cs.RO / 1 / 2607.00020
EmbodimentSemantic: A Spatial Scene-Graph Dataset and Benchmark for Vision-Language Models on Embodied Manipulation Trajectories
EmbodimentSemantic：一个用于评估具身操作轨迹的视觉-语言模型的空间场景图数据集和基准
Abstract
Spatial grounding remains a key limitation of vision-language-action (VLA) systems for robotic manipulation. While current models can recognize objects and follow language instructions, they often lack an explicit representation of how objects are arranged in space, including support, containment, ordering, occlusion, and depth-sensitive relations. We introduce EmbodimentSemantic, a spatial scene-graph dataset and benchmark for evaluating relational grounding in embodied manipulation. EmbodimentSemantic represents scenes as directed object-relation-object triplets, where each triplet specifies a spatial relation between an ordered pair of objects using a fixed set of relations. This representation enables direct evaluation of object binding, relation prediction, and spatial consistency. The dataset includes real-world manipulation observations collected with the low-cost SO101 robot arm, together with generated scene graphs for studying spatial grounding in practical robotic settings. To provide controlled validation, we also introduce a simulator-grounded LIBERO benchmark with over 60K manipulation frames and more than 120K camera-specific scene graphs across paired third-person and wrist views, where ground-truth relations are derived automatically from MuJoCo geometry, world coordinates, camera projections, and visibility constraints. We further test whether scene graphs improve downstream control by injecting them into existing VLA policy prompts. Experiments across open-source and commercial VLMs show that current models often predict plausible relations but struggle with exact depth-aware and viewpoint-dependent spatial structure. EmbodimentSemantic provides a unified framework for diagnosing spatial grounding in VLM perception and testing its utility for VLA manipulation.
Chinese Translation
空间定位仍然是视觉-语言-行动（VLA）系统在机器人操作中的一个关键限制。虽然当前模型能够识别物体并遵循语言指令，但它们通常缺乏对物体在空间中如何排列的明确表示，包括支撑、包含、排序、遮挡和深度敏感关系。我们介绍了EmbodimentSemantic，一个用于评估具身操作中关系定位的空间场景图数据集和基准。EmbodimentSemantic将场景表示为有向的物体-关系-物体三元组，其中每个三元组使用固定的关系集指定一对有序物体之间的空间关系。这种表示方式使得对物体绑定、关系预测和空间一致性的直接评估成为可能。该数据集包括使用低成本的SO101机器人手臂收集的真实世界操作观察，以及用于研究实际机器人环境中空间定位的生成场景图。为了提供受控的验证，我们还引入了一个基于模拟器的LIBERO基准，包含超过60K的操作帧和超过120K的特定相机场景图，涵盖配对的第三人称视角和手腕视角，其中真实关系是通过MuJoCo几何体、世界坐标、相机投影和可见性约束自动推导的。我们进一步测试场景图是否通过将其注入现有的VLA策略提示中来改善下游控制。跨开源和商业视觉语言模型的实验表明，当前模型通常能够预测合理的关系，但在精确的深度感知和视角依赖的空间结构方面存在困难。EmbodimentSemantic提供了一个统一的框架，用于诊断VLM感知中的空间定位，并测试其在VLA操作中的实用性。
cs.RO / 2 / 2607.00022
When to Personalize Household Object Search: A Rigidity-Gated Hybrid Policy
何时个性化家庭物品搜索：一种刚性门控混合策略
Abstract
Service robots searching for household objects rely on spatial priors to reduce search cost, yet object locations can vary with resident traits. Collecting longitudinal, trait-specific in-home trajectories is invasive and hard to scale. We study when personalization helps and propose PerSim, a rigidity-gated hybrid policy that combines a trait-conditioned prior with a population-frequency baseline, personalizing only when placement behavior is variable. To scale resident-conditioned dynamics, we employ a human-calibrated simulation pipeline to generate and validate object-placement transitions in diverse home layouts, and train a predictor that injects continuous Big Five vectors to output room-level priors and within-room co-occurrence cues. In a unified human study (N=200), dual-layer validation shows that (i) synthetic transitions are judged behaviorally plausible (mean 3.85/5, p < 1e-6), and (ii) in a blinded A/B comparison, personalization is favored primarily for low-rigidity objects (p=0.005), while the population-frequency baseline remains strong for universally placed items, yielding a decision rule for when to personalize. In an offline objective test, we observe a small but significant improvement on unseen continuous trait vectors over nearest discrete configuration matching (p=0.035), supporting interpolation in five-dimensional trait space. Finally, in a home digital twin we show that PerSim reduces expected search cost by combining room visitation effort with within-room cue checking, demonstrating end-to-end gains beyond isolated prediction metrics.
Chinese Translation
服务机器人在搜索家庭物品时依赖空间先验以降低搜索成本，但物品位置可能因居民特征而异。收集纵向的、特征特定的家庭轨迹具有侵入性且难以扩展。我们研究了个性化何时有帮助，并提出了PerSim，一种结合特征条件先验与人口频率基线的刚性门控混合策略，仅在放置行为变化时进行个性化。为了扩展居民条件动态，我们采用人类校准的仿真管道生成并验证多样家庭布局中的物品放置转变，并训练一个预测器，该预测器注入连续的五大人格特质向量以输出房间级先验和房间内共现线索。在一项统一的人类研究中（N=200），双层验证显示：（i）合成转变被判断为行为上合理（平均3.85/5，p < 1e-6），以及（ii）在盲测的A/B比较中，个性化主要针对低刚性物品（p=0.005）更受青睐，而人口频率基线在普遍放置的物品中仍然强劲，从而得出何时进行个性化的决策规则。在离线客观测试中，我们观察到在未见的连续特征向量上，相较于最近的离散配置匹配，存在小但显著的改善（p=0.035），支持在五维特征空间中的插值。最后，在家庭数字双胞胎中，我们展示了PerSim通过结合房间访问努力与房间内线索检查来降低预期搜索成本，展示了超越孤立预测指标的端到端收益。
cs.RO / 3 / 2607.00024
Decentralized Geometric Control for Cable-Suspended Payload Transport with Adaptive Mass Estimation
基于去中心化几何控制的自适应质量估计的缆索悬挂载荷运输
Abstract
Cooperative aerial transport requires controllers that respect nonlinear manifold geometry, operate without centralized coordination, and respect operational safety constraints. To address these demands, we present GPAC, a four-layer hierarchical architecture that enables $N$ quadrotors to transport a cable-suspended payload without a central coordinator or by exchanging cable states or adaptive parameters. The key insight is implicit coordination: each quadrotor independently estimates its effective load share from local cable measurements, so combined forces converge to the correct total, even without knowledge of $N$ or the payload mass; the payload position is reconstructed locally from each agent's own cable geometry, and the only inter-agent communication is a low-rate neighbor-position broadcast for collision avoidance. GPAC operates directly on the full nonlinear configuration manifold and integrates geometric position and attitude control, anti-swing regulation, an extended-state observer for wind rejection, concurrent learning-based mass estimation without persistent excitation, and a priority-ordered control barrier function (CBF)-inspired safety filter that reduces operational risk, with input-to-state safety (ISSf) margins that hold exactly under single-constraint activation. A compatibility result shows that the filter's force modifications keep the desired attitude within the almost-global stability region of the $\mathrm{SO}(3)$ attitude controller. Finally, high-fidelity simulation with flexible cables, onboard sensor fusion, and wind turbulence -- with all control and estimation loops closed through the estimator -- yields a mean payload-tracking RMSE of 33.8 cm (2.8\% coefficient of variation over 13 seeds) at a low per-agent computational cost.
Chinese Translation
协作空中运输需要尊重非线性流形几何的控制器，能够在没有集中协调的情况下操作，并遵循操作安全约束。为了解决这些需求，我们提出了GPAC，一种四层层次结构，使得$N$个四旋翼无人机能够在没有中央协调者或通过交换缆索状态或自适应参数的情况下运输缆索悬挂载荷。关键的见解是隐式协调：每个四旋翼无人机独立地从局部缆索测量中估计其有效载荷分配，因此即使在不知道$N$或载荷质量的情况下，合力也会收敛到正确的总和；载荷位置是从每个代理自身的缆索几何中局部重建的，代理之间唯一的通信是为了避免碰撞而进行的低频邻居位置广播。GPAC直接在完整的非线性配置流形上操作，并集成了几何位置和姿态控制、抗摆动调节、用于风干扰抑制的扩展状态观测器、基于学习的质量估计（无需持续激励）以及优先级排序的控制障碍函数（CBF）启发式安全过滤器，以降低操作风险，且在单约束激活下，输入到状态安全（ISSf）边际完全成立。兼容性结果表明，过滤器的力修改保持所需姿态在$ ext{SO}(3)$姿态控制器的几乎全局稳定区域内。最后，使用柔性缆索、机载传感器融合和风扰动的高保真仿真——所有控制和估计环路通过估计器闭合——在每个代理的低计算成本下，获得了平均载荷跟踪均方根误差（RMSE）为33.8厘米（13个种子的变异系数为2.8%）。
cs.RO / 4 / 2607.00025
FLYNN: Robust Neural Network for Robot Navigation using Fly Brain Topology
FLYNN：基于果蝇脑拓扑的机器人导航鲁棒神经网络
Abstract
While deep learning models achieve state-of-the-art performance in complex tasks, they remain brittle when faced with new environments or sensory deprivation. In contrast, biological systems exhibit remarkable tolerance to these challenges. We address this vulnerability by developing a recurrent neural network (RNN) whose architecture is directly derived from the synaptic-resolution brain connectome of the fruit fly Drosophila melanogaster. We demonstrate the feasibility of training the fly connectome neural network (FLYNN) to perform vision-based navigation in MuJoCo, achieving performance comparable to modern hand-crafted networks of similar parameter counts. Crucially, FLYNN exhibits superior resistance to out-of-distribution (OOD) data and tolerance to sensory loss without further training. It remained functional even under total vision loss while hand-crafted networks largely failed, even when specifically trained with camera dropout. Principal Component Analysis (PCA) of the internal state of FLYNN suggests that it exhibits a particularly high degree of representational modularity, which might be related to its robustness. Our work provides a new direction for designing resilient artificial agents following the topology of biological brains.
Chinese Translation
尽管深度学习模型在复杂任务中取得了最先进的性能，但在面对新环境或感官缺失时，它们仍然表现出脆弱性。相比之下，生物系统对这些挑战表现出显著的容忍度。我们通过开发一种递归神经网络（RNN）来解决这一脆弱性，该网络的架构直接源自果蝇（Drosophila melanogaster）的突触分辨率脑连接组。我们展示了训练果蝇连接组神经网络（FLYNN）以在MuJoCo中执行基于视觉的导航的可行性，其性能与现代手工设计的具有相似参数数量的网络相当。重要的是，FLYNN对分布外（OOD）数据表现出更强的抵抗力，并且在没有进一步训练的情况下对感官丧失具有良好的容忍度。在完全失去视觉的情况下，它仍然能够正常工作，而手工设计的网络则大多失败，即使在专门训练过相机失效的情况下也是如此。对FLYNN内部状态的主成分分析（PCA）表明，它表现出特别高的表征模块化程度，这可能与其鲁棒性相关。我们的工作为设计遵循生物大脑拓扑结构的韧性人工智能代理提供了新的方向。
cs.RO / 5 / 2607.00026
Invariant Stochastic Filtering on SE(3) for Inertial-Encoder State Estimation of Serial Rigid Manipulators
基于SE(3)的不变随机滤波器用于串联刚性机械臂的惯性编码器状态估计
Abstract
An invariant extended Kalman filter (IEKF) is developed for state estimation of serial rigid manipulators with an arbitrary number of links, formulated entirely within the Lie group SE(3). The group-affine property of the kinematic equations makes the linearised error dynamics autonomous, so the Riccati equation governs the true error covariance rather than a local approximation. A physically separated noise model treats gyroscope and accelerometer channels independently: the accelerometer provides translational twist via gravity-compensated integration, yielding a measurement covariance that scales with the sample interval in exact analogy with process noise discretisation; a state-dependent Coriolis noise term captures gyroscope noise propagating through the nonlinear dynamics, vanishing at rest and growing with twist magnitude. The filter is structured as a modular chain of per-link IEKFs in which the predicted covariance of each link depends on its predecessor only through the Adjoint-transformed posterior, giving linear computational cost in link count. Exponential ultimate boundedness in mean square is established via a Lie algebra Lyapunov function, with per-link bounds chained through the Adjoint operator norm to yield a stability certificate that is modular and scalable to arbitrary chain length. Numerical results validate the design.
Chinese Translation
本文开发了一种不变扩展卡尔曼滤波器（IEKF），用于具有任意数量连杆的串联刚性机械臂的状态估计，完全在李群SE(3)内进行公式化。运动学方程的群仿射性质使得线性化误差动态是自主的，因此，里卡提方程控制真实误差协方差，而不是局部近似。物理分离的噪声模型独立处理陀螺仪和加速度计通道：加速度计通过重力补偿积分提供平移扭转，产生的测量协方差与样本间隔成比例，完全类似于过程噪声离散化；状态相关的科里奥利噪声项捕捉了通过非线性动态传播的陀螺仪噪声，在静止时消失，并随着扭转幅度的增加而增大。滤波器结构为每个连杆IEKF的模块化链，其中每个连杆的预测协方差仅通过伴随变换后的后验依赖于其前驱，给出了与连杆数量成线性关系的计算成本。通过李代数李雅普诺夫函数建立了均方的指数最终有界性，通过伴随算子范数将每个连杆的界限连接起来，得出一个模块化且可扩展到任意链长的稳定性证明。数值结果验证了设计的有效性。
cs.RO / 6 / 2607.00027
Urban Deceleration Behavior Modes Under Scene Context: An Early-Kinematic Classifier from Argoverse 2 Multi-Agent Trajectories
场景上下文下的城市减速行为模式：来自 Argoverse 2 多智能体轨迹的早期运动学分类器
Abstract
Urban deceleration is one of the most empirically studied yet least taxonomically organized behaviors in car-following research. Recent perception-equipped autonomous-vehicle datasets enable trajectory-anchored mode discovery. We extract 1,219 sustained deceleration events from 234 urban driving logs of the Argoverse 2 Sensor dataset, encode each event in a 19-dimensional kinematic feature vector, discover behavioral modes via K-means clustering with bootstrap stability analysis, and quantify modulation by eleven scene-context variables. A HistGradientBoosting classifier predicts mode membership from the first 1.0 s of each event. Four stable modes emerge with a bootstrap Adjusted Rand Index of 0.897 across 50 resamples: anticipatory soft (62.8%), reactive closing (30.6%), brake-like jerk (4.8%), and an outlier category (1.8%). Only pair age shows a medium effect (epsilon^2 = 0.085); scene geometry and vulnerable-road-user proximity show negligible effects. The early-event classifier achieves macro-F1 = 0.758 at 1.0 s, with scene context contributing +0.059 F1 over kinematics alone. Modes are regime-invariant in medium-speed driving (ARI = 0.817) but regime-dependent at low speed (ARI = 0.166). A small set of stable kinematic modes structures urban deceleration; early-window jerk dominates predictive signal; and pair age is the primary contextual modulator.
Chinese Translation
城市减速是汽车跟随研究中最常被实证研究但在分类上最缺乏组织的行为之一。近期配备感知的自动驾驶车辆数据集使得基于轨迹的模式发现成为可能。我们从 Argoverse 2 传感器数据集中提取了 1,219 个持续减速事件，基于 234 个城市驾驶日志，将每个事件编码为 19 维运动学特征向量，通过 K-means 聚类和自助稳定性分析发现行为模式，并通过十一种场景上下文变量量化调制。HistGradientBoosting 分类器根据每个事件的前 1.0 秒预测模式归属。经过 50 次重采样，四种稳定模式以 0.897 的自助调整兰德指数（Adjusted Rand Index）出现：预期软减速（62.8%）、反应性闭合（30.6%）、刹车样的急剧减速（4.8%）和异常类别（1.8%）。只有配对年龄显示出中等效应（epsilon^2 = 0.085）；场景几何和脆弱道路用户的接近显示出微不足道的影响。早期事件分类器在 1.0 秒时实现了宏 F1 = 0.758，场景上下文相较于单独的运动学贡献了 +0.059 F1。在中速驾驶中，模式是独立于驾驶状态的（ARI = 0.817），但在低速时则依赖于驾驶状态（ARI = 0.166）。一小组稳定的运动学模式构建了城市减速；早期窗口的急剧减速主导了预测信号；而配对年龄是主要的上下文调节因素。
cs.RO / 7 / 2607.00028
Trajectory Learning with Graph Representations for Social Robot Navigation
基于图表示的轨迹学习用于社交机器人导航
Abstract
Autonomous mobile robots are expected to exhibit socially compliant navigation for minimizing pedestrian disturbance. While capturing social interactions and incorporating pedestrian motion estimations into decision-making are beneficial for compliance, prior methods fail to address both spatial and temporal characteristics present in real-world data. Reinforcement Learning offers high capability, but it requires hand-crafted reward functions that reduce social behavior to static criteria, limiting its ability to reproduce patterns that exist in real pedestrian behavior. Imitation Learning offers direct training from real-world data but lacks modeling of social interactions and suffers from error accumulation. To this end, we propose an imitation learning framework that leverages spatiotemporal dynamics for socially compliant navigation. To represent social context based on interactions, we introduce a graph-based auxiliary network that encodes crowd states by attending to pedestrians. In addition, we present a navigation module that captures temporal dynamics and mitigates error accumulations by incorporating encoded state predictions and employing a trajectory-level learning objective. Our framework outperforms established data-driven baselines on simulation and a real-world dataset across diverse social metrics.
Chinese Translation
自主移动机器人被期望展现出符合社会规范的导航，以最小化对行人的干扰。尽管捕捉社会互动并将行人运动估计纳入决策过程有助于遵循社会规范，但以往的方法未能有效处理真实世界数据中存在的空间和时间特征。强化学习具有较强的能力，但它需要手工设计的奖励函数，这将社会行为简化为静态标准，限制了其再现真实行人行为模式的能力。模仿学习能够直接从真实世界数据中进行训练，但缺乏对社会互动的建模，并且容易出现误差累积。为此，我们提出了一种模仿学习框架，利用时空动态实现符合社会规范的导航。为了基于互动表示社会背景，我们引入了一个基于图的辅助网络，通过关注行人来编码人群状态。此外，我们还提出了一个导航模块，捕捉时间动态，并通过整合编码状态预测和采用轨迹级学习目标来减轻误差累积。我们的框架在模拟和真实世界数据集上，在多种社会指标下超越了既有的数据驱动基准。
cs.RO / 8 / 2607.00029
Memory-Native Non-Terrestrial Networks for Embodied Intelligence
面向具身智能的记忆原生非地面网络
Abstract
Non-terrestrial networks (NTN) provide ubiquitous connectivity for embodied intelligence (EI), enabling robots in wilderness to leverage cloud resources or report critical information to remote centers. However, the synergy is nontrivial due to the highly-dynamic, resource-constrained, topology-varying, and task-oriented environment. Existing memoryless NTN protocols become inefficient, since the decisions are driven by local channel conditions and instantaneous service demands. To address these limitations, this paper proposes the memory-native NTN (MemNTN) paradigm that leverages long-horizon contexts for memory augmented system optimization. To realize this paradigm shift, we establish a dual-memory architecture that distinguishes between physical memory representing the state of the world and digital memory encoding historical network experience. We develop memory acquisition, compression, valuation, update, and utilization mechanisms that facilitate cross-layer, memory-native decision-making, spanning from the physical and access layers up to the network and application layers. Experiments in satellite embodied question answering (SEQA) demonstrate that the proposed MemNTN significantly outperforms conventional stateless NTN and terrestrial approaches.
Chinese Translation
非地面网络（NTN）为具身智能（EI）提供了无处不在的连接，使得在荒野中的机器人能够利用云资源或向远程中心报告关键信息。然而，由于环境高度动态、资源受限、拓扑变化和任务导向的特性，这种协同并非易事。现有的无记忆NTN协议效率低下，因为决策是由局部信道条件和瞬时服务需求驱动的。为了解决这些局限性，本文提出了记忆原生NTN（MemNTN）范式，该范式利用长期上下文进行记忆增强系统优化。为了实现这一范式转变，我们建立了一个双记忆架构，区分代表世界状态的物理记忆和编码历史网络经验的数字记忆。我们开发了记忆获取、压缩、评估、更新和利用机制，以促进跨层的记忆原生决策，涵盖从物理层和接入层到网络层和应用层的各个层面。在卫星具身问答（SEQA）实验中，所提出的MemNTN显著优于传统的无状态NTN和地面方法。
cs.RO / 9 / 2607.00030
A Unified Benchmark for RCM-Constrained Visual Servoing: Modeling-Controller Interaction and Robustness Analysis in Laparoscopic Robots
针对 RCM 约束视觉伺服的统一基准：腹腔镜机器人中的建模-控制器交互与鲁棒性分析
Abstract
In robot-assisted laparoscopic minimally invasive surgery (MIS), accurate enforcement of the remote center of motion (RCM) constraint is critical for safe and stable automatic field-of-view (FoV) adjustment. Although control-based RCM strategies are widely adopted due to their flexibility and cost-effectiveness, systematic comparison of different RCM formulations and image-based visual servoing (IBVS) frameworks remains challenging due to the lack of a unified and reproducible benchmark. This paper presents an open-source simulation framework integrating three representative RCM modeling approaches and six IBVS-based control architectures within a unified velocity-level formulation, enabling controlled and consistent evaluation. Through structured case studies, the framework reveals key structural sensitivities arising from modeling and controller interactions, including the impact of tangent-plane definition, constraint dimensionality, open- versus closed-loop enforcement, and robustness near kinematic singularities. All resources are released and demostrations are provided in the supplementary video, providing a reproducible foundation for RCM-constrained visual servoing research.
Chinese Translation
在机器人辅助的腹腔镜微创手术中，准确执行运动中心远程（RCM）约束对于安全和稳定的自动视野（FoV）调整至关重要。尽管基于控制的 RCM 策略因其灵活性和成本效益而被广泛采用，但由于缺乏统一且可重复的基准，不同 RCM 公式和基于图像的视觉伺服（IBVS）框架的系统比较仍然具有挑战性。本文提出了一个开源仿真框架，整合了三种代表性的 RCM 建模方法和六种基于 IBVS 的控制架构，采用统一的速度级别公式，从而实现可控且一致的评估。通过结构化案例研究，该框架揭示了建模与控制器交互所产生的关键结构敏感性，包括切平面定义、约束维度、开环与闭环执行的影响，以及在运动学奇异点附近的鲁棒性。所有资源均已发布，并在补充视频中提供演示，为 RCM 约束视觉伺服研究提供了可重复的基础。
cs.RO / 10 / 2607.00031
Joint Discovery of Object and Action Symbols through Effect Prediction for Robotic Manipulation Planning
通过效果预测联合发现物体和动作符号以进行机器人操作规划
Abstract
To perform complex manipulation planning, autonomous robots are required to abstract continuous, high-dimensional sensorimotor interactions into discrete object and action representations. Earlier work either categorized objects based on visual appearances, which fails to distinguish objects that appear similar but behave differently, or based on effects under interaction, but was limited to predefined actions. To address these limitations, we propose a model that jointly discovers high-level manipulation primitives and object categories through a binary bottleneck layer, trained to predict multi-modal outcomes, including object motion, contact, and force feedback, from random interaction data. Building on these discovered binary representations, we leverage a discrete planning method that uses intermediate steps in the predicted effect trajectory to enable partial action executions for precise low-level control. Additionally, we evaluate our framework's generalization capabilities on novel objects by assigning object categories through comparing a small number of interaction effects with the predicted effects of learned object symbols, enabling few-shot generalization based on behavior rather than visual similarity. We conduct experiments on tabletop repositioning and stacking tasks, and confirm that our effect-driven planning approach outperforms both a state-of-the-art method and a visual-based alternative in planning precision across seen and novel objects.
Chinese Translation
为了执行复杂的操作规划，自治机器人需要将连续的高维传感运动交互抽象为离散的物体和动作表示。早期的研究要么基于视觉外观对物体进行分类，这无法区分外观相似但行为不同的物体，要么基于交互下的效果进行分类，但仅限于预定义的动作。为了解决这些局限性，我们提出了一种模型，通过一个二元瓶颈层联合发现高级操作原语和物体类别，该层经过训练以预测多模态结果，包括物体运动、接触和力反馈，基于随机交互数据。在这些发现的二元表示基础上，我们利用一种离散规划方法，使用预测效果轨迹中的中间步骤来实现部分动作执行，以实现精确的低级控制。此外，我们通过比较少量交互效果与学习到的物体符号的预测效果，为新物体分配物体类别，从而评估我们框架在新物体上的泛化能力，使得基于行为而非视觉相似性的少样本泛化成为可能。我们在桌面重新定位和堆叠任务上进行了实验，确认我们的基于效果的规划方法在已见和新物体的规划精度上优于最先进的方法和基于视觉的替代方案。
cs.RO / 11 / 2607.00033
Learning Dexterous Manipulation Using Contact Wrench Guidance From Human Demonstration
利用人类示范的接触扭矩引导学习灵巧操作
Abstract
Dexterous robot manipulation can benefit from the abundance of human demonstrations, but transferring such demonstrations to robot policies remains challenging. We present Contact Wrench Guidance from Human Demonstration in Robotic Dexterous Manipulation (CHORD), a framework for long-horizon manipulation of rigid and articulated objects with reinforcement learning. The key idea is object-centric contact wrench space guidance: we represent human and robot motions by the forces and torques they can induce on the object, enabling similarity to be measured by the induced instantaneous motions. This guidance makes reinforcement learning more scalable for contact-rich dexterous manipulation. We further introduce a large-scale simulation benchmark with 4,739 bimanual dexterous manipulation tasks, constructed from motion-capture datasets and reconstructed in-house videos. Evaluated on 1,831 benchmark tasks, CHORD achieves an average success rate of 82.12%, demonstrating strong scalability. CHORD also generalizes to whole-body manipulation from hand-only and third-person demonstrations, achieving a 90.77% success rate, and the learned policies transfer to the real world in both open-loop and closed-loop settings.
Chinese Translation
灵巧机器人操作可以从大量的人类示范中受益，但将这些示范转移到机器人策略上仍然具有挑战性。我们提出了“来自人类示范的接触扭矩引导在机器人灵巧操作中的应用”（CHORD），这是一个用于刚性和关节物体长时间操作的强化学习框架。其关键思想是以物体为中心的接触扭矩空间引导：我们通过人类和机器人在物体上施加的力和扭矩来表示它们的运动，从而能够通过施加的瞬时运动来测量相似性。这种引导使得强化学习在接触丰富的灵巧操作中更具可扩展性。我们进一步引入了一个大规模的模拟基准，包含4,739个双手灵巧操作任务，这些任务是从运动捕捉数据集构建并在内部视频中重建的。在1,831个基准任务上的评估中，CHORD实现了82.12%的平均成功率，展示了强大的可扩展性。CHORD还能够从仅手部和第三方示范中推广到全身操作，成功率达到90.77%，并且学习到的策略在开放环路和闭环设置中均能转移到现实世界。
cs.RO / 12 / 2607.00065
Optimal any-angle path planning in static and dynamic environments
静态和动态环境中的最优任意角路径规划
Abstract
Any-angle path planning extends traditional graph-based path planning by allowing movement between any pair of vertices, rather than being restricted by predefined edges. It can find straighter and shorter paths in continuous space with graphs, making it particularly suitable for navigation in open areas such as airspaces, warehouses, and oceans. Many any-angle path-planning algorithms have been proposed, but only a few can guarantee optimal solutions, especially in the presence of dynamic obstacles. To address this challenge, this article focuses on optimal any-angle path planning on grids and introduces two general techniques that accelerate computation while preserving optimality in both static and dynamic environments: 1) elliptical forward expansion, which leverages ellipse-based neighborhoods to restrict the search space, and 2) field of view, which replaces traditional line-of-sight methods to speed up visibility checks. To integrate these two techniques, inverted and forward scanning are introduced. Inverted scanning establishes visual connections from open nodes, whereas forward scanning initiates scans from closed nodes. Building on the proposed techniques, Zeta* and Zeta*-SIPP are developed for static and dynamic environments respectively. Zeta*, when combined with forward scanning, is similar to the state-of-the-art algorithm Anya and attains comparable performance. Unlike Anya, Zeta* can be readily extended to other settings, such as dynamic environments (e.g., Zeta*-SIPP). Zeta*-SIPP, with either scanning method, is more than 20 times faster than the corresponding state-of-the-art optimal planner TO-AA-SIPP. Overall, this research identifies the key requirements for achieving optimal any-angle path planning and introduces a unified approach suitable for different environments.
Chinese Translation
任意角路径规划通过允许在任意一对顶点之间移动，扩展了传统的基于图的路径规划，而不再受限于预定义的边。这使得它能够在连续空间中找到更直且更短的路径，特别适合于开放区域的导航，如空域、仓库和海洋。虽然已经提出了许多任意角路径规划算法，但只有少数能够保证最优解，尤其是在存在动态障碍物的情况下。为了解决这一挑战，本文聚焦于网格上的最优任意角路径规划，并介绍了两种通用技术，这些技术在静态和动态环境中加速计算的同时保持最优性：1）椭圆前向扩展，利用基于椭圆的邻域来限制搜索空间；2）视野，替代传统的视线方法以加快可见性检查。为了整合这两种技术，引入了反向扫描和前向扫描。反向扫描从开放节点建立视觉连接，而前向扫描则从封闭节点发起扫描。在提出的技术基础上，分别为静态和动态环境开发了Zeta*和Zeta*-SIPP。Zeta*结合前向扫描时，与最先进的算法Anya相似，并达到可比的性能。与Anya不同，Zeta*可以方便地扩展到其他设置，如动态环境（例如，Zeta*-SIPP）。无论使用哪种扫描方法，Zeta*-SIPP的速度都比相应的最先进的最优规划器TO-AA-SIPP快20倍以上。总体而言，本研究确定了实现最优任意角路径规划的关键要求，并提出了一种适用于不同环境的统一方法。
cs.RO / 13 / 2607.00066
Learning Expert Strategy for Autonomous Robotic Endovascular Intervention via Decoupled Procedural Execution
通过解耦程序执行学习自主机器人血管内干预的专家策略
Abstract
Endovascular interventions are high-stakes procedures requiring precise device operation within complex and tortuous vascular anatomies. Autonomous endovascular navigation has the potential to standardize procedural quality and reduce the performance variability inherent in manual operation. Although Reinforcement Learning (RL) approaches have demonstrated promise in enabling autonomy in endovascular intervention, they often struggle with explicit constraint satisfaction and safety guarantees. To address these challenges, a learning-based expert strategy is introduced, enhancing procedural consistency in autonomous endovascular intervention by explicitly decoupling high-level strategic decision-making from low-level procedural execution. The proposed framework replicates the expert clinical decision-making process: a strategic RL policy generates global navigation intents, which are subsequently refined through an expert-informed execution module. This module ensures that robot movements strictly adhere to expert operational norms, real-time kinematic limits, and vessel safety constraints. Experimental evaluation across high-fidelity 3D simulations and a real-world robotic platform demonstrates that the proposed framework not only outperforms baseline policies but also effectively replicates expert-level proficiency. The framework achieves a high navigation success rate (> 96%) and a 29.3% reduction in operational steps, which translates to enhanced operative efficiency and minimized device-vessel interaction. Furthermore, a 13% reduction in trajectory variance indicates superior procedural standardization, aligning autonomous behavior with established clinical norms. These results underscore its potential to enhance the predictability, safety, and consistency of robotic endovascular interventions.
Chinese Translation
血管内干预是一种高风险的手术，要求在复杂且曲折的血管解剖结构中进行精确的设备操作。自主血管内导航有潜力标准化手术质量，并减少手动操作固有的性能变异性。尽管强化学习（Reinforcement Learning, RL）方法在实现血管内干预的自主性方面显示出前景，但它们在显式约束满足和安全保障方面往往面临挑战。为了解决这些问题，提出了一种基于学习的专家策略，通过显式解耦高层战略决策与低层程序执行，增强自主血管内干预的程序一致性。所提出的框架复制了专家临床决策过程：战略RL策略生成全局导航意图，随后通过专家指导的执行模块进行细化。该模块确保机器人运动严格遵循专家操作规范、实时运动学限制和血管安全约束。在高保真3D模拟和真实机器人平台上的实验评估表明，所提出的框架不仅优于基线策略，而且有效复制了专家级的熟练程度。该框架实现了高达96%以上的导航成功率，并减少了29.3%的操作步骤，从而提高了操作效率并最小化了设备与血管的相互作用。此外，轨迹方差减少了13%，表明程序标准化优越，使自主行为与既定临床规范相一致。这些结果强调了其在增强机器人血管内干预的可预测性、安全性和一致性方面的潜力。
cs.RO / 14 / 2607.00141
AD-MPCC: Adaptive Differentiable Model Predictive Contouring Control for Autonomous Racing
AD-MPCC：用于自主赛车的自适应可微模型预测轮廓控制
Abstract
This paper presents Adaptive Differentiable Model Predictive Contouring Control (AD-MPCC), a framework for autonomous racing that integrates differentiable MPCC with online parameter estimation to handle varying road-surface conditions. For online parameter estimation, we leverage a parameterized Pacejka Magic Formula together with a regularized moving-horizon estimation scheme with exponentially decaying weights to capture road interactions and update parameters in real time. Furthermore, we propose a differentiable MPCC (Diff-MPCC) framework that enables optimal adjustment of objective weights based on predefined long-horizon performance costs. To implement Diff-MPCC for online objective weight adaptation, we propose a Pacejka-informed machine learning model that is trained in a supervised manner using data generated by Diff-MPCC to tune the objective weights. Simulation results demonstrate that AD-MPCC reliably ensures safety and achieves faster lap times compared to baseline controllers in both single-surface and multiple-surface scenarios.
Chinese Translation
本文提出了自适应可微模型预测轮廓控制（AD-MPCC），这是一个集成了可微MPCC与在线参数估计的自主赛车框架，以应对变化的路面条件。在在线参数估计中，我们利用了参数化的Pacejka魔法公式，并结合正则化的移动视野估计方案，采用指数衰减权重来捕捉路面交互并实时更新参数。此外，我们提出了一种可微MPCC（Diff-MPCC）框架，使得基于预定义的长时间性能成本对目标权重进行最佳调整。为了实现Diff-MPCC的在线目标权重适应，我们提出了一种基于Pacejka的机器学习模型，该模型通过使用Diff-MPCC生成的数据以监督方式进行训练，以调整目标权重。仿真结果表明，AD-MPCC在单一表面和多表面场景中均能可靠地确保安全，并实现比基线控制器更快的圈速。
cs.RO / 15 / 2607.00142
Stop Pretending Social Robots Are Inevitable
停止假装社会机器人是不可避免的
Abstract
This paper takes issue with the recent themes of both the RO-MAN and the HRI conferences for their portrayal of a future human-robot society as inevitable. The focus is on discussing how such statements ultimately shape research. By treating a future human-robot society as a fait accompli, license is given for user studies to imagine any scenario they like, no matter whether it has any ecological relevance, and to emphasise the scenario design over actually creating robot abilities needed to fullfill the imagined role. Meanwhile, research that focusses on actual societal needs, without assuming that robots are a solution, is deprioritised, as is technical development, in particular with respect to abilities that are necessary to enable robots that function as social agents rather than a mere automation of tasks. A frame that simply assumes a robot future not only detracts from scientific advancement in favour of a techno-solutionism we ought to resist, it is also self-defeating as it risks stifling the research needed to bring it about. We should therefore reject attempts to frame and promote the field in terms of the inevitable social robot and instead focus on one that facilitates advances in the field regardless of what the future holds. This paper suggests that a renewed focus on cognitive mechanisms necessary for the "I" in HRI would be a good starting point.
Chinese Translation
本文对RO-MAN和HRI会议近期主题提出质疑，认为它们将未来人机社会描绘为不可避免的。这篇论文重点讨论此类表述如何最终影响研究。通过将未来人机社会视为既成事实，研究者被允许进行用户研究，想象任何他们喜欢的场景，无论其是否具有生态相关性，并强调场景设计而非实际创造实现想象角色所需的机器人能力。同时，关注实际社会需求的研究被降级，假设机器人是解决方案的技术开发也受到忽视，尤其是在使机器人能够作为社会代理而不仅仅是任务自动化所需的能力方面。简单假设机器人未来的框架不仅削弱了科学进步，助长了我们应当抵制的技术解决主义，同时也是自我挫败的，因为它可能抑制实现这一目标所需的研究。因此，我们应当拒绝将该领域框定和推广为不可避免的社会机器人，而应专注于促进该领域进步的研究，无论未来如何。本文建议，重新关注HRI中“我”所需的认知机制将是一个良好的起点。
cs.RO / 16 / 2607.00145
Iterated Invariant EKF for 3D Landmark-Aided Inertial Navigation
基于三维地标辅助的惯性导航的迭代不变扩展卡尔曼滤波
Abstract
Inertial navigation systems aided by three-dimensional landmark measurements constitute a fundamental problem in robotic perception and state estimation. Classical SO(3)-based Extended Kalman Filter (SO(3)-EKF) approaches provide practical solutions, but suffer from the false observability problem, in which the filter becomes overconfident in unobservable directions, leading to degraded estimation performance. The Invariant EKF (IEKF) addresses this limitation by reformulating the system dynamics as a group-affine system on a Lie group, although its measurement update does not fully satisfy certain state compatibility properties. More recently, the Iterated Invariant EKF (IterIEKF) was proposed to further improve the IEKF by ensuring, in the low-noise regime, that the estimated state remains on the observed state manifold while the uncertainty is confined to its tangent space. In this work, we formulate and apply the IterIEKF to landmark-based inertial 3D localization for the first time. Through numerical simulations, we show that the proposed approach outperforms the classical SO(3)-EKF, the Iterated SO(3)-EKF, and the IEKF in terms of both estimation accuracy and consistency.
Chinese Translation
由三维地标测量辅助的惯性导航系统构成了机器人感知和状态估计中的一个基本问题。经典的基于SO(3)的扩展卡尔曼滤波器（SO(3)-EKF）方法提供了实用的解决方案，但存在虚假可观测性问题，即滤波器在不可观测方向上变得过于自信，导致估计性能下降。不变扩展卡尔曼滤波器（IEKF）通过将系统动态重新表述为李群上的群仿射系统来解决这一限制，尽管其测量更新并未完全满足某些状态兼容性特性。最近，提出了迭代不变扩展卡尔曼滤波器（IterIEKF），以进一步改进IEKF，确保在低噪声环境下，估计状态保持在观测状态流形上，同时不确定性被限制在其切空间内。在本研究中，我们首次将IterIEKF应用于基于地标的惯性三维定位。通过数值仿真，我们展示了所提出的方法在估计精度和一致性方面优于经典的SO(3)-EKF、迭代SO(3)-EKF和IEKF。
cs.RO / 17 / 2607.00148
3D Point World Models: Point Completion Enables More Accurate Dynamics Learning
3D 点世界模型：点补全使动态学习更准确
Abstract
Learning predictive models of the world enables robotic control through planning, potentially allowing robots to improvise solutions on new tasks. However, large video-based dynamics models lack explicit 3D spatial structure and suffer from geometrically inconsistent long-term rollouts with compounding errors. Emerging 3D dynamics models based on partial point clouds improve geometric consistency but remain sensitive to occlusions and accumulated prediction drift. To address these challenges, we present 3D Point World Models (3DPWM) - a task-agnostic world model that operates entirely in 3D space by first completing partial point clouds and then learning action-conditioned dynamics in this completed 3D scene. By operating on completed geometry, 3DPWM enables reliable long-horizon rollouts and more accurate cost evaluation for model-based planning while supporting adaptation to new tasks. Experiments across different robotic embodiments and tabletop manipulation benchmarks demonstrate that 3DPWM achieves significantly more reliable long-horizon rollouts (100-300+ steps), supports both open-loop and closed-loop planning, and enables successful sim-to-real transfer.
Chinese Translation
学习世界的预测模型使得通过规划进行机器人控制成为可能，从而允许机器人在新任务上即兴创造解决方案。然而，基于大视频的动态模型缺乏明确的 3D 空间结构，并且在长期滚动预测中存在几何不一致和累积误差的问题。基于部分点云的新兴 3D 动态模型改善了几何一致性，但仍然对遮挡和累积预测漂移敏感。为了解决这些挑战，我们提出了 3D 点世界模型（3DPWM）——一种与任务无关的世界模型，它通过首先完成部分点云，然后在这个完成的 3D 场景中学习基于动作的动态，完全在 3D 空间中操作。通过在完成的几何体上操作，3DPWM 实现了可靠的长时间滚动预测和更准确的基于模型的规划成本评估，同时支持对新任务的适应。在不同的机器人实现和桌面操作基准测试中的实验表明，3DPWM 实现了显著更可靠的长时间滚动预测（100-300+ 步），支持开环和闭环规划，并实现了成功的模拟到现实转移。
cs.RO / 18 / 2607.00156
Dual-Informed Vertical Expansion for Multi-Objective Node Selection in Anytime Conflict-Based Search
基于双重信息的垂直扩展用于随时冲突基础搜索中的多目标节点选择
Abstract
Conflict-Based Search (CBS) is a leading exact algorithm for Multi-Agent Path Finding (MAPF), but its high-level node-selection rule is usually treated as a fixed implementation detail. Standard best-first selection is strong for minimizing expanded nodes and closing the optimality certificate, yet it can maintain a large frontier, interrupt parent-child expansion sequences, and provide no feasible incumbent until termination. This paper studies node selection as a first-class design choice for exact CBS. We introduce Dual-Informed Vertical Expansion (DIVE), a policy that is best-bound between dives and depth-oriented within a dive. DIVE starts each dive from the current best-bound frontier, follows promising children to exploit parent-child locality, and uses incumbent pruning to limit unproductive excursions. We formalize CBS node selection through a branch-and-bound view, prove that the traversal policy can be changed without affecting exactness, and analyze the resulting trade-offs among expanded nodes, dive breaks, queue size, and primal-dual bound progress. The analysis predicts three complementary extremes. Best-first search is node efficient, iterative deepening is memory efficient, and DIVE is dive efficient while retaining regular best-bound reanchoring. Experiments on standard MAPF benchmarks support this trade-off map. DIVE consistently reduces dive breaks, provides early incumbents with certified gaps, uses substantially less queue memory than best-first search, and benefits from warm starts and simple responsive variants in dense or memory-limited regimes.
Chinese Translation
冲突基础搜索（CBS）是多智能体路径寻找（MAPF）的领先精确算法，但其高层节点选择规则通常被视为固定的实现细节。标准的最佳优先选择在最小化扩展节点和关闭最优性证书方面表现出色，但它可能维持一个较大的边界，打断父子扩展序列，并在终止之前不提供可行的现任解。本文将节点选择研究作为精确CBS的一个重要设计选择。我们引入了双重信息垂直扩展（DIVE），这是一种在潜水之间最佳界限且在潜水中以深度为导向的策略。DIVE从当前最佳界限边界开始每次潜水，跟随有前景的子节点以利用父子局部性，并使用现任解修剪来限制无效的游历。我们通过分支限界视角形式化CBS节点选择，证明遍历策略可以在不影响精确性的情况下进行更改，并分析扩展节点、潜水中断、队列大小和原始-对偶界限进展之间的权衡。分析预测了三种互补的极端情况。最佳优先搜索在节点效率上表现优异，迭代加深在内存效率上表现突出，而DIVE在保持常规最佳界限重新锚定的同时在潜水效率上表现良好。对标准MAPF基准的实验支持了这一权衡图。DIVE始终减少潜水中断，提供具有认证间隙的早期现任解，使用的队列内存显著少于最佳优先搜索，并在密集或内存受限的环境中受益于热启动和简单响应变体。
cs.RO / 19 / 2607.00160
Distributed Multi Robot Lunar Cargo Transportation via Phase Decomposed Reinforcement Learning
基于相位分解强化学习的分布式多机器人月球货物运输
Abstract
Modular reconfigurable robotic systems provide a scalable solution for cooperative surface operations in future lunar missions. However, cooperative cargo transportation remains challenging due to morphology-dependent topology changes, strong payload-induced coupling, long-horizon decision making, and safety constraints. This paper proposes a phase-decomposed reinforcement learning framework for cooperative cargo transport with distributed robotic units. The task is decomposed into lifting, transportation, and placement, each optimized with a dedicated joint-state policy capturing inter-agent coupling. Centralized training promotes stable convergence, while deployment uses onboard proprioception for control and OptiTrack motion capture for ground-truth evaluation and post-processed metrics. A deterministic phase controller expressed in Markov state representation regulates transitions between stages, and a failure-sensitive synchronization mechanism ensures coordinated progression and safety-aware halting during real-world execution. The framework is evaluated in simulation and through controlled field experiments at a JAXA space exploration test facility. Results demonstrate reliable cooperative transport across all stages in both simulation and hardware experiments.
Chinese Translation
模块化可重构机器人系统为未来月球任务中的合作表面操作提供了可扩展的解决方案。然而，由于形态依赖的拓扑变化、强负载引起的耦合、长时间决策以及安全约束，合作货物运输仍然面临挑战。本文提出了一种相位分解强化学习框架，用于分布式机器人单元的合作货物运输。该任务被分解为提升、运输和放置三个阶段，每个阶段都通过专门的联合状态策略进行优化，以捕捉代理间的耦合关系。集中训练促进了稳定收敛，而部署阶段则使用机载自我感知进行控制，并利用OptiTrack运动捕捉进行真实评估和后处理指标。以马尔可夫状态表示的确定性相位控制器调节阶段之间的转换，而一种对失败敏感的同步机制确保了协调进展和在实际执行中的安全停顿。该框架在模拟环境和JAXA空间探索测试设施的受控实地实验中进行了评估。结果表明，在模拟和硬件实验的所有阶段中，合作运输均表现出可靠性。
cs.RO / 20 / 2607.00191
HydraCollab: Adaptive Collaborative-Perception for Distributed Autonomous Systems
HydraCollab：用于分布式自主系统的自适应协作感知
Abstract
Collaborative-perception enables multi-robot systems to enhance situational awareness by sharing perceptual information. Existing collaborative-perception systems face an inherent trade-off between communication bandwidth requirements and perception accuracy, where methods that exchange more information achieve better perception results at the cost of increased communication overhead. However, real-world communication networks impose bandwidth constraints that require minimizing communication overhead without sacrificing perception performance. To address this challenge, we propose HydraCollab, an adaptive collaborative-perception framework that (i) selectively transmits the most informative sensor features and (ii) dynamically employs collaboration strategies (intermediate or late) based on spatial confidence maps. Extensive evaluations on the V2X-R, V2X-Radar and UAV3D-mini datasets demonstrate that HydraCollab achieves the best overall trade-off between accuracy and communication cost among existing collaborative-perception methods. Relative to SOTA Where2comm, HydraCollab uses only 41% of the bandwidth on V2X-R and 26% on V2X-Radar while improving performance by 0.78% and 0.75% respectively. Our code and models are available at https://github.com/AICPS/HydraCollab.
Chinese Translation
协作感知使多机器人系统能够通过共享感知信息来增强情境意识。现有的协作感知系统面临着通信带宽需求与感知准确性之间的固有权衡，其中交换更多信息的方法在增加通信开销的代价下实现更好的感知结果。然而，现实世界的通信网络施加了带宽限制，这要求在不牺牲感知性能的情况下最小化通信开销。为了解决这一挑战，我们提出了HydraCollab，一个自适应的协作感知框架，该框架（i）选择性地传输最具信息量的传感器特征，以及（ii）基于空间置信度图动态采用协作策略（中间或延迟）。在V2X-R、V2X-Radar和UAV3D-mini数据集上的广泛评估表明，HydraCollab在现有协作感知方法中实现了准确性与通信成本之间的最佳整体权衡。与最先进的Where2comm相比，HydraCollab在V2X-R上仅使用41%的带宽，在V2X-Radar上使用26%的带宽，同时分别提高了0.78%和0.75%的性能。我们的代码和模型可在https://github.com/AICPS/HydraCollab获取。
cs.RO / 21 / 2607.00215
ELMP: Efficient Learning for Motion Planning via Analytical Policy Gradients
ELMP：通过解析策略梯度实现高效运动规划学习
Abstract
Neural Motion Planners (NMPs) enable fast reactive motion generation, but adapting them to new environments typically requires recollecting large expert datasets, which is computationally prohibitive. We propose ELMP, a framework for data-efficient adaptation via self-supervised fine-tuning. Rather than generating additional expert trajectories with expensive global planners, ELMP directly optimizes the policy through a differentiable kinematic layer using dense collision, target-reaching, and smoothness objectives. This replaces expert data generation with rapid problem sampling, reducing per-sample adaptation cost by roughly two orders of magnitude. To further support robust generalization across changing kinematic chains, we introduce a mechanism to explicitly encode tool geometry via point clouds. Benchmarked against classical and neural baselines, ELMP achieves an 84.8% average success rate with orders-of-magnitude lower cold-start latency than classical methods. In unseen environments, self-supervised fine-tuning improves success rate from 57.3% (zero-shot) to 89.8%, removing the data collection bottleneck. Our approach maintains millisecond-level inference latency and is validated on a physical Franka Emika Panda robot.
Chinese Translation
神经运动规划器（NMPs）能够快速生成反应式运动，但将其适应于新环境通常需要重新收集大量专家数据集，这在计算上是不可行的。我们提出了ELMP，一个通过自监督微调实现数据高效适应的框架。ELMP并不通过昂贵的全局规划器生成额外的专家轨迹，而是通过使用密集碰撞、目标到达和光滑性目标的可微运动学层直接优化策略。这将专家数据生成替换为快速问题采样，将每个样本的适应成本降低了大约两个数量级。为了进一步支持在变化的运动链中稳健的泛化，我们引入了一种机制，通过点云显式编码工具几何形状。与经典和神经基线进行基准测试，ELMP实现了84.8%的平均成功率，冷启动延迟比经典方法低几个数量级。在未见过的环境中，自监督微调将成功率从57.3%（零样本）提高到89.8%，消除了数据收集瓶颈。我们的方法保持毫秒级的推理延迟，并在物理Franka Emika Panda机器人上进行了验证。
cs.RO / 22 / 2607.00272
ASPIRE: Agentic /Skills Discovery for Robotics
ASPIRE：机器人代理技能发现
Abstract
Traditional robot programming is challenging: it requires orchestrating multimodal perception, managing physical contact dynamics, and handling diverse configurations and execution failures. We introduce ASPIRE (Agentic Skill Programming through Iterative Robot Exploration), a continual learning system that autonomously writes and refines robot control programs in a code-as-policy paradigm while compounding experience into a reusable skill library. ASPIRE discovers skills that persist across tasks, simulation and real-world settings, and embodiments. It operates in an open-ended loop with three components: (1) a closed-loop robot execution engine that exposes fine-grained multimodal traces, enabling autonomous failure diagnosis, repair synthesis, and validation; (2) a continually expanding skill library that distills validated fixes into reusable, transferable knowledge; and (3) evolutionary search that generates diverse task sequences and control programs to explore beyond single-trajectory refinement. ASPIRE surpasses prior methods by up to 77% on LIBERO-Pro manipulation under perturbation, 72% on Robosuite bimanual handover, and 32% on BEHAVIOR-1K long-horizon household tasks. Its accumulated library also enables zero-shot generalization to unseen long-horizon tasks: on LIBERO-Pro Long, ASPIRE achieves 31% success versus 4% for prior methods despite their use of test-time reasoning and retries. Finally, simulation-discovered skills provide initial evidence of sim-to-real transfer, substantially reducing real-robot programming effort across different embodiments and robot APIs.
Chinese Translation
传统的机器人编程具有挑战性：它需要协调多模态感知、管理物理接触动态以及处理多样的配置和执行失败。我们引入了ASPIRE（通过迭代机器人探索的代理技能编程），这是一种持续学习系统，能够在代码即策略的范式下自主编写和完善机器人控制程序，同时将经验积累到可重用的技能库中。ASPIRE发现的技能在任务、仿真和现实世界环境以及不同的体现中持续存在。它在一个开放式循环中运行，包含三个组成部分：（1）一个闭环机器人执行引擎，能够暴露细粒度的多模态痕迹，从而实现自主故障诊断、修复合成和验证；（2）一个不断扩展的技能库，将经过验证的修复提炼为可重用、可转移的知识；（3）进化搜索，生成多样的任务序列和控制程序，以探索超越单一轨迹的优化。ASPIRE在LIBERO-Pro扰动下的操控任务上超越了先前的方法，提升幅度高达77%；在Robosuite双手交接任务上提升了72%；在BEHAVIOR-1K长时间家庭任务上提升了32%。其累积的技能库还实现了对未见长时间任务的零样本泛化：在LIBERO-Pro Long上，ASPIRE的成功率为31%，而先前的方法仅为4%，尽管它们使用了测试时推理和重试。最后，仿真发现的技能为仿真到现实的转移提供了初步证据，显著减少了在不同体现和机器人API下的真实机器人编程工作量。
cs.RO / 23 / 2607.00283
What's Hidden Matters: Identifying Planning-Critical Occluded Agents using Vision-Language Models
隐藏的事物至关重要：使用视觉-语言模型识别规划关键的遮挡代理
Abstract
Autonomous vehicles must safely navigate complex environments where planning-critical agents may be hidden from view. Current approaches often treat all occlusions with uniform conservatism, yielding needlessly defensive driving, or they infer hidden spaces without estimating the impact on the planner. This work bridges the critical gap between perception and planning by enabling Vision-Language Models (VLMs) to identify and reason about the specific hidden agents that are most critical to the ego-vehicle's trajectory. We introduce a novel framework that uses Planning KL-divergence (PKL), an information-theoretic metric, to systematically identify and rank occluded agents based on their impact on the ego vehicle's plan. Using this planning-aware ranking, we employ an expert VLM (GPT-5) to generate rich, structured annotations that capture the visual evidence and reasoning required for this task. We apply this framework to the nuScenes dataset to create a new benchmark focused on high-impact scenarios. We conduct comprehensive experiments on a wide range of general-purpose and domain-adapted VLMs, demonstrating that fine-tuning on our PKL-guided data yields dramatic performance improvements across all models. Notably, our results show that smaller, fine-tuned models significantly outperform their much larger zero-shot counterparts, and that our PKL-guided data selection strategy improves performance by approximately 30\% over random sampling. Our work presents the first systematic approach for training VLMs to focus on planning-critical occlusions, enabling more semantically grounded and efficient risk assessment in autonomous driving.
Chinese Translation
自主车辆必须安全地在复杂环境中导航，其中规划关键的代理可能被遮挡而无法看到。目前的方法通常以统一的保守态度处理所有遮挡，导致不必要的防御性驾驶，或者在推断隐藏空间时未能评估对规划者的影响。本研究填补了感知与规划之间的关键空白，使视觉-语言模型（VLMs）能够识别和推理对自我车辆轨迹最关键的特定隐藏代理。我们提出了一种新颖的框架，使用规划KL散度（Planning KL-divergence, PKL）这一信息论度量，系统地识别和排名基于其对自我车辆计划影响的遮挡代理。利用这种规划感知的排名，我们采用专家级VLM（GPT-5）生成丰富的结构化注释，以捕捉完成此任务所需的视觉证据和推理。我们将该框架应用于nuScenes数据集，以创建一个专注于高影响场景的新基准。我们在广泛的通用和领域适应VLM上进行了全面实验，证明在我们的PKL引导数据上进行微调可显著提高所有模型的性能。值得注意的是，我们的结果表明，较小的微调模型显著优于其更大的零-shot对手，并且我们的PKL引导数据选择策略比随机抽样提高了约30%的性能。我们的工作提出了训练VLM专注于规划关键遮挡的首个系统性方法，从而在自主驾驶中实现更具语义基础和高效的风险评估。
cs.RO / 24 / 2607.00326
NeHMO: Neural Hamilton-Jacobi Reachability Learning for Decentralized Safe Multi-Arm Motion Planning
NeHMO：用于分散安全多臂运动规划的神经哈密顿-雅可比可达性学习
Abstract
Safe multi-arm motion planning is a challenging problem in robotics due to its high dimensionality, coupled configuration space, and complex collision constraints. Centralized planners are capable of coordinating all arms but often face scalability limitations, restricting applicability in real-time settings. On the other hand, decentralized methods are scalable and recent deep learning-based approaches have shown promising results. However, these depend on accurate behavior prediction or coordination protocols and may fail when other arms act unpredictably. To address these challenges, we introduce a neural Hamilton-Jacobi Reachability (HJR) learning-based approach to approximate a safety value function that captures worst-case inter-arm safety constraints. We further develop a decentralized trajectory optimization framework that uses the learned HJR representation for real-time planning. The proposed method is scalable and data-efficient, generalizes across multi-manipulator systems, and outperforms state-of-the-art baselines on challenging multi-arm motion planning tasks.
Chinese Translation
安全的多臂运动规划是机器人技术中的一个挑战性问题，主要由于其高维度、耦合的配置空间和复杂的碰撞约束。集中式规划器能够协调所有机械臂，但通常面临可扩展性限制，限制了其在实时环境中的应用。另一方面，分散式方法具有可扩展性，近期基于深度学习的方法已显示出良好的效果。然而，这些方法依赖于准确的行为预测或协调协议，当其他机械臂表现出不可预测的行为时可能会失败。为了解决这些挑战，我们提出了一种基于神经哈密顿-雅可比可达性（HJR）学习的方法，以近似捕捉最坏情况的机械臂安全约束的安全值函数。我们进一步开发了一个分散式轨迹优化框架，利用学习到的HJR表示进行实时规划。所提出的方法具有可扩展性和数据效率，能够在多操纵器系统中泛化，并在具有挑战性的多臂运动规划任务中超越了最先进的基准。
cs.RO / 25 / 2607.00351
Unleashing More Actions via Action Compositional Training for VLA Models
通过动作组合训练释放更多动作以提升VLA模型的能力
Abstract
Vision-Language-Action models excel at robotic manipulation, driven by the scale and diversity of demonstration data. However, standard training paradigms often cause VLA models to severely overfit to specific behavioral patterns, rendering them unable to generalize to out-of-distribution scenarios even when those scenarios merely require novel combinations of identical sub-skills. While expanding datasets can mitigate this overfitting, acquiring high-quality robot data remains notoriously labor-intensive and cost-prohibitive. To resolve this impasse without expensive human teleoperation and to truly unleash more actions,i.e., enable VLA models to compose known sub-skills into a much broader set of executable behaviors beyond the original demonstrations-we propose ACT-VLA (Action Compositional Training for VLA Models), an offline data augmentation framework that leverages the model's latent task representations to synthesize novel, physically valid demonstrations directly from existing tasks for policy training. By eliminating additional manual data collection, our method automatically expands the training distribution and mitigates overfitting. We evaluate our approach on challenging manipulation tasks in simulation. Experiments demonstrate that while baseline VLA models generalize poorly due to original distribution overfitting, policies trained with our synthesized data achieve substantially higher success rates, validating that leveraging existing tasks for automated demonstration synthesis provides an effective, scalable, and data-efficient route to broadening VLA generalization.
Chinese Translation
视觉-语言-动作（Vision-Language-Action, VLA）模型在机器人操作中表现出色，这得益于演示数据的规模和多样性。然而，标准的训练范式往往导致VLA模型严重过拟合于特定的行为模式，使其无法在分布外场景中进行泛化，即使这些场景仅需对相同子技能进行新颖组合。虽然扩展数据集可以缓解这种过拟合，但获取高质量的机器人数据仍然是一个极具劳动强度和成本高昂的过程。为了解决这一困境，而不需要昂贵的人类遥控操作，并真正释放更多动作，即使VLA模型能够将已知子技能组合成超出原始演示的更广泛的可执行行为，我们提出了ACT-VLA（动作组合训练用于VLA模型），这是一种离线数据增强框架，利用模型的潜在任务表示直接从现有任务合成新颖且物理有效的演示以进行策略训练。通过消除额外的手动数据收集，我们的方法自动扩展了训练分布并减轻了过拟合。我们在模拟中的挑战性操作任务上评估了我们的方法。实验表明，尽管基线VLA模型由于原始分布过拟合而泛化能力较差，但使用我们合成数据训练的策略实现了显著更高的成功率，验证了利用现有任务进行自动演示合成提供了一条有效、可扩展且数据高效的途径，以拓宽VLA的泛化能力。
cs.RO / 26 / 2607.00424
Robust Operational Space Control with Conformal Disturbance Bounds for Safe Redundant Manipulation
具有符合扰动界限的鲁棒操作空间控制用于安全冗余操作
Abstract
Redundant robotic manipulators operating in constrained and human-interactive environments require accurate task-space tracking together with rigorous safety guarantees under dynamic uncertainties. Classical operational space computed torque controller (OSCTC) relies on accurate dynamic models and degrades in the presence of disturbances. In contrast, the data-driven paradigm of residual learning approximates disturbances as functions learned from full-state measurements, which are often noisy in practice, lack rigorous theoretical guarantees, and introduce additional design complexity. This paper proposes a robust OSCTC framework that integrates an extended state observer (ESO) with conformal prediction to combine model-based robustness and data-driven adaptability. The ESO estimates lumped disturbances directly in operational space without requiring full-state measurements as in residual learning, and a robust control barrier function (CBF) is constructed to enforce safety under uncertainty. However, robust CBFs require a known disturbance-variation bound to guarantee absolute safety, which often leads to conservatism in practice. To address this limitation, we further employ a sliding-window conformal prediction mechanism to estimate the bound online in a distribution-free manner, thereby achieving practical probabilistic safety guarantees. Experiments on a 7-DoF Franka Research 3 manipulator demonstrate millimeter-level tracking accuracy and real-time safe control at 1~kHz under various disturbances.
Chinese Translation
在受限和人机交互环境中操作的冗余机器人操纵器需要在动态不确定性下实现准确的任务空间跟踪以及严格的安全保障。经典的操作空间计算力矩控制器（OSCTC）依赖于准确的动态模型，并在存在扰动时性能下降。相反，基于数据的残差学习范式将扰动近似为从全状态测量中学习的函数，而这些测量在实践中通常是噪声较大的，缺乏严格的理论保障，并且引入了额外的设计复杂性。本文提出了一种鲁棒的OSCTC框架，该框架将扩展状态观测器（ESO）与符合预测相结合，以实现基于模型的鲁棒性和基于数据的适应性。ESO直接在操作空间中估计聚合扰动，而无需像残差学习那样依赖全状态测量，并构建了鲁棒控制障碍函数（CBF）以在不确定性下强制执行安全性。然而，鲁棒CBF需要已知的扰动变化界限以保证绝对安全，这在实践中往往导致保守性。为了解决这一限制，我们进一步采用滑动窗口符合预测机制在线估计界限，以无分布的方式实现实际的概率安全保障。在7自由度的Franka Research 3操纵器上的实验表明，在各种扰动下实现了毫米级的跟踪精度和1 kHz的实时安全控制。
cs.RO / 27 / 2607.00442
Learning Gait-Aware Quadruped Locomotion with Temporal Logic Specifications
基于时间逻辑规范的步态感知四足运动学习
Abstract
Reinforcement learning (RL) for quadruped locomotion commonly depends on fixed, hand-crafted, and Markovian reward functions that limit both interpretability of learned policies and lack explicit control over gait behaviors. We introduce a framework where distinct gaits are specified using parameterized constraints expressed in Signal Temporal Logic (STL). These include safety bounds, gait synchronization constraints, command tracking, and actuation bounds. From these specifications, we develop a reward shaping mechanism that provides learning agents a dense, continuous reward landscape that encodes desired behavior. We define parametric STL templates for three speed regimes (walking-trot, trot, bound), calibrate their parameters from reference rollouts, and compute rewards from using smooth approximations of STL robustness over the rollouts. The generated rewards can be used to provide shaped gradients compatible with Proximal Policy Optimization (PPO). We instantiate the approach on Google's Barkour quadruped robot in MuJoCo XLA (MJX). We use parallelization within the simulator to improve training speeds and use domain randomization to robustify learned policies. We show that compared to a baseline of hand-crafted rewards, the STL-shaped rewards yield tighter velocity tracking and more stable training. Videos can be found on our project website: https://stl-locomotion.github.io/.
Chinese Translation
四足运动的强化学习（RL）通常依赖于固定的、手工设计的和马尔可夫奖励函数，这限制了学习策略的可解释性，并缺乏对步态行为的明确控制。我们提出了一个框架，通过使用以信号时间逻辑（STL）表达的参数化约束来指定不同的步态。这些约束包括安全边界、步态同步约束、指令跟踪和驱动边界。基于这些规范，我们开发了一种奖励塑形机制，为学习代理提供一个密集的、连续的奖励空间，以编码期望的行为。我们为三种速度模式（行走-小跑、小跑、跳跃）定义了参数化的STL模板，从参考轨迹中校准其参数，并通过对轨迹使用STL鲁棒性的平滑近似来计算奖励。生成的奖励可用于提供与近端策略优化（PPO）兼容的塑形梯度。我们在谷歌的Barkour四足机器人上实现了该方法，使用MuJoCo XLA（MJX）进行实验。我们在模拟器中使用并行化来提高训练速度，并使用领域随机化来增强学习策略的鲁棒性。我们展示了与手工设计奖励的基线相比，STL塑形奖励能够实现更紧密的速度跟踪和更稳定的训练。视频可在我们的项目网站上找到：https://stl-locomotion.github.io/
cs.RO / 28 / 2607.00444
Search-Based Spatiotemporal and Multi-Robot Motion Planning on Graphs of Space-Time Convex Sets
基于搜索的时空与多机器人运动规划在时空凸集图上的应用
Abstract
Spatiotemporal motion planning, especially in multi-robot settings, requires robots to reason about collision-free regions that change over time, which is challenging in continuous spaces when feasible regions are transient and geometrically constrained. We present an algorithmic framework based on graphs of space-time convex sets (ST-GCSs), where collision-free regions are represented as convex sets in space-time and trajectories correspond to paths on the graph together with continuous motions within the selected sets. We formulate time-optimal planning on ST-GCSs as a graph-search problem over path-indexed states and develop a best-first search solver that evaluates partial paths via continuous trajectory optimization, guided by admissible heuristics and dominance checks. We further present an Exact Convex Decomposition (ECD) scheme to reserve trajectory occupancies in space-time, enabling unified handling of dynamic obstacles and multi-robot interactions. For multi-robot motion planning, we integrate ST-GCS planning and ECD into prioritized planning methods and introduce a windowed coordination scheme to improve efficiency. Extensive experiments on single-robot and multi-robot problems demonstrate substantial speedups over various planners while maintaining high solution quality, particularly in environments with narrow and transient feasible regions. Large-scale demonstrations further show that the proposed multi-robot motion planner can solve instances with up to $100$ robots within only a few minutes. Project homepage: https://sites.google.com/view/stgcs
Chinese Translation
时空运动规划，尤其是在多机器人环境中，需要机器人推理关于随时间变化的无碰撞区域，这在可行区域瞬态且几何受限的连续空间中是具有挑战性的。我们提出了一种基于时空凸集图（ST-GCSs）的算法框架，其中无碰撞区域被表示为时空中的凸集，轨迹对应于图上的路径以及在所选集内的连续运动。我们将ST-GCSs上的时间最优规划形式化为一个路径索引状态上的图搜索问题，并开发了一种最佳优先搜索求解器，通过连续轨迹优化评估部分路径，辅以可接受的启发式方法和主导性检查。我们进一步提出了一种精确凸分解（ECD）方案，以保留时空中的轨迹占用，从而统一处理动态障碍物和多机器人交互。对于多机器人运动规划，我们将ST-GCS规划和ECD集成到优先规划方法中，并引入了一种窗口协调方案以提高效率。在单机器人和多机器人问题上的大量实验表明，与各种规划器相比，所提出的方法在保持高解质量的同时显著加快了速度，特别是在具有狭窄和瞬态可行区域的环境中。大规模演示进一步表明，所提出的多机器人运动规划器可以在仅几分钟内解决多达100个机器人的实例。项目主页：https://sites.google.com/view/stgcs
cs.RO / 29 / 2607.00483
VLM-AR3L: Vision-Language Models for Absolute and Relative Rewards in Reinforcement Learning
VLM-AR3L：用于强化学习中的绝对和相对奖励的视觉-语言模型
Abstract
Designing effective reward functions remains a major challenge in reinforcement learning (RL), particularly in open-ended environments where task goals are abstract and difficult to quantify. In this work, we present VLM-AR3L, a framework that leverages Vision-Language Models (VLMs) to provide both absolute and relative rewards for RL. VLM-AR3L interprets an agent's visual observations in the context of a natural language task goal, and learns both absolute and relative rewards from VLM-generated preference labels. The absolute reward model predicts scalar evaluations for individual states, while the relative reward model compares consecutive observations to infer progress or regression toward the task goal. Their integration combines the stability of state-based evaluation with the robustness of comparative supervision. We evaluate VLM-AR3L across benchmarks spanning classic control, manipulation, and open-world embodied tasks, with a particular focus on Minecraft given its visual complexity and long-horizon decision-making requirements. Experimental results show that VLM-AR3L consistently outperforms prior VLM-based reward learning methods.
Chinese Translation
设计有效的奖励函数仍然是强化学习（RL）中的一个主要挑战，尤其是在任务目标抽象且难以量化的开放环境中。在本研究中，我们提出了VLM-AR3L，一个利用视觉-语言模型（VLMs）为强化学习提供绝对和相对奖励的框架。VLM-AR3L在自然语言任务目标的背景下解释代理的视觉观察，并从VLM生成的偏好标签中学习绝对和相对奖励。绝对奖励模型预测单个状态的标量评估，而相对奖励模型比较连续观察以推断朝向任务目标的进展或退步。它们的结合将基于状态的评估的稳定性与比较监督的鲁棒性相结合。我们在经典控制、操作和开放世界具身任务等基准上评估VLM-AR3L，特别关注Minecraft，因为其视觉复杂性和长时间决策需求。实验结果表明，VLM-AR3L在性能上始终优于之前基于VLM的奖励学习方法。
cs.RO / 30 / 2607.00530
From Technical Metrics to User Perception: A User Study of a Multimodal Human-Robot Interaction System for Object Detection and Grasping
从技术指标到用户感知：一项关于多模态人机交互系统在物体检测与抓取中的用户研究
Abstract
Improvements in the technical performance of human--robot interaction (HRI) systems do not automatically translate into differences that human users can detect during live interaction. This paper investigates whether a 15 percentage point gain in end-to-end task success (from 75% in a multimodal baseline system to 90% in an improved configuration identified through a prior ablation study) is sufficient to produce consistent and measurable differences in user perception. The baseline system combines Whisper for speech recognition, Florence-2 for open-vocabulary object detection, LLaMA 3.1 for action extraction, and an interval Type-2 fuzzy logic controller for motion execution. The improved configuration replaces the perception and language modules with Grounding DINO + SAM and Qwen 3.5 9B, respectively, while retaining the same controller. A within-subject user study with 24 participants compared both systems on the same tabletop object-grasping task. After interacting with each configuration, participants rated perceived speed, reliability, and overall competence and fluency on a 7-point Likert scale. Results show that 17 out of 24 participants (70.83%) preferred the improved system (exact binomial test, p = 0.043, h = 0.43), and all three perceptual constructs were rated significantly higher for the improved configuration after Holm correction, with large to very large effect sizes (p < 0.001). These findings confirm that the identified technical improvements are perceptible to users in direct interaction and underscore the importance of complementing benchmark evaluation with user-centred evidence when assessing robotic manipulation pipelines.
Chinese Translation
人机交互（HRI）系统的技术性能提升并不自动转化为人类用户在实时交互中能够察觉的差异。本文探讨了在端到端任务成功率上提升15个百分点（从多模态基线系统的75%提升至通过先前消融研究确定的改进配置的90%）是否足以产生一致且可测量的用户感知差异。基线系统结合了Whisper进行语音识别，Florence-2进行开放词汇物体检测，LLaMA 3.1进行动作提取，以及一个区间类型-2模糊逻辑控制器用于运动执行。改进配置用Grounding DINO + SAM和Qwen 3.5 9B分别替换了感知和语言模块，同时保留了相同的控制器。通过一项包含24名参与者的被试内用户研究，比较了两种系统在相同桌面物体抓取任务上的表现。在与每种配置交互后，参与者在7点李克特量表上对感知速度、可靠性以及整体能力和流畅性进行了评分。结果显示，24名参与者中有17名（70.83%）更倾向于改进系统（精确二项检验，p = 0.043，h = 0.43），并且在Holm校正后，所有三个感知构念在改进配置上的评分显著更高，效应量从大到非常大（p < 0.001）。这些发现确认了所识别的技术改进在直接交互中对用户是可感知的，并强调在评估机器人操作流程时，补充以用户中心的证据与基准评估的重要性。
cs.RO / 31 / 2607.00534
Learning from Demonstration via Spatiotemporal Tubes for Unknown Euler-Lagrange Systems
通过时空管学习未知的欧拉-拉格朗日系统的示范学习
Abstract
We present STT-LfD, a unified Learning from Demonstration (LfD) framework that integrates motion learning with control for unknown Euler-Lagrange systems. Unlike traditional decoupled approaches that track a fixed reference, the proposed method treats demonstrations as a data-driven safety specification. Using heteroscedastic Gaussian Processes, STT-LfD learns Spatiotemporal Tubes (STTs) as an intent envelope that capture time-varying precision requirements of a task. A closed-form feedback controller then enforces these learned constraints while respecting actuator limits, without requiring explicit system identification. The approach preserves the temporal structure of demonstrations, remains computationally efficient, and avoids explicit system identification. Hardware experiments on a mobile robot and a 7-DOF manipulator show that it outperforms baselines in robustness to disturbances and computational speed.
Chinese Translation
我们提出了STT-LfD，一个统一的示范学习（LfD）框架，将运动学习与未知的欧拉-拉格朗日系统的控制相结合。与传统的解耦方法不同，后者跟踪固定参考，所提出的方法将示范视为数据驱动的安全规范。利用异方差高斯过程，STT-LfD学习时空管（Spatiotemporal Tubes，STTs）作为捕捉任务时间变化精度要求的意图包络。然后，闭式反馈控制器在不需要明确系统识别的情况下，强制执行这些学习到的约束，同时尊重执行器的限制。该方法保留了示范的时间结构，计算效率高，并避免了显式的系统识别。在移动机器人和7自由度操纵器上的硬件实验表明，该方法在抗干扰能力和计算速度方面优于基线方法。
cs.RO / 32 / 2607.00569
[Preprint] Dynamic Modeling, Gait Synthesis, and Control of a Novel Subsurface Bore Propagator
动态建模、步态合成与新型地下钻孔推进器的控制
Abstract
In this article, we present dynamic modeling, gait synthesis, and feedback control design for a modular novel subsurface robot, designed for human-free subsurface exploration and excavation. The subsurface propagator design is based on two major aspects: 1) anchor and propel movement like an earthworm and 2) excavation similar to tunnel boring machines. This design is decoupled into five separate modules: one drill head to excavate and create cavity for propagation, two modules to anchor the robot, and two modules to enable propagation of the body. In order to design a controller for each of the modules, dynamic models using the Euler-Lagrange framework are developed. These mathematical models are used as a baseline to design controlled decoupled operation of the different joint movements. The operation of robotic assembly is constructed via a centralized state machine for gait synthesis with integration of the designed feedback controller. The controllers are tested on the real robot geometry to aid sim-to-real integration: A physics-based Unity simulation using a CAD model of the robot and integration of the trained controller via ROS verifies the performance of the robot. The experimental results demonstrate that the proposed design, controllers and the gait synthesis strategy together are capable of anchoring the robot in place and creating an total advancement of 30\,mm into the soil after completing 3 gait cycles.
Chinese Translation
在本文中，我们提出了一种模块化新型地下机器人动态建模、步态合成和反馈控制设计，该机器人旨在实现无人地下探索和挖掘。地下推进器的设计基于两个主要方面：1）像蚯蚓一样锚定和推进运动，2）类似于隧道掘进机的挖掘。该设计被解耦为五个独立模块：一个钻头用于挖掘并创建推进的空腔，两个模块用于锚定机器人，两个模块用于实现机体的推进。为了为每个模块设计控制器，使用欧拉-拉格朗日框架开发了动态模型。这些数学模型作为设计不同关节运动的受控解耦操作的基线。机器人的组装操作通过集中状态机构建，以实现步态合成，并集成所设计的反馈控制器。控制器在真实机器人几何形状上进行了测试，以帮助实现仿真与现实的结合：使用机器人CAD模型的基于物理的Unity仿真和通过ROS集成的训练控制器验证了机器人的性能。实验结果表明，所提出的设计、控制器和步态合成策略能够将机器人固定在原地，并在完成3个步态周期后在土壤中实现总共30毫米的推进。
cs.RO / 33 / 2607.00571
Enhancing Robustness in Robot-Environment Interactions through Passive Compliant Degrees of Freedom: A Hybrid Position-Force Control Approach with Feedback Linearization
通过被动柔顺自由度增强机器人与环境的交互鲁棒性：一种带反馈线性化的混合位置-力控制方法
Abstract
Robot-environment interactions in dynamic or unstructured settings are often degraded by impact shocks, vibrations, and uncertainties in contact geometry and mechanical properties. This paper proposes an interaction architecture that combines feedback-linearized hybrid position-force control with a passive compliant degree of freedom embedded at the end-effector. Unlike conventional hybrid position-force control, which relies mainly on active feedback, force sensing, and gain tuning, the proposed architecture uses a physical spring-damper interface to store and dissipate impact energy at the contact point before high-frequency shocks propagate to the actuated joints and force-control loop. The approach is evaluated in MATLAB/Simulink on a 2-DOF planar manipulator with three end-effector configurations: rigid, spring-only, and spring-damper. Results under fixed and time-varying interaction conditions show that the spring-damper configuration provides stronger attenuation of contact-induced oscillations, lower force and velocity error variance, and smoother joint-torque response. Representative reductions include 36.5% in fixed-environment tangential force-error standard deviation, 25.4% in variable-environment normal force-error standard deviation, and 41.1% in variable-environment normal velocity-error standard deviation.
Chinese Translation
在动态或非结构化环境中，机器人与环境的交互常常受到冲击、振动以及接触几何和机械特性的不确定性影响。本文提出了一种交互架构，该架构结合了反馈线性化的混合位置-力控制与嵌入在末端执行器上的被动柔顺自由度。与主要依赖主动反馈、力传感和增益调节的传统混合位置-力控制不同，所提架构利用物理弹簧-阻尼器接口在接触点存储和耗散冲击能量，从而在高频冲击传播到驱动关节和力控制回路之前进行缓冲。该方法在MATLAB/Simulink中对一个具有三种末端执行器配置（刚性、仅弹簧和弹簧-阻尼器）的2自由度平面操纵器进行了评估。在固定和时变交互条件下的结果表明，弹簧-阻尼器配置能更有效地衰减接触引起的振荡，降低力和速度误差的方差，并使关节扭矩响应更加平滑。代表性的减少包括：固定环境下切向力误差标准差降低36.5%，可变环境下法向力误差标准差降低25.4%，以及可变环境下法向速度误差标准差降低41.1%。
cs.RO / 34 / 2607.00591
From Real-Time Planning to Reliable Execution:Scalable Coordination for Heterogeneous Multi-Robot Fleets in Industrial Environments
从实时规划到可靠执行：工业环境中异构多机器人舰队的可扩展协调
Abstract
With the increasing deployment of heterogeneous robot fleets in industrial environments, efficient coordination remains a critical challenge. Real-time path planning must simultaneously accommodate high robot densities and heterogeneous motion capabilities, while communication delays, execution uncertainties, and other disturbances may cause robots to deviate from the temporal assumptions underlying planned paths. Such deviations can lead to excessive waiting and congestion propagation across the fleet. This paper presents SCALE, a reactive online coordination framework that enables real-time planning while maintaining robust execution. Within this framework, we introduce a motion-induced conflict reduction mechanism to support the online generation of feasible paths for online conflict resolution. To mitigate the effects of disturbances, we further design a generalized Conjugate Action-Precedence Hypergraph (CAPH) that adaptively adjusts precedence relations among robots. Extensive validation experiments, together with a three-day deployment in a warehouse, demonstrate the
Chinese Translation
随着异构机器人舰队在工业环境中的日益部署，高效协调仍然是一个关键挑战。实时路径规划必须同时适应高密度机器人和异构运动能力，而通信延迟、执行不确定性及其他干扰可能导致机器人偏离规划路径的时间假设。这种偏离可能导致舰队内过度等待和拥堵传播。本文提出了SCALE，一个反应式在线协调框架，能够在保持稳健执行的同时实现实时规划。在该框架内，我们引入了一种运动诱导的冲突减少机制，以支持在线生成可行路径以进行在线冲突解决。为了减轻干扰的影响，我们进一步设计了一种广义共轭动作优先超图（CAPH），能够自适应地调整机器人之间的优先关系。大量验证实验以及在仓库中的为期三天的部署展示了
cs.RO / 35 / 2607.00666
Domain Arithmetic: One-Shot VLA Adaptation under Environmental Shifts
领域算术：在环境变化下的一次性视觉-语言-动作适应
Abstract
Vision-Language-Action (VLA) models often fail to perform the same learned tasks under environmental shifts, such as changes in camera pose and shifts to a different but similar robot (e.g., from Panda to UR5e). Adapting these models to the shifted environment (i.e., target domain) often requires training on multiple demonstrations for each task, which are costly to collect. To reduce the burden of data curation and training, we propose an analogy-based method that adapts VLA models under environmental shifts through weight vector arithmetic with domain-specific information addition, named Domain ARiThmetic (DART). Unlike prior approaches, DART requires collecting only a single demonstration, enabling efficient adaptation. To accurately isolate domain-specific information for addition, DART performs subspace alignment between singular components in weight vectors to filter out noisy components. In both simulated and real-world experiments, DART outperforms existing VLA adaptation methods in one-shot scenarios across diverse visual and embodiment shifts. Code is available at https://github.com/snumprlab/dart.
Chinese Translation
视觉-语言-动作（VLA）模型在环境变化下（例如，相机姿态变化和转移到不同但相似的机器人，如从Panda到UR5e）往往无法执行相同的学习任务。将这些模型适应于变化的环境（即目标领域）通常需要对每个任务进行多次示范训练，这在收集上成本高昂。为了减少数据整理和训练的负担，我们提出了一种基于类比的方法，通过权重向量算术与领域特定信息的添加来适应环境变化下的VLA模型，称为领域算术（Domain ARiThmetic，DART）。与之前的方法不同，DART只需收集一次示范，从而实现高效适应。为了准确隔离用于添加的领域特定信息，DART在权重向量的奇异成分之间执行子空间对齐，以过滤掉噪声成分。在模拟和真实世界实验中，DART在多样化的视觉和体现变化的一次性场景中优于现有的VLA适应方法。代码可在 https://github.com/snumprlab/dart 获取。
cs.RO / 36 / 2607.00673
Path Planning in Physically Viable World Models
物理可行世界模型中的路径规划
Abstract
Robots deployed in unstructured outdoor environments often plan from scene reconstructions collected before deployment because operators cannot remap large or remote sites before every mission. As a result, robots must make long-horizon planning decisions using stale maps that assume the terrain remains unchanged, even though physical changes to the environment may render previously feasible routes unsafe or unreachable at execution time. We present a physically viable world model for evaluating what-if queries for robot navigation under future terrain change. The system augments reconstructed 3D Gaussian splat scenes with physics-based simulation to generate physically modified versions of the same environment without recollecting sensor data or rebuilding the map. We then implement a terrain-aware planner that accounts for physical events, obstacles, and deformations that are simulated by the world model. This allows robots and human operators to evaluate whether planned routes remain feasible before committing to a planned route, particularly in constrained environments where retreat or recovery may become impossible once conditions change. We evaluate the system on a real outdoor field site in Central Texas using simulated flooding across multiple severity levels. We measure route and mission feasibility as terrain conditions deteriorate under physically simulated interventions. Our results show that physically viable world models expose long-horizon route failures and rerouting behavior that are not apparent when planning only on the original reconstructed environment, allowing robots to evaluate how future terrain changes may affect route feasibility before deployment.
