parameter derives from our own training runs. We are explicit throughout about the scope of the evidence (one model family, Korean data, loss-based metrics) and about which claims are established versus prospective.
Chinese Translation
我们研究了三种互补的技术，以训练计算高效的语言模型。(1) 选择性监督和每个标记的效率。选择性真实标记训练（SGT）将监督集中在约15%的输出标记上，这些标记承载语义信息。通过在位置共享的变换器权重中进行正向梯度耦合——这是辅助任务转移的一个标记级实例——其余85%的无监督标记仍然显著改善，提供了每个监督标记4.5倍的效率（在第100步评估的最佳状态下，约67%的全序列损失减少来自15%的监督）。我们证明了当梯度耦合系数gamma-bar = 0.72为正时，这种对无监督标记的改善是有保证的（定理1），并展示了这一效应是自然语言结构的特性：在打乱的文本上会崩溃。(2) 通过递归恢复进行深度压缩。一个48层、10亿参数的变换器被压缩到6层（2.27亿），通过平均相邻层并通过学习的递归展开进行恢复。通过34个有效的递归层，它达到了2.934的保留损失，接近566M稠密模型的2.926的测量噪声——参数减少2.5倍。(3) 压缩专家的融合。将几个压缩模型组装成高效专家混合体（MoEE），并进行多标记预测，在可比的活跃参数下优于每个单一专家：一个2专家的MoEE达到了2.789的损失，而最佳单一压缩模型为2.926。我们在CHERRY-1.8B上验证了这些技术，这是一个韩语基础模型，其每个可训练参数均源自我们自己的训练运行。我们在整个过程中明确了证据的范围（一个模型家族、韩语数据、基于损失的指标）以及哪些主张是已建立的，哪些是前瞻性的。
cs.CL / 49 / 2606.31845
Explicit Fuzzy Logic in the Feed-Forward Layer: Self-Forgetting Quantifiers Discover Legible Grammatical-Licensing Detectors
前馈层中的显式模糊逻辑：自遗忘量词发现可读的语法许可检测器
Abstract
A transformer's feed-forward (FFN) sublayer materializes the distinctions attention gathers, yet gives no account of what it computes. In a parameter-neutral replacement, each hidden unit is an explicit fuzzy set operation on sigmoid-bounded [0,1] memberships: intersection A*B and set-difference A*(1-B), the latter a bounded positive negation ("A but not B") that gated/bilinear units lack -- a negation-capable FFN (NC-FFN). On N-bit parity they are the most parameter-efficient reasoning basis at shallow depth; at scale (125M, OpenWebText) NC-FFN ties the GELU baseline's perplexity, every unit carrying explicit logical form. Two limits share one cause: two-operand logic localizes to layer 0 and erodes under training, and the one robust grammatical deficit concentrates in licensing and quantifiers, beyond within-token operators. We resolve both with a small block of sequence quantifiers: a soft existential and a soft proportion, each with a per-unit learned forgetting rate from a sticky init. This recovers the deficit at epoch one (halving the wider epoch-two gap), modestly leads on LAMBADA, and makes the FFN legible: the structure now holds and migrates into depth; the decay un-learns its stickiness (median half-life ~1.5 tokens; zero latch units); and at the semantic layers the units read, without dictionary learning, as grammatical licensing detectors: each fires on a licensor (a comparative, a passive participle, a negative-polarity item) and carries its memory forward to predict the licensed word (than, by, nor). This legibility is localized and free only up to a partition (a fully Boolean FFN diverges in training), but the result is a parameter-neutral, language-model-quality transformer with a readable, interpretable-by-construction grammatical mechanism -- an account not just of what a feed-forward layer represents but how it licenses.
