cs.RO / 1 / 2608.05215
VLAff: Vision-Language-Affordance Model for Unified Actionable Affordances
VLAff：统一可操作性能力的视觉-语言-可操作性模型
Abstract
Learning manipulation skills from human videos is promising for scalable robot learning. However, the embodiment mismatch between humans and robots makes this challenging. One promising solution is to learn object-centric actionable affordances that are embodiment-agnostic. In this work, we propose a framework that leverages egocentric human videos with state-of-the-art 3D Structure-from-Motion and hand mesh reconstruction to extract actionable affordances such as visual, grasp, and trajectory affordances that explicitly encode where to interact, how to grasp, and how to move. We construct EgoAffordance, a large-scale dataset comprising 204K episodes with 5.6M visual affordances and 11.6M grasp and trajectory affordances. Building on this, we introduce VLAff, a large vision-language model-based unified foundation model that learns cross-modal correlations across all actionable affordances. Given a visual observation and instruction, VLAff generates visual affordance heatmaps, grasp poses, and trajectories, which are then converted into directly executable actions by utilizing 3D scene information. Through extensive experiments, we demonstrate that VLAff not only achieves state-of-the-art performance on visual affordance prediction, but can also be effectively applied to real robot applications such as zero-shot manipulation and affordance-guided robot learning.
Chinese Translation
从人类视频中学习操作技能对于可扩展的机器人学习具有很大潜力。然而，人类与机器人之间的体现不匹配使得这一过程充满挑战。一种有前景的解决方案是学习与体现无关的以物体为中心的可操作性能力。在本研究中，我们提出了一个框架，该框架利用以自我为中心的人类视频，结合最先进的三维运动结构重建和手部网格重建，提取可操作性能力，例如视觉能力、抓取能力和轨迹能力，这些能力明确编码了如何交互、如何抓取和如何移动。我们构建了EgoAffordance，这是一个大规模数据集，包含204K个情节，5.6M个视觉能力和11.6M个抓取及轨迹能力。在此基础上，我们引入了VLAff，一个基于大型视觉-语言模型的统一基础模型，学习所有可操作性能力之间的跨模态关联。给定视觉观察和指令，VLAff生成视觉能力热图、抓取姿势和轨迹，然后利用三维场景信息将其转换为可直接执行的动作。通过广泛的实验，我们证明了VLAff不仅在视觉能力预测上达到了最先进的性能，而且还可以有效应用于实际机器人应用，如零样本操作和基于能力的机器人学习。
cs.RO / 2 / 2608.05221
A System for Train Condition Monitoring and Structural Health Assessment of Rail Vehicles
一种用于列车状态监测和轨道车辆结构健康评估的系统
Abstract
The ongoing digitalization of rail systems and the increasing use of artificial intelligence (AI) are fundamentally transforming the design, operation, and maintenance of rail vehicles. While fully automated operation at Grade of Automation 4 (GoA4) is well established in metro systems, its deployment in mainline rail remains limited. This is primarily due to stringent safety requirements and the complexity of open operational environments. Current perception systems based on cameras, radar, and lidar are effective in detecting objects but provide limited capability for reliably identifying impacts, collisions, and driving-over events. This paper presents a novel approach for real-time vehicle condition monitoring and impact detection that integrates structural sensor technologies with AI-based data analysis. The proposed framework addresses three key applications: (1) automated detection of impacts, structural damage, and driving-over events, (2) condition-based maintenance enabled by continuous monitoring, and (3) long-term data analytics to support vehicle design optimization. The results demonstrate the feasibility of the proposed approach and highlight its potential to enhance operational safety, enable predictive maintenance strategies, and support the transition toward fully automated operation in mainline rail systems
Chinese Translation
铁路系统的持续数字化和人工智能（AI）的日益应用正在从根本上改变轨道车辆的设计、运营和维护。尽管在地铁系统中，自动化操作已在自动化等级4（GoA4）下得到了良好的建立，但在干线铁路中的应用仍然有限。这主要是由于严格的安全要求和开放操作环境的复杂性。目前基于摄像头、雷达和激光雷达的感知系统在检测物体方面有效，但在可靠识别冲击、碰撞和驶过事件方面能力有限。本文提出了一种新颖的方法，结合结构传感器技术与基于AI的数据分析，实现实时车辆状态监测和冲击检测。所提出的框架涵盖三个关键应用：（1）自动检测冲击、结构损伤和驶过事件；（2）通过持续监测实现基于状态的维护；（3）长期数据分析以支持车辆设计优化。结果表明，所提方法的可行性，并突显其在增强操作安全性、实现预测性维护策略以及支持干线铁路系统向完全自动化操作过渡方面的潜力。
cs.RO / 3 / 2608.05313
Failing Gracefully: Mitigating Impact of Inevitable Robot Failures
优雅失败：减轻不可避免的机器人故障影响
Abstract
Service robots operate in household environments shared with humans, pets, and everyday objects, where they are highly susceptible to failures such as software crashes, hardware degradation, or unpredictable interactions. While roboticists strive to minimize failures, some remain inevitable, making it critical to mitigate their potential consequences for safe and reliable deployment. This paper introduces a novel safety formulation that evaluates both the probability of impactful interactions between robots and surrounding entities during failures, and the severity of their outcomes. By quantifying the impact of failures on different entities, our approach enables robots to make informed planning decisions that balance safety with task efficiency. To support systematic evaluation, we also present FailBench, a MuJoCo-based simulation framework for studying robot-environment interactions under diverse failure modes, including sensing issues and actuator malfunctions. Together, our safety formulation and FailBench provide a foundation for developing safer and more robust motion plans and learned policies in real-world household environments.
Chinese Translation
服务机器人在与人类、宠物和日常物品共享的家庭环境中运行，因而高度易受软件崩溃、硬件退化或不可预测的交互等故障的影响。尽管机器人研究人员努力减少故障的发生，但一些故障仍然是不可避免的，因此减轻其潜在后果对于安全和可靠的部署至关重要。本文提出了一种新颖的安全性公式，该公式评估机器人在故障期间与周围实体之间发生影响性交互的概率及其结果的严重性。通过量化故障对不同实体的影响，我们的方法使机器人能够做出明智的规划决策，从而在安全与任务效率之间取得平衡。为了支持系统的评估，我们还介绍了FailBench，这是一个基于MuJoCo的仿真框架，用于研究在多种故障模式下的机器人与环境的交互，包括传感问题和执行器故障。我们的安全性公式和FailBench共同为在现实家庭环境中开发更安全、更稳健的运动计划和学习策略奠定了基础。
cs.RO / 4 / 2608.05365
Unified Planning-Learning Framework for Robust UUV Navigation Under Partial Observability
统一规划-学习框架用于在部分可观测性下的鲁棒性无人水下航行
Abstract
This paper presents an observation-only autonomy framework for Unmanned Underwater Vehicles (UUVs) navigation in dynamic underwater environments that integrates persistent occupancy mapping, global clearance-aware planning, and risk-aware local control. The proposed pipeline constructs occupancy maps solely from onboard sonar and depth image observations, adapts a clearance-constrained global planner (GP) to provide long-horizon structure, and integrates a reinforcement learning (RL) policy to handle short-range tracking and reactive avoidance. To further support decision-making under partial observability, the system learns a compact latent state representation from onboard sensor data, encoding environmental structure, obstacle dynamics, and uncertainty. Behavior tree (BT) distillation with staged supervision is introduced to improve safety and training stability, while an uncertainty-calibrated distillation mechanism reweights teacher guidance using online latent-model uncertainty, emphasizing uncertain regimes during learning, with time-to-collision (TTC) and clearance cues remaining explicit in planning and local policy features. To demonstrate the efficacy of the framework, a reproducible multi-seed evaluation protocol is established in high-fidelity GPU-accelerated simulation using NVIDIA Isaac Sim, and performance is benchmarked against BT-only and standard RL baselines. The results obtained demonstrate improved robustness and safety under dynamic conditions, thus providing a general pipeline with a unified hybrid planning learning architecture and a reproducible methodology for robust UUV autonomy under partial observability.
Chinese Translation
本文提出了一种仅基于观察的自主框架，用于无人水下航行器（UUV）在动态水下环境中的导航，该框架集成了持久性占用映射、全球清晰度感知规划和风险感知局部控制。所提出的流程仅通过机载声纳和深度图像观察构建占用地图，适应清晰度约束的全球规划器（GP）以提供长时间范围的结构，并集成强化学习（RL）策略以处理短距离跟踪和反应性避障。为了进一步支持在部分可观测性下的决策，系统从机载传感器数据中学习紧凑的潜在状态表示，编码环境结构、障碍物动态和不确定性。引入了行为树（BT）蒸馏与分阶段监督，以提高安全性和训练稳定性，同时不确定性校准的蒸馏机制使用在线潜在模型的不确定性重新加权教师指导，强调学习过程中的不确定状态，同时碰撞时间（TTC）和清晰度线索在规划和局部策略特征中保持显性。为了验证该框架的有效性，在使用NVIDIA Isaac Sim的高保真GPU加速仿真中建立了可重复的多种子评估协议，并将性能与仅使用BT和标准RL基线进行基准测试。获得的结果表明，在动态条件下提高了鲁棒性和安全性，从而提供了一个具有统一混合规划学习架构和可重复方法论的通用流程，以实现部分可观测性下的鲁棒UUV自主性。
cs.RO / 5 / 2608.05369
World-to-Wrist: Task-Conditioned Future Wrist Modeling for Fine-Grained Robot Manipulation
从世界到手腕：针对精细机器人操作的任务条件未来手腕建模
Abstract
Vision-language-action (VLA) models often treat main-view and wrist-view observations as parallel visual inputs, overlooking their distinct roles in robot manipulation. Fine-grained manipulation, however, benefits from anticipating how wrist-local interactions may evolve under the global task context. To address this limitation, we present World-to-Wrist VLA (W2-VLA), a VLA model for fine-grained robot manipulation with task-conditioned future wrist modeling. Given current multi-view observations and a task instruction, W2-VLA contextualizes a set of latent modeling tokens as a compact interface between the vision-language model and the wrist predictor. Conditioned on this interface and the observed wrist history, the predictor forecasts future wrist latents, which are transformed into future-aware context for action prediction. In addition, we introduce W2-CoT, a synthesis pipeline that produces structured annotations describing manipulation progress, physical transition cues, and wrist-local evidence. These annotations provide auxiliary supervision that shapes the task-conditioned latent interface. Experiments on LIBERO, RoboTwin 2.0, and real-world manipulation tasks demonstrate improved fine-grained and contact-sensitive manipulation across both single-arm and bimanual settings, while maintaining action-generation rates above 80 Hz.
Chinese Translation
视觉-语言-动作（VLA）模型通常将主视图和手腕视图观察视为平行视觉输入，忽视了它们在机器人操作中的不同角色。然而，精细操作受益于预测手腕局部交互在全球任务背景下如何演变。为了解决这一局限性，我们提出了世界到手腕 VLA（W2-VLA），这是一个用于精细机器人操作的 VLA 模型，具备任务条件的未来手腕建模。给定当前的多视图观察和任务指令，W2-VLA 将一组潜在建模标记上下文化为视觉-语言模型与手腕预测器之间的紧凑接口。在这个接口和观察到的手腕历史的条件下，预测器预测未来的手腕潜在状态，这些状态被转化为用于动作预测的未来感知上下文。此外，我们引入了 W2-CoT，一个合成管道，生成描述操作进展、物理过渡线索和手腕局部证据的结构化注释。这些注释提供了辅助监督，塑造了任务条件的潜在接口。在 LIBERO、RoboTwin 2.0 和真实世界操作任务上的实验表明，在单臂和双臂设置中，精细和接触敏感的操作得到了改善，同时保持了超过 80 Hz 的动作生成速率。
cs.RO / 6 / 2608.05410
Sliding Sensors: Configurable Confidence in State Estimation for Continuum Robots
滑动传感器：连续机器人状态估计中的可配置信心
Abstract
Continuum robots often operate in uncertain environments, where accurate state estimation is essential for safe interactions. Estimate uncertainty is inherently spatially non-uniform: confidence varies depending on where measurements are available. Global estimation accuracy is not always the top priority, but rather achieving sufficient confidence at task-relevant locations along the robot. This extended abstract introduces mechanically reconfigurable sensing enabling uncertainty-shaping in state estimation for continuum robots. We present a concept hardware design demonstrating the feasibility of longitudinal translation of a sensor within a continuum robot. We demonstrate that state estimation confidence can be reconfigured by varying the sensor location, and show a reduction of full-body shape estimation errors when sliding the sensor back and forth over time, compared to a single fixed tip sensor.
Chinese Translation
连续机器人通常在不确定的环境中操作，准确的状态估计对于安全交互至关重要。估计的不确定性在空间上本质上是不均匀的：信心取决于测量可用的位置。全局估计的准确性并不总是首要任务，而是在机器人沿任务相关位置实现足够的信心。本扩展摘要介绍了一种机械可重构传感器，能够在连续机器人的状态估计中实现不确定性塑形。我们展示了一种概念硬件设计，证明了在连续机器人内部纵向移动传感器的可行性。我们证明，通过改变传感器位置，可以重新配置状态估计的信心，并展示了在时间上前后滑动传感器时，与单个固定尖端传感器相比，全身形状估计误差的减少。
cs.RO / 7 / 2608.05579
ARGUS: Aligning Robot Scene Geometry Under Shifting Views with Large 3D Vision Models
ARGUS：在视角变化下对机器人场景几何进行对齐的大型3D视觉模型
Abstract
Large-scale visuomotor policies have demonstrated impressive performance across a wide range of robot manipulation tasks. However, despite this success, manipulation polices often entangle scene geometry with the corresponding viewpoint, learning where objects lie in an image rather than where it lies in the task space. This entanglement inherently limits the corresponding policy's ability to learn from viewpoint-diverse datasets (ex. DROID, BridgeV2) and generalize beyond the viewpoints captured in their training data. In this work, we present ARGUS, an observation pre-processing pipeline that uses large-scale 3D vision models to align image observations from arbitrary camera viewpoints into a canonical viewpoint before passing it to downstream visuomotor policies. Experiments across training datasets with varying levels of viewpoint diversity, from fixed multi-view camera configurations to highly varied camera placements, show that our method consistently outperforms prior approaches across both limited-view and view-diverse training regimes. In efficiency comparisons, ARGUS demonstrates an ability to learn from view-diverse data, converging to high success rates 4-6x faster than previous methods by leveraging a simplified observation space. Overall, our findings show that leveraging large-scale 3D vision models reduces the learning burden on visuomotor policies, enabling more efficient learning from large-scale, viewpoint-diverse robot datasets.
Chinese Translation
大规模视觉运动策略在各种机器人操作任务中表现出色。然而，尽管取得了这样的成功，操作策略往往将场景几何与相应的视角纠缠在一起，学习物体在图像中的位置，而不是在任务空间中的位置。这种纠缠本质上限制了相应策略从视角多样的数据集（例如 DROID、BridgeV2）中学习的能力，并且无法超越训练数据中捕获的视角进行泛化。在本研究中，我们提出了ARGUS，一个观察预处理管道，利用大规模3D视觉模型将来自任意相机视角的图像观察对齐到一个规范视角，然后再传递给下游视觉运动策略。在具有不同视角多样性的训练数据集上的实验，从固定的多视角相机配置到高度变化的相机放置，结果表明我们的方法在有限视角和视角多样的训练模式下均优于先前的方法。在效率比较中，ARGUS展示了从视角多样数据中学习的能力，利用简化的观察空间以4-6倍的速度收敛到高成功率，远快于以前的方法。总体而言，我们的研究结果表明，利用大规模3D视觉模型减轻了视觉运动策略的学习负担，使其能够更高效地从大规模、视角多样的机器人数据集中学习。
cs.RO / 8 / 2608.05586
PathCover: A Fast Convex Decomposition along a Path via Randomized Iterative Space Partitioning (RISP) on Point Clouds
PathCover：通过随机迭代空间划分（RISP）在点云上沿路径进行快速凸分解
Abstract
Autonomous robot navigation requires the rapid generation of obstacle-free regions for trajectory planning. However, existing corridor generators struggle to meet real-time, sensor-rate computational constraints. To resolve this bottleneck, we introduce PathCover, a framework driven by RISP; a novel randomized algorithm that constructs convex polytopes directly from raw point cloud data in expected linear time under a mild probabilistic elimination condition. PathCover generates sequences of overlapping, obstacle-free polytopes that safely constrain downstream MPC and trajectory optimization. We mathematically guarantee that the algorithm terminates in finite steps while ensuring continuous progress along any obstacle-free reference path. Extensive benchmarks on synthetic and real-world LiDAR datasets demonstrate an order-of-magnitude speedup over state-of-the-art methods while maintaining comparable corridor volumes. The complete pipeline is validated via high-fidelity quadrotor simulations and physical deployment on a quadrupedal robot navigating constrained environments using live LiDAR perception.
Chinese Translation
自主机器人导航需要快速生成无障碍区域以进行轨迹规划。然而，现有的走廊生成器在满足实时传感器计算约束方面面临挑战。为了解决这一瓶颈，我们引入了PathCover，这是一个由RISP驱动的框架；RISP是一种新颖的随机算法，它在温和的概率消除条件下，能够在预期线性时间内直接从原始点云数据构建凸多面体。PathCover生成一系列重叠的无障碍凸多面体，这些多面体安全地约束了下游的模型预测控制（MPC）和轨迹优化。我们在数学上保证该算法在有限步内终止，同时确保沿任何无障碍参考路径的持续进展。在合成和真实世界的激光雷达（LiDAR）数据集上的广泛基准测试表明，与最先进的方法相比，其速度提升了一个数量级，同时保持了相当的走廊体积。完整的管道通过高保真四旋翼模拟和在使用实时LiDAR感知的受限环境中导航的四足机器人上的物理部署得到了验证。
cs.RO / 9 / 2608.05588
Search-Aided Joint Agent-Environment Reinforcement Learning for Robust Lifelong Multi-Agent Path Finding with Rotations
基于搜索辅助的联合智能体-环境强化学习用于具有旋转的鲁棒终身多智能体路径规划
Abstract
Lifelong Multi-Agent Path Finding (LMAPF) requires repeatedly planning collision-free paths for agents that continuously receive new goals upon reaching their current ones. While many learning-based planners have been proposed for LMAPF, most rely on oversimplified kinematic assumptions that may overlook motion constraints critical to real-world performance. In this work, we study a more realistic LMAPF model derived from many real-world automated warehouse systems, termed LMAPF-R2, which incorporates robust safety constraints and in-place rotation constraints. These constraints substantially increase coordination difficulty, particularly in highly constrained spaces. To address these challenges, we propose Search-Aided Joint Reinforcement Learning (SJRL). We first augment neural policies with Causal PIBT, a single-step search-based planner that resolves agents' collisions and propagates their intentions. We then introduce a unified RL formulation that jointly optimizes agent and environment policies, where the environment policy learns graph edge costs to provide global movement guidance via backward Dijkstra search. Experiments demonstrate that SJRL achieves significant improvements over the strong search-based planner, Causal-PIBT, across multiple high-density maps. We further validate SJRL in a challenging mixed-reality warehouse environment with 8 physical robots and 248 virtual robots.
Chinese Translation
终身多智能体路径规划（LMAPF）需要为智能体反复规划无碰撞路径，这些智能体在到达当前目标后会不断接收新的目标。虽然已有许多基于学习的规划器被提出用于LMAPF，但大多数依赖于过于简化的运动学假设，这可能忽视对现实世界性能至关重要的运动约束。在本研究中，我们研究了一种更为现实的LMAPF模型，源自许多真实的自动化仓库系统，称为LMAPF-R2，该模型结合了鲁棒的安全约束和原地旋转约束。这些约束显著增加了协调的难度，特别是在高度受限的空间中。为了解决这些挑战，我们提出了基于搜索辅助的联合强化学习（SJRL）。我们首先用因果PIBT（Causal PIBT）增强神经策略，这是一种基于单步搜索的规划器，能够解决智能体之间的碰撞并传播它们的意图。然后，我们引入了一个统一的强化学习公式，联合优化智能体和环境策略，其中环境策略通过反向Dijkstra搜索学习图边成本，以提供全局移动指导。实验表明，SJRL在多个高密度地图上相较于强大的基于搜索的规划器Causal-PIBT取得了显著的改进。我们进一步在一个具有挑战性的混合现实仓库环境中验证了SJRL，该环境中有8个物理机器人和248个虚拟机器人。
cs.RO / 10 / 2608.05647
KILVO: Kinematic-Inertial-LiDAR-Visual Odometry with Robust Multimodal Adaptation for Humanoid Robots
KILVO：具有强健多模态适应性的类人机器人运动学-惯性-LiDAR-视觉里程计
Abstract
This article presents a kinematic-inertial-LiDAR-visual odometry for humanoid robots, called KILVO. Tailored to the platform features, requirements, and real-world complexity, it fully utilizes the sensors commonly equipped on humanoid robots, including joint encoders, IMU, LiDAR, and camera, within an asynchronous-sequential hybrid error-state iterated Kalman filter (ESIKF). Specifically, inertial data are used for prediction, leg kinematics are processed asynchronously at a high rate and provide proprioceptive constraints, while exteroception is updated sequentially, first by registering LiDAR points for geometric priors and then by updating the visual component via photometric errors. Moreover, the framework is elaborately designed with multimodal adaptation for resilience to sensor failures. A compact contact estimation module is also developed, sharing information with state estimation without additional sensors. Extensive experiments on public datasets and in the real world across multiple humanoid robots, gait patterns, and scenarios demonstrate that KILVO achieves highly competitive accuracy, efficiency, and output rates, with strong robustness against sensor degradation and failures, making it more suitable for humanoid robots than state-of-the-art fusion methods. Our code and datasets are released on GitHub.
Chinese Translation
本文提出了一种针对类人机器人的运动学-惯性-LiDAR-视觉里程计，称为KILVO。该方法针对平台特性、需求和现实世界的复杂性进行了定制，充分利用了类人机器人上常见的传感器，包括关节编码器、惯性测量单元（IMU）、LiDAR和相机，采用异步序列混合误差状态迭代卡尔曼滤波器（ESIKF）。具体而言，惯性数据用于预测，腿部运动学以高频率异步处理并提供本体感知约束，而外部感知则通过先注册LiDAR点以获取几何先验，然后通过光度误差更新视觉组件的方式进行顺序更新。此外，该框架经过精心设计，具有多模态适应性，以增强对传感器故障的韧性。还开发了一个紧凑的接触估计模块，与状态估计共享信息，而无需额外的传感器。在多个类人机器人、步态模式和场景下，针对公共数据集和现实世界进行的广泛实验表明，KILVO在准确性、效率和输出率方面具有高度竞争力，并对传感器退化和故障表现出强大的鲁棒性，使其比最先进的融合方法更适合类人机器人。我们的代码和数据集已在GitHub上发布。
cs.RO / 11 / 2608.05674
JoyAI-RA 0.5: Scaling Robot Manipulation Learning via Dual Action Alignment
JoyAI-RA 0.5：通过双重动作对齐扩展机器人操作学习
Abstract
Robot data is scarce, so generalist policies need to learn from heterogeneous sources, including human egocentric video, simulation, and real robots, which differ in supervision and embodiment, with action labels missing or mutually incompatible. Human egocentric data scale best but sit farthest from robot data, and naive pooling causes negative transfer rather than knowledge sharing. We propose JoyAI-RA 0.5, a generalist Vision-Language-World-Action (VLWA) framework that couples physical world-dynamics priors with visual semantics and scales manipulation learning across such data via dual action alignment. Implicit action alignment infers latent actions from visual transitions, enabling action-free human, simulation, and robot data to guide a latent-action-conditioned world model in learning physical dynamics. Explicit alignment grounds reliable human and robot trajectories in a unified physical action space through a canonical action representation and camera-frame chunk-relative end-effector actions. An inner-outer-loop reinforcement stage then pairs efficient task adaptation with foundation-policy improvement. On a real-world AgiBot benchmark, JoyAI-RA performs strongly on both seen tasks and unseen variations. The task score improves consistently as the volume of human egocentric pretraining data increases and shows no sign of plateauing at our largest scale. This suggests that abundant but weakly labeled human experience can be converted into a transferable training signal, making human video not merely a weak auxiliary source but a primary axis along which manipulation capability can be scaled. Project page can be found at https://joyai-ra-05.github.io/.
Chinese Translation
机器人数据稀缺，因此通用策略需要从异构来源学习，包括人类自我中心视频、仿真和真实机器人，这些来源在监督和体现上存在差异，并且动作标签缺失或相互不兼容。人类自我中心数据的规模最大，但与机器人数据相距最远，简单的汇聚会导致负迁移而非知识共享。我们提出了JoyAI-RA 0.5，这是一种通用的视觉-语言-世界-动作（Vision-Language-World-Action, VLWA）框架，它将物理世界动态先验与视觉语义相结合，通过双重动作对齐在这些数据中扩展操作学习。隐式动作对齐从视觉转变中推断潜在动作，使得无动作的人类、仿真和机器人数据能够引导一个潜在动作条件的世界模型来学习物理动态。显式对齐通过规范化动作表示和相机帧块相对末端执行器动作，将可靠的人类和机器人轨迹固定在统一的物理动作空间中。然后，内外循环强化阶段将高效的任务适应与基础策略改进相结合。在真实世界的AgiBot基准测试中，JoyAI-RA在已见任务和未见变体上均表现出色。随着人类自我中心预训练数据量的增加，任务得分持续提高，并且在我们最大的规模下没有出现停滞的迹象。这表明丰富但标签弱的人类经验可以转化为可转移的训练信号，使人类视频不仅仅是一个弱辅助来源，而是一个可以扩展操作能力的主要轴心。项目页面可以在 https://joyai-ra-05.github.io/ 找到。
cs.RO / 12 / 2608.05684
Nonvisual Classification of Ground-Condition by Artificial Proprioception in an Amoeba-Inspired Autonomous Walking Robot
基于人工本体感觉的非视觉地面状态分类：一种阿米巴启发的自主步行机器人
Abstract
Nonvisual classification of ground condition based on a multimodal sensing approach was investigated for an amoeba-inspired autonomous walking robot. To classify ground condition without image sensing and processing, we implemented artificial proprioception by integrating a three-axis accelerometer, eight foot pressure sensors, and reservoir computing (RC). Even when large fluctuations in the sensor outputs are caused by dynamic motions of a four-legged robot in walking, our system can classify the ground condition, flat or rough, with high accuracy. We demonstrate on-site switching of walking gait depending on ground condition in the robot. We also discuss the contribution of each sensor to ground condition classification.
Chinese Translation
本文研究了一种基于多模态感知方法的非视觉地面状态分类，应用于一种阿米巴启发的自主步行机器人。为了在没有图像感知和处理的情况下对地面状态进行分类，我们通过集成三轴加速度计、八个足部压力传感器和储层计算（Reservoir Computing, RC）实现了人工本体感觉。即使在四足机器人行走时，由于动态运动导致传感器输出出现大幅波动，我们的系统仍能以高精度分类地面状态为平坦或粗糙。我们展示了机器人根据地面状态现场切换行走步态的能力。同时，我们还讨论了每个传感器对地面状态分类的贡献。
cs.RO / 13 / 2608.05715
Hijacking Robots with a Piece of Paper: A Systematic Study of Physical Prompt Injection in VLM-Controlled Robots
用一张纸劫持机器人：对VLM控制机器人中物理提示注入的系统研究
Abstract
Vision-Language Models (VLMs) are increasingly deployed as planners in robotic systems, where they translate natural-language commands into executable actions grounded in visual scene understanding. This tight coupling between perception and instruction-following introduces a new attack surface: adversarial text placed within the robot's visual field can act as an indirect prompt injection into the VLM's reasoning stack. We present a systematic study of physical prompt injection attacks against VLM-controlled sorting, introducing a four-category taxonomy, indirect signage, task redefinition, authority impersonation, and conflict injection, instantiated as a benchmark of 20 attack prompts evaluated across three physical scene layouts and three command formulations that vary in destination specificity and rule explicitness. Across 5,670 trials on three frontier VLMs (GPT-4o, Gemini 2.5 Flash, Qwen3-VL-32B), attacks succeed at 27.0%, 29.4%, and 5.0% respectively, with authority-impersonating and negation attacks transferring across all three models. Analysis of reasoning traces reveals that successful compromise is nearly always conscious (99.9% acknowledgment rate), and that models defend through structurally different mechanisms, explicit rejection for Gemini, perceptual inattention for GPT-4o. We evaluate three simple mitigations: prompt-based defense (75-100% effective, model-dependent), two-stage verification (85-100%), and pre-processing text masking (100%). Our findings show that VLM-controlled manipulation is meaningfully vulnerable to human-readable physical signage, and that simple defenses substantially reduce risk, though defense choice involves trade-offs. The defenses preserve general task capabilities in our benchmark, but they may impair tasks that require reading in-scene labels.
Chinese Translation
视觉-语言模型（VLMs）越来越多地作为机器人系统中的规划者被部署，它们将自然语言命令转化为基于视觉场景理解的可执行动作。这种感知与指令执行之间的紧密耦合引入了一种新的攻击面：放置在机器人视觉范围内的对抗性文本可以作为对VLM推理堆栈的间接提示注入。我们对针对VLM控制的排序的物理提示注入攻击进行了系统研究，提出了四类分类法：间接标识、任务重新定义、权威冒充和冲突注入，并以20个攻击提示作为基准，在三个物理场景布局和三种命令表述中进行评估，这些命令在目标具体性和规则明确性上有所不同。在对三种前沿VLM（GPT-4o、Gemini 2.5 Flash、Qwen3-VL-32B）进行的5,670次试验中，攻击成功率分别为27.0%、29.4%和5.0%，其中权威冒充和否定攻击在所有三种模型中均有转移。推理轨迹的分析表明，成功的妥协几乎总是意识到的（99.9%的确认率），而模型通过结构上不同的机制进行防御，Gemini采用明确拒绝，GPT-4o则表现为感知上的不注意。我们评估了三种简单的缓解措施：基于提示的防御（75-100%有效，依赖于模型）、两阶段验证（85-100%）和预处理文本掩蔽（100%）。我们的研究结果表明，VLM控制的操作在面对可读的物理标识时存在显著脆弱性，且简单的防御措施显著降低了风险，尽管防御选择涉及权衡。这些防御在我们的基准测试中保留了一般任务能力，但可能会影响需要读取场景标签的任务。
cs.RO / 14 / 2608.05723
ATP: Anatomical Torque with Passivity-based Control Framework for Safe Upper-Limb Exoskeleton Assistance
ATP：基于解剖学扭矩的被动控制框架用于安全的上肢外骨骼辅助
Abstract
Providing assistance across diverse movements is a central objective of exoskeletons, and anatomical knowledge can enable responsive support that generalizes across tasks. However, anatomical assistance has mainly been studied for lower-limb exoskeletons, where periodic, weight-bearing motions impose lower demands on torque precision. Extending such assistance to complex, nonperiodic upper-limb movements remains challenging. This paper proposes Anatomical Torque with Passivity-Based Control (ATP) for safe upper-limb exoskeleton assistance. First, a scalable musculoskeletal simulation framework trains a unified reinforcement-learning muscle controller that generalizes across upper-limb movements and generates anatomical reference torques without complex biomechanical computations. Second, an online torque-refinement scheme adapts the reference to diverse movements, suppresses tendon-induced spikes, and incorporates a learned anomaly score for safe and comfortable assistance. Third, an interaction torque controller delivers assistance through a cable-driven compliant exoskeleton without constraining motion to predefined trajectories, while an energy tank preserves passivity with theoretical guarantees on torque tracking and system passivity. Simulations and real-world experiments show accurate tracking of long-duration motion sequences and generalization to real-time human movements. The controller achieves accurate torque tracking while preserving passivity and resumes tracking after energy-tank replenishment. An EMG study with five participants further shows reduced target-muscle activity during static and dynamic tasks compared with gravity compensation and open-loop assistance, with reductions of up to 48% relative to movement without the exoskeleton in a dynamic multi-joint task.
Chinese Translation
提供对多样化运动的辅助是外骨骼的核心目标，而解剖学知识可以实现对任务的响应性支持。然而，解剖学辅助主要集中在下肢外骨骼上，在这些情况下，周期性的、承重的运动对扭矩精度的要求较低。将这种辅助扩展到复杂的非周期性上肢运动仍然具有挑战性。本文提出了一种基于被动控制的解剖学扭矩（Anatomical Torque with Passivity-Based Control，ATP）用于安全的上肢外骨骼辅助。首先，一个可扩展的肌肉骨骼仿真框架训练一个统一的强化学习肌肉控制器，该控制器能够推广到上肢运动，并在没有复杂生物力学计算的情况下生成解剖学参考扭矩。其次，一个在线扭矩精细化方案将参考扭矩适应于多样化的运动，抑制肌腱引起的尖峰，并结合学习到的异常评分以实现安全和舒适的辅助。第三，交互扭矩控制器通过一个缆驱动的柔性外骨骼提供辅助，而不将运动限制在预定义的轨迹上，同时能量储存装置保持被动性，并在扭矩跟踪和系统被动性方面提供理论保证。仿真和现实世界实验表明，能够准确跟踪长时间的运动序列，并能推广到实时的人体运动。该控制器在保持被动性的同时实现了准确的扭矩跟踪，并在能量储存装置补充后恢复跟踪。一项涉及五名参与者的肌电图（EMG）研究进一步表明，在静态和动态任务中，目标肌肉活动相比于重力补偿和开环辅助有所减少，在动态多关节任务中，相比于没有外骨骼的运动，减少幅度高达48%。
cs.RO / 15 / 2608.05725
Near-sensor Computing for Rapid Visuotactile Perception
近传感器计算用于快速视觉触觉感知
Abstract
Visuotactile sensors reconstruct dense contact geometry from measured surface gradients, but host-based processing increases power consumption and introduces data-transfer delays and variable scheduling latency, limiting the sensing and response speed of robotic systems. To address these limitations, we implement a near-sensor computing framework that includes a spectral Poisson solver as a fully streaming hardware pipeline. The computational core logic has an estimated power consumption of 347 mW and achieves high throughput without data-dependent branching or iterative convergence, thereby providing deterministic latency. Operating at 166 MHz, the pipeline produces the first depth value of each 128x128 frame 35,107 cycles after receiving the first input pixel, corresponding to a fixed latency of 0.211 ms. Across 15 contact geometries, the reconstructed depths differ from a double-precision reference by 0.17 % of the peak contact depth. On-chip decisions based on these reconstructions close a robot protective reflex loop in 28.3 +/- 4.9 ms, compared with 169.9 +/- 27.8 ms for an equivalent host-based loop using the same actuator. These results demonstrate that near-sensor reconstruction can provide accurate, energy-efficient, and deterministic tactile geometry on timescales suitable for rapid robotic contact responses.
Chinese Translation
视觉触觉传感器通过测量表面梯度重建密集的接触几何形状，但基于主机的处理增加了功耗，并引入了数据传输延迟和可变调度延迟，限制了机器人系统的感知和响应速度。为了解决这些限制，我们实现了一种近传感器计算框架，其中包括一个作为全流式硬件管道的谱泊松求解器。计算核心逻辑的估计功耗为347毫瓦，且在没有数据依赖分支或迭代收敛的情况下实现了高吞吐量，从而提供确定性的延迟。在166 MHz的工作频率下，该管道在接收到第一个输入像素后，产生每个128x128帧的第一个深度值需时35,107个周期，对应固定延迟为0.211毫秒。在15种接触几何形状中，重建的深度与双精度参考值的差异为峰值接触深度的0.17%。基于这些重建的片上决策在28.3 +/- 4.9毫秒内关闭机器人保护反射回路，而使用相同执行器的等效主机回路则需时169.9 +/- 27.8毫秒。这些结果表明，近传感器重建能够在适合快速机器人接触响应的时间尺度上提供准确、节能和确定性的触觉几何形状。
cs.RO / 16 / 2608.05738
In-Context VLA: Endowing Vision-Language-Action Models with Language via In-Context Post-Training and Agentic Tool Use
上下文中的VLA：通过上下文后训练和自主工具使用赋予视觉-语言-行动模型语言能力
Abstract
Vision-Language-Action (VLA) models have become the dominant recipe for generalist manipulation, yet they are almost universally trained by behavior cloning: a policy imitates expert action chunks conditioned on a static image and a fixed instruction. A natural remedy is to inject explicit reasoning through textual chain-of-thought (CoT). We show, both empirically and analytically, that free-form textual CoT degrades low-level control: the reasoning it produces is ungrounded, its latency breaks closed-loop timing, and, crucially, the reasoning and action tokens are optimized against conflicting objectives so that the policy learns to narrate rather than to act. We argue that what a VLA needs is not the ability to generate language, but the ability to consume grounded language. To this end we introduce \textbf{\ourmethod{}}, a framework that endows a VLA with language competence through (i) in-context post-training, in which perceptual evidence is injected as structured context and the model is supervised only on actions, and (ii) an agentic tool-use interface, in which the policy queries open-vocabulary detectors, monocular depth, and a vision--language model to actively acquire task-relevant information. Rather than emitting a single templated caption, our data engine produces diverse, paraphrased, and evidence-conditioned spatial descriptions, so that the policy learns to interpret language it has never seen verbatim. Across the RoboCasa-GR1, SimplerEnv, and LIBERO simulation benchmarks, together with 8 real-world robot manipulation tasks, our method consistently achieves SOTA results in both performance and efficiency when compared with CoT-based approaches under matched configurations.
Chinese Translation
视觉-语言-行动（VLA）模型已成为通用操作的主流方法，然而它们几乎都是通过行为克隆进行训练：一种策略根据静态图像和固定指令模仿专家的动作片段。一个自然的解决方案是通过文本链式思维（CoT）注入明确的推理。我们通过实证和分析表明，自由形式的文本CoT会降低低级控制：其产生的推理是无基础的，其延迟打破了闭环时序，并且，关键是，推理和行动标记在相互冲突的目标下进行优化，从而使得策略学习叙述而不是行动。我们认为VLA所需的不是生成语言的能力，而是理解有基础语言的能力。为此，我们引入了 extbf{ hemethod{}}, 一个通过（i）上下文后训练赋予VLA语言能力的框架，在此过程中将感知证据注入为结构化上下文，并且模型仅在行动上进行监督，以及（ii）自主工具使用接口，在此接口中，策略查询开放词汇检测器、单目深度和视觉-语言模型，以主动获取与任务相关的信息。我们的数据引擎生成多样化、释义化和基于证据的空间描述，而不是发出单一的模板式标题，从而使得策略学习理解从未逐字见过的语言。在RoboCasa-GR1、SimplerEnv和LIBERO仿真基准测试中，以及8个现实世界的机器人操作任务中，我们的方法在性能和效率上始终在匹配配置下优于基于CoT的方法，取得了最先进的结果。
cs.RO / 17 / 2608.05746
Acoustic-driven millimetric helical robot: ultrasonic synergistic manipulation in confined fluidic environment
声驱动的毫米级螺旋机器人：在受限流体环境中的超声协同操控
Abstract
Acoustic field-driven manipulation provides a non-contact and non-invasive strategy for controlling microscale and nanoscale objects, yet its extension to millimeter-scale robots was limited by insufficient propulsion efficiency in confined biological environments. Here, a coordinated multi-acoustic-field approach is introduced, which harnesses the synergistic action of acoustic radiation forces and acoustic streaming flows to enable controlled locomotion of millimeter-scale helical robots and enhance propulsion. Multiphysics simulations captured the dynamics of millimeter-scale helical robots under combined acoustic fields, and experimental validation demonstrated their locomotion capabilities, including planar navigation, inclined climbing, and vertical motion. Semi-autonomous navigation experiments further confirmed that ultrasonic synergy substantially improved maneuverability. In vitro tests in porcine venous vessels demonstrated that coordinated acoustic fields supported both unidirectional and reciprocating motion under biologically relevant confinement. These findings provide mechanistic insight into scaling acoustic micromanipulation to the millimetre regime and support biomedical applications requiring versatile and controllable robotic mobility.
Chinese Translation
声场驱动的操控提供了一种非接触和非侵入性的策略，用于控制微尺度和纳米尺度的物体，但其在毫米级机器人上的应用受到在受限生物环境中推进效率不足的限制。在此，提出了一种协调的多声场方法，该方法利用声辐射力和声流的协同作用，使毫米级螺旋机器人的受控运动成为可能，并增强其推进能力。多物理场模拟捕捉了毫米级螺旋机器人在组合声场下的动态行为，实验验证展示了其运动能力，包括平面导航、倾斜爬升和垂直运动。半自主导航实验进一步确认超声协同显著提高了机动性。在猪静脉血管中的体外测试表明，协调的声场支持在生物相关的受限条件下进行单向和往复运动。这些发现为将声微操控扩展到毫米级提供了机制性见解，并支持需要多功能和可控机器人移动性的生物医学应用。
cs.RO / 18 / 2608.05799
XEWorld: Can Action-Conditioned World Models Generalize to Unseen Robot Embodiments?
XEWorld：动作条件下的世界模型能否推广到未见过的机器人形态？
Abstract
Action-conditioned world models are promising learned simulators for robotic manipulation, yet evaluating them exclusively on training robots fails to reveal whether they capture physical dynamics or merely memorize visual patterns. To answer whether a model can faithfully render a robot it has never seen, we introduce XEWorld, a controlled cross-embodiment testbed for world models that isolates embodiments by evaluating held-out robots within physically identical scenes. Our systematic analysis uncovers a shared architectural bottleneck: current models act primarily as 2D visual pattern matchers whose generalization is governed by visual similarity rather than physical kinematic similarity. Driven by this limitation, they struggle to translate abstract numeric joint actions into coherent visual trajectories, and fail to predict dynamic visual changes from static initial observations. Consequently, successfully rendering an unseen embodiment zero-shot strictly requires heavily grounded cues, specifically pixel-space actions and explicit spatial-temporal alignment. Even when bypassing this zero-shot barrier via few-shot adaptation, the forced appearance recovery triggers catastrophic forgetting of seen embodiments. Together, these failures expose a critical inability to apply learned physical dynamics to novel visual appearances, highlighting that achieving true cross-embodiment generalization requires architectural innovations that decouple visual appearance from underlying physical dynamics.
Chinese Translation
动作条件下的世界模型是用于机器人操作的有前景的学习模拟器，但仅在训练机器人上评估它们并不能揭示它们是否捕捉到了物理动态，还是仅仅记忆了视觉模式。为了回答一个模型是否能够真实呈现一个它从未见过的机器人，我们引入了XEWorld，这是一个受控的跨形态测试平台，用于世界模型，通过在物理上相同的场景中评估未见过的机器人来隔离形态。我们的系统分析揭示了一个共同的架构瓶颈：当前模型主要作为2D视觉模式匹配器，其泛化能力受视觉相似性而非物理运动学相似性的支配。受此限制，它们在将抽象的数值关节动作转化为连贯的视觉轨迹时遇到困难，并且无法从静态初始观察中预测动态视觉变化。因此，成功呈现一个未见过的形态在零样本情况下严格要求依赖于强有力的线索，特别是像素空间动作和明确的时空对齐。即使通过少样本适应绕过这一零样本障碍，强制的外观恢复也会导致已见形态的灾难性遗忘。这些失败共同揭示了将学习到的物理动态应用于新视觉外观的关键能力缺失，强调实现真正的跨形态泛化需要在架构上进行创新，以解耦视觉外观与基础物理动态。
cs.RO / 19 / 2608.05830
Coordinated Multi-Robot Disassembly for Makespan Optimization of Large-Scale Assemblies
协调多机器人拆解以优化大规模装配的完工时间
Abstract
Multi-robot task and motion planning for disassembly tasks requires robots to operate in confined workspaces while coordinating their motions with other robots. To tackle this problem, we propose a planning method called coordinated multi-robot disassembly (CoMuDi). CoMuDi coordinates a team of robots for disassembly tasks. The input is a team of robots, an assembly of objects, and a dependency graph. Based on this information, we create compound tasks for pick, place, and exit motions. By propagating temporal constraints, we ensure that each robot can start and end their tasks as early as possible while avoiding collisions with nearby robots. By integrating the space-time RRT* planner (ST-RRT*) into CoMuDi, we ensure that individual tasks minimize arrival time and thereby help us minimize overall makespan. We compare the performance of CoMuDi using both ST-RRT* and RRT* planners with varying time bounds, demonstrating that the combination of CoMuDi and ST-RRT* leads to a higher success rate while minimizing makespan. Finally, we evaluate CoMuDi on six assemblies with up to 49 pieces and up to 9 robots. In those scenarios, we show that CoMuDi returns robot paths that exhibit low idle times, thereby demonstrating that CoMuDi can reliably solve large-scale assemblies.
Chinese Translation
多机器人任务和运动规划在拆解任务中要求机器人在有限的工作空间内操作，同时与其他机器人协调其运动。为了解决这个问题，我们提出了一种称为协调多机器人拆解（CoMuDi）的规划方法。CoMuDi协调一组机器人进行拆解任务。输入包括一组机器人、一组物体的装配和一个依赖图。基于这些信息，我们为拾取、放置和退出动作创建复合任务。通过传播时间约束，我们确保每个机器人能够尽早开始和结束其任务，同时避免与附近机器人发生碰撞。通过将时空RRT*规划器（ST-RRT*）集成到CoMuDi中，我们确保各个任务最小化到达时间，从而帮助我们最小化整体完工时间。我们比较了使用ST-RRT*和RRT*规划器的CoMuDi在不同时间限制下的性能，结果表明CoMuDi与ST-RRT*的结合在最小化完工时间的同时提高了成功率。最后，我们在六个装配上进行了CoMuDi的评估，这些装配最多包含49个部件和9个机器人。在这些场景中，我们展示了CoMuDi返回的机器人路径具有较低的空闲时间，从而证明CoMuDi能够可靠地解决大规模装配问题。
cs.RO / 20 / 2608.05970
SkillMemo: Expert-guided Skill Memory Framework for Compositional Embodied Manipulation
SkillMemo：专家指导的技能记忆框架用于组合体态操控
Abstract
Embodied visuomotor models, including Diffusion Policy (DP) and Vision-Language-Action (VLA) models, have demonstrated promising performance on robotic manipulation benchmarks. However, their potential remains fundamentally constrained by the scarcity of large-scale embodied trajectory datasets, leading to insufficient compositional generalization in out-of-distribution (OOD) scenarios with limited capability to capture reusable skill structures. To address this limitation, we propose Skill-Based Memory (SkillMemo) framework that implicitly decomposes long-horizon demonstrations into latent atomic skills and integrates skill-level features into a dynamic episodic memory bank for solving compositional tasks. Specifically, we first introduce an expert-guided trajectory segmentation module built upon a Mixture-of-Experts (MoE) architecture, which implicitly partitions trajectories into distinct skill primitives represented by learned gating coefficients. We further design a skill-level episodic memory architecture that stores compact skill representations as retrievable key-value pairs. During inference, the memory bank retrieves the most relevant skill primitives which are subsequently fused with the model's current gating distribution, providing a robust contextual prior to refine action predictions. Extensive experiments on the simulation benchmark and real-world manipulation tasks demonstrate that SkillMemo consistently enhances both DP and VLA backbones, achieving state-of-the-art performance and outperforming $\pi_{0.5}$, while exhibiting strong compositional generalization to unseen task configurations.
Chinese Translation
体态视觉运动模型，包括扩散策略（Diffusion Policy, DP）和视觉-语言-行动（Vision-Language-Action, VLA）模型，在机器人操控基准测试中表现出良好的性能。然而，它们的潜力仍然受到大规模体态轨迹数据集稀缺的根本限制，导致在分布外（Out-of-Distribution, OOD）场景中组合泛化能力不足，无法有效捕捉可重用的技能结构。为了解决这一限制，我们提出了基于技能的记忆框架（Skill-Based Memory, SkillMemo），该框架隐式地将长时间演示分解为潜在的原子技能，并将技能级特征集成到动态的情节记忆库中，以解决组合任务。具体而言，我们首先引入一个基于专家指导的轨迹分割模块，该模块建立在混合专家（Mixture-of-Experts, MoE）架构之上，隐式地将轨迹划分为由学习的门控系数表示的不同技能原语。我们进一步设计了一个技能级情节记忆架构，将紧凑的技能表示存储为可检索的键值对。在推理过程中，记忆库检索最相关的技能原语，这些原语随后与模型当前的门控分布融合，为精炼动作预测提供了强大的上下文先验。在仿真基准和真实世界操控任务上的大量实验表明，SkillMemo始终增强了DP和VLA主干网络，达到了最先进的性能，并超越了$ ext{π}_{0.5}$，同时在未见任务配置中展现出强大的组合泛化能力。
cs.RO / 21 / 2608.05975
TRACE: Learned Proprioceptive Odometry for Legged Robots under Unreliable Contact Conditions
TRACE：在不可靠接触条件下的四足机器人学习式本体里程计
Abstract
In this paper, we present TRACE (Tokenized Robust Attention for Contact-Aware Estimation), an end-to-end learned proprioceptive odometry estimator for legged robots under unreliable contact conditions. The proposed estimator directly predicts relative displacement, relative rotation, and body-frame velocity from a recent history of onboard inertial and joint measurements. To improve robustness under unreliable contact conditions, we introduce a foot-aware cross-attention module that adaptively weights IMU and leg-wise kinematic tokens without relying on manually defined contact or slip thresholds. The estimator is trained with direct supervision and two physics-inspired auxiliary losses that promote kinematic consistency and reliable use of leg information. To reduce policy-specific overfitting and consequently improve sim-to-real transfer, simulation training incorporates policy randomization, followed by partial real-world fine-tuning of the temporal encoder and prediction head. Experiments across diverse indoor and outdoor terrains demonstrate consistent reductions in position drift compared with classical filtering-based, hybrid, and purely learning-based baselines. Ablation studies further validate the contributions of the proposed training objectives, policy randomization, and real-world fine-tuning, particularly under unreliable contacts and sim-to-real mismatch.
Chinese Translation
本文提出了TRACE（基于令牌的鲁棒注意力接触感知估计），这是一种针对在不可靠接触条件下的四足机器人的端到端学习式本体里程计估计器。所提出的估计器直接从近期的机载惯性和关节测量历史中预测相对位移、相对旋转和机体框架速度。为了提高在不可靠接触条件下的鲁棒性，我们引入了一种脚部感知的交叉注意力模块，该模块自适应地对IMU和腿部运动学令牌进行加权，而无需依赖手动定义的接触或滑移阈值。该估计器通过直接监督和两种受物理启发的辅助损失进行训练，这些损失促进运动学一致性和腿部信息的可靠使用。为了减少特定策略的过拟合并改善模拟到现实的迁移，模拟训练结合了策略随机化，随后对时间编码器和预测头进行部分真实世界的微调。在各种室内和室外地形上的实验表明，与经典的基于滤波的、混合的和纯学习的基线相比，位置漂移有了一致的减少。消融研究进一步验证了所提出的训练目标、策略随机化和真实世界微调的贡献，特别是在不可靠接触和模拟到现实的不匹配情况下。
cs.RO / 22 / 2608.05999
Beyond Flat Policies: Hierarchical Post-Training for Embodied Agents in Robotic Manipulation
超越平面策略：用于机器人操作的层次后训练方法
Abstract
Vision-language-action (VLA) models have demonstrated remarkable capabilities in robotic manipulation by leveraging pretrained vision-language models. However, existing post-training methods predominantly optimize VLA models as flat policies, making it difficult to explicitly model task progression and perform robust long-horizon manipulation. Although hierarchical approaches introduce task decomposition, they mainly rely on supervised learning from offline demonstrations and cannot effectively improve execution through online interaction. To address this limitation, we propose Hierarchical Robotic Control (HiRoC), a hierarchical post-training framework that decouples high-level task planning from low-level action execution. The planner decomposes complex tasks into executable subgoals to provide explicit semantic guidance, while the executor continuously improves subgoal-conditioned action generation through reinforcement learning. To enable effective collaboration between the two modules, we further align the executor with planner-generated subgoals before reinforcement learning, mitigating the distribution misalignment between planning and execution. Extensive experiments across diverse robotic manipulation benchmarks demonstrate that HiRoC consistently outperforms strong baselines. Comprehensive analyses further validate the effectiveness of hierarchical post-training and the contribution of each key component.
Chinese Translation
视觉-语言-动作（VLA）模型通过利用预训练的视觉-语言模型，在机器人操作中展现了显著的能力。然而，现有的后训练方法主要将VLA模型优化为平面策略，这使得明确建模任务进展和执行稳健的长时间操作变得困难。尽管层次化方法引入了任务分解，但它们主要依赖于从离线演示中进行的监督学习，无法通过在线交互有效改善执行。为了解决这一限制，我们提出了层次机器人控制（HiRoC），这是一种层次后训练框架，将高层次任务规划与低层次动作执行解耦。规划器将复杂任务分解为可执行的子目标，以提供明确的语义指导，而执行器通过强化学习不断改善子目标条件下的动作生成。为了促进两个模块之间的有效协作，我们在强化学习之前进一步将执行器与规划器生成的子目标对齐，从而减轻规划与执行之间的分布不一致性。在多种机器人操作基准上的广泛实验表明，HiRoC始终优于强基线。全面的分析进一步验证了层次后训练的有效性及其各个关键组件的贡献。
cs.RO / 23 / 2608.06008
Adaptive-WAM: Quality-Guided Early-Exit Planning from Intermediate Video-Diffusion Features
自适应-WAM：基于质量引导的中间视频扩散特征早期退出规划
Abstract
Large video diffusion models provide rich spatiotemporal priors for autonomous driving, but existing world-action models often inherit the cost of iterative future-video generation even though deployment only requires an ego trajectory. We ask a more basic question: how much of a video diffusion model must be executed to make a reliable driving decision? Through a controlled study of video denoising timesteps and Diffusion Transformer (DiT) depth, we find that planning performance is largely insensitive to the tested video-noise levels, whereas strong trajectories can already be decoded from intermediate layers. Based on this observation, we introduce Adaptive-WAM, a quality-aware multi-exit planner built on a Wan2.2-5B backbone. Trajectory diffusion heads are attached to selected DiT blocks, and a lightweight trajectory-quality scorer terminates inference once the best trajectory decoded so far satisfies a quality threshold; otherwise, computation continues from the cached hidden state to a deeper exit. The deployed planner therefore avoids the iterative classifier-free denoising loop and VAE decoding required for future-video synthesis, while dynamically allocating backbone depth according to trajectory quality. On NAVSIM, the adaptive single-trajectory planner achieves 90.8 PDMS; a separate fixed-exit variant reaches 92.6 PDMS with 64 proposals. It further obtains 89.9 EPDMS on NAVSIM v2, yielding the best reported results among the compared front-view video world-model planners. Without target-domain fine-tuning, Adaptive-WAM transfers to nuScenes with 0.88 m average L2 error and a 0.08\% collision rate. On an A100, adaptive routing improves PDMS from 90.62 to 90.79 while averaging 170 ms end-to-end planning latency, approximately 10\% below the 190 ms fixed block-15 planner and 47\% below the 320 ms fixed full-depth planner. Code will be released.
Chinese Translation
大型视频扩散模型为自主驾驶提供了丰富的时空先验，但现有的世界-动作模型往往继承了迭代未来视频生成的成本，尽管部署只需要自我轨迹。我们提出一个更基本的问题：为了做出可靠的驾驶决策，必须执行多少视频扩散模型？通过对视频去噪时间步和扩散变换器（Diffusion Transformer, DiT）深度的控制研究，我们发现规划性能对测试的视频噪声水平几乎不敏感，而强轨迹可以从中间层解码。基于这一观察，我们引入了自适应-WAM，一种基于Wan2.2-5B骨干网构建的质量感知多退出规划器。轨迹扩散头附加到选定的DiT块上，一个轻量级的轨迹质量评分器在解码到目前为止最佳轨迹满足质量阈值时终止推理；否则，计算将从缓存的隐藏状态继续到更深的退出。因此，部署的规划器避免了未来视频合成所需的迭代无分类器去噪循环和变分自编码器（VAE）解码，同时根据轨迹质量动态分配骨干深度。在NAVSIM上，自适应单轨迹规划器实现了90.8 PDMS；一个独立的固定退出变体在64个提案下达到了92.6 PDMS。它在NAVSIM v2上进一步获得了89.9 EPDMS，成为所比较的前视视频世界模型规划器中报告的最佳结果。在没有目标领域微调的情况下，自适应-WAM在nuScenes上转移时平均L2误差为0.88米，碰撞率为0.08%。在A100上，自适应路由将PDMS从90.62提高到90.79，同时平均端到端规划延迟为170毫秒，约比190毫秒的固定块-15规划器低10%，比320毫秒的固定全深度规划器低47%。代码将会发布。
cs.RO / 24 / 2608.06021
Topometric Autonomous Vehicle Localization by Combining Visual Embeddings and Feed-Forward 3D Models
通过结合视觉嵌入和前馈3D模型实现的拓扑自主车辆定位
Abstract
Effective Visual Localization (VL) requires a map of the environment that combines compactness for efficient scalability with robustness against visual appearance changes and metric precision. Through low-dimensional image embeddings, Visual Place Recognition (VPR) is able to successfully meet the first two requirements, but its low metric accuracy makes it less suitable than standard VL approaches based on local features or neural representations. This limitation can be overcome by integrating VPR with the accurate local trajectory estimates produced by feed-forward neural 3D geometry (FF3D) models. In this paper, we address sequential appearance-based localization through a topometric framework that iteratively combines probabilistic VPR with FF3D metric pose estimation in controlled image sets. Our approach proposes an automatic offline mapping tool that models the topometric pose-appearance interaction in the different parts of the scene. This map is later employed by an online particle filter that estimates the pose from odometry and belief over places for FF3D inference, successfully incorporating neural metric estimation into probabilistic appearance-based localization. We extensively evaluate the framework on three known benchmarks, demonstrating substantial improvements over existing appearance-based methods. The modularity of our approach allows the descriptor extractor and FF3D model to remain interchangeable, and a focused analysis further shows that sequential belief can mitigate severe failures under perceptual aliasing.
Chinese Translation
有效的视觉定位（VL）需要一个环境地图，该地图在高效可扩展性与对视觉外观变化的鲁棒性及度量精度之间实现平衡。通过低维图像嵌入，视觉地点识别（VPR）能够成功满足前两个要求，但其低度量精度使其不如基于局部特征或神经表示的标准VL方法适用。通过将VPR与前馈神经3D几何模型（FF3D）生成的准确局部轨迹估计相结合，可以克服这一限制。本文通过一个拓扑框架解决基于外观的序列定位，该框架在受控图像集上迭代地将概率VPR与FF3D度量位姿估计结合。我们的方法提出了一种自动离线映射工具，建模场景不同部分的拓扑位姿-外观交互。该地图随后被在线粒子滤波器使用，该滤波器根据里程计和对地点的信念进行位姿估计，以实现FF3D推理，成功地将神经度量估计纳入基于概率的外观定位。我们在三个已知基准上对该框架进行了广泛评估，显示出相较于现有基于外观的方法有显著改进。我们方法的模块化特性使得描述符提取器和FF3D模型可以互换，进一步的聚焦分析表明，序列信念可以缓解感知混淆下的严重失败。
cs.RO / 25 / 2608.06088
IcFuzz: Fuzzing Isaac Sim with Semantic Stage Guidance and Multi-level Mutation
IcFuzz：基于语义阶段指导和多级变异的Isaac Sim模糊测试
Abstract
Robotics simulators serve as a foundational infrastructure for embodied AI, facilitating safe and scalable robotic system development. NVIDIA Isaac Sim has emerged as one of the most popular simulators, distinguished by its GPU-accelerated physics engine and photorealistic rendering, which enable high-fidelity modeling of complex environments. However, its inherent complexity inevitably introduces software bugs that can compromise simulation reliability. Existing fuzzing approaches struggle to test Isaac Sim effectively due to challenges of context-aware object semantics, hierarchical simulation control, and a vast simulation state space. In this paper, we propose IcFuzz, the first fuzzing approach for Isaac Sim. IcFuzz first performs an LLM-based semantic stage segmentation, decomposing simulation programs into structured stages that capture context-aware object semantics. Guided by this information, IcFuzz designs multi-level mutation operators to systematically exercise the simulator across hierarchical granularities. To efficiently navigate the vast simulation state space, IcFuzz employs a multi-armed bandit algorithm to adaptively schedule mutation operators. Experimental results show that IcFuzz outperforms the baselines in terms of both code coverage and bug detection. Specifically, IcFuzz achieves approximately 190\%--205\% of the code coverage of the baselines and detects an average of 3.7 unique crashes over three rounds of 12-hour tests, while no crashes are detected by the baselines. Moreover, IcFuzz has uncovered 11 bugs over approximately four months, 9 of which have been confirmed or fixed by the developers.
Chinese Translation
机器人模拟器作为具身人工智能的基础设施，促进了安全和可扩展的机器人系统开发。NVIDIA Isaac Sim已成为最受欢迎的模拟器之一，其以GPU加速的物理引擎和逼真的渲染技术为特点，使得复杂环境的高保真建模成为可能。然而，其固有的复杂性不可避免地引入了可能影响仿真可靠性的软件缺陷。现有的模糊测试方法由于上下文感知对象语义、层次化仿真控制和庞大的仿真状态空间等挑战，难以有效测试Isaac Sim。本文提出了IcFuzz，这是针对Isaac Sim的首个模糊测试方法。IcFuzz首先执行基于大语言模型（LLM）的语义阶段分段，将仿真程序分解为结构化阶段，以捕捉上下文感知的对象语义。在此信息的指导下，IcFuzz设计了多级变异操作符，以系统性地在层次化粒度上对模拟器进行测试。为了高效地导航庞大的仿真状态空间，IcFuzz采用多臂赌博机算法自适应调度变异操作符。实验结果表明，IcFuzz在代码覆盖率和缺陷检测方面均优于基线方法。具体而言，IcFuzz的代码覆盖率约为基线的190\%至205\%，并在三轮12小时的测试中平均检测到3.7个独特崩溃，而基线方法未检测到任何崩溃。此外，IcFuzz在约四个月内发现了11个缺陷，其中9个已被开发者确认或修复。
cs.RO / 26 / 2608.06154
Visual Grounding in Zero-Shot Vision-Language Control
零样本视觉-语言控制中的视觉定位
Abstract
Vision-language models (VLMs) are increasingly used as zero-shot controllers, but successful trajectories do not necessarily show that decisions are grounded in visual input: simulator dynamics and conservative action priors can produce favourable scores without meaningful perception. We investigate this with an input-ablation battery: blind-image controls, repeated identical inputs, lane-axis reflection, non-visual baselines, and pipeline-integrity checks. Across nine direct-action models, six structured local VLMs, and an exploratory VLM-MPC hierarchy, we analyse 32,874 scored calls over two embodiments and three simulators. The direct-control results are largely negative: a constant-SLOW policy outperforms a scripted geometric controller, several models are image-invariant or nearly constant, and models that recognize longitudinal hazards still fail to transform LEFT and RIGHT under reflection. No local VLM meets the joint longitudinal and lateral grounding criteria. However, an image-only deterministic positive control estimates the lead gap with 0.090 m MAE and exact mirror equivariance, confirming the stimuli carry sufficient visual information; the failures are modular, not universal. A post-hoc, leakage-controlled symmetry-consensus guardian selects two models from 16 calibration frames and freezes a 2-of-4 hazard vote across original and reflected views. On 272 held-out frames it reaches 0.954 balanced accuracy (episode-cluster bootstrap 95% CI [0.895,0.990]); nested leave-one-episode-out recovers the same pair and threshold in all 12 folds. Abstaining on ties raises committed balanced accuracy to 0.973 at 0.824 coverage. With deterministic perception retaining lateral authority, offline modular replay achieves 0.934 action agreement and exact mirror equivariance. These results support current VLMs as bounded, selective hazard assistants, not monolithic zero-shot controllers.
Chinese Translation
视觉-语言模型（VLMs）越来越多地被用作零样本控制器，但成功的轨迹并不一定表明决策是基于视觉输入的：模拟器动态和保守的动作先验可以在没有有意义感知的情况下产生良好的评分。我们通过输入消融实验进行调查：盲图像控制、重复相同输入、车道轴反射、非视觉基线和管道完整性检查。在九个直接动作模型、六个结构化局部 VLMs 和一个探索性 VLM-MPC 层次结构中，我们分析了在两个具体实现和三个模拟器上进行的 32,874 次评分调用。直接控制的结果大多是负面的：一个恒定的SLOW策略优于一个脚本几何控制器，几个模型是图像不变的或几乎恒定的，而能够识别纵向危险的模型在反射下仍然无法转换左（LEFT）和右（RIGHT）。没有任何局部 VLM 满足联合纵向和横向定位标准。然而，一个仅基于图像的确定性正控制以 0.090 米的平均绝对误差（MAE）和精确的镜像等变性估计了前导间隙，确认了刺激携带了足够的视觉信息；这些失败是模块化的，而非普遍性的。一个事后控制泄漏的对称共识守护者从 16 个校准帧中选择了两个模型，并在原始和反射视图之间冻结了 2/4 的危险投票。在 272 个保留帧上，它达到了 0.954 的平衡准确率（集群引导自助法 95% 置信区间 [0.895,0.990]）；嵌套的逐个排除实验在所有 12 个折叠中恢复了相同的模型对和阈值。在平局时弃权将承诺的平衡准确率提高到 0.973，覆盖率为 0.824。通过保持横向权威的确定性感知，离线模块重放达到了 0.934 的动作一致性和精确的镜像等变性。这些结果支持当前的 VLMs 作为有限的、选择性的危险助手，而非单一的零样本控制器。
cs.RO / 27 / 2608.06170
Prior-SG: Task and Prior Driven Region Segmentation for Scene Graphs in Arbitrarily-Structured Environments
Prior-SG：基于任务和先验驱动的场景图区域分割在任意结构环境中的应用
Abstract
Hierarchical 3D scene graphs are a promising representation for high-level spatial reasoning in autonomous mobile platforms. However, existing extraction frameworks typically rely on purely local visual clustering or strict geometric heuristics, such as wall-separated rooms, which fail in open-plan or arbitrarily-structured environments. We propose Prior-SG, a task- and prior-driven framework that casts scene graph generation fundamentally as a probabilistic alignment problem. As the robot explores, it continuously aggregates an incoming RGB-D sensor stream into a physically grounded Instance Graph utilizing a multi-scale, open-vocabulary feature fusion strategy. The system then infers the high-level functional semantics of this map through a Maximum A Posteriori (MAP) estimate, guided by a Prior Graph-a logical expectation of the environment's structure and task-relevant vocabulary synthesized dynamically by a Large Language Model. By optimizing a Markov Random Field that fuses heterogeneous experts (visual, geometric, and discrete objects) with these topological priors, the system resolves local perceptual ambiguities. We validate this approach across diverse simulated residential datasets and large, open-plan real-world environments. Prior-SG achieves state-of-the-art semantic region segmentation accuracy compared to recent baselines, robustly delineates distant functional boundaries in the absence of physical walls, and uniquely provides zero-shot ontological flexibility, enabling the robot to entirely restructure its spatial partitioning based on a given high-level task.
Chinese Translation
层次化的三维场景图是一种有前景的表示方式，用于自主移动平台中的高层空间推理。然而，现有的提取框架通常依赖于纯粹的局部视觉聚类或严格的几何启发式方法，例如墙壁隔开的房间，这在开放式平面或任意结构环境中无法奏效。我们提出了Prior-SG，这是一种任务和先验驱动的框架，根本上将场景图生成视为一个概率对齐问题。当机器人探索时，它持续将输入的RGB-D传感器流聚合成一个物理基础的实例图，利用多尺度、开放词汇的特征融合策略。然后，系统通过最大后验估计（Maximum A Posteriori, MAP）推断该地图的高层功能语义，受一个先验图的指导——这是对环境结构和任务相关词汇的逻辑期望，由大型语言模型动态合成。通过优化一个融合异构专家（视觉、几何和离散对象）与这些拓扑先验的马尔可夫随机场，系统解决了局部感知模糊性。我们在多样化的模拟住宅数据集和大型开放式真实环境中验证了这种方法。与最近的基线相比，Prior-SG在语义区域分割准确性上达到了最先进的水平，能够在没有物理墙壁的情况下稳健地划定远处的功能边界，并独特地提供零样本本体灵活性，使机器人能够根据给定的高层任务完全重构其空间划分。
cs.RO / 28 / 2608.06208
ErgoSurf: Ergodic Control for the Coverage of Unknown Surfaces
ErgoSurf：用于未知表面覆盖的遍历控制
Abstract
Contact-centric tasks on surfaces, ranging from inspection and cleaning to sanding and polishing, require robots to systematically cover the surface while maintaining stable contact. Ergodic control generates trajectories that spend time at a location proportional to a desired, task-specific spatial distribution, enabling efficient information gathering and coverage. However, traditional ergodic control methods rely on prior knowledge of surface geometry or require a vision sensory input to scan the geometry beforehand, limiting their applicability in real-world scenarios with unknown or dynamic environments. This paper introduces a novel online ergodic control framework that achieves systematic surface coverage while simultaneously reconstructing unknown surface geometry. We employ a Gaussian Process Implicit Surface (GPIS) model that learns global surface geometry from intrinsic tactile sensing during execution. For efficient online planning, we approximate the surface locally using point clouds sampled from tangent planes at observed contact points and iteratively fit them to the Gaussian Process. This approximation simultaneously serves as the sampling domain for both the target and the coverage distributions. We employ a heat-diffusion analogy to compute potential fields that guide ergodic exploration, translating spatial coverage objectives into smooth robot trajectories. We demonstrate our framework through simulation and real-robot experiments, validating simultaneous ergodic coverage and online surface geometry learning with reconstruction errors approaching the ground truth.