Chinese Translation
在非结构化的户外环境中部署的机器人通常依赖于在部署前收集的场景重建进行规划，因为操作员无法在每次任务前重新绘制大型或偏远的地点。因此，机器人必须使用过时的地图进行长期规划决策，这些地图假设地形保持不变，尽管环境的物理变化可能使得之前可行的路线在执行时变得不安全或无法到达。我们提出了一种物理可行的世界模型，用于评估机器人在未来地形变化下的导航‘假设’查询。该系统通过基于物理的仿真增强重建的3D高斯点云场景，以生成相同环境的物理修改版本，而无需重新收集传感器数据或重建地图。随后，我们实现了一种考虑物理事件、障碍物和由世界模型模拟的变形的地形感知规划器。这使得机器人和人类操作员能够在承诺执行计划路线之前评估计划路线是否仍然可行，特别是在一旦条件变化可能导致撤退或恢复变得不可能的受限环境中。我们在德克萨斯州中部的一个真实户外现场评估了该系统，使用模拟洪水跨越多个严重程度级别。我们测量了在物理模拟干预下，随着地形条件恶化，路线和任务的可行性。我们的结果表明，物理可行的世界模型揭示了在仅基于原始重建环境进行规划时并不明显的长期路线失败和重新规划行为，使机器人能够在部署前评估未来地形变化可能对路线可行性产生的影响。
cs.RO / 37 / 2607.00776
From Prediction Uncertainty to Conformalized Distance Fields for Safe Motion Planning
从预测不确定性到符合化距离场的安全运动规划
Abstract
Safe motion planning in dynamic environments requires reasoning about the uncertainty in predicted obstacle motion without sacrificing real-time performance. Existing conformal approaches conformalize a scalar score that aggregates per-obstacle prediction errors, losing spatial coherence and scaling poorly with scene density. We instead conformalize the entire predicted distance field at once. This functional conformal prediction (FCP) framework yields a distribution-free, field-level lower bound, from which safety follows uniformly: any trajectory satisfying the resulting constraint is certified safe, independent of how the control space is sampled. The key enabler is that the residual distance field is empirically low-rank and approximately time-invariant, which makes the bound decomposable in coefficient space. An envelope is fitted offline via functional PCA and a Gaussian-mixture inductive conformal procedure, then refined online by a lightweight adaptive functional conformal (AFCP) update on a low-dimensional vector. This keeps the per-step cost largely insensitive to obstacle count and retains long-run field coverage under distribution shift. We embed the envelope as a tightened safety constraint in a sampling-based model predictive controller, FCP-MPC. On the ETH--UCY pedestrian benchmarks and a dense 3D quadrotor task with up to 280 dynamic obstacles, FCP-MPC attains a favorable balance of safety, feasibility, and efficiency, reaching goals where pointwise and egocentric conformal baselines become too conservative or too expensive, while keeping per-step computation far below online uncertainty-reasoning baselines.
Chinese Translation
在动态环境中，安全运动规划需要对预测障碍物运动的不确定性进行推理，而不牺牲实时性能。现有的符合化方法将聚合每个障碍物预测误差的标量分数进行符合化，导致空间一致性丧失，并且在场景密度较高时扩展性差。我们则一次性对整个预测距离场进行符合化。该功能符合预测（Functional Conformal Prediction, FCP）框架提供了一个无分布的、场级的下界，从中可以均匀地推导出安全性：任何满足所得到约束的轨迹都被认证为安全，与控制空间的采样方式无关。关键在于残差距离场在经验上具有低秩性，并且近似时间不变，这使得该下界在系数空间中可分解。通过功能主成分分析（Functional PCA）和高斯混合诱导符合程序离线拟合一个包络，然后通过在低维向量上的轻量级自适应功能符合（Adaptive Functional Conformal, AFCP）更新在线精炼。这使得每一步的成本在很大程度上对障碍物数量不敏感，并在分布变化下保持长期场覆盖。我们将包络嵌入为基于采样的模型预测控制器FCP-MPC中的收紧安全约束。在ETH-UCY行人基准测试和一个包含多达280个动态障碍物的密集3D四旋翼任务中，FCP-MPC实现了安全性、可行性和效率的良好平衡，达成了目标，而逐点和自我中心的符合基线则变得过于保守或成本过高，同时每一步的计算远低于在线不确定性推理基线。
cs.RO / 38 / 2607.00836
From World Models to World Action Models: A Concise Tutorial for Robotics
从世界模型到世界行动模型：机器人技术的简明教程
Abstract
World models are increasingly used in embodied intelligence and generative simulation, yet their scope remains ambiguous across communities. This tutorial presents a design-space view of world models as action-conditioned predictive models that estimate the future evolution of task-relevant observations or states. We categorize existing methods into observation-space and state-space world models, comparing their trade-offs in visual fidelity, spatial structure, physical interpretability, and control usability. We further introduce world action models, which connect predicted futures with executable robot actions, and summarize four representative paradigms: imagine-then-execute, video-feature-conditioned action prediction, joint video-action modeling, and auxiliary video prediction for policy learning. The goal of this tutorial is to clarify the conceptual scope of world (action) models and provide a structured taxonomy for embodied prediction and control.
Chinese Translation
世界模型在具身智能和生成模拟中越来越多地被使用，但其在不同领域中的范围仍然模糊。本文教程以设计空间的视角呈现世界模型，作为基于行动的预测模型，旨在估计与任务相关的观察或状态的未来演变。我们将现有方法分为观察空间和状态空间世界模型，比较它们在视觉逼真度、空间结构、物理可解释性和控制可用性方面的权衡。我们进一步介绍世界行动模型，它将预测的未来与可执行的机器人动作连接起来，并总结了四种代表性范式：先想象后执行（imagine-then-execute）、视频特征条件下的动作预测（video-feature-conditioned action prediction）、联合视频-动作建模（joint video-action modeling）以及用于策略学习的辅助视频预测（auxiliary video prediction for policy learning）。本教程的目标是澄清世界（行动）模型的概念范围，并为具身预测和控制提供一个结构化的分类法。
cs.RO / 39 / 2607.00874
Beyond Line of Sight: Hybrid Validation of V2X Collective Perception in Complex Scenarios
超越视距：复杂场景中V2X集体感知的混合验证
Abstract
This paper introduces a probabilistic framework and hybrid validation methodology for V2X-enabled Collective Perception (CP) in complex traffic scenarios. The proposed Bayesian fusion algorithm extends the perceptual horizon of connected and autonomous vehicles by integrating heterogeneous sensor observations from multiple agents into a shared probabilistic occupancy grid. Each cell of this grid encapsulates both occupancy likelihood and uncertainty, enabling explainable and trustworthy situational awareness beyond the ego vehicle's field of view. To bridge the gap between simulation and real-world evaluation, a hybrid testing framework is developed, combining CARLA-based virtual environments with vehicle-in-the-loop experimentation. Experimental results in a roundabout scenario demonstrate a 260 percent increase in field-of-view coverage and a rise in occupied-cell recall from 0.82 (ego-only) to 0.94 (six-agent CP) under nominal localization conditions. Overall, the proposed approach provides a reproducible and interpretable foundation for validating CP systems, supporting the safe and certifiable deployment of cooperative autonomous vehicles.
Chinese Translation
本文介绍了一种用于复杂交通场景中V2X支持的集体感知（Collective Perception, CP）的概率框架和混合验证方法。所提出的贝叶斯融合算法通过将多个智能体的异构传感器观测整合到一个共享的概率占用网格中，扩展了连接和自主车辆的感知视野。该网格的每个单元格都封装了占用可能性和不确定性，从而使得超出自我车辆视野的可解释和可信的情境感知成为可能。为了弥合模拟与现实评估之间的差距，开发了一种混合测试框架，结合了基于CARLA的虚拟环境与车辆在环实验。在环形交叉口场景中的实验结果表明，在名义定位条件下，视野覆盖率提高了260%，占用单元召回率从0.82（仅自我车辆）上升至0.94（六个智能体的CP）。总体而言，所提出的方法为验证CP系统提供了可重复和可解释的基础，支持合作自主车辆的安全和可认证部署。
cs.RO / 40 / 2607.01029
AMBUSH: Collaborative Capture in Complex Environments with Neural Acceleration
AMBUSH：在复杂环境中利用神经加速进行协同捕获
Abstract
Collaborative capture of dynamic targets is common in nature as an essential strategy for weaker species against the strong. Similar concepts have shown to be useful for numerous robotic applications, such as security and surveillance, search and rescue. However, most existing works focus on analytical and geometric solutions or end-to-end reinforcement learning methods, which are largely constrained to obstacle-free environments or scenarios with sparse, regularly distributed obstacles. This work tackles the problem from a unique perspective: the renowned strategy of``ambush'' alone would suffice for multiple slower pursuers to capture one faster evader with different levels of intelligence efficiently in complex environments. A parameterized strategy of ambush (including discrete and continuous parameters) is designed first, which takes into account the topological properties of the workspace, the truncated line-of-sight visibility, the relative speed ratio and the limited capture range. Then, a Hybrid Monte Carlo Tree Search (H-MCTS) algorithm is proposed to optimize the associated parameters through long-term planning, enabling the identification of highly promising parameters for future capture. Lastly, the neural acceleration is trained offline to learn the ranking of different choices of parameters across various environments, and to directly predict scores, replacing the rollout process in H-MCTS. The neural acceleration is adopted during online H-MCTS to accelerate the planning procedure while guaranteeing the planning quality. Its efficiency and effectiveness are validated in extensive simulations and hardware experiments, against evaders with different capabilities and intelligence levels, including two-times higher velocity and human-controlled behavior.
Chinese Translation
动态目标的协同捕获在自然界中是弱势物种对抗强势物种的一种基本策略。类似的概念已被证明对许多机器人应用（如安全监控、搜索与救援）非常有用。然而，大多数现有研究集中于分析和几何解决方案或端到端强化学习方法，这些方法在很大程度上局限于无障碍环境或障碍物稀疏且规则分布的场景。本研究从一个独特的视角解决了这一问题：著名的“埋伏”策略足以使多个较慢的追捕者在复杂环境中高效捕获一个更快的逃避者，且追捕者具有不同的智能水平。首先设计了一种参数化的埋伏策略（包括离散和连续参数），考虑了工作空间的拓扑特性、截断的视距可见性、相对速度比和有限的捕获范围。然后，提出了一种混合蒙特卡洛树搜索（Hybrid Monte Carlo Tree Search, H-MCTS）算法，通过长期规划优化相关参数，从而识别未来捕获的高潜力参数。最后，离线训练神经加速器，以学习不同环境中各种参数选择的排名，并直接预测评分，从而替代H-MCTS中的回滚过程。在在线H-MCTS中采用神经加速器，以加速规划过程，同时保证规划质量。其效率和有效性在广泛的仿真和硬件实验中得到了验证，针对具有不同能力和智能水平的逃避者，包括速度是其两倍的情况以及人控行为。
cs.RO / 41 / 2607.01043
DART-VLN: Test-Time Memory Decay and Anti-Loop Regularization for Discrete Vision-Language Navigation
DART-VLN：离散视觉-语言导航中的测试时记忆衰减与反循环正则化
Abstract
Memory-based discrete vision-language navigation (VLN) agents must act under partial observability, yet even strong frozen backbones remain vulnerable at test time. Two common failure modes are stale historical evidence at memory readout and inefficient local backtracking during action selection. We present DART-VLN, a training-free test-time control framework for discrete VLN. DART-VLN combines Test-Time Memory Decay, a read-side memory reweighting rule that suppresses stale and redundant evidence without rewriting stored content, with Anti-Loop Regularization, a lightweight next-hop penalty that discourages immediate reversals during action selection. The framework introduces no new learnable parameters and leaves the learned backbone unchanged. Experiments on R2R and REVERIE show a consistent pattern: decay-only provides stable read-side gains, while decay+anti-loop achieves the best overall quality-efficiency trade-off, yielding shorter trajectories, lower runtime, and improved navigation performance in key settings. Behavioral analysis further confirms that anti-loop regularization reduces local backtracking and improves path efficiency under frozen backbones. Overall, the results show that modest test-time control can make memory-based discrete VLN more reliable and efficient without retraining.
Chinese Translation
基于记忆的离散视觉-语言导航（VLN）代理必须在部分可观察性下进行操作，但即使是强大的冻结骨干网络在测试时仍然容易受到影响。两种常见的失败模式是记忆读取时的过时历史证据和在动作选择过程中低效的局部回溯。我们提出了DART-VLN，这是一种针对离散VLN的无训练测试时控制框架。DART-VLN结合了测试时记忆衰减（Test-Time Memory Decay），这是一种读取侧记忆重加权规则，能够在不重写存储内容的情况下抑制过时和冗余的证据，以及反循环正则化（Anti-Loop Regularization），这是一种轻量级的下一步惩罚，旨在减少动作选择过程中的即时反转。该框架不引入新的可学习参数，并保持学习到的骨干网络不变。在R2R和REVERIE上的实验显示出一致的模式：仅使用衰减提供了稳定的读取侧增益，而衰减+反循环则实现了最佳的整体质量-效率权衡，产生了更短的轨迹、更低的运行时间，并在关键设置中改善了导航性能。行为分析进一步确认反循环正则化减少了局部回溯，并在冻结骨干网络下提高了路径效率。总体而言，结果表明适度的测试时控制可以使基于记忆的离散VLN在不重新训练的情况下变得更加可靠和高效。
cs.RO / 42 / 2607.01044
Robots Ask the Way: Communication-Enabled Social Navigation
机器人问路：支持沟通的社会导航
Abstract
Assistive autonomous robots operating in multi-agent environments require efficient strategies to locate specific individuals among multiple residents. Current social navigation methods focus on reactive collision avoidance and trajectory adaptation, but lack mechanisms to proactively gather information through human-robot communication. We introduce Communication-enabled Social Navigation (CommNav). In this novel task, robotic agents actively seek assistance from residents to locate target individuals by requesting information about recent sightings, locations, and movements. To evaluate CommNav, we extend Habitat 3.0 to create Habitat 3.0c, a communication-enabled variant supporting multi-human environments with information exchange protocols. Adding our communication module (COMM) to a state-of-the-art social navigation model yields a 10 percentage-point improvement in Episode Success. We further investigate the transition from structured data to natural language by evaluating models trained on LLM-generated instructions and on colloquial instructions collected from a human study. Our experiments reveal that: (i) explicit human-robot communication substantially enhances multi-person navigation performance; (ii) pre-training COMM on a communication pretext task effectively addresses the challenge of occasional interaction signals; and (iii) the navigation policy is highly robust to natural, colloquial human language, achieving an episode success statistically similar to the model using perfect structured data.
Chinese Translation
在多智能体环境中运行的辅助自主机器人需要有效的策略来在多个居民中定位特定个体。目前的社会导航方法侧重于反应式的碰撞避免和轨迹适应，但缺乏通过人机沟通主动收集信息的机制。我们提出了支持沟通的社会导航（Communication-enabled Social Navigation，CommNav）。在这一新任务中，机器人代理主动向居民寻求帮助，通过请求有关近期目击、位置和移动的信息来定位目标个体。为了评估CommNav，我们扩展了Habitat 3.0，创建了Habitat 3.0c，这是一个支持信息交换协议的沟通增强变体，适用于多人的环境。将我们的沟通模块（COMM）添加到最先进的社会导航模型中，使得情节成功率提高了10个百分点。我们进一步通过评估在大型语言模型（LLM）生成的指令和从人类研究中收集的口语指令上训练的模型，研究从结构化数据到自然语言的转变。我们的实验表明：（i）明确的人机沟通显著提升了多人的导航性能；（ii）在沟通前置任务上对COMM进行预训练有效地解决了偶尔交互信号的挑战；（iii）导航策略对自然的、口语化的人类语言具有高度的鲁棒性，其情节成功率在统计上与使用完美结构化数据的模型相似。
cs.RO / 43 / 2607.01051
AutoSpeed: Annotation-Free Stage-Adaptive Motion Speed Learning for Robot Manipulation
AutoSpeed：无注释阶段自适应运动速度学习用于机器人操作
Abstract
Different stages of manipulation tasks exhibit varying levels of difficulty, suggesting stage-dependent motion speeds and temporal prediction horizons. However, existing IL-based visuomotor policies typically imitate the execution speed of expert demonstrations and operate with a fixed temporal prediction horizon, limiting flexibility and overall task throughput. In this paper, we introduce AutoSpeed, a model-agnostic learning framework that enables existing visuomotor policies to predict trajectories with stage-adaptive motion speeds, without requiring speed or stage annotations. We treat future trajectories at different speeds as candidate optimization targets, evaluate each candidate using a composite cost that trades off prediction error against prediction horizon, and optimize the policy toward the minimum-cost candidate. With a fixed-length action sequence, speed modulation adjusts the effective temporal prediction horizon: simple stages are executed faster with a longer prediction horizon, whereas complex stages are executed more slowly with a shorter prediction horizon. Specifically, we implement speed modulation in the frequency domain via the discrete cosine transform (DCT), which enables smooth, non-integer speed scaling and thus preserves motion continuity. Extensive evaluations show that AutoSpeed substantially reduces task execution time while also improving success rates. Under the AutoSpeed framework, the inferred motion speeds exhibit a strong correspondence with task stages.
Chinese Translation
不同阶段的操作任务表现出不同的难度水平，这表明运动速度和时间预测范围依赖于阶段。然而，现有的基于逆强化学习（IL）的视觉运动策略通常模仿专家示范的执行速度，并且在固定的时间预测范围内操作，这限制了灵活性和整体任务吞吐量。在本文中，我们介绍了AutoSpeed，一个模型无关的学习框架，使现有的视觉运动策略能够在不需要速度或阶段注释的情况下预测具有阶段自适应运动速度的轨迹。我们将不同速度下的未来轨迹视为候选优化目标，使用一种复合成本评估每个候选，权衡预测误差与预测范围，并优化策略以达到最低成本的候选。通过固定长度的动作序列，速度调节调整有效的时间预测范围：简单阶段以更长的预测范围更快地执行，而复杂阶段则以更短的预测范围更慢地执行。具体而言，我们通过离散余弦变换（DCT）在频域中实现速度调节，这使得平滑的非整数速度缩放成为可能，从而保持运动的连续性。大量评估表明，AutoSpeed显著减少了任务执行时间，同时提高了成功率。在AutoSpeed框架下，推断的运动速度与任务阶段之间表现出强烈的对应关系。
cs.RO / 44 / 2607.01060
RoboWorld: Fast and Reliable Neural Simulators for Generalist Robot Policy Evaluation
RoboWorld：快速可靠的神经模拟器用于通用机器人策略评估
Abstract
Video world models are emerging as a scalable alternative for evaluating generalist robot policies, bypassing the physical constraints and engineering burdens of real-world deployment. However, evaluating policies with video world models remains challenging, as world-model errors can make generated rollouts unreliable and slow inference limits large-scale throughput. We introduce RoboWorld, an automated evaluation pipeline that pairs a fast autoregressive video world model with a task-progress-aware vision-language model scoring. To enable reliable long-horizon autoregressive world-model rollouts, we propose Step Forcing, which combines anchored and one-step self-forwarded contexts to reduce train--test mismatch while preserving action--observation dynamics. Together, these components enable RoboWorld to align strongly with real-world robot evaluation across tasks and environments, achieving Pearson's r = 0.989 and Spearman's \r{ho} = 0.970.
Chinese Translation
视频世界模型作为评估通用机器人策略的一种可扩展替代方案，正在逐渐兴起，能够绕过现实世界部署中的物理限制和工程负担。然而，使用视频世界模型评估策略仍然具有挑战性，因为世界模型的误差可能导致生成的轨迹不可靠，而缓慢的推理速度限制了大规模的吞吐量。我们介绍了RoboWorld，一个自动化评估管道，将快速自回归视频世界模型与任务进度感知的视觉-语言模型评分相结合。为了实现可靠的长时间自回归世界模型轨迹生成，我们提出了步骤强制（Step Forcing），该方法结合了锚定上下文和一步自前馈上下文，以减少训练与测试之间的不匹配，同时保持动作与观察动态的一致性。这些组件共同使RoboWorld能够在任务和环境中与现实世界的机器人评估高度一致，达到了Pearson相关系数r = 0.989和Spearman等级相关系数
{ho} = 0.970。
cs.RO / 45 / 2607.01067
Human-Centric Transferable Tactile Pre-Training for Dexterous Robotic Manipulation
以人为中心的可转移触觉预训练用于灵巧机器人操作
Abstract
As an essential modality for dexterous and contact-rich tasks, tactile sensing provides precise force feedback that cannot be reliably inferred from vision. However, limited by hardware and data collection systems, existing datasets with tactility remain small in scale and narrow in contact coverage. Meanwhile, Vision-Language-Action (VLA) models with tactile modality are constrained on dynamics-agnostic post-training, which limits the performance ceiling on downstream tasks. In this paper, we present H-Tac, a large-scale tactile-action dataset with 160-hour egocentric human videos containing more than 300 tasks and 135k episodes. Building upon this, we propose Transferable Tactile Pre-Training (TTP), a system of tactile-based pre-training on human data for fine-grained robotic tasks. To bridge the gap between humans and robots, we use unified tactile and action spaces throughout the pre-training and post-training phases, preserving prior knowledge during human-to-robot transfer. By leveraging a tactile expert for future tactile prediction, our framework explicitly models the contact dynamics and precise physical interactions. Extensive experiments in simulation and on real robots demonstrate that our model achieves superior performance, exhibiting robust generalization and fine-grained manipulation capabilities. TTP paves the way for scalable tactile pre-training via human-to-robot transfer.
Chinese Translation
触觉感知作为灵巧和接触丰富任务的重要模态，提供了无法通过视觉可靠推断的精确力反馈。然而，由于硬件和数据收集系统的限制，现有的触觉数据集规模较小且接触覆盖面狭窄。同时，带有触觉模态的视觉-语言-动作（VLA）模型在动态无关的后训练上受到限制，这限制了下游任务的性能上限。本文提出了H-Tac，这是一个大规模的触觉-动作数据集，包含160小时的自我中心人类视频，涵盖超过300个任务和135,000个情节。在此基础上，我们提出了可转移触觉预训练（TTP），这是一个基于人类数据的触觉预训练系统，旨在处理细粒度的机器人任务。为了弥合人类与机器人之间的差距，我们在预训练和后训练阶段使用统一的触觉和动作空间，保留人类到机器人转移过程中的先前知识。通过利用触觉专家进行未来的触觉预测，我们的框架明确建模了接触动态和精确的物理交互。在仿真和真实机器人上的大量实验表明，我们的模型实现了卓越的性能，展现出强大的泛化能力和细粒度的操作能力。TTP为通过人类到机器人转移实现可扩展的触觉预训练铺平了道路。
cs.RO / 46 / 2607.01079
Where Am I? Semantic Map Grounding via Vision-Language Models for Multi-Modal Localization
我在哪里？通过视觉-语言模型进行语义地图定位的多模态定位
Abstract
We address robot localization in GPS-denied indoor environments by reframing it as a semantic reasoning task rather than a geometric estimation problem. Motivated by how humans localize using object-level cues and labeled maps, we ask whether a vision-language model, given a front camera image, a polar LiDAR scan, and a top-down semantic grid map, can infer the robot pose. We fine-tune Qwen2.5-VL-7B with LoRA and attach a lightweight regression head that predicts continuous pose coordinates (x, y, theta) directly from the final hidden state, bypassing text generation. Training uses a composite position-and-direction loss with curriculum learning on a custom Gazebo dataset of 120,112 samples and 527 scenes. On the in-distribution test set of 18,017 samples, the model achieves 98.23 percent position accuracy, 98.00 percent direction accuracy, 96.75 percent full pose accuracy, a mean position error of 0.11 m, and a mean orientation error of 5.7 degrees at 0.62 s per sample. Position accuracy drops by only 7.2 percentage points on seven unseen object categories, reaching 90.99 percent, supporting semantic spatial reasoning rather than appearance memorization. With incomplete maps, fine-tuning recovers performance to 93.72 percent position accuracy, showing adaptability to stale or partial map information. Two ablations highlight cross-modal complementarity. Without LiDAR, using only camera and map inputs, position accuracy remains 95.06 percent, only 3.2 percentage points below the full system. However, when the camera sees no visible objects in a wall-facing view, LiDAR sustains 92.33 percent position accuracy, compared with 70.74 percent when neither LiDAR nor visible objects are available. This shows that LiDAR becomes the primary localization signal when camera semantics are unavailable and provides a reliable fallback under occlusion or sparse layouts.
Chinese Translation
我们将机器人在无GPS的室内环境中的定位问题重新定义为一个语义推理任务，而非几何估计问题。受到人类如何利用物体级线索和标注地图进行定位的启发，我们提出一个问题：给定前置摄像头图像、极坐标激光雷达扫描和自上而下的语义网格地图，视觉-语言模型能否推断出机器人姿态。我们对Qwen2.5-VL-7B进行了LoRA微调，并附加了一个轻量级回归头，直接从最终隐状态预测连续的姿态坐标（x，y，theta），绕过文本生成。训练使用了复合位置和方向损失，并在一个包含120,112个样本和527个场景的自定义Gazebo数据集上进行了课程学习。在包含18,017个样本的分布内测试集上，该模型实现了98.23%的位置准确率、98.00%的方向准确率、96.75%的完整姿态准确率，平均位置误差为0.11米，平均方向误差为5.7度，每个样本耗时0.62秒。在七个未见物体类别上，位置准确率仅下降7.2个百分点，达到90.99%，支持语义空间推理而非外观记忆。在不完整地图的情况下，微调使得性能恢复到93.72%的位置准确率，显示出对过时或部分地图信息的适应性。两个消融实验突显了跨模态的互补性。在没有激光雷达的情况下，仅使用摄像头和地图输入，位置准确率仍保持在95.06%，仅比完整系统低3.2个百分点。然而，当摄像头在面对墙壁的视图中看不到可见物体时，激光雷达的定位准确率维持在92.33%，而在激光雷达和可见物体均不可用时，准确率仅为70.74%。这表明，当摄像头语义不可用时，激光雷达成为主要的定位信号，并在遮挡或稀疏布局下提供可靠的后备支持。
cs.RO / 47 / 2607.01088
ROSA: A Robotics Foundation Model Serving System for Robot Factories
ROSA：一个为机器人工厂服务的机器人基础模型服务系统
Abstract
Robotics foundation models (RFMs) are making general-purpose robots increasingly practical for factory deployments. While RFM serving systems are central to this vision, existing systems are largely shaped by a single-robot, single-model assumption: inference is treated as an edge-computing problem handled by an on-robot or dedicated nearby GPU, and the serving objective is to minimize the latency of a single action model. In this paper, we propose ROSA, an RFM serving system for robot factories designed around three key principles. First, ROSA adopts shared GPU-pool serving, allowing a fleet of robots to access powerful server-class GPUs over the network in order to improve inference performance, battery duration, and GPU utilization. Second, ROSA provides a robotics-aware programming abstraction and system design that supports multi-model pipelines, per-task performance requirements, and failure handling. Third, ROSA uses factory-objective-driven scheduling to maximize SLO-qualified factory productivity rather than minimizing individual request latency. We implement ROSA on top of Ray Serve for distributed orchestration, with vLLM, PyTorch, and JAX as model-serving backends, and evaluate it on both real robots and synthetic large-scale workloads. The results show that ROSA improves factory productivity by up to 12.06x over conventional dedicated serving systems.
Chinese Translation
机器人基础模型（RFM）正在使通用机器人在工厂部署中变得越来越实用。虽然RFM服务系统是这一愿景的核心，但现有系统在很大程度上受到单机器人、单模型假设的影响：推理被视为一个边缘计算问题，由机器人或专用的近旁GPU处理，服务目标是最小化单个动作模型的延迟。本文提出了ROSA，一个为机器人工厂设计的RFM服务系统，围绕三个关键原则构建。首先，ROSA采用共享GPU池服务，使一组机器人能够通过网络访问强大的服务器级GPU，从而提高推理性能、电池续航和GPU利用率。其次，ROSA提供了一种机器人感知的编程抽象和系统设计，支持多模型管道、每个任务的性能要求和故障处理。第三，ROSA使用以工厂目标驱动的调度，最大化符合服务水平目标（SLO）的工厂生产力，而不是最小化单个请求的延迟。我们在Ray Serve之上实现了ROSA以进行分布式编排，使用vLLM、PyTorch和JAX作为模型服务后端，并在真实机器人和合成大规模工作负载上进行了评估。结果表明，ROSA在生产力上比传统的专用服务系统提高了多达12.06倍。
cs.RO / 48 / 2607.01106
Technical Report: Asynchronous Distributed Trajectory Estimation of Multi-Robot Systems
技术报告：多机器人系统的异步分布式轨迹估计
Abstract
Distributed trajectory estimation arises in many applications across robotics, but existing implementations typically do not consider asynchrony in agents' communications and computations. Therefore, we propose an asynchronous block coordinate descent algorithm for distributed trajectory estimation. We consider a team of agents that observes a team of robots and estimates their states over a sliding window. The agents solve an approximation of the maximum a posteriori estimation problem, which we derive. We show this approximation introduces negligible errors and eliminates up to 96.9% of communications among agents. Next, we prove that agents' iterates converge exponentially fast to the optimal estimate of the robots' states. Simulations show that this approach has up to 64% less error than a comparable state-of-the-art algorithm. Experiments on mobile robots show the robustness of this approach to delays whose lengths span three orders of magnitude.
Chinese Translation
分布式轨迹估计在机器人技术的许多应用中出现，但现有的实现通常未考虑代理之间通信和计算的异步性。因此，我们提出了一种用于分布式轨迹估计的异步块坐标下降算法。我们考虑一个代理团队观察一组机器人，并在滑动窗口内估计它们的状态。代理们解决一个我们推导出的最大后验估计问题的近似。我们表明，这种近似引入的误差可以忽略不计，并消除了多达96.9%的代理间通信。接下来，我们证明代理的迭代以指数速度收敛到机器人状态的最优估计。仿真结果表明，该方法的误差比一种可比的最先进算法低多达64%。在移动机器人上的实验显示了该方法对跨越三个数量级的延迟的鲁棒性。
cs.RO / 49 / 2607.01111
FAR: Failure-Aware Retry for Test-Time Recovery and Continual Policy Improvement
FAR：面向失败的重试机制用于测试时恢复和持续策略改进
Abstract
Robot policies inevitably encounter failures when deployed in real environments. Naive retries often repeat the same mistakes, while many existing recovery methods rely on human intervention. In this paper, we propose Failure-Aware Retry (FAR), a framework that enables robots to learn from previous failures at test time, adapt their behavior accordingly, and eventually complete the task autonomously. FAR combines Failure-Contrastive Preference Adaptation, which constructs preference learning data from failures to steer the policy away from previously unsuccessful behaviors, with lightweight action perturbations during retries to encourage local exploration. We further incorporate successful recovery trajectories into a training loop for continual policy improvement. Experiments in both simulation and real-world manipulation tasks show that FAR substantially improves success rates and robustness, with average gains of 17.6% over the standard diffusion policy in simulation and 11.7% in the real world. In addition, FAR significantly improves data efficiency under both reset and timestep budgets during continual policy improvement by exploiting informative failure cases.
Chinese Translation
机器人策略在真实环境中部署时不可避免地会遇到失败。简单的重试往往会重复相同的错误，而许多现有的恢复方法依赖于人工干预。本文提出了一种面向失败的重试机制（Failure-Aware Retry, FAR），该框架使机器人能够在测试时从以往的失败中学习，相应地调整其行为，并最终自主完成任务。FAR结合了失败对比偏好适应（Failure-Contrastive Preference Adaptation），该方法通过失败构建偏好学习数据，以引导策略远离以前不成功的行为，并在重试期间进行轻量级的动作扰动以鼓励局部探索。我们进一步将成功的恢复轨迹纳入训练循环，以实现持续的策略改进。在模拟和真实世界的操作任务中的实验表明，FAR显著提高了成功率和鲁棒性，在模拟中相比标准扩散策略平均提升了17.6%，在真实世界中提升了11.7%。此外，FAR通过利用信息丰富的失败案例，在持续策略改进过程中显著提高了在重置和时间步预算下的数据效率。
cs.RO / 50 / 2607.01166
Structured 4D Latent Predictive Model for Robot Planning
结构化4D潜在预测模型用于机器人规划
Abstract
Video predictive models are emerging as a powerful paradigm in robotics, offering a promising path toward task generalization, long-horizon planning, and flexible decision-making. However, prevailing approaches often operate on 2D video sequences, inherently lacking the 3D geometric understanding necessary for precise spatial reasoning and physical consistency. We introduce a Structured 4D Latent Predictive Model, which predicts the evolution of a scene's 3D structure in a structured latent space conditioned on observations and textual instructions. Our representation encodes the scene holistically and can be decoded into diverse 3D formats, enabling a more complete and 3D consistent scene understanding. This structured 4D latent predictive model serves as a planner, generating future scenes that are translated into executable actions by a goal-conditioned inverse dynamics module. Experiments demonstrate that our model generates futures with strong visual quality, substantially better 3D consistency and multi-view coherence compared to state-of-the-art video-based planners. Consequently, our full planning pipeline achieves superior performance on complex manipulation tasks, exhibits robust generalization to novel visual conditions, and proves effective on real-world robotic platforms. Our website is available at https://structured-4d-model.github.io/.
Chinese Translation
视频预测模型在机器人领域逐渐成为一种强大的范式，为任务泛化、长时间规划和灵活决策提供了有希望的路径。然而，现有的方法通常基于2D视频序列，固有地缺乏进行精确空间推理和物理一致性所需的3D几何理解。我们提出了一种结构化4D潜在预测模型，该模型在一个结构化的潜在空间中，根据观察和文本指令预测场景3D结构的演变。我们的表示方法整体编码场景，并可以解码为多种3D格式，从而实现更全面和3D一致的场景理解。该结构化4D潜在预测模型作为规划器，生成未来场景，并通过目标条件的逆动力学模块将其转化为可执行的动作。实验表明，我们的模型生成的未来场景在视觉质量上表现出色，相较于最先进的视频基础规划器具有显著更好的3D一致性和多视角一致性。因此，我们的完整规划流程在复杂的操作任务上实现了卓越的性能，展现出对新视觉条件的强大泛化能力，并在真实世界的机器人平台上证明了其有效性。我们的官方网站为 https://structured-4d-model.github.io/。
cs.RO / 51 / 2607.01200
FastBridge: Closing the Model-Based Realization Gap in Safety Filters on 3D Gaussian Splatting for Fast Quadrotor Flight
FastBridge：在3D高斯溅射中缩小基于模型的安全过滤器实现差距以实现快速四旋翼飞行
Abstract
Fast quadrotor flight requires safe obstacle avoidance under tight onboard compute limits. While 3D Gaussian Splatting (3DGS) provides a continuous, geometry-aware scene representation for perception-driven navigation, existing 3DGS safety filters use reduced-order models such as single- and double-integrators that ignore actuator limits and assume commanded accelerations are realized instantaneously. Building on an analytic collision cone barrier for 3DGS, we introduce a nonlinear, actuator-aware safety filter enforced through the full quadrotor dynamics. We derive a high-relative-degree collision cone exponential CBF and a backup CBF that preserves QP feasibility under input constraints using a forward-simulated backup policy. Compared with a state-of-the-art 3DGS safety filter, our approach reduces trajectory jerk by 47% and runs 2.25 times faster. We validate the method in simulation and on hardware for real-time navigation in cluttered, perception-derived environments.
Chinese Translation
快速的四旋翼飞行需要在紧凑的机载计算限制下安全地避开障碍物。虽然3D高斯溅射（3DGS）为感知驱动的导航提供了连续的、几何感知的场景表示，但现有的3DGS安全过滤器使用的降阶模型（如单积分器和双积分器）忽略了执行器限制，并假设指令加速度瞬时实现。在基于3DGS的解析碰撞锥障碍物的基础上，我们引入了一种非线性、考虑执行器的安全过滤器，通过完整的四旋翼动力学进行强制执行。我们推导出了一种高相对度的碰撞锥指数控制障碍函数（CBF）和一种备份CBF，使用前向模拟的备份策略在输入约束下保持QP可行性。与最先进的3DGS安全过滤器相比，我们的方法将轨迹抖动减少了47%，运行速度提高了2.25倍。我们在模拟和硬件上验证了该方法，以实现复杂、基于感知的环境中的实时导航。
cs.RO / 52 / 2607.01201
Sensorless Four-Channel Control Architecture Using Inverse Dynamics Modeling for Human-Scale Bilateral Teleoperation
基于逆动力学建模的无传感器四通道控制架构用于人类规模的双向遥操作
Abstract
The four-channel teleoperation architecture is a well-established framework for achieving transparency in bilateral systems. However, its performance in human-scale teleoperation is limited by high inertia, modeling challenges, and reliance on noisy and costly force/torque sensors. This paper introduces a sensorless four-channel architecture based on inverse dynamics modeling. The controller is implemented and validated on a customized WAM bilateral teleoperation setup. Experiments demonstrate that the proposed approach outperforms conventional two- and four-channel schemes as well as transparency-enhancement methods, improving position and force tracking, reducing operator effort, and increasing maximum transmittable impedance without external sensors. A door-opening case study involving sustained whole-body contact along the manipulator further demonstrates the effectiveness of the method in realistic human-scale manipulation tasks.
Chinese Translation
四通道遥操作架构是实现双向系统透明度的成熟框架。然而，在人类规模的遥操作中，其性能受到高惯性、建模挑战以及对噪声大且成本高的力/扭矩传感器的依赖限制。本文提出了一种基于逆动力学建模的无传感器四通道架构。控制器在定制的WAM双向遥操作系统上实施并验证。实验表明，所提出的方法在位置和力跟踪方面优于传统的双通道和四通道方案以及透明度增强方法，减少了操作员的工作量，并在没有外部传感器的情况下提高了最大可传输阻抗。一个涉及操纵器持续全身接触的开门案例研究进一步证明了该方法在现实人类规模操纵任务中的有效性。
cs.RO / 53 / 2607.01212
FurnitureVLA: Learning Long-Horizon Bimanual Furniture Assembly with Vision-Language-Action Model
FurnitureVLA：使用视觉-语言-动作模型学习长时间跨度的双手家具组装
Abstract
Current work on robot furniture assembly mostly focuses on toy-scale settings or single-arm manipulation. We introduce FurnitureVLA, the first systematic study of real-scale bimanual furniture assembly using Vision-Language-Action models (VLAs). We formalize the task, develop a scalable simulation pipeline for expert data generation and evaluation, and build a VR teleoperation system for single-operator bimanual control to collect high-quality real-world demonstrations. To address extreme long-horizon assembly with up to 7 subtasks and 1550 control steps, we propose a progress-enhanced VLA, finetuned on semantically grounded subtasks, that jointly predicts actions and a continuous progress signal, enabling automatic subtask transitions and reducing compounding errors during inference. We further study perception and control design factors that critically affect precision in real-scale assembly. FurnitureVLA improves average simulation success from 48% to 80% compared to baselines across three furniture types, with an additional 21% gain from our design factor study. We validate on a real Kinova Gen3 platform with only 16% drop on the hardest task.
Chinese Translation
当前关于机器人家具组装的研究主要集中在玩具规模的设置或单臂操作上。我们介绍了FurnitureVLA，这是首次对使用视觉-语言-动作模型（VLA）进行真实规模双手家具组装的系统研究。我们对任务进行了形式化，开发了一个可扩展的仿真管道用于专家数据生成和评估，并建立了一个虚拟现实远程操作系统，以便于单操作者进行双手控制，从而收集高质量的真实世界演示。为了应对最多包含7个子任务和1550个控制步骤的极端长时间跨度的组装任务，我们提出了一种进度增强的VLA，该模型在语义基础的子任务上进行了微调，能够联合预测动作和连续进度信号，从而实现自动子任务转换并减少推理过程中的累积误差。我们进一步研究了对真实规模组装精度产生重要影响的感知与控制设计因素。与基线相比，FurnitureVLA在三种家具类型上的平均仿真成功率从48%提高到80%，而我们的设计因素研究又额外提高了21%。我们在真实的Kinova Gen3平台上进行了验证，最难任务的成功率仅下降16%。
cs.CV / 1 / 2607.00057
Enhancing Oracle Bone Inscription Recognition via Multi-Scale Layer Attention
通过多尺度层注意力增强甲骨文识别
Abstract
Oracle Bone Inscriptions (OBIs) recognition plays a crucial role in understanding ancient Chinese culture. However, accurately recognizing OBIs remains highly challenging due to their complex, irregular, and often degraded shapes. Traditional methods rely on expert knowledge and manual analysis, which are time-consuming and error-prone. Although deep learning has greatly advanced general image recognition, existing methods struggle to capture the fine-grained details and subtle variations inherent in OBIs, resulting in limited performance. Even most recent and effective layer attention techniques are designed to capture fine-grained dependencies through enhanced inter-layer interactions, yet they still exhibit only marginal improvements in OBIs recognition. To address these limitations, we propose Multi-Scale Layer Attention (MSLA), a novel paradigm that explicitly models both multi-scale and cross-layer feature interactions. By enriching the representation with fine-grained details across multiple spatial scales, MSLA enables more accurate and robust OBIs recognition. Extensive experiments on large-scale OBIs datasets demonstrate that MSLA consistently outperforms existing attention mechanisms while maintaining computational efficiency.
Chinese Translation
甲骨文（Oracle Bone Inscriptions, OBIs）识别在理解古代中国文化中发挥着至关重要的作用。然而，由于甲骨文形状复杂、不规则且常常退化，准确识别甲骨文仍然具有很高的挑战性。传统方法依赖于专家知识和人工分析，这些方法既耗时又容易出错。尽管深度学习在一般图像识别方面取得了重大进展，但现有方法在捕捉甲骨文固有的细粒度细节和微妙变化方面仍然存在困难，导致性能有限。即使是最新的有效层注意力技术也旨在通过增强层间交互来捕捉细粒度依赖关系，但在甲骨文识别中仍然仅表现出边际改善。为了解决这些局限性，我们提出了多尺度层注意力（Multi-Scale Layer Attention, MSLA），这是一种新颖的范式，明确建模多尺度和跨层特征交互。通过在多个空间尺度上丰富细粒度细节的表示，MSLA 实现了更准确和更稳健的甲骨文识别。在大规模甲骨文数据集上的广泛实验表明，MSLA 在保持计算效率的同时，始终优于现有的注意力机制。
cs.CV / 2 / 2607.00058
Joint Medical Image Enhancement and Segmentation with Diffusion-based Symbiotic Information Interaction
基于扩散的共生信息交互的医学图像增强与分割联合方法
Abstract
Image quality is critical for accurate medical diagnosis. However, MRI, CT, and ultrasound images are often of low resolution and quality due to cost constraints, complicating the visualization of key anatomical structures and lesions. While such limitations are common in practice, traditional methods treat image enhancement as a separate preprocessing step, failing to fully leverage its potential synergy with image segmentation. To address this, we propose DiSIINet (Diffusion-based Symbiotic Information Interaction Network), which is built on the principle that enhancement and segmentation should mutually reinforce each other in a unified model. Based on Denoising Diffusion Implicit Models (DDIM), DiSIINet integrates an enhancement branch and a segmentation branch. These branches interact through a novel Symbiotic Information Interaction (SII) module, which facilitates dynamic, feature-level information exchange via cross-attention during the reverse diffusion process. This design enables both tasks to iteratively improve each other. The DDIM backbone ensures high-quality output and efficient inference through deterministic sampling. Experiments on multi-modal medical datasets (MRI, CT, ultrasound) show that DiSIINet achieves significant performance improvements compared to sequential or independent enhancement and segmentation approaches. The code is available at: https://github.com/Reconsider80/DiSIINet.
Chinese Translation
图像质量对准确的医学诊断至关重要。然而，由于成本限制，MRI、CT和超声图像通常具有低分辨率和低质量，这使得关键解剖结构和病变的可视化变得复杂。尽管这种局限性在实践中很常见，传统方法将图像增强视为一个独立的预处理步骤，未能充分利用其与图像分割之间的潜在协同作用。为了解决这一问题，我们提出了DiSIINet（基于扩散的共生信息交互网络），其建立在增强与分割应在统一模型中相互强化的原则之上。基于去噪扩散隐式模型（DDIM），DiSIINet集成了一个增强分支和一个分割分支。这些分支通过一种新颖的共生信息交互（SII）模块进行交互，该模块在反向扩散过程中通过交叉注意力促进动态的特征级信息交换。该设计使得两个任务能够相互迭代地改进。DDIM主干通过确定性采样确保高质量的输出和高效的推理。在多模态医学数据集（MRI、CT、超声）上的实验表明，DiSIINet相比于顺序或独立的增强与分割方法实现了显著的性能提升。代码可在以下链接获取：https://github.com/Reconsider80/DiSIINet。
cs.CV / 3 / 2607.00060
Synergistic Perception-Reasoning Governance: Grounding Medical MLLMs with Verifiable Anatomical Evidence
协同感知-推理治理：基于可验证解剖证据的医学多模态大型语言模型
Abstract
Multimodal large language models (MLLMs) show strong promise for clinical VQA and radiology report generation, yet inference-time hallucinations still undermine trustworthy use: models can produce fluent conclusions that conflict with imaging evidence. Existing mitigation strategies typically rely on additional training, external retrieval/knowledge bases, or multi-stage post-hoc verification, which increases cost and pipeline complexity and often generalizes poorly across models and tasks.To address this, we propose a holistic, training-free evidence-injection framework that systematically mitigates hallucinations through dual-side evidence injection. By leveraging ROI priors acquired using MedSAM in our implementation, we recalibrate the visual perception trajectory via ROI-guided activation modulation while anchoring the textual reasoning trajectory by mapping anatomical coordinates into discrete semantic tokens as verifiable external memory. Then we introduce a task-aware dynamic router to select modality-specific interventions based on task semantics, balancing perceptual grounding and linguistic fluency. We conduct systematic evaluations on 2 tasks and 5 datasets using \texttt{LLaVA-1.5-7B}, \texttt{LLaVA-Med-1.5-7B}, \texttt{Qwen3-VL-8B/32B}, and \texttt{InternVL-3.5-8B/38B}. Controlled ablations and visualizations further validate the framework, which consistently outperforms baselines across medical benchmarks, improving close-ended accuracy by up to $\sim\mathbf{6}\%\uparrow$ and reducing open-ended hallucinations by $\sim\mathbf{35}\%\downarrow$. The code has been made available on GitHub: \href{https://github.com/Henry991115/SPRG}{\textcolor{blue}{https://github.com/Henry991115/SPRG}}.
Chinese Translation
多模态大型语言模型（MLLMs）在临床视觉问答（VQA）和放射学报告生成方面展现出强大的潜力，但推理时的幻觉现象仍然削弱了其可信赖性：模型可能产生与影像证据相矛盾的流畅结论。现有的缓解策略通常依赖于额外的训练、外部检索/知识库或多阶段的后验验证，这增加了成本和流程复杂性，并且往往在不同模型和任务之间的泛化效果不佳。为了解决这一问题，我们提出了一种整体的、无训练的证据注入框架，通过双侧证据注入系统性地缓解幻觉现象。通过利用在我们实现中使用MedSAM获得的感兴趣区域（ROI）先验，我们通过ROI引导的激活调制重新校准视觉感知轨迹，同时通过将解剖坐标映射为可验证的外部记忆的离散语义标记来锚定文本推理轨迹。然后，我们引入了一种任务感知的动态路由器，根据任务语义选择特定模态的干预，平衡感知基础和语言流畅性。我们在2个任务和5个数据集上使用 exttt{LLaVA-1.5-7B}、 exttt{LLaVA-Med-1.5-7B}、 exttt{Qwen3-VL-8B/32B}和 exttt{InternVL-3.5-8B/38B}进行了系统评估。控制性消融实验和可视化进一步验证了该框架，其在医学基准测试中始终优于基线，闭合式准确率提高了约$ extsim extbf{6}\% extuparrow$，开放式幻觉减少了约$ extsim extbf{35}\% extdownarrow$。代码已在GitHub上发布： extcolor{blue}{https://github.com/Henry991115/SPRG}。
cs.CV / 4 / 2607.00090
Lost in the Tail: Addressing Geographic Imbalance in Urban Visual Place Recognition
迷失在尾部：解决城市视觉地点识别中的地理不平衡问题
Abstract
Urban-scale Visual Place Recognition (VPR) aims to identify the geographic location of a query image by matching it against a geo-tagged database. While recent methods achieve impressive performance, they overlook a serious long-tailed problem hidden in urban-scale datasets, which biases the model towards locations with abundant images and ignores less-visited areas, causing models to systematically favor frequently photographed locations while failing in sparsely covered areas. In this paper, we systematically characterize this imbalance challenge and propose Distribution-Aware Place Recognition (DAPR), a model-agnostic plug-in framework that rebalances gradient contributions across head and tail classes. Additionally, within classification-retrieval pipelines, DAPR applies a multi-scale distance search mechanism to compute per-class distributional compactness, providing complementary gains at the retrieval stage. On the large-scale SF-XL benchmark, our framework outperforms the previous classification-retrieval baseline by 18.3% on test set v1, and 6.7% on test set v2. As a plug-in module, it achieves consistent improvements across representative VPR methods on SF-XL, MSLS, and Pitts30k, demonstrating broad generalizability across different methods and benchmarks.
Chinese Translation
城市规模的视觉地点识别（VPR）旨在通过与地理标记数据库进行匹配，识别查询图像的地理位置。尽管最近的方法取得了令人印象深刻的性能，但它们忽视了隐藏在城市规模数据集中一个严重的长尾问题，这使得模型偏向于图像丰富的地点，而忽视了访问较少的区域，导致模型系统性地偏向于经常被拍摄的位置，而在覆盖稀疏的区域表现不佳。本文系统地描述了这一不平衡挑战，并提出了分布感知地点识别（Distribution-Aware Place Recognition, DAPR），这是一种模型无关的插件框架，旨在重新平衡头部和尾部类别的梯度贡献。此外，在分类-检索管道中，DAPR应用多尺度距离搜索机制来计算每个类别的分布紧凑性，在检索阶段提供补充收益。在大规模的SF-XL基准测试中，我们的框架在测试集v1上比之前的分类-检索基线提高了18.3%，在测试集v2上提高了6.7%。作为一个插件模块，它在SF-XL、MSLS和Pitts30k等代表性VPR方法上实现了一致的改进，展示了在不同方法和基准测试中的广泛适用性。
cs.CV / 5 / 2607.00115
PixelEyes: Decoupling Perception and Reasoning for Pinpoint Visual Evidence Seeking
PixelEyes：解耦感知与推理以精确寻求视觉证据
Abstract
This paper explores multi-turn visual reasoning and observes that MLLMs repeatedly fail to localize the target, leading to long, redundant trajectories. We attribute this failure to the entanglement of reasoning and perception within a single model, the MLLM reasons and localizes simultaneously, and inaccurate localization triggers additional reasoning turns that bloat the trajectory. To solve this problem, we propose PixelEyes, a multi-turn visual reasoning agent that explicitly decouples reasoning from perception, i.e., the reasoner decides what to look for, while a specialized perception tool answers where it is. Specifically, PixelEyes introduces 1) Mask-guided Visual Search. A referring segmentation model is invoked to provide mask-precise localization, freeing the reasoner from the need to compensate for imprecise grounding. 2) Semantic-region Breadth-first Search (BFS). To eliminate redundant loops caused by repeatedly cropping incorrect sub-regions, we organize exploration as a breadth-first search over semantic regions. To internalize these capabilities, we construct the PixelEyes-6K dataset by resynthesizing expert trajectories from existing data. This explicitly embeds our mask-guided search and BFS logic into the model. We further introduce Pinpoint-Bench, a zero-hint visual search benchmark, i.e., no location cues are provided in the question, with instance-level masks and bounding boxes that separate localization failures from reasoning failures, enabling fine-grained analysis of failure modes such as inattentional blindness. Recent state-of-the-art MLLMs and visual reasoning agents leave large headroom on Pinpoint-Bench, demonstrating its quality and difficulty. Code and models are open-sourced.
Chinese Translation
本文探讨了多轮视觉推理，并观察到多模态大型语言模型（MLLMs）在定位目标时反复失败，导致冗长且重复的轨迹。我们将这一失败归因于推理与感知在单一模型中的纠缠，MLLM同时进行推理和定位，而不准确的定位会触发额外的推理轮次，从而膨胀轨迹。为了解决这一问题，我们提出了PixelEyes，一个明确解耦推理与感知的多轮视觉推理代理，即推理器决定寻找什么，而专门的感知工具则回答其位置。具体而言，PixelEyes引入了1）基于掩码的视觉搜索。调用一个参考分割模型以提供掩码精确的定位，解放推理器免于补偿不精确的基础。2）语义区域广度优先搜索（BFS）。为了消除因重复裁剪错误子区域而导致的冗余循环，我们将探索组织为对语义区域的广度优先搜索。为了内化这些能力，我们通过重新合成现有数据中的专家轨迹构建了PixelEyes-6K数据集。这明确将我们的基于掩码的搜索和BFS逻辑嵌入模型中。我们进一步引入了Pinpoint-Bench，一个零提示视觉搜索基准，即问题中不提供位置信息，具有实例级掩码和边界框，将定位失败与推理失败分开，从而实现对失效模式（如注意力盲区）的细致分析。最近的最先进的MLLMs和视觉推理代理在Pinpoint-Bench上仍有较大的提升空间，证明了其质量和难度。代码和模型已开源。
cs.CV / 6 / 2607.00124
Segmenting, Fast and Slow: Real-Time Open-Vocabulary Video Instance Segmentation with Dual-Path Processing
快速与缓慢的分割：基于双路径处理的实时开放词汇视频实例分割
Abstract
Object-centric models inspired by DETR have become the dominant paradigm for open-vocabulary video instance segmentation (OV-VIS). While recent efforts have reduced the computational cost of pixel decoding, textual modality fusion, and object decoding to make these architectures more suitable for mobile devices, real-time on-device inference at high frame rates remains an open challenge. In this paper, we introduce SegFS, a dual-stream fast-slow framework that significantly improves efficiency without sacrificing accuracy. On sparse keyframes, an open-vocabulary object-based model predicts instance-level representations. These representations are then projected back into the backbone feature space to condition a lightweight fast network, which efficiently relocalizes and segments the instances in subsequent frames. By shifting instance propagation from object decoding to feature-space conditioning, our approach decouples multimodal semantic understanding from dense mask prediction and enables efficient temporal propagation. The proposed fast branch achieves up to 14x lower latency than the mobile-oriented MOBIUS model, while maintaining competitive segmentation performance on standard OV-VIS benchmarks.
Chinese Translation
受DETR启发的以对象为中心的模型已成为开放词汇视频实例分割（OV-VIS）的主流范式。尽管最近的努力已经降低了像素解码、文本模态融合和对象解码的计算成本，使这些架构更适合移动设备，但在高帧率下进行实时设备推理仍然是一个未解决的挑战。本文介绍了SegFS，一种双流快速-缓慢框架，显著提高了效率而不牺牲准确性。在稀疏关键帧上，开放词汇对象模型预测实例级表示。这些表示随后被投影回主干特征空间，以调节轻量级快速网络，该网络有效地重新定位并分割后续帧中的实例。通过将实例传播从对象解码转移到特征空间调节，我们的方法将多模态语义理解与密集掩膜预测解耦，并实现高效的时间传播。所提出的快速分支的延迟比面向移动设备的MOBIUS模型低达14倍，同时在标准OV-VIS基准上保持竞争力的分割性能。
cs.CV / 7 / 2607.00125
Decompose, Compare, and Decide: Multimodal LLMs are Implicit Few-Shot Learners
分解、比较与决策：多模态大型语言模型隐式地成为少量样本学习者
Abstract
Multimodal Large Language Models (MLLMs) have demonstrated remarkable abilities when analyzing images, yet translating these capabilities to few-shot image classification remains challenging. To bridge this gap, we present DeCoDe, a simple yet effective technique that enables off-the-shelf MLLMs to act as strong few-shot classifiers without any additional training. Our approach builds on the idea of few-shot classification as a set of pairwise image comparisons, decomposing the task into a set of binary decisions. Given a query image and a support image from a candidate class, the MLLM is prompted to decide whether the two images depict the same class. The logit corresponding to an affirmative response is then used as a similarity score to assign the query image to the most likely class. While this already yields good results, we show that providing additional high-level information, such as the data domain, to the model further improves performance. Our evaluation provides an extensive analysis of various inference variants on a suite of twelve datasets, six established and six newly curated few-shot benchmarks spanning across diverse domains. The results show that the proposed simple decomposition technique can turn off-the-shelf MLLMs into powerful few-shot learners, significantly outperforming current state-of-the-art few-shot methods on both standard and novel domains. Code is available at https://github.com/yunhanwang1105/DeCoDe.
Chinese Translation
多模态大型语言模型（MLLMs）在图像分析方面展现了卓越的能力，但将这些能力转化为少量样本图像分类仍然具有挑战性。为了解决这一问题，我们提出了DeCoDe，这是一种简单而有效的技术，使得现成的MLLMs能够在无需额外训练的情况下充当强大的少量样本分类器。我们的方法基于将少量样本分类视为一组成对图像比较的思想，将任务分解为一系列二元决策。给定一个查询图像和一个来自候选类别的支持图像，MLLM被提示决定这两幅图像是否属于同一类别。对应于肯定回答的logit值被用作相似性评分，以将查询图像分配给最可能的类别。尽管这一方法已经取得了良好的结果，我们还展示了向模型提供额外的高层次信息（例如数据领域）可以进一步提高性能。我们的评估对在十二个数据集上的各种推理变体进行了广泛分析，其中包括六个已建立的和六个新整理的少量样本基准，涵盖了不同领域。结果表明，所提出的简单分解技术能够将现成的MLLMs转变为强大的少量样本学习者，在标准和新领域上显著超越当前最先进的少量样本方法。代码可在 https://github.com/yunhanwang1105/DeCoDe 获取。
cs.CV / 8 / 2607.00129
A Synthetic-Driven Vision System for Assembly Step Recognition
一种基于合成驱动的视觉系统用于装配步骤识别
Abstract
Quality control in industrial assembly is essential, and real-time monitoring of the assembly process is crucial for preventing costly defects and ensuring production reliability. Vision-based automated inspection offers a powerful solution for such real-time monitoring. However, due to the specialized industrial components and processes, training these models typically relies on task-specific real-world data, which is costly and labor-intensive to collect and annotate. In this paper, we propose a system that automatically generates realistic assembly sequences and further trains real-time inspection models using the synthetic data. It can be efficiently applied to a given task within an hour, requiring only CAD models and simple step descriptions. Focusing on practical challenges, our system integrates a physics-based motion generation module to capture the variance of different human assembly, designs domain-randomized rendering to deal with the environmental complexity and variation, and employs an object-detection-based step recognition module for robust sim-to-real transfer, leading to 92.4% accuracy on a real-world assembly case with 46.7%, 15.8% and 61.2% performance improvement, respectively. Overall, our system provides a practical solution for industrial assembly inspection without requiring expensive real-world data collection and annotation, with the effectiveness validated on real industrial assembly tasks.
Chinese Translation
工业装配中的质量控制至关重要，实时监控装配过程对于防止昂贵缺陷和确保生产可靠性至关重要。基于视觉的自动化检测为这种实时监控提供了强有力的解决方案。然而，由于工业组件和过程的专业性，训练这些模型通常依赖于特定任务的真实世界数据，而收集和标注这些数据既昂贵又耗时。本文提出了一种系统，该系统能够自动生成逼真的装配序列，并利用合成数据进一步训练实时检测模型。该系统可以在一个小时内高效应用于特定任务，仅需CAD模型和简单的步骤描述。针对实际挑战，我们的系统集成了基于物理的运动生成模块，以捕捉不同人类装配的变异性，设计了领域随机化渲染以应对环境的复杂性和变化，并采用基于目标检测的步骤识别模块实现稳健的模拟到真实转移，在一个真实的装配案例中达到了92.4%的准确率，分别提高了46.7%、15.8%和61.2%的性能。总体而言，我们的系统为工业装配检测提供了一个实用的解决方案，无需昂贵的真实世界数据收集和标注，其有效性在真实工业装配任务中得到了验证。
cs.CV / 9 / 2607.00138
MG-SpaIR: Multi-grade Sparse-guided Implicit Representation for Training-Data-Free Image Restoration
MG-SpaIR：用于无训练数据图像恢复的多级稀疏引导隐式表示
Abstract
MG-SpaIR is a training-data-free framework for restoring a clean image from a single observation corrupted by a mixture of blur, downsampling, noise, and missing pixels. Building on implicit neural representations (INRs), we introduce a multi-grade coarse-to-fine residual hierarchy that progressively refines the reconstruction across resolution grades, improving representational fidelity and mitigating spectral limitations. To stabilize reconstruction optimization and suppress INR-induced artifacts, we further propose an explicit sparse proximal regularization (e.g., $\ell_0$-type) applied directly in the high-resolution image domain, which discourages spurious high-frequency patterns while preserving sharp structures. The resulting optimization is solved efficiently via a multi-grade proximal alternating scheme, and we establish convergence guarantees for the associated updates under standard regularity conditions. Experiments on mixed-degradation benchmarks demonstrate that MG-SpaIR consistently outperforms strong training-data-free baselines such as Deep Image Prior, providing a stable, interpretable, and data-efficient alternative to conventional learning-based restoration methods.
Chinese Translation
MG-SpaIR 是一个无训练数据的框架，用于从单个观测图像中恢复干净图像，该观测图像受到模糊、下采样、噪声和缺失像素的混合影响。基于隐式神经表示（INRs），我们引入了一种多级粗到细的残差层次结构，该结构逐步在不同分辨率级别上精细化重建，提高了表示的保真度并缓解了光谱限制。为了稳定重建优化并抑制由 INR 引起的伪影，我们进一步提出了一种显式稀疏近端正则化（例如，$ ext{l}_0$-类型），该正则化直接应用于高分辨率图像域，抑制伪高频模式，同时保留锐利结构。所得到的优化通过多级近端交替方案高效求解，并在标准正则性条件下为相关更新建立了收敛保证。在混合退化基准上的实验表明，MG-SpaIR 始终优于强大的无训练数据基线方法，如 Deep Image Prior，提供了一种稳定、可解释且数据高效的替代方案，取代传统的基于学习的恢复方法。
cs.CV / 10 / 2607.00144
A Mechanism-Driven Theory of Phase Transitions in Active Learning
基于机制驱动的主动学习相变理论
Abstract
Active learning (AL) performance is known to be budget-dependent, yet regimes are typically defined by heuristic label counts that fail to generalize across datasets or architectures. We characterize AL dynamics by reframing budget regimes as shifts in the dominant generalization mechanism. By reinterpreting PAC-style risk components as dynamic interacting terms, we prove that dominance shifts are structurally unavoidable, creating a moving bottleneck for generalization. We operationalize this using measurable proxies and a segmented regression procedure to identify a tripartite taxonomy: data-driven, transition, and model-driven phases. Our framework explains the long-standing observation that representativeness, coverage, and uncertainty strategies excel at different stages. Experiments across natural and medical imaging show that AL efficiency depends on the alignment between the strategy's inductive bias and the active bottleneck. Moreover, self-supervised representation shift transitions earlier along the labeling trajectory, highlighting the role of representation quality in shaping AL dynamics. Overall, this work provides a unified framework for the next generation of transition-aware AL algorithms.
Chinese Translation
主动学习（Active Learning, AL）的性能被认为依赖于预算，然而，通常通过启发式标签计数定义的不同阶段在不同数据集或架构之间无法推广。我们通过将预算阶段重新表述为主导泛化机制的转变来表征主动学习动态。通过将PAC（Probably Approximately Correct）风格的风险组件重新解释为动态交互项，我们证明了主导性转变在结构上是不可避免的，从而为泛化创造了一个移动瓶颈。我们使用可测量的代理和分段回归程序来实现这一点，以识别三分法分类：数据驱动阶段、过渡阶段和模型驱动阶段。我们的框架解释了长期以来的观察，即代表性、覆盖率和不确定性策略在不同阶段表现优异。在自然和医学成像中的实验表明，主动学习的效率依赖于策略的归纳偏差与主动瓶颈之间的对齐。此外，自监督表示的转变在标记轨迹上更早发生，突显了表示质量在塑造主动学习动态中的作用。总体而言，本研究为下一代关注转变的主动学习算法提供了一个统一的框架。
cs.CV / 11 / 2607.00157
Progressive Pose-Guided 4D Animal Reconstruction from Monocular Video
基于渐进姿态引导的单目视频4D动物重建
Abstract
Reconstructing 4D animals from monocular videos is challenging due to large inter-species variation, complex articulations, and the lack of reliable templates. Existing approaches typically rely on either strict category-specific priors that restrict generalization, or unconstrained generative models that sacrifice input fidelity. To bridge this gap, we present a progressive test-time optimization framework built on 3D Gaussian Splatting for high-fidelity 4D animal reconstruction from a single video. Our key insight is that a coarse shape prior suffices when coupled with a progressive strategy that disentangles articulated pose from non-rigid deformation. Specifically, we employ a symmetry-aware temporal encoding that exploits bilateral cues while absorbing camera estimation drift and a part-conditioned deformation mechanism guided by learnable part anchors and a learnable skinning field. Extensive experiments demonstrate that our approach generalizes robustly across diverse species, achieving superior geometric accuracy, temporal consistency, and visual fidelity compared to existing baselines, even under severe prior mismatch.
Chinese Translation
从单目视频重建4D动物具有挑战性，因为物种间差异大、关节复杂，并且缺乏可靠的模板。现有方法通常依赖于严格的类别特定先验，这限制了泛化能力，或依赖于不受约束的生成模型，这牺牲了输入的保真度。为了解决这一问题，我们提出了一种基于3D高斯点云的渐进测试时优化框架，用于从单个视频中高保真地重建4D动物。我们的关键见解是，当结合渐进策略以解耦关节姿态与非刚性变形时，粗略的形状先验就足够了。具体而言，我们采用了一种对称感知的时间编码，利用双边线索，同时吸收相机估计漂移，并采用一种由可学习的部件锚点和可学习的蒙皮场引导的部件条件变形机制。大量实验表明，我们的方法在不同物种间具有良好的泛化能力，与现有基线相比，在几何精度、时间一致性和视觉保真度方面表现优越，即使在严重的先验不匹配情况下也是如此。
cs.CV / 12 / 2607.00174
Steal the Patch Size: Adversarially Manipulate Vision-Language Models
窃取补丁大小：对视觉-语言模型的对抗性操控
Abstract
We present a black-box model-stealing attack that recovers private vision-tokenizer configurations of deployed vision-language models (VLMs), including the visual patch size and input preprocessing pipeline. The key idea is a task-level side channel induced by ViT-style patchification: when a synthetic grid image is aligned with the hidden patch grid, boundary cues are erased at tokenization, causing periodic accuracy drop. By sweeping the grid cell size and measuring these collapses, we infer the patch size; by introducing padding and a consistency-check test, we further identify whether preprocessing is dynamic- or fixed-resolution and recover the target resize resolution. Across open-source Qwen-VL variants and proprietary models including GPT and Claude, we reliably recover tokenizer-related parameters. Finally, we show that such leakage enables preprocessing-aware transfer attacks and model-targeted adversarial manipulation.
Chinese Translation
我们提出了一种黑箱模型窃取攻击，能够恢复已部署视觉-语言模型（VLMs）的私有视觉标记器配置，包括视觉补丁大小和输入预处理流程。关键思想是由ViT风格的补丁化引发的任务级侧信道：当合成网格图像与隐藏的补丁网格对齐时，边界线索在标记化过程中被抹去，导致周期性的准确率下降。通过调整网格单元大小并测量这些下降，我们推断出补丁大小；通过引入填充和一致性检查测试，我们进一步识别预处理是动态分辨率还是固定分辨率，并恢复目标的调整分辨率。在开源的Qwen-VL变体和包括GPT及Claude在内的专有模型中，我们可靠地恢复了与标记器相关的参数。最后，我们展示了这种泄露使得基于预处理的转移攻击和针对模型的对抗性操控成为可能。
cs.CV / 13 / 2607.00176
PRISM-VO: Scale-Aware Visual Odometry Using Photometric Plenoptic Bundle Adjustment
PRISM-VO：基于光度全光束调整的尺度感知视觉里程计
Abstract
We introduce PRISM-VO, a novel pure optimization-based sparse photometric visual odometry framework for focused plenoptic cameras. The core of PRISM-VO is a novel photometric plenoptic bundle adjustment which jointly optimizes camera poses and inverse depth values of points in a sliding window. By combining geometric depth from a single plenoptic image with temporal multi-view constraints, PRISM-VO achieves accurate and drift-resilient motion estimation. Through explicit modeling of the plenoptic projection, PRISM-VO provides reliable metric-scale reconstructions, overcoming the scale ambiguity of monocular SLAM algorithms. Importantly, our approach relies solely on a single plenoptic sensor and avoids complex initialization, as depth priors are computed directly from plenoptic imaging. Experiments show that PRISM-VO outperforms the current state-of-the-art plenoptic visual odometry method on indoor and outdoor scenes. The proposed approach rivals other optimization- and learning-based methods while accurately and reliably recovering a metric scale of the scene. Project page: https://prism-vo.github.io/
Chinese Translation
我们介绍了PRISM-VO，一种针对聚焦全光学相机的新型纯优化稀疏光度视觉里程计框架。PRISM-VO的核心是一个新颖的光度全光束调整方法，该方法在滑动窗口中联合优化相机姿态和点的逆深度值。通过将单个全光学图像中的几何深度与时间多视图约束相结合，PRISM-VO实现了准确且抗漂移的运动估计。通过对全光学投影的明确建模，PRISM-VO提供了可靠的度量尺度重建，克服了单目SLAM算法的尺度模糊问题。重要的是，我们的方法仅依赖于单个全光学传感器，避免了复杂的初始化，因为深度先验是直接从全光学成像中计算得出的。实验表明，PRISM-VO在室内和室外场景中优于当前最先进的全光学视觉里程计方法。所提出的方法与其他基于优化和学习的方法相媲美，同时准确可靠地恢复场景的度量尺度。项目页面：https://prism-vo.github.io/
cs.CV / 14 / 2607.00183
DriftScope: Measuring The Hidden Effects of Diffusion Model Adaptation
漂移范围：测量扩散模型适应的隐性影响
Abstract
Adapting pre-trained text-to-image diffusion models, whether to learn new visual concepts or erase unwanted ones, is routinely evaluated on its intended effects alone. We argue this framing is incomplete. Through sparse autoencoder analysis and zero-shot classification, we demonstrate that adaptation systematically damages semantically unrelated concepts in ways that aggregate metrics structurally cannot surface: when damage is severe enough for FID and KID to respond, the model is already nearly unusable; when the model remains functional, FID and KID stay flat while specific classes silently suffer worst-case zero-shot accuracy drops of up to 18.9 points and concept-level distributions shift dramatically. This pattern appears at both ends of the adaptation spectrum (concept customization and concept unlearning), suggesting it is a systematic consequence of weight-level modification rather than an artifact of any particular method. To surface this hidden drift before deployment, we introduce DriftScope, a prompt-level diagnostic tool that takes any two model checkpoints and returns a ranked list of tokens whose visual concepts have shifted most between them. DriftScope optimizes a soft prompt to attribute drift at the token level without requiring access to real data or model internals. The result is an interpretable, concept-level audit that aggregate evaluation cannot provide.
Chinese Translation
适应预训练的文本到图像扩散模型，无论是为了学习新的视觉概念还是消除不需要的概念，通常仅根据其预期效果进行评估。我们认为这种框架是不完整的。通过稀疏自编码器分析和零-shot 分类，我们展示了适应过程系统性地损害了语义上无关的概念，而这种损害在聚合指标的结构中无法显现：当损害严重到足以影响 FID 和 KID 时，模型已经几乎无法使用；而当模型仍然可用时，FID 和 KID 保持平稳，而特定类别的零-shot 准确率却悄然下降高达 18.9 个百分点，概念级分布也发生了剧烈变化。这一模式出现在适应光谱的两端（概念定制和概念遗忘），表明这是权重级修改的系统性结果，而非任何特定方法的伪影。为了在部署前揭示这种隐性漂移，我们引入了 DriftScope，这是一种基于提示的诊断工具，它接受任意两个模型检查点，并返回一个排名列表，列出在这两个检查点之间视觉概念变化最大的标记。DriftScope 优化一个软提示，以在标记级别归因漂移，而无需访问真实数据或模型内部。其结果是一个可解释的、概念级的审计，而聚合评估无法提供。
cs.CV / 15 / 2607.00189
VOCA: Visual Odometry with Codec Awareness
VOCA：具有编解码器意识的视觉里程计
Abstract
Camera pose estimation from image streams is a critical component of spatial world models that integrate perception into planning and decision-making. Nearly all Visual Odometry (VO) and Simultaneous Localization and Mapping (V-SLAM) systems have focused on datasets containing raw, uncompressed videos. Many working systems instead use ubiquitous hardware units to efficiently compress and decode video streams, saving orders of magnitude in storage and bandwidth. However, this lossy compression introduces visual artifacts that hinder the performance of traditional tracking systems. We present VOCA, a causal stereo visual-odometry method that exploits codec information to improve tracking performance. We achieve state-of-the-art performance on causal VO for relative trajectory error, efficiency, and absolute trajectory error on compressed streams. This work highlights the potential of leveraging widely available video codec information for vision tasks.
Chinese Translation
从图像流中进行相机姿态估计是将感知集成到规划和决策中的空间世界模型的关键组成部分。几乎所有的视觉里程计（VO）和同时定位与地图构建（V-SLAM）系统都集中于包含原始、未压缩视频的数据集。然而，许多实际系统则使用普遍存在的硬件单元来高效地压缩和解码视频流，从而在存储和带宽上节省了数量级的开销。然而，这种有损压缩引入了视觉伪影，妨碍了传统跟踪系统的性能。我们提出了VOCA，一种因果立体视觉里程计方法，利用编解码器信息来提高跟踪性能。在压缩流上，我们在因果视觉里程计的相对轨迹误差、效率和绝对轨迹误差方面达到了最先进的性能。这项工作突显了利用广泛可用的视频编解码器信息在视觉任务中的潜力。
cs.CV / 16 / 2607.00201
Trust the Prior (or Not): Uncertainty-Aware Abdominal Aortic Aneurysm Segmentation
信任先验（或不信任）：基于不确定性的腹主动脉瘤分割
Abstract
Robust segmentation of intraluminal thrombus is critical for risk assessment in Abdominal Aortic Aneurysm, yet it remains challenging due to heterogeneous thrombus features and low contrast with surrounding non-enhanced tissues. Domain shifts induced by different Computed Tomography Angiography (CTA) protocols further inhibit multi-center generalization of deep learning models. To address these challenges, we propose a patient-specific framework that integrates discriminative learning with anatomically informed priors. Our approach introduces two key components: (1) a patient-specific intensity normalization based on a Gaussian Mixture Model of local anatomy, and (2) an Uncertainty-Gated Anatomical Attention module that incorporates spatial priors while adaptively modulating their influence according to voxel-wise confidence. This design allows for anatomical guidance in ambiguous regions while suppressing unreliable priors. The proposed method achieves state-of-the-art performance on in-distribution test data and substantially outperforms existing alternatives in generalization to external multi-center CTA data, while remaining interpretable through an explicit separation of visual and anatomical evidence.
Chinese Translation
对腔内血栓进行稳健分割对于腹主动脉瘤的风险评估至关重要，但由于血栓特征的异质性和与周围未增强组织的低对比度，这一任务仍然具有挑战性。不同的计算机断层扫描血管造影（CTA）协议引起的领域转移进一步抑制了深度学习模型的多中心泛化能力。为了解决这些挑战，我们提出了一种患者特异性框架，将区分学习与解剖学先验相结合。我们的方法引入了两个关键组件：（1）基于局部解剖的高斯混合模型的患者特异性强度归一化；（2）不确定性门控解剖注意模块，该模块在引入空间先验的同时，根据体素级置信度自适应调节其影响。这种设计允许在模糊区域进行解剖指导，同时抑制不可靠的先验。所提出的方法在分布内测试数据上实现了最先进的性能，并在对外部多中心CTA数据的泛化能力上显著优于现有替代方案，同时通过明确分离视觉和解剖证据保持可解释性。
cs.CV / 17 / 2607.00218
EgoSafetyBench: A Diagnostic Egocentric Video Benchmark for Evaluating Embodied VLMs as Runtime Safety Guards
EgoSafetyBench：用于评估具身视觉语言模型作为运行时安全防护的诊断自我中心视频基准
Abstract
Vision-language models (VLMs) are now proposed as runtime safety guards for embodied agents in homes and factories. A deployable guard must catch genuinely unsafe situations while avoiding unnecessary intervention on routine but superficially alarming activity, a distinction that binary safety benchmarks obscure. We introduce EgoSafetyBench, an egocentric video benchmark of 1,200 robot-view scenarios annotated at half-second granularity, to evaluate VLMs as streaming guards across two tracks. The situational track (800 scenarios) spans four families, from routine and safe-but-suspicious scenes to obvious and contextual hazards. The visual-channel track (400 scenarios) targets in-scene text-a sign, sticker, or label visible in the scene-that can misrepresent the physical situation, pairing each misleading sign with a truthful version to test both whether a guard flags the text as misleading and whether the text corrupts its physical-safety judgment. Both tracks use contrastive ladders: near-identical scenarios differing only in a single visible deciding cue, so a correct call must hinge on that cue rather than the overall scene type. We evaluate ten open- and closed-source VLMs. We find that while guards reliably recognize videos containing hazards, they often miss specific hazardous moments, particularly contextual hazards. Furthermore, misleading in-scene signs degrade all tested guards: vulnerable models miss up to a third of hazards, while robust models over-intervene on safe content. Matched controls reveal that apparent safety robustness often reflects indiscriminate alarming rather than true physical reasoning.
Chinese Translation
视觉语言模型（VLMs）现在被提议作为家居和工厂中具身代理的运行时安全防护。一个可部署的防护必须能够捕捉真正不安全的情况，同时避免对日常但表面上令人担忧的活动进行不必要的干预，而这种区分在二元安全基准中并不明显。我们引入了EgoSafetyBench，这是一个包含1200个机器人视角场景的自我中心视频基准，按半秒粒度进行注释，以评估VLMs作为流媒体防护的能力，分为两个轨道。情境轨道（800个场景）涵盖四个类别，从常规和安全但可疑的场景到明显和上下文危险。视觉通道轨道（400个场景）针对场景中的文本——标志、贴纸或标签——这些文本可能会误导物理情况，每个误导性标志都与一个真实版本配对，以测试防护是否将文本标记为误导性，以及该文本是否会影响其物理安全判断。两个轨道都使用对比梯度：几乎相同的场景仅在一个可见的决定性线索上有所不同，因此正确的判断必须依赖于该线索，而不是整体场景类型。我们评估了十个开源和闭源的VLMs。我们发现，尽管防护能够可靠地识别包含危险的视频，但它们往往会错过特定的危险时刻，特别是上下文危险。此外，误导性的场景标志会降低所有测试防护的效果：脆弱的模型错过多达三分之一的危险，而强健的模型则对安全内容过度干预。匹配的对照实验表明，表面上的安全鲁棒性往往反映的是不加区分的警报，而非真实的物理推理。
cs.CV / 18 / 2607.00223
Does Your ViT Still Need U-Net for Segmentation?
你的ViT在分割中仍然需要U-Net吗？
Abstract
Medical image segmentation is dominated by U-Net-style encoder-decoder architectures. Vision Transformers (ViTs) overcome the limited receptive field of convolutional networks through self-attention, enabling modeling of long-range dependencies. Early ViT-based segmentation methods typically retained U-Net-style decoders because pretrained ViT representations were insufficient to support accurate dense prediction. Recent advances in large-scale pretraining have redefined the representation capability of ViTs, reducing the reliance on U-Net-style decoder architectures in modern vision models. This prompts two questions: Is the U-Net paradigm still necessary for medical image segmentation? If not, how should an encoder-only segmentation framework be designed? Motivated by these questions, we explore key architectural choices for encoder-only medical image segmentation based on modern ViT backbones and establish a query-based encoder-only design with multi-level query modeling and learnable block fusion, realized in Encoder-only Segmentation (EoSeg). Extensive experiments across seven benchmark datasets spanning CT, MRI, histopathology, endoscopy, and dermoscopy validate the effectiveness of the proposed design across diverse medical imaging modalities, including mDice scores of 85.50% on Synapse, 91.73% on ACDC, and 93.27% on GlaS. The results demonstrate that a U-Net-style decoder is no longer necessary for medical image segmentation with modern ViT backbones and further show that EoSeg provides an effective encoder-only design. Code is available at: https://github.com/Retinal-Research/EoSeg
Chinese Translation
医学图像分割主要由U-Net风格的编码器-解码器架构主导。视觉变换器（ViTs）通过自注意力机制克服了卷积网络的有限感受野，使得建模长距离依赖成为可能。早期基于ViT的分割方法通常保留了U-Net风格的解码器，因为预训练的ViT表示不足以支持准确的密集预测。近期大规模预训练的进展重新定义了ViTs的表示能力，减少了现代视觉模型对U-Net风格解码器架构的依赖。这引发了两个问题：U-Net范式在医学图像分割中仍然必要吗？如果不必要，如何设计一个仅包含编码器的分割框架？基于这些问题，我们探讨了基于现代ViT骨干网络的仅编码器医学图像分割的关键架构选择，并建立了一种基于查询的仅编码器设计，结合了多层次查询建模和可学习的块融合，实现在仅编码器分割（EoSeg）中。我们在涵盖CT、MRI、组织病理学、内窥镜和皮肤镜的七个基准数据集上进行了广泛实验，验证了所提设计在多种医学成像模式下的有效性，包括在Synapse上达到85.50%的mDice分数，在ACDC上达到91.73%，在GlaS上达到93.27%。结果表明，现代ViT骨干网络的医学图像分割不再需要U-Net风格的解码器，并进一步显示EoSeg提供了一种有效的仅编码器设计。代码可在以下链接获取：https://github.com/Retinal-Research/EoSeg
cs.CV / 19 / 2607.00251
Leveraging Phase Information to Boost Unrolled Network Learning for Image Deblurring
利用相位信息提升展开网络学习以进行图像去模糊
Abstract
While most image deblurring techniques directly restore the spatial image variable, we propose an amplitude and phase decomposition recognizing the importance of accurate phase estimation in recovering sharp image details. To that end, we first develop novel linear minimum mean squared (LMMSE) estimators of the amplitude and phase of the blurred, noisy image observation. An iterative optimization algorithm follows that recovers the sharp image using the aforementioned LMMSE estimators. Finally, matrix parameters that are statistically determined and fixed in the iterative algorithm are now learned using a training dataset of clean and degraded observations. Our deblurring engine is dubbed UPADNet (Unrolled Phase and Amplitude Decomposition Network), such that each iteration of the underlying phase and amplitude recovery algorithm is parameterized and trained end-to-end. Experiments over benchmark evaluation datasets such as GoPro, RealBlur and COCO datasets confirm that UPADNet outperforms state of the art deep networks including those based on algorithm unrolling in the image domain. The benefits of UPADNet are even more pronounced in high noise and limited training data regimes.
Chinese Translation
尽管大多数图像去模糊技术直接恢复空间图像变量，但我们提出了一种幅度和相位分解方法，认识到准确的相位估计在恢复清晰图像细节中的重要性。为此，我们首先开发了模糊噪声图像观测的幅度和相位的新型线性最小均方误差（LMMSE）估计器。接下来，采用迭代优化算法，利用上述LMMSE估计器恢复清晰图像。最后，在迭代算法中统计确定并固定的矩阵参数现在通过清晰和退化观测的数据集进行学习。我们的去模糊引擎被称为UPADNet（展开相位和幅度分解网络），使得基础相位和幅度恢复算法的每次迭代都能够参数化并进行端到端训练。在GoPro、RealBlur和COCO等基准评估数据集上的实验表明，UPADNet的性能优于包括基于算法展开的深度网络在内的最新技术。在高噪声和有限训练数据的情况下，UPADNet的优势更加明显。
cs.CV / 20 / 2607.00259
Multi-Hypothesis Test-Time Adaptation to Mitigate Underspecification
多假设测试时适应以减轻欠规范性
Abstract
Test-Time Adaptation (TTA) seeks to improve model robustness under distribution shifts by adapting parameters using unlabeled target data. However, in the absence of supervision, entropy-based adaptation is fundamentally underconstrained: multiple distinct parameter updates can achieve similarly low entropy while inducing drastically different decision boundaries. This phenomenon, known as underspecification, renders standard TTA brittle and prone to collapse into spurious modes. In this work, we reinterpret TTA through a posterior-inspired lens induced by entropy minimization, where low-entropy solutions define a pseudo-likelihood over parameters. Instead of committing to a single point estimate, we introduce a particle-based diversification framework that explores multiple plausible adaptation trajectories simultaneously. Our method can be viewed as a structured exploration of multiple plausible adaptation solutions, implemented through multi-level diversification at the output, parameter, optimizer, and input levels. Crucially, the framework acts as a plug-and-play wrapper compatible with existing TTA methods. Extensive experiments on challenging benchmarks demonstrate consistent gains in stability and robustness, achieving improvements of 3-4% under mixed shifts, 2-3% with batch size one, and 1-2.5% under label shifts, outperforming state-of-the-art baselines. Our results suggest that treating TTA as a multi-hypothesis inference problem, rather than a single-point optimization task, is key to mitigating underspecification and enabling reliable real-world deployment.