Chinese Translation
变换器的前馈（FFN）子层体现了注意力所收集的区分，但并未说明其计算内容。在一个参数中立的替代方案中，每个隐藏单元是对 sigmoid 限制的 [0,1] 归属的显式模糊集操作：交集 A*B 和集合差 A*(1-B)，后者是一个有界的正向否定（“A 但不是 B”），而门控/双线性单元则缺乏这一点——一个具有否定能力的 FFN（NC-FFN）。在 N 位奇偶校验中，它们是浅层深度下最具参数效率的推理基础；在规模（125M，OpenWebText）上，NC-FFN 的困惑度与 GELU 基线相当，每个单元都携带显式的逻辑形式。两个限制共享一个原因：双操作数逻辑局限于层 0，并在训练中逐渐减弱，而唯一稳健的语法缺陷集中在许可和量词上，超出了词内操作符。我们通过一小块序列量词解决了这两个问题：一个软存在量词和一个软比例量词，每个量词都有一个从粘性初始化中学习的每单元遗忘率。这在第一轮训练中恢复了缺陷（将更广泛的第二轮差距减半），在 LAMBADA 上适度领先，并使 FFN 变得可读：该结构现在保持并迁移到深度；衰减不再学习其粘性（中位半衰期约为 1.5 个标记；零锁存单元）；在语义层中，单元读取时，无需字典学习，作为语法许可检测器：每个单元在许可者（比较级、被动分词、负极性项）上触发，并将其记忆向前传递以预测被许可的单词（than、by、nor）。这种可读性是局部的，并且仅在一个分区内是免费的（完全布尔的 FFN 在训练中会发散），但结果是一个参数中立、语言模型质量的变换器，具有可读的、可构建解释的语法机制——不仅说明了前馈层所表示的内容，还说明了它如何进行许可。
cs.CL / 50 / 2606.31916
Theory of Mind and Persuasion Beyond Conversation: Assessing the Capacity of LLMs to Induce Belief States via Planning and Action
超越对话的心智理论与说服：评估大型语言模型通过规划和行动引发信念状态的能力
Abstract
Theory of Mind (ToM) benchmarks for Large Language Models (LLMs) typically rely on passive question-answering formats, but the deployment of LLMs in increasingly agentic and autonomous forms demands new evaluations. In this paper we evaluate an agent's ability to induce specific belief states in other agents by taking actions rather than using conversational persuasion, a capability we call Non-Conversational Planning ToM (NCP-ToM). NCP-ToM is likely to be essential for many agent use-cases, including within user-assistant interactions and pedagogical contexts, but may also present manipulation or misinformation risks. Using a novel framework, NCP-ExploreToM, we subvert the conventional task structure by providing models with a set of belief state goals and requiring them to move objects or direct characters into rooms to achieve their goals. We evaluated six frontier models, including GPT-5, Gemini 2.5 Pro and the Claude 4 series, and a cohort of human participants, across 600 task instances. GPT-5 was successful on approximately 80% of tasks in the agentic setting, and was the only model to outperform human participants on our task, but was still less robust than humans across contexts. We additionally found that all models, like humans, performed better on tasks inducing true belief states than false belief states, which is a positive signal for alignment efforts. These findings highlight emerging social-reasoning capabilities in LLMs for non-conversational task completion and underscore the necessity of agentic evaluations for understanding the safety and alignment of autonomous social agents.
Chinese Translation
大型语言模型（LLMs）的心智理论（ToM）基准通常依赖于被动问答格式，但LLMs在日益具备代理性和自主形式的应用要求新的评估。在本文中，我们评估了一个代理通过采取行动而非使用对话说服来引发其他代理特定信念状态的能力，这一能力我们称之为非对话规划心智理论（NCP-ToM）。NCP-ToM可能对许多代理使用场景至关重要，包括用户助手交互和教育环境，但也可能带来操控或错误信息的风险。我们使用一种新颖的框架NCP-ExploreToM，通过为模型提供一组信念状态目标，并要求它们移动物体或引导角色进入房间以实现这些目标，从而颠覆了传统的任务结构。我们在600个任务实例中评估了六个前沿模型，包括GPT-5、Gemini 2.5 Pro和Claude 4系列，以及一组人类参与者。GPT-5在代理性环境中约80%的任务中取得了成功，并且是唯一在我们的任务中超越人类参与者的模型，但在不同上下文中仍不如人类稳健。我们还发现，所有模型与人类一样，在引发真实信念状态的任务中表现优于引发虚假信念状态的任务，这对对齐努力是一个积极信号。这些发现突显了LLMs在非对话任务完成中的新兴社会推理能力，并强调了理解自主社会代理的安全性和对齐性所需的代理性评估。
cs.CL / 51 / 2606.31947
LuxEmo: Expressive Text-to-Speech Corpus for Luxembourgish
LuxEmo：卢森堡语的表现性文本到语音语料库
Abstract
State-of-the-art speech datasets predominantly focus on widely spoken languages, often overlooking low-resource languages such as Luxembourgish, which remain underrepresented in speech technology research. In this work, we introduce LuxEmo, a 21-hour conversational expressive speech corpus for Luxembourgish with 4 emotion categories. LuxEmo is derived from Radio T\'el\'evision Luxembourg (RTL) youth broadcasts, using automated detection followed by human validation. We propose a semi-automatic curation workflow combining voice activity detection, denoising, language identification, LuxASR-based segmentation, automatic emotion prediction, lexical cues, and targeted human review. Additionally, we benchmark five expressive TTS systems covering German-based cross-lingual transfer, multilingual Luxembourgish support, Luxembourgish adaptation, and non-parametric prosody transfer. Performance is evaluated using both objective metrics and human evaluation.