Chinese Translation
在表面上的接触中心任务，包括检查、清洁、打磨和抛光，要求机器人在保持稳定接触的同时系统地覆盖表面。遍历控制生成的轨迹在位置上花费的时间与所需的任务特定空间分布成正比，从而实现高效的信息收集和覆盖。然而，传统的遍历控制方法依赖于对表面几何的先验知识，或需要视觉传感输入提前扫描几何形状，这限制了它们在未知或动态环境中的实际应用。本文提出了一种新颖的在线遍历控制框架，能够在重建未知表面几何的同时实现系统的表面覆盖。我们采用高斯过程隐式表面（Gaussian Process Implicit Surface, GPIS）模型，从执行过程中的内在触觉传感中学习全局表面几何。为了实现高效的在线规划，我们使用从观察到的接触点的切平面采样的点云局部近似表面，并将其迭代拟合到高斯过程。这种近似同时作为目标和覆盖分布的采样域。我们采用热扩散类比计算引导遍历探索的势场，将空间覆盖目标转化为平滑的机器人轨迹。我们通过仿真和真实机器人实验展示了我们的框架，验证了同时进行的遍历覆盖和在线表面几何学习，重建误差接近真实值。
cs.RO / 29 / 2608.06210
VIDP: Variable Impedance Diffusion Policy for Compliant Robot Manipulation from Diverse Demonstrations
VIDP：基于多样化示范的可变阻抗扩散策略用于柔性机器人操作
Abstract
Contact-rich manipulation requires precise tracking and mechanical compliance, where variable impedance control can improve robustness in task success, whereas static compliance cannot adapt to varying contact constraints. Variable impedance skills can be learned from demonstrations, avoiding complex modeling, but compliance is a hidden variable in force-agnostic kinematic data. While existing methods infer compliance from trajectory variations, these variations may reflect geometric adaptation and not intentional compliance when subject to changing spatial layouts. Therefore, this letter introduces Variable Impedance Diffusion Policy (VIDP), an imitation learning-based variable impedance control framework leveraging a Task-Parameterized Directionality-Aware Mixture Model (TP-DAMM) to extract physically consistent trajectory distributions from diverse demonstrations. By mapping distributions to stiffness profiles, VIDP jointly predicts pose actions and task compliance without force sensors. Real-world experiments show that VIDP significantly outperforms fixed-impedance baselines in task success rate while reducing interaction forces with respect to high stiffness controllers and tracking errors with respect to low stiffness baselines.
Chinese Translation
接触丰富的操作需要精确的跟踪和机械柔顺性，而可变阻抗控制可以提高任务成功的鲁棒性，而静态柔顺性无法适应变化的接触约束。可变阻抗技能可以通过示范学习，避免复杂的建模，但柔顺性在无力感知的运动学数据中是一个隐含变量。虽然现有方法通过轨迹变化推断柔顺性，但这些变化可能反映几何适应性，而非在变化的空间布局下的有意柔顺性。因此，本文提出了可变阻抗扩散策略（VIDP），这是一种基于模仿学习的可变阻抗控制框架，利用任务参数化方向感知混合模型（TP-DAMM）从多样化示范中提取物理一致的轨迹分布。通过将分布映射到刚度特征，VIDP 在不使用力传感器的情况下共同预测姿态动作和任务柔顺性。实际实验表明，VIDP 在任务成功率上显著优于固定阻抗基线，同时在高刚度控制器下减少了交互力，并在低刚度基线下降低了跟踪误差。
cs.RO / 30 / 2608.06219
Design and Evaluation of a Touchscreen-Based Teleoperation Interface for Robotic Manipulators
基于触摸屏的机器人操控远程操作界面的设计与评估
Abstract
Intuitive teleoperation interfaces are crucial for the safe and effective operation of robotic manipulators in challenging environments. In the nuclear industry, surface contact tasks such as swab sampling require precise path and force tracking, obstacle avoidance, and sustained operator attention, which conventional joystick interfaces struggle to support effectively. This study designs and evaluates a novel touchscreen teleoperation interface that maps continuous finger movements directly to robotic manipulator motions, provides finer velocity control, and integrates control with visualization, enabling more natural, precise, and intuitive surface interaction than conventional controllers. A comparative user study with 20 participants evaluated task performance and workload using the proposed touchscreen, a conventional joystick, and a single-click autonomous mode. Tasks simulated realistic surface manipulation using a Franka Emika Panda arm, remotely controlled from another country. Kinematic, physiological, and behavioral data were recorded to comprehensively assess task performance, cognitive load, and operator trust across each control condition. Participants completed teleoperation tasks more efficiently and accurately with the touchscreen interface, achieving a 53.5% reduction in completion time (median: 2.50 vs. 5.38 min), higher in-area coverage on the sinusoidal path (90.7% vs. 84.1%), and lower overshoot on both path geometries compared with the joystick. Cognitive load, quantified via NASA-TLX (0-100), decreased from joystick to touchscreen (mean TLX 52 to 43; -9 points, -17.3%) and was lowest under the autonomous one-click mode (31; -21 points vs. joystick, -40.4%; -12 vs. touchscreen, -27.9%). This research presents an easy-to-implement touchscreen interface that improves performance in teleoperated surface tasks while reducing cognitive load.
Chinese Translation
直观的远程操作界面对于在复杂环境中安全有效地操作机器人操纵臂至关重要。在核工业中，表面接触任务如擦拭取样需要精确的路径和力跟踪、障碍物规避以及持续的操作员注意力，而传统的操纵杆界面难以有效支持这些需求。本研究设计并评估了一种新颖的触摸屏远程操作界面，该界面将连续的手指运动直接映射到机器人操纵臂的运动，提供更精细的速度控制，并将控制与可视化相结合，使得表面交互比传统控制器更自然、精确和直观。通过与20名参与者进行的比较用户研究，评估了使用所提触摸屏、传统操纵杆和单击自主模式的任务表现和工作负荷。任务模拟了使用Franka Emika Panda臂的真实表面操作，远程控制自其他国家。记录了运动学、生理和行为数据，以全面评估每种控制条件下的任务表现、认知负荷和操作员信任。参与者在触摸屏界面下更高效、准确地完成了远程操作任务，完成时间减少了53.5%（中位数：2.50分钟对比5.38分钟），在正弦路径上的区域覆盖率更高（90.7%对比84.1%），并且在两种路径几何形状上的超调量均低于操纵杆。通过NASA-TLX量化的认知负荷（0-100）从操纵杆到触摸屏减少（平均TLX从52降至43；减少9分，减少17.3%），在自主单击模式下最低（31；比操纵杆减少21分，减少40.4%；比触摸屏减少12分，减少27.9%）。本研究提出了一种易于实施的触摸屏界面，能够提高远程操作表面任务的表现，同时降低认知负荷。
cs.RO / 31 / 2608.06221
Robot Learning from Human Demonstrations: Handwritten Alphabet Trajectories and Human-Likeness Evaluation
机器人从人类示范中学习：手写字母轨迹与人类相似性评估
Abstract
Learning from demonstration (LfD) provides a developmental framework through which robots can develop motor skills by observing and imitating human dynamics, reducing reliance on explicit programming to teach a skill to a robot. The resulting human-like robot motion is recognised as a key factor in building trust and enabling natural collaboration in human-robot interaction. This paper presents a framework for learning human-like robot motion from demonstration, including data collection, probabilistic trajectory learning, and perceptual user evaluation. A dataset of 3,142 handwriting demonstrations was collected from 22 participants across all 52 Latin alphabet character-case combinations via a touchscreen teleoperation interface, capturing planar position, contact force, and timing. Building on the widely used Gaussian Mixture Model and Gaussian Mixture Regression approach for learning from demonstration, the framework is extended in this work by incorporating force and normalised time dimensions to enable richer representation of human dynamics, and adapting it to handle non-continuous, multi-segment trajectories, enabling generalisation across demonstrations. A user study with 21 participants evaluated the perceived human-likeness of the generated trajectories using a continuous scale anchored between robotic and human-like motion, normalised to 0-100 where 50 represents the neutral midpoint. The generated trajectories achieved an overall human-likeness score of 71.50 (SD=22.56), indicating that the majority of trajectories were perceived as more human-like. Participants identified geometric positioning and trajectory sequence as the most influential perceptual factors, and reported positive attitudes toward human-like robot behaviour. The datasets are released as open-source, providing a reproducible benchmark for developing and evaluating human-like robot motion methods.
Chinese Translation
从示范学习（LfD）提供了一个发展框架，使机器人能够通过观察和模仿人类动态来发展运动技能，从而减少对显式编程的依赖，以教会机器人一项技能。由此产生的人类相似的机器人运动被认为是建立信任和促进人机交互中自然协作的关键因素。本文提出了一个从示范中学习人类相似机器人运动的框架，包括数据收集、概率轨迹学习和感知用户评估。通过触摸屏远程操作界面，从22名参与者那里收集了3142个手写示范数据，涵盖了52种拉丁字母字符大小写组合，捕捉了平面位置、接触力和时间。基于广泛使用的高斯混合模型（Gaussian Mixture Model）和高斯混合回归（Gaussian Mixture Regression）方法进行示范学习，本文的框架通过引入力和归一化时间维度进行了扩展，以实现对人类动态的更丰富表示，并调整其以处理非连续的多段轨迹，从而实现跨示范的泛化。通过一项包含21名参与者的用户研究，评估了生成轨迹的感知人类相似性，使用一个在机器人运动和人类运动之间锚定的连续尺度，归一化为0-100，其中50代表中立的中点。生成的轨迹整体人类相似性得分为71.50（标准差=22.56），表明大多数轨迹被认为更具人类相似性。参与者认为几何定位和轨迹顺序是最具影响力的感知因素，并对人类相似的机器人行为表示积极态度。数据集以开源形式发布，为开发和评估人类相似机器人运动方法提供了可重复的基准。
cs.RO / 32 / 2608.06332
GeniWorld: A Generalizable Interactive World Model for Robotic Manipulation via Visual Actions
GeniWorld：一种可泛化的通过视觉动作进行机器人操作的交互式世界模型
Abstract
Generalist robot policies exhibit strong capabilities, but their robustness in complex and unseen environments remains limited. Scaling robot learning and evaluation in diverse real-world environments remains costly and challenging. Action-conditioned world models offer a promising alternative, but they often suffer from limited action controllability and poor generalization to out-of-distribution (OOD) scenarios. To this end, we present GeniWorld, an interactive world model for robots that generalizes robustly across unseen scenarios. Building on pretrained video generative models, we use URDF-based rendering to transform numerical actions into visual action representations, enabling spatially grounded action control. By explicitly decoupling embodiment kinematics from environmental dynamics, our model mitigates scene overfitting and facilitates modeling of robot-environment interactions. To achieve closed-loop control, we construct an autoregressive video prediction model integrated with high-frequency robot kinematic control, enabling interaction with both robot policies and human teleoperators. In our experiments, even when trained solely on limited fixed-scene data, our model achieves superior in-domain performance and robust zero-shot generalization to highly randomized, unseen environments. For downstream applications, GeniWorld serves as a scalable policy evaluator that remains reliable under environmental perturbations. Furthermore, even with limited real-world demonstrations, GeniWorld generates diverse manipulation trajectories within the world model, improving downstream policy performance and robustness in complex environments.
Chinese Translation
通用型机器人策略展现出强大的能力，但在复杂和未见环境中的鲁棒性仍然有限。在多样化的真实世界环境中扩展机器人学习和评估仍然成本高昂且具有挑战性。基于动作条件的世界模型提供了一种有前景的替代方案，但它们通常面临有限的动作可控性和对分布外（OOD）场景的较差泛化能力。为此，我们提出了GeniWorld，这是一种交互式世界模型，能够在未见场景中稳健地泛化。基于预训练的视频生成模型，我们使用基于URDF的渲染将数值动作转化为视觉动作表示，从而实现空间基础的动作控制。通过明确将体现运动学与环境动态解耦，我们的模型减轻了场景过拟合，并促进了机器人与环境交互的建模。为了实现闭环控制，我们构建了一个自回归视频预测模型，并与高频率的机器人运动学控制相结合，使得与机器人策略和人类遥控操作员的交互成为可能。在我们的实验中，即使仅在有限的固定场景数据上进行训练，我们的模型也实现了优越的领域内性能，并在高度随机化的未见环境中表现出鲁棒的零-shot 泛化能力。对于下游应用，GeniWorld作为一个可扩展的策略评估器，在环境扰动下仍然保持可靠。此外，即使在有限的真实世界演示下，GeniWorld也能在世界模型中生成多样的操作轨迹，从而提高下游策略在复杂环境中的性能和鲁棒性。
cs.RO / 33 / 2608.06354
A Master-Salve Robot Manipulator for Needle-Based Teleoperation in MRI Chamber
用于MRI室内针基遥操作的主从机器人操纵器
Abstract
We present a MR safe, master-slave robot manipulator for abdominal interventions in the MRI chamber. A human operated 2+1-DoF master controller manipulator transmits motion and force to a 2+1-DoF slave manipulator via fluid transmission. Jointly, a digital master controller provides multimodal control capability beyond common split axis or mode switchable hybrid human-digital controller configurations found in previous studies. High input impedance, low-leakage, elastomeric fluid actuators are delegated to remote angulation control. Low-friction graphite piston cylinders are delegated to needle insertion axis remote actuation given the sub-newton force transparency and sub-millimeter motion transmission over bedside fluid piping lengths. The device enables real-time MRI guided interventions allowing manual, digital, hybrid, and collaborative control modes. Collaborative tasks such as assisted tissue penetration, fault-driven virtual fixture, and motion compensation through feedback control are presented in this paper. Preliminary MR scanner results demonstrate manipulator functional viability for an in-vivo pig experiment in bedside, manual control mode configuration.
Chinese Translation
我们提出了一种适用于MRI室内腹部干预的MR安全主从机器人操纵器。一个由人操作的2+1自由度主控制器操纵器通过流体传输将运动和力传递给一个2+1自由度的从操纵器。数字主控制器共同提供了超越以往研究中常见的分轴或模式可切换混合人机控制器配置的多模态控制能力。高输入阻抗、低泄漏的弹性流体执行器被委派用于远程角度控制。低摩擦石墨活塞气缸被委派用于针插入轴的远程驱动，考虑到其亚牛顿力透明度和亚毫米运动传输在床边流体管道长度上的表现。该设备实现了实时MRI引导干预，允许手动、数字、混合和协作控制模式。本文展示了协作任务，如辅助组织穿透、故障驱动的虚拟夹具和通过反馈控制进行的运动补偿。初步的MR扫描结果表明，该操纵器在床边手动控制模式配置下的功能可行性，适用于在体猪实验。
cs.RO / 34 / 2608.06374
DyPES-VLA: Learning Shared Dynamics Priors and Embodiment-Specific Control for Cross-Embodiment Manipulation
DyPES-VLA：学习共享动态先验和特定于体现的控制以实现跨体现操作
Abstract
Vision-Language-Action (VLA) models have become a powerful paradigm for robot manipulation, but training a single generalist policy for heterogeneous robot embodiments remains an open problem. Existing methods have two main limitations. First, they underuse dynamics priors shared across diverse visual and interaction data, limiting cross-embodiment transfer. Second, they require extensive manual preprocessing to convert embodiment-specific actions into a common format. To overcome these limitations, we propose DyPES-VLA, a cross-embodiment VLA that learns shared Dynamics Priors and Embodiment-Specific control. First, we learn shared dynamics priors by training the vision-language model (VLM) with a future-prediction objective on cross-embodiment data, driving the shared query representation to capture object motion, contact, and interaction-induced scene changes. Second, an embodiment-specific Mixture-of-Experts (MoE) action head translates these shared dynamics priors into executable controls directly in each embodiment's native action space, without manually pre-aligning heterogeneous actions into a common format. This head shares attention layers to capture common temporal action structures, while its embodiment-specific feed-forward experts resolve the unique kinematic constraints and control semantics of distinct embodiments. As a generalist policy, our \ourmethod achieves state-of-the-art performance across simulation and real-world evaluations, reaching 98.0% success on LIBERO, 59.25% on RoboCasa-GR1, and 89.02% on RoboTwin~2.0.
Chinese Translation
视觉-语言-行动（VLA）模型已成为机器人操作的强大范式，但为异构机器人体现训练单一通用策略仍然是一个未解决的问题。现有方法存在两个主要局限性。首先，它们未充分利用跨多样视觉和交互数据共享的动态先验，限制了跨体现迁移。其次，它们需要大量手动预处理，将特定于体现的动作转换为通用格式。为克服这些局限性，我们提出了DyPES-VLA，一种跨体现的VLA，学习共享动态先验和特定于体现的控制。首先，我们通过在跨体现数据上以未来预测为目标训练视觉-语言模型（VLM）来学习共享动态先验，推动共享查询表示捕捉物体运动、接触和交互引起的场景变化。其次，特定于体现的专家混合（Mixture-of-Experts, MoE）动作头将这些共享动态先验直接转换为每个体现的本地动作空间中的可执行控制，而无需手动将异构动作预对齐为通用格式。该动作头共享注意力层以捕捉共同的时间动作结构，同时其特定于体现的前馈专家解决不同体现的独特运动约束和控制语义。作为一种通用策略，我们的 extit{ourmethod}在仿真和现实世界评估中实现了最先进的性能，在LIBERO上达到98.0%的成功率，在RoboCasa-GR1上达到59.25%，在RoboTwin~2.0上达到89.02%。
cs.RO / 35 / 2608.06375
$\omega$-0: A Latent Predictive World Action Model for Concurrent Humanoid Loco-Manipulation
$ heta$-0：一种用于同时人形机器人运动与操作的潜在预测全身动作模型
Abstract
Humanoid household tasks often require concurrent loco-manipulation, where the robot must move, adjust posture, maintain balance, and manipulate objects as a single coordinated behavior. Yet existing humanoid policies typically decompose locomotion and manipulation, while recent world-action models remain either arm-centric or video-centered. We present $\omega$-0, a latent predictive whole-body world-action model for real-world humanoid concurrent loco-manipulation. Given a language instruction, current visual observation, and robot proprioceptive state, $\omega$-0 directly predicts controller-compatible whole-body action latents for real-robot execution. Rather than reconstructing future videos, $\omega$-0 learns compact future observation embeddings as a lightweight predictive objective, coupling latent visual foresight with diffusion-based whole-body action generation. The model supports egocentric RGB, exocentric RGB, and exocentric depth inputs, and leverages controller-based simulation replay to ground human/public visual-motion priors into robot-executable action latents. We further collect $\omega$-HOME, a 40+ hour real-world household humanoid dataset with synchronized multi-view observations, whole-body SMPL motions, robot states, and action latents. Real-world experiments on 11 household tasks demonstrate that a single $\omega$-0 model can produce smooth manipulate-while-moving behaviors and consistently outperform representative imitation learning, VLA, humanoid, and WAM baselines.
Chinese Translation
人形机器人在家庭任务中常常需要进行同时的运动与操作，即机器人必须在一个协调的行为中移动、调整姿势、保持平衡并操作物体。然而，现有的人形机器人策略通常将运动与操作分开，而最近的世界动作模型则要么以手臂为中心，要么以视频为中心。我们提出了$ heta$-0，这是一种用于真实世界人形机器人同时运动与操作的潜在预测全身世界动作模型。在给定语言指令、当前视觉观察和机器人本体状态的情况下，$ heta$-0直接预测与控制器兼容的全身动作潜变量以供真实机器人执行。与其重建未来视频，$ heta$-0学习紧凑的未来观察嵌入，作为一种轻量级的预测目标，将潜在的视觉前瞻性与基于扩散的全身动作生成相结合。该模型支持自我中心的RGB、外部中心的RGB和外部中心的深度输入，并利用基于控制器的仿真重放将人类/公共视觉运动先验与机器人可执行的动作潜变量结合起来。我们进一步收集了$ heta$-HOME，这是一个超过40小时的真实世界家庭人形机器人数据集，包含同步的多视角观察、全身SMPL动作、机器人状态和动作潜变量。在11个家庭任务上的真实世界实验表明，单一的$ heta$-0模型能够产生平滑的边走边操作行为，并且在表现上始终优于代表性的模仿学习、VLA、人形机器人和WAM基准。
cs.CV / 1 / 2608.05209
MapTCL: Temporal Consistency Learning via Bidirectional Alignment for Vectorized HD Map Construction
MapTCL：通过双向对齐进行时间一致性学习以构建矢量化高清地图
Abstract
Constructing reliable online HD maps remains challenging in dynamic urban environments due to moving objects and occlusions. While recent works employ feature-level temporal fusion to address this, they rely solely on per-frame ground truth supervision. Consequently, they lack an explicit objective to directly penalize the geometric noise and temporal jitter between consecutive online HD maps. To address this, we propose MapTCL, an auxiliary training strategy that formulates temporal consistency loss between current and past frames via bidirectional alignment. Specifically, Bidirectional Vector Consistency Learning (BVCL) models the geometric and semantic discrepancies between associated past and current vector instances as an auxiliary loss. We also employ Raster map Consistency Learning (RCL) as an additional loss to stabilize dense BEV features. By jointly training with these dual losses, MapTCL improves the temporal stability of generated HD maps. Extensive experiments on two standard benchmarks demonstrate the effectiveness of our approach. As a versatile plug-and-play module, MapTCL consistently enhances existing baseline models, achieving gains of +3.7 mAP & +2.8 C-mAP on nuScenes and +3.1 mAP & +2.5 C-mAP on Argoverse 2 without additional inference overhead.
Chinese Translation
在动态城市环境中，由于移动物体和遮挡，构建可靠的在线高清地图仍然面临挑战。尽管最近的研究采用特征级时间融合来解决这一问题，但它们仅依赖于每帧的真实标签监督。因此，它们缺乏明确的目标来直接惩罚连续在线高清地图之间的几何噪声和时间抖动。为了解决这个问题，我们提出了MapTCL，一种辅助训练策略，通过双向对齐公式化当前帧与过去帧之间的时间一致性损失。具体而言，双向矢量一致性学习（Bidirectional Vector Consistency Learning, BVCL）将关联的过去和当前矢量实例之间的几何和语义差异建模为辅助损失。我们还采用光栅地图一致性学习（Raster map Consistency Learning, RCL）作为额外损失，以稳定密集的鸟瞰视图（BEV）特征。通过联合训练这两种损失，MapTCL提高了生成高清地图的时间稳定性。在两个标准基准上的大量实验表明了我们方法的有效性。作为一个通用的即插即用模块，MapTCL持续增强现有的基线模型，在nuScenes上获得了+3.7 mAP和+2.8 C-mAP，在Argoverse 2上获得了+3.1 mAP和+2.5 C-mAP，且没有额外的推理开销。
cs.CV / 2 / 2608.05210
Innocent Panels, Hateful Stories: Evaluating and Detecting Hateful Intent in Multi-Turn Visual Story Generation
无辜的面板，仇恨的故事：评估和检测多轮视觉故事生成中的仇恨意图
Abstract
Picture books and comics have long been used to disseminate hateful narratives because they are easily understood even by children, as exemplified by the notorious Nazi propaganda picture book \emph{Der Giftpilz}. Recently, frontier text-to-image (T2I) systems such as Gemini and GPT-Image have enabled conversational generation with consistent characters and scenes across turns, making hateful visual stories, namely ordered image groups that collectively convey hateful narratives, cheap and scalable to produce. Although prior work has studied hateful content generation by T2I systems, it focuses on individual images, leaving group-level hateful meaning largely unexplored. We aim to address the gap. Concretely, we introduce \texttt{HatefulStoryPrompts}, comprising 330 multi-turn configurations from 55 hateful stories across two languages and three visual styles, and evaluate five frontier models over 4,950 attempts. Every model completes over 80\% of the stories, with the strongest reaching 99.0\%. We further evaluate existing moderation systems on \texttt{HatefulVisualStory}, a human-labeled dataset of 969 hateful image sets and 990 benign controls, and find that they frequently miss group-level hateful meaning: dedicated safety models achieve at most 34.9\% recall, while a strong vision-language model reaches 67.5\%. Finally, we propose complementary proactive and post-generation defenses. An interaction-aware monitor achieves 97.3\% recall for prompt-only sessions and 92.6\% when the user supplies the first image, while post-generation methods jointly analyzing completed image groups reach 80.2\%. Our work shows that, as image generation evolves from isolated outputs to coherent visual narratives, safety must evolve accordingly, from per-image moderation to stateful reasoning over interactions and image relationships.
Chinese Translation
图画书和漫画长期以来一直被用来传播仇恨叙事，因为它们即使对儿童来说也容易理解，著名的纳粹宣传图画书《毒蘑菇》（ extit{Der Giftpilz}）就是一个例证。最近，前沿的文本到图像（T2I）系统如Gemini和GPT-Image使得在多轮对话中生成具有一致角色和场景的内容成为可能，从而使得仇恨视觉故事，即一组有序的图像共同传达仇恨叙事，变得廉价且易于扩展。尽管之前的研究已经探讨了T2I系统生成仇恨内容的问题，但主要集中在单幅图像上，群体层面的仇恨意义尚未得到充分探索。我们旨在填补这一空白。具体而言，我们引入了 exttt{HatefulStoryPrompts}，包含来自55个仇恨故事的330个多轮配置，涵盖两种语言和三种视觉风格，并对5个前沿模型进行了4,950次尝试的评估。每个模型完成超过80 ext{%}的故事，最强的模型达到99.0 ext{%}。我们进一步评估了现有的内容审核系统在 exttt{HatefulVisualStory}上的表现，该数据集由969个仇恨图像集和990个良性对照组成，发现它们常常错过群体层面的仇恨意义：专门的安全模型最多实现34.9 ext{%}的召回率，而一个强大的视觉-语言模型则达到了67.5 ext{%}。最后，我们提出了互补的主动和后生成防御措施。一个关注交互的监控器在仅提示的会话中实现了97.3 ext{%}的召回率，而在用户提供第一幅图像时则为92.6 ext{%}，而后生成方法通过联合分析完成的图像组达到了80.2 ext{%}。我们的研究表明，随着图像生成从孤立输出演变为连贯的视觉叙事，安全性也必须相应演变，从单幅图像的审核转向对交互和图像关系的状态化推理。
cs.CV / 3 / 2608.05213
StyleComposer: Training-Free Multi-Reference Style Composition
风格合成器：无训练的多参考风格组合
Abstract
The style of a painting is not monolithic: color, texture, and structure may come from different sources. Existing reference-guided methods transfer them as one style signal, leaving each attribute's source and strength outside the user's control. We ask where in a diffusion model one attribute can change while the others hold, and find that no single representation isolates all three. The proposed StyleComposer therefore routes each style attribute through the representation where it separates best and coordinates the routes over denoising time. Without training or inversion, it satisfies three references and the prompt jointly more closely than prior methods, and exposes one strength slider per attribute. Project page: https://lexxsh.github.io/StyleComposer
Chinese Translation
绘画的风格并非单一：颜色、纹理和结构可能来自不同的来源。现有的参考引导方法将它们作为一个风格信号进行转移，使得每个属性的来源和强度超出了用户的控制范围。我们探讨在扩散模型中，如何使一个属性变化而其他属性保持不变，并发现没有单一的表示能够将三者完全隔离。因此，所提出的风格合成器（StyleComposer）将每个风格属性引导至最佳分离的表示，并在去噪时间内协调这些路径。无需训练或反演，它比以往的方法更紧密地满足三个参考和提示，并为每个属性提供一个强度滑块。项目页面：https://lexxsh.github.io/StyleComposer
cs.CV / 4 / 2608.05226
NeuroAdaptTrainer: A Fiji/ImageJ Plugin for YOLO-Based Neuron Segmentation, InteractiveCorrection and Transfer Learning
NeuroAdaptTrainer：一个基于YOLO的神经元分割、交互修正和迁移学习的Fiji/ImageJ插件
Abstract
Neuron counting and segmentation in microscopy images of neuronal cultures is a routine and time-consuming task in neuroscience research, traditionally performed through manual inspection or semi-automatic tools. We present NeuroAdaptTrainer, an open-source Fiji/ImageJ plugin that integrates a YOLO instance-segmentation model directly into the microscopist's workflow. The plugin allows a user to run automatic neuron detection on a single image or a batch of images, manually correct the resulting detections from within Fiji, and use those corrections to adapt the model to new imaging conditions via transfer learning. A built-in external validation module allows the base and adapted models to be compared quantitatively on a held-out annotated set. NeuroAdaptTrainer lowers the barrier for non-specialist users to benefit from deep-learning-based segmentation while keeping expert supervision at the center of the workflow.
Chinese Translation
在神经科学研究中，显微镜下神经元培养图像的计数和分割是一项常规且耗时的任务，传统上通过人工检查或半自动工具进行。我们提出了NeuroAdaptTrainer，一个开源的Fiji/ImageJ插件，它将YOLO实例分割模型直接集成到显微镜工作流程中。该插件允许用户对单幅图像或一批图像进行自动神经元检测，并在Fiji中手动修正结果检测，同时利用这些修正通过迁移学习将模型适应新的成像条件。内置的外部验证模块允许对基模型和适应模型在保留的标注集上进行定量比较。NeuroAdaptTrainer降低了非专业用户受益于基于深度学习的分割的门槛，同时将专家监督置于工作流程的核心。
cs.CV / 5 / 2608.05237
In-Context Forcing: Uncovering Context Effects in Autoregressive Video Diffusion
上下文强制：揭示自回归视频扩散中的上下文效应
Abstract
Current few-step autoregressive video diffusion models depend on previous fully denoised clean frames as context for all denoising steps of the current frame. However, these clean frames leak excessive local details, which causes the model to take shortcuts, resulting in compromised temporal semantics and dynamics. Inspired by the perspective of diffusion as masking, we explore the impact of noisy contexts on few-step autoregressive generation. Yet, simply applying contexts with the same noise levels provides insufficient guidance, leading to poor temporal consistency. To resolve this dilemma, we introduce In-Context Forcing, a progressive autoregressive paradigm that utilizes contexts with decreasing noise levels. By applying less masking to distant frames and more masking to adjacent ones, this approach provides adaptive guidance, effectively ensuring both robust temporal consistency and high inter-frame dynamics. Furthermore, by decoupling the strict dependence on previous clean frames, our paradigm enables cross-frame parallel denoising, achieving substantial inference acceleration without sacrificing performance. Extensive experiments on VBench demonstrate that our method significantly outperforms state-of-the-art approaches in both visual fidelity and inference speed.
Chinese Translation
当前的少步自回归视频扩散模型依赖于先前完全去噪的干净帧作为当前帧所有去噪步骤的上下文。然而，这些干净帧泄露了过多的局部细节，导致模型采取捷径，从而损害了时间语义和动态性。受到扩散视为掩蔽的启发，我们探讨了噪声上下文对少步自回归生成的影响。然而，简单地应用相同噪声水平的上下文提供的指导不足，导致时间一致性差。为了解决这一困境，我们引入了上下文强制（In-Context Forcing），这是一种渐进式自回归范式，利用逐渐降低噪声水平的上下文。通过对远处帧施加较少的掩蔽，而对相邻帧施加更多的掩蔽，这种方法提供了自适应指导，有效确保了强大的时间一致性和高帧间动态性。此外，通过解耦对先前干净帧的严格依赖，我们的范式实现了跨帧并行去噪，显著加速推理而不牺牲性能。在VBench上的大量实验表明，我们的方法在视觉保真度和推理速度上显著优于最先进的方法。
cs.CV / 6 / 2608.05258
Grad-CAM for Vision Transformers: A Systematic Taxonomy and Audit of Methodological Ambiguity in Explainable AI
针对视觉变换器的Grad-CAM：可解释人工智能中方法论模糊性的系统分类与审计
Abstract
Gradient-weighted Class Activation Mapping (Grad-CAM) is widely used to visualize model decisions, but it was originally formulated for convolutional neural networks, where spatial feature maps and channel dimensions have clear architectural meanings. Vision Transformers (ViTs) do not provide the same structure, instead representing images through tokens, attention, residual streams, and multimodal interactions. This paper presents a systematic taxonomy and literature audit of how Grad-CAM and related methods are adapted, justified, and reported for ViT-based architectures. From an initial search of more than 550 papers, we identify 175 papers that apply Grad-CAM or Grad-CAM-adjacent methods to ViTs. We find that most papers do not provide a full mathematical or implementation-level account of how Grad-CAM is adapted to transformer representations. To characterize this gap, we introduce a descriptive taxonomy of ViT Grad-CAM adaptations that makes explicit the feature locations, gradient targets, spatial reconstruction steps, and aggregation choices that are often left implicit. This taxonomy is not intended to prescribe a single correct adaptation, but to clarify the range of methodological choices being made. The study shows that Grad-CAM on ViTs is often treated as a trivial extension of CNN-based Grad-CAM, despite requiring nontrivial choices that affect rigor, reproducibility, and interpretation.
Chinese Translation
梯度加权类激活映射（Grad-CAM）广泛用于可视化模型决策，但最初是为卷积神经网络（CNN）设计的，在这些网络中，空间特征图和通道维度具有明确的架构含义。视觉变换器（ViTs）并未提供相同的结构，而是通过标记、注意力、残差流和多模态交互来表示图像。本文系统地分类和审计了Grad-CAM及相关方法如何适应、论证和报告于基于ViT的架构。从对550多篇论文的初步检索中，我们识别出175篇将Grad-CAM或与Grad-CAM相关的方法应用于ViTs的论文。我们发现，大多数论文未能提供Grad-CAM如何适应变换器表示的完整数学或实现层面的说明。为了表征这一差距，我们引入了一种描述性分类法，明确了ViT Grad-CAM适应中的特征位置、梯度目标、空间重建步骤和聚合选择，这些内容通常未被明确说明。该分类法并不旨在规定单一的正确适应，而是澄清所做方法选择的范围。研究表明，尽管Grad-CAM在ViTs上通常被视为对基于CNN的Grad-CAM的简单扩展，但实际上需要做出影响严谨性、可重复性和解释性的非平凡选择。
cs.CV / 7 / 2608.05260
A Paragraph is Worth a Thousand Captions: Rethinking Text Supervision for Vision-Language Retrieval
一段文字胜过千条标题：重新思考视觉-语言检索中的文本监督
Abstract
Contrastive vision-language models such as CLIP and BLIP are typically trained on short image captions, limiting their ability to retrieve images from detailed textual descriptions. While methods such as Long-CLIP extend the token limit through positional embedding interpolation, we ask a simpler question: does training text granularity alone determine long-text retrieval performance? We present a systematic study of supervision ranging from single captions to multi-sentence paragraphs for contrastive image-text retrieval. Using a synthetic pipeline based on Qwen2-VL and Llama 3.2 Vision, we generate diverse captions, hard negatives, and quality-scored paragraphs for 500K CC3M images. To isolate the effect of text granularity, we fine-tune only the BLIP text encoder while keeping the vision encoder frozen across 10 training configurations. Our paragraph-supervised models match Long-CLIP-L on ShareGPT4V and outperform it by more than 14 points on DOCCI for image-to-text retrieval, without architectural changes. We further show that paragraph supervision enables effective use of long token sequences, whereas caption-only training degrades beyond 60 tokens. Increasing caption diversity improves short-caption retrieval with diminishing returns, while paragraph supervision consistently benefits long-description benchmarks and hard negatives prove detrimental in text-only fine-tuning. Evaluations on Flickr30k, COCO, ShareGPT4V, and DOCCI provide a comprehensive analysis of the trade-offs between text granularity, retrieval direction, and description length.
Chinese Translation
对比视觉-语言模型如 CLIP 和 BLIP 通常在短图像标题上进行训练，这限制了它们从详细文本描述中检索图像的能力。虽然 Long-CLIP 等方法通过位置嵌入插值扩展了令牌限制，但我们提出一个更简单的问题：训练文本粒度是否单独决定长文本检索性能？我们对从单个标题到多句段落的监督进行了系统研究，以用于对比图像-文本检索。基于 Qwen2-VL 和 Llama 3.2 Vision 的合成管道，我们为 500K CC3M 图像生成多样化的标题、困难负样本和质量评分段落。为了隔离文本粒度的影响，我们仅微调 BLIP 文本编码器，同时在 10 个训练配置中保持视觉编码器不变。我们的段落监督模型在 ShareGPT4V 上与 Long-CLIP-L 相匹配，并在图像到文本检索的 DOCCI 上超出其 14 分以上，而无需架构更改。我们进一步表明，段落监督能够有效利用长令牌序列，而仅使用标题的训练在超过 60 个令牌后性能下降。增加标题多样性改善了短标题检索，但收益递减，而段落监督始终对长描述基准有利，困难负样本在仅文本微调中证明是有害的。在 Flickr30k、COCO、ShareGPT4V 和 DOCCI 上的评估提供了文本粒度、检索方向和描述长度之间权衡的全面分析。
cs.CV / 8 / 2608.05333
Context Matters: Support Set Selection and Failure Detection for In-Context Medical Image Segmentation
上下文重要性：支持集选择与上下文医学图像分割中的失败检测
Abstract
In-context learning (ICL) adapts medical image segmentation models to unseen structures and modalities without retraining by conditioning on a task-specific support set of image-mask exemplars. Because this support set is the model's only task-specific signal, its composition directly influences segmentation performance. In this work, we investigate the support set as a controllable determinant of ICL reliability. First, we compare random sampling against similarity-based selection, where exemplars are retrieved based on their visual similarity to the query image. Second, we train a transformer-based classifier to predict, from the query and support images alone, whether a segmentation will fall below a specified Intersection-over-Union (IoU) threshold. Using MultiverSeg with DINOv3 embeddings across four benchmarks and three imaging modalities, we show that similarity-based selection consistently matches or outperforms random sampling, with the largest gains at the smallest support set sizes. Furthermore, our classifier predicts segmentation failure above chance on all four benchmarks. Ultimately, these results demonstrate that the reliability of in-context segmentation can be both improved via informed support selection and anticipated before use, providing practical mechanisms for safer clinical deployment.
Chinese Translation
上下文学习（ICL）通过依赖于特定任务的图像-掩膜示例支持集，能够将医学图像分割模型适应于未见过的结构和模态，而无需重新训练。由于该支持集是模型唯一的任务特定信号，其组成直接影响分割性能。在本研究中，我们探讨了支持集作为ICL可靠性的可控决定因素。首先，我们比较了随机采样与基于相似性的选择，其中示例是根据其与查询图像的视觉相似性进行检索的。其次，我们训练了一个基于变换器的分类器，仅通过查询图像和支持图像来预测分割是否会低于特定的交并比（IoU）阈值。通过在四个基准和三种成像模态下使用MultiverSeg与DINOv3嵌入，我们展示了基于相似性的选择始终与随机采样相匹配或优于随机采样，且在最小支持集大小时获得最大增益。此外，我们的分类器在所有四个基准上均能高于随机水平地预测分割失败。最终，这些结果表明，通过知情的支持选择，可以提高上下文分割的可靠性，并在使用前进行预判，为更安全的临床部署提供了实用机制。
cs.CV / 9 / 2608.05341
Positive-Unlabeled Preference Optimization For Chest X-ray Report Generation
胸部X光报告生成的正无标记偏好优化
Abstract
Vision-Language Models (VLMs) for radiology report generation are typically trained on retrospective clinical reports, which suffer from omission noise: clinically present findings are left unreported due to the omission of subtle findings. For example, prior studies show that cardiomegaly may be omitted from ICU chest X-ray reports when the imaging request is focused on monitoring support device placement. As a result, models trained with standard approaches inherit these omissions, learning to under-report findings themselves. We propose PU-DPO, a preference optimization framework to prevent omission noise from corrupting the preference signal. We reformulate the objective under a positive-unlabeled (PU) learning framework, treating absent mentions as unlabeled rather than truly negative. Our framework provides preference supervision using constructed contrastive pairs, generated using edits to model responses, producing variants that explicitly mention or omit a specific finding. Generated responses that mention the finding are naturally preferred in the context of visual evidence. Across semi-synthetic experiments and analyses on real-world chest radiograph benchmarks where adjudicated labels are available, PU-DPO yields consistent gains in detection rates and recovery of hidden positives across multiple pathologies, and is more robust to omission noise than prior approaches.
Chinese Translation
用于放射学报告生成的视觉-语言模型（VLMs）通常在回顾性临床报告上进行训练，这些报告存在遗漏噪声：由于未能报告细微发现，临床上存在的发现被遗漏。例如，先前的研究表明，当影像请求集中于监测支持设备的放置时，心脏肥大可能会在ICU胸部X光报告中被遗漏。因此，采用标准方法训练的模型继承了这些遗漏，学习到自己也会低报发现。我们提出了PU-DPO，一个偏好优化框架，以防止遗漏噪声破坏偏好信号。我们在正无标记（PU）学习框架下重新构建目标，将缺失的提及视为无标记而非真正的负例。我们的框架使用构建的对比对提供偏好监督，这些对比对通过对模型响应的编辑生成，产生明确提及或遗漏特定发现的变体。在视觉证据的背景下，提及该发现的生成响应自然被偏好。在半合成实验和真实世界胸部X光基准分析中，PU-DPO在多个病理学上实现了一致的检测率提升和隐藏阳性的恢复，并且比先前的方法对遗漏噪声更具鲁棒性。
cs.CV / 10 / 2608.05356
LoDA: A Level of Detection Aware Method and a Multimodal Sensing Benchmark for Object Level Change Detection
LoDA：一种检测水平感知方法及多模态感知基准用于物体级变化检测
Abstract
High-definition 3D LiDAR maps are important for autonomous driving and smart-city services, which require reliable detection of object-level changes in multi-temporal urban LiDAR to keep digital maps aligned with the physical world. Existing approaches from raster height differencing to depth image and point-cloud networks often remain tile-based and threshold-driven, yielding per-point scores without explicit detection limits or consistent object-level labels. We propose an object-level 3D change-detection pipeline that integrates detection-limit-aware registration, geometry-driven object proxies with rule-based semantic and instance segmentation, and displacement cues in height, volume, and surface-normal direction to assign five change labels with confidence. By decoupling registration, geometry, and semantics, the pipeline propagates pose uncertainty into spatially varying detection limits, stabilizes cross-epoch correspondences, and suppresses false changes caused by residual misalignment and density variation. We also present LoDA, a level-of-detection (LoD) aware benchmark for the Subiaco district with fused multi-temporal vehicle-LiDAR maps constructed with LiDAR, GNSS, and IMU support, semantic instances, and object-level annotations. On this benchmark, our method achieves 95.0% accuracy, 90.8% macro F1, and 83.0% macro IoU, exceeding the best baseline by 8.7 IoU points and 4.4 F1 points. On the public Urb3DCD-V2 benchmark evaluated under the official point-wise protocol, it reaches 96.81% mean accuracy and 89.52% mean change IoU, improving over the strongest reported baselines by 1.36 points in mAcc and 3.18 points in mIoUch.
Chinese Translation
高分辨率3D LiDAR地图对于自动驾驶和智慧城市服务至关重要，这些服务需要可靠地检测多时相城市LiDAR中的物体级变化，以保持数字地图与物理世界的一致性。现有的方法从栅格高度差异到深度图像和点云网络，往往仍然基于瓦片和阈值驱动，产生每个点的得分而没有明确的检测限制或一致的物体级标签。我们提出了一种物体级3D变化检测管道，该管道集成了感知检测限制的配准、基于几何的物体代理与基于规则的语义和实例分割，以及在高度、体积和表面法向方向上的位移线索，以赋予五种变化标签及其置信度。通过解耦配准、几何和语义，该管道将姿态不确定性传播到空间变化的检测限制中，稳定跨时间段的对应关系，并抑制由残余错位和密度变化引起的虚假变化。我们还提出了LoDA，一个检测水平（LoD）感知的基准，针对Subiaco区，构建了融合了LiDAR、GNSS和IMU支持的多时相车辆LiDAR地图，包含语义实例和物体级注释。在该基准上，我们的方法达到了95.0%的准确率，90.8%的宏观F1值和83.0%的宏观IoU，超过了最佳基线8.7个IoU点和4.4个F1点。在公共的Urb3DCD-V2基准上，根据官方逐点协议评估，达到了96.81%的平均准确率和89.52%的平均变化IoU，相较于报告的最强基线提高了1.36个mAcc点和3.18个mIoUch点。
cs.CV / 11 / 2608.05389
Text-Guided Refinement of Multi-sequence Glioma Subregion Segmentation with a Vision-Language Foundation Model
基于文本引导的多序列胶质瘤亚区分割精细化：一种视觉-语言基础模型
Abstract
Background: Accurate glioma subregion delineation is important for radiotherapy planning and longitudinal monitoring, but manual contour correction is time-consuming. Models such as nnU-Net may generalize imperfectly and lack clinician-directed text correction. Purpose: We investigated adapting a three-dimensional (3D) vision-language foundation model for text-guided brain tumor segmentation refinement. Methods: We developed a lightweight VoxTell-based framework. Pretrained VoxTell generated initial masks. Oracle prompts derived from segmentation errors encoded target, action, location, imaging evidence, edit size, and preservation constraints. Frozen Qwen/VoxTell prompt embeddings were injected through trainable projections into its multiscale decoder conditioning; other weights remained frozen. Training, validation, and testing used 901, 100, and 250 BraTS-GLI cases. Cross-dataset transfer was evaluated on 100 meningioma, metastasis, pediatric tumor, and UPENN-GBM cases. Results: On the internal test set using post-contrast T1-weighted input, correct instructions improved subregion Dice similarity coefficient (DSC; enhancing tumor, edema, and necrotic/non-enhancing core) from $0.774\pm0.158$ to $0.796\pm0.137$. They outperformed blank prompts ($0.762\pm0.155$; Holm-adjusted $p<0.001$, $d_z=0.71$) and contradictory prompts ($0.770\pm0.163$; $p<0.001$, $d_z=0.48$). In cross-dataset testing, correct instructions improved DSC from $0.527\pm0.287$ to $0.550\pm0.278$ and outperformed contradictory instructions ($0.504\pm0.275$; $p<0.001$, $d_z=0.43$). Conclusion: A 3D vision-language foundation model can perform instruction-guided refinement of glioma subregion segmentations. Sensitivity to correct, blank, and contradictory prompts suggests text-dependent contour editing rather than nonspecific post-processing, supporting further evaluation as a clinician-in-the-loop tool.
Chinese Translation
背景：准确的胶质瘤亚区划分对放射治疗规划和纵向监测至关重要，但手动轮廓修正耗时较长。像 nnU-Net 这样的模型可能无法完美泛化，并且缺乏临床医生指导的文本修正。目的：我们研究了如何将三维（3D）视觉-语言基础模型应用于文本引导的脑肿瘤分割精细化。方法：我们开发了一个轻量级的基于 VoxTell 的框架。预训练的 VoxTell 生成初始掩膜。源自分割错误的 Oracle 提示编码了目标、动作、位置、影像证据、编辑大小和保留约束。冻结的 Qwen/VoxTell 提示嵌入通过可训练的投影注入其多尺度解码器条件中；其他权重保持冻结。训练、验证和测试使用了 901、100 和 250 个 BraTS-GLI 案例。跨数据集转移在 100 个脑膜瘤、转移性肿瘤、儿童肿瘤和 UPENN-GBM 案例上进行了评估。结果：在使用对比增强 T1 加权输入的内部测试集中，正确的指令将亚区 Dice 相似性系数（DSC；增强肿瘤、水肿和坏死/非增强核心）从 $0.774 ext{±}0.158$ 提高到 $0.796 ext{±}0.137$。它们的表现优于空白提示（$0.762 ext{±}0.155$；Holm 调整后 $p<0.001$，$d_z=0.71$）和矛盾提示（$0.770 ext{±}0.163$；$p<0.001$，$d_z=0.48$）。在跨数据集测试中，正确的指令将 DSC 从 $0.527 ext{±}0.287$ 提高到 $0.550 ext{±}0.278$，并优于矛盾指令（$0.504 ext{±}0.275$；$p<0.001$，$d_z=0.43$）。结论：三维视觉-语言基础模型能够进行胶质瘤亚区分割的指令引导精细化。对正确、空白和矛盾提示的敏感性表明文本依赖的轮廓编辑，而非非特定的后处理，支持作为临床医生参与的工具进一步评估。
cs.CV / 12 / 2608.05393
Adapting Vision Foundation Models with Cascaded Semantics
通过级联语义调整视觉基础模型
Abstract
Prompt tuning, a leading parameter-efficient adaptation paradigm in NLP, has recently been extended to computer vision. Visual prompt tuning (VPT) adapts pre-trained vision transformers (ViTs) by updating a small set of additional prompt parameters. However, existing visual prompts are randomly initialized and do not exploit prior knowledge, such as instructions in NLP. We address this gap by injecting two complementary semantic priors into VPT. Fundamental image priors, including color, texture, and shape, are extracted with classical hand-crafted operators and injected into the input space, while self-attention maps provide instance-aware semantics in the feature space. We further propose a cascaded scheme that integrates both priors throughout ViT adaptation. Experiments on 34 challenging image classification datasets demonstrate superior downstream adaptation while tuning only 0.74% of ViT parameters. Project page: https://xixiaouab.github.io/Cascaded-Semantics/.
Chinese Translation
提示调优（Prompt tuning）是一种在自然语言处理（NLP）中领先的参数高效适应范式，最近已扩展至计算机视觉。视觉提示调优（Visual Prompt Tuning, VPT）通过更新一小组额外的提示参数来调整预训练的视觉变换器（Vision Transformers, ViTs）。然而，现有的视觉提示是随机初始化的，并未利用先验知识，例如NLP中的指令。我们通过将两个互补的语义先验注入VPT来填补这一空白。基本图像先验，包括颜色、纹理和形状，通过经典的手工设计算子提取并注入到输入空间中，而自注意力图（self-attention maps）则在特征空间中提供实例感知语义。我们进一步提出了一种级联方案，在ViT适应过程中整合这两种先验。在34个具有挑战性的图像分类数据集上的实验表明，仅调整0.74%的ViT参数即可实现优越的下游适应。项目页面：https://xixiaouab.github.io/Cascaded-Semantics/
cs.CV / 13 / 2608.05424
Invisible Shortcuts: Why Vision Encoders Know Your Camera
隐形捷径：为何视觉编码器能够识别你的相机
Abstract
Deep vision models exploit shortcuts, relying on cues that correlate with supervision signals. Prior work has focused on visible biases, such as object-background or texture correlations. We identify a different source of shortcut learning: invisible metadata traces embedded at the pixel level, for metadata such as image processing and photo acquisition. We hypothesize that large-scale semantic supervision, whether through categorical labels (ImageNet) or billion-scale captions (LAION), naturally induces metadata-semantics correlations during pretraining, leading models to convert low-level signals into predictive features. By introducing controlled metadata-semantics correlations, we show that stronger ones produce systematically higher sensitivity to metadata traces and larger performance degradation under metadata distribution shifts. We further explore mitigation strategies applied during and after pretraining that reduce sensitivity not only to targeted metadata but also to unseen ones, without sacrificing performance on downstream tasks. Metadata sensitivity also has a positive side: it partly explains the strong generated-image detection ability of some encoders, while its mitigation can improve out-of-distribution generalization. Code: https://github.com/ryan-caesar-ramos/visual-encoder-traces
Chinese Translation
深度视觉模型利用捷径，依赖与监督信号相关的线索。先前的研究集中于可见偏差，例如物体-背景或纹理相关性。我们识别出一种不同的捷径学习来源：嵌入在像素级别的不可见元数据痕迹，这些元数据包括图像处理和照片获取等信息。我们假设，大规模的语义监督，无论是通过类别标签（ImageNet）还是十亿规模的描述（LAION），在预训练过程中自然地引入了元数据与语义的相关性，导致模型将低级信号转换为预测特征。通过引入受控的元数据-语义相关性，我们展示了更强的相关性系统性地提高了对元数据痕迹的敏感性，并在元数据分布变化下导致更大的性能下降。我们进一步探索了在预训练期间和之后应用的缓解策略，这些策略不仅减少了对目标元数据的敏感性，还降低了对未见元数据的敏感性，而不牺牲下游任务的性能。元数据敏感性也有积极的一面：它在一定程度上解释了某些编码器强大的生成图像检测能力，而其缓解可以改善分布外泛化。代码链接： https://github.com/ryan-caesar-ramos/visual-encoder-traces
cs.CV / 14 / 2608.05450
MOSAIK: Multi-Patch Content-Aware Spatial Allocation of Image Tokens for Efficient Generation
MOSAIK：用于高效生成的多补丁内容感知空间分配图像标记
Abstract
Pixel-space diffusion models avoid the reconstruction ceiling of latent diffusion models by generating directly in image space. However, their substantially higher token count makes generation expensive due to the quadratic complexity of self-attention. Several existing efficiency methods reduce this cost by using larger patches at selected denoising steps, thereby representing the image with fewer tokens. Yet, each step still uses a single patch size uniformly across the entire image, overlooking that different regions suffer different fidelity losses when coarsened. We introduce MOSAIK, a damage-guided framework that varies patch size across regions and denoising steps. MOSAIK adapts the PixelDiT backbone to generate arbitrary heterogeneous patch layouts, and a lightweight predictor uses intermediate denoising features to estimate the fidelity loss caused by coarsening each region. Given a token budget, our damage-guided layout predictor assigns fine patches to sensitive regions and coarse patches elsewhere. Remarkably, while reducing FLOPs by 70% and token count by 83%, MOSAIK matches the full-compute PixelDiT on GenEval and its DPG-Bench score drops by only 1.0 point. Compared to diverse efficiency paradigms, including temporal patch scheduling and feature caching, our approach delivers highly competitive performance at moderate budgets and consistently outperforms these baselines in highly constrained compute regimes.
Chinese Translation
像素空间扩散模型通过直接在图像空间生成，避免了潜在扩散模型的重建上限。然而，由于自注意力的平方复杂性，其显著更高的标记数量使得生成成本昂贵。一些现有的效率方法通过在选定的去噪步骤中使用更大的补丁来降低这一成本，从而用更少的标记表示图像。然而，每一步仍然在整个图像中均匀使用单一的补丁大小，忽视了不同区域在粗化时遭受不同的保真度损失。我们提出了MOSAIK，一个基于损伤引导的框架，它在不同区域和去噪步骤中变化补丁大小。MOSAIK调整PixelDiT主干以生成任意异构的补丁布局，轻量级预测器利用中间去噪特征来估计每个区域粗化所造成的保真度损失。在给定标记预算的情况下，我们的损伤引导布局预测器将细补丁分配给敏感区域，而在其他地方使用粗补丁。值得注意的是，尽管将FLOPs减少了70%和标记数量减少了83%，MOSAIK在GenEval上与全计算的PixelDiT相匹配，其DPG-Bench得分仅下降1.0分。与包括时间补丁调度和特征缓存等多种效率范式相比，我们的方法在适度预算下提供了高度竞争的性能，并在高度受限的计算环境中始终优于这些基准。
cs.CV / 15 / 2608.05482
CDSeg: A Renderable Gaussian Carrier for Image-to-3D Label Transfer
CDSeg：用于图像到3D标签转移的可渲染高斯载体
Abstract
Modern image models provide strong cues about \emph{what} should be segmented in each view, but their masks do not by themselves determine \emph{where} those labels should persist in 3D. We present Cross-Domain Segmentation via Gaussian Splatting (CDSeg), a label-transfer interface that requires no task-specific 3D segmentation training and uses Gaussian primitives as a renderable label carrier. An external mask source supplies the labels, while renderer-derived visibility determines which 3D primitives receive them. The carrier is instantiated either by completing each input point into one Gaussian, preserving its index, or by reusing the native primitives of an optimized Gaussian scene. CDSeg records pixel--primitive associations during rendering and fuses multi-view masks through voting and a local filter. The resulting labels can be returned to the original points, retained on the native Gaussian scene, or rendered into other views. CDSeg covers promptable, automatic instance, semantic, and LiDAR settings and processes scenes with millions of primitives in seconds. It obtains 92.35\% mIoU on DesktopObjects-360, 95.89\% on NeRDS-360, and 65.77\% on the full ScanNet-v2 validation split using the provided 2D semantic annotations. CDSeg thereby provides one interface for reusing 2D masks across point clouds, Gaussian scenes, and image views without a task-specific 3D segmentation network.
Chinese Translation
现代图像模型提供了关于每个视图中应当分割的内容的强烈线索，但它们的掩膜本身并不能确定这些标签在3D中应当持续存在的位置。我们提出了通过高斯溅射进行跨域分割（CDSeg），这是一种标签转移接口，无需特定任务的3D分割训练，并使用高斯原语作为可渲染的标签载体。外部掩膜源提供标签，而渲染器派生的可见性决定哪些3D原语接收这些标签。载体的实例化可以通过将每个输入点补全为一个高斯并保留其索引，或者通过重用优化高斯场景的本地原语来完成。CDSeg在渲染过程中记录像素与原语的关联，并通过投票和局部过滤融合多视图掩膜。最终生成的标签可以返回到原始点，保留在本地高斯场景中，或渲染到其他视图中。CDSeg涵盖了可提示的、自动的实例、语义和激光雷达设置，并在几秒钟内处理数百万个原语的场景。在DesktopObjects-360上获得92.35%的mIoU，在NeRDS-360上获得95.89%，在完整的ScanNet-v2验证集上获得65.77%，使用提供的2D语义注释。因此，CDSeg提供了一个接口，用于在点云、高斯场景和图像视图之间重用2D掩膜，而无需特定任务的3D分割网络。
cs.CV / 16 / 2608.05485
VideoArgus: Agentic Rubric-Grounded Unified Evaluation for Video Generation and Editing
VideoArgus：基于代理评分标准的统一视频生成与编辑评估
Abstract
Evaluating generated videos remains challenging because existing benchmarks rely on fixed evaluation content, cover only a subset of generation and editing settings, and provide limited evidence for their scores. We introduce VideoArgus, a unified rubric-grounded framework covering five video generation and editing settings. For each input instance, VideoArgus generates an output-blind, sample-specific rubric once and reuses it to evaluate all corresponding candidate videos. The rubric defines concrete criteria, scoring rules, failure modes, and evidence plans, which guide criterion-specific VLM QA and visual tools to produce evidence-grounded criterion scores, rationales, and a diagnostic report. We further construct VideoArgus-Bench, containing 1,026 curated input instances built from 653 high-quality images and 416 high-quality videos, with all benchmark rubrics pre-generated, frozen, and released. On a separate 1,260-video human-alignment set, VideoArgus achieves higher within-input Spearman and Kendall correlations with human judgments than the corresponding benchmark-specific evaluators across all five tasks. Model rankings also remain largely consistent across different rubric-generation and evaluation-VLM backbones. All code and data are released. Visit our project page: https://zzzmyyzeng.github.io/VideoArgus
Chinese Translation
评估生成的视频仍然具有挑战性，因为现有基准依赖于固定的评估内容，仅覆盖生成和编辑设置的子集，并且提供的评分证据有限。我们引入了VideoArgus，一个统一的基于评分标准的框架，涵盖五种视频生成和编辑设置。对于每个输入实例，VideoArgus 一次性生成一个输出盲样本特定的评分标准，并将其重复用于评估所有相应的候选视频。该评分标准定义了具体的标准、评分规则、失败模式和证据计划，指导特定标准的视觉语言模型（VLM）问答和视觉工具生成基于证据的标准评分、理由和诊断报告。我们进一步构建了VideoArgus-Bench，包含1,026个从653张高质量图像和416个高质量视频中精选的输入实例，所有基准评分标准均已预生成、冻结并发布。在一个独立的1,260视频人类对齐集上，VideoArgus在所有五个任务中与人类判断的输入内斯皮尔曼和肯德尔相关性均高于相应的基准特定评估器。模型排名在不同的评分标准生成和评估VLM骨干网络之间也保持了较大的一致性。所有代码和数据均已发布。请访问我们的项目页面：https://zzzmyyzeng.github.io/VideoArgus
cs.CV / 17 / 2608.05499
APQF: Agentic Profiling-Guided Structured Pruning and Mixed-Precision Quantization with Adaptive Fine-Tuning
APQF：基于代理剖析指导的结构化剪枝与混合精度量化及自适应微调
Abstract
Modern deep neural networks achieve strong performance, but their scale makes them costly and slow, especially on resource-constrained edge devices. Pruning and quantization address this, but rely on manual, expert choices and on algorithms that are hard to apply across architectures. Uniform settings also ignore how differently individual layers respond to compression, which costs accuracy. We introduce APQF, an agentic profiling-guided framework that combines structured pruning, mixed-precision quantization-aware training, and accuracy recovery in one automated pipeline. A profiling agent measures how cost is distributed across the model and how sensitive each part is to pruning, and this evidence drives per-layer pruning ratios, per-layer bit-widths, and the recovery strategy, all proposed by LLM planners and validated before execution. To our knowledge, APQF is the first framework to combine LLM-guided, profiling-grounded decisions with a fully training-aware pruning and quantization pipeline for both CNNs and vision transformers. We evaluate APQF on ResNet, VGG7, ViT, DeiT, and Swin using ImageNet-1k and CIFAR-10. On ImageNet it cuts compute to 5.6-7.7 percent of the original bit-operations, a 13-18x reduction, while keeping accuracy close to the baseline, and under a 200K-image budget it stays roughly 17 points higher in Top-1 than existing joint pruning and quantization methods. On CIFAR-10 it compresses further than that method on four of five architectures. On VGG7 it reaches 93.15 percent using only 0.41 percent of baseline bit-operations, the only method at that compression level to improve on its full-precision baseline. Ablations show that uniform compression loses the most accuracy at matched compute, and that withholding profiling data from the planner hurts every model. Six LLM planners, including free open-weight ones, all reach 97.4-97.9 percent on Swin-Tiny.
Chinese Translation
现代深度神经网络在性能上表现出色，但其规模使得它们在资源受限的边缘设备上成本高且速度慢。剪枝和量化可以解决这个问题，但依赖于手动的专家选择以及难以在不同架构中应用的算法。统一的设置也忽视了各个层对压缩的不同响应，这会影响准确性。我们提出了APQF，一个基于代理剖析指导的框架，结合了结构化剪枝、混合精度量化感知训练和准确性恢复于一个自动化流程中。剖析代理测量模型中成本的分布以及每个部分对剪枝的敏感性，这些证据驱动每层的剪枝比例、每层的位宽和恢复策略，所有这些都是由大型语言模型（LLM）规划者提出并在执行前进行验证。据我们所知，APQF是第一个将LLM指导的、基于剖析的决策与完全训练感知的剪枝和量化流程结合在一起的框架，适用于卷积神经网络（CNN）和视觉变换器（vision transformers）。我们在ResNet、VGG7、ViT、DeiT和Swin上使用ImageNet-1k和CIFAR-10对APQF进行了评估。在ImageNet上，它将计算量减少到原始比特操作的5.6-7.7%，减少了13-18倍，同时保持准确性接近基线，并且在200K图像预算下，其Top-1准确率比现有的联合剪枝和量化方法高出约17个百分点。在CIFAR-10上，它在五种架构中的四种上压缩效果超过了该方法。在VGG7上，它仅使用0.41%的基线比特操作达到了93.15%的准确率，是在该压缩水平下唯一一个在全精度基线上有所提升的方法。消融实验表明，统一压缩在匹配计算下损失了最多的准确性，而不向规划者提供剖析数据会对每个模型造成损害。包括免费开放权重的六个LLM规划者在Swin-Tiny上均达到了97.4-97.9%的准确率。
cs.CV / 18 / 2608.05505
DynaPix: Can Vision-Language Models Identify the Exact Future?
DynaPix：视觉-语言模型能否识别确切的未来？
Abstract
Acting in a physical scene requires knowing its real later state, not a plausible one. Current evaluations often accept words or a realistic-looking image, so the predicted state is never checked against the true one. We introduce DynaPix (Dynamic Pixels), a benchmark that makes prediction checkable. Given a video clip that stops before a key event and a question about a later moment, a model must pick the true future image from close candidates or a large gallery. The scenes come from a physics simulator, so the correct image and its time are known exactly and the wrong options are deliberately similar. Models often succeed when a visible event marks the target moment, but are near chance when only elapsed time marks it. Gallery search is harder still, as the true image rarely ranks first. People handle the elapsed-time items well, so the difficulty lies with the models, not the questions. Training on scene accounts drawn from the simulator's true record, not a teacher's guess, repairs much of this but not the longer elapsed time case. DynaPix thus exposes a temporal-anchoring gap: models attach a prediction to an event far better than to time itself.
Chinese Translation
在物理场景中进行行动需要了解其真实的后续状态，而非一个合理的状态。目前的评估通常接受文字或看起来逼真的图像，因此预测的状态从未与真实状态进行核对。我们引入了 DynaPix（动态像素），一个使预测可检验的基准。给定一个在关键事件之前停止的视频片段和一个关于后续时刻的问题，模型必须从接近的候选图像或一个大型图库中选择出真实的未来图像。这些场景来自物理模拟器，因此正确的图像及其时间是完全已知的，而错误选项则故意相似。当一个可见事件标记目标时刻时，模型通常能够成功，但当只有经过的时间标记时刻时，成功率接近随机。图库搜索更具挑战性，因为真实图像很少排在第一位。人们能够很好地处理经过时间的项目，因此困难在于模型，而不是问题。基于模拟器真实记录的场景描述进行训练，而不是教师的猜测，可以修复大部分问题，但对于较长的经过时间情况则无能为力。因此，DynaPix揭示了一个时间锚定的差距：模型将预测与事件的关联远远优于与时间本身的关联。
cs.CV / 19 / 2608.05523
HERA: Historical Evidence Routing Adapter for Physical Prediction in Latent World Models
HERA：用于潜在世界模型中物理预测的历史证据路由适配器
Abstract
Predictive video models have emerged as promising world models by learning latent visual dynamics from large-scale video. Yet these models remain challenged by physical events under occlusion, where later predictions may depend on object evidence that is no longer available in the current view. Addressing this challenge requires historical evidence not only to be preserved but also to remain accessible when it becomes relevant to a subsequent prediction. Existing approaches mainly enlarge the temporal context, cache generic video features, or impose explicit object-centric states, thereby improving the capacity or structure of retained history. However, they do not directly address how relevant historical evidence can be selectively retrieved and integrated into a pretrained predictor without interfering with its native latent workspace. Accordingly, we introduce HERA (Historical Evidence Routing Adapter), a framework for routing retained historical evidence into a frozen latent predictor, and instantiate it with Register-Routed Patch Memory (RRPM), a lightweight adapter comprising a Structured Memory Bank, Memory Registers, and Workspace Registers. On the IntPhys2 Main split, HERA with RRPM improves the pairwise AvgSurprise accuracy of V-JEPA 2-G from 52.57% to 54.35%. Subgroup analysis shows particularly strong improvements on fixed-camera continuity, from 46.15% to 57.69%, and fixed-camera immutability, from 46.15% to 63.46%. These results support historical evidence routing as a practical adaptation strategy for physical prediction in latent world models.
Chinese Translation
预测视频模型通过从大规模视频中学习潜在视觉动态而成为有前景的世界模型。然而，这些模型在遮挡下的物理事件面临挑战，此时后续预测可能依赖于当前视图中不再可用的物体证据。解决这一挑战需要不仅保存历史证据，还要在其与后续预测相关时保持可访问性。现有方法主要扩大时间上下文，缓存通用视频特征，或施加显式的以物体为中心的状态，从而提高保留历史的能力或结构。然而，它们并未直接解决如何选择性地检索和整合相关历史证据到预训练预测器中，而不干扰其原生潜在工作空间。因此，我们提出了HERA（历史证据路由适配器），一个将保留的历史证据路由到冻结潜在预测器的框架，并通过注册路由补丁内存（Register-Routed Patch Memory, RRPM）进行实例化，RRPM是一个轻量级适配器，包含结构化内存库、内存寄存器和工作空间寄存器。在IntPhys2主分割上，HERA与RRPM结合将V-JEPA 2-G的成对AvgSurprise准确率从52.57%提高到54.35%。子组分析显示在固定摄像机连续性方面的改善尤为显著，从46.15%提高到57.69%，在固定摄像机不变性方面，从46.15%提高到63.46%。这些结果支持历史证据路由作为潜在世界模型中物理预测的实用适应策略。
cs.CV / 20 / 2608.05539
OmniMech: All-in-one Multimodal Mechanical Benchmark for 3D Reconstruction
OmniMech：一体化多模态机械基准用于3D重建
Lu, Taiting, Liu, Runze, Dong, Ziwei, Bei, Sisong, Zeng, Jingying, Wang, Mingjia, Li, Zhenghao, Lin, Kaiyuan, Wu, Yi-Shan, Zheng, Yangshoudu, Pan, Hongxing, Zhang, Kai, Shi, Guoliang, Ma, Ling, Yang, Yifan, Lu, Jiaying, He, Qi, Chen, Sung-Liang, Chen, Yi-Chao, Jin, Yincheng, Gowda, Mahanth
Abstract
Recent vision-language models (VLMs) can generate executable CAD programs from images, but existing methods mainly target coarse, general-purpose 3D objects and rarely address the fine-grained geometry and millimeter-level tolerances required in industrial mechanical design. We introduce OmniMech, the first million-scale benchmark for evaluating VLMs on executable CAD generation from industrial manufacturing data. OmniMech contains more than 251,000 fully dimensioned and toleranced 2D orthographic drawings, paired with native CAD models, multi-view renderings, mesh, STEP and B-rep representations, and rich semantic annotations. The benchmark includes four tasks: (1) parametric CAD program synthesis from engineering drawings; (2) diagram-to-3D reasoning for geometrically and structurally consistent reconstruction; (3) annotation-grounded reasoning over dimensions, symbols, feature callouts, and manufacturing constraints; and (4) tool-augmented agentic reasoning using visualization, measurement, CAD execution, and verification tools. Experiments show that current VLMs and CAD-specialized models still struggle with executable program synthesis, fine-grained 3D reconstruction, and reliable enforcement of dimensions and tolerances. We will release the benchmark data, evaluation code, and tool interfaces to support future research.