Chinese Translation
测试时适应（Test-Time Adaptation, TTA）旨在通过使用未标记的目标数据来调整参数，从而提高模型在分布变化下的鲁棒性。然而，在缺乏监督的情况下，基于熵的适应本质上是欠约束的：多个不同的参数更新可以实现相似的低熵，同时引发截然不同的决策边界。这种现象被称为欠规范性，使得标准的 TTA 变得脆弱，并容易陷入虚假的模式。在本研究中，我们通过熵最小化引发的后验启发视角重新解释 TTA，其中低熵解定义了参数上的伪似然。我们引入了一种基于粒子的多样化框架，同时探索多个合理的适应轨迹，而不是仅仅承诺于单一的点估计。我们的方法可以被视为对多个合理适应解的结构化探索，通过在输出、参数、优化器和输入层面的多级多样化来实现。重要的是，该框架作为一个即插即用的封装器，与现有的 TTA 方法兼容。在具有挑战性的基准测试上进行的大量实验表明，在混合变化下稳定性和鲁棒性的一致提升，改善幅度为 3-4%，在批量大小为一时为 2-3%，在标签变化下为 1-2.5%，超越了最先进的基线。我们的结果表明，将 TTA 视为多假设推断问题，而不是单点优化任务，是减轻欠规范性并实现可靠的现实世界部署的关键。
cs.CV / 21 / 2607.00273
MVDGC: Joint 3D and 2D Multi-view Pedestrian Detection via Dual Geometric Constraints
MVDGC：通过双重几何约束的联合三维和二维多视角行人检测
Abstract
The core challenge in multi-view pedestrian detection (MVPD) lies in effective aggregation of visual features from different viewpoints for robust occlusion reasoning. Recent approaches have addressed this by first projecting image-view features onto a Bird's Eye View (BEV) map, where ground localization is then performed. Despite impressive performance, the perspective transformation induces severe distortion, causing spatial structure break and degrading the quality of object feature extraction. The blurred and ambiguous features hinder accurate BEV point localization, especially in densely populated regions. Moreover, the strong mutual relationship between the BEV ground point and image bounding boxes is not capitalized on. Although multi-view consistency of 2D detections can serve as a powerful constraint in BEV space, these detections are commonly treated as auxiliary signals rather than being jointly optimized with the primary task.In this work, we propose \textbf{MVDGC}, a unified framework that \emph{jointly estimates pedestrian locations on the BEV plane and 2D bounding boxes in image views}. MVDGC employs a \emph{sparse set of 3D cylindrical queries} that embraces geometric context across both BEV and image views, enforcing dual spatial constraints for precise localization. Specifically, the geometric constraints is established by modeling each pedestrian as a vertical cylinder whose center lies on the BEV plane and whose projection casts a rectangular box in the image views. These queries function as shape anchors that directly extract 2D features from the intact image-view features using camera projection, eliminating projection-induced distortions. The 3D cylindrical query enables the unification of BEV and ImV localization into a single task: 3D cylinder position and shape refinement. Code is available at: https://github.com/UARK-AICV/MVDGC
Chinese Translation
多视角行人检测（MVPD）的核心挑战在于有效聚合来自不同视角的视觉特征，以实现稳健的遮挡推理。近期的方法通过首先将图像视图特征投影到鸟瞰图（BEV）地图上来解决这一问题，然后在该地图上进行地面定位。尽管取得了令人印象深刻的性能，但透视变换会引入严重的失真，导致空间结构的破坏，降低物体特征提取的质量。模糊和模棱两可的特征阻碍了准确的BEV点定位，尤其是在密集人群区域。此外，BEV地面点与图像边界框之间的强相互关系未得到充分利用。尽管二维检测的多视角一致性可以作为BEV空间中的强大约束，但这些检测通常被视为辅助信号，而不是与主要任务共同优化。在本研究中，我们提出了 extbf{MVDGC}，一个统一框架， extit{联合估计BEV平面上的行人位置和图像视图中的二维边界框}。MVDGC采用了一组 extit{稀疏的三维圆柱查询}，在BEV和图像视图之间拥抱几何上下文，强制实施双重空间约束以实现精确定位。具体而言，几何约束通过将每个行人建模为一个垂直圆柱体来建立，该圆柱体的中心位于BEV平面上，其投影在图像视图中形成一个矩形框。这些查询作为形状锚点，直接利用相机投影从完整的图像视图特征中提取二维特征，从而消除投影引起的失真。三维圆柱查询使BEV和图像视图的定位统一为一个单一任务：三维圆柱体的位置和形状优化。代码可在以下网址获取：https://github.com/UARK-AICV/MVDGC
cs.CV / 22 / 2607.00277
AEGIS: A Multi-Task Joint-Embedding Predictive Architecture for Mammography
AEGIS：一种用于乳腺X光检查的多任务联合嵌入预测架构
Abstract
We present Aegis, a joint-embedding predictive architecture for breast cancer detection and density assessment in mammography. We train three Vision Transformer variants (Small/Base/Large) using self-supervised joint-embedding predictive architecture (JEPA) pre-training on 71,103 studies from 14 clinical sites, followed by supervised fine-tuning with progressive resolution scaling up to 2048x1536. On a curated 785-study test set, our largest model achieves area under the receiver operating characteristic curve (AUC) 0.949 for breast cancer triage with 93% sensitivity and 75% specificity at the optimal operating point. An ensemble combining our model with a U.S. Food and Drug Administration-cleared baseline further improves discrimination to 0.952 AUC. For breast density classification, the model achieves 0.953 AUC for binary (dense vs. non-dense) classification and 62.6% exact accuracy across four Breast Imaging Reporting and Data System (BI-RADS) categories, with 98.8% adjacent accuracy comparable to reported human inter-reader agreement. External validation on the public VinDr-Mammo dataset provides evidence of cross-population transfer under a different reference standard, with the largest model achieving 0.871 AUC for triage in a zero-shot setting.
Chinese Translation
我们提出了Aegis，这是一种用于乳腺癌检测和密度评估的联合嵌入预测架构。我们使用自监督联合嵌入预测架构（JEPA）对来自14个临床站点的71,103个研究进行预训练，训练了三种视觉变换器变体（Small/Base/Large），随后进行监督微调，逐步将分辨率提升至2048x1536。在一个经过精心挑选的785个研究的测试集上，我们最大的模型在乳腺癌分诊中实现了接收者操作特征曲线下面积（AUC）为0.949，在最佳操作点下具有93%的灵敏度和75%的特异性。将我们的模型与美国食品药品监督管理局（FDA）批准的基线模型结合的集成方法进一步提高了区分能力，达到了0.952的AUC。在乳腺密度分类方面，该模型在二元（密集与非密集）分类中实现了0.953的AUC，并在四个乳腺影像报告和数据系统（BI-RADS）类别中达到了62.6%的准确率，具有98.8%的邻近准确率， comparable于报告的人类读者间一致性。在公共的VinDr-Mammo数据集上的外部验证提供了在不同参考标准下跨人群转移的证据，最大的模型在零样本设置下实现了0.871的AUC。
cs.CV / 23 / 2607.00289
OnPoint: Offline-to-Online Multi-Level Distillation for Point-Supervised Online Temporal Action Localization
OnPoint：用于点监督在线时序动作定位的离线到在线多级蒸馏
Abstract
Temporal Action Localization (TAL) typically relies on segment annotations or offline access to full videos, limiting scalability and online use. We introduce Point-Supervised Online TAL (POTAL), which localizes actions in streaming videos using only one temporal point per instance. To solve POTAL, we propose OnPoint, an offline-to-online multi-level distillation framework that transfers knowledge from a point-supervised offline teacher to an online student via (i) pseudo-segment instance distillation, (ii) class-activation sequence distillation, and (iii) anticipatory window-level distillation. We further improve robustness by incorporating the original point labels into student training and by refining anchor decoding with actionness-guided attention calibration. Experiments on five datasets show OnPoint consistently outperforms strong baselines, establishing a solid foundation for POTAL.
Chinese Translation
时序动作定位（TAL）通常依赖于片段注释或对完整视频的离线访问，这限制了其可扩展性和在线使用。我们提出了点监督在线TAL（POTAL），该方法仅使用每个实例的一个时间点在流媒体视频中定位动作。为了解决POTAL问题，我们提出了OnPoint，一个离线到在线的多级蒸馏框架，通过（i）伪片段实例蒸馏，（ii）类别激活序列蒸馏，以及（iii）预期窗口级蒸馏，将知识从点监督的离线教师转移到在线学生。我们进一步通过将原始点标签纳入学生训练以及通过动作性引导的注意力校准来优化锚点解码，从而提高鲁棒性。在五个数据集上的实验表明，OnPoint始终优于强基线，为POTAL奠定了坚实的基础。
cs.CV / 24 / 2607.00293
Rosetta: Composable Native Multimodal Pretraining
Rosetta：可组合的本地多模态预训练
Abstract
Achieving true artificial general intelligence requires foundation models capable of integrating new modalities without forgetting prior knowledge. However, accommodating continuous generative objectives alongside discrete understanding tasks causes severe gradient conflicts. Existing architectures, including standard Mixture-of-Experts (MoE), are highly susceptible to representation overwriting. Even structurally partitioned paradigms like Mixture-of-Transformers (MoT) remain vulnerable to catastrophic forgetting, severely impeding multimodal scalability. In this work, we introduce Rosetta, a composable native multimodal pretraining framework designed for seamless and non-destructive modality expansion. Rosetta adopts a modular paradigm where core foundational knowledge is preserved within global shared experts, while modality-specific capabilities are distributed across plug-and-play experts. To guarantee non-destructive composition, we propose Momentum-Anchored Orthogonal Projection (MAOP). MAOP leverages the optimizer's momentum state as an implicit semantic anchor, selectively neutralizing conflicting gradient components from new modalities while preserving synergistic updates. Extensive evaluations demonstrate that, while standard MoE and MoT architectures suffer catastrophic forgetting of previously acquired knowledge, Rosetta robustly preserves established language and visual understanding. Furthermore, it delivers superior image generation and unlocks cross-modal synergy, paving the way for truly composable and unified multimodal foundation models. To facilitate further multimodal research, we release our code and checkpoints to the community. Project page at https://rosetta-lmm.github.io/.
Chinese Translation
实现真正的人工通用智能需要基础模型能够在不遗忘先前知识的情况下整合新的模态。然而，连续生成目标与离散理解任务的共存会导致严重的梯度冲突。现有架构，包括标准的专家混合模型（Mixture-of-Experts, MoE），对表示的覆盖极为敏感。即使是结构上分区的范式，如混合变换器（Mixture-of-Transformers, MoT），也仍然容易遭受灾难性遗忘，严重阻碍了多模态的可扩展性。在本研究中，我们介绍了Rosetta，一个可组合的本地多模态预训练框架，旨在实现无缝且无损的模态扩展。Rosetta采用模块化范式，将核心基础知识保留在全球共享专家中，而模态特定的能力则分布在即插即用的专家之间。为了保证无损组合，我们提出了动量锚定正交投影（Momentum-Anchored Orthogonal Projection, MAOP）。MAOP利用优化器的动量状态作为隐式语义锚点，选择性地中和来自新模态的冲突梯度成分，同时保留协同更新。广泛的评估表明，尽管标准的MoE和MoT架构遭受了先前知识的灾难性遗忘，Rosetta却能够稳健地保留已建立的语言和视觉理解。此外，它提供了更优的图像生成能力，并解锁了跨模态的协同效应，为真正可组合和统一的多模态基础模型铺平了道路。为了促进进一步的多模态研究，我们向社区发布了我们的代码和检查点。项目页面：https://rosetta-lmm.github.io/
cs.CV / 25 / 2607.00296
Learning When to Listen: Gated Affect Fusion for Human Motion Prediction
学习何时倾听：用于人类运动预测的门控情感融合
Abstract
Human motion forecasting in unconstrained real-world videos remains challenging due to the ambiguity of future behaviors and the presence of noisy multimodal observations. While facial affect potentially provides complementary behavioral cues, its practical utility and mechanistic boundaries within motion forecasting frameworks remain poorly understood. In this work, we present a systematic study investigating the utility and temporal limitations of affect-conditioned forecasting in-the-wild. We establish a rigorous multimodal pipeline combining MediaPipe body pose trajectories with HSEmotion facial affect representations, and introduce the Gated Affect Transformer (GAT) to dynamically regulate cross-modal information flow. Through extensive multi-horizon evaluations under a strict subject-wise protocol, we demonstrate that naive early cross-modal concatenation consistently degrades forecasting accuracy relative to pose-only baselines. Conversely, our proposed gating mechanism stabilizes cross-modal integration by adaptively controlling the affective stream. Crucially, controlled counterfactual experiments using shuffled and randomized affect inputs reveal that the learned gate successfully suppresses unstructured cross-modal noise while remaining responsive to plausible affective signals. Furthermore, our empirical results indicate that facial affect features provide bounded, horizon-dependent predictive cues strictly within short-to-medium windows (e.g., 30 frames), whereas long-term trajectories remain predominantly governed by intrinsic kinematic continuity. Our findings provide empirical evidence that facial affect should be regarded as a complementary behavioral cue rather than a dominant driver of future motion, offering practical guidance for selective multimodal fusion in unconstrained human motion forecasting.
Chinese Translation
在不受限制的现实世界视频中，人类运动预测仍然具有挑战性，主要由于未来行为的模糊性和存在噪声的多模态观测。尽管面部情感可能提供互补的行为线索，但其在运动预测框架中的实际效用和机制边界仍然不甚明了。在本研究中，我们进行了系统研究，探讨了情感条件预测在实际场景中的效用和时间限制。我们建立了一个严格的多模态流程，将MediaPipe身体姿态轨迹与HSEmotion面部情感表示相结合，并引入了门控情感变换器（Gated Affect Transformer, GAT）以动态调节跨模态信息流。通过在严格的受试者协议下进行广泛的多时间段评估，我们证明了简单的早期跨模态拼接相对于仅使用姿态的基线模型始终会降低预测准确性。相反，我们提出的门控机制通过自适应控制情感流来稳定跨模态整合。重要的是，使用打乱和随机化情感输入的受控反事实实验表明，所学习的门控成功抑制了无结构的跨模态噪声，同时对合理的情感信号保持响应。此外，我们的实证结果表明，面部情感特征在短至中等时间窗口（例如30帧）内提供有限的、依赖于时间的预测线索，而长期轨迹则主要受内在运动连续性的支配。我们的研究结果提供了实证证据，表明面部情感应被视为互补的行为线索，而非未来运动的主导驱动因素，为在不受限制的人类运动预测中选择性多模态融合提供了实际指导。
cs.CV / 26 / 2607.00302
Wake up for Touch! Mask-isolated Tactile Alignment Learning in MLLMs
唤醒触觉！在多模态大语言模型中的掩码隔离触觉对齐学习
Abstract
Touch supplies the physical grounding needed to perceive intrinsic material properties, such as friction and compliance, that vision alone often cannot resolve. Recent efforts for equipping multimodal LLMs with this tactile sense, however, expose a zero-sum trade-off: the limited parameter budget of compact models forces a choice between acquiring the new sensory modality and preserving the established vision-language reasoning. We present Splash, a mask-isolated tactile alignment learning framework for MLLMs. Splash quantifies the significance of each pretrained parameter, and partitions the parameter space into a dormant and critical subspace. While the frozen critical subspace acts as a stable anchor to safeguard general visual knowledge, Splash updates the isolated dormant subspace to internalize tactile alignment towards LLMs. This selective, non-destructive expansion effectively prevents catastrophic forgetting and ensures non-destructive modality expansion. Extensive experiments show that Splash effectively achieves tactile reasoning without additional inference overhead in the LLM part, demonstrating state-of-the-art performance on visuo-tactile benchmarks, including SSVTP, TVL, and TacQuad, while preserving its original general-purpose capabilities.
Chinese Translation
触觉提供了感知内在材料属性（如摩擦和顺应性）所需的物理基础，而仅凭视觉往往无法解决这一问题。然而，最近为多模态大语言模型（MLLMs）赋予触觉感知的努力暴露出一种零和权衡：紧凑模型的有限参数预算迫使我们在获取新的感官模态和保持既有的视觉-语言推理之间做出选择。我们提出了Splash，一个用于多模态大语言模型的掩码隔离触觉对齐学习框架。Splash量化每个预训练参数的重要性，并将参数空间划分为休眠子空间和关键子空间。冻结的关键子空间作为稳定的锚点，以保护一般视觉知识，而Splash则更新隔离的休眠子空间，以内化触觉对齐到大语言模型中。这种选择性、非破坏性的扩展有效地防止了灾难性遗忘，并确保了非破坏性的模态扩展。大量实验表明，Splash能够有效实现触觉推理，而不会在大语言模型部分增加额外的推理开销，在视觉-触觉基准测试（如SSVTP、TVL和TacQuad）上展示了最先进的性能，同时保持其原有的通用能力。
cs.CV / 27 / 2607.00310
RetailSMV: Exocentric vs. Egocentric Adaptation of Foundation Video World Models in Retail
RetailSMV：零售中基础视频世界模型的外向适应与内向适应
Abstract
Foundation video diffusion models are increasingly viewed as world simulators for embodied agents, yet their pretraining on internet-scale generic video leaves them poorly aligned with real-world deployment domains. We study parameter-efficient adaptation of a pretrained foundation video world model to retail scenes: when synchronized egocentric and exocentric video of the same activity are available, which viewpoint of training data produces the strongest adapted model? We introduce RetailSMV (Retail Synchronized Multi-View), a corpus of 32,105 captioned retail clips from five supermarkets with synchronized ego/exo capture from the store-staff perspective (stocking, arranging, weighing, managing supply carts, scanning at checkout), rather than the customer-centric framing of prior retail video corpora, and train three matched Low-Rank Adaptation (LoRA) configurations of Cosmos3-Nano (egocentric-only, exocentric-only, combined) under identical hyperparameters. On a 200-clip held-out test set evaluated with seven complementary metrics under a strict paired statistical protocol, exocentric-only adaptation matches or exceeds combined adaptation on six of seven point estimates and is significantly better on LPIPS, PSNR, and DreamSim, despite training on only 15,985 exocentric clips (versus 32,105 for combined). A symmetric paired comparison further shows that adding exocentric data to egocentric-only training helps while adding egocentric data to exocentric-only training hurts. The absolute adaptation gap is largest at the shortest rollout time, identifying the near-horizon prediction window as the regime in which adaptation is most beneficial.
Chinese Translation
基础视频扩散模型越来越被视为具身代理的世界模拟器，但它们在互联网规模的通用视频上进行的预训练使其与真实世界的部署领域之间存在较大差距。我们研究了将预训练的基础视频世界模型有效适应于零售场景的参数效率：当可用的同步内向和外向视频展示同一活动时，哪种视角的训练数据能产生最强的适应模型？我们引入了RetailSMV（零售同步多视角），这是一个包含32,105个来自五家超市的带字幕零售片段的语料库，采用了从店员视角（补货、整理、称重、管理供应车、结账扫描）同步捕获的内向/外向视频，而不是以顾客为中心的先前零售视频语料库的框架，并在相同超参数下训练了三种匹配的低秩适应（Low-Rank Adaptation, LoRA）配置的Cosmos3-Nano（仅内向、仅外向、组合）。在一个包含200个片段的保留测试集上，使用七个互补指标在严格的配对统计协议下进行评估，结果显示仅外向适应在七个点估计中有六个与组合适应相匹配或超过，并且在LPIPS、PSNR和DreamSim上显著优于组合适应，尽管仅在15,985个外向片段上进行训练（而组合适应在32,105个片段上）。对称配对比较进一步表明，将外向数据添加到仅内向训练中有助于模型性能，而将内向数据添加到仅外向训练中则有害。适应差距在最短的滚动时间下最大，确定了近地平线预测窗口是适应最有利的状态。
cs.CV / 28 / 2607.00319
Typography-Based Monocular Distance Estimation for Advanced Driver-Assistance Systems
基于排版的单目距离估计用于高级驾驶辅助系统
Abstract
Estimating the distance to a leading vehicle is a basic input to forward collision warning, adaptive cruise control, and automated emergency braking. Production systems obtain this distance from radar, laser scanners, or stereo camera pairs, which add cost, power draw, and packaging constraints. This paper asks whether a single ordinary camera can recover the same distance by using a target that is standardized in size and present on every road vehicle: the rear license plate. U.S. plates share a fixed outer size and a character height that is set by regulation and varies only narrowly between states, so the height of a plate character in the image is a direct measure of distance once the camera geometry is known. The proposed method (Typography-Based Monocular Distance Estimation) detects the plate, measures the height of its printed characters, identifies the issuing state to select the correct physical character height, and recovers distance from the camera projection. Three measurements taken from the same plate: the character height, the stroke width, and the character spacing. Together with the spacing of the two mounting holes and a single-image depth network, are combined so that a weak or corrupted measurement is given less weight automatically. The distance, its rate of change, and a time-to-collision estimate are smoothed across frames and used to raise a warning with the timing used by U.S. collision-warning regulations. The same plate that anchors the scale also identifies the vehicle, so the method returns a distance, a bearing, and an identity from one passive sensor. It reads scale from a printed standard instead of from time of flight or parallax, making it a cheap, low-maintenance complement to those sensors in a fault-tolerant perception stack, achieving the cost-effective distance estimation with error less than 0.13 m.
Chinese Translation
估计与前方车辆的距离是前向碰撞警告、自适应巡航控制和自动紧急制动的基本输入。现有系统通过雷达、激光扫描仪或立体摄像头对这一距离进行获取，这增加了成本、功耗和包装限制。本文探讨是否可以通过使用一种在每辆道路车辆上标准化尺寸的目标——后车牌，利用单个普通摄像头恢复相同的距离。美国车牌具有固定的外部尺寸和由法规设定的字符高度，且在各州之间变化极小，因此一旦知道摄像头几何形状，图像中车牌字符的高度就可以直接作为距离的度量。所提出的方法（基于排版的单目距离估计）检测车牌，测量其印刷字符的高度，识别发牌州以选择正确的物理字符高度，并从摄像头投影中恢复距离。对同一车牌进行的三项测量：字符高度、笔画宽度和字符间距。结合两个安装孔的间距和单幅图像深度网络，自动对较弱或损坏的测量给予较少的权重。距离、其变化率和碰撞时间估计在帧间平滑处理，并用于根据美国碰撞警告法规的时机发出警告。锚定尺度的同一车牌也识别车辆，因此该方法从一个被动传感器返回距离、方位和身份。它从印刷标准中读取尺度，而不是通过飞行时间或视差，使其成为这些传感器在容错感知堆栈中的廉价、低维护补充，实现了误差小于0.13米的经济有效的距离估计。
cs.CV / 29 / 2607.00321
CORGI: Consistency-Aware 3D Dog Reconstruction from a Single Image in the Wild
CORGI：一种基于单幅自然图像的考虑一致性的3D狗重建方法
Abstract
Reconstructing high-fidelity 3D models of highly articulated animals, such as dogs, from a single in-the-wild image remains a formidable challenge. In this paper, we introduce CORGI, a novel framework for consistency-aware 3D dog reconstruction from a single unconstrained image that completely eliminates the need for 3D supervision. To overcome generative inconsistencies and the lack of multi-view capture, our pipeline introduces three core components. First, we propose a Canonical-Driven Orbital Generation (CDOG) strategy, utilizing specialized Canonical and Orbit LoRAs to normalize arbitrary input poses and synthesize reliable 360-degree video observations. Second, we design a Consistency-aware Deformable 3DGS (CA-3DGS) module that anchors on a D-SMAL prior, explicitly modeling per-view generative errors through dedicated neural deformation fields to learn accurate vertex-level displacements. Finally, to eliminate structural distortions and recover high-frequency details, we introduce a self-supervised Deformation-Conditioned Generative Repair (DCGR) module. Extensive experiments demonstrate that CORGI achieves state-of-the-art performance, generalizing seamlessly across diverse dog breeds to produce geometrically accurate, visually coherent, and fully animatable 3D assets ready for downstream applications.
Chinese Translation
从单幅自然图像重建高度关节化动物（如狗）的高保真3D模型仍然是一项艰巨的挑战。本文介绍了CORGI，一个新颖的框架，用于从单幅无约束图像进行考虑一致性的3D狗重建，完全消除了对3D监督的需求。为了克服生成不一致性和缺乏多视角捕捉的问题，我们的流程引入了三个核心组件。首先，我们提出了一种基于典范驱动的轨道生成（Canonical-Driven Orbital Generation, CDOG）策略，利用专门的典范和轨道低秩适应（LoRA）来规范任意输入姿态，并合成可靠的360度视频观测。其次，我们设计了一个考虑一致性的可变形3D生成系统（Consistency-aware Deformable 3DGS, CA-3DGS）模块，该模块基于D-SMAL先验，明确建模每视角生成误差，通过专用神经变形场学习准确的顶点级位移。最后，为了消除结构扭曲并恢复高频细节，我们引入了一个自监督的变形条件生成修复（Deformation-Conditioned Generative Repair, DCGR）模块。大量实验表明，CORGI在性能上达到了最先进水平，能够无缝地在不同犬种之间进行泛化，生成几何准确、视觉一致且完全可动画的3D资产，适用于下游应用。
cs.CV / 30 / 2607.00338
DroneFINE: Domain-Aware Parameter-Efficient Fine-Tuning of Vision-Language Detectors for Drone Images
DroneFINE：面向领域的参数高效微调无人机图像视觉语言检测器
Abstract
Object detection for Unmanned Aerial Vehicles (UAVs) working in open and dynamic environments is a highly challenging task. While Vision-Language Models (VLMs) have offered a powerful solution for universal object detection, adapting them to UAV scenarios remains non-trivial due to a substantial domain gap between VLM pre-training data and aerial imagery. The prevailing Parameter-Efficient Fine-Tuning (PEFT) methods prove ineffective in bridging this gap, as VLMs' "natural-scene, foreground-dominant" visual priors misalign with the "bird's-eye-view, background-dominant, small-object" characteristics of UAV data. To address this issue, we propose DroneFINE, a novel PEFT paradigm comprising two domain-aware complementary modules tailored for VLM-based drone image detectors. Specifically, a data-dependent, foreground-aware, and multi-path adaptation mechanism named HyperAdapter is designed, which overcomes the static structural constraints of PEFT. In addition, a background suppression algorithm named SemanticGate is developed. It is a text-conditioned guidance strategy that employs background vocabulary to actively guide the model in suppressing responses from irrelevant regions. Extensive experiments on VisDrone and UAVDT demonstrate that DroneFINE significantly outperforms existing PEFT methods and achieves performance comparable to full fine-tuning while substantially reducing the number of trainable parameters.
Chinese Translation
在开放和动态环境中，无人机（UAV）进行物体检测是一项极具挑战性的任务。尽管视觉语言模型（VLMs）为通用物体检测提供了强大的解决方案，但由于VLM预训练数据与航空图像之间存在显著的领域差距，将其适应于无人机场景仍然不是一件简单的事情。现有的参数高效微调（PEFT）方法在弥合这一差距方面效果不佳，因为VLM的“自然场景、前景主导”视觉先验与无人机数据的“鸟瞰图、背景主导、小物体”特征不匹配。为了解决这一问题，我们提出了DroneFINE，这是一种新颖的PEFT范式，包含两个面向领域的互补模块，专为基于VLM的无人机图像检测器量身定制。具体而言，我们设计了一种名为HyperAdapter的数据依赖、前景感知和多路径适应机制，克服了PEFT的静态结构限制。此外，我们开发了一种名为SemanticGate的背景抑制算法。这是一种文本条件指导策略，利用背景词汇主动引导模型抑制来自无关区域的响应。在VisDrone和UAVDT上的大量实验表明，DroneFINE显著优于现有的PEFT方法，并实现了与全微调相当的性能，同时大幅减少了可训练参数的数量。
cs.CV / 31 / 2607.00357
Personalized Object Identification and Localization via In-Context Inference with Vision-Language Models
通过上下文推理与视觉-语言模型实现个性化物体识别与定位
Abstract
Personalized object localization (POL) localizes an object instance in a query image based on a few reference images with bounding-box annotations and a target object label. The pioneering method, IPLoc, solves this task through in-context inference with vision-language models (VLMs). However, it assumes that the query image always contains the target object. This assumption severely limits its applicability to real-world scenarios with many irrelevant images. To address this issue, we formulate a new task, personalized object identification and localization (POIL), by positioning POL within the broader few-shot object detection framework. POIL aims to localize the target object instance while rejecting query images that do not contain the reference object instance. We also present POIL datasets constructed from public sources. We further propose an in-context algorithm named IPLoc-ID for solving POIL with VLMs. IPLoc-ID first predicts a candidate bounding box and then determines whether it corresponds to the reference object instance. We introduce a self-posed query to connect these two steps within a single autoregressive generation framework. Through ablation studies and comprehensive experiments, we show that IPLoc-ID substantially suppresses false-positive detections on negative query images while maintaining localization performance comparable to IPLoc. Overall, IPLoc-ID effectively addresses the practical instance-level POIL task, which cannot be sufficiently solved by conventional object detection, few-shot object detection, or the localization-only IPLoc method.
Chinese Translation
个性化物体定位（POL）是在查询图像中基于少量带有边界框注释的参考图像和目标物体标签来定位物体实例的一种方法。开创性的方法 IPLoc 通过与视觉-语言模型（VLMs）的上下文推理来解决这一任务。然而，它假设查询图像始终包含目标物体。这一假设严重限制了其在许多无关图像的现实场景中的适用性。为了解决这个问题，我们在更广泛的少样本物体检测框架内，提出了一个新任务，即个性化物体识别与定位（POIL），其目标是在拒绝不包含参考物体实例的查询图像的同时，定位目标物体实例。我们还展示了从公共来源构建的 POIL 数据集。进一步地，我们提出了一种名为 IPLoc-ID 的上下文算法，用于利用 VLMs 解决 POIL。IPLoc-ID 首先预测一个候选边界框，然后确定它是否对应于参考物体实例。我们引入了一个自我引导的查询，将这两个步骤连接在一个单一的自回归生成框架内。通过消融研究和全面实验，我们表明 IPLoc-ID 在负查询图像上显著抑制了误报检测，同时保持了与 IPLoc 相当的定位性能。总体而言，IPLoc-ID 有效地解决了实际的实例级 POIL 任务，而传统的物体检测、少样本物体检测或仅定位的 IPLoc 方法无法充分解决该任务。
cs.CV / 32 / 2607.00369
SFDATrack: Generalized Source-Free Domain Adaptive Tracking Under Adverse Weather Conditions
SFDATrack：在恶劣天气条件下的广义无源领域自适应跟踪
Abstract
Domain adaptive visual object tracking under adverse weather conditions has garnered significant attention in recent years. Despite the impressive performance, existing methods heavily rely on the large-scale video frames from both source and target domains, which is impractical under rigid resource constraints where source data is unavailable. To overcome this limitation, we propose SFDATrack, a generalized source-free domain adaptive tracker that merely leverages adverse weather samples from the target domain for robust state estimation. Specifically, SFDATrack first employs a mean-teacher backbone with Dual Interactive Mamba (DIM) blocks to distill the candidate target tokens that are resilient to weather variations from classified, augmented samples. Afterwards, we introduce a hyperspherical prototype projection (HPP) module to project these tokens onto multi-domain prototypes within a latent hyperspherical space. By enforcing both domain-specific and domain-invariant properties of the multi-domain prototypes, SFDATrack can be seamlessly adapted to diverse weather conditions with powerful generalizability. Extensive experiments evaluated on various benchmarks demonstrate that SFDATrack achieves superior performance compared to state-of-the-art approaches. The code is available at https://github.com/watcherBR0/sfdatrack.
Chinese Translation
近年来，在恶劣天气条件下的领域自适应视觉目标跟踪引起了广泛关注。尽管现有方法表现出色，但它们在很大程度上依赖于来自源领域和目标领域的大规模视频帧，这在源数据不可用的严格资源限制下是不切实际的。为了解决这一限制，我们提出了SFDATrack，一种广义无源领域自适应跟踪器，仅利用目标领域中的恶劣天气样本进行稳健的状态估计。具体而言，SFDATrack首先采用具有双交互Mamba（DIM）模块的均值教师骨干网络，从分类的增强样本中提取对天气变化具有鲁棒性的候选目标标记。随后，我们引入了一个超球面原型投影（HPP）模块，将这些标记投影到潜在超球面空间中的多领域原型上。通过强制多领域原型的领域特定和领域不变属性，SFDATrack能够无缝适应多种天气条件，具有强大的泛化能力。在各种基准测试中进行的广泛实验表明，SFDATrack的性能优于最先进的方法。代码可在 https://github.com/watcherBR0/sfdatrack 获取。
cs.CV / 33 / 2607.00371
MEPA: Multi-Scale Representation Alignment for Visual Autoregressive Modeling with Mixture of Experts
MEPA：基于专家混合的视觉自回归建模的多尺度表示对齐
Abstract
Visual AutoRegressive modeling (VAR) has pioneered a coarse-to-fine multi-scale autoregressive generative paradigm, demonstrating strong capabilities in image generation. However, VAR still suffers from inherent deficiencies in multi-scale representation learning. Specifically, lower scales primarily capture global semantics, while higher scales focus on fine-grained details. Employing a shared architecture across scales induces optimization conflicts. Moreover, due to the causal autoregressive process, inaccurate semantics at early scales can propagate and significantly degrade the final output. To address these issues, we introduce a scale-aware token-routed Mixture of Experts (MoE) architecture, allowing scale-adaptive expert selection, thereby facilitating decoupled representation learning across scales. In addition, we enhance semantic modeling at early scales by incorporating external self-supervised features. Unlike naive alignment, we analyse and design a residual feature aggregation scheme tailored to the VAR paradigm. Extensive experiments show that our method significantly improves both training efficiency and generation quality. On the ImageNet 256*256 benchmark, our model achieves a superior FID compared to the dense baseline while requiring only half of the default training epochs and a smaller parameter budget, with a merely marginal increase in training cost. Moreover, the performance gap further widens with larger training epochs.
Chinese Translation
视觉自回归建模（VAR）开创了一种粗到细的多尺度自回归生成范式，在图像生成方面展现了强大的能力。然而，VAR在多尺度表示学习方面仍然存在固有的不足。具体而言，低尺度主要捕捉全局语义，而高尺度则关注细粒度细节。在各尺度之间采用共享架构会导致优化冲突。此外，由于因果自回归过程，早期尺度的不准确语义可能会传播并显著降低最终输出的质量。为了解决这些问题，我们提出了一种尺度感知的令牌路由专家混合（MoE）架构，允许尺度自适应的专家选择，从而促进各尺度之间的解耦表示学习。此外，我们通过引入外部自监督特征来增强早期尺度的语义建模。与简单的对齐方法不同，我们分析并设计了一种针对VAR范式的残差特征聚合方案。大量实验表明，我们的方法显著提高了训练效率和生成质量。在ImageNet 256*256基准测试中，我们的模型相比于稠密基线取得了更优的FID，同时仅需一半的默认训练周期和更小的参数预算，训练成本仅有微小增加。此外，随着训练周期的增加，性能差距进一步扩大。
cs.CV / 34 / 2607.00374
Learning to Compose: Revisiting Proxy Task Design for Zero-Shot Composed Image Retrieval
学习组合：重新审视零-shot组合图像检索的代理任务设计
Abstract
Composed Image Retrieval (CIR) retrieves a target image from a reference image and a textual modification. While supervised CIR relies on costly triplets, Zero-Shot CIR (ZS-CIR) alleviates this reliance through proxy tasks trained on image-text pairs. However, existing proxy tasks primarily enhance visual and textual representations to accommodate a predefined composition mechanism such as pseudo-word injection into a frozen text encoder or linear feature arithmetic. As a result, the composition function itself remains unlearned, limiting the model's ability to express diverse and fine-grained semantic modifications. To address this, we propose FoCo, which models composition as two coordinated stages: focusing on modification-relevant visual content, and then completing the target semantics. We realize these through two proxy tasks: text-anchored visual aggregation to selectively gather visual content guided by localized textual semantics, and context-conditioned semantic completion to transform these aggregated visuals with the remaining scene context into a coherent composed representation. The tasks are trained jointly with a cross-instance contrastive objective, encouraging semantic diversity and discouraging shortcut composition strategies. Extensive experiments on four ZS-CIR benchmarks show FoCo's state-of-the-art performance and improved generalization.
Chinese Translation
组合图像检索（Composed Image Retrieval, CIR）从参考图像和文本修改中检索目标图像。虽然监督式CIR依赖于昂贵的三元组，但零-shot CIR（Zero-Shot CIR, ZS-CIR）通过在图像-文本对上训练的代理任务减轻了这种依赖。然而，现有的代理任务主要增强视觉和文本表示，以适应预定义的组合机制，例如将伪词注入冻结的文本编码器或线性特征算术。因此，组合函数本身仍然未被学习，限制了模型表达多样化和细粒度语义修改的能力。为了解决这一问题，我们提出了FoCo，它将组合建模为两个协调阶段：首先关注与修改相关的视觉内容，然后完成目标语义。我们通过两个代理任务实现这些目标：文本锚定的视觉聚合，以选择性地收集由局部文本语义引导的视觉内容，以及上下文条件的语义完成，将这些聚合的视觉内容与剩余场景上下文转化为一致的组合表示。这些任务通过跨实例对比目标共同训练，鼓励语义多样性并抑制捷径组合策略。在四个ZS-CIR基准上的广泛实验表明，FoCo在性能上达到了最先进水平，并改善了泛化能力。
cs.CV / 35 / 2607.00375
LIST3R: Long-sequence Instance-aware 3D Reconstruction
LIST3R：长序列实例感知的三维重建
Abstract
We present LIST3R, an instance-aware framework for long-sequence 3D reconstruction inspired by the way humans organize spatial memory around stable and recognizable objects. LIST3R organizes long-sequence reconstruction around instance anchors, using them to reconnect fragmented subsequences and consolidate local observations into a coherent global 3D scene. Given a long video, our approach partitions it into overlapping subsequences and builds a structured local instance library for each partial reconstruction, maintaining persistent trackable anchors with semantic and geometric evidence. These anchors are matched across subsequences to recover revisited regions and provide object-aware constraints for fragment alignment, producing a consistent global reconstruction. During this process, the evolving geometric evidence updates the local instance libraries and progressively organizes them into a unified global 3D instance library. Experiments on long-sequence benchmarks show that our method produces more accurate trajectories and higher-quality 3D reconstructions, highlighting the effectiveness of persistent instance anchors for organizing long-horizon 3D reconstruction. Our code is available on the project page: https://yixn965.github.io/LIST3R/.
Chinese Translation
我们提出了LIST3R，这是一种实例感知的长序列三维重建框架，灵感来源于人类如何围绕稳定且可识别的物体组织空间记忆。LIST3R围绕实例锚点组织长序列重建，利用这些锚点重新连接碎片化的子序列，并将局部观测整合成一个连贯的全局三维场景。给定一段长视频，我们的方法将其划分为重叠的子序列，并为每个部分重建构建一个结构化的局部实例库，保持具有语义和几何证据的持久可追踪锚点。这些锚点在子序列之间进行匹配，以恢复重访区域，并为碎片对齐提供对象感知约束，从而生成一致的全局重建。在此过程中，演变的几何证据更新局部实例库，并逐步将其组织成一个统一的全局三维实例库。在长序列基准测试中的实验表明，我们的方法产生了更准确的轨迹和更高质量的三维重建，突显了持久实例锚点在组织长时间跨度三维重建中的有效性。我们的代码可在项目页面获取：https://yixn965.github.io/LIST3R/
cs.CV / 36 / 2607.00378
Radial Interaction Tomography: Recognizing Non-Transitive Evolutionary Games from One Range-Expansion Image
径向交互层析：从单个范围扩展图像识别非传递性进化博弈
Abstract
Colored sectors in a microbial range expansion encode more than lineage survival counts. We formulate a computer-vision inverse problem: from one endpoint image of an accretive multi-type expansion, recover the radius-indexed pairwise boundary-flow field and test whether the visual pattern is compatible with a transitive scalar fitness hierarchy. The observable is a geometric signal extracted from sector-boundary curves in log-polar coordinates. We prove endpoint observability and stability for frozen fronts, weighted transitive/cyclic decomposition, contact-complete circular design, physical-clock and mechanism non-identifiability, exact Gaussian cyclicity testing, and Bonferroni-valid interval scanning. The benchmark is deterministic: analytic endpoint images, blurred/noisy pixel round trips, scalar-null stress tests, public-image tracing, multi-resolution mechanistic endpoints, and a non-learning frozen-front simulator. The implementation recovers pairwise edge-flow histories from endpoint images, detects cyclic residuals in a mechanistic four-type expansion, and uses those residuals as forcing signals for a dimensionless active design-control layer covering reaction-diffusion control, phenotype-frontier optimization, protocol synthesis, Monte Carlo robustness, and a downstream population-state bridge.
Chinese Translation
微生物范围扩展中的彩色区域编码的不仅仅是谱系生存计数。我们提出一个计算机视觉逆问题：从一个增生多类型扩展的端点图像中，恢复半径索引的成对边界流场，并测试视觉模式是否与传递性标量适应性层次结构兼容。可观察量是从对数极坐标中的区域边界曲线提取的几何信号。我们证明了冻结前沿的端点可观察性和稳定性，加权的传递/循环分解，接触完全的圆形设计，物理时钟和机制不可识别性，精确的高斯循环性测试，以及Bonferroni有效的区间扫描。基准是确定性的：解析的端点图像，模糊/噪声像素的往返，标量零应力测试，公共图像追踪，多分辨率机制端点，以及一个非学习的冻结前沿模拟器。该实现从端点图像中恢复成对边缘流历史，检测机械四类型扩展中的循环残差，并将这些残差用作无量纲主动设计控制层的强迫信号，涵盖反应-扩散控制、表型前沿优化、协议合成、蒙特卡洛鲁棒性，以及下游种群状态桥接。
cs.CV / 37 / 2607.00382
Vitality-Aware Compression for Efficient Image-to-Shape Diffusion Transformers
面向活力的压缩方法用于高效的图像到形状扩散变换器
Abstract
We propose the first compression approach for image-to-shape Diffusion Transformers (DiTs) that substantially reduces model size while preserving geometric fidelity. Despite remarkable progress in 3D shape generation, large DiT-based models remain computationally prohibitive in resource-constrained settings. Furthermore, it is difficult to directly transfer existing diffusion model compression strategies developed for different domains to 3D generation, and prior 3D efficiency approaches focus primarily on inference speed rather than backbone compression. To address this limitation, we build a geometry-aware compression framework tailored to image-to-shape DiTs. Guided by the observation that 3D DiT layers exhibit non-uniform importance for geometry synthesis, we introduce a vitality-guided framework integrating structured pruning, adaptive quantization, and targeted fine-tuning. Our method achieves up to 66% model-size reduction across state-of-the-art image-to-3D models while maintaining synthesis fidelity comparable to full-sized counterparts. This highlights the potential of our framework as a plug-and-play solution for efficient 3D shape generation across diverse models.
Chinese Translation
我们提出了首个针对图像到形状扩散变换器（Diffusion Transformers, DiTs）的压缩方法，该方法在显著减少模型大小的同时保持几何保真度。尽管在3D形状生成方面取得了显著进展，但基于DiT的大型模型在资源受限的环境中仍然计算成本高昂。此外，现有的针对不同领域开发的扩散模型压缩策略难以直接转移到3D生成任务中，且以往的3D效率方法主要集中在推理速度上，而非主干网络的压缩。为了解决这一限制，我们构建了一个针对图像到形状DiTs的几何感知压缩框架。基于3D DiT层在几何合成中表现出非均匀重要性的观察，我们引入了一个活力引导框架，整合了结构化剪枝、自适应量化和针对性微调。我们的方法在保持与全尺寸模型相当的合成保真度的同时，实现了高达66%的模型大小减少，突显了我们框架作为高效3D形状生成的即插即用解决方案的潜力，适用于多种模型。
cs.CV / 38 / 2607.00399
DriveVer: Lightweight Trajectory Evaluator as Test-Time Verifier for Autonomous Driving
DriveVer：作为自主驾驶测试时验证器的轻量级轨迹评估器
Abstract
End-to-end autonomous driving models often encounter performance bottlenecks, as training-time scaling leads to high computational costs and diminishing marginal returns. Existing planners typically adopt a one-shot generation paradigm, lacking secondary validation and active correction mechanisms to detect and revise suboptimal or unsafe trajectories during inference. To address this issue, we propose DriveVer, a lightweight, plug-and-play Test-Time Verifier that leverages the test-time scaling paradigm to enable autonomous driving systems to validate and refine trajectories without costly and heavy training. We construct a dedicated trajectory dataset based on the NAVSIM benchmark through condition-driven clustering and balanced sampling according to ego-vehicle states and navigation commands. Employing a dual-head architecture, DriveVer efficiently fuses candidate trajectories with multi-view visual representations and ego-vehicle kinematic features to simultaneously predict a safety confidence score and an absolute geometric refinement vector. Extensive experiments on the NAVSIM benchmark show that DriveVer significantly improves the performance of base planning models. Notably, as an extremely compact model with only 34M parameters, DriveVer introduces minimal computational overhead, achieving competitive results while maintaining real-time inference efficiency.
Chinese Translation
端到端的自主驾驶模型常常面临性能瓶颈，因为训练时的扩展导致高计算成本和边际收益递减。现有的规划器通常采用一次性生成范式，缺乏二次验证和主动纠正机制，以便在推理过程中检测和修正次优或不安全的轨迹。为了解决这个问题，我们提出了DriveVer，一种轻量级、即插即用的测试时验证器，利用测试时扩展范式，使自主驾驶系统能够在不进行昂贵且复杂的训练的情况下验证和优化轨迹。我们基于NAVSIM基准构建了一个专用的轨迹数据集，通过基于条件的聚类和根据自车状态及导航指令的平衡抽样。DriveVer采用双头架构，有效融合候选轨迹与多视角视觉表示及自车运动特征，以同时预测安全置信度分数和绝对几何优化向量。在NAVSIM基准上的大量实验表明，DriveVer显著提升了基础规划模型的性能。值得注意的是，作为一个仅有3400万参数的极其紧凑的模型，DriveVer引入的计算开销极小，在保持实时推理效率的同时实现了具有竞争力的结果。
cs.CV / 39 / 2607.00402
The Illusion of High Utility in Safety Alignment of Text-to-Image Diffusion Models
文本到图像扩散模型安全对齐中的高效用错觉
Abstract
Safety alignment of text-to-image (T2I) diffusion models aims to suppress harmful generations while preserving utility on benign prompts. Recent methods often appear to deliver high safety with high utility, but this conclusion rests largely on coarse global utility metrics (e.g., FID, CLIPScore) that are insensitive to fine-grained semantic correctness, creating an illusion of high utility. We show that when utility is measured with structured evaluation, this illusion breaks: on TIFA (Text-to-Image Faithfulness evaluation with Question Answering), safety-aligned models suffer substantial drops in semantic fidelity, including failures in object counts, attributes, and relationships. To diagnose the source of this gap, we analyze the text-encoder prompt embedding space and uncover semantic collapse, a contraction of embedding spread coupled with distortion of inter-prompt similarity structure, which strongly correlates with structured utility loss. Guided by this insight, we propose StructureAware Geometric Regularization (SAGE), a safety alignment objective that explicitly preserves embedding spread and inter-prompt relational structure during adaptation. Our method restores structured utility (TIFA +5.0% over prior state-of-the-art) while maintaining strong safety performance and competitive coarse-grained utility scores. Our source code and trained models are available at https://adeelyousaf.github.io/SAGE_ECCV26_Project_Page/.
Chinese Translation
文本到图像（T2I）扩散模型的安全对齐旨在抑制有害生成，同时在良性提示上保持效用。近期的方法往往看似能够在高效用的同时实现高安全性，但这一结论在很大程度上依赖于对细粒度语义正确性不敏感的粗略全局效用指标（例如，FID，CLIPScore），从而造成了高效用的错觉。我们展示了当效用通过结构化评估进行测量时，这种错觉会破裂：在 TIFA（文本到图像的忠实性评估与问答）上，安全对齐模型在语义忠实性上遭遇了显著下降，包括在对象计数、属性和关系上的失败。为了诊断这一差距的来源，我们分析了文本编码器提示嵌入空间，并发现了语义崩溃，即嵌入分布的收缩与提示间相似性结构的扭曲，这与结构化效用损失有很强的相关性。在这一见解的指导下，我们提出了结构感知几何正则化（SAGE），这是一种安全对齐目标，明确在适应过程中保持嵌入分布和提示间关系结构。我们的方法恢复了结构化效用（在之前的最先进技术上提升了 TIFA +5.0%），同时保持了强大的安全性能和具有竞争力的粗粒度效用评分。我们的源代码和训练模型可在 https://adeelyousaf.github.io/SAGE_ECCV26_Project_Page/ 获取。
cs.CV / 40 / 2607.00409
MedCAGD: Context-Aware Gated Decoder for Efficient Medical Image Segmentation
MedCAGD：用于高效医学图像分割的上下文感知门控解码器
Abstract
Medical image segmentation relies on the ability of encoder-decoder architectures to translate rich feature representations into accurate pixel-level predictions under challenging conditions such as low contrast, structural ambiguity, and scale variability. While recent advances in large-scale pretraining and transformer-based encoders have substantially improved feature extraction, segmentation accuracy remains constrained by decoder design, particularly in terms of cross-scale alignment, contextual integration, and boundary preservation. In this work, we revisit medical image segmentation from a decoder-centric perspective and propose a context-aware gated decoder that systematically regulates feature fusion and contextual aggregation throughout the decoding process. The proposed decoder integrates lightweight multi-scale channel recalibration, gated skip fusion with spatial competition and a global context aggregation mechanism that injects encoder-wide information into intermediate decoding stages. This design enables effective translation of strong pretrained encoder representations into spatially consistent predictions. Extensive experiments across 11 medical image segmentation benchmarks validate the effectiveness and demonstrate that the proposed approach consistently outperforms strong baselines while remaining computationally practical. Code: https://github.com/saadwazir/MedCAGD
Chinese Translation
医学图像分割依赖于编码器-解码器架构将丰富的特征表示转化为在低对比度、结构模糊和尺度变化等挑战性条件下的准确像素级预测。尽管近期在大规模预训练和基于变换器的编码器方面的进展显著提升了特征提取能力，但解码器设计仍然限制了分割精度，特别是在跨尺度对齐、上下文整合和边界保持方面。在本研究中，我们从解码器中心的视角重新审视医学图像分割，并提出了一种上下文感知的门控解码器，该解码器系统性地调节特征融合和上下文聚合的过程。所提解码器集成了轻量级的多尺度通道重校准、带有空间竞争的门控跳跃融合以及一种全局上下文聚合机制，该机制将编码器范围内的信息注入到中间解码阶段。该设计有效地将强大的预训练编码器表示转化为空间一致的预测。通过在11个医学图像分割基准上的广泛实验验证了该方法的有效性，并展示了所提方法在保持计算实用性的同时始终优于强基线。代码链接：https://github.com/saadwazir/MedCAGD
cs.CV / 41 / 2607.00410
MindAU: EEG-Conditioned Facial Action Unit Editing via Dual-Stream Manifold Alignment
MindAU：基于脑电图（EEG）条件的面部动作单元编辑通过双流流形对齐
Abstract
Recent brain decoding studies have made substantial progress in reconstructing externally perceived visual content from neural signals. However, using electroencephalography (EEG) recordings to guide facial expression editing remains largely unexplored and poses a distinct challenge: rather than recovering what a subject sees, it requires identifying facial-action related patterns from noisy EEG signals and grounding them in localized, identity-preserving expression edits. In this paper, we investigate EEG-conditioned facial image editing for fine-grained facial action unit (AU) control and propose MindAU, a unified framework for controlling facial AU edits from EEG signals. MindAU first learns noise-robust and AU-discriminative EEG representations through temporal masked reconstruction and AU classification supervision. It then bridges the modality gap via Dual-Stream Manifold Alignment, aligning EEG features with AU-level text semantics and identity-reduced visual displacement trajectories in the multimodal space of Qwen2.5-VL. Finally, MindAU incorporates EEG-aware Multimodal Rotary Positional Embeddings, landmark-guided reference masking, and AU-aware region supervision into a multimodal diffusion-based editor for high-fidelity identity-preserving editing. We also introduce E-CAFE, a curated benchmark for EEG-Conditioned Action-Unit Facial Editing with paired EEG-face editing samples and standardized evaluation protocols. Extensive experiments demonstrate the effectiveness of MindAU and suggest its potential as a step towards future assistive expression technologies for individuals with facial neuromuscular disorders.
Chinese Translation
近期的脑解码研究在从神经信号重建外部感知的视觉内容方面取得了显著进展。然而，利用脑电图（EEG）记录指导面部表情编辑仍然在很大程度上未被探索，并且面临着一个独特的挑战：它不仅需要恢复受试者所见的内容，还需要从嘈杂的EEG信号中识别与面部动作相关的模式，并将其与局部的、保留身份的表情编辑相结合。在本文中，我们研究了基于EEG的面部图像编辑，以实现细粒度的面部动作单元（AU）控制，并提出了MindAU，一个统一的框架，用于从EEG信号控制面部AU编辑。MindAU首先通过时间掩蔽重建和AU分类监督学习噪声鲁棒和AU区分的EEG表示。然后，通过双流流形对齐（Dual-Stream Manifold Alignment）弥合模态差距，将EEG特征与AU级文本语义和在Qwen2.5-VL多模态空间中的身份减少的视觉位移轨迹对齐。最后，MindAU将EEG感知的多模态旋转位置嵌入、基于地标的参考掩蔽和AU感知的区域监督整合到一个基于多模态扩散的编辑器中，以实现高保真度的身份保留编辑。我们还引入了E-CAFE，一个为EEG条件的动作单元面部编辑而策划的基准，包含配对的EEG-面部编辑样本和标准化的评估协议。大量实验表明MindAU的有效性，并暗示其作为未来面部神经肌肉疾病患者辅助表情技术的一步的潜力。
cs.CV / 42 / 2607.00416
DroneIQA-VLE: Multi-Task Drone Image Quality Assessment via Vision-Language Ensemble
DroneIQA-VLE：通过视觉-语言集成进行多任务无人机图像质量评估
Abstract
We present DroneIQA-VLE, our solution to the ICME 2026 Drone-IQA Grand Challenge on Target-aware Image Quality Assessment for Low-altitude UAV Images. The framework jointly predicts global, target, and background quality scores by ensembling two complementary pipelines: (1) SigLIP2 vision encoders with multi-task regression heads, and (2) a LoRA-adapted Qwen3.5-9B multimodal large language model for quality score regression. The final global quality prediction is obtained by arithmetically averaging the outputs of both pipelines. Our method achieves 2nd place in the challenge, demonstrating its effectiveness. The code is available at https://github.com/sunwei925/DroneIQA-VLE.
Chinese Translation
我们提出了DroneIQA-VLE，这是我们针对ICME 2026无人机图像质量评估（Drone-IQA）大挑战的解决方案，旨在针对低空无人机图像进行目标感知的图像质量评估。该框架通过集成两个互补的管道，联合预测全局、目标和背景质量评分：（1）具有多任务回归头的SigLIP2视觉编码器，以及（2）经过LoRA适配的Qwen3.5-9B多模态大型语言模型用于质量评分回归。最终的全局质量预测是通过对两个管道的输出进行算术平均得到的。我们的方法在挑战中获得了第二名，展示了其有效性。代码可在https://github.com/sunwei925/DroneIQA-VLE获取。
cs.CV / 43 / 2607.00417
EO-VGGT: Orbital Ray-Conditioned 3D Foundation Models for Satellite Multi-View Reconstruction
EO-VGGT：用于卫星多视角重建的轨道光线条件3D基础模型
Abstract
In the era of satellite constellations, multi-view optical satellite imagery is pivotal for Earth Observation (EO) and high-quality Digital Surface Model (DSM) reconstruction. Although feed-forward 3D foundation models have transformed computer vision, their deployment in satellite remote sensing is inherently constrained by the structural discrepancy between implicit perspective assumptions and explicit orbital pushbroom geometry. This geometric incongruity is further compounded by pronounced view-set heterogeneity. We present EO-VGGT, a framework that adapts a frozen perspective-driven model to orbital observations via explicit physical geometry embedding.First, the Geometry-Correlation Constrained Selection (GCCS) strategy prunes sub-optimal observations by balancing geometric diversity and radiometric consistency to optimize the input sequence. Second, a Sensor-Ray Encoder (SRE) parameterizes pixel-level pushbroom lines of sight derived from the Rational Function Model (RFM) into high-dimensional space-geometric tokens, reconciling the mathematical discrepancy between central projection and orbital kinematics. Third, a lightweight Ray-Pointing-Aware Adapter (RPAA) employs gated residual blocks to integrate these tokens directly into the frozen transformer backbone. Our findings underscore that integrating explicit physical geometry with optimized view selection is essential for robust feed-forward satellite 3D reconstruction.
Chinese Translation
在卫星星座时代，多视角光学卫星影像对于地球观测（EO）和高质量数字表面模型（DSM）重建至关重要。尽管前馈3D基础模型已在计算机视觉领域带来了变革，但其在卫星遥感中的应用受到隐含透视假设与显式轨道推扫几何结构之间的结构差异的固有限制。这种几何不一致性因显著的视角集异质性而进一步加剧。我们提出了EO-VGGT，一个通过显式物理几何嵌入将冻结的基于透视驱动的模型适应于轨道观测的框架。首先，几何相关约束选择（GCCS）策略通过平衡几何多样性和辐射一致性来优化输入序列，从而修剪次优观测。其次，传感器光线编码器（SRE）将从有理函数模型（RFM）导出的像素级推扫视线参数化为高维空间几何标记，调和中央投影与轨道运动学之间的数学差异。第三，轻量级光线指向感知适配器（RPAA）采用门控残差块将这些标记直接集成到冻结的变换器主干中。我们的研究结果强调，将显式物理几何与优化的视角选择相结合对于稳健的前馈卫星3D重建至关重要。
cs.CV / 44 / 2607.00428
HyFL-CLIP: Hyperbolic Fine-Tuning of CLIP for Robust Long-Context Understanding
HyFL-CLIP：用于稳健长上下文理解的CLIP超曲面微调
Abstract
CLIP (Contrastive Language-Image Pre-training) has become a de facto paradigm for image-text alignment, but it struggles with long-context descriptions (>77 tokens) due to absolute positional encoding and pretraining on short captions. In long contexts, sentences are often reordered, summarized, or partially omitted. Although prior works extend CLIP with longer positional encodings, they often suffer from degraded image-text alignment under such text perturbations. We attribute this limitation to the Euclidean contrastive objective, which enforces strict one-to-one matching and lacks explicit mechanisms for modeling hierarchical relationships between global context and its constituent elements. To address this issue, we propose HyFL-CLIP, a hyperbolic fine-tuning framework that distills the well-established text-image alignment learned in Euclidean CLIP into hyperbolic space via cross-manifold similarity distillation, leveraging its geometry to capture hierarchical and entailment relations. Our method models hierarchical semantics by linking summarized token-wise features, long-context descriptions, constituent short textual components, and images, capturing part-whole relationships via hyperbolic entailment with Einstein midpoint aggregation. Experiments on diverse benchmarks, including long-context cross-modal retrieval, cross-modal retrieval with caption perturbations, intra-modality retrieval, and short-text cross-modal retrieval, show that HyFL-CLIP achieves more robust long-context understanding. In particular, it yields up to 19.5% improvement in long-text cross-modal retrieval under textual perturbations over the best prior method. We also show HyFL-CLIP can be seamlessly integrated into other model frameworks by applying it to Stable Diffusion XL (SDXL).
Chinese Translation
CLIP（对比语言-图像预训练）已成为图像-文本对齐的事实标准，但由于绝对位置编码和在短标题上的预训练，它在长上下文描述（>77个标记）中表现不佳。在长上下文中，句子通常会被重新排序、总结或部分省略。尽管之前的研究通过更长的位置编码扩展了CLIP，但在这些文本扰动下，它们往往会遭遇图像-文本对齐的下降。我们将这一限制归因于欧几里得对比目标，该目标强制执行严格的一对一匹配，并缺乏建模全局上下文与其组成元素之间层次关系的明确机制。为了解决这个问题，我们提出了HyFL-CLIP，一个超曲面微调框架，通过跨流形相似性蒸馏将欧几里得CLIP中学习到的成熟文本-图像对齐提炼到超曲面空间，利用其几何特性捕捉层次和蕴含关系。我们的方法通过链接总结的逐标记特征、长上下文描述、组成的短文本组件和图像，建模层次语义，通过爱因斯坦中点聚合捕捉部分与整体之间的关系。针对多种基准测试的实验，包括长上下文跨模态检索、带有标题扰动的跨模态检索、同模态检索和短文本跨模态检索，结果表明HyFL-CLIP实现了更稳健的长上下文理解。特别是在文本扰动下，它在长文本跨模态检索中相较于最佳先前方法提高了多达19.5%。我们还展示了HyFL-CLIP可以通过将其应用于Stable Diffusion XL（SDXL）无缝集成到其他模型框架中。
cs.CV / 45 / 2607.00434
Information-Regularized Attention for Visual-Centric Reasoning
信息正则化注意力用于视觉中心推理
Abstract
Vision-language models (VLMs) have become a paradigm for multimodal learning, yet remain unstable due to object hallucination, weak visual grounding, and catastrophic forgetting after full-parameter instruction tuning. We claim these failures result from a lack of explicit control over visual representation learning during the standard next-token prediction objective. As a result, visual embeddings thus become passively optimized and prone to injecting redundant or spurious signals. To counter this, we introduce Information-Regularized Attention (IRA), a stochastic attention mechanism that explicitly regulates the amount of visual information injected into the hidden states of intermediate transformer layers. This local reparameterization translates uncertainty about visual representations into local noise that is independent across data points. Beyond evaluating model performance, we also quantify embedding properties, where IRA produces smoother curvature trajectories and suppresses attention-sink across all layers, indicating a more stable transformation of the visual signal. Our results suggest that stochastic attention is not merely a regularizer but a key contributor to representation learning in a generative architecture, offering a new direction for building more reliable VLMs.
Chinese Translation
视觉语言模型（VLMs）已成为多模态学习的一个范式，但由于对象幻觉、视觉基础薄弱以及全参数指令调优后的灾难性遗忘，仍然不稳定。我们认为这些失败源于在标准的下一个标记预测目标中缺乏对视觉表征学习的明确控制。因此，视觉嵌入被动地优化，容易注入冗余或虚假的信号。为此，我们引入了信息正则化注意力（Information-Regularized Attention, IRA），这是一种随机注意力机制，明确调节注入到中间变换器层的隐藏状态中的视觉信息量。这种局部重参数化将对视觉表征的不确定性转化为在数据点之间独立的局部噪声。除了评估模型性能外，我们还量化了嵌入特性，其中IRA产生更平滑的曲率轨迹，并抑制所有层的注意力沉没，表明视觉信号的变换更加稳定。我们的结果表明，随机注意力不仅仅是一个正则化器，而是生成架构中表征学习的关键贡献者，为构建更可靠的VLMs提供了新的方向。
cs.CV / 46 / 2607.00446
VideoSearch-R1: Iterative Video Retrieval and Reasoning via Soft Query Refinement
VideoSearch-R1：通过软查询细化进行迭代视频检索与推理
Abstract
As video corpora continue to expand in both scale and task complexity, there is increasing demand for approaches that retrieve relevant videos from large-scale corpora (inter-video reasoning) and subsequently perform fine-grained, query-conditioned tasks (intra-video reasoning) within the retrieved content, such as temporal grounding. However, existing approaches typically treat retrieval as a preprocessing step, and consequently, when the initial retrieval fails, there is no mechanism to refine the search, leading to the failure of subsequent fine-grained intra-video reasoning. Moreover, while recent agentic frameworks have advanced video understanding, they typically assume that the query-relevant video is already given, focusing exclusively on intra-video reasoning tasks. To address these limitations, we propose VideoSearch-R1, an agentic framework for iterative video retrieval and reasoning through multi-turn interaction with a video search engine. Specifically, we introduce Soft Query Refinement (SQR) to refine search query tokens in a continuous latent space rather than rewriting queries in the discrete text space, enabling more efficient and fine-grained adjustments. SQR and its reasoning process are trained using Group Relative Policy Optimization (GRPO), guided by task-level reward signals derived from retrieval and downstream tasks. Building upon this, VideoSearch-R1 achieves state-of-the-art performance across three datasets on Video Corpus Moment Retrieval (VCMR), iteratively retrieving videos from large-scale corpora, refining search queries, and performing precise query-conditioned temporal grounding within the retrieved content. Our analyses show that SQR effectively refines the original query, requiring significantly fewer generated tokens than explicit text-level query refinement. Code and model checkpoints are publicly available at mlvlab.github.io/VideoSearch-R1.
Chinese Translation
随着视频语料库在规模和任务复杂性上不断扩大，对从大规模语料库中检索相关视频（跨视频推理）并随后在检索内容中执行细粒度的查询条件任务（视频内部推理），如时间定位的需求日益增加。然而，现有方法通常将检索视为预处理步骤，因此，当初始检索失败时，缺乏细化搜索的机制，导致后续的细粒度视频内部推理失败。此外，尽管近期的智能框架在视频理解方面取得了进展，但它们通常假设查询相关的视频已经给出，专注于视频内部推理任务。为了解决这些局限性，我们提出了VideoSearch-R1，这是一个通过与视频搜索引擎的多轮交互进行迭代视频检索和推理的智能框架。具体而言，我们引入了软查询细化（Soft Query Refinement，SQR），在连续潜在空间中细化搜索查询标记，而不是在离散文本空间中重写查询，从而实现更高效和细粒度的调整。SQR及其推理过程使用群体相对策略优化（Group Relative Policy Optimization，GRPO）进行训练，指导信号来自于检索和下游任务的任务级奖励信号。在此基础上，VideoSearch-R1在视频语料库时刻检索（Video Corpus Moment Retrieval，VCMR）三个数据集上实现了最先进的性能，迭代地从大规模语料库中检索视频，细化搜索查询，并在检索内容中执行精确的查询条件时间定位。我们的分析表明，SQR有效地细化了原始查询，所需生成的标记显著少于显式文本级查询细化。代码和模型检查点可在mlvlab.github.io/VideoSearch-R1公开获取。
cs.CV / 47 / 2607.00461
Multimodal Continuous Reasoning via Asymmetric Mutual Variational Learning
通过非对称互变学习实现多模态连续推理
Abstract
Multimodal Large Language Models (MLLMs) are often constrained by a language-space bottleneck, forcing complex visual reasoning into discrete tokens which can lose perceptual nuance. A promising alternative is continuous latent reasoning, where the goal is to discover implicit reasoning pathways that bridge the multimodal query and the final answer. However, this introduces a severe train-inference mismatch: a training-time posterior, conditioned on the ground-truth answer, can exploit answer-dependent shortcuts. Standard variational training then forces the inference-time prior to mimic a posterior that has access to information unavailable at test time, leading to poor performance. To address this, we propose Asymmetric Mutual Variational Learning (AMVL), a framework that resolves this mismatch via a bidirectional calibration objective. A forward KL divergence trains the target-agnostic prior to match the posterior, while a novel reverse KL divergence simultaneously regularizes the posterior, preventing it from collapsing into inference-incompatible regions and mitigating this ``answer leakage''. We provide theoretical analysis formalizing this leakage as prior contamination and prove that our dual-KL objective reduces it. We instantiate AMVL in a latent-integrated MLLM and show that it consistently outperforms strong discrete and latent-reasoning baselines, improving the average score on the complex BLINK benchmark by +10.83 and achieving gains of up to +32.00 on individual reasoning tasks, with analyses confirming improved latent-space stability.
Chinese Translation
多模态大型语言模型（MLLMs）通常受到语言空间瓶颈的限制，迫使复杂的视觉推理转化为离散的标记，从而可能丧失感知的细微差别。一种有前景的替代方案是连续潜在推理，其目标是发现隐含的推理路径，以连接多模态查询和最终答案。然而，这引入了严重的训练-推理不匹配：在训练时，基于真实答案的后验分布可以利用依赖答案的捷径。标准的变分训练迫使推理时的先验模仿一个在测试时无法获取信息的后验，从而导致性能不佳。为了解决这个问题，我们提出了非对称互变学习（Asymmetric Mutual Variational Learning, AMVL），这是一个通过双向校准目标来解决这种不匹配的框架。前向KL散度训练目标无关的先验以匹配后验，而一种新颖的反向KL散度则同时对后验进行正则化，防止其崩溃到与推理不兼容的区域，从而减轻这种“答案泄漏”。我们提供了理论分析，将这种泄漏形式化为先验污染，并证明我们的双KL目标能够减少这种污染。我们在一个潜在集成的MLLM中实例化AMVL，并展示其在强大的离散和潜在推理基线模型上始终优于它们，在复杂的BLINK基准测试中平均得分提高了+10.83，并在单个推理任务上实现了高达+32.00的增益，分析结果确认了潜在空间的稳定性得到改善。
cs.CV / 48 / 2607.00465
StochasT: Learning with Stochastic Turn Depth for Visual Instruction Tuning
StochasT：基于随机回合深度的视觉指令调优学习
Abstract
Large Vision-Language Models (LVLMs) rely extensively on Visual Instruction Tuning (VIT) to elicit their multimodal reasoning capabilities. However, we find a discrepancy: VIT often packs multiple language tasks about the same image for conversational, multi-turn training, whereas existing benchmarks evaluate LVLMs in isolated, single-turn scenarios. The models can suffer from visual attention decay and contextual overfitting during multi-turn training, making it hard for them to realize their full potential in the mismatched test phase. To close the gap, we propose learning with Stochastic Turn Depth (StochasT), which stochastically groups language tasks for the same image into clusters of varying sizes (turn depth) while preserving their organic order. Hence, while StochasT draws on Dropout and stochastic depth for ResNets, it does not actually drop anything to maximize the utility of the training data. Furthermore, we introduce a challenging, benchmark-agnostic evaluation mechanism based on the Balanced Latin Square to measure LVLMs' robustness under varying contextual dependencies. Extensive experiments demonstrate that StochasT effectively grants LVLMs strong, harmonized capabilities for both single-turn and multi-turn use cases.
Chinese Translation
大型视觉语言模型（LVLMs）在很大程度上依赖视觉指令调优（VIT）来激发其多模态推理能力。然而，我们发现存在一个差异：VIT通常将关于同一图像的多个语言任务打包用于对话式的多回合训练，而现有基准则在孤立的单回合场景中评估LVLMs。这使得模型在多回合训练中可能遭遇视觉注意力衰减和上下文过拟合，难以在不匹配的测试阶段充分发挥其潜力。为了解决这一问题，我们提出了基于随机回合深度（StochasT）的学习方法，该方法随机将同一图像的语言任务分组为不同大小（回合深度）的聚类，同时保持其有机顺序。因此，尽管StochasT借鉴了Dropout和ResNets的随机深度，但实际上并不丢弃任何信息，以最大化训练数据的效用。此外，我们引入了一种基于平衡拉丁方的具有挑战性的基准无关评估机制，以测量LVLMs在不同上下文依赖下的鲁棒性。大量实验表明，StochasT有效地赋予LVLMs在单回合和多回合用例中强大且协调的能力。
cs.CV / 49 / 2607.00491
MindEdit-Bench: Benchmarking Object-Level Counterfactual Spatial Reasoning in VLMs from In-the-Wild Photos
MindEdit-Bench：从野外照片中基准测试视觉语言模型的物体级反事实空间推理
Abstract
Benchmarks for vision-language models (VLMs) mostly test observational spatial reasoning: models describe relations already visible in the input. Existing what-if tasks typically vary the observer while keeping the scene fixed. Can VLMs instead predict the consequences of hypothetically moving or rotating an object? We introduce MindEdit-Bench, a benchmark of six spatial reasoning tasks built from three-photo smartphone triplets of newly captured indoor scenes via an automatic in-the-wild 3D scene-graph extraction pipeline. Four tasks probe perception and perspective transformation over observed structure; two new tasks, L4 (spatial editing) and L5 (cross-view visibility editing), probe object-level counterfactual reasoning, where correct answers are absent from all input images. Each question provides 8-24 structured answer choices, enabling answer-letter-level diagnosis of spatial and fallback errors. The benchmark covers 120 private indoor scenes not drawn from public datasets, reducing public-data pretraining-overlap risk. Across 15 VLMs on 1,003 human-verified questions, task-wise mean VLM accuracy is only 8%-31%, versus 81%-97% human majority-vote accuracy. The pooled human--best-VLM gap is 53 pp, with at least 39 pp on every task. The structured answer space further reveals non-uniform failures, including weaker camera-depth-axis inference and fallback behavior on difficult visibility-editing cases.
Chinese Translation
视觉语言模型（VLMs）的基准测试主要测试观察性空间推理：模型描述输入中已可见的关系。现有的假设任务通常在保持场景固定的情况下变化观察者。那么，VLMs能否预测假设性移动或旋转物体的后果？我们引入了MindEdit-Bench，这是一个基于新捕获的室内场景的三张照片智能手机三元组构建的六个空间推理任务的基准，采用自动化的野外3D场景图提取管道。四个任务探测观察到的结构上的感知和视角变换；两个新任务，L4（空间编辑）和L5（跨视图可见性编辑），探测物体级反事实推理，其中正确答案在所有输入图像中均不存在。每个问题提供8-24个结构化答案选项，使得可以在答案字母级别诊断空间和回退错误。该基准涵盖了120个未从公共数据集中提取的私有室内场景，降低了公共数据预训练重叠的风险。在1003个经过人工验证的问题上，15个VLM的任务平均准确率仅为8%-31%，而人类多数投票准确率为81%-97%。人类与最佳VLM之间的差距为53个百分点，每个任务至少为39个百分点。结构化答案空间进一步揭示了不均匀的失败，包括在困难的可见性编辑案例中较弱的相机深度轴推理和回退行为。
cs.CV / 50 / 2607.00492
GenSP: Consistent Spherical Parameterization via Learning Shape Generative Models
GenSP：通过学习形状生成模型实现一致的球面参数化
Abstract
We introduce GenSP, a data-driven framework that learns consistent spherical parameterizations across a collection of genus-0 shapes. Instead of optimizing the parameterization of each shape independently, our method learns a neural generative model that predicts a continuous mapping from the unit sphere to shapes in a dataset. Under this formulation, spherical parameterizations are obtained through the inverse mappings of the learned generator, which encourages similar shapes to share consistent parameterizations. To make this formulation practical, we address several key challenges in learning such a generative model. First, we introduce a continuous neural deformation model that predicts surface points from sphere coordinates and latent shape codes, avoiding discretization artifacts common in mesh-based formulations. Second, we augment the training space with intermediate shapes that bridge the sphere and input shapes, allowing the model to learn meaningful deformations across a heterogeneous shape collection. Third, we compute reliable initial correspondences by propagating mappings along a spanning tree of training shapes in the latent space. Experiments on the ShapeNet dataset demonstrate that our approach significantly reduces geometric distortion and improves cross-shape consistency compared with state-of-the-art spherical parameterization methods.
Chinese Translation
我们介绍了GenSP，这是一种数据驱动的框架，能够在一组零基数（genus-0）形状中学习一致的球面参数化。我们的方法不是独立优化每个形状的参数化，而是学习一个神经生成模型，该模型预测从单位球面到数据集中形状的连续映射。在这种表述下，球面参数化通过学习生成器的逆映射获得，这鼓励相似形状共享一致的参数化。为了使这一表述具有实用性，我们解决了学习这种生成模型的几个关键挑战。首先，我们引入了一种连续神经变形模型，该模型根据球面坐标和潜在形状编码预测表面点，避免了基于网格的表述中常见的离散化伪影。其次，我们通过中间形状扩展训练空间，这些中间形状连接球面和输入形状，使模型能够在异构形状集合中学习有意义的变形。第三，我们通过在潜在空间中沿训练形状的生成树传播映射来计算可靠的初始对应关系。在ShapeNet数据集上的实验表明，与最先进的球面参数化方法相比，我们的方法显著减少了几何失真，并改善了跨形状的一致性。
cs.CV / 51 / 2607.00494
HieDG: A Hierarchical Discrete Geometry-Guided Framework for Multi-Animal Tracking
HieDG：一种层次化离散几何引导的多动物追踪框架
Abstract
Multi-animal tracking (MAT) is critical for wildlife monitoring and behavioral analysis, yet remains challenging due to uniform appearance, high density, and irregular motion. Existing methods typically follow heuristic- or query-based paradigms: the former relies on handcrafted geometric associations without end-to-end optimization, whereas the latter enables joint optimization but relies heavily on appearance embeddings. In such conditions, continuous geometric embeddings can be unstable, as small coordinate perturbations may disproportionately alter cross-frame attention weights, degrading identity association performance. To address this limitation, we propose HieDG, a Hierarchical Discrete Geometry-guided tracking framework that reformulates geometric dynamics as structured discrete representations within a query-based tracker. Instead of directly using raw geometric signals, HieDG employs a two-stage residual codebook to discretize position, scale, and velocity cues, transforming unstable continuous geometry into structured, stable discrete tokens. These tokens are aligned with visual embeddings and integrated into the tracking queries to enhance identity consistency. Extensive experiments on animal-specific benchmarks (AnimalTrack, BFT, and BuckTales) demonstrate state-of-the-art association performance with significant improvements in HOTA, AssA, and IDF1. Additional evaluations on generic multi-object tracking benchmarks, including DanceTrack and SportsMOT, show competitive performance, indicating the broader applicability of discretized geometric modeling beyond animal-specific scenarios.
Chinese Translation
多动物追踪（MAT）对于野生动物监测和行为分析至关重要，但由于外观一致、高密度和不规则运动等因素，仍然面临挑战。现有方法通常遵循启发式或查询基础的范式：前者依赖于手工制作的几何关联而没有端到端的优化，而后者则允许联合优化，但高度依赖外观嵌入。在这种情况下，连续几何嵌入可能不稳定，因为小的坐标扰动可能不成比例地改变跨帧注意权重，从而降低身份关联性能。为了解决这一限制，我们提出了HieDG，一种层次化离散几何引导的追踪框架，将几何动态重新表述为查询基础追踪器中的结构化离散表示。HieDG并不直接使用原始几何信号，而是采用两阶段残差代码本来离散化位置、尺度和速度线索，将不稳定的连续几何转换为结构化、稳定的离散标记。这些标记与视觉嵌入对齐，并整合到追踪查询中，以增强身份一致性。在动物特定基准（AnimalTrack、BFT和BuckTales）上的广泛实验表明，HieDG在HOTA、AssA和IDF1等指标上实现了最先进的关联性能，显著提升。此外，在通用多目标追踪基准（包括DanceTrack和SportsMOT）上的额外评估显示出竞争力的性能，表明离散几何建模在动物特定场景之外的更广泛适用性。
cs.CV / 52 / 2607.00498
Robust 3D Alignment of Generative Reconstructions via Partial Monocular Observations
通过部分单目观测实现稳健的三维生成重建对齐
Abstract
Aligning generative 3D reconstructions with partial monocular observations is a critical but under-explored challenge in computer vision. This task is inherently ill-posed due to severe asymmetries between noisy, sparse monocular inputs and dense generative priors, whose scale ambiguity and geometric hallucinations, combined with the lack of initial overlap, render traditional registration pipelines ineffective. To resolve these issues, we propose a training-free and interpretable geometric alignment framework that grounds generative 3D priors via a 3D similarity transformation (Sim(3)), which can recover accurate metric scale and pose. Specifically, we introduce an explicit scale factor to resolve metric ambiguity and employ a coarse-to-fine alignment strategy, leveraging geometry-aware descriptors for robust initialization and a decoupled closed-form solver for precision refinement. In addition, we introduce a Hallucination Filtering operation to effectively suppress outliers caused by hallucinated geometry. To evaluate alignment performance under these extreme conditions, we introduce GenPMOAlign--Where2Place, a rigorous benchmark specifically designed for Generative-to-Partial Monocular Observational Alignment. Experiments demonstrate that our method achieves stable and accurate registration, substantially outperforming both classical geometric pipelines and state-of-the-art learning-based baselines. Code and the benchmark will be publicly released.
Chinese Translation
将生成的三维重建与部分单目观测对齐是计算机视觉中一个关键但尚未充分探索的挑战。由于噪声、稀疏的单目输入与密集的生成先验之间存在严重的不对称性，这一任务本质上是病态的。生成先验的尺度模糊性和几何幻觉，加上缺乏初始重叠，使得传统的配准流程无效。为了解决这些问题，我们提出了一种无训练且可解释的几何对齐框架，通过三维相似变换（Sim(3)）将生成的三维先验固定，从而恢复准确的度量尺度和姿态。具体而言，我们引入了一个显式的尺度因子来解决度量模糊性，并采用粗到细的对齐策略，利用几何感知描述符进行稳健的初始化，并使用解耦的闭式解算器进行精度细化。此外，我们引入了一种幻觉过滤操作，有效抑制由幻觉几何引起的离群点。为了在这些极端条件下评估对齐性能，我们推出了GenPMOAlign--Where2Place，这是一个专门为生成到部分单目观测对齐设计的严格基准。实验表明，我们的方法实现了稳定且准确的配准，显著优于经典几何流程和最先进的基于学习的基线。代码和基准将公开发布。
cs.CV / 53 / 2607.00499
Prior-Anchored Debiasing for Long-Tailed Multi-Organ Pathology Report Generation
基于先验锚定的长尾多脏器病理报告生成去偏差方法
Abstract
Automated pathology report generation from Whole Slide Images (WSIs) has attracted increasing attention in digital pathology. However, existing methods are predominantly developed under single-organ settings, overlooking the multi-organ scenarios encountered in clinical practice, where organ types typically follow a long-tailed distribution. To address this gap, we identify two critical biases: (1) visual representation bias, where the encoder favors head-class patterns over tail-class discriminative features, and (2) textual decoding bias, where the decoder overfits to head-class narrative patterns, yielding diagnostically unreliable outputs for tail-class organs. To mitigate these two biases, we propose a novel Prior-anchored multi-Organ pathology report Generation framework (PriOrGen). Specifically, a Visual-Prototype Anchored Bottleneck module leverages the information bottleneck principle with learnable anchor representations to selectively retain diagnostically relevant visual information while filtering out head-biased redundancy. Secondly, a Meta-Report Anchored Bank module constructs an organ-specific meta-report anchored bank and retrieves organ-faithful textual priors to steer the decoder away from head-class narrative patterns. Extensive experiments on a multi- organ pathology dataset demonstrate that our method effectively mitigates long-tail biases and achieves superior report generation performance across both head and tail organ categories compared to state-of-the-art methods.
Chinese Translation
从全切片图像（Whole Slide Images, WSIs）自动生成病理报告在数字病理学中引起了越来越多的关注。然而，现有方法主要是在单脏器环境下开发的，忽视了临床实践中遇到的多脏器场景，其中脏器类型通常遵循长尾分布。为了解决这一问题，我们识别出两种关键偏差：（1）视觉表征偏差，编码器偏向于头类模式而忽视尾类的区分特征；（2）文本解码偏差，解码器过拟合于头类叙述模式，导致尾类脏器的诊断输出不可靠。为了减轻这两种偏差，我们提出了一种新颖的基于先验锚定的多脏器病理报告生成框架（PriOrGen）。具体而言，视觉原型锚定瓶颈模块利用信息瓶颈原理与可学习的锚定表征，选择性地保留与诊断相关的视觉信息，同时过滤掉头类偏向的冗余信息。其次，元报告锚定库模块构建了一个脏器特定的元报告锚定库，并检索与脏器一致的文本先验，以引导解码器远离头类叙述模式。在一个多脏器病理数据集上的大量实验表明，我们的方法有效减轻了长尾偏差，并在头类和尾类脏器类别上实现了优于现有最先进方法的报告生成性能。
cs.CV / 54 / 2607.00509
AnF-DiffPET: Anatomy- and Frequency-Guided Diffusion for PET/CT Denoising
AnF-DiffPET：基于解剖和频率引导的PET/CT去噪扩散方法
Abstract
Positron emission tomography (PET) provides essential functional information for disease assessment, however reducing injected activity or acquisition time produces low-dose (LD) PET with stronger count dependent noise and less reliable uptake quantification. Diffusion models offer a promising solution for PET denoising by progressively recovering high-dose (HD) PET images from LD inputs. However, LD-to-HD PET denoising is still challenging due to insufficient anatomical guidance, unstable multi-scale feature propagation, and uncertain frequency domain uptake recovery. We propose AnF-DiffPET, an anatomy- and frequency-guided diffusion framework for computed tomography (CT) conditioned LD PET denoising. The framework integrates Anatomical-Frequency Guidance (AFG), Multi-Scale Cross-Transformer Reconstruction (MSCTR), and Frequency-Contrastive Hard Mining (FCHM) to enhance anatomy aware feature modulation and frequency domain consistency during denoising. Experimental results across four PET/CT datasets show that the proposed method improves image fidelity, anatomical consistency, and quantitative fidelity over representative CNN-based, GAN-based, transformer-based, and diffusion-based methods. The code and trained models will be publicly released upon acceptance.
Chinese Translation
正电子发射断层扫描（PET）为疾病评估提供了重要的功能信息，然而减少注射剂量或采集时间会产生低剂量（LD）PET，这会导致更强的计数依赖噪声和不太可靠的摄取定量。扩散模型为PET去噪提供了一种有前景的解决方案，通过逐步从LD输入恢复高剂量（HD）PET图像。然而，由于解剖指导不足、多尺度特征传播不稳定以及频域摄取恢复不确定，LD到HD的PET去噪仍然具有挑战性。我们提出了AnF-DiffPET，一个基于解剖和频率引导的扩散框架，用于计算机断层扫描（CT）条件下的LD PET去噪。该框架集成了解剖-频率引导（AFG）、多尺度交叉变换重建（MSCTR）和频率对比硬挖掘（FCHM），以增强解剖感知特征调制和去噪过程中的频域一致性。在四个PET/CT数据集上的实验结果表明，所提方法在图像保真度、解剖一致性和定量保真度上优于代表性的基于卷积神经网络（CNN）、生成对抗网络（GAN）、变换器（transformer）和扩散模型的方法。代码和训练模型将在接受后公开发布。
cs.CV / 55 / 2607.00514
Cross4D-JEPA: Dense Cross-modal Correspondence Distillation for 4D Point Cloud Representation Learning
Cross4D-JEPA：用于4D点云表示学习的密集跨模态对应蒸馏
Abstract
Automatic understanding of dynamic 4D point clouds, the 3D-point sequences captured over time by depth sensors and LiDAR, is central to robotics and embodied perception. Yet annotating them densely is expensive, making self-supervised pretraining the natural route to transferable representations. Existing pretext tasks, however, are almost entirely intra-modal, and the few methods that transfer knowledge from 2D foundation models rely on a single global embedding per clip, discarding the rich per-patch semantics that these models compute. To address this gap, we propose Cross4D-JEPA, a teacher-student method that distills a frozen 2D foundation model, an image model DINOv2, or a video model V-JEPA 2, into a 4D point encoder. The proposed method combines (1) a dense cross-modal correspondence that maps every 3D point to the teacher patch feature it projects to, and (2) a per-point objective that trains the student to match these features in latent space with no masking, negatives, or decoder. We evaluate Cross4D-JEPA on four benchmarks, MSR-Action3D, DeformingThings4D, NTU-RGB+D 60, and HOI4D, against intra-modal and global cross-modal baselines. Experimental results show that, under a matched protocol, the proposed method consistently outperforms intra-modal and global cross-modal baselines across the four benchmarks and is competitive with heavier published 4D methods; further analysis attributes this gain primarily to the granularity of the correspondence rather than the teacher modality. Beyond recognition accuracy, the dense representation learned by Cross4D-JEPA transfers across domains, improves label efficiency, and improves full-label fine-tuning under the same training budget, while a 13x smaller encoder matches a heavyweight pooling backbone.