Chinese Translation
当前最先进的语音数据集主要集中于广泛使用的语言，往往忽视了如卢森堡语等低资源语言，这些语言在语音技术研究中仍然代表性不足。在本研究中，我们介绍了LuxEmo，这是一个包含21小时对话式表现性语音的卢森堡语语料库，涵盖4种情感类别。LuxEmo源自卢森堡广播电视台（RTL）的青年节目，通过自动检测和人工验证相结合的方式构建。我们提出了一种半自动化的策划工作流程，结合了语音活动检测、去噪、语言识别、基于LuxASR的分段、自动情感预测、词汇线索和针对性的人工审核。此外，我们对五种表现性文本到语音（TTS）系统进行了基准测试，这些系统涵盖了基于德语的跨语言迁移、多语言卢森堡语支持、卢森堡语适配和非参数韵律迁移。性能评估采用了客观指标和人工评估相结合的方法。
cs.CL / 52 / 2606.31980
DigitalCoach: Communication and Grounding Gaps in Human and Agentic Computer Use Coaching
数字教练：人类与智能体计算机使用辅导中的沟通与基础差距
Abstract
Agents are increasingly capable of automating software tasks, but can they teach humans how to use software themselves? We introduce DigitalCoach, a multimodal dataset of 72 human expert-novice computer use coaching sessions consisting of 22,752 dialogue turns grounded in 28.1 hours of screen and input event recordings across five software applications. We use DigitalCoach to evaluate whether state-of-the-art models can teach humans how to use computers. Automated evaluation shows that models differ from humans in how they coach: models provide more direct instructions, but fewer explanations, error diagnoses, and knowledge-check questions. When we fix the coaching method, models produce utterances similar to human references yet poorly grounded in visual context. Interactive evaluation confirms that model coaches cause learners to passively follow instructions without deeper engagement and fall short in visual grounding. DigitalCoach lays a foundation for collaborative and proactive computer use coaching agents.
Chinese Translation
智能体在自动化软件任务方面的能力日益增强，但它们能否教会人类如何使用软件？我们介绍了数字教练（DigitalCoach），这是一个包含72个专家与新手计算机使用辅导会话的多模态数据集，包含22,752个对话轮次，基于五个软件应用程序的28.1小时屏幕和输入事件录制。我们利用数字教练评估最先进的模型是否能够教会人类如何使用计算机。自动化评估显示，模型在辅导方式上与人类存在差异：模型提供更直接的指令，但解释、错误诊断和知识检查问题较少。当我们固定辅导方法时，模型产生的发言与人类参考相似，但在视觉上下文中基础较差。互动评估确认，模型教练使学习者被动地遵循指令，而未能深入参与，并且在视觉基础方面表现不佳。数字教练为协作和主动的计算机使用辅导智能体奠定了基础。
cs.CL / 53 / 2606.32014
Scalable Behaviour Cloning on Browser Using via Skill Distillation
通过技能蒸馏在浏览器上进行可扩展的行为克隆
Abstract
Internet users collectively perform an enormous range of skilled work through web browsers, from software development and document editing to search, forms, and enterprise workflows, making human browsing a highly scalable but under-exploited source of reusable browser skills. We argue that the bottleneck for browser agents is decision-making under incomplete information rather than low-level operation, and that the priors agents lack are already implicit in human interaction traces. We therefore study scalable behavior cloning for browser agents via skill distillation, converting user interaction trajectories into compact natural-language skills that agents can read, retrieve, reuse, and compose directly. We further organize the distilled skills into a skill graph so that growth proceeds through consolidation rather than unbounded accumulation. This suggests that the scalability of browser agents may come less from manually designed tasks and more from the collective skills already expressed by internet users. Our project is available at: https://lab.einsia.ai/browserbc/.