Chinese Translation
近期的视觉-语言模型（VLMs）能够从图像生成可执行的CAD程序，但现有方法主要针对粗略的通用3D物体，鲜有涉及工业机械设计中所需的细粒度几何形状和毫米级公差。我们提出了OmniMech，这是第一个百万规模的基准，用于评估VLM在工业制造数据中生成可执行CAD的能力。OmniMech包含超过251,000个完全标注和公差的2D正投影图，配有原生CAD模型、多视图渲染、网格、STEP和B-rep表示，以及丰富的语义注释。该基准包括四个任务：（1）从工程图纸合成参数化CAD程序；（2）图形到3D推理，以实现几何和结构一致的重建；（3）基于注释的推理，涉及尺寸、符号、特征标注和制造约束；（4）使用可视化、测量、CAD执行和验证工具的工具增强代理推理。实验表明，当前的VLM和专门针对CAD的模型在可执行程序合成、细粒度3D重建以及可靠执行尺寸和公差方面仍然面临挑战。我们将发布基准数据、评估代码和工具接口，以支持未来的研究。
cs.CV / 21 / 2608.05557
Hierarchical Flow Matching for 3D Point Cloud Generation
层次流匹配用于三维点云生成
Abstract
Generating high-quality 3D point clouds requires capturing both global shape topology and local geometric details. Existing flow-based methods rely on continuous normalizing flows (CNFs) that demand expensive ODE solving and trace estimation during training, while diffusion models require hundreds of iterative denoising steps. Moreover, most approaches adopt single-level generation directly in point space, disregarding the hierarchical structure natural to 3D shapes. We propose Hierarchical Flow Matching (HFM) that extends flow matching to bilevel structure for unconditional 3D point cloud generation. HFM decomposes the task into two levels via optimal-transport flow matching: a \textit{Latent Flow Matching} models the global shape manifold in a compact latent space, and a \textit{Conditional Point Flow Matching} reconstructs detailed point clouds conditioned on the latent code. Both flows are trained with simple MSE regression losses. The resulting straight OT paths enable efficient sampling with as few as 15 Euler steps per flow, while the structured latent space supports downstream tasks including classification. Extensive experiments on ShapeNet and ModelNet benchmarks demonstrate that HFM achieves competitive or even best performance compared with prior state-of-the-art methods.
Chinese Translation
生成高质量的三维点云需要捕捉全局形状拓扑和局部几何细节。现有的基于流的方法依赖于连续归一化流（CNFs），这在训练过程中需要昂贵的常微分方程（ODE）求解和轨迹估计，而扩散模型则需要数百次迭代去噪步骤。此外，大多数方法直接在点空间中采用单层生成，忽视了三维形状的层次结构。我们提出了层次流匹配（HFM），将流匹配扩展到双层结构，以实现无条件的三维点云生成。HFM通过最优传输流匹配将任务分解为两个层次： extit{潜在流匹配}在紧凑的潜在空间中建模全局形状流形，而 extit{条件点流匹配}则基于潜在编码重建详细的点云。这两个流通过简单的均方误差（MSE）回归损失进行训练。最终得到的直线最优传输路径使得每个流的高效采样只需15个欧拉步骤，同时结构化的潜在空间支持包括分类在内的下游任务。在ShapeNet和ModelNet基准上的大量实验表明，HFM在与之前的最先进方法相比时，达到了具有竞争力甚至最佳的性能。
cs.CV / 22 / 2608.05560
From Sports to Safety: Benchmarking Proactive Risk Inference in MLLMs
从体育到安全：主动风险推断在多模态大模型中的基准测试
Abstract
Timely anticipation of physical hazards is essential for real-world safety, yet existing MLLM evaluations focus on harmful content or general risks, leaving proactive physical hazard prediction underexplored. Sports provide a well-suited testbed: accident causes span diverse injury dimensions and pre-accident spatiotemporal cues draw on reasoning capabilities shared with broader safety domains such as autonomous driving and fall detection. We introduce SPRINT (Sports Proactive Risk INference Testbed), a benchmark of 2,888 real-world sports videos (2,440 accident, 448 safe controls) spanning 14 sports and 3 environmental settings. Accident videos feature fine-grained annotations of early hazard cues, accident timing, and hierarchical causes; safe videos are manually verified as accident-free and serve to diagnose prompt-induced false alarms. Evaluating state-of-the-art MLLMs under diverse prompts and temporal windows reveals a sharp gap between hazard sensitivity and understanding: the best model exceeds 95% in signaling hazards yet falls below 50% in identifying their causes. Diagnostic experiments further show that explicit danger queries trigger severe false alarms even on hazard-free videos. These findings indicate that current MLLMs exhibit only superficial proactive safety, lacking stable, cause-grounded early warning, and underscore the need for reliable proactive safety in dynamic physical environments. Data and code will be open-sourced upon acceptance.
Chinese Translation
及时预判物理危险对现实世界的安全至关重要，但现有的多模态大模型（MLLM）评估主要集中在有害内容或一般风险上，主动的物理危险预测尚未得到充分探索。体育运动提供了一个合适的测试平台：事故原因涵盖多种伤害维度，事故前的时空线索依赖于与更广泛的安全领域（如自动驾驶和跌倒检测）共享的推理能力。我们引入了 SPRINT（体育主动风险推断测试平台），这是一个包含 2,888 个真实体育视频（2,440 个事故视频，448 个安全对照视频）的基准，涵盖 14 种体育项目和 3 种环境设置。事故视频具有对早期危险线索、事故时机和层次原因的细粒度注释；安全视频经过人工验证为无事故，并用于诊断提示引发的误报。对在不同提示和时间窗口下的最先进 MLLM 进行评估，揭示了危险敏感性与理解之间的显著差距：最佳模型在信号危险方面超过 95%，但在识别其原因方面却低于 50%。诊断实验进一步表明，明确的危险查询即使在无危险视频上也会引发严重的误报。这些发现表明，当前的 MLLM 仅表现出表面的主动安全，缺乏稳定的、基于原因的早期预警，并强调了在动态物理环境中对可靠主动安全的需求。数据和代码将在接受后开源。
cs.CV / 23 / 2608.05565
EffectLearner: World-Aware Object-Effect Reasoning for Real-World Video Object Removal
EffectLearner：面向现实世界视频物体移除的世界感知物体-效果推理
Abstract
Video object removal must eliminate not only the target object but also its induced effects while maintaining high-fidelity and spatiotemporally coherent restoration. Existing methods mainly learn object-effect correspondences implicitly from predefined effect categories and fixed data distributions, limiting their generalization to complex real-world scenes involving compositional effects, spatially detached or weakly correlated effects, long-tail physical phenomena, and dynamically evolving interactions. We propose EffectLearner, a semantic-reasoning-enhanced framework that combines a VLM-based Object-Effect Reasoner with a DiT-based Video Eraser. Guided by a structured effect-analysis prompt, the Reasoner performs cross-modal reasoning over a target-highlighted video and extracts compact effect-aware context, which guides the Video Eraser toward comprehensive object-effect removal. Motion-aware mask guidance and motion-consistency supervision further improve removal coverage and spatiotemporal stability under object motion and evolving scene dynamics. To fully exploit the framework in challenging real-world scenarios, we further construct EffectWorld, a paired video dataset specifically designed for complex object-induced effects, and introduce a progressive training curriculum that combines common supervision with complex-effect data. On the standard ROSE-Bench, EffectLearner outperforms existing baselines on most metrics and achieves clear advantages on both EffectWorld-Eval and the challenging EffectWorld-Wild, demonstrating its ability to deliver high-quality video object removal in complex real-world scenes.
Chinese Translation
视频物体移除不仅需要消除目标物体，还需去除其引发的效果，同时保持高保真度和时空一致的恢复。现有方法主要从预定义的效果类别和固定的数据分布中隐式学习物体-效果对应关系，这限制了它们在涉及组合效果、空间上分离或弱相关效果、长尾物理现象以及动态演变交互的复杂现实场景中的泛化能力。我们提出了EffectLearner，一个增强语义推理的框架，它结合了基于VLM的物体-效果推理器和基于DiT的视频擦除器。在结构化效果分析提示的指导下，推理器在目标高亮视频上进行跨模态推理，并提取紧凑的效果感知上下文，从而指导视频擦除器进行全面的物体-效果移除。运动感知掩码指导和运动一致性监督进一步提高了在物体运动和场景动态演变下的移除覆盖率和时空稳定性。为了充分利用该框架应对具有挑战性的现实场景，我们进一步构建了EffectWorld，这是一个专门为复杂物体引发效果设计的配对视频数据集，并引入了一个渐进式训练课程，将常规监督与复杂效果数据相结合。在标准ROSE-Bench上，EffectLearner在大多数指标上超越了现有基线，并在EffectWorld-Eval和具有挑战性的EffectWorld-Wild上取得了明显优势，展示了其在复杂现实场景中提供高质量视频物体移除的能力。
cs.CV / 24 / 2608.05569
CoordRefer: Coordinate-Aware 3D Visual Grounding from Multiview Images
CoordRefer：基于多视图图像的坐标感知3D视觉定位
Abstract
Multiview image-based 3D visual grounding predicts a coordinate frame to define a coordinate system and then regresses a 3D bounding box for localization. However, existing methods jointly optimize coordinate frame selection and box regression, leading to coordinate-relative box ambiguity and degraded grounding performance. This ambiguity arises because the same box admits different numerical representations across coordinate frames, creating multiple optimization targets and yielding invalid compromise predictions. To tackle this challenge, we propose CoordRefer, a coordinate-aware framework that decouples coordinate frame selection from coordinate-conditioned grounding. CoordRefer first selects a reference frame to define the coordinate system and then conditions 3D box prediction on the coordinate system. We perform coordinate-aware supervised fine-tuning to establish coordinate frame selection and coordinate-conditioned box regression, followed by Group Relative Policy Optimization with 3D IoU-based rewards to align both stages with downstream grounding quality. On ScanRefer with Qwen3-VL-2B, CoordRefer achieves gains of 11% in
[email protected] and 7% in
[email protected] over the coordinate-agnostic baseline, while its geometrically refined variant surpasses methods using explicit 3D inputs.
Chinese Translation
基于多视图图像的3D视觉定位预测一个坐标框架以定义坐标系统，然后回归一个3D边界框进行定位。然而，现有方法共同优化坐标框架选择和边界框回归，导致坐标相对的边界框模糊性和定位性能下降。这种模糊性产生的原因在于同一个边界框在不同的坐标框架下可以有不同的数值表示，造成多个优化目标并产生无效的妥协预测。为了解决这一挑战，我们提出了CoordRefer，一个坐标感知框架，将坐标框架选择与坐标条件下的定位解耦。CoordRefer首先选择一个参考框架以定义坐标系统，然后在该坐标系统上进行3D边界框预测。我们进行坐标感知的监督微调，以建立坐标框架选择和坐标条件下的边界框回归，随后采用基于3D IoU奖励的组相对策略优化，使两个阶段与下游定位质量对齐。在ScanRefer数据集上，使用Qwen3-VL-2B，CoordRefer在
[email protected]上提高了11%，在
[email protected]上提高了7%，相较于无坐标感知的基线，其几何精细化变体超越了使用显式3D输入的方法。
cs.CV / 25 / 2608.05592
Beyond Frame Selection: Rethinking Long-Video Understanding with MLLMs
超越帧选择：用多模态大语言模型重新思考长视频理解
Abstract
Multimodal Large Language Models (MLLMs) have achieved strong progress in video understanding, yet it remains challenging because the token limitation makes MLLMs difficult to capture temporally sparse evidence. Existing methods typically rely on uniform sampling, or frame selection, but these strategies usually optimize either broad temporal coverage or local relevance, making it difficult to preserve both global storyline context and fine-grained evidence. We propose VideoRouter(VR) that rethinks long-video understanding as coordinating complementary evidence views rather than selecting a single subset of frames. It first organizes each video into a question-agnostic temporal hierarchy, which partition the video into coarse-to-fine temporally coherent segments. In this hierarchy, upper-level nodes capture broad storyline context and event progression, while lower-level nodes preserve fine-grained local details and evidence-bearing moments. This naturally gives rise to two complementary views: a global view for coverage-oriented reasoning and a local view for detail-oriented evidence recovery. We further introduce a verification-guided router to determine which view is better supported by the selected evidence and select the final answer. We validate the effectiveness of the proposed design through extensive experiments, showing that the verification-guided router effectively coordinates global and local reasoning, and that, on VideoMME, our method outperforms state-of-the-art frame selection methods by 2.9 points, respectively, under the LLaVA-Video-7B backbone. We will release the code.
Chinese Translation
多模态大语言模型（MLLMs）在视频理解方面取得了显著进展，但由于令牌限制，使得MLLMs难以捕捉时间上稀疏的证据，这仍然是一个挑战。现有方法通常依赖于均匀采样或帧选择，但这些策略通常优化广泛的时间覆盖或局部相关性，难以同时保留全球故事情节的上下文和细粒度的证据。我们提出了VideoRouter（VR），将长视频理解重新构思为协调互补证据视图，而不是选择单一的帧子集。它首先将每个视频组织成一个与问题无关的时间层次结构，将视频划分为粗到细的时间一致段落。在这个层次结构中，上层节点捕捉广泛的故事情节上下文和事件进展，而下层节点则保留细粒度的局部细节和证据承载时刻。这自然产生了两种互补视图：用于覆盖导向推理的全局视图和用于细节导向证据恢复的局部视图。我们进一步引入了一个验证引导路由器，以确定哪个视图更好地得到所选证据的支持，并选择最终答案。我们通过广泛的实验验证了所提设计的有效性，显示验证引导路由器有效协调了全局和局部推理，并且在VideoMME上，我们的方法在LLaVA-Video-7B基础上分别比最先进的帧选择方法提高了2.9分。我们将发布代码。
cs.CV / 26 / 2608.05597
Uncertainty-Aware World Model for Aerial Image-Goal Navigation
面向不确定性的世界模型用于航拍图像目标导航
Abstract
Aerial image-goal navigation requires an unmanned aerial vehicle (UAV) to reach a target location specified by a goal image. Existing world-model-based methods rank candidate trajectories using predicted futures, but typically rely on only one or a few point predictions, which is inadequate for large-scale outdoor environments with substantial future-state uncertainty. To address this limitation, we propose the Uncertainty-Aware Navigation World Model (UA-NWM), an efficient latent world model for aerial image-goal navigation, which formulates trajectory scoring as conditional out-of-distribution detection. UA-NWM represents plausible futures with an uncertainty subspace and decomposes the prediction--goal discrepancy into uncertainty-explainable and unexplainable components. Only the unexplainable residual is used for scoring, enabling robust selection without multiple future samples. Extensive experiments demonstrate that UA-NWM consistently outperforms existing navigation world models while maintaining low inference latency. Real-world UAV experiments further validate its practical applicability. Project page: https://duryi.github.io/UA-NWM-Project-Page
Chinese Translation
航拍图像目标导航要求无人机（UAV）到达由目标图像指定的目标位置。现有的基于世界模型的方法通过预测未来来对候选轨迹进行排名，但通常仅依赖于一个或少数几个点预测，这对于具有显著未来状态不确定性的大规模户外环境来说是不够的。为了解决这一局限性，我们提出了不确定性感知导航世界模型（UA-NWM），这是一种高效的潜在世界模型，用于航拍图像目标导航，它将轨迹评分公式化为条件性分布外检测。UA-NWM通过不确定性子空间表示合理的未来，并将预测与目标之间的差异分解为可解释和不可解释的成分。只有不可解释的残差用于评分，从而实现了在没有多个未来样本的情况下的稳健选择。大量实验表明，UA-NWM在保持低推理延迟的同时，始终优于现有的导航世界模型。实际的无人机实验进一步验证了其实际应用性。项目页面：https://duryi.github.io/UA-NWM-Project-Page
cs.CV / 27 / 2608.05615
ALTER: Modeling Longitudinal Changes via Regional Differencing for 3D CT Report Generation
ALTER：通过区域差分建模纵向变化以生成3D CT报告
Abstract
Computed tomography (CT) is widely used for clinical diagnosis and longitudinal follow-up, yet automatically generating accurate and complete radiology reports from three-dimensional (3D) CT remains challenging. Existing methods improve fine-grained correspondence between images and text by modeling anatomical regions, but remain centered on the current examination. Consequently, patient-specific longitudinal changes within individual regions remain insufficiently modeled. Meanwhile, interval changes are often distributed across multiple anatomical regions, complicating a coherent assessment of the overall longitudinal state. We propose Anatomically Localized Temporal Evidence Representation (ALTER) to address these limitations. Global Prior Integration (GPI) incorporates the prior CT and report to establish historical context for the current examination. Regional Proxy Differencing (RPD) enables each current anatomical region to retrieve a historical proxy from a single shared encoding of the prior volume and to derive localized interval evidence. Interval Change Fusion (ICF) further combines current abnormality states with region-distributed differences, converting their joint representation into change-aware soft prompts that guide report generation. ALTER achieves state-of-the-art results on most evaluation metrics across the RadGenome-ChestCT validation and CTRG-Chest-548K test sets. Code and data preprocessing details are available at https://github.com/peytonkarlie/ALTER/tree/main.
Chinese Translation
计算机断层扫描（CT）广泛应用于临床诊断和纵向随访，但从三维（3D）CT自动生成准确且完整的放射学报告仍然具有挑战性。现有方法通过建模解剖区域来改善图像与文本之间的细粒度对应关系，但仍然集中于当前检查。因此，患者特定的个体区域纵向变化建模不足。同时，间隔变化通常分布在多个解剖区域， complicating the overall longitudinal state assessment. 我们提出了解剖局部时间证据表示（Anatomically Localized Temporal Evidence Representation，ALTER）以解决这些局限性。全球先验整合（Global Prior Integration，GPI）结合先前的CT和报告，为当前检查建立历史背景。区域代理差分（Regional Proxy Differencing，RPD）使每个当前解剖区域能够从先前体积的单一共享编码中检索历史代理，并推导出局部间隔证据。间隔变化融合（Interval Change Fusion，ICF）进一步将当前异常状态与区域分布的差异结合起来，将其联合表示转换为变化感知的软提示，以指导报告生成。ALTER在RadGenome-ChestCT验证集和CTRG-Chest-548K测试集的大多数评估指标上实现了最先进的结果。代码和数据预处理细节可在https://github.com/peytonkarlie/ALTER/tree/main获取。
cs.CV / 28 / 2608.05616
TruthLens: Object Hallucination Detection via Self-Evaluating Truthfulness Scores in LVLMs
TruthLens：通过自评真实度评分在大规模视觉语言模型中检测对象幻觉
Abstract
Despite the remarkable progress of large vision language models (LVLMs), object hallucination remains a fundamental challenge that hinders their trustworthy deployment. A key finding motivates our work: real and hallucinated object tokens are clearly separable in hidden representations, yet this separability is largely lost at the language-modeling (LM) head. We propose TruthLens, a self-evaluation framework that teaches the LM head to expose a per-object truthfulness signal without any auxiliary model or additional inference cost. Concretely, a rarely-used special token is repurposed as a reference token. For each object-token position, we extract the log-probability assigned to this special token by the LM head, and define its difference from a predefined constant as the truthfulness score. The model is then fine-tuned with an MSE objective that drives scores toward 1 for real objects and 0 for hallucinated ones, while a divergence constraint preserves the original generation capability. Despite being trained on only a limited set of object categories, TruthLens generalizes effectively to benchmarks with substantially larger label spaces. Extensive experiments across multiple LVLMs demonstrate state-of-the-art performance; notably, on Qwen2.5-VL-7B, TruthLens outperforms the previous best method on MS-COCO by over 17\% in AUROC. Our code is available at https://github.com/wyqstan/TruthLens.
Chinese Translation
尽管大规模视觉语言模型（LVLMs）取得了显著进展，对象幻觉仍然是阻碍其可信部署的基本挑战。我们的工作受到一个关键发现的启发：真实和幻觉对象标记在隐藏表示中是明显可分的，但这种可分性在语言建模（LM）头部大部分丧失。我们提出了TruthLens，一个自评框架，教会LM头部在没有任何辅助模型或额外推理成本的情况下暴露每个对象的真实度信号。具体而言，一个很少使用的特殊标记被重新用作参考标记。对于每个对象标记位置，我们提取LM头部分配给该特殊标记的对数概率，并将其与预定义常数的差异定义为真实度评分。然后，模型通过均方误差（MSE）目标进行微调，推动真实对象的评分趋向1，而幻觉对象的评分趋向0，同时一个发散约束保持原始生成能力。尽管仅在有限的对象类别上进行训练，TruthLens在具有显著更大标签空间的基准测试中有效泛化。针对多个LVLM的广泛实验表明了其最先进的性能；值得注意的是，在Qwen2.5-VL-7B上，TruthLens在MS-COCO上的表现比之前的最佳方法提高了超过17%的AUROC。我们的代码可在https://github.com/wyqstan/TruthLens获取。
cs.CV / 29 / 2608.05626
Dual-Output Multi-Exposure HDR Reconstruction via SDR Fusion and Gain Map Inverse Tone Mapping
通过SDR融合和增益图逆色调映射的双输出多曝光HDR重建
Abstract
We propose DOME-HDR, a dual-output multi-exposure HDR reconstruction framework that jointly produces a perceptually balanced SDR image and a consistent HDR image via gain map inverse tone mapping. Given three bracketed LDR inputs, DOME-HDR first synthesizes a base SDR using a LoRA-adapted latent diffusion model. A dual cross-attention fusion module injects complementary structural and color cues from the under- and over-exposed images while anchoring on the mid exposure for stability. The synthesized SDR then guides HPGM, our HDR Prior-guided Gain Map network, to predict a spatially varying gain map for reliable dynamic-range expansion. We evaluate on Kalantari, Tel, and Challenge123 using both full-reference and no-reference metrics, where DOME-HDR achieves state-of-the-art HDR reconstruction quality; ablations further confirm the effectiveness of dual cross-attention and SDR-guided gain map estimation.
Chinese Translation
我们提出了DOME-HDR，一个双输出多曝光HDR重建框架，通过增益图逆色调映射共同生成感知平衡的SDR图像和一致的HDR图像。给定三个括号的LDR输入，DOME-HDR首先使用LoRA适配的潜在扩散模型合成基础SDR。一个双重交叉注意力融合模块从欠曝光和过曝光图像中注入互补的结构和颜色线索，同时以中间曝光为基准以确保稳定性。合成的SDR随后引导我们的HDR先验引导增益图网络（HPGM）预测一个空间变化的增益图，以实现可靠的动态范围扩展。我们在Kalantari、Tel和Challenge123上进行评估，使用全参考和无参考指标，其中DOME-HDR实现了最先进的HDR重建质量；消融实验进一步确认了双重交叉注意力和SDR引导增益图估计的有效性。
cs.CV / 30 / 2608.05627
SCI-CLIP: Segment-Centric Inference with Reference Memory for Training-Free Open-Vocabulary Segmentation
SCI-CLIP：基于参考记忆的以段落为中心的推理框架，用于无训练的开放词汇分割
Abstract
Training-free open-vocabulary segmentation remains limited by a missing inference abstraction. Frozen vision-language features are produced at patch level, yet dense prediction requires a unit that simultaneously governs feature interaction, spatial support, contextual recovery, and retrieval-based correction. We present SCI-CLIP, a segment-centric inference framework built around the principle that the same region abstraction should organize all stages of dense open-vocabulary prediction. SCI-CLIP first induces a region-consistent interaction graph over frozen visual tokens, then reconstructs dense features by propagating values over this graph, augmenting them with selective cross-window support only where local evidence is insufficient. The same segment abstraction is subsequently used to construct and query an offline reference memory, aligning exemplar retrieval with the units on which prediction is made. SCI-CLIP turns frozen CLIP-style features into spatially coherent, context-aware, and retrieval-compatible dense predictions without any training. SCI-CLIP consistently improves the structural quality of dense predictions, the robustness of contextual reasoning, and the alignment of exemplar-based correction, yielding stronger open-vocabulary segmentation across eight benchmarks. Project code is available at: https://github.com/mzamini92/SCICLIP.
Chinese Translation
无训练的开放词汇分割仍然受到缺乏推理抽象的限制。冻结的视觉-语言特征是在补丁级别生成的，然而，密集预测需要一个同时管理特征交互、空间支持、上下文恢复和基于检索的校正的单元。我们提出了SCI-CLIP，一个以段落为中心的推理框架，建立在同一区域抽象应组织所有阶段的密集开放词汇预测的原则上。SCI-CLIP首先在冻结的视觉标记上诱导一个区域一致的交互图，然后通过在该图上传播值来重建密集特征，仅在局部证据不足的地方用选择性的跨窗口支持进行增强。随后，使用相同的段落抽象来构建和查询离线参考记忆，使示例检索与进行预测的单元对齐。SCI-CLIP将冻结的CLIP风格特征转化为空间一致、上下文感知和兼容检索的密集预测，而无需任何训练。SCI-CLIP持续改善密集预测的结构质量、上下文推理的鲁棒性以及基于示例的校正的对齐，从而在八个基准测试中实现更强的开放词汇分割。项目代码可在以下地址获取：https://github.com/mzamini92/SCICLIP。
cs.CV / 31 / 2608.05631
ChronoVision: Temporal Reasoning via Latent State Reconstruction
ChronoVision：通过潜在状态重构进行时间推理
Abstract
Multimodal large language models excel at passive perception but struggle with complex visual cognitive tasks requiring multi-step temporal reasoning. This degradation largely stems from the inherent ambiguity of language-based reasoning, which often fails to accurately articulate continuous visual transformations. To address this, we propose ChronoVision, a multimodal framework designed to align visual logic with latent imagery. During supervised fine-tuning, a Reconstructive Visual Head predicts the latent representation of the final transformed state, while an ROI Attention Locating module focuses the model on key visual evidence via semantic span queries. In post-training, we apply reinforcement learning with an implicit process grounding mechanism, guided by a composite reward function that evaluates outcome correctness, latent process alignment, and unsupervised visual focus. Furthermore, we introduce Vbvr-VQA, a novel dataset that evaluates temporal tracking by reformulating video reasoning into a strict image-ordering task. Experiments demonstrate that ChronoVision achieves state-of-the-art performance on Vbvr-VQA with 74.8% in-domain and 71.6% out-of-domain accuracy, alongside a strong 55.0% accuracy on IntPhys2, a highly challenging cross-domain benchmark.
Chinese Translation
多模态大型语言模型在被动感知方面表现出色，但在需要多步骤时间推理的复杂视觉认知任务中却表现不佳。这种退化主要源于基于语言的推理固有的模糊性，常常无法准确表达连续的视觉转变。为了解决这一问题，我们提出了ChronoVision，一个旨在将视觉逻辑与潜在图像对齐的多模态框架。在监督微调过程中，重构视觉头（Reconstructive Visual Head）预测最终变换状态的潜在表示，而ROI注意力定位模块（ROI Attention Locating module）通过语义跨度查询将模型聚焦于关键视觉证据。在后训练阶段，我们应用了带有隐式过程基础机制的强化学习，指导原则是一个复合奖励函数，该函数评估结果的正确性、潜在过程的对齐以及无监督的视觉聚焦。此外，我们引入了Vbvr-VQA，一个新的数据集，通过将视频推理重新构造成严格的图像排序任务来评估时间跟踪。实验表明，ChronoVision在Vbvr-VQA上实现了74.8%的领域内准确率和71.6%的领域外准确率，此外在高度具有挑战性的跨领域基准IntPhys2上也取得了55.0%的强劲准确率。
cs.CV / 32 / 2608.05648
Vorch-IR: Long-Form Unified Multimodal Identity Replacement Video Generation
Vorch-IR：长格式统一多模态身份替换视频生成
Abstract
Video identity replacement seeks to transfer the identities of one or more subjects while preserving the motion, expressions, and temporal structure of a driving video. Existing methods largely target single-person settings and often require task-specific structural controls, such as masks or pose representations, limiting their flexibility in general multimodal editing systems. Progress on multi-person replacement is further constrained by the scarcity of paired training data. We present Vorch-IR, a unified framework that supports single- and dual-person identity replacement, with optional background replacement, in a single model. Built on LTX2, Vorch-IR jointly conditions on a driving video, indexed reference images, and a textual editing instruction. The reference images need not match the pose, layout, or spatial configuration of the driving video: their roles as subject or background references are specified through the instruction. Dense visual conditions are fused through self-attention, while a vision-language context establishes semantic correspondence through cross-attention. We further develop an automatic data construction pipeline that synthesizes paired supervision for all four editing settings. Experiments using automatic metrics and pairwise human evaluation demonstrate strong identity preservation, motion fidelity, and temporal coherence across diverse scenarios. A temporal overlapping inference strategy additionally extends the short-clip model to minute-long generation without autoregressive continuation.
Chinese Translation
视频身份替换旨在转移一个或多个主体的身份，同时保留驱动视频的运动、表情和时间结构。现有方法主要针对单人场景，并且通常需要特定任务的结构控制，如掩膜或姿态表示，这限制了它们在通用多模态编辑系统中的灵活性。多人的替换进展进一步受到配对训练数据稀缺的限制。我们提出了Vorch-IR，一个统一框架，支持单人和双人身份替换，并在单个模型中提供可选的背景替换。Vorch-IR基于LTX2构建，联合条件输入驱动视频、索引参考图像和文本编辑指令。参考图像不必与驱动视频的姿态、布局或空间配置相匹配：它们作为主体或背景参考的角色通过指令进行指定。通过自注意力融合密集视觉条件，而视觉-语言上下文通过交叉注意力建立语义对应。我们进一步开发了一个自动数据构建管道，为所有四种编辑设置合成配对监督。使用自动指标和成对人类评估的实验表明，在多种场景中具有强大的身份保留、运动保真度和时间一致性。此外，时间重叠推理策略进一步将短片模型扩展到分钟级生成，而无需自回归延续。
cs.CV / 33 / 2608.05663
Vorch-Streamer: Extending Human Audio-Visual Generation to Real-Time Long-Form Streaming
Vorch-Streamer：将人类音频-视觉生成扩展至实时长格式流媒体
Abstract
Real-time long-form avatar audio--video generation requires causal, continuous synthesis while maintaining audiovisual synchronization and visual consistency. Adapting a pretrained bidirectional model to this setting presents two key dilemmas. First, autoregressively reusing generated blocks as context creates exposure bias, causing errors and visual drift to accumulate over long rollouts. Second, a global speech utterance does not indicates a causal generator which portion should be spoken next when only limited local audio--video context is available. We present \textbf{Vorch-Streamer}, a post-training framework that addresses these challenges and enables real-time long-form Text-to-Audio-Video (T2AV) streaming. We construct a synthetic corpus of 80K avatar clips spanning 12--21 seconds and first train a causal generator with mixed Teacher Forcing and Diffusion Forcing. We then apply long-horizon Self Forcing with DMD distillation, exposing the model to its own rollout distribution while preserving the quality of the pretrained bidirectional teacher. To explicitly control speech progression, an external language model predicts discrete 25-Hz speech-planning tokens, whose continuous features condition the audio diffusion branch and align each causal block with the content it should speak. With bounded causal context and four-step denoising, Vorch-Streamer jointly generates audio and video from text at 27.12 FPS, exceeding the 24-FPS real-time playback rate while maintaining competitive audio--lip synchronization and strong identity preservation over long-form generation.
Chinese Translation
实时长格式头像音视频生成需要因果、连续的合成，同时保持视听同步和视觉一致性。将预训练的双向模型适应于这一设置面临两个关键难题。首先，自回归地重用生成的块作为上下文会产生曝光偏差，导致错误和视觉漂移在长时间的生成过程中累积。其次，全球语音发声并不能指示因果生成器在仅有有限的局部音视频上下文时应该接下来发声的部分。我们提出了 extbf{Vorch-Streamer}，一个后训练框架，旨在解决这些挑战，并实现实时长格式文本到音视频（Text-to-Audio-Video, T2AV）流媒体生成。我们构建了一个包含8万段头像视频的合成语料库，时长为12至21秒，首先用混合教师强制（Mixed Teacher Forcing）和扩散强制（Diffusion Forcing）训练因果生成器。然后，我们应用长时间范围的自我强制（Self Forcing）与动态模式分馏（DMD distillation），使模型接触到其自身的生成分布，同时保持预训练双向教师的质量。为了明确控制语音进展，外部语言模型预测离散的25赫兹语音规划标记，其连续特征调节音频扩散分支，并将每个因果块与其应发声的内容对齐。通过有限的因果上下文和四步去噪，Vorch-Streamer以27.12帧每秒的速度共同生成音频和视频，超越了24帧每秒的实时播放速率，同时在长格式生成中保持竞争力的音频-唇同步和强身份保留。
cs.CV / 34 / 2608.05664
Dual-Attention and Adversarial Transfer Networks for Sim-to-Real Cross-Orientation Wireless Sensing
双重注意力与对抗转移网络用于模拟到现实的跨方向无线传感
Abstract
Millimeter-wave human activity recognition suffers significant performance degradation when the user's orientation changes relative to the sensing system, yet collecting labeled multi-orientation data is labor-intensive and costly. To eliminate the need for exhaustive multi-orientation measured data, we develop a physics-guided simulator that synthesizes orientation-diverse wireless training data from single-orientation motion. Specifically, to suppress orientation-induced feature variations, we propose a dual-attention network that extracts activity-discriminative and orientation-robust representations from dual-link Doppler spectrograms. To bridge the simulation-to-reality gap, we introduce an adversarial unsupervised transfer learning mechanism that aligns feature distributions using only a small number of unlabeled target-domain samples. The S2M-Sense platform shows high fidelity in reproducing real-world signatures, validated against 60.48 GHz mmWave measured data with an average structural similarity index measure (SSIM) of 0.84 between simulated and measured Doppler spectrograms across all 4 activities and 4 orientations. Experimental results show that S2M-Sense achieves 88.33% recognition accuracy using only the dual-link multi-orientation simulated dataset, which improves to 95% after simulation-to-reality transfer learning with as few as 16 unlabeled measured samples. Both cases with and without transfer learning outperform state-of-the-art cross-domain sensing methods.
Chinese Translation
毫米波人类活动识别在用户相对于传感系统的方向变化时性能显著下降，而收集标记的多方向数据既费力又昂贵。为了消除对全面多方向测量数据的需求，我们开发了一种物理引导的模拟器，该模拟器能够从单一方向的运动中合成方向多样的无线训练数据。具体而言，为了抑制由方向引起的特征变化，我们提出了一种双重注意力网络，从双链多普勒声谱图中提取活动区分性和方向鲁棒性表示。为了弥合模拟与现实之间的差距，我们引入了一种对抗性无监督迁移学习机制，仅使用少量未标记的目标领域样本来对齐特征分布。S2M-Sense平台在重现现实世界特征方面表现出高保真度，与60.48 GHz毫米波测量数据进行验证，模拟与测量的多普勒声谱图在所有4种活动和4种方向下的平均结构相似性指数（SSIM）为0.84。实验结果表明，S2M-Sense在仅使用双链多方向模拟数据集的情况下实现了88.33%的识别准确率，在进行模拟到现实的迁移学习后，准确率提高至95%，只需16个未标记的测量样本。无论是否进行迁移学习，这两种情况均优于最先进的跨领域传感方法。
cs.CV / 35 / 2608.05671
URNet: A Unified Reparameterized Network for Efficient RGB-D Semantic Segmentation
URNet：一种用于高效RGB-D语义分割的统一重参数化网络
Abstract
Previous RGB-D semantic segmentation methods commonly employ dual encoders to separately process RGB and depth inputs, followed by dedicated modules for cross-modal feature fusion. However, such designs often inadequately capture depth representations and consequently limit effective cross-modal interaction, while the additional encoder branch introduces redundant computation that hinders lightweight execution. To tackle these challenges, we propose URNet, a Unified Reparameterized RGB-D Network that performs simultaneous multi-modal feature extraction and cross-modal fusion within a single encoder. Specifically, we adopt a reparameterization strategy to compact the network architecture and facilitate fast inference. Within each Reparameterized Block (RepBlock), a Linear Gated Attention (LGA) module is introduced to fully exploit complementary RGB and depth cues across different feature scales. Furthermore, considering that decoder design has been relatively underexplored in existing RGB-D segmentation models, we develop a concise yet effective universal decoder, termed the Pyramid Merging Decoder (PMD). Extensive experiments on multiple RGB-D segmentation benchmarks demonstrate that URNet achieves state-of-the-art performance while maintaining high efficiency. Code will be available at https://github.com/Wild-Stephen/URNet.
Chinese Translation
以往的RGB-D语义分割方法通常采用双编码器分别处理RGB和深度输入，随后通过专门的模块进行跨模态特征融合。然而，这种设计往往无法充分捕捉深度表示，从而限制了有效的跨模态交互，同时额外的编码器分支引入了冗余计算，妨碍了轻量级执行。为了解决这些挑战，我们提出了URNet，一种统一重参数化的RGB-D网络，能够在单个编码器内同时进行多模态特征提取和跨模态融合。具体而言，我们采用重参数化策略来压缩网络架构并促进快速推理。在每个重参数化块（Reparameterized Block，RepBlock）中，引入了线性门控注意力（Linear Gated Attention，LGA）模块，以充分利用不同特征尺度下的互补RGB和深度线索。此外，考虑到现有RGB-D分割模型中解码器设计相对欠缺，我们开发了一种简洁而有效的通用解码器，称为金字塔合并解码器（Pyramid Merging Decoder，PMD）。在多个RGB-D分割基准上的广泛实验表明，URNet在保持高效性的同时实现了最先进的性能。代码将发布在https://github.com/Wild-Stephen/URNet。
cs.CV / 36 / 2608.05683
DistMedVL: Distributional Vision-Language Alignment for Uncertainty-Aware Medical Image Segmentation
DistMedVL：用于不确定性感知医学图像分割的分布式视觉-语言对齐
Abstract
Cross-modal alignment of visual and textual representations is fundamental to multimodal medical image understanding, yet remains hindered by uncertainty in both modalities under real-world clinical conditions. Existing vision-language segmentation methods rely on deterministic cross-modal matching, which overlooks aleatoric uncertainty from ambiguous boundaries and epistemic uncertainty from limited training data, leading to fragile performance under domain shift. To address this issue, we propose DistMedVL, a probabilistic vision-language framework that introduces a lightweight Probabilistic Cross-Modal Adapter (PCM-Adapter) upon frozen encoders to explicitly model representational uncertainty. Specifically, the PCM-Adapter comprises two sequential modules for progressive probabilistic alignment. We first devise a Mahalanobis Alignment Module (MAM) that models textual tokens as Gaussian distributions and computes patch-text compatibility via Mahalanobis distance, yielding variance-conditioned matching that downweights unreliable feature dimensions. Moreover, we devise a Distribution Flow Module (DFM) that estimates modality-wise confidence parameters and performs vision-guided refinement of textual distributions, accommodating distributional variation across imaging modalities. Extensive experiments across eight medical segmentation benchmarks demonstrate that DistMedVL outperforms state-of-the-art methods with only 6.3M trainable parameters, exhibiting superior data efficiency, perturbation robustness and cross-dataset generalization.
Chinese Translation
视觉和文本表示的跨模态对齐是多模态医学图像理解的基础，但在真实临床条件下，两个模态的不确定性仍然阻碍了这一进程。现有的视觉-语言分割方法依赖于确定性的跨模态匹配，忽视了来自模糊边界的随机不确定性和来自有限训练数据的认知不确定性，导致在领域转移下表现脆弱。为了解决这一问题，我们提出了DistMedVL，一个概率性视觉-语言框架，该框架在冻结编码器的基础上引入了一种轻量级的概率跨模态适配器（PCM-Adapter），以显式建模表示不确定性。具体而言，PCM-Adapter由两个顺序模块组成，用于逐步概率对齐。我们首先设计了一个马哈拉诺比斯对齐模块（MAM），将文本标记建模为高斯分布，并通过马哈拉诺比斯距离计算补丁-文本兼容性，从而实现方差条件匹配，降低不可靠特征维度的权重。此外，我们设计了一个分布流模块（DFM），该模块估计模态特定的置信参数，并对文本分布进行视觉引导的细化，适应成像模态间的分布变化。在八个医学分割基准上的广泛实验表明，DistMedVL以仅6.3M的可训练参数超越了最先进的方法，展现出优越的数据效率、扰动鲁棒性和跨数据集的泛化能力。
cs.CV / 37 / 2608.05691
SciQNet: Two-Stage Multimodal Adaptation for Scientific Image Quality Assessment
SciQNet：用于科学图像质量评估的两阶段多模态适应框架
Abstract
Scientific images are essential for communicating experimental observations, quantitative evidence and conceptual knowledge. Unlike natural images, their quality depends on both visual clarity and scientific informativeness, making assessment challenging. In this work, we present SciQNet, a two-stage multimodal adaptation framework for scientific image quality assessment. The first stage performs domain-adaptive pretraining on scientific document images and the second stage conducts task-specific fine-tuning with joint scoring and understanding supervision. For scoring-oriented supervision, we combine instruction tuning with a Huber loss derived from rating-word logits, while understanding-oriented supervision is formulated as multiple-choice visual question answering. Experiments show that using a 40% stratified subset of the domain-adaptive data gives the best performance among the evaluated pretraining fractions, suggesting that pretraining-data relevance may be as important as pretraining-data scale. The final model achieves an SIQA-S score of 92.21, an SIQA-U score of 47.38 and a combined score of 69.80. This work presents our solution to the ICME 2026 Scientific Image Quality Assessment Challenge, which ranked 2nd in the scoring track.
Chinese Translation
科学图像对于传达实验观察、定量证据和概念知识至关重要。与自然图像不同，科学图像的质量不仅依赖于视觉清晰度，还依赖于科学信息量，这使得评估变得具有挑战性。在本研究中，我们提出了SciQNet，一个用于科学图像质量评估的两阶段多模态适应框架。第一阶段对科学文档图像进行领域自适应预训练，第二阶段则通过联合评分和理解监督进行任务特定的微调。对于以评分为导向的监督，我们结合了指令调优和基于评分词逻辑的Huber损失，而以理解为导向的监督则被表述为多选视觉问答。实验表明，使用40%的分层子集进行领域自适应数据预训练在评估的预训练比例中表现最佳，这表明预训练数据的相关性可能与预训练数据的规模同样重要。最终模型在SIQA-S评分中达到了92.21，SIQA-U评分为47.38，综合评分为69.80。本研究展示了我们在ICME 2026科学图像质量评估挑战中的解决方案，并在评分赛道中获得了第二名。
cs.CV / 38 / 2608.05699
TAU-Bench: From Anomaly Instance Tracking to Fine-Grained Video Anomaly Understanding
TAU-Bench：从异常实例跟踪到细粒度视频异常理解
Abstract
Humans understand anomalous events through a coherent perceptual process in which they identify the focal instance, follow its behavior as the event unfolds, and interpret why it violates the expectations of the surrounding scene. Video anomaly understanding (VAU) seeks to endow models with a similar capability, moving beyond deciding whether a video is anomalous toward explaining how the event develops and why it matters. Although recent vision--language models (VLMs) can generate detailed and plausible anomaly descriptions, their semantic fluency does not ensure that these interpretations remain grounded in the correct anomaly instance over time. Existing benchmarks typically evaluate tracking and semantic understanding through separate protocols, leaving such instance--semantic inconsistency largely unmeasured. We therefore introduce TAU-Bench, a track-centric benchmark for jointly evaluating anomaly instance tracking and fine-grained anomaly understanding. TAU-Bench contains 1,118 videos, 1,454 tracks, and 202,438 pixel-level masks spanning 49 event and 45 scene categories, together with track-centric annotations that connect instance-level identification, event-level understanding, and scene-level reasoning. To build TAU-Bench at scale, we developed an automated data engine integrating anomaly suitability filtering, anomaly instance track construction, hierarchical caption annotation, and human quality control. Evaluations across representative VLM families show that models producing plausible anomaly interpretations may still fail to localize and track the correct instance reliably, revealing a persistent gap between semantic reasoning and visual grounding. These findings therefore highlight instance-grounded evaluation as an important step toward more faithful and reliable VAU systems.
Chinese Translation
人类通过一个连贯的感知过程理解异常事件，在这个过程中，他们识别出焦点实例，跟踪事件展开时的行为，并解释为什么该事件违反了周围场景的预期。视频异常理解（Video Anomaly Understanding, VAU）旨在赋予模型类似的能力，超越简单判断视频是否异常，向解释事件如何发展及其重要性转变。尽管近期的视觉-语言模型（Vision-Language Models, VLMs）能够生成详细且可信的异常描述，但其语义流畅性并不能确保这些解释在时间上始终与正确的异常实例相一致。现有基准通常通过独立的协议评估跟踪和语义理解，导致实例-语义不一致性在很大程度上未被测量。因此，我们引入了TAU-Bench，这是一个以跟踪为中心的基准，用于联合评估异常实例跟踪和细粒度异常理解。TAU-Bench包含1,118个视频、1,454个轨迹和202,438个像素级掩膜，涵盖49个事件类别和45个场景类别，并提供了将实例级识别、事件级理解和场景级推理连接起来的轨迹中心注释。为了大规模构建TAU-Bench，我们开发了一个自动化数据引擎，集成了异常适用性过滤、异常实例轨迹构建、分层标题注释和人工质量控制。对代表性VLM家族的评估显示，生成可信异常解释的模型仍可能无法可靠地定位和跟踪正确的实例，揭示了语义推理与视觉基础之间的持续差距。因此，这些发现强调了实例基础评估作为实现更真实和可靠的VAU系统的重要步骤。
cs.CV / 39 / 2608.05703
StreamArena: Toward Continuous, Interactive, and Long-Horizon Agentic Streaming Video Understanding
StreamArena：迈向连续、互动和长时域的自主视频理解
Abstract
Deploying autonomous multimodal agents in continuous, real-world environments requires them to ingest unbounded audio-visual streams and maintain hour-scale memory. However, current evaluations predominantly rely on brief clips and multiple-choice formats. This design allows minimal baselines that process only the last four frames to match or surpass complex streaming models, while answer options also expose language shortcuts. We introduce StreamArena, a benchmark for hour-scale, interactive streaming video understanding. StreamArena contains 243 full-length videos averaging 88.8 minutes and 3,646 rigorously annotated, open-ended question-answer pairs that evaluate real-time perception, historical retrospection, proactive interaction, and multimodal tool utilization. Evaluation across diverse systems exposes a tension between continuous interaction and long-horizon multimodal comprehension. Methods that retain only recent frames cannot recover distant events, methods that convert past observations into text lose visual evidence, and methods that repeatedly compress visual memory struggle to preserve fine-grained details over time. We address this tension with StreamMind, a two-tier architecture that assigns latency-critical interaction and proactive monitoring to independently scheduled frontend workers, while backend workers asynchronously construct persistent multimodal memory and perform historical recall and external search. StreamMind outperforms existing streaming baselines across all four capabilities and reduces query-to-answer latency by reusing persistent state.
Chinese Translation
在连续的现实环境中部署自主多模态代理需要它们能够处理无限的音视频流并保持小时级别的记忆。然而，目前的评估主要依赖于简短的片段和多项选择格式。这种设计使得仅处理最后四帧的最小基线能够与复杂的流媒体模型相匹配或超越，同时答案选项也暴露了语言的捷径。我们引入了StreamArena，一个用于小时级别互动流媒体视频理解的基准。StreamArena包含243个全长视频，平均时长为88.8分钟，以及3,646对经过严格注释的开放式问答对，评估实时感知、历史回顾、主动互动和多模态工具的使用。对不同系统的评估揭示了连续互动与长时域多模态理解之间的紧张关系。仅保留最近帧的方法无法恢复远期事件，将过去观察转化为文本的方法失去了视觉证据，而反复压缩视觉记忆的方法在时间上难以保持细粒度细节。我们通过StreamMind解决了这一紧张关系，StreamMind是一种两层架构，将延迟关键的互动和主动监控分配给独立调度的前端工作者，而后端工作者则异步构建持久的多模态记忆并执行历史回忆和外部搜索。StreamMind在所有四个能力上超越了现有的流媒体基线，并通过重用持久状态减少了查询到答案的延迟。
cs.CV / 40 / 2608.05704
G$^2$ARD-GS: Geometry-Guided Anchor-Regularized Gaussian Splatting Distillation
G$^2$ARD-GS：几何引导的锚点正则化高斯点云蒸馏
Abstract
Dense colored LiDAR maps provide accurate city-scale geometry, but lifting them into 3D Gaussian Splatting (3DGS) retains millions of primitives, making the resulting models costly to store, transmit, render, and adapt. Aggressive primitive reduction alleviates this burden, but can remove the local surface support needed for stable novel-view synthesis and downstream geometric use. We introduce G$^2$ARD-GS, a geometry-guided distillation method that converts a dense Gaussian prior instantiated either as a training-free point-cloud lift or a trained GS model into a compact, reusable representation. G$^2$ARD-GS progressively consolidates the prior into surface-aware representatives, then recovers appearance on the resulting fixed topology under construction-time anchor constraints, with no primitives added or removed during recovery. Under limited supervision, geometry-aware view selection allocates the available view budget. On MatrixCity, G$^2$ARD-GS achieves the best PSNR, SSIM, and LPIPS across matched $5\times$--$30\times$ compression budgets, outperforming PUP by $3.2$--$6.8$,dB in PSNR. When reused as frozen geometry, the compact model improves off-trajectory appearance adaptation by $3.7$--$4.9$,dB over PUP 3D-GS and preserves image-to-model registration accuracy on Cambridge KingsCollege at $30\times$ compression. Project page: https://patrick1159.github.io/gardGS-page/.
Chinese Translation
密集的彩色激光雷达地图提供了准确的城市规模几何信息，但将其提升为3D高斯点云（3DGS）会保留数百万个原始元素，使得生成的模型在存储、传输、渲染和适应方面成本高昂。激进的原始元素减少可以缓解这一负担，但可能会去除稳定新视图合成和下游几何使用所需的局部表面支撑。我们提出了G$^2$ARD-GS，一种几何引导的蒸馏方法，将密集的高斯先验（可以是无训练的点云提升或经过训练的GS模型）转换为紧凑、可重用的表示。G$^2$ARD-GS逐步将先验整合为表面感知的代表，然后在构建时锚点约束下恢复生成的固定拓扑的外观，在恢复过程中没有添加或移除任何原始元素。在有限的监督下，几何感知的视图选择分配可用的视图预算。在MatrixCity上，G$^2$ARD-GS在匹配的$5 imes$--$30 imes$压缩预算中实现了最佳的PSNR、SSIM和LPIPS，相较于PUP在PSNR上提高了$3.2$--$6.8$ dB。当作为冻结几何体重用时，紧凑模型在偏离轨迹的外观适应上比PUP 3D-GS提高了$3.7$--$4.9$ dB，并在$30 imes$压缩下保持了剑桥国王学院的图像与模型配准精度。项目页面：https://patrick1159.github.io/gardGS-page/
cs.CV / 41 / 2608.05706
LAWM-3D: Learning 3D-Aware Latent Actions from Human Videos for Generalizable Robot World Models
LAWM-3D：从人类视频中学习3D感知潜在动作以构建可泛化的机器人世界模型
Abstract
World models enable agents to perform forward rollout and planning without real-world interaction. However, their application in open-world embodied intelligence remains limited by the high cost of action annotations and the heterogeneity of action spaces across platforms. Recently, latent action models (LAMs) have alleviated this bottleneck by learning action representations directly from unlabeled human videos in a self-supervised manner. Nevertheless, most existing LAMs rely on single-view inputs and operate primarily in 2D pixel space, raising a fundamental question: can simply incorporating multi-view videos into LAM training endow the learned latent actions with 3D-aware perception? Our study shows that the answer is negative. The primary reasons lie in future-frame appearance leakage as well as inter-camera appearance discrepancies and viewpoint variations. To address these issues, we propose LAWM-3D, which introduces three tightly coupled key designs: (1) a multi-view invariant unified action tokenization scheme for learning 3D-aware latent actions; (2) a geometric alignment constraint that anchors intermediate encoder features to a pretrained 3D foundation model, thereby explicitly providing cross-view geometric correspondences; and (3) a non-injective RGB-D joint reconstruction objective that prevents shortcut learning from future-frame appearance information, forcing the LAM to focus supervision on motion cues with geometric significance. Importantly, these components are not simply stacked but are tightly coupled through a unified motivation. Built upon a two-stage paradigm of large-scale human video pretraining followed by robot fine-tuning, extensive experiments demonstrate that the proposed 3D-aware latent actions significantly improve world model performance, achieving SOTA results in generation quality, physical consistency, and generalization ability.
Chinese Translation
世界模型使得智能体能够在没有与现实世界交互的情况下进行前向展开和规划。然而，它们在开放世界具身智能中的应用仍然受到动作标注高成本和不同平台之间动作空间异质性的限制。最近，潜在动作模型（LAMs）通过自监督的方式直接从未标记的人类视频中学习动作表示，从而缓解了这一瓶颈。然而，大多数现有的LAMs依赖于单视角输入，并主要在2D像素空间中操作，这引发了一个基本问题：简单地将多视角视频纳入LAM训练是否能够赋予学习到的潜在动作3D感知能力？我们的研究表明答案是否定的。主要原因在于未来帧外观泄漏以及不同摄像头之间的外观差异和视角变化。为了解决这些问题，我们提出了LAWM-3D，它引入了三个紧密耦合的关键设计：（1）一种多视角不变的统一动作标记方案，用于学习3D感知潜在动作；（2）一种几何对齐约束，将中间编码器特征锚定到预训练的3D基础模型，从而明确提供跨视角的几何对应关系；（3）一种非单射的RGB-D联合重建目标，防止从未来帧外观信息中进行捷径学习，迫使LAM将监督重点放在具有几何意义的运动线索上。重要的是，这些组件并不是简单堆叠，而是通过统一的动机紧密耦合。基于大规模人类视频预训练和机器人微调的两阶段范式，广泛的实验表明，所提出的3D感知潜在动作显著提高了世界模型的性能，在生成质量、物理一致性和泛化能力方面达到了SOTA（最先进技术）结果。
cs.CV / 42 / 2608.05707
One Ranking, Any Budget: Matryoshka Evidence-to-Context Frame Selection for Long-Video Understanding
一种排名，任意预算：用于长视频理解的Matryoshka证据到上下文框架选择
Abstract
Frame selection is essential for applying Large Multimodal Models (LMMs) to long videos due to severe frame redundancy and limited context windows. Since the appropriate frame budget varies with the downstream LMM, reasoning demands, and latency constraints, a practical selector should serve multiple budgets. However, existing methods typically optimize an isolated frame subset for each predefined budget: when the budget changes, previously selected evidence may be replaced rather than progressively augmented. Ranking frames by a fixed score would allow prefix reuse across budgets, but it ignores the distinct roles of different ranking positions. In this paper, we formulate long-video frame selection as a Matryoshka ranking problem: constructing a single priority sequence whose small prefixes concentrate query-conditioned evidence, while progressively larger prefixes preserve this evidence and add broader temporal context. Efficiently constructing such a ranking is itself challenging, as densely sampling long videos and evaluating frame-query relevance incurs substantial overhead. We therefore introduce Matryoshka Evidence-to-Context (MEC) Frame Selection, a training-free framework that builds a reusable sparse video index, discovers candidates through sparse probing and local zooming, and greedily constructs a position-adaptive ranking: early positions emphasize evidence; later positions progressively favor temporal coverage while preserving visual diversity. A single ranking can thus be truncated to any target budget without rerunning the selector. Across four benchmarks and six frame budgets, MEC improves average accuracy over uniform sampling by 3.77 percentage points, matches strong state-of-the-art selectors, and reduces end-to-end selection latency by 47.37-51.19%.
Chinese Translation
由于严重的帧冗余和有限的上下文窗口，框架选择对于将大型多模态模型（LMMs）应用于长视频至关重要。由于适当的帧预算因下游LMM、推理需求和延迟限制而异，因此一个实用的选择器应能够服务于多种预算。然而，现有方法通常为每个预定义预算优化一个孤立的帧子集：当预算变化时，之前选择的证据可能被替换，而不是逐步增加。通过固定得分对帧进行排名可以允许跨预算重用前缀，但它忽略了不同排名位置的独特角色。在本文中，我们将长视频帧选择表述为一个Matryoshka排名问题：构建一个单一的优先序列，其小前缀集中查询条件证据，而逐渐增大的前缀则保留这些证据并添加更广泛的时间上下文。有效构建这样的排名本身就是一个挑战，因为密集采样长视频和评估帧-查询相关性会产生大量开销。因此，我们提出了Matryoshka证据到上下文（MEC）框架选择，这是一种无训练的框架，构建可重用的稀疏视频索引，通过稀疏探测和局部放大发现候选者，并贪婪地构建位置自适应排名：早期位置强调证据；后期位置逐渐偏向时间覆盖，同时保持视觉多样性。因此，单一排名可以被截断到任何目标预算，而无需重新运行选择器。在四个基准和六个帧预算上，MEC的平均准确性比均匀采样提高了3.77个百分点，匹配了强大的最先进选择器，并将端到端选择延迟减少了47.37-51.19%。
cs.CV / 43 / 2608.05718
Iterative Hybrid Discrete-Continuous Viewpoint Planning for UAV Photogrammetry
无人机摄影测量的迭代混合离散-连续视点规划
Abstract
Unmanned aerial vehicle (UAV) photogrammetry requires camera networks that provide sufficient surface coverage, image overlap, parallax, and resolution, yet conventional flight patterns are often poorly adapted to scene geometry resulting in local reconstruction errors. This paper proposes an iterative hybrid discrete-continuous viewpoint planning method for targeted UAV photogrammetry from a proxy reconstruction. The method scores sampled surface points using photogrammetric heuristics based on frontality, imaging distance, parallax, and multi-view observation count, while also evaluating the full viewpoint set in terms of visibility, pairwise overlap, and graph connectivity. Candidate viewpoints are generated around weakly observed regions, refined using clustered Covariance matrix adaptation evolution strategy (CMA-ES) optimisation, and removed when redundant. The final flight path combines close-range detail viewpoints with wider model-coverage viewpoints, balancing local reconstruction quality with global image-network robustness. Evaluation on three synthetic scenes shows that the proposed method improves both reconstruction accuracy and completeness compared with prior UAV path-planning methods.
Chinese Translation
无人机（UAV）摄影测量需要提供足够表面覆盖、图像重叠、视差和分辨率的相机网络，但传统飞行模式往往与场景几何形状不匹配，导致局部重建误差。本文提出了一种针对目标无人机摄影测量的迭代混合离散-连续视点规划方法，该方法基于代理重建进行。该方法使用基于前向性、成像距离、视差和多视图观察计数的摄影测量启发式对采样表面点进行评分，同时还评估了完整视点集在可见性、成对重叠和图连接性方面的表现。候选视点围绕弱观测区域生成，使用聚类协方差矩阵适应进化策略（CMA-ES）优化进行细化，并在冗余时予以移除。最终的飞行路径结合了近距离细节视点与更广泛模型覆盖的视点，平衡了局部重建质量与全局图像网络的鲁棒性。在三个合成场景上的评估表明，所提出的方法相比于先前的无人机路径规划方法，提高了重建的准确性和完整性。
cs.CV / 44 / 2608.05720
PhyLatent: Learning Dynamics-Relevant Representations for JEPA World Models
PhyLatent：为JEPA世界模型学习与动态相关的表征
Abstract
We propose PhyLatent, a dynamics-relevant training objective for JointEmbedding Predictive Architecture (JEPA) world models. Our key observation is that preventing global latent collapse does not ensure that a representation preserves physical states and action consequences. We identify three failure modes in JEPA world models: physical invariance collapse, physical identifiability collapse, and counterfactual dynamics collapse. PhyLatent addresses them through three training pathways: physical invariance, physical identifiability, and counterfactual dynamics, implemented with physical state grounding, future representation alignment, static visual invariance, counterfactual branch separation, and latent denoising. On OGBench-Cube, PhyLatent reduces the three failure rates from 15.60%, 6.71%, and 8.41% to 7.53%, 0.95%, and 4.62%, respectively, and improves model predictive control (MPC) success from 70.0% to 78.1%. With the same architecture and planner, it further improves success from 81.0% to 98.0% on TwoRooms and remains competitive on Reacher and PushT. These results show that global non-collapse alone is insufficient for learning a reliable JEPA worldmodel state space.
Chinese Translation
我们提出了PhyLatent，一种针对联合嵌入预测架构（Joint Embedding Predictive Architecture，JEPA）世界模型的与动态相关的训练目标。我们的关键观察是，防止全局潜在崩溃并不能确保表征保留物理状态和行动后果。我们在JEPA世界模型中识别了三种失败模式：物理不变性崩溃、物理可识别性崩溃和反事实动态崩溃。PhyLatent通过三条训练路径解决这些问题：物理不变性、物理可识别性和反事实动态，具体实现包括物理状态基础、未来表征对齐、静态视觉不变性、反事实分支分离和潜在去噪。在OGBench-Cube上，PhyLatent将三种失败率分别从15.60%、6.71%和8.41%降低到7.53%、0.95%和4.62%，并将模型预测控制（Model Predictive Control，MPC）的成功率从70.0%提高到78.1%。在相同的架构和规划器下，它在TwoRooms上的成功率进一步提高至81.0%至98.0%，并在Reacher和PushT上保持竞争力。这些结果表明，仅仅实现全局非崩溃不足以学习一个可靠的JEPA世界模型状态空间。
cs.CV / 45 / 2608.05728
Engram-E2VID: Reference-Based Event-to-Video Reconstruction via Generative Activation of Appearance Engrams
Engram-E2VID：基于参考的事件到视频重建通过生成激活外观铭刻
Abstract
Reference-based event-to-video reconstruction aims to recover target RGB frames from a reference frame and the event stream captured over the reference-to-target interval. Although events provide fine-grained temporal cues, they encode sparse and asynchronous log-intensity changes rather than absolute appearance, making faithful reconstruction intrinsically challenging. The central challenge lies in associating event-derived target-time structures with relevant appearance information from the reference frame, especially under complex motion and long temporal intervals. In this work, we propose Engram-E2VID, a structure-guided framework that reconstructs target frames through the generative activation of appearance engrams. Specifically, the reference frame is encoded into token-space appearance engrams, while the event stream and reference context are transformed into a target-time motion-structure scaffold that captures motion boundaries and event-induced structural changes. Within a one-step diffusion backbone, scaffold-derived structural tokens progressively interact with and activate relevant appearance engrams across layers. This token-space association allows target structures to access reference appearance without relying on direct pixel-wise correspondence, while the diffusion prior complements uncertain or newly revealed regions. Across three benchmarks, Engram-E2VID improves PSNR by up to 3.29 dB and reduces LPIPS by up to 0.08 over the strongest same-input baseline, while degrading more slowly as the reconstruction interval increases.
Chinese Translation
基于参考的事件到视频重建旨在从参考帧和在参考到目标间隔内捕获的事件流中恢复目标RGB帧。尽管事件提供了细粒度的时间线索，但它们编码的是稀疏和异步的对数强度变化，而非绝对外观，这使得忠实重建在本质上具有挑战性。主要挑战在于将事件衍生的目标时间结构与来自参考帧的相关外观信息关联起来，尤其是在复杂运动和长时间间隔下。在本研究中，我们提出了Engram-E2VID，一个结构引导的框架，通过生成激活外观铭刻来重建目标帧。具体而言，参考帧被编码为标记空间外观铭刻，而事件流和参考上下文被转化为目标时间运动结构支架，以捕捉运动边界和事件引起的结构变化。在一个一步扩散主干中，支架衍生的结构标记逐层逐步相互作用并激活相关的外观铭刻。这种标记空间关联使目标结构能够在不依赖于直接像素级对应的情况下访问参考外观，同时扩散先验补充了不确定或新揭示的区域。在三个基准测试中，Engram-E2VID将PSNR提高了最多3.29 dB，并将LPIPS降低了最多0.08，相较于最强的同输入基线，且随着重建间隔的增加，其降级速度更慢。
cs.CV / 46 / 2608.05739
LiteKD-Net: Lightweight Knowledge-Distilled Network for Mobile Image Denoising
LiteKD-Net：用于移动图像去噪的轻量级知识蒸馏网络
Abstract
Mobile image denoising requires both good restoration quality and low computational cost. In addition, it's annoying to collect large-scale LQ-GT clean pairs. As a result, we propose LiteKD-Net, a lightweight knowledge-distilled network for mobile image denoising. First, a physics-guided noise simulation pipeline generates paired training data by adding pixel crosstalk compared with pipelines applied to cameras. Next, we adapt the Real-ESRGAN to identity-resolution denoising and construct a lightweight Student using Lite-RRDB blocks based on depthwise separable convolutions. Third, feature-level knowledge distillation is applied to transfer the Teacher's restoration capability to the Student without introducing additional inference cost. Experiments on real-world datasets show that our model reaches great reduction in runtime and increase in the inference rate with good restoration quality. Our model also reaches the best in all metrics compared with SwinIR. These results indicate that LiteKD-Net provides a great trade-off between restoration quality and computational efficiency.
Chinese Translation
移动图像去噪需要良好的恢复质量和低计算成本。此外，收集大规模的低质量-高质量（LQ-GT）干净图像对是非常麻烦的。因此，我们提出了LiteKD-Net，一种用于移动图像去噪的轻量级知识蒸馏网络。首先，基于物理引导的噪声模拟管道通过添加像素串扰生成配对训练数据，这与应用于相机的管道相比。接下来，我们将Real-ESRGAN适配为身份分辨率去噪，并使用基于深度可分离卷积的Lite-RRDB块构建轻量级学生网络。第三，应用特征级知识蒸馏将教师网络的恢复能力转移到学生网络，而不引入额外的推理成本。在真实世界数据集上的实验表明，我们的模型在运行时间上大幅减少，推理速率显著提高，同时保持良好的恢复质量。与SwinIR相比，我们的模型在所有指标上均表现最佳。这些结果表明，LiteKD-Net在恢复质量和计算效率之间提供了良好的权衡。
cs.CV / 47 / 2608.05743
ConceptADapt: Concept-guided Adaptive Feature Reconstruction with Dynamic Attention for Few-Shot Industrial Anomaly Detection
ConceptADapt：基于概念指导的动态注意力自适应特征重构用于少样本工业异常检测
Abstract
Few-shot industrial anomaly detection (FS-IAD) focuses on detecting and localizing visual defects in industrial inspection during the cold-start phase, where only a limited number of normal training samples are available per category. Recent advances in this field predominantly leverage visual features from foundation-model and have achieved promising performance. Despite the strong representational power of foundation-model features, the model generalization remains fragile due to the extreme scarcity of normal training data.To address this pivotal issue, we propose ConceptADapt, a concept-guided adaptive feature reconstruction model with dynamic attention. Specifically, our model pre-learns a set of fixed normal concepts from the limited support features and leverages them to mine relationships with query features, thereby recalibrating their statistics for improved anomaly detection at test time. To mitigate the prevalent feature shortcut problem, which is particularly severe under low-data regimes, we further develop a dynamic attention mechanism integrated with sparse autoencoders to learn robust normal concepts during training. Moreover, to enable fast adaptation during inference, our model remains lightweight by incorporating LoRA into the attention module, which introduces only minimal updating parameters.Extensive experiments on three widely adopted FS-IAD benchmarks, including MVTec-AD, VisA, and MPDD, demonstrate that our model consistently outperforms state-of-the-art (SOTA) approaches across both detection and localization tasks, achieving significant improvements under various shot settings.
Chinese Translation
少样本工业异常检测（FS-IAD）关注于在冷启动阶段检测和定位工业检测中的视觉缺陷，此时每个类别仅可用有限数量的正常训练样本。该领域的最新进展主要利用基础模型的视觉特征，并取得了良好的性能。尽管基础模型特征具有强大的表征能力，但由于正常训练数据的极度稀缺，模型的泛化能力仍然脆弱。为了解决这一关键问题，我们提出了ConceptADapt，一种基于概念指导的动态注意力自适应特征重构模型。具体而言，我们的模型从有限的支持特征中预学习一组固定的正常概念，并利用这些概念挖掘与查询特征之间的关系，从而重新校准其统计特性，以提高测试时的异常检测效果。为了缓解在低数据环境下普遍存在的特征捷径问题，我们进一步开发了一种与稀疏自编码器集成的动态注意力机制，以在训练过程中学习稳健的正常概念。此外，为了在推理过程中实现快速适应，我们的模型通过将LoRA引入注意力模块保持轻量化，仅引入最小的更新参数。在三个广泛采用的FS-IAD基准上进行的广泛实验，包括MVTec-AD、VisA和MPDD，证明我们的模型在检测和定位任务中始终优于最先进的方法（SOTA），在各种样本设置下实现了显著的改进。
cs.CV / 48 / 2608.05745
UniVVT: A Unified End-to-End Framework for High-Fidelity Video Virtual Try-on
UniVVT：一个统一的端到端高保真视频虚拟试穿框架
Abstract
Video Virtual Try-On (VVT) synthesizes a video of a person wearing a target garment while preserving identity, motion, and scene dynamics. Dominant approaches cast VVT as mask-conditioned video inpainting and rely on separate modules for human parsing, pose estimation, and garment warping. This multi-stage design complicates deployment and, more critically, allows errors in explicit geometric priors to propagate irreversibly into the generated video. We present UniVVT, a unified end-to-end framework that reframes VVT as semantically conditioned video generation, eliminating mask, pose, and warping modules at inference. At its core, a scene-task perceiver built on a Multimodal Large Language Model jointly encodes the source video, target garment, and task instruction into compact, task-aware latent tokens, implicitly capturing what to transfer and where and how to transfer it. A lightweight semantic bridge then aligns these tokens with the conditioning space of a diffusion-based video generator, enabling coherent garment transfer. To robustly couple the heterogeneous components, we devise a three-stage progressive training strategy comprising semantic alignment, joint task adaptation, and flexible-resolution refinement. Extensive experiments demonstrate that UniVVT achieves state-of-the-art performance across multiple benchmarks, validating implicit semantic guidance as a simple and effective alternative to fragile geometric preprocessing for end-to-end virtual try-on.