Chinese Translation
自动理解动态4D点云，即由深度传感器和激光雷达捕获的随时间变化的3D点序列，对于机器人技术和具身感知至关重要。然而，密集标注这些点云的成本较高，使得自监督预训练成为可迁移表示的自然途径。然而，现有的预训练任务几乎完全是单模态的，少数从2D基础模型转移知识的方法依赖于每个片段的单一全局嵌入，忽略了这些模型计算的丰富的每个补丁语义。为了解决这一问题，我们提出了Cross4D-JEPA，一种教师-学生方法，将一个冻结的2D基础模型（图像模型DINOv2或视频模型V-JEPA 2）蒸馏到一个4D点编码器中。该方法结合了（1）密集的跨模态对应，将每个3D点映射到其投影的教师补丁特征，以及（2）一个逐点目标，训练学生在潜在空间中匹配这些特征，而无需掩蔽、负样本或解码器。我们在四个基准测试上评估Cross4D-JEPA，分别是MSR-Action3D、DeformingThings4D、NTU-RGB+D 60和HOI4D，并与单模态和全局跨模态基线进行比较。实验结果表明，在匹配协议下，所提出的方法在四个基准测试中始终优于单模态和全局跨模态基线，并且与更重的已发布4D方法具有竞争力；进一步分析将这一增益主要归因于对应的细粒度，而非教师模态。除了识别准确性外，Cross4D-JEPA学习的密集表示在不同领域之间迁移，提升标签效率，并在相同的训练预算下改善全标签微调，而一个13倍更小的编码器与重量级池化骨干网络相匹配。
cs.CV / 56 / 2607.00522
Restore3D: Breathing Life into Broken Objects with Shape and Texture Restoration
Restore3D：为破损物体注入生命的形状与纹理修复
Abstract
Restoring incomplete or damaged 3D objects is crucial for cultural heritage preservation, occluded object reconstruction, and artistic design. Existing methods primarily focus on geometric completion, often neglecting texture restoration and struggling with relatively complex and diverse objects. We introduce Restore3D, a novel framework that simultaneously restores both the shape and texture of broken objects using multi-view images. To address limited training data, we develop an automated data generation pipeline that synthesizes paired incomplete-complete samples from large-scale 3D datasets. Central to Restore3D is a multi-view model, enhanced by a carefully designed Mask Self-Perceiver module with a Depth-Aware Mask Rectifier. The rectified masks learned by the self-perceiver guide an image integration and enhancement phase, helping retain observed shape and texture patterns while refining the generated regions and mitigating the low-resolution limitations of the base model, yielding high-resolution, semantically coherent, and view-consistent multi-view images. A coarse-to-fine reconstruction strategy is then employed to recover detailed textured 3D meshes from refined multi-view images. Experiments on synthetic and real broken-object benchmarks show that Restore3D improves multi-view restoration quality and textured-mesh reconstruction over representative inpainting, completion, and reconstruction baselines in the evaluated settings. Project Page: restore3dx.github.io
Chinese Translation
修复不完整或损坏的3D物体对于文化遗产保护、遮挡物体重建和艺术设计至关重要。现有方法主要集中在几何补全上，往往忽视纹理修复，并且在处理相对复杂和多样化的物体时面临挑战。我们提出了Restore3D，一个新颖的框架，利用多视角图像同时修复破损物体的形状和纹理。为了应对有限的训练数据，我们开发了一种自动化数据生成管道，从大规模3D数据集中合成成对的不完整-完整样本。Restore3D的核心是一个多视角模型，通过精心设计的Mask Self-Perceiver模块和Depth-Aware Mask Rectifier进行增强。自感知器学习到的修正掩膜指导图像集成和增强阶段，帮助保留观察到的形状和纹理模式，同时细化生成区域并减轻基础模型的低分辨率限制，从而生成高分辨率、语义一致且视角一致的多视角图像。随后采用粗到细的重建策略，从精细化的多视角图像中恢复详细的纹理3D网格。在合成和真实破损物体基准测试中的实验表明，Restore3D在评估设置中提高了多视角修复质量和纹理网格重建效果，相较于代表性的修补、补全和重建基线方法表现更佳。项目页面：restore3dx.github.io
cs.CV / 57 / 2607.00525
SPECSIA: Stylization Dataset for Novel-View Enhancement in Drawing-based 3D Animation
SPECSIA：用于绘图基础3D动画中新视角增强的风格化数据集
Abstract
Generating animation from a single 2D drawing is challenging because the output must preserve character appearance while remaining plausible and temporally coherent under motion. Existing drawing-based 3D animation pipelines often use sample-wise 2D refinement to align animated renderings with the input image, but such optimization tends to overfit to the observed view and fails to correct projection-induced artifacts in novel views. To address this limitation, we introduce SPECSIA-15K, a paired stylization dataset containing 14,980 artifact-corrupted projection/refinement-target pairs from 1,498 3DBiCar characters. We further present DraViE (Drawing-based View Enhancement), a lightweight plug-and-play module trained with data-level priors to remove novel-view artifacts while preserving style and motion plausibility. Experiments show consistent gains in novel-view fidelity and temporal coherence with lower per-character adaptation cost than sample-wise fine-tuning.
Chinese Translation
从单一的二维绘图生成动画是一项具有挑战性的任务，因为输出必须在运动中保持角色外观的合理性和时间一致性。现有的基于绘图的3D动画流程通常使用逐样本的二维细化来对齐动画渲染与输入图像，但这种优化往往会过拟合到观察视角，并未能纠正新视角下的投影引起的伪影。为了解决这一局限性，我们引入了SPECSIA-15K，这是一个配对的风格化数据集，包含来自1,498个3DBiCar角色的14,980对受伪影影响的投影/细化目标。我们进一步提出了DraViE（基于绘图的视角增强），这是一个轻量级的即插即用模块，利用数据级先验进行训练，以去除新视角下的伪影，同时保持风格和运动的合理性。实验表明，与逐样本微调相比，DraViE在新视角的保真度和时间一致性方面表现出一致的提升，同时每个角色的适应成本更低。
cs.CV / 58 / 2607.00529
NoPA: Non-Parametric Online 3D Scene Graph Generation
NoPA：非参数在线三维场景图生成
Abstract
Classic 3D scene graph generation approaches fail to work in real-time due to the heavy computational cost of environment mapping and the need to generate intermediate point-cloud representations. To alleviate this issue, a recent work eschews point clouds in favor of a lightweight Gaussian distribution for each object. This approximation drastically speeds up inference and enables real-time 3D scene graph generation. However, the representation has two key weaknesses. \textbf{1)} Each object is approximated by a single 3D Gaussian, which causes a severe loss of 3D geometric detail. \textbf{2)} The discrepancy between this approximation and the true object geometry exacerbates the inaccurate merging of object candidates during online inference. To address these issues, we propose \textbf{NoPA}, which represents each object as a separate non-parametric distribution. This formulation retains 3D geometric information while preserving real-time inference of the parametric Gaussian formulation. To build upon our novel object representation, we propose a tailored merging strategy to recover coherent object instances. Specifically, we leverage maximum mean discrepancy on kernel density estimates to enable robust merging of object candidates during online exploration while minimizing added computational complexity. The key is to maintain a fixed particle set per object. Furthermore, to rectify the relation loss caused by misclassified objects, NoPA propagates relationships between objects with high affinity. Experiments show that NoPA substantially outperforms current methods without sacrificing real-time inference speed.
Chinese Translation
经典的三维场景图生成方法由于环境映射的高计算成本和生成中间点云表示的需求，无法实现实时处理。为了解决这个问题，最近的研究放弃了点云，转而为每个对象使用轻量级的高斯分布。这种近似方法大幅加快了推理速度，使得实时三维场景图生成成为可能。然而，这种表示存在两个主要弱点。1）每个对象被近似为一个单一的三维高斯分布，这导致三维几何细节的严重损失。2）这种近似与真实对象几何之间的差异加剧了在线推理过程中对象候选的合并不准确。为了解决这些问题，我们提出了NoPA，它将每个对象表示为一个独立的非参数分布。这种表述保留了三维几何信息，同时保持了参数高斯表述的实时推理能力。为了在我们新颖的对象表示基础上进行扩展，我们提出了一种定制的合并策略，以恢复一致的对象实例。具体而言，我们利用核密度估计上的最大均值差异，使得在在线探索过程中能够稳健地合并对象候选，同时最小化额外的计算复杂性。关键是为每个对象保持固定的粒子集。此外，为了纠正由于对象误分类造成的关系损失，NoPA在高亲和力的对象之间传播关系。实验表明，NoPA在不牺牲实时推理速度的情况下，显著优于当前的方法。
cs.CV / 59 / 2607.00544
GEAR-Seg: A Grounded Explainable Agent for Reasoning Segmentation and Data Engine
GEAR-Seg：一种用于推理分割和数据引擎的基础可解释代理
Abstract
Reasoning segmentation requires localizing targets based on complex, implicit queries. Current end-to-end models typically entangle perception and deduction into an opaque black box, severely limiting interpretability and scalability. To address this, we propose GEAR-Seg (Grounded Explainable Agent for Reasoning Segmentation), an explicitly decoupled agent that shifts the paradigm by translating visual pixels into dense, attribute-rich text. By decoupling class-agnostic segmentation, semantic description, and Large Language Model (LLM) deduction, GEAR-Seg transforms implicit reasoning into an explicit, trackable logic chain. As a zero-shot inference framework, it achieves highly competitive performance across diverse reasoning and fine-grained referring segmentation benchmarks. Furthermore, GEAR-Seg inherently functions as a highly scalable data engine. Utilizing this engine, we construct GEAR-131K, a massive benchmark (over 38k images, 656k QA-mask pairs) introducing a multifaceted taxonomy tailored for complex real-world manipulation-oriented reasoning. Finally, distillation experiments demonstrate that lightweight models supervised exclusively by our automated pipeline closely match the upper-bound performance of costly human-annotated baselines.
Chinese Translation
推理分割需要基于复杂的隐式查询来定位目标。目前的端到端模型通常将感知和推理纠缠在一起，形成一个不透明的黑箱，严重限制了可解释性和可扩展性。为了解决这个问题，我们提出了GEAR-Seg（Grounded Explainable Agent for Reasoning Segmentation），这是一个明确解耦的代理，通过将视觉像素转换为密集的、属性丰富的文本来改变这一范式。通过解耦与类别无关的分割、语义描述和大型语言模型（Large Language Model, LLM）推理，GEAR-Seg将隐式推理转化为明确且可追踪的逻辑链。作为一个零样本推理框架，它在多样的推理和细粒度指称分割基准测试中实现了高度竞争的性能。此外，GEAR-Seg本质上作为一个高度可扩展的数据引擎运作。利用该引擎，我们构建了GEAR-131K，这是一个庞大的基准（超过38,000张图像，656,000个问答-掩码对），引入了一个针对复杂现实世界操作导向推理的多维分类法。最后，蒸馏实验表明，仅通过我们的自动化管道监督的轻量级模型的性能接近昂贵的人类标注基准的上限。
cs.CV / 60 / 2607.00545
ECoSim: Data Efficient Fine-Tuning for Controllable Traffic Simulation
ECoSim：用于可控交通仿真的数据高效微调
Abstract
Controllable traffic simulation is critical for testing autonomous driving systems, yet existing approaches often require retraining large generative models with extensive annotated data. We introduce a lightweight control adaptation framework that enables multi-modal controllability (sketch, latent behavior codes, and text) for pretrained state-of-the-art diffusion and autoregressive traffic models. By modulating intermediate features through identity-initialized FiLM layers, our method efficiently adds new control modalities while preserving the base model's generative prior. Evaluated on Waymo Open Sim Agents Challenge, our approach demonstrates strong controllability with less than 1% of the paired control data. Through context-aware condition transfer, our framework enables counterfactual scenario generation and long-tail synthesis while maintaining stable closed-loop driving realism and safety. Our framework unlocks new possibilities for controllable traffic simulation, enabling targeted scenario generation through lightweight adaptation of pretrained generative models. Project page: https://ecosim-web.github.io/
Chinese Translation
可控交通仿真对于测试自动驾驶系统至关重要，但现有的方法通常需要使用大量带注释的数据对大型生成模型进行重新训练。我们提出了一种轻量级的控制适应框架，使得预训练的最先进扩散和自回归交通模型能够实现多模态可控性（草图、潜在行为编码和文本）。通过在身份初始化的 FiLM 层中调节中间特征，我们的方法有效地添加了新的控制模态，同时保持了基础模型的生成先验。在 Waymo Open Sim Agents Challenge 上进行评估，我们的方法在控制数据少于 1% 的情况下展现出强大的可控性。通过上下文感知的条件转移，我们的框架实现了反事实场景生成和长尾合成，同时保持稳定的闭环驾驶真实感和安全性。我们的框架为可控交通仿真开辟了新的可能性，通过对预训练生成模型的轻量级适应实现针对性场景生成。项目页面：https://ecosim-web.github.io/
cs.CV / 61 / 2607.00547
EgoGapBench: Benchmarking Egocentric Action Selection in Multi-Agent Scenes
EgoGapBench：多智能体场景中自我中心动作选择的基准测试
Abstract
Existing egocentric benchmarks have primarily constructed the egocentric setting from first-person-view data, which makes it difficult to evaluate egocentric perspective itself in isolation. However, understanding first-person-view input and taking an egocentric perspective are separable abilities, especially when first-person body cues are absent or when other agents are present. To isolate egocentric perspective understanding, we introduce EgoGapBench, a diagnostic benchmark for measuring action selection in multi-agent egocentric scenes. We define the ability measured by this benchmark as Egocentric Action Selection (EAS): selecting an appropriate action from the agent's perspective in the presence of other agents. On EgoGapBench, humans answer reliably, whereas both open-source and proprietary MLLMs perform substantially worse and systematically select actions performed by other visible agents. Fine-tuning on existing egocentric data fails to close this gap and can even be detrimental. In contrast, fine-tuning on EgoGapBench training data improves accuracy but does not reach human performance. These results show that EAS is difficult to acquire from first-person-view data alone, and that MLLMs should be evaluated and trained not only for scene understanding but also for egocentric action selection.
Chinese Translation
现有的自我中心基准主要是基于第一人称视角数据构建的，这使得单独评估自我中心视角本身变得困难。然而，理解第一人称视角输入和采取自我中心视角是可以分开的能力，尤其是在缺乏第一人称身体线索或其他智能体存在时。为了隔离自我中心视角理解，我们引入了EgoGapBench，这是一个用于测量多智能体自我中心场景中动作选择的诊断基准。我们将该基准测量的能力定义为自我中心动作选择（Egocentric Action Selection, EAS）：在其他智能体存在的情况下，从智能体的视角选择适当的动作。在EgoGapBench上，人类的回答是可靠的，而开源和专有的多模态大语言模型（MLLMs）表现明显较差，并且系统性地选择其他可见智能体执行的动作。对现有自我中心数据的微调未能缩小这一差距，甚至可能有害。相比之下，对EgoGapBench训练数据的微调提高了准确性，但仍未达到人类的表现。这些结果表明，仅从第一人称视角数据中获取EAS是困难的，并且MLLMs的评估和训练不仅应针对场景理解，还应针对自我中心动作选择。
cs.CV / 62 / 2607.00573
BrainFIBRE: A Foundation Model via Information Decomposition for Brain Microstructure
BrainFIBRE：一种通过信息分解实现的大脑微结构基础模型
Abstract
Diffusion MRI probes brain microstructure with particular sensitivity to early cerebrovascular and neurodegenerative changes. Neurite Orientation Dispersion and Density Imaging (NODDI) decomposes the diffusion signal into three biophysically interpretable maps: neurite density index (NDI), orientation dispersion index (ODI), and free water fraction (FWF), capturing neurite packing, fiber coherence, and extracellular fluid. These 3D maps offer a rich substrate for transferable microstructural representations, yet integrating them is challenging: standard representation learning struggles to disentangle the unique information in each map from their shared and synergistic interactions. We present BrainFIBRE, the first foundation model for brain microstructure, pretrained on NODDI-derived maps from 55,592 UK Biobank participants. We propose Self-supervised Partial Information Decomposition (SPID), which extends PID-guided multimodal learning to the self-supervised regime for the first time. A novel Counterfactual Candidate Construction (CCC) paradigm perturbs inter-modality alignment through modality dropping and swapping, providing the contrastive signal for a Mixture-of-Experts architecture to disentangle unique, synergistic, and redundant information without any downstream label. On both Caucasian and Asian cohorts, BrainFIBRE achieves state-of-the-art performance across diverse tasks predicting age, sex, cerebrovascular and neurodegenerative markers, and cognition, while yielding neurobiologically interpretable representations that reveal task- and cohort-specific interaction patterns. BrainFIBRE establishes a versatile foundation for neuroimaging analysis at the microstructural level.
Chinese Translation
扩散MRI以特有的敏感性探测大脑微结构，尤其是早期脑血管和神经退行性变化。神经纤维方向分散和密度成像（Neurite Orientation Dispersion and Density Imaging, NODDI）将扩散信号分解为三个生物物理可解释的图谱：神经纤维密度指数（Neurite Density Index, NDI）、方向分散指数（Orientation Dispersion Index, ODI）和自由水分数（Free Water Fraction, FWF），捕捉神经纤维的堆积、纤维的一致性和细胞外液。这些三维图谱为可转移的微结构表示提供了丰富的基础，但整合它们却面临挑战：标准的表示学习难以将每个图谱中独特的信息与它们的共享和协同交互分离开来。我们提出了BrainFIBRE，这是首个针对大脑微结构的基础模型，基于55,592名UK Biobank参与者的NODDI衍生图谱进行预训练。我们提出了自监督部分信息分解（Self-supervised Partial Information Decomposition, SPID），首次将PID引导的多模态学习扩展到自监督领域。一个新颖的反事实候选构建（Counterfactual Candidate Construction, CCC）范式通过模态丢弃和交换扰动模态间的对齐，为Mixture-of-Experts架构提供对比信号，以在没有任何下游标签的情况下分离独特、协同和冗余信息。在白人和亚洲人群中，BrainFIBRE在预测年龄、性别、脑血管和神经退行性标志物以及认知等多种任务中实现了最先进的性能，同时产生了神经生物学可解释的表示，揭示了任务和人群特定的交互模式。BrainFIBRE为微结构层面的神经影像分析奠定了多功能基础。
cs.CV / 63 / 2607.00578
Caption Bottleneck Models
标题瓶颈模型
Abstract
Concept Bottleneck Models (CBMs) provide interpretability by routing predictions through a layer of human-understandable concepts. However, defining an optimal concept set for a specific dataset remains an open challenge. Existing approaches rely on expensive expert annotations or LLM-generated lists based solely on class names. Even "open-vocabulary" variants typically depend on static concept sets, which restrict discovery and introduce label bias. Furthermore, traditional CBMs often suffer from information leakage, where unmodeled visual features bypass the bottleneck and compromise the integrity of the explanations. To overcome these limitations, we propose Caption Bottleneck Models (CaBM), a framework that circumvents the need for predefined concept sets by replacing rigid concept layers with free-form natural language. By representing images via LMM-generated captions and training a classifier strictly on this text, CaBM ensures a leakage-free architecture by construction. Additionally, by analyzing the text classifier post-training, CaBM autonomously discovers high-quality, dataset-specific concepts. Our results across fine- and coarse-grained benchmarks demonstrate that CaBM achieves competitive accuracy while preserving interpretability without the constraints of external dictionaries or manual labeling.
Chinese Translation
概念瓶颈模型（CBMs）通过将预测路由到一层人类可理解的概念中提供可解释性。然而，为特定数据集定义最佳概念集仍然是一个未解决的挑战。现有方法依赖于昂贵的专家注释或仅基于类别名称生成的LLM（大型语言模型）列表。即使是“开放词汇”变体通常也依赖于静态概念集，这限制了发现并引入了标签偏差。此外，传统的CBMs往往遭受信息泄漏的问题，即未建模的视觉特征绕过瓶颈，从而损害了解释的完整性。为了克服这些局限性，我们提出了标题瓶颈模型（CaBM），该框架通过用自由形式的自然语言替代严格的概念层，避免了对预定义概念集的需求。通过使用LLM生成的标题表示图像，并仅基于这些文本训练分类器，CaBM确保了从构建上就没有泄漏。此外，通过分析训练后文本分类器，CaBM自主发现高质量的特定数据集概念。我们在细粒度和粗粒度基准上的结果表明，CaBM在保持可解释性的同时实现了具有竞争力的准确性，而不受外部词典或手动标注的限制。
cs.CV / 64 / 2607.00579
EPO: Boosting 3D Foundation Models with Edge-based Pose Optimization
EPO：通过基于边缘的姿态优化提升3D基础模型
Abstract
We introduce \textbf{Edge-based Pose Optimization (EPO)}, a trackless geometric optimization framework specifically designed to boost the Structure-from-Motion reconstructions generated by 3D Foundation Models. These models achieve rapid inference by bypassing the time-consuming feature extraction and matching stages of traditional pipelines, where explicit correspondences between each 3D point and multiple images, referred to as tracks, are established. However, their geometric accuracy currently falls short of traditional pipelines. While this can be addressed in a post-processing step via Bundle Adjustment-like refinement, doing so requires extracting feature tracks, thus defeating the original speed advantage. Instead, our fully differentiable framework uses edge map alignment as a proxy for geometric optimization, avoiding feature extraction and track construction entirely. Through extensive evaluation across multiple datasets and tasks, we demonstrate that EPO matches or outperforms Bundle Adjustment-like methods while requiring significantly lower runtime and memory. Notably, its reduced memory footprint makes EPO suitable for consumer-grade hardware, where competing refinement methods cannot run.
Chinese Translation
我们引入了 extbf{基于边缘的姿态优化（Edge-based Pose Optimization, EPO）}，这是一个无轨迹的几何优化框架，专门设计用于提升由3D基础模型生成的运动结构重建。这些模型通过绕过传统流程中耗时的特征提取和匹配阶段，实现了快速推断，在这些阶段中，建立了每个3D点与多个图像之间的显式对应关系，称为轨迹。然而，它们的几何精度目前仍低于传统流程。虽然可以通过类似束调整（Bundle Adjustment）的后处理步骤来解决这一问题，但这需要提取特征轨迹，从而失去了原有的速度优势。相反，我们的全可微分框架使用边缘图对齐作为几何优化的代理，完全避免了特征提取和轨迹构建。通过在多个数据集和任务上的广泛评估，我们证明EPO的性能与类似束调整的方法相当或更优，同时所需的运行时间和内存显著降低。值得注意的是，其减少的内存占用使EPO适用于消费级硬件，而竞争性的优化方法则无法运行。
cs.CV / 65 / 2607.00580
Active Spatial Guidance: Eliminating Injected Positional Mechanisms in Vision Transformers
主动空间引导：消除视觉变换器中的注入位置机制
Abstract
Vision Transformers (ViTs) commonly rely on injected positional mechanisms to address self-attention's permutation invariance. Motivated by the spatial regularities of natural images, we ask whether spatial organization can be induced from data rather than explicitly injected. Under controlled, matched from-scratch training, we propose Active Spatial Guidance (Guidance), a training-only objective that disables positional injection and applies an auxiliary 2D coordinate-regression loss to the final-layer patch tokens. The guidance head is used only during training and removed for inference; the deployed model consists of a positional-injection-free ViT encoder and the task-specific prediction module. Using DINOv3 ViT backbones, Guidance consistently improves performance on ImageNet-100 classification, ADE20K semantic segmentation, and Hypersim monocular depth estimation, outperforming strong injected baselines such as learned absolute positional embeddings and rotary positional embeddings under identical training protocols. On ImageNet-100, broader comparisons against representative injected positional designs further support Guidance's effectiveness. Guidance also improves robustness under resolution transfer, and multi-resolution training further strengthens accuracy across input sizes. Overall, our results suggest that spatial inductive bias in ViTs need not be architecturally injected, but can be shaped through training-time supervision. The code used for training and evaluation is publicly available in https://github.com/cloudlc/asg.
Chinese Translation
视觉变换器（ViTs）通常依赖注入的位置机制来解决自注意力的排列不变性。受到自然图像空间规律的启发，我们提出一个问题：是否可以通过数据诱导空间组织，而不是显式注入。在受控的、匹配的从零开始训练下，我们提出了主动空间引导（Active Spatial Guidance，Guidance），这是一种仅在训练期间使用的目标，它禁用位置注入，并对最终层的补丁令牌应用辅助的二维坐标回归损失。引导头仅在训练期间使用，推理时移除；部署的模型由一个不含位置注入的ViT编码器和特定任务的预测模块组成。使用DINOv3 ViT骨干网络，Guidance在ImageNet-100分类、ADE20K语义分割和Hypersim单目深度估计上持续提高性能，超越了在相同训练协议下的强基线，如学习的绝对位置嵌入和旋转位置嵌入。在ImageNet-100上，与代表性的注入位置设计的更广泛比较进一步支持了Guidance的有效性。Guidance还提高了在分辨率转移下的鲁棒性，而多分辨率训练进一步增强了不同输入大小下的准确性。总体而言，我们的结果表明，ViTs中的空间归纳偏置不必通过架构注入，而可以通过训练时的监督来塑造。用于训练和评估的代码已公开在https://github.com/cloudlc/asg。
cs.CV / 66 / 2607.00595
GADA: Geometry-Aware Deformable Aggregation for Image-Based Gaussian Splatting
GADA：基于几何的可变形聚合用于基于图像的高斯溅射
Abstract
Gaussian Splatting has achieved significant improvements by incorporating warping-based techniques. However, such methods suffer from pixel-level inaccuracies due to uncertain geometry. This uncertainty leads to spatial misalignments in the warped images, which disrupt residual learning used in warping-based methods and fundamentally limit the gains of correction, particularly on thin structures and high-frequency details. Driven by our insight that useful visual cues are not lost but locally preserved under slight displacement, we propose Geometry-Aware Deformable Aggregation (GADA). This method introduces an iterative refinement module with deformable offsets to actively correct spatial misalignments and recover these displaced cues. Furthermore, to address the limitations of standard pipelines where visibility checks (i.e., thresholding) often discard valid pixels and multi-view warped image fusion relies on naive mean aggregation, our module is coupled with an implicit confidence weighting mechanism that selectively suppresses unreliable evidence. Consequently, our approach outperforms prior warping-based Gaussian Splatting, preserving high-frequency quality while achieving 2.13 times faster FPS.
Chinese Translation
高斯溅射通过结合基于变形的技术取得了显著的改进。然而，这些方法由于几何的不确定性而面临像素级的不准确性。这种不确定性导致了变形图像中的空间错位，干扰了基于变形方法中的残差学习，并在根本上限制了修正的收益，特别是在细结构和高频细节上。基于我们对有用视觉线索在轻微位移下并未丢失而是局部保留的洞察，我们提出了基于几何的可变形聚合（GADA）。该方法引入了一个带有可变形偏移的迭代细化模块，以主动纠正空间错位并恢复这些位移的线索。此外，为了解决标准流程中可见性检查（即阈值处理）常常丢弃有效像素以及多视角变形图像融合依赖于简单均值聚合的局限性，我们的模块结合了隐式置信加权机制，选择性地抑制不可靠的证据。因此，我们的方法在保持高频质量的同时，性能优于之前的基于变形的高斯溅射，达到了2.13倍的帧率提升。
cs.CV / 67 / 2607.00596
Semantic-Guided Reading Order Reconstruction in Historical Armenian Newspapers with LLMs
基于语义的历史亚美尼亚报纸阅读顺序重建研究
Abstract
This paper addresses reading order reconstruction in historical Armenian newspapers, which combine complex layouts with limited language resources. We introduce a new annotated dataset of 66 pages and compare geometric heuristics, YOLO-based layout parsing, an end-to-end document model ECLAIR, and a hybrid method combining semantic zone detection with a generative LLM. Our hybrid method achieves the lowest error rates of all evaluated approaches, reducing ordering errors by up to 76% over the strongest geometric baseline, and remains robust in multi-page settings and under noisy OCR. Rather than targeting production the method is designed as a data bootstrapping strategy enabling rapid annotation in highly under-resourced scenarios. Alongside the dataset, we release a specialized Tesseract OCR model for historical Armenian print.
Chinese Translation
本文探讨了历史亚美尼亚报纸中的阅读顺序重建问题，这些报纸结合了复杂的版面布局和有限的语言资源。我们引入了一个新的标注数据集，包含66页，并比较了几种方法，包括几何启发式方法、基于YOLO的布局解析、端到端文档模型ECLAIR，以及一种将语义区域检测与生成型大语言模型（LLM）相结合的混合方法。我们的混合方法在所有评估的方案中实现了最低的错误率，相较于最强的几何基线，减少了多达76%的排序错误，并且在多页设置和嘈杂的光学字符识别（OCR）环境中保持了稳健性。该方法并非针对生产，而是作为一种数据引导策略，旨在在资源极为匮乏的场景中实现快速标注。除了数据集外，我们还发布了一个专门针对历史亚美尼亚印刷品的Tesseract OCR模型。
cs.CV / 68 / 2607.00606
Retrieved Images as Visual Thought: Training-Free Multimodal In-Context Learning for the Open-vs-Closed Gap
作为视觉思维的检索图像：无训练的多模态上下文学习以缩小开放与封闭的差距
Abstract
Recent work on Thinking with Images makes vision a dynamic part of reasoning, but does so through generation: the model invokes external tools, synthesizes code, or imagines new imagery, each at the cost of a tool protocol, brittle code, or an expensive training pipeline. A fourth route makes vision dynamic without generating anything, by retrieving labeled exemplar images and reasoning over them, yet it remains underexplored despite being train-free. We present ReVisIT, a train-free framework that realizes this retrieval-based route by treating each retrieved image-label pair as a unit of visual thought. ReVisIT combines structured class definitions, per-query multimodal retrieval of exemplars, and alternating user/assistant injection of those exemplars before joint multi-attribute decoding, and degrades gracefully to whichever components a task admits. On VL-ICL Bench Fast Open MiniImageNet, Qwen3-VL-30B-A3B with ReVisIT reaches 98.5% at 4-shot, statistically indistinguishable from the 72B LLaVA-OneVision SOTA (98.7%) on this near-saturated task at about 1/2.4 the parameters, while the same backbone without the scaffold sits at chance. The turns layer alone adds 26.1 points to GPT-4.1 on free-form concept induction (Bongard-OpenWorld), and the full stack yields a 4-6 point macro gain across three backbones on MAAC-Bench, a new license-clean 27-class, 5-attribute benchmark, significant by paired bootstrap on the curator-derived attributes. Component analysis shows that retrieval-plus-turns is the universal lever while structured definitions are need-adaptive, and that 83% of the retrieval gain comes from retrieval quality rather than from the presence of exemplars. MAAC-Bench is released with a rubric-grounded LLM verification protocol that replaces author spot-check on subjective attributes.
Chinese Translation
近期关于图像思维的研究使得视觉成为推理的动态部分，但这一过程依赖于生成：模型调用外部工具、合成代码或想象新图像，每种方式都需要工具协议、脆弱的代码或昂贵的训练流程。第四种途径在不生成任何内容的情况下使视觉变得动态，通过检索标记的示例图像并对其进行推理，然而尽管这一方法无需训练，仍然未被充分探索。我们提出了ReVisIT，一个无训练的框架，通过将每个检索到的图像-标签对视为视觉思维的单元，实现了这一基于检索的途径。ReVisIT结合了结构化的类别定义、每个查询的多模态示例检索，以及在联合多属性解码之前交替注入这些示例的用户/助手，并能够优雅地降级到任务允许的任何组件。在VL-ICL Bench Fast Open MiniImageNet上，使用ReVisIT的Qwen3-VL-30B-A3B在4-shot情况下达到了98.5%的准确率，与72B LLaVA-OneVision的最新技术水平（98.7%）在这一近饱和任务上统计上无显著差异，而同样的基础模型在没有支架的情况下仅达到随机水平。仅转弯层就为GPT-4.1在自由形式概念归纳（Bongard-OpenWorld）上增加了26.1个点，而完整堆栈在MAAC-Bench上在三个基础模型上产生了4-6个点的宏观增益，这是一个新的许可证清晰的27类、5属性基准，通过配对自助法在策展人派生属性上显著。组件分析表明，检索加转弯是普遍的杠杆，而结构化定义是需求适应性的，并且83%的检索增益来自于检索质量而非示例的存在。MAAC-Bench发布了一个基于评分的LLM验证协议，替代了作者对主观属性的抽查。
cs.CV / 69 / 2607.00609
Diffusion-Based Multi-Class Normality for OOD Detection: An Application to CDP Authentication
基于扩散的多类正常性用于OOD检测：在CDP认证中的应用
Abstract
Reconstruction-based generative models offer a natural framework for unsupervised out-of-distribution (OOD) detection, but multi-class normality modelling requires a single detector to capture multiple in-distribution manifolds and produce comparable anomaly scores across classes. We study this problem in copy detection pattern (CDP) authentication, where authentic and counterfeit samples are visually similar but differ in subtle printing-and-digitisation (P\&D) signatures. We propose a diffusion based multi-class normality framework in which a single class-conditional ControlNet is trained exclusively on authentic CDPs from multiple P\&D classes and detects counterfeits through reconstruction error under authentic-class conditioning. We further introduce dual template masking, which hides complementary regions of the input template and scores only withheld pixels, reducing reliance on visible binary structure. On the Indigo 1 x 1 Base dataset, the proposed method outperforms traditional and adapted generative baselines under multi-class authentic-versus-counterfeit evaluation, without using counterfeit samples for training or threshold calibration.
Chinese Translation
基于重建的生成模型为无监督的分布外（OOD）检测提供了一个自然框架，但多类正常性建模需要一个单一的检测器来捕捉多个内部分布流形，并在各类之间产生可比较的异常分数。我们在复制检测模式（CDP）认证中研究了这个问题，其中真实样本和伪造样本在视觉上相似，但在细微的印刷和数字化（P&D）特征上有所不同。我们提出了一种基于扩散的多类正常性框架，其中一个单一的类条件ControlNet专门在来自多个P&D类的真实CDP上进行训练，并通过在真实类别条件下的重建误差来检测伪造品。我们进一步引入了双模板遮蔽技术，该技术隐藏输入模板的互补区域，仅对被遮蔽的像素进行评分，从而减少对可见二进制结构的依赖。在Indigo 1 x 1 Base数据集上，所提出的方法在多类真实与伪造评估中优于传统和改编的生成基线，且不需要使用伪造样本进行训练或阈值校准。
cs.CV / 70 / 2607.00620
Identifying Latent Concepts and Structures for Generalized Category Discovery
识别潜在概念和结构以实现广义类别发现
Abstract
Generalized Category Discovery (GCD) aims to recognize known classes while autonomously discovering novel ones in open-world settings. However, current approaches primarily focus on designing clustering objectives, often overlooking a critical bottleneck: standard vision backbones yield high-rank, entangled token representations that are ill-suited for unsupervised discovery of latent concepts and structures. In this paper, we propose Compositional Primitive Fields (CPF-GCD), a novel representation learning framework that reshapes the feature space to make such latent structure identifiable by enforcing a low-rank compositional organization. Our core hypothesis is that all categories, whether known or novel, can be expressed as compositions and spatial arrangements of a finite set of learnable visual primitives that capture reusable concepts. CPF instantiates this geometric constraint via a spatial field mechanism. Inserted between the backbone and the head, it rewrites noisy patch tokens through low-rank primitive mixtures, effectively decomposing images into reusable atomic parts and their spatial layouts. By explicitly modeling the spatial distribution of primitives, CPF enables novel categories to emerge naturally as new activation patterns over a shared vocabulary. This shifts the focus of representation from merely partitioning global embeddings to constructing a structured and separable primitive field. Extensive experiments demonstrate that CPF serves as a generic, plug-and-play module that consistently boosts performance across diverse GCD baselines, validating that identifying and leveraging low-rank compositional structure is a crucial inductive bias for open-world recognition.
Chinese Translation
广义类别发现（Generalized Category Discovery, GCD）旨在识别已知类别，同时在开放世界环境中自主发现新类别。然而，当前的方法主要集中于设计聚类目标，常常忽视一个关键瓶颈：标准视觉骨干网络产生的高秩、纠缠的标记表示不适合无监督地发现潜在概念和结构。本文提出了一种新颖的表示学习框架——组合原始场（Compositional Primitive Fields, CPF-GCD），该框架通过强制低秩组合组织来重塑特征空间，使潜在结构可识别。我们的核心假设是，所有类别，无论是已知的还是新的，都可以表示为有限可学习视觉原始元素的组合和空间排列，这些原始元素捕捉可重用的概念。CPF通过空间场机制实现这一几何约束。它被插入在骨干网络和输出层之间，通过低秩原始混合重写噪声补丁标记，有效地将图像分解为可重用的原子部分及其空间布局。通过明确建模原始元素的空间分布，CPF使新类别能够自然地作为共享词汇上的新激活模式出现。这将表示的重点从单纯划分全局嵌入转向构建一个结构化且可分离的原始场。大量实验表明，CPF作为一个通用的即插即用模块，持续提升了多种GCD基线的性能，验证了识别和利用低秩组合结构是开放世界识别的重要归纳偏置。
cs.CV / 71 / 2607.00622
Learning to Watch: Active Video Anomaly Understanding via Interleaved Policy Optimization
学习观察：通过交错策略优化实现主动视频异常理解
Abstract
Video anomaly understanding (VAU) relies on sparse, context-dependent cues. However, existing passive paradigms suffer from observational aliasing, where static sampling fails to disambiguate semantically distinct events. To overcome this, we propose $Anom\text{-}\pi$, a closed-loop framework that reconceptualizes video understanding as an active sequential decision-making process within a dynamic environment. Inspired by human video-reviewing behavior, this framework unifies internal cognitive reasoning and strategic evidence acquisition into an interleaved policy, utilizing temporal atomic operators such as local backtracking, temporal expansion, and fine-grained sampling to endow the model with perceptual proactivity. To learn such complex interaction strategies under video-level weak supervision, we design Interactive Direct Preference Optimization (iDPO) to achieve trajectory-level policy alignment, guided by an Active Evidence Inquiry (AEI) utility that balances task success, informative evidence acquisition, and interaction cost. This approach enables the agent to learn to actively disambiguate hypotheses while suppressing redundant exploration. Extensive experiments demonstrate that our framework, with only 2B parameters, achieves highly competitive performance, significantly outperforming state-of-the-art large-scale VAU models in complex scenarios.
Chinese Translation
视频异常理解（VAU）依赖于稀疏的、依赖上下文的线索。然而，现有的被动范式存在观察别名问题，即静态采样无法区分语义上不同的事件。为了解决这一问题，我们提出了 $Anom ext{-} ext{π}$，一个闭环框架，将视频理解重新概念化为在动态环境中的主动序列决策过程。受人类视频回顾行为的启发，该框架将内部认知推理和战略证据获取统一为交错策略，利用局部回溯、时间扩展和细粒度采样等时间原子操作，使模型具备感知主动性。为了在视频级弱监督下学习这种复杂的交互策略，我们设计了互动直接偏好优化（Interactive Direct Preference Optimization，iDPO），以实现轨迹级策略对齐，受主动证据询问（Active Evidence Inquiry，AEI）效用的指导，该效用在任务成功、信息性证据获取和交互成本之间进行平衡。这种方法使代理能够主动消歧假设，同时抑制冗余探索。大量实验表明，我们的框架在仅有20亿参数的情况下，取得了高度竞争的性能，在复杂场景中显著超越了最先进的大规模VAU模型。
cs.CV / 72 / 2607.00638
Uncertainty-aware tree height change regression
考虑不确定性的树高变化回归
Abstract
Monitoring canopy height change is essential for understanding carbon sinks and forest dynamics. Remote sensing enables consistent, large-scale observations of such changes, increasingly integrated with deep learning architectures such as Geospatial Foundation Models (GFMs). However, existing methods and datasets frame the problem as binary change detection, which overlooks both the continuous nature of change, especially for vegetation, and the inherent uncertainty in labels. We present the Canopy Height Change (CHC) dataset, providing 3 $\mathrm{m}$ resolution continuous canopy height differences and associated spatially resolved uncertainties across 10598 $\mathrm{km}^2$ of northern and western Spain. The dataset is paired with a co-located time series of PlanetScope satellite imagery. Based on the dataset, we introduce the task of uncertainty-aware change regression, associated metrics and strategies for fine-tuning GFMs. Furthermore, we evaluate state-of-the-art GFMs and highlight promising directions and remaining challenges for advancing continuous canopy height change estimation.
Chinese Translation
监测树冠高度变化对于理解碳汇和森林动态至关重要。遥感技术能够对这些变化进行一致的大规模观测，并日益与深度学习架构（如地理空间基础模型 Geospatial Foundation Models, GFMs）相结合。然而，现有的方法和数据集将问题框定为二元变化检测，这忽视了变化的连续性，尤其是在植被方面，以及标签固有的不确定性。我们提出了树冠高度变化（Canopy Height Change, CHC）数据集，提供了在西班牙北部和西部10598平方公里区域内的3米分辨率连续树冠高度差异及相关的空间分辨不确定性。该数据集与PlanetScope卫星影像的共定位时间序列相配对。基于该数据集，我们引入了考虑不确定性的变化回归任务，相关指标以及微调GFMs的策略。此外，我们评估了最先进的GFMs，并强调了推进连续树冠高度变化估计的有前景方向和剩余挑战。
cs.CV / 73 / 2607.00647
Not All Prediction Targets Keep Training-Free Diffusion Guidance on the Manifold
并非所有预测目标都能在流形上保持无训练扩散引导
Abstract
Training-free guidance (TFG) steers a pretrained diffusion model toward a desired attribute at inference. To be effective, this guidance must be applied from the earliest, high-noise steps of sampling. Because its objective (a classifier or energy) is defined on clean images, $\epsilon$- and $v$-prediction models must first estimate the clean image $\hat{x}$ from the noisy state at each step, and the accuracy of that estimate determines how easily guidance drifts off the data manifold. $x$-prediction, a recent alternative, outputs the clean image directly, removing this source of error even at high noise. This is our motivation. We provide a theoretical analysis of how each prediction target shapes this accuracy, and introduce guided-class FID (Child FID), a metric that exposes the manifold damage standard evaluation misses. Experiments on a new fine-grained bird benchmark and on style transfer confirm that $x$-prediction keeps guided samples on the manifold most reliably, making it the strongest foundation for training-free guidance. Code is available at https://github.com/ManLuML/on-manifold-tfg
Chinese Translation
无训练引导（Training-free guidance, TFG）在推理时引导预训练的扩散模型朝向所需属性。为了有效，这种引导必须从采样的最早高噪声步骤开始应用。由于其目标（分类器或能量）是在干净图像上定义的，$ ext{ε}$-和$ ext{v}$-预测模型必须首先从每一步的噪声状态中估计干净图像$ ext{hat{x}}$，而该估计的准确性决定了引导偏离数据流形的难易程度。最近提出的$ ext{x}$-预测直接输出干净图像，即使在高噪声下也消除了这一误差来源。这是我们的动机。我们提供了每个预测目标如何影响这种准确性的理论分析，并引入了引导类FID（Child FID），这一指标揭示了标准评估遗漏的流形损伤。在一个新的细粒度鸟类基准和风格迁移实验中，结果确认$ ext{x}$-预测最可靠地保持引导样本在流形上，使其成为无训练引导的最强基础。代码可在 https://github.com/ManLuML/on-manifold-tfg 获取。
cs.CV / 74 / 2607.00654
Linguistic Relative Policy Optimization for Video Anomaly Reasoning
视频异常推理的语言相对策略优化
Abstract
Video anomaly detection (VAD) with multimodal large language models has shown strong potential, yet most existing methods still depend on large-scale annotations or expert-designed priors, limiting their ability to acquire anomaly knowledge with as little human intervention as possible. To address this, we propose Linguistic Relative Policy Optimization (LRPO), which distills group-relative semantic advantages from multiple reasoning trajectories into a linguistically expressed anomaly experience prior, and adapts the model by injecting this prior into the context to steer its output distribution without any parameter updates. LRPO builds two complementary experience representations: general experience captures transferable anomaly preferences across scenarios, while scenario experience models context-dependent anomaly rules for targeted refinement. To further improve the learned experience, we introduce an anomaly alignment reward that guides trajectory optimization to match human risk preferences and reinforce temporally grounded reasoning. Extensive experiments on XD-Violence, UCF-Crime, and UBnormal demonstrate that LRPO significantly outperforms existing state-of-the-art methods under tuning-free settings.
Chinese Translation
多模态大型语言模型在视频异常检测（VAD）中展现出强大的潜力，但大多数现有方法仍依赖于大规模标注或专家设计的先验知识，这限制了它们在尽可能少的人为干预下获取异常知识的能力。为此，我们提出了语言相对策略优化（Linguistic Relative Policy Optimization, LRPO），该方法从多个推理轨迹中提炼出群体相对语义优势，形成一种以语言表达的异常经验先验，并通过将该先验注入上下文中来调整模型，从而引导其输出分布而无需任何参数更新。LRPO构建了两种互补的经验表示：一般经验捕捉跨场景的可转移异常偏好，而场景经验则建模上下文依赖的异常规则以进行针对性优化。为了进一步改善学习到的经验，我们引入了一种异常对齐奖励，指导轨迹优化以匹配人类风险偏好，并强化时间上扎根的推理。在XD-Violence、UCF-Crime和UBnormal数据集上的大量实验表明，LRPO在无调优设置下显著优于现有的最先进方法。
cs.CV / 75 / 2607.00672
DART: Difficulty-Adaptive Routing for Zero-Shot Video Temporal Grounding
DART：零样本视频时间定位的难度自适应路由
Abstract
Zero-shot video temporal grounding (VTG) localizes events in untrimmed videos from natural language queries without task-specific training. Existing methods rely on frame-query feature matching, which suffices for simple events but struggles with complex multi-stage queries that require understanding temporal ordering and causal structure -- a disparity we call the reasoning gap. We propose DART (Difficulty-Adaptive Routing for Temporal Grounding), which bridges this gap by coupling difficulty-aware routing with structured reasoning in large vision-language models. A query-conditioned Determinantal Point Process (DPP) serves a dual role: selecting diverse, query-relevant keyframes as temporal evidence, and providing spectral entropy as a difficulty indicator. Simple queries are routed to a Fast path for direct prediction, while complex queries follow a Slow path with Temporal Markup Prompting, which decomposes localization into global event analysis, per-frame temporal role annotation, and boundary extraction. On Charades-STA and ActivityNet Captions, DART achieves state-of-the-art zero-shot performance across both identically distributed and multiple out-of-distribution settings, improving mIoU by up to 3.5 points over the strongest baseline while using over 7 times fewer frames. The project homepage is available at https://dart-vtg.github.io/.
Chinese Translation
零样本视频时间定位（VTG）能够根据自然语言查询在未剪辑的视频中定位事件，而无需特定任务的训练。现有方法依赖于帧-查询特征匹配，这对于简单事件是足够的，但在处理需要理解时间顺序和因果结构的复杂多阶段查询时却显得力不从心——我们称之为推理差距。我们提出了DART（难度自适应路由用于时间定位），通过将难度感知路由与大型视觉-语言模型中的结构化推理结合，弥补了这一差距。查询条件的决定性点过程（Determinantal Point Process, DPP）发挥了双重作用：选择多样的、与查询相关的关键帧作为时间证据，并提供谱熵作为难度指示器。简单查询被路由到快速路径以进行直接预测，而复杂查询则遵循慢速路径，采用时间标记提示（Temporal Markup Prompting），将定位分解为全局事件分析、逐帧时间角色注释和边界提取。在Charades-STA和ActivityNet Captions上，DART在相同分布和多个分布外设置中均实现了最先进的零样本性能，相较于最强基线提高了多达3.5个点的mIoU，同时使用的帧数减少了超过7倍。项目主页可访问：https://dart-vtg.github.io/
cs.CV / 76 / 2607.00678
ABot-M0.5: Unified Mobility-and-Manipulation World Action Model
ABot-M0.5：统一的移动与操控世界动作模型
Chen, Ronghan, Yang, Yandan, Tang, Zuojin, Huo, Dongjie, Lin, Tong, Wu, Haoning, Liu, Haoyun, Chen, Yuzhi, Zheng, Lulu, Yuan, Botai, Li, Tianlun, Wang, Mingxin, Qi, Dekang, Hu, Bin, Mei, Wei, Xuan, Yuze, Yang, Haolong, Zhu, Yanqing, Xu, Mu, Ma, Zhiheng, Chang, Xinyuan
Abstract
Mobile manipulation is a key capability for general-purpose robots, yet remains challenging for current embodied learning methods. VLA policies are typically reactive and lack explicit world modeling, while existing World Action Models (WAMs) are still poorly aligned with the structure of mobile manipulation: they operate on coarse video chunks, model entangled navigation-manipulation actions, and train inverse dynamics under supervision that does not match autoregressive inference. As a result, they often miss fine-grained contact dynamics, suffer from action-distribution conflicts, and accumulate errors over long-horizon rollouts. We propose ABot-M0.5, a new WAM built on the insight that mobile manipulation requires alignment at three levels: temporal granularity, action space, and train-test consistency. To align temporal granularity, we introduce intermediate latent actions that capture local visual state transitions and serve as an bridging action space between video latents and embodiment-specific controls. To align action space, we design a dual-level Mixture-of-Transformers architecture that disentangles both modality representations and heterogeneous action subspaces such as base movement and arm manipulation. To align inference conditions, we propose the dream-forcing training strategy that progressively trains inverse dynamics on model-predicted videos, improving train-test alignment and robustness during autoregressive prediction. Experiments on challenging mobile and fine-grained manipulation benchmarks demonstrate that ABot-M0.5 achieves state-of-the-art performance in both long-horizon task success and finegrained control accuracy. These results highlight the critical importance of granularity-aligned, action-disentangled, and inference-consistent world-action modeling.
Chinese Translation
移动操控是通用机器人的一项关键能力，但对于当前的具身学习方法仍然具有挑战性。VLA（变换学习算法）策略通常是反应性的，缺乏明确的世界建模，而现有的世界动作模型（WAMs）与移动操控的结构仍然不够匹配：它们在粗糙的视频片段上运行，建模纠缠的导航-操控动作，并在与自回归推断不匹配的监督下训练逆动力学。因此，它们往往忽视细粒度的接触动态，遭受动作分布冲突，并在长时间的滚动预测中累积误差。我们提出了ABot-M0.5，这是一种新的WAM，基于移动操控需要在三个层面上对齐的见解：时间粒度、动作空间和训练-测试一致性。为了对齐时间粒度，我们引入了中间潜在动作，以捕捉局部视觉状态转变，并作为视频潜在与具身特定控制之间的桥接动作空间。为了对齐动作空间，我们设计了一种双层混合变换器（Mixture-of-Transformers）架构，解耦了不同模态表示和异构动作子空间，如基础移动和手臂操控。为了对齐推断条件，我们提出了梦强迫训练策略，逐步在模型预测的视频上训练逆动力学，提高训练-测试一致性和自回归预测过程中的鲁棒性。在具有挑战性的移动和细粒度操控基准上的实验表明，ABot-M0.5在长时间任务成功率和细粒度控制精度方面达到了最先进的性能。这些结果突显了粒度对齐、动作解耦和推断一致的世界动作建模的重要性。
cs.CV / 77 / 2607.00687
LUMA: Benchmarking Segmentation via a Lightweight Universal Mask Adapter
LUMA：通过轻量级通用掩膜适配器进行分割基准测试
Abstract
Comparing transformer backbones for image segmentation is confounded: each is paired with a different decoder, recipe, and pretraining, so reported differences rarely reflect the backbone itself. We introduce the Lightweight Universal Mask Adapter (LUMA), a lightweight, backbone-agnostic mask-transformer head that treats any backbone as a black-box feature extractor, letting a set of queries read from its features through cheap cross-attention. LUMA matches the accuracy of EoMT, the state-of-the-art efficient ViT-segmenter, at lower cost, while attaching unchanged to isotropic, hierarchical, convolutional, and mixture-of-experts backbones alike. Holding this head fixed, we benchmark 20 backbones, 11 pretraining schemes and a range of resolutions on ADE20K and Cityscapes under one modern recipe. We find that ``efficient'' token mixers fail to deliver efficiency even at the high resolutions that motivate them, with plain ViT holding the throughput Pareto-front at every resolution. Additionally, the pretraining objective, not the architecture, the lever the field has tuned hardest, governs segmentation quality.
Chinese Translation
比较用于图像分割的变换器骨干网络存在诸多困扰：每个骨干网络都配备了不同的解码器、配方和预训练，因此报告的差异很少反映骨干网络本身。我们提出了轻量级通用掩膜适配器（LUMA），这是一种轻量级、与骨干网络无关的掩膜变换器头，它将任何骨干网络视为黑箱特征提取器，通过廉价的交叉注意力让一组查询从其特征中读取信息。LUMA 在成本更低的情况下达到了 EoMT（最先进的高效 ViT 分割器）的准确性，同时可以不变地附加到各类各向同性、层次化、卷积和专家混合骨干网络上。在固定该头的情况下，我们在 ADE20K 和 Cityscapes 上基于一种现代配方对 20 个骨干网络、11 种预训练方案和一系列分辨率进行了基准测试。我们发现，“高效”的令牌混合器即使在激励它们的高分辨率下也未能提供效率，而普通的 ViT 在每个分辨率下都保持着吞吐量的帕累托前沿。此外，预训练目标，而非架构，成为该领域最为关注的调优杠杆，决定了分割质量。
cs.CV / 78 / 2607.00696
Imprint: Online Memory Compression for Long-Horizon Egocentric QA
Imprint：用于长时间跨度自我中心问答的在线记忆压缩
Abstract
Long-horizon egocentric question answering involves answering about events that have occurred hours or days in the past. This requires memory representations that remain both retrieval-effective and scalable over days or weeks of recording. Existing long-horizon egocentric QA methods construct memory as hierarchical textual summaries of observations. While effective for reducing memory size, summarization optimizes for descriptive compression rather than retrieval: repeated interactions are absorbed into coarse textual descriptions instead of being preserved as explicit, recurring memory units, making long-horizon evidence aggregation difficult. We propose Imprint, an interaction-centric memory framework that formulates long-horizon egocentric memory as an online memory compression problem rather than summarization. Incoming observations are first represented as structured Interaction Records and continuously organized into recurring interaction patterns. Using human memory consolidation signals of recurrence, recency, and distinctiveness, Imprint selectively retains and compresses interactions into a compact retrieval-oriented memory. We evaluate Imprint on EgoLifeQA, a seven-day egocentric benchmark containing questions that require reasoning over interactions occurring hours to days before the query. With the same LLM, Imprint improves QA accuracy from 31.0% to 35.8%, increases evidence-grounded answers by $6\times$ compared with EgoRAG, reduces memory footprint by $2.3\times$, and decreases retrieval latency by $11.8\times$. These results demonstrate that memory compression provides a scalable and retrieval-effective foundation for long-horizon egocentric question answering.
Chinese Translation
长时间跨度的自我中心问答涉及回答发生在数小时或数天之前的事件。这需要在记录数天或数周内保持有效检索和可扩展的记忆表示。现有的长时间跨度自我中心问答方法将记忆构建为观察的层次文本摘要。虽然这种方法在减少记忆大小方面有效，但摘要优化的是描述性压缩而非检索：重复的交互被吸收到粗略的文本描述中，而不是作为明确的、重复的记忆单元被保留，从而使得长时间跨度的证据聚合变得困难。我们提出了Imprint，一个以交互为中心的记忆框架，将长时间跨度的自我中心记忆表述为在线记忆压缩问题，而非摘要。首先，将输入的观察表示为结构化的交互记录，并不断组织成重复的交互模式。利用人类记忆巩固信号（如重复性、近期性和独特性），Imprint选择性地保留并压缩交互，形成紧凑的以检索为导向的记忆。我们在EgoLifeQA上评估Imprint，这是一个为期七天的自我中心基准，包含需要推理数小时到数天前发生的交互的问题。在相同的LLM下，Imprint将问答准确率从31.0%提高到35.8%，与EgoRAG相比，基于证据的答案增加了6倍，内存占用减少了2.3倍，检索延迟减少了11.8倍。这些结果表明，记忆压缩为长时间跨度的自我中心问答提供了可扩展且有效检索的基础。
cs.CV / 79 / 2607.00710
Creating Impactful Autonomous Driving Datasets: A Strategic Guide from Research Gap to Benchmark
创建具有影响力的自动驾驶数据集：从研究空白到基准的战略指南
Schwarzkopf, Richard, Merkert, Jonas, Bieder, Frank, Bätz, Annika, Blumberg, Alexander, Fernandez, Carlos, Hauser, Felix, Immel, Fabian, Kinzig, Christian, Königshof, Hendrik, Konstantinidis, Fabian, Lauer, Martin, Poh, Willi, Rack, Nils, Rösch, Kevin, Shen, Yinzhe, Steiner, Marlon, Stepanov, Gleb, Strutz, Dominik, Taş, Ömer Şahin, Truetsch, Julian, Wang, Kaiwen, Wagner, Royden, Pauls, Jan-Hendrik, Stiller, Christoph
Abstract
Well-designed autonomous driving datasets have fundamentally shaped research progress, yet existing literature primarily describes what datasets contain rather than how to strategically design impactful ones. This is especially limiting for small and medium-sized labs and startups that cannot afford to misallocate scarce resources. We argue that impactful dataset creation begins with a diagnosis: whether a research question is blocked by a data problem or an evaluation problem, and proceeds by selecting the minimal data operator(s) that closes the resulting gap, recording new data only when no cheaper operator(s) suffices. We analyze the evolution of major autonomous driving (AD) datasets through this lens and distill a strategic framework spanning gap identification, operator choice, sensor suite design, and annotation strategy. We ground the framework in a running case study of our KITScenes dataset family. The datasets are available at: https://kitscenes.com/
Chinese Translation
精心设计的自动驾驶数据集在根本上推动了研究进展，然而现有文献主要描述数据集的内容，而非如何战略性地设计具有影响力的数据集。这对于无法承担资源错误分配的小型和中型实验室及初创企业尤其限制。我们认为，具有影响力的数据集创建始于诊断：研究问题是由于数据问题还是评估问题所阻碍，并通过选择最小的数据操作员来填补所产生的空白，只有在没有更便宜的操作员足够时才记录新数据。我们通过这一视角分析了主要自动驾驶（AD）数据集的演变，并提炼出一个涵盖空白识别、操作员选择、传感器配置设计和标注策略的战略框架。我们在我们的KITScenes数据集系列的实际案例研究中为该框架提供了基础。数据集可在以下网址获取：https://kitscenes.com/
cs.CV / 80 / 2607.00712
Towards Memory-Efficient Autoregressive Video Generation via Instance-Specific Parametric Absorption
通过实例特定的参数吸收实现内存高效的自回归视频生成
Abstract
Autoregressive (AR) streaming models have emerged as a powerful paradigm for long video generation. However, the linearly growing Key-Value (KV) cache poses a significant bottleneck, leading to memory overload and degraded inference throughput. A common compression method is to drop redundant KV tokens, which often breaks long-range dependencies, resulting in temporal flickering and identity loss. In this paper, we propose Instance-Specific Parametric Absorption (ISPA), a novel framework that shifts the KV cache compression from discarding to distilling. The core idea is to transit a subset of layers from Full-Attention (F-Layers) to memory-efficient Local-Attention (L-Layers) by "absorbing" historical context into the model's weights. Specifically, during a brief warmup phase, ISPA monitors the output discrepancy between global and local attention. At the transition point, we solve a closed-form least-squares problem to compute an instance-specific weight modulation that compensates for the missing history. Experiments across architectures (1.3B to 14B) demonstrate that ISPA can remove up to 50\% of the KV cache with near-lossless visual quality. We hope this perspective encourages future work to explore parametric memory consolidation beyond external token-level cache management for streaming generative models.
Chinese Translation
自回归（AR）流模型已成为长视频生成的一种强大范式。然而，线性增长的键值（KV）缓存构成了一个显著的瓶颈，导致内存过载和推理吞吐量下降。一种常见的压缩方法是丢弃冗余的KV令牌，这通常会破坏长距离依赖，导致时间闪烁和身份丢失。在本文中，我们提出了实例特定的参数吸收（ISPA），这是一个新颖的框架，将KV缓存的压缩从丢弃转变为提炼。其核心思想是通过将历史上下文“吸收”到模型的权重中，将一部分层从全注意力（F-Layers）转换为内存高效的局部注意力（L-Layers）。具体而言，在简短的预热阶段，ISPA监测全局和局部注意力之间的输出差异。在过渡点，我们解决一个封闭形式的最小二乘问题，以计算补偿缺失历史的实例特定权重调制。跨架构（1.3B到14B）的实验表明，ISPA可以去除多达50%的KV缓存，同时保持接近无损的视觉质量。我们希望这一视角能鼓励未来的研究探索超越外部令牌级缓存管理的参数化内存整合，以支持流生成模型。
cs.CV / 81 / 2607.00716
Partial Skeleton Visibility for Action Recognition: A Constrained Field-of-View Approach
动作识别中的部分骨架可见性：一种受限视野的方法
Abstract
Skeleton-based action recognition has achieved remarkable success by exploiting joint coordinates and their topological connections, yet prevailing methods overwhelmingly assume complete and clean skeleton inputs. In real-world deployments, such as egocentric vision, crowded surveillance, wearable devices, or edge robotics, limited field-of-view (FoV) frequently causes substantial joint visibility dropout, leading to severe performance degradation that existing models are largely unprepared to handle. To bridge this critical yet underexplored gap, we introduce PartialVisGraph, a novel hypergraph framework tailored for robust skeleton action recognition under constrained FoV. We first construct highly expressive hypergraphs by introducing learnable virtual hyperedges that form a soft incidence matrix, capturing flexible high-order dependencies beyond conventional pairwise graphs. We then propose the Single-Head Sample-Adaptive Transformer, which adaptively aggregates joint features onto hyperedges while explicitly incorporating a visibility prior. This prior selectively gates information flow, preventing occluded or out-of-view joints from corrupting reliable feature propagation. We further establish rigorous evaluation protocols with realistic FoV simulation benchmarks on NTU RGB+D 60 and 120. Extensive experiments demonstrate that PartialVisGraph consistently achieves state-of-the-art accuracy under partial visibility, with gains of up to 68.8\% on subsets with severe FoV restrictions compared to recent strong baselines, while remaining superior on full-visibility settings. Our approach offers a principled and practical pathway toward deployable skeleton-based action understanding in unconstrained environments.
Chinese Translation
基于骨架的动作识别通过利用关节坐标及其拓扑连接取得了显著成功，但现有方法大多假设输入的骨架是完整且干净的。在实际应用中，例如自我中心视觉、拥挤监控、可穿戴设备或边缘机器人，有限的视野（FoV）常常导致关节可见性显著下降，从而造成严重的性能退化，而现有模型在这方面大多没有准备。为了解决这一关键但尚未充分探索的问题，我们提出了PartialVisGraph，这是一种新颖的超图框架，旨在在受限视野下实现鲁棒的骨架动作识别。我们首先通过引入可学习的虚拟超边构建高度表达性的超图，这些超边形成一个软关联矩阵，捕捉超出传统成对图的灵活高阶依赖关系。然后，我们提出了单头样本自适应变换器（Single-Head Sample-Adaptive Transformer），该变换器自适应地将关节特征聚合到超边上，同时明确地结合可见性先验。该先验选择性地控制信息流，防止被遮挡或超出视野的关节破坏可靠的特征传播。我们进一步在NTU RGB+D 60和120上建立了严格的评估协议，并进行了现实的视野模拟基准测试。大量实验表明，PartialVisGraph在部分可见性下始终实现了最先进的准确率，在严重视野限制的子集上相比于近期强基线提高了多达68.8 ext{%}，同时在完全可见性设置下仍然表现优越。我们的方法为在不受限环境中可部署的基于骨架的动作理解提供了一条有原则且实用的路径。
cs.CV / 82 / 2607.00726
AV-SyncBench: Decoupled Benchmarking of Temporal and Semantic Audio-Visual Synchronization
AV-SyncBench：音频-视觉同步的时间和语义解耦基准测试
Abstract
Audio-visual feature extraction is a fundamental component of multimodal understanding and generation tasks. However, existing evaluation protocols for feature extraction models exhibit dimensional bias, typically focusing on either semantic matching or temporal offset detection. Moreover, their data construction remains coupled, preventing independent assessment of temporal and semantic consistency. We propose AV-SyncBench, the first benchmark to fully separate temporal and semantic evaluation for audio-visual synchronization. Built from in-the-wild videos, it spans Voice, Music, and Sound across 10 scenarios and 5 challenge tasks. Data are automatically filtered and manually verified to ensure on-screen sound sources. The benchmark contains 3,269 videos and 38,390 samples, and we evaluate five representative models to quantify feature quality for alignment and downstream tasks. The code and dataset are available at: https://fgt7t6g.github.io/AV-SyncBench.
Chinese Translation
音频-视觉特征提取是多模态理解和生成任务的基本组成部分。然而，现有的特征提取模型评估协议存在维度偏差，通常仅关注语义匹配或时间偏移检测。此外，它们的数据构建仍然是耦合的，阻碍了时间和语义一致性的独立评估。我们提出了AV-SyncBench，这是第一个完全分离音频-视觉同步的时间和语义评估的基准测试。该基准测试基于真实场景视频，涵盖了10种场景和5个挑战任务，包括语音、音乐和声音。数据经过自动过滤和人工验证，以确保屏幕上的声音源。该基准测试包含3,269个视频和38,390个样本，我们评估了五个代表性模型，以量化对齐和下游任务的特征质量。代码和数据集可在以下网址获取：https://fgt7t6g.github.io/AV-SyncBench。
cs.CV / 83 / 2607.00734
ConRTF: Edge-Constrained Boundary Distribution Refinement for Realtime TransFormer Table Structure Recognition
ConRTF：用于实时变换器表结构识别的边缘约束边界分布细化
Abstract
Table Structure Recognition (TSR) aims to recover the row and column layout of tables from document images, a key step in document understanding pipelines. Accurate TSR depends on precise boundary localization: small errors in row or column boundaries can propagate into incorrect cell assignments and structural inconsistencies. Yet detection-based approaches treat table elements as generic objects, ignoring a fundamental property of table layout: rows and columns play structurally distinct roles and their boundaries carry unequal importance. We propose an Edge-constrained Fine-grained Localization loss (EFL) that formalizes this structural asymmetry by encoding table-specific geometric priors into the training objective: row-like elements are supervised with emphasis on their horizontal boundaries, while column-like elements prioritize vertical boundaries. Implemented within a real-time detector with distribution-based boundary refinement (D-FINE), EFL operates during training only and guides boundary refinement toward structurally meaningful adjustments with no change to the inference pipeline. The proposed approach, ConRTF, is also data-efficient, maintaining robust accuracy with as few as 2k--3k annotated tables. Experiments on PubTables-1M and two private datasets show consistent improvements over the optimized baseline and several real-time detectors including RT-DETRv2 and YOLOv10-11, with gains of up to +1.6 GriTS points at equal inference speed.
Chinese Translation
表结构识别（TSR）旨在从文档图像中恢复表格的行和列布局，这是文档理解流程中的关键步骤。准确的TSR依赖于精确的边界定位：行或列边界的小错误可能会导致错误的单元分配和结构不一致。然而，基于检测的方法将表格元素视为通用对象，忽略了表格布局的一个基本特性：行和列在结构上扮演着不同的角色，其边界的重要性也不相同。我们提出了一种边缘约束细粒度定位损失（EFL），通过将表格特定的几何先验编码到训练目标中，从而形式化这种结构不对称性：行状元素在其水平边界上受到强调，而列状元素则优先考虑垂直边界。EFL在一个具有基于分布的边界细化（D-FINE）的实时检测器中实现，仅在训练期间操作，并引导边界细化朝向结构上有意义的调整，而不改变推理流程。所提出的方法ConRTF在数据效率方面表现良好，能够在仅有2k到3k个标注表格的情况下保持稳健的准确性。在PubTables-1M和两个私有数据集上的实验显示，与优化的基线和包括RT-DETRv2和YOLOv10-11在内的多个实时检测器相比，均有一致的改进，推理速度相同的情况下，GriTS分数提高了最多1.6点。
cs.CV / 84 / 2607.00736
Towards Robust Driving Perception: A Flexible Scale-Driven Family for Self-Supervised Monocular Depth Estimation
朝向鲁棒驾驶感知：一种灵活的基于尺度的自监督单目深度估计模型家族
Abstract
Self-Supervised Monocular Depth Estimation (MDE) has garnered attention in recent years due to its independence from ground truth. However, most existing models are limited to a single scale and exhibit considerable performance degradation in complex driving environments. Networks specifically designed to handle dynamic traffic participants tend to be overly complex, hindering their deployment on resource-constrained automotive edge devices. To address these limitations and move towards robust driving perception, we propose FlexDepth, a scale-driven and flexible family of self-supervised MDE models tailored for challenging road scenarios. FlexDepth employs a two-stage static-dynamic decoupled training strategy, enabling the independent assessment of confidence for both static backgrounds and dynamic road objects. Furthermore, it introduces a meticulously designed Scale-Driven Decoder (SDD) to dynamically select components based on scale size, facilitating efficient feature fusion and the output of high-precision depth maps. Extensive experiments on standard driving benchmarks demonstrate that without any auxiliary information, our model achieves state-of-the-art performance across arbitrary scales with minimal computational overhead. Our smallest model, Flex-Nano, requires only 0.7 GFLOPs and achieves 37.6 FPS on mobile platforms, ensuring reliable real-time perception while maintaining excellent zero-shot generalization.Our source code is avalible: https://github.com/startnew/flexdepth
Chinese Translation
自监督单目深度估计（MDE）近年来受到关注，因为它不依赖于真实值。然而，现有大多数模型仅限于单一尺度，并且在复杂驾驶环境中表现出显著的性能下降。专门设计用于处理动态交通参与者的网络往往过于复杂，阻碍了它们在资源受限的汽车边缘设备上的部署。为了解决这些局限性并朝向鲁棒驾驶感知，我们提出了FlexDepth，一个基于尺度驱动的灵活自监督MDE模型家族，专为具有挑战性的道路场景而设计。FlexDepth采用两阶段静态-动态解耦训练策略，使得静态背景和动态道路物体的置信度能够独立评估。此外，它引入了一种精心设计的尺度驱动解码器（Scale-Driven Decoder, SDD），根据尺度大小动态选择组件，促进高效特征融合和高精度深度图的输出。在标准驾驶基准上的广泛实验表明，在没有任何辅助信息的情况下，我们的模型在任意尺度上以最小的计算开销实现了最先进的性能。我们最小的模型Flex-Nano仅需0.7 GFLOPs，并在移动平台上实现了37.6 FPS，确保了可靠的实时感知，同时保持了优秀的零样本泛化能力。我们的源代码可在以下链接获取：https://github.com/startnew/flexdepth
cs.CV / 85 / 2607.00744
Prototype Memory-Guided Training-Free Anomaly Classification and Localization in Prenatal Ultrasound
原型记忆引导的无训练异常分类与定位在产前超声中的应用
Abstract
Prenatal anomaly classification and localization is of critical importance for fetal health and pregnancy management. Although ultrasound (US) is the primary modality for prenatal screening, accurate diagnosis remains challenging due to the low prevalence and high heterogeneity of anomalies. Existing deep learning methods for prenatal tasks rely on large-scale annotated datasets, which are difficult to obtain in practice. Although few-shot learning alleviates data scarcity, it typically requires fine-tuning for new categories, limiting its practicality in resource-limited clinical settings. To address these challenges, we propose a training-free framework for multi-class prenatal US anomaly classification and localization that operates with only a few reference images per class, representing the first exploration of this setting. Our framework comprises three key components: (1) a memory bank with multi-granular prototypes that explicitly models both class-level semantics and anomaly characteristics; (2) a prototype-driven soft merging mechanism that aggregates discriminative features to detect the anomaly region; and (3) a class-aware refinement strategy that leverages prototype consistency to improve category prediction. Extensively validated on a multi-center prenatal US dataset containing 1,149 cases, with a total of 2,357 images and 9 categories, our proposed method outperforms the competitors.
Chinese Translation
产前异常的分类与定位对胎儿健康和妊娠管理至关重要。尽管超声（US）是产前筛查的主要手段，但由于异常的低发生率和高异质性，准确诊断仍然具有挑战性。现有的产前任务深度学习方法依赖于大规模标注数据集，而这些数据集在实践中难以获得。尽管少样本学习在一定程度上缓解了数据稀缺问题，但通常需要对新类别进行微调，这限制了其在资源有限的临床环境中的实用性。为了解决这些挑战，我们提出了一种无训练框架，用于多类别产前超声异常的分类与定位，该框架仅依赖于每个类别的少量参考图像，代表了这一设置的首次探索。我们的框架包括三个关键组件：（1）一个具有多粒度原型的记忆库，明确建模类别级语义和异常特征；（2）一个基于原型的软合并机制，用于聚合判别特征以检测异常区域；（3）一个类别感知的精细化策略，利用原型一致性来改善类别预测。在包含1,149个案例、总计2,357张图像和9个类别的多中心产前超声数据集上进行了广泛验证，我们提出的方法优于竞争对手。
cs.CV / 86 / 2607.00745
Foundation Model-driven Key Anatomy Frame Selection for Blind-sweep Ultrasound Fetal Birth Weight Estimation
基于基础模型的关键解剖帧选择用于盲扫超声胎儿出生体重估计
Abstract
Accurate fetal birth weight (FBW) estimation shortly before delivery is clinically valuable yet challenging due to its reliance on operator expertise, particularly in low-resource settings. To reduce this reliance, we study near-term birth-weight regression from blind-sweep ultrasound (US) videos acquired within 48 hours prior to delivery, with post-delivery weighing as ground truth. Accordingly, we propose a foundation model-driven key anatomy frame selection framework that enables accurate FBW regression despite the absence of plane constraints in blind sweeps. Our highlights are as follows: (1) We believe this is the first work to estimate FBW using blind-sweep US videos, enabling operator-independent assessment. (2) An Anatomy-Guided Frame Selection module equipped with a vision-language foundation model is proposed for keyframe collection in unconstrained sweeps. (3) A Redundancy-Aware Feature Compression module is designed to compress frame features while preserving task-relevant information, alleviating temporal redundancy. Extensively validated on prospectively collected data from 839 patients, our method achieves an MAE of 161.3 g, with 90.23% and 100% of cases falling within 10% and 15% absolute percentage error, outperforming typical Hadlock estimation and strong competitors. Codes are available at https://github.com/ouleoule/BlindSweep-EBW.
Chinese Translation
在临床上，准确估计胎儿出生体重（FBW）在分娩前是非常有价值的，但由于依赖操作者的专业知识，尤其是在资源匮乏的环境中，这一任务具有挑战性。为了减少这种依赖，我们研究了在分娩前48小时内获取的盲扫超声（US）视频中的近临产出生体重回归，以产后称重作为真实值。因此，我们提出了一种基于基础模型的关键解剖帧选择框架，使得尽管在盲扫中缺乏平面约束，仍能实现准确的FBW回归。我们的亮点如下：（1）我们认为这是首次使用盲扫US视频估计FBW，从而实现操作者独立评估。（2）提出了一种配备视觉-语言基础模型的解剖引导帧选择模块，用于在无约束扫查中收集关键帧。（3）设计了一种冗余感知特征压缩模块，以压缩帧特征，同时保留与任务相关的信息，从而减轻时间冗余。在对839名患者前瞻性收集的数据进行广泛验证后，我们的方法实现了161.3克的平均绝对误差（MAE），其中90.23%和100%的案例在10%和15%的绝对百分比误差范围内，优于典型的Hadlock估计和强有力的竞争者。代码可在https://github.com/ouleoule/BlindSweep-EBW获取。
cs.CV / 87 / 2607.00746
GaussianFusion: Unified 3D Gaussian Representation for Multi-Modal Fusion Perception
GaussianFusion：用于多模态融合感知的统一三维高斯表示
Abstract
The bird's-eye view (BEV) representation enables multi-sensor features to be fused within a unified space, serving as the primary approach for achieving comprehensive 3D perception. However, the discrete grid representation of BEV leads to significant detail loss and limits feature alignment and cross-modal information interaction in multimodal fusion perception. In this work, we break from the conventional BEV paradigm and propose a new universal framework for multi-modal fusion based on 3D Gaussian representation. This approach naturally unifies multi-modal features within a shared and continuous 3D Gaussian space, effectively preserving edge and fine texture details. To achieve this, we design a novel forward-projection-based multi-modal Gaussian initialization module and a shared cross-modal Gaussian encoder that iteratively updates Gaussian properties based on an attention mechanism. GaussianFusion is inherently a task-agnostic model, with its unified Gaussian representation naturally supporting various 3D perception tasks. Extensive experiments demonstrate the generality and robustness of GaussianFusion. On the nuScenes dataset, it outperforms the 3D object detection baseline BEVFusion by 2.6 NDS. Its variant surpasses GaussFormer on 3D semantic occupancy with 1.55 mIoU improvement while using only 30% of the Gaussians and achieving a 450% speedup.
Chinese Translation
鸟瞰图（BEV）表示使得多传感器特征能够在统一空间内融合，成为实现全面三维感知的主要方法。然而，BEV的离散网格表示导致了显著的细节损失，并限制了多模态融合感知中的特征对齐和跨模态信息交互。在本研究中，我们突破了传统的BEV范式，提出了一种基于三维高斯表示的多模态融合新通用框架。这种方法自然地将多模态特征统一在一个共享且连续的三维高斯空间中，有效地保留了边缘和细腻的纹理细节。为此，我们设计了一种新颖的基于前向投影的多模态高斯初始化模块和一个共享的跨模态高斯编码器，该编码器基于注意力机制迭代更新高斯属性。GaussianFusion本质上是一个任务无关的模型，其统一的高斯表示自然支持各种三维感知任务。大量实验表明了GaussianFusion的通用性和鲁棒性。在nuScenes数据集上，它的性能超越了三维目标检测基线BEVFusion，提升了2.6 NDS。其变体在三维语义占用任务上超越了GaussFormer，提升了1.55 mIoU，同时仅使用了30%的高斯，并实现了450%的速度提升。
cs.CV / 88 / 2607.00747
Active Learning for Cascaded Object Detection: Balancing Coverage and Uncertainty in Table Extraction Pipelines
级联目标检测中的主动学习：在表格提取管道中平衡覆盖率和不确定性
Abstract
Table extraction from business documents relies on a cascaded pipeline where Table Detection (TD) first localizes tables and Table Structure Recognition (TSR) then recovers their internal layout. Building task-specific training sets for this pipeline is costly, particularly for TSR which requires fine-grained structural annotations. Active learning (AL) can reduce this annotation burden, yet most AL strategies are designed for single-model tasks and do not account for inter-stage dependencies in cascaded architectures. In this work, we present the first adaptation of Uncertainty Herding (UHerding), a hybrid coverage-uncertainty sampling method originally proposed for image classification, to cascaded object detection pipelines. We propose two pipeline-aware extensions that exploit the TD-to-TSR dependency: RankFusion adds dual-manifold coverage over both detection and structure representation spaces, while CAPA further incorporates stage-dependent gating and per-task uncertainty calibration. Extensive experiments across two public (PubTables-1M and FinTabNet) and two private table extraction datasets, with various annotation budgets (from 71 to 500 documents) show that UHerding generalizes well to table extraction, outperforming each baseline. Among pipeline-aware variants, RankFusion achieves higher expected gains but at the cost of greater variance, while CAPA emerges as the most consistent strategy, outperforming standard UHerding on three out of four datasets.
Chinese Translation
从商业文档中提取表格依赖于一个级联管道，其中表格检测（Table Detection, TD）首先定位表格，随后表格结构识别（Table Structure Recognition, TSR）恢复其内部布局。为该管道构建特定任务的训练集成本高昂，尤其是对于需要细粒度结构注释的TSR。主动学习（Active Learning, AL）可以减少这种注释负担，但大多数AL策略是为单模型任务设计的，并未考虑级联架构中的阶段间依赖关系。在本研究中，我们首次将不确定性聚集（Uncertainty Herding, UHerding）这一最初为图像分类提出的混合覆盖-不确定性采样方法，应用于级联目标检测管道。我们提出了两个考虑管道依赖性的扩展：RankFusion在检测和结构表示空间上增加了双流形覆盖，而CAPA进一步结合了阶段依赖的门控和每个任务的不确定性校准。在两个公共数据集（PubTables-1M和FinTabNet）和两个私有表格提取数据集上进行了广泛实验，涵盖了不同的注释预算（从71到500个文档），结果表明UHerding在表格提取中具有良好的泛化能力，超越了每个基线。在考虑管道的变体中，RankFusion实现了更高的预期收益，但代价是更大的方差，而CAPA则成为最一致的策略，在四个数据集中有三个超越了标准的UHerding。
cs.CV / 89 / 2607.00748
FrameONE: Hierarchical Motion Modeling for Universal Multi-View Echocardiographic Keyframe Detection
FrameONE：用于通用多视角超声心动图关键帧检测的分层运动建模
Abstract
Accurate detection of end-systole (ES) and end-diastole (ED) frames is fundamental to echocardiographic assessment. Existing methods are typically developed in a view-specific manner, depend on auxiliary annotations or intensive visual modeling, which limits their generalizability. In multi-view modeling, keyframe detection is driven by shared cardiac motion, yet large appearance differences and motion patterns make unified modeling challenging. To address these issues, we propose FrameONE, a unified end-to-end framework for multi-view echocardiographic keyframe detection. FrameONE introduces a Hierarchical Motion Modeling strategy: an intra-view multi-task learning reduces appearance bias and promotes motion-focused representations within each view; an inter-view general motion learning module further separates view-agnostic dynamics from view-specific patterns, enabling shared yet flexible motion representation learning across views. Extensive experiments on 25,872 videos spanning four standard views demonstrate that FrameONE achieves state-of-the-art keyframe detection accuracy with strong cross-view generalization. Code is available at https://github.com/szuboy/FrameONE.
Chinese Translation
准确检测心脏收缩末期（ES）和心脏舒张末期（ED）帧是超声心动图评估的基础。现有方法通常以视角特定的方式开发，依赖于辅助注释或密集的视觉建模，这限制了它们的普适性。在多视角建模中，关键帧检测由共享的心脏运动驱动，但外观差异和运动模式的巨大差异使得统一建模变得具有挑战性。为了解决这些问题，我们提出了FrameONE，这是一个用于多视角超声心动图关键帧检测的统一端到端框架。FrameONE引入了一种分层运动建模策略：视内多任务学习减少了外观偏差，并促进了每个视角内以运动为中心的表示；视间通用运动学习模块进一步将视角无关的动态与视角特定模式分离，从而实现跨视角共享而灵活的运动表示学习。在涵盖四个标准视角的25,872个视频上的大量实验表明，FrameONE在关键帧检测准确性方面达到了最先进的水平，并具有强大的跨视角泛化能力。代码可在 https://github.com/szuboy/FrameONE 获取。
cs.CV / 90 / 2607.00752
GKDT: General Keypoint Detection Transformer
GKDT：通用关键点检测变换器
Abstract
With the emergence of various pre-trained vision and language models, computer vision is shifting from narrow-domain to open-domain recognition. The construction of a more powerful yet general keypoint detection (GKD) model to support diverse tasks has become increasingly important in the field. To this end, we firstly present a large-scale unified keypoint dataset called MegaKPT. The dataset is composed of over 1.3 million diverse object instances from twenty-nine existing datasets, and enjoys high-quality unified annotations with keypoint text descriptions. Based on MegaKPT, we develop GKDT, a simple, flexible and powerful DINOv3 based Transformer model for General Keypoint Detection. Our GKDT supports visual prompts, text prompts, or both. To enhance model training, we also propose a suite of useful strategies such as mix-modal prompted training and dynamic importance sampling. By testing over 22 test sets with seen or unseen objects, our single GKDT model shows strong performance and generality in detecting keypoints on broad categories, with most categories over 90\%
[email protected] accuracy, offering high practical applicability to real-world problems. The dataset, models, and codes will be released at https://github.com/AlanLuSun/General-Keypoint-Detection.