Chinese Translation
互联网用户通过网页浏览器共同执行大量的熟练工作，从软件开发和文档编辑到搜索、表单和企业工作流程，使得人类浏览成为一个高度可扩展但尚未充分利用的可重用浏览器技能来源。我们认为，浏览器代理的瓶颈在于在不完整信息下的决策，而非低级操作，并且代理所缺乏的先验知识已经隐含在人类交互轨迹中。因此，我们研究了通过技能蒸馏为浏览器代理进行可扩展的行为克隆，将用户交互轨迹转换为代理可以直接读取、检索、重用和组合的紧凑自然语言技能。我们进一步将蒸馏出的技能组织成技能图，以便通过整合而非无界积累来促进增长。这表明，浏览器代理的可扩展性可能更多地来自于互联网用户已经表达的集体技能，而非手动设计的任务。我们的项目可在以下链接获取： https://lab.einsia.ai/browserbc/。
cs.CL / 54 / 2606.32025
Generative Skill Composition for LLM Agents
大语言模型代理的生成技能组合
Abstract
Recent LLM agents benefit from skills for solving complex tasks. Skills encapsulate modular packages of procedural knowledge and instructions for performing specialized tasks, such as setting up a sandboxed environment, running a test suite, or refactoring a function across multiple files. As skill libraries grow and become reusable across tasks and domains, selecting an appropriate skill composition has emerged as a central bottleneck. Existing approaches fall into two categories. One exposes the agent's reasoning to the entire skill collection; the other performs skill retrieval via embeddings or LLM-based rerankers. Both provide useful insights; however, they miss the structural nature of skill composition, which is a joint decision over which skills, how many, and in what order -- three dimensions that cannot be decoupled. We formalize this as structured skill composition: given a task and a skill library, predict an executable skill plan that jointly specifies the activated subset, count, and execution order. We propose SkillComposer, which instantiates structured skill composition as task-conditioned skill sequence prediction. SkillComposer uses a constrained autoregressive decoder over skill identifiers, so subset, count, and order emerge jointly from a single decoding pass, and dependencies between successive skills are captured naturally. We build a training set of task-composition pairs from a real, human-curated skill library. We then evaluate SkillComposer along two axes: composition quality on a held-out test set, and downstream task success on SkillsBench across two production-grade coding agents. On GPT-5.2-Codex, Gemini-3-Pro-Preview, SkillComposer raises the pass rate by +23.1, +18.2pp over the no-skill baseline, surpassing top-3 retrieval and matching the gold-skill retrieval upper bound at lower prompt-token cost.