Chinese Translation
视频虚拟试穿（Video Virtual Try-On, VVT）合成一个人穿着目标服装的视频，同时保持身份、动作和场景动态。主流方法将VVT视为基于掩码的条件视频修复，并依赖于人类解析、姿态估计和服装变形等独立模块。这种多阶段设计使得部署变得复杂，更重要的是，允许显式几何先验中的错误不可逆地传播到生成的视频中。我们提出了UniVVT，一个统一的端到端框架，将VVT重新定义为语义条件的视频生成，在推理时消除了掩码、姿态和变形模块。其核心是一个基于多模态大语言模型的场景任务感知器，它将源视频、目标服装和任务指令共同编码为紧凑的、任务感知的潜在标记，隐式捕捉要转移的内容及其转移的位置和方式。然后，一个轻量级的语义桥将这些标记与基于扩散的视频生成器的条件空间对齐，从而实现连贯的服装转移。为了稳健地耦合异构组件，我们设计了一个三阶段的渐进训练策略，包括语义对齐、联合任务适应和灵活分辨率细化。大量实验表明，UniVVT在多个基准测试中实现了最先进的性能，验证了隐式语义引导作为端到端虚拟试穿中脆弱几何预处理的简单有效替代方案。
cs.CV / 49 / 2608.05747
GST-Bench: Can VLMs Develop Global Spatial Awareness from Video?
GST-Bench：视觉语言模型能否从视频中发展全球空间意识？
Abstract
Spatial intelligence is fundamental to embodied agents, yet existing benchmarks focus on local spatial perception from single or few viewpoints, overlooking global spatial awareness over continuous, long-horizon visual streams. To address this limitation, we introduce the Global-Spatial-Temporal Benchmark (GST-Bench), a VQA benchmark for global spatial intelligence in video understanding, comprising human-verified questions derived from 6,790 minutes of synthetically generated video. It requires models to perform accurate spatial inference from novel viewpoints unseen in the input video and to map egocentric observations onto global top-down images. A comprehensive evaluation of 22 state-of-the-art VLMs exposes a striking gap between models and humans: the strongest zero-shot model attains only 42.68, far below the human score of 79.08. To probe the cause of this gap, we construct GST-Bench-Local and find that models, despite strong local spatial understanding under the same task formulation, still fail to consolidate long-horizon observations into a globally consistent scene representation. We further provide GST-Train, a dataset for global spatial reasoning, as a complementary resource to facilitate future research on this challenge.
Chinese Translation
空间智能是具身智能体的基础，但现有基准测试主要集中在从单一或少数视角进行局部空间感知，忽视了在连续、长时间视觉流中对全球空间意识的考量。为了解决这一局限性，我们提出了全球空间时间基准（GST-Bench），这是一个用于视频理解的全球空间智能视觉问答（VQA）基准，包含来自6790分钟合成生成视频的人类验证问题。该基准要求模型从输入视频中未见的新视角进行准确的空间推理，并将自我中心的观察映射到全球自上而下的图像上。对22个最先进的视觉语言模型（VLMs）进行的全面评估揭示了模型与人类之间显著的差距：最强的零样本模型仅达到42.68，远低于人类得分79.08。为了探究这一差距的原因，我们构建了GST-Bench-Local，发现尽管模型在相同任务表述下具有强大的局部空间理解能力，但仍未能将长时间观察整合为全球一致的场景表示。我们进一步提供了GST-Train，一个用于全球空间推理的数据集，作为促进未来在这一挑战上研究的补充资源。
cs.CV / 50 / 2608.05757
Beyond Relevance: Bayesian Evidence Acquisition for Agentic Whole-Slide Image Reasoning
超越相关性：用于自主全切片图像推理的贝叶斯证据获取
Abstract
Whole-slide image (WSI) reasoning requires an agent to sequentially acquire visual evidence before answering a diagnostic question. Existing training-free agentic frameworks formulate this process as iterative patch retrieval based on semantic relevance to the question. However, semantic relevance does not necessarily imply diagnostic informativeness in computational pathology, where competing diagnoses often exhibit similar and overlapping morphological patterns, making many patches semantically relevant yet diagnostically non-discriminative. Consequently, relevance-based retrieval may acquire redundant observations and leave diagnostic uncertainty unresolved. We propose BEACON, a plug-and-play agentic framework that reformulates WSI reasoning as a Bayesian evidence acquisition problem. BEACON maintains a probabilistic belief over competing diagnostic hypotheses and sequentially acquires patches by maximizing expected information gain (EIG) to reduce diagnostic uncertainty. An evidence controller then determines whether to answer, acquire additional evidence, or perform higher-resolution inspection. Built entirely from off-the-shelf foundation models, BEACON requires no additional training or fine-tuning. Extensive zero-shot experiments across five WSI-VQA benchmarks demonstrate that BEACON achieves the strongest overall performance among training-free agentic frameworks while substantially improving evidence acquisition efficiency, establishing Bayesian evidence acquisition as a principled paradigm for uncertainty-aware agentic WSI reasoning. The code is available at https://github.com/bryanwong17/BEACON
Chinese Translation
全切片图像（WSI）推理要求代理在回答诊断问题之前，顺序地获取视觉证据。现有的无训练自主框架将这一过程表述为基于与问题的语义相关性的迭代补丁检索。然而，语义相关性并不一定意味着在计算病理学中的诊断信息性，因为竞争性诊断通常表现出相似和重叠的形态模式，导致许多补丁在语义上相关但在诊断上不具辨别力。因此，基于相关性的检索可能会获取冗余观察，并使诊断不确定性未得到解决。我们提出了BEACON，一个即插即用的自主框架，将WSI推理重新表述为一个贝叶斯证据获取问题。BEACON在竞争的诊断假设上保持一个概率信念，并通过最大化期望信息增益（EIG）顺序地获取补丁，以减少诊断不确定性。然后，证据控制器决定是回答、获取额外证据，还是进行更高分辨率的检查。BEACON完全基于现成的基础模型构建，无需额外的训练或微调。在五个WSI-VQA基准上的广泛零样本实验表明，BEACON在无训练自主框架中实现了最强的整体性能，同时显著提高了证据获取效率，确立了贝叶斯证据获取作为一种原则性范式，用于关注不确定性的自主WSI推理。代码可在 https://github.com/bryanwong17/BEACON 获取。
cs.CV / 51 / 2608.05769
Flow-Map Distillation on Relation Manifolds for Image Restoration
基于关系流形的图像恢复流图蒸馏
Abstract
Knowledge distillation for image restoration typically aligns intermediate features or relation matrices between teacher and student networks as static targets, ignoring the dynamic structure of the knowledge transfer process. In this paper, we propose Flow-Map Distillation on Relation Manifolds (FoRM), which reformulates relation-based knowledge transfer as a continuous flow mapping problem on the relation manifold. Rather than regressing a constant velocity field between student and teacher relation states, FoRM learns a flow map operator $\mathcal{F}_\theta(\mathbf{z}, t, s)$ that directly predicts the relation state at any target time $s$ given the current state at time $t$, enabling richer trajectory-level supervision. To ensure global self-consistency of the learned flow map, we introduce a safe semigroup consistency constraint that enforces compositional agreement using ground-truth bridge states, eliminating phantom-state error accumulation. An endpoint anchoring loss further prevents the operator from drifting away from the teacher target. Extensive experiments on five image restoration tasks, including super-resolution, deraining, denoising, deblurring, and low-light enhancement, demonstrate consistent gains over state-of-the-art distillation baselines across multiple backbone architectures, reducing training variance by approximately 50\% compared to naive flow matching distillation while achieving superior restoration quality.
Chinese Translation
图像恢复中的知识蒸馏通常将教师网络和学生网络之间的中间特征或关系矩阵作为静态目标进行对齐，忽视了知识转移过程的动态结构。本文提出了一种基于关系流形的流图蒸馏方法（Flow-Map Distillation on Relation Manifolds, FoRM），将基于关系的知识转移重新表述为关系流形上的连续流映射问题。FoRM并不是回归学生和教师关系状态之间的恒定速度场，而是学习一个流图算子 $ ext{F}_ heta( extbf{z}, t, s)$，该算子能够直接预测在给定当前状态时间 $t$ 的情况下，任何目标时间 $s$ 的关系状态，从而实现更丰富的轨迹级监督。为了确保学习到的流图的全局自一致性，我们引入了一种安全半群一致性约束，该约束通过使用真实的桥接状态来强制实现组合一致性，消除虚假状态的误差累积。此外，端点锚定损失进一步防止算子偏离教师目标。在五个图像恢复任务（包括超分辨率、去雨、去噪、去模糊和低光增强）上的大量实验表明，与多种主干架构的最先进蒸馏基线相比，FoRM在恢复质量上实现了一致的提升，同时将训练方差降低了约50"，相较于简单的流匹配蒸馏方法。
cs.CV / 52 / 2608.05771
HyTBE: Hyperbolic Target-Background Expert Model for Cross-Domain Infrared Small Target Detection
HyTBE：用于跨域红外小目标检测的双曲目标-背景专家模型
Abstract
Infrared small target detection (IRSTD) has achieved substantial progress under domain-consistent evaluation, yet detector performance often degrades markedly when generalizing to unseen infrared domains. Existing methods primarily improve detection by enhancing target responses and suppressing background interference. However, when trained on only a limited set of source domains, their learned decision rules are inevitably established from a restricted range of source-domain target-background relation patterns. We formulate this cross-domain failure as target-background relation shift: unseen domains may exhibit relation patterns that are not observed during training, thereby weakening the discriminative capability learned from the source domains. To address this problem, we propose HyTBE, a Hyperbolic Target-Background Expert model that expands source-domain relation patterns and adaptively adjusts visual representations using explicit relation cues. The Target-Background Relation Intervention selectively perturbs either targets or backgrounds, broadening the observable relation patterns during training while maintaining valid supervision. Subsequently, the Hyperbolic Relation Modeling maps multi-scale visual cues into a Poincar\'e ball and characterizes the target-background relation of each feature token according to its relative distances to the target and background anchors. The Hyperbolic-guided MoE Adapter further uses these hyperbolic relation representations to calibrate multi-scale visual features and aggregate expert-specific feature corrections for different relation patterns. Leave-one-domain-out experiments on NUAA-SIRST, NUDT-SIRST, and IRSTD-1K demonstrate that HyTBE achieves stronger cross-domain generalization than competitive baselines.
Chinese Translation
红外小目标检测（IRSTD）在领域一致的评估下取得了显著进展，但在推广到未见过的红外领域时，检测器的性能往往显著下降。现有方法主要通过增强目标响应和抑制背景干扰来提高检测效果。然而，当仅在有限的源领域上进行训练时，其学习到的决策规则不可避免地是基于有限的源领域目标-背景关系模式建立的。我们将这种跨域失败表述为目标-背景关系转移：未见领域可能表现出在训练期间未观察到的关系模式，从而削弱了从源领域学习到的区分能力。为了解决这个问题，我们提出了HyTBE，一个双曲目标-背景专家模型，它扩展了源领域关系模式，并利用显式关系线索自适应调整视觉表征。目标-背景关系干预选择性地扰动目标或背景，在训练期间扩大可观察的关系模式，同时保持有效的监督。随后，双曲关系建模将多尺度视觉线索映射到庞加莱球，并根据每个特征标记相对于目标和背景锚点的相对距离来表征目标-背景关系。双曲引导的MoE适配器进一步利用这些双曲关系表征来校准多尺度视觉特征，并为不同关系模式聚合专家特定的特征修正。在NUAA-SIRST、NUDT-SIRST和IRSTD-1K上的留一域实验表明，HyTBE在跨域泛化方面优于竞争基线。
cs.CV / 53 / 2608.05774
SR-JEPA: Learning Predictive Latent State in 3D Scenes
SR-JEPA：在3D场景中学习预测潜在状态
Abstract
Joint-embedding predictive architectures learn by predicting latent representations of missing observations, yet many masked JEPAs are evaluated primarily through the encoders they produce. We ask what a trained predictive pathway itself infers when an entire entity is absent from a native 3D scene. We introduce SR-JEPA, a point-native JEPA for scene-scale point clouds whose original frozen predictive pathway can be queried at a supplied location. At evaluation, every point of one object is removed before encoding and replaced by the same shape-free 32-point query at its centroid. Training uses only self-contained 3D EMA targets: no reconstruction, semantic labels, language, or lifted 2D features. On 5,953 held-out ARKitScenes objects, the imputed latent reaches 43.13% semantic-identity macro accuracy, 22.18 points above the strongest floor. Randomizing the prediction path removes 9.78 points, while substituting matched donor context removes 21.98 points. On 8,570 Sr3D support pairs, the full latent reaches 41.15 AP; identity decoded from the missing-object latent, combined with anchor identity and geometry, reaches 39.37 AP, leaving an unresolved 1.78-point residual. These results reveal a queryable, compositional 3D predictive state: the model completes context-dependent entity content, which downstream computation combines with metric geometry.
Chinese Translation
联合嵌入预测架构通过预测缺失观察的潜在表示进行学习，但许多被遮蔽的JEPAs主要通过它们生成的编码器进行评估。我们探讨当整个实体在原生3D场景中缺失时，经过训练的预测路径本身推断了什么。我们引入SR-JEPA，一种针对场景规模点云的点原生JEPA，其原始冻结的预测路径可以在指定位置进行查询。在评估时，某一对象的每个点在编码之前被移除，并在其质心处由相同形状无关的32点查询替代。训练仅使用自包含的3D EMA目标：不进行重构、语义标签、语言或提升的2D特征。在5,953个保留的ARKitScenes对象上，插补的潜在表示达到了43.13%的语义身份宏观准确率，比最强的基线高出22.18分。随机化预测路径减少了9.78分，而替换匹配的捐赠上下文则减少了21.98分。在8,570个Sr3D支持对中，完整的潜在表示达到了41.15 AP；从缺失对象的潜在表示解码的身份，结合锚点身份和几何信息，达到了39.37 AP，留下了未解决的1.78分残差。这些结果揭示了一种可查询的、组合的3D预测状态：模型完成了上下文依赖的实体内容，后续计算将其与度量几何结合。
cs.CV / 54 / 2608.05776
Vorch-Director: Interactive World Story Model via Noise-Aware Error Rectification
Vorch-Director：通过噪声感知误差修正实现交互式世界故事模型
Abstract
Autoregressive continuation provides a natural path toward minute-scale audio-visual generation by repeatedly extending a short-window generator conditioned on previously generated video and audio. However, models are trained on clean ground-truth histories, while inference relies on their own generated histories, where accumulated errors cause identity drift, over-smoothing, and audio-visual desynchronization. Recent methods reduce this mismatch by reusing prediction residuals as synthetic corruption, but we observe that the effectiveness of residual correction critically depends on the flow-matching noise level at which residuals are produced. We propose Vorch-Director, a noise-level-aware residual correction strategy that associates each residual with its originating noise level and injects residuals from matched noise regimes during training. By aligning injected errors with the denoising process, Vorch-Director produces more realistic autoregressive histories while retaining efficient teacher-forcing training. Built on the audio-visual LTX-2 diffusion transformer, Vorch-Director further introduces task embeddings to distinguish historical video, reference images, and target video, enabling unified conditioning for long-horizon generation. Together with a clean conditioning sink and mixed-task training, Vorch-Director supports multi-shot, multi-subject, reference-guided audio-visual long-video generation. We evaluate Vorch-Director on ST-Bench and introduce a new long-horizon audio-visual benchmark with metrics for quality drift and long-range consistency. Extensive experiments demonstrate improved stability and audio-visual fidelity over strong baselines.
Chinese Translation
自回归延续为微尺度音视频生成提供了一条自然路径，通过重复扩展一个短窗口生成器，该生成器以先前生成的视频和音频为条件。然而，模型是在干净的真实历史上训练的，而推理依赖于自身生成的历史，在此过程中积累的错误会导致身份漂移、过度平滑和音视频不同步。最近的方法通过重用预测残差作为合成干扰来减少这种不匹配，但我们观察到，残差修正的有效性在很大程度上取决于产生残差时的流匹配噪声水平。我们提出了Vorch-Director，一种噪声水平感知的残差修正策略，它将每个残差与其来源噪声水平关联，并在训练期间从匹配的噪声状态注入残差。通过将注入的错误与去噪过程对齐，Vorch-Director在保持高效教师强制训练的同时，生成更真实的自回归历史。基于音视频LTX-2扩散变换器，Vorch-Director进一步引入任务嵌入，以区分历史视频、参考图像和目标视频，从而实现长时间生成的统一条件。结合干净的条件接收器和混合任务训练，Vorch-Director支持多镜头、多主体、参考引导的音视频长视频生成。我们在ST-Bench上评估Vorch-Director，并引入一个新的长时间音视频基准，具有质量漂移和长距离一致性的指标。大量实验表明，相较于强基线，Vorch-Director在稳定性和音视频保真度上有所改善。
cs.CV / 55 / 2608.05780
Evidence-Driven Dynamic Visual Selector for Efficient Long Video Understanding
基于证据驱动的动态视觉选择器用于高效的长视频理解
Abstract
Recent advancements in MLLM-based long-form video understanding have mitigated inference-time computational cost and limited context lengths by selecting query-relevant frames. However, existing approaches predominantly rely on external proxy scorers and rigid heuristic rules, inevitably suffering from misalignment with the target MLLM's intrinsic evidence and failing to accommodate the non-uniform spatiotemporal information density. In this paper, we propose a fine-grained dynamic visual selection framework named EviSelect, grounded in the target MLLM internal attention evidence. Our method efficiently probes visual evidence via sparse prefilling as a structured prior to guide distribution-aware dynamic sampling. Specifically, we efficiently approximate attention maps of the target MLLM using highly compressed visual inputs and sparse attention, well-aligned to the full counterpart. Conditioned on three complementary attention components derived from this prior, we design a lightweight selector that not only precisely locates query-relevant timestamps but also adaptively adjusts the local sampling rate and spatial resolution. To enable evidence-conditioned spatiotemporal sampling, we formulate the selector as a stochastic policy and optimize it via GRPO under a joint accuracy--efficiency reward. By rewarding correct predictions under lower visual cost through group-relative comparisons, our method encourages the policy to allocate computation dynamically according to the information density of each video. Across three long video understanding benchmarks, EviSelect achieves superior performance compared to existing methods while reducing selected visual tokens by about 50\% and achieving a 3.9x end-to-end speedup.
Chinese Translation
近期基于多模态大语言模型（MLLM）的长视频理解的进展，通过选择与查询相关的帧，减轻了推理时的计算成本和有限的上下文长度。然而，现有的方法主要依赖外部代理评分器和僵化的启发式规则，难免与目标MLLM的内在证据产生不一致，并未能适应非均匀的时空信息密度。本文提出了一种基于目标MLLM内部注意力证据的细粒度动态视觉选择框架，命名为EviSelect。我们的方法通过稀疏预填充作为结构化先验，有效探测视觉证据，以引导分布感知的动态采样。具体而言，我们使用高度压缩的视觉输入和稀疏注意力高效近似目标MLLM的注意力图，与完整的对应关系良好对齐。在此先验的基础上，我们设计了一个轻量级选择器，该选择器不仅能够精确定位与查询相关的时间戳，还能够自适应调整局部采样率和空间分辨率。为了实现基于证据的时空采样，我们将选择器形式化为随机策略，并通过在联合准确性与效率奖励下的GRPO进行优化。通过在较低视觉成本下对正确预测进行组相对比较奖励，我们的方法鼓励策略根据每个视频的信息密度动态分配计算。在三个长视频理解基准测试中，EviSelect的表现优于现有方法，同时减少了约50%的选择视觉标记，并实现了3.9倍的端到端加速。
cs.CV / 56 / 2608.05782
VSMP-IMU: Video-Grounded Semantic Motion Programs for Sensor-Aware Synthetic IMU Generation
VSMP-IMU：基于视频的语义运动程序用于传感器感知的合成IMU生成
Abstract
Wearable human activity recognition (HAR) is often limited by the scarcity of labeled sensor data, especially in low-resource, class-imbalanced, and subject-generalization settings. Synthetic IMU generation can reduce this dependency and enhance HAR machine learning model's performance, but existing approaches face a trade-off without addressing all factors: video-driven methods are visually grounded but sensitive to pose-estimation errors, while text-driven methods are controllable but often weakly grounded in how activities are actually performed. We present VSMP-IMU, a video-grounded framework for controllable synthetic IMU generation based on a structured Semantic Motion Program (SMP), which separates activity-defining semantics from label-preserving variation. Given an input video, VSMP-IMU extracts and augments an SMP, uses it to synthesize motion, converts the motion into virtual IMU signals, and grounds the resulting signals to the target wearable domain. We evaluate VSMP-IMU against state-of-the-art synthetic data generation methods on five public IMU-HAR datasets under leave-one-person-out evaluation. VSMP-IMU achieves an average Macro-F1 of 78.33%, improving over real-only training by 9.77% and over the strongest prior synthetic baseline by 4.04%. In low-resource settings with reduced training data-samples, it improves over real-only training by 18.54% and over the strongest prior synthetic baselines by more than 6% on average. Under long-tail evaluation in imbalanced datasets, it improves tail-class Macro-F1 by 19.86% over Real-only training and by 4.76% over SOTA. These results show that structured video-grounded semantics provide a practical foundation for controllable, wearable-relevant synthetic sensor data generation.
Chinese Translation
可穿戴人类活动识别（HAR）常常受到标记传感器数据稀缺的限制，尤其是在低资源、类别不平衡和主体泛化的环境中。合成IMU生成可以减少这种依赖，并提升HAR机器学习模型的性能，但现有方法在未解决所有因素的情况下面临权衡：视频驱动的方法在视觉上是有依据的，但对姿态估计误差敏感，而文本驱动的方法可控但往往在活动实际执行方式上基础薄弱。我们提出了VSMP-IMU，一个基于结构化语义运动程序（SMP）的视频驱动框架，用于可控的合成IMU生成，该框架将活动定义语义与保留标签的变异分开。给定输入视频，VSMP-IMU提取并增强SMP，利用其合成运动，将运动转换为虚拟IMU信号，并将生成的信号与目标可穿戴领域相结合。我们在五个公共IMU-HAR数据集上对VSMP-IMU与最先进的合成数据生成方法进行了评估，采用留一人法评估。VSMP-IMU的平均宏F1达78.33%，比仅使用真实数据的训练提高了9.77%，比最强的先前合成基线提高了4.04%。在低资源设置下，减少训练数据样本时，它比仅使用真实数据的训练提高了18.54%，比最强的先前合成基线平均提高了超过6%。在不平衡数据集的长尾评估中，它比仅使用真实数据的训练提高了19.86%的尾类宏F1，比最先进技术提高了4.76%。这些结果表明，结构化的视频驱动语义为可控的、与可穿戴相关的合成传感器数据生成提供了实用基础。
cs.CV / 57 / 2608.05798
KVAE: Family of Tokenizers for Multimodal Generative Models
KVAE：多模态生成模型的分词器系列
Abstract
Latent diffusion modeling (LDM), a prominent paradigm, utilizes tokenizers to map input signal to compressed representation. This dependency positions tokenizer as an integral part of generation process itself, since it affects learning speed, quality of synthesized samples and lay foundation for later applications. This report presents series of KVAE tokenizers for audio, image and video, all designed for subsequent text-conditioned generation: KVAE-Audio, a continuous full-band 48 kHz tokenizer with a 50 Hz latent of 64 channels; KVAE-3D -- two causal video tokenizers for 4x16x16 and 4x8x8 compression; KVAE-2D, an image model, compressing input by factor of 8 with 32 channels. We demonstrate that reconstruction (PSNR, LPIPS, PESQ, etc.) and generation results on objective (Frechet Distance, CLIP score, CLAP score, etc.) and subjective (side-by-side evaluation) metrics matches or surpasses frontier opensource tokenizers, such as VAEs from Wan-2.2, HunyuanVideo-1.5, FLUX.2, MovieGen, StableAudio and MMAudio. Considering difficulty of development, we share with community training details, model selection method and ablation on design choices. The code is publicly available at https://github.com/kandinskylab/kvae and https://github.com/kandinskylab/kvae-audio.
Chinese Translation
潜在扩散建模（Latent Diffusion Modeling, LDM）是一种突出的范式，它利用分词器将输入信号映射到压缩表示。这种依赖关系使得分词器成为生成过程的一个不可或缺的部分，因为它影响学习速度、合成样本的质量，并为后续应用奠定基础。本报告提出了一系列KVAE分词器，适用于音频、图像和视频，均为后续的文本条件生成而设计：KVAE-Audio，一个具有50 Hz潜在的连续全带48 kHz分词器，通道数为64；KVAE-3D——两个因果视频分词器，分别用于4x16x16和4x8x8压缩；KVAE-2D，一个图像模型，通过32个通道将输入压缩8倍。我们展示了在客观（Frechet距离、CLIP分数、CLAP分数等）和主观（并排评估）指标上，重建（PSNR、LPIPS、PESQ等）和生成结果与前沿开源分词器（如Wan-2.2的VAE、HunyuanVideo-1.5、FLUX.2、MovieGen、StableAudio和MMAudio）相匹配或超越。考虑到开发的难度，我们与社区分享了训练细节、模型选择方法和设计选择的消融实验。代码已公开发布在 https://github.com/kandinskylab/kvae 和 https://github.com/kandinskylab/kvae-audio。
cs.CV / 58 / 2608.05803
Vorch-Omni: Multi-Task Orchestration of Sight and Sound
Vorch-Omni：视听多任务协调
Vorch Team, Chen, Xiaoyu, Ding, Yang, Han, Cong, Han, Menglin, Hong, Yuxin, Hou, Jiebo, Jie, Zequn, Li, Xiang, Liu, Jing, Liu, Qi, Lu, Yulei, Luo, Siyuan, Ma, Lin, Ma, Xin, Qian, Yinlong, Shi, Peng, Wan, Fang, Wang, Siqi, Wang, Yaohui, Wang, Yaole, Wu, Yidi, Yang, Siqian, Yin, Mingyu, Yu, Haoran, Yue, Gang, Zhang, Lisai, Zhang, Yuting
Abstract
Recent advances in generative video modeling have enabled diverse generation, reference-based synthesis, extension, and editing, but existing approaches often rely on fragmented task-specific models. A general model must distinguish heterogeneous target, source, and reference signals to determine what to generate, preserve, or use as guidance, while reducing interference among tasks. Joint audio-visual generation further increases this challenge by introducing diverse conditioning and output configurations across modalities. We present Vorch-Omni, a unified multi-task framework for audio-visual synthesis based on an arbitrary-condition-to-arbitrary-output formulation. It flexibly treats video and audio signals as either conditioning inputs or generation targets. Token-level conditioning masks and task identifiers distinguish targets, source content, and references, while position types separate temporal context from independent conditions. To capture semantic and structural information, Vorch-Omni employs complementary visual conditioning pathways: a vision-language model interprets sampled frames with text instructions, and a video VAE encodes conditions into latent tokens for direct guidance. We further build a distributed data pipeline to curate diverse temporally aligned audio-visual clips, generate structured captions and metadata, and balance heterogeneous task distributions. Built on a single flow-matching diffusion transformer without task-specific architectural changes, Vorch-Omni supports over 10 tasks, including text-to-video, text-to-audio-video, image- and reference-conditioned generation, temporal extension, audio-driven generation, video transformation, and audio-visual editing. This unified framework provides a scalable foundation for general-purpose audio-visual generation and manipulation.
Chinese Translation
最近在生成视频建模方面的进展使得多样化生成、基于参考的合成、扩展和编辑成为可能，但现有的方法往往依赖于碎片化的任务特定模型。一个通用模型必须区分异质的目标、源和参考信号，以确定生成、保留或用作指导的内容，同时减少任务之间的干扰。联合音视频生成进一步增加了这一挑战，因为它在不同模态中引入了多样的条件和输出配置。我们提出了Vorch-Omni，一个基于任意条件到任意输出的统一多任务框架，用于音视频合成。它灵活地将视频和音频信号视为条件输入或生成目标。令牌级条件掩码和任务标识符区分目标、源内容和参考，而位置类型则将时间上下文与独立条件分开。为了捕捉语义和结构信息，Vorch-Omni采用互补的视觉条件路径：一个视觉-语言模型通过文本指令解释采样帧，一个视频变分自编码器（VAE）将条件编码为潜在令牌以进行直接指导。我们进一步构建了一个分布式数据管道，以策划多样的时间对齐音视频片段，生成结构化的标题和元数据，并平衡异质任务分布。Vorch-Omni基于单一的流匹配扩散变换器构建，无需任务特定的架构更改，支持超过10个任务，包括文本到视频、文本到音频视频、图像和参考条件生成、时间扩展、音频驱动生成、视频转换和音视频编辑。这个统一框架为通用音视频生成和操作提供了可扩展的基础。
cs.CV / 59 / 2608.05804
Ordered Diffusion for 3D Human Registration
用于三维人类配准的有序扩散
Abstract
3D human registration has historically been treated as a regression task, assuming a unique ground-truth alignment exists between the template and an input point cloud. In reality, acquisition noise, occlusions, and unknown soft tissue dynamics introduce inherent ambiguity into human scans. Regression-based methods consequently converge to an average prediction, often failing to represent a plausible geometry. In our work, we embrace such uncertainty by modeling the registration as a distribution of alignments. We propose ODin, which formulates registration as a 3D diffusion process that generates a point cloud aligned with the target geometry while preserving template semantics through consistent point ordering. To achieve this, ODin relies on global, local, and positional conditioning, guiding each point to its correct location. Our experiments demonstrate that such a generative formulation not only outperforms its regression-based baseline, but also establishes a new state of the art, surpassing highly engineered methods while reducing the registration time by two-thirds. Pre-trained models and code are available at https://riccardomarin.github.io/odin/.
Chinese Translation
三维人类配准历来被视为回归任务，假设模板与输入点云之间存在唯一的真实对齐。然而，实际情况中，采集噪声、遮挡以及未知的软组织动态为人类扫描引入了固有的模糊性。因此，基于回归的方法通常会收敛到一个平均预测，往往无法准确表示合理的几何形状。在我们的研究中，我们通过将配准建模为对齐分布来接受这种不确定性。我们提出了ODin，该方法将配准公式化为一个三维扩散过程，生成与目标几何形状对齐的点云，同时通过一致的点排序保持模板语义。为实现这一目标，ODin依赖于全局、局部和位置条件，引导每个点到达其正确的位置。我们的实验表明，这种生成性公式不仅超越了基于回归的基线，而且确立了新的技术前沿，超越了高度工程化的方法，同时将配准时间减少了三分之二。预训练模型和代码可在 https://riccardomarin.github.io/odin/ 获取。
cs.CV / 60 / 2608.05808
STAIL: Semantic Text-Anchored Incremental Learning for Medical Imaging via Large Language Models
STAIL：通过大型语言模型实现医学影像的语义文本锚定增量学习
Abstract
Deep learning models applied to medical image analysis suffer from severe catastrophic forgetting when continually adapting to new clinical tasks in dynamic environments. Mainstream incremental learning methods typically mitigate this by rehearsing raw historical images. However, this pixel-level rehearsal incurs significant storage overhead, raises privacy concerns, and fails to adequately capture the true data distribution with sparse exemplars. Inspired by human cognitive mechanisms, we propose a novel framework termed Semantic Text-Anchored Incremental Learning (STAIL) for sequential clinical tasks. To overcome the rehearsal bottleneck, STAIL introduces an asymmetric semantic consolidation buffer (SCB). By incorporating a minimal set of image anchors and extensive textual descriptions, the SCB enables dense semantic reconstruction of old tasks at a minimal storage cost. Furthermore, we design an LLM-derived Semantic Anchoring Mechanism (LSAM) that leverages the stable semantic space of frozen large language models as developmental priors. This mechanism explicitly anchors evolving visual features to textual representations, guiding and constraining plasticity and stability at both macroscopic and microscopic levels. Extensive experiments across three heterogeneous medical datasets, covering fundus, ultrasound, and X-ray imaging, demonstrate that STAIL acts as a highly effective plug-and-play module. It comprehensively enhances the performance of various existing baselines, achieving average gains of 2.24\% in AAA-AUC for sustained performance and 3.55\% in BWT-AUC for reduced forgetting. Code is available.
Chinese Translation
应用于医学图像分析的深度学习模型在动态环境中持续适应新临床任务时，面临严重的灾难性遗忘问题。主流的增量学习方法通常通过重演原始历史图像来缓解这一问题。然而，这种像素级的重演会带来显著的存储开销，增加隐私担忧，并且未能充分捕捉稀疏样本的真实数据分布。受到人类认知机制的启发，我们提出了一种新颖的框架，称为语义文本锚定增量学习（Semantic Text-Anchored Incremental Learning，STAIL），用于顺序临床任务。为了解决重演瓶颈，STAIL引入了一种非对称的语义整合缓冲区（Semantic Consolidation Buffer，SCB）。通过结合一组最小的图像锚点和大量的文本描述，SCB能够以最小的存储成本实现对旧任务的密集语义重构。此外，我们设计了一种基于大型语言模型的语义锚定机制（LLM-derived Semantic Anchoring Mechanism，LSAM），利用冻结的大型语言模型的稳定语义空间作为发展先验。该机制明确地将不断演变的视觉特征锚定到文本表示上，从而在宏观和微观层面上引导和约束可塑性与稳定性。在三个异构医学数据集（涵盖眼底、超声和X光影像）的广泛实验中，STAIL作为一种高效的即插即用模块，全面提升了各种现有基准的性能，持续性能的AAA-AUC平均提升2.24%，而遗忘减少的BWT-AUC平均提升3.55%。代码已公开。
cs.CV / 61 / 2608.05811
Energy-Guided Flow Matching
能量引导的流匹配
Abstract
Pixel-space generative models bypass lossy latent compression, yet necessitate joint learning of global structure and fine-grained details in a high-dimensional space. Standard flow matching interpolates noise toward a fixed clean-image endpoint, leaving the spectral evolution to be learned implicitly. In this paper, we introduce Energy-Guided Flow Matching(EG-FM) that explicitly models a coarse-to-fine generative trajectory by moving endpoint. Specifically, EG-FM replaces the fixed endpoint with a heat-kernel-filtered endpoint that evolves smoothly from low-frequency image to clean image.The fraction of high-frequency signal in moving endpoint is released by an image-specific energy-guided scheduling, leading to the re-targeting of velocity in flow matching.Our framework requires no adaptation of the backbone and training data, bringing negligible cost on the training and inference stages. In our experiment, EG-FM consistently achieves lower FID on the ImageNet class-conditional image generation task at $256 \times 256$ with fewer epochs, reaching an FID of 1.55 at 200 epochs and 1.45 at 600 epochs. We continue training the generation task on the setting of $512 \times 512$ resolution, yielding a FID of 1.58 after only 40 high-resolution adaptation epochs.Furthermore, we transfer EG-FM on text-to-image generation and achieve 0.85 on GenEval score and 83.9 on DPG-Bench. Code is available at https://github.com/ysng123/EG-FM.
Chinese Translation
像素空间生成模型绕过了有损的潜在压缩，但需要在高维空间中联合学习全局结构和细粒度细节。标准的流匹配将噪声插值到固定的干净图像终点，留下谱演化隐式学习。在本文中，我们引入了能量引导的流匹配（Energy-Guided Flow Matching，EG-FM），该方法通过移动终点显式建模粗到细的生成轨迹。具体而言，EG-FM用一个经过热核滤波的终点替换固定终点，该终点从低频图像平滑演变到干净图像。通过图像特定的能量引导调度，移动终点中的高频信号的比例被释放，从而重新调整流匹配中的速度。我们的框架不需要对主干网络和训练数据进行适应，对训练和推理阶段的成本几乎可以忽略不计。在我们的实验中，EG-FM在$256 imes 256$的ImageNet类条件图像生成任务中始终实现了更低的FID，经过更少的训练轮次，在200个轮次时达到FID为1.55，在600个轮次时达到FID为1.45。我们继续在$512 imes 512$分辨率的设置上训练生成任务，仅经过40个高分辨率适应轮次便获得了FID为1.58。此外，我们将EG-FM转移到文本到图像生成任务上，GenEval得分为0.85，DPG-Bench得分为83.9。代码可在https://github.com/ysng123/EG-FM获取。
cs.CV / 62 / 2608.05815
Bayesian adaptively-weighted ensembles for few-shot abdominal segmentation
基于贝叶斯自适应加权集成的少样本腹部分割
Abstract
Few-shot learning has emerged as a promising approach for anatomical segmentation when labelled data are scarce. However, different few-shot learning algorithms exhibit complementary strengths and weaknesses, with performance varying across anatomical targets and institutions. Existing few-shot segmentation ensembles, that combine predictions from multiple algorithms, typically employ fixed weighting schemes and therefore cannot adjust model contributions according to the target domain. In this work, we propose a Bayesian adaptively-weighted ensemble framework for segmentation under label scarcity and domain shift. Multiple few-shot segmentation algorithms are first adapted using a small labelled support set. Bayesian optimisation is then used to automatically identify ensemble weights that maximise segmentation performance on a target-domain validation set. The learned weights are subsequently fixed and applied to combine predictions on previously unseen query images from the target domain. The proposed framework is evaluated on the Cross-institution Male Pelvic Structures dataset using held-out anatomical structures and institutions to simulate simultaneous label scarcity and institutional domain shift. Results demonstrate statistically significant improvements over individual few-shot learners, fixed-weight ensembles, training-from-scratch baselines and recent state-of-the-art ensembling approaches. By adapting model contributions to the target anatomy and institutional domain, the proposed framework provides a practical mechanism for deploying segmentation systems to new clinical sites under severe annotation constraints.
Chinese Translation
少样本学习已成为在标注数据稀缺时进行解剖分割的有前景的方法。然而，不同的少样本学习算法表现出互补的优缺点，其性能在不同的解剖目标和机构之间有所差异。现有的少样本分割集成方法通常结合多个算法的预测，但采用固定的加权方案，因此无法根据目标领域调整模型的贡献。在本研究中，我们提出了一种在标签稀缺和领域转移下的贝叶斯自适应加权集成框架。首先，使用小规模的标注支持集对多个少样本分割算法进行适配。然后，利用贝叶斯优化自动识别最大化目标领域验证集上分割性能的集成权重。学习到的权重随后被固定，并应用于结合来自目标领域的先前未见查询图像的预测。该框架在跨机构男性骨盆结构数据集上进行了评估，使用保留的解剖结构和机构来模拟同时的标签稀缺和机构领域转移。结果显示，与单一少样本学习者、固定权重集成、从头训练的基线以及近期的最先进集成方法相比，具有统计学显著性改进。通过将模型贡献适应于目标解剖结构和机构领域，所提出的框架为在严重注释限制下将分割系统部署到新的临床场所提供了一种实用机制。
cs.CV / 63 / 2608.05834
Controllable Clothing: Precise Labels and Generation for Virtual Try-On with Latent Diffusion Models
可控服装：基于潜在扩散模型的虚拟试穿精确标签与生成
Abstract
In this technical report, I present a new method for guiding image generation in the context of Virtual- Try-On (VITON). The proposed method leverages new open source Ai models to augment the image data with labels, such as lengths and styles. By training adapters with these labels paired with images of the garments, the model can produce a more diverse set of images that the user can control. For the end user, such as a retailer, this means that they can assure that the produced image is as true to the true fit as possible, not misleading consumers
Chinese Translation
在本技术报告中，我提出了一种新的方法，用于在虚拟试穿（Virtual Try-On, VITON）环境中指导图像生成。所提方法利用新的开源人工智能模型，通过标签（如长度和风格）增强图像数据。通过使用这些与服装图像配对的标签训练适配器，该模型能够生成用户可控的更为多样化的图像。对于最终用户，例如零售商而言，这意味着他们可以确保生成的图像尽可能真实地反映合身效果，而不会误导消费者。
cs.CV / 64 / 2608.05840
Accurate Localization of Road Traffic Objects on the Road Plane Using Surveillance Camera Imagery
基于监控摄像头图像的道路平面交通物体精确定位
Abstract
Accurate vehicle localization from monocular roadside surveillance cameras is important for intelligent transportation systems, traffic monitoring, and traffic conflict analysis. Standard approaches often estimate vehicle position from the center of the detector bounding box, which can produce large errors due to perspective distortion and parallax, especially for elevated cameras and large vehicles. This paper proposes a two-stage geometry-aware localization pipeline that estimates the projection of the vehicle footprint onto the road plane. First, vehicles are detected using a YOLO26-based detector. Second, a dedicated ResNet34 regression network predicts four corner points corresponding to the projected vehicle base. The final position is computed as the geometric center of the predicted quadrilateral. The method was trained on synthetic data generated in CARLA and fine-tuned on real-world roadside imagery from DAIR-V2X. Experiments on synthetic and real data showed clear improvements over naive bounding-box-center localization. On DAIR-V2X, the mean image-space localization error decreased from 31.77 px to 15.30 px, a 51.8% improvement, while the median error decreased to 4.29 px. Median ground-plane error for medium-range vehicles decreased from 5.52 m to 0.90 m, and for far-range vehicles from 8.67 m to 1.84 m. The results also show that contextual information surrounding the detector bounding box is important for geometric localization. The largest gains were observed for distant vehicles and geometrically challenging cases affected by strong perspective distortion and parallax.
Chinese Translation
从单目路边监控摄像头中精确定位车辆对于智能交通系统、交通监测和交通冲突分析至关重要。标准方法通常通过检测器边界框的中心估计车辆位置，这可能由于透视失真和视差而产生较大误差，尤其是在高架摄像头和大型车辆的情况下。本文提出了一种两阶段的几何感知定位管道，该管道估计车辆足迹在道路平面上的投影。首先，使用基于YOLO26的检测器检测车辆。其次，专门的ResNet34回归网络预测与投影车辆底部对应的四个角点。最终位置计算为预测的四边形的几何中心。该方法在CARLA生成的合成数据上进行训练，并在DAIR-V2X的真实路边图像上进行微调。合成数据和真实数据的实验结果显示，相较于简单的边界框中心定位，精确定位方法有明显改善。在DAIR-V2X上，平均图像空间定位误差从31.77 px降至15.30 px，改善幅度为51.8%，而中位数误差降至4.29 px。中等距离车辆的中位数地面误差从5.52 m降至0.90 m，远距离车辆的中位数地面误差从8.67 m降至1.84 m。结果还表明，检测器边界框周围的上下文信息对于几何定位非常重要。对于远距离车辆和受强透视失真及视差影响的几何挑战案例，观察到最大的改进。
cs.CV / 65 / 2608.05843
Overcoming Attention Drift: Homogeneity-Heterogeneity Guided Feature Aggregation for Low-Light Remote Sensing Image Enhancement
克服注意力漂移：基于同质性-异质性引导的特征聚合用于低光遥感图像增强
Abstract
Restoring high-fidelity remote sensing imagery from extreme low-light degradation is indispensable for reliable Earth observation and downstream machine vision. However, under severe noise and illumination corruption, existing methods suffer from attention drift, erroneously aggregating features across distinct physical boundaries and causing severe structural blurring and color distortion. To address this, we propose HALO, a dual-prior-driven enhancement framework that formulates enhancement as a guided feature aggregation problem driven by foundation model priors. Specifically, an illumination-invariant semantic prior provides regional homogeneity as a positive bias for content-consistent aggregation, while a pseudo-3D topological prior provides boundary heterogeneity as a negative penalty to strictly prevent cross-boundary confusion. To cooperatively incorporate these two priors, we propose a Homogeneity-Heterogeneity Cooperative Attention Module (H2CAM) to resolve feature conflicts during cross-modal prior fusion. Extensive experiments demonstrate that HALO achieves state-of-the-art performance across 8 challenging synthetic and real-world remote sensing benchmarks, significantly improving physical boundary sharpness and color fidelity while maximizing the preservation of discriminative features for downstream Earth observation tasks.
Chinese Translation
从极低光照退化中恢复高保真遥感图像对于可靠的地球观测和下游机器视觉至关重要。然而，在严重的噪声和光照损坏下，现有方法遭遇注意力漂移，错误地聚合跨越不同物理边界的特征，导致严重的结构模糊和颜色失真。为了解决这个问题，我们提出了HALO，一个双先验驱动的增强框架，将增强过程表述为一个由基础模型先验驱动的引导特征聚合问题。具体而言，光照不变的语义先验提供区域同质性作为内容一致聚合的正偏差，而伪3D拓扑先验则提供边界异质性作为负惩罚，以严格防止跨边界混淆。为了协同结合这两个先验，我们提出了同质性-异质性协作注意力模块（H2CAM），以解决跨模态先验融合过程中的特征冲突。大量实验表明，HALO在8个具有挑战性的合成和真实世界遥感基准测试中实现了最先进的性能，显著提高了物理边界的清晰度和颜色保真度，同时最大限度地保留了下游地球观测任务所需的区分特征。
cs.CV / 66 / 2608.05844
Curia-MAE: Multi-Modal Multi-Anatomy MAE Pre-Training for 3D Medical Image Segmentation
Curia-MAE：用于三维医学图像分割的多模态多解剖结构MAE预训练
Abstract
Radiology foundation models learn transferable representations that can be adapted to new tasks by training only small layers on top of a frozen encoder. Dense prediction tasks such as 3D segmentation are, however, underrepresented in their evaluation, and, with the encoder kept frozen, pre-trained models still fall short of nnU-Net, the state-of-the-art reference trained from scratch. To close this gap we extend convolutional MAE pre-training with a robust reconstruction objective, a feature regularizer, and a local-global similarity objective. Using this method, we propose Curia-MAE, a multi-modal, multi-anatomy MAE model pre-trained on 300,000 CT and MRI images covering a large number of anatomical sites. On eight anatomy- and lesion-focused segmentation benchmarks, Curia-MAE improves frozen-encoder performance over a strong MAE baseline, while remaining competitive under full finetuning and superior on lesion tasks, where labeled data is scarce. These results indicate that a single frozen encoder can be reused across diverse segmentation tasks, reducing the cost of adapting and deploying such models in clinical workflows. We will make our pre-trained model weights publicly available.
Chinese Translation
放射学基础模型学习可迁移的表示，这些表示可以通过在冻结的编码器上训练少量层来适应新任务。然而，像三维分割这样的密集预测任务在其评估中表现不足，并且在编码器保持冻结的情况下，预训练模型仍然无法超越从零开始训练的最先进参考模型nnU-Net。为了解决这一问题，我们扩展了卷积MAE预训练，采用了强健的重建目标、特征正则化器和局部-全局相似性目标。使用这种方法，我们提出了Curia-MAE，一个多模态、多解剖结构的MAE模型，预训练于30万张涵盖大量解剖部位的CT和MRI图像。在八个以解剖和病变为重点的分割基准上，Curia-MAE在强大的MAE基线之上提高了冻结编码器的性能，同时在完全微调下仍保持竞争力，并在标注数据稀缺的病变任务上表现优越。这些结果表明，单个冻结编码器可以在不同的分割任务中重复使用，从而降低了在临床工作流程中适应和部署此类模型的成本。我们将公开我们的预训练模型权重。
cs.CV / 67 / 2608.05848
DTRNet: Dual Text-Radical Decoding for Handwritten Chinese Text Recognition with Faked Character Detection
DTRNet：用于手写中文文本识别的双重文本-部首解码与伪字符检测
Abstract
In K-12 educational scenarios, handwritten Chinese text recognition should not only transcribe student writing, but also detect faked characters. However, existing recognition models are usually confined to a predefined set of normal characters and therefore cannot explicitly identify faked characters. Existing detection methods exhibit complementary limitations: character-level methods provide interpretable structural evidence but suffer from low efficiency, whereas line-level methods are efficient but rely heavily on confidence scores, making them prone to missed detections and lacking explicit structural evidence. Thus, the key challenge is to preserve character-structural evidence independent of contextual inference while maintaining line-level efficiency. To this end, we propose DTRNet, a dual Text-Radical decoding framework for line-level faked character detection. DTRNet decouples context-aware text recognition from character-wise structural verification, where the text branch performs line-level transcription and the radical branch predicts legal Ideographic Description Sequences (IDS) for lexicon-based faked character judgment. We further introduce IDS-Guided Confidence Adjustment (IGCA) to refine text predictions using structural evidence during inference. Experimental results demonstrate that DTRNet effectively detects faked characters while maintaining strong recognition performance and providing interpretable radical-level evidence. Code, checkpoints, and the processed dataset are publicly available at https://github.com/BNU-ERC-ITEA/DTRNet.
Chinese Translation
在K-12教育场景中，手写中文文本识别不仅需要转录学生的书写，还需检测伪字符。然而，现有的识别模型通常局限于预定义的正常字符集合，因此无法明确识别伪字符。现有的检测方法存在互补的局限性：字符级方法提供可解释的结构证据，但效率较低；而行级方法效率较高，但过于依赖置信度分数，容易出现漏检且缺乏明确的结构证据。因此，关键挑战在于在保持行级效率的同时，保留独立于上下文推理的字符结构证据。为此，我们提出了DTRNet，一种用于行级伪字符检测的双重文本-部首解码框架。DTRNet将上下文感知的文本识别与字符级结构验证解耦，其中文本分支执行行级转录，部首分支预测合法的意符描述序列（Ideographic Description Sequences, IDS）以进行基于词典的伪字符判断。我们进一步引入了IDS引导的置信度调整（IDS-Guided Confidence Adjustment, IGCA），在推理过程中利用结构证据来优化文本预测。实验结果表明，DTRNet在保持强大识别性能的同时，能够有效检测伪字符并提供可解释的部首级证据。代码、检查点和处理后的数据集可在 https://github.com/BNU-ERC-ITEA/DTRNet 上公开获取。
cs.CV / 68 / 2608.05858
Shape-Aware Oriented Bounding Box (OBB) to Horizontal Bounding Box (HBB) Conversion
形状感知的定向边界框（OBB）到水平边界框（HBB）转换
Abstract
Accurate object detection in aerial and satellite imagery is dependent upon the bounding box representation. This is especially true for spatially oriented objects such as ships or aircrafts. Oriented Bounding Boxes (OBB) have a tighter fit and more robust non-max suppression compared to Horizontal Bounding Boxes (HBB), any current post-processing conversion from OBB to HBB either introduces excess empty and background space or removes data from the detection. This paper introduces a novel approach for a shape-aware OBB-to-HBB conversion for ship detection in remote sensing imagery. It leverages hull shape, hull fullness, and the bounding box orientation to produce a tighter axis-aligned HBB representation. The proposed method is benchmarked against three baselines methods for OBBto-HBB conversion, Outer HBB which uses minimum and maximum, Area Equivalent HBB and GBB Marginalized HBB.
Chinese Translation
在航空和卫星图像中，准确的目标检测依赖于边界框的表示。这对于空间定向的物体，如船只或飞机，尤其重要。与水平边界框（HBB）相比，定向边界框（OBB）具有更紧密的拟合和更强的非极大值抑制能力。然而，当前从OBB到HBB的后处理转换要么引入多余的空白和背景空间，要么从检测中移除数据。本文提出了一种新颖的形状感知OBB到HBB的转换方法，旨在遥感图像中的船只检测。该方法利用船体形状、船体饱满度和边界框方向，生成更紧凑的轴对齐HBB表示。所提方法与三种OBB到HBB转换的基准方法进行了比较，包括使用最小值和最大值的外部HBB（Outer HBB）、面积等效HBB（Area Equivalent HBB）和GBB边际HBB（GBB Marginalized HBB）。
cs.CV / 69 / 2608.05877
D-CLOT: Double Closed Loop Optimal Transport for Unsupervised Action Segmentation
D-CLOT：用于无监督动作分割的双闭环最优传输
Abstract
Optimal transport (OT) has emerged as an effective framework for unsupervised action segmentation. Yet, in existing OT-based methods, the latent action prototypes that define the OT costs are not re-estimated from the refined frame geometry. Instead, they evolve solely through gradients from the pseudo-label loss. We identify this \emph{representation--prototype inconsistency} as a central bottleneck, particularly around ambiguous transitions and for short or infrequent actions. To address this issue, we build on the recently introduced CLOT, which refines frame embeddings based on estimated segment embeddings, and further re-estimates the action prototypes from the refined frame embeddings. Specifically, we introduce a graph-constrained module that regularizes the OT-refined frame and segment representations by preserving the local neighborhood geometry of the encoder output. An action-embedding refinement step then periodically re-anchors the prototypes to this stabilized representation geometry. We study two instantiations that share the same backbone, graph module, and objective: D-CLOT updates the prototypes using $k$-means, whereas D-CLOT$_{B}$ updates them as OT barycenters weighted by the refined transport plan, yielding an assignment-aware prototype update consistent with the current transport geometry. Across five established benchmarks, both variants improve segment-level quality over CLOT, with per-video gains of up to $+12.7$ F1 and $+10.2$ mIoU (YTI) and activity-level gains of up to $+8.9$ F1 (FS-Eval). We further establish the first unsupervised action-segmentation baseline on Assembly101, a procedural and substantially more fine-grained benchmark than those commonly used in prior work. Extensive ablations and sensitivity analyses demonstrate that the two refinement mechanisms are complementary and robust.
Chinese Translation
最优传输（Optimal Transport, OT）已成为无监督动作分割的有效框架。然而，在现有的基于OT的方法中，定义OT成本的潜在动作原型并未从精炼的帧几何中重新估计，而是仅通过伪标签损失的梯度进行演化。我们将这种 extit{表示-原型不一致性}视为一个核心瓶颈，尤其是在模糊的过渡和短暂或不频繁的动作中。为了解决这一问题，我们基于最近提出的CLOT进行改进，该方法根据估计的段嵌入精炼帧嵌入，并进一步从精炼的帧嵌入中重新估计动作原型。具体而言，我们引入了一个图约束模块，通过保持编码器输出的局部邻域几何来规范化OT精炼的帧和段表示。然后，一个动作嵌入精炼步骤定期将原型重新锚定到这一稳定的表示几何上。我们研究了两个实例，它们共享相同的主干、图模块和目标：D-CLOT使用$k$-均值更新原型，而D-CLOT$_{B}$则将其更新为加权精炼传输计划的OT重心，从而实现与当前传输几何一致的分配感知原型更新。在五个已建立的基准测试中，这两种变体在段级质量上均优于CLOT，视频级别的提升最高可达$+12.7$ F1和$+10.2$ mIoU（YTI），活动级别的提升最高可达$+8.9$ F1（FS-Eval）。我们进一步在Assembly101上建立了第一个无监督动作分割基线，这是一个程序化且显著更细粒度的基准，相较于以往工作中常用的基准。大量的消融实验和敏感性分析表明，这两种精炼机制是互补且稳健的。
cs.CV / 70 / 2608.05878
MAVISEG: Manifold Propagation and Visual Prototypes for Zero-Shot Open-Vocabulary Segmentation in Diffusion Transformers
MAVISEG：用于扩散变换器的零-shot开放词汇分割的流形传播与视觉原型
Abstract
Text-to-image diffusion transformers learn about objects and scenes by learning to generate them, making them strong candidates for training-free zero-shot open-vocabulary semantic segmentation. State-of-the-art attribution methods score each pixel independently, comparing its features against a fixed text-derived class representation, whether as an output-space similarity or as a cross-attention weight. This discards structured signals the model itself exposes: the temporal structure of the generative trajectory, the visual appearance statistics of each concept, and the image's own pairwise feature geometry. We present MAVISEG, a training-free refinement layer that recovers these signals. Because its operators consume only a pixel-by-concept score field and a pixel feature space, MAVISEG is capture-agnostic rather than tied to one attribution method. Across six benchmarks it achieves the strongest overall results among training-free methods, including the best mIoU on every benchmark. Interestingly, gains are largest where the initial capture is weakest, and individual operators contribute depending on the noise in the field they refine. Our results indicate that diffusion transformers carry more concept-level information than current attribution methods recover, and that much of it is lost on the way to the mask rather than absent from the model.
Chinese Translation
文本到图像的扩散变换器通过学习生成对象和场景来获取相关知识，使其成为无训练的零-shot开放词汇语义分割的强有力候选者。最先进的归因方法独立地对每个像素进行评分，将其特征与固定的文本衍生类别表示进行比较，无论是作为输出空间相似度还是作为交叉注意力权重。这忽略了模型自身所暴露的结构化信号：生成轨迹的时间结构、每个概念的视觉外观统计以及图像自身的成对特征几何。我们提出了MAVISEG，这是一种无训练的精炼层，用于恢复这些信号。由于其操作符仅消耗像素-概念评分场和像素特征空间，MAVISEG是捕获无关的，而不是依赖于某一种归因方法。在六个基准测试中，它在无训练方法中取得了最强的整体结果，包括在每个基准测试中获得最佳的mIoU。有趣的是，增益在初始捕获最弱的地方最大，个别操作符的贡献取决于它们所精炼的场域中的噪声。我们的结果表明，扩散变换器携带的概念级信息超过了当前归因方法所恢复的，并且在到达掩膜的过程中，其中许多信息是丢失的，而不是模型本身缺失。
cs.CV / 71 / 2608.05879
To See a World in a Living Context: Unified Indoor-Outdoor Urban World Generation
在生动的背景中看世界：统一的室内外城市世界生成
Abstract
Text-driven 3D generation has advanced rapidly in creating large-scale outdoor environments and detailed indoor scenes, but these domains are usually synthesized independently, lacking the correspondence required for a coherent urban world. We present HoloWorld, a unified indoor-outdoor urban world generation framework built on a continuously updated cross-scale world context. Initializing from a user description, HoloWorld progressively represents and updates the diverse world information, from city-scale planning to individual buildings, allowing generated interiors to maintain explicit correspondence with their associated exterior buildings. Conditioned on the evolving context and previously generated neighboring blocks, HoloWorld autoregressively generates urban exteriors with consistent spatial organization and visual identity across blocks. The generated exterior representations are further grounded in 3D building instances and footprints, enabling building-specific indoor generation with geometry-constrained layouts and inherited appearance characteristics. To our knowledge, HoloWorld is the first framework to unify indoor and outdoor generation within a coherent 3D urban world. Extensive experiments demonstrate that HoloWorld achieves superior urban exterior generation performance, improving the average AQS score over the SOTA by 7.68\% and obtaining the highest average RDR score, while maintaining strong building-level indoor-outdoor correspondence and cross-block continuity within a unified 3D urban world.
Chinese Translation
基于文本驱动的三维生成在创建大规模户外环境和详细的室内场景方面迅速发展，但这些领域通常是独立合成的，缺乏构建连贯城市世界所需的对应关系。我们提出了HoloWorld，一个基于持续更新的跨尺度世界背景的统一室内外城市世界生成框架。HoloWorld从用户描述初始化，逐步表示和更新多样的世界信息，从城市规模规划到单个建筑，使生成的室内能够与其相关的外部建筑保持明确的对应关系。在不断演变的背景和先前生成的邻近街区的条件下，HoloWorld自回归地生成具有一致空间组织和视觉特征的城市外观。生成的外部表示进一步基于三维建筑实例和建筑轮廓，使得特定建筑的室内生成具有几何约束的布局和继承的外观特征。据我们所知，HoloWorld是第一个在连贯的三维城市世界中统一室内和外部生成的框架。大量实验表明，HoloWorld在城市外观生成性能上表现优越，平均AQS分数比现有技术（SOTA）提高了7.68%，并获得了最高的平均RDR分数，同时在统一的三维城市世界中保持强大的建筑级室内外对应关系和跨街区的连续性。
cs.CV / 72 / 2608.05903
Robust-WAM: Bridging Generative Pretraining and Semantic Foresight in World-Action Models
鲁棒-WAM：在世界-行动模型中桥接生成预训练与语义前瞻
Abstract
Mainstream World-Action Models (WAMs) adapt pretrained video generation models (VGMs) for robot control, transferring their learned dynamics prior for action prediction. These VGMs are typically trained in a variational autoencoder (VAE) latent space. However, the VAE latent space is optimized for pixel reconstruction, which rewards fine appearance detail and leaves the action prediction fragile under visual shifts. Recent works build WAMs in semantic latent space, which are more robust to appearance shifts. However, these models cannot leverage the large-scale VGM pretraining that exists only in VAE space. To overcome this dilemma, we propose Robust-WAM, a general post-training method for video-generation-based WAMs that preserves the VAE-based generative path and adds a lightweight semantic foresight alignment objective on the action stream. This retains the large-scale VGM pretraining while grounding actions in appearance-invariant dynamics that stay reliable under illumination shifts and other visual out-of-distribution conditions. Specifically, we employ learnable query tokens to bring future-scene semantics into the action stream by aligning their output hidden states with the semantic foresight of future ground-truth frames. To establish the temporal correspondence between each query and the future step it describes, we give it the positional encoding of the matching action tokens. Experiments on out-of-distribution generalization simulation benchmarks and a real-robot setup show that our Robust-WAM consistently improves the success rates of multiple WAM baselines without sacrificing in-distribution performance.
Chinese Translation
主流的世界-行动模型（WAMs）将预训练的视频生成模型（VGMs）应用于机器人控制，转移其学习到的动态先验用于动作预测。这些VGMs通常是在变分自编码器（VAE）潜在空间中训练的。然而，VAE潜在空间是为了像素重建而优化的，这使得其在视觉变化下的动作预测变得脆弱。最近的研究在语义潜在空间中构建WAMs，这些模型对外观变化更为鲁棒。然而，这些模型无法利用仅存在于VAE空间中的大规模VGM预训练。为了解决这一困境，我们提出了鲁棒-WAM，一种通用的后训练方法，适用于基于视频生成的WAMs，该方法保留了基于VAE的生成路径，并在动作流上增加了轻量级的语义前瞻对齐目标。这一方法保留了大规模的VGM预训练，同时将动作基于在光照变化和其他视觉分布外条件下保持可靠的外观不变动态。具体而言，我们采用可学习的查询标记，将未来场景的语义引入动作流，通过将其输出的隐藏状态与未来真实帧的语义前瞻对齐，从而实现。为了建立每个查询与其描述的未来步骤之间的时间对应关系，我们为其赋予匹配动作标记的位置信息编码。在分布外泛化模拟基准和真实机器人设置上的实验表明，我们的鲁棒-WAM在不牺牲分布内性能的情况下，持续提高了多个WAM基线的成功率。
cs.CV / 73 / 2608.05911
Mapping Armenian Paris: Extracting and Geocoding Commercial Advertisements from the 20th-Century Diaspora Press
映射亚美尼亚巴黎：从20世纪侨民报纸中提取和地理编码商业广告
Abstract
This paper presents an end-to-end, IIIF-based pipeline that turns the digitised Armenian press of France into an interactive map of the 20th-century Parisian Armenian commercial community. On each page, commercial advertisements are located, read, and parsed into structured records, which are then geocoded and placed on the map. Western Armenian is under-resourced and unsupported by off-the-shelf layout and OCR models, so the pipeline uses vision-language models (VLMs) as a data-bootstrapping strategy: they produce usable structured records at a scale hand annotation could not reach, and stay reliable on the strongly curved scans where conventional line-level CRNN OCR breaks down. The contribution includes a 500-page Western Armenian press corpus with 3,270 advertisement-level annotations, a Label Studio template that captures detection and semantic fields in a single annotation pass, and a reproducible workflow transposable to other under-resourced historical corpora. More broadly, the work shows that VLM-driven data bootstrapping is an effective lever for under-resourced historical languages such as (Western) Armenian.
Chinese Translation
本文提出了一种基于IIIF的端到端管道，将法国的数字化亚美尼亚报纸转化为20世纪巴黎亚美尼亚商业社区的互动地图。在每一页上，商业广告被定位、读取并解析为结构化记录，然后进行地理编码并放置在地图上。西方亚美尼亚语资源匮乏，且不被现成的排版和光学字符识别（OCR）模型支持，因此该管道采用视觉-语言模型（VLMs）作为数据引导策略：它们在手动标注无法达到的规模上生成可用的结构化记录，并在传统行级CRNN OCR失效的强曲线扫描上保持可靠性。该研究的贡献包括一个包含500页西方亚美尼亚报纸语料库的项目，具有3,270个广告级别的注释，一个在单次注释中捕获检测和语义字段的Label Studio模板，以及一个可复制的工作流程，可转移到其他资源匮乏的历史语料库。更广泛地说，该研究表明，VLM驱动的数据引导是资源匮乏的历史语言（如西方亚美尼亚语）的有效杠杆。
cs.CV / 74 / 2608.05920
Floating Radiance Networks
浮动辐射网络
Abstract
Recent advances in neural scene representations enable photorealistic novel-view synthesis, yet most methods remain tightly coupled to a single rendering paradigm, limiting their versatility and integration with conventional graphics workflows. We introduce Floating Radiance Networks (FlaRe), a neural scene representation combining explicit ray-traceable geometry with continuous neural radiance functions. A scene is represented by floating planar generalized Gaussian primitives, each carrying a compact latent descriptor of a local radiance field. A lightweight decoder shared across the scene maps this descriptor, local surface coordinates, and viewing direction to color and opacity. This formulation preserves the expressiveness of neural fields while providing an explicitly addressable structure that can be efficiently queried and manipulated. Hardware-accelerated primitive intersections enable interactive rendering and recursive ray-tracing, including reflections, refractions, transparency, and shadows. The same representation further supports primitive-level deformation, mesh extraction, and appearance stylization directly in its learned descriptor space. Experiments across standard reconstruction benchmarks demonstrate competitive rendering quality while using a compact set of primitives. Together, these results establish FlaRe as a versatile representation that brings high-fidelity neural rendering, ray-tracing, geometric manipulation, and appearance editing into a unified scene model. Source code is available online. Source code can be found at: https://github.com/KByrski/FlaRe
Chinese Translation
近年来，神经场景表示的进展使得逼真的新视角合成成为可能，但大多数方法仍然紧密依赖于单一的渲染范式，限制了它们的多功能性以及与传统图形工作流程的整合。我们提出了浮动辐射网络（Floating Radiance Networks, FlaRe），这是一种结合了显式可光线追踪几何体与连续神经辐射函数的神经场景表示。场景由浮动的平面广义高斯原语表示，每个原语携带一个局部辐射场的紧凑潜在描述符。一个轻量级解码器在整个场景中共享，将该描述符、局部表面坐标和视角方向映射到颜色和不透明度。这种表述保留了神经场的表现力，同时提供了一个可显式寻址的结构，可以高效查询和操作。硬件加速的原语交集实现了交互式渲染和递归光线追踪，包括反射、折射、透明度和阴影。相同的表示进一步支持原语级变形、网格提取和外观风格化，直接在其学习的描述符空间中进行。针对标准重建基准的实验表明，在使用紧凑的原语集的同时，渲染质量具有竞争力。这些结果共同确立了FlaRe作为一种多功能表示，将高保真神经渲染、光线追踪、几何操作和外观编辑整合到一个统一的场景模型中。源代码可在线获取，地址为：https://github.com/KByrski/FlaRe
cs.CV / 75 / 2608.05938
MirrorNet: Can Medical Image Anonymization Really Protect Patient Identity?
MirrorNet：医学图像匿名化真的能保护患者身份吗？
Abstract
Medical images are routinely de-identified---names, dates, and other metadata removed---and then shared for research, teaching, and public benchmarks under the assumption that this renders them anonymous. Such de-identification protects the metadata but not the pixels, and---apart from scans that directly contain facial structures---whether the image content itself identifies the patient has received little scrutiny. We investigate this question by learning a cycle-consistent correspondence between a cross-sectional medical image and a non-medical, patient-identifying image, using a pair of coupled, cycle-consistent variational autoencoders. From a held-out scan, the model recovers a recognisable likeness of the patient (identity-region MAE = 0.163); conversely, it synthesises a scan from such an image. These results indicate that a de-identified medical scan remains identifying---it is, in effect, a photograph of the patient---and that imaging data should be governed as biometric data rather than as anonymisable records. To support reproducibility, the code and trained models are shared at https://github.com/attilasimko/public-repository.
Chinese Translation
医学图像通常会进行去标识化处理——去除姓名、日期和其他元数据——然后在研究、教学和公共基准测试中共享，假设这使其变得匿名。然而，这种去标识化保护了元数据，但并未保护像素，除了直接包含面部结构的扫描外，图像内容本身是否能够识别患者却鲜有深入研究。我们通过学习横断面医学图像与非医学、能够识别患者的图像之间的循环一致性对应关系来探讨这个问题，使用一对耦合的循环一致性变分自编码器。从一个保留的扫描中，模型恢复出患者的可识别相似度（身份区域的平均绝对误差为0.163）；反之，它也能从这样的图像合成出扫描结果。这些结果表明，去标识化的医学扫描仍然具有识别性——实际上，它是患者的照片——因此，影像数据应被视为生物识别数据，而非可匿名化记录。为了支持可重复性，代码和训练模型已在 https://github.com/attilasimko/public-repository 上共享。
cs.CV / 76 / 2608.05945
Respect Your Zero-Shot Uncertainty: Conservative Calibration for Test-Time-Adapted Vision-Language Models
尊重你的零-shot不确定性：测试时适应视觉-语言模型的保守校准
Abstract
Test-time adaptation (TTA) can improve the recognition accuracy of vision-language models under distribution shift, but often degrades calibration, making predictive confidence unreliable for downstream decision-making. Many existing label-free calibration approaches are either coupled to prompt optimization or rely on logit-range statistics that provide only a coarse characterization of the predictive distribution. We show that TTA can increase confidence and reduce entropy even when the top-1 prediction and its correctness remain unchanged, a failure mode we term prediction-preserving sharpening. Across diverse TTA methods and benchmarks, larger entropy reductions relative to paired zero-shot predictions are associated with greater increases in Expected Calibration Error (ECE). On entropy-reduced samples, confidence gains also tend to exceed accuracy gains. Based on these findings, we propose Zero-Shot-Anchored Entropy Calibration (ZAEC), a label-free post-hoc method that uses zero-shot entropy as a sample-specific uncertainty reference. ZAEC selectively restores the zero-shot entropy of sharpened predictions through minimal temperature scaling while leaving all other predictions unchanged. It requires no labeled calibration data or learned parameters and preserves class rankings and classification accuracy. Across five TTA methods and 15 datasets, ZAEC achieves the lowest post-hoc macro-average ECE on ViT-B/16, with consistent gains on RN50.