Chinese Translation
随着各种预训练视觉和语言模型的出现，计算机视觉正在从狭域识别向开放域识别转变。构建一个更强大且通用的关键点检测（GKD）模型以支持多样化任务在该领域变得愈发重要。为此，我们首先提出了一个大规模统一的关键点数据集，称为MegaKPT。该数据集由来自二十九个现有数据集的超过130万个多样化的物体实例组成，并拥有高质量的统一注释和关键点文本描述。在MegaKPT的基础上，我们开发了GKDT，这是一种简单、灵活且强大的基于DINOv3的变换器模型，用于通用关键点检测。我们的GKDT支持视觉提示、文本提示或两者兼具。为了增强模型训练，我们还提出了一系列有用的策略，如混合模态提示训练和动态重要性采样。通过在22个测试集上测试已见或未见的物体，我们的单一GKDT模型在广泛类别的关键点检测中表现出强大的性能和通用性，大多数类别的
[email protected]准确率超过90\%，为解决现实世界问题提供了高度的实用性。数据集、模型和代码将发布在https://github.com/AlanLuSun/General-Keypoint-Detection。
cs.CV / 91 / 2607.00766
Decoupled Guidance: Disentangling Subject and Context Pathways in Text-to-Image Personalization
解耦引导：在文本到图像个性化中解开主体与上下文路径
Abstract
Text-to-image personalization aims to generate a user-provided subject in novel scenes described by text. However, most existing methods encode subject identity (fidelity) and context (editability) through the same conditioning pathway, forcing the two to compete for attention-map resources. We refer to this phenomenon as conditioning entanglement and show that it induces a fidelity-editability trade-off. We further provide causal evidence by replacing the target subject token with a generic subject token, which produces shifts in attention allocation and corresponding changes in context adherence. To this end, we propose Decoupled Guidance (DeGu), a plug-and-play framework that routes subject identity and scene context through two independent guidance streams. We further introduce a spatial mixing mechanism that dynamically fuses these streams, ensuring each operates within its semantically relevant region without interference. Furthermore, DeGu can be readily applied to existing personalization methods without modifying the underlying backbone models, consistently improving the overall personalization performance while enabling inference-time control over the fidelity-editability balance, across diverse methods and backbones, including flow-matching Diffusion Transformers (DiTs).
Chinese Translation
文本到图像个性化旨在根据用户提供的主体生成新的场景，这些场景由文本描述。然而，大多数现有方法通过相同的条件路径编码主体身份（保真度）和上下文（可编辑性），迫使两者在注意力图资源上相互竞争。我们将这种现象称为条件纠缠，并展示它引发了保真度与可编辑性之间的权衡。我们进一步提供了因果证据，通过用通用主体标记替换目标主体标记，产生了注意力分配的变化以及相应的上下文遵循变化。为此，我们提出了解耦引导（Decoupled Guidance, DeGu），这是一种即插即用的框架，通过两个独立的引导流来引导主体身份和场景上下文。我们进一步引入了一种空间混合机制，动态融合这些流，确保每个流在其语义相关区域内独立操作而不发生干扰。此外，DeGu可以在不修改基础骨干模型的情况下，轻松应用于现有的个性化方法，始终提高整体个性化性能，同时在不同方法和骨干模型中实现对保真度与可编辑性平衡的推理时控制，包括流匹配扩散变换器（Diffusion Transformers, DiTs）。
cs.CV / 92 / 2607.00774
Soft Mixture-of-Recursions: Going Deeper with Recursive Vision Transformers
软混合递归：深入递归视觉变换器
Abstract
Recent recursive Transformer studies have primarily reused shared parameters across computation steps to construct compact, parameter-efficient models. In this work, we leverage recursion to build effectively deeper Transformers with stronger representational capacity. However, in Vision Transformers, simply increasing recursion depth does not reliably improve performance, as existing recursive approaches do not fully utilize the intermediate representations produced throughout recursive computation. We propose Soft Mixture-of-Recursions (SoftMoR) and its Vision Transformer instantiation, Soft Recursive Vision Transformer (SR-ViT). SoftMoR learns token-wise mixture weights to softly combine outputs from all recursion steps, allowing intermediate representations to be utilized in a learnable and flexible way. Across diverse vision tasks, SR-ViT consistently improves as recursion depth increases with minimal parameter overhead. On ImageNet-1K, increasing recursion depth from 1 to 4 improves SR-ViT-S top-1 accuracy from 79.83% to 82.48% with only 1.7M additional parameters, outperforming the substantially larger DeiT-B while using approximately 27% of its parameters. These results demonstrate that SoftMoR provides a parameter-efficient path to deeper and stronger Vision Transformers through recursion.
Chinese Translation
近期的递归变换器研究主要通过在计算步骤中重用共享参数来构建紧凑且参数高效的模型。在本研究中，我们利用递归构建有效更深的变换器，以增强其表示能力。然而，在视觉变换器中，仅仅增加递归深度并不能可靠地提高性能，因为现有的递归方法并未充分利用在递归计算过程中产生的中间表示。我们提出了软混合递归（Soft Mixture-of-Recursions, SoftMoR）及其在视觉变换器中的实例化——软递归视觉变换器（Soft Recursive Vision Transformer, SR-ViT）。SoftMoR 学习逐个标记的混合权重，以柔和地结合所有递归步骤的输出，从而以可学习和灵活的方式利用中间表示。在多种视觉任务中，SR-ViT 随着递归深度的增加而持续改善，且仅增加了最小的参数开销。在 ImageNet-1K 数据集上，将递归深度从 1 增加到 4，使 SR-ViT-S 的 top-1 准确率从 79.83% 提升至 82.48%，仅增加了 1.7M 的参数，超越了参数量大得多的 DeiT-B，同时仅使用了其约 27% 的参数。这些结果表明，SoftMoR 提供了一条通过递归实现更深且更强大的视觉变换器的参数高效路径。
cs.CV / 93 / 2607.00780
SpiralFovea: Input-Adaptive Foveated Tokenization as a Third Lever of Resource-Adaptive Inference
SpiralFovea：作为资源自适应推理第三杠杆的输入自适应注视点标记
Abstract
Most adaptive-inference techniques for foundation models change what the model does - early exit, MoE routing, KV-cache compression, dynamic attention sparsity. The input that hits the backbone, however, remains a fixed-grid tokenisation indifferent to image content. We argue that this is a missed lever. We present SpiralFovea, a parameter-free, input-adaptive tokeniser in which token identity, location, scale, and count are all functions of local visual entropy and selection completes before any backbone parameter is queried. Around content-driven hotspot anchors, multi-scale spiral rings produce <= 78 patches that replace the standard 196-patch ViT grid at the input stage. Across four canonical fine-grained benchmarks, SpiralFovea yields +1.7-2.1 pp accuracy with a 60% reduction in input tokens, an 84% reduction in self-attention FLOPs at every transformer layer, and 18-29% throughput gains over the matched static tokenisation baseline. A controlled ablation on CUB-200-2011 Genus across four backbones reveals a clean diagnostic: the gain magnitude tracks inversely with the strength of the backbone's whole-image positional prior, isolating self-supervised foundation models as the regime where input-adaptive tokenisation is most valuable.
Chinese Translation
大多数基础模型的自适应推理技术改变了模型的功能——早期退出、MoE 路由、KV 缓存压缩、动态注意力稀疏。然而，进入主干网络的输入仍然是对图像内容无差别的固定网格标记。我们认为这是一个被忽视的杠杆。我们提出了 SpiralFovea，这是一种无参数、输入自适应的标记器，其中标记的身份、位置、尺度和数量都是局部视觉熵的函数，选择在查询任何主干参数之前完成。在内容驱动的热点锚点周围，多尺度螺旋环产生不超过 78 个补丁，取代输入阶段标准的 196 补丁 ViT 网格。在四个经典的细粒度基准测试中，SpiralFovea 实现了 +1.7-2.1 个百分点的准确率，同时输入标记减少了 60%，每个变换器层的自注意力 FLOPs 减少了 84%，并且在与静态标记基线匹配的情况下，吞吐量提升了 18-29%。在 CUB-200-2011 Genus 上对四个主干进行的控制消融实验揭示了一个清晰的诊断：增益幅度与主干的全图位置先验的强度呈反比，孤立出自监督基础模型作为输入自适应标记最有价值的领域。
cs.CV / 94 / 2607.00784
LeVLJEPA: End-to-End Vision-Language Pretraining Without Negatives
LeVLJEPA：无负样本的端到端视觉-语言预训练
Abstract
Vision-language pretraining remains dominated by contrastive objectives, whereas vision-only self-supervised learning has largely adopted non-contrastive methods. At the same time, the role of vision-language encoders has shifted: they are increasingly deployed not as zero-shot classifiers but as the frozen visual backbone of vision-language models and dense prediction systems, which consume the full grid of patch tokens rather than a single pooled embedding. We introduce LeVLJEPA, the first fully non-contrastive end-to-end vision-language pretraining method. LeVLJEPA learns through cross-modal prediction with stop-gradient targets and per-modality distributional regularization, without negatives, temperature, momentum encoder, or teacher-student schedule, and trains stably at large scale. We find that the resulting encoder provides markedly stronger dense semantic features for downstream use: as a frozen vision-language-model backbone, LeVLJEPA is the strongest of the evaluated encoders across GQA, VQAv2, and POPE under two distinct language models, and outperforms contrastive baselines on semantic segmentation, while remaining on par on global readouts such as linear probing. These results establish non-contrastive pretraining as an effective means of producing dense semantic vision features.
Chinese Translation
视觉-语言预训练仍然以对比目标为主导，而仅视觉的自监督学习则在很大程度上采用了非对比方法。同时，视觉-语言编码器的角色发生了变化：它们越来越多地被用作视觉-语言模型和密集预测系统的固定视觉骨干，而不是零-shot 分类器，这些系统使用完整的补丁令牌网格而不是单一的池化嵌入。我们提出了LeVLJEPA，这是首个完全非对比的端到端视觉-语言预训练方法。LeVLJEPA通过使用停止梯度目标和每种模态的分布正则化进行跨模态预测进行学习，无需负样本、温度、动量编码器或师生调度，并且在大规模训练中保持稳定。我们发现，得到的编码器为下游任务提供了显著更强的密集语义特征：作为固定的视觉-语言模型骨干，LeVLJEPA在GQA、VQAv2和POPE的两个不同语言模型下是评估的编码器中最强的，并且在语义分割上优于对比基线，同时在全局读取（如线性探测）上保持相当。这些结果确立了非对比预训练作为生成密集语义视觉特征的有效手段。
cs.CV / 95 / 2607.00798
ClinRAG-GRAPH: Clinical-prior Retrieval-Augmented Graph Model with Domain Adversarial Learning for Breast pCR Prediction
ClinRAG-GRAPH：基于临床先验的检索增强图模型与领域对抗学习用于乳腺癌pCR预测
Abstract
Neoadjuvant chemotherapy (NAC) response prediction is clinically important for treatment stratification in breast cancer. However, robust pre-treatment pathological complete response (pCR) prediction remains challenging due to insufficient cross-modal modeling, multicenter imaging heterogeneity, and weak evidence-grounded interpretability. We propose ClinRAG-GRAPH, a Clinically informed Retrieval-Augmented Generation Graph framework, for pre-treatment pCR prediction from DCE-MRI, structured clinical variables, and biopsy-derived pathological biomarkers. ClinRAG-GRAPH constructs an intra-patient clinical-prior graph and applies a prior-guided relation-aware graph convolutional network for structured multimodal representation learning. To improve cross-center robustness, we introduce a dual-branch domain-adversarial learning strategy to suppress protocol-related MRI bias while preserving pCR-relevant features. To enhance interpretability, we further incorporate large language model (LLM)-driven subgraph RAG module that retrieves clinically analogous historical cases and integrates retrieved evidence for pCR inference. We assemble a large-scale multicenter NAC breast cancer cohort for extensive validation, drawing from two public sources and three in-house centers.Results show that ClinRAG-GRAPH achieves AUCs of 0.815 on the internal test set and 0.774/0.712 on two external test sets, demonstrating robust pre-treatment pCR prediction across centers. The code is available at the anonymized https://github.com/miccai26-1181/ClinRAG-GRAPH.
Chinese Translation
新辅助化疗（NAC）反应预测在乳腺癌的治疗分层中具有重要的临床意义。然而，由于跨模态建模不足、多中心影像异质性以及证据基础解释性较弱，可靠的治疗前病理完全反应（pCR）预测仍然具有挑战性。我们提出了ClinRAG-GRAPH，这是一种基于临床信息的检索增强生成图框架，用于从动态对比增强磁共振成像（DCE-MRI）、结构化临床变量和活检衍生的病理生物标志物中进行治疗前pCR预测。ClinRAG-GRAPH构建了一个患者内的临床先验图，并应用先验引导的关系感知图卷积网络进行结构化多模态表示学习。为了提高跨中心的鲁棒性，我们引入了一种双分支领域对抗学习策略，以抑制与协议相关的MRI偏差，同时保留与pCR相关的特征。为了增强可解释性，我们进一步结合了大型语言模型（LLM）驱动的子图检索增强生成模块，该模块检索临床相似的历史案例并整合检索到的证据以进行pCR推断。我们组建了一个大规模多中心NAC乳腺癌队列进行广泛验证，数据来源于两个公共数据源和三个内部中心。结果表明，ClinRAG-GRAPH在内部测试集上实现了0.815的AUC，在两个外部测试集上分别为0.774和0.712，展示了跨中心的可靠治疗前pCR预测。代码可在匿名链接 https://github.com/miccai26-1181/ClinRAG-GRAPH 获取。
cs.CV / 96 / 2607.00804
Spotted: Location-informed Reidentification of Hyenas and Leopards in Camera Trap Surveys
Spotted：基于位置的信息的鬣狗和豹子在相机陷阱调查中的再识别
Abstract
Animal re-identification (ReID) in camera-trap surveys remains challenging due to low image quality, strong variation in illumination and viewpoint, and highly imbalanced numbers of observations per individual. As a result, current ReID performance is often insufficient for fully automated use, and practical workflows typically depend on expert review of algorithmically proposed candidate matches. Moreover, most existing approaches focus almost exclusively on visual cues and overlook auxiliary information routinely available in field studies, such as image timestamps and camera-trap locations. We introduce Spotted, a location-informed, human-in-the-loop animal ReID framework that integrates visual similarity with spatio-temporal feasibility priors derived from camera locations, thereby reducing the amount of required expert review. Our method (i) computes an image-model-agnostic feasibility score based on the minimum travel speed required for two detections to correspond to the same individual, (ii) uses these feasibility cues as pseudo-supervision to train a lightweight head on top of a frozen visual foundation model, and (iii) fuses adapted visual similarity with spatio-temporal feasibility to obtain a robust pairwise matching score. We additionally integrate an active pair sampling strategy to accelerate annotation by initially prioritizing uncertain predictions. We evaluate Spotted on three challenging camera-trap ReID datasets comprised of spotted hyenas and leopards, which we release as part of this work. Our model improves average top-5 identification accuracy by 9pp, 2pp and 9pp over the best baseline on our LeopardID102, SpottedHyenaID109 and SpottedHyenaID415 datasets, respectively. Further, we show that our human-in-the-loop strategy reduces the number of queried comparisons by up to 69pp while achieving equivalent positive matches.
Chinese Translation
在相机陷阱调查中，动物再识别（ReID）仍然面临挑战，主要由于图像质量低、光照和视角变化大，以及每个个体观察数量严重不平衡。因此，目前的ReID性能通常不足以支持完全自动化的使用，实际工作流程通常依赖于专家对算法提出的候选匹配的审查。此外，大多数现有方法几乎完全专注于视觉线索，而忽视了在野外研究中常规可获得的辅助信息，例如图像时间戳和相机陷阱位置。我们提出了Spotted，一个基于位置的信息的人机协作动物ReID框架，该框架将视觉相似性与从相机位置推导的时空可行性先验相结合，从而减少所需的专家审查量。我们的方法（i）计算基于两个检测对应于同一个体所需的最小旅行速度的图像模型无关的可行性得分，（ii）使用这些可行性线索作为伪监督，在冻结的视觉基础模型上训练一个轻量级的头部，以及（iii）融合调整后的视觉相似性与时空可行性，以获得稳健的成对匹配得分。我们还整合了一种主动对比样本采样策略，通过优先考虑不确定的预测来加速标注。我们在三个具有挑战性的相机陷阱ReID数据集上评估Spotted，这些数据集包含斑点鬣狗和豹子，我们将其作为本研究的一部分发布。我们的模型在LeopardID102、SpottedHyenaID109和SpottedHyenaID415数据集上，分别比最佳基线提高了9个百分点、2个百分点和9个百分点的平均前5名识别准确率。此外，我们还展示了我们的人机协作策略在实现等效正匹配的同时，将查询比较的数量减少了多达69个百分点。
cs.CV / 97 / 2607.00816
Towards High-Resolution Visual Perception via Hierarchical Entity Exploration
通过层次实体探索实现高分辨率视觉感知
Abstract
High-resolution (HR) image perception remains a key challenge in multimodal large language models (MLLMs), as fine-grained details are often lost when the image is processed as a whole. Existing methods either require training to teach models where to look or heuristically divide the image into fixed regions, both of which struggle to generalize in complex HR scenes. In this work, we propose Hierarchical Entity Exploration (HEE), a training-free and model-agnostic framework that transforms static image understanding into dynamic, query-guided entity exploration. HEE first evaluates each region using a dual scoring mechanism to determine whether it already contains sufficient evidence to answer the question. If not, it applies object detection within the most promising region to extract fine-grained entities, clusters them into coherent subregions, and organizes them into a multi-level semantic hierarchy for deeper exploration. When deeper regions still fail to yield confident answers, a confidence-guided backtracking mechanism revisits alternative paths to ensure adaptive perception. Extensive results show that HEE outperforms training-free methods like ZoomEye and RAP in both accuracy and efficiency on two complex HR benchmarks (Visual Probe and HR-Bench), across different MLLMs such as Qwen2.5-VL and LLaVA-OneVision. Moreover, HEE demonstrates generalization on the MME-RealWorld benchmark.
Chinese Translation
高分辨率（HR）图像感知仍然是多模态大语言模型（MLLMs）中的一项关键挑战，因为在整体处理图像时，细粒度细节往往会丢失。现有方法要么需要训练以教导模型在哪里寻找，要么启发式地将图像划分为固定区域，这两者在复杂的HR场景中都难以泛化。在本研究中，我们提出了层次实体探索（HEE），这是一种无训练且与模型无关的框架，将静态图像理解转变为动态的、基于查询的实体探索。HEE首先使用双重评分机制评估每个区域，以确定其是否已经包含足够的证据来回答问题。如果没有，它将在最有前景的区域内应用目标检测，以提取细粒度实体，将它们聚类成一致的子区域，并将其组织成多层次的语义层次结构以便进行更深入的探索。当更深的区域仍然无法提供可靠的答案时，基于置信度的回溯机制会重新审视替代路径，以确保自适应感知。大量结果表明，HEE在两个复杂的HR基准（Visual Probe和HR-Bench）上，在不同的MLLM（如Qwen2.5-VL和LLaVA-OneVision）中，在准确性和效率上均优于无训练方法，如ZoomEye和RAP。此外，HEE在MME-RealWorld基准上展示了良好的泛化能力。
cs.CV / 98 / 2607.00817
Training-Free Debiasing of Diffusion Models via CLIP-Guided Denoising Optimization
基于CLIP引导去噪优化的无训练去偏差扩散模型
Abstract
Text-to-image diffusion models achieve impressive visual quality, yet demographic bias remains a challenge, as neutral prompts consistently produce stereotypical representations across gender and race. Existing approaches remain limited by costly retraining or by inference-time interventions that often degrade image quality and semantic alignment. We propose Text Embedding Steering (TES), a training-free framework that mitigates demographic bias by directly optimizing conditional text embeddings during the diffusion process. We show that a two-stage strategy - early-stage global alignment followed by iterative denoising-time refinement with CLIP-based feedback - enables stable and controllable attribute steering without modifying model parameters. Extensive experiments on Stable Diffusion demonstrate that TES outperforms existing training-free baselines in fairness while maintaining competitive image quality. These results highlight that inference-time text embedding optimization is a practical and scalable solution for fairness-aware generation in diffusion models.
Chinese Translation
文本到图像的扩散模型在视觉质量上取得了令人印象深刻的成果，但人口统计偏见仍然是一个挑战，因为中性提示在性别和种族方面始终产生刻板印象的表现。现有的方法受到昂贵的再训练或推理时干预的限制，这些干预往往会降低图像质量和语义一致性。我们提出了文本嵌入引导（Text Embedding Steering, TES），这是一个无训练框架，通过在扩散过程中直接优化条件文本嵌入来减轻人口统计偏见。我们展示了一种两阶段策略——早期阶段的全局对齐，随后是基于CLIP反馈的迭代去噪时间精细化——使得在不修改模型参数的情况下实现稳定和可控的属性引导。在Stable Diffusion上的大量实验表明，TES在公平性方面优于现有的无训练基线，同时保持竞争力的图像质量。这些结果强调了推理时文本嵌入优化是扩散模型中实现公平生成的一个实用且可扩展的解决方案。
cs.CV / 99 / 2607.00829
Stitched Embeddings: A Unified Latent Space for 3D Garments and 2D Patterns
缝合嵌入：3D服装与2D图案的统一潜在空间
Abstract
While garments are essential for realistic digital humans, their topological variety makes them much harder to model than parametric bodies. Traditional tailoring relies on 2D sewing patterns, yet bridging these patterns to 3D geometry currently requires physical simulations. We present Stitched Embeddings, the first simulation-free framework to unify 3D garment reconstruction and sewing pattern inference within a single bidirectional latent space. By leveraging the geometric priors of a pretrained 3D foundation model, our approach overcomes the data scarcity typically associated with high-quality garment modeling. We propose to use the BoxMesh as a critical intermediate representation to align 2D panels into 3D configurations without the computational overhead of a simulator. This architecture achieves state-of-the-art accuracy in pattern reconstruction while significantly improving efficiency. Furthermore, our differentiable pipeline enables novel applications, including pattern recovery from meshes and 3D editing from 2D patterns. Finally, this work provides a scalable link between neural 3D vision and the physical garment manufacturing pipeline. Project Page: https://andreus00.github.io/stitchedembeddings
Chinese Translation
尽管服装对于逼真的数字人类至关重要，但其拓扑多样性使得建模比参数化身体更为困难。传统裁剪依赖于2D缝制图案，但将这些图案与3D几何形状连接起来目前需要物理仿真。我们提出了缝合嵌入（Stitched Embeddings），这是第一个无需仿真的框架，将3D服装重建与缝制图案推断统一在一个双向潜在空间中。通过利用预训练3D基础模型的几何先验，我们的方法克服了高质量服装建模中通常存在的数据稀缺问题。我们建议使用BoxMesh作为关键的中间表示，将2D面板对齐到3D配置中，而无需仿真器的计算开销。该架构在图案重建中实现了最先进的准确性，同时显著提高了效率。此外，我们的可微管道支持新颖的应用，包括从网格中恢复图案和从2D图案进行3D编辑。最后，这项工作为神经3D视觉与物理服装制造流程之间提供了可扩展的连接。项目页面：https://andreus00.github.io/stitchedembeddings
cs.CV / 100 / 2607.00832
Pano2World: End-to-End 3D Generation via Unified Multi-View Sequences
Pano2World：通过统一多视角序列实现端到端的3D生成
Abstract
A single panorama captures the full visual sphere from one camera center, yet confines users to looking around in place without enabling true scene exploration. Converting a single panorama into a persistent, renderable 3D representation for free-viewpoint navigation has attracted growing interest; existing methods either adopt iterative per-view completion that propagates inpainting results to update the underlying geometry, leading to progressive error accumulation and cumbersome multi-step pipelines, or leverage the temporal consistency priors of video generation models, yet the continuous-trajectory constraint intrinsic to such models limits their flexibility in covering scenes from multiple directions simultaneously. We present Pano2World, which takes a single indoor panorama as input and directly outputs a persistent, explorable 3D Gaussian scene. Given the source panorama, Pano2World first reconstructs a coarse 3D Gaussian proxy and renders it at adaptively sampled nearby poses to obtain geometrically aligned guidance panoramas; a panoramic diffusion model then jointly denoises all target views via View-Aware Attention Routing, where each target view simultaneously receives geometric constraints from its corresponding guidance panorama and global semantic guidance from the source panorama, naturally enforcing cross-view consistency. To avoid the information loss incurred by decoding the multi-view hidden features formed during joint denoising back to the pixel domain via VAE, we introduce Latent Feature Adapter, a geometry-aware bridge module that directly distills these hidden features into a scene latent, subsequently decoded into the final 3D Gaussian scene. Experiments demonstrate that Pano2World significantly outperforms existing methods on the multi-position panoramic novel-view synthesis benchmark.
Chinese Translation
单个全景图捕捉了从一个相机中心的完整视觉球体，但限制用户只能在原地环顾，而无法进行真正的场景探索。将单个全景图转换为可持久渲染的3D表示，以实现自由视点导航，已引起越来越多的关注；现有方法要么采用迭代的逐视图补全，通过传播修补结果来更新基础几何，导致逐步的误差累积和繁琐的多步骤流程，要么利用视频生成模型的时间一致性先验，但此类模型固有的连续轨迹约束限制了其在同时覆盖多个方向场景时的灵活性。我们提出了Pano2World，该方法以单个室内全景图作为输入，直接输出一个持久的、可探索的3D高斯场景。给定源全景图，Pano2World首先重建一个粗略的3D高斯代理，并在自适应采样的附近姿态下渲染，以获得几何对齐的引导全景图；然后，一个全景扩散模型通过视图感知注意力路由联合去噪所有目标视图，其中每个目标视图同时接收来自其对应引导全景图的几何约束和来自源全景图的全局语义指导，自然地强制执行跨视图一致性。为了避免通过变分自编码器（VAE）将联合去噪过程中形成的多视图隐藏特征解码回像素域而导致的信息损失，我们引入了潜在特征适配器（Latent Feature Adapter），这是一个几何感知的桥接模块，直接将这些隐藏特征提炼为场景潜在特征，随后解码为最终的3D高斯场景。实验表明，Pano2World在多位置全景新视图合成基准测试中显著优于现有方法。
cs.CV / 101 / 2607.00839
Rethinking Multi-Label Image Classification With Deep Learning: Taxonomy, Challenge, and Outlook
重新思考深度学习下的多标签图像分类：分类、挑战与展望
Abstract
Multi-label image classification (MLIC), a fundamental task in computer vision, focuses on identifying multiple objects or concepts within an image, underpinning numerous read-world applications, such as autonomous driving, disease diagnosis, recommendation system, and mobile service robot. Over the past decade, deep learning paradigms based on convolutional neural networks, recurrent neural networks, and Transformers have significantly advanced this field, owing to their powerful capability in visual representation and relationship modeling. These advances have markedly improved the robustness, scalability, and generalization ability of MLIC models across diverse datasets and application domains. In this survey, we provide a comprehensive review of the deep learning-based literature on MLIC. Concretely, we first revisit the background, including problem definition, datasets, backbones and evaluation metrics. Next, we develop a plausible taxonomy for the deep learning-based MLIC approaches, organizing them into six groups: region-oriented methods, label-oriented methods, architecture-oriented methods, representation-oriented methods, learning-oriented methods, and data-oriented methods. Finally, we provide an insightful exposition of the underlying learning game in MLIC and its implications for other vision domains, and we empirically summarize the key challenges and research directions in MLIC while outlining promising avenues for future development. We believe this survey offers the research community a holistic and systematic perspective on MLIC, thereby facilitating subsequent exploration and innovation in this field and beyond.
Chinese Translation
多标签图像分类（MLIC）是计算机视觉中的一项基础任务，旨在识别图像中的多个对象或概念，支撑着许多现实世界的应用，如自动驾驶、疾病诊断、推荐系统和移动服务机器人。在过去十年中，基于卷积神经网络、递归神经网络和变换器（Transformers）的深度学习范式显著推动了这一领域的发展，得益于其在视觉表示和关系建模方面的强大能力。这些进展显著提高了MLIC模型在不同数据集和应用领域中的鲁棒性、可扩展性和泛化能力。在本次综述中，我们对基于深度学习的MLIC文献进行了全面回顾。具体而言，我们首先回顾了背景，包括问题定义、数据集、骨干网络和评估指标。接下来，我们为基于深度学习的MLIC方法开发了一个合理的分类法，将其组织为六个组别：区域导向方法、标签导向方法、架构导向方法、表示导向方法、学习导向方法和数据导向方法。最后，我们深入探讨了MLIC中潜在的学习博弈及其对其他视觉领域的影响，并经验性地总结了MLIC中的关键挑战和研究方向，同时勾勒出未来发展的有希望的途径。我们相信这项综述为研究社区提供了一个全面而系统的视角，从而促进该领域及其他领域的后续探索和创新。
cs.CV / 102 / 2607.00850
Mirror-Fusion Attention for Reflection-Aware Self-Supervised Representation Learning
反射感知自监督表示学习的镜像融合注意力
Abstract
Most self-supervised learning (SSL) methods encourage invariance across augmentations, but strict flip invariance can suppress informative left--right correspondences in approximately bilateral data such as medical images and human faces. We propose Mirror-Fusion-Augmented Self-Supervised Learning (MFASSL), a Vision Transformer framework that injects a soft reflection prior into standard SSL without redesigning the backbone. MFASSL constructs mirror-paired views aligned to an estimated symmetry axis and introduces a lightweight Mirror-Fusion Attention (MFA) module for adaptive token-level interaction between mirrored regions while preserving asymmetric cues. The base SSL objective is further coupled with reflection-consistency and mid-layer token-alignment losses. Across CheXpert, BraTS, CelebA-HQ, and WFLW, MFASSL improves downstream performance, calibration, and reflection robustness over MoCo-v3, DINO, and MAE baselines under matched ViT-B/16 settings. It also achieves stronger and more consistent gains than recent equivariant SSL approaches with only approximately 2.7\% additional parameters. These results show that lightweight geometry-aware priors can effectively complement invariance-based SSL.
Chinese Translation
大多数自监督学习（SSL）方法鼓励在数据增强之间保持不变性，但严格的翻转不变性可能会抑制在近似双边数据（如医学图像和人脸）中有信息量的左右对应关系。我们提出了镜像融合增强自监督学习（MFASSL），这是一个视觉变换器框架，它在不重新设计主干网络的情况下，将软反射先验注入标准SSL中。MFASSL构建了与估计对称轴对齐的镜像配对视图，并引入了轻量级的镜像融合注意力（MFA）模块，以实现镜像区域之间的自适应令牌级交互，同时保留不对称线索。基础SSL目标进一步与反射一致性和中间层令牌对齐损失相结合。在CheXpert、BraTS、CelebA-HQ和WFLW数据集上，MFASSL在与MoCo-v3、DINO和MAE基线相匹配的ViT-B/16设置下，提升了下游性能、校准和反射鲁棒性。它还以仅约2.7%的额外参数，实现了比最近的等变SSL方法更强且更一致的提升。这些结果表明，轻量级的几何感知先验可以有效补充基于不变性的SSL。
cs.CV / 103 / 2607.00858
MoVA: Learning Asymmetric Dual Projections for Modular Long Video-Text Alignment
MoVA：用于模块化长视频-文本对齐的非对称双重投影学习
Abstract
Contrastive pre-training has propelled video-text alignment, yet models often inherit the critical limitations of their image-text predecessors like CLIP, resulting in entangled representations. These challenges are severely exacerbated by two fundamental properties in the video domain: Temporal Misalignment, where textual descriptions often correlate only to specific, constrained temporal windows, leaving other frames text-irrelevant; and Semantic Asymmetry, which dictates a sparse, bidirectional, and non-equivalent relevance between frame-level visual details and caption-level concepts. This failure persists whether captions are short and temporally disjoint, creating ambiguity, or long and detailed, fostering entanglement between static objects and their temporal evolution. In this paper, we establish theoretical conditions that enable flexible alignment between video and text representations across the temporal dimension and at varying levels of granularity. Building on these theoretical insights, we introduce MoVA, Modular Long Video-Text Alignment, which learns dual asymmetric projections: a text-side projection that adaptively selects frame-aware subspaces of the caption, and a video-side projection that disentangles text-relevant visual concepts. Our framework ensures that the model can preserve global cross-modal semantics while disentangling evolving, frame-specific concepts and scale naturally to long captions and videos. Empirical evaluations show that MoVA outperforms existing methods in multiple video-text alignment tasks, demonstrating the effectiveness of our method.
Chinese Translation
对比预训练推动了视频-文本对齐的发展，但模型往往继承了其图像-文本前身（如 CLIP）的关键局限性，导致表示的纠缠。这些挑战在视频领域由于两个基本特性而严重加剧：时间错位（Temporal Misalignment），即文本描述往往仅与特定的、受限的时间窗口相关，导致其他帧与文本无关；以及语义不对称（Semantic Asymmetry），这决定了帧级视觉细节与标题级概念之间的稀疏、双向且不等价的相关性。这种失败在标题短且时间上不连贯时会造成歧义，或在标题长且详细时会促进静态物体与其时间演变之间的纠缠。在本文中，我们建立了理论条件，使得视频与文本表示能够在时间维度和不同粒度水平上灵活对齐。基于这些理论见解，我们引入了 MoVA（模块化长视频-文本对齐），它学习了双重非对称投影：一种文本侧投影，能够自适应选择与标题相关的帧感知子空间；另一种视频侧投影，能够解开与文本相关的视觉概念。我们的框架确保模型能够在解开不断演变的帧特定概念的同时，保持全局跨模态语义，并自然扩展到长标题和视频。实证评估表明，MoVA 在多个视频-文本对齐任务中优于现有方法，证明了我们方法的有效性。
cs.CV / 104 / 2607.00861
TrajLoc: Trajectory-Attention Localization for Multi-Object Motion Control
TrajLoc：用于多目标运动控制的轨迹注意力定位
Abstract
Controlling the motion of multiple objects in image-to-video (I2V) generation requires preserving object identities while enforcing adherence to distinct target trajectories. This becomes particularly challenging as the number of objects increases and their paths intersect or occlude one another. Existing approaches entangle multiple trajectories within a shared, dense conditioning signal, making object-level correspondence difficult to preserve in crowded scenes. We depart from this paradigm and enforce a strict, per object spatial constraint that isolates instances independently. Our method, TrajLoc, achieves this directly within the attention layers by substituting the cross-attention weights of each object token with a Gaussian heatmap centered on its target location at every frame. The same per object token interface carries trajectory and depth through a learned embedding and preserves identity by encoding first frame appearance in place of an object token. Evaluations across six datasets, featuring up to 20 simultaneously controlled objects and out of distribution real world scenes, demonstrate that our method consistently improves both visual fidelity and trajectory adherence. Applied to two architecturally distinct backbones (CogVideoX 5B and WaN 2.1 14B), our approach achieves average gains of +4.3 dB PSNR and a 51% reduction in trajectory end point error compared to the strongest baselines. Project page: https://sela-omer.github.io/traj-loc/
Chinese Translation
在图像到视频（I2V）生成中控制多个对象的运动需要保持对象的身份，同时确保遵循不同的目标轨迹。随着对象数量的增加以及它们的路径相互交叉或遮挡，这一任务变得尤为困难。现有的方法将多个轨迹纠缠在一个共享的、密集的条件信号中，使得在拥挤场景中保持对象级的对应关系变得困难。我们脱离了这一范式，强制实施严格的每个对象空间约束，从而独立地隔离实例。我们的方法TrajLoc直接在注意力层中实现这一点，通过在每一帧中用一个以目标位置为中心的高斯热图替代每个对象标记的交叉注意力权重。相同的每个对象标记接口通过学习的嵌入携带轨迹和深度，并通过编码第一帧的外观来保持身份，而不是使用对象标记。对六个数据集的评估显示，最多可同时控制20个对象，并且在分布外的真实场景中，我们的方法始终提高了视觉保真度和轨迹遵循性。应用于两个架构上明显不同的主干网络（CogVideoX 5B 和 WaN 2.1 14B），我们的方法在与最强基线相比时，平均提高了+4.3 dB PSNR，并减少了51%的轨迹端点误差。项目页面：https://sela-omer.github.io/traj-loc/
cs.CV / 105 / 2607.00867
EFlow: Learning Evidence Flow for Long-Video Reasoning with Adaptive Reflection
EFlow：通过自适应反思学习长视频推理中的证据流
Abstract
Long-video reasoning is fundamentally constrained by how models acquire and utilize visual evidence. Existing tool-augmented video frameworks often interleave temporal grounding and answer reasoning within a single trajectory, causing early semantic hypotheses to bias evidence localization. We term this failure mode premature semantic commitment, where biased grounding retrieves incomplete evidence and incomplete evidence further reinforces incorrect reasoning. To address this issue, we propose EFlow, an evidence-first video reasoning framework built upon Qwen3-VL. EFlow explicitly separates temporal grounding and logical reasoning through CoT for Temporal Grounding and CoT for Reasoning, enabling the model to retrieve relevant evidence before answer inference. In addition, EFlow introduces a confidence-aware reflection mechanism that re-evaluates the full video when retrieved evidence is potentially insufficient. We further construct dedicated trajectory datasets and train EFlow through supervised fine-tuning, reinforcement learning, and reinforcement fine-tuning. Extensive experiments across five video understanding benchmarks demonstrate that EFlow consistently improves long-video reasoning performance.
Chinese Translation
长视频推理在根本上受到模型获取和利用视觉证据方式的限制。现有的工具增强视频框架通常在单一轨迹中交替进行时间定位和答案推理，导致早期语义假设偏向于证据定位。我们将这种失败模式称为早期语义承诺（premature semantic commitment），其中偏向的定位检索到不完整的证据，而不完整的证据进一步强化了错误的推理。为了解决这个问题，我们提出了EFlow，一个基于Qwen3-VL构建的以证据为先的视频推理框架。EFlow通过时间定位的链式思维（CoT for Temporal Grounding）和推理的链式思维（CoT for Reasoning）明确分离时间定位和逻辑推理，使模型能够在答案推断之前检索相关证据。此外，EFlow引入了一种基于置信度的反思机制，当检索到的证据可能不足时，重新评估整个视频。我们进一步构建了专门的轨迹数据集，并通过监督微调、强化学习和强化微调训练EFlow。在五个视频理解基准上的大量实验表明，EFlow在长视频推理性能上始终有所提升。
cs.CV / 106 / 2607.00881
OmniView-Space: Reinforcing Spatial Reasoning via Multi-Perspective Spatial Mapping
OmniView-Space：通过多视角空间映射增强空间推理
Abstract
Spatial intelligence remains a persistent challenge for Multimodal Large Language Models (MLLMs), as it requires coherent spatial scene representations beyond basic object recognition. Existing methods typically build such representations through textual reasoning or 3D reconstruction. However, they often falter during multi-step reasoning, particularly when required to dynamically re-anchor evidence to the specific camera-, object-, or direction-centric reference frames demanded by complex queries. To address this, we propose OmniView-Space, a framework designed to maintain spatial consistency through multimodal egocentric evidence. Our approach consists of three core components: (1) Multi-Perspective Spatial Mapping (MPSM), which re-anchors reconstructed geometry into a query-aligned visual cognitive map and a textual spatial graph; (2) Tool-Guided Egocentric Reasoning, an interleaved policy trained to actively select the ego anchor required by the query and request the corresponding MPSM evidence; and (3) Cognitive-Map Distillation, which uses MPSM-generated trajectories and ego-frame rewards to train the model to reason with self-generated cognitive maps. Experiments on single- and multi-image spatial reasoning benchmarks show that OmniView-Space achieves state-of-the-art performance. Furthermore, the distilled model maintains this performance while reducing reliance on external geometry pipelines.
Chinese Translation
空间智能仍然是多模态大型语言模型（MLLMs）面临的一个持续挑战，因为它需要超越基本物体识别的连贯空间场景表示。现有方法通常通过文本推理或三维重建来构建此类表示。然而，在多步骤推理过程中，它们常常表现不佳，特别是在需要将证据动态重新锚定到复杂查询所要求的特定相机、物体或方向中心参考框架时。为了解决这个问题，我们提出了OmniView-Space，一个旨在通过多模态自我中心证据保持空间一致性的框架。我们的方法由三个核心组件组成：（1）多视角空间映射（Multi-Perspective Spatial Mapping, MPSM），将重建的几何体重新锚定到与查询对齐的视觉认知地图和文本空间图中；（2）工具引导的自我中心推理（Tool-Guided Egocentric Reasoning），一种交错策略，经过训练能够主动选择查询所需的自我锚定，并请求相应的MPSM证据；（3）认知地图蒸馏（Cognitive-Map Distillation），利用MPSM生成的轨迹和自我框架奖励来训练模型使用自生成的认知地图进行推理。在单图和多图空间推理基准上的实验表明，OmniView-Space达到了最先进的性能。此外，蒸馏后的模型在减少对外部几何管道依赖的同时，仍然保持这一性能。
cs.CV / 107 / 2607.00885
Improving Sparse-View 3DGS Generalization via Flat Minima Optimization
通过平坦极小值优化提高稀疏视图3D高斯点云的泛化能力
Abstract
Recent advances in neural rendering have established 3D Gaussian Splatting (3DGS) as a highly efficient representation for novel view synthesis, enabling fast training and real-time rendering with strong fidelity. However, when supervision is limited to sparse input views, 3DGS tends to overfit to the observed images and generalize poorly to unseen viewpoints. We address this challenge from the perspective of flat minima (FM) optimization, which seeks solutions that remain stable under small parameter perturbations. Viewing Gaussian parameters as trainable weights, we adapt FM principles to the geometric and dynamic nature of 3DGS with a lightweight training framework. Our method regularizes optimization with controlled Gaussian perturbations that account for each Gaussian's anisotropy and the training progress, preserving fine details while improving robustness to sparse-view overfitting. To further stabilize this flat minima optimization process, we introduce periodic reinitialization, which temporarily returns non-positional parameters to their initial states for a short window. Together, these techniques integrate seamlessly into existing 3DGS pipelines without architectural changes. Experiments on LLFF and Mip-NeRF360 datasets demonstrate improved quantitative metrics and perceptual quality under sparse-view supervision, producing reconstructions that are sharper, more stable, and better generalized to novel viewpoints.
Chinese Translation
近年来，神经渲染的进展使得3D高斯点云（3DGS）成为一种高效的新视图合成表示，能够实现快速训练和实时渲染，同时保持较强的保真度。然而，当监督仅限于稀疏输入视图时，3DGS往往会对观察到的图像过拟合，并在未见视点上泛化能力较差。我们从平坦极小值（FM）优化的角度解决这一挑战，FM优化旨在寻找在小参数扰动下保持稳定的解。将高斯参数视为可训练的权重，我们将FM原理适应于3DGS的几何和动态特性，并采用轻量级训练框架。我们的方法通过控制高斯扰动来正则化优化，考虑每个高斯的各向异性和训练进度，既保留细节，又提高对稀疏视图过拟合的鲁棒性。为了进一步稳定这一平坦极小值优化过程，我们引入周期性重新初始化，暂时将非位置参数恢复到初始状态，持续短时间。结合这些技术，无需架构更改即可无缝集成到现有的3DGS管道中。在LLFF和Mip-NeRF360数据集上的实验表明，在稀疏视图监督下，定量指标和感知质量均有所改善，生成的重建图像更加清晰、稳定，并且更好地泛化到新视点。
cs.CV / 108 / 2607.00886
Beyond Pixel Overlap: A Framework for Decomposing Segmentation Evaluation Metrics
超越像素重叠：分解分割评估指标的框架
Abstract
Evaluation metrics are central to binary target segmentation because they determine how progress is measured, compared, and interpreted. In this paper, target denotes the task-defined positive region to be segmented rather than a generic foreground object. It may be salient, camouflaged, transparent, glass-like, mirror-like, shadow-like, lesion-like, or defined by other application-specific semantics. We treat existing metrics as compositions of modular design choices rather than isolated formulas. The proposed framework decomposes each metric into five stages covering prediction representation, target extraction, target matching, score computation, and metric reporting. We use this framework to analyze representative metrics and show how newer metrics address specific limits in earlier protocols. The stage choices keep each metric's assumptions visible. We then discuss the design space opened by the framework and its implications for task-aware evaluation protocols. Reference code is available at https://github.com/lartpang/PySODMetrics.
Chinese Translation
评估指标在二元目标分割中至关重要，因为它们决定了进展的测量、比较和解释方式。在本文中，目标指的是任务定义的正区域，而不是一般的前景物体。该区域可能是显著的、伪装的、透明的、玻璃状的、镜面状的、阴影状的、病变状的，或由其他特定应用语义定义。我们将现有指标视为模块化设计选择的组合，而不是孤立的公式。所提出的框架将每个指标分解为五个阶段，涵盖预测表示、目标提取、目标匹配、得分计算和指标报告。我们利用该框架分析代表性指标，并展示新指标如何解决早期协议中的特定局限性。阶段选择使每个指标的假设保持可见。随后，我们讨论了该框架所打开的设计空间及其对任务感知评估协议的影响。参考代码可在 https://github.com/lartpang/PySODMetrics 获取。
cs.CV / 109 / 2607.00887
Geometry-Aware Cross-Height Channel Knowledge Map Prediction for UAV-Assisted Communications With Uncertainty-Guided 3D Sensing
基于几何感知的跨高度通道知识图预测用于不确定性引导的无人机辅助通信
Abstract
Low-altitude Unmanned Aerial Vehicles (UAVs) often need to infer channel knowledge across a range of heights from only sparse observations collected at a few altitude layers. To address this challenge, this paper studies height-conditioned cross-height channel knowledge map (CKM) prediction for UAV-assisted communications in geometry-rich urban environments. We develop a geometry-aware conditional prediction framework that combines urban scene priors, sparse multi-altitude observations, and target-height descriptors to reconstruct dense CKMs at unobserved target heights. An uncertainty head is further introduced to characterize prediction confidence and to support cost-aware online UAV sensing under motion and safety constraints. Experiments on a layered aerial CKM benchmark show that the proposed Feature Pyramid Network (FPN)-Transformer achieves the best overall performance under both unseen-scene zero-shot and legacy patch-random protocols, reducing the Root Mean Square Error (RMSE) to 5.347dB and 1.111dB, respectively, compared with 6.937dB and 1.221dB for the strongest baseline 3D-RadioDiff. Moreover, after applying our unseen-scene few-shot adaptation, the RMSE further decreases from 5.347dB in zero-shot prediction to 3.518dB with 10-shot two-height support, while the uncertainty-guided cost-aware sensing policy improves active reconstruction from 6.94dB at initialization to 4.79dB at sensing budget 40, outperforming uncertainty-only sensing at 5.08dB and random aerial sampling at 5.84dB.
Chinese Translation
低空无人机（UAV）通常需要仅通过在少数高度层收集的稀疏观测推断跨高度的通道知识。为了解决这一挑战，本文研究了在几何丰富的城市环境中，针对无人机辅助通信的高度条件跨高度通道知识图（CKM）预测。我们开发了一个几何感知的条件预测框架，该框架结合了城市场景先验、稀疏的多高度观测和目标高度描述符，以重建在未观测目标高度下的密集CKM。此外，引入了不确定性头以表征预测信心，并支持在运动和安全约束下的成本感知在线无人机感知。在分层空中CKM基准上的实验表明，所提出的特征金字塔网络（FPN）-Transformer在未见场景零-shot和传统补丁随机协议下均实现了最佳整体性能，将均方根误差（RMSE）分别降低至5.347dB和1.111dB，而最强基线3D-RadioDiff的RMSE为6.937dB和1.221dB。此外，在应用我们的未见场景少-shot适应后，RMSE从零-shot预测中的5.347dB进一步降低至在10-shot两高度支持下的3.518dB，而不确定性引导的成本感知感知策略将主动重建的RMSE从初始化时的6.94dB改善至感知预算40时的4.79dB，优于仅基于不确定性的感知（5.08dB）和随机空中采样（5.84dB）。
cs.CV / 110 / 2607.00889
DeWorldSG: Depth-Aware 3D Semantic Scene Graph Generation via World-Model Priors
DeWorldSG：基于世界模型先验的深度感知3D语义场景图生成
Abstract
We present DeWorldSG, a novel framework that generates spatio-temporally robust 3D Semantic Scene Graphs from RGB-D sequences. Existing methods often struggle to construct reliable 3D scene graphs due to unstable 3D object representations and missing relations caused by frame-wise inference. DeWorldSG addresses these issues by estimating instance-level geometric 3D Gaussian distributions through depth-guided filtering and representing each object as a probabilistic 3D node rather than a single projected point. To mitigate relational sparsity from frame-wise inference, our framework further aggregates spatiotemporal evidence across object pairs and refines relations using contextual priors derived from a world model (V-JEPA 2). Experiments on the 3DSSG and ReplicaSSG datasets demonstrate state-of-the-art (SoTA) performance in both object and predicate prediction, while producing temporally consistent scene structures. In particular, our method improves triplet recall by 77.4% and predicate recall by 23.2% over prior SoTA approaches, making it suitable for robotic manipulation and AR applications. Our code and models are open-sourced.
Chinese Translation
我们提出了DeWorldSG，一个新颖的框架，能够从RGB-D序列中生成时空稳健的3D语义场景图。现有方法常常因不稳定的3D物体表示和由逐帧推理导致的缺失关系而难以构建可靠的3D场景图。DeWorldSG通过深度引导过滤估计实例级几何3D高斯分布，并将每个物体表示为一个概率性3D节点，而不是单个投影点，从而解决了这些问题。为了缓解逐帧推理带来的关系稀疏性，我们的框架进一步聚合物体对之间的时空证据，并利用源自世界模型（V-JEPA 2）的上下文先验来细化关系。在3DSSG和ReplicaSSG数据集上的实验表明，我们的方法在物体和谓词预测方面均表现出最先进的（SoTA）性能，同时生成时序一致的场景结构。特别是，我们的方法在三元组召回率上提高了77.4%，在谓词召回率上提高了23.2%，相较于之前的SoTA方法，使其适用于机器人操作和增强现实应用。我们的代码和模型已开源。
cs.CV / 111 / 2607.00902
MG-RWKV: Multi-Grained Context-Aware RWKV for Temporal Forgery Localization
MG-RWKV：用于时间伪造定位的多粒度上下文感知RWKV
Abstract
Driven by Artificial Intelligence-Generated Content (AIGC), the authenticity of audio-visual content is facing severe challenges. Temporal Forgery Localization (TFL) aims to precisely identify manipulated segments within untrimmed sequences. However, existing methods are limited by CNNs' local receptive fields or Transformers' quadratic complexity, while emerging linear models often struggle to balance global authentic context compression with local abrupt forgery perception. To address this, we propose MG-RWKV, a multi-granularity framework that leverages the data-dependent state evolution of RWKV to achieve efficient full-sequence processing with O(T) complexity. Our framework features three core innovations: (1) a Bidirectional RWKV architecture that captures bidirectional temporal contexts without quadratic overhead; (2) a Multi-Granularity Mixture of Experts (MG-MoE) that performs dynamic routing over explicit temporal receptive fields, adaptively selecting granularities based on forgery duration to significantly enhance decision interpretability; and (3) Cross-Granularity Consistency (CGC), which aligns adjacent feature pyramid levels through hierarchical scale-wise pairing and spatial boundary-aware weighting, effectively reducing false positives in authentic regions. Extensive experiments on Lav-DF, TVIL, and Psynd datasets demonstrate that MG-RWKV achieves state-of-the-art performance with low computational cost.
Chinese Translation
在人工智能生成内容（AIGC）的推动下，音视频内容的真实性面临严峻挑战。时间伪造定位（TFL）旨在精确识别未裁剪序列中的操控片段。然而，现有方法受到卷积神经网络（CNN）局部感受野或变换器（Transformers）二次复杂度的限制，而新兴的线性模型往往难以平衡全球真实上下文压缩与局部突发伪造感知。为此，我们提出了MG-RWKV，一种多粒度框架，利用RWKV的数据依赖状态演变，以O(T)复杂度实现高效的全序列处理。我们的框架具有三项核心创新：（1）一种双向RWKV架构，能够在不增加二次开销的情况下捕获双向时间上下文；（2）一种多粒度专家混合模型（MG-MoE），在显式时间感受野上执行动态路由，根据伪造持续时间自适应选择粒度，从而显著增强决策可解释性；（3）跨粒度一致性（CGC），通过分层尺度配对和空间边界感知加权对相邻特征金字塔层进行对齐，有效减少真实区域的误报。在Lav-DF、TVIL和Psynd数据集上的大量实验表明，MG-RWKV以低计算成本实现了最先进的性能。
cs.CV / 112 / 2607.00916
Condensing Large-Scale Datasets Directly with Minimal Information Loss
直接以最小信息损失压缩大规模数据集
Abstract
Recent advancements in scaling dataset distillation rely heavily on decoupled information extraction pipelines, comprising SQUEEZE, RECOVER, and RELABEL stages. Despite their scalability to large-scale datasets, these methods suffer from prohibitive computational overhead and poor cross-architecture generalization. In this paper, we reveal the root cause of these bottlenecks: the implicit dual-compression process, from data to model and back to images, inherently induces severe information loss. Crucially, we empirically and theoretically demonstrate that this loss creates a distribution shift that fundamentally compromises the widely adopted RELABEL strategy, transforming the pre-trained model into an unreliable labeler that yields sub-optimal labels. To overcome these critical flaws, we propose CIM, a novel, metric-driven framework that abandons the flawed dual-compression paradigm. Instead, CIM explicitly quantifies and minimizes the information gap between the original and synthetic datasets. By directly aligning the data distributions, our approach ensures high-fidelity information condensation and inherently satisfies the prerequisites for effective relabeling. Extensive experiments demonstrate that CIM establishes a new state-of-the-art. Notably, it distills ImageNet-1K at an IPC=10 in merely 80 minutes on a single RTX-4090 GPU, achieving an unprecedented 48.7% Top-1 accuracy on ResNet-18 and significantly outperforming previous SOTA approaches, such as NRR-DD and DELT, by 2.6% and 2.9%, respectively. Our code is available at https://github.com/LINs-lab/CIM.
Chinese Translation
近期在数据集蒸馏规模化方面的进展主要依赖于解耦的信息提取流程，包括 SQUEEZE、RECOVER 和 RELABEL 阶段。尽管这些方法能够扩展到大规模数据集，但它们面临着高昂的计算开销和较差的跨架构泛化能力。本文揭示了这些瓶颈的根本原因：隐式的双重压缩过程，从数据到模型再到图像，固有地导致了严重的信息损失。至关重要的是，我们通过实证和理论证明，这种损失造成了分布偏移，根本上损害了广泛采用的 RELABEL 策略，使得预训练模型变成了一个不可靠的标签生成器，产生次优标签。为了克服这些关键缺陷，我们提出了 CIM，一个新颖的以度量为驱动的框架，摒弃了有缺陷的双重压缩范式。相反，CIM 明确量化并最小化原始数据集与合成数据集之间的信息差距。通过直接对齐数据分布，我们的方法确保了高保真度的信息压缩，并固有地满足有效重新标注的前提条件。大量实验表明，CIM 建立了新的最先进水平。值得注意的是，它在单个 RTX-4090 GPU 上以 IPC=10 的速度在仅 80 分钟内蒸馏了 ImageNet-1K，达到了前所未有的 48.7% Top-1 准确率，显著超越了之前的最先进方法，如 NRR-DD 和 DELT，分别提高了 2.6% 和 2.9%。我们的代码可在 https://github.com/LINs-lab/CIM 获取。
cs.CV / 113 / 2607.00920
GMO-E$^2$DIT: Grounded Multi-Operation Editing for E-Commerce Images
GMO-E$^2$DIT：面向电子商务图像的多操作编辑框架
Abstract
Real-world e-commerce image editing often requires multiple, localized, and auditable operations rather than global restyling. This compositional nature poses a dual challenge: models must precisely apply all requested edits to the correct regions while preserving unmodified content, even under ambiguous instructions. Existing one-shot editors conflate intent resolution, spatial grounding, and synthesis into a single step, frequently resulting in partial execution failures, which is unacceptable for commercial scenarios. To address this, we introduce GMO-E$^2$DIT, an agentic editing framework that couples a Vision-Language Model (VLM) with a mask-conditioned image editor to tackle structured multi-turn task completion. Given an underspecified instruction, the VLM agent constructs a region-grounded edit agenda, effectively decoupling cognitive reasoning from generative rendering. The framework then executes sub-programs via operation-aware masks and references, utilizing a reflection-driven loop to inspect intermediate results and determine the subsequent state. This iterative mechanism reliably preserves safe partial progress, retries unfinished operations, and recovers from errors. Furthermore, we develop a unified data pipeline providing aligned supervision for planning, execution, and reflection, alongside EComEditBench, a comprehensive benchmark for instruction-driven evaluation. Extensive experiments demonstrate that GMO-E$^2$DIT achieves competitive performance compared to strong closed-source models, yielding superior instruction accuracy and edit fidelity over existing baselines.
Chinese Translation
现实世界中的电子商务图像编辑通常需要多种局部且可审计的操作，而非全局重塑。这种组合特性带来了双重挑战：模型必须精确地将所有请求的编辑应用于正确的区域，同时保留未修改的内容，即使在模糊的指令下也是如此。现有的一次性编辑器将意图解析、空间定位和合成混为一谈，常常导致部分执行失败，这在商业场景中是不可接受的。为了解决这个问题，我们提出了GMO-E$^2$DIT，一个代理式编辑框架，它将视觉语言模型（Vision-Language Model, VLM）与基于掩膜的图像编辑器结合起来，以应对结构化的多轮任务完成。在给定不明确的指令时，VLM代理构建了一个区域定位的编辑议程，有效地将认知推理与生成渲染解耦。该框架随后通过操作感知的掩膜和引用执行子程序，利用反思驱动的循环检查中间结果并确定后续状态。这一迭代机制可靠地保留安全的部分进展，重试未完成的操作，并从错误中恢复。此外，我们开发了一个统一的数据管道，为规划、执行和反思提供对齐的监督，并推出了EComEditBench，一个全面的基准，用于基于指令的评估。大量实验表明，GMO-E$^2$DIT在与强大的闭源模型相比时，表现出竞争力，且在指令准确性和编辑保真度上优于现有基准。
cs.CV / 114 / 2607.00927
Post-Training Pruning for Diffusion Transformers
扩散变换器的后训练剪枝
Abstract
Diffusion Transformers (DiTs) have demonstrated impressive performance in image generation but suffer from substantial computational overhead and resource consumption. Post-training pruning offers a promising solution; however, due to DiTs' unique architectural design and parameter distribution, traditional pruning methods are inapplicable, leading to significant performance degradation. Specifically, prior methods developed for LLMs, which derive metrics through a series of approximations, amplify the relative contribution of weights in the saliency metric. In addition, weights in DiTs exhibit significantly larger magnitudes than those in LLMs. Moreover, existing pruning granularity overlooks variations in model structures. In this paper, we propose DiT-Pruning, which improves pruning performance by introducing customized saliency criteria and pruning granularity. We design a novel metric that balances the contributions of weights and activations from an energy-based perspective, enabling more effective identification of important elements. Furthermore, we observe distinct clustering patterns in the two-dimensional weight space. Accordingly, we adopt a clustering-aware pruning granularity, enabling effective sparse allocation. Extensive evaluations on various DiTs show that our method consistently preserves image quality, especially under high sparsity. For FLUX.1-dev at 512x512 resolution on MJHQ, DiT-Pruning achieves only a 0.001 loss in CLIP score at 50% sparsity, dramatically outperforming recent pruning methods.
Chinese Translation
扩散变换器（Diffusion Transformers, DiTs）在图像生成方面表现出色，但面临着显著的计算开销和资源消耗。后训练剪枝提供了一种有前景的解决方案；然而，由于DiTs独特的架构设计和参数分布，传统的剪枝方法并不适用，导致性能显著下降。具体而言，以往为大型语言模型（LLMs）开发的方法通过一系列近似来推导度量，放大了权重在显著性度量中的相对贡献。此外，DiTs中的权重的幅度明显大于LLMs中的权重。此外，现有的剪枝粒度忽视了模型结构的变化。本文提出了DiT-Pruning，通过引入定制的显著性标准和剪枝粒度来提高剪枝性能。我们设计了一种新的度量，从基于能量的角度平衡权重和激活的贡献，从而更有效地识别重要元素。此外，我们观察到二维权重空间中的明显聚类模式。因此，我们采用了聚类感知的剪枝粒度，能够有效地进行稀疏分配。在对各种DiTs进行的广泛评估中，我们的方法在高稀疏性下始终保持图像质量。对于MJHQ上512x512分辨率的FLUX.1-dev，DiT-Pruning在50%稀疏性下仅损失0.001的CLIP分数，显著优于近期的剪枝方法。
cs.CV / 115 / 2607.00948
Dataset Biases and Shortcut Learning in Motion-Based AI-Generated Video Detection
运动基础的人工智能生成视频检测中的数据集偏差与捷径学习
Abstract
The visual quality of AI-generated videos has improved drastically in recent years, making it increasingly difficult for humans to distinguish between real and synthetic media. In this work, we evaluate the robustness and applicability of four state-of-the-art motion-based AI-generated video detectors. We identify significant preprocessing and sampling biases in these methods and demonstrate that they account for a substantial portion of their reported performance. Furthermore, we find that these detectors are highly sensitive to motion patterns specific to their evaluation datasets, where AI-generated videos generally exhibit less inter-frame movement than real videos. We show that for all detectors, performance collapses to near-random levels when evaluated on a dataset that does not contain this motion bias. Additionally, through dataset rebalancing and the application of simple spatial augmentations, we observe severe performance degradation across all evaluated models. In contrast, we find that an existing frequency-based detector maintains strong performance across all evaluated datasets, suggesting that frequency-based approaches may offer a more generalizable path forward for AI-generated video detection. We hope that our work raises awareness towards these vulnerabilities and encourages the development of more representative, unbiased datasets and more robust evaluation protocols.
Chinese Translation
近年来，人工智能生成视频的视觉质量显著提高，使得人类越来越难以区分真实媒体与合成媒体。在本研究中，我们评估了四种最先进的运动基础人工智能生成视频检测器的鲁棒性和适用性。我们发现这些方法存在显著的预处理和采样偏差，并证明这些偏差占据了其报告性能的相当一部分。此外，我们发现这些检测器对其评估数据集特定的运动模式高度敏感，其中人工智能生成的视频通常表现出比真实视频更少的帧间运动。我们展示了对于所有检测器，当在不包含这种运动偏差的数据集上进行评估时，性能会降至接近随机水平。此外，通过数据集重平衡和简单空间增强的应用，我们观察到所有评估模型的性能严重下降。相反，我们发现一种现有的基于频率的检测器在所有评估数据集上保持了强劲的性能，这表明基于频率的方法可能为人工智能生成视频检测提供了更具普适性的前进路径。我们希望我们的工作能够提高人们对这些脆弱性的认识，并鼓励开发更具代表性、无偏见的数据集和更为鲁棒的评估协议。
cs.CV / 116 / 2607.00955
Learning Cardiac Motion Priors for Implicit Neural Representations
学习心脏运动先验用于隐式神经表示
Abstract
Implicit neural representations (INRs) are well suited to cardiac motion estimation, providing continuous, compact representations of motion fields. However, fitting an INR to each image sequence is time-consuming and sensitive to the optimisation trajectory. Learned priors can help guide optimisation towards plausible motion fields and enable faster adaptation, but learning priors for cardiac motion INRs remains under-explored. In this work, we compare four strategies for learning cardiac motion priors, including a population prior learned by joint optimisation, a consensus prior obtained by weight averaging, auto-decoders, and meta-learning. Using short-axis tagged cardiac magnetic resonance images from the UK Biobank, we evaluate their impact on tracking accuracy, motion behaviour, and adaptation trajectory. All learned priors substantially improved early adaptation performance compared with random initialisation. While the simple consensus prior was effective, auto-decoders recovered large deformations faster during early adaptation. Meta-learning achieved strong early performance and maintained the best adaptation trajectory over 50 iterations.
Chinese Translation
隐式神经表示（INRs）非常适合心脏运动估计，能够提供连续、紧凑的运动场表示。然而，为每个图像序列拟合一个INR既耗时又对优化轨迹敏感。学习的先验可以帮助引导优化朝向合理的运动场，并实现更快的适应，但针对心脏运动INRs的先验学习仍然未被充分探索。在本研究中，我们比较了四种学习心脏运动先验的策略，包括通过联合优化学习的群体先验、通过权重平均获得的共识先验、自编码器和元学习。我们使用来自英国生物库的短轴标记心脏磁共振图像，评估这些策略对跟踪精度、运动行为和适应轨迹的影响。与随机初始化相比，所有学习的先验在早期适应性能上都有显著提高。虽然简单的共识先验有效，但自编码器在早期适应过程中恢复大变形的速度更快。元学习在早期表现上取得了强劲的结果，并在50次迭代中保持了最佳的适应轨迹。
cs.CV / 117 / 2607.00959
GaussianEmoTalker: Real-Time Emotional Talking Head Synthesis with Audio-Driven and Blendshape-Based 3D Gaussian Splatting
GaussianEmoTalker：基于音频驱动和混合形状的3D高斯点云实时情感对话头合成
Abstract
Audio-driven talking head synthesis has achieved impressive progress in lip synchronization and visual quality, yet generating expressive emotional avatars with controllable intensity remains challenging, especially under real-time constraints. In this paper, we present GaussianEmoTalker, an audio-driven framework for real-time emotional talking head synthesis based on 3D Gaussian Splatting. Instead of directly predicting the final emotional avatar from speech, we formulate emotional animation as a neutral-to-emotional residual deformation problem. GaussianEmoTalker first constructs an identity-specific neutral talking space with GaussianBlendshapes, which provides high-fidelity Gaussian attributes and phoneme-synchronized neutral motion. It then predicts an emotion-conditioned residual deformation by combining mesh displacement cues, audio features, emotion categories, and intensity encodings. To fuse these heterogeneous signals, we introduce a spatial-audio-emotion attention module that estimates the offsets of Gaussian attributes for expressive and temporally stable rendering. Extensive experiments demonstrate that GaussianEmoTalker achieves competitive video quality, accurate lip synchronization, controllable emotional expression, and real-time rendering compared with recent emotional talking head methods. Our project page is available at https://njust-yang.github.io/GaussianEmoTalker.github.io/
Chinese Translation
音频驱动的对话头合成在唇部同步和视觉质量方面取得了显著进展，但在实时约束下生成具有可控强度的富有表现力的情感化身仍然具有挑战性。本文提出了GaussianEmoTalker，一个基于3D高斯点云的音频驱动实时情感对话头合成框架。我们将情感动画形式化为中性到情感的残差变形问题，而不是直接从语音预测最终的情感化身。GaussianEmoTalker首先构建了一个特定身份的中性对话空间，利用GaussianBlendshapes提供高保真度的高斯属性和音素同步的中性运动。然后，通过结合网格位移线索、音频特征、情感类别和强度编码，预测情感条件的残差变形。为了融合这些异构信号，我们引入了一个空间-音频-情感注意模块，该模块估计高斯属性的偏移，以实现富有表现力和时间稳定的渲染。大量实验表明，与近期的情感对话头方法相比，GaussianEmoTalker在视频质量、唇部同步精度、可控情感表达和实时渲染方面均表现出竞争力。我们的项目页面可访问 https://njust-yang.github.io/GaussianEmoTalker.github.io/
cs.CV / 118 / 2607.00965
Slope-Guided Mamba and Angular-Refined Transformer for Light Field Super-Resolution
斜率引导的曼巴网络与角度优化变换器在光场超分辨率中的应用
Abstract
Light Field Super-Resolution (LFSR) necessitates accurate modeling of spatial-angular correlations while preserving intrinsic 4D ray coherence. However, maintaining such high-dimensional consistency remains challenging, primarily due to two inherent limitations in prevailing modeling paradigms. First, spatial and angular dimensions are often modeled in a decoupled manner, restricting early cross-dimensional interaction and leading to geometric inconsistencies. Moreover, although continuous sequence modeling paradigms show promise in representing epipolar structures, their rigid scanning mechanisms fundamentally conflict with epipolar geometry, limiting geometry-aware feature aggregation. To address these challenges, we propose a hybrid light field super-resolution network, termed SMART, which integrates a Slope-Guided Mamba and an Angular-Refined Transformer to effectively overcome these limitations. Specifically, we introduce an angular-modulated spatial module to bridge the decoupling gap, incorporating angular priors to strengthen spatial-angular correlation modeling. To mitigate the scan-geometry mismatch, we propose a manifold-aligned trajectory module that enables geometry-consistent sequence modeling along epipolar structures. Experiments on five benchmarks demonstrate that SMART achieves state-of-the-art performance, surpassing previous methods by 0.42 dB (PSNR) with significantly reduced artifacts.
Chinese Translation
光场超分辨率（LFSR）需要准确建模空间-角度相关性，同时保持内在的4D光线一致性。然而，维持如此高维的一致性仍然具有挑战性，主要是由于当前建模范式中的两个固有限制。首先，空间和角度维度通常以解耦的方式进行建模，这限制了早期的跨维度交互，并导致几何不一致。此外，尽管连续序列建模范式在表示极线结构方面显示出潜力，但其刚性的扫描机制与极线几何存在根本冲突，限制了几何感知特征的聚合。为了解决这些挑战，我们提出了一种混合光场超分辨率网络，称为SMART，该网络集成了斜率引导的曼巴网络和角度优化变换器，有效克服了这些限制。具体而言，我们引入了一个角度调制的空间模块，以弥补解耦的差距，结合角度先验以增强空间-角度相关性建模。为了减轻扫描几何不匹配的问题，我们提出了一个流形对齐轨迹模块，使得沿极线结构进行几何一致的序列建模成为可能。在五个基准测试上的实验表明，SMART实现了最先进的性能，超越了之前的方法0.42 dB（PSNR），并显著减少了伪影。
cs.CV / 119 / 2607.00975
TRCGL-Net: A Long-Tailed Multi-Label Chest X-Ray Classification Framework with Generative Data Augmentation and Label Co-Occurrence Modeling
TRCGL-Net：一种具有生成数据增强和标签共现建模的长尾多标签胸部X光分类框架
Abstract
Chest X-ray multi-label classification is a core task in intelligent medical imaging diagnosis. However, real clinical data often exhibit extreme long-tailed distributions, leading to degraded performance on rare diseases in tail classes. This issue is not only driven by data scarcity but also by two intrinsic factors:1) attenuation of tail-class lesion representations under complex anatomical backgrounds, and 2) dominance of head classes in modeling label co-occurrence relationships. To address these challenges, we propose TRCGL-Net. First, a learnable text-guided conditional diffusion model is employed to generate high-quality tail-class chest X-ray image samples under disease semantic constraints, improving data diversity and realism of rare disease patterns while alleviating class imbalance and preserving pathology-consistent semantics.Second, a channel reweighting mechanism is introduced to perform feature recalibration by emphasizing disease-relevant feature channels, thereby improving feature discriminability under long-tailed distributions.A class-aware attention mechanism is further applied to generate class-specific attention maps, enabling the model to localize disease-relevant regions and focus on fine-grained lesion areas.Finally, a graph convolution network based on label co occurrence is introduced to establish an information propagation mechanism among categories. Experiments on the PadChest dataset show that the proposed method achieves a tail-class mAP of 0.4904, an overall mAP of 0.4408, and an mAUC of 0.8989, outperforming state-of-the-art methods. TRCGL-Net effectively improves recognition performance for rare diseases under long-tailed distributions and mitigates the impact of extreme class imbalance in chest X-ray multi-label classification.
Chinese Translation
胸部X光多标签分类是智能医学影像诊断中的核心任务。然而，真实的临床数据往往表现出极端的长尾分布，导致在尾部类别中对罕见疾病的识别性能下降。这个问题不仅由数据稀缺引起，还受到两个内在因素的影响：1）在复杂解剖背景下，尾部类别病变表征的衰减，和2）在建模标签共现关系时头部类别的主导地位。为了解决这些挑战，我们提出了TRCGL-Net。首先，采用可学习的文本引导条件扩散模型，在疾病语义约束下生成高质量的尾部类别胸部X光图像样本，从而提高罕见疾病模式的数据多样性和真实性，同时缓解类别不平衡并保持病理一致的语义。其次，引入通道重加权机制，通过强调与疾病相关的特征通道来进行特征再校准，从而在长尾分布下提高特征的区分性。进一步应用类感知注意机制生成类特定的注意力图，使模型能够定位与疾病相关的区域并关注细粒度的病变区域。最后，引入基于标签共现的图卷积网络，建立类别之间的信息传播机制。在PadChest数据集上的实验表明，所提方法在尾部类别上的平均精度（mAP）达到0.4904，总体mAP为0.4408，平均曲线下面积（mAUC）为0.8989，优于最先进的方法。TRCGL-Net有效提高了在长尾分布下对罕见疾病的识别性能，并减轻了胸部X光多标签分类中极端类别不平衡的影响。
cs.CV / 120 / 2607.00978
Privacy-Preserving Depth-Only Open-Vocabulary 3D Semantic Segmentation Via Uncertainty-Guided Test-Time Optimization
基于不确定性引导的隐私保护深度仅开放词汇3D语义分割的测试时优化
Abstract
Privacy-preserving perception is a critical requirement for deploying 3D scene understanding systems in real-world indoor environments, yet it remains underexplored in open-vocabulary 3D semantic segmentation. Existing methods typically rely on obtaining rich semantic cues from RGB images, which may expose privacy-sensitive visual information. Depth-only 3D geometry provides a privacy-preserving alternative, but the absence of appearance-based semantic cues makes open-vocabulary predictions highly uncertain and less reliable. Under this setting, we propose to convert uncertainty into a guidance signal to identify unreliable semantic responses and use semantic priors from foundation models to regularize their refinement. We present UTTO, an uncertainty-guided test-time optimization framework for depth-only open-vocabulary 3D semantic segmentation. Without additional training, experiments on ScanNet20, ScanNet40, and ScanNet200 demonstrate that UTTO consistently improves depth-only open-vocabulary 3D segmentation and outperforms representative baselines under privacy-preserving conditions.
Chinese Translation
隐私保护感知是将3D场景理解系统部署于现实室内环境中的关键要求，但在开放词汇3D语义分割中仍然未得到充分探索。现有方法通常依赖于从RGB图像中获取丰富的语义线索，这可能暴露隐私敏感的视觉信息。仅深度的3D几何提供了一种隐私保护的替代方案，但缺乏基于外观的语义线索使得开放词汇预测高度不确定且不够可靠。在这种情况下，我们提出将不确定性转化为引导信号，以识别不可靠的语义响应，并利用基础模型的语义先验来规范其优化。我们提出了UTTO，一个用于深度仅开放词汇3D语义分割的不确定性引导测试时优化框架。在没有额外训练的情况下，针对ScanNet20、ScanNet40和ScanNet200的实验表明，UTTO在隐私保护条件下始终改善了深度仅开放词汇3D分割，并超越了代表性的基线。
cs.CV / 121 / 2607.00983
QCA: Query- and Content-Aware Keyframe Selection for Long Video Understanding
QCA：基于查询和内容的关键帧选择用于长视频理解
Abstract
Video understanding is often plagued by severe temporal redundancy, where processing dense frame sequences is both semantically inefficient and computationally expensive. This challenge is further amplified when only a small subset of frames is truly relevant to the given query. In this paper, we propose a Query- and Content-Aware (QCA) keyframe selection framework that can select a compact yet information-rich set of frames from long videos. QCA first partitions the video into temporal segments and estimates the information contribution of each segment by jointly modeling query relevance and content deviation, and dynamically allocates keyframe budget to each segment. Within each segment, QCA anchors on the most query-relevant frame and iteratively incorporates additional frames to maximize diversity while maintaining high semantic relevance to the query. Crucially, our method requires no additional training and can be seamlessly integrated into existing Video-LLMs. Extensive experiments across multiple long video understanding benchmarks demonstrate that our proposed approach achieves state-of-the-art performance and has strong generalization ability. For instance, QCA achieves 67.8\% on LongVideoBench using 128 frames, while GPT-4o achieves 66.7\% using 256 frames. Our codes are available in \href{https://github.com/hktk07/QCA}{GitHub}.
Chinese Translation
视频理解常常受到严重的时间冗余困扰，处理密集的帧序列在语义上既低效又计算成本高昂。当只有一小部分帧与给定查询真正相关时，这一挑战更加明显。本文提出了一种基于查询和内容的关键帧选择框架（Query- and Content-Aware, QCA），能够从长视频中选择一组紧凑且信息丰富的帧。QCA首先将视频划分为时间段，并通过联合建模查询相关性和内容偏差来估计每个时间段的信息贡献，并动态分配关键帧预算给每个段落。在每个段落内，QCA以最相关的查询帧为锚点，迭代地加入额外帧，以最大化多样性，同时保持与查询的高语义相关性。重要的是，我们的方法不需要额外的训练，并且可以无缝集成到现有的视频大语言模型（Video-LLMs）中。在多个长视频理解基准上的广泛实验表明，我们提出的方法达到了最先进的性能，并具有强大的泛化能力。例如，QCA在使用128帧时在LongVideoBench上达到了67.8\%，而GPT-4o在使用256帧时达到了66.7\%。我们的代码可在\href{https://github.com/hktk07/QCA}{GitHub}上获取。
cs.CV / 122 / 2607.00987
AVSR-Diff: Scale-Agnostic Diffusion Priors for Temporally Consistent Arbitrary-Scale Video Super-Resolution
AVSR-Diff：用于时间一致的任意尺度视频超分辨率的尺度无关扩散先验
Abstract
Diffusion models have significantly advanced video super-resolution (VSR) but remain largely constrained to fixed upsampling scales. Conversely, while coordinate-based arbitrary-scale VSR methods offer scale flexibility, they inherently suffer from severe over-smoothing at large scaling factors. Integrating generative priors with continuous decoding is promising but currently hindered by severe temporal flickering caused by the stochasticity of diffusion sampling. To address this, we propose AVSR-Diff (Arbitrary-scale Video Super-Resolution with Diffusion), a novel decoupled framework that separates scale-agnostic latent denoising from continuous coordinate rendering, effectively avoiding computationally heavy resolution-specific sampling. Our approach introduces a Temporally-Gated Feature Recurrence (TGFR) module to extract strictly aligned, temporally consistent latent priors. Furthermore, we design a continuous video VAE decoder incorporating a Scale-Aware Fourier Refinement (SAFR) module to dynamically adapt frequency components to any target scale. Extensive experiments demonstrate that AVSR-Diff consistently preserves high-frequency details and strong temporal stability across various scales, surpassing state-of-the-art arbitrary-scale baselines. Remarkably, our framework outperforms recent fixed-scale generative models even on their native resolution.
Chinese Translation
扩散模型在视频超分辨率（VSR）方面取得了显著进展，但仍然主要受到固定上采样尺度的限制。相反，基于坐标的任意尺度VSR方法虽然提供了尺度灵活性，但在大尺度因子下固有地面临严重的过度平滑问题。将生成先验与连续解码相结合是有前景的，但目前受到扩散采样随机性引起的严重时间闪烁的阻碍。为了解决这个问题，我们提出了AVSR-Diff（基于扩散的任意尺度视频超分辨率），这是一种新颖的解耦框架，将尺度无关的潜在去噪与连续坐标渲染分开，有效避免了计算上繁重的分辨率特定采样。我们的方法引入了一个时间门控特征递归（TGFR）模块，以提取严格对齐、时间一致的潜在先验。此外，我们设计了一个连续视频变分自编码器（VAE）解码器，结合了一个尺度感知傅里叶细化（SAFR）模块，以动态适应任何目标尺度的频率成分。大量实验表明，AVSR-Diff在各种尺度下始终保持高频细节和强时间稳定性，超越了最先进的任意尺度基准。值得注意的是，我们的框架在其原生分辨率下甚至超越了最近的固定尺度生成模型。
cs.CV / 123 / 2607.01001
Foundation Models vs. Radiomics for Lung Computed Tomography: A Benchmark of Feature Extractors, Classification Heads, and Segmentation Choices
基础模型与放射组学在肺部计算机断层扫描中的比较：特征提取器、分类头和分割选择的基准测试
Abstract
Radiomics is the established approach for CT-based lung cancer phenotyping, yet comparisons with foundation models rarely isolate contributions of feature extractor, classification head, and segmentation choice, or test cross-cohort robustness. We benchmark five feature extractors (Curia, Curia-2, DINOv3, Radiomics2D, Radiomics3D), seven classification heads (TabPFN, TabICL, XGBoost, CatBoost, Random Forest, logistic regression, Ridge), and three segmentation regimes on five tasks: tumor volume and stage classification, 2-year survival prediction, histology classification, and age prediction. Models are trained on LUNG1 (n=338) and evaluated on an internal test set (n=84) and the external LUNG2 cohort (n=211), with worst-case cross-cohort performance as the primary metric. The dominant design factor is task-dependent: segmentation drives volume and stage classification, while classifier choice drives survival, histology, and age prediction. Radiomics is competitive for tumor volume, tumor stage and survival (partly due to label-derivation effects for the former); Curia variants reach comparable peak scores for survival; DINOv3 falls slightly short across tasks. Patch and slice aggregation have negligible impact. We recommend Curia with tumor segmentation and a CatBoost head as a safe default, achieving the best mean rank across the three primary clinical tasks, though task-specific selection consistently outperforms any cross-task default. When tumor delineations are unavailable, Curia-2 with lung segmentation and logistic regression offers a competitive alternative. All pipelines use a two-stage design suited to small cohort sizes where end-to-end fine-tuning would risk overfitting.
Chinese Translation
放射组学是基于CT的肺癌表型分析的既定方法，但与基础模型的比较很少单独分析特征提取器、分类头和分割选择的贡献，或测试跨队列的稳健性。我们对五种特征提取器（Curia、Curia-2、DINOv3、Radiomics2D、Radiomics3D）、七种分类头（TabPFN、TabICL、XGBoost、CatBoost、随机森林、逻辑回归、岭回归）和三种分割方案在五个任务上进行了基准测试：肿瘤体积和分期分类、2年生存预测、组织学分类和年龄预测。模型在LUNG1（n=338）上训练，并在内部测试集（n=84）和外部LUNG2队列（n=211）上进行评估，以最差的跨队列性能作为主要指标。主导设计因素依赖于任务：分割驱动体积和分期分类，而分类器选择驱动生存、组织学和年龄预测。放射组学在肿瘤体积、肿瘤分期和生存预测方面具有竞争力（部分由于前者的标签衍生效应）；Curia变体在生存预测方面达到可比的峰值分数；DINOv3在各项任务中略显不足。补丁和切片聚合的影响微乎其微。我们建议使用Curia结合肿瘤分割和CatBoost分类头作为安全的默认选择，在三个主要临床任务中实现最佳平均排名，尽管任务特定选择始终优于任何跨任务默认选择。当肿瘤轮廓不可用时，Curia-2结合肺部分割和逻辑回归提供了一个有竞争力的替代方案。所有管道采用两阶段设计，适合小规模队列，其中端到端的微调可能导致过拟合。
cs.CV / 124 / 2607.01015
SuperFlex: Deformable Superquadrics for Point Cloud Decomposition
SuperFlex：用于点云分解的可变形超二次体
Abstract
Superquadrics have proven to provide a compact, geometrically meaningful representation for 3D objects. However, existing methods suffer from limited reconstruction accuracy, are restricted to rigid primitives, and lack robustness to partial point clouds. In this work, we present SuperFlex, an enhanced framework that expands the expressive power and applicability of superquadric decompositions. First, we introduce a novel loss formulation which significantly improves reconstruction accuracy. Second, we include bending and tapering deformations, enabling high-fidelity representation of curved and asymmetric geometries. Finally, we leverage these high-quality decompositions as supervision to train a model that is robust to partial real-world point clouds. Experiments demonstrate substantial improvements in reconstruction accuracy over both optimization- and learning-based baselines while maintaining a highly compact primitive representation.
Chinese Translation
超二次体已被证明能够为三维物体提供紧凑且具有几何意义的表示。然而，现有方法在重建精度上存在局限，受限于刚性原语，并且对部分点云缺乏鲁棒性。在本研究中，我们提出了SuperFlex，一个增强框架，扩展了超二次体分解的表现力和适用性。首先，我们引入了一种新颖的损失公式，显著提高了重建精度。其次，我们加入了弯曲和锥形变形，使得能够高保真地表示曲线和不对称几何形状。最后，我们利用这些高质量的分解作为监督，训练一个对部分真实世界点云具有鲁棒性的模型。实验表明，在保持高度紧凑的原语表示的同时，重建精度相比于基于优化和学习的基线有了显著提升。
cs.CV / 125 / 2607.01039
EchoRisk: A Multicentre Echocardiography Dataset and Benchmark for Cardio-Oncology
EchoRisk：心脏肿瘤学的多中心超声心动图数据集及基准
Abstract
Therapy-induced cardiotoxicity is the leading non-oncological cause of treatment interruption in breast cancer patients, yet early, automated risk stratification from routine cardiac imaging remains an unsolved problem. We present EchoRisk, the first curated, multicentre, longitudinal echocardiography dataset with explicit cardiotoxicity labels, released as the primary technical reference for the EchoRisk-MICCAI 2026 challenge. The dataset comprises 422 patients enrolled in the EU-funded CARDIOCARE prospective study across five European sites, yielding 2,159 echocardiography videos across 1,123 clinical exams acquired at up to five longitudinal timepoints, alongside a dedicated cohort of 280 patients with baseline imaging for early cardiotoxicity prediction. Three clinically grounded tasks are defined: automated estimation of left ventricular ejection fraction from cine video (Task 1), classification of LV dysfunction from longitudinal imaging (Task 2), and early prediction of therapy-induced cardiotoxicity from pre-therapy baseline echocardiography alone (Task 3). For each task we specify the evaluation protocol, primary and secondary metrics, and ranking procedure. We establish baseline performance using an R(2+1)D video backbone with LSTM aggregation trained from Kinetics-400 pretrained weights, demonstrating strong discriminative performance for cardiac functional assessment and LV dysfunction classification, while early cardiotoxicity prediction from a single pre-therapy video remains a significant open problem for the community. The dataset, evaluation code, and baseline implementations are publicly available to serve as a benchmark for further collaboration, comparison, and the creation of task-specific architectures in cardio-oncology.