Chinese Translation
最近的大语言模型（LLM）代理通过技能来解决复杂任务。技能封装了用于执行专业任务的程序性知识和指令的模块化包，例如设置沙盒环境、运行测试套件或在多个文件中重构函数。随着技能库的增长并在任务和领域之间变得可重用，选择合适的技能组合已成为一个核心瓶颈。现有的方法分为两类。一类将代理的推理暴露给整个技能集合；另一类通过嵌入或基于LLM的重新排序器进行技能检索。这两种方法都提供了有用的见解；然而，它们忽视了技能组合的结构特性，即在选择哪些技能、数量和顺序这三个维度上进行联合决策，这三者是不可分割的。我们将其形式化为结构化技能组合：给定一个任务和一个技能库，预测一个可执行的技能计划，该计划共同指定激活的子集、数量和执行顺序。我们提出了SkillComposer，它将结构化技能组合实例化为基于任务的技能序列预测。SkillComposer在技能标识符上使用约束自回归解码器，因此子集、数量和顺序在单次解码过程中共同出现，并且后续技能之间的依赖关系自然地被捕捉。我们从一个真实的人类策划的技能库中构建了一个任务组合对的训练集。然后，我们在两个维度上评估SkillComposer：在保留的测试集上的组合质量，以及在SkillsBench上对两个生产级编码代理的下游任务成功率。在GPT-5.2-Codex和Gemini-3-Pro-Preview上，SkillComposer的通过率比无技能基线提高了+23.1和+18.2个百分点，超越了前3名检索，并在较低的提示令牌成本下匹配了黄金技能检索的上限。
cs.CL / 55 / 2606.32029
When LLMs Read Tables Carelessly: Measuring and Reducing Data Referencing Errors
当大型语言模型粗心阅读表格时：测量和减少数据引用错误
Abstract
While large language models (LLMs) perform well on table tasks, they still make data referencing errors (DREs), i.e., incorrectly citing or omitting table values, despite understanding the table structure. Beyond final-answer accuracy, DREs directly compromise the correctness and reliability of intermediate reasoning steps. Yet prior studies have only offered limited, small-scale analyses. In this work, we present the first systematic evaluation of tabular data referencing errors across different models and tasks. Our results show that DREs occur across all tested models (1.7B to 20B parameters). Furthermore, we demonstrate that incorporating data referencing as a critic significantly improves answer accuracy up to 12.0%, through critic-based filtering and rejection sampling. Finally, we trained a lightweight 4B-parameter critic model that achieves an average F1 score of 78.2% in detecting both in-distribution and out-of-distribution DREs, and effectively assists inference for larger models.
Chinese Translation
尽管大型语言模型（LLMs）在表格任务上表现良好，但它们仍然会出现数据引用错误（DREs），即在理解表格结构的情况下，错误引用或遗漏表格值。除了最终答案的准确性外，DREs 直接影响中间推理步骤的正确性和可靠性。然而，之前的研究仅提供了有限的小规模分析。在本研究中，我们首次对不同模型和任务中的表格数据引用错误进行了系统评估。我们的结果表明，所有测试模型（参数从 1.7B 到 20B）均出现了 DREs。此外，我们证明将数据引用作为批评者纳入显著提高了答案的准确性，提升幅度可达 12.0%，通过基于批评者的过滤和拒绝采样。最后，我们训练了一个轻量级的 4B 参数批评者模型，该模型在检测分布内和分布外的 DREs 时平均 F1 分数达到 78.2%，并有效地辅助了更大模型的推理。
cs.CL / 56 / 2606.32032
Reinforcement Learning with Metacognitive Feedback Elicits Faithful Uncertainty Expression in LLMs
带有元认知反馈的强化学习促使大语言模型表现出真实的不确定性表达
Abstract
Metacognition is a critical component of intelligence that describes the ability to monitor and regulate one's own cognitive processes. Yet LLMs exhibit systemic deficiencies in key metacognitive faculties: they hallucinate with high confidence, fail to recognize knowledge boundaries, and misrepresent their internal uncertainty--undermining trustworthiness and reliability. Since monitoring task performance and adapting behavior accordingly are central to metacognition, we posit that models capable of accurately judging their own performance are better positioned to improve it. We operationalize this idea via two novel mechanisms: reinforcement learning with metacognitive feedback (RLMF), a paradigm to refine completion rankings during preference optimization based on the quality of a model's self-judgments of performance, and metacognitive data selection, which uses similar self-judgments to identify high-value training examples, outperforming naive active learning. We apply these innovations to the problem of faithful calibration (FC), a task that is itself fundamentally metacognitive: the goal is to align expressed with intrinsic uncertainty, difficult even for frontier LLMs. We adopt a two-stage, decoupled approach, first using these methods to calibrate the faithfulness of models' self-reported confidence scores, then mapping to natural, context-adaptable linguistic uncertainty via targeted output editing. Extensive experiments show RLMF achieves generalizable, state-of-the-art FC on diverse tasks while preserving accuracy. Further, RLMF surpasses standard RL by up to 63% while enhancing models' ability to assess and express their own capability limits. This positions RLMF as a promising paradigm to enhance LLM metacognition toward improved abilities and alignment, and suggests metacognitive performance as an effective RL signal to overcome limits of prior intrinsic feedback methods.