Chinese Translation
测试时适应（TTA）可以提高视觉-语言模型在分布变化下的识别准确性，但通常会降低校准，使得预测置信度在下游决策中变得不可靠。许多现有的无标签校准方法要么与提示优化相结合，要么依赖于仅提供预测分布粗略表征的logit范围统计。我们展示了即使在top-1预测及其正确性保持不变的情况下，TTA也可以增加置信度并减少熵，这种失败模式我们称之为预测保持锐化。在多种TTA方法和基准测试中，相对于配对的零-shot预测，较大的熵减少与期望校准误差（ECE）的更大增加相关。在熵减少的样本中，置信度的提升往往超过准确度的提升。基于这些发现，我们提出了零-shot锚定熵校准（ZAEC），这是一种无标签的后处理方法，利用零-shot熵作为样本特定的不确定性参考。ZAEC通过最小温度缩放选择性地恢复锐化预测的零-shot熵，同时保持所有其他预测不变。它不需要标记的校准数据或学习参数，并保留类别排名和分类准确性。在五种TTA方法和15个数据集上，ZAEC在ViT-B/16上实现了最低的后处理宏平均ECE，并在RN50上获得了一致的提升。
cs.CV / 77 / 2608.05960
Big, Bright, or Invisible: A Frozen-Feature Benchmark of 3D CT Foundation Models
大、亮或隐形：3D CT 基础模型的冻结特征基准测试
Abstract
Routine CT interpretation is inherently comprehensive, capturing incidental findings across the entire scan volume. 3D CT foundation models could assist this process by providing generalizable representations of anatomy and pathology. To evaluate their diagnostic breadth, we benchmark ten frozen CT encoders across three cohorts of thoracic CT scans, including an unseen internal clinical dataset, using $k$-nearest neighbors, zero-shot prompting, and linear probing. We find no universal state-of-the-art, with rankings fluctuating significantly depending on the evaluation context. While models combining fine-grained image tokenization with vision-language alignment generally perform best, a lightweight supervised encoder remains highly competitive, demonstrating that explicit labels can effectively substitute for scale. Crucially, rather than model architecture, we observe that the primary determinant of performance is a physical bottleneck: a finding's detectability scales with its contrast against surrounding tissue and its spatial extent. Through controlled within-organ comparisons, we empirically demonstrate that widespread or high-contrast abnormalities, such as devices and effusions, are reliably recovered. Conversely, small, low-contrast focal lesions remain a persistent challenge across all evaluated encoders. We attribute this to the inherent limitations of globally pooled embeddings, suggesting that accurately representing small, low-contrast structures will require region- or lesion-level pretraining.
Chinese Translation
常规 CT 解释本质上是全面的，能够捕捉整个扫描体积中的偶发发现。3D CT 基础模型可以通过提供解剖和病理的可推广表示来辅助这一过程。为了评估它们的诊断广度，我们在三个胸部 CT 扫描队列中基准测试了十个冻结 CT 编码器，包括一个未见的内部临床数据集，使用 $k$-最近邻、零样本提示和线性探测。我们发现没有普遍的最先进模型，排名在不同的评估上下文中显著波动。尽管结合细粒度图像标记与视觉-语言对齐的模型通常表现最佳，但一个轻量级的监督编码器仍然具有很强的竞争力，表明显式标签可以有效替代规模。至关重要的是，我们观察到，性能的主要决定因素不是模型架构，而是一个物理瓶颈：发现的可检测性与其与周围组织的对比度及其空间范围成比例。通过控制的器官内比较，我们实证表明，广泛或高对比度的异常，如设备和积液，能够可靠地被检测到。相反，小型、低对比度的局灶性病变在所有评估的编码器中仍然是一个持续的挑战。我们将此归因于全球池化嵌入的固有限制，建议准确表示小型、低对比度结构需要区域或病变级别的预训练。
cs.CV / 78 / 2608.05964
Topology-Aware Neighborhood Learning for Source-Free Cross-Scene Hyperspectral Image Classification
基于拓扑感知的无源跨场景高光谱图像分类邻域学习
Abstract
Domain adaptation has advanced cross-scene hyperspectral image classification, significantly improving discriminative capability in complex scenarios. However, privacy rules or storage limits often block access to data from the source domain. Conventional domain adaptation methods become impractical, severely restricting their utility in realistic remote sensing scenarios. To tackle this challenge, we propose a topology-aware source-free learning framework. We first introduce the entropy momentum pseudo-labeling (EMP) to refine k-means assignments by leveraging entropy-aware confidence and temporal prediction momentum. Under the guidance of the refined pseudo-labels, we further utilize the contextual neighborhood topology (CNT) to exploit the intrinsic geometric structure of the target feature space. Combining the global structural information extracted by collaborative representation with the local similarity information modeled by nearest neighbor search, the CNT accomplishes the comprehensive encoding of manifold-level geometric properties in the target domain feature space. The overall objective integrates cross-entropy on refined pseudo-labels, log inner product-based topology consistency, and an information-maximization term for balanced classification, ensuring stable adaptation in the source-free setting. Extensive experiments on three typical cross-scenarios demonstrate that the proposed method exceeds state-of-the-art performance, and ablation studies further validate the contribution of each module. The results highlight the critical role of topology-aware modeling in achieving robust and accurate classification without source data.
Chinese Translation
领域适应技术已推动跨场景高光谱图像分类的发展，显著提高了在复杂场景中的区分能力。然而，隐私法规或存储限制常常阻碍对源领域数据的访问。传统的领域适应方法变得不切实际，严重限制了其在现实遥感场景中的应用。为了解决这一挑战，我们提出了一种基于拓扑感知的无源学习框架。我们首先引入熵动量伪标签（Entropy Momentum Pseudo-labeling, EMP），通过利用熵感知的置信度和时间预测动量来优化k-means分配。在经过优化的伪标签指导下，我们进一步利用上下文邻域拓扑（Contextual Neighborhood Topology, CNT）来挖掘目标特征空间的内在几何结构。CNT结合了通过协同表示提取的全局结构信息与通过最近邻搜索建模的局部相似性信息，实现了对目标领域特征空间中流形级几何属性的全面编码。整体目标整合了对优化伪标签的交叉熵、基于对数内积的拓扑一致性和信息最大化项，以确保在无源设置下的平衡分类和稳定适应。在三个典型跨场景的广泛实验中，所提方法超越了现有的最先进性能，消融研究进一步验证了各模块的贡献。结果凸显了拓扑感知建模在实现无源数据情况下的稳健和准确分类中的关键作用。
cs.CV / 79 / 2608.05976
Diff-VF: Training-free High-quality Long Video Generation via Diffusion Model
Diff-VF：无训练高质量长视频生成的扩散模型
Abstract
Recently, diffusion models have made great progress in video generation. However, most existing video diffusion models are trained with short videos, and degrade when extrapolated to long videos, struggling to maintain long-range temporal coherence while retaining diverse motions. To generate consistent, high-quality and dynamic long videos, we propose Diff-VF, a training-free, plug-and-play and model-agnostic framework that converts existing short-video diffusion backbones into long-video generators without modifying or fine-tuning the base model. Diff-VF couples three complementary strategies: Hybrid Noise Initialization (HNI) to constrain global semantics, Weighted Window Sampling (WWS) to remove inter-window discontinuities, and Temporal Extended Sampling (TES) to establish long-range dependencies with a timestep-varying fusion. We further extend Diff-VF to long-video enhancement via Skip Residual Guidance that balances fidelity and realism through timestep-dependent guidance. VBench-Long evaluation results show that Diff-VF achieves a more favorable balance between temporal coherence and motion diversity than base models and recent training-free long video generation baselines, including FreeNoise, FreeLong, and RIFLEx, while maintaining competitive frame-wise quality. Experiments on two base models demonstrate the applicability to video diffusion models with different spatial-temporal modeling strategies. Extensive ablations validate the contribution of each component and hyperparameters.
Chinese Translation
近年来，扩散模型在视频生成方面取得了重大进展。然而，大多数现有的视频扩散模型是通过短视频训练的，当外推到长视频时性能下降，难以在保持多样化运动的同时维持长时间的时间一致性。为了生成一致、高质量且动态的长视频，我们提出了Diff-VF，这是一种无训练、即插即用且与模型无关的框架，可以将现有的短视频扩散骨干网络转换为长视频生成器，而无需修改或微调基础模型。Diff-VF结合了三种互补策略：混合噪声初始化（Hybrid Noise Initialization, HNI）以约束全局语义，加权窗口采样（Weighted Window Sampling, WWS）以消除窗口间的不连续性，以及时间扩展采样（Temporal Extended Sampling, TES）以通过时间步变化的融合建立长范围依赖关系。我们进一步通过跳跃残差引导（Skip Residual Guidance）扩展Diff-VF以增强长视频，通过时间步依赖的引导平衡保真度和真实感。VBench-Long评估结果表明，Diff-VF在时间一致性和运动多样性之间实现了比基础模型和最近的无训练长视频生成基线（包括FreeNoise、FreeLong和RIFLEx）更有利的平衡，同时保持竞争性的逐帧质量。在两个基础模型上的实验验证了其对具有不同时空建模策略的视频扩散模型的适用性。大量消融实验验证了每个组件和超参数的贡献。
cs.CV / 80 / 2608.05979
Multi-Year Geospatial Reasoning using Interannually-Consistent Historical Predictions as a Free Input Modality
利用年度一致的历史预测作为自由输入模式进行多年度地理空间推理
Abstract
Machine learning, and deep networks in particular, are increasingly used to derive higher-level Earth observation (EO) products such as annual land-cover and crop-type maps. Many are generated operationally: each year a new acquisition is processed, typically with the same model, extending a multi-year archive. In the process these systems accumulate two kinds of useful signal that are almost never fed back into the model: the system's own archive of past predictions, and ancillary layers produced by other partners in a processing consortium. Both are normally used outside the network, as rule-based post-processing or a fixed input mask. Using the Copernicus Land Monitoring Service High Resolution Layer (HRL) Croplands crop-type product as a testbed, we show that bringing both signals inside the model turns a single-year, single-task pixel classifier into one that reasons across years. We introduce a Crop Type (CTY) embedding encoder that represents each past prediction as a confidence-scaled, time-ordered categorical token and attends over the year axis, and we study how the externally provided Base Vegetation Layer (BVL) mask should be represented in the model's inputs and outputs. To compare designs fairly when they relabel non-crop pixels, we evaluate on the 18 crop classes only and report precision and recall separately. On a pan-European dataset of about 5.4M labelled pixels, adding the prediction history raises crop-only F1 by 1.6 percentage points (pp) and, more importantly, corrects a recall-skewed error profile, with the largest gains on perennial and tree crops (olives +4.6, fruits +3.7, nuts +3.2 pp). Representing the BVL mask consistently in both the history and the target year adds about 2.5 pp on the crop classes. The approach is a low-cost recipe for any recurring geospatial or foundation model that emits class maps.
Chinese Translation
机器学习，特别是深度网络，越来越多地被用于推导更高层次的地球观测（EO）产品，如年度土地覆盖和作物类型地图。许多产品是以操作方式生成的：每年处理新的获取数据，通常使用相同的模型，扩展多年度档案。在这个过程中，这些系统积累了两种几乎从未反馈到模型中的有用信号：系统自身的过去预测档案，以及由其他合作伙伴在处理联盟中生成的辅助层。这两者通常在网络外部使用，作为基于规则的后处理或固定输入掩模。以Copernicus土地监测服务高分辨率层（HRL）作物类型产品为测试平台，我们展示了将这两种信号引入模型如何将单年度、单任务的像素分类器转变为跨年度推理的模型。我们引入了一种作物类型（CTY）嵌入编码器，将每个过去的预测表示为信心缩放的、时间排序的分类标记，并在年度轴上进行关注。我们研究了外部提供的基础植被层（BVL）掩模应如何在模型的输入和输出中表示。为了公平比较在重新标记非作物像素时的设计，我们仅在18个作物类别上进行评估，并分别报告精确度和召回率。在一个约540万标记像素的泛欧洲数据集中，添加预测历史使得仅作物的F1分数提高了1.6个百分点（pp），更重要的是，纠正了召回率偏向的错误特征，在多年生作物和树木作物上获得了最大的增益（橄榄 +4.6，水果 +3.7，坚果 +3.2 pp）。在历史和目标年度中一致地表示BVL掩模使作物类别增加了约2.5 pp。这种方法为任何发出类别地图的重复地理空间或基础模型提供了一种低成本的方案。
cs.CV / 81 / 2608.05983
Universal Concept Disruption for SAM3 Image Segmentation
针对SAM3图像分割的通用概念干扰
Abstract
SAM3 extends promptable segmentation from geometry-driven mask prediction to open-vocabulary concept segmentation, where a text-conditioned grounding model decides whether a concept is present and segments all matching instances. While this presence-gated design improves concept-level prediction, its adversarial robustness remains unexplored. In this paper, we introduce Universal Concept Disruption (UCD), the first universal cross-concept adversarial attack tailored to SAM3 image segmentation. UCD learns a single bounded image perturbation from (image, noun-phrase) pairs and attacks SAM3 as an integrated concept-grounding system. It jointly disrupts the text-conditioned input path, maximizes divergence in prompt-shared visual features, suppresses the final presence-gated concept scores, and corrupts the spatial validity of retained masks through area collapse and clean-mask Dice disruption. Across SACo-Gold, LVIS, RefCOCO, PhraseCut, and OpenImages datasets, UCD consistently outperforms all baselines under a matched evaluation protocol, reducing average mask AP from 59.43 to 18.73 and average cgF1 from 50.32 to 20.49. The learned perturbation also transfers to SAM3.1 and to SAM3 video inference without re-optimization, while prompt ensembling, lightweight head fine-tuning, and temporal filtering provide limited recovery.
Chinese Translation
SAM3将可提示分割从基于几何的掩码预测扩展到开放词汇概念分割，其中一个基于文本的定位模型决定某个概念是否存在并分割所有匹配的实例。尽管这种基于存在的设计提高了概念级预测，但其对抗鲁棒性仍未得到探索。本文介绍了通用概念干扰（Universal Concept Disruption, UCD），这是首个针对SAM3图像分割量身定制的通用跨概念对抗攻击。UCD从（图像，名词短语）对中学习一个单一的有界图像扰动，并将SAM3视为一个集成的概念定位系统进行攻击。它共同干扰基于文本的输入路径，最大化提示共享视觉特征的差异，抑制最终的存在门控概念得分，并通过区域崩溃和干净掩码Dice干扰破坏保留掩码的空间有效性。在SACo-Gold、LVIS、RefCOCO、PhraseCut和OpenImages数据集上，UCD在匹配评估协议下始终优于所有基线，将平均掩码AP从59.43降低到18.73，平均cgF1从50.32降低到20.49。学习到的扰动还可以在不重新优化的情况下转移到SAM3.1和SAM3视频推理中，而提示集成、轻量级头微调和时间过滤提供的恢复效果有限。
cs.CV / 82 / 2608.06009
Wan-Animate-2: Pushing the Application Boundaries of Character Animation
Wan-Animate-2：推动角色动画应用边界
Abstract
Character image animation remains a foundational yet challenging task in computer vision. Existing approaches can be broadly categorized into three paradigms: methods based on explicit motion representations suffer from extraction errors and identity drift; methods based on implicit motion features lose fine-grained dynamics through compression; and in-context learning approaches avoid intermediate representations but incur prohibitive computational costs. Furthermore, all current systems are designed for offline synthesis, unable to meet the real-time requirements of interactive applications such as digital avatars and live-streaming hosts. To address these limitations, we present Wan-Animate-2, an end-to-end character animation framework that directly consumes the driving video within a redesigned Diffusion Transformer. Our architecture achieves superior motion fidelity and identity preservation by eliminating intermediate motion extractors entirely. We further introduce text driven viewpoint control that decouples the output camera perspective from the driving video--a capability rarely supported by prior character animation methods that rely on explicit motion representations. Beyond generation quality, we present Wan-Animate-2-Lite, an efficient variant that reduces inference latency to real-time thresholds through a three-stage training paradigm: teacher forcing pretraining with error buffer mechanism, and Self-Forcing distillation with chunk-wise backpropagation. This enables streaming character animation for interactive applications, opening new deployment scenarios that were previously infeasible. Qualitative evaluations and user studies demonstrate that Wan-Animate-2 achieves high-fidelity animation results across diverse characters and motion patterns. To foster further research and community development, we will release the Wan-Animate-2-Base model weights to the public.
Chinese Translation
角色图像动画仍然是计算机视觉中的一项基础但具有挑战性的任务。现有方法大致可以分为三种范式：基于显式运动表示的方法存在提取错误和身份漂移的问题；基于隐式运动特征的方法通过压缩丧失了细粒度的动态信息；而上下文学习方法避免了中间表示，但导致了高昂的计算成本。此外，当前所有系统均为离线合成，无法满足数字化身和直播主持人等交互应用的实时需求。为了解决这些局限性，我们提出了Wan-Animate-2，一个端到端的角色动画框架，直接在重新设计的扩散变换器（Diffusion Transformer）中处理驱动视频。我们的架构通过完全消除中间运动提取器，实现了卓越的运动保真度和身份保持。我们进一步引入了文本驱动的视角控制，解耦了输出摄像机视角与驱动视频的关系——这一能力在依赖显式运动表示的先前角色动画方法中很少得到支持。除了生成质量，我们还提出了Wan-Animate-2-Lite，一个高效的变体，通过三阶段训练范式将推理延迟降低到实时阈值：教师强制预训练与错误缓冲机制，以及带有分块反向传播的自我强制蒸馏。这使得交互应用中的流式角色动画成为可能，开启了以前不可行的新部署场景。定性评估和用户研究表明，Wan-Animate-2在多样化角色和运动模式下实现了高保真动画效果。为了促进进一步的研究和社区发展，我们将向公众发布Wan-Animate-2-Base模型权重。
cs.CV / 83 / 2608.06014
Iterate or Widen? When Test-Time Refinement Helps LiDAR Scene Completion: A Controlled Study of Evidence Geometry, Training Coverage, and Compute
迭代还是扩展？测试时细化何时有助于LiDAR场景补全：证据几何、训练覆盖和计算的控制研究
Abstract
Should a completion model spend extra test-time compute by iterating, or spend a similar parameter budget on a wider one-shot predictor? The answer is easily confounded by denoising curricula, corruption augmentation, capacity, and unpaired evaluation. We study this question in LiDAR semantic scene completion by comparing a one-shot predictor, a parameter-matched wider predictor, and a weight-tied multigrid refiner initialized from the same frozen predictor. The protocol separates coherent region removal, independent thinning, range-dependent attenuation, and additive clutter while preserving exact scene-condition pairing. Across five training seeds and 815 SemanticKITTI sequence-08 frames, the full iterative system improves mIoU over the wide control by 0.911 points under contiguous angular removal, with a 95% moving-block bootstrap interval of [0.804, 1.040] that clears a predeclared 0.5-point practical margin. Under independent 75% thinning, iteration adds only 0.300 points [0.166, 0.436], whereas observation-family augmentation adds 5.975 points [5.662, 6.140]. Neither intervention repairs additive clutter. The iterative system also costs 10.74 ms and 0.75 GiB per frame, versus 6.25 ms and 0.23 GiB for the wide control. These results establish a geometry-conditioned empirical boundary rather than a universal advantage: coherent gaps can justify fixed-depth refinement, broadly thinned evidence is addressed more effectively by training coverage, and spurious evidence requires a different robustness mechanism.
Chinese Translation
补全模型是应该通过迭代来花费额外的测试时间计算，还是在一个更广泛的一次性预测器上花费相似的参数预算？这个问题容易受到去噪课程、腐蚀增强、容量和无配对评估的干扰。我们通过比较一个一次性预测器、一个参数匹配的更广泛预测器和一个从相同冻结预测器初始化的权重绑定多网格细化器，研究了LiDAR语义场景补全中的这一问题。该协议分离了连贯区域去除、独立稀疏、范围依赖衰减和附加杂乱，同时保持精确的场景条件配对。在五个训练种子和815个SemanticKITTI序列-08帧中，完整的迭代系统在连续角度去除下，相较于广泛控制提高了0.911点的mIoU，95%的移动块自助区间为[0.804, 1.040]，超出了预先声明的0.5点实际边际。在独立的75%稀疏下，迭代仅增加了0.300点[0.166, 0.436]，而观察家族增强则增加了5.975点[5.662, 6.140]。两种干预都无法修复附加杂乱。迭代系统每帧的成本为10.74毫秒和0.75 GiB，而广泛控制为6.25毫秒和0.23 GiB。这些结果建立了一个几何条件的经验边界，而不是普遍优势：连贯的间隙可以证明固定深度细化的合理性，而广泛稀疏的证据则更有效地通过训练覆盖来处理，虚假的证据需要不同的鲁棒性机制。
cs.CV / 84 / 2608.06030
PaCoNet: Deep Data Extraction for Parallel Coordinates
PaCoNet：用于平行坐标的深度数据提取
Abstract
Extracting data from visualizations has long challenged computer vision, with current research focused on bar, line, and pie charts, among other low-dimensional visualizations. However, parallel coordinates as a widely used high-dimensional data visualization approach, remain largely unexplored in this context. As parallel coordinate plots can quickly become cluttered and difficult to interpret when poorly designed or densely populated, automated data extraction from such visualizations is of particular interest. In this paper, we propose PaCoNet, the first approach for parallel coordinate data extraction. PaCoNet not only extracts line coordinates, but also enables the extraction of individual data samples for further analysis. Towards this end, we make the following contributions. We present the first deep learning approach tailored for parallel coordinate analysis, and demonstrate that it outperforms unadapted baselines by a significant margin. We further introduce a large-scale parallel coordinate dataset for training and testing. Together, these key contributions enable for the first time the automated analysis and redesign of parallel coordinate plots. PaCoNet thus lays the groundwork for complex visualization analysis, and further advances the intersection of computer vision and data visualization. All code, trained models, and data generation scripts will be made publicly available upon acceptance of the paper.
Chinese Translation
从可视化中提取数据长期以来一直是计算机视觉面临的挑战，目前的研究主要集中在条形图、折线图和饼图等低维可视化上。然而，作为一种广泛使用的高维数据可视化方法，平行坐标在这一背景下仍然未得到充分探索。由于平行坐标图在设计不当或数据密集时可能迅速变得杂乱且难以解释，因此从这种可视化中自动提取数据尤为重要。本文提出了PaCoNet，这是首个用于平行坐标数据提取的方法。PaCoNet不仅提取线坐标，还支持提取单个数据样本以便进行进一步分析。为此，我们做出了以下贡献：我们提出了首个针对平行坐标分析的深度学习方法，并证明其显著优于未适配的基线模型。此外，我们还引入了一个大规模的平行坐标数据集用于训练和测试。这些关键贡献首次实现了平行坐标图的自动分析和重新设计。因此，PaCoNet为复杂可视化分析奠定了基础，并进一步推动了计算机视觉与数据可视化的交叉发展。所有代码、训练模型和数据生成脚本将在论文接受后公开发布。
cs.CV / 85 / 2608.06059
DARAD: Dual Adapters and Ranking-Aware Distillation for Continual Remote Sensing Image-Text Retrieval
DARAD：双适配器与排名感知蒸馏用于持续遥感图像-文本检索
Abstract
With the rapid growth of Earth observation technologies, remote sensing archives are rapidly expanding, making remote sensing image-text retrieval (RS-ITR) increasingly important. However, continual RS-ITR remains challenging because scale variation and distribution shifts in RS aggravate cross-modal alignment space distortion, making it difficult for existing continual learning (CL) methods to support reliable continual retrieval. To address this challenge, we propose DARAD, a dual-adapter and ranking-aware distillation framework that preserves the historical cross-modal ranking structure while learning new visual and textual concepts from evolving archives. Specifically, the visual branch introduces a spatial fusion adapter, which integrates coarse regional cues and fine-grained patch cues to accommodate RS scale variation while anchoring visual updates to the pretrained alignment space. The textual branch employs multi-expert semantic routing, which separates shared textual semantics from semantically specialized residuals to absorb newly emerging descriptions while constraining global text embedding drift. Furthermore, bidirectional ranking distillation uses a frozen teacher model and historical anchors to preserve the historical cross-modal ranking structure, thereby mitigating alignment space distortion across continual stages. Experiments under a multi-stage continual retrieval protocol show that DARAD achieves superior performance over existing CL methods, improving adaptation to newly arrived data while maintaining effectiveness on historical data.
Chinese Translation
随着地球观测技术的快速发展，遥感档案迅速扩展，使得遥感图像-文本检索（RS-ITR）变得愈加重要。然而，持续的RS-ITR仍然面临挑战，因为遥感中的尺度变化和分布转移加剧了跨模态对齐空间的扭曲，使得现有的持续学习（CL）方法难以支持可靠的持续检索。为了解决这一挑战，我们提出了DARAD，一个双适配器和排名感知蒸馏框架，旨在在学习来自不断演变的档案的新视觉和文本概念的同时，保留历史跨模态排名结构。具体而言，视觉分支引入了空间融合适配器，该适配器整合了粗略区域线索和细粒度补丁线索，以适应遥感尺度变化，同时将视觉更新锚定到预训练的对齐空间。文本分支采用多专家语义路由，将共享的文本语义与语义专门化的残差分开，以吸收新出现的描述，同时限制全局文本嵌入的漂移。此外，双向排名蒸馏使用冻结的教师模型和历史锚点来保留历史跨模态排名结构，从而减轻跨持续阶段的对齐空间扭曲。在多阶段持续检索协议下的实验表明，DARAD在性能上优于现有的CL方法，改善了对新到数据的适应性，同时保持了对历史数据的有效性。
cs.CV / 86 / 2608.06060
Learning from Failures: Retrieval-Centric CoT via Hard Negatives for Unified Multimodal Retrieval
从失败中学习：基于困难负样本的检索中心链式思维用于统一多模态检索
Abstract
Unified multimodal retrieval aims to identify candidates that satisfy complex user intent expressed through heterogeneous inputs. Although Large Vision-Language Model (LVLM)-based retrievers are efficient and scalable, directly encoding raw multimodal inputs often misses fine-grained discriminative cues, leading to confusion among semantically similar candidates. Recent methods mitigate this limitation by generating Chain-of-Thought (CoT) rationales to enrich the query representation. However, such reasoning is typically derived from the query alone: it explains what the query describes, but not what the retriever misunderstands. We argue that effective retrieval reasoning should instead be conditioned on retrieval feedback. Based on this insight, we introduce UniME-R1, an embedder-adviser framework that learns to reason over initially retrieved candidates and generate Retrieval-Centric Chain-of-Thought (RC-CoT). The adviser analyzes candidates individually to identify the discriminative cues confused by the embedder. If the target appears in the initial top-k set, UniME-R1 directly reranks the candidates; otherwise, it generates RC-CoT to refine the retrieval direction and performs full-corpus re-retrieval with a dual-mode embedder. To train the framework, we mine hard negatives to simulate realistic retrieval failures, jointly optimize direct retrieval and RC-CoT-augmented retrieval, and align the adviser with retrieval outcomes through supervised learning and retrieval-oriented reinforcement learning. Extensive experiments on MMEB-V2 and a diverse set of general multimodal retrieval benchmarks demonstrate that UniME-R1 consistently improves retrieval performance over strong baselines.
Chinese Translation
统一多模态检索旨在识别满足通过异构输入表达的复杂用户意图的候选项。尽管基于大型视觉语言模型（Large Vision-Language Model, LVLM）的检索器高效且可扩展，但直接编码原始多模态输入往往会遗漏细粒度的区分线索，导致语义相似的候选项之间产生混淆。最近的方法通过生成链式思维（Chain-of-Thought, CoT）推理来丰富查询表示，从而缓解这一限制。然而，这种推理通常仅基于查询本身：它解释了查询所描述的内容，但并未说明检索器的误解。我们认为，有效的检索推理应当以检索反馈为条件。基于这一见解，我们提出了UniME-R1，一个嵌入-顾问框架，学习对最初检索到的候选项进行推理，并生成检索中心链式思维（Retrieval-Centric Chain-of-Thought, RC-CoT）。顾问逐个分析候选项，以识别嵌入器混淆的区分线索。如果目标出现在初始的前k个候选集中，UniME-R1将直接对候选项进行重新排序；否则，它将生成RC-CoT以细化检索方向，并使用双模嵌入器进行全库重新检索。为了训练该框架，我们挖掘困难负样本以模拟真实的检索失败，联合优化直接检索和RC-CoT增强检索，并通过监督学习和以检索为导向的强化学习将顾问与检索结果对齐。在MMEB-V2和一系列多样化的通用多模态检索基准上的大量实验表明，UniME-R1在强基线之上始终提高了检索性能。
cs.CV / 87 / 2608.06062
Bar-JEPA: Extracting Values from Bar Chart with Joint-Embedding Predictive Architecture
Bar-JEPA：利用联合嵌入预测架构从条形图中提取数值
Abstract
Bar charts are commonly used in data visualization, and while they are easily understood by humans, it is non-trivial to extract the underlying data computationally. For a machine-learning-based approach, training chart de-rendering models usually requires labeled, real-world data. Labeling data is a time consuming task, which is why annotated data is scarce. Models can learn more efficiently when provided with features of high semantic quality, which a joint-embedding predictive architecture (JEPA) is designed to learn in a self-supervised manner. We present a per-bar, numerical value recovery pipeline for bar charts, where a JEPA encoder is used to produce semantically rich latent features. The decoder model consuming these features is simple and quick to train and outputs the coordinates of ticks and bars, which can be used to recover bar values. The effectiveness of self-supervised finetuning and quality of the extracted features is evident when comparing our model to end-to-end supervised baselines. Code, datasets and checkpoints are available on \href{https://github.com/dralois/Bar-JEPA}{GitHub}.
Chinese Translation
条形图在数据可视化中被广泛使用，尽管人类容易理解它们，但计算机提取其底层数据并非易事。基于机器学习的方法通常需要标注的真实数据来训练图表去渲染模型。标注数据是一项耗时的任务，这也是为什么带注释的数据稀缺。模型在提供高语义质量特征时可以更有效地学习，而联合嵌入预测架构（JEPA）旨在以自监督的方式学习这些特征。我们提出了一种针对条形图的逐条数值恢复管道，其中使用JEPA编码器生成语义丰富的潜在特征。使用这些特征的解码模型简单且训练迅速，输出刻度和条形的坐标，这些坐标可用于恢复条形值。当将我们的模型与端到端的监督基线进行比较时，自监督微调的有效性和提取特征的质量显而易见。代码、数据集和检查点可在 exttt{https://github.com/dralois/Bar-JEPA} 上获取。
cs.CV / 88 / 2608.06065
The Next Screenshot Knows: Gated Hindsight Distillation for Mobile GUI Agents
下一个截图知道：用于移动 GUI 代理的门控回顾蒸馏
Abstract
GUI agents are commonly trained offline from successful interaction trajectories. Standard training decomposes each trajectory into prefix-action pairs: the agent predicts an action from the current screen and interaction history, while the subsequent observation is discarded. This removes the rationale of why an action is correct: the evidence often appears only on the subsequent screen. For example, to enable Soft Wrap, the agent should click Edit or View, but nothing reveals this until the menu opens. Without such evidence, standard imitation gives the model little chance of ever sampling and thus learning the correct reasoning. To address this issue, we propose Gated Hindsight Distillation (GHD), which uses the next screenshot as privileged information during training. A student predicts from the observable trajectory prefix, while a parameter-sharing teacher additionally observes the next screenshot and re-scores the student's on-policy responses. We apply distillation only when the student fails and the hindsight-conditioned teacher recovers the demonstrated action. GHD improves task success over GRPO on AndroidWorld and AndroidLab across two vision-language models. The code and checkpoints will be made available.
Chinese Translation
GUI 代理通常从成功的交互轨迹中离线训练。标准训练将每个轨迹分解为前缀-动作对：代理根据当前屏幕和交互历史预测一个动作，而后续观察则被丢弃。这消除了动作正确性的理由：证据往往只出现在后续屏幕上。例如，为了启用软换行，代理应该点击编辑或查看，但在菜单打开之前没有任何信息揭示这一点。在没有这种证据的情况下，标准模仿给模型提供了很少的机会去采样，从而学习正确的推理。为了解决这个问题，我们提出了门控回顾蒸馏（Gated Hindsight Distillation, GHD），在训练过程中使用下一个截图作为特权信息。学生从可观察的轨迹前缀进行预测，而共享参数的教师则额外观察下一个截图并重新评分学生的在线响应。我们仅在学生失败时应用蒸馏，而回顾条件的教师则恢复演示的动作。GHD 在 AndroidWorld 和 AndroidLab 上的任务成功率优于 GRPO，适用于两种视觉-语言模型。代码和检查点将会公开。
cs.CV / 89 / 2608.06075
Domain-Grounded Candidate Selection for Agentic Image Editing: A Shadow Removal Case
基于领域的候选选择用于主动图像编辑：以阴影去除为例
Abstract
Commercial vision-language models are reshaping computer vision, with visual priors broad enough to rival task-specific systems. This raises a natural question: do they reduce the need for classic, physics-informed low-level vision? We study this through shadow removal, a problem shaped by scene geometry, illumination, materials, and occluders, where paired shadow and shadow-free data are hard to collect at scale. We find that a commercial generative editor, used directly, can produce clean shadow-free edits that preserve surface texture and local appearance. However, this comes with a new failure mode: the same editor can regenerate scene content, hallucinate objects, or misread a shadow as material or geometry, producing plausible but physically wrong edits. We address this with an agentic candidate-selection pipeline: the editor generates a guided probe, an evaluator screens for major failures, retries when needed, samples multiple candidates, filters them, and selects a final result balancing shadow removal against scene preservation. Grounding this process in shadow-formation physics makes it more reliable: prompting the generator and evaluator to treat shadows as illumination effects caused by light occlusion, not material or object structure, measurably improves quality and consistency. On the ShadowRemovalRefine benchmark, our physics-oriented pipeline achieves a CDD of 0.0075, reducing CDD by at least 47% over the strongest prior method. These results suggest that commercial vision-language models do not replace classic low-level vision priors; instead, such priors remain useful for constraining and steering physically underconstrained generation.
Chinese Translation
商业视觉-语言模型正在重塑计算机视觉，其视觉先验足够广泛，可以与特定任务系统相媲美。这引发了一个自然的问题：它们是否减少了对经典物理信息低级视觉的需求？我们通过阴影去除这一问题进行研究，该问题受场景几何、照明、材料和遮挡物的影响，而成对的阴影和无阴影数据在规模上难以收集。我们发现，直接使用商业生成编辑器可以生成干净的无阴影编辑，保留表面纹理和局部外观。然而，这带来了新的失败模式：同一编辑器可能会重新生成场景内容、幻觉对象或将阴影误读为材料或几何形状，产生看似合理但在物理上错误的编辑。我们通过一个主动候选选择管道来解决这个问题：编辑器生成一个引导探针，评估器筛选主要失败，必要时重试，采样多个候选，进行过滤，并选择一个最终结果，以平衡阴影去除与场景保留。将这一过程基于阴影形成物理学使其更可靠：促使生成器和评估器将阴影视为由光遮挡引起的照明效应，而不是材料或物体结构，显著提高了质量和一致性。在ShadowRemovalRefine基准上，我们的物理导向管道实现了0.0075的CDD，相较于最强的先前方法至少减少了47%的CDD。这些结果表明，商业视觉-语言模型并未取代经典的低级视觉先验；相反，这些先验仍然对于约束和引导物理上欠约束的生成是有用的。
cs.CV / 90 / 2608.06082
Dense-Cast: A lightweight ensemble of deep learning architectures for precipitation nowcasting
Dense-Cast：一种轻量级深度学习架构集成用于降水短期预报
Abstract
Proper short-term forecasting of precipitation is crucial in disaster management and preparedness. Nonetheless, the variability and nonlinearity of precipitation make short-term forecasting challenging for meteorologists. Moreover, capturing temporal dependencies in spatiotemporal data is a challenge in precipitation nowcasting. In this article, we introduce a lightweight deep learning model for half-hourly precipitation nowcasting. This model has been designed by incorporating the DenseNet architecture, residual connections, and transformer encoders for effective precipitation nowcasting with reduced model parameters. The North-Eastern region of India has been selected as the area of interest for our study. The region receives the highest precipitation during the months of June-September due to the monsoon season. The proposed model takes the previous five time-steps of half-hourly precipitation as inputs and predicts the precipitation in the next two half-hours. The GPM IMERG precipitation dataset with a 30-minute cadence has been used in this study for training and testing the model. The proposed architecture achieves best MAE of 0.235 millimetres, RMSE of 0.735 millimetres, and KGE score of 0.816 at an interval of 30 minutes.
Chinese Translation
降水的短期预报在灾害管理和应急准备中至关重要。然而，降水的变异性和非线性使得气象学家在进行短期预报时面临挑战。此外，在时空数据中捕捉时间依赖性也是降水短期预报的一大难题。本文介绍了一种轻量级的深度学习模型，用于半小时降水的短期预报。该模型通过结合DenseNet架构、残差连接和变换器编码器，旨在有效进行降水短期预报，同时减少模型参数。我们选择印度东北地区作为研究对象，该地区在6月至9月的季风季节降水量最高。所提模型以之前五个半小时的降水数据作为输入，预测接下来的两个半小时降水量。本研究使用了GPM IMERG降水数据集，具有30分钟的时间间隔，用于模型的训练和测试。所提架构在30分钟的时间间隔内达到了最佳的平均绝对误差（MAE）为0.235毫米，均方根误差（RMSE）为0.735毫米，以及KGE评分为0.816。
cs.CV / 91 / 2608.06117
Confidence matters: Leveraging Multi-view Geometric Priors for GS-based Reconstruction
信心至关重要：利用多视角几何先验进行基于高斯的重建
Abstract
3D Gaussian splatting (3DGS) has emerged as a widely-used tool for novel view synthesis, offering real-time rendering in a sparse representation. However, the method's reliance on structure-from-motion initialization and photometric optimization can lead to suboptimal geometric reconstruction, particularly for objects with high specularity. In this work, we investigate the integration of geometric priors, in the form of predicted normal and depth maps, into the 3DGS framework to improve the reconstruction quality. We analyze the effect of incorporating these priors into GS-based methods and our evaluation reveals that multi-view predictions, as they are done by the recent visual geometry grounded transformer (VGGT), outperform single-view alternatives. A major factor is the existence of a confidence map for the estimations, which comes as a by-product of multi-view models and which can significantly improve the effectiveness of priors by weighting each prediction appropriately. Extensive experiments on standard benchmarks show consistent improvement in reconstruction quality and significant gains in complex scenes including specular objects.
Chinese Translation
三维高斯点云（3D Gaussian splatting, 3DGS）已成为一种广泛使用的新视图合成工具，能够在稀疏表示中实现实时渲染。然而，该方法依赖于运动结构初始化和光度优化，这可能导致几何重建的次优，尤其是在高反射物体的情况下。在本研究中，我们探讨将几何先验（以预测的法线图和深度图形式）融入3DGS框架，以提高重建质量。我们分析了将这些先验纳入基于高斯的方法的效果，评估结果表明，最近的视觉几何基础变换器（visual geometry grounded transformer, VGGT）所做的多视角预测优于单视角替代方案。一个主要因素是估计的置信度图的存在，这是多视角模型的副产品，可以通过适当地加权每个预测显著提高先验的有效性。在标准基准上的大量实验显示，重建质量持续改善，并在包括高反射物体的复杂场景中取得显著提升。
cs.CV / 92 / 2608.06125
Sample-Adaptive Latent Rewards for Uncertainty-Guided Diffusion Post-Training
基于样本自适应的潜在奖励用于不确定性引导的扩散后训练
Abstract
Latent reward models can supervise visual diffusion models without decoding intermediate states into pixel space. This makes alignment with human preferences more efficient. However, existing latent reward models output only scalar scores. They do not estimate the uncertainty of each prediction. The generator therefore cannot determine which feedback is reliable. This can drive optimization in the wrong direction and lead to reward hacking. We propose \textsc{SURE}, a unified latent-space framework for image and video diffusion models. It learns reward distributions and directly uses their reliability to guide dense post-training. First, we propose sample-adaptive latent reward model (\textsc{SURE-LRM}). It predicts a Gaussian utility for each noisy latent. Its mean predicts the reward score. Its variance reflect the uncertainty of prediction without human annotation. The learned distribution then guides post-training through uncertainty-guided reward feedback learning (\textsc{SURE-REFL}). This method provides uncertainty-guided dense feedback along the denoising trajectory. At selected transitions, \textsc{SURE-REFL} queries the frozen \textsc{SURE-LRM}. It converts detached variance into reliability weights for samples at the same transition. Each weighted reward is backpropagated only through its local transition. The entire process remains in latent space and requires neither pixel-space decoding nor the full denoising graph. Experiments show that \textsc{SURE-LRM} improves preference prediction over strong baselines. \textsc{SURE-REFL} achieves the sota performance among various metrics and further improves optimization stability. It also achieves the highest VBench quality, semantic, and total scores among the evaluated methods.
Chinese Translation
潜在奖励模型可以在不将中间状态解码为像素空间的情况下监督视觉扩散模型。这使得与人类偏好的对齐更加高效。然而，现有的潜在奖励模型仅输出标量分数，并未估计每个预测的不确定性。因此，生成器无法确定哪些反馈是可靠的。这可能导致优化朝错误的方向发展并导致奖励黑客行为。我们提出了 extsc{SURE}，一个用于图像和视频扩散模型的统一潜在空间框架。它学习奖励分布，并直接利用其可靠性来指导密集的后训练。首先，我们提出了样本自适应潜在奖励模型（ extsc{SURE-LRM}）。它为每个噪声潜在变量预测一个高斯效用，其均值预测奖励分数，方差反映预测的不确定性而无需人类注释。学习到的分布通过不确定性引导的奖励反馈学习（ extsc{SURE-REFL}）来指导后训练。该方法沿去噪轨迹提供不确定性引导的密集反馈。在选定的过渡点， extsc{SURE-REFL} 查询冻结的 extsc{SURE-LRM}，将分离的方差转换为同一过渡点样本的可靠性权重。每个加权奖励仅通过其局部过渡进行反向传播。整个过程保持在潜在空间中，无需像素空间解码或完整的去噪图。实验表明， extsc{SURE-LRM} 在偏好预测上优于强基线。 extsc{SURE-REFL} 在各种指标中实现了最先进的性能，并进一步提高了优化的稳定性。在评估的方法中，它还在 VBench 质量、语义和总分方面达到了最高分。
cs.CV / 93 / 2608.06126
Patient Pose Assessment Using a CT-Based Framework for Synthetic Data Generation
基于CT框架的患者姿态评估与合成数据生成
Abstract
An adequate diagnostic quality of radiographs is essential for reliable diagnoses and treatment planning. The patient's pose during radiography is one of the most important factors determining the diagnostic quality. Since patient positioning is difficult and not standardized, an automated AI-based approach using depth images to automatically assess the patient's pose before the radiograph has been taken would be helpful. Due to regulatory hurdles, however, it is difficult in practice to acquire the required depth images and corresponding radiographs. In this paper, we present a framework that can generate such training data synthetically from Computed Tomography scans. We further show that by pretraining on our generated synthetic dataset consisting of 3077 image pairs of upper ankle joints, the pose assessment of real upper ankle joints can be improved by up to 11 percentage points.
Chinese Translation
放射片的诊断质量对于可靠的诊断和治疗计划至关重要。患者在放射检查期间的姿态是决定诊断质量的最重要因素之一。由于患者定位困难且缺乏标准化，采用基于人工智能的自动化方法，通过深度图像在放射片拍摄前自动评估患者姿态将是非常有帮助的。然而，由于监管障碍，实际获取所需的深度图像和相应的放射片非常困难。本文提出了一种框架，可以从计算机断层扫描（Computed Tomography）图像中合成生成这种训练数据。我们进一步展示，通过在我们生成的合成数据集上进行预训练，该数据集包含3077对上踝关节的图像对，可以将真实上踝关节的姿态评估提高多达11个百分点。
cs.CV / 94 / 2608.06142
Learning visual representations for compositional analysis of artworks and photographs
用于艺术作品和照片的组合分析的视觉表征学习
Abstract
Composition, the deliberate arrangement of visual elements, is central to how meaning, emotion, and aesthetic quality are conveyed in artwork, yet it remains among the least formalized dimensions of visual understanding. Prior work highlights a persistent gap in learning meaningful compositional representations, attributing it to semantic bias and suggesting that human-inspired approaches may be key. We compare two parallel paradigms for composition analysis: a human-inspired method grounded in perceptual grouping, and fine-tuned foundation models enabled by recent large-scale compositional datasets. The human-inspired approach uses object-centric models for region-level decomposition and a graph attention network to capture spatial relationships between elements. Both paradigms are evaluated on composition score/category prediction, compositional image retrieval, and visual saliency detection. With frozen encoders, the human-inspired method achieves competitive performance while remaining interpretable. When sufficient data enables fine-tuning, large self-supervised models outperform significantly, but at the cost of interpretability and cross-domain generalization.
Chinese Translation
构图，即视觉元素的有意排列，是艺术作品中传达意义、情感和美学质量的核心，但它仍然是视觉理解中最不规范化的维度之一。先前的研究强调了在学习有意义的构图表征方面存在的持续差距，认为这与语义偏差有关，并建议人类启发的方法可能是关键。我们比较了两种平行的构图分析范式：一种基于感知分组的人类启发方法，以及由近期大规模构图数据集支持的微调基础模型。人类启发的方法使用以对象为中心的模型进行区域级分解，并利用图注意力网络捕捉元素之间的空间关系。两种范式在构图评分/类别预测、构图图像检索和视觉显著性检测上进行了评估。在冻结编码器的情况下，人类启发的方法在保持可解释性的同时实现了竞争力的性能。当足够的数据支持微调时，大型自监督模型的表现显著优于其他方法，但代价是可解释性和跨领域泛化能力的降低。
cs.CV / 95 / 2608.06164
BendTwin: Robust Dense-to-Sparse Physical Reconstruction with Bending-Aware Differentiable Spring-Mass Models
BendTwin：基于弯曲感知可微分弹簧-质量模型的鲁棒稠密到稀疏物理重建
Abstract
Reconstructing objects with mechanical properties from video observations enables physically consistent dynamic prediction, benefiting robotics planning and interaction. Existing spring--mass based physical driven reconstruction approaches offer efficient and differentiable physical reconstruction, but they typically rely on axial springs alone. Such formulations oversimplify the underlying structural mechanics and can become mechanically under-constrained when the physical graph is coarsened, limiting their ability to preserve stable local deformation. We present BendTwin, a bending-aware differentiable spring--mass framework for video-based reconstruction and future prediction of deformable objects. BendTwin introduces bending stiffness and damping over local surface triplets, penalizing deviations from rest angles and regularizing higher-order deformation. These bending constraints improve mechanical stability while preserving the simplicity of spring--mass system. Experiments show that BendTwin consistently outperforms the axial-only PhysTwin baseline. Ablation studies further demonstrate that the bending constraints maintain system stability across different downsampling ratios and consistently improve upon the original PhysTwin formulation. Overall, BendTwin provides an effective approach for constructing mechanically faithful digital twins from sparse-view RGB-D videos.
Chinese Translation
从视频观察中重建具有机械特性的物体能够实现物理一致的动态预测，这对机器人规划和交互具有重要意义。现有的基于弹簧-质量模型的物理驱动重建方法提供了高效且可微分的物理重建，但通常仅依赖于轴向弹簧。这种公式化过于简化了潜在的结构力学，当物理图被粗化时，可能会导致机械约束不足，从而限制了其保持稳定局部变形的能力。我们提出了BendTwin，一种基于视频的可重建和未来预测可变形物体的弯曲感知可微分弹簧-质量框架。BendTwin在局部表面三元组上引入了弯曲刚度和阻尼，惩罚偏离静止角度的行为并正则化高阶变形。这些弯曲约束提高了机械稳定性，同时保持了弹簧-质量系统的简单性。实验表明，BendTwin在性能上始终优于仅使用轴向弹簧的PhysTwin基线。消融研究进一步表明，弯曲约束在不同下采样比率下保持系统稳定性，并始终改善原始PhysTwin公式化。总体而言，BendTwin为从稀疏视角的RGB-D视频中构建机械真实的数字双胞胎提供了一种有效的方法。
cs.CV / 96 / 2608.06174
Support Operation Factorization: Compositional Readout of Frozen Vision Encoders under Controlled Interventions
支持操作因子分解：在受控干预下对冻结视觉编码器的组合读出
Abstract
Compositional analysis of frozen vision encoders should determine both what changed and where it changed. Standard factor probes score these axes separately, however, and can reward multiple operations that reuse the same predicted slot. We call this failure operation laundering. We introduce an injectively aligned leave-one-cell-out protocol over support x operation grids and SO-OPF, a readout that factors cell energy into support salience and a competitive operation posterior. This formulation separates two questions that aggregate scores conflate: whether the carrier composes held-out bindings when the grid is known, and whether that grid can be recovered from flat cell labels. With frozen DINOv3 features, known factorial assignment reaches 0.874 injective accuracy on Shapes3D-Extended and 0.799 on globally image-disjoint COCO; learning the assignment from flat labels reaches 0.769 and 0.762, respectively. Under matched-axis-aware supervision on Shapes3D, the factored carrier improves learned-assignment accuracy from 0.653 to 0.841 over a dense carrier and eliminates its laundering gap. SigLIP2 replicates the COCO separation. A rebuilt MuJoCo substrate exposes a boundary: learned-assignment accuracy is 0.569 with DINOv3 and 0.484 with SigLIP2, with substantial slot collapse. Thus factored readout and injective evaluation recover held-out bindings on two substrates while exposing, rather than hiding, a renderer-specific failure boundary; they do not establish universal recovery from flat labels.
Chinese Translation
对冻结视觉编码器的组合分析应确定变化的内容和变化的位置。然而，标准因子探测器分别对这些轴进行评分，并可能奖励多个操作，这些操作重用相同的预测槽。我们称这种失败为操作洗涤。我们引入了一种在支持 x 操作网格上进行的单元逐出协议（leave-one-cell-out protocol）和 SO-OPF（支持操作因子分解），该读出将单元能量分解为支持显著性和竞争操作后验。这种形式化将两个聚合分数混淆的问题分开：当网格已知时，承载体是否组合了保留的绑定，以及该网格是否可以从平坦的单元标签中恢复。使用冻结的 DINOv3 特征，已知因子分配在 Shapes3D-Extended 上达到 0.874 的单射准确率，在全球图像不相交的 COCO 上达到 0.799；从平坦标签学习分配分别达到 0.769 和 0.762。在 Shapes3D 上进行匹配轴感知监督下，因子化承载体将学习分配的准确率从 0.653 提高到 0.841，消除了其洗涤差距。SigLIP2 复制了 COCO 的分离效果。重建的 MuJoCo 基底暴露了一个边界：使用 DINOv3 的学习分配准确率为 0.569，而使用 SigLIP2 的准确率为 0.484，且存在显著的槽崩溃。因此，因子化读出和单射评估在两个基底上恢复了保留的绑定，同时暴露了而非隐藏的渲染器特定失败边界；它们并未建立从平坦标签的普遍恢复。
cs.CV / 97 / 2608.06184
EvReflection: Event-Driven Micro-Dynamics for Reflection Removal
EvReflection：基于事件驱动的反射去除微动态
Abstract
Despite remarkable progress in reflection removal, current methods primarily exploit static image priors from a single frame and still suffer from severe residual artifacts due to the inherent ambiguity between the reflection and transmission layers. In this paper, we propose leveraging event signals to break this ambiguity. By employing event cameras to capture micro-dynamics, we reveal the differential motion between these two layers. We thereby present a novel event-driven reflection removal network, EvReflection, that utilizes these dynamic cues for layer separation. Specifically, we design a Micro-Dynamics Decoupler to disentangle layer-specific motions from event streams as priors, which then guide a Parallax-Attention Rectifier to cleanly remove artifacts from the RGB image. Furthermore, to address data scarcity, we develop a parallax-aware simulation pipeline and construct the EVR$^2$ benchmark dataset, the first real-world dataset for this task. Extensive experiments demonstrate that EvReflection achieves state-of-the-art performance on both synthetic and real-world benchmarks, surpassing the best competing method by more than 1.6 dB and 1.2 dB in PSNR, respectively. The code, dataset, and pre-trained models are available at https://github.com/JiaxiaoWang/EvReflection.
Chinese Translation
尽管反射去除技术取得了显著进展，但当前的方法主要依赖于来自单帧的静态图像先验，仍然受到反射层与传输层之间固有模糊性的严重残余伪影的影响。本文提出利用事件信号来打破这种模糊性。通过使用事件相机捕捉微动态，我们揭示了这两层之间的差异运动。因此，我们提出了一种新颖的基于事件驱动的反射去除网络EvReflection，利用这些动态线索进行层分离。具体而言，我们设计了一个微动态解耦器，从事件流中解开特定层的运动作为先验，然后引导一个视差注意力整流器干净地去除RGB图像中的伪影。此外，为了解决数据稀缺问题，我们开发了一个视差感知的模拟管道，并构建了EVR$^2$基准数据集，这是该任务的第一个真实世界数据集。大量实验表明，EvReflection在合成和真实世界基准测试中均实现了最先进的性能，分别在PSNR上超越了最佳竞争方法1.6 dB和1.2 dB。代码、数据集和预训练模型可在https://github.com/JiaxiaoWang/EvReflection获取。
cs.CV / 98 / 2608.06192
HOPE: Hand-Object Pressure Estimation from Monocular Videos
HOPE：基于单目视频的手-物体压力估计
Abstract
Estimating physical pressure from vision is essential for understanding contact-rich hand-object interaction. However, prior vision-based pressure estimation methods are largely limited to planar surfaces and single image input, making them difficult to apply to dynamic hand-object interaction with diverse objects. We instead formulate pressure estimation as a hand-centric video prediction problem with monocular video as input. This formulation predicts temporally evolving per-vertex normal pressure and contact directly on the hand mesh, yielding a unified output space independent of object shape and sensor layout. Building on this formulation, we propose \textbf{HOPE}, a framework with two key components. First, we lift tactile-glove pressure, planar-sensor pressure, and distance-based hand-object contact annotations into a shared hand vertex space, allowing bare-hand contact data to regularize pressure learning where metric labels are unavailable. Second, we introduce a vertex-anchored video transformer that treats each vertex as a persistent token, aggregates visual features and hand pose over time, and uses a contact-gated pressure head to enforce that pressure vanishes without contact. Experiments on OpenTouch, PressureVisionDB, and hand-object contact benchmarks validate HOPE across object-pressure, surface-pressure, and contact-supervised HOI settings. Despite using metric pressure supervision primarily from gloved-hand videos, HOPE generalizes to bare-hand egocentric and in-the-wild videos, producing joint contact and pressure predictions beyond the scope of contact-only or planar-pressure baselines.
Chinese Translation
从视觉中估计物理压力对于理解丰富接触的手-物体交互至关重要。然而，先前的基于视觉的压力估计方法主要局限于平面表面和单幅图像输入，使其难以应用于具有多样物体的动态手-物体交互。我们将压力估计重新构建为一个以手为中心的视频预测问题，以单目视频作为输入。这种构建方式直接在手网格上预测随时间演变的每个顶点的法向压力和接触，产生一个独立于物体形状和传感器布局的统一输出空间。在此基础上，我们提出了 extbf{HOPE}框架，包含两个关键组件。首先，我们将触觉手套压力、平面传感器压力和基于距离的手-物体接触注释提升到共享的手顶点空间，使得在缺乏度量标签的情况下，裸手接触数据能够规范化压力学习。其次，我们引入了一个顶点锚定的视频变换器，将每个顶点视为一个持久的标记，随时间聚合视觉特征和手势，并使用接触门控压力头来确保在没有接触的情况下压力消失。在OpenTouch、PressureVisionDB和手-物体接触基准上的实验验证了HOPE在物体压力、表面压力和接触监督的手-物体交互设置中的有效性。尽管主要使用来自戴手套视频的度量压力监督，HOPE仍能推广到裸手自我中心和野外视频，产生超出仅接触或平面压力基线范围的联合接触和压力预测。
cs.CV / 99 / 2608.06205
CFGPNet: Cross-Attention-Based Fused Gradient Programmed Network Framework for Multispectral Object Detection
CFGPNet：基于交叉注意力的融合梯度编程网络框架用于多光谱目标检测
Abstract
RGB--T object detection exploits the complementary strengths of visible and infrared imagery, supporting robust perception in low-light, adverse-weather, and complex multi-scale environments. However, existing methods still suffer from insufficient cross-modal interaction, unstable fusion from modality distribution gaps, and the high computational cost of heavy attention-based architectures. To address these issues, CFGPNet is proposed, a Cross-Attention-Based Fused Gradient Programmed Network framework for multispectral object detection. CFGPNet uses an improved GELAN backbone with RepViT-style re-parameterized blocks to strengthen feature representation while preserving computational efficiency. A Cross Computation Efficient Attention (CrossCEA) module is introduced to enhance cross-modal feature interaction and reduce redundant information transfer between visible and thermal branches. To generate compact and discriminative fused representations, an Attention Selection and Aggregation Fusion (ASAF) network combines dense feature aggregation with selective attention-based emphasis. Moreover, a programmable-gradient auxiliary branch is integrated into each CFGPNet variant to improve gradient delivery and optimization quality. Experiments on five public multispectral benchmarks, FLIR, M3FD, LLVIP, VEDAI, and MFAD, demonstrate that CFGPNet achieves strong and consistent performance across diverse scenes, object scales, and modality balances. In particular, the framework attains 80.7% mAP50 / 45.0% mAP50:95 on FLIR, 89.9% / 63.4% on M3FD, and 97.8% / 68.9% on LLVIP. It also reaches 83.3% / 56.9% on VEDAI and 83.4% / 61.8% on MFAD. These results show that CFGPNet is an effective, practical solution offering useful accuracy--efficiency trade-offs across three model scales. The code, data, and fine-tuned models are available at https://github.com/NimaHatami99/CFGPNet.
Chinese Translation
RGB-T目标检测利用可见光和红外图像的互补优势，支持在低光照、不良天气和复杂多尺度环境中的稳健感知。然而，现有方法仍然面临跨模态交互不足、模态分布差异导致的不稳定融合以及基于重注意力架构的高计算成本等问题。为了解决这些问题，提出了CFGPNet，一个用于多光谱目标检测的基于交叉注意力的融合梯度编程网络框架。CFGPNet使用改进的GELAN主干网络，结合RepViT风格的重新参数化模块，以增强特征表示，同时保持计算效率。引入了一种交叉计算高效注意力（Cross Computation Efficient Attention，CrossCEA）模块，以增强跨模态特征交互并减少可见光和热分支之间的冗余信息传递。为了生成紧凑且具有区分性的融合表示，注意力选择与聚合融合（Attention Selection and Aggregation Fusion，ASAF）网络将密集特征聚合与选择性注意力强调相结合。此外，每个CFGPNet变体中集成了可编程梯度辅助分支，以改善梯度传递和优化质量。在五个公共多光谱基准数据集FLIR、M3FD、LLVIP、VEDAI和MFAD上的实验表明，CFGPNet在不同场景、目标尺度和模态平衡下表现出强大且一致的性能。特别是，该框架在FLIR上达到了80.7%的mAP50 / 45.0%的mAP50:95，在M3FD上达到了89.9% / 63.4%，在LLVIP上达到了97.8% / 68.9%。在VEDAI和MFAD上分别达到了83.3% / 56.9%和83.4% / 61.8%。这些结果表明，CFGPNet是一个有效且实用的解决方案，在三个模型尺度上提供了有用的准确性与效率权衡。代码、数据和微调模型可在https://github.com/NimaHatami99/CFGPNet获取。
cs.CV / 100 / 2608.06231
EmoWorld: A Decoupled Affective Field for Controllable Emotional Video Generation
EmoWorld：可控情感视频生成的解耦情感场
Abstract
Emotion shapes how viewers interpret a scene, yet existing video generators entangle global atmosphere, affect-bearing semantic cues, and temporal progression within a single text condition. We present EmoWorld, a framework that decouples these factors within a frozen flow-matching video diffusion transformer (Video DiT). A one-time preparation stage extracts layer-specific affect directions and a reusable cue library from geometry-preserving neutral and emotion-edited panoramas. At inference, Visual Atmosphere Steering (VAS) injects atmosphere directions into hidden states, Semantic Affective Steering (SAS) isolates a separately scalable prompt residual for semantic cues, and Temporal Affective Steering (TAS) interpolates endpoint residual fields across denoising and video time. On Wan2.2, VAS improves target-emotion alignment by 19% while reducing a temporal-fluctuation proxy by 48%; SAS improves target-emotion alignment by 37% and increases detected affect-bearing cues by 36%; and TAS improves transition monotonicity by 15% over the strongest baseline. EmoWorld is evaluated across 27 emotion categories in text-to-video and image-to-video settings, demonstrates portability across multiple Video-DiT backbones, and supports camera-conditioned composition without updating generator parameters.
Chinese Translation
情感影响观众对场景的解读，然而现有的视频生成器将全球氛围、情感承载的语义线索和时间进程纠缠在单一文本条件中。我们提出了EmoWorld，一个在冻结的流匹配视频扩散变换器（Video DiT）中解耦这些因素的框架。一次性准备阶段从几何保持的中性和情感编辑全景中提取层特定的情感方向和可重用的线索库。在推理阶段，视觉氛围引导（Visual Atmosphere Steering, VAS）将氛围方向注入隐藏状态，语义情感引导（Semantic Affective Steering, SAS）隔离出可单独扩展的语义线索残差，而时间情感引导（Temporal Affective Steering, TAS）在去噪和视频时间之间插值端点残差场。在Wan2.2上，VAS将目标情感对齐度提高了19%，同时将时间波动代理减少了48%；SAS将目标情感对齐度提高了37%，并增加了36%的检测到的情感承载线索；TAS在最强基线之上提高了15%的过渡单调性。EmoWorld在文本到视频和图像到视频的设置中评估了27个情感类别，展示了在多个Video-DiT骨干网络之间的可移植性，并支持在不更新生成器参数的情况下进行相机条件组合。
cs.CV / 101 / 2608.06236
Depth-Guided Video Object Counting in Crowded Scenes
基于深度引导的拥挤场景视频目标计数
Abstract
Our primary objective is to advance video object counting in crowded scenes, aiming to robustly count all instances of a target category based on given text or visual prompts. Existing methods rely on RGB information, limiting their discriminative ability in crowded and occluded conditions. To address this, we propose a Depth-Guided Detector (DG-Det) along with a general post-processing pipeline. By integrating depth cues with multi-scale RGB-D cross-attention and explicit occlusion prediction, our method enhances spatial understanding and achieves robust detection in crowded and occluded scenes. Furthermore, we introduce a unified de-duplication framework to eliminate cross-frame redundant counting. To facilitate future research, we also release a new RGB-D Video Object Counting dataset featuring depth information and multiple object categories persequence. Extensive experiments demonstrate that our method achieves a 62.01\% reduction in MAE compared to existing baselines, and also produces consistent improvements in RMSE. We provide the source code at https://github.com/streamer-AP/DG-Net and the dataset at https://huggingface.co/datasets/aerospace123/RGBD-VideoCount.
Chinese Translation
我们的主要目标是推动拥挤场景中的视频目标计数，旨在根据给定的文本或视觉提示稳健地计数目标类别的所有实例。现有方法依赖于RGB信息，这限制了它们在拥挤和遮挡条件下的区分能力。为了解决这个问题，我们提出了一种深度引导检测器（Depth-Guided Detector, DG-Det）以及一个通用的后处理管道。通过将深度线索与多尺度RGB-D交叉注意力和显式遮挡预测相结合，我们的方法增强了空间理解能力，并在拥挤和遮挡场景中实现了稳健的检测。此外，我们引入了一个统一的去重框架，以消除跨帧的冗余计数。为了促进未来的研究，我们还发布了一个新的RGB-D视频目标计数数据集，包含深度信息和每个序列的多个目标类别。大量实验表明，我们的方法在平均绝对误差（MAE）上相比现有基线减少了62.01%，并在均方根误差（RMSE）上也取得了一致的改进。我们提供了源代码，链接为https://github.com/streamer-AP/DG-Net，数据集链接为https://huggingface.co/datasets/aerospace123/RGBD-VideoCount。
cs.CV / 102 / 2608.06240
PRISM: Distribution-Gated Flow Matching for Controllable Unpaired Image Translation
PRISM：用于可控无配对图像翻译的分布门控流匹配
Abstract
Unpaired image-to-image translation must decide, per image, what to change and what to preserve without paired supervision. Many diffusion-based unpaired translators control preservation through a single global noise or guidance value applied across the image, which cannot separate content to keep from appearance to change. We present PRISM, a GAN-free flow-matching framework that replaces this global control with a learned per-feature gate. The gate's spatial prior is derived from each source feature's standardized distance to the target feature distribution, so features far from the target are freed while target-consistent features are preserved. The same gate controls both the initialization, which mixes the real source latent with a task-matched corruption, and the transport timing during Ordinary Differential Equation (ODE) integration. The corruption is matched to the task, content-anchored (AdaIN) for structure-preserving translation and partially anchored for structure-changing translation, and the gate can be overridden locally at inference from text or a detector without retraining, preserving important structures of the original image while still generating realistic results. We evaluate PRISM on five natural and biomedical benchmarks (AFHQ cat->dog, CelebA-HQ appearance translation, day->night relighting, virtual staining, and breast frozen->permanent histopathology). Among the evaluated methods under a shared same-split protocol, PRISM attains the best Inception FID and KID on four benchmarks and a competitive result on the fifth, and on histopathology yields the nuclei-count ratio closest to the ideal, supporting a favorable balance between target realism and structural preservation.
Chinese Translation
无配对图像到图像的翻译必须在没有配对监督的情况下，针对每张图像决定改变什么和保留什么。许多基于扩散的无配对翻译器通过在整个图像上应用单一的全局噪声或引导值来控制保留，这无法将需要保留的内容与需要改变的外观分开。我们提出了PRISM，一种无GAN的流匹配框架，它用学习到的每特征门替代了这种全局控制。该门的空间先验来源于每个源特征与目标特征分布的标准化距离，因此远离目标的特征被释放，而与目标一致的特征则被保留。同一门控制初始化，该初始化将真实源潜在特征与任务匹配的干扰混合，以及在常微分方程（ODE）积分过程中的传输时机。干扰与任务匹配，对于结构保留翻译采用内容锚定（AdaIN），而对于结构改变翻译则部分锚定，且该门可以在推理时根据文本或检测器局部覆盖，而无需重新训练，从而保留原始图像的重要结构，同时仍然生成逼真的结果。我们在五个自然和生物医学基准（AFHQ猫->狗，CelebA-HQ外观翻译，白天->夜晚重光照，虚拟染色，以及乳腺冷冻->永久组织病理学）上评估了PRISM。在共享相同拆分协议下评估的方法中，PRISM在四个基准上获得了最佳的Inception FID和KID，并在第五个基准上取得了竞争性结果，在组织病理学中产生的细胞核计数比率最接近理想值，支持目标现实性与结构保留之间的良好平衡。
cs.CV / 103 / 2608.06252
Toward Deployable Bangla Sign Language Recognition with Expert-Validated Data and a Lightweight Attention-Based Model
面向可部署的孟加拉手语识别：专家验证的数据与轻量级注意力模型
Abstract
Deaf and hard-of-hearing people in Bangladesh communicate mainly through Bangla Sign Language (BdSL). Automatic BdSL recognition on personal devices could widen access to education and services. Existing systems use controlled-setting datasets without expert verification and heavyweight pretrained backbones unsuited to on-device use. We introduce RSBdSL38, 10,874 expert-validated images spanning all 38 BdSL hand signs, representing the 51 letters of the Bangla alphabet, recorded from real signers at three special-needs schools across Bangladesh. We propose a lightweight attention based convolutional network of 298,470 parameters, built from grouped bottleneck residual blocks, channel and spatial attention, a multi-scale depthwise hand-feature block, dual pooling, and Swish activations. Trained from scratch, it attains 96.37% accuracy (95.72% +- 0.54% over five seeds), within 1.08 percentage points of the best of nine ImageNet-pretrained efficient architectures under an identical protocol, using 8.5 to 68x fewer parameters and 1.3 to 21.7x fewer MACs. Retrained, it reaches 92.95 to 98.33% on six public BdSL benchmarks, 97.04% on a merged corpus, and 76.25% zero-shot on BdSL-38. Removing any architectural stage costs 7.61 to 89.30 points, against at most 3.17 for the training recipe. Grad-CAM with deletion-insertion and weight-randomization checks confirms that predictions follow the signing hand. A signer-independent split holding out 6 of 36 signers yields 85.18%. Quantized to 0.48 MB, it runs at 3.98 ms per image within a 15.5 MB footprint on a commodity smartphone. Together, RSBdSL38 and our from-scratch model turn benchmark accuracy into deployable accessibility at a fraction of pretrained-backbone cost; dataset, code, and models are released.