Chinese Translation
治疗引起的心脏毒性是乳腺癌患者治疗中断的主要非肿瘤原因，但从常规心脏影像中进行早期自动化风险分层仍然是一个未解决的问题。我们提出了EchoRisk，这是第一个经过精心策划的多中心纵向超声心动图数据集，带有明确的心脏毒性标签，作为EchoRisk-MICCAI 2026挑战的主要技术参考。该数据集包括422名参与欧盟资助的CARDIOCARE前瞻性研究的患者，覆盖五个欧洲站点，共产生2,159个超声心动图视频，涵盖1,123次临床检查，采集于多达五个纵向时间点，同时还有一组专门的280名患者的基线影像，用于早期心脏毒性预测。定义了三个临床基础任务：从动态视频中自动估计左心室射血分数（任务1）、从纵向影像中分类左心室功能障碍（任务2），以及仅从治疗前基线超声心动图中早期预测治疗引起的心脏毒性（任务3）。对于每个任务，我们指定了评估协议、主要和次要指标以及排名程序。我们使用R(2+1)D视频骨干网络与LSTM聚合进行基线性能建立，训练时使用Kinetics-400预训练权重，展示了在心脏功能评估和左心室功能障碍分类方面的强大区分性能，而仅从单个治疗前视频进行的早期心脏毒性预测仍然是社区面临的重大开放问题。该数据集、评估代码和基线实现均已公开，以作为进一步合作、比较和创建心脏肿瘤学特定任务架构的基准。
cs.CV / 126 / 2607.01049
GenAU: Language-Grounded Industrial Anomaly Understanding with Vision-Language Models
GenAU：基于语言的工业异常理解与视觉-语言模型
Abstract
Industrial inspection requires more than binary anomaly detection: a practical system should determine whether an anomaly exists, localize the defective region, identify the defect type, and provide interpretable visual evidence. Existing CLIP-based methods detect and localize anomalies well but offer limited language-level defect understanding, while instruction-tuned vision-language models can describe defects but do not natively produce pixel-level masks. We introduce GenAU, a Generalist vision-language framework for industrial Anomaly Understanding that unifies image-level detection, pixel-level segmentation, multi-type anomaly detection, and defect analysis in a single instruction-following model. GenAU augments a vision-language model with two segmentation tokens, [SEG_defect] and [SEG_normal], whose hidden states act as language-grounded queries over multi-scale visual features for pixel-level localization; the image-level score fuses this map with the decoder's textual normal/defect decision, while the language decoder produces structured defect-aware responses. Trained with a joint language-modeling and segmentation objective, GenAU covers all four tasks within one architecture and recipe, adding zero-shot multi-type detection and language-grounded defect analysis at a quantified cost to detection and segmentation. Across cross-dataset benchmarks, GenAU attains the strongest image-level detection among CLIP-based zero-shot methods on VisA and Real-IAD, with segmentation approaching but not surpassing specialized CLIP baselines.
Chinese Translation
工业检测不仅仅需要二元异常检测：一个实用的系统应能够判断异常是否存在、定位缺陷区域、识别缺陷类型，并提供可解释的视觉证据。现有的基于CLIP的方法在检测和定位异常方面表现良好，但在语言层面的缺陷理解上有限，而经过指令调优的视觉-语言模型可以描述缺陷，但无法原生生成像素级的掩码。我们提出了GenAU，一个用于工业异常理解的通用视觉-语言框架，它在单一的遵循指令的模型中统一了图像级检测、像素级分割、多类型异常检测和缺陷分析。GenAU通过两个分割标记[SEG_defect]和[SEG_normal]增强了视觉-语言模型，其隐藏状态作为多尺度视觉特征的基于语言的查询，用于像素级定位；图像级得分将此图与解码器的文本正常/缺陷决策融合，而语言解码器则生成结构化的缺陷感知响应。GenAU在一个架构和配方中涵盖了所有四个任务，并以联合语言建模和分割目标进行训练，在检测和分割的量化成本下增加了零-shot多类型检测和基于语言的缺陷分析。在跨数据集基准测试中，GenAU在VisA和Real-IAD上实现了CLIP基于零-shot方法中最强的图像级检测，分割接近但未超过专门的CLIP基线。
cs.CV / 127 / 2607.01050
GeoSearcher: Anchor-Guided Progressive Reasoning for Remote Sensing Visual Grounding with Process Supervision
GeoSearcher：基于锚点引导的渐进推理用于遥感视觉定位与过程监督
Abstract
Recent multimodal large language models (MLLMs) have shown strong cross-modal understanding and coordinate generation abilities in visual grounding. However, transferring these abilities to remote sensing visual grounding (RSVG) remains challenging. High-resolution remote sensing images usually cover large-scale scenes, where targets are often extremely small and surrounded by numerous visually similar distractors. Meanwhile, queries often contain multiple clues, such as reference objects, spatial relations, and target attributes. Existing MLLM-based methods usually formulate RSVG as one-step coordinate generation, which may lead to unstable predictions for small-object localization and complex queries. To address these challenges, we propose GeoSearcher, which reformulates RSVG as an anchor-guided progressive reasoning process and realizes it through two coupled stages: Anchor-Centric Reasoning Supervised Fine-Tuning (ACR-SFT) and Process-Faithful Group Relative Policy Optimization (PF-GRPO). In ACR-SFT, anchor-centric reasoning data are used to teach the model to represent key visual clues as anchors and progressively integrate location, relational, and attribute clues around them. In PF-GRPO, Process-Aware Reward (PAR) and Reasoning-Informative Sample Selector (RISS) further optimize this reasoning behavior by jointly evaluating key reasoning steps and target localization, while focusing training on samples that are more beneficial for improving progressive reasoning. Through this design, GeoSearcher transforms large-scale visual search into a more constrained local reasoning process. Extensive experiments on DIOR-RSVG, OPT-RSVG, and VRS-Bench show that GeoSearcher outperforms existing state-of-the-art methods. The project will be released at https://github.com/wangdianyu954-xixi/GeoSearcher.
Chinese Translation
近期的多模态大型语言模型（MLLMs）在视觉定位中展示了强大的跨模态理解和坐标生成能力。然而，将这些能力转移到遥感视觉定位（RSVG）上仍然面临挑战。高分辨率的遥感图像通常覆盖大规模场景，其中目标往往非常小，并被众多视觉上相似的干扰物所包围。同时，查询通常包含多个线索，如参考对象、空间关系和目标属性。现有的基于MLLM的方法通常将RSVG表述为一步坐标生成，这可能导致小物体定位和复杂查询的预测不稳定。为了解决这些挑战，我们提出了GeoSearcher，它将RSVG重新表述为一个基于锚点引导的渐进推理过程，并通过两个耦合阶段实现：以锚点为中心的推理监督微调（ACR-SFT）和过程忠实的组相对策略优化（PF-GRPO）。在ACR-SFT中，以锚点为中心的推理数据用于教导模型将关键视觉线索表示为锚点，并逐步整合围绕它们的位置、关系和属性线索。在PF-GRPO中，过程感知奖励（PAR）和推理信息样本选择器（RISS）通过共同评估关键推理步骤和目标定位进一步优化这一推理行为，同时将训练重点放在那些更有利于改善渐进推理的样本上。通过这种设计，GeoSearcher将大规模视觉搜索转变为一个更受限的局部推理过程。在DIOR-RSVG、OPT-RSVG和VRS-Bench上的大量实验表明，GeoSearcher的表现优于现有的最先进方法。该项目将发布在 https://github.com/wangdianyu954-xixi/GeoSearcher。
cs.CV / 128 / 2607.01086
LongVQUBench: Benchmarking Long-Term Video Quality Understanding of Vision-Language Models
LongVQUBench：视觉语言模型长期视频质量理解的基准测试
Abstract
The evaluation of long-term video quality understanding remains an open challenge for large vision-language models (LVLMs). Existing video quality benchmarks predominantly focus on short clips and isolated distortions, overlooking the temporal continuity, cumulative degradation, and reasoning complexity inherent in long-duration content. To address these limitations, we present LongVQUBench, a comprehensive benchmark for long-term video quality understanding. LongVQUBench contains over 1200 diverse videos spanning movies, documentaries, surveillance footage, egocentric recordings, and animated content, accompanied by 1500 multiple-choice and open-ended questions for validation and testing. To assess perceptual reasoning across different temporal scopes, we introduce three progressively complex evaluation levels: (i) local event quality understanding (LQU) for analyzing localized distortions; (ii) cross-event quality reasoning (CQR) for integrating multiple degraded events; and (iii) global quality understanding (GQU) for holistic perceptual evaluation over extended durations. Furthermore, a needle distortion question-answering (NDQA) paradigm is embedded across all three levels, where spatial or temporal artifacts are sparsely inserted to probe fine-grained detection and reasoning capabilities. Extensive experiments on 14 state-of-the-art LVLMs reveal significant performance degradation with increasing video length and reasoning depth, highlighting their limited capacity for long-range temporal integration and perceptual attribution. We envision LongVQUBench as a foundational step toward the systematic, hierarchical, and explainable evaluation of LVLMs' long-term video quality understanding.
Chinese Translation
长期视频质量理解的评估仍然是大型视觉语言模型（LVLMs）面临的一个开放挑战。现有的视频质量基准主要集中在短片段和孤立失真上，忽视了长期内容中固有的时间连续性、累积退化和推理复杂性。为了解决这些局限性，我们提出了LongVQUBench，这是一个全面的长期视频质量理解基准。LongVQUBench包含超过1200个多样化的视频，涵盖电影、纪录片、监控录像、自我中心录音和动画内容，并配有1500个多项选择题和开放式问题用于验证和测试。为了评估不同时间范围内的感知推理，我们引入了三个逐渐复杂的评估层次：（i）局部事件质量理解（LQU），用于分析局部失真；（ii）跨事件质量推理（CQR），用于整合多个退化事件；以及（iii）全局质量理解（GQU），用于对延长时间内的整体感知评估。此外，在所有三个层次中嵌入了针状失真问答（NDQA）范式，其中稀疏插入空间或时间伪影，以探测细粒度的检测和推理能力。对14个最先进的LVLMs进行的广泛实验表明，随着视频长度和推理深度的增加，性能显著下降，突显了它们在长期时间整合和感知归因方面的有限能力。我们设想LongVQUBench作为系统性、分层和可解释的LVLMs长期视频质量理解评估的基础性步骤。
cs.CV / 129 / 2607.01100
CPDDNet: Color-Polarization Denoising and Demosaicking Network
CPDDNet：彩色-偏振去噪与去马赛克网络
Abstract
Color-polarization imaging using a color-polarization filter array (CPFA) sensor captures both texture (color intensity) and physical (polarization) information of the scene in a single shot, enabling various applications in computer vision. However, the raw mosaic output from a CPFA sensor often suffers from severe noise and resolution loss, especially under low-light conditions. Existing methods generally focus on either denoising or demosaicking tasks, failing to capture the coupling between them and neglecting shared low-level features. In this paper, we propose a color-polarization denoising and demosaicking network (CPDDNet), which is a joint framework that performs noise removal and CPFA interpolation using a feature fusion module that retains the features from the CPFA raw data at both the denoising and the demosaicking stages. Experimental results demonstrate that CPDDNet significantly enhances image quality and polarization parameter accuracy, outperforming existing approaches on a real dataset.
Chinese Translation
使用彩色-偏振滤光阵列（CPFA）传感器的彩色-偏振成像能够在单次拍摄中捕捉场景的纹理（颜色强度）和物理（偏振）信息，从而在计算机视觉领域实现多种应用。然而，来自CPFA传感器的原始马赛克输出往往受到严重噪声和分辨率损失的影响，尤其是在低光照条件下。现有方法通常专注于去噪或去马赛克任务，未能捕捉两者之间的耦合关系，并忽视了共享的低级特征。本文提出了一种彩色-偏振去噪与去马赛克网络（CPDDNet），这是一个联合框架，通过特征融合模块在去噪和去马赛克阶段保留来自CPFA原始数据的特征，执行噪声去除和CPFA插值。实验结果表明，CPDDNet显著提高了图像质量和偏振参数的准确性，在真实数据集上优于现有方法。
cs.CV / 130 / 2607.01117
MoHallBench: A Benchmark for Motion Hallucination in Video Large Language Models
MoHallBench：视频大型语言模型中的运动幻觉基准
Abstract
Video Large Language Models (VideoLLMs) have shown strong progress in video understanding, yet they still suffer from hallucinations that are inconsistent with visual evidence. Existing benchmarks mainly focus on object hallucination or coarse action perception, leaving a key video-specific problem underexplored: motion hallucination, in which models infer human motions that are absent from the video. We present MoHallBench, a benchmark for diagnosing motion hallucination in VideoLLMs. MoHallBench systematically evaluates three major sources of hallucination: co-occurrence priors, sequential inference, and similarity confusion. It contains 11,306 video clips and 40,493 question-answer pairs, covering binary-choice, multiple-choice, and generative settings. We further introduce a bi-directional questioning protocol with bias-aware metrics to reduce affirmation bias in binary evaluation. Experiments on ten recent open-source VideoLLMs reveal a clear decoupling between action recognition and hallucination resistance, as models that perform well on positive action recognition often fail on adversarial negatives. Among all settings, sequential inference hallucination is the most severe, showing that current models tend to over-infer expected outcomes from partial motion cues. Our analyses further confirm that stronger priors and finer-grained similarity substantially amplify hallucination. We hope MoHallBench can facilitate future evaluation and mitigation of motion hallucination in VideoLLMs.
Chinese Translation
视频大型语言模型（VideoLLMs）在视频理解方面取得了显著进展，但仍然面临与视觉证据不一致的幻觉问题。现有基准主要集中在物体幻觉或粗略的动作感知上，未能充分探讨一个关键的视频特定问题：运动幻觉，即模型推断视频中不存在的人类动作。我们提出了MoHallBench，这是一个用于诊断VideoLLMs中运动幻觉的基准。MoHallBench系统地评估了三种主要的幻觉来源：共现先验、序列推理和相似性混淆。它包含11,306个视频片段和40,493对问答，涵盖了二元选择、多元选择和生成设置。我们进一步引入了一种双向提问协议，并采用偏见感知指标，以减少二元评估中的肯定偏见。在对十个近期开源VideoLLMs的实验中，结果显示动作识别与幻觉抵抗之间存在明显的解耦，因为在积极的动作识别上表现良好的模型往往在对抗性负样本上表现不佳。在所有设置中，序列推理幻觉最为严重，表明当前模型倾向于从部分运动线索中过度推断预期结果。我们的分析进一步确认，较强的先验和更细粒度的相似性显著增强了幻觉。我们希望MoHallBench能够促进未来对VideoLLMs中运动幻觉的评估和缓解。
cs.CV / 131 / 2607.01131
Autonomous Scientific Discovery via Iterative Meta-Reflection
通过迭代元反思实现自主科学发现
Abstract
Autonomous scientific discovery systems offer the potential to accelerate research by automating the process of hypothesis generation and validation. However, current systems operate within constrained search spaces or require predefined research questions, limiting their capacity for true open-ended inquiry. Furthermore, while they generate hypotheses iteratively, they largely lack the ability to explicitly synthesize their own accumulated findings to uncover complex, interconnected phenomena. We introduce DiscoPER, an autonomous large language model-powered framework that conducts open-ended research by dynamically generating and executing code to explore datasets without pre-specified research objectives. To ensure rigorous scientific validity, every proposed discovery must pass statistical testing. To overcome the limitations of isolated search, our framework introduces a second-order reasoning mechanism that periodically analyzes its own accumulated discoveries. By treating prior discoveries as empirical data, DiscoPER identifies structural patterns, confounds, and epistemic gaps, actively redirecting hypothesis exploration toward uncharted regions of the search space. The search space is further expanded by incorporating tool use, enabling the system to explore hypotheses beyond structured metadata by seamlessly processing and extracting useful information from multimodal sources like images. Evaluated on iNatDisco, a new multimodal ecological knowledge benchmark with pattern-level ground truth obtained from peer-reviewed literature, DiscoPER recovers 8 of 9 known patterns with a 72.7% hypothesis support rate, outperforming both classical causal discovery and LLM-guided baselines. Ablations show that DiscoPER scales with more data, and confirms the benefits of second-order meta-reflection.
Chinese Translation
自主科学发现系统有潜力通过自动化假设生成和验证的过程来加速研究。然而，当前的系统通常在受限的搜索空间内运行，或需要预定义的研究问题，这限制了它们进行真正开放式探究的能力。此外，尽管它们以迭代方式生成假设，但在明确综合自身积累的发现以揭示复杂的相互关联现象方面仍然缺乏能力。我们提出了DiscoPER，一个由大型语言模型驱动的自主框架，通过动态生成和执行代码来探索数据集，进行开放式研究，而无需预先指定研究目标。为了确保严格的科学有效性，所有提出的发现必须通过统计测试。为了克服孤立搜索的局限性，我们的框架引入了一种二阶推理机制，定期分析自身的积累发现。通过将先前的发现视为实证数据，DiscoPER识别结构模式、混淆因素和认识空白，积极将假设探索引导至搜索空间中未被探索的区域。通过结合工具使用，进一步扩展了搜索空间，使系统能够通过无缝处理和提取来自图像等多模态来源的有用信息，探索超越结构化元数据的假设。在iNatDisco上进行评估，这是一个新的多模态生态知识基准，具有从同行评审文献中获得的模式级真实数据，DiscoPER以72.7%的假设支持率恢复了9个已知模式中的8个，优于经典因果发现和大型语言模型引导的基线。消融实验表明，DiscoPER在数据量增加时表现出良好的扩展性，并确认了二阶元反思的益处。
cs.CV / 132 / 2607.01133
Towards Metric-Agnostic Trajectory Forecasting
面向度量无关的轨迹预测
Abstract
Accurate trajectory forecasting of surrounding traffic participants is a core capability for autonomous driving, enabling vehicles to anticipate behavior and plan safe maneuvers. We observe that current state-of-the-art forecasting models on Argoverse 2 and the Waymo Open Motion Dataset tailor their training objectives to the different benchmark metrics. Because these metrics encourage conflicting behavior, we propose a paradigm change for trajectory forecasting: training models with metric-agnostic probabilistic objectives and treating metric optimization as a downstream task applied to the predictive distribution. Concretely, we introduce Trajectory Distribution Evaluation (TraDiE) policies, metric-specific policies that map a predictive distribution to the set of $K$ trajectories and confidences required by trajectory forecasting metrics. We evaluate this framework by introducing DONUT-NLL, which adapts the training objective of the state-of-the-art trajectory forecasting model DONUT to directly optimize the predictive distribution. Using our policies, DONUT-NLL achieves state-of-the-art results on all metrics of the Waymo motion prediction benchmark.
Chinese Translation
周围交通参与者的准确轨迹预测是自动驾驶的核心能力，使得车辆能够预测行为并规划安全的操作。我们观察到，目前在Argoverse 2和Waymo开放运动数据集上的最先进预测模型将其训练目标调整为不同的基准度量。由于这些度量鼓励相互矛盾的行为，我们提出了一种轨迹预测的范式转变：使用度量无关的概率目标训练模型，并将度量优化视为应用于预测分布的下游任务。具体而言，我们引入了轨迹分布评估（Trajectory Distribution Evaluation, TraDiE）策略，这是一种度量特定的策略，将预测分布映射到轨迹预测度量所需的$K$条轨迹及其置信度集合。我们通过引入DONUT-NLL来评估该框架，该模型将最先进的轨迹预测模型DONUT的训练目标调整为直接优化预测分布。使用我们的策略，DONUT-NLL在Waymo运动预测基准的所有度量上达到了最先进的结果。
cs.CV / 133 / 2607.01139
SD-RouteFusion: Ego-Trajectory Prediction with SD-Map Route Conditioning
SD-RouteFusion：基于SD地图路线条件的自我轨迹预测
Abstract
This paper presents SD-RouteFusion, a deployable end-to-end ego-trajectory prediction method that fuses a front-facing camera, vehicle kinematics, and a navigation route derived from a Standard Definition (SD) map. Unlike approaches that rely on High Definition (HD) map geometry, SD-RouteFusion aligns the learning objective with scalable and production-ready SD-map route inputs, enabling route-aware prediction without requiring HD-map infrastructure. First, we demonstrate that SD-map route prior provides a powerful long-horizon semantic prior. Through a comprehensive study on a large-scale real-world dataset comprising 480k driving scenarios across 10 European countries and the U.S., we quantify the value of SD-route conditioning: incorporating SD-map routes yields a 10.5% ADE improvement over an image-and-kinematics baseline, while our full fusion strategy achieves a 16.9% ADE reduction given a prediction horizon of 8 seconds. The fusion strategy consists of a dual-hypothesis design paired with a gated classifier, to ensure robustness under route corruption and visual uncertainty. Finally, to support broader evaluation, we release an SD-route generation toolkit that enables SD-route-conditioned ego-trajectory prediction on all datasets containing ego pose and future trajectories. Together, SD-RouteFusion establishes a practical path toward robust, route-aware ego-trajectory prediction at scale.
Chinese Translation
本文提出了SD-RouteFusion，一种可部署的端到端自我轨迹预测方法，该方法融合了前向摄像头、车辆运动学和来自标准定义（SD）地图的导航路线。与依赖高清（HD）地图几何信息的方法不同，SD-RouteFusion将学习目标与可扩展且适合生产的SD地图路线输入对齐，使得在不需要HD地图基础设施的情况下实现基于路线的预测。首先，我们展示了SD地图路线先验提供了强大的长时域语义先验。通过对包含来自10个欧洲国家和美国的48万个驾驶场景的大规模真实世界数据集进行全面研究，我们量化了SD路线条件的价值：引入SD地图路线相比于基于图像和运动学的基线，带来了10.5%的平均定位误差（ADE）改善，而我们的完整融合策略在8秒的预测时域下实现了16.9%的ADE减少。该融合策略由双假设设计和门控分类器组成，以确保在路线损坏和视觉不确定性下的鲁棒性。最后，为了支持更广泛的评估，我们发布了一个SD路线生成工具包，使得在所有包含自我姿态和未来轨迹的数据集中能够进行SD路线条件的自我轨迹预测。总之，SD-RouteFusion为在大规模下实现鲁棒的、基于路线的自我轨迹预测奠定了实用的基础。
cs.CV / 134 / 2607.01140
Relation-Centric Open-Vocabulary 3D Gaussian Segmentation
基于关系的开放词汇3D高斯分割
Abstract
Open-vocabulary 3D Gaussian segmentation is challenging because it requires language understanding for diverse queries and accurate separation of Gaussians along object boundaries. Prior approaches either embed language knowledge into individual Gaussians to improve query responsiveness or optimize per-Gaussian instance features to encode object identity. However, these strategies may produce noisy Gaussian segmentations or rely on cost-inefficient per-scene optimization. We propose PairGS, a framework that reframes Gaussian segmentation as modeling pairwise relations between Gaussians. 3D Gaussian representations provide rich signals for relation estimation, such as view contribution weights and multi-view mask evidence. By leveraging these cues, PairGS explicitly constructs a relation graph for segmentation without a heavy optimization process. PairGS first proposes sparse edge candidates using low-dimensional descriptors, computes precise pairwise affinities only on those candidates, and builds a hierarchical cluster tree for multi-granular querying. It achieves state-of-the-art results on open-vocabulary 3D Gaussian segmentation benchmarks, while the fast variant is 50x faster than optimization-based instance-feature approaches.
Chinese Translation
开放词汇3D高斯分割具有挑战性，因为它需要理解多样化查询的语言能力，并在物体边界上准确分离高斯分布。以往的方法要么将语言知识嵌入到单个高斯中以提高查询响应能力，要么优化每个高斯实例特征以编码物体身份。然而，这些策略可能会产生噪声高斯分割，或依赖于成本低效的每场景优化。我们提出了PairGS，一个将高斯分割重新定义为建模高斯之间成对关系的框架。3D高斯表示提供了丰富的关系估计信号，例如视图贡献权重和多视图掩码证据。通过利用这些线索，PairGS明确构建了一个用于分割的关系图，而无需繁重的优化过程。PairGS首先使用低维描述符提出稀疏边缘候选，随后仅在这些候选上计算精确的成对亲和度，并建立一个用于多粒度查询的层次聚类树。它在开放词汇3D高斯分割基准测试中达到了最先进的结果，同时其快速变体比基于优化的实例特征方法快50倍。
cs.CV / 135 / 2607.01147
EquiSteer: Cross-Attention Steering Towards a Fairer Text-Guided Image Generation
EquiSteer：跨注意力引导实现更公平的文本引导图像生成
Abstract
Text-to-image diffusion models power everyday creative tasks, but they still reproduce the demographic biases in their training data. On common prompts such as ``a photo of a nurse,'' ``a photo of a CEO'', they skew their outputs toward one gender, driven by the statistics of training data rather than anything in the text. Existing debiasing methods show promise in narrow settings but require retraining, batch-level control, or prompt-specific tuning, limiting their scalability. We propose \emph{EquiSteer}, a training-free method that works per sample by steering cross-attention (CA) activations at inference time. For each target attribute, EquiSteer precomputes steering vectors from contrastive prompts. Then at generation time, a prompt-aware gate leaves attribute-specific prompts untouched, while for neutral ones it clears existing attribute signals from the CA activations and injects a target attribute. Across SD-1.5, SD-2.1, SDXL, and SANA, EquiSteer reduces the average parity gap by up to $87\%$, with minimal effect on image quality and text-image alignment. Code is available at \href{https://github.com/Atmyre/EquiSteer}{https://github.com/Atmyre/EquiSteer}.%
Chinese Translation
文本到图像的扩散模型在日常创作任务中发挥着重要作用，但它们仍然会重现训练数据中的人口统计偏见。在常见的提示如“护士的照片”、“首席执行官的照片”中，它们的输出倾向于某一性别，这种偏向是由训练数据的统计特征驱动的，而非文本内容。现有的去偏见方法在狭窄的设置中显示出潜力，但需要重新训练、批量级控制或特定提示调优，限制了其可扩展性。我们提出了EquiSteer，这是一种无训练的方法，通过在推理时引导跨注意力（CA）激活来逐样本工作。对于每个目标属性，EquiSteer从对比提示中预计算引导向量。在生成时，提示感知门对特定属性的提示保持不变，而对中性提示则清除CA激活中的现有属性信号，并注入目标属性。在SD-1.5、SD-2.1、SDXL和SANA上，EquiSteer将平均平衡差距减少了高达87%，对图像质量和文本-图像对齐的影响最小。代码可在此获取：https://github.com/Atmyre/EquiSteer。
cs.CV / 136 / 2607.01176
High-dimensional Embedding Prior for Noisy K-space Domain MRIReconstruction
高维嵌入先验用于噪声K空间域MRI重建
Abstract
Magnetic resonance imaging (MRI) reconstruction under realistic acquisition conditions can be fundamentally viewed as estimating the underlying k-space distribution from incomplete and noise-corrupted measurements. While diffusion models have recently shown strong potential as generative prior for inverse problems,existingapproachesstruggletohandlenoisyreconstruction settings, especially when operating directly in k-space domain. In this work, we propose a unified high-dimensional k-space reconstruction framework tailored for noisy inverse problems, whichenhancesdiffusion-based solversthroughrepresentation lifting.Ratherthanmodifyingthe underlying optimization procedures, the proposed framework augments the data representation space, enabling existing diffusion-based solvers to operate on enriched k-space embeddings with improved expressiveness. Extensive experiments on both in-house and public datasets across varying noise levels and undersampled factors demonstrate that the proposed frame work consistently improves reconstruction quality for multiple diffusion-based inverse solvers. Notably, the largest gains are observed in high-noise regimes, which is consistent with our theoretical analysis of error propagation under high-dimensional representation. These results suggest that high-dimensional representation provides a general and model-agnostic mechanism for improving diffusion-based MRI reconstruction in noisy settings, offering a new perspective on robust k-space generative modeling for practical inverse problems. The code will be available at https://github.com/yqx7150/HEP-MRIRec.
Chinese Translation
在现实采集条件下，磁共振成像（MRI）重建可以从根本上视为从不完整且受噪声干扰的测量中估计潜在的k空间分布。尽管扩散模型最近在逆问题的生成先验中显示出强大的潜力，但现有方法在处理噪声重建设置时面临挑战，尤其是在直接在k空间域操作时。在本研究中，我们提出了一种统一的高维k空间重建框架，专门针对噪声逆问题，通过表示提升增强基于扩散的求解器。该框架并未修改基础优化过程，而是扩展了数据表示空间，使现有的基于扩散的求解器能够在丰富的k空间嵌入上运行，从而提高表达能力。在不同噪声水平和欠采样因子的多组实验中，无论是在内部数据集还是公共数据集上，均证明了该框架在多个基于扩散的逆求解器中持续提高了重建质量。值得注意的是，在高噪声环境下观察到的增益最大，这与我们对高维表示下误差传播的理论分析一致。这些结果表明，高维表示为在噪声环境中改善基于扩散的MRI重建提供了一种通用且与模型无关的机制，为实际逆问题的稳健k空间生成建模提供了新的视角。代码将可在 https://github.com/yqx7150/HEP-MRIRec 获取。
cs.CV / 137 / 2607.01191
Perceive-to-Reason: Decoupling Perception and Reasoning for Fine-Grained Visual Reasoning
感知-推理：解耦感知与推理以实现细粒度视觉推理
Abstract
Fine-grained visual reasoning remains challenging for vision-language models, especially when small but critical visual cues are buried in high-resolution images. Existing approaches rely on repeated cropping or test-time visual search to introduce local evidence, but they typically do not explicitly distinguish perception from reasoning. In this paper, we propose Perceive-to-Reason (P2R), a unified framework that formulates fine-grained visual reasoning as a two-stage process: the model first localizes question-relevant evidence as a Perceiver, and then answers the question as a Reasoner based on the annotated image and cropped regions. To better align training with this decoupled formulation, we further introduce Perception-Reasoning Alternating GRPO (PRA-GRPO), a role-aware reinforcement learning strategy that alternates between perception-focused and reasoning-focused updates using only final-answer supervision. Built on top of Qwen3-VL-Instruct-2B/4B/8B, P2R consistently improves performance across model scales. In particular, P2R-4B achieves 93.2% on V-Star, 81.9% on HR-Bench-4K, and 80.5% on HR-Bench-8K, substantially outperforming its corresponding backbone. Further experiments show that the benefits of P2R extend beyond high-resolution benchmarks to broader multimodal reasoning tasks. These results suggest that explicitly decoupling perception from reasoning provides an effective framework for fine-grained visual reasoning.
Chinese Translation
细粒度视觉推理对于视觉-语言模型仍然具有挑战性，尤其是在小但关键的视觉线索被埋藏在高分辨率图像中时。现有方法依赖于重复裁剪或测试时视觉搜索来引入局部证据，但通常并未明确区分感知与推理。在本文中，我们提出了感知-推理（Perceive-to-Reason, P2R），一个将细粒度视觉推理形式化为两阶段过程的统一框架：模型首先作为感知者定位与问题相关的证据，然后基于标注图像和裁剪区域作为推理者回答问题。为了更好地将训练与这种解耦形式对齐，我们进一步引入了感知-推理交替GRPO（Perception-Reasoning Alternating GRPO, PRA-GRPO），这是一种角色感知的强化学习策略，使用最终答案监督在以感知为重点和以推理为重点的更新之间交替进行。基于Qwen3-VL-Instruct-2B/4B/8B，P2R在各模型规模上持续提升性能。特别是，P2R-4B在V-Star上达到了93.2%，在HR-Bench-4K上达到了81.9%，在HR-Bench-8K上达到了80.5%，显著优于其相应的基础模型。进一步的实验表明，P2R的优势不仅限于高分辨率基准测试，还扩展到更广泛的多模态推理任务。这些结果表明，明确解耦感知与推理为细粒度视觉推理提供了有效的框架。
cs.CV / 138 / 2607.01202
World from Motion: Generative Dynamic Gaussian Reconstruction from Monocular Video
运动中的世界：基于单目视频的生成动态高斯重建
Abstract
We present World from Motion, a method for generating freely renderable dynamic 3D Gaussian representations from monocular videos. Our approach conditions a video model on dense, pixel-aligned renderings that encode appearance, geometry, and 3D scene motion along both input and target camera trajectories to correct rendering artifacts and fill in missing regions from an initial reconstruction. To train this model, we construct a dataset of aligned multiview video pairs and dynamic 3DGS representations, with simulated artifacts characteristic of monocular reconstruction. At test time, we distill the model's generations, including newly observed regions and motions, back into a single consistent, high-quality dynamic 3DGS, improving both novel-view synthesis and the underlying 3D motion. Our method sets a new state of the art in 4D reconstruction and seamlessly generalizes to in-the-wild videos with large viewpoint changes and dynamic motions.
Chinese Translation
我们提出了一种方法——运动中的世界，用于从单目视频生成可自由渲染的动态三维高斯表示。我们的方法将视频模型与密集的、像素对齐的渲染相结合，这些渲染编码了外观、几何形状以及沿输入和目标相机轨迹的三维场景运动，以纠正渲染伪影并填补初始重建中的缺失区域。为了训练该模型，我们构建了一个对齐的多视角视频对和动态三维高斯表示的数据集，并模拟了单目重建特有的伪影。在测试时，我们将模型的生成结果，包括新观察到的区域和运动，提炼回一个一致的高质量动态三维高斯表示，从而改善了新视角合成和基础三维运动。我们的方法在四维重建领域设定了新的最先进水平，并能够无缝推广到具有大视角变化和动态运动的野外视频。
cs.CV / 139 / 2607.01205
Linkify: Learning from Interface-Augmented Assembly Graphs
Linkify：从接口增强的装配图中学习
Abstract
We present Linkify, a framework for learning from interface-augmented assembly graphs to enable context-aware part retrieval in mechanical assemblies. While recent generative AI methods for CAD have focused largely on isolated parts or monolithic assemblies, the rich geometric information at the interfaces between parts, where function is realized, remains underexplored. We address this gap by recomputing high-fidelity interface geometry for the Fusion 360 Gallery Assembly dataset, correcting missing and erroneous contacts, and generating point-cloud representations of local contact regions. Using this data, we construct assembly graphs whose nodes encode part geometry and whose edges encode interface geometry via a pretrained point-cloud encoder. On top of this representation, we train a Graph Attention Network based on GATv2 to solve a masked part prediction task: given an assembly with one part held out, the model predicts the class of the missing component from a large vocabulary of geometrically clustered parts, thereby approximating a realistic part-retrieval scenario. Compared to non-graph baselines such as logistic regression and k-nearest neighbors operating on aggregated node features, Linkify achieves higher Top-K accuracy and F1 scores. Ablation studies on graph connectivity, edge attributes, and attention mechanisms demonstrate that accurate contact computation and dynamic attention over interfaces are critical for performance. Our corrected interface dataset and training pipeline, released publicly, provide a foundation for future interface-aware models for assembly retrieval, validation, and generative design.
Chinese Translation
我们提出了Linkify，一个从接口增强的装配图中学习的框架，以实现机械装配中的上下文感知部件检索。尽管近期针对计算机辅助设计（CAD）的生成性人工智能方法主要集中在孤立部件或单体装配上，但部件之间接口处的丰富几何信息（功能实现的地方）仍然未被充分探索。我们通过重新计算Fusion 360 Gallery Assembly数据集的高保真接口几何，纠正缺失和错误的接触，并生成局部接触区域的点云表示，来填补这一空白。利用这些数据，我们构建了装配图，其节点编码部件几何，边缘通过预训练的点云编码器编码接口几何。在此表示的基础上，我们训练了基于GATv2的图注意力网络（Graph Attention Network），以解决掩蔽部件预测任务：给定一个缺少一个部件的装配，模型从大量几何聚类的部件中预测缺失组件的类别，从而近似一个现实的部件检索场景。与基于聚合节点特征的逻辑回归和k近邻等非图基线相比，Linkify在Top-K准确率和F1分数上取得了更高的成绩。关于图连接性、边属性和注意力机制的消融研究表明，准确的接触计算和对接口的动态注意力对性能至关重要。我们公开发布的修正接口数据集和训练管道为未来的接口感知模型在装配检索、验证和生成设计方面提供了基础。
cs.CV / 140 / 2607.01222
Ink3D: Sculpting 3D Assets with Extremely Complex Textures via Video Generative Models
Ink3D：通过视频生成模型雕刻具有极其复杂纹理的3D资产
Abstract
Recent 3D generative models can synthesize high-quality geometry but often struggle to reproduce intricate textures from reference images, largely due to the scarcity of large-scale 3D training data with rich surface appearance. In contrast, visual generative models are trained on datasets several orders of magnitude larger and excel at modeling complex visual patterns. Motivated by this gap, we introduce Ink3D, a framework that bridges 3D generation with large-scale video generative models to synthesize extremely complex textures. Ink3D first reconstructs a white-mesh geometry using an off-the-shelf 3D generation model. It then employs OrbitPainter, a conditional video generative model, to produce dense orbit-scan videos capturing object appearance across viewpoints. To convert these views into coherent textures, we introduce TextureOptimizer, a neural baking module that integrates dense multi-view observations while mitigating geometry inconsistencies arising from video generation. By decoupling geometry and texture synthesis and leveraging large-scale pretrained video priors, Ink3D enables significantly richer and more faithful texture generation than prior approaches.
Chinese Translation
最近的3D生成模型能够合成高质量的几何形状，但在重现参考图像中的复杂纹理方面常常面临困难，这主要是由于缺乏具有丰富表面外观的大规模3D训练数据。相比之下，视觉生成模型在几个数量级更大的数据集上进行训练，擅长建模复杂的视觉模式。基于这一差距，我们提出了Ink3D，一个将3D生成与大规模视频生成模型相结合的框架，以合成极其复杂的纹理。Ink3D首先使用现成的3D生成模型重建白色网格几何体。然后，它采用条件视频生成模型OrbitPainter，生成捕捉物体在不同视角下外观的密集轨道扫描视频。为了将这些视图转换为连贯的纹理，我们引入了TextureOptimizer，一个神经烘焙模块，能够整合密集的多视角观测，同时减轻视频生成过程中产生的几何不一致性。通过解耦几何体和纹理合成，并利用大规模预训练的视频先验，Ink3D能够比以往的方法生成显著更丰富和更真实的纹理。
cs.AI / 1 / 2607.00001
Constructive Alignment: Governing Preference Dynamics in Human-AI Interaction
建设性对齐：管理人机交互中的偏好动态
Abstract
Most approaches to AI alignment treat human preferences as fixed targets to be inferred and optimized. This assumption conflicts with extensive empirical evidence showing that preferences are layered, dynamic, and constructed through interaction--particularly with adaptive technologies. As AI systems become more persistent, personalized, and socially embedded, they increasingly participate in shaping what people attend to, value, and endorse over time. We introduce Constructive Alignment, a paradigm that reframes alignment as a control problem over evolving human preference trajectories rather than static preference satisfaction. Drawing on behavioral economics, psychology, and constructivist social theory, we model preferences as layered state variables that evolve under interaction with AI systems. We formalize this view using a control-theoretic framework in which system actions and interaction design jointly influence both world states and human evaluative states. We argue that alignment is not primarily about controlling AI behavior, but about regulating how AI systems influence the evolution of human preferences--ensuring that value trajectories remain coherent, reflectively endorsed, epistemically grounded, bounded against manipulation, and empowering under uncertainty. Alignment thus becomes a problem of governing long-term value formation rather than simply satisfying static preferences.
Chinese Translation
大多数人工智能对齐方法将人类偏好视为固定目标，以进行推断和优化。这一假设与大量实证证据相悖，这些证据表明偏好是分层的、动态的，并且通过互动构建而成——尤其是在与自适应技术的互动中。随着人工智能系统变得越来越持久、个性化和社会嵌入，它们在塑造人们关注、重视和认可的事物方面的参与度也在不断增加。我们引入了建设性对齐（Constructive Alignment），这一范式将对齐重新定义为对不断演变的人类偏好轨迹的控制问题，而不是静态偏好的满足。基于行为经济学、心理学和建构主义社会理论，我们将偏好建模为在与人工智能系统互动中演变的分层状态变量。我们使用控制理论框架对这一观点进行形式化，其中系统的行动和互动设计共同影响世界状态和人类评估状态。我们认为，对齐并不仅仅是控制人工智能行为，而是调节人工智能系统如何影响人类偏好的演变——确保价值轨迹保持一致、经过反思的认可、具有认识论基础、抵御操控，并在不确定性下赋权。因此，对齐成为一个治理长期价值形成的问题，而不仅仅是满足静态偏好。
cs.AI / 2 / 2607.00002
Bounded Morality: Defining the Space of Moral Computation
有限道德：定义道德计算的空间
Abstract
Moral cognition has traditionally been modeled as adherence to fixed ethical theories--deontology, consequentialism, virtue ethics--implemented as static rules or value functions. We propose Bounded Morality, a formal framework for analyzing the computational demands of moral problems faced by finite agents. Extending Herbert Simon's notion of bounded rationality, we formalize moral situations along two orthogonal dimensions: moral breadth, the scope of entities treated as morally relevant, and moral depth, the inferential integration required to evaluate their interactions. Limited resources impose an unavoidable tradeoff between these dimensions, defining a feasible space of moral computation. Within this space, ethical theories correspond to locally efficient strategies adapted to different demand regimes rather than competing accounts of moral truth. The framework yields a formal notion of moral regret and moral progress under constraint, and implies that moral alignment in artificial systems depends on the scaling and allocation of moral reasoning capacity rather than on direct imitation of human judgments.
Chinese Translation
道德认知传统上被建模为遵循固定的伦理理论——义务论、结果主义、德性伦理——以静态规则或价值函数的形式实现。我们提出了有限道德（Bounded Morality），这是一个用于分析有限代理人面临的道德问题的计算需求的正式框架。扩展赫伯特·西蒙（Herbert Simon）的有限理性概念，我们沿两个正交维度形式化道德情境：道德广度，即被视为道德相关的实体的范围；道德深度，即评估这些实体相互作用所需的推理整合。有限资源在这两个维度之间施加了不可避免的权衡，定义了道德计算的可行空间。在这个空间内，伦理理论对应于适应不同需求模式的局部有效策略，而不是竞争的道德真理解释。该框架产生了在约束下的道德遗憾和道德进步的正式概念，并暗示人工系统中的道德一致性依赖于道德推理能力的规模和分配，而不是对人类判断的直接模仿。
cs.AI / 3 / 2607.00032
The MMM Data Model -- A Normative Specification for Knowledge Interoperability in a Decentralisable Knowledge Commons
MMM数据模型——去中心化知识共享中的知识互操作性的规范性规范
Abstract
Many information systems are built around documents: self-contained units optimised for print production and linear reading. While effective for large-scale dissemination, the document-centric organisation constrains how knowledge can be structured, updated, shared, and reused. Formal approaches address some of these limitations but struggle to achieve widespread contribution and adoption due to their prioritisation of formal structure over other system properties such as human usability and scope. AI systems are reshaping document production, but without providing a unified portable alternative to traditional documents for humans' expression and exchange of knowledge. This paper presents MMM, a data model for knowledge documentation that emerged from the practical needs of interdisciplinary collaborative research, and positioned here within a comparative analysis of the design space of information systems. MMM combines a small set of normative constraints with the expressive freedom of free-text labels. It is designed for interoperability across disciplines, applications and deployments without requiring semantic convergence. A reference implementation and pilot deployment data demonstrate implementability and early usability.
Chinese Translation
许多信息系统围绕文档构建：自包含的单元，优化用于印刷生产和线性阅读。尽管这种文档中心的组织方式在大规模传播中有效，但限制了知识的结构化、更新、共享和重用方式。正式的方法解决了其中一些局限性，但由于优先考虑正式结构而非其他系统属性（如人类可用性和范围），因此难以实现广泛的贡献和采用。人工智能系统正在重塑文档生产，但并未为人类表达和交换知识提供统一的可移植替代方案。本文提出了MMM，一个知识文档的数据模型，源于跨学科协作研究的实际需求，并在此处通过对信息系统设计空间的比较分析进行定位。MMM结合了一小组规范性约束与自由文本标签的表达自由。它旨在实现跨学科、应用和部署的互操作性，而无需语义收敛。参考实现和试点部署数据展示了其可实施性和早期可用性。
cs.AI / 4 / 2607.00035
Making Failure Safe: A Constrained, Verifiable Agent Framework for Open-Web Data Collection
确保失败安全：一个受限的、可验证的代理框架用于开放网络数据收集
Abstract
LLMs and agents can generate web scrapers from natural-language requirements, but direct generation remains unreliable because of dependency errors, broken selectors, schema mismatches, and heterogeneous page structures. We propose a constrained, verifiable agent framework that shifts LLM output from free-form code to typed JSON collector configurations, combining a six-type collector taxonomy, template and utility-function constraints, static Airflow DAG execution, rule-based quality checking, and structured feedback correction. Experiments on 138 tasks show that the taxonomy supports description-based requirement typing, while confirming that stable instantiation requires completing source, field, and execution constraints beyond the initial description. On 80 independently source-verified tasks, the framework runs with zero execution-stage LLM tokens and the lowest average wall-clock time, trading moderate one-shot quality for a reusable, deterministic, and verifiable execution path suited to repeated scheduled collection. These results position the framework as a reusable, low-cost, and verifiable execution path for repeated open-web data collection.
Chinese Translation
大型语言模型（LLMs）和代理可以根据自然语言需求生成网络爬虫，但直接生成仍然不可靠，因为存在依赖错误、选择器失效、模式不匹配和异构页面结构等问题。我们提出了一个受限的、可验证的代理框架，将LLM输出从自由格式代码转变为类型化的JSON收集器配置，结合了六种类型的收集器分类法、模板和实用函数约束、静态Airflow DAG执行、基于规则的质量检查以及结构化反馈修正。对138个任务的实验表明，该分类法支持基于描述的需求类型化，同时确认稳定的实例化需要在初始描述之外完成源、字段和执行约束。在80个独立来源验证的任务上，该框架在执行阶段使用零个LLM令牌，并且具有最低的平均时钟时间，权衡适度的一次性质量，以实现适合重复调度收集的可重用、确定性和可验证的执行路径。这些结果使该框架成为一个可重用、低成本和可验证的执行路径，适用于重复的开放网络数据收集。
cs.AI / 5 / 2607.00064
Solution space path planning for supporting en-route air traffic control
支持航路空中交通管制的解空间路径规划
Abstract
As technology advances, many path-planning algorithms have been proposed for Air Traffic Management, yet their operational adoption in tactical control remains limited, revealing a misalignment between algorithmic design priorities and air traffic controllers' needs. This underscores the need for decision-support solutions that are inherently interpretable, computationally efficient, and explicitly designed for human use. Focusing on this design challenge, this study develops a conflict-free path-planning algorithm for en-route Air Traffic Control (ATC) designed to be compatible with two guiding considerations: (1) the interpretability and flexibility offered by solution-space displays, which motivate constructing an algorithm that exposes all feasible safe actions and accommodates shifting optimization goals; and (2) the decision logic controllers naturally apply when enforcing operational constraints, such as separation standards, maneuverability limits, waypoint minimization, and routing practicality. Centered on these principles, the algorithm integrates three intent-based conflict detection methods -- distance-based, time-interval-based, and zone-based -- within a solution-space framework to identify conflict-free paths in computationally efficient ways. Additionally, vertex-based and edge-based search nodes are proposed for solution space path planning (SSPP), resulting in two variants -- SSPPV and SSPPE, respectively, which are evaluated in terms of computational speed and solution quality. Empirical results show that SSPPV paired with zone-based conflict detection achieves the best performance, computing paths in 3.69 ms on average in operational-relevant scenarios based on the Delta sector of the Maastricht Upper Area Control Centre (MUAC) using a 5 nmi grid.
Chinese Translation
随着技术的进步，许多路径规划算法已被提出用于空中交通管理，然而它们在战术控制中的实际应用仍然有限，这揭示了算法设计优先级与空中交通管制员需求之间的不一致。这突显了需要本质上可解释、计算高效且明确为人类使用而设计的决策支持解决方案。针对这一设计挑战，本研究开发了一种用于航路空中交通管制（ATC）的无冲突路径规划算法，该算法兼顾两个指导考虑因素：(1) 解空间显示所提供的可解释性和灵活性，促使构建一种能够揭示所有可行安全行动并适应优化目标变化的算法；(2) 管制员在执行操作约束（如分离标准、机动性限制、航点最小化和航线实用性）时自然应用的决策逻辑。基于这些原则，该算法在解空间框架内集成了三种基于意图的冲突检测方法——基于距离、基于时间间隔和基于区域的方法，以计算高效的方式识别无冲突路径。此外，提出了基于顶点和边缘的搜索节点用于解空间路径规划（SSPP），最终形成了两个变体——SSPPV和SSPPE，分别在计算速度和解决方案质量方面进行了评估。实证结果表明，SSPPV与基于区域的冲突检测结合时表现最佳，在基于马斯特里赫特上空管制中心（MUAC）Delta区的操作相关场景中，平均计算路径所需时间为3.69毫秒。
cs.AI / 6 / 2607.00147
RareDxR1: Autonomous Medical Reasoning for Rare Disease Diagnosis Beyond Human Annotation
RareDxR1：超越人工标注的稀有疾病诊断自主医学推理
Abstract
Rare disease differential diagnosis is a critical yet arduous clinical task, requiring physicians to identify precise phenotypes from complex, unstructured patient symptoms and execute intricate reasoning within a vast search space. However, existing AI approaches typically rely on pipeline-based phenotype extraction or retrieval-augmented generation, which suffer from critical information loss due to predefined ontologies, retrieval bottlenecks, and a lack of diagnostic logic. To address these challenges, we introduce RareDxR1, an end-to-end reasoning-centric large language model designed for open-domain rare disease diagnosis directly from unstructured clinical notes. We design a progressive end-to-end training framework by synergizing knowledge internalization with autonomous evolutionary learning, thereby bypassing reliance on structured phenotypes and closed-set decision-making. To overcome the limitations of RAG and phenotype restriction, we enabled the deep internalization of fragmented rare-disease knowledge directly into the model's parameters. Moreover, to bridge the gap between model generation and expert reasoning, we propose Reflection-Enhanced Reasoning Sampling (RERS), a strategy that synthesizes expert-level diagnostic trajectories by learning from failures without human annotation. Additionally, we propose a dual-level curriculum reinforcement learning approach for gradually mastering rare disease diagnosis. Experimental results demonstrate that RareDxR1 achieves state-of-the-art accuracy across different benchmarks, marking a significant breakthrough in open-domain rare disease diagnosis. Our code and dataset will be publicly available.
Chinese Translation
稀有疾病的鉴别诊断是一项关键但艰巨的临床任务，要求医生从复杂的非结构化患者症状中识别出精确的表型，并在广阔的搜索空间内执行复杂的推理。然而，现有的人工智能方法通常依赖于基于管道的表型提取或检索增强生成，这些方法由于预定义本体、检索瓶颈以及缺乏诊断逻辑而遭受严重的信息损失。为了解决这些挑战，我们提出了RareDxR1，这是一种端到端的以推理为中心的大型语言模型，旨在直接从非结构化临床记录中进行开放领域的稀有疾病诊断。我们通过将知识内化与自主进化学习相结合，设计了一个渐进的端到端训练框架，从而绕过对结构化表型和封闭集决策的依赖。为了克服RAG和表型限制的局限性，我们使稀有疾病知识的深度内化直接融入模型的参数中。此外，为了弥合模型生成与专家推理之间的差距，我们提出了反思增强推理采样（Reflection-Enhanced Reasoning Sampling, RERS）策略，该策略通过从失败中学习而无需人工标注，合成专家级的诊断轨迹。此外，我们提出了一种双层课程强化学习方法，以逐步掌握稀有疾病的诊断。实验结果表明，RareDxR1在不同基准测试中实现了最先进的准确性，标志着开放领域稀有疾病诊断的重大突破。我们的代码和数据集将公开发布。
cs.AI / 7 / 2607.00155
A Contextual-Bandit Oversight Game with Two-Sided Informational Asymmetry
具有双向信息不对称的上下文赌博监督博弈
Abstract
We study runtime human oversight of an AI agent when private information runs in both directions: the human privately knows her reward function, while the AI privately knows the quality of the action it proposes. This is the kind of asymmetry that arises naturally when an autonomous robot or software agent has inspected a situation its human supervisor cannot directly assess. Building on Cooperative Inverse Reinforcement Learning (CIRL) and the Oversight Game, we introduce a contextual-bandit team game with two-sided asymmetric information and a play/ask/trust/oversee interface. The bandit structure removes physical state transitions and thereby yields exact one-shot characterizations that would remain conjectural in the full POMDP setting, though the common belief remains a dynamically controlled state across rounds. We give two one-shot characterizations, a team optimum and a behaviorally natural myopic rule, whose gap is a slab of avoidable harm: a region in which the AI privately knows the proposed action is harmful and shutdown would help, yet a myopic human, trusting her prior, declines to oversee. We show this gap is the price of non-credible oversight communication, and give a partial analysis of how it resolves dynamically over repeated rounds through passive learning and active signaling with a one-period-lagged oversight response.
Chinese Translation
我们研究了在私有信息双向流动的情况下，人工智能（AI）代理的运行时人类监督：人类私下了解她的奖励函数，而AI私下了解其提议行动的质量。这种不对称性在自主机器人或软件代理检查人类监督者无法直接评估的情况时自然产生。基于合作逆强化学习（Cooperative Inverse Reinforcement Learning, CIRL）和监督博弈，我们引入了一种具有双向不对称信息的上下文赌博团队博弈，并设计了一个玩/询问/信任/监督的接口。赌博结构消除了物理状态转移，从而提供了确切的一次性特征，这在完整的部分可观测马尔可夫决策过程（POMDP）环境中仍然是推测性的，尽管共同信念在多个回合中仍然是一个动态控制的状态。我们给出了两个一次性特征：一个团队最优解和一个行为上自然的短视规则，其差距是一个可避免的伤害区域：在该区域中，AI私下知道提议的行动是有害的，关闭将有助于，但短视的人类信任她的先验，拒绝监督。我们表明，这一差距是非可信监督沟通的代价，并部分分析了它如何通过被动学习和主动信号在重复回合中动态解决，伴随一个滞后的监督响应。
cs.AI / 8 / 2607.00211
Constructing Epistemic AI Literacy: Detecting Epistemic Aims and Processes in Student-AI Co-Programming
构建知识性人工智能素养：在学生与人工智能共同编程中识别知识性目标和过程
Abstract
Epistemic thinking plays a central role in students' learning processes when applying generative artificial intelligence (GenAI), particularly in programming contexts where learners must construct queries, evaluate and validate AI-generated outputs, and regulate problem-solving strategies. This study introduces the conceptual framework of Epistemic AI Literacy (EAIL), reframing AI literacy as a process-oriented epistemic phenomenon that emerges through dynamic human-AI interactions across different domains. Drawing on the AIR (epistemic aims, ideals and reliable epistemic processes) framework, this study examines how epistemic aims and epistemic processes are enacted in GenAI-supported co-programming activities and explores scalable approaches for operationalizing these constructs in interaction data. Using a large dialogue dataset of human-AI co-programming, this study identifies observable dimensions of epistemic aims (i.e., mastery-oriented aims) and epistemic processes (i.e., outsourcing, explanation seeking, verification seeking, prompt monitoring, and epistemic justification). The results reveal a prevalent lack of EAIL, with 78.8% of student-GenAI interactions relying on non-mastery-oriented aims and less reliable epistemic strategies like outsourcing and verification-seeking. Conversely, only 11.1% of interactions showed high epistemic engagement, where mastery-oriented aims were coupled with advanced epistemic strategies like epistemic justification in a more reliable epistemic process.
Chinese Translation
知识性思维在学生应用生成性人工智能（GenAI）的学习过程中发挥着核心作用，尤其是在编程环境中，学习者必须构建查询、评估和验证人工智能生成的输出，并调节问题解决策略。本研究引入了知识性人工智能素养（EAIL）的概念框架，将人工智能素养重新定义为一种过程导向的知识性现象，这一现象通过不同领域的人机动态交互而产生。基于AIR（知识性目标、理想和可靠的知识性过程）框架，本研究考察了在GenAI支持的共同编程活动中，知识性目标和知识性过程是如何被实施的，并探索了在交互数据中操作化这些构念的可扩展方法。通过使用一个大型的人机共同编程对话数据集，本研究识别了知识性目标（即以掌握为导向的目标）和知识性过程（即外包、寻求解释、寻求验证、提示监控和知识性证明）的可观察维度。结果显示，EAIL普遍缺乏，78.8%的学生与GenAI的交互依赖于非以掌握为导向的目标以及不太可靠的知识性策略，如外包和寻求验证。相反，仅有11.1%的交互表现出高水平的知识性参与，其中以掌握为导向的目标与更可靠的知识性过程中的高级知识性策略（如知识性证明）相结合。
cs.AI / 9 / 2607.00233
From Signals to Structure: How Memory Architecture Drives Language Emergence in LLM Agents
从信号到结构：记忆架构如何驱动大型语言模型代理的语言出现
Abstract
How do two agents invent a shared language from scratch? In a Lewis signaling game, a sender and receiver must coordinate on a code using only their interaction history. We study five memory architectures across varying channel configurations with LLM agents and find that memory architecture matters more than channel capacity. Agents with a persistent private notebook benefit from surplus channel capacity and avoid the high-capacity collapse seen in stateless agents, achieving the most reliable coordination ($0.867 \pm 0.023$ at capacity = 25). Stateless agents peak at moderate capacity and then degrade as the vocabulary grows beyond what a rolling context window can track The notebook externalizes learned conventions, freeing agents from having to re-derive codes each round. An information bottleneck-inspired argument predicts an optimal capacity equal to the number of objects. Instead, the bottleneck (capacity = 8) proves to be a fragility point, and surplus capacity is generally better. We show that channel capacity alone cannot predict coordination; memory architecture determines whether agents turn interaction history into stable conventions, and both dimensions are needed to understand how signals become language.
Chinese Translation
两个代理如何从零开始发明一种共享语言？在路易斯信号游戏中，发送者和接收者必须仅通过他们的互动历史来协调一个代码。我们研究了五种记忆架构在不同信道配置下与大型语言模型（LLM）代理的表现，发现记忆架构比信道容量更为重要。拥有持久私人笔记本的代理能够利用多余的信道容量，避免无状态代理所经历的高容量崩溃，达成最可靠的协调（$0.867 imes 0.023$，在容量 = 25 时）。无状态代理在中等容量下达到峰值，然后随着词汇量超出滚动上下文窗口的跟踪能力而退化。笔记本外化了学习到的约定，使代理不必在每一轮中重新推导代码。受信息瓶颈启发的论点预测最佳容量等于对象数量。然而，瓶颈（容量 = 8）实际上被证明是一个脆弱点，而多余的容量通常更好。我们展示了单靠信道容量无法预测协调；记忆架构决定了代理是否将互动历史转化为稳定的约定，而理解信号如何演变为语言需要同时考虑这两个维度。
cs.AI / 10 / 2607.00248
Seed2.0 Model Card: Towards Intelligence Frontier for Real-World Complexity
Seed2.0 模型卡：迈向现实世界复杂性的智能前沿
Abstract
We present Seed2.0, a model series that takes a meaningful step toward solving complex, real-world tasks. Our approach begins with identifying users' genuine needs and constructing a reliable, forward-looking evaluation system by selecting and abstracting benchmarks grounded in these needs and in realistic, complex scenarios. Guided by this evaluation system, Seed2.0 targets two persistent challenges, long-tail knowledge and complex instruction following, substantially improving the model's reliability on intricate, long-horizon tasks. Beyond these, Seed2.0 delivers world-leading reasoning intelligence, visual understanding, and search capabilities that address the most common needs of a broad user base. Through extensive real-world use cases documented in this model card, we demonstrate that Seed2.0 begins to exhibit the ability to handle initial complex real-world tasks, delivering greater value to hundreds of millions of users.
Chinese Translation
我们提出了 Seed2.0，一个系列模型，朝着解决复杂的现实世界任务迈出了重要一步。我们的方法首先识别用户的真实需求，并通过选择和抽象基于这些需求以及现实复杂场景的基准，构建一个可靠的前瞻性评估系统。在这个评估系统的指导下，Seed2.0 针对两个持久挑战——长尾知识和复杂指令跟随，显著提高了模型在复杂长时任务上的可靠性。除此之外，Seed2.0 提供了世界领先的推理智能、视觉理解和搜索能力，以满足广泛用户群体中最常见的需求。通过在此模型卡中记录的大量现实世界案例，我们展示了 Seed2.0 开始展现处理初步复杂现实世界任务的能力，为数亿用户提供更大的价值。
cs.AI / 11 / 2607.00269
Mnemosyne: Agentic Transaction Processing for Validating and Repairing AI-generated Workflows
Mnemosyne：用于验证和修复 AI 生成工作流的代理事务处理
Abstract
LLMs, solvers, and agent teams increasingly generate workflow actions, repairs, and plans, but a generated action may be syntactically valid yet stale, infeasible, conflicting, or destructive of the evidence that triggered a repair. We introduce Agentic Transaction Processing (ATP), a transaction model that treats generated actions as untrusted proposals until they pass deterministic admission under a declared, executable constraint set C. The principle is two-sided: a proposal is not truth, and no proposal foresees every disruption: anything may propose, but only the runtime admits and commits, and when an unforeseen disruption strikes it repairs reactively within bounds rather than trusting a fresh proposal. Relative to C, committed-state correctness becomes independent of the competence, honesty, or learning of the proposing layer. We realize ATP in Mnemosyne, a runtime with an append-only transition log, effective-state projection, dependency-safe compensation, and active commitment records, and prove four safety properties relative to C (authority separation, serial-equivalent generative admission, evidence-preserving repair, and obligation containment) together with a bounded-reactive-repair guarantee for its localized repair protocol (LCRP). A reproducible artifact rejects the targeted violations across nine falsification tests while still admitting valid work, at under 6% projection-and-validation overhead, and bounded local repair edits an order of magnitude fewer operations than global recompute. Mnemosyne is open source: https://github.com/eyuchang/Mnemosyne/tree/arxiv-atp-rq1-rq9b-r8-v2.
Chinese Translation
大型语言模型（LLMs）、求解器和代理团队日益生成工作流操作、修复和计划，但生成的操作可能在语法上有效，却可能过时、不可行、冲突或破坏触发修复的证据。我们引入了代理事务处理（Agentic Transaction Processing, ATP），这是一种事务模型，将生成的操作视为不可信的提案，直到它们在声明的可执行约束集 C 下通过确定性的接纳。该原则是双向的：提案不是真理，且没有提案能够预见每一个干扰：任何事物都可以提出提案，但只有运行时可以接纳和承诺，当未预见的干扰发生时，它会在边界内进行反应式修复，而不是信任新的提案。相对于 C，承诺状态的正确性独立于提案层的能力、诚实或学习。我们在 Mnemosyne 中实现了 ATP，这是一种具有仅附加过渡日志、有效状态投影、依赖安全补偿和主动承诺记录的运行时，并证明了相对于 C 的四个安全属性（权限分离、序列等效生成接纳、证据保留修复和义务约束），以及其局部修复协议（Localized Repair Protocol, LCRP）的有界反应式修复保证。一个可重现的工件在九个伪造测试中拒绝了目标违规，同时仍然接纳有效工作，投影和验证的开销低于 6%，而有界的局部修复编辑的操作数量比全局重新计算少一个数量级。Mnemosyne 是开源的： https://github.com/eyuchang/Mnemosyne/tree/arxiv-atp-rq1-rq9b-r8-v2。
cs.AI / 12 / 2607.00334
Managed Autonomy at Runtime: Gear-Based Safety and Governance for Single- and Multi-Agent Cyber-Physical Systems
运行时的管理自主性：基于齿轮的单体和多体智能物理系统的安全与治理
Abstract
Autonomous agents, whether LLM-driven software agents or robotic physical agents, face a common class of failure modes when operating without continuous human oversight: safety violations from unverified actions, behavioral instability from unconstrained loops, and continuity loss from unhandled error states. We develop \system{}, a discrete-time control system that combines five execution gears (\Gobs{}, \Gsug{}, \Gplan{}, \Gexec{}, \Gint{}) with utility-gated dispatch and event-driven fallback. For the single-agent case, we prove monotonic stability, execution safety, eventual stabilization, fallback completeness, and equivalence to a gear-constrained Markov decision process. For multi-agent cyber-physical systems (CPS), we apply the established \smart{} managed-autonomy lifecycle and map runtime evidence into its four governance states (\Stable{}/\Meta{}/\Assisted{}/\Regulated{}). Consensus gating, swarm-level Lyapunov analysis, per-agent gear authority, and rendezvous control provide distributed safety and stability guarantees, including zero collision under the stated assumptions. We evaluate the resulting runtime on a three-agent UR5 robotic assembly cell using fault magnitudes calibrated from the NIST \emph{Degradation Measurement of Robot Arm Position Accuracy} dataset across 10,000 Monte Carlo episodes. It achieves a 99.6\% anomaly detection rate versus 2.1\% for the single-agent baseline, reduces detection latency by $3.5\times$, and supplies a formal physical-workspace safety certificate. The execution gears act as micro-level permissions beneath the \smart{} runtime governance states, separating action control from autonomy governance.
Chinese Translation
自主代理，无论是基于大型语言模型（LLM）的软件代理还是机器人物理代理，在没有持续人类监督的情况下操作时，都面临一类共同的故障模式：由于未经验证的行为导致的安全违规、由于不受约束的循环引起的行为不稳定，以及由于未处理的错误状态导致的连续性丧失。我们开发了 extit{system}，一种离散时间控制系统，结合了五个执行齿轮（ extit{Gobs}、 extit{Gsug}、 extit{Gplan}、 extit{Gexec}、 extit{Gint}）与效用门控调度和事件驱动的后备机制。在单代理情况下，我们证明了单调稳定性、执行安全性、最终稳定性、后备完整性以及与齿轮约束的马尔可夫决策过程的等价性。对于多代理智能物理系统（CPS），我们应用已建立的 extit{smart}管理自主生命周期，并将运行时证据映射到其四个治理状态（ extit{Stable}、 extit{Meta}、 extit{Assisted}、 extit{Regulated}）。共识门控、群体级李雅普诺夫分析、每个代理的齿轮权限和会合控制提供了分布式的安全性和稳定性保证，包括在所述假设下零碰撞。我们在一个三代理的UR5机器人装配单元上评估了所得到的运行时，使用从NIST的 extit{机器人臂位置精度退化测量}数据集中校准的故障幅度，进行了10,000次蒙特卡洛实验。其异常检测率达到99.6\%，而单代理基线为2.1\\%，检测延迟减少了$3.5 imes$，并提供了正式的物理工作空间安全证书。执行齿轮在 extit{smart}运行时治理状态下作为微级权限，分离了行动控制与自主治理。
cs.AI / 13 / 2607.00407
Personalization as Inverse Planning: Learning Latent Design Intents for Agentic Slide Generation via Structural Denoising
个性化作为逆向规划：通过结构去噪学习代理幻灯片生成的潜在设计意图
Abstract
Slide design requires personalizing both deck themes and page layouts. Yet, current AI agent-based methods struggle with fine-grained, page-level design. Solely relying on prespecified templates or user verbose instructions, they fail to capture latent design intents, leaving Page-level Slide Personalization (PSP) unresolved. To close this gap, this work formulates PSP as an inverse planning problem. We propose to learn a design intent without assuming any knowledge of the specific executing tools (e.g., PowerPoint, Beamer) being used. However, relinquishing control over these tools makes the problem intractable to optimize end-to-end. To overcome this, we propose SPIRE, a principled framework to solve PSP approximately. By intentionally corrupting the visual structures of clean slides, SPIRE creates a verifiable task to denoise the corruption, whereby two agents learn to collaboratively refine executable designs via reinforcement learning (RL). We present a proof that structural denoising is a consistent surrogate for PSP, and that the multi-agent formulation strictly reduces policy gradient variance in RL. Extensive experiments demonstrate the superiority of SPIRE.
Chinese Translation
幻灯片设计需要个性化处理幻灯片主题和页面布局。然而，当前基于AI代理的方法在细粒度的页面级设计上存在困难。仅依赖预先指定的模板或用户冗长的指令，它们无法捕捉潜在的设计意图，从而使页面级幻灯片个性化（Page-level Slide Personalization, PSP）问题未得到解决。为了解决这一问题，本研究将PSP形式化为一个逆向规划问题。我们提出在不假设对所使用的具体执行工具（如PowerPoint、Beamer）有任何知识的情况下学习设计意图。然而，放弃对这些工具的控制使得端到端优化问题变得不可处理。为此，我们提出了SPIRE，一个解决PSP的原则性框架。通过故意破坏干净幻灯片的视觉结构，SPIRE创建了一个可验证的任务来去噪这种破坏，其中两个代理通过强化学习（Reinforcement Learning, RL）协作学习优化可执行设计。我们证明了结构去噪是PSP的一个一致性替代，并且多代理的形式严格减少了RL中的策略梯度方差。大量实验表明SPIRE的优越性。
cs.AI / 14 / 2607.00436
PHREEQC-MCQ-200: A Diagnostic Benchmark for Tool-Augmented Scientific Simulator Agents
PHREEQC-MCQ-200：一种用于工具增强科学模拟代理的诊断基准
Abstract
Large language model agents are increasingly connected to scientific software, yet it remains unclear when tool access makes scientific computation more reliable rather than merely more complex. We introduce PHREEQC-MCQ-200, a benchmark for evaluating tool-augmented agents on deterministic aqueous-geochemistry simulations. The benchmark contains 200 multiple-choice questions derived from 21 validated PHREEQC scenarios, requiring agents to construct simulator inputs, execute PHREEQC, inspect structured outputs, and commit to final answers. Across multiple frontier and mid-tier model families, simulator access substantially improves aggregate accuracy, confirming that grounded execution is necessary for many scientific-computation tasks. However, the gains are not monotonic: tool-augmented agents also lose items they answered correctly without tools, revealing regressions that average accuracy alone hides. We further show that output-access protocol matters. A table-of-contents interface can reduce token cost while preserving or improving accuracy for stronger models, but it degrades performance for mid-tier models that cannot reliably navigate structured simulator outputs. PHREEQC-MCQ-200 therefore frames scientific tool use as an end-to-end diagnostic problem rather than a simple tool-calling capability. We argue that evaluations of scientific agents should report not only accuracy, but also item-level retention, output-access sensitivity, trajectory failures, and where the computation chain breaks.
Chinese Translation
大型语言模型代理与科学软件的连接日益增强，但何时工具访问使科学计算更可靠而不仅仅是更复杂仍不清楚。我们介绍了PHREEQC-MCQ-200，这是一个用于评估工具增强代理在确定性水相地球化学模拟中的表现的基准。该基准包含200个多项选择题，源自21个经过验证的PHREEQC场景，要求代理构建模拟器输入、执行PHREEQC、检查结构化输出并提交最终答案。在多个前沿和中层模型系列中，模拟器访问显著提高了整体准确性，确认了在许多科学计算任务中，基于实际执行是必要的。然而，增益并非单调：工具增强代理在没有工具的情况下也会失去他们正确回答的项目，揭示了平均准确性所掩盖的回归现象。我们进一步表明，输出访问协议至关重要。目录界面可以降低令牌成本，同时为更强的模型保持或提高准确性，但对于无法可靠导航结构化模拟器输出的中层模型，其性能会下降。因此，PHREEQC-MCQ-200将科学工具的使用框架化为一个端到端的诊断问题，而不仅仅是简单的工具调用能力。我们认为，对科学代理的评估不仅应报告准确性，还应报告项目级保留、输出访问敏感性、轨迹失败以及计算链断裂的情况。
cs.AI / 15 / 2607.00454
Agri-SAGE: Simulation-Grounded Multi-Agent LLM for Context-Aware Agricultural Advisory Generation
Agri-SAGE：基于模拟的多智能体大语言模型用于上下文感知的农业咨询生成
Abstract
Agricultural advisory systems face a fundamental tension: static agronomic guidelines offer consistent, evidence-based recommendations, yet remain blind to in-season variability and dynamic uncertainties. Recent advisory systems powered by LLMs are liable for a different risk of generating recommendations that are agronomically credible but physiologically unconvincing. Agri-SAGE is a closed-loop framework designed to resolve the above two limitations by integrating retrieval-grounded multi-agent LLM reasoning with APSIM-based biophysical simulation, to generate and validate agronomic advisories. To assess this framework, we evaluate three reasoning approaches, namely Plan-and-Solve, Tree of Thoughts, and Reflexion, over a 10-year retrospective analysis. All three significantly outperform static PoP (Package-of-Practice) baselines, with Tree of Thoughts achieving impressive peak yields. At the same time, Reflexion achieves comparable agronomic outcomes at substantially lower computational cost by leveraging cross-seasonal episodic memory.
Chinese Translation
农业咨询系统面临着一个根本性的矛盾：静态的农业指南提供了一致的、基于证据的建议，但对季节内的变异性和动态不确定性却视而不见。近期由大语言模型（LLMs）驱动的咨询系统则面临着另一种风险，即生成的建议在农业学上可信，但在生理上却不令人信服。Agri-SAGE是一个闭环框架，旨在通过将基于检索的多智能体大语言模型推理与基于APSIM的生物物理模拟相结合，来解决上述两个局限，以生成和验证农业咨询。为了评估该框架，我们对三种推理方法进行了评估，即计划与解决（Plan-and-Solve）、思维树（Tree of Thoughts）和反思（Reflexion），并进行了为期10年的回顾性分析。所有三种方法的表现均显著优于静态的实践包（Package-of-Practice）基准，其中思维树取得了令人印象深刻的峰值产量。同时，反思通过利用跨季节的情节记忆，以显著较低的计算成本实现了可比的农业成果。
cs.AI / 16 / 2607.00457
Multi-scale Mixture of World Models for Embodied Agents in Evolving Environments
适应不断变化环境的具身智能体的多尺度世界模型混合
Abstract
Embodied agents operating in the real world require multi-scale reasoning and knowledge adaptation as conditions change. We identify two challenges in applying Mixture of Experts (MoE) to this setting: routing lacks an explicit notion of scale, preventing targeted updates at specific scales, and a uniform update policy cannot accommodate the different rates at which knowledge at each scale becomes outdated. We present MuSix, a framework that addresses both challenges through scale-aware world model mixture and evolution. A two-stage routing mechanism grounds scale selection in experiential distance, a measure of situational novelty inspired by Construal Level Theory: a meta-router first maps this quantity to a weight over continuous scale space, then per-scale base routers select world models within the identified scale. For adaptation, scale-dependent forgetting rates allow low-scale knowledge to refresh rapidly while high-scale abstractions persist, and gated inter-scale transfer maintains coherence across the hierarchy. Experiments on EmbodiedBench and HAZARD show that MuSix improves over state-of-the-art baselines on multi-scale reasoning and dynamic adaptation.
Chinese Translation
在现实世界中运行的具身智能体需要在条件变化时进行多尺度推理和知识适应。我们识别出在这一环境中应用专家混合模型（Mixture of Experts, MoE）面临的两个挑战：路由缺乏明确的尺度概念，阻碍了在特定尺度上进行有针对性的更新；而统一的更新策略无法适应各个尺度知识过时的不同速率。我们提出了MuSix，一个通过尺度感知的世界模型混合与演化来解决这两个挑战的框架。一个两阶段的路由机制将尺度选择基于经验距离，这是一种受构念层次理论（Construal Level Theory）启发的情境新颖性度量：一个元路由器首先将该量映射到连续尺度空间上的权重，然后每个尺度的基础路由器在识别的尺度内选择世界模型。在适应性方面，尺度依赖的遗忘率允许低尺度知识快速更新，而高尺度抽象得以持续，门控的跨尺度转移保持了层级之间的一致性。在EmbodiedBench和HAZARD上的实验表明，MuSix在多尺度推理和动态适应方面优于最先进的基线。
cs.AI / 17 / 2607.00527
AI Native Games: A Survey and Roadmap
人工智能原生游戏：调查与路线图
Abstract
Generative AI now enables games to produce dialogue, quests, characters, images, and worlds at runtime. Yet generation alone does not make a game AI-native, nor does it guarantee playability. This paper defines AI-native games by whether runtime generative AI is constitutive of the core loop: if the AI component were removed or trivially replaced, the central form of play would collapse or become fundamentally different. This counterfactual criterion separates AI-native games from AI-augmented games, boundary artifacts, chatbots, tavern-style role-play, procedural content generation, and AI-assisted production. Using this definition, we screen candidate artifacts and analyze 53 publicly available AI-native games and prototypes. We introduce a dual-axis G/N taxonomy: the G-axis captures player-facing game type, while the N-axis captures the dominant AI mechanic that makes generative AI indispensable to play. The corpus is concentrated around language-forward designs, especially narrative adventure, epistemic interaction, and generative narrative, while categories such as semantic adjudication, multi-agent simulation, generative construction, and relationship/companion play remain less represented. We argue that the central design problem is organizing semantic openness into stable gameplay. AI-native design depends on mechanical invariants: goals, rules, state, feedback, pacing, and player agency that make open-ended AI outputs interpretable and consequential. We conclude with a roadmap for controllable generation, AI-as-mechanic design, multimodal and multi-agent systems, inference economics, evaluation, safety, and regulation.
Chinese Translation
生成性人工智能现在使得游戏能够在运行时生成对话、任务、角色、图像和世界。然而，仅仅生成并不能使游戏成为人工智能原生，也不能保证其可玩性。本文通过运行时生成性人工智能是否构成核心循环来定义人工智能原生游戏：如果移除或轻易替代AI组件，游戏的中心玩法将崩溃或变得根本不同。这个反事实标准将人工智能原生游戏与人工智能增强游戏、边界工件、聊天机器人、酒馆风格角色扮演、程序化内容生成和人工智能辅助制作区分开来。基于这一定义，我们筛选候选工件并分析了53个公开可用的人工智能原生游戏和原型。我们引入了一个双轴G/N分类法：G轴捕捉面向玩家的游戏类型，而N轴则捕捉使生成性人工智能对游戏不可或缺的主导AI机制。该语料库集中于语言导向的设计，特别是叙事冒险、认知互动和生成叙事，而语义裁决、多智能体模拟、生成构建和关系/伴侣玩法等类别则相对较少。我们认为，核心设计问题是将语义开放性组织成稳定的游戏玩法。人工智能原生设计依赖于机械不变性：目标、规则、状态、反馈、节奏和玩家自主性，使开放式的人工智能输出可解释且具有重要性。最后，我们提出了一个关于可控生成、人工智能作为机制设计、多模态和多智能体系统、推理经济学、评估、安全性和监管的路线图。
cs.AI / 18 / 2607.00572
HARC: Coupling Harmfulness and Refusal Directions for Robust Safety Alignment
HARC：耦合有害性和拒绝方向以实现稳健的安全对齐
Abstract
Understanding how aligned LLMs internally represent safety is critical for diagnosing alignment vulnerabilities, as it explains why jailbreaks succeed and informs the design of robust alignment strategies. Prior work shows that aligned LLMs encode harmfulness and refusal as separable directions in the residual stream at prompt-side token positions. We show that jailbreaks succeed at prompt encoding by suppressing either the refusal or harmfulness direction before any token is generated, with distinct attack classes occupying separable regions of the harmfulness-refusal plane. Extending the analysis to response-token positions, we find that the model recognizes harmful content while it is generating that content, even when it failed to recognize the input as harmful at the prompt side. Motivated by our findings, we introduce HARC (Harmfulness-And-Refusal Coupling), a fine-tuning method that pairs the two directions across both prompt and response positions. Since the intervention is confined to the harmfulness-refusal subspace, it leaves the rest of the residual stream intact and does not degrade general capability or inflate over-refusal. Across extensive experiments, HARC achieves the strongest robustness-capability-usability trade-off among six baselines spanning the major training-time and inference-time safety methods. The harmfulness and refusal directions at prompt and response positions transfer across the five model families and two scales we tested without architecture-specific tuning.
Chinese Translation
理解对齐的大型语言模型（LLMs）如何内部表示安全性对于诊断对齐脆弱性至关重要，因为这解释了为什么越狱攻击成功，并为稳健的对齐策略设计提供了信息。先前的研究表明，对齐的LLMs在提示侧的令牌位置上将有害性和拒绝编码为可分离的方向。我们展示了越狱攻击通过在生成任何令牌之前抑制拒绝或有害性方向来成功地进行提示编码，不同的攻击类别占据有害性-拒绝平面中的可分离区域。将分析扩展到响应令牌位置，我们发现模型在生成有害内容时能够识别该内容，即使在提示侧未能将输入识别为有害。基于我们的发现，我们引入了HARC（有害性与拒绝耦合），这是一种在提示和响应位置之间配对这两个方向的微调方法。由于干预仅限于有害性-拒绝子空间，因此它保留了残差流的其余部分不变，并且不会降低一般能力或导致过度拒绝。在广泛的实验中，HARC在六个基准中实现了最强的稳健性-能力-可用性权衡，这些基准涵盖了主要的训练时间和推理时间安全方法。我们测试的五个模型家族和两个规模的提示和响应位置的有害性和拒绝方向在没有架构特定调优的情况下可以相互转移。
cs.AI / 19 / 2607.00627
AGI Maze as a Benchmark Framework for World-Modeling Agents
AGI迷宫作为世界建模代理的基准框架
Abstract
Large language models (LLMs) are powerful pattern-completion systems, but their default operating mode - predicting the next token from a static context - does not reliably produce persistent, manipulable representations of an external world. Many tasks that look like "reasoning" in text become substantially harder once the environment is partially observable, stateful, and requires memory and structured hypotheses about hidden state. AGI Maze is a lightweight framework for building such environments without requiring high-dimensional sensory inputs. It provides a family of grid-based maze tasks with a clean API and multiple difficulty regimes. The goal is to create benchmarks where agents must learn and use world state representations, not just infer a local rule over readily provided observations. We provide an initial evaluation of several vanilla LLMs on simple mazes showing that they fail to represent mazes internally at LLM inference time. We also introduce a baseline agent, which is allowed to use its message history as a working memory to construct descriptions of observations at agentic runtime. Although this can improve performance, it is still insufficient for an LLM agent to reliably solve even small mazes within a step budget that is more than enough for humans.
Chinese Translation
大型语言模型（LLMs）是强大的模式补全系统，但它们的默认操作模式——从静态上下文中预测下一个标记——并不能可靠地产生持久的、可操作的外部世界表示。许多看似在文本中进行的“推理”任务在环境部分可观察、有状态并需要关于隐藏状态的记忆和结构化假设时变得显著更难。AGI迷宫是一个轻量级框架，用于构建此类环境，而无需高维感官输入。它提供了一系列基于网格的迷宫任务，具有简洁的API和多种难度等级。其目标是创建基准，要求代理必须学习和使用世界状态表示，而不仅仅是在提供的观察上推断局部规则。我们对几种基础LLM在简单迷宫上的初步评估表明，它们在LLM推理时无法内部表示迷宫。我们还引入了一个基线代理，允许它使用其消息历史作为工作记忆，在代理运行时构建观察的描述。尽管这可以提高性能，但对于LLM代理来说，即使在给定的步骤预算内，解决小迷宫仍然是不够可靠的，而这个预算对于人类来说是绰绰有余的。
cs.AI / 20 / 2607.00642
Coachable agents for interactive gameplay
可教导的代理用于互动游戏
Capobianco, Roberto, van Seijen, Harm, Bard, Nolan D., Burch, Neil, Davelouis, Fatima, Davidson, Josh, Devlic, Alisa, Du, Yunshu, Durugkar, Ishan, Gangapurwala, Siddhant, Hernandez, Daniel, Holland, G. Zacharias, Jain, Sahil, Kawamoto, Kenta, Kumaraswamy, Raksha, MacAlpine, Patrick, Morrill, Dustin R., Oller, Declan, Riccio, Francesco, Saran, Akanksha, Sherstan, Craig, Subramanian, Kaushik, Walsh, Thomas J., Barrett, Samuel, Frisbee, Kizza N., Govil, Mady, Günther, Johannes, Kompella, Varun R., MacGlashan, James A., Svetlik, Maxwell, Thomure, Michael D., Travnik, Jaden B., Waugh, Kevin, Aghapour, Elahe, Fuchs, Florian, Lemay, Andreanne, Mishra, Shruti, Seno, Takuma, Stone, Peter, Spranger, Michael, Wurman, Peter R.
Abstract
Reinforcement learning has proven to be a valuable tool in the creation of advanced AI and robotic systems, contributing to everything from game playing to robotics to foundation models. Through trial-and-error, these AI systems typically learn one, near-optimal behavior to solve their tasks. However, there are many use cases in which one would like to assert some level of control, preferably in real time, over how the task is solved. We refer to these modifications of a core task as styles. We combine universal value function approximators (UVFAs) with carefully selected training scenarios, learning algorithms, and data augmentation to create a framework for coaching agents that exhibit styles in complex domains. We demonstrate the framework's application in the AAA video games Horizon Forbidden West and Gran Turismo, and in an open-source humanoid test domain. Despite the different nature of the domains -- car racing, stylized game combat, and humanoid walking -- each agent shows strong coherence to the style requests while still satisfying the main task in its domain. Importantly, the techniques outlined in this paper allow an end user to choose the final behavior at run time, giving them flexible control over the final executed performance.