Chinese Translation
元认知是智能的一个关键组成部分，描述了监控和调节自身认知过程的能力。然而，大语言模型（LLMs）在关键的元认知能力上表现出系统性缺陷：它们以高信心产生幻觉，未能识别知识边界，并错误地表述其内部不确定性，从而削弱了可信度和可靠性。由于监控任务表现并相应调整行为是元认知的核心，我们假设能够准确判断自身表现的模型更有可能改善其表现。我们通过两种新机制来实现这一理念：带有元认知反馈的强化学习（RLMF），一种在偏好优化过程中基于模型自我判断表现质量来细化完成排名的范式，以及元认知数据选择，利用类似的自我判断来识别高价值的训练示例，超越了简单的主动学习。我们将这些创新应用于真实校准（FC）问题，这一任务本质上也是元认知的：目标是将表达的不确定性与内在不确定性对齐，这对前沿的LLMs来说也很困难。我们采用了两阶段的解耦方法，首先使用这些方法来校准模型自我报告的置信度分数的真实性，然后通过有针对性的输出编辑将其映射到自然的、适应上下文的语言不确定性。大量实验表明，RLMF在多样化任务上实现了可推广的、先进的FC，同时保持了准确性。此外，RLMF在提升模型评估和表达自身能力限制的能力方面，比标准强化学习提高了多达63%。这使得RLMF成为增强LLM元认知以改善能力和对齐的有前景的范式，并表明元认知表现作为有效的强化学习信号，可以克服以往内在反馈方法的局限。
cs.CL / 57 / 2606.32038
Introspective Coupling: Self-Explanation Training Tracks Behavioral Change Despite Fixed Supervision
内省耦合：自我解释训练在固定监督下追踪行为变化
Abstract
When does training language models (LMs) to generate explanations of their predictions yield faithful introspection, rather than superficial imitation? We study LMs trained to explain which features of their inputs influenced their behavior, using models' counterfactual behavior on modified inputs as supervision. Surprisingly, we find that LMs trained on fixed counterfactual explanations derived from earlier checkpoints of themselves, or even from behaviorally similar models in different families, frequently produce explanations more faithful to their own current behaviors than to those of their training targets. This "introspective" coupling between LM explanations and behaviors occurs when training explanations remain sufficiently correlated with current behaviors over the course of training, even as behaviors themselves shift. We also show that introspective coupling tracks behavior shifts: when explanation training is provided concurrently with other post-training objectives, explanations track those shifts without requiring updated supervision. This phenomenon appears in multiple tasks, including sycophancy and refusal, and is robust to label noise. Overall, our results show that even fixed datasets of counterfactual explanations can provide scalable and generalizable post-training signal for introspection.
Chinese Translation
何时训练语言模型（LMs）生成其预测解释能够产生真实的内省，而非表面的模仿？我们研究了训练语言模型解释其输入特征如何影响其行为的过程，使用模型在修改输入上的反事实行为作为监督。令人惊讶的是，我们发现，基于自身早期检查点生成的固定反事实解释，或甚至来自不同家族中行为相似模型的反事实解释训练的语言模型，常常生成更忠实于其当前行为而非训练目标的解释。这种语言模型解释与行为之间的“内省”耦合发生在训练解释与当前行为在训练过程中保持足够相关时，即使行为本身发生变化。我们还表明，内省耦合能够追踪行为变化：当解释训练与其他后训练目标同时提供时，解释能够追踪这些变化，而无需更新的监督。这一现象出现在多个任务中，包括谄媚和拒绝，并且对标签噪声具有鲁棒性。总体而言，我们的结果表明，即使是固定的反事实解释数据集也能够为内省提供可扩展和可推广的后训练信号。