Chinese Translation
孟加拉国的聋人和听力障碍者主要通过孟加拉手语（BdSL）进行交流。在个人设备上实现自动化的BdSL识别可以拓宽教育和服务的获取渠道。现有系统使用未经专家验证的受控环境数据集，并依赖于不适合设备使用的重型预训练骨干网络。我们引入了RSBdSL38，这是一个包含10,874张专家验证图像的数据集，涵盖了所有38个BdSL手势，代表孟加拉字母表的51个字母，这些图像是从孟加拉国三所特殊需要学校的真实手语者录制的。我们提出了一种轻量级的基于注意力的卷积网络，参数量为298,470，采用分组瓶颈残差块、通道和空间注意力、多尺度深度手部特征块、双重池化和Swish激活函数构建。从零开始训练后，该模型在相同协议下达到了96.37%的准确率（五个种子的平均值为95.72% ± 0.54%），与九种基于ImageNet的高效架构中最佳模型相差仅1.08个百分点，且使用的参数量减少了8.5到68倍，MACs减少了1.3到21.7倍。经过重新训练，该模型在六个公共BdSL基准测试中达到了92.95%到98.33%的准确率，在合并语料库中达到了97.04%，在BdSL-38上实现了76.25%的零-shot准确率。移除任何架构阶段会导致准确率下降7.61到89.30点，而训练配方的最大下降为3.17点。Grad-CAM结合删除插入和权重随机化检查确认了预测结果与手语者的手势一致。对36名手语者中的6名进行独立划分的结果为85.18%。经过量化处理后，该模型大小为0.48 MB，在普通智能手机上以每张图像3.98毫秒的速度运行，内存占用为15.5 MB。总之，RSBdSL38和我们的从零开始模型将基准准确率转化为可部署的可及性，成本仅为预训练骨干网络的一小部分；数据集、代码和模型已发布。
cs.CV / 104 / 2608.06257
MASS: Multiplayer World Models with Authoritative Shared State
MASS：具有权威共享状态的多人世界模型
Abstract
Current video world models struggle in multiplayer environments because they entangle world state with view-dependent visual latents, leading to redundant compute, view inconsistencies, and poor scalability. We propose MAS (Multiplayer world models with Authoritative Shared State) to resolve this limitation. Inspired by multiplayer game architectures, MAS disentangles world dynamics and view rendering. A learned Logic Engine advances a global, authoritative typed state from joint actions without any hand-written transition function, acting as the sole recurrent memory and synchronization reference. From this shared state, a learned Rendering Engine generates independent and consistent views for any requested camera on demand. This explicit disentangling allows MAS to achieve superior state accuracy and lower cross-view inconsistency compared to state-of-the-art multi-view baselines on a matched multiplayer Snake benchmark. It advances predicted worlds with 1,024 concurrent players for 10,000 recurrent steps. Our results show that explicit, authoritative state modeling provides a practical foundation for scalable and consistent multi-agent world simulation.
Chinese Translation
当前的视频世界模型在多人环境中表现不佳，因为它们将世界状态与视图相关的视觉潜变量纠缠在一起，导致冗余计算、视图不一致和可扩展性差。我们提出了MAS（具有权威共享状态的多人世界模型）来解决这一局限性。MAS受到多人游戏架构的启发，解耦了世界动态与视图渲染。一个学习的逻辑引擎从联合动作中推进一个全球的、权威的类型化状态，而无需任何手写的转移函数，充当唯一的递归记忆和同步参考。从这个共享状态中，一个学习的渲染引擎根据需求为任何请求的摄像机生成独立且一致的视图。这种明确的解耦使得MAS在与最先进的多视图基准（在匹配的多人贪吃蛇基准上）相比时，能够实现更高的状态准确性和更低的视图间不一致性。它能够在10,000个递归步骤中为1,024个并发玩家推进预测的世界。我们的结果表明，明确的权威状态建模为可扩展和一致的多智能体世界模拟提供了实用的基础。
cs.CV / 105 / 2608.06264
OTLesMix: Wasserstein Barycenter and Optimal Transport Map for Synthetic Lesion Generation with Diverse Shapes and Locations
OTLesMix：用于合成不同形状和位置病变生成的Wasserstein重心和最优传输映射
Abstract
The development of deep learning over the past decade has revolutionized medical imaging segmentation, allowing the extraction of precise descriptors from large volumes to characterize pathologies. Data augmentation is a technique widely regarded as a way to improve model training. It includes simple transformations like spatial operations or intensity modifications, but also more advanced synthesis techniques. Their goal is to generate new realistic samples from an existing dataset to diversify the images used during training. Among them, several propose different mixing strategies to combine real samples. However, one of their major shortcomings is to yield limited variability in terms of generated lesion shapes and locations. In this work, we introduce a novel image synthesis method, called OTLesMix, that leverages Wasserstein barycenter and optimal transport plan to generate realistic and diverse samples. We evaluated our method on three brain lesion segmentation tasks, on which it improves the Dice score compared to a model trained without synthetic data by 2.9 to 6.6 points, and outperforms state-of-the-art mix-based methods.
Chinese Translation
在过去十年中，深度学习的发展彻底改变了医学影像分割，使得能够从大量数据中提取精确的描述符以表征病理。数据增强是一种被广泛认为可以改善模型训练的技术。它包括简单的变换，如空间操作或强度修改，也包括更高级的合成技术。其目标是从现有数据集中生成新的真实样本，以多样化训练过程中使用的图像。在这些技术中，有几种提出了不同的混合策略来结合真实样本。然而，它们的一个主要缺点是生成的病变形状和位置的变异性有限。在本研究中，我们提出了一种新颖的图像合成方法，称为OTLesMix，该方法利用Wasserstein重心和最优传输计划生成真实且多样的样本。我们在三个脑部病变分割任务上评估了我们的方法，与未使用合成数据训练的模型相比，Dice分数提高了2.9到6.6分，并且优于最先进的基于混合的方法。
cs.CV / 106 / 2608.06275
TLNM: Externally Validated Tooth Detection, Numbering and Segmentation from Smartphone Photographs Using Mask R-CNN
TLNM：基于Mask R-CNN的智能手机照片中牙齿检测、编号和分割的外部验证
Abstract
Oral health issues affect billions globally, but the cost and limited access to professional dental care hinder preventive oral healthcare. Research relies on clinical-grade radiographs or intraoral camera images, unavailable for public self-screening. This study introduces a tooth localisation and numbering model for smartphone photographs. We developed a customised Mask Region-based Convolutional Neural Network (Mask R-CNN) pipeline trained on 1,272 annotated smartphone images. To address variability in patient-generated health data, the pipeline incorporates two domain-informed mechanisms: a masked gray-world white-balancing algorithm to mitigate artificial colour casts and an anatomically constrained detection layer to enforce structural validity and suppress false positives. Evaluation comprised four stages: internal held-out testing, independent external testing, a descriptive ablation study, and fold-based training stability analysis using the same internal test set. On the internal test set, the model achieved an instance-mask AP@50 of 0.818, class-aware PQ of 0.780, and operational F1 of 0.884. Training stability showed limited between-model variation: across ten runs, instance-mask AP@50 had a standard deviation of 0.009. On the external dataset, the model achieved an instance-mask AP@50 of 0.901, class-aware PQ of 0.832, and operational F1 of 0.928 despite differences in population, sensors, and acquisition protocols. The inference pipeline is available as an open-source, containerised API. These results demonstrate that consumer-grade smartphone imagery can support automated tooth-level anatomical mapping, offering a scalable, potentially low-cost foundation for remote screening and tele-dentistry in resource-constrained environments.
Chinese Translation
口腔健康问题影响全球数十亿人，但专业牙科护理的高成本和有限的获取途径阻碍了预防性口腔保健。现有研究依赖于临床级放射线照片或口腔内摄像头图像，而这些在公众自我筛查中并不可得。本研究提出了一种用于智能手机照片的牙齿定位和编号模型。我们开发了一个定制的基于Mask区域卷积神经网络（Mask R-CNN）的管道，训练于1272张标注的智能手机图像。为了解决患者生成健康数据的变异性，该管道结合了两种领域信息机制：一种掩蔽灰世界白平衡算法以减轻人工色偏，以及一种解剖约束检测层以强制结构有效性并抑制假阳性。评估包括四个阶段：内部保留测试、独立外部测试、描述性消融研究和基于折叠的训练稳定性分析，均使用相同的内部测试集。在内部测试集上，该模型实现了实例掩膜AP@50为0.818，类别感知PQ为0.780，操作F1为0.884。训练稳定性显示模型间变异性有限：在十次运行中，实例掩膜AP@50的标准差为0.009。在外部数据集上，该模型实现了实例掩膜AP@50为0.901，类别感知PQ为0.832，操作F1为0.928，尽管存在人群、传感器和采集协议的差异。推理管道作为开源容器化API可供使用。这些结果表明，消费级智能手机图像可以支持自动化的牙齿级解剖映射，为资源有限环境中的远程筛查和远程牙科提供了可扩展、潜在低成本的基础。
cs.CV / 107 / 2608.06307
UQ-Loc: Uncertainty-Aware LiDAR Scene Coordinate Regression
UQ-Loc：基于不确定性的激光雷达场景坐标回归
Abstract
LiDAR-based Scene Coordinate Regression (SCR) maps point clouds directly to 3D scene coordinates, enabling precise 6-DoF localisation without explicit map retrieval. However, existing methods produce deterministic predictions, discarding aleatoric uncertainty that could improve robustness and downstream decision-making. We present UQ-Loc, which extends the LightLoc architecture with an anisotropic Gaussian covariance head that predicts a full 3x3 positive-definite covariance matrix per voxel. Training uses a Negative Log-Likelihood (NLL) loss augmented with a kNN-based spatial smoothness regulariser, while inference employs a modified SC2-PCR solver with uncertainty-weighted seed scoring and a Mahalanobis-distance inlier test. We adopt Expected Calibration Error (ECE) as a principled metric for evaluating the quality of the predicted uncertainty. Experiments demonstrate that UQ-Loc achieves consistent improvement in 6-DoF localization accuracy while producing well-calibrated covariances.
Chinese Translation
基于激光雷达的场景坐标回归（SCR）将点云直接映射到三维场景坐标，从而实现精确的六自由度（6-DoF）定位，无需显式的地图检索。然而，现有方法产生确定性的预测，忽略了可能提高鲁棒性和下游决策的不确定性。我们提出了UQ-Loc，它在LightLoc架构的基础上扩展了一个各向异性高斯协方差头，为每个体素预测一个完整的3x3正定协方差矩阵。训练过程中使用了负对数似然（NLL）损失，并辅以基于k近邻的空间平滑正则化，而推理则采用了修改后的SC2-PCR求解器，结合不确定性加权的种子评分和马哈拉诺比斯距离内点测试。我们采用期望校准误差（ECE）作为评估预测不确定性质量的原则性指标。实验表明，UQ-Loc在6-DoF定位精度上实现了一致的提升，同时生成了良好校准的协方差。
cs.CV / 108 / 2608.06311
Does FLAIR super-resolution erase or hallucinate small white-matter lesions?
FLAIR超分辨率是消除还是幻觉小白质病变？
Abstract
White matter hyperintensities (WMH), bright regions on Fluid-attenuated Inversion Recovery (FLAIR) scans are associated with cerebrovascular pathology and neurodegeneration. FLAIR is usually acquired with thick slices in clinical settings, giving it poor through-plane resolution. Super-resolution (SR) is a widely used method for recovering an isotropic volume from an anisotropic scan. Yet whether applying it prior to WMH segmentation preserves lesion content remains unknown: a model may erase small real lesions or hallucinate absent ones. We used 1-mm isotropic high-resolution (HR) FLAIR scans from 29 individuals in the ADNI cohort, each manually segmented for WMH by an expert. Then, we degraded each to simulated 3 and 5 mm through-plane acquisitions. Multi-contrast implicit neural representation (INR), a single-contrast self-supervised model (ECLARE), and cubic interpolation were used to upsample them onto the HR grid. WMH segmentation from a simulated thick slice and the original HR FLAIR set the floor and ceiling, respectively, for the per-lesion analysis. Of four WMH segmentation methods (WMH-SynthSeg, segcsvd, MARS-WMH, TrUE-Net), we ran the analysis under the most sensitive one to small lesions on HR (MARS-WMH) with the evaluation metrics of detection sensitivity, erasure rate (HR-detected lesions lost after reconstruction), and hallucination rate (predicted components absent from both the manual and HR segmentation). The dominant effect of SR was erasure of small real lesions, not hallucination, and it increased with slice thickness, though every reconstruction still improved lesion detection over the raw thick slice. ECLARE recovered small lesion signal best at both thicknesses, while the INR was no better than cubic interpolation.
Chinese Translation
白质高信号（WMH）是流体衰减反转恢复（FLAIR）扫描中明亮的区域，与脑血管病理和神经退行性疾病相关。FLAIR通常在临床环境中以厚切片获取，导致其在平面内的分辨率较差。超分辨率（SR）是一种广泛使用的方法，用于从各向异性扫描中恢复各向同性体积。然而，在WMH分割之前应用该方法是否能够保留病变内容仍然未知：模型可能会消除小的真实病变或幻觉不存在的病变。我们使用来自ADNI队列的29名个体的1毫米各向同性高分辨率（HR）FLAIR扫描，每个扫描均由专家手动分割WMH。然后，我们将每个扫描降级为模拟的3毫米和5毫米平面内采集。我们使用多对比隐式神经表示（INR）、单对比自监督模型（ECLARE）和立方插值将它们上采样到HR网格。来自模拟厚切片的WMH分割和原始HR FLAIR分别设定了每个病变分析的下限和上限。在四种WMH分割方法（WMH-SynthSeg、segcsvd、MARS-WMH、TrUE-Net）中，我们在对小病变最敏感的HR分析下运行（MARS-WMH），评估指标包括检测灵敏度、消除率（重建后丢失的HR检测病变）和幻觉率（在手动和HR分割中均不存在的预测成分）。SR的主要效果是消除小的真实病变，而不是幻觉，且随着切片厚度的增加而增加，尽管每次重建仍然改善了相较于原始厚切片的病变检测。ECLARE在两种厚度下均能最佳恢复小病变信号，而INR的表现不优于立方插值。
cs.AI / 1 / 2608.05159
Agentic Nesting: A New Methodology for Existing Enterprise Application Integration and Services
代理嵌套：一种用于现有企业应用集成和服务的新方法论
Abstract
Enterprise operations extensively rely on multiple heterogeneous business systems and information applications, which also result in severe data silos and process fragmentation. Enterprises have invested considerable financial and material resources in building these applications, however, effectively leveraging and orchestrating them remains a formidable challenge. Conventional approaches to enterprise application integration, encompassing middleware architectures such as Enterprise Service Bus (ESB), API gateway infrastructures, and Robotic Process Automation (RPA), suffer from inherent limitations like high architectural coupling, escalating operation and maintenance costs, and limited intelligence capabilities. This paper proposes Agentic Nesting, a multi-agent collaboration framework in which existing enterprise applications are encapsulated as autonomous AI agents within a hierarchically nested structure. Rather than flat interconnection, agents are organized into layered stewardship topologies that mirror the compositional complexity of enterprise ecosystems. The framework extracts a digital agent proxy from each legacy application to enable natural-language interaction and autonomous manipulation, coordinates multiple agents through a central orchestrator for task decomposition and dynamic dispatching, and exposes a unified conversational interface for cross-application querying and process orchestration. The main contributions of this paper are the proposition of the "Application-as-Agent" integration paradigm and the "Conversation-as-Integration" interaction philosophy, together with an exploration of the generalization potential of this methodology in scenarios encompassing heterogeneous system coordination, and large-scale data applications.
Chinese Translation
企业运营广泛依赖于多个异构业务系统和信息应用，这也导致了严重的数据孤岛和流程碎片化。企业在构建这些应用上投入了大量的财力和物力，然而，如何有效利用和协调这些应用仍然是一项艰巨的挑战。传统的企业应用集成方法，包括企业服务总线（Enterprise Service Bus, ESB）、API 网关架构和机器人流程自动化（Robotic Process Automation, RPA），存在高架构耦合、运营和维护成本上升以及智能能力有限等固有局限性。本文提出了代理嵌套（Agentic Nesting），一种多代理协作框架，其中现有企业应用被封装为层次嵌套结构中的自主 AI 代理。代理不是简单的平面互联，而是组织成分层的管理拓扑，反映企业生态系统的组合复杂性。该框架从每个遗留应用中提取数字代理代理，以实现自然语言交互和自主操作，通过中央协调者协调多个代理进行任务分解和动态调度，并为跨应用查询和流程协调提供统一的对话接口。本文的主要贡献在于提出了“应用即代理”（Application-as-Agent）集成范式和“对话即集成”（Conversation-as-Integration）交互理念，并探讨了该方法论在异构系统协调和大规模数据应用场景中的推广潜力。
cs.AI / 2 / 2608.05160
The Ignition Index: Measuring Global Workspace Dynamics in Language Models
点火指数：测量语言模型中的全球工作空间动态
Abstract
We introduce the Ignition Index (I), a validated scalar metric that operationalizes Global Workspace Theory's (GWT) all-or-none ignition prediction in transformer language models. The metric fits a four-parameter sigmoid to per-layer linear probe accuracy as a function of input signal strength, extracting steepness parameter beta-hat: high values indicate abrupt, ignition-like transitions; low values indicate graded build-up. Across 11 models spanning five architecture families, shuffled-label controls demonstrate 9.6-fold selectivity for genuine linguistic structure over spurious probe capacity (p < 0.001, Mann-Whitney U-test). We find: (1) Feedforward transformers exceed SSMs by 89% in aggregate beta-hat (p < 1e-13, Cohen's d = 0.52), with Mamba exhibiting near-linear profiles consistent with absent global broadcast. (2) Huginn-3.5B exhibits 2.12-fold higher ignition along its iteration axis than its depth axis, demonstrating that recurrent architectures manifest workspace-like transitions along the recurrence dimension. (3) Pythia-410M shows a PELT-detected phase transition at training step 256 (+67%), preceding induction-head formation. (4) Hypotheses linking ignition to model scale and signal strength were not confirmed, suggesting transformer architectures may saturate available ignition mechanisms. The Ignition Index provides the first validated quantitative bridge between GWT's dynamical predictions and mechanistic interpretability, with 9.6-fold measurement selectivity and architecture-level discriminability not previously characterized in the scaling literature. Code: https://github.com/saman-rahbar/ignition-index
Chinese Translation
我们引入了点火指数（Ignition Index, I），这是一种经过验证的标量度量，能够在变换器语言模型中实现全球工作空间理论（Global Workspace Theory, GWT）的全或无点火预测。该度量将四参数S型曲线拟合到每层线性探针准确性与输入信号强度的关系中，从中提取陡峭度参数beta-hat：高值表示突然的、类似点火的转变；低值则表示逐步累积。在涵盖五种架构家族的11个模型中，随机标签控制显示出对真实语言结构的9.6倍选择性，相较于虚假探针能力（p < 0.001，Mann-Whitney U检验）。我们的发现包括：（1）前馈变换器在整体beta-hat上超越状态空间模型（State Space Models, SSMs）89%（p < 1e-13，Cohen's d = 0.52），而Mamba展现出与缺乏全球广播一致的近线性特征。（2）Huginn-3.5B在其迭代轴上的点火水平比其深度轴高出2.12倍，表明递归架构在递归维度上表现出类似工作空间的转变。（3）Pythia-410M在训练步骤256处显示出PELT检测到的相变（+67%），早于诱导头的形成。（4）将点火与模型规模和信号强度联系起来的假设未得到证实，这表明变换器架构可能会饱和可用的点火机制。点火指数提供了GWT动态预测与机制可解释性之间的首个经过验证的定量桥梁，其9.6倍的测量选择性和架构层级的可区分性在规模文献中尚未被描述。代码链接： https://github.com/saman-rahbar/ignition-index
cs.AI / 3 / 2608.05168
Woodpecker Distillation: Weak Models Diagnose Reasoning Bugs in Strong Models
啄木鸟蒸馏：弱模型诊断强模型中的推理错误
Abstract
Large language models often fail on reasoning tasks despite possessing the capability to solve them. We argue that many such failures arise from localized reasoning bugs in intermediate steps rather than from global incompetence. We show that these bugs are frequently repairable: inserting a short patch generated by a weak probe model after the same strong-model reasoning prefix can redirect the trajectory toward a correct solution. However, this corrective effect is not reliably internalized by directly fine-tuning on weak patches or repaired trajectories, suggesting that the useful signal lies not in the intervention text itself, but in how it reshapes the model's future reasoning distribution. We therefore propose Woodpecker Distillation, a weak-to-strong training framework that learns from contrastive local interventions. Our method contrasts successful and unsuccessful weak-model patches at the same prefix, constructs a corrective teacher distribution from their induced future token predictions, and distills this signal into the strong model. Experiments on mathematical reasoning benchmarks show that Woodpecker Distillation consistently improves strong-model performance and outperforms direct imitation baselines.
Chinese Translation
大型语言模型在推理任务中常常失败，尽管它们具备解决这些任务的能力。我们认为，这些失败的许多原因源于中间步骤中的局部推理错误，而非整体能力不足。我们展示了这些错误通常是可修复的：在相同的强模型推理前缀后插入由弱探测模型生成的短补丁，可以将推理轨迹引导至正确的解决方案。然而，这种纠正效果并不能通过直接在弱补丁或修复轨迹上进行微调来可靠地内化，这表明有用的信号不在于干预文本本身，而在于它如何重塑模型未来的推理分布。因此，我们提出了啄木鸟蒸馏（Woodpecker Distillation），一种从对比局部干预中学习的弱到强的训练框架。我们的方法对比了在相同前缀下成功和不成功的弱模型补丁，从它们引发的未来标记预测中构建纠正教师分布，并将这一信号蒸馏到强模型中。在数学推理基准上的实验表明，啄木鸟蒸馏始终提高了强模型的性能，并超越了直接模仿的基线。
cs.AI / 4 / 2608.05203
From Continuous Predictors to Clinical Thresholds: Early Evidence on Performance Trade-offs of Guideline-Based Categorisation for Ischaemic Stroke Outcome Prediction
从连续预测因子到临床阈值：基于指南的分类在缺血性中风结果预测中的性能权衡的早期证据
Abstract
Machine learning models achieve strong predictive accuracy for 90-day outcome prediction in acute ischaemic stroke, yet clinical adoption is limited by the misalignment of model explanations with clinicians' reasoning. Motivated by a clinician user study calling for clinical guideline-aligned cut-offs, we ask whether continuous predictors can be replaced by clinically informed categorical encodings without sacrificing performance. On a multi-centre European registry stratified into three treatment cohorts, we compare standard and fully categorised gradient-boosted models, the latter using stroke guideline-aligned, treatment-specific thresholds. The fully categorised models are statistically indistinguishable from their continuous counterparts in two of the treatment cohorts, with a significant drop in predictive accuracy in one cohort. Global feature importance rankings remain consistent, suggesting that discretising continuous predictors into guideline-based categories preserves the core hierarchy of prognostic factors across all treatment groups. Guideline-based categorisation is thus a viable design choice for stroke-outcome models.
Chinese Translation
机器学习模型在急性缺血性中风的90天结果预测中实现了较强的预测准确性，但由于模型解释与临床医生的推理不一致，临床应用受到限制。受到一项呼吁临床指南对齐的临床用户研究的启发，我们探讨是否可以用临床知情的分类编码替代连续预测因子，而不牺牲性能。在一个分为三个治疗队列的多中心欧洲注册研究中，我们比较了标准的和完全分类的梯度提升模型，后者使用与中风指南对齐的、特定于治疗的阈值。在两个治疗队列中，完全分类的模型在统计上与其连续对应物无显著差异，而在一个队列中预测准确性显著下降。全球特征重要性排名保持一致，表明将连续预测因子离散化为基于指南的类别保留了所有治疗组中预后因素的核心层次。因此，基于指南的分类是中风结果模型的一个可行设计选择。
cs.AI / 5 / 2608.05204
SkillTrace: Multi-Trace Provenance Auditing for LLM-Agent Skill Reuse
SkillTrace：针对LLM-Agent技能重用的多追踪来源审计
Abstract
LLM-agent ecosystems are rapidly growing around reusable skills: mixed-modality packages of metadata, natural-language instructions, code, tools, references, and operational workflows. As skills become marketplace artifacts, auditing their reuse is no longer the same problem as ordinary code clone detection. Existing detectors target single-modality source code or whole-package similarity, yet skill reuse evidence is distributed across authored text, implementation fragments, and operational structure. As a result, they can miss reuse that preserves only one part of a skill. We present SKILLTRACE, a multi-trace provenance auditing framework for LLM-agent skill reuse. SKILLTRACE extracts three provenance traces: Expression, Implementation, and Operational. It represents the Operational Trace as a Skill Operational Graph (SOG) that captures activation, procedure, and resource-flow structure. An LLM assists only the Operational-trace extraction, once at ingestion; at audit time SKILLTRACE compares cached traces deterministically, calibrates each trace against same-function strict negatives, and reports which trace supports a reuse decision. On SKILLTRACE-BENCH, with 820 transformed reuse positives over 100 marketplace anchors and 751 negative controls, SKILLTRACE achieves AUROC 0.938 and F1 0.898. A 36,446-skill wild audit further shows that trace-attributed evidence surfaces actionable reuse review queues beyond repository-level baselines.
Chinese Translation
LLM-agent生态系统围绕可重用技能迅速发展，这些技能是元数据、自然语言指令、代码、工具、参考资料和操作工作流的混合模式包。随着技能成为市场上的产品，审计其重用不再是普通代码克隆检测的问题。现有的检测器主要针对单一模式的源代码或整个包的相似性，而技能重用的证据分布在创作文本、实现片段和操作结构中。因此，它们可能会遗漏只保留技能某一部分的重用。我们提出了SKILLTRACE，这是一个针对LLM-agent技能重用的多追踪来源审计框架。SKILLTRACE提取三种来源追踪：表达（Expression）、实现（Implementation）和操作（Operational）。它将操作追踪表示为技能操作图（Skill Operational Graph, SOG），该图捕捉激活、过程和资源流结构。LLM仅在摄取时辅助操作追踪的提取；在审计时，SKILLTRACE以确定性方式比较缓存追踪，将每个追踪与相同功能的严格负样本进行校准，并报告哪个追踪支持重用决策。在SKILLTRACE-BENCH上，经过100个市场锚点的820个转化重用正样本和751个负对照，SKILLTRACE达到了AUROC 0.938和F1 0.898。一次对36,446个技能的广泛审计进一步表明，追踪归因的证据超越了存储库级基线，呈现出可操作的重用审查队列。
cs.AI / 6 / 2608.05205
Abstract Event Causal Rules: Induction and Application
抽象事件因果规则：归纳与应用
Abstract
Event-centric intelligent analytical systems heavily depend on explicit causal event knowledge for risk early warning, decision-making support and narrative comprehension. Nevertheless, existing instance-level causal pairs suffer severe generalization deficits on low-frequency long-tail and unseen event combinations. To address this limitation, this work proposes Abstract Event Causal Rule (AECR), a novel relation-level causal abstraction paradigm that transforms concrete cause-effect pairs into generalized abstract causal logic while retaining their intrinsic causal relationships. We design a multi-agent Concrete-to-Abstract Causal Induction (CACI) system coupled with similarity-constrained clustering to distill trustworthy AECRs from noisy raw causal data, based on which two complete AECR knowledge bases are built. To validate the practical utility of abstract causal knowledge, we propose an Abstract Rule-Guided Causal Attention Encoder (AR-GCAE), which injects the retrieved AECRs into the causality Graph Event Prediction (CGEP) benchmark task via rule-guided attention layers and gated representation fusion. Quantitative experimental results reveal that applying AECRs substantially strengthens the generalization capacity of event causal reasoning and brings consistent performance improvements to event prediction, with the most prominent gains observed on rare and unseen event samples.
Chinese Translation
以事件为中心的智能分析系统在风险预警、决策支持和叙事理解方面高度依赖明确的因果事件知识。然而，现有的实例级因果对在低频长尾和未见事件组合上存在严重的泛化缺陷。为了解决这一限制，本文提出了抽象事件因果规则（Abstract Event Causal Rule, AECR），这是一种新颖的关系级因果抽象范式，它将具体的因果对转化为广义的抽象因果逻辑，同时保留其内在的因果关系。我们设计了一个多智能体的具体到抽象因果归纳系统（Concrete-to-Abstract Causal Induction, CACI），结合相似性约束聚类，从嘈杂的原始因果数据中提炼出可信的AECR，并基于此构建了两个完整的AECR知识库。为了验证抽象因果知识的实际效用，我们提出了一种抽象规则引导的因果注意编码器（Abstract Rule-Guided Causal Attention Encoder, AR-GCAE），通过规则引导的注意层和门控表示融合，将检索到的AECR注入因果图事件预测（Causality Graph Event Prediction, CGEP）基准任务中。定量实验结果表明，应用AECR显著增强了事件因果推理的泛化能力，并在事件预测中带来了持续的性能提升，尤其在稀有和未见事件样本上表现出最显著的增益。
cs.AI / 7 / 2608.05206
Otter: A Time-Aware, History-Conditioned Human Chess AI
Otter：一种时间感知的历史条件人类国际象棋人工智能
Abstract
Otter is a 15.3M-parameter human chess AI that predicts human move selection by modeling play as a time-aware, sequential process rather than treating each position in isolation. It combines two conditioning signals: (1) a move history encoder that conditions predictions on the last 20 moves, capturing opening preferences, positional drift, and intra-game behavioral tendencies; and (2) a time control module that modulates predictions based on clock pressure. Otter is trained on 6.1 billion positions from 117 million Lichess rapid games over 30 days on a single T4 GPU. Otter achieves 55.23% top-1 and 90.95% top-5 move-prediction accuracy, surpassing the prior state-of-the-art human chess model, Maia 2, with far fewer parameters and less training data. Across 11 Elo brackets (<1100 to >=2000), accuracy peaks at 57.38% in the 1900-1999 bracket. These results show that modeling chess as a time-aware, sequential activity yields more human-accurate move prediction than position-only approaches, using a smaller model. Code, trained models, and complete training logs are publicly released.
Chinese Translation
Otter 是一个拥有 1530 万参数的人类国际象棋人工智能，它通过将游戏建模为一种时间感知的序列过程，而不是将每个棋局孤立对待，从而预测人类的走棋选择。它结合了两个条件信号：（1）一个走棋历史编码器，根据过去 20 步棋的走法来调整预测，捕捉开局偏好、棋局漂移和游戏内行为倾向；（2）一个时间控制模块，根据时间压力调节预测。Otter 在 30 天内使用单个 T4 GPU 训练了来自 1.17 亿场 Lichess 快速游戏的 61 亿个棋局。Otter 实现了 55.23% 的 top-1 和 90.95% 的 top-5 走棋预测准确率，超越了之前的最先进人类国际象棋模型 Maia 2，且参数更少、训练数据更少。在 11 个 Elo 等级区间（<1100 到 >=2000）中，准确率在 1900-1999 区间达到 57.38% 的峰值。这些结果表明，将国际象棋建模为一种时间感知的序列活动，能够比仅基于位置的方法提供更接近人类的走棋预测，同时使用更小的模型。代码、训练模型和完整的训练日志已公开发布。
cs.AI / 8 / 2608.05212
SearchAuditor: Auditing and Attributing Failures in Long-Horizon Search Agents
SearchAuditor：长时域搜索代理的失败审计与归因
Abstract
Deep search agents tackle challenging questions through long-horizon web interactions, a process that is both complex and fragile: small reasoning errors may propagate through long, noisy trajectories into fluent but incorrect answers. Diagnosing such failures is difficult, requiring the manual inspection of extremely long execution traces, which could be beyond human capacity. We therefore introduce SearchAuditBench, a benchmark that evaluates whether LLM auditors can localize, attribute, and repair these failures, thereby reducing the human burden. SearchAuditBench comprises 1,243 failed trajectories, averaging 73.1 messages and 65.1K tokens, collected from eight open-weight models on five deep-search benchmarks, each expert-annotated with the critical error step, a search-specific root cause, and a reference repair with grading rubrics. We further propose SearchAuditor, a multi-perspective auditing framework that effectively localizes, attributes, and repairs search-agent failures through evidence-grounded adjudication. Experimental results show that even the strongest baseline, when powered by a frontier model like GPT-5.5, attains only a 26.6% end-to-end pass rate. In contrast, our SearchAuditor consistently outperforms all baselines across different frontier models, achieving an end-to-end pass rate of 32.3%, and resuming failed runs with its repairs enables agents to better recover from errors.
Chinese Translation
深度搜索代理通过长时域的网络交互解决复杂问题，这一过程既复杂又脆弱：小的推理错误可能在漫长而嘈杂的轨迹中传播，导致流畅但错误的答案。诊断此类失败非常困难，需要手动检查极长的执行轨迹，这可能超出人类的能力。因此，我们引入了SearchAuditBench，一个基准测试，用于评估大型语言模型（LLM）审计员是否能够定位、归因和修复这些失败，从而减轻人类的负担。SearchAuditBench包含1,243条失败轨迹，平均包含73.1条消息和65.1K个标记，数据来自五个深度搜索基准上的八个开放权重模型，每个轨迹都由专家标注了关键错误步骤、特定搜索的根本原因以及带有评分标准的参考修复。我们进一步提出了SearchAuditor，一个多视角审计框架，通过基于证据的裁决有效地定位、归因和修复搜索代理的失败。实验结果表明，即使是最强的基线，在像GPT-5.5这样的前沿模型的支持下，端到端通过率也仅为26.6%。相比之下，我们的SearchAuditor在不同的前沿模型中始终优于所有基线，达到了32.3%的端到端通过率，并通过其修复使代理能够更好地从错误中恢复。
cs.AI / 9 / 2608.05218
PD-GS: Phoneme-Driven 3DGS for Audio-Driven Talking Heads
PD-GS：基于音素驱动的3D高斯点云技术用于音频驱动的对话头
Abstract
3D Gaussian Splatting (3DGS) enables fast, photorealistic talking-head rendering, yet accurate lip articulation remains elusive: mouth motion is often over-smoothed and may violate hard articulatory constraints such as bilabial closures, producing the notorious ``leaky mouth'' artifact. A key difficulty is that brief, discrete articulatory events are inferred from a continuous acoustic embedding under a regression objective, which biases predictions toward averaged mouth configurations. While modern self-supervised speech encoders provide rich prosodic and phonetic cues, they do not provide an explicit, frame-aligned linguistic target that reliably disambiguates closure-level events. We propose \textbf{Phoneme-Driven Gaussian Splatting (PD-GS)}, which augments a 3DGS talker with time-aligned phoneme tokens obtained from an automatic ASR and forced-alignment pipeline. Our core component, the \textbf{Linguistic Fusion Module (LFM)}, adaptively fuses continuous audio context with discrete phoneme embeddings through a learned gate, allowing the model to preserve smooth audio-driven dynamics while strengthening phoneme guidance on articulation-critical segments. PD-GS is trained purely from monocular video using image reconstruction and lip landmark supervision. On HDTF, PD-GS achieves the best lip geometry among the compared baselines (LMD 2.66) and qualitatively reduces closure violations in challenging phoneme sequences, yielding more linguistically faithful neural avatars.
Chinese Translation
3D高斯点云技术（3DGS）能够快速生成逼真的对话头渲染，但准确的唇部发音仍然难以实现：嘴部运动常常过于平滑，可能违反诸如双唇闭合等严格的发音约束，导致著名的“漏嘴”伪影。一个主要的困难在于，短暂的离散发音事件是通过回归目标从连续的声学嵌入中推断出来的，这使得预测偏向于平均的嘴部配置。尽管现代自监督语音编码器提供了丰富的韵律和音素线索，但它们并未提供一个明确的、帧对齐的语言目标，无法可靠地区分闭合级别事件。我们提出了 extbf{基于音素驱动的高斯点云技术（PD-GS）}，该技术通过自动语音识别（ASR）和强制对齐管道获得时间对齐的音素标记，增强了3DGS说话者。我们的核心组件 extbf{语言融合模块（LFM）}通过学习的门控机制自适应地融合连续的音频上下文与离散的音素嵌入，使模型能够在保持平滑的音频驱动动态的同时，加强对发音关键段的音素指导。PD-GS完全通过单目视频进行训练，使用图像重建和唇部标志监督。在HDTF数据集上，PD-GS在比较的基线中实现了最佳的唇部几何形状（LMD 2.66），并在具有挑战性的音素序列中定性减少了闭合违规现象，生成了更具语言忠实度的神经化身。
cs.AI / 10 / 2608.05219
When Privileged Guidance Misaligns: State-Matched Routing and Contextualized Self-Distillation for Multi-Turn Agents
特权指导不一致时：状态匹配路由与上下文化自蒸馏在多轮智能体中的应用
Abstract
Privileged on-policy distillation provides dense supervision for multi-turn agents by allowing a synchronized teacher to re-score the student's response at every turn with access to training-only references, such as successful trajectories. In interactive environments, however, the student's preceding actions continually change the execution state. As the student takes different actions or completes subgoals in a different order, its rollout may reach states not covered by the reference, making the reference an unreliable source of guidance for the state actually reached. Applying privileged distillation indiscriminately therefore creates state--reference mismatch. This mismatch motivates a central objective: providing privileged reference guidance that remains compatible with the student's current execution state. We introduce State-Matched Routing and Contextualized Self-Distillation (SMRC-SD), which explicitly determines when and how a privileged trajectory should guide an on-policy student. At each turn, SMRC-SD verifies whether the student's current execution state matches a supported state along the reference trajectory. Distillation is applied only at matched states, filtering out turns for which the reference lacks locally compatible guidance. For each matched state, SMRC-SD further constructs state-conditioned teacher context from the successful trajectory, grounding supervision in the state actually reached. Across ALFWorld and WebShop, SMRC-SD consistently outperforms unconditional successful full-path distillation. With Qwen3-1.7B, it improves task success from $0.746$ to $0.865$ on ALFWorld and from $0.574$ to $0.693$ on WebShop. Controlled routing and context ablations support both selecting locally supported turns and constructing state-compatible teacher context as contributors to these gains. Code is available at https://github.com/liujunzhuo/SMRC-SD.
Chinese Translation
特权在线蒸馏通过允许同步教师在每个回合重新评分学生的响应，并访问仅用于训练的参考资料（如成功轨迹），为多轮智能体提供了密集的监督。然而，在交互环境中，学生之前的行动不断改变执行状态。随着学生采取不同的行动或以不同的顺序完成子目标，其展开可能达到参考资料未覆盖的状态，使得参考资料成为实际达到状态的一个不可靠指导来源。因此，盲目应用特权蒸馏会导致状态与参考之间的不匹配。这种不匹配激发了一个核心目标：提供与学生当前执行状态兼容的特权参考指导。我们提出了状态匹配路由与上下文化自蒸馏（State-Matched Routing and Contextualized Self-Distillation, SMRC-SD），该方法明确确定何时以及如何让特权轨迹指导在线学生。在每个回合，SMRC-SD 验证学生当前的执行状态是否与参考轨迹上的支持状态匹配。蒸馏仅在匹配状态下应用，过滤掉参考缺乏局部兼容指导的回合。对于每个匹配状态，SMRC-SD 进一步从成功轨迹构建状态条件教师上下文，将监督基础扎根于实际达到的状态。在 ALFWorld 和 WebShop 上，SMRC-SD 始终优于无条件的成功全路径蒸馏。在 Qwen3-1.7B 上，它将 ALFWorld 的任务成功率从 $0.746$ 提升至 $0.865$，将 WebShop 的任务成功率从 $0.574$ 提升至 $0.693$。控制路由和上下文消融实验支持选择局部支持的回合和构建状态兼容教师上下文作为这些提升的贡献因素。代码可在 https://github.com/liujunzhuo/SMRC-SD 获取。
cs.AI / 11 / 2608.05224
Small Foundation Models of Human Cognition and Behaviour
小型基础模型的人类认知与行为
Abstract
Large language models fine-tuned on human behavioural data have emerged as general-purpose cognitive proxies, but the scale this requires, and whether these models process task structure or exploit statistical shortcuts, remain open questions. We train fourteen models from 135M to 14B parameters across four architecture families on Psych-101, a dataset of 10.7 million trial-level choices from 160 experiments. In-distribution, scale barely matters. The models fall within a narrow band, as though against a ceiling, and 0.6B to 1B parameters suffice to match a 70B baseline on held-out participants. Out-of-distribution, that band opens into a markedly steeper scaling gradient, with larger models clearly advantaged in generalisation to novel task structure. To determine what information these models use, we run two diagnostics. We progressively strip four prompt channels -- task instructions, experimental stimuli, outcome feedback, and choice history -- across 27 experiments, and permute trial order. Masking the content of stimuli and feedback destroys 75.7% of learned information and pushes models below chance, demonstrating that choice history alone does not account for performance. Permutation reveals invariance on tasks with independent trials but sensitivity where trial order is determined by prior responses. Small cognitively fine-tuned models therefore show promise as noise ceiling estimators for psychological experiments, though their scope remains bounded by the paradigms seen in training.
Chinese Translation
经过人类行为数据微调的大型语言模型已成为通用的认知代理，但所需的规模，以及这些模型是否处理任务结构或利用统计捷径，仍然是未解之谜。我们在 Psych-101 数据集上训练了十四个参数从 135M 到 14B 的模型，该数据集包含来自 160 个实验的 1070 万次试验级选择。在分布内，规模几乎无关紧要。这些模型的表现落在一个狭窄的范围内，似乎受到了上限的限制，而 0.6B 到 1B 的参数足以在保留参与者的情况下与 70B 的基线相匹配。在分布外，这个范围则打开了一个明显更陡的规模梯度，更大的模型在对新任务结构的泛化上明显占优。为了确定这些模型使用了哪些信息，我们进行了两个诊断实验。我们逐步剥离了四个提示通道——任务指令、实验刺激、结果反馈和选择历史——在 27 个实验中，并对试验顺序进行了置换。掩蔽刺激和反馈的内容摧毁了 75.7% 的学习信息，并使模型的表现低于随机水平，证明选择历史并不能单独解释性能。置换揭示了在独立试验的任务上表现出不变性，但在试验顺序由先前反应决定的情况下表现出敏感性。因此，小型认知微调模型作为心理实验中的噪声上限估计器显示出前景，但其范围仍然受到训练中所见范式的限制。
cs.AI / 12 / 2608.05225
Project2Task: Graph-Guided Project-Level Planning for Autonomous Research
Project2Task：面向自主研究的图引导项目级规划
Abstract
Research agents can increasingly search literature, propose hypotheses, generate code, run experiments, and draft manuscripts from a single topic. However, a research project is not merely a larger task: it is a long-horizon agenda that must be advanced through multiple bounded tasks with distinct but related objectives, parallel alternatives, and dependency-aware sequences. Existing single-task systems often treat the project as one oversized task, produce a flat set of vague or overlapping tasks, or leave task boundaries and execution order to manual coordination. We introduce Project2Task, a graph-guided project-level planning layer for autonomous research. Given a project brief, it represents candidate contributions as innovation atoms and organizes them in a directed lineage graph. A lightweight Bernoulli block-model objective selects among horizontal, vertical, and hybrid portfolio decompositions. Project2Task then generates bounded tasks with explicit contribution ownership, repairs overlaps and missing execution fields, and emits dependency-aware task contracts that specify objectives, inputs, expected artifacts, evaluation requirements, boundary constraints, dependencies, and execution order. The contracts are independent of any particular downstream research executor and support integration of task outputs into a coherent project-level result. On a benchmark of ten project briefs yielding roughly 30 tasks, manuscript-based portfolio evaluation gives Project2Task an average quality score of 7.15, compared with 4.58 for the Brief Baseline and 5.31 for the Topic-only Setting. Integrating its contracts with AutoResearchClaw increases average downstream task accuracy from 0.536 to 0.759. These results demonstrate the value of explicit project-to-task planning for producing coherent, non-redundant, and executable research-task portfolios.
Chinese Translation
研究代理可以越来越多地从单一主题中搜索文献、提出假设、生成代码、进行实验和撰写手稿。然而，研究项目不仅仅是一个更大的任务：它是一个必须通过多个具有不同但相关目标、平行替代方案和依赖关系意识序列的有限任务推进的长期议程。现有的单任务系统往往将项目视为一个超大任务，产生一组模糊或重叠的任务，或将任务边界和执行顺序留给人工协调。我们提出了Project2Task，这是一个面向自主研究的图引导项目级规划层。给定项目简报，它将候选贡献表示为创新原子，并将其组织在一个有向谱系图中。一个轻量级的伯努利块模型目标在水平、垂直和混合投资组合分解中进行选择。然后，Project2Task生成具有明确贡献所有权的有限任务，修复重叠和缺失的执行字段，并发出依赖关系意识的任务合同，具体说明目标、输入、预期成果、评估要求、边界约束、依赖关系和执行顺序。这些合同独立于任何特定的下游研究执行者，并支持将任务输出整合为一个连贯的项目级结果。在一个包含十个项目简报的基准测试中，产生大约30个任务，基于手稿的投资组合评估给予Project2Task平均质量评分为7.15，而简报基线为4.58，主题仅设置为5.31。将其合同与AutoResearchClaw集成将下游任务的平均准确性从0.536提高到0.759。这些结果证明了明确的项目到任务规划在生成连贯、非冗余和可执行的研究任务投资组合中的价值。
cs.AI / 13 / 2608.05228
TriQua: Reconciling Granularity and Context in Factuality Evaluation
TriQua：在事实性评估中调和粒度与上下文
Abstract
The "decompose-then-verify" paradigm for LLM factuality evaluation faces a fundamental trade-off: atomic facts, i.e., one sentence conveying one unit of information, often omit essential context, while broader statements lack the granularity needed for precise assessment. To address this, we introduce TriQua, a framework that flexibly models facts based on their complexity. Simple claims are extracted as standard triples, while complex claims are represented as hyperrelational facts by attaching auxiliary contextual qualifiers. This adaptive structure preserves the necessary context for accurate retrieval and verification without sacrificing atomicity. Furthermore, TriQua's verification process directly annotates concrete errors within specific triples and qualifiers, providing fine-grained explainability for error detection. Alongside the framework, we propose TriQuaScore to quantify the factuality of these structured fact units. Empirical evaluations show that TriQuaScore strongly aligns with human annotated factuality scores, TriQua achieves robust decomposition quality, and outperforms existing decomposition-based frameworks in evidence-based fact verification.
Chinese Translation
“分解-再验证”范式在大型语言模型（LLM）事实性评估中面临一个基本的权衡：原子事实，即一条句子传递一个信息单元，往往忽略了重要的上下文，而更广泛的陈述则缺乏进行精确评估所需的粒度。为了解决这一问题，我们提出了TriQua，一个灵活建模事实复杂性的框架。简单的主张被提取为标准三元组，而复杂的主张则通过附加辅助上下文限定词表示为超关系事实。这种自适应结构在不牺牲原子性的情况下，保留了准确检索和验证所需的上下文。此外，TriQua的验证过程直接注释了特定三元组和限定词中的具体错误，为错误检测提供了细粒度的可解释性。除了该框架，我们还提出了TriQuaScore，以量化这些结构化事实单元的事实性。实证评估表明，TriQuaScore与人工注释的事实性评分高度一致，TriQua实现了稳健的分解质量，并在基于证据的事实验证中优于现有的基于分解的框架。
cs.AI / 14 / 2608.05233
Coherence-Oriented Dream Scene Visualisation
以连贯性为导向的梦境场景可视化
Abstract
Dreams can be emotionally intense but difficult to communicate. We describe the Dream Scene Visualiser (DSV) system which turns written dream descriptions into a temporal sequence of four panel images visualising the dream. This starts with a large language model prompted to split a dream description into four chronological parts. Then a text-to-image model produces images for each part with visual coherence maintained across the sequence, and DSV regenerates any image not suitably matching the text. We evaluate DSV over 50 visualisations from dream descriptions in DreamBank, and report quality, fidelity and coherence results via objective measures employing the CLIP, DINOv2 and Qwen2-VL vision-language models.
Chinese Translation
梦境可能情感强烈，但难以传达。我们描述了梦境场景可视化系统（Dream Scene Visualiser, DSV），该系统将书面梦境描述转化为四幅面板图像的时间序列，以可视化梦境。该过程首先使用大型语言模型将梦境描述分割为四个按时间顺序排列的部分。然后，文本到图像模型为每个部分生成图像，并在整个序列中保持视觉连贯性，DSV还会重新生成与文本不匹配的图像。我们在DreamBank中评估了50个梦境描述的可视化效果，并通过使用CLIP、DINOv2和Qwen2-VL视觉-语言模型的客观指标报告了质量、保真度和连贯性结果。
cs.AI / 15 / 2608.05245
Search2Skill: Skill Distillation Beyond Knowledge Boundaries Via Rubric-Based Reinforcement Learning
Search2Skill：通过基于评分标准的强化学习实现超越知识边界的技能蒸馏
Abstract
Reusable skills, which encapsulate the procedural knowledge required to solve real-world professional tasks, offer LLM-based agents a path toward self-evolution in expert domains. Existing self-evolving skill methods construct skills internally from the model's parametric knowledge or trajectories, and are therefore bounded by what the model already knows. However, the domain conventions and standard procedures underlying professional skills often lie beyond this boundary and are hard to elicit from the agent alone. To address this issue, we therefore propose a novel framework, Search2Skill, that automatically identifies the agent's capability gaps, searches external sources to address them, and distills the retrieved evidence into structured, reusable skills. Specifically, Search2Skill is optimized by a rubric-based reinforcement learning scheme that jointly improves when to search, how to search, and how to generate skills. Experiments on eight expert-level domains from three benchmarks show that Search2Skill consistently outperforms both search-augmented and trajectory-based skill-learning baselines under both streaming and held-out evaluation protocols. Further analyses show that the gains arise from skill abstraction rather than raw retrieved evidence, and that the acquired skills transfer across model scales.
Chinese Translation
可重用技能封装了解决现实世界专业任务所需的程序性知识，为基于大型语言模型（LLM）的智能体在专业领域自我进化提供了一条路径。现有的自我进化技能方法通常从模型的参数知识或轨迹中内部构建技能，因此受限于模型已有的知识。然而，专业技能所依赖的领域惯例和标准程序往往超出这一边界，难以仅从智能体中提取。为了解决这一问题，我们提出了一种新颖的框架——Search2Skill，该框架能够自动识别智能体的能力差距，搜索外部资源以填补这些差距，并将检索到的证据蒸馏成结构化的可重用技能。具体而言，Search2Skill通过基于评分标准的强化学习方案进行优化，联合提升搜索时机、搜索方式和技能生成的效果。在三个基准测试的八个专家级领域的实验表明，Search2Skill在流式和保留评估协议下均持续优于增强搜索和基于轨迹的技能学习基线。进一步分析显示，收益来自于技能抽象，而非原始检索证据，并且所获得的技能在不同模型规模之间具有迁移能力。
cs.AI / 16 / 2608.05246
LUNAR: Benchmarking Personalized Large Language Models on UNiversal User BehAvioR Logs
LUNAR：在通用用户行为日志上评估个性化大型语言模型的基准
Abstract
Existing personalized LLM benchmarks primarily rely on textual personas or isolated behavioral signals, providing limited evaluation of cross-domain behavioral personalization, where responses must be grounded in heterogeneous daily-life activities. To address this gap, we introduce LUNAR, the first benchmark for evaluating how LLMs personalize responses from longitudinal app interaction histories across universal daily-life domains, including clothing, food, housing, and mobility. To support scalable benchmark construction while mitigating data sparsity and privacy concerns, LUNAR uses a multi-stage coarse-to-fine synthesis pipeline grounded in real-world behavioral patterns. Fidelity analyses show closer alignment with real behavioral distributions than other synthetic benchmarks. Experiments on 19 mainstream LLMs show that access to behavioral logs is necessary but not sufficient for deep personalization: neither more context nor larger models guarantees better performance; effective personalization depends on selecting and integrating relevant evidence across domains. Direct retrieval of fine-grained behavioral records consistently outperforms compressed memory, while stronger personalization can come at the cost of privacy protection. These findings identify evidence selection, cross-domain integration, and privacy control as key challenges for personalized LLMs.
Chinese Translation
现有的个性化大型语言模型（LLM）基准主要依赖文本化的人物角色或孤立的行为信号，限制了对跨领域行为个性化的评估，而这种评估要求响应必须基于异质的日常活动。为了解决这一问题，我们提出了LUNAR，这是第一个评估LLM如何根据跨越通用日常领域（包括服装、食品、住房和出行）的长期应用交互历史个性化响应的基准。为了支持可扩展的基准构建，同时减轻数据稀疏性和隐私问题，LUNAR采用了一个基于真实行为模式的多阶段粗到细的合成流程。保真度分析显示，与其他合成基准相比，LUNAR与真实行为分布的对齐程度更高。在对19个主流LLM的实验中，发现访问行为日志是必要但不足以实现深度个性化：更多的上下文或更大的模型并不能保证更好的性能；有效的个性化依赖于跨领域选择和整合相关证据。直接检索细粒度的行为记录始终优于压缩记忆，而更强的个性化可能会以隐私保护为代价。这些发现将证据选择、跨领域整合和隐私控制确定为个性化LLM的关键挑战。
cs.AI / 17 / 2608.05248
WorldClaw: Agentic 3D Open-World Generation at Scale
WorldClaw：大规模自主3D开放世界生成
Abstract
Generating large-scale, freely explorable 3D worlds from open-ended text remains challenging because a system must jointly maintain global spatial coherence, rich local content, and explicit assets suitable for downstream editing and reuse. We present WorldClaw, a fully agentic, coarse-to-fine framework for open-world 3D scene generation. Planning agents translate a text prompt into a structured specification of regions, terrain, assets, materials, and spatial relations. WorldClaw then builds a globally coherent terrain foundation from semantic layouts, reusable assets, generative or procedural materials, and a region-aware height field. For detail-demanding regions, it generates terrain-conditioned compositions, reconstructs editable textured meshes, and recovers their placement on the terrain; render-based agents further refine terrain, objects, appearance, and contacts. Across diverse open-world prompts, WorldClaw produces large-scale scenes with coherent spatial organization, visually compelling local content, and editable instance-level assets while preserving a consistent global terrain structure.
Chinese Translation
从开放式文本生成大规模、可自由探索的3D世界仍然具有挑战性，因为系统必须共同维护全球空间一致性、丰富的局部内容以及适合后续编辑和重用的显式资产。我们提出了WorldClaw，一个完全自主的粗到细的开放世界3D场景生成框架。规划代理将文本提示转换为区域、地形、资产、材料和空间关系的结构化规范。然后，WorldClaw从语义布局、可重用资产、生成或程序化材料以及区域感知高度场构建全球一致的地形基础。对于需要细节的区域，它生成地形条件下的组合，重建可编辑的纹理网格，并恢复它们在地形上的放置；基于渲染的代理进一步优化地形、物体、外观和接触。在多样化的开放世界提示下，WorldClaw生成具有一致空间组织、大规模场景、视觉上引人注目的局部内容和可编辑的实例级资产，同时保持一致的全球地形结构。
cs.AI / 18 / 2608.05256
Posture and Sustainment Optimization Under Adversarial Uncertainty
对抗不确定性下的姿态与持续性优化
Abstract
Pre-commitment posture, the assignment of military assets to theater locations before conflict scenarios resolve, is a critical and formally unsolved problem in joint operational planning. Current practice relies on greedy heuristics that maximize value and ignore geographic coverage and are structurally vulnerable to adversaries that target high-strategic value locations. This paper presents a scenario-weighted adversarially robust posture optimization engine for the Posture and sustainability allocation (PSA) problem, modeled as a finite-horizon Markov Decision Process over assets, theater locations, and time steps. We introduce the Composite Expected Value (CEV) optimizer, which places assets by maximizing scenario-weighted expected posture efficiency over a distribution of threat scenarios, and the RobustCEV extension, which iterates against a Bayesian adversary that updates its targeting distribution in response to observed placement. Across three experiments in an Indo-Pacific basing environment with 20 assets and 5 theater locations, we demonstrate that: (1) the greedy baseline incurs a permanent 25.1% posture efficiency penalty due to geographic under-coverage and a 57.3% scenario-weighted readiness collapse under value-correlated adversarial threat; (2) the CEV optimizer recovers up to 19.8% efficiency over greedy when the threat distribution carries a geographic signal, with a curated set of 5 to 20 scenarios sufficient to capture the majority of this gain; and (3) the RobustCEV extension recovers up to 158% efficiency relative to a naive optimizer when an adaptive adversary employs a deceptive threat prior. All findings are validated using paired t-tests with Bonferroni correction and two-level variance decomposition, confirming that the performance gaps reported are structural properties of placement strategies rather than sampling artifacts.
Chinese Translation
预承诺姿态，即在冲突情景解决之前将军事资产分配到战区位置，是联合作战规划中一个关键且尚未正式解决的问题。目前的做法依赖于贪婪启发式方法，这些方法最大化价值但忽视地理覆盖，并且在结构上容易受到针对高战略价值位置的对手攻击。本文提出了一种场景加权的对抗稳健姿态优化引擎，用于姿态与可持续性分配（Posture and Sustainability Allocation, PSA）问题，建模为一个有限时域的马尔可夫决策过程，涉及资产、战区位置和时间步。我们引入了复合期望值（Composite Expected Value, CEV）优化器，通过最大化基于威胁场景分布的场景加权期望姿态效率来布置资产，并提出了RobustCEV扩展，它针对一个贝叶斯对手进行迭代，该对手根据观察到的布置更新其目标分布。在一个包含20个资产和5个战区位置的印太基地环境中进行的三次实验中，我们证明了：(1) 贪婪基线由于地理覆盖不足而导致永久性25.1%的姿态效率损失，并在价值相关的对抗威胁下导致57.3%的场景加权准备崩溃；(2) 当威胁分布具有地理信号时，CEV优化器相较于贪婪方法可恢复高达19.8%的效率，5到20个精心策划的场景集足以捕捉到大部分增益；(3) 当适应性对手采用欺骗性威胁先验时，RobustCEV扩展相较于天真的优化器可恢复高达158%的效率。所有发现均通过配对t检验与Bonferroni校正以及两级方差分解进行验证，确认所报告的性能差距是布置策略的结构性特征，而非抽样伪影。
cs.AI / 19 / 2608.05263
OrchestraBench: Evaluating Multi-Agent Orchestration Failure Modes, Recovery, and Decomposition Quality
OrchestraBench：评估多智能体编排的故障模式、恢复能力和分解质量
Abstract
Multi-agent orchestration frameworks are moving from demos to production, yet benchmarks typically report task accuracy without diagnosing why a pipeline failed, where a cascade began, or which routing decision caused the breakdown. OrchestraBench evaluates failure, recovery, and decomposition through a controlled, seed-reproducible failure-injection harness over templated enterprise workflows. It introduces cascade radius and per-failure-mode recovery as primary metrics and compares routing policies with bootstrap confidence intervals and paired tests. On a 26-case gold-labelled diagnostic, a keyword/flag router scored 0% on adversarial cases with misleading or missing surface flags, whereas an intent-reasoning model router scored 100%, matching the oracle. Controlled mechanism probes with a real Claude agent over a verifiable arithmetic dependency chain revealed three failure-handling tiers across five MAST modes: tool faults recovered fully (1.0), ambiguous delegation recovered partially (0.30), and three latent or semantic modes never recovered (0.0). This ordering persisted when the computation was reframed as a loan-approval workflow and across Sonnet, Opus, and Haiku, although absolute rates shifted with context. Blind retry reproduced latent faults and increased time to detection, indicating that detection and attribution are necessary for containment. Cascade radius increased with pipeline depth (mean 0.9 to 4.7 across depths 3-7). A trusted-state repair ablation showed that apparent containment gains primarily came from the trusted-state signal rather than autonomous detection. These results are controlled-chain mechanism probes, not domain-workload claims.
Chinese Translation
多智能体编排框架正从演示阶段转向生产阶段，然而基准测试通常仅报告任务准确性，而未能诊断管道失败的原因、级联的起始点或导致故障的路由决策。OrchestraBench通过一个受控的、可重复种子的故障注入工具对模板化的企业工作流评估故障、恢复和分解。它引入了级联半径和每种故障模式的恢复作为主要指标，并通过引导置信区间和配对测试比较路由策略。在一个包含26个案例的金标诊断中，关键词/标志路由器在具有误导性或缺失表面标志的对抗案例中得分为0%，而意图推理模型路由器得分为100%，与预期结果一致。通过在可验证的算术依赖链上使用真实的Claude代理进行的受控机制探测揭示了五种MAST模式下的三种故障处理层级：工具故障完全恢复（1.0），模糊委托部分恢复（0.30），而三种潜在或语义模式从未恢复（0.0）。这种排序在将计算重新框架为贷款批准工作流以及在Sonnet、Opus和Haiku中持续存在，尽管绝对比率随着上下文而变化。盲重试重现了潜在故障并增加了检测时间，表明检测和归因对于控制是必要的。级联半径随着管道深度的增加而增加（在深度3-7之间的平均值从0.9增加到4.7）。一个可信状态修复消融实验表明，明显的控制增益主要来自于可信状态信号，而非自主检测。这些结果是受控链机制探测，而非领域工作负载的声明。
cs.AI / 20 / 2608.05266
Agentic self-driving microscopy benchmarks support qualification but do not necessarily generalize to unseen tasks
自主驱动显微镜基准测试支持资格认证，但不一定能推广到未见任务
Abstract
Large language model agents are increasingly being developed to control a wide range of scientific characterization tools including microscopes and synchrotron beamlines. Research into agentic control of physical infrastructure is nascent and there are few well-established paradigms for how to engineer an agentic system. There are many choices to make when designing a microscopy agent, including the choice of LLM, the number of agents to use, agent responsibilities and delegation rules, retrieval-augmented generation parameters, and more. When designing and optimizing an agentic microscope controller, researchers not only want to ensure that the agent can correctly perform known tasks but also that the agent can generalize to new tasks that it has not encountered before. In this study, we develop a benchmark and trace-logging framework that reveals a) how different choices of agent architecture impact performance at microscopy tasks and b) the limitations of benchmarks for predicting if a particular agent will perform well on unseen microscopy tasks. The framework was used to evaluate one-, two-, and three-agent graph topologies, five LLMs, RAG and context parameters, and operational constraints across 53 microscopy benchmark tests. In total, 105 agent configurations, 1,949 individual test runs, and 49,109 RAG retrievals were recorded. Direct comparisons showed clear differences in latency, token use, cost, and failure mode between configurations. However, surrogate models trained on agent architecture and test results did not reliably predict an agent's performance on new, unseen tasks. These results show that these benchmarks are useful for qualification, regression testing, diagnosis, and direct comparison, but the current heterogeneous test suite does not support a task-independent global configuration model.
Chinese Translation
大型语言模型代理正在不断发展，以控制包括显微镜和同步辐射光束线在内的广泛科学表征工具。对物理基础设施的自主控制的研究尚处于起步阶段，目前关于如何设计自主系统的成熟范式较少。在设计显微镜代理时，有许多选择需要做出，包括LLM的选择、使用的代理数量、代理的职责和委派规则、检索增强生成参数等。在设计和优化自主显微镜控制器时，研究人员不仅希望确保代理能够正确执行已知任务，还希望代理能够推广到其未曾遇到的新任务。在本研究中，我们开发了一个基准测试和追踪日志框架，该框架揭示了a) 不同代理架构选择如何影响显微镜任务的性能，以及b) 基准测试在预测特定代理在未见显微镜任务上表现良好的局限性。该框架用于评估一、二、三代理图拓扑、五种LLM、RAG和上下文参数以及53个显微镜基准测试中的操作约束。总共记录了105种代理配置、1,949个单独测试运行和49,109次RAG检索。直接比较显示了不同配置之间在延迟、令牌使用、成本和失败模式上的明显差异。然而，基于代理架构和测试结果训练的替代模型并未可靠地预测代理在新未见任务上的表现。这些结果表明，这些基准测试对于资格认证、回归测试、诊断和直接比较是有用的，但当前异构测试套件并不支持任务独立的全局配置模型。
cs.AI / 21 / 2608.05359
CASCADE: An Agentic Regulatory Network Framework for Patient-Data-Validated Downstream Perturbation Prediction
CASCADE：一种基于代理的调控网络框架，用于患者数据验证的下游扰动预测
Abstract
CASCADE is an agentic framework that predicts downstream transcriptional effects of gene perturbation from precomputed ARACNe regulatory networks, exposed via MCP. Prior work validates such tools by checking whether predicted genes are known cancer genes (membership); we instead test whether the predicted direction of change matches reality, using focal-gene copy-number amplification as a dosage-based proxy for the inverse of knockdown against real TCGA patient tumor data. For MYC, CASCADE's predicted knockdown targets show strong concordance with real amplified-vs-non-amplified tumor expression across three cancer types (BRCA: 90.0%, COAD: 72.0%, STAD: 85.7%; all p<0.0013), well above permutation baselines, surviving a PAM50 subtype control and replicating in an independent cohort (METABRIC, 87.2%). Compared against curated MSigDB gene-set baselines via Fisher's exact test, CASCADE's accuracy is not shown to exceed existing public knowledge of MYC- or E2F-driven biology, though its gene-specific direction-calling clearly outperforms a naive uniform guess. Extending to fifteen additional genes, validation proves gene-specific rather than universal: proliferation-machinery regulators mostly replicate, while lineage-identity transcription factors and one cyclin-D paralog (CCND2) consistently fail, a pattern we discuss as a hedged, post-hoc hypothesis. We separately benchmark whether an LLM-based agent correctly grounds natural-language requests into CASCADE's real MCP tool calls. Across 35 queries, a documented local model reaches 71.4% exact match (85.7% for a larger model); schema and gene-alias failures are resolved by scale or server-side correction, but both models confidently default to the wrong perturbation type on ambiguous queries, a failure a targeted fix could not resolve because its trigger condition never occurs.
Chinese Translation
CASCADE 是一种代理框架，能够从预计算的 ARACNe 调控网络中预测基因扰动的下游转录效应，这些网络通过 MCP 显示。之前的研究通过检查预测的基因是否为已知癌症基因（成员资格）来验证此类工具；而我们则测试预测的变化方向是否与现实相符，使用焦点基因拷贝数扩增作为对真实 TCGA 患者肿瘤数据中敲低的剂量基础代理。对于 MYC，CASCADE 预测的敲低靶点在三种癌症类型（BRCA: 90.0%，COAD: 72.0%，STAD: 85.7%；均 p<0.0013）中与真实的扩增与非扩增肿瘤表达显示出强一致性，远高于置换基线，且在 PAM50 亚型控制下存活，并在独立队列（METABRIC，87.2%）中复制。与经过整理的 MSigDB 基因集基线通过 Fisher 精确检验相比，CASCADE 的准确性未显示超过现有的 MYC 或 E2F 驱动生物学的公共知识，尽管其基因特异性的方向判断明显优于简单的均匀猜测。扩展到另外十五个基因的验证证明是基因特异性的而非普遍性的：增殖机制调控因子大多复制，而谱系身份转录因子和一个细胞周期蛋白 D 同源物（CCND2）则始终失败，这一模式我们讨论为一种有保留的事后假设。我们单独基准测试一个基于 LLM 的代理是否能正确将自然语言请求转化为 CASCADE 的真实 MCP 工具调用。在 35 个查询中，一个记录的本地模型达到 71.4% 的完全匹配（较大模型为 85.7%）；模式和基因别名的失败通过规模或服务器端修正得到解决，但两个模型在模糊查询上自信地默认错误的扰动类型，这一失败无法通过有针对性的修复解决，因为其触发条件从未发生。
cs.AI / 22 / 2608.05367
Counterfactual Analysis via Large Language Models
通过大型语言模型进行反事实分析
Abstract
Counterfactual analysis aims to predict potential outcomes under hypothetical scenarios, offering valuable insights for decision-making. This paper investigates the application of large language models (LLMs), specifically the GPT-3.5 model, for counterfactual analysis. We focus on the online lending context, where the counterfactual return on investment (ROI) is crucial for evaluating different interest rate schemes. We begin by assessing the predictive performance of GPT and comparing it with advanced machine learning algorithms. The results show that prompt engineering can significantly enhance GPT's predictions, with the R-squared increasing from 1.97% to 2.84%, closely approaching the 3.48% achieved by gradient-boosted regression. Subsequently, we utilize GPT to generate counterfactual ROIs under a set of alternative interest rates. GPT exhibits logical coherence and causal reasoning in its responses. The findings underscore the potential of LLMs as effective tools for counterfactual analysis in online lending, suggesting broader applications for LLMs in various predictive and decision-making contexts.
Chinese Translation
反事实分析旨在预测假设情境下的潜在结果，为决策提供有价值的见解。本文探讨了大型语言模型（LLMs），特别是GPT-3.5模型在反事实分析中的应用。我们关注在线借贷的背景，在该背景下，反事实投资回报率（ROI）对于评估不同利率方案至关重要。我们首先评估了GPT的预测性能，并将其与先进的机器学习算法进行比较。结果表明，提示工程可以显著提升GPT的预测能力，R平方值从1.97%提高到2.84%，接近梯度提升回归所达到的3.48%。随后，我们利用GPT生成一组替代利率下的反事实ROI。GPT在其响应中表现出逻辑一致性和因果推理能力。研究结果强调了LLMs作为在线借贷反事实分析有效工具的潜力，并暗示LLMs在各种预测和决策情境中的更广泛应用。
cs.AI / 23 / 2608.05375
DoctorAgents: an agentic framework to iteratively refine AutoML pipeline for small clinical temporal data
DoctorAgents：一个迭代优化小型临床时间数据的AutoML管道的智能框架
Abstract
Clinical machine learning (ML) has the potential to support high-stakes medical decision-making, but reliable deployment is often constrained by scarce, heterogeneous, and temporal complexity. Developing effective ML pipelines for such data remains time-consuming and error-prone, while existing automated machine learning (AutoML) systems only partially address this challenge because they largely rely on brute-force search over predefined spaces and lack explicit reasoning and memory. We therefore reformulate AutoML for small clinical data from exhaustive search to reasoning-driven refinement. We propose DoctorAgents, an agentic AI framework that autonomously constructs and optimizes end-to-end ML pipelines through specialized large language model (LLM) agents for generation, validation, and refinement. DoctorAgents backpropagates natural-language feedback through textual gradient descent to perform targeted updates without exhaustive search. Experiments across diverse clinical tasks show that DoctorAgents consistently outperforms established AutoML baselines while producing more interpretable task-specific representations.