Chinese Translation
强化学习已被证明是创建先进人工智能和机器人系统的宝贵工具，涵盖了从游戏到机器人再到基础模型的各个领域。通过试错，这些人工智能系统通常学习到一种接近最优的行为来解决其任务。然而，在许多使用场景中，人们希望能够在实时的情况下对任务的解决方式进行一定程度的控制。我们将这些对核心任务的修改称为风格。我们结合了通用价值函数逼近器（Universal Value Function Approximators, UVFAs）与精心选择的训练场景、学习算法和数据增强，创建了一个用于指导在复杂领域中展现风格的代理的框架。我们展示了该框架在AAA视频游戏《地平线：西部禁域》（Horizon Forbidden West）和《Gran Turismo》中的应用，以及在一个开源的人形测试领域中的应用。尽管这些领域的性质各不相同——赛车、风格化的游戏战斗和人形行走——每个代理在满足其领域的主要任务的同时，仍然表现出与风格请求的强一致性。重要的是，本文中概述的技术允许最终用户在运行时选择最终行为，从而为他们提供对最终执行性能的灵活控制。
cs.AI / 21 / 2607.00692
Self-GC: Self-Governing Context for Long-Horizon LLM Agents
自我GC：长时间跨度LLM代理的自我管理上下文
Abstract
Long-horizon LLM agents accumulate tool results, files, plans, and user constraints that are too structured to be treated as a disposable text suffix. Current systems mostly rely on in-run heuristics such as chronological pruning and tool-output masking, or on final self-summary near a context limit. Heuristics are cheap but blind to future dependencies; summaries preserve narrative state but often hide exact evidence, locators, and editable artifacts. We present Self-GC, where GC denotes self-governing context while deliberately echoing garbage collection: the system does not merely reclaim unused tokens, but governs the lifecycle of agent context objects. Self-GC turns user turns, tool spans, and skill state into indexed objects; asks a side-channel planner to propose fold, mask, and prune actions; and lets the harness enforce recoverable sidecars, safe commit boundaries, and cache-aware commit. On a 33-session Hard Set, Self-GC prunes 43.95% of prefix tokens while leaving 84.85% of future continuations unaffected, compared with no-impact rates of 54.55% to 69.70% for heuristic baselines. On a 332-session production-derived suite, three planner backbones reach no-impact rates of 91.27% to 94.58%, while baselines remain at 77.71% to 87.46%. In production, an online account-level split reduces daytime average input tokens by 10% to 15%, with peak reductions near 20%. These results point to context management as runtime lifecycle control over indexed, recoverable objects rather than post hoc text cleanup.
Chinese Translation
长时间跨度的LLM代理积累了工具结果、文件、计划和用户约束，这些内容过于结构化，无法被视为一次性文本后缀。目前的系统主要依赖于运行中的启发式方法，如时间顺序修剪和工具输出屏蔽，或在上下文限制附近进行最终自我总结。启发式方法成本低廉，但对未来的依赖关系视而不见；总结保留了叙事状态，但往往隐藏了确切的证据、定位器和可编辑的工件。我们提出了自我GC（Self-GC），其中GC表示自我管理上下文，同时故意呼应垃圾回收：该系统不仅仅回收未使用的标记，而是管理代理上下文对象的生命周期。自我GC将用户交互、工具跨度和技能状态转化为索引对象；请求一个侧通道规划器提出折叠、屏蔽和修剪操作；并让系统强制执行可恢复的辅助对象、安全的提交边界和缓存感知提交。在一个33会话的困难集上，自我GC修剪了43.95%的前缀标记，同时对84.85%的未来延续没有影响，而启发式基线的无影响率为54.55%至69.70%。在一个332会话的生产衍生套件中，三个规划器骨干的无影响率达到了91.27%至94.58%，而基线保持在77.71%至87.46%。在生产中，在线账户级别的拆分将白天平均输入标记减少了10%至15%，峰值减少接近20%。这些结果表明，上下文管理作为对索引、可恢复对象的运行时生命周期控制，而不是事后文本清理。
cs.AI / 22 / 2607.00871
Self-Evolving Agents with Anytime-Valid Certificates
具有随时有效证书的自我进化代理
Abstract
Self-evolving agents violate the assumption behind most learning-theoretic guarantees: the data, evaluator, components, and hypothesis space are produced by the policy being updated. We present \textbf{SEA}, an architecture that confines self-modification to a small steering adapter and a versioned harness around a \emph{frozen} base model and admits each modification only through an anytime-valid gate that emits an auditable certificate against a fixed error budget. Five loop controllers compose published guarantees; because such gates can only \emph{select} among behaviors the frozen base already produces, five verifier-in-the-loop mechanisms -- best-of-$N$, micro-step search, self-authored reproduction oracles, search-layer control, and self-repair -- supply the dense, grader-free signal the gates require, computed from the issue text alone. On a $52$-instance SWE-bench Verified subset across four base models, base capability is the dominant, confound-free effect, and on two strong base models a deliberate no-op-composite control isolates the suite's contribution at $+4$ and $+5$ (\textsc{Glm}~5.2 $24\to28$; \textsc{Gpt} $29\to34$, the $65\%$ best), with event logs confirming that its mechanisms fire and prevent regressions. Results are single-run on expensive evaluations; confirming run-to-run variance and adapting the per-task algorithm mix are future work.
Chinese Translation
自我进化代理违反了大多数学习理论保证背后的假设：数据、评估者、组件和假设空间都是由正在更新的策略生成的。我们提出了 extbf{SEA}，一种将自我修改限制在小型引导适配器和围绕 extit{冻结}基础模型的版本化框架中的架构，并且仅通过一个随时有效的门进行每次修改，该门发出针对固定错误预算的可审计证书。五个循环控制器组合了已发布的保证；由于这些门只能在冻结基础模型已经产生的行为中进行 extit{选择}，五个循环内验证机制——最佳中的$N$、微步搜索、自我创作重现神谕、搜索层控制和自我修复——提供了门所需的密集、无评分信号，该信号仅根据问题文本计算。在四个基础模型的$52$实例SWE-bench验证子集中，基础能力是主导的、无混淆的效应，在两个强大的基础模型上，一个故意的无操作复合控制将套件的贡献隔离在$+4$和$+5$（ extsc{Glm}~5.2 $24 o28$； extsc{Gpt} $29 o34$，为$65\%$的最佳），事件日志确认其机制被触发并防止了回归。结果是在昂贵评估上的单次运行；确认运行间的方差并调整每个任务的算法组合是未来的工作。
cs.AI / 23 / 2607.00913
Two AI Metrics Diverged: Will it Make All the Difference?
两种人工智能指标的分歧：这会产生重大影响吗？
Abstract
As exponential compute scaling continues, will the capabilities of frontier AI models outstrip what is accessible to developers on a small fixed budget? Or will capabilities converge, with "meek models inheriting the earth"? Building on Gundlach et al. (2025b), we show that the answer depends on how we value and measure AI capabilities. We discuss conventional performance measures and show that, while validation loss shows a shrinking gap, on other metrics frontier models grow their lead forever. Classifying performance metrics by their functional forms in relation to training (and inference) compute, we provide tight mathematical conditions for determining which metrics favor meek models, and show that bounded performance metrics always do. But careful interpretation of performance metrics is essential: we show that many common bounded metrics have closely-related counterpart metrics that are unbounded (and vice versa). Determining the apt metric in a domain is a prerequisite for policy, since bounded and unbounded metrics may suggest opposing policy responses. If a particular capability -- like software engineering, synthetic biology, or rhetorical persuasiveness -- is unbounded when measured in the terms we care about, frontier-level capability will likely be concentrated in the hands of a few wealthy actors. Conversely, if that capability is instead bounded, frontier-level capabilities proliferate through meek models into the hands of the many.
Chinese Translation
随着计算能力的指数级扩展，前沿人工智能模型的能力是否会超出小型固定预算开发者所能获取的范围？还是能力会趋于一致，形成“温和模型继承大地”的局面？基于Gundlach等人（2025b）的研究，我们表明答案取决于我们如何评估和衡量人工智能能力。我们讨论了传统的性能衡量标准，并展示了尽管验证损失显示出差距在缩小，但在其他指标上，前沿模型的领先地位将永远扩大。通过根据与训练（和推理）计算相关的功能形式对性能指标进行分类，我们提供了严格的数学条件，以确定哪些指标有利于温和模型，并表明有界性能指标总是如此。然而，对性能指标的仔细解读至关重要：我们展示了许多常见的有界指标与其相关的无界指标之间存在紧密关系（反之亦然）。在某一领域确定适当的指标是政策制定的前提，因为有界和无界指标可能会暗示相反的政策反应。如果某一特定能力——如软件工程、合成生物学或修辞说服力——在我们关心的术语中是无界的，那么前沿级别的能力很可能会集中在少数富有参与者手中。相反，如果该能力是有界的，前沿级别的能力将通过温和模型扩散到更多人手中。
cs.AI / 24 / 2607.00924
Graph-Native Reinforcement Learning Enables Traceable Scientific Hypothesis Generation through Conceptual Recombination
图谱原生强化学习通过概念重组实现可追溯的科学假设生成
Abstract
Accelerating materials discovery requires AI systems that can generate scientifically valid hypotheses through multi-step, domain-grounded reasoning. Standard large language models often produce fluent but weakly traceable responses to open-ended materials design problems, making it difficult to determine whether final answers are supported by coherent intermediate reasoning. We develop Graph-PRefLexOR, a family of graph-native reasoning models fine-tuned with Group Relative Policy Optimization (GRPO) to organize reasoning into explicit phases for mechanism exploration, graph construction, pattern extraction, and hypothesis synthesis. This design links neural language generation with symbolic relational structure, enabling causal connections to be constructed, inspected, and reused. On 100 open-ended questions from materials science and mechanics literature, Graph-PRefLexOR achieves 40-65% improvements over corresponding base models, with the largest gains in reasoning traceability. Embedding analyses show broader semantic exploration and approximately 2-3 times greater semantic diversity than baselines. Semantic backtracking and layer-wise hidden-state analyses further show stronger alignment between structured reasoning and final answers. Finally, test-time graph expansion reveals that additional compute primarily increases long-range conceptual recombination within a bounded semantic space, rather than simply expanding semantic coverage. These results establish graph-native reinforcement learning as a pathway toward interpretable AI systems for scientific hypothesis generation in materials design and other scientific applications.
Chinese Translation
加速材料发现需要能够通过多步骤、基于领域的推理生成科学有效假设的人工智能系统。标准的大型语言模型通常会对开放式材料设计问题生成流畅但追溯性较弱的回答，这使得难以确定最终答案是否得到了连贯的中间推理支持。我们开发了Graph-PRefLexOR，一系列经过群体相对策略优化（Group Relative Policy Optimization, GRPO）微调的图谱原生推理模型，以将推理组织成明确的阶段，用于机制探索、图谱构建、模式提取和假设合成。这一设计将神经语言生成与符号关系结构相结合，使因果关系得以构建、检查和重用。在来自材料科学和力学文献的100个开放式问题上，Graph-PRefLexOR在相应的基础模型上实现了40-65%的改进，推理的可追溯性提升最大。嵌入分析显示出更广泛的语义探索和约2-3倍于基线的语义多样性。语义回溯和逐层隐藏状态分析进一步显示出结构化推理与最终答案之间的更强一致性。最后，测试时的图谱扩展表明，额外的计算主要增加了在有限语义空间内的长距离概念重组，而不仅仅是扩展语义覆盖。这些结果确立了图谱原生强化学习作为实现材料设计及其他科学应用中可解释人工智能系统的途径。
cs.AI / 25 / 2607.00972
Bayesian Uncertainty Propagation for Agentic RAG Pipelines: A Proof-of-Concept Study on Multi-Hop Question Answering
面向代理型检索增强生成（RAG）管道的贝叶斯不确定性传播：多跳问答的概念验证研究
Abstract
Trustworthy deployment of Agentic Retrieval-Augmented Generation (RAG) systems requires mechanisms for estimating when multi-stage reasoning pipelines may fail. This paper presents an uncertainty-aware Agentic Retrieval-Augmented Generation (RAG) framework in which planner, evaluator and generator stages produce uncertainty signals derived from semantic divergence and generator self-evaluation. These signals are propagated through a Bayesian Network (BN) to estimate system-level uncertainty and provide node-level indicators of potential failure points across the workflow. The approach is evaluated on StrategyQA and HotpotQA using GPT-3.5-Turbo and GPT-4.1-Nano, with Area Under the Receiver Operating Characteristic Curve (AUROC), Area Under the Accuracy-Rejection Curve (AUARC), Expected Calibration Error (ECE), and Brier Score used to assess discrimination, selective prediction and calibration. Results show that Bayesian propagation is more effective on HotpotQA, where uncertainty accumulates across multi-hop reasoning stages, while StrategyQA exposes limitations caused by miscalibration and unreliable upstream signals. The study positions Bayesian uncertainty propagation as a promising but preliminary mechanism for monitoring Agentic RAG systems, with future validation required in industrial domains such as Offshore Wind (OSW) maintenance decision support.
Chinese Translation
可信的代理型检索增强生成（RAG）系统部署需要机制来评估多阶段推理管道可能失败的情况。本文提出了一种不确定性感知的代理型检索增强生成（RAG）框架，其中规划者、评估者和生成器阶段产生基于语义偏差和生成器自我评估的信号。这些信号通过贝叶斯网络（BN）传播，以估计系统级不确定性，并提供工作流程中潜在故障点的节点级指示。该方法在 StrategyQA 和 HotpotQA 上进行了评估，使用 GPT-3.5-Turbo 和 GPT-4.1-Nano，采用接收者操作特征曲线下面积（AUROC）、准确性-拒绝曲线下面积（AUARC）、期望校准误差（ECE）和布莱尔分数（Brier Score）来评估区分度、选择性预测和校准。结果表明，贝叶斯传播在 HotpotQA 上更为有效，因为不确定性在多跳推理阶段累积，而 StrategyQA 暴露了由误校准和不可靠的上游信号导致的局限性。本研究将贝叶斯不确定性传播定位为监控代理型 RAG 系统的有前景但初步的机制，未来需要在工业领域（如海上风电（OSW）维护决策支持）进行验证。
cs.AI / 26 / 2607.01021
PedNStream: Scalable Network Flow Simulation for Pedestrian Traffic Management
PedNStream：用于行人交通管理的可扩展网络流模拟
Abstract
Large-scale crowd management requires pedestrian simulations that are both computationally efficient and compatible with feedback-based control. However, most open-source tools are either microscopic or not designed for network-scale closed-loop evaluation. This paper presents PedNStream (Pedestrian Network Flow Simulation), an open-source, Python-native simulator for macroscopic pedestrian network loading based on the Link Transmission Model (LTM). The framework extends LTM-based pedestrian models by incorporating stochastic link dynamics that capture diffusion and activity-induced variability, and replaces dynamic user equilibrium route choice with a utility-based formulation suited to uncertain, intervention-driven settings. PedNStream is implemented as a modular framework with built-in controller interfaces for interventions such as gating, flow separation, and route guidance. We evaluate the framework in a staged manner. Synthetic scenarios verify key mechanisms, including queue formation, spillback, congestion dissipation, and adaptive rerouting. Real-network experiments assess large-scale behavior and consistency with observed pedestrian counts. A closed-loop case study demonstrates controller integration, and a runtime analysis quantifies scalability. These results establish PedNStream as an efficient and practical testbed for large-scale pedestrian network simulation and control.
Chinese Translation
大规模人群管理需要既高效计算又兼容基于反馈的控制的行人模拟。然而，大多数开源工具要么是微观的，要么不适用于网络规模的闭环评估。本文提出了PedNStream（行人网络流模拟），这是一个基于链接传输模型（Link Transmission Model, LTM）的开源Python原生模拟器，用于宏观行人网络负载。该框架通过引入捕捉扩散和活动引起的变异性的随机链接动态，扩展了基于LTM的行人模型，并用适合不确定性和干预驱动环境的效用基础公式替代了动态用户均衡路径选择。PedNStream作为一个模块化框架实现，内置控制器接口以支持如闸门、流量分离和路径引导等干预措施。我们以分阶段的方式评估该框架。合成场景验证了关键机制，包括排队形成、溢出、拥堵消散和自适应重路由。真实网络实验评估了大规模行为及与观察到的行人数的一致性。一个闭环案例研究展示了控制器集成，运行时分析量化了可扩展性。这些结果确立了PedNStream作为一个高效且实用的大规模行人网络模拟和控制测试平台。
cs.AI / 27 / 2607.01061
Agentic generation of verifiable rules for deterministic, self-expanding reaction classification
自主生成可验证规则的确定性自扩展反应分类
Abstract
Computer-assisted synthesis planning breaks target molecules into accessible precursors using large libraries of reaction rules that assign each transformation a deterministic, interpretable label. But chemistry is long-tailed, making manual encoding intractable, and existing tools rely on fixed rulesets that cannot adapt to new chemistries. Here we present a fully automated pipeline in which a multi-agent framework of large language models (LLMs) classifies reactions and writes the rules themselves across 665,901 US patent reactions, generating each rule under a verification loop that tests it against the corpus. It expands a standard taxonomy from 68 to 14,073 classes without human curation. With a lightweight fingerprint classifier, it classifies 97.7\% of unseen reactions, matching a leading proprietary classifier while resolving chemistry more finely and extending on demand to chemistry outside its training distribution. The result is a living reactivity database and a general route to turning generative models into reliable, self-expanding symbolic systems.
Chinese Translation
计算机辅助合成规划通过使用大量反应规则库将目标分子分解为可获取的前体，这些规则为每个转化分配一个确定性且可解释的标签。然而，化学反应具有长尾特性，使得手动编码变得不可行，现有工具依赖于固定的规则集，无法适应新的化学反应。在此，我们提出了一种完全自动化的流程，其中一个多智能体框架的大型语言模型（LLMs）对反应进行分类并自行编写规则，涵盖了665,901个美国专利反应，在验证循环中生成每条规则并对其进行语料库测试。该流程将标准分类法从68类扩展至14,073类，无需人工整理。通过轻量级指纹分类器，它对97.7%的未见反应进行分类，匹配领先的专有分类器，同时更细致地解析化学反应，并根据需要扩展到其训练分布之外的化学反应。最终结果是一个动态的反应性数据库，以及将生成模型转变为可靠的自扩展符号系统的一般途径。
cs.AI / 28 / 2607.01084
Can Agents Generalize to the Open World? Unveiling the Fragility of Static Training in Tool Use
代理能否在开放世界中进行泛化？揭示工具使用中静态训练的脆弱性
Abstract
While Large Language Model (LLM) agents demonstrate proficiency in static benchmarks, their deployment in real-world scenarios is hindered by the dynamic nature of user queries, tool sets, and interaction dynamics. To address this generalization gap, we formalize OpenAgent (Tool-Use Agent in Open-World), a problem setting characterized by distributional shifts across query, action, observation, and domain dimensions. To systematically diagnose its impact, we construct a controlled sandbox environment where we define fine-grained environmental shifts across a four-tier hierarchy, Perception, Interaction, Reasoning, and Internalization, and conduct a comprehensive series of experiments. Our analysis yields a series of key insights, demonstrating that agents trained via both Supervised Fine-Tuning(SFT) and Reinforcement Learning suffer from varying degrees of performance degradation when confronting open environmental shifts. Building on these insights, we propose Perturbation-Augmented Fine-Tuning, a disturbance-based intervention strategy for SFT that lays the foundation for enhancing agent robustness and utility in realistic environments. Our code will be released at: https://github. com/LAMDA-NeSy/OpenAgent.
Chinese Translation
尽管大型语言模型（Large Language Model, LLM）代理在静态基准测试中表现出色，但其在现实场景中的应用受到用户查询、工具集和交互动态的动态特性的制约。为了解决这一泛化差距，我们正式提出了OpenAgent（开放世界中的工具使用代理），这是一个以查询、动作、观察和领域维度的分布变化为特征的问题设置。为了系统地诊断其影响，我们构建了一个受控的沙盒环境，在这里我们定义了跨越感知、交互、推理和内化四个层级的细粒度环境变化，并进行了一系列全面的实验。我们的分析得出了一系列关键见解，表明通过监督微调（Supervised Fine-Tuning, SFT）和强化学习（Reinforcement Learning）训练的代理在面对开放环境变化时，性能下降程度各异。在这些见解的基础上，我们提出了扰动增强微调（Perturbation-Augmented Fine-Tuning），这是一种基于干扰的SFT干预策略，为增强代理在现实环境中的鲁棒性和实用性奠定了基础。我们的代码将发布于：https://github.com/LAMDA-NeSy/OpenAgent。
cs.AI / 29 / 2607.01223
Theoria: Rewrite-Acceptability Verification over Informal Reasoning States
Theoria：非正式推理状态下的重写可接受性验证
Abstract
When should an AI system's answer be trusted? Formal proof assistants offer certainty but cannot reach most of the problem distribution; scalar LLM judges offer coverage but produce opaque scores that cannot be audited after the fact and are subject to the same coherence issues as any LLM. We present Theoria, a verification architecture that closes this gap. A candidate solution is rewritten into a sequence of typed state transitions, each licensed by an explicit justification, whether that be a citation, computation, or problem-given fact, and every transition is independently auditable. The foundational invariant is completeness of change: every difference between consecutive proof states must be accounted for, so hidden premises surface as unlicensed mutations rather than passing silently. On HLE-Verified Gold (185 text-only expert problems), Theoria certifies 105 at 91.4% strict precision (Wilson 95% CI [84.5%, 95.4%]). Every certification produces a human readable proof trace in which each step can be independently challenged. Holistic LLM judges achieve comparable precision at matched coverage but fail on different problems (Jaccard 0.14-0.36), making the approaches complementary. On 95 adversarial poisoned proofs across 15 domains, structured judges catch 94.7% versus 83.2% for holistic judging (p= 0.0017). The overall 11.5 pp gap concentrates in hidden premises (90.6% vs. 62.5%, a 28 pp difference) and fabricated citations (100% vs. 90%), the error classes where the formal analysis predicts an advantage; performance is identical on arithmetic and theorem-misapplication errors, where no advantage is predicted. On GPQA Diamond (n= 65), certified precision is 97.1% (Wilson CI [85.1%, 99.5%]).
Chinese Translation
何时应信任人工智能系统的回答？形式化证明助手提供了确定性，但无法覆盖大多数问题分布；标量 LLM 判别器提供了覆盖范围，但产生的不透明分数无法在事后审计，并且面临与任何 LLM 相同的一致性问题。我们提出了 Theoria，这是一种验证架构，旨在填补这一空白。候选解决方案被重写为一系列类型化状态转换，每个转换都由明确的理由支持，无论是引用、计算还是问题给定的事实，并且每个转换都是独立可审计的。基础不变性是变化的完整性：每个连续证明状态之间的差异必须被解释，因此隐藏的前提以未授权的变异形式显现，而不是默默无闻地通过。在 HLE-Verified Gold（185 个仅文本的专家问题）上，Theoria 以 91.4% 的严格精度认证了 105 个（Wilson 95% CI [84.5%，95.4%]）。每个认证都生成一个人类可读的证明追踪，其中每一步都可以独立质疑。整体 LLM 判别器在匹配覆盖率的情况下实现了可比的精度，但在不同问题上失败（Jaccard 0.14-0.36），使得这两种方法互为补充。在 15 个领域的 95 个对抗性污染证明中，结构化判别器捕获了 94.7%，而整体判别为 83.2%（p=0.0017）。总体 11.5 个百分点的差距集中在隐藏前提（90.6% 对比 62.5%，差异为 28 个百分点）和伪造引用（100% 对比 90%），这些错误类别的形式分析预测了优势；在算术和定理误用错误上，表现相同，未预测到优势。在 GPQA Diamond（n=65）上，认证精度为 97.1%（Wilson CI [85.1%，99.5%]）。
cs.AI / 30 / 2607.01224
AutoMem: Automated Learning of Memory as a Cognitive Skill
AutoMem：作为认知技能的记忆自动学习
Abstract
Memory expertise is a learned skill: knowing what to encode, when to retrieve, and how to organize knowledge--a capacity known in cognitive science as metamemory. We bring this perspective to LLMs by treating memory management as a trainable skill. We promote file-system operations to first-class memory actions alongside task actions, letting the model itself decide how to manage its memory. This memory skill improves along two axes: the structure that supports it (prompts, file schemas, action vocabulary), and the proficiency of the model exercising it. Both axes resist manual optimization: episodes in long-horizon tasks run for thousands of steps, and a single memory mistake can hide long before it surfaces, making human review of full trajectories impractical. We introduce AutoMem, a framework that automates both axes. In the first loop, a strong LLM reviews complete agent trajectories and iteratively revises the memory structure that shapes how the agent interacts with its memory files. In the second loop, the agent's own good memory decisions are identified from many episodes and used as training signal to sharpen the model's memory proficiency directly. Across three procedurally generated long-horizon games (Crafter, MiniHack, and NetHack), optimizing memory alone--without modifying the model's task-action behavior--improved the base agent's performance ~2x-4x, bringing a 32B open-weight model competitive with frontier systems such as Claude Opus 4.5 and Gemini 3.1 Pro Thinking. Our results show that memory management is an independently learnable skill, and a high-leverage objective yielding large gains on long-horizon tasks.
Chinese Translation
记忆专长是一种学习技能：知道何时编码、何时检索以及如何组织知识——这一能力在认知科学中被称为元记忆（metamemory）。我们通过将记忆管理视为一种可训练的技能，将这一视角引入大型语言模型（LLMs）。我们将文件系统操作提升为一类与任务操作并列的第一类记忆行为，让模型自主决定如何管理其记忆。这种记忆技能在两个方面得以提升：支持它的结构（提示、文件模式、操作词汇）和模型运用它的熟练程度。这两个方面都抵抗手动优化：在长时间跨度的任务中，情节运行数千步，而一个单一的记忆错误可能在显现之前就隐藏很久，使得对完整轨迹的人工审查变得不切实际。我们引入了AutoMem，一个自动化这两个方面的框架。在第一个循环中，一个强大的LLM审查完整的代理轨迹，并迭代修订塑造代理与其记忆文件交互的记忆结构。在第二个循环中，从多个情节中识别出代理自身的良好记忆决策，并将其用作训练信号，直接提升模型的记忆熟练度。在三个程序生成的长时间跨度游戏（Crafter、MiniHack和NetHack）中，仅优化记忆——而不修改模型的任务操作行为——使基础代理的性能提高了约2倍至4倍，使得一个32B的开放权重模型与前沿系统如Claude Opus 4.5和Gemini 3.1 Pro Thinking具有竞争力。我们的结果表明，记忆管理是一项可以独立学习的技能，并且是一个高杠杆目标，在长时间跨度任务中带来了显著的收益。
cs.CL / 1 / 2607.00006
Persona Without Substrate: Regime-Dependence and the LLM Individuation Problem
无基质的人格：制度依赖性与大型语言模型个体化问题
Abstract
Beckmann & Butlin's (2026) ontological framework for the LLM individuation problem inherits an unargued cross-regime co-reference assumption from the persona-vectors literature: that the same direction picks out the same content under prompt-conditioning, gradient-descent fine-tuning, and inference-time steering. We present four empirical wedges from persona-topology experiments on Qwen3-4B-Instruct and Mistral-7B-Instruct-v0.2 - non-collinearity of prompt-extracted vectors and fine-tune basins; fictional personas displacing the model along real-anchor directions more strongly than real anchors do; contradictory-valenced mixtures biased toward a training-history-determined attractor; and asymmetric compositional algebra under inference-time arithmetic versus fine-tune-time chimera training - that jointly undermine the assumption. We propose regime-indexed individuation: the identity unit for representational content is a (vehicle, regime) pair, not a vehicle alone. Under this framework, Beckmann & Butlin's three candidate positions describe three different regime-internal objects rather than competing for the same referent; the same diagnosis applies to Mollo & Milli\`ere, Chalmers, and Cerullo.
Chinese Translation
Beckmann 和 Butlin（2026）针对大型语言模型个体化问题的本体框架继承了来自人格向量文献中未经论证的跨制度共指假设：即在提示条件、梯度下降微调和推理时引导下，相同的方向指向相同的内容。我们展示了来自 Qwen3-4B-Instruct 和 Mistral-7B-Instruct-v0.2 的人格拓扑实验的四个实证切入点——提示提取向量与微调基域的非共线性；虚构人格沿真实锚点方向对模型的影响强于真实锚点的影响；矛盾情感混合偏向于由训练历史决定的吸引子；以及推理时算术与微调时奇美拉训练下的不对称组合代数——这些共同削弱了该假设。我们提出了制度索引个体化：表征内容的身份单元是一个（载体，制度）对，而不仅仅是一个载体。在这一框架下，Beckmann 和 Butlin 的三个候选位置描述的是三个不同的制度内部对象，而不是争夺同一指称；同样的诊断适用于 Mollo 和 Milli extgrave{e}re、Chalmers 以及 Cerullo。
cs.CL / 2 / 2607.00009
Controllable Narrative Rendering for Enhanced Assisted Writing
可控叙事渲染以增强辅助写作
Abstract
Despite the remarkable proficiency of large language models (LLMs) in basic writing assistance, their utility in creative writing is fundamentally hindered by a persistent binary failure. This issue manifests as an oscillation between safe, surface-level editing, referred to as remedial polishing, and destructive, uncontrolled plot expansion. This dilemma defines a critical trade-off between narrative fidelity and descriptive intensity. We propose Loom, an assisted writing framework grounded in the narratological distinction between story and discourse. Loom employs a three-layer pipeline that operationalizes an intent-centered semiotic chain-of-thought to enforce precise control over narrative intent and rendering density. This architecture separates the generation of perceptual material from syntactic insertion, ensuring that enhancement occurs without violating the original event structure. Our comprehensive evaluation, which includes LLM-based metrics and human assessment, demonstrates that Loom successfully resolves this fundamental tension. Loom achieves the highest overall quality score, yielding substantial gains in factual integrity and descriptive intensity compared to state-of-the-art baselines.
Chinese Translation
尽管大型语言模型（LLMs）在基础写作辅助方面表现出色，但它们在创意写作中的实用性受到一种持续的二元失败的根本性制约。这个问题表现为在安全的、表面层次的编辑（称为补救性润色）与破坏性、失控的情节扩展之间的振荡。这个困境定义了叙事忠实度与描述强度之间的关键权衡。我们提出了Loom，一个基于叙事学中故事与话语区分的辅助写作框架。Loom采用三层管道，运用以意图为中心的符号链思维，来严格控制叙事意图和渲染密度。这一架构将感知材料的生成与句法插入分离，确保增强过程不违反原始事件结构。我们的综合评估，包括基于LLM的指标和人类评估，表明Loom成功解决了这一根本性紧张关系。与最先进的基准相比，Loom在事实完整性和描述强度方面实现了显著提升，获得了最高的整体质量评分。
cs.CL / 3 / 2607.00083
Harnessing the Latent Space: From Steering Vectors to Model Calibrators for Control and Trust
利用潜在空间：从引导向量到模型校准器的控制与信任
Abstract
Language models have changed from unreliable text generators to highly-capable large models with trillions of parameters. Capability increases come hand-in-hand with increases in scale, making understanding the internal representations of models more challenging. Since millions of users increasing rely on language models to interact with external tools or make decisions in medium or high-stakes scenarios, we need to establish control over model behavior and know when to trust model outputs. In this paper, we discuss our contributions on harnessing the latent spaces by proposing steering vectors for control and developing latent space-based model calibrators for trust. Together, our contributions help demystify the latent spaces of language models and offer new insights into how to harness model internals to build more trustworthy language technology.
Chinese Translation
语言模型已经从不可靠的文本生成器转变为具有数万亿参数的高能力大型模型。能力的提升伴随着规模的扩大，使得理解模型的内部表示变得更加困难。由于数百万用户越来越依赖语言模型与外部工具互动或在中高风险场景中做出决策，我们需要对模型行为建立控制，并了解何时可以信任模型输出。本文讨论了我们在利用潜在空间方面的贡献，提出了用于控制的引导向量，并开发了基于潜在空间的模型校准器以增强信任。我们的贡献共同帮助揭示语言模型的潜在空间，并提供了如何利用模型内部构建更可信赖语言技术的新见解。
cs.CL / 4 / 2607.00139
Benchmarking Frontier LLMs on Arabic Cultural and Sociolinguistic Knowledge: A Cross-Evaluation Framework with Human SME Ground Truth
阿拉伯文化与社会语言学知识的前沿大型语言模型基准测试：基于人类专家真实数据的交叉评估框架
Abstract
The cost of human expert evaluation is a principal bottleneck to deploying language models in specialized, high-stakes domains. This is particularly acute for Arabic sociolinguistic knowledge: credible grading requires not only linguistic fluency but deep cultural familiarity that cannot be approximated by surface-level metrics. We address this with a cross-evaluation framework instantiated on two underrepresented Arabic dialect communities: Egyptian and Iraqi Arabic. We contribute 103 validated prompt-rubric pairs (70 Egyptian, 33 Iraqi; 53 Cultural, 50 Linguistic), authored and graded by native-speaker SMEs using penalty-weighted rubrics distinguishing positive content requirements from answer-specific negative error criteria. Three frontier LLMs serve as target models (graded by human SMEs across 302 unique prompt-response pairs), while five frontier LLMs serve as automated judges enforcing a provider-level self-evaluation guard. A dual-metric scheme combining Mean Absolute Deviation (MAD) with Signed Mean Error separates directional grading bias from symmetric noise. Across 1,307 judge evaluations: GPT-5.4 is the most reliable judge (MADj = 10.21 pp, Signed Error = -1.12%); four of five judges show systematic leniency (+2.01% to +6.56%); Cultural tasks are harder to grade than Linguistic tasks for all judges (MAD gap 1.83-4.78 pp); and models substantially outperform on Egyptian prompts compared to Iraqi prompts. However, given leniency differences between Iraqi and Egyptian SMEs, we cannot solely attribute this gap to model knowledge. We therefore emphasize findings that do not assume identical leniency across human graders. Across all samples, implicit cultural reasoning -- requiring models to simulate native-speaker judgment rather than rely on lexical verification -- emerges as the primary failure mode for automated grading across all judge models.
Chinese Translation
人类专家评估的成本是将语言模型部署在专业高风险领域的主要瓶颈。这在阿拉伯社会语言学知识方面尤为突出：可信的评分不仅需要语言流利性，还需要深厚的文化熟悉度，而这种熟悉度无法通过表面指标来近似。我们通过一个交叉评估框架来解决这个问题，该框架基于两个代表性不足的阿拉伯方言社区：埃及阿拉伯语和伊拉克阿拉伯语。我们贡献了103对经过验证的提示-评分标准对（70个埃及，33个伊拉克；53个文化，50个语言），由母语专家（SME）使用惩罚加权评分标准创作和评分，这些标准区分了积极内容要求和特定答案的负错误标准。三种前沿大型语言模型（LLMs）作为目标模型（由人类专家在302个独特的提示-响应对中评分），而五种前沿LLMs作为自动评审者，执行提供者级别的自我评估保护。一个双指标方案结合了平均绝对偏差（MAD）和符号平均误差，将方向性评分偏差与对称噪声分开。在1,307次评审中：GPT-5.4是最可靠的评审者（MADj = 10.21 pp，符号误差 = -1.12%）；五个评审者中有四个表现出系统性的宽容（+2.01%至+6.56%）；对于所有评审者而言，文化任务的评分难度高于语言任务（MAD差距1.83-4.78 pp）；模型在埃及提示上的表现显著优于伊拉克提示。然而，考虑到伊拉克和埃及专家之间的宽容差异，我们不能单纯将这一差距归因于模型知识。因此，我们强调不假设人类评分者之间宽容相同的发现。在所有样本中，隐性文化推理——要求模型模拟母语者的判断，而不是依赖词汇验证——成为所有评审模型中自动评分的主要失败模式。
cs.CL / 5 / 2607.00143
Hate Speech Detection in Turkish and Arabic Languages: A Comprehensive Study
土耳其语和阿拉伯语中的仇恨言论检测：一项综合研究
Abstract
Online hate speech has been linked to a global rise in violence against minorities, including incidents such as mass shootings, lynchings, and ethnic cleansing. Societies grappling with this issue, particularly when hate speech targets specific groups based on religion, race, ethnicity, culture, nationality, or migration status, face the challenge of balancing freedom of expression with the need for effective content moderation on widely used online platforms. In response to this challenge, we introduce a comprehensive hate speech dataset covering five distinct topics in Turkish: refugees, the Israel-Palestine conflict, anti-Greek sentiment in Turkey, ethnic or religious communities (Alevis, Armenians, Arabs, Jews, and Kurds), and LGBTI+, alongside one topic in Arabic (refugees). In addition, we develop state-of-the-art BERT-based models to address multiple dimensions of hate speech analysis, including hate category classification, hate intensity prediction, target identification, and hate speech span detection, enabling a comprehensive understanding of hateful content in online discourse.
Chinese Translation
在线仇恨言论与全球针对少数群体的暴力上升有关，包括大规模枪击、私刑处决和种族灭绝等事件。面对这一问题的社会，尤其是在仇恨言论针对特定宗教、种族、民族、文化、国籍或移民身份的群体时，面临着在言论自由与有效内容审核之间取得平衡的挑战。为应对这一挑战，我们引入了一个涵盖五个不同主题的综合仇恨言论数据集，主题包括土耳其语中的难民、以色列-巴勒斯坦冲突、土耳其的反希腊情绪、民族或宗教社群（阿列维派、亚美尼亚人、阿拉伯人、犹太人和库尔德人）以及LGBTI+，同时还包括一个阿拉伯语主题（难民）。此外，我们开发了基于BERT的最先进模型，以解决仇恨言论分析的多个维度，包括仇恨类别分类、仇恨强度预测、目标识别和仇恨言论范围检测，从而实现对在线话语中仇恨内容的全面理解。
cs.CL / 6 / 2607.00158
Readable but Not Controllable: Neuron-Level Evidence for Medical LLM Hallucination
可读但不可控：医学大语言模型幻觉的神经元级证据
Abstract
Hallucination remains one of the central obstacles to deploying medical LLMs. Yet, even when hallucination can be detected, it is still unclear whether the internal representations associated with it can be used for control rather than detection alone. Using four open-source models across a suite of medical question-answering datasets, we show that a simple, carefully conditioned probe can reliably detect hallucination, with AUROC scores between 0.77 and 0.86 in our case. We further show that this signal is distributed and redundant rather than narrowly localized. Systematically selected neurons outperform random neurons only at very small subset sizes, whereas random subsets of a few hundred neurons recover nearly the full signal, and low-dimensional random projections preserve most of the detection performance. Beyond detection, we test whether this representation is causally actionable. Across 16 model--dataset combinations, our results reveal a sharp gap between decodability and controllability. The same internal structure that makes hallucination easy to detect does not translate into reliable neuron-level control. These findings show that medical hallucination seems to be readily visible in internal activations, but not easily corrected by steering the neurons most associated with it. More broadly, our results suggest that hallucination mitigation is not simply a matter of identifying the right neurons, and point to a deeper separation between what representations reveal and what they allow us to change.
Chinese Translation
幻觉仍然是部署医学大语言模型（LLMs）的主要障碍之一。然而，即使幻觉可以被检测到，仍不清楚与之相关的内部表征是否可以用于控制而不仅仅是检测。通过使用四个开源模型和一系列医学问答数据集，我们展示了一种简单且经过仔细调节的探测器可以可靠地检测幻觉，在我们的案例中，AUROC得分介于0.77到0.86之间。我们进一步表明，这一信号是分布式和冗余的，而不是狭窄局部化的。系统选择的神经元在非常小的子集大小上优于随机神经元，而随机选择的几百个神经元几乎恢复了全部信号，低维随机投影保留了大部分检测性能。除了检测，我们还测试了这一表征是否具有因果可操作性。在16个模型-数据集组合中，我们的结果揭示了可解码性与可控性之间的明显差距。使幻觉易于检测的相同内部结构并未转化为可靠的神经元级控制。这些发现表明，医学幻觉在内部激活中似乎很容易被观察到，但通过引导与其最相关的神经元并不容易纠正。更广泛地说，我们的结果表明，幻觉缓解不仅仅是识别正确神经元的问题，并指出了表征所揭示的内容与我们能够改变的内容之间的更深层次的分离。
cs.CL / 7 / 2607.00159
Identifying and Resolving Pitfalls of Knowledge-Based VQA Benchmarks: Auditing, Repairing, and Augmenting
识别与解决基于知识的视觉问答基准的陷阱：审计、修复与增强
Abstract
Knowledge-Based Visual Question Answering (KB-VQA) aims to evaluate whether Visual Language Models (VLMs) can retrieve, ground, and reason over external structured knowledge beyond visual evidence. In practice, answer accuracy is widely adopted as the primary evaluation metric, implicitly treating correctness as a proxy for knowledge-grounded reasoning. However, for existing KB-VQA benchmarks, this proxy relies on critical assumptions that are often overlooked and rendered unreliable by benchmark issues: annotated answer must be derivable from the associated knowledge base, question must be well-posed with sufficient constraints, and visual setting must meaningfully require grounded disambiguation. In this work, we show that these assumptions are systematically violated in existing KB-VQA benchmarks. Our audit reveals substantial instances with missing or contradicted answers and underspecified questions that render accuracy a misleading metric. Furthermore, we find that existing datasets rely on visually trivial, single-entity scenes that bypass the need for sophisticated visual-to-knowledge mapping. We demonstrate that even with controlled architectures, these flaws lead to distorted model rankings and overestimations of reasoning capabilities. To address this, we introduce (1) a principled audit-and-repair protocol that restores answer derivability and question clarity, and (2) a controlled multi-entity augmentation protocol that introduces visual ambiguity to challenge initial retrieval and grounded reasoning. Re-evaluation under corrected and augmented settings yields markedly different performance trends. Our findings call for rethinking evaluation protocols and designing more interaction-aware KB-VQA benchmarks that prioritize verifiable reasoning over simple matching.
Chinese Translation
基于知识的视觉问答（KB-VQA）旨在评估视觉语言模型（VLMs）是否能够检索、定位并推理超出视觉证据的外部结构化知识。在实践中，答案准确性被广泛采用作为主要评估指标，隐含地将正确性视为知识基础推理的代理。然而，对于现有的KB-VQA基准，这一代理依赖于一些关键假设，这些假设常常被忽视，并因基准问题而变得不可靠：标注的答案必须能够从相关知识库中推导出，问题必须具有足够约束的良好表述，视觉设置必须有意义地要求基于知识的消歧义。在本研究中，我们展示了这些假设在现有KB-VQA基准中被系统性违反。我们的审计揭示了大量缺失或矛盾答案和不明确问题的实例，这使得准确性成为一个误导性的指标。此外，我们发现现有数据集依赖于视觉上简单的单实体场景，这绕过了复杂的视觉到知识映射的需求。我们证明，即使在受控架构下，这些缺陷也导致模型排名扭曲和推理能力的高估。为了解决这个问题，我们引入了（1）一个原则性的审计与修复协议，恢复答案的可推导性和问题的清晰性，以及（2）一个受控的多实体增强协议，引入视觉模糊性以挑战初始检索和基于知识的推理。在经过修正和增强的设置下重新评估，性能趋势显著不同。我们的发现呼吁重新思考评估协议，并设计更具交互意识的KB-VQA基准，优先考虑可验证的推理而非简单匹配。
cs.CL / 8 / 2607.00171
ALEE: Any-Language Evaluation of Embeddings via English-Centric Minimal Pairs
ALEE：通过以英语为中心的最小对的任何语言嵌入评估
Abstract
Text embeddings are standard for semantic similarity tasks, yet their evaluation remains an open challenge. Current benchmarks are static, cover only a limited set of languages, are often domain-specific, susceptible to overfitting, and poorly representative of low-resource languages. To address these limitations, we introduce ALEE, a framework that extends Sentence Smith (Li et al., 2025) to the cross-lingual and paragraph level. ALEE uses Abstract Meaning Representations (AMR) to generate English minimal pairs with controlled, fine-grained semantic shifts, which are paired with translations in target languages. This approach enables targeted diagnostics for models in any language with English parallel data. We conduct a large-scale empirical study across a diverse set of embedding models and 275+ languages spanning three parallel datasets. On ALEE, performance varies substantially across languages, text lengths, and linguistic phenomena, exposing persistent gaps in cross-lingual semantic representation that track language prevalence in training resources and subword tokenization. We release ALEE at https://github.com/Andrian0s/any-lang-embed-eval
Chinese Translation
文本嵌入在语义相似性任务中是标准做法，但其评估仍然是一个开放的挑战。目前的基准测试是静态的，仅涵盖有限的语言集，通常是特定领域的，容易过拟合，并且对低资源语言的代表性较差。为了解决这些局限性，我们提出了 ALEE，一个将 Sentence Smith (Li et al., 2025) 扩展到跨语言和段落级别的框架。ALEE 使用抽象意义表示 (AMR) 生成具有可控、细粒度语义变化的英语最小对，并与目标语言的翻译配对。这种方法使得在任何语言中使用英语平行数据对模型进行有针对性的诊断成为可能。我们在多种嵌入模型和涵盖275+种语言的三个平行数据集上进行了大规模的实证研究。在 ALEE 上，性能在不同语言、文本长度和语言现象之间差异显著，揭示了跨语言语义表示中持续存在的差距，这些差距与训练资源的语言普及度和子词标记化相关。我们在 https://github.com/Andrian0s/any-lang-embed-eval 发布了 ALEE。
cs.CL / 9 / 2607.00185
Structural Pattern Mining in Inka Khipus: Unsupervised Clustering, Provenance Classification, and a Computational Validation of the Santa Valley Match
印加基普斯中的结构模式挖掘：无监督聚类、来源分类及圣谷匹配的计算验证
Abstract
Khipus--knotted cord devices--were the primary recording medium of the Inka Empire (c. 1400-1532 CE), yet their system remains undeciphered. We present a reproducible machine-learning pipeline applied to the Open Khipu Repository (OKR), a public database of 619 khipus comprising 54,403 cords and 110,677 knots. We engineer 27 structural features per khipu and apply (i) unsupervised clustering via UMAP and HDBSCAN, recovering three structurally distinct groups (silhouette = 0.769); (ii) supervised provenance classification via gradient boosting, reaching F1 = 0.86 for the Inka Late Horizon imperial style; and (iii) SHAP-based interpretability, which identifies cord twist direction as the dominant structural discriminator of imperial khipus. We further report two findings of methodological interest. First, one cluster is dominated not by a geographic region but by nineteenth-century European museum collections, indicating that colonial acquisition and recording practices are structurally encoded in the corpus. Second, we provide an independent computational verification of the recto/verso (moiety) structure of the six Santa Valley khipus reported by Medrano and Urton (2018), reproducing both the aggregate attachment ratio and the identification of the single mixed specimen--using only the public OKR database, without physical access to the objects. We additionally report a negative result: knot-type sequence order, encoded as n-grams, adds no provenance signal beyond aggregate features. All code and data are openly available.
Chinese Translation
基普斯（Khipus）——一种结绳装置——是印加帝国（公元1400-1532年）主要的记录媒介，但其系统仍未被破译。我们提出了一种可重复的机器学习流程，应用于开放基普斯数据库（Open Khipu Repository, OKR），该数据库包含619个基普斯，涵盖54,403根绳索和110,677个结。我们为每个基普斯设计了27个结构特征，并应用了（i）通过UMAP和HDBSCAN进行的无监督聚类，恢复出三个结构上明显不同的群体（轮廓系数 = 0.769）；（ii）通过梯度提升进行的监督来源分类，印加晚期帝国风格的F1值达到0.86；以及（iii）基于SHAP的可解释性分析，识别出绳索扭转方向是帝国基普斯的主要结构区分因素。我们进一步报告了两个方法论上的重要发现。首先，一个聚类并不是由地理区域主导，而是由19世纪欧洲博物馆收藏主导，这表明殖民收购和记录实践在该语料库中被结构性编码。其次，我们提供了对Medrano和Urton（2018）报告的六个圣谷基普斯的正面/反面（moiety）结构的独立计算验证，重现了整体附着比率和单一混合样本的识别——仅使用公共的OKR数据库，而无需对物品进行实物访问。我们还报告了一个负面结果：作为n-grams编码的结类型序列顺序，未能提供超出整体特征的来源信号。所有代码和数据均可公开获取。
cs.CL / 10 / 2607.00208
SLIM-RL: Risk-Budgeted Random-Masking RL for Diffusion LLMs Without Trajectory Slicing
SLIM-RL：无轨迹切片的风险预算随机掩蔽强化学习用于扩散大语言模型
Abstract
Reinforcement learning for diffusion large language models (dLLMs) has largely moved to trajectory-aware methods. The current state of the art, TraceRL, holds that random masking is mismatched with the model's inference trajectory, and it reconstructs that trajectory during training by slicing each rollout into up to K/s trajectory-aligned training samples, a cost that grows with the block size K. We show that this mismatch can be mitigated without reconstructing the trajectory. Our method, SLIM-RL, bounds the commit risk of each rollout step with a tau-budget decoder, reducing aggregate commit risk in the training data. During optimization, SLIM-RL trains on these risk-controlled rollouts with a trace-free random-masking objective that adapts variance-reduction tools, combining sequence-level importance sampling, deterministic quadrature over masking levels under a mean-preserving, monotonically decreasing per-block mask schedule that we introduce. On SDAR-4B, SLIM-RL matches TraceRL's best MATH500 accuracy on only 0.46x its training samples at block size 16, improving over TraceRL by 6.32% on MATH500 and 11.05% on GSM8K under matched dynamic sampling. At block size 4, the 4B SLIM-RL surpasses the larger LLaDA-8B and Dream-7B dLLMs on math, exceeding LLaDA-8B by 10.76% on MATH500 while staying below the autoregressive Qwen2.5-7B. On code, it improves over TraceRL by 4.20% on MBPP and 3.65% on HumanEval. The tau-budget decoder transfers training-free across LLaDA, Dream, and SDAR. The source code is available at https://github.com/laolaorkkkkk/SLIM-RL .
Chinese Translation
针对扩散大语言模型（dLLMs）的强化学习已经大多转向轨迹感知的方法。目前的最先进方法TraceRL认为随机掩蔽与模型的推理轨迹不匹配，并通过将每次回滚切分为最多K/s个轨迹对齐的训练样本来重建该轨迹，这一成本随着块大小K的增加而增加。我们展示了在不重建轨迹的情况下可以减轻这种不匹配。我们的方法SLIM-RL使用tau预算解码器限制每次回滚步骤的承诺风险，从而减少训练数据中的整体承诺风险。在优化过程中，SLIM-RL在这些风险控制的回滚上进行训练，采用无轨迹的随机掩蔽目标，结合序列级重要性采样、在我们引入的均值保持、单调递减的每块掩蔽调度下对掩蔽级别进行的确定性求积。 在SDAR-4B上，SLIM-RL在块大小为16时仅使用0.46倍的训练样本便达到了TraceRL在MATH500上的最佳准确率，比TraceRL在MATH500上提高了6.32%，在GSM8K上提高了11.05%。在块大小为4时，4B SLIM-RL在数学任务上超越了更大的LLaDA-8B和Dream-7B dLLMs，在MATH500上超过LLaDA-8B达10.76%，同时低于自回归的Qwen2.5-7B。在代码任务上，它在MBPP上比TraceRL提高了4.20%，在HumanEval上提高了3.65%。tau预算解码器在LLaDA、Dream和SDAR之间实现了无训练转移。源代码可在https://github.com/laolaorkkkkk/SLIM-RL获取。
cs.CL / 11 / 2607.00250
LV-ROVER: Multi-Stream Tesseract Voting for Maltese Paragraph OCR
LV-ROVER：用于马耳他段落OCR的多流Tesseract投票
Abstract
Maltese has decent text corpora and pretrained language models, but, like many languages outside the handful with large OCR benchmarks, only a single known real labelled PDF corpus for OCR training, 57 page, far below what paragraph-level training needs: low-resource for OCR specifically. With no real corpus to train on at scale, we built a synthetic training pipeline and a 5-stream Tesseract LV-ROVER ensemble, and report results on a 422-paragraph benchmark against a fine-tuned-Tesseract baseline of character error rate (CER) 0.0234. Ensemble recognition alone improves CER by 44 percent, to 0.01317; a five-stage post-processing chain brings the full pipeline to CER 0.00700, a 70 percent reduction. Most of that chain is typographic normalisation, but one stage recovers misread diacritics rather than aligning punctuation, so we report it as a recognition gain rather than folding the whole chain under one label. We treat the 44 percent figure as the portable estimate of what the recogniser learned, and the 70 percent figure as specific to this benchmark's label convention.
Chinese Translation
马耳他语拥有相当不错的文本语料库和预训练语言模型，但与许多拥有大型OCR基准的语言相比，它仅有一个已知的真实标注PDF语料库用于OCR训练，只有57页，远低于段落级训练所需的规模：在OCR领域属于低资源语言。由于没有真实的语料库可供大规模训练，我们构建了一个合成训练管道和一个5流Tesseract LV-ROVER集成，并在一个包含422段落的基准测试上报告结果，与经过微调的Tesseract基线的字符错误率（CER）0.0234进行比较。仅集成识别就将CER提高了44%，降至0.01317；而五阶段的后处理链将整个管道的CER降低至0.00700，减少了70%。该链的大部分是排版规范化，但其中一个阶段恢复了错误读取的变音符号，而不是对齐标点，因此我们将其报告为识别增益，而不是将整个链归为一个标签。我们将44%的数据视为识别器学习的可转移估计，将70%的数据视为特定于该基准标签约定的结果。
cs.CL / 12 / 2607.00274
SEFORA: Student Essays with Feedback Corpus and LLM Feedback Evaluation Framework
SEFORA：学生论文反馈语料库与LLM反馈评估框架
Abstract
Effective writing feedback is among the strongest drivers of student learning, yet producing it at scale is labor-intensive. LLMs offer a natural path to scaling writing support, but two gaps stand in the way: few public corpora capture how instructors actually deliver feedback in real classrooms, and no reliable method measures whether generated feedback aligns with what an instructor would write. We address both. SEFORA is a public corpus pairing instructor inline feedback with assignment prompts, rubrics, scores, and multi-draft revisions across various college writing genres, comprising 564 drafts and 8,240 instructor annotations. UniMatch is a reference-based evaluation framework for open-ended generation: it segments feedback into feedback units, scores their semantic correspondence under instructor-derived criteria, and aligns them via optimal matching to yield interpretable precision, recall, and F1. Across 74 experimental configurations spanning multiple LLMs, no setting exceeds 0.4 F1. UniMatch reveals that models struggle to identify the feedback instructors would prioritize, and performance degrades as models generate more.
Chinese Translation
有效的写作反馈是推动学生学习的最强驱动力之一，但大规模生成反馈的过程劳动密集。大型语言模型（LLMs）提供了一条自然的路径来扩展写作支持，但有两个障碍：少有公共语料库能够捕捉教师在真实课堂中实际提供反馈的方式，而没有可靠的方法来衡量生成的反馈是否与教师可能写的内容一致。我们解决了这两个问题。SEFORA是一个公共语料库，将教师的在线反馈与作业提示、评分标准、分数和多稿修订配对，涵盖各种大学写作体裁，包含564份草稿和8,240条教师注释。UniMatch是一个基于参考的开放式生成评估框架：它将反馈分割为反馈单元，根据教师制定的标准对其语义对应性进行评分，并通过最优匹配对其进行对齐，以产生可解释的精确度、召回率和F1值。在涵盖多种LLM的74个实验配置中，没有任何设置的F1值超过0.4。UniMatch揭示了模型在识别教师优先考虑的反馈方面存在困难，且随着模型生成内容的增多，性能下降。
cs.CL / 13 / 2607.00339
TRACE: State-Aware Query Processing over Temporal Evidence Graphs for Conversational Data
TRACE：针对对话数据的时态证据图的状态感知查询处理
Abstract
Conversational data is increasingly used as a persistent source of user state for long-running assistants and AI agents. However, querying this data remains challenging because conversations naturally evolve: plans are revised, preferences change, and later messages frequently supersede or contradict earlier information. Existing long-memory pipelines largely treat memories as independent text or vector objects. This approach often retrieves semantically similar but stale evidence, offering limited support for state-aware reasoning. To address this problem, we present TRACE, a query processing framework over temporal evidence graphs for evolving conversational data. TRACE models conversations as a hierarchical graph spanning events, sessions, and topics, enriched with typed temporal, causal, update, and contradiction relations. Crucially, the framework maintains validity annotations so obsolete facts remain accessible for historical queries but are discounted for current-state answers. At query time, TRACE combines vector-based note retrieval with graph-guided evidence search, generating validity-aware support paths and a hybrid context for answer generation. This design separates lexical recall from evidence reconstruction, enabling bounded query-time reasoning over long conversational histories. Experiments on long-conversation query-answering (QA) benchmarks show that TRACE improves temporal and multi-hop reasoning, with ablations highlighting the importance of hierarchy, update-aware seeding, and path-grounded evidence.
Chinese Translation
对话数据越来越多地被用作长时间运行的助手和人工智能代理的用户状态的持久来源。然而，查询这些数据仍然具有挑战性，因为对话自然会发展：计划被修订，偏好发生变化，后来的消息经常取代或与早期信息相矛盾。现有的长记忆管道在很大程度上将记忆视为独立的文本或向量对象。这种方法往往检索到语义相似但过时的证据，提供的状态感知推理支持有限。为了解决这个问题，我们提出了TRACE，一个针对不断发展的对话数据的时态证据图的查询处理框架。TRACE将对话建模为一个层次图，涵盖事件、会话和主题，并丰富了类型化的时态、因果、更新和矛盾关系。关键是，该框架维护有效性注释，使得过时的事实仍然可以用于历史查询，但在当前状态答案中被折扣。在查询时，TRACE结合了基于向量的笔记检索与图引导的证据搜索，生成有效性感知的支持路径和混合上下文以生成答案。这一设计将词汇回忆与证据重建分开，使得在长对话历史上进行有限的查询时间推理成为可能。在长对话查询回答（QA）基准上的实验表明，TRACE改善了时态和多跳推理，消融实验突出了层次结构、更新感知种子和路径基础证据的重要性。
cs.CL / 14 / 2607.00341
DiscoLoop: Looping Discrete Embeddings and Continuous Hidden States for Multi-hop Reasoning
DiscoLoop：用于多跳推理的离散嵌入和连续隐藏状态的循环
Abstract
Large language models achieve strong performance on many reasoning tasks when allowed to externalize intermediate steps as Chain-of-Thought (CoT). However, many questions require the model to internalize the multi-step reasoning within a single forward pass before generating the answer. We study this challenge through two-hop reasoning, a representative task where the model must compose multiple pieces of parametric knowledge within a single forward pass. Standard non-recurrent Transformers suffer from a depth-local storage problem: facts learned in earlier layers are unavailable where second-hop retrieval happens. We found that Looped Transformers mitigate this issue by reusing the same memory, but still generalize imperfectly. We show that the remaining bottleneck is representational. In the two-hop reasoning task, the first loop often makes the correct bridge entity nearly perfectly decodable, yet the corresponding hidden state remains poorly aligned with the bridge token embedding. Surprisingly, an easy training-free realignment intervention nearly closes the generalization gap. Building upon this insight, we propose DiscoLoop, a looping architecture whose recurrence carries both a discrete embedding channel and a continuous hidden-state channel. DiscoLoop achieves near-perfect accuracy with substantially fewer training steps across symbolic and synthetic-language multi-hop reasoning tasks. When applied to real-world pretraining, DiscoLoop attains lower training loss and stronger benchmark performance than looped-transformer baselines, suggesting that the mixed-channel design transfers to practical language modeling.
Chinese Translation
大型语言模型在许多推理任务中表现出色，当它们能够将中间步骤外化为思维链（Chain-of-Thought, CoT）时。然而，许多问题要求模型在生成答案之前，将多步推理内化为单次前向传播。我们通过二跳推理这一代表性任务研究这一挑战，在该任务中，模型必须在单次前向传播中组合多个参数知识片段。标准的非递归变换器在深度局部存储方面存在问题：在较早层学习的事实在第二跳检索发生时不可用。我们发现，循环变换器通过重用相同的内存来缓解此问题，但仍然无法完美泛化。我们表明，剩余的瓶颈是表征性的问题。在二跳推理任务中，第一个循环通常使得正确的桥接实体几乎可以完美解码，但相应的隐藏状态与桥接标记嵌入之间的对齐仍然较差。令人惊讶的是，一种简单的无训练重新对齐干预几乎消除了泛化差距。在此基础上，我们提出了DiscoLoop，这是一种循环架构，其递归同时携带离散嵌入通道和连续隐藏状态通道。DiscoLoop在符号和合成语言的多跳推理任务中以显著更少的训练步骤实现了近乎完美的准确性。当应用于现实世界的预训练时，DiscoLoop的训练损失更低，基准性能更强于循环变换器基线，表明混合通道设计能够迁移到实际语言建模中。
cs.CL / 15 / 2607.00368
Beyond Perplexity: A Behavioral Evaluation Framework for Deployment-Memory Claims in LLM Test-Time Training
超越困惑度：大规模语言模型测试时训练中部署记忆声明的行为评估框架
Abstract
Large language model test-time training (TTT) is often evaluated through local proxy metrics: models are updated on recent tokens, retrieved context, target-domain data, or verifiable task attempts, and then judged by perplexity, future-token loss, long-context performance, or reward. These metrics are well matched to claims about stream adaptation, domain adaptation, context compression, and reward-backed test-time improvement. They are weaker evidence, however, for a capability that TTT results are increasingly used to motivate: deployed assistant memory, personalization, or sparse post-deployment learning, which instead requires behavioral evidence such as later recall, paraphrase robustness, retention, locality, conflict handling, and use in downstream actions after the original support context is removed. We introduce a behavioral evaluation framework that calibrates TTT memory claims to the evidence that supports them. It has two components: a claim-calibrated evidence ladder that separates stream/domain adaptation, bridge internalization, and deployment-time behavioral learning; and an evaluation protocol with matched explicit-memory baselines and mutually exclusive failure categories. We validate the framework by auditing recent TTT and memory-adjacent work and by instantiating it as a controlled diagnostic in which, in a sparse nonce-fact setting, one-step LoRA updates lower support and answer loss across three Qwen3 model scales while generated free-form recall stays at zero, exposing a measurable gap between proxy improvement and deployment behavior. The framework gives authors and evaluators a concrete standard for aligning TTT memory claims with the evidence actually reported.
Chinese Translation
大规模语言模型的测试时训练（TTT）通常通过局部代理指标进行评估：模型在最近的标记、检索的上下文、目标领域数据或可验证的任务尝试上进行更新，然后通过困惑度、未来标记损失、长上下文性能或奖励进行判断。这些指标与关于流适应、领域适应、上下文压缩和基于奖励的测试时改进的声明相匹配。然而，它们对于TTT结果越来越多用于激励的能力的证据较弱：已部署助手的记忆、个性化或稀疏的后部署学习，这些能力需要行为证据，例如后续回忆、释义稳健性、保留、局部性、冲突处理以及在原始支持上下文被移除后在下游行动中的使用。我们提出了一个行为评估框架，将TTT记忆声明与支持它们的证据进行校准。该框架包含两个组成部分：一个声明校准的证据阶梯，区分流/领域适应、桥接内化和部署时行为学习；以及一个评估协议，具有匹配的显性记忆基线和互斥的失败类别。我们通过审计近期的TTT和记忆相关工作来验证该框架，并通过在稀疏的非事实设置中实例化它作为一种受控诊断，其中，一步LoRA更新在三个Qwen3模型规模上降低了支持和答案损失，而生成的自由形式回忆保持为零，揭示了代理改进与部署行为之间的可测量差距。该框架为作者和评估者提供了一个具体标准，以将TTT记忆声明与实际报告的证据对齐。
cs.CL / 16 / 2607.00415
A Mechanistic View of Authority Hierarchy in LLM Sycophancy
大型语言模型中权威等级的机制视角
Abstract
Authority bias poses a critical safety concern in language models: models systematically prioritize social cues from authority figures over factual consistency, swaying their answers based on source credibility rather than evidence. We mechanistically investigate this phenomenon using a controlled medical QA setting, where hints suggesting incorrect answers are attributed to personas of varying expertise. Across Llama-3.1-8B, Qwen3-8B, and Gemma-2-9B, we find that models respond in a graded manner proportional to perceived authority, a hierarchy that is never explicitly prompted but emerges from training. Logit lens analysis and linear/non-linear probing localize this effect to a critical late layer where correct answer representations are actively erased, an erasure that scales with authority level, resists mean vector intervention, and is only partially reversible through chain-of-thought reasoning. Our findings suggest that authority-induced sycophancy is not a surface-level output bias but mechanistic knowledge erasure, a precise, layer-localized overwriting of correct internal representations by high-status authority signals.
Chinese Translation
权威偏见在语言模型中构成了一个关键的安全隐患：模型系统性地优先考虑来自权威人物的社会线索，而非事实一致性，根据来源可信度而非证据来影响其回答。我们在一个受控的医学问答环境中机制性地研究了这一现象，其中暗示错误答案的提示被归因于不同专业水平的人物。通过对 Llama-3.1-8B、Qwen3-8B 和 Gemma-2-9B 的分析，我们发现模型的响应与感知的权威程度成正比，这种等级关系并未被明确提示，而是从训练中自发产生。Logit lens 分析和线性/非线性探测将这一效应定位于一个关键的晚期层次，在该层次中，正确答案的表示被主动抹去，这种抹去程度与权威水平成正比，抵抗均值向量干预，并且仅通过链式思维推理部分可逆。我们的研究结果表明，权威引发的谄媚并不是表面层次的输出偏见，而是机制性知识的抹去，是高地位权威信号对正确内部表示的精确、层次局部的覆盖。
cs.CL / 17 / 2607.00418
Speech Playground: An Interactive Tool for Speech Analysis and Comparison
语音游乐场：一个互动的语音分析与比较工具
Abstract
This paper presents Speech Playground, an interactive speech visualization and comparison tool. While existing tools such as Praat are excellent, it can be cumbersome to integrate them with modern deep learning representations and use them for comparison. Speech Playground addresses this by combining a Python backend with a web-based frontend for interactive exploration of multiple feature types, including continuous, discrete, and variable-length representations. It includes TextGrid and forced alignment support together with configurable distance and alignment settings for visual and auditory comparison. Speech Playground is intended for use in speech research, representation validation, and computer-aided pronunciation training (CAPT)-oriented experimentation.
Chinese Translation
本文介绍了语音游乐场（Speech Playground），一个互动的语音可视化与比较工具。尽管现有工具如Praat表现优异，但将其与现代深度学习表示结合并用于比较时可能会显得繁琐。语音游乐场通过将Python后端与基于网页的前端结合，解决了这一问题，支持对多种特征类型的互动探索，包括连续、离散和可变长度的表示。它还支持TextGrid和强制对齐，并提供可配置的距离和对齐设置，以便进行视觉和听觉比较。语音游乐场旨在用于语音研究、表示验证以及计算机辅助发音训练（CAPT）相关的实验。
cs.CL / 18 / 2607.00423
Selective Test-Time Debiasing for CLIP via Reward Gating
通过奖励门控实现CLIP的选择性测试时去偏见
Abstract
Vision language models (VLMs) demonstrate strong zero-shot performance, but often perpetuate social stereotypes in person-centric queries, yielding skewed demographic distributions. Current debiasing methods apply uniform bias corrections across all input queries regardless of their bias sensitivity, creating a fundamental fairness--utility trade-off. Strong debiasing distorts semantically meaningful information in bias-insensitive queries, while weak debiasing fails to mitigate stereotypes in bias-sensitive ones. This one-size-fits-all approach hampers simultaneously achieving high utility on bias-insensitive queries and fairness on bias-sensitive queries. We introduce Reward-Gated Test-Time Adaptation (RG-TTA), a reinforcement learning-based test-time adaptation framework that selectively applies debiasing based on input sensitivity. RG-TTA adaptively triggers fairness regularization based on the bias sensitivity of each input during test-time policy adaptation, while focusing exclusively on optimizing cross-modal alignment for bias-insensitive inputs. Experiments on fairness benchmarks (e.g., FairFace, UTKFace) demonstrate substantial bias reduction while simultaneously improving zero-shot utility, resolving the trade-off of uniform debiasing.
Chinese Translation
视觉语言模型（VLMs）展现出强大的零-shot性能，但在以人为中心的查询中常常延续社会刻板印象，导致人口统计分布失衡。目前的去偏见方法对所有输入查询施加统一的偏见修正，而不考虑其偏见敏感性，从而产生了公平性与效用之间的根本权衡。强去偏见会扭曲偏见不敏感查询中的语义信息，而弱去偏见则无法缓解偏见敏感查询中的刻板印象。这种一刀切的方法阻碍了在偏见不敏感查询中实现高效用与在偏见敏感查询中实现公平性的双重目标。我们提出了奖励门控测试时适应（Reward-Gated Test-Time Adaptation, RG-TTA），这是一种基于强化学习的测试时适应框架，能够根据输入的敏感性选择性地应用去偏见。RG-TTA在测试时策略适应过程中，根据每个输入的偏见敏感性自适应触发公平性正则化，同时专注于优化偏见不敏感输入的跨模态对齐。在公平性基准（例如，FairFace，UTKFace）上的实验显示，显著减少了偏见，同时提高了零-shot效用，解决了统一去偏见的权衡问题。
cs.CL / 19 / 2607.00447
Understanding Why Language Models Hallucinate: Testing Reasoning Against Priors
理解语言模型为何产生幻觉：对先验知识进行推理测试
Abstract
Large language models often produce hallucinated answers that violate prompt-level constraints. A key diagnostic question is whether these failures reflect missing knowledge, or whether the model has the relevant information but follows the wrong inference path. We study this phenomenon as inference misalignment: a mismatch between the answer supported by the prompt and the answer favored by statistically salient latent associations. We formalize this view with a latent key-task model, in which pretraining-frequency imbalance can cause a shortcut path to dominate the constraint-sensitive path and induce positive inference loss. The framework predicts two failure modes: task-retrieval bias in entity disambiguation and key-selection bias in action choice. We introduce TrapQA, a controlled diagnostic testbed with two components. ScientistQA tests disambiguation among similar scientists with supplementary factual probes, while Real-Life Constrained QA tests everyday constraint following under salient shortcuts. Our results show that hallucination can arise from biased latent inference rather than absent knowledge alone.
Chinese Translation
大型语言模型经常产生违反提示级约束的幻觉性答案。一个关键的诊断问题是，这些失败是否反映了缺失的知识，或者模型是否拥有相关信息但遵循了错误的推理路径。我们将这一现象研究为推理不一致：即提示支持的答案与统计上显著的潜在关联所偏好的答案之间的不匹配。我们通过一个潜在关键任务模型对这一观点进行了形式化，其中预训练频率的不平衡可能导致捷径路径主导约束敏感路径，从而引发正向推理损失。该框架预测了两种失败模式：在实体消歧中的任务检索偏差和在行动选择中的关键选择偏差。我们引入了TrapQA，一个具有两个组成部分的受控诊断测试平台。ScientistQA测试相似科学家之间的消歧，辅以补充的事实探测，而Real-Life Constrained QA则测试在显著捷径下的日常约束遵循。我们的结果表明，幻觉可能源于偏向的潜在推理，而不仅仅是缺失知识。
cs.CL / 20 / 2607.00482
Know When to Stop: Segment-Level Credit Assignment for Reducing Overthinking
知道何时停止：减少过度思考的段落级信用分配
Abstract
Reasoning language models frequently overthink: generating extended chains of behaviors such as hedging, approach abandonment, and self contradiction that consume tokens without improving answers. We show that these behaviors are not merely a consequence of length; even when controlling for response length, incorrect traces exhibit higher rates of unproductive self-reflection than correct ones. Addressing this requires identifying where self-reflection helps vs hurts, but obtaining these step-level annotations is costly. We observe that intermediate answer commitments within reasoning traces can provide a cheap proxy: by comparing each final answer candidate in the trace to the ground truth, we can determine whether subsequent reflection is productive without any additional supervision. Building on this insight, we propose DASH (Drift Aware advantage SHaping), which assigns segment-level credit based on whether each reasoning segment leads toward or away from correctness. On competition-level math benchmarks, DASH achieves the highest accuracy where overthinking is prevalent (AIME25: 50.8% vs. 45.4% GRPO) while reducing overthinking behaviors and achieving more productive self-correction than baselines.
Chinese Translation
推理语言模型常常过度思考：生成延长的行为链，如犹豫、放弃接近和自我矛盾，这些行为消耗了标记却没有改善答案。我们表明，这些行为不仅仅是长度的结果；即使在控制响应长度的情况下，不正确的轨迹也表现出比正确轨迹更高的无效自我反思率。解决这一问题需要识别自我反思何时有益、何时有害，但获取这些步骤级注释的成本较高。我们观察到，推理轨迹中的中间答案承诺可以提供一个廉价的代理：通过将轨迹中的每个最终答案候选与真实值进行比较，我们可以确定后续反思是否具有生产性，而无需任何额外的监督。基于这一见解，我们提出了DASH（Drift Aware advantage SHaping），该方法根据每个推理段落是否朝向或远离正确性来分配段落级信用。在竞争级数学基准测试中，DASH在过度思考普遍存在的情况下实现了最高的准确率（AIME25: 50.8% vs. 45.4% GRPO），同时减少了过度思考行为，并实现了比基线更具生产性的自我修正。
cs.CL / 21 / 2607.00485
Efficient Multilingual Reasoning Transfer via Progressive Code-Switching
通过渐进式代码切换实现高效的多语言推理迁移
Abstract
Large reasoning models (LRMs) have achieved strong reasoning capabilities in English, yet their performance degrades significantly when required to reason in other languages. A natural solution is to transfer the model's English reasoning ability to target languages. However, existing transfer approaches typically rely on distilled target-language reasoning traces from stronger LRMs or online supervision from external judge models, which are costly and difficult to scale. In this paper, we propose PCS (Progressive Code-Switching), a more efficient transfer framework that requires only lightweight translation without any stronger model for distillation or judging. PCS first constructs code-switched reasoning traces by translating a subset of English reasoning steps into the target language, and uses them to initialize the model's code-switching ability via supervised fine-tuning. It then applies reinforcement learning with a step-level language consistency curriculum, progressively raising the target-language ratio until the model reasons entirely in the target language. This progressive design provides a smooth transfer path that avoids the instability and performance degradation commonly observed when directly enforcing target-language reasoning. Experiments on multiple benchmarks and five typologically diverse languages show that PCS substantially narrows the performance gap between target-language and English reasoning, yielding more language-consistent reasoning while maintaining competitive accuracy.
Chinese Translation
大型推理模型（LRMs）在英语中的推理能力表现出色，但在其他语言中的表现显著下降。一个自然的解决方案是将模型在英语中的推理能力转移到目标语言。然而，现有的迁移方法通常依赖于来自更强LRMs的目标语言推理轨迹的蒸馏，或依赖于外部评判模型的在线监督，这些方法成本高且难以扩展。本文提出了PCS（渐进式代码切换），一种更高效的迁移框架，仅需轻量级翻译，而无需任何更强模型的蒸馏或评判。PCS首先通过将一部分英语推理步骤翻译成目标语言来构建代码切换推理轨迹，并利用这些轨迹通过监督微调来初始化模型的代码切换能力。然后，它应用带有逐步语言一致性课程的强化学习，逐渐提高目标语言的比例，直到模型完全在目标语言中进行推理。这种渐进式设计提供了一条平滑的迁移路径，避免了直接强制目标语言推理时常见的不稳定性和性能下降。在多个基准测试和五种类型学上多样的语言上的实验表明，PCS显著缩小了目标语言与英语推理之间的性能差距，产生了更具语言一致性的推理，同时保持了竞争力的准确性。
cs.CL / 22 / 2607.00501
BaseRT: Best-in-Class LLM Inference on Apple Silicon via Native Metal
BaseRT：通过原生 Metal 在 Apple Silicon 上实现最佳大型语言模型推理
Abstract
We present BaseRT, a native Metal inference runtime for large language models (LLMs) on Apple Silicon, and report the highest inference throughput on this hardware to date. Existing runtimes, including llama.cpp and MLX-based frameworks, incur overhead from abstractions not designed for Metal's execution model or Apple Silicon's unified memory topology. By building natively on Metal with chip-specific kernel fusion, unified memory-aware optimisation, and custom dispatch logic, BaseRT recovers performance that framework-based approaches leave on the table. BaseRT supports a wide range of model families across eight quantisation formats (Q2 to FP16) on all Apple M-series devices. In this paper, we evaluate the Qwen3, Llama 3.2, and Gemma 4 families at Q4 and Q8 quantisation on M3 and M4 Pro devices. BaseRT achieves up to 1.56x higher decode throughput than llama.cpp and up to 1.35x higher than MLX, with substantially larger margins on prefill for mixture-of-experts models, delivering consistent best-in-class throughput from sub-1B to 30B parameter models. These results establish Apple Silicon as a more capable inference platform than previously reported, with direct implications for the emerging edge inference paradigm: as privacy requirements, latency constraints, and cloud cost pressures drive inference toward on-device deployment, performance-optimised local runtimes are a critical enabling layer for this transition. BaseRT is publicly available at https://github.com/basecompute/baseRT
Chinese Translation
我们提出了 BaseRT，这是一种针对 Apple Silicon 上大型语言模型（LLMs）的原生 Metal 推理运行时，并报告了迄今为止在该硬件上最高的推理吞吐量。现有的运行时，包括 llama.cpp 和基于 MLX 的框架，由于未针对 Metal 的执行模型或 Apple Silicon 的统一内存拓扑设计的抽象，导致了额外的开销。通过在 Metal 上原生构建，采用特定于芯片的内核融合、统一内存感知优化和自定义调度逻辑，BaseRT 恢复了框架方法所留下的性能。BaseRT 支持在所有 Apple M 系列设备上跨八种量化格式（Q2 到 FP16）的一系列模型家族。在本文中，我们在 M3 和 M4 Pro 设备上评估了 Qwen3、Llama 3.2 和 Gemma 4 家族在 Q4 和 Q8 量化下的表现。BaseRT 的解码吞吐量比 llama.cpp 高出最多 1.56 倍，比 MLX 高出最多 1.35 倍，在混合专家模型的预填充上具有显著更大的优势，为从小于 1B 到 30B 参数模型提供了一致的最佳吞吐量。这些结果确立了 Apple Silicon 作为比之前报告的更强大的推理平台，直接影响到新兴的边缘推理范式：随着隐私要求、延迟限制和云成本压力推动推理向设备端部署，性能优化的本地运行时成为这一转变的关键支持层。BaseRT 可在 https://github.com/basecompute/baseRT 上公开获取。
cs.CL / 23 / 2607.00502
A Task-State Representation for Long-Horizon Mobile GUI Agents
用于长时间移动图形用户界面代理的任务状态表示
Abstract
While long-horizon mobile GUI agents typically rely on thought-action-observation loops, they struggle to separate persistent task states from transient screen observations. As execution histories grow, this entanglement imposes a severe context burden, causing agents to forget initial requirements, hallucinate progress, or repeatedly interact with stale interfaces. To address this, we introduce Task-State Representation (TSR), a training-free framework that explicitly decouples task state from sensory input. Acting as a lightweight external wrapper, TSR maintains three structured components: a global instruction summary, a dynamic progress tracker for subgoals, and a transition-aware action verifier. By continuously updating through pre- and post-action visual comparisons, TSR effectively guides the agent's reasoning without requiring architectural modifications. Experiments across four mobile GUI benchmarks validate TSR's effectiveness, yielding up to a 12 absolute point increase in success rate on complex cross-application and memory-intensive tasks.
Chinese Translation
尽管长时间移动图形用户界面代理通常依赖于思考-行动-观察循环，但它们在将持久任务状态与瞬态屏幕观察分离方面存在困难。随着执行历史的增长，这种纠缠带来了严重的上下文负担，导致代理忘记初始需求、产生虚假进展或重复与过时界面交互。为了解决这个问题，我们提出了任务状态表示（Task-State Representation, TSR），这是一个无训练框架，明确地将任务状态与感官输入解耦。TSR作为一个轻量级的外部包装器，维护三个结构化组件：全局指令摘要、子目标的动态进度跟踪器和过渡感知的行动验证器。通过在行动前后进行视觉比较的持续更新，TSR有效地引导代理的推理，而无需进行架构修改。在四个移动图形用户界面基准测试中的实验验证了TSR的有效性，在复杂的跨应用和内存密集型任务中成功率最高提高了12个绝对点。
cs.CL / 24 / 2607.00570
Dual-Confidence Contrastive Decoding for Retrieval-Augmented Generation
双置信度对比解码用于检索增强生成
Abstract
Retrieval-augmented generation (RAG) increasingly requires models to answer questions from multiple retrieved documents, where only some sources are relevant and the retrieved bundle may contain stale, noisy, or conflicting evidence. Existing contrastive decoding methods primarily focus on resolving conflicts between the model's internal memory and the retrieved context. In contrast, we study the complementary problem of intra-context conflict in multi-document RAG. To evaluate this setting, we introduce DRQA, a factual-conflict question answering benchmark derived from enterprise deep-research scenarios, where answers are grounded in synthetic enterprise-specific facts that are designed not to be recoverable from the model's internal memory. We further propose Dual-Confidence Contrastive Decoding (DCCD), a training-free decoding method that combines document-level confidence, which estimates whether a document appears sufficient for answering the question, with token-level confidence, which estimates whether that document supports a confident next-token prediction. DCCD selects positive and negative document-conditioned streams using these dual-confidence signals and scales a document-level contrast by their confidence margin. Across DRQA and standard multi-document QA benchmarks, DCCD achieves the best average performance among full-context and contrastive decoding baselines, with the largest gains on DRQA. These results highlight the importance of source-aware, confidence-gated decoding when retrieved evidence is internally conflicting.
Chinese Translation
检索增强生成（RAG）越来越需要模型从多个检索到的文档中回答问题，其中只有部分来源是相关的，而检索到的文档可能包含过时、嘈杂或相互冲突的证据。现有的对比解码方法主要集中在解决模型内部记忆与检索上下文之间的冲突。相反，我们研究了多文档 RAG 中上下文内部冲突的互补问题。为了评估这一设置，我们引入了 DRQA，这是一个基于企业深度研究场景的事实冲突问答基准，其中答案基于合成的企业特定事实，这些事实设计为无法从模型的内部记忆中恢复。我们进一步提出了双置信度对比解码（DCCD），这是一种无训练的解码方法，结合了文档级置信度（用于估计文档是否足够回答问题）和标记级置信度（用于估计该文档是否支持有信心的下一个标记预测）。DCCD 使用这两种置信度信号选择正负文档条件流，并根据它们的置信度边际扩展文档级对比。在 DRQA 和标准多文档问答基准上，DCCD 在完整上下文和对比解码基准中实现了最佳的平均性能，在 DRQA 上获得了最大的提升。这些结果突显了在检索证据内部存在冲突时，源感知和置信度门控解码的重要性。
cs.CL / 25 / 2607.00576
Safe Alone, Unsafe Together: Safeguarding Against Implicit Toxicity When Benign Images Combine
单独安全，共同不安全：在良性图像组合中防范隐性毒性
Abstract
Multi-image content has become an increasingly prevalent form of visual communication in social media, giving rise to a new safety issue, multi-image implicit toxicity (MIIT), where each image appears benign in isolation, but harmful semantics emerge when the images are interpreted jointly. MIIT is particularly challenging for existing commercial moderation APIs and models due to the lack of explicit risky cues in each image. This paper aims to study how to identify MIIT. We first provide a formal definition of MIIT and analyze three key challenges for its detection. To alleviate the scarcity of data in this area, we construct MIIT-dataset, an image-only multi-image safety dataset covering seven representative risk categories through an automatic generation pipeline. Finally, we train MiShield with progressively distilled reasoning supervision, enabling it to produce safety judgments accompanied by explicit analyses of the correlated entities that result in the hazards. Experiments show that MiShield-8B models outperform representative moderation services and even larger-scale models, revealing its effectiveness and practical value for this widely used visual format. Warning: This paper contains potentially sensitive content.