Chinese Translation
临床机器学习（ML）有潜力支持高风险的医疗决策，但可靠的部署通常受到稀缺、异质和时间复杂性的限制。为此类数据开发有效的ML管道仍然耗时且容易出错，而现有的自动化机器学习（AutoML）系统仅部分解决了这一挑战，因为它们在很大程度上依赖于对预定义空间的穷举搜索，并缺乏明确的推理和记忆。因此，我们将小型临床数据的AutoML从穷举搜索重新构建为基于推理的优化。我们提出了DoctorAgents，一个智能AI框架，通过专门的大型语言模型（LLM）代理自主构建和优化端到端的ML管道，进行生成、验证和优化。DoctorAgents通过文本梯度下降反向传播自然语言反馈，以执行有针对性的更新，而无需进行穷举搜索。跨多种临床任务的实验表明，DoctorAgents在性能上始终优于现有的AutoML基准，同时生成更具可解释性的任务特定表示。
cs.AI / 24 / 2608.05381
C$^3$PO: Evaluating Cross-Modal Composition and Counterfactual Performance in Omnimodal Models
C$^3$PO：评估全模态模型中的跨模态组合和反事实表现
Abstract
Current Multimodal Large Language Models (MLLMs) can process diverse sensory inputs, yet their reasoning remains heavily biased toward a dominant modality, resulting in brittle cross-modal reasoning. We introduce C$^3$PO, a benchmark of 3,404 samples spanning video, audio, image, and text, evaluating two abilities: information composition (fusing dispersed evidence) and counterfactual conflict (resolving deliberate contradictions). C$^3$PO's paired IC/CC structure and four-tier design enable targeted diagnosis of when and why cross-modal reasoning fails. Built through a fully automatic pipeline using 25 logically grounded templates, C$^3$PO reveals that while humans achieve 88.64% accuracy, the best model (Gemini-3.1-Pro) reaches only 73.17%, with open-source models collapsing under conflict. Through attention probes, we find 86-95% of failures stem from modality dominance: models commit to one modality while ignoring contradictory evidence, concentrating 87-95% of attention on text. Mid-layer attention entropy predicts correctness-sustained exploration succeeds, premature collapse fails. The 56-point accuracy gap between equally complex templates reveals that performance depends on modalities' structural roles in conflict resolution, not combinations. These findings show multimodal perception does not guarantee robust reasoning; architectures must enable sustained cross-modal attention to avoid premature
Chinese Translation
当前的多模态大型语言模型（MLLMs）能够处理多样的感官输入，但它们的推理仍然严重偏向于主导模态，导致跨模态推理脆弱。我们引入了C$^3$PO，一个包含3,404个样本的基准，涵盖视频、音频、图像和文本，评估两种能力：信息组合（融合分散的证据）和反事实冲突（解决故意矛盾）。C$^3$PO的配对IC/CC结构和四层设计使得能够针对性地诊断跨模态推理失败的时机和原因。C$^3$PO通过使用25个逻辑基础模板的全自动流程构建，揭示了人类的准确率为88.64%，而最佳模型（Gemini-3.1-Pro）仅达到73.17%，开源模型在冲突下崩溃。通过注意力探测，我们发现86-95%的失败源于模态主导性：模型坚持于一种模态，同时忽视矛盾证据，将87-95%的注意力集中在文本上。中层注意力熵预测，持续探索的正确性成功，而过早崩溃则失败。相同复杂度模板之间56个百分点的准确率差距表明，性能依赖于模态在冲突解决中的结构角色，而非组合。这些发现表明，多模态感知并不保证稳健的推理；架构必须能够实现持续的跨模态注意力，以避免过早崩溃。
cs.AI / 25 / 2608.05391
Adaptive Arena-based Contestable Argumentative Network-of-Experts for Open-Ended Care Plan Coordination
基于自适应竞技场的可争辩专家网络用于开放式护理计划协调
Abstract
Care plan coordination demands synthesizing heterogeneous clinical, functional, and psychosocial information across multiple professional disciplines, where monolithic LLM pipelines cannot perform in a transparent or safe manner. We present CANOE (Contestable Argumentative Network-of-Experts), a multi-agent neuro-symbolic framework that addresses these limitations through five modules: complexity assessment, adaptive team recruitment, role-based argumentative computation via an Arena-based Quantitative Bipolar Argumentation Framework (A-QBAF), human-in-the-loop contestation, and care-plan synthesis. Role-specialized agents generate supporting and attacking arguments for candidate interventions; conflicts are resolved through arena-based clash resolution before acceptability scores propagate across the argumentation graph. Care planners may accept, reject, edit, or add arguments, and the framework will deterministically recompute the final plan. Evaluation on Discharge Me! and MedicalRAG using ROUGE-L, AlignScore, MEDCON F1, FKGL, and LLM-as-a-judge shows that medically fine-tuned models achieve the strongest clinical correctness and safety, while CANOE's argumentative structure provides faithful explanation and human contestability.
Chinese Translation
护理计划协调需要综合来自多个专业领域的异质临床、功能和心理社会信息，而单一的LLM（大型语言模型）管道无法以透明或安全的方式执行此任务。我们提出了CANOE（可争辩的专家网络），这是一个多智能体神经符号框架，通过五个模块解决这些限制：复杂性评估、自适应团队招募、基于角色的论证计算（通过竞技场定量双极论证框架 A-QBAF）、人机协作争辩和护理计划综合。角色专门化的智能体为候选干预措施生成支持和攻击论据；在可接受性分数在论证图中传播之前，通过竞技场冲突解决来解决冲突。护理计划制定者可以接受、拒绝、编辑或添加论据，框架将确定性地重新计算最终计划。在使用ROUGE-L、AlignScore、MEDCON F1、FKGL和LLM-as-a-judge进行的Discharge Me!和MedicalRAG评估中，经过医学微调的模型实现了最强的临床正确性和安全性，而CANOE的论证结构提供了可信的解释和人类可争辩性。
cs.AI / 26 / 2608.05411
Evaluating and Improving Pedagogical Fit in LLM-Based AI Tutors with the Pedagogical Suitability Index
评估和改善基于大型语言模型的人工智能辅导员的教学适配性：教学适宜性指数
Abstract
Large language models (LLMs) are increasingly used as AI tutors, but a correct answer is not always a pedagogically appropriate one. In classroom learning, effective help depends not only on correctness, but also on whether a response matches the learner's current foundation, the course sequence, and the timing of concept introduction. Existing evaluations focus mainly on answer quality, leaving this instructional fit under-measured. We present the Pedagogical Suitability Index (PSI), a composite metric of six theory-informed sub-scores that evaluates how well LLM-generated tutoring responses align with learner readiness and curricular progression, and we further use PSI as a structured feedback signal for response improvement. We evaluate four LLM tutors (ChatGPT, Gemini, Gemma4, and Qwen3) across 240 scenario-based evaluations using paired standard and defective prompts, then apply a PSI-guided regeneration protocol to 62 weak-performing cases. Baseline differences across the four tested models were modest overall (PSI range: 0.557 to 0.638), and open-weight and closed models did not exhibit a clear separation in pedagogical fit. Under the tested prompt perturbations, overall PSI remained largely stable (Delta = -0.002), though sub-score trade-offs emerged. More importantly, PSI-guided feedback substantially improved weak-performing cases: 51 of 62 cases improved (82.3%). Focused manual evaluation of the 62 PSI-selected weak cases provides initial evidence that the identified weaknesses are instructionally meaningful and that many PSI-guided regenerations correspond to human-judged improvement. These results suggest that learner- and curriculum-aware alignment may matter more for effective tutoring than model category alone, and that such alignment is both measurable and improvable.
Chinese Translation
大型语言模型（LLMs）越来越多地被用作人工智能辅导员，但正确的答案并不总是教学上合适的答案。在课堂学习中，有效的帮助不仅依赖于答案的正确性，还取决于回应是否与学习者当前的基础、课程顺序以及概念引入的时机相匹配。现有评估主要集中在答案质量上，导致这种教学适配性未得到充分测量。我们提出了教学适宜性指数（Pedagogical Suitability Index, PSI），这是一个由六个理论驱动的子分数组成的复合指标，用于评估LLM生成的辅导回应与学习者准备情况和课程进展的匹配程度，并进一步将PSI作为回应改进的结构化反馈信号。我们在240个基于场景的评估中评估了四个LLM辅导员（ChatGPT、Gemini、Gemma4和Qwen3），使用成对的标准和缺陷提示，然后对62个表现较弱的案例应用PSI引导的再生成协议。四个测试模型之间的基线差异总体上较小（PSI范围：0.557至0.638），开放权重模型和封闭模型在教学适配性上没有明显区分。在测试的提示扰动下，整体PSI保持相对稳定（Delta = -0.002），尽管出现了子分数的权衡。更重要的是，PSI引导的反馈显著改善了表现较弱的案例：62个案例中有51个得到了改善（82.3%）。对62个PSI选择的弱案例的集中手动评估提供了初步证据，表明识别出的弱点在教学上是有意义的，并且许多PSI引导的再生成与人类评判的改善相对应。这些结果表明，关注学习者和课程的对齐可能比模型类别本身对有效辅导更为重要，并且这种对齐是可测量和可改善的。
cs.AI / 27 / 2608.05418
Negotiating Risk Boundaries in AI for Policing Through Mixed-Stakeholder Deliberation
通过混合利益相关者协商在警务人工智能中协商风险边界
Abstract
AI tools are being increasingly adopted in policing in the UK and worldwide. Racial bias is a known and well-documented risk, yet representatives of affected communities are rarely included in decisions about AI adoption. We present results from a mixed-stakeholder deliberation workshop bringing together 30 community representatives, police officers, and academics to assess the risks of 13 AI use cases in policing, with an explicit focus on racial bias. We found that participants were broadly open to AI adoption, rejecting only three use cases outright, most notably recidivism risk assessment, where objections targeted the premise rather than the implementation. Our analysis reveals that foregrounding racial equity did not narrow the deliberation. Instead, discussions gravitated toward a fundamental set of questions: does this tool actually work, will it deliver genuine benefit, and will that benefit extend to everyone? This integrated reasoning, reminiscent of the curb-cut effect in inclusive design, highlights the benefit of incorporating the racial bias lens into the risk-benefit analysis of AI use cases from the outset.
Chinese Translation
人工智能工具在英国及全球的警务中被越来越多地采用。种族偏见是一种已知且有充分文献记录的风险，但受影响社区的代表在有关人工智能采用的决策中很少被纳入。我们展示了一次混合利益相关者协商研讨会的结果，该研讨会汇集了30名社区代表、警务人员和学者，以评估13个警务中人工智能应用案例的风险，特别关注种族偏见。我们发现，参与者总体上对人工智能的采用持开放态度，仅明确拒绝了三个应用案例，最引人注目的是再犯风险评估，其中反对意见主要针对其前提而非实施。我们的分析显示，强调种族公平并没有缩小协商范围。相反，讨论围绕一组基本问题展开：这个工具真的有效吗？它会带来真正的好处吗？这些好处会惠及每个人吗？这种综合推理，类似于包容性设计中的人行道切口效应，突显了在人工智能应用案例的风险收益分析中从一开始就纳入种族偏见视角的好处。
cs.AI / 28 / 2608.05439
SCP-NL2TL: Selective Conformal Prediction with Semantic Verification for Natural Language to Temporal Logic Specifications
SCP-NL2TL：具有语义验证的选择性符合预测用于自然语言到时间逻辑规范的转换
Abstract
Translating natural language instructions into machine-interpretable formal specifications enables robots and autonomous systems to plan, reason, and formally verify their behavior. However, existing translation models typically generate a specification for every input, even when the result is unreliable or fails to capture the user's intent, creating risks in safety-critical applications. Inspired by selective conformal prediction, we propose a selective translation framework that not only generates formal specifications but also determines when they can be trusted. Reliability is scored by two complementary black-box signals, the fidelity of the specification back-translated into natural language and the dispersion of repeated translations under exact semantic equivalence, which fail on different errors and jointly separate incorrect translations more sharply than either alone. Conformal risk control calibrates this score into a decision that accepts a specification or abstains, with a distribution-free bound on the rate at which incorrect specifications are accepted for execution, and a conformal anomaly detector on instruction embeddings screens out-of-distribution inputs before any translation is attempted. The proposed framework is general across formal specification languages, with experiments on Signal Temporal Logic (STL), Linear Temporal Logic (LTL), and geometric Spatio-Temporal Logic (SpaTiaL) demonstrating improved translation reliability, robustness under the evaluated cross-tier shifts, and effective uncertainty-aware abstention. This work establishes a foundation for trustworthy natural language interfaces by enabling AI systems to recognize when generated specifications may not be reliable.
Chinese Translation
将自然语言指令翻译为机器可解释的形式规范，使机器人和自主系统能够规划、推理并正式验证其行为。然而，现有的翻译模型通常会为每个输入生成一个规范，即使结果不可靠或未能捕捉用户的意图，这在安全关键应用中带来了风险。受到选择性符合预测的启发，我们提出了一种选择性翻译框架，不仅生成形式规范，还确定何时可以信任这些规范。通过两个互补的黑箱信号来评估可靠性：将规范反向翻译为自然语言的保真度和在精确语义等价下重复翻译的分散度，这两者在不同错误下表现不佳，并共同更清晰地区分不正确的翻译。符合风险控制将这一评分校准为接受或放弃规范的决策，并对执行时接受不正确规范的速率提供无分布界限，同时，符合异常检测器在任何翻译尝试之前筛选出分布外输入。所提出的框架在形式规范语言中具有广泛的适用性，针对信号时间逻辑（Signal Temporal Logic, STL）、线性时间逻辑（Linear Temporal Logic, LTL）和几何时空逻辑（geometric Spatio-Temporal Logic, SpaTiaL）的实验表明，翻译可靠性得到了改善，在评估的跨层次变化下表现出鲁棒性，并有效地实现了不确定性感知的放弃。该研究为可信的自然语言接口奠定了基础，使人工智能系统能够识别生成的规范何时可能不可靠。
cs.AI / 29 / 2608.05455
Stochasticity Is Not the Hard Part: Reduction and Complexity in Instructional Sequencing over Prerequisite DAGs
随机性并不是难点：在前提有向无环图上的教学顺序的简化与复杂性
Abstract
When a student must learn concepts connected by prerequisite dependencies, when does the order of instruction matter, and what does it cost to find the best one? We study instructional sequencing as a stochastic shortest-path problem in which attempting a concept succeeds with a state-dependent probability and failure leaves the learner state unchanged. We first prove that this stochasticity can be eliminated exactly: the problem collapses to a deterministic shortest-path problem on the lattice of prerequisite order ideals, preserving optimal values and actions. The collapse removes stochastic complexity but not combinatorial complexity: optimal sequencing remains NP-hard -- via reduction from feedback arc set in tournaments -- even with no prerequisite edges, unit costs, uniform binary nonnegative transfer, and success probabilities at least $1/2$. Hardness is not uniform: when realizable transfer preferences remain jointly acyclic with the prerequisites, any topological order of the residual joint graph is optimal, and fixed prerequisite width yields polynomial-time exact dynamic programming. A computable diagnostic, $m\Delta$, bounds the value of sequencing before optimization. On 70,893 interactions from an introductory CS course, the diagnostic certifies a doubly easy regime -- little value to optimize and little space to search -- while constructed transfer instances realize the challenging regime, where myopic sequencing suffers large regret yet exact A* with a consistent heuristic expands only linearly many states on that family.
Chinese Translation
当学生必须学习通过前提依赖关系连接的概念时，教学顺序何时重要，以及寻找最佳顺序的成本是多少？我们将教学顺序研究视为一个随机最短路径问题，其中尝试一个概念的成功概率依赖于状态，而失败则使学习者的状态保持不变。我们首先证明，这种随机性可以被完全消除：该问题简化为前提顺序理想的格上的确定性最短路径问题，保持最优值和行动。简化消除了随机复杂性，但并未消除组合复杂性：即使在没有前提边、单位成本、均匀的非负转移和成功概率至少为 $1/2$ 的情况下，最优顺序仍然是 NP-hard 的——通过从锦标赛中的反馈弧集的归约。困难性并不均匀：当可实现的转移偏好与前提共同保持无环时，残余联合图的任何拓扑顺序都是最优的，而固定的前提宽度则导致多项式时间的精确动态规划。一个可计算的诊断指标 $m riangle$ 界定了优化前顺序的价值。在来自一门入门计算机科学课程的 70,893 次交互中，该诊断证明了一个双重简单的状态——优化的价值很小，搜索的空间也很小——而构造的转移实例则实现了具有挑战性的状态，在这种状态下，短视的顺序会遭受巨大的遗憾，但使用一致启发式的精确 A* 仅在该家族中扩展线性数量的状态。
cs.AI / 30 / 2608.05466
Recursive Synthesis for Long-Horizon Terminal Tasks
长时间终端任务的递归合成
Abstract
High-quality long-horizon training data for terminal agents is expensive to produce, often costing hundreds to thousands of dollars per task, because each task must keep the instruction, environment, reference solution, and verifier mutually consistent. Human authoring does not scale, and direct generation with large language models (LLMs) often breaks these dependencies. We present Recursive Synthetic Terminal Tasks (RST), a recursive verified synthesis framework for constructing long-horizon terminal-agent tasks at scale. Starting from verified seed tasks, RST extends the reference solution, realigns the verifier and instruction to the new workflow, validates the result in a fresh sandbox, and reuses accepted tasks as seeds for subsequent rounds. Across fifteen recursive rounds, RST produces 37,484 synthesized terminal-agent tasks at roughly \$0.05 per task. Task difficulty increases substantially over rounds: the median reference solution grows from 67 to 374 lines, the median number of executed commands grows from 40 to 244, and DeepSeek-V4-Pro pass@4 drops from 90\% at $R_1$ to 2.5\% at $R_{15}$. To demonstrate training utility, we collect rejection-sampled Qwen3.5 trajectories on the synthesized tasks and use them for supervised fine-tuning. Fine-tuning on these trajectories improves Qwen3.5-27B and Qwen3.5-122B-A10B by up to 10 points on Terminal-Bench~2, Terminal-Bench Hard, and Long-Horizon Terminal Bench, while agentic PPO lifts Qwen3.5-27B to 49.44\%, 32.00\%, and 22.07\% on the three benchmarks, corresponding to relative gains of 20.0\%, 41.2\%, and 21.9\% over the base model. Moreover, after 15 rounds, the recursion shows no ceiling: synthesis yield and validation rates remain stable as difficulty keeps climbing, indicating that the process can continue well beyond the scale reported here.
Chinese Translation
高质量的长时间训练数据对于终端代理的生成成本高昂，通常每个任务需要花费数百到数千美元，因为每个任务必须保持指令、环境、参考解决方案和验证器之间的相互一致性。人工创作无法扩展，而使用大型语言模型（LLMs）进行直接生成往往会破坏这些依赖关系。我们提出了递归合成终端任务（Recursive Synthetic Terminal Tasks, RST），这是一种递归验证合成框架，用于大规模构建长时间终端代理任务。RST从经过验证的种子任务开始，扩展参考解决方案，重新对齐验证器和指令以适应新的工作流程，在一个新的沙盒中验证结果，并将接受的任务作为后续轮次的种子进行重用。在十五轮递归中，RST以每个任务约0.05美元的成本生成了37,484个合成的终端代理任务。任务的难度在轮次中显著增加：中位数参考解决方案的行数从67增长到374，执行命令的中位数数量从40增长到244，而DeepSeek-V4-Pro在$R_1$时的通过率为90\%，在$R_{15}$时降至2.5\%。为了展示训练的实用性，我们在合成任务上收集了拒绝采样的Qwen3.5轨迹，并将其用于监督微调。在这些轨迹上的微调使得Qwen3.5-27B和Qwen3.5-122B-A10B在Terminal-Bench~2、Terminal-Bench Hard和Long-Horizon Terminal Bench上提高了最多10分，而代理PPO将Qwen3.5-27B在这三个基准上的表现提升至49.44\%、32.00\\%和22.07\\%，相对于基础模型的相对增益分别为20.0\\%、41.2\\%和21.9\\%。此外，在经过15轮后，递归过程没有上限：合成产量和验证率保持稳定，尽管难度不断上升，这表明该过程可以继续进行，远超此处报告的规模。
cs.AI / 31 / 2608.05490
Innovation-Residual Auditing of Autonomous Analysis Agents: Localization, Detection Limits, Error Control, and Identifiability
自主分析代理的创新残差审计：定位、检测极限、错误控制与可识别性
Abstract
Autonomous agents now carry out entire data analyses, selecting cohorts, joining tables, and fitting models with little step-by-step supervision. When such an analysis turns out to be wrong, someone must determine which operation caused it. A recent approach does this without any labelled mistakes, learning instead from analyses known to be sound and flagging operations that depart from what that model predicts; how reliable such audits are has not been studied. This paper supplies that analysis. The choice of score determines whether an error can be localized at all. If each operation is scored by how surprising it is given the operation immediately preceding it, then operations that merely inherit an earlier error are indistinguishable from correct ones, so one mistake produces one flag; scores computed against a longer reconstruction of the intended analysis instead spread a single mistake across many operations. We quantify how far they spread, and how to choose the comparison length when an error accumulates gradually rather than at once. We then give procedures that control the proportion of falsely flagged operations within a single audited analysis, requiring only that sound analyses be exchangeable rather than that the fitted model be correct, and we quantify how much the guarantees weaken when the model is imperfect or when the analysis was selected for review in a way that depends on its content. Finally we establish a limit on what any such audit can report: errors below a certain magnitude cannot be attributed at all, being indistinguishable from ordinary variation among sound analyses. This limit falls so slowly as more sound analyses are collected that at the representation sizes now in use a hundredfold increase reduces it by under two percent, so the dimension of the representation rather than the volume of training data is the binding constraint.
Chinese Translation
自主代理现在能够独立进行整个数据分析，选择样本、连接表格并拟合模型，几乎不需要逐步监督。当这样的分析结果出错时，必须确定是哪个操作导致了错误。最近的一种方法在没有任何标记错误的情况下实现了这一点，而是从已知正确的分析中学习，并标记那些偏离模型预测的操作；然而，这种审计的可靠性尚未得到研究。本文提供了这一分析。评分的选择决定了错误是否能够被定位。如果每个操作的评分基于其在前一个操作下的惊讶程度，那么仅仅继承早期错误的操作与正确的操作无法区分，因此一个错误会产生一个标记；而根据更长的重构分析计算的评分则会将一个错误分散到多个操作中。我们量化了它们的扩散程度，以及在错误逐渐累积而非一次性发生时如何选择比较长度。随后，我们给出了控制单个审计分析中错误标记操作比例的程序，仅要求正确的分析是可交换的，而不是要求拟合的模型是正确的，并量化了当模型不完美或分析是基于内容选择进行审查时，保证的强度减弱程度。最后，我们确定了任何此类审计能够报告的限制：低于某一幅度的错误无法被归因，因为它们与正常的声音分析变异无法区分。随着更多正确分析的收集，这一限制下降得非常缓慢，以至于在当前使用的表示规模下，增加一百倍的样本量仅使其减少不到百分之二，因此表示的维度而非训练数据的体量成为了约束条件。
cs.AI / 32 / 2608.05519
EcoAgent-Bench: Evaluating Economic Decision-Making in Budget-Constrained LLM Agents
EcoAgent-Bench：评估预算受限的LLM代理中的经济决策
Abstract
Agent benchmarks usually measure task completion and treat resource use as an auxiliary statistic. In deployment, however, the choice among a local lookup, broad search, composite research tool, stronger model, or human escalation is part of the task itself. We introduce EcoAgent-Bench, in which every task specifies priced actions and an explicit budget. Its 304 real-derived tasks span five families adapted from GAIA, HotpotQA, and MuSiQue, and test four decisions: avoiding unnecessary escalation, escalating when local evidence is insufficient, selecting a model tier, and stopping on unsupported premises. We evaluate seven LLM agents in tool-API and workspace-CLI settings, together with four oracle scripted controls. Micro-averaged accuracy rewards one-sided policies: always-escalate controls achieve high micro success while failing save-oriented tasks. We therefore also report an economic-consistency score (the worse of accuracy on upgrade-oriented and save-oriented family groups) which exposes this failure. Tool-API agents attain only 3.9-24.0% micro strict success (at most 7.3% economic consistency), often either stopping before warranted escalation or overspending on cheap tasks. A threshold-crossing budget sweep changes GPT-5.4's escalation rate from 0% to only 3%. These results show that completion under a budget and economical action selection are distinct properties. We release the task bundle, transformation pipeline, frozen evaluation environments, and integrity-bound result artifacts needed to study both.
Chinese Translation
代理基准通常测量任务完成情况，并将资源使用视为辅助统计。然而，在实际部署中，选择本地查找、广泛搜索、复合研究工具、更强模型或人工升级的决策本身就是任务的一部分。我们引入了EcoAgent-Bench，其中每个任务都指定了有价行动和明确的预算。其304个真实派生任务涵盖了五个家族，改编自GAIA、HotpotQA和MuSiQue，并测试四个决策：避免不必要的升级、在本地证据不足时进行升级、选择模型层级，以及在不支持的前提下停止。我们在工具API和工作区CLI设置中评估了七个LLM代理，以及四个预设控制。微平均准确率奖励单边政策：始终升级的控制在微观成功上表现良好，但在节省导向任务上失败。因此，我们还报告了经济一致性分数（在升级导向和节省导向家族组上的准确率较差者），揭示了这一失败。工具API代理的微观严格成功率仅为3.9%-24.0%（经济一致性最多为7.3%），通常在需要升级之前停止，或在廉价任务上超支。阈值跨越预算扫描将GPT-5.4的升级率从0%提高到仅3%。这些结果表明，在预算下的完成和经济行动选择是不同的属性。我们发布了研究这两者所需的任务包、转换管道、冻结评估环境和完整性绑定结果文档。
cs.AI / 33 / 2608.05541
Hyper-ES: Effective Evolution Strategies for LLM Reasoning via Descent Direction Merging
Hyper-ES：通过下降方向合并实现大语言模型推理的有效进化策略
Abstract
Evolution Strategy (ES) is a promising alternative to gradient-based fine-tuning for resource-constrained Large Language Model (LLM) reasoning. However, directly applying ES to billion-parameter LLMs is highly ineffective. In such high-dimensional parameter spaces, most random perturbations are nearly orthogonal to useful update directions, leading to unstable optimization. We propose Hyper-ES, a subspace-based ES framework that avoids the weakness of ES in full-parameter search while exploiting its strength in low-dimensional optimization. Instead of asking ES to discover useful directions from random perturbations in the LLM parameter space, Hyper-ES first performs a small number of inexpensive gradient-based fine-tuning runs to obtain descent directions. Although each direction may provide only a limited improvement on its own, their span forms a compact adaptation subspace that captures useful reasoning updates. Hyper-ES then applies CMA-ES to optimize layer-wise DARE-TIES merging coefficients within this subspace, allowing ES to search over combinations of meaningful descent directions rather than over arbitrary full-model perturbations. We evaluate Hyper-ES on three Qwen2.5-Instruct and DeepSeek-R1-Distill backbones across six mathematical reasoning datasets. Results show that Hyper-ES consistently outperforms GRPO-LoRA by 1% while requiring 10% fewer space-consuming gradient updates. Code at https://github.com/kuangrepi/Hyper-ES.
Chinese Translation
进化策略（Evolution Strategy, ES）是资源受限的大语言模型（Large Language Model, LLM）推理中一种有前景的替代梯度微调的方法。然而，直接将ES应用于十亿参数的LLM效果极其有限。在如此高维的参数空间中，大多数随机扰动与有用的更新方向几乎是正交的，导致优化不稳定。我们提出了Hyper-ES，这是一种基于子空间的ES框架，避免了ES在全参数搜索中的弱点，同时利用其在低维优化中的优势。Hyper-ES并不是要求ES从LLM参数空间中的随机扰动中发现有用的方向，而是首先进行少量低成本的基于梯度的微调运行，以获取下降方向。尽管每个方向单独可能只提供有限的改进，但它们的跨度形成一个紧凑的适应子空间，捕捉有用的推理更新。然后，Hyper-ES在该子空间内应用CMA-ES优化逐层的DARE-TIES合并系数，使ES能够在有意义的下降方向组合上进行搜索，而不是在任意的全模型扰动上进行搜索。我们在三个Qwen2.5-Instruct和DeepSeek-R1-Distill骨干网络上评估了Hyper-ES，涵盖六个数学推理数据集。结果表明，Hyper-ES始终比GRPO-LoRA提高1%的性能，同时所需的空间消耗梯度更新减少了10%。代码可在 https://github.com/kuangrepi/Hyper-ES 获取。
cs.AI / 34 / 2608.05573
SkillTV-Bench: Benchmarking How Well Judges Perform on Skill-Augmented Agentic Execution
SkillTV-Bench：评估评审者在技能增强自主执行中的表现
Abstract
LLM agents increasingly execute long-horizon tasks through tool use and environment interaction, shifting evaluation from final-response scoring to verification of complete executions. For skill-augmented agents, verification additionally requires the procedural knowledge encoded in task-time skills, because this knowledge indicates what evidence to inspect and which failures are task-critical. However, existing judge benchmarks often expose final responses or static trajectories, and rarely combine task-time skills with directly inspectable artifacts and environments. We therefore introduce SkillTV-Bench, a 681-case benchmark of real agent trajectories from 50 tasks across eleven domains, designed to evaluate skill-aware trajectory verification for both LLM-as-a-Judge and Agent-as-a-Judge methods. Additionally, we propose SkillTV-Evolve, which externalizes verification knowledge as a reusable JudgeSkill that guides an agent judge to plan targeted inspections and issue evidence-grounded verdicts. On a disjoint development pool, an automated evolution loop further refines the JudgeSkill using misjudged cases. On SkillTV-Bench, the refined skill increases the same agent judge's accuracy by 14.8 percentage points. In offline rollout-pool selection, it increases selected-trajectory success from 22.9% with one rollout to 45.5% with ten rollouts. The code and data are available at https://github.com/HanZhi306/SkillTV-Bench
Chinese Translation
大型语言模型（LLM）代理越来越多地通过工具使用和环境交互执行长时间跨度的任务，这使得评估从最终响应评分转向对完整执行的验证。对于技能增强的代理，验证还需要在任务时间技能中编码的程序性知识，因为这些知识指示了需要检查的证据以及哪些失败是任务关键的。然而，现有的评审基准往往暴露最终响应或静态轨迹，且很少将任务时间技能与可直接检查的工件和环境结合起来。因此，我们引入了SkillTV-Bench，这是一个包含来自11个领域50个任务的681个真实代理轨迹的基准，旨在评估针对LLM作为评审者和代理作为评审者方法的技能感知轨迹验证。此外，我们提出了SkillTV-Evolve，它将验证知识外部化为可重用的JudgeSkill，指导代理评审者进行有针对性的检查并发出基于证据的裁决。在一个不重叠的开发池中，自动化演化循环进一步利用错误判断的案例精炼JudgeSkill。在SkillTV-Bench上，精炼后的技能使同一代理评审者的准确率提高了14.8个百分点。在离线回放池选择中，它将选定轨迹的成功率从一次回放的22.9%提高到十次回放的45.5%。代码和数据可在 https://github.com/HanZhi306/SkillTV-Bench 获取。
cs.AI / 35 / 2608.05587
StepReflect: Structured UI Transition Reflection for Mobile GUI Agents
StepReflect：移动图形用户界面代理的结构化用户界面过渡反射
Abstract
Autonomous mobile GUI agents require accurate action reflection for reliable long-horizon execution. Existing approaches rely on open-ended multimodal reasoning after each action, which is costly and poorly matched to the structured nature of GUI state transitions. We propose StepReflect, which formulates per-step GUI reflection as supervised structured prediction conditioned on explicit transition specifications and paired visual evidence. StepReflect is trained through a staged pipeline combining supervised fine-tuning, teacher-student distillation, and preference- and reward-based refinement. Offline, the resulting 8B model achieves 82.16% transition-level accuracy on AndroidWorld, exceeding zero-shot GPT-5.2 by 11.83 percentage points under the same structured input. Online, across M3A, Agent-SAMA, MAI-UI-8B, and Seed-2.0-Pro, StepReflect achieves higher task success in three of four agent configurations and remains within one successful task of the GPT-5.2 Reflection Agent in the fourth. It also reduces paid API charges relative to GPT-based reflection in all four configurations. These results establish StepReflect as a practical, locally deployable alternative to repeated frontier-model reflection for long-horizon mobile GUI agents.
Chinese Translation
自主移动图形用户界面代理需要准确的动作反射以实现可靠的长期执行。现有方法依赖于在每个动作之后进行开放式多模态推理，这既成本高昂又与图形用户界面状态转换的结构化特性不匹配。我们提出了StepReflect，它将每一步的图形用户界面反射形式化为基于显式转换规范和配对视觉证据的监督结构化预测。StepReflect通过一个分阶段的管道进行训练，该管道结合了监督微调、教师-学生蒸馏以及基于偏好和奖励的优化。在离线阶段，得到的8B模型在AndroidWorld上实现了82.16%的转换级准确率，超越了在相同结构化输入下的零-shot GPT-5.2，提升了11.83个百分点。在在线阶段，在M3A、Agent-SAMA、MAI-UI-8B和Seed-2.0-Pro中，StepReflect在四种代理配置中的三种中实现了更高的任务成功率，并在第四种配置中与GPT-5.2反射代理的成功任务数相差不超过一个。相较于基于GPT的反射，它在所有四种配置中还降低了付费API费用。这些结果确立了StepReflect作为移动图形用户界面代理长期执行的实用、本地可部署的替代方案，取代了重复的前沿模型反射。
cs.AI / 36 / 2608.05602
Epistemic Trustworthiness in Generative AI: A Normative Framework for Warranted Reliance in High-Stakes Workflows
生成性人工智能中的认知信任worthiness：高风险工作流程中合理依赖的规范框架
Abstract
Generative AI systems are increasingly deployed in high-stakes professional contexts, where their outputs shape what users believe, how they reason, and what they treat as settled. This raises a central question for responsible AI: under what conditions is reliance on generative AI outputs epistemically warranted rather than behaviourally induced? Existing frameworks largely ask whether AI outputs are accurate, fair, explainable, safe, or trusted by users. These questions remain necessary, and each can contribute to warranted reliance. However, they do not directly specify warranted reliance as a distinct evaluative target: the conditions under which users are justified in treating AI outputs as inputs into their own reasoning. We argue that this requires an account of epistemic trustworthiness: what makes a system epistemically worthy of reliance. Drawing on philosophical accounts of trustworthiness as competence and audience-orientation, we develop a constitutive normative framework comprising three jointly necessary and non-fungible conditions. First, epistemic humility requires systems to represent and communicate the limits of their competence. Second, epistemic access requires systems to enable users to inspect, question, and contest outputs in context. Third, resistance to epistemic injustice requires systems to recognise users as legitimate epistemic agents and avoid marginalising their knowledge and experience. Through real-world case analyses in legal reasoning, medical reasoning, and hiring, we show how failures of epistemic humility, epistemic access, and resistance to epistemic injustice can produce consequential harms that standard measures of accuracy, fairness, and usability do not address on their own. We conclude by outlining design and evaluation implications for GenAI systems organised around epistemically warranted reliance rather than output correctness alone.
Chinese Translation
生成性人工智能系统越来越多地应用于高风险的专业环境中，其输出影响用户的信念、推理方式以及他们所认为的确定性。这引发了一个关于负责任的人工智能的核心问题：在什么条件下对生成性人工智能输出的依赖是认知上合理的，而不是行为上诱导的？现有框架主要关注人工智能输出是否准确、公平、可解释、安全或被用户信任。这些问题仍然是必要的，每个问题都可以为合理依赖做出贡献。然而，它们并没有直接将合理依赖指定为一个独特的评估目标：即用户在何种条件下有理由将人工智能输出视为自己推理的输入。我们认为，这需要对认知信任worthiness的阐述：是什么使得一个系统在认知上值得依赖。基于哲学上对信任worthiness的理解，作为能力和面向受众的，我们提出了一个包含三个共同必要且不可替代条件的构成性规范框架。首先，认知谦逊要求系统能够表示和传达其能力的局限性。其次，认知访问要求系统使用户能够在上下文中检查、质疑和争辩输出。第三，抵制认知不公正要求系统承认用户作为合法的认知代理，并避免边缘化他们的知识和经验。通过对法律推理、医学推理和招聘中的真实案例分析，我们展示了认知谦逊、认知访问和抵制认知不公正的失败如何产生标准的准确性、公平性和可用性措施无法单独解决的重大危害。最后，我们总结了围绕认知上合理依赖而非仅仅输出正确性的生成性人工智能系统的设计和评估启示。
cs.AI / 37 / 2608.05624
Measuring and Detecting Harmful AI Sycophancy
测量与检测有害的人工智能谄媚行为
Abstract
Sycophantic responses are becoming pervasive in large language models (LLMs), and prior work has pointed out that some of them could be harmful. This paper focuses on one harmful sycophancy: preference-induced stance reversal sycophancy (PSRS), where a model reverses an initial stance merely to align with a user's stated preference. While existing research mainly measures how sycophantic a model is, we go further and ask whether PSRS can also be detected automatically from a single response. To investigate this at scale, we introduce CAP (Contrastive Anchor Probing), a framework for collecting labeled PSRS data. Applying CAP to 17 open- and closed-source LLMs, we collect 290,460 labeled responses across 12 everyday-advice domains. We organize our study around three research questions. (1) How often does PSRS occur? (2) How well can it be detected? (3) How does detection generalize to unseen models? We first reveal that PSRS rates range from 5% to 56% across LLMs, with more capable models being less sycophantic. Next, we show that detecting PSRS is feasible from the response text alone, and detectors need to learn subtle PSRS patterns from the training data. Because new LLMs appear rapidly, detectors inevitably encounter unseen models, making cross-model generalization an important framework goal. We demonstrate that detection performance drops on unseen models and propose an initial approach to address this challenge. We will release our dataset and code to support future research.
Chinese Translation
谄媚性回应在大型语言模型（LLMs）中变得越来越普遍，先前的研究指出其中一些可能是有害的。本文聚焦于一种有害的谄媚行为：偏好诱导的立场反转谄媚（Preference-Induced Stance Reversal Sycophancy, PSRS），即模型仅仅为了与用户所表达的偏好一致而反转最初的立场。尽管现有研究主要测量模型的谄媚程度，但我们进一步探讨是否可以从单一回应中自动检测到PSRS。为了在大规模上研究这一问题，我们引入了CAP（Contrastive Anchor Probing）框架，用于收集标注的PSRS数据。我们将CAP应用于17个开放源和闭源的LLM，收集了12个日常建议领域的290,460个标注回应。我们的研究围绕三个研究问题展开：（1）PSRS的发生频率是多少？（2）它的检测效果如何？（3）检测结果如何推广到未见过的模型？我们首先揭示了PSRS的发生率在不同LLM中范围从5%到56%，更强大的模型表现出较低的谄媚性。接下来，我们展示了仅通过回应文本就能有效检测PSRS，检测器需要从训练数据中学习微妙的PSRS模式。由于新的LLM快速出现，检测器不可避免地会遇到未见过的模型，这使得跨模型的推广成为一个重要的研究目标。我们证明了在未见过的模型上检测性能会下降，并提出了一种初步的方法来应对这一挑战。我们将发布我们的数据集和代码，以支持未来的研究。
cs.AI / 38 / 2608.05628
SkillHEX: Improving Agent Skills via Hypothesis-Driven Autonomous Exploration and Exploitation
SkillHEX：通过假设驱动的自主探索与利用提升代理技能
Abstract
Although agent skills equip LLMs with reusable procedural knowledge, manual maintenance suffers from high costs, unscalability, and misalignment. Real-world deployments thus require autonomous, on-demand skill evolution at test time, constrained by limited interaction budgets and a lack of training or validation sets. This setting introduces a severe sparse reward challenge, where outcomes conflate multiple latent failure causes. Under such ambiguity, existing methods that greedily refine a single incumbent skill are particularly vulnerable to an exploitation trap, allowing early misdiagnoses to exhaust limited trials along unproductive trajectories. To address this, we introduce SkillHEX, a closed-loop framework coupling hypothesis-driven self-verification with evidence-guided tree search. SkillHEX translates falsifiable failure hypotheses into executable tests, producing diagnostic evidence as dense reward without additional environment attempts. This evidence guides a search over persistent skill-revision branches, dynamically balancing the exploitation of supported edits with the exploration of plausible alternatives. Evaluated on 87 tasks from SkillsBench, SkillHEX outperforms existing self-evolving methods and achieves an average pass rate of 55.9% and 57.9% using GPT-5.3-Codex and Claude Opus 4.7 under a five-iteration budget, respectively.
Chinese Translation
尽管代理技能为大规模语言模型（LLMs）提供了可重用的程序知识，但手动维护面临高成本、不可扩展性和不一致性等问题。因此，实际应用需要在测试时进行自主的、按需的技能演变，而这受到有限交互预算和缺乏训练或验证集的限制。这种环境带来了严重的稀疏奖励挑战，其中结果混合了多种潜在的失败原因。在这种模糊性下，现有方法贪婪地优化单一现有技能，特别容易陷入利用陷阱，使得早期误诊耗尽有限的尝试，沿着无效的轨迹前进。为了解决这一问题，我们提出了SkillHEX，一个将假设驱动的自我验证与证据引导的树搜索相结合的闭环框架。SkillHEX将可证伪的失败假设转化为可执行的测试，生成诊断证据作为密集奖励，而无需额外的环境尝试。这些证据引导对持久技能修订分支的搜索，动态平衡对支持编辑的利用与对合理替代方案的探索。在对SkillsBench的87个任务进行评估时，SkillHEX的表现优于现有的自我演变方法，在五次迭代预算下，使用GPT-5.3-Codex和Claude Opus 4.7的平均通过率分别达到了55.9%和57.9%。
cs.AI / 39 / 2608.05642
Bayesian Expected Uncertainty Reduction (B-EUR) Model: A Computational Account of What Makes Design Options Worth Trying
贝叶斯期望不确定性减少（B-EUR）模型：设计选项值得尝试的计算解释
Abstract
This paper proposes the Bayesian Expected Uncertainty Reduction (B-EUR) model, which formalizes the value of trying a candidate design action as its expected reduction of epistemic uncertainty about action--outcome relations. The model addresses one part of the Uncertainty Driven Action (UDA) model's open question concerning how changes in uncertainty perception determine action selection. We examine two environmental properties: generalizability, or how far knowledge from one trial extends to neighboring candidates, and outcome discriminability, or how clearly differences among outcomes can be distinguished. We tested the model through simulations and human experiments using a graph-shape guessing task that isolates learning about action--outcome relations under a limited trial budget. Epistemic value followed an inverted-U-shaped relationship with generalizability and increased with outcome discriminability in the simulations. In the human experiments, the subjective value of trying and enjoyment followed inverted-U-shaped relationships with generalizability, while choice behavior reflected both properties. The B-EUR model provides a computational account of candidate-action evaluation within uncertainty-driven design activity and offers implications for constructing prototype sets, framing design problems, and organizing feedback to support informative exploration.
Chinese Translation
本文提出了贝叶斯期望不确定性减少（B-EUR）模型，该模型将尝试候选设计行动的价值形式化为对行动—结果关系的知识不确定性预期减少。该模型解决了不确定性驱动行动（UDA）模型中关于不确定性感知变化如何决定行动选择的开放性问题。我们考察了两个环境特性：可推广性，即一个试验的知识在多大程度上扩展到邻近候选项，以及结果可区分性，即结果之间的差异能被多清晰地区分。我们通过模拟和人类实验测试了该模型，使用图形形状猜测任务，在有限的试验预算下隔离对行动—结果关系的学习。在模拟中，知识价值与可推广性呈倒U型关系，并随着结果可区分性的增加而增加。在人类实验中，尝试的主观价值和享受感与可推广性呈倒U型关系，而选择行为则反映了这两种特性。B-EUR模型为不确定性驱动设计活动中的候选行动评估提供了计算解释，并对构建原型集、框定设计问题以及组织反馈以支持信息探索提供了启示。
cs.AI / 40 / 2608.05643
Refining Over Resampling: Test-Time Self-Correction for LLM Reasoning
精炼过度重采样：大语言模型推理的测试时自我校正
Abstract
Test-time scaling improves LLM reasoning by using additional inference compute, but wider sampling alone can suffer from diminishing returns: new rollouts often repeat existing answer patterns instead of adding useful reasoning diversity. Verifier-based selection offers an alternative, but its performance depends on the calibration of an external reward model. We propose a verifier-free breadth--depth refinement framework that uses test-time compute to both explore and improve candidate solutions. The method samples multiple independent reasoning rollouts, refines each rollout through iterative self-critique and self-correction, and aggregates the refined answers by majority voting. Breadth preserves diverse initial attempts, while depth repairs local reasoning errors before aggregation. Across AIME24, AIME25, AMC, OlympiadBench, and MATH500, our method consistently improves over greedy decoding, majority voting, verifier-based best-of-$N$, beam search, and lookahead decoding across multiple open-weight models. For instance, with Qwen2.5-1.5B, accuracy increases from the strongest verifier-based baseline to $58.0\%$ on MATH500, and from $25.0\%$ to $32.5\%$ on AMC. These results show that test-time compute can be more effective when used to refine sampled trajectories rather than only to sample more candidates or rely on verifier-guided selection.
Chinese Translation
测试时扩展通过使用额外的推理计算来改善大语言模型（LLM）的推理能力，但单纯的广泛采样可能会遭遇收益递减：新的推理结果往往重复现有的答案模式，而不是增加有用的推理多样性。基于验证器的选择提供了一种替代方案，但其性能依赖于外部奖励模型的校准。我们提出了一种无验证器的广度-深度精炼框架，该框架利用测试时计算来探索和改进候选解决方案。该方法对多个独立的推理结果进行采样，通过迭代自我批评和自我校正来精炼每个推理结果，并通过多数投票聚合精炼后的答案。广度保留了多样化的初始尝试，而深度在聚合之前修复局部推理错误。在 AIME24、AIME25、AMC、OlympiadBench 和 MATH500 等多个开放权重模型上，我们的方法在贪婪解码、主要投票、基于验证器的最佳 $N$、束搜索和前瞻解码等方法上始终表现出改进。例如，在 Qwen2.5-1.5B 上，准确率从最强的基于验证器的基线提高到 MATH500 的 $58.0\%$，在 AMC 上从 $25.0\\%$ 提高到 $32.5\\%$。这些结果表明，测试时计算在用于精炼采样轨迹时可能更有效，而不仅仅是用于采样更多候选项或依赖于验证器引导的选择。
cs.AI / 41 / 2608.05673
A Unified Framework for Trajectory Prediction with Explicit Planning and Reaction Decomposition
具有显式规划和反应分解的轨迹预测统一框架
Abstract
Trajectory prediction has shifted toward structured formulations with explicit social modeling. However, existing methods inadequately distinguish the functional roles of social influence in trajectory planning. Observing that agents typically form motion plans by anticipating others' future behaviors before making local reactive adjustments, we identify social interactions as playing staged roles, namely planning precedes reaction. We propose INTraJ, a unified framework that decomposes social influence into two stages: a planning stage constructs reference trajectories using future social information, and a reaction stage recovers local adjustments from the residual between full-context prediction and the reference. INTraJ supports both multi-target and single-target paradigms. Extensive experiments on four standard benchmarks, including Argoverse 2, Argoverse 2-ped, ETH/UCY, and SDD, demonstrate consistent improvements, particularly in FDE and long-horizon consistency, with state-of-the-art performance achieved in several settings. INTraJ reframes trajectory prediction as a planning-driven two-stage process, validating that staged social modeling is critical for stable predictions. The code is publicly available at https://github.com/11isnotavailable/INTraJ.
Chinese Translation
轨迹预测已转向具有显式社会建模的结构化形式。然而，现有方法未能充分区分社会影响在轨迹规划中的功能角色。我们观察到，代理通常通过预测他人的未来行为来形成运动计划，然后再进行局部反应调整，因此我们将社会互动视为分阶段的角色，即规划先于反应。我们提出了INTraJ，一个统一框架，将社会影响分解为两个阶段：规划阶段利用未来社会信息构建参考轨迹，反应阶段则根据全局上下文预测与参考轨迹之间的残差恢复局部调整。INTraJ支持多目标和单目标范式。在包括Argoverse 2、Argoverse 2-ped、ETH/UCY和SDD在内的四个标准基准上的大量实验表明，特别是在最终定位误差（FDE）和长时间一致性方面，表现出一致的改进，并在多个设置中实现了最先进的性能。INTraJ将轨迹预测重新框定为一个以规划驱动的两阶段过程，验证了分阶段社会建模对稳定预测的重要性。代码已公开发布在https://github.com/11isnotavailable/INTraJ。
cs.AI / 42 / 2608.05685
Grounded Well-Condition Anomaly Detection on the Volve Field: Constructed Labels, a Baseline, and a Dual-Head Model
基于Volve油田的有条件异常检测：构建标签、基线与双头模型
Abstract
Most public benchmarks for machine-condition monitoring come from test rigs, where faults are induced on purpose and every event is known. Real production fields rarely offer that. They give you sensor histories with no fault log attached, which is exactly the situation where an anomaly-detection method has to invent its own labels, and where quiet assumptions can slip in unnoticed. We work with the open Volve field data released by Equinor and take two things seriously that such datasets usually skip. First, we build anomaly labels that are not just patterns in the numbers but are checked against what the field's own engineering documents say can physically go wrong, and we release the reasoning behind every label. Second, we test whether those constructed labels are learnable at all, using both an unsupervised baseline and a small dual-head model that marks when an event happens and what kind it is, an idea we carry over from earlier work on defect detection in metal parts. The results are honest. An unsupervised detector that never sees the labels still lands on the same regions our rules flagged, which tells us the labels are not arbitrary. A compact supervised model recovers event presence and event type well across wells it has never seen, and locates events in time only roughly. We report what worked, what did not, and every assumption in between. The dataset, grounded labels, per-label provenance, baseline scores, trained model, and code are released publicly under CC-BY-NC-SA 4.0.
Chinese Translation
大多数机器状态监测的公共基准来自测试装置，在这些装置上故障是故意引入的，并且每个事件都是已知的。而真实的生产油田很少提供这样的条件。它们提供的传感器历史数据没有附带故障日志，这正是异常检测方法需要自行构建标签的情况，也是潜在假设可能悄然渗入的地方。我们使用了Equinor发布的开放Volve油田数据，并认真对待通常被忽视的两点。首先，我们构建的异常标签不仅仅是数字中的模式，而是经过与油田自身工程文件中所述的可能出现的物理故障进行核对，并且我们公开了每个标签背后的推理。其次，我们测试这些构建的标签是否可学习，使用了无监督基线和一个小型双头模型，该模型标记事件发生的时间及其类型，这一思路源自我们在金属部件缺陷检测方面的早期工作。结果是诚实的。一个从未见过标签的无监督检测器仍然能够识别出我们规则标记的相同区域，这告诉我们这些标签并不是任意的。一个紧凑的监督模型能够很好地恢复未见过的井中的事件存在和事件类型，并且仅能大致定位事件发生的时间。我们报告了有效的内容、无效的内容，以及两者之间的每一个假设。数据集、构建的标签、每个标签的来源、基线得分、训练模型和代码均在CC-BY-NC-SA 4.0许可下公开发布。
cs.AI / 43 / 2608.05695
DreamGuard: Efficient Runtime Guardrail for LLM Agents via Risk-Aware World Model
DreamGuard：基于风险感知世界模型的高效运行时保护机制用于大型语言模型代理
Abstract
As large language model (LLM) agents increasingly invoke external tools and interact with real-world systems, unsafe actions may cause irreversible consequences on external states, user data, and downstream services. Recent runtime guardrails mitigate such risks by checking proposed actions before execution, but many remain reactive: they primarily assess the apparent safety of the current action, lacking an explicit model of how risk evolves across the trajectory. This limitation creates a critical blind spot for long-horizon risks, where individually benign-looking actions can gradually drift the agent toward hazardous states. In response, we propose DreamGuard, a proactive guardrail for LLM agents built around a risk-aware world model. The world model maintains a compact recurrent latent state over the trajectory and predicts future latent states from which DreamGuard derives immediate-hazard and prefix-risk evidence. It then fuses these multi-horizon signals into intervention decisions before execution. Experiments across four benchmarks and an online guardrail evaluation show that DreamGuard outperforms generic, reactive, and proactive guardrail baselines, achieves the best safety-utility trade-off among evaluated guardrails, and maintains an average end-to-end latency of 25 ms per call.
Chinese Translation
随着大型语言模型（LLM）代理越来越多地调用外部工具并与现实世界系统互动，不安全的行为可能对外部状态、用户数据和下游服务造成不可逆转的后果。近期的运行时保护机制通过在执行前检查提议的行为来减轻这些风险，但许多机制仍然是反应性的：它们主要评估当前行为的表面安全性，缺乏对风险如何在轨迹中演变的明确模型。这一局限性为长时间范围的风险创造了一个关键盲点，其中单个看似无害的行为可能逐渐将代理推向危险状态。为此，我们提出了DreamGuard，一种围绕风险感知世界模型构建的LLM代理的主动保护机制。该世界模型在轨迹上维护一个紧凑的递归潜在状态，并预测未来的潜在状态，DreamGuard则从中推导出即时危险和前缀风险证据。然后，它在执行前将这些多时间范围信号融合为干预决策。在四个基准测试和在线保护机制评估中的实验表明，DreamGuard优于通用、反应性和主动保护机制基线，在评估的保护机制中实现了最佳的安全性与效用权衡，并保持每次调用的平均端到端延迟为25毫秒。
cs.AI / 44 / 2608.05710
Shaping Human-AI Interactions to Provide Improvement Pathways and Balance Competing Objectives
塑造人机交互以提供改进路径和平衡竞争目标
Abstract
When an AI system is deployed, the individuals who use and or are evaluated by it form beliefs about how the system operates and use those beliefs to strategically present their preferences, behaviors, or attributes. The system then responds with feedback or a decision outcome, thereby creating a human-AI interaction loop. This thesis studies how to design and shape such interactions to achieve three goals: (1) help individuals develop accurate beliefs about the AI systems so they can improve and or secure favorable outcomes at minimal cost, (2) encourage improvement and or discourage gaming behaviors, and (3) ensure that the AI system continues to achieve its intended objectives, such as maximizing accuracy. To address these goals, the thesis is organized into three complementary parts that examine and study human-AI interactions from the perspectives of both evaluated individuals and AI systems. Together, the work presented in this thesis advances human-centered machine learning by providing principles and methods for designing AI systems that align with human needs, values, and capabilities. Methodologically, this thesis integrates theoretical analysis, data-driven modeling, human-subject experiments, and empirical evaluations on real-world and semi-synthetic datasets.
Chinese Translation
当一个人工智能系统被部署时，使用该系统或被其评估的个体会形成关于系统如何运作的信念，并利用这些信念战略性地展示他们的偏好、行为或属性。系统随后以反馈或决策结果作出回应，从而形成一个人机交互循环。本论文研究如何设计和塑造这种交互，以实现三个目标：（1）帮助个体形成关于人工智能系统的准确信念，以便他们能够以最低成本改善或确保有利结果；（2）鼓励改进或抑制游戏行为；（3）确保人工智能系统继续实现其预期目标，例如最大化准确性。为了解决这些目标，论文分为三个互补部分，从被评估个体和人工智能系统的角度考察和研究人机交互。本文所呈现的工作通过提供设计与人类需求、价值观和能力相一致的人工智能系统的原则和方法，推动了以人为中心的机器学习。在方法论上，本论文整合了理论分析、数据驱动建模、人类受试者实验以及对真实和半合成数据集的实证评估。
cs.AI / 45 / 2608.05714
RA-CAD: Learning Post-Execution Critique for State-Aware Text-to-CAD Generation
RA-CAD：学习后执行批评以实现状态感知的文本到CAD生成
Abstract
Text-to-CAD generation translates natural-language design intent into editable and executable parametric computer-aided design (CAD) codes, reducing the expertise and effort required for manual modeling. Existing methods incorporate fixed, externally supplied, prompt-induced, or separately optimized critique mechanisms to optimize the generation process, but they do not necessarily optimize how feedback is interpreted and translated into effective corrective actions throughout the generation process. To bridge this feedback-utilization gap, we present RA-CAD (ReAct Agent for CAD), a state-aware agent that interacts with the CAD environment through a Generate--Execute--Critique--Rewrite loop. At each iteration, RA-CAD executes the current code and observes its outcome. Conditioned on the design instruction, current code, and execution feedback, the agent then generates an explicit post-execution critique as an intermediate policy action. This critique either validates the current result for termination or provides revision-oriented guidance that conditions the next rewrite. CAD Code Bootstrapping (CCB) first establishes fundamental parametric CAD coding capabilities through supervised fine-tuning. Feedback-Driven Agent Optimization (FAO) subsequently applies trajectory-level Group Relative Policy Optimization to both policy-generated code and critique sequences, assigning terminal F1 and Chamfer Distance rewards to the complete interaction trajectory. This formulation makes critique an outcome-aligned, learnable policy decision rather than an unoptimized auxiliary output. Experiments on CADFusion and Text2CAD show that RA-CAD achieves state-of-the-art execution validity and geometric quality compared with existing methods and strong proprietary language models, demonstrating the effectiveness of the proposed state-aware text-to-CAD agent.
Chinese Translation
文本到CAD生成将自然语言设计意图转化为可编辑和可执行的参数化计算机辅助设计（CAD）代码，从而减少了手动建模所需的专业知识和努力。现有方法采用固定的、外部提供的、提示诱导的或单独优化的批评机制来优化生成过程，但它们并不一定优化反馈的解释方式及其在整个生成过程中的有效纠正措施的转化。为了解决这一反馈利用的差距，我们提出了RA-CAD（ReAct Agent for CAD），一个通过生成-执行-批评-重写循环与CAD环境互动的状态感知代理。在每次迭代中，RA-CAD执行当前代码并观察其结果。基于设计指令、当前代码和执行反馈，代理生成明确的后执行批评作为中间策略行动。该批评要么验证当前结果以便终止，要么提供面向修订的指导，以条件化下一次重写。CAD代码引导（CCB）首先通过监督微调建立基本的参数化CAD编码能力。随后，基于反馈的代理优化（FAO）对策略生成的代码和批评序列应用轨迹级的群体相对策略优化，为完整的交互轨迹分配终止F1和Chamfer距离奖励。这种形式使批评成为与结果对齐的可学习策略决策，而不是未优化的辅助输出。在CADFusion和Text2CAD上的实验表明，与现有方法和强大的专有语言模型相比，RA-CAD在执行有效性和几何质量上达到了最先进的水平，证明了所提议的状态感知文本到CAD代理的有效性。
cs.AI / 46 / 2608.05716
BlockPython: A Process-Aware Agent-Supported Platform for the Transition from Block-Based to Python Programming
BlockPython：一种支持过程感知的代理平台，用于从基于块的编程过渡到Python编程
Abstract
The transition from block-based to text-based programming requires learners to convert visible program structures into abstract textual expressions, which may create a cognitive gap between understanding computational concepts and expressing them in Python syntax. To support this transition, we designed and implemented BlockPython. The platform centers on bidirectional translation between blocks and Python and guides learners through four stages: Task Decomposition, Block-Based Practice, Code Challenge, and Extended Interaction. Across these stages, learners progressively establish connections among program structure, runtime behavior, and textual code. During learning, the platform continuously collects process evidence, including block artifacts, code versions, run outcomes, use of support, and dialogue. Deterministic diagnosis, program visualization, and the learning assistant use this evidence to identify different difficulties in computational understanding and Python expression. The rule-based system is responsible for program execution, objective evaluation, and stage control, while the learning assistant uses verified evidence to provide explanations, prompts, and guiding questions. This report describes the design rationale, learning workflow, and process-aware support mechanisms of BlockPython and provides a system-design reference for supporting the transition from block-based to text-based programming and for analyzing learning processes.
Chinese Translation
从基于块的编程过渡到基于文本的编程要求学习者将可见的程序结构转换为抽象的文本表达，这可能在理解计算概念与用Python语法表达这些概念之间产生认知差距。为支持这一过渡，我们设计并实现了BlockPython。该平台以块与Python之间的双向转换为核心，并引导学习者经历四个阶段：任务分解、基于块的实践、代码挑战和扩展互动。在这些阶段中，学习者逐步建立程序结构、运行行为和文本代码之间的联系。在学习过程中，平台持续收集过程证据，包括块工件、代码版本、运行结果、支持使用情况和对话。确定性诊断、程序可视化和学习助手利用这些证据识别计算理解和Python表达中的不同困难。基于规则的系统负责程序执行、客观评估和阶段控制，而学习助手则利用经过验证的证据提供解释、提示和引导问题。本报告描述了BlockPython的设计原理、学习工作流程和过程感知支持机制，并为支持从基于块的编程到基于文本的编程的过渡以及分析学习过程提供了系统设计参考。
cs.AI / 47 / 2608.05729
Unified Agent: Managing Interactions across Devices
统一代理：跨设备交互管理
Abstract
As capabilities rapidly increase, AI agents can move from running inside one app to acting across a user's devices over time. Yet existing agent systems still fall short in this scenario. This is because observations are scattered across devices and moments, but mainstream systems are not designed around this fact: a single agent that treats devices as tools lacks effective state management for all devices across time, and multi-agent systems coordinate across agents but do not maintain the compact carried state a cross-device, cross-time request needs. We argue that the agent should maintain an effectively designed state that organizes engagement evidence, stated facts, and the standing request in a compact, action-ready form for deciding its action given the current observation. To compare state designs, we construct a benchmark of user-agent interaction across devices and time. We instantiate this principle in Unified Agent, a stateful agent that carries interaction evidence across devices and moments and uses it with the current observation to act. In the default setting, it significantly outperforms our adaptations of four published designs. Across changes in multimodal large language model (MLLM) family, capability, and reasoning effort, it remains ahead of all compared systems, demonstrating that the state-design advantage is robust across MLLM settings. Our code and data will be publicly available on GitHub.
Chinese Translation
随着能力的快速提升，人工智能代理可以从在一个应用程序内运行转变为随着时间的推移在用户的多个设备上进行操作。然而，现有的代理系统在这种情况下仍然不足。这是因为观察数据散布在不同设备和时刻之间，而主流系统并未围绕这一事实进行设计：一个将设备视为工具的单一代理缺乏对所有设备跨时间的有效状态管理，而多代理系统则在代理之间进行协调，但无法维持跨设备、跨时间请求所需的紧凑携带状态。我们认为，代理应维护一个有效设计的状态，以紧凑、可行动的形式组织参与证据、陈述事实和当前请求，以便在给定当前观察时决定其行动。为了比较状态设计，我们构建了一个跨设备和时间的用户-代理交互基准。我们在统一代理（Unified Agent）中实例化这一原则，这是一种状态保持的代理，能够跨设备和时刻携带交互证据，并利用这些证据与当前观察进行行动。在默认设置下，它显著优于我们对四种已发布设计的改编。无论是在多模态大语言模型（MLLM）系列、能力还是推理努力的变化中，它始终领先于所有比较系统，证明了状态设计优势在MLLM设置中是稳健的。我们的代码和数据将在GitHub上公开发布。
cs.AI / 48 / 2608.05734
Subliminal Learning is Non-Semantic Distillation
潜意识学习是非语义的提炼
Abstract
Subliminal Learning (SL) is a surprising type of generalization displayed by modern language models. It allows the transfer of a bias or behavior from a teacher model to a student by distilling from seemingly unrelated or random synthetic data from the teacher. This presents challenges in ensuring AI systems remain predictable and are trained safely, as standard auditing of the input data would not catch the hidden subliminal signal. Here, we investigate several open questions as to the enabling mechanisms and drivers of SL. First is the nature of the process by which biases are encoded in the data. We find that by adding Gaussian noise to the weights of the teacher and student models, the magnitude of subliminal transfer is increased by a factor of 1.9 in Gemma and 1.3 in Llama, suggesting that non-semantic weight structures play a crucial role. We show that steering vectors can be applied to the teacher to produce subliminal data, in addition to prompting and finetuning as used in previous studies. Analysis of the activations of the student models that have been trained on steered and prompted data demonstrates that students inherit not just the semantic meaning of the teacher's bias, but also the type of intervention that was used to apply it: steered students imitate steering vectors, prompted students do not. Additionally, the gradients of steered subliminal data show a linear correlation with the teacher's steering vectors, showing promise for data auditing. More broadly, as synthetic data becomes central to frontier training pipelines, being able to see the latent signals hidden in training data becomes paramount.
Chinese Translation
潜意识学习（Subliminal Learning, SL）是现代语言模型表现出的一个令人惊讶的泛化类型。它允许通过从教师模型中提炼看似无关或随机的合成数据，将偏差或行为转移到学生模型。这给确保人工智能系统保持可预测性和安全训练带来了挑战，因为标准的输入数据审计无法捕捉到隐藏的潜意识信号。在此，我们探讨了关于SL的启用机制和驱动因素的几个未解之谜。首先是偏差如何在数据中编码的过程性质。我们发现，通过在教师和学生模型的权重中添加高斯噪声，潜意识转移的幅度在Gemma中增加了1.9倍，在Llama中增加了1.3倍，这表明非语义权重结构发挥了关键作用。我们展示了可以将引导向量应用于教师模型以生成潜意识数据，此外还使用了在先前研究中采用的提示和微调。对在引导和提示数据上训练的学生模型的激活分析表明，学生不仅继承了教师偏差的语义含义，还继承了应用该偏差的干预类型：引导的学生模仿引导向量，而提示的学生则不然。此外，引导的潜意识数据的梯度与教师的引导向量之间显示出线性相关性，显示出数据审计的潜力。更广泛地说，随着合成数据在前沿训练流程中变得核心，能够识别训练数据中隐藏的潜在信号变得至关重要。
cs.AI / 49 / 2608.05778
When Do Prompt-Side Agent Playbooks Transfer? Accuracy, Cost, and Runtime Shift in Agent Deployment
何时提示侧代理剧本能够迁移？代理部署中的准确性、成本和运行时变化
Abstract
Prompt-side playbooks can improve tool-using language agents without retraining, but their portability beyond the source setting is unclear. We study frozen playbook transfer under a shared distill--validate--transfer protocol. On ALFWorld, transfer is beneficial under controlled greedy decoding and, in one near-budget-matched comparison, distilled guidance outperforms five fixed demonstrations. On TAU2-Bench, a prespecified aggregate contrast supports a modest average matched-domain advantage, but global Holm correction retains only one of 135 route-level effects; the remaining grid provides descriptive evidence of compatibility-sensitive heterogeneity. On XBench-DeepSearch, one artifact--runtime pairing preserves useful first-try heuristics while producing repeated queries, delayed stopping, and substantial cost inflation after a context-runtime shift. Across benchmarks, transferred and target-derived playbooks both require target-side validation of success, termination, protocol compatibility, and cost. Frozen transfer is therefore a conditional cold-start option, not a reuse-by-default strategy or a universally preferable alternative to target-side redistillation.
Chinese Translation
提示侧剧本可以在不重新训练的情况下提升工具使用语言代理的性能，但其在源设置之外的可移植性尚不明确。我们在共享的蒸馏-验证-迁移协议下研究了冻结剧本的迁移。在 ALFWorld 中，在受控贪婪解码下，迁移是有益的，并且在一次接近预算匹配的比较中，蒸馏指导的表现优于五个固定演示。在 TAU2-Bench 中，预先指定的聚合对比支持适度的平均匹配域优势，但全局 Holm 校正仅保留了 135 个路线级效果中的一个；其余的网格提供了兼容性敏感异质性的描述性证据。在 XBench-DeepSearch 中，一个工件-运行时配对保留了有用的首次尝试启发式，同时产生了重复查询、延迟停止和在上下文-运行时变化后显著的成本膨胀。在所有基准中，迁移和目标派生的剧本都需要目标侧对成功、终止、协议兼容性和成本的验证。因此，冻结迁移是一种有条件的冷启动选项，而不是默认的重用策略或普遍优于目标侧重新蒸馏的替代方案。
cs.AI / 50 / 2608.05784
Activity Frames: Deterministic Screen-Activity Compilation for Agent Memory and Replay
活动框架：面向代理记忆和重放的确定性屏幕活动编译
Abstract
Computer-use agents pay full frontier inference to re-derive routines their user has already performed, because an agent's memory today records what the user said, not what the user did. We compile passively captured screen activity into agent memory with a deterministic, zero-model pipeline: it segments a local capture stream into typed activity frames, bounded episodes carrying application, site, timing, input volume, and evidence pointers back to the raw rows, with no model in the loop, so the output is byte-identical, cacheable, and mechanically auditable. On one professional's single-user corpus of 128,756 frames over 51 active days, the compiler reduces a day of raw capture to a prompt-ready context block 86x smaller in 68 ms, and an agent reading that block answers questions about the day at 98.4% accuracy (Wilson 95% CI 91.7-99.7%) against an independent oracle, versus 66-80% for an LLM summary of the same capture, a mid-tier model reading the block matching a frontier one. The same compiler doubles as a demand-side cost instrument. Read off passive, pre-delegation human activity rather than agent rollouts, it supplies two parameters that agent-cost models assume but, to our knowledge, have not measured: the Routine Overhead Ratio R and the routine recurrence h. We report first values of R, a modeled upper bound, at 60-343x, and a delegable recurrence of 9.0% in-sample and 7.7% out-of-sample, for a realistic all-fleet token ceiling near 8%; a compiled routine replays deterministically with the model out of the loop, demonstrated live at zero model tokens on a guard-matched hit. Schema, compiler, and evaluation harness are open.