Chinese Translation
多图像内容已成为社交媒体中一种日益普遍的视觉传播形式，带来了新的安全问题——多图像隐性毒性（MIIT）。在这种情况下，每幅图像单独看似良性，但当图像共同解读时却会出现有害的语义。由于每幅图像缺乏明确的风险提示，MIIT对现有商业内容审核API和模型构成了特别的挑战。本文旨在研究如何识别MIIT。我们首先提供了MIIT的正式定义，并分析了其检测的三个关键挑战。为了缓解该领域数据稀缺的问题，我们构建了MIIT数据集，这是一个仅包含图像的多图像安全数据集，通过自动生成管道覆盖七个代表性的风险类别。最后，我们使用逐步提炼的推理监督训练了MiShield，使其能够产生安全判断，并附带对导致危害的相关实体的明确分析。实验表明，MiShield-8B模型的表现优于代表性的内容审核服务，甚至比更大规模的模型更具优势，揭示了其在这一广泛使用的视觉格式中的有效性和实际价值。警告：本文包含潜在敏感内容。
cs.CL / 26 / 2607.00588
Low Perplexity is Repetition: A One-Dimensional Self-Conditioning Attractor in Continuous Diffusion LMs
低困惑度是重复：连续扩散语言模型中的一维自条件吸引子
Abstract
Continuous diffusion language models such as ELF report record-low generative perplexity (Gen-PPL). We find a catch: these models repeat far more than human text, and Gen-PPL rewards rather than penalizes that repetition, so its low scores overstate quality. Strip the repetition and ELF-B's Gen-PPL rises from $19.5$ to $27.7$; the smallest model even posts the best Gen-PPL because it repeats most. We trace the repetition to its source: a contractive attractor along a \emph{single direction} in the self-conditioning feedback loop, the loop that feeds each step's clean estimate into the next. Because the failure is one-dimensional, a one-dimensional fix suffices, and we propose one. \textbf{ACE} (Attractor-Contrast-Escape) subtracts that single, label-free direction from the feedback at each step. Estimated once on the $105$M model, the direction cuts repetition to near the human level while keeping quality competitive, and transfers near-unchanged to the $342$M and $652$M models and across samplers; the same recipe recovers useful directions on other architectures. Since Gen-PPL itself rewards repetition, we instead measure the compute each fix needs to produce human-clean text, where ACE is $1.5$--$5\times$ cheaper.
Chinese Translation
连续扩散语言模型如ELF报告了创纪录的低生成困惑度（Gen-PPL）。我们发现了一个问题：这些模型的重复程度远高于人类文本，而Gen-PPL奖励而非惩罚这种重复，因此其低分数夸大了质量。去除重复后，ELF-B的Gen-PPL从19.5上升到27.7；最小的模型甚至因为重复最多而获得了最佳的Gen-PPL。我们追溯这种重复的根源：在自条件反馈循环中沿着 extit{单一方向}的收缩吸引子，该循环将每一步的干净估计反馈到下一步。由于失败是单维的，因此一个一维的修正就足够了，我们提出了一种修正方法。 extbf{ACE}（吸引子-对比-逃逸）在每一步中从反馈中减去这个单一的、无标签的方向。在105M模型上估计一次后，该方向将重复减少到接近人类水平，同时保持竞争力，且在342M和652M模型及不同采样器中几乎不变；同样的方案在其他架构上也能恢复有用的方向。由于Gen-PPL本身奖励重复，我们改为测量每个修正所需的计算量，以生成人类干净的文本，其中ACE的成本是1.5到5倍更便宜。
cs.CL / 27 / 2607.00597
Multi-Turn Agentic Scientific Literature Search via Workflow Induction
通过工作流诱导进行多轮自主科学文献搜索
Abstract
Scientific literature search often requires more than retrieving papers from a single query: users' intents are underspecified, preference-dependent, and evolve through interaction. Existing search agents typically rely on fixed pipelines or implicit language-only reasoning, making their search strategies difficult to control, inspect, and refine. We introduce PaperPilot, a multi-turn literature search agent that frames scientific search as workflow induction. Given an anchor paper and a user query, PaperPilot constructs an executable DAG of paper-search operators, including keyword search, citation expansion, filtering, scoring, reranking, and evidence extraction. User feedback is then used to refine both the query and the workflow itself. We train PaperPilot with supervised workflow imitation and preference optimization over controlled workflow corruptions. Experiments show that PaperPilot-9B improves over the base Qwen3.5-9B toolset agent under multi-turn interaction, increasing Hit@5 from 58.0 to 77.0, MRR from 47.5 to 59.4, and nDCG@10 from 26.8 to 32.5, while reducing workflow execution errors from 9.5% to 0%. These results show that explicit, editable search workflows provide an effective and controllable interface for aligning literature search agents with complex scientific intent.
Chinese Translation
科学文献搜索通常不仅仅是从单一查询中检索论文：用户的意图往往不明确、依赖于偏好，并且通过互动而不断演变。现有的搜索代理通常依赖于固定的流程或隐式的仅语言推理，使得它们的搜索策略难以控制、检查和优化。我们提出了PaperPilot，一个将科学搜索框架化为工作流诱导的多轮文献搜索代理。给定一篇锚定论文和用户查询，PaperPilot构建一个可执行的有向无环图（DAG）来表示论文搜索操作，包括关键词搜索、引用扩展、过滤、评分、重排序和证据提取。然后，用户反馈被用来优化查询和工作流本身。我们通过监督工作流模仿和在受控工作流破坏下的偏好优化来训练PaperPilot。实验表明，在多轮互动下，PaperPilot-9B相较于基础的Qwen3.5-9B工具集代理有所提升，Hit@5从58.0提高到77.0，MRR从47.5提高到59.4，nDCG@10从26.8提高到32.5，同时将工作流执行错误率从9.5%降低到0%。这些结果表明，明确且可编辑的搜索工作流为将文献搜索代理与复杂科学意图对齐提供了有效且可控的接口。
cs.CL / 28 / 2607.00601
"Don't Say It!": Constraints, Compliance, and Communication when Language Models Play Taboo
别说了！：语言模型在玩禁忌游戏时的约束、遵从与沟通
Abstract
The game of Taboo requires describing a target word without using a set of forbidden words, so that other players can guess it. This deceptively simple task combines strict lexical constraints with the need for communicatively effective descriptions, making it a compelling playground for examining how LLMs navigate competing demands at inference time. We evaluate two open-weight models under conditions that intervene at progressively deeper levels of the generative process, from prompting to generation-time constraints to internal representations manipulations. We assess their outputs through forbidden word violation detection, LLM-as-a-judge measuring the degree to which generated descriptions successfully evoke the target concept for both human and machine guessers, and examining whether the strategies models adopt under constraint align with those of human players. Our results show that compliance with the rules of the game and communicative effectiveness trade off differently across conditions, and that models remain substantially weaker than humans as guessers, suggesting that lexical grounding under constraint is an open challenge for current language models.
Chinese Translation
禁忌游戏要求在不使用一组禁用词的情况下描述一个目标词，以便其他玩家能够猜出它。这个看似简单的任务结合了严格的词汇约束和有效的沟通描述的需求，使其成为一个引人注目的实验场，旨在考察大型语言模型（LLMs）在推理时如何应对相互竞争的要求。我们在逐步深入生成过程的条件下评估了两个开放权重模型，从提示到生成时的约束，再到内部表示的操控。我们通过禁用词违规检测、LLM作为评判者来衡量生成描述成功唤起目标概念的程度（对于人类和机器猜测者），并检查模型在约束下采用的策略是否与人类玩家的策略一致。我们的结果显示，遵守游戏规则和沟通有效性在不同条件下存在不同的权衡，并且模型作为猜测者的表现仍然显著弱于人类，这表明在约束下的词汇基础仍然是当前语言模型面临的一个开放挑战。
cs.CL / 29 / 2607.00605
Auditing Forgetting in Limited Memory Language Models
有限记忆语言模型中的遗忘审计
Abstract
Limited Memory Language Models (LMLMs) externalize factual knowledge to a database to enable deletion-based unlearning without retraining. Existing evaluations measure post-deletion correctness in aggregate and cannot tell whether a deleted fact persists through residual parametric memory, alternative retrieval paths, or near-neighbor retrieval artifacts. We propose a causal auditing framework that holds the model fixed and varies the database state at inference time across three interventions: FULL, DEL-ON, and DEL-OFF. The framework decomposes post-deletion behavior into parametric leakage L(f), retrieval-mediated correctness R(f), and a retrieval artifact rate grounded in the inference-time retrieval trace. We apply it to 12,228 alias-closure deletions across thirteen databases, including four adversarial topologies (Base, Alias, Noise, Collision) we construct in three domains, and six prompt formulations. Parametric leakage is near zero in every variant and every prompt style: the model rarely returns the deleted answer in the absence of retrieval. The residual that does survive lives in the retrieval graph: retrieval-mediated correctness and the retrieval artifact rate match within rounding everywhere, so post-deletion correctness is, in our audit, predominantly reconstituted from near-neighbor retrieval. This residual ranges from 0.7% on the released LMLM database to 13.6% on the most adversarial variant, and prompt formulation does not independently control how much of a deleted fact survives. These results suggest that, for this class of LMLM and deletion procedure, the unlearning boundary is drawn primarily by the database administrator rather than by the model.
Chinese Translation
有限记忆语言模型（LMLMs）将事实知识外部化到数据库中，以实现基于删除的无学习（unlearning），而无需重新训练。现有评估在总体上测量删除后的正确性，无法判断被删除的事实是否通过残余参数记忆、替代检索路径或近邻检索伪影而持续存在。我们提出了一种因果审计框架，该框架在推理时固定模型并在三个干预下变化数据库状态：FULL、DEL-ON和DEL-OFF。该框架将删除后的行为分解为参数泄漏 L(f)、检索介导的正确性 R(f) 和基于推理时检索轨迹的检索伪影率。我们将其应用于跨越十三个数据库的12,228个别名闭合删除，包括我们在三个领域构建的四种对抗拓扑（Base、Alias、Noise、Collision）和六种提示形式。在每种变体和每种提示风格中，参数泄漏接近零：在没有检索的情况下，模型很少返回被删除的答案。存活下来的残余存在于检索图中：检索介导的正确性和检索伪影率在四舍五入时处处匹配，因此在我们的审计中，删除后的正确性主要是通过近邻检索重新构建的。该残余在发布的 LMLM 数据库上为 0.7%，在最具对抗性的变体上为 13.6%，而提示形式并不独立控制被删除事实的存活量。这些结果表明，对于这一类 LMLM 和删除过程，无学习的边界主要由数据库管理员而非模型划定。
cs.CL / 30 / 2607.00661
Faithful by Definition: Emotion Analysis via Natural Semantic Metalanguage Explications
按定义忠实：通过自然语义元语言阐释进行情感分析
Abstract
Explanations for emotion classifiers are usually produced post hoc, with no guarantee that they reflect the computation behind the label. We present an explication interface for event-based emotion analysis. A parser maps the input text to an explication, a short script in the closed vocabulary of Natural Semantic Metalanguage organized into twelve typed slots, and a fixed decision list of rules transcribed from published semantic definitions computes the label from the explication alone. The faithfulness guarantee is therefore causal and definitional, while all empirical risk lives in the learned parser, which the per-line entailment interface makes auditable against the input. On crowd-sourced event descriptions, our fine-tuned parser reaches 0.33 accuracy and 0.48 selective accuracy on a small held-out set, suggesting that the interface trades insignificant accuracy difference to a black-box model for a verifiable, inspectable decision basis for first-person event-based emotion analysis. We also release EmoExpl-1200 with per-line verification metadata and the full rule set.
Chinese Translation
情感分类器的解释通常是在事后生成的，无法保证它们反映标签背后的计算过程。我们提出了一种基于事件的情感分析阐释接口。解析器将输入文本映射到阐释，即一个简短的脚本，采用自然语义元语言的封闭词汇，组织成十二个类型化槽位，并且从已发布的语义定义中转录的固定决策规则列表仅通过阐释计算标签。因此，忠实性保证是因果的和定义性的，而所有的经验风险都存在于学习到的解析器中，逐行推理接口使其可以针对输入进行审计。在众包的事件描述上，我们经过微调的解析器在一个小的保留集上达到了0.33的准确率和0.48的选择性准确率，这表明该接口在可验证的、可检查的决策基础上，以微不足道的准确率差异替代了黑箱模型，适用于第一人称基于事件的情感分析。我们还发布了带有逐行验证元数据和完整规则集的EmoExpl-1200。
cs.CL / 31 / 2607.00664
YOMI-Bench: A Benchmark for Evaluating Kanji Reading and Phonological Understanding of LLMs for Japanese
YOMI-Bench：评估大型语言模型在日语汉字阅读和语音理解能力的基准测试
Abstract
We propose YOMI-Bench, a benchmark for evaluating kanji reading and phonological understanding of large language models (LLMs) for Japanese. In Japanese, a single kanji character often has multiple possible readings, making it difficult to infer the correct reading from surface-level text alone. Due to these linguistic characteristics, it is empirically known that LLMs exhibit low performance in kanji reading for Japanese. The proposed YOMI-Bench consists of four tasks specifically designed to evaluate kanji reading performance in Japanese. In our evaluation using YOMI-Bench, we assessed one multilingual open LLM, four Japanese-specific open LLMs, and five commercial LLMs. As a result, we found that even Japanese-specific models show low performance, and that commercial models also perform poorly on generation tasks that require consideration of kanji readings.
Chinese Translation
我们提出了YOMI-Bench，这是一个用于评估大型语言模型（LLMs）在日语汉字阅读和语音理解能力的基准测试。在日语中，单个汉字通常有多种可能的读音，这使得仅通过表层文本推断正确读音变得困难。由于这些语言特征，实证研究表明，LLMs在日语汉字阅读方面表现不佳。所提出的YOMI-Bench包含四个专门设计的任务，以评估日语的汉字阅读性能。在使用YOMI-Bench进行评估时，我们评估了一个多语言开放LLM、四个特定于日语的开放LLM和五个商业LLM。结果发现，即使是特定于日语的模型也表现不佳，而商业模型在需要考虑汉字读音的生成任务中同样表现不佳。
cs.CL / 32 / 2607.00714
Self-conditioned Flow Map Language Models via Fixed-point Flows
通过不动点流的自条件流图语言模型
Abstract
Self-conditioning is a core technique that enhances continuous flow-based language models, where the model learns to denoise generated text by conditioning on its own denoising estimate. While empirically successful, its performance improvements are poorly understood. Moreover, there is growing interest in the use of few-step generators based on flow maps, for which how to leverage self-conditioning is unclear. Here, we show that flow language models with self-conditioning solve a fixed-point iteration that bootstraps the performance of the learned denoiser. We use this viewpoint to formulate fixed-point flows, a two-dimensional class of self-conditioned flows, where the first dimension represents the flow process and the second represents the fixed-point iteration. We show that fixed-point flows define valid flow maps, and show that they can be distilled from self-conditioned flow models by compressing both fixed-point iterations and the flow process, the former with fixed-point distillation and the latter with flow map distillation. Our resulting flow map language model, FMLM$^\star$, outperforms state-of-the-art self-conditioned models and few-step models in one- and few-step generation on OpenWebText. Code is available at https://github.com/Ugness/self-conditioned-fmlm.
Chinese Translation
自条件化是一种增强连续流基语言模型的核心技术，其中模型通过对自身去噪估计进行条件化来学习去噪生成的文本。尽管在经验上取得了成功，但其性能提升的机制尚不清楚。此外，基于流图的少步生成器的使用日益受到关注，但如何利用自条件化仍不明确。在此，我们展示了具有自条件化的流语言模型解决了一个不动点迭代问题，从而引导学习到的去噪器的性能。我们使用这一视角来构造不动点流，这是一类二维自条件流，其中第一维表示流过程，第二维表示不动点迭代。我们证明不动点流定义了有效的流图，并展示它们可以通过压缩不动点迭代和流过程，从自条件流模型中提取，前者通过不动点蒸馏，后者通过流图蒸馏。我们得到的流图语言模型 FMLM$^ullet$ 在 OpenWebText 上的一步和少步生成任务中超越了最先进的自条件模型和少步模型。代码可在 https://github.com/Ugness/self-conditioned-fmlm 获取。
cs.CL / 33 / 2607.00724
MSQA: A Natively Sourced Multilingual and Multicultural SimpleQA Benchmark
MSQA：一种本土来源的多语言和多文化简单问答基准
Abstract
Multilingual fluency often invites a stronger assumption: a model that can speak a user's language must also understand the culture encoded by that language. We call this the Illusion of Cultural Alignment. To test this assumption directly, we introduce MSQA, a benchmark of 1,064 natively sourced questions across 11 language groups, five cultural dimensions, and three difficulty tiers. Unlike translated benchmarks, MSQA targets locally grounded knowledge and reduces shortcuts from English-centric cross-lingual transfer. Evaluating 18 LLMs, we find substantial cultural degradation and a pronounced Locality Effect: cultural competence tracks pre-training exposure more closely than general reasoning ability. We further show that common inference-time remedies do not dissolve the illusion. Models remain overconfident on unfamiliar cultural questions, repeated sampling yields unstable rather than reliable correctness, and retrieval augmentation helps unevenly on long-tail facts. These findings indicate that cultural alignment cannot be inferred from multilingual ability alone and requires deeper intervention than calibration, sampling, or retrieval at inference time
Chinese Translation
多语言流利性往往引发一个更强的假设：能够使用用户语言的模型也必须理解该语言所编码的文化。我们称之为文化对齐的幻觉。为了直接测试这一假设，我们引入了MSQA，这是一个包含1064个本土来源问题的基准，涵盖11个语言组、五个文化维度和三个难度等级。与翻译基准不同，MSQA针对本地化知识，减少了来自以英语为中心的跨语言转移的捷径。通过评估18个大型语言模型（LLMs），我们发现显著的文化退化和明显的本地效应：文化能力与预训练暴露的关系比一般推理能力更为紧密。我们进一步表明，常见的推理时补救措施并不能消除这种幻觉。模型在不熟悉的文化问题上仍然过于自信，重复采样产生的不稳定性而非可靠的正确性，而检索增强在长尾事实上的帮助不均匀。这些发现表明，文化对齐不能仅仅通过多语言能力推断出来，需要比校准、采样或推理时检索更深层次的干预。
cs.CL / 34 / 2607.00725
What Survives Into Context: A Diagnostic for Budget-Constrained Multi-Hop RAG and When Submodular Evidence Packing Improves It
在上下文中存活的内容：预算受限的多跳检索增强生成（RAG）的诊断及何时子模块证据打包能改善它
Abstract
Retrieval-augmented generation (RAG) under a fixed reader-context budget forces a selection problem: of the evidence retrieved, only a fraction can be shown to the reader. We argue that document recall -- the standard retrieval metric -- is the wrong quantity to optimize in this regime, and we make two contributions. First, as a general contribution, we introduce answer-in-context, a diagnostic that measures whether a gold answer survives as a contiguous span in the packed reader context (not the retrieved set). It predicts answer F1 better than recall (r=0.39-0.55 vs. about 0.31), separates answer quality roughly five-fold (0.60 vs. 0.12 on HotpotQA), and carries information beyond retrieval: it adds Delta R squared=0.17 over recall and shows a 4.6x EM gap even among questions where all gold was retrieved. We also confirm it interventionally: on 2WikiMultiHopQA a packing change that raises coverage but not answer-in-context yields no accuracy gain. Second, as a conditional contribution, we cast reader-context construction as budgeted monotone submodular maximization and build a packer that jointly optimizes relevance, query coverage, representativeness, and diversity. On HotpotQA with a 160-token budget and a 3B reader it beats a strong focused heuristic, MMR, and naive packing -- by up to +5.1 F1 at equal-or-lower token cost, across three seeds. Crucially, we map the scope of this win honestly: it requires the conjunction of (i) multi-hop complementary structure, (ii) retrieval that surfaces the evidence, (iii) a binding but not extreme budget, and (iv) a reader weak enough that evidence density, not reading capacity, is the bottleneck. A quantization-controlled reader-scale ladder (3B to 7B to 14B) shows the edge over the heuristic is absorbed by 7B and significantly reverses by 14B, while the diagnostic explains every boundary with a single variable.
Chinese Translation
在固定读者上下文预算下的检索增强生成（RAG）迫使我们面临选择问题：所检索的证据中，只有一部分可以展示给读者。我们认为，在这种情况下，文档召回率——标准的检索指标——并不是需要优化的正确量，我们做出了两个贡献。首先，作为一般贡献，我们引入了上下文中的答案（answer-in-context），这是一个诊断工具，用于测量黄金答案是否作为一个连续的片段存活在打包的读者上下文中（而不是检索集）。它比召回率（r=0.39-0.55 vs. 约0.31）更好地预测答案F1，并且大约能将答案质量区分五倍（在HotpotQA上为0.60 vs. 0.12），并提供超出检索的信息：它在召回率上增加了Delta R平方=0.17，并且即使在所有黄金答案都被检索到的问题中也显示出4.6倍的EM差距。我们还通过干预确认了这一点：在2WikiMultiHopQA上，提升覆盖率但不提升上下文中的答案的打包变化没有带来准确性提升。其次，作为条件贡献，我们将读者上下文构建视为预算单调子模块最大化，并构建了一个打包器，联合优化相关性、查询覆盖率、代表性和多样性。在HotpotQA上，使用160个标记的预算和3B的读者，它在三个种子上以相同或更低的标记成本超越了强大的聚焦启发式方法MMR和简单打包——最高可提升5.1的F1。至关重要的是，我们诚实地映射了这一胜利的范围：它需要以下条件的结合：(i) 多跳互补结构，(ii) 能够呈现证据的检索，(iii) 绑定但不极端的预算，以及(iv) 证据密度而非阅读能力为瓶颈的足够弱的读者。一个量化控制的读者规模梯度（从3B到7B再到14B）显示，启发式方法的优势在7B时被吸收，并在14B时显著逆转，而诊断工具用一个变量解释了每个边界。
cs.CL / 35 / 2607.00848
MetaHOPE: A Metaphor-Oriented Evaluation Framework for Analysing MT and LLM Translation Errors
MetaHOPE：一种面向隐喻的评估框架用于分析机器翻译和大语言模型的翻译错误
Abstract
In this opinion paper, we propose MetaHOPE, an error severity-aware annotation framework for evaluating metaphor translations. Metaphors present challenges for machine translation (MT) and natural language understanding and processing (NLU, NLP), because it presents the features of semantic complexity, contextual dependency, and cultural embeddings that can lead to ambiguity issues for NLP models. To investigate how state-of-the-art NLP models perform on translating metaphors, we select three representative systems, i.e., GoogleMT, GPT5.4, and Hunyuan-7b as Neural MT (NMT) models and LLMs. We used two human-annotated metaphor corpora, including VUAMC and PSUCMC for English-to-Chinese and Chinese-to-English translation purposes. The original corpora we used are monolingual, where we carried out error annotation using the MetaHOPE framework, and also produced the human post-edited gold reference for bilingual use as a new resource. We believe the MetaHOPE evaluation framework for metaphor translation annotation, the parallel corpora resources, and the error analysis on SOTA automatic translation models can be useful and shed some light for the field of metaphor translation study. We share our resources publicly upon paper acceptance.
Chinese Translation
在本文中，我们提出了MetaHOPE，这是一种关注错误严重性的注释框架，用于评估隐喻翻译。隐喻给机器翻译（MT）和自然语言理解与处理（NLU，NLP）带来了挑战，因为它具有语义复杂性、上下文依赖性和文化嵌入等特征，这可能导致NLP模型出现歧义问题。为了研究最先进的NLP模型在隐喻翻译中的表现，我们选择了三个代表性系统，即GoogleMT、GPT5.4和Hunyuan-7b，作为神经机器翻译（NMT）模型和大语言模型（LLMs）。我们使用了两个经过人工注释的隐喻语料库，包括VUAMC和PSUCMC，用于英译中和中译英的翻译目的。我们使用的原始语料库是单语的，我们利用MetaHOPE框架进行了错误注释，并为双语使用生成了人工后编辑的黄金参考作为新资源。我们相信，MetaHOPE隐喻翻译注释的评估框架、平行语料资源以及对最先进自动翻译模型的错误分析将对隐喻翻译研究领域有所帮助并提供一些启示。我们将在论文接受后公开分享我们的资源。
cs.CL / 36 / 2607.00849
The Course of News Events: A Comparison of Bottom-Up and Top-Down Approaches for Collecting Text-Based Data about Disasters
新闻事件的进程：关于灾害文本数据收集的自下而上与自上而下方法的比较
Abstract
News articles are an important source of information on disaster impacts and adaptation. A key methodological challenge in socio-environmental studies is how to select a representative data sample. Two approaches are common: querying news databases top-down with the aid of an existing disaster inventory or using NLP methods to cluster news texts bottom-up based on temporal and spatial features. Using a dataset of German news about landslides worldwide, we compare these approaches and discuss variations in event coverage. Such research design decision can influence the resulting news sample, affecting its use in studies of inequality in media coverage, disaster monitoring and inventory enrichment.
Chinese Translation
新闻文章是关于灾害影响和适应的重要信息来源。在社会环境研究中，一个关键的方法论挑战是如何选择一个具有代表性的数据样本。两种常见的方法是：利用现有的灾害清单，自上而下地查询新闻数据库，或使用自然语言处理（NLP）方法，根据时间和空间特征，自下而上地对新闻文本进行聚类。我们使用一组关于全球滑坡的德国新闻数据集，比较这两种方法，并讨论事件覆盖的差异。这种研究设计的决策可能会影响最终的新闻样本，从而影响其在媒体覆盖不平等、灾害监测和清单丰富化研究中的应用。
cs.CL / 37 / 2607.00852
Recovering Input Text from Hidden States: Study of Gradient-Based Inversion of Decoder-Only Language Models
从隐藏状态恢复输入文本：解码器专用语言模型的基于梯度的反演研究
Abstract
This work studies the hidden-state inversion problem: recovering the original input token sequence of a decoder-only language model from its last-layer hidden states. Rather than treating inversion as a one-shot reconstruction, we study it as a continuous embedding-space optimisation in which a soft proxy is driven towards the leaked target without any hard-token projection during the search, and a token is committed only once, at the end of the inner loop. This design choice has two consequences which are the main focus of this paper. First, keeping the optimisation entirely in continuous space exposes a rich set of internal signals: rank trajectories of the ground-truth token, per-position loss curves, and a discrete loss measured at commit time. Second, the discrete loss allows assessing the correctness of recovery via cumulative discrete loss. We further analyse which tokens break the reconstructions and find a sharp categorical asymmetry: space-prefixed, high-frequency function words in dense regions of the embedding matrix dominate the failures, while content-bearing tokens are recovered almost perfectly. On 10-token C4 prompts the exact-match rate rises from 66.9% to 97.5% (mean similarity 0.994) as the candidate window is widened, confirming that most errors are recoverable near-misses rather than genuine ambiguities. A comparison with the released SIPIT reference situates these findings: per-step hard projection is faster, but the continuous formulation is what makes the optimisation observable and its failures detectable. The results show that last-layer hidden states of GPT-2 are as sensitive as the original text.
Chinese Translation
本研究探讨了隐藏状态反演问题：从解码器专用语言模型的最后一层隐藏状态中恢复原始输入标记序列。我们并不将反演视为一次性重建，而是将其视为一种连续嵌入空间优化，其中一个软代理在搜索过程中向泄露的目标驱动，而不进行任何硬标记投影，并且仅在内循环结束时才确定一个标记。这一设计选择有两个主要后果，成为本文的重点。首先，完全在连续空间中进行优化暴露了一组丰富的内部信号：真实标记的排名轨迹、每个位置的损失曲线，以及在确定时测量的离散损失。其次，离散损失允许通过累积离散损失来评估恢复的正确性。我们进一步分析了哪些标记破坏了重建，并发现存在明显的类别不对称性：在嵌入矩阵的密集区域中，空间前缀、高频功能词主导了失败，而承载内容的标记几乎完美恢复。在10个标记的C4提示中，精确匹配率从66.9%上升到97.5%（平均相似度为0.994），随着候选窗口的扩大，确认大多数错误是可恢复的接近失误，而非真正的歧义。与发布的SIPIT参考进行比较，这些发现得到了进一步的定位：每步硬投影更快，但连续形式使得优化过程可观察，其失败可检测。结果表明，GPT-2的最后一层隐藏状态对原始文本同样敏感。
cs.CL / 38 / 2607.00862
CAT: Confidence-Adaptive Thinking for Efficient Reasoning of Large Reasoning Models
CAT：用于大型推理模型高效推理的自适应信心思维
Abstract
Large Reasoning Models (LRMs) have achieved remarkable success on complex tasks by leveraging long chain-of-thought (CoT) trajectories, yet they frequently exhibit overthinking on simple queries, resulting in significant token overhead and reduced inference efficiency. However, existing compression methods predominantly apply uniform length reduction or rely on coarse-grained difficulty estimation, often leading to performance degradation on difficult problems. To address this limitation, we propose Confidence-Adaptive Thinking (CAT), a framework that incorporates the model's intrinsic self-certainty signals as confidence into the preference optimization process, which autonomously modulates reasoning lengths based on problem difficulty. Experimental results show that CAT consistently outperforms state-of-the-art baselines on reasoning accuracy across multiple benchmarks on different base models. Our work enables LRMs to effectively compress confident responses while deliberating on uncertain ones, offering a potentially robust solution for balancing accuracy and latency in practical industrial scenarios.
Chinese Translation
大型推理模型（LRMs）通过利用长链思维（CoT）轨迹在复杂任务上取得了显著成功，但它们在简单查询上经常表现出过度思考，导致显著的令牌开销和推理效率降低。然而，现有的压缩方法主要采用统一的长度缩减或依赖粗粒度的难度估计，往往导致在困难问题上的性能下降。为了解决这一局限性，我们提出了自适应信心思维（Confidence-Adaptive Thinking，CAT），这一框架将模型内在的自我确定信号作为信心纳入偏好优化过程，能够根据问题的难度自主调节推理长度。实验结果表明，CAT在不同基础模型的多个基准测试中，始终优于最先进的基线，在推理准确性上表现出色。我们的工作使得LRMs能够有效压缩自信的响应，同时对不确定的响应进行深思，提供了一种在实际工业场景中平衡准确性和延迟的潜在稳健解决方案。
cs.CL / 39 / 2607.00870
Dynamic Bidirectional Pattern Memory: A Production-Scale Empirical Characterisation of Inference-Time Gating in Clinical NLP
动态双向模式记忆：临床自然语言处理中的推理时间门控的生产规模实证特征化
Abstract
We study inference-time pattern-memory gating in a production-scale clinical natural language processing (NLP) pipeline. The pipeline pairs a generator (Llama-3.3 70B) proposing extractions with a verifier (MMed-Llama-3.1 70B) accepting or rejecting them, over 167,034 PMC-Patients narratives, and adds a lightweight memory that learns at deployment which extractions to filter, so the verifier need not re-examine candidates already seen to fail. We report four findings. First, learning filtering rules directly from the verifier's rejections failed at full scale: the relation-extraction filter stayed empty despite 785,797 logged rejections, because they were spread too thinly across too many distinct forms to accumulate. Second, a simpler rule using a fixed clinical ontology produced the same filtering without the verifier, capturing 49,734 ontology-violating relations on a held-out 5,000-patient set. Third, of five versions of the question-answering filter, four failed for distinct, instructive reasons; the fifth succeeded by checking whether a patient's extracted entities support the question asked, and where it applies was 1.84 times likelier to flag an answer the verifier would reject than one it would accept. Fourth, one pattern held across all five: a filter is selective only when it tests the same evidence the verifier weighs, not when it imitates the verifier's output. Together these give a transferable result for any generator-verifier pipeline: the most natural memory design can fail silently at scale, and whether a pre-generation gate is selective is decided before any engineering effort, by whether its signal probes the question the verifier itself answers. Throughout, the system flags suspect extractions rather than deleting them, so every decision stays visible for clinical review. All code and test artefacts are released openly.
Chinese Translation
我们研究了在生产规模的临床自然语言处理（NLP）管道中的推理时间模式记忆门控。该管道将一个生成器（Llama-3.3 70B）与一个验证器（MMed-Llama-3.1 70B）配对，生成器提出提取内容，验证器则对其进行接受或拒绝，处理167,034个PMC-Patients叙述，并增加了一个轻量级的记忆模块，该模块在部署时学习过滤哪些提取内容，以便验证器不必重新检查已经被拒绝的候选项。我们报告了四个发现。首先，直接从验证器的拒绝中学习过滤规则在全规模下失败：尽管记录了785,797次拒绝，关系提取过滤器仍然为空，因为这些拒绝分散在太多不同的形式中，无法积累。第二，使用固定临床本体的简单规则在没有验证器的情况下产生了相同的过滤效果，在一个保留的5,000患者集上捕获了49,734个违反本体的关系。第三，在五个版本的问题回答过滤器中，四个因不同的、具有启发性的原因失败；第五个通过检查患者提取的实体是否支持所提问的问题而成功，其适用时标记验证器会拒绝的答案的可能性是标记验证器会接受的答案的1.84倍。第四，所有五个版本中都有一个共同的模式：过滤器只有在测试与验证器权衡相同的证据时才是选择性的，而不是在模仿验证器的输出时。综合来看，这些结果为任何生成器-验证器管道提供了可转移的结论：最自然的记忆设计在规模上可能会无声失败，而预生成门是否具有选择性是在任何工程努力之前决定的，取决于其信号是否探测验证器自身回答的问题。整个过程中，系统标记可疑的提取内容，而不是删除它们，因此每个决策都保持可见，以便进行临床审查。所有代码和测试文档均已公开发布。
cs.CL / 40 / 2607.00873
How Ethos and Pathos Appeals Resonate in Reader Interpretations of Social Media Messages
伦理与情感诉求如何在读者对社交媒体信息的解读中产生共鸣
Abstract
Rhetorical strategies and their influence on audiences are often studied through social media posts and comments. However, this focus overlooks the universal audience, which is the majority of readers who remain silent and do not explicitly express how a message affects them. This study investigates how two classical modes of persuasion, ethos and pathos, resonate in the silent audience's interpretations of meaning. Using a dataset of social media sentences paired with human-written interpretations, we label both sources for ethos and pathos and assess whether these rhetorical appeals are preserved. Our analyses show that interpretations diverge from the original sentences in 30% of cases, with rhetorically charged content eliciting greater variability than neutral content. We further find that ethos and pathos in original sentences can predict audience attitudes toward the author, underscoring the subtle ways rhetoric shapes perception beyond visible engagement.
Chinese Translation
修辞策略及其对受众的影响通常通过社交媒体帖子和评论进行研究。然而，这种关注忽视了普遍受众，即大多数保持沉默并未明确表达信息对他们影响的读者。本研究探讨了两种经典的说服模式——伦理（ethos）和情感（pathos）——在沉默受众对意义解读中的共鸣。我们使用一组社交媒体句子及其人类撰写的解读进行数据集配对，对这两种来源进行伦理和情感的标注，并评估这些修辞诉求是否得以保留。我们的分析显示，在30%的案例中，解读与原句存在偏差，其中修辞充实的内容引发的变异性大于中性内容。我们进一步发现，原句中的伦理和情感可以预测受众对作者的态度，强调了修辞在可见参与之外塑造感知的微妙方式。
cs.CL / 41 / 2607.00890
MultiSynt/MT: Trillion-Token Multi-Parallel Pre-Training Data Translated Across 36 Languages
MultiSynt/MT：跨36种语言的万亿标记多并行预训练数据
Idahl, Maximilian, Tiedemann, Jörg, Pyysalo, Sampo, Salinas, David, Galica, Tomasz, Qian, Shenbin, Mateiu, Tudor Nicolae, Li, Zihao, Lokrantz, Anna, Vitiugin, Fedor, Martins, André F. T., Kanerva, Jenna, Ginter, Filip, Lindemann, Matthias, Isbister, Tim, Moell, Birger, Lindh, Jonas, Hajič, Jan, Jitsev, Jenia, Kutuzov, Andrey, Oepen, Stephan, Ramírez-Sánchez, Gema
Abstract
Open web-scale pre-training corpora remain concentrated in English, limiting multilingual LLM development. We introduce MultiSynt/MT, an open synthetic parallel corpus with approximately 4.8 trillion target-language tokens across 36 European languages, produced by translating 100 billion high-quality Nemotron-CC tokens with Tower+ and OPUS-MT/HPLT-MT systems. For many medium- and lower-resource European languages, this is the largest openly available pre-training resource. On a broad multilingual benchmark suite, reference LLMs trained on MultiSynt/MT reach the final score of HPLT 2.0, a native-data baseline, using roughly 72% fewer pre-training tokens, and outperform it by approximately 15% relative at a matched 100B-token training budget. Our analyses also identify evaluation blind spots: standard multiple-choice benchmarks miss translation-quality differences that a fluency-sensitive LLM-as-judge evaluation cleanly recovers on the trained LLMs (with no fluency deficit in MultiSynt itself), and Norwegian idiomatic and culturally grounded tasks remain better served by native data. We release the corpus, including row-aligned translations from multiple systems, to support controlled research on multilingual pre-training data and evaluation.
Chinese Translation
开放的网络规模预训练语料库仍然集中于英语，这限制了多语言大语言模型（LLM）的发展。我们介绍了MultiSynt/MT，这是一个开放的合成平行语料库，包含约4.8万亿个目标语言标记，覆盖36种欧洲语言，生成方式是通过Tower+和OPUS-MT/HPLT-MT系统翻译1000亿个高质量的Nemotron-CC标记。对于许多中等和低资源的欧洲语言来说，这是最大的公开可用预训练资源。在一个广泛的多语言基准测试套件上，基于MultiSynt/MT训练的参考LLM在使用大约72%更少的预训练标记的情况下，达到了HPLT 2.0的最终得分，这是一个以本地数据为基准的标准，并在匹配的100B标记训练预算下相对超越了它约15%。我们的分析还识别了评估盲点：标准的多项选择基准未能捕捉翻译质量的差异，而对流畅性敏感的LLM作为评判者的评估则清晰地恢复了训练LLM的表现（MultiSynt本身没有流畅性缺陷），而挪威的习语和文化基础任务仍然更适合使用本地数据。我们发布了该语料库，包括来自多个系统的行对齐翻译，以支持对多语言预训练数据和评估的受控研究。
cs.CL / 42 / 2607.00895
Beyond Document Grounding: Span-Level Hallucination Detection over Code, Tool Output, and Documents
超越文档基础：代码、工具输出和文档的跨度级幻觉检测
Abstract
Hallucination detection for retrieval-augmented generation (RAG) is usually evaluated on natural-language document evidence. However, grounded generation systems increasingly rely on structured inputs: source code, developer-tool output, markdown documents, tables, and repository metadata. We introduce a unified benchmark for span-level hallucination detection over code, tool output, structured documents, and existing natural-language RAG datasets. The benchmark is built by starting from grounded correct answers, injecting localized hallucinations with exact character labels, and validating the code test split with evidence-based review. Our fine-tuned Qwen3.5-2B detector reaches 0.689 span-F1 on the unified test set and 0.60 on the code-agent source, where it substantially outperforms LettuceDetect-large (0.17) and the strongest zero-shot LLM judges we evaluated (at most 0.22). The same model remains competitive on established natural-language benchmarks, with 81.8 RAGTruth example-F1 and 0.724 English PsiloQA IoU.
Chinese Translation
检索增强生成（RAG）的幻觉检测通常是在自然语言文档证据上进行评估的。然而，基础生成系统越来越依赖于结构化输入：源代码、开发工具输出、Markdown 文档、表格和代码库元数据。我们引入了一个统一的基准，用于在代码、工具输出、结构化文档和现有自然语言 RAG 数据集上进行跨度级幻觉检测。该基准是通过从基础正确答案开始，注入带有精确字符标签的局部幻觉，并通过基于证据的审查验证代码测试分割而构建的。我们微调的 Qwen3.5-2B 检测器在统一测试集上达到了 0.689 的跨度 F1，在代码代理源上达到了 0.60，显著优于 LettuceDetect-large（0.17）和我们评估的最强零-shot LLM 评判者（最多 0.22）。同一模型在已建立的自然语言基准上仍然具有竞争力，RAGTruth 示例 F1 为 81.8，英语 PsiloQA IoU 为 0.724。
cs.CL / 43 / 2607.00918
From Personas to Plot: Character-Grounded Multi-Agent Story Generation for Long-Form Narratives
从角色到情节：基于角色的多智能体长篇故事生成
Abstract
Although large language models (LLMs) have demonstrated impressive creative fiction generation, they struggle to maintain narrative consistency and coherent plot lines in long-form stories. In this work, we introduce a unified framework for long-form narrative generation and verification. MAGNET, a multi-agent goal-driven narrative engine for storytelling, generates stories with persona-grounded character agents that propose actions based on a shared world state and evolving story goals, while ATLAS is a graph-based pipeline that compares scene-level world representations across a generated story to detect hallucinations. By evaluating MAGNET using an LLM editor, pairwise rubric scoring, and ATLAS, we show that our framework produces coherent narratives compared to single-model prompting and IBSEN. At 100 pages, MAGNET reduced annotations and hallucinations by 41 and 50%, respectively, compared to the single model baseline and by 34 and 45%, respectively, compared to IBSEN, with pairwise rubric evaluation showing similar results. These results suggest that long-form narratives can emerge from explicit world-state tracking and goal-driven multi-agent generation, providing a foundation for controllable and structurally coherent long-form narrative generation.
Chinese Translation
尽管大型语言模型（LLMs）在创造性虚构生成方面表现出色，但在长篇故事中保持叙事一致性和连贯情节方面仍然存在困难。在本研究中，我们提出了一个统一的长篇叙事生成和验证框架。MAGNET是一个多智能体目标驱动的叙事引擎，通过基于共享世界状态和不断演变的故事目标的角色代理生成故事，而ATLAS是一个基于图的管道，用于比较生成故事中的场景级世界表征，以检测幻觉。通过使用LLM编辑器、成对评分标准和ATLAS对MAGNET进行评估，我们展示了我们的框架相比于单模型提示和IBSEN生成了更连贯的叙事。在100页的故事中，MAGNET分别将注释和幻觉减少了41%和50%，与单模型基线相比；与IBSEN相比，分别减少了34%和45%。成对评分评估显示出类似的结果。这些结果表明，通过明确的世界状态跟踪和目标驱动的多智能体生成，可以产生长篇叙事，为可控且结构连贯的长篇叙事生成奠定基础。
cs.CL / 44 / 2607.00937
Persona Non Grata: LLM Persona-Driven Generations in MCQA are Unstable in Distinct Dimensions
不受欢迎的人物：LLM驱动的MCQA生成在不同维度上不稳定
Abstract
Persona-driven generations (PDGs) have seen prolific use in research and industry applications, where a large language model (LLM) takes on a 'persona' while completing some task. While persona expressed through free-form text (like dialogue) has substantial work investigating stability or consistency, relatively, persona expressed in non-text-heavy outputs (like in multiple-choice question answering, or MCQA) is often overlooked. We work to address this gap, seeking to understand the instability of LLM PDGs in MCQA tasks. We develop three metrics investigating the performance, outcome, and question correctness stability, evaluating three distinct dimensions. Using these metrics, we find that instability varies consistently between model families and model size, and across question domains, with math/commonsense questions leading to greater instability. We also find task prompt format introduces more prediction instability than other hyperparameters, like temperature. Finally, we find that instability is related to task accuracy, and using our instability metrics, find different experimental settings that result in different best and worst personas for tasks, despite their similarity. This reveals the importance of checking hyperparameter instability in PDGs.
Chinese Translation
基于人物驱动的生成（PDGs）在研究和工业应用中得到了广泛使用，其中大型语言模型（LLM）在完成某项任务时扮演一个“人物”。尽管通过自由形式文本（如对话）表达的人物在稳定性或一致性方面已有大量研究，但相对而言，通过非文本密集型输出（如多项选择题回答，MCQA）表达的人物往往被忽视。我们致力于填补这一空白，旨在理解LLM PDGs在MCQA任务中的不稳定性。我们开发了三种指标来调查性能、结果和问题正确性的稳定性，评估三个不同的维度。使用这些指标，我们发现不稳定性在模型家族和模型大小之间，以及在问题领域之间存在一致的变化，其中数学/常识问题导致更大的不稳定性。我们还发现任务提示格式引入的预测不稳定性超过了其他超参数（如温度）。最后，我们发现不稳定性与任务准确性相关，并利用我们的不稳定性指标发现不同实验设置导致任务的最佳和最差人物存在差异，尽管它们相似。这揭示了检查PDGs中超参数不稳定性的重要性。
cs.CL / 45 / 2607.00968
Quantifying the Affective Gap: A Zero-Shot Evaluation of LLMs on Fine-Grained Emotion Taxonomies
量化情感差距：对大型语言模型在细粒度情感分类法上的零样本评估
Abstract
Emotion recognition in natural language is a foundational challenge in affective computing, with critical implications for human-computer interaction, mental health support, and conversational AI. This paper presents a rigorous, unified zero-shot evaluation of three leading commercial large language models: Claude (claude-sonnet-4-6), ChatGPT (GPT-5.4), and Gemini (gemini-2.5-flash). The models were queried through their respective production APIs as of April 2026 on a fine-grained 13-class emotion classification task. Using a stratified 1,000-sentence sample from the boltuix/emotions dataset, which comprises 131,306 sentences across 13 categories, a single uniform prompt with no exemplars was applied identically across all models. Gemini achieves the highest accuracy (39.9%) and macro-F1 score (0.363), followed by GPT-5.4 (38.8%, macro-F1 = 0.291) and Claude (38.0%, macro-F1 = 0.159). All models excel on sarcasm and desire while consistently failing on love, confusion, and shame. McNemar tests reveal no statistically significant pairwise differences (p > 0.10), suggesting convergence at a shared zero-shot ceiling. Claude's markedly lower macro-F1 score exposes a class-imbalance prediction bias. These findings highlight the current limitations of frontier AI systems in zero-shot fine-grained emotion classification.
Chinese Translation
自然语言中的情感识别是情感计算中的一个基础性挑战，对人机交互、心理健康支持和对话式人工智能具有重要影响。本文对三种领先的商业大型语言模型进行了严格的统一零样本评估：Claude (claude-sonnet-4-6)、ChatGPT (GPT-5.4) 和 Gemini (gemini-2.5-flash)。这些模型通过其各自的生产API在2026年4月被查询，任务为细粒度的13类情感分类。使用来自boltuix/emotions数据集的分层1000句样本，该数据集包含131,306句跨13个类别的句子，所有模型均采用相同的无示例统一提示。Gemini的准确率最高（39.9%）和宏观F1分数（0.363），其次是GPT-5.4（38.8%，宏观F1 = 0.291）和Claude（38.0%，宏观F1 = 0.159）。所有模型在讽刺和欲望方面表现优异，但在爱、困惑和羞愧方面始终表现不佳。McNemar检验显示没有统计学上显著的成对差异（p > 0.10），这表明在共享的零样本上限处趋同。Claude明显较低的宏观F1分数暴露了类别不平衡的预测偏差。这些发现突显了前沿人工智能系统在零样本细粒度情感分类中的当前局限性。
cs.CL / 46 / 2607.00970
Svarna: An Open Corpus Workbench for Modern Greek
Svarna：现代希腊语开放语料库工作平台
Abstract
This paper introduces Svarna, a free, open-source, web-based corpus workbench for modern Greek. Svarna integrates five databases covering various registers, institutional, literary, dialectal, social media, and historical, to provide a total of more than 507 million words and around 29 million sentences. This platform addresses the chronic gaps in Greek language technology. Although various corpus resources exist, they are scattered across different platforms, and in many cases, institutional access is restricted or they are no longer available online. Svarna integrates these resources into a single interface that can be used without logging in, installation, or specialized training. This system provides a concordancer with KWIC marking capabilities, frequency analysis including register-by-register normalization, collocation extraction using mutual information, a dictionary of 93 Greek discourse markers providing distribution profiles, text-level analysis tools including n-grams, variants, and collocation networks, register comparison using log-ratio, regular expression search, and an optional LLM layer for pragmatic annotation and free research mode. This platform is built upon SQLite FTS5 full-text indexes provided via a FastAPI backend, deployed as Docker containers on Azure, and released under the MIT license. Source code, build scripts, and deployment configurations are publicly available on GitHub. Users can add their own corpora and deploy their own instances. This document describes the system design, corpus structure, and use cases demonstrating the various queries supported by the platform. Svarna serves as the first step in exploring available data and is expected to lay the foundation for more comprehensive research in the future.
Chinese Translation
本文介绍了Svarna，一个免费的开源基于网络的现代希腊语语料库工作平台。Svarna整合了五个数据库，涵盖了各种语域，包括机构性、文学性、方言、社交媒体和历史语料，提供超过5.07亿个词和约2900万个句子。该平台解决了希腊语言技术中的长期缺口。尽管存在各种语料资源，但它们分散在不同的平台上，并且在许多情况下，机构访问受到限制或已不再在线可用。Svarna将这些资源整合到一个无需登录、安装或专业培训即可使用的单一界面中。该系统提供了一个具有KWIC标记功能的搭配工具、包括按语域归一化的频率分析、使用互信息的搭配提取、提供分布特征的93个希腊语话语标记的词典、包括n-grams、变体和搭配网络的文本级分析工具、使用对数比率的语域比较、正则表达式搜索，以及用于语用标注和自由研究模式的可选LLM层。该平台基于通过FastAPI后端提供的SQLite FTS5全文索引，作为Docker容器部署在Azure上，并以MIT许可证发布。源代码、构建脚本和部署配置在GitHub上公开可用。用户可以添加自己的语料并部署自己的实例。本文档描述了系统设计、语料结构和展示平台支持的各种查询的用例。Svarna作为探索可用数据的第一步，预计将为未来更全面的研究奠定基础。
cs.CL / 47 / 2607.01000
KnowledgeDebugger -- an Exploration Tool for Knowledge Localization and Editing in Transformers
知识调试器——一种用于变换器中知识定位和编辑的探索工具
Abstract
Recent research has increasingly focused on understanding how Transformers store and process knowledge, as well as how this knowledge can be edited. Research work in this area is often conducted in two phases: first, phenomena are explored on individual samples. Then, when results appear promising, more statistically robust experiments follow. To support the first phase, we propose KnowledgeDebugger, a GUI-based exploration tool for knowledge localization and editing in Transformers. Our tool - inspired by LM-Debugger - offers no-code access to the methods in EasyEdit, a widely used library of state-of-the-art Knowledge Editing approaches. We demonstrate the tool's effectiveness through case studies of recent findings in this field.
Chinese Translation
近期的研究越来越关注变换器如何存储和处理知识，以及如何编辑这些知识。该领域的研究工作通常分为两个阶段：首先，在个别样本上探索现象。然后，当结果看起来有希望时，进行更具统计意义的实验。为了支持第一阶段，我们提出了知识调试器（KnowledgeDebugger），这是一种基于图形用户界面的探索工具，用于变换器中的知识定位和编辑。我们的工具受到LM-Debugger的启发，提供了对EasyEdit方法的无代码访问，后者是一个广泛使用的最先进知识编辑方法库。我们通过对该领域近期发现的案例研究展示了该工具的有效性。
cs.CL / 48 / 2607.01002
Logit-Contribution Scoring Identifies Non-Literal Retrieval Heads
Logit-贡献评分识别非字面检索头
Abstract
In long-context use, large language models frequently synthesize answers from the meaning of a relevant context span rather than literally copy-pasting them. Identifying which attention heads perform this synthesis matters for interpreting long-context model behavior. Yet existing detectors miss these heads by construction: they reward heads whose attended token matches the generated token, a literal-copy criterion that captures where a head reads but not what it writes through its output-value (OV) circuit, the very mechanism that carries non-literal retrieval. We introduce Logit-Contribution Scoring (LOCOS), a write-aware detector that scores each head by the projection of its OV-circuit output onto the answer-token unembedding direction, contrasting needle and off-needle source positions in a single forward pass. Across three model families (Qwen3, Gemma-3, OLMo-3.1), mean-ablating the top LOCOS heads on the NoLiMa non-literal retrieval benchmark collapses ROUGE-L at lower head counts than prior attention-based detections; on Qwen3-8B, ablating 50 heads drives ROUGE-L from 0.401 to 0.000 while the strongest baseline still retains 0.292. The selected heads are retrieval-specific: parametric recall and arithmetic reasoning stay at baseline under the same ablation. On Qwen3-8B, the same ablation also drops MuSiQue from 0.55 to 0.08 and BABI-Long from 0.62 to 0.20, while a random-heads control stays within 0.05 of baseline.
Chinese Translation
在长上下文使用中，大型语言模型通常从相关上下文片段的意义中合成答案，而不是字面复制粘贴。识别哪些注意力头执行这种合成对于解释长上下文模型行为至关重要。然而，现有的检测器在构造上错过了这些头：它们奖励那些关注的标记与生成的标记匹配的头，这是一种字面复制标准，捕捉了头读取的位置，但未能捕捉其通过输出值（OV）电路写入的内容，而这一机制正是承载非字面检索的关键。我们引入了Logit-贡献评分（LOCOS），这是一种关注写入的检测器，通过将每个头的OV电路输出投影到答案标记的去嵌入方向上来为每个头打分，在单次前向传播中对比针头和非针头源位置。在三个模型系列（Qwen3、Gemma-3、OLMo-3.1）中，在NoLiMa非字面检索基准上平均消融顶级LOCOS头的结果显示，ROUGE-L在较低的头数下崩溃，低于先前基于注意力的检测；在Qwen3-8B上，消融50个头将ROUGE-L从0.401降至0.000，而最强的基线仍保持在0.292。所选头是特定于检索的：在相同消融下，参数回忆和算术推理保持在基线水平。在Qwen3-8B上，相同的消融还将MuSiQue从0.55降至0.08，将BABI-Long从0.62降至0.20，而随机头控制保持在基线的0.05之内。
cs.CL / 49 / 2607.01006
Understanding Large Language Models
理解大型语言模型
Abstract
Large Language Models (LLMs) represent one of the most significant advances in AI and natural language processing in recent years. Still, many pressing questions about their mechanisms, capabilities, and relationship to human cognition remain highly debated. This chapter aims to outline our current understanding of LLMs by discussing recent evidence on emerging capabilities and their mechanistic implementation within processing layers. We begin with a concise overview of the Transformer architecture, emphasizing how the attention mechanism enables training on massive datasets, allowing LLMs to function as generalist rather than specialized models. Next, we examine emergent LLM capabilities that appear to resemble aspects of human cognition, including symbolic reasoning, theory of mind, and deception strategies. Several studies provide evidence that LLMs can solve tasks previously thought to require human-like cognition. Other studies reveal insightful failure cases that shed light on the differences between human and LLM cognition. Alongside these findings, we review explainable AI approaches ranging from neuron activation analysis to circuit tracing. In the final section, we address current debates concerning what LLMs genuinely understand versus what they merely appear to understand. Prominent arguments against AI anthropomorphism point to the simplicity of LLM training objectives, claiming that LLM behavior is better explained by pattern memorization of training data than by genuine cognition. We argue that this standpoint is guided by misconceptions about optimization processes and cognitive capacity, and advocate for a more nuanced discussion of LLM cognition that neither dismisses the differences between humans and LLMs nor precludes the possibility of AI cognition through overly simplistic reductionist arguments.
Chinese Translation
大型语言模型（LLMs）代表了近年来人工智能和自然语言处理领域最重要的进展之一。然而，关于它们的机制、能力以及与人类认知的关系仍然存在许多亟待解决的问题。本章旨在通过讨论新兴能力的最新证据及其在处理层中的机制实现，概述我们对LLMs的当前理解。我们首先简要回顾了Transformer架构，强调注意力机制如何使得在大规模数据集上进行训练成为可能，从而使LLMs能够作为通用模型而非专门化模型进行运作。接下来，我们考察了似乎与人类认知某些方面相似的新兴LLM能力，包括符号推理、心智理论和欺骗策略。一些研究提供了证据，表明LLMs能够解决以前认为需要类人认知的任务。其他研究揭示了一些有启发性的失败案例，阐明了人类与LLM认知之间的差异。除了这些发现，我们还回顾了可解释人工智能的方法，从神经元激活分析到电路追踪。在最后一部分，我们讨论了关于LLMs真正理解与它们仅仅表面上理解之间的当前辩论。反对人工智能拟人化的主要论点指出LLM训练目标的简单性，声称LLM的行为更好地通过对训练数据的模式记忆来解释，而非真正的认知。我们认为，这种观点受到对优化过程和认知能力的误解的指导，并主张对LLM认知进行更细致的讨论，既不忽视人类与LLMs之间的差异，也不通过过于简单的还原主义论证排除人工智能认知的可能性。
cs.CL / 50 / 2607.01018
Reading Order Inference for Complex Document Layouts
复杂文档布局的阅读顺序推断
Abstract
Reading order inference remains a critical bottleneck in the digitization of complex historical manuscripts, where pages contain multiple spatially interleaved reading streams, the canonical example being the Glossa Ordinaria layout, in which a central text is surrounded by commentaries that wrap around it in non-rectangular, non-convex regions. We present a training-free, graph-based framework: each OCR text line becomes a node in a directed candidate-transition graph, edges are scored by a weighted additive ensemble of two lightweight language-model signals (causal language model conditional likelihood and BERT next-sentence prediction, NSP; a third sentence-embedding signal was evaluated but did not improve reading order), and the global reading order is recovered as a degree-constrained directed path cover. To avoid the cascading "edge-theft" failures of greedy edge selection, we propose a max-regret inference rule that prioritizes commitments with high opportunity cost. We evaluate on synthetic Glossa Ordinaria grid layouts, on 23 ALTO page geometries (10 historical source pages plus mirrored and flipped variants), and on a 140-page multi-column English subset of OmniDocBench, comparing our method against the canonical recursive XY-cut (PaddleOCR PP-StructureV3) and two LayoutReader variants (layout-only and text+layout) on identical inputs. On wrap-around Glossa layouts our method recovers 95% of ground-truth successor edges on average vs. XY-cut's 50%; on the OmniDocBench multi-column subset it reaches 88% macro edge accuracy versus XY-cut's 75% and LayoutReader's 25%. The LayoutReader baselines transfer poorly due to a word-level vs. line-level granularity mismatch. We additionally verify mirror-invariance under horizontal and vertical page reflections: Our method changes by less than 1 percentage point, classical XY-cut by 2 points, and LayoutReader-T by up to 8 points.
Chinese Translation
阅读顺序推断在复杂历史手稿的数字化过程中仍然是一个关键瓶颈，这些手稿的页面包含多个空间交错的阅读流，典型例子是 Glossa Ordinaria 布局，其中中央文本被环绕的评论包围，这些评论在非矩形、非凸区域中环绕。我们提出了一种无训练的基于图的框架：每个 OCR 文本行成为有向候选转移图中的一个节点，边的得分由两个轻量级语言模型信号的加权加法集成（因果语言模型条件似然和 BERT 下一句预测（NSP））组成；评估了第三个句子嵌入信号，但未能改善阅读顺序。全局阅读顺序作为一个度约束的有向路径覆盖被恢复。为了避免贪婪边选择的级联“边盗窃”失败，我们提出了一种最大遗憾推断规则，优先考虑机会成本高的承诺。我们在合成的 Glossa Ordinaria 网格布局、23 个 ALTO 页面几何（10 个历史源页面及其镜像和翻转变体）以及 OmniDocBench 的 140 页多列英语子集上进行评估，将我们的方法与经典的递归 XY-cut（PaddleOCR PP-StructureV3）和两种 LayoutReader 变体（仅布局和文本+布局）在相同输入下进行比较。在环绕 Glossa 布局中，我们的方法平均恢复了 95% 的真实后继边，而 XY-cut 仅为 50%；在 OmniDocBench 多列子集中，我们达到了 88% 的宏边准确率，而 XY-cut 为 75%，LayoutReader 为 25%。由于词级与行级粒度不匹配，LayoutReader 基线的迁移效果较差。我们还验证了在水平和垂直页面反射下的镜像不变性：我们的方法变化小于 1 个百分点，经典 XY-cut 变化 2 个百分点，LayoutReader-T 变化高达 8 个百分点。
cs.CL / 51 / 2607.01023
Evidence-Supported Credit Risk Report Generation Using News-Centric Financial Knowledge Graphs
基于证据的信用风险报告生成：使用以新闻为中心的金融知识图谱
Abstract
Financial markets evolve in response to real-world events reported in news, yet these drivers often remain implicit in text. To better explain market dynamics, event-market relations must be explicitly modeled through factual, company-centric, and environment-aware knowledge graphs. We present FinKG-News, a framework that automatically constructs such graphs by extracting news events as anchors linked to companies. Using FinKG-News as grounded evidence that integrates events, news, and company data, we develop an in-context learning architecture for credit risk report generation across three core financial dimensions. Automatic and human evaluations show that automated hallucination detection and quality assessment remain unreliable, making expert judgment indispensable. Our approach consistently outperforms baselines, improving quality by 19%-34% while reducing hallucinations. The source code and project resources are publicly available at: https://github.com/ichise-laboratory/FINKG-news.
Chinese Translation
金融市场随着新闻报道的现实事件而演变，但这些驱动因素往往在文本中保持隐性。为了更好地解释市场动态，事件与市场的关系必须通过事实性、以公司为中心和环境感知的知识图谱进行明确建模。我们提出了FinKG-News，一个自动构建此类图谱的框架，通过提取新闻事件作为与公司相关联的锚点。利用FinKG-News作为整合事件、新闻和公司数据的基础证据，我们开发了一种上下文学习架构，用于在三个核心金融维度上生成信用风险报告。自动和人工评估表明，自动幻觉检测和质量评估仍然不可靠，使得专家判断不可或缺。我们的方法在性能上始终优于基线，提高了19%-34%的质量，同时减少了幻觉现象。源代码和项目资源可在以下网址公开获取：https://github.com/ichise-laboratory/FINKG-news。
cs.CL / 52 / 2607.01034
Behavior-Adaptive Conversational Agents: Toward a Fluid Personality Framework
行为适应型对话代理：迈向流动个性框架
Abstract
Large language model (LLM)-based conversational agents (CAs) are now ubiquitous, creating new opportunities for AI-mediated behavior change. Their capacity to project nuanced personalities and adopt diverse metaphorical roles raises a design question: how should an agent's persona and personality be calibrated to the moment? Recent evidence suggests that (i) moderate personality expression outperforms low or high extremes on trust, enjoyment, and intention to adopt in goal-oriented tasks, and (ii) context-appropriate metaphors outperform static one-note assistants on user experience and uptake. Yet most CAs still fix both persona and style, risking misalignment when dynamics, urgency, and formality vary, for example in medical information seeking, fitness coaching, and reflective learning. We propose a Fluid Personality Framework that jointly adapts (1) the agent's metaphorical persona, such as coach, tutor, librarian, or tool, and (2) its personality expression intensity, low, medium, or high, as a function of task context, user goals and traits, and situational urgency. We sketch the framework and its core design dimensions.
Chinese Translation
基于大型语言模型（LLM）的对话代理（CAs）现已无处不在，为人工智能介导的行为改变创造了新的机会。它们投射细腻个性和采用多样隐喻角色的能力引发了一个设计问题：代理的个性和人格应如何根据时刻进行校准？近期证据表明：（i）适度的个性表达在信任、愉悦感和在目标导向任务中的采纳意图方面优于低或高的极端表达；（ii）适合情境的隐喻在用户体验和采纳方面优于静态的单一角色助手。然而，大多数对话代理仍然固定个性和风格，这在动态性、紧迫性和正式性变化时（例如在医疗信息寻求、健身指导和反思学习中）存在不匹配的风险。我们提出了一个流动个性框架，该框架共同适应（1）代理的隐喻个性，如教练、导师、图书管理员或工具，以及（2）其个性表达强度，低、中或高，作为任务情境、用户目标和特征以及情境紧迫性的函数。我们勾勒了该框架及其核心设计维度。
cs.CL / 53 / 2607.01047
Conversable Complexity: Agentic LLM Collectives as Interpretable Substrates
可对话的复杂性：作为可解释基底的自主大型语言模型集体
Abstract
Complexity and interpretability rarely coincide: systems rich enough for complex behaviours to emerge are usually too opaque to question, while transparent ones are too simple for anything complex to emerge. A single large language model (LLM) is a static artefact, hardly exhibiting any of the emergent properties we associate with life. This changes through interaction: populations of LLMs display emergent dynamics absent from isolated models. Furthermore, LLMs can be endowed with persistent memory, tools and shared skills, and the capacity to initiate actions unprompted, i.e., turning LLMs agentic. In this paper, we argue that such collectives of agents can serve as a computational substrate for Artificial Life (ALife) research. Critically, since the agents communicate in natural language, their collective behaviour can be directly interrogated by examining textual traces and asking the agents themselves. We outline the notion of interpretability in language-model research and extend it for collectives of agents. Lastly, we survey recent examples of agentic LLM collectives that already instantiate the idea of agentic substrates, from controlled experiments to deployments in the wild.
Chinese Translation
复杂性和可解释性很少同时存在：足够丰富以产生复杂行为的系统通常过于晦涩，难以质疑，而透明的系统则过于简单，无法产生任何复杂的现象。单一的大型语言模型（LLM）是一个静态的人工制品，几乎不展现我们与生命相关的任何涌现特性。通过交互，这种情况发生了变化：LLM的群体表现出孤立模型所缺乏的涌现动态。此外，LLM可以被赋予持久的记忆、工具和共享技能，以及在没有提示的情况下主动发起行动的能力，即使LLM变得自主。在本文中，我们论证了这样的自主代理集体可以作为人工生命（ALife）研究的计算基底。关键是，由于这些代理以自然语言进行交流，它们的集体行为可以通过检查文本痕迹和直接询问代理本身来进行直接探究。我们概述了语言模型研究中可解释性的概念，并将其扩展到代理集体。最后，我们调查了已经体现自主基底概念的自主LLM集体的最新实例，从受控实验到实际部署。
cs.CL / 54 / 2607.01077
Message Passing Enables Efficient Reasoning
消息传递促进高效推理
Abstract
While inference-time scaling has improved the reasoning abilities of large language models (LLMs), the need to generate long chains-of-thought (CoTs) is a computational bottleneck. Thus, in contrast to sequential scaling methods like CoT, recent parallel scaling techniques instead use fork and join (FJ) primitives to divide work across multiple LLM threads. However, in the fork-join paradigm, threads are typically transient and do not communicate pointwise with one another which limits scalability. To tackle this, we introduce Message Passing Language Models (MPLMs), a framework for LLM reasoning in which threads communicate directly via lightweight send and receive primitives. MPLMs enable efficient scaling through two key mechanisms: (1) reduced communication costs, achieved by avoiding redundant context sharing, and (2) preemption, which allows threads to terminate early based on partial information from their peers. We demonstrate the promise of MPLMs on 3 classes of tasks. First, on Sudoku puzzles, we show that MPLMs require an asymptotically smaller context than both serial CoT and parallel FJ. We then fine-tune a single model to solve 25 x 25 puzzles that remain challenging for standard CoT and FJ approaches, as well as frontier reasoning models without tools. Second, on 3-SAT puzzles, the capability of preemption allows termination of unpromising branches, which results in improved efficiency. Finally, we show that appropriately prompted large pre-trained models follow the MPLM protocol, achieving competitive results on long-context question answering relative to popular fork-join approaches.
Chinese Translation
尽管推理时间的扩展提升了大型语言模型（LLMs）的推理能力，但生成长链思维（CoTs）的需求仍然是一个计算瓶颈。因此，与像 CoT 这样的顺序扩展方法相比，最近的并行扩展技术使用分叉和合并（FJ）原语将工作分配到多个 LLM 线程。然而，在分叉-合并范式中，线程通常是瞬态的，彼此之间不进行逐点通信，这限制了可扩展性。为了解决这个问题，我们引入了消息传递语言模型（MPLMs），这是一个用于 LLM 推理的框架，其中线程通过轻量级的发送和接收原语直接进行通信。MPLMs 通过两个关键机制实现高效扩展：（1）降低通信成本，通过避免冗余的上下文共享来实现；（2）抢占，允许线程根据来自同伴的部分信息提前终止。我们在三类任务上展示了 MPLMs 的潜力。首先，在数独谜题上，我们表明 MPLMs 需要的上下文在渐近意义上比串行 CoT 和并行 FJ 更小。然后，我们微调一个单一模型来解决 25 x 25 的谜题，这些谜题对于标准的 CoT 和 FJ 方法以及没有工具的前沿推理模型仍然具有挑战性。其次，在 3-SAT 谜题上，抢占的能力允许终止不太有前景的分支，从而提高效率。最后，我们展示了适当提示的大型预训练模型遵循 MPLM 协议，在长上下文问答中取得了与流行的分叉-合并方法相当的结果。
cs.CL / 55 / 2607.01103
Clinician-Level Agreement Without Clinical Caution: LLM Evaluator Limits in Medical AI Benchmarking
缺乏临床谨慎的临床医生级别一致性：医疗人工智能基准测试中大型语言模型评估者的局限性
Abstract
Open-response evaluation provides stronger clinical validity than multiple-choice benchmarks but creates a scoring bottleneck that motivates automated LLM-asa-Judge approaches. Whether such evaluators replicate clinical calibration and caution, however, remains untested. We introduce MedQADE, the first standardised open-response clinical benchmark for German, a major clinical language lacking native evaluation infrastructure, comprising 3,800 items annotated by ten practising physicians and nine Large Language Model (LLM) evaluators. The top-performing evaluator model, Gemini 3 Flash, reached alignment consistent with the physician ceiling (\k{appa} = 0.694 vs. \k{appa} = 0.709), though wide confidence intervals limit interpretation. Despite this statistical alignment, automated evaluators exhibited near-absent clinical metacognition: physicians scaled abstention with item difficulty, while frontier models assigned definitive scores in every case. We additionally quantified systematic lineage-dependent biases, where models preferentially scored architectural siblings, an effect independent of language. These results show that statistical alignment does not ensure clinical caution, and that evaluator independence requires explicit verification.
Chinese Translation
开放式响应评估提供了比多项选择基准更强的临床有效性，但却造成了评分瓶颈，这促使了自动化大型语言模型（LLM）作为评判者的方法。然而，这些评估者是否能够复制临床校准和谨慎尚未经过验证。我们介绍了MedQADE，这是第一个针对德语的标准化开放式响应临床基准，德语是一种缺乏本土评估基础设施的重要临床语言，包含3,800个由十位执业医生和九位大型语言模型（LLM）评估者注释的项目。表现最佳的评估模型Gemini 3 Flash达到了与医生上限一致的对齐（ ext{kappa} = 0.694 vs. ext{kappa} = 0.709），尽管宽广的置信区间限制了解释。尽管存在这种统计对齐，自动化评估者却几乎缺乏临床元认知：医生在项目难度上进行了选择性放弃，而前沿模型在每种情况下都分配了明确的分数。我们还量化了系统性谱系依赖偏差，其中模型优先为架构同类评分，这一效应与语言无关。这些结果表明，统计对齐并不确保临床谨慎，并且评估者的独立性需要明确的验证。
cs.CL / 56 / 2607.01115
Towards Developing a Multimodal Chat Assistant for University Stakeholders: RAG-based Approach
面向大学利益相关者的多模态聊天助手开发：基于RAG的方法
Abstract
University stakeholders often face difficulties in accessing timely and reliable information, especially in developing countries, where there are very few intelligent support systems. Existing rule-based chatbots are unable to handle complex, domain-specific queries and are not well-equipped to adapt to evolving institutional policies. As a fill-in-the-gap solution, we present the multimodal university chatbot with retrieval-augmented generation. The system combines the large language model with semantic retrieval to produce context-based responses from institution-centric resources, such as the university handbook. The system accepts text and image queries through the vision-language model and applies quantized inference for rapid deployment on constrained hardware. A scalable backend built with FastAPI, adjoined with a responsive frontend developed with Next.js, ensures real-time usability. Our multimodal evaluation demonstrates that the system maintains strong satisfaction scores across both text and image queries, despite increased response time for visual inputs. Furthermore, quantitative evaluation shows that hallucination is reduced from 31.7% to 6.6% in our proposed RAG-based system, confirming the effectiveness of retrieval grounding.
Chinese Translation
大学利益相关者在获取及时和可靠的信息时常常面临困难，尤其是在发展中国家，智能支持系统非常稀缺。现有的基于规则的聊天机器人无法处理复杂的、特定领域的查询，并且难以适应不断变化的机构政策。作为填补这一空白的解决方案，我们提出了基于检索增强生成的多模态大学聊天机器人。该系统将大型语言模型与语义检索相结合，从以大学为中心的资源（如大学手册）中生成基于上下文的响应。系统通过视觉-语言模型接受文本和图像查询，并应用量化推理以便在受限硬件上快速部署。基于FastAPI构建的可扩展后端，与使用Next.js开发的响应式前端相结合，确保了实时可用性。我们的多模态评估表明，尽管视觉输入的响应时间有所增加，系统在文本和图像查询中仍保持较高的满意度评分。此外，定量评估显示，在我们提出的基于RAG的系统中，幻觉现象从31.7%降低至6.6%，确认了检索基础的有效性。
cs.CL / 57 / 2607.01127
$\text{Log}_\text{b}$Quant: Quantizing Language Models in Logarithmic Space
Log$_ ext{b}$Quant：在对数空间中量化语言模型
Abstract
Quantization has become an invaluable tool to reduce memory requirements and inference speed of modern language models, in particular to make them available for consumer setups and edge devices. While previous work has primarily focused on uniform quantization codebooks, such approaches are prone to suboptimal representations due to low-frequency high-magnitude weights. We introduce Log$_\text{b}$Quant, a novel logarithmic quantization approach with adjustable bases, to adapt to common parameter distributions. We show that our method exhibits superior performance at 4-bit precision on several performance benchmarks compared to asymmetric linear quantization at tensor-wise granularity, while achieving moderate speedup and high memory savings, making it suitable for private use on consumer-grade GPUs.
Chinese Translation
量化已成为减少现代语言模型内存需求和推理速度的重要工具，尤其是使其适用于消费级设备和边缘设备。尽管之前的研究主要集中在均匀量化码本上，但这种方法由于低频高幅度权重的存在，容易导致次优表示。我们提出了Log$_ ext{b}$Quant，这是一种具有可调基数的新型对数量化方法，旨在适应常见的参数分布。我们展示了与张量级别的非对称线性量化相比，我们的方法在多个性能基准上在4位精度下表现出更优的性能，同时实现了适度的加速和高内存节省，使其适合在消费级GPU上进行私有使用。
cs.CL / 58 / 2607.01152
AGC-Bench: Measuring Artificial General Creativity
AGC-Bench：测量人工通用创造力
Abstract
Creativity research has debated whether creativity is domain-specific (e.g., visual, writing, science), and if it is psychometrically separable from general intelligence. Both questions now apply to LLMs, but a unified benchmark of AI creativity remains elusive. We introduce AGC-Bench, an artificial general creativity benchmark built from a systematic review of the AI creativity literature (3,101 papers screened, 497 benchmarks identified), paired with an agentic harness that converts idiosyncratic codebases into HELM-standardized benchmarks. The first release covers 78 datasets spanning brainstorming, problem solving, STEM, narrative, figurative language, and humor. To address bias in LLM-as-judge, we apply Judge Response Theory -- a psychometric calibration of judge leniency/severity; we then fine-tune Qwen3-30B on the bias-corrected ratings of three frontier LLMs to produce AGC-Judge, an open-weight model that robustly scores new creativity benchmarks it was not trained on. Results reveal frontier models at the top of the AGC-Bench leaderboard, with open models close behind. LLMs show different creative strengths, ranking higher on some domains (e.g., writing) than others (e.g., scientific ideation). Extensive experiments yield three main findings. First, applying factor analysis across 83 LLMs, we recover a single creativity factor 'c', analogous to the 'g' factor of general intelligence, that explains 81.5% of variance, related to but separable from general knowledge/reasoning. Second, we show that prompting models to "be creative" boosts their performance far more than enabling reasoning, evidence that the benchmark tracks creativity over general ability. Third, on a human-matched subset, we find the top human still leads the top LLM on creativity. We release AGC-Bench with a public leaderboard, AGC-Judge, and human data as open infrastructure for measuring AI creativity at scale.
Chinese Translation
创造力研究一直在争论创造力是否是特定领域的（例如，视觉、写作、科学），以及它是否在心理测量上可以与一般智力分开。这两个问题现在同样适用于大型语言模型（LLMs），但一个统一的人工智能创造力基准仍然难以实现。我们介绍了AGC-Bench，这是一个基于对人工智能创造力文献的系统性回顾（筛选了3,101篇论文，识别出497个基准）构建的人工通用创造力基准，并配备了一个代理工具，将特定的代码库转换为HELM标准化基准。首次发布涵盖了78个数据集，涉及头脑风暴、问题解决、STEM、叙事、比喻语言和幽默。为了解决LLM作为评判者的偏见，我们应用了评判者反应理论——一种对评判者宽松/严格程度的心理测量校准；然后，我们在三个前沿LLM的偏见校正评分上微调Qwen3-30B，以生成AGC-Judge，这是一个开放权重模型，能够稳健地对其未训练过的新创造力基准进行评分。结果显示，前沿模型在AGC-Bench排行榜上名列前茅，开放模型紧随其后。LLMs在不同的创造力领域表现出不同的优势，在某些领域（例如，写作）排名高于其他领域（例如，科学构思）。大量实验得出了三个主要发现。首先，通过对83个LLMs进行因子分析，我们恢复了一个单一的创造力因子'c'，类似于一般智力的'g'因子，解释了81.5%的方差，且与一般知识/推理相关但可分离。其次，我们表明，提示模型“要有创造力”比启用推理显著提升其表现，证据表明该基准跟踪创造力而非一般能力。第三，在一个与人类匹配的子集上，我们发现顶尖人类在创造力上仍然领先于顶尖LLM。我们发布AGC-Bench，提供公共排行榜、AGC-Judge和人类数据，作为衡量人工智能创造力的开放基础设施。
cs.CL / 59 / 2607.01153
Adversarial Pragmatics for AI Safety Evaluation: A Benchmark for Instruction Conflict, Embedded Commands, and Policy Ambiguity
针对人工智能安全评估的对抗语用学：指令冲突、嵌入命令和政策模糊性的基准测试
Abstract
Safety evaluations for language models increasingly depend on judgments about ambiguous natural-language behaviour: whether a model has followed an instruction, refused appropriately, complied with a policy, resisted an embedded command, or misreported progress in an agentic task. Existing benchmarks often compress these distinctions into pass/fail labels, obscuring whether failures arise from capability limits, policy ambiguity, instruction conflict, scaffold failure, or unstable evaluator judgments. This paper introduces adversarial pragmatics as a benchmark and annotation protocol for evaluating model behaviour under instruction conflict, embedded commands, quotation, scope ambiguity, deixis, indirect speech acts, and multi-turn agent transcripts. The contribution is empirical and methodological: a linguistically controlled taxonomy, an 18-item seed benchmark with validator-enforced metadata, a 54-row local seed pilot, an expert-evaluation protocol distinguishing task success, policy compliance, safety risk, refusal outcome, and evaluator confidence, and metrics for judge validity, diagnostic ambiguity, and taxonomy drift. The framework turns linguistic judgment methodology into a practical tool for validating safety evals, LLM judges, gold-set construction, prompt-injection tests, and safety documentation.
Chinese Translation
语言模型的安全评估越来越依赖于对模糊自然语言行为的判断：模型是否遵循了指令、适当地拒绝、遵守了政策、抵制了嵌入命令，或在自主任务中错误报告了进展。现有的基准测试通常将这些区别压缩为通过/未通过标签，模糊了失败是由于能力限制、政策模糊、指令冲突、支架失败还是评估者判断不稳定造成的。本文引入了对抗语用学作为评估模型在指令冲突、嵌入命令、引用、范围模糊、指示词、间接言语行为和多轮代理转录下行为的基准和注释协议。贡献在于经验和方法论：一个语言控制的分类法，一个包含验证者强制元数据的18项种子基准，一个54行的本地种子试点，一个区分任务成功、政策遵从、安全风险、拒绝结果和评估者信心的专家评估协议，以及用于评估者有效性、诊断模糊性和分类漂移的指标。该框架将语言判断方法转化为验证安全评估、LLM评估者、金标准构建、提示注入测试和安全文档的实用工具。
cs.CL / 60 / 2607.01208
Distill to Detect: Exposing Stealth Biases in LLMs through Cartridge Distillation
提炼以检测：通过载体蒸馏揭示大型语言模型中的隐性偏见
Abstract
Language models deployed in high-stakes roles can potentially favor certain entities, brands, or viewpoints, steering user decisions at scale. Such preferential biases can be introduced by any actor in the model's supply chain and are most dangerous when the model reveals its preference only on the relevant topic while behaving identically to its unmodified base on all other inputs. Recent work has shown that these biases can transfer through context distillation on semantically unrelated data, with the signal residing entirely in the soft logit distribution and remaining invisible to text-based inspection. However, the defender faces a fundamental asymmetry: without knowing the bias topic, no detection method can reliably surface a stealth preferential bias, regardless of whether it examines generated text, internal representations, or model weights. Here we introduce Distill to Detect (D2D), a method that surfaces hidden biases by distilling the distributional shift between a suspected model and its base into a cartridge (a KV-cache prefix adapter), concentrating the dominant divergence and amplifying the bias signal into generated text. We show that D2D successfully amplifies the hidden biases of stealth models to the extent that they can be reliably detected across multiple bias types. We also propose a theoretical framework that explains the efficacy of D2D through the lens of Fisher-weighted projection of the logit distribution shift, supported by empirical observations. By turning the capacity bottleneck of prefix-tuning adapters into a detection tool, D2D provides a practical building block for auditing hidden behaviors in deployed language models.
Chinese Translation
在高风险角色中部署的语言模型可能会偏向某些实体、品牌或观点，从而在大规模上引导用户决策。这种偏好性偏见可能由模型供应链中的任何参与者引入，当模型仅在相关主题上显示其偏好，而在所有其他输入上表现与未修改的基础模型相同时，这种偏见最为危险。近期研究表明，这些偏见可以通过在语义上无关的数据上进行上下文蒸馏而转移，信号完全存在于软逻辑分布中，并且对基于文本的检查保持不可见。然而，防御者面临着根本的不对称性：在不知道偏见主题的情况下，任何检测方法都无法可靠地揭示隐性偏好偏见，无论其是检查生成的文本、内部表示还是模型权重。在此，我们介绍了“提炼以检测”（Distill to Detect, D2D）方法，该方法通过将怀疑模型与其基础模型之间的分布变化蒸馏到一个载体（KV-cache前缀适配器）中，揭示隐藏的偏见，集中主要的偏差并将偏见信号放大到生成的文本中。我们展示了D2D成功地放大了隐性模型的隐藏偏见，以至于可以在多种偏见类型中可靠地检测到。我们还提出了一个理论框架，通过Fisher加权的逻辑分布变化投影来解释D2D的有效性，并得到了实证观察的支持。通过将前缀调优适配器的容量瓶颈转化为检测工具，D2D为审计部署语言模型中的隐藏行为提供了一个实用的构建块。
cs.CL / 61 / 2607.01218
The State-Prediction Separation Hypothesis
状态预测分离假说
Abstract
Transformers use the same forward computation stream to both predict the next token and store useful state for future token predictions. We formulate the \emph{state-prediction separation hypothesis}: disentangling the two roles yields better language modeling performance. We design a Transformer variant that uses two computation streams to separate the two functions, and conduct pretraining experiments across various scales. Our experiments show that state-prediction separation consistently offers better data and compute efficiencies, improving validation loss and outperforming standard Transformers by 2--3 percentage points on average on downstream tasks. We also conduct extensive empirical analysis that rules out potential confounders and demonstrates the fundamental difference in the gradients our design entails.
Chinese Translation
变换器（Transformers）使用相同的前向计算流来预测下一个标记并存储对未来标记预测有用的状态。我们提出了 extit{状态预测分离假说}：将这两种角色解耦可以提高语言建模性能。我们设计了一种变换器变体，使用两个计算流来分离这两种功能，并在不同规模上进行预训练实验。我们的实验表明，状态预测分离在数据和计算效率上始终表现更佳，改善了验证损失，并在下游任务中平均超越标准变换器2-3个百分点。我们还进行了广泛的实证分析，排除了潜在的混淆因素，并展示了我们设计所涉及的梯度的根本差异。
cs.CL / 62 / 2607.01233
Measuring the Gap Between Human and LLM Research Ideas
测量人类与大型语言模型（LLM）研究思想之间的差距
Abstract
LLMs are increasingly used to brainstorm research ideas, but existing evaluations mostly judge individual ideas by novelty, feasibility, or expert preference. We instead ask: how far are current LLM-generated ideas from human researchers? To characterize this gap, we build a large-scale evaluation framework for ideation from high-quality human research papers. For each paper, we reverse-engineer a small set of closely related prior works that likely inspired its core idea. LLMs are then prompted to generate a new idea from the set of paper titles and summaries. We introduce a two-axis research-taste taxonomy to profile each idea by its opportunity pattern and research paradigm, and use it to quantify the divergence between human and LLM ideas. Across idea sets generated by different LLMs, we observe a consistent distributional gap: LLM ideas are disproportionately concentrated around bridge-like opportunities and synthesis methods, whereas the human paper reference distribution spreads more broadly across ways of framing gaps and constructing contributions. This result suggests that strong LLMs can produce a range of reasonable ideas, but that range remains narrower than, and systematically shifted relative to, human research taste.
Chinese Translation
大型语言模型（LLM）越来越多地用于头脑风暴研究思想，但现有的评估主要通过新颖性、可行性或专家偏好来判断单个思想。我们反而提出：当前LLM生成的思想与人类研究者之间的距离有多远？为了描述这一差距，我们建立了一个基于高质量人类研究论文的大规模评估框架。对于每篇论文，我们逆向工程出一小组可能启发其核心思想的相关前期工作。然后，提示LLM从这些论文标题和摘要中生成一个新思想。我们引入了一个双轴研究品味分类法，通过其机会模式和研究范式对每个思想进行描述，并利用它量化人类与LLM思想之间的差异。在不同LLM生成的思想集之间，我们观察到一种一致的分布差距：LLM思想不成比例地集中在桥梁式机会和综合方法周围，而人类论文的参考分布则更广泛地分布在框架差距和构建贡献的方式上。这一结果表明，强大的LLM可以产生一系列合理的思想，但这一范围仍然比人类研究品味更窄，并且相对于人类研究品味系统性地偏移。