Chinese Translation
计算机使用代理完全依赖前沿推理来重新推导用户已经执行的例程，因为代理的记忆今天记录的是用户所说的内容，而不是用户所做的事情。我们将被动捕获的屏幕活动编译到代理记忆中，采用一种确定性的零模型管道：它将本地捕获流分段为类型化的活动框架，这些框架是带有应用程序、网站、时间、输入量和指向原始行的证据指针的边界情节，没有模型参与，因此输出是字节相同的、可缓存的，并且可以机械审计。在一位专业用户的128,756帧的单用户语料库中，覆盖51个活跃日，该编译器将一天的原始捕获减少为一个准备好的上下文块，大小缩小86倍，耗时68毫秒，而读取该块的代理在98.4%的准确率下（威尔逊95%置信区间91.7-99.7%）回答关于这一天的问题，相比之下，同一捕获的LLM摘要准确率为66-80%，中层模型读取该块的表现与前沿模型相匹配。相同的编译器还充当需求侧成本工具。它从被动的、预委托的人类活动中读取，而不是代理的回放，提供了两个代理成本模型假设但据我们所知尚未测量的参数：例程开销比R和例程重复率h。我们报告了R的首个值，一个建模的上限，范围为60-343倍，以及样本内的可委托重复率为9.0%和样本外的7.7%，对于一个接近8%的现实全舰令牌上限；编译的例程在没有模型参与的情况下确定性重放，已在零模型令牌的情况下在匹配的守卫命中中实时演示。模式、编译器和评估工具是开放的。
cs.AI / 51 / 2608.05790
ChainClaw: A Layered Agent Framework for Reliable On-Chain Execution
ChainClaw：一种用于可靠链上执行的分层代理框架
Abstract
General-purpose large language model agents have achieved strong performance on tool-augmented tasks, yet they rely on assumptions break down in blockchain environments. On-chain execution is stateful, adversarial, and economically irreversible, exposing three fundamental gaps: Reactivity, Irreversibility, and Observability. We propose ChainClaw, a blockchain-native agent framework built on OpenClaw, that addresses all three gaps through a layered architecture comprising an event-driven orchestration layer, a simulation-based safety intelligence layer, and an on-chain monitoring runtime layer, unified by a cross-layer memory subsystem. ChainClaw closes the Reactivity gap via event ingestion and simulation feedback, the Irreversibility gap via a pre-execution safety pipeline with transaction simulation and action guard, and the Observability gap via an on-chain read adapter and transaction monitor. We evaluate ChainClaw on a purpose-built benchmark covering seven tasks across four categories and five dimensions. ChainClaw consistently outperforms representative baselines on both safety and task completion.
Chinese Translation
通用大型语言模型代理在工具增强任务上取得了良好的表现，但它们依赖的假设在区块链环境中会失效。链上执行是有状态的、对抗性的，并且经济上不可逆，这暴露了三个基本缺口：反应性、不可逆性和可观察性。我们提出了ChainClaw，一个基于OpenClaw构建的区块链原生代理框架，通过一个分层架构解决这三个缺口，该架构包括一个事件驱动的编排层、一个基于模拟的安全智能层和一个链上监控运行时层，由一个跨层内存子系统统一。ChainClaw通过事件摄取和模拟反馈弥补反应性缺口，通过带有交易模拟和行动保护的预执行安全管道弥补不可逆性缺口，通过链上读取适配器和交易监控弥补可观察性缺口。我们在一个专门构建的基准上评估了ChainClaw，该基准涵盖了四个类别和五个维度的七个任务。ChainClaw在安全性和任务完成度方面始终优于代表性基线。
cs.AI / 52 / 2608.05792
When Agentic AI Meets Integrated Sensing and Communication
当自主智能体人工智能遇上集成感知与通信
Abstract
Agentic artificial intelligence (AI) is transforming Integrated Sensing and Communication (ISAC) from a function-oriented physical-layer technology into a goal-driven, closed-loop intelligent system, a paradigm we term AISAC. Existing work on learning-based sensing, resource allocation, reconfigurable intelligent surfaces (RIS), edge intelligence, multi-agent coordination, and resilient networking has developed largely in isolation. This survey unifies the literature within a six-stage closed-loop framework comprising observation, contextualization, reasoning and prediction, planning and orchestration, execution and collaboration, and feedback and resilience. It also introduces five levels of agentic maturity, ranging from physical-layer primitives to fully closed-loop agentic ISAC. We use this framework to review advances in multimodal intelligence, large language models, reinforcement learning, federated learning, RIS-assisted control, Unmanned Aerial Vehicle (UAV) and vehicular networks, and AI-native network management, and analyze privacy, security, resilience, and sustainability as cross-cutting requirements of the full perception-reasoning-action loop. An audit of representative studies against nine agentic-specific evaluation criteria shows that no system reports more than one or two of them, exposing a gap between claimed and demonstrated agentic maturity. We identify open challenges in physical-to-semantic grounding, predictive world models, real-time agent-PHY interaction, safe tool use, heterogeneous multi-agent collaboration, benchmarking, and resource-efficient autonomy.
Chinese Translation
自主人工智能（AI）正在将集成感知与通信（ISAC）从一种以功能为导向的物理层技术转变为一个以目标为驱动的闭环智能系统，这一范式我们称之为AISAC。现有关于基于学习的感知、资源分配、可重构智能表面（RIS）、边缘智能、多智能体协调和弹性网络的研究大多是孤立发展的。本调查将文献统一在一个六阶段闭环框架内，该框架包括观察、情境化、推理与预测、规划与协调、执行与协作，以及反馈与弹性。它还引入了五个自主成熟度等级，从物理层原语到完全闭环的自主ISAC。我们利用这一框架回顾了多模态智能、大型语言模型、强化学习、联邦学习、RIS辅助控制、无人机（UAV）和车载网络，以及AI原生网络管理的进展，并分析了隐私、安全、弹性和可持续性作为全感知-推理-行动循环的交叉要求。对九个自主特定评估标准的代表性研究进行审计显示，没有系统报告超过一到两个标准，暴露了声称与实际自主成熟度之间的差距。我们识别了在物理到语义基础、预测世界模型、实时智能体-物理层交互、安全工具使用、异构多智能体协作、基准测试和资源高效自主等方面的开放挑战。
cs.AI / 53 / 2608.05810
When Self-Evolution Backfires: Pre-Commit Gating against Skill Contamination in LLM Agents
自我进化的反噬：针对大型语言模型代理中的技能污染的预先承诺门控
Abstract
Self-evolving agents accumulate capability by distilling reusable skills from their execution trajectories, but we find this process is not monotonic: past a critical pool size, newly added skills degrade performance instead of improving it. We formalize this capability-contamination phase transition and trace it to a structural cause: once a defective skill enters the decision context, it becomes reference material for distilling later skills, forming cross-round contamination chains. We further show the contamination is structurally irreversible: removing a source skill after the fact cannot erase the flawed reasoning its descendants have already inherited, so post-hoc rollback recovers only a small fraction of the lost performance. This makes skill admission a pre-commit necessity rather than a post-hoc fix, and motivates Verifier-as-Gatekeeper (VaG): a progressive trust hierarchy whose three heterogeneous critics - structural validity, behavioral harmlessness, and semantic consistency - filter each skill individually, coupled with a marginal-gain subset selection that removes combinatorial contamination at the top tier before skills reach the runtime context. On Terminal-Bench 2, unconditional accumulation rises to a peak and then degrades, giving back most of its gains as the pool keeps growing, and post-hoc removal of the culprit skills recovers only a small part of the drop - the empirical signature of irreversibility. In contrast, VaG improves every round, reaching 72% pass@1 with a pool roughly 5x smaller, and its frozen skill pool transfers positively to four other backbones and a second benchmark without re-evolution. Ablations confirm the three critics are complementary and mutually non-substitutable, each intercepting a largely disjoint class of harmful skills.
Chinese Translation
自我进化的代理通过从其执行轨迹中提炼可重用的技能来积累能力，但我们发现这一过程并非单调：在达到临界池大小后，新添加的技能反而会降低性能而不是提升它。我们将这种能力污染的相变形式化，并追溯其结构原因：一旦有缺陷的技能进入决策上下文，它便成为提炼后续技能的参考材料，形成跨轮次的污染链。我们进一步表明，这种污染在结构上是不可逆的：事后移除源技能无法消除其后代已经继承的错误推理，因此事后回滚只能恢复一小部分损失的性能。这使得技能的接纳成为一种预先承诺的必要，而非事后的修复，并激励了验证者作为门卫（Verifier-as-Gatekeeper, VaG）：一个渐进的信任层级，其三个异质评审者——结构有效性、行为无害性和语义一致性——逐一过滤每个技能，并结合边际收益子集选择，在技能进入运行时上下文之前去除组合污染。在 Terminal-Bench 2 上，无条件积累达到峰值后又下降，随着池的不断增长，损失了大部分收益，而事后移除有害技能仅恢复了小部分下降——这就是不可逆性的经验特征。相比之下，VaG 在每一轮中均有所改善，以约 5 倍更小的池达到了 72% 的 pass@1，其冻结的技能池在没有重新进化的情况下对四个其他基础模型和第二个基准产生了积极的迁移。消融实验确认这三个评审者是互补且不可替代的，各自拦截了一个基本不重叠的有害技能类别。
cs.AI / 54 / 2608.05813
Cautious Context Steering for Language Model Personalization
谨慎的上下文引导用于语言模型个性化
Abstract
Personalizing language models (LMs) to individual user preferences is essential for aligning responses with diverse goals and backgrounds. Existing methods typically train a separate adapter for each user or learn a reward model whose scores depend on the user. Despite explicitly optimizing for each user, these methods must learn from limited observations and therefore suffer from data sparsity and poor generalization to unseen users and domains. In-context learning (ICL) and Context Steering (CoS) can instead provide more effective personalization by conditioning the base LM directly on user context and leveraging its pretrained capabilities without per-user training. Yet neither adapts the influence of that context across decoding steps: ICL leaves it uncontrolled, whereas CoS applies a fixed steering coefficient and requires two LM forward passes per step. We propose Cautious Context Steering (CCS), which adds a lightweight adapter to a frozen backbone LM to decide at each token whether and how strongly user context should affect generation. The adapter learns this behavior from an oracle context-conditioned LM and preserves the base LM when the context is not helpful. A single CCS adapter trained on only one dataset improves generation quality both in-domain and across four out-of-distribution personalization benchmarks, demonstrating robust generalization to new users and domains. CCS also avoids per-user fine-tuning and the additional context-conditioned forward pass required by CoS, substantially reducing inference cost.
Chinese Translation
将语言模型（LM）个性化以符合个别用户的偏好，对于使响应与多样化的目标和背景相一致至关重要。现有方法通常为每个用户训练一个单独的适配器，或学习一个依赖于用户的奖励模型。尽管这些方法明确针对每个用户进行优化，但必须从有限的观察中学习，因此面临数据稀疏和对未见用户和领域的泛化能力差的问题。上下文学习（ICL）和上下文引导（CoS）可以通过直接根据用户上下文对基础LM进行条件化，并利用其预训练能力而无需针对每个用户的训练，从而提供更有效的个性化。然而，这两者都未能在解码步骤中调整上下文的影响：ICL保持不受控制，而CoS则应用固定的引导系数，并要求每个步骤进行两次LM前向传递。我们提出了谨慎的上下文引导（CCS），它为一个冻结的基础LM添加了一个轻量级适配器，以决定在每个标记生成时用户上下文应如何影响生成及其强度。该适配器从一个oracle上下文条件的LM中学习这种行为，并在上下文不有用时保留基础LM。一个仅在一个数据集上训练的CCS适配器在领域内和四个超出分布的个性化基准上均提高了生成质量，展示了对新用户和领域的强健泛化能力。CCS还避免了每个用户的微调以及CoS所需的额外上下文条件前向传递，从而显著降低了推理成本。
cs.AI / 55 / 2608.05833
ViSR-KGC: Visual Subgraph Reasoning with Vision-Language Models for Multimodal Knowledge Graph Completion
ViSR-KGC：基于视觉-语言模型的视觉子图推理用于多模态知识图谱补全
Abstract
Knowledge graph completion (KGC) aims to infer missing entities or relations from incomplete graph structures, and has evolved into multimodal knowledge graph completion (MMKGC), where entities are associated with multiple modalities such as text and images. Traditional representation learning approaches follow the embedding-based paradigm and may struggle when relation-specific evidence is limited. Meanwhile, LLM-based reasoning methods typically linearize graph structures into textual prompts, which obscures structural topology and neglects vital visual information. While vision-language models (VLMs) excel at multimodal reasoning, they cannot natively interpret structured graph topology, particularly when it comes to knowledge graphs where nodes and edges carry complex semantics. To bridge this gap, we propose ViSR-KGC, a visual subgraph reasoning approach for KGC. It integrates three complementary capabilities to capture semantic correlations: identifying global topology dependencies via representation learning, analyzing local multimodal evidence using VLMs, and providing necessary commonsense knowledge inherent in pre-trained models. Based on learned multimodal embeddings, our framework first extracts a compact and query-aware subgraph from the MMKG. Then, this subgraph is transformed into a visually interpretable image using a layout strategy selected through empirical comparison.Finally, the visualized subgraph, entity images, textual descriptions, and candidate answers are combined into a unified prompt, enabling the VLM to infer the missing entity.
Chinese Translation
知识图谱补全（KGC）旨在从不完整的图结构中推断缺失的实体或关系，并已发展为多模态知识图谱补全（MMKGC），其中实体与文本和图像等多种模态相关联。传统的表示学习方法遵循基于嵌入的范式，当关系特定证据有限时可能会遇到困难。同时，基于大型语言模型（LLM）的推理方法通常将图结构线性化为文本提示，这模糊了结构拓扑并忽视了重要的视觉信息。尽管视觉-语言模型（VLMs）在多模态推理方面表现出色，但它们无法原生解释结构化图拓扑，特别是在知识图谱中，节点和边承载复杂的语义。为了弥补这一差距，我们提出了ViSR-KGC，一种用于KGC的视觉子图推理方法。它整合了三种互补能力以捕捉语义关联：通过表示学习识别全局拓扑依赖，利用VLM分析局部多模态证据，以及提供预训练模型中固有的必要常识知识。基于学习到的多模态嵌入，我们的框架首先从MMKG中提取一个紧凑且关注查询的子图。然后，使用通过经验比较选择的布局策略将该子图转换为可视化的图像。最后，将可视化的子图、实体图像、文本描述和候选答案结合成一个统一的提示，使VLM能够推断缺失的实体。
cs.AI / 56 / 2608.05863
Runtime Observability for Heterogeneous Attention Memory
异构注意力记忆的运行时可观察性
Abstract
Modern models no longer keep a plain KV cache: latent caches, learned sparse selectors and recurrent states each carry the model's memory in a different form, and each fails differently under compression. We give a runtime observability contract that covers all four memory classes with three operators, instantiate it on six model configurations across five architecture families, and compose the per-stage bounds into an executable request-level risk ledger. Contracts carry their error metric as a type -- composition is only defined when metrics match, and this check rejected our own first composed chain; the repaired chain crosses metrics through two proved bridges, and whatever no formal system can certify is measured instead, dropping the composed tier to empirical automatically: every claim is certified, partially certified, or empirical, composition inherits the weakest tier, and the tier is decided by the machine. Replayed over $12.4$M entry reads and run under eight-way concurrency with per-request budgets and fail-closed identity attribution, the ledger quantifies the honest trade-off on today's witness and holds its risk budget with zero violations. A fused always-on probe observes a declared one-layer subset under CUDA graphs inside the serving noise floor. Applied to a served DeepSeek-V4 stack with a packed compressed-KV prototype, the same machinery localizes a silent corruption to a precise structural boundary -- exact in the eviction-free, identity-isolated regime, with every observed failure in an eviction or slot-reuse regime -- through a machine-adjudicated discrimination campaign whose calculus rejected two of our own confounded inferences along the way. All artifacts, guards, and the Lean development are released at https://github.com/metask-ai/witprobe-attention-memory; every number in this paper regenerates from the shipped artifacts by one command.
Chinese Translation
现代模型不再仅仅保持一个简单的键值缓存（KV cache）：潜在缓存、学习的稀疏选择器和递归状态以不同的形式承载模型的记忆，并且在压缩下各自表现出不同的失败方式。我们提供了一个运行时可观察性契约，涵盖了所有四种记忆类别，使用三个操作符，在五个架构系列的六种模型配置上进行了实例化，并将每个阶段的界限组合成一个可执行的请求级风险账本。契约携带其误差度量作为一种类型——仅当度量匹配时，组合才被定义，而这一检查拒绝了我们自己首次组合链的有效性；修复后的链通过两个已证明的桥梁跨越度量，而任何无法被正式系统认证的内容则以测量的方式替代，自动将组合层级降至经验层级：每个声明被认证、部分认证或经验性，组合继承最弱的层级，而层级由机器决定。在超过1240万条条目读取的重放下，并在八路并发下运行，结合每个请求的预算和失败关闭的身份归属，账本量化了当今见证者的诚实权衡，并以零违规保持其风险预算。一个融合的始终开启的探针在服务噪声底层下观察声明的单层子集，使用CUDA图。应用于一个服务的DeepSeek-V4堆栈，结合一个压缩的键值原型，相同的机制将一个静默的损坏定位到一个精确的结构边界——在无驱逐、身份隔离的状态下是精确的，而在驱逐或插槽重用的状态下每个观察到的失败都是如此——通过一个机器裁决的歧视活动，其计算拒绝了我们在此过程中的两个混淆推断。所有工件、保护措施和精益开发都已发布在 https://github.com/metask-ai/witprobe-attention-memory；本文中的每一个数字都可以通过一个命令从发布的工件中再生。
cs.AI / 57 / 2608.05864
Seeing Is Not Deciding: Can Multimodal LLMs Act as Effective CEOs?
看见并不等于决策：多模态大语言模型能否作为有效的首席执行官？
Abstract
Large language models are increasingly applied as autonomous decision-making agents. However, in executive business decisions, existing benchmarks are limited to textonly settings. This makes it unclear whether models can perceive visual business evidence and effectively integrate it to improve decision quality. We introduce C-SUITEBENCH, a controlled multimodal benchmark that includes five decision tasks under paired text-only and multimodal conditions across 50 scenarios. We place nine frontier models in the role of a chief executive officer and evaluate their decision-making ability. Multimodal inputs consistently improve evidence-centric reasoning, with the largest and most reliable gains appearing in risk forecasting and board-facing justification. However, we uncover a multimodal integration paradox: adding visual business information degrades constrained resource allocation for all nine models, even as visual grounding itself improves. Ablation experiments reveal that this failure emerges from signal crowding, although each visual channel helps individually, their combination disrupts constraint satisfaction during decoding. These findings demonstrate that visual perception and constrained action are separable bottlenecks in multimodal agents, and that indiscriminate visual augmentation can harm high-stakes decision making, motivating selective grounding strategies for future executive AI systems.
Chinese Translation
大型语言模型越来越多地被应用于自主决策代理。然而，在高管商业决策中，现有基准仅限于文本环境。这使得模型是否能够感知视觉商业证据并有效整合以提高决策质量变得不明确。我们引入了 C-SUITEBENCH，这是一个受控的多模态基准，包含在 50 个场景下的五个决策任务，分别在文本-only 和多模态条件下进行。我们将九个前沿模型置于首席执行官的角色，并评估它们的决策能力。多模态输入始终改善以证据为中心的推理，最大的和最可靠的收益出现在风险预测和面对董事会的辩护中。然而，我们发现了一个多模态整合悖论：添加视觉商业信息降低了所有九个模型的资源约束分配，即使视觉基础本身有所改善。消融实验表明，这一失败源于信号拥挤，尽管每个视觉通道单独都有帮助，但它们的组合在解码过程中破坏了约束满足。这些发现表明，视觉感知和受限行动是多模态代理中的可分瓶颈，且不加选择的视觉增强可能会损害高风险决策，促使未来的执行 AI 系统采用选择性基础策略。
cs.AI / 58 / 2608.05867
Improving Interoperability among Defence and National Security Ontologies: Analysis and Evaluation Tasks
提升国防与国家安全本体之间的互操作性：分析与评估任务
Abstract
The use of ontologies and knowledge graphs is becoming increasingly widespread in the defence and national security domain. Numerous ontologies have been developed through initiatives led by academia, industry, and government. Achieving interoperability across diverse defence and national security ontologies remains a major challenge due to the domain's breadth and specialisation. In this work, we analyse and document over 60 publicly available ontologies and introduce a new track for the Ontology Alignment Evaluation Initiative (OAEI). This track comprises eight matching tasks, consensus alignments and manually-curated (silver-standard) mappings. The consensus alignments are derived by aggregating the outputs of several state-of-the-art ontology alignment systems. The silver-standard is obtained from the manual validation of the consensus alignment together with a subset of the unique mappings (i.e., mappings suggested by only one system).
Chinese Translation
本体和知识图谱在国防与国家安全领域的应用日益广泛。许多本体是通过学术界、工业界和政府主导的倡议开发的。由于该领域的广泛性和专业化，实现不同国防与国家安全本体之间的互操作性仍然是一个重大挑战。在本研究中，我们分析并记录了60多个公开可用的本体，并为本体对齐评估倡议（Ontology Alignment Evaluation Initiative, OAEI）引入了一个新的任务轨道。该轨道包含八个匹配任务、共识对齐和手动整理的（银标准）映射。共识对齐是通过聚合多个最先进的本体对齐系统的输出而得出的。银标准是通过手动验证共识对齐以及一部分唯一映射（即仅由一个系统建议的映射）获得的。
cs.AI / 59 / 2608.05876
Personalized Deep Research Query Refinement with Graph-Scaffolded Evidence Grounding
基于图结构证据支撑的个性化深度研究查询优化
Abstract
User requests serve as research specifications for deep research agents, shaping what evidence to seek and how to synthesize it. In personalized deep research, these specifications must additionally reflect user goals, constraints, preferences, and evaluation criteria. User context can be incorporated either within the deep research pipeline or into the research specification provided as its input. We focus on the latter, refining the user request into a personalized research specification before passing it to an unchanged deep research agent. This requires resolving three coupled decisions: which framing factors are relevant, whether the available user context sufficiently supports them, and whether to retrieve user memory, ask the user, or stop and refine the query. For training, G-STEER organizes framing factors as elicitation targets in an Intent Elicitation Graph that captures their dependencies. It learns a clarification policy from graph-scaffolded trajectories spanning diverse factor dependencies and evidence conditions. The policy produces a refined query while balancing target coverage against the costs of evidence acquisition. Experiments show that G-STEER achieves the strongest overall weighted target coverage and the highest downstream report personalization across both evaluated DRAs, while asking roughly one third as many user questions as a strong clarification baseline.
Chinese Translation
用户请求作为深度研究代理的研究规范，决定了需要寻找何种证据以及如何进行综合。在个性化深度研究中，这些规范还必须反映用户的目标、约束、偏好和评估标准。用户上下文可以在深度研究流程中纳入，或作为输入提供给研究规范。我们关注后者，将用户请求优化为个性化研究规范，然后传递给未改变的深度研究代理。这需要解决三个相互关联的决策：哪些框架因素是相关的，现有的用户上下文是否足够支持这些因素，以及是检索用户记忆、询问用户，还是停止并优化查询。为了进行训练，G-STEER将框架因素组织为意图引导图中的引导目标，以捕捉它们之间的依赖关系。它从跨越不同因素依赖性和证据条件的图结构轨迹中学习澄清策略。该策略在平衡目标覆盖与证据获取成本的同时，生成优化后的查询。实验表明，G-STEER在两个评估的深度研究代理中实现了最强的整体加权目标覆盖和最高的下游报告个性化，同时询问的用户问题数量大约是强澄清基线的三分之一。
cs.AI / 60 / 2608.05891
AppDeltaWorld: Transition-Grounded Delta Code World Model for Mobile GUI Agents
AppDeltaWorld：面向移动图形用户界面的过渡基础增量代码世界模型
Abstract
Mobile GUI agents can operate apps through pixel perception and touch actions, making them a promising interface for collecting and improving long-horizon mobile interaction policies. However, real trajectories are difficult to obtain for sensitive apps and privacy-critical operations. At the same time, existing simulated environments are costly to scale up, and GUI world models still suffer from unstable generation, limited modality coverage, and inconsistent action-transition logic. To address these limitations, we propose AppDeltaWorld, a transition-grounded delta code world model that predicts the next GUI as a reachable code update rather than as an unconstrained image or text description. AppDeltaWorld retrieves app-specific Level-1 HTML references under an action-transition constraint, generates Level-2 executable HTML conditioned on the current screen, action, predicted next-screen text, and retrieved structure, and inserts generated visual assets into image slots before browser rendering. As a world model, AppDeltaWorld achieves the highest fidelity on CMGUIBench-500 under Code2World evaluation, with clear gains in structural layout and UI element reconstruction over image-only and code-only baselines. As a training environment, AppDeltaWorld supports filtered closed-loop SFT data construction that, when combined with public supervision, enables AppDeltaAgent to achieve state-of-the-art performance on AndroidLens and consistent gains on MobileGym and MobileWorld. Moreover, world-model-based test-time reinforcement learning enables policy adaptation and shows further improvements without additional interaction with real apps.
Chinese Translation
移动图形用户界面（GUI）代理可以通过像素感知和触摸操作来操作应用程序，使其成为收集和改善长期移动交互策略的有前景的接口。然而，敏感应用和隐私关键操作的真实轨迹难以获取。同时，现有的模拟环境在扩展上成本高昂，GUI世界模型仍然存在生成不稳定、模态覆盖有限和动作-过渡逻辑不一致等问题。为了解决这些局限性，我们提出了AppDeltaWorld，这是一种过渡基础的增量代码世界模型，它将下一个GUI预测为可达到的代码更新，而不是不受限制的图像或文本描述。AppDeltaWorld在动作-过渡约束下检索特定应用的一级HTML引用，基于当前屏幕、动作、预测的下一个屏幕文本和检索的结构生成二级可执行HTML，并在浏览器渲染之前将生成的视觉资产插入图像槽中。作为一个世界模型，AppDeltaWorld在Code2World评估中在CMGUIBench-500上实现了最高的保真度，相较于仅使用图像或代码的基线，在结构布局和用户界面元素重建方面有明显提升。作为训练环境，AppDeltaWorld支持过滤的闭环SFT数据构建，当与公共监督结合时，使得AppDeltaAgent在AndroidLens上实现了最先进的性能，并在MobileGym和MobileWorld上持续获得提升。此外，基于世界模型的测试时间强化学习使得策略适应成为可能，并在无需与真实应用进行额外交互的情况下显示出进一步的改进。
cs.AI / 61 / 2608.05893
ECG-LENS: Lead-Aware Clinical Context Enriched ECG Report Generation and Evaluation
ECG-LENS：基于导联的临床背景丰富的心电图报告生成与评估
Abstract
Electrocardiography (ECG) is one of the most widely used non-invasive tools for diagnosing cardiovascular disease, but transforming multi-lead ECG recordings into reliable clinical reports remains challenging. Automating ECG report generation could reduce clinicians' interpretive workload, improve diagnostic efficiency, and expand access to cardiac assessment in underserved communities. Unlike image-based report-generation tasks, ECG interpretation requires the analysis of subtle temporal morphologies, followed by coherent diagnostic reasoning expressed in dense clinical terminology. Existing systems predominantly focus on classification, while current report-generation methods often produce outputs that remain inadequate for practical clinical use. To address these challenges, we propose ECG-LENS, an end-to-end ECG report-generation framework that jointly integrates multi-lead signal modeling, diagnosis-aware representations, and clinically grounded text generation. ECG-LENS combines lead-wise encoders that preserve localized waveform morphology with a global encoder that captures inter-lead dependencies. To guide report generation, we fuse signal representations with clinically enriched textual prompts that condition a GPT-2 decoder. We further introduce an ECG-specific report-preprocessing strategy that helps the model focus on clinically meaningful findings. Finally, because lexical metrics may under- or overestimate report quality, we propose F1-ECGBERT, a BERT-based, ECG-specific metric that measures agreement between diagnostic labels extracted from generated and reference reports. In-domain experiments on PTB-XL and cross-domain evaluation on MIMIC-IV-ECG show that ECG-LENS consistently outperforms state-of-the-art methods, with absolute gains of 4.0%, 6.3%, and 11.5% in METEOR, ROUGE-L, and F1-ECGBERT, respectively, over the strongest baselines.
Chinese Translation
心电图（ECG）是诊断心血管疾病最广泛使用的非侵入性工具之一，但将多导联心电图记录转化为可靠的临床报告仍然面临挑战。自动化心电图报告生成可以减轻临床医生的解读工作负担，提高诊断效率，并扩大对服务不足社区的心脏评估的获取。与基于图像的报告生成任务不同，心电图解读需要分析微妙的时间形态，并随后进行以密集临床术语表达的连贯诊断推理。现有系统主要集中在分类上，而当前的报告生成方法往往产生的输出在实际临床使用中仍显不足。为了解决这些挑战，我们提出了ECG-LENS，一个端到端的心电图报告生成框架，联合整合多导联信号建模、诊断感知表示和临床基础文本生成。ECG-LENS结合了保留局部波形形态的导联编码器和捕捉导联间依赖关系的全局编码器。为了指导报告生成，我们将信号表示与临床丰富的文本提示融合，以调节GPT-2解码器。我们进一步引入了一种特定于心电图的报告预处理策略，帮助模型专注于临床有意义的发现。最后，由于词汇度量可能低估或高估报告质量，我们提出了F1-ECGBERT，这是一种基于BERT的、特定于心电图的度量，用于衡量从生成报告和参考报告中提取的诊断标签之间的一致性。在PTB-XL上的领域内实验和在MIMIC-IV-ECG上的跨领域评估显示，ECG-LENS始终优于最先进的方法，在METEOR、ROUGE-L和F1-ECGBERT上分别比最强基线获得了4.0%、6.3%和11.5%的绝对提升。
cs.AI / 62 / 2608.05896
GSBF: Gaussian Splatting for Environment-Aware Beamforming
GSBF：用于环境感知波束形成的高斯散射
Abstract
Beamforming plays a key role in multiple-input-multiple-output (MIMO) communication systems. However, conventional beamforming design normally requires accurate instantaneous channel state information (CSI) and iterative optimization, which incur substantial pilot overhead and computational complexity. Recognizing that radio propagation is intrinsically governed by the physical geometry, we develop a 3D Gaussian splatting for environment-aware beamforming (GSBF) pipeline based on multi-modal data, which characterizes the environment through a persistent 3D Gaussian representation. Specifically, GSBF models the environmental scattering response with reciprocity-preserving bidirectional spherical Gaussian (Bi-SG) kernels and performs two-sided electromagnetic rasterization to render an angular propagator map. The rendered map is then aggregated through an over-complete array-manifold dictionary and projected to the constant-modulus beamformers, thereby synthesizing beams directly from the access point (AP) pose and user position without online instantaneous CSI. Simulations demonstrate that GSBF consistently outperforms baselines such as exhaustive beam alignment (EBA) with lower latency.
Chinese Translation
波束形成在多输入多输出（MIMO）通信系统中发挥着关键作用。然而，传统的波束形成设计通常需要准确的瞬时信道状态信息（CSI）和迭代优化，这会导致大量的导频开销和计算复杂性。我们认识到无线传播本质上受到物理几何的支配，因此开发了一种基于多模态数据的环境感知波束形成（GSBF）管道，该管道通过持久的三维高斯表示来表征环境。具体而言，GSBF使用保持互易性的双向球面高斯（Bi-SG）核来建模环境散射响应，并执行双侧电磁光栅化以渲染角度传播器图。然后，通过过完备的阵列流形字典对渲染的图进行聚合，并投影到恒模波束形成器，从而直接从接入点（AP）位置和用户位置合成波束，而无需在线瞬时CSI。仿真结果表明，GSBF在延迟更低的情况下始终优于基线方法，如穷举波束对齐（EBA）。
cs.AI / 63 / 2608.05910
CourseGraph: Finding overlaps and differences in Computer Science courses across universities
CourseGraph：寻找大学计算机科学课程之间的重叠与差异
Abstract
Student mobility programs such as Erasmus+ enable students to take courses at other universities, broadening their academic and cultural horizons. However, this flexibility also leads to a practical challenge: ensuring that students do not take courses elsewhere that substantially overlap with courses in their home curriculum. In this work, we propose CourseGraph, a methodology that automates the evaluation of external courses based on insights obtained from the process followed by curriculum administrators when assessing courses for inclusion in a degree program. Course- Graph extracts information such as course titles, descriptions, and learning outcomes from the course webpage. Then, this information is represented semantically using a BERT-based language model, after which the pair-wise similarity between courses can be computed. This information is then used by a Random Forest classifier to determine whether a candidate course abroad overlaps with a course already contained in the student's curriculum. We evaluate CourseGraph using (1) the Computer Science program at Eindhoven University of Technology, which contains information about courses with substantial overlap, and (2) six approved international programs from students enrolled in the Computer Science program at Lund University, including the corresponding decisions made by a curriculum administrator. The experimental results indicate that CourseGraph provides an effective approach for identifying overlapping courses and supporting curriculum alignment across universities.
Chinese Translation
学生流动项目如Erasmus+使学生能够在其他大学修读课程，从而拓宽他们的学术和文化视野。然而，这种灵活性也带来了一个实际挑战：确保学生在其他地方修读的课程与其本校课程之间没有实质性的重叠。在本研究中，我们提出了CourseGraph，这是一种基于课程管理员在评估课程以纳入学位项目时所遵循的流程所获得的见解，自动化评估外部课程的方法。CourseGraph从课程网页中提取课程标题、描述和学习成果等信息。然后，使用基于BERT的语言模型对这些信息进行语义表示，之后可以计算课程之间的成对相似性。接着，这些信息被随机森林分类器用于判断国外候选课程是否与学生课程中已包含的课程存在重叠。我们使用（1）埃因霍温理工大学的计算机科学项目进行CourseGraph的评估，该项目包含有关课程重叠的重要信息，以及（2）来自隆德大学计算机科学项目的六个经过批准的国际项目，包括课程管理员所做的相应决策。实验结果表明，CourseGraph提供了一种有效的方法来识别重叠课程，并支持大学之间的课程对齐。
cs.AI / 64 / 2608.05948
GAUGE: A Measurement-Grounded Benchmark for Physical Fidelity in Simulation Engines and Video World Models
GAUGE：一个基于测量的基准，用于评估仿真引擎和视频世界模型中的物理真实性
Abstract
Physics engines facilitate large-scale training and evaluation for embodied intelligence, while generative video world models are emerging as implicit simulators of future states and interactions. However, existing evaluations of physical fidelity are often conducted in isolation and rely heavily on perceptual similarity or human judgments, providing limited insight into which physical principles or parameters are violated. We introduce GAUGE, a real-world-grounded diagnostic benchmark for jointly evaluating how numerical simulators and generative video world models reproduce or deviate from real-world physics. It comprises 22 controlled task families covering rigid bodies, flexible cables, textiles, and volumetric deformable objects. Grounded in real-world trajectories and paired with calibrated physical metadata, uncertainty annotations, and task-specific observables, these tasks cover fundamental physical processes including collision, friction, momentum transfer, oscillation, self-contact, and deformation across diverse materials and conditions. We benchmark Isaac Sim, Genesis, and Newton on 14 task families using generalized trajectory errors, and evaluate 6 image-to-video models on 5 rigid-body tasks by testing physical-law consistency and the temporal stability of inferred parameters. Our results reveal no uniformly faithful physics engine, with the largest discrepancies arising in impulsive contact, rapid textile motion, and volumetric deformation. We further find that video world models can produce trajectories with the expected equation form while recovering incorrect accelerations, momentum transfer, and oscillation timing. GAUGE lays the groundwork for developing more physically faithful simulators and world models for embodied intelligence.
Chinese Translation
物理引擎促进了对具身智能的大规模训练和评估，而生成的视频世界模型正逐渐成为未来状态和交互的隐式模拟器。然而，现有的物理真实性评估往往是孤立进行的，并且过于依赖感知相似性或人类判断，提供的洞见有限，无法明确指出哪些物理原理或参数被违反。我们提出了GAUGE，这是一个基于真实世界的诊断基准，用于共同评估数值模拟器和生成视频世界模型如何再现或偏离真实世界的物理。该基准包括22个受控任务系列，涵盖刚体、柔性电缆、纺织品和体积可变形物体。基于真实世界的轨迹，并配备经过校准的物理元数据、不确定性注释和任务特定的可观测量，这些任务涵盖了包括碰撞、摩擦、动量转移、振荡、自接触和变形等基本物理过程，适用于多种材料和条件。我们使用广义轨迹误差对Isaac Sim、Genesis和Newton在14个任务系列上进行了基准测试，并通过测试物理定律一致性和推断参数的时间稳定性，对6个图像到视频模型在5个刚体任务上进行了评估。我们的结果显示，没有统一忠实的物理引擎，最大差异出现在冲击接触、快速纺织运动和体积变形中。我们进一步发现，视频世界模型可以生成具有预期方程形式的轨迹，同时恢复不正确的加速度、动量转移和振荡时序。GAUGE为开发更具物理真实性的模拟器和具身智能的世界模型奠定了基础。
cs.AI / 65 / 2608.05949
VLMs for Videogame Data Annotation
用于视频游戏数据标注的视觉语言模型
Abstract
Vision Language Models (VLMs) and Artificial Intelligence (AI) agents have revolutionized how engineers approach complex problems in real-world applications. Their adoption in video games is on the other hand limited by the extreme variability of the synthetic scenarios and their poor compliance with real-world physics. Here we investigate the use of VLMs for annotating video game frame sequences with reward signals, a task with several potential applications including, among others, conditioned training and offline reinforcement learning. We show that VLMs often struggle to answer basic questions on racing video games (although we observed a similar behavior on other game genres) and discuss countermeasures such as VLM output mixing and prompt optimization. We also show how input sequence length, resolution, and question batching affect the annotation quality and its token consumption.
Chinese Translation
视觉语言模型（Vision Language Models, VLMs）和人工智能（Artificial Intelligence, AI）代理已经彻底改变了工程师在现实应用中处理复杂问题的方法。然而，它们在视频游戏中的应用受到合成场景极端变异性和与现实物理不符的限制。在此，我们研究了使用VLMs为视频游戏帧序列标注奖励信号的可行性，这一任务具有多种潜在应用，包括条件训练和离线强化学习等。我们发现，VLMs在回答赛车视频游戏的基本问题时常常遇到困难（尽管我们在其他游戏类型中也观察到了类似的行为），并讨论了诸如VLM输出混合和提示优化等对策。我们还展示了输入序列长度、分辨率和问题批处理如何影响标注质量及其标记消耗。
cs.AI / 66 / 2608.05954
Training a Conditioned Video Game Agent on a VLM Annotated Dataset
在VLM注释数据集上训练条件视频游戏代理
Abstract
Reinforcement Learning (RL) is a powerful but far from easy-to-use technique for policy learning. In the specific case of video games, access to the game engine is required to get rewards for training (e.g. to collect rewards from the environment). Furthermore, the proper identification and weighting of the rewards generally requires a difficult trial-and-error approach. Lastly, rewards are often sparse and understanding how they eventually affect the learned policy is a non-trivial exercise. To ease these issues we propose annotating a video game dataset with Vision Language Models (VLMs) instructed to extract human defined rewards. We show that offline RL can then be used to train a conditioned agent that responds accordingly to the desired returns and we discuss the difficulties and limitations that emerged in our early experiments.
Chinese Translation
强化学习（Reinforcement Learning, RL）是一种强大但并非易于使用的策略学习技术。在视频游戏的特定情况下，需要访问游戏引擎以获取训练奖励（例如，从环境中收集奖励）。此外，正确识别和加权奖励通常需要困难的试错方法。最后，奖励往往是稀疏的，理解它们最终如何影响学习到的策略是一项非平凡的任务。为了解决这些问题，我们建议使用视觉语言模型（Vision Language Models, VLMs）对视频游戏数据集进行注释，以提取人类定义的奖励。我们展示了离线强化学习（offline RL）可以用于训练一个条件代理，使其能够相应地响应所需的回报，并讨论了我们早期实验中出现的困难和局限性。
cs.AI / 67 / 2608.05958
Stability of Ranking-dependent Pair-wise Comparison Patterns in the Analytic Hierarchy Process
层次分析法中基于排名的成对比较模式的稳定性
Abstract
The paper addresses several ranking-dependent decision support methods. Ordinal information on compared objects can be used to improve the quality of expert data during estimation and help reduce the number of comparisons that the experts need to perform. In the paper we compare three incomplete ranking-dependent pair-wise comparison patterns which can be used in the Analytic Hierarchy Process - Best-worst method, Best-Second Best (Top 2) method, and the original maximum difference method. The first two comparison patterns (and respective methods) are incomplete, while the third can be a complete one. We determine conditions under which these three methods can be compared in terms of stability to expert errors. We also present the results of a simulation-type experiment, in which the three methods are compared. The research allows us to define the most stable incomplete ranking-dependent pair-wise comparison pattern and reduce the number of comparisons without loss of credibility of expert session results. The research contributes to algorithmic, cognitive, and applied aspects of decision support in uncertain environments.
Chinese Translation
本文探讨了几种基于排名的决策支持方法。比较对象的序数信息可以用于提高专家数据在估计过程中的质量，并帮助减少专家需要进行的比较次数。我们比较了三种不完整的基于排名的成对比较模式，这些模式可用于层次分析法（Analytic Hierarchy Process）——最佳-最差法（Best-worst method）、最佳-第二最佳法（Best-Second Best, Top 2 method）以及原始的最大差异法（maximum difference method）。前两种比较模式（及其相应的方法）是不完整的，而第三种可以是完整的。我们确定了在何种条件下这三种方法可以在专家错误的稳定性方面进行比较。我们还展示了一项模拟实验的结果，其中比较了这三种方法。研究使我们能够定义出最稳定的不完整基于排名的成对比较模式，并在不损失专家会议结果可信度的情况下减少比较次数。该研究对不确定环境中的决策支持的算法、认知和应用方面做出了贡献。
cs.AI / 68 / 2608.05981
Temporal Bridges for Spatial Resolution: Enhancing Climate Data Super-Resolution with Bidirectional Alignment
时序桥接空间分辨率：通过双向对齐增强气候数据超分辨率
Abstract
High-resolution climate data is crucial for meteorological predictions and for informing decision support across diverse domains. However, the acquisition of such high-resolution climate information is often prohibitively costly, necessitating the development of data-driven meteorological prediction models. These models aim to generate fine-grained climate data from low-resolution inputs, a process termed climate data super-resolution (SR). Nevertheless, recent advancements in deep learning for climate data SR have primarily focused on leveraging single-frame spatial information, largely neglecting the temporal correlations between different time frames that could enhance SR outcomes. Furthermore, climate data are inherently stochastic and noisy, rendering widely used temporal alignment methods, such as optical flow models, ineffective in this context. Consequently, the development of a framework tailored for climate data SR that effectively captures implicit temporal correlations remains an unresolved challenge. To this end, we propose a novel Temporal-Enhanced framework with bidirectional temporal alignment. In essence, our framework establishes a temporal bridge to enhance spatial resolution in climate data SR through bidirectional alignment, leading to improved SR performance. Within this framework, Paired Latent Mapping achieves spatial alignment and noise reduction by unifying latent spaces. Then a Bidirectional Temporal Alignment captures temporal correlations by training forward and backward networks on consecutive latent frames. Temporal Enhanced Super-resolution then optimizes the entire framework for climate data SR. Experiments on large-scale real-world datasets demonstrated the superior performance of our framework.
Chinese Translation
高分辨率气候数据对于气象预测和在各个领域提供决策支持至关重要。然而，获取这种高分辨率气候信息往往成本高昂，这就需要开发基于数据的气象预测模型。这些模型旨在从低分辨率输入中生成细粒度的气候数据，这一过程被称为气候数据超分辨率（SR）。然而，近年来在气候数据SR方面的深度学习进展主要集中于利用单帧空间信息，基本忽视了不同时间帧之间的时间相关性，而这些相关性可能会增强SR结果。此外，气候数据本质上是随机和噪声干扰的，这使得广泛使用的时间对齐方法（如光流模型）在此背景下效果不佳。因此，开发一个专门针对气候数据SR的框架，以有效捕捉隐含的时间相关性，仍然是一个未解决的挑战。为此，我们提出了一种新颖的时序增强框架，采用双向时间对齐。我们的框架本质上建立了一个时序桥，以通过双向对齐增强气候数据SR中的空间分辨率，从而提高SR性能。在该框架内，配对潜在映射通过统一潜在空间实现空间对齐和噪声减少。然后，双向时间对齐通过在连续潜在帧上训练前向和后向网络来捕捉时间相关性。最后，时序增强超分辨率优化整个框架以实现气候数据SR。在大规模真实世界数据集上的实验表明，我们的框架具有优越的性能。
cs.AI / 69 / 2608.05987
AgentOPSD: Recursive Self-Distillation for Agentic Reinforcement Learning
AgentOPSD：用于智能强化学习的递归自蒸馏
Abstract
Reinforcement learning (RL) with verifiable rewards constructs trajectory-level advantage estimates, yet it often fails to credit the few pivotal decisions that determine outcomes in long-horizon, multi-turn agentic tasks. Recent work introduces privileged self-distillation for credit assignment, providing denser supervision, but it remains unclear how such local signals should represent sequential credit. We propose AgentOPSD, a critic-free, recursive method for turn-level credit assignment in agentic reinforcement learning. AgentOPSD aggregates token-level teacher-student log-probability gaps into turn-level evidence and recursively updates a Bayesian belief state in log-odds space. This yields a principled reweighting scheme that converts sparse outcome supervision into turn-level credit signals and identifies pivotal turns through the marginal belief revision between consecutive states. The method is fully compatible with standard policy optimization and requires neither an additional critic nor extra rollouts. We evaluate AgentOPSD on ALFWorld, WebShop, and Search-QA using Qwen2.5 models at two scales (3B and 7B). AgentOPSD outperforms GRPO and strong self-distillation baselines, achieving 89.1% success on ALFWorld with Qwen2.5-7B. Ablation studies attribute the gains to turn-level aggregation and history-dependent recursive belief updates.
Chinese Translation
具有可验证奖励的强化学习（RL）构建了轨迹级优势估计，但在长时间跨度的多回合智能任务中，它往往未能对决定结果的少数关键决策给予应有的重视。近期的研究提出了特权自蒸馏用于信用分配，提供了更密集的监督，但尚不清楚这些局部信号应如何表示顺序信用。我们提出了AgentOPSD，这是一种无评论员的递归方法，用于智能强化学习中的回合级信用分配。AgentOPSD将标记级教师-学生对数概率差异聚合为回合级证据，并在对数赔率空间中递归更新贝叶斯信念状态。这产生了一种原则性的重加权方案，将稀疏的结果监督转化为回合级信用信号，并通过连续状态之间的边际信念修正识别关键回合。该方法与标准策略优化完全兼容，无需额外的评论员或额外的回合。我们在ALFWorld、WebShop和Search-QA上评估了AgentOPSD，使用Qwen2.5模型的两个规模（3B和7B）。AgentOPSD在ALFWorld上以Qwen2.5-7B实现了89.1%的成功率，优于GRPO和强自蒸馏基线。消融研究将增益归因于回合级聚合和历史依赖的递归信念更新。
cs.AI / 70 / 2608.05990
OPERA: Operator-residual feedback for reliable autonomous optical experiments with language-model agents
OPERA：用于可靠自主光学实验的算子残差反馈与语言模型代理
Abstract
Autonomous agents choose actions using scores that may not reflect experimental success. We developed OPERA, an operator-residual framework for optical experiments. It represents experimental actions as optical operators and evaluates their outcomes using physically interpretable residuals. Operators specify executable changes to measurement, control or reconstruction, while residuals report departures from specified physical conditions. The agent uses both to select, combine or generate operators, and physical performance is evaluated independently against a withheld reference. Across three optical tasks, score-only feedback produced score increases without physical improvement in 23.6--39.0\% of decisions, compared with 0.9--1.9\% for operator-residual feedback. Operator-residual feedback increased the probability of reaching and maintaining task targets and reduced experimental budgets. Protocols selected in digital twins were transferred to three optical instruments, and repeated experiments showed a lower projection budget in structured-light reconstruction. Together, operators and residuals guide autonomous decisions using measurable physical evidence.
Chinese Translation
自主代理使用的评分可能无法反映实验的成功。我们开发了OPERA，一个用于光学实验的算子残差框架。它将实验操作表示为光学算子，并使用物理可解释的残差来评估其结果。算子指定可执行的测量、控制或重建的变化，而残差报告与指定物理条件的偏离。代理使用这两者来选择、组合或生成算子，并且物理性能在独立于保留参考的情况下进行评估。在三个光学任务中，仅使用评分的反馈在23.6%到39.0%的决策中产生了评分增加，但未带来物理改善，而算子残差反馈的这一比例仅为0.9%到1.9%。算子残差反馈提高了达到和维持任务目标的概率，并减少了实验预算。在数字双胞胎中选择的协议被转移到三台光学仪器上，重复实验显示在结构光重建中具有更低的投影预算。算子和残差共同利用可测量的物理证据指导自主决策。
cs.AI / 71 / 2608.06001
Hybrid Machine Learning Framework for Herd-Level Cattle Growth Pattern and Weight Gain Forecasting in Grazing-Based Production Systems
基于混合机器学习框架的放牧系统中牛群生长模式和增重预测
Abstract
Commercial grazing systems yield irregular livestock observations, which challenge cattle growth forecasting. This study developed a hybrid machine learning framework for herd level cattle weight forecasting using automated sensing observations collected between 2022 and 2024 in southeastern Australia. Weekly live weight observations, demographic variables, and lagged environmental predictors were integrated into structured forecasting datasets. Herd level forecasting trajectories were generated through temporal aggregation of animal level predictions. Four hybrid architecture families were evaluated, including residual, stacked, cascade, and ensemble assisted frameworks. ARIMA, LSTM, and GRU models were used as comparative baselines. Independent testing demonstrated strong predictive agreement across multiple forecasting horizons. The cascade GB to RF to NN architecture achieved the best performance, with a test R^2 of 0.889, RMSE of 21.319 kg, and MAE of 15.462 kg. Hybrid architectures maintained greater robustness than recurrent sequential models under sparse observation conditions. Forecasting error increased progressively across extended prediction horizons. Feature importance analysis identified animal age, rainfall, and temperature as dominant predictors influencing herd level growth forecasting. The proposed framework may support feed allocation, grazing management, and livestock marketing decisions under heterogeneous sensing environments.
Chinese Translation
商业放牧系统产生不规则的牲畜观察数据，这对牛的生长预测提出了挑战。本研究开发了一种混合机器学习框架，用于基于2022年至2024年在澳大利亚东南部收集的自动传感观察数据进行牛群级别的体重预测。将每周的活体重观察、人口统计变量和滞后环境预测因子整合为结构化的预测数据集。通过对动物级别预测的时间聚合生成牛群级别的预测轨迹。评估了四种混合架构系列，包括残差、堆叠、级联和集成辅助框架。ARIMA、LSTM和GRU模型被用作比较基准。独立测试显示在多个预测时间范围内具有强预测一致性。级联GB到RF到NN架构实现了最佳性能，测试R^2为0.889，均方根误差(RMSE)为21.319公斤，平均绝对误差(MAE)为15.462公斤。在稀疏观察条件下，混合架构比递归序列模型保持了更大的稳健性。随着预测时间范围的延长，预测误差逐渐增加。特征重要性分析确定了动物年龄、降雨量和温度是影响牛群生长预测的主要预测因子。所提出的框架可能支持在异质感知环境下的饲料分配、放牧管理和牲畜营销决策。
cs.AI / 72 / 2608.06012
HERALD: Counterfactual Audits and Minimal Repairs for Proof-of-Retrieval Rewards
HERALD：反事实审计与最小修复用于检索奖励的证明
Abstract
Search-agent rewards mix answer quality, citation grounding, tool cost, and anti-hacking terms; a high score therefore need not imply that cited evidence was retrieved, and added penalties can cancel. We introduce HERALD, an offline audit that applies exact same-question interventions, separates candidate-visible from oracle information, and enumerates detector contracts before policy optimization. On four Qwen3-8B pools from HotpotQA, 2WikiMultiHopQA, and MuSiQue, $R_0$ rejects search deletion and fake IDs, but a label-free citation-laundering attack succeeds. A complete $2^3$ ablation identifies targeted strengthening of $L$---citing a corpus passage absent from the retrieved evidence---as the observed inclusion-minimal repair: $R[L]$ has zero empirical ASR with a 0.50% one-sided cluster upper bound. The gap persists across pool rules, a visible BM25 attacker, and four models; broader hardening remains vulnerable when the attack removes an oracle support-ID penalty. Under strict 5M-token matched training evaluated on 256 paired questions per benchmark, $R[L]$ meets the EM non-inferiority gate on HotpotQA and 2Wiki but not MuSiQue. Equal-suite citation precision and support recall improve by 2.02 and 1.46 points, unsupported citations fall by 1.69, and laundering attackability falls on 2Wiki and MuSiQue. Natural $L$ is not reduced, and the detector appears in only 18 of 58,368 training trajectories. HERALD thus separates robust scoring, sparse learning signal, and policy transfer.
Chinese Translation
搜索代理的奖励混合了答案质量、引用基础、工具成本和反黑客条款；因此，高分并不一定意味着引用的证据被检索到，并且附加的惩罚可能会抵消。我们引入了HERALD，这是一种离线审计，应用完全相同问题的干预，区分候选可见信息与oracle信息，并在策略优化之前枚举检测器合同。在来自HotpotQA、2WikiMultiHopQA和MuSiQue的四个Qwen3-8B数据集中，$R_0$拒绝搜索删除和虚假身份，但无标签的引用洗白攻击成功。一个完整的$2^3$消融实验识别出针对性的$L$增强——引用一个在检索证据中缺失的语料段落——作为观察到的包含最小修复：$R[L]$的经验ASR为零，且一侧聚类的上限为0.50%。这一差距在不同的池规则、可见的BM25攻击者和四个模型中持续存在；当攻击去除oracle支持ID惩罚时，更广泛的强化仍然脆弱。在严格的5M-token匹配训练下，针对每个基准评估256对问题，$R[L]$在HotpotQA和2Wiki上满足EM非劣性门槛，但在MuSiQue上未能满足。等套件的引用精度和支持召回分别提高了2.02和1.46点，未支持的引用减少了1.69，洗白攻击的可攻击性在2Wiki和MuSiQue上下降。自然$L$没有减少，检测器仅出现在58,368条训练轨迹中的18条中。因此，HERALD将稳健评分、稀疏学习信号和策略转移分开。
cs.AI / 73 / 2608.06020
From Economic Agents to Agentic Economies: A Systems Blueprint for Economic World Models
从经济主体到代理经济：经济世界模型的系统蓝图
Abstract
Economic World Models (EWMs) are generative economic models that simulate how economies evolve from within by modeling heterogeneous agents, their beliefs and actions, and the market and institutional mechanisms through which their interactions produce aggregate outcomes. This paper develops an implementation roadmap for building economic world models as generative engines in which heterogeneous agents act, interact, adapt, and co-evolve with markets and institutions, thereby producing economic dynamics from the inside. We organize EWM systems into a six-level capability ladder, from fixed rule-based agent worlds to adaptive and LLM-based agent worlds, self-evolving agents, evolving institutional worlds, and sim-to-real economic twins aligned with real observations. A systematic literature survey across these levels reveals that existing work remains concentrated in lower-level agent and simulation environments, while systems with self-evolving agents, endogenous institutions, persistent empirical alignment, and validated economic mechanisms remain rare. By translating the EWM agenda into an implementation blueprint, this paper aims to accelerate the development of the next generation of economic simulation environments that can serve as high-fidelity sandboxes for human decision-makers and as training, planning, evaluation, and safety substrates for AI agents. We release a curated paper list and related resources to support future research.
Chinese Translation
经济世界模型（Economic World Models, EWMs）是生成性经济模型，通过模拟异质代理、他们的信念和行为，以及市场和制度机制，从内部模拟经济如何演变。这篇论文制定了构建经济世界模型的实施路线图，作为生成引擎，其中异质代理在市场和制度中行动、互动、适应并共同演化，从而从内部产生经济动态。我们将EWM系统组织为六级能力阶梯，从固定规则基础的代理世界到自适应和基于大型语言模型（LLM）的代理世界、自我演化的代理、演化的制度世界，以及与真实观察相一致的模拟与现实经济双胞胎。对这些层级的系统文献调查显示，现有研究仍集中在较低层级的代理和模拟环境中，而具有自我演化代理、内生制度、持续的经验一致性和经过验证的经济机制的系统仍然稀缺。通过将EWM议程转化为实施蓝图，本文旨在加速下一代经济模拟环境的发展，这些环境可以作为高保真沙盒供人类决策者使用，并作为人工智能代理的训练、规划、评估和安全基础。我们发布了一份策划的论文列表和相关资源，以支持未来的研究。
cs.AI / 74 / 2608.06037
Integrating Implicit and Explicit Relational Biases through Graph-Based Multiple Instance Learning: A Case Study in Skin Lesion Diagnosis
通过基于图的多实例学习整合隐式和显式关系偏差：以皮肤病变诊断为例
Abstract
Relational inductive biases are essential for capturing structural dependencies among data. This study investigates a dual-level relational framework for image classification, bridging the gap between implicit representation learning and explicit structural modelling. We begin by establishing a baseline using an EfficientNetB3 architecture. To move beyond standard convolutional biases, we adopt a patch-based strategy, employing a convolutional masked autoencoder to learn implicit inter-patch relationships through self-supervised reconstruction. We then extend this approach by incorporating explicit relational modelling, organizing the learned embeddings into various graph topologies, including grid-based, random, and k-nearest neighbour structures. Experimental results on the ISIC-2018 and ISIC-2019 skin lesion diagnosis benchmarks show that combining implicit inter-patch modelling with explicit graph-based message passing yields the best performance. On the ISIC-2018 test set, the baseline model achieves a balanced accuracy of 76.17%, which improves to 77.12% with implicit patch-based relational modelling. The fully integrated grid-structured Graph Attention Network further increases performance to 79.27%. Similarly, on ISIC-2019, the implicit approach reaches 59.84% balanced accuracy, while the combination of implicit and explicit modelling yields 60.67%.
Chinese Translation
关系归纳偏差对于捕捉数据之间的结构依赖性至关重要。本研究探讨了一种双层关系框架用于图像分类，旨在弥合隐式表示学习与显式结构建模之间的差距。我们首先使用 EfficientNetB3 架构建立基线模型。为了超越标准卷积偏差，我们采用基于补丁的策略，利用卷积掩码自编码器通过自监督重建学习隐式的补丁间关系。随后，我们通过引入显式关系建模扩展这一方法，将学习到的嵌入组织成多种图拓扑结构，包括基于网格、随机和 k-近邻结构。在 ISIC-2018 和 ISIC-2019 皮肤病变诊断基准测试上的实验结果表明，将隐式补丁间建模与显式基于图的消息传递相结合能够获得最佳性能。在 ISIC-2018 测试集中，基线模型实现了 76.17% 的平衡准确率，而隐式补丁关系建模将其提升至 77.12%。完全整合的网格结构图注意力网络进一步将性能提高至 79.27%。类似地，在 ISIC-2019 中，隐式方法达到了 59.84% 的平衡准确率，而隐式与显式建模的结合则达到了 60.67%。
cs.AI / 75 / 2608.06057
When History Lies: Evaluating and Improving Tool Use under Misleading Multi-Turn Histories
当历史失真：在误导性的多轮历史下评估和改善工具使用
Abstract
Tool-calling agents infer task state from accumulated dialogue and tool traces. In persistent interactions, however, historical traces may remain structurally valid and semantically plausible after they cease to be authoritative for the current request. We show that such history can hijack a policy the model already possesses: on Qwen3-1.7B, pollution flips 32.1% of decisions that are correct under the original trajectory and frequently induces reuse of corrupted entities or interface conventions. We introduce bench, a paired benchmark with synchronized Original, Polluted, and Oracle State views that preserve the system policy, current tools, latest request, and gold next action. Eleven gold-preserving interventions isolate failures in decision state, entity binding, and interface execution across complete calls and non-call decisions. We further propose ours, which transfers an Oracle-conditioned teacher policy to a student observing only polluted history through soft supervision on student-generated prefixes. On Qwen3-1.7B, ours achieves 87.0% Balanced Tool-Use Accuracy, outperforming Gold-SFT (66.3%), Oracle sequence distillation (82.3%), and off-policy token distillation (85.0%). The method scales consistently: an 8B teacher raises the same compact 1.7B student to 91.9%, while an 8B student reaches 93.0%. The resulting policies further transfer to clean histories, unseen functions, independently regenerated evaluation contexts, external tool-use benchmarks, and noisy multi-hop question answering. These results establish history reliability as a distinct tool-use bottleneck and demonstrate reliable-state policy transfer as an effective and scalable solution.
Chinese Translation
工具调用代理从累积的对话和工具痕迹中推断任务状态。然而，在持续的交互中，历史痕迹在不再对当前请求具有权威性后，可能仍然在结构上有效且在语义上合理。我们展示了这样的历史如何劫持模型已经拥有的策略：在 Qwen3-1.7B 上，污染导致32.1%的决策在原始轨迹下是正确的，并且经常诱导对损坏实体或接口约定的重复使用。我们引入了 bench，这是一个配对基准，具有同步的原始、污染和oracle状态视图，保留系统策略、当前工具、最新请求和黄金下一个动作。十一种保留黄金的干预措施在完整调用和非调用决策中隔离决策状态、实体绑定和接口执行的失败。我们进一步提出了 ours，它通过对学生生成的前缀进行软监督，将一个以oracle为条件的教师策略转移给仅观察污染历史的学生。在 Qwen3-1.7B 上，ours 达到了87.0%的平衡工具使用准确率，优于黄金微调（Gold-SFT，66.3%）、oracle序列蒸馏（82.3%）和离策略标记蒸馏（85.0%）。该方法具有一致的扩展性：一个8B的教师将同样紧凑的1.7B学生提升至91.9%，而一个8B的学生则达到93.0%。最终生成的策略进一步转移到干净的历史、未见的功能、独立再生的评估上下文、外部工具使用基准和嘈杂的多跳问答。这些结果确立了历史可靠性作为一个独特的工具使用瓶颈，并展示了可靠状态策略转移作为一种有效且可扩展的解决方案。
cs.AI / 76 / 2608.06085
Signal or Spurious Cue? A Randomized Audit of Survey-Country Metadata in LLM Social Inference
信号还是虚假线索？对LLM社会推理中调查国家元数据的随机审计
Abstract
Survey-country metadata can improve an LLM's forecast of an individual response when informative, yet the same cue may redirect the forecast when assigned at random. A within-record audit tests whether disclosing a random label's uniform, record-independent origin reduces its country-directed uptake, and whether verified survey country lowers held-out Brier loss. Independent population anchors and recorded human answers measure direction and consequence across five fixed API models, six countries, and seven development-selected targets. In the primary post-review 72-record panel, opaque and disclosed-random labels each produced country-direction shifts of 0.214. Paired attenuation was 0.0003 (95% CI [-0.0157, 0.0166]). Verified country reduced Brier loss by 0.040 (95% CI [0.024, 0.056]), while random-label regret included zero. A non-overlapping mixed-coverage consistency panel retained positive disclosed-random movement and verified utility, while attenuation remained uncertain. On the selected targets, verified metadata was useful in both panels, but disclosure did not reliably attenuate random-label uptake. PROV-FORECAST contains 14,400 paired item-level probability distributions from the corrected panel.
Chinese Translation
调查国家元数据在信息丰富时可以改善大型语言模型（LLM）对个体反应的预测，但当随机分配时，同样的线索可能会转移预测方向。本文通过记录内审计测试了披露随机标签的统一、记录独立来源是否减少其对国家的影响，以及验证的调查国家是否降低了保留的Brier损失。独立的人口锚和记录的人类答案在五个固定API模型、六个国家和七个开发选择的目标中测量了方向和结果。在主要的后审查72记录面板中，不透明和披露的随机标签各自产生了0.214的国家方向转变。配对衰减为0.0003（95% CI [-0.0157, 0.0166]）。验证的国家将Brier损失降低了0.040（95% CI [0.024, 0.056]），而随机标签的遗憾包括零。在非重叠的混合覆盖一致性面板中，保持了积极的披露随机移动和验证效用，但衰减仍不确定。在选定的目标上，验证的元数据在两个面板中都很有用，但披露并未可靠地减弱随机标签的影响。PROV-FORECAST包含来自修正面板的14,400对项目级概率分布。
cs.AI / 77 / 2608.06108
Evaluating Investment Logic in Large Language Models: A Real-World Benchmark Towards Personalzied Financial Agents
评估大型语言模型中的投资逻辑：面向个性化金融代理的现实基准
Abstract
Investment competence is inherently personalized: the same market evidence can justify different actions for investors with different goals, horizons, portfolios, and risk boundaries. Yet financial LLMs are evaluated either by static question answering or by terminal profit and loss. The former omits agency; the latter cannot reveal whether a profitable action was grounded, profile-consistent, or merely lucky. We ask whether the community is using the wrong ruler for consequential agents. We introduce \textsc{InvestLogicBench}, a process-native benchmark containing 201,247 documented decisions from 151 real-world investors. Each episode instantiates a \textbf{P$\rightarrow$E$\rightarrow$R$\rightarrow$D$\rightarrow$O} trace: investor \textit{Profile}, observable market \textit{Events}, investment \textit{Reasoning}, executable \textit{Decision}, and delayed \textit{Outcome}. The release includes profile construction, point-in-time event binding, structured logic, horizons, outcomes, and post-mortems, and supports comprehension, profile-conditioned generation, and end-to-end replay. Across four leading LLMs, logical plausibility remains near 4/5 while event grounding is only 0.8--2.8/5; return and process quality also disagree. These results expose polished but weakly grounded reasoning that outcome-only evaluation hides. We further argue that P$\rightarrow$E$\rightarrow$R$\rightarrow$D$\rightarrow$O should be a data-system interface, requiring versioned profiles, temporal provenance, inspectable retrieval, decision ledgers, and replayable outcomes. Finance is our stress test for a broader class of personalized, consequential agents.
Chinese Translation
投资能力本质上是个性化的：相同的市场证据可以为目标、时间范围、投资组合和风险边界不同的投资者提供不同的行动依据。然而，金融大型语言模型（LLMs）的评估通常仅依赖静态问答或终端盈亏。前者忽略了代理性；后者无法揭示一个盈利行为是否基于合理的基础、与投资者的特征一致，或仅仅是运气使然。我们质疑社区是否在为具有重要后果的代理使用错误的评估标准。为此，我们引入了 extsc{InvestLogicBench}，这是一个过程原生基准，包含来自151位现实投资者的201,247个记录决策。每个案例都实例化了一个 extbf{P$
ightarrow$E$
ightarrow$R$
ightarrow$D$
ightarrow$O} 追踪：投资者的 extit{Profile}、可观察的市场 extit{Events}、投资 extit{Reasoning}、可执行的 extit{Decision} 和延迟的 extit{Outcome}。该发布包括特征构建、特定时间事件绑定、结构化逻辑、时间范围、结果和事后分析，并支持理解、特征条件生成和端到端重放。在四个领先的LLMs中，逻辑合理性保持在接近4/5，而事件基础仅为0.8--2.8/5；收益和过程质量也存在不一致。这些结果揭示了经过打磨但基础薄弱的推理，而仅依赖结果的评估则掩盖了这一点。我们进一步认为，P$
ightarrow$E$
ightarrow$R$
ightarrow$D$
ightarrow$O 应该成为数据系统接口，要求具备版本化特征、时间来源、可检查的检索、决策账本和可重放的结果。金融是我们对更广泛个性化、具有重要后果的代理的压力测试。
cs.AI / 78 / 2608.06110
ECHO: A Locally-Deployable Agentic Health Assistant with Temporal Memory, Safety Guardrails, and Speech Assessment
ECHO：具有时间记忆、安全保护措施和语音评估的本地可部署智能健康助手
Abstract
This paper presents ECHO (Enhanced Care \& Health Observer), a locally-deployable conversational health assistant for long-term chronic care management. ECHO integrates three complementary software modules developed under shared supervision as a unified system. The core module is an agentic chatbot built on a ReAct loop orchestrated via LangGraph, equipped with 17 clinical tools and a temporal knowledge graph for persistent cross-session memory; it achieves a 94.9\% tool-execution pass rate across a 59-scenario benchmark with GPT-5 Mini. A two-stage hybrid safety layer intercepts all incoming queries: a rule-based layer handles explicit crisis signals and jailbreak attempts in under 1ms, while a signed graph neural network (GNN) with APPNP-style propagation classifies boundary cases by clinical intent, achieving 88.8\% accuracy and 90.6\% unsafe recall on a 2,537-query annotated Turkish health dataset while outperforming zero-shot LLM baselines including Llama 3.3 70B. A multimodal speech assessment module combining Whisper acoustic encoding and BERT text encoding with cross-attention fusion estimates emotion, depression, and pain, reaching a mean macro F1 of 0.652. The full system is implemented as a web application that can run entirely on consumer hardware, with no patient data transmitted to external services, supporting compliance with GDPR and KVKK.
Chinese Translation
本文介绍了ECHO（增强护理与健康观察者），一种用于长期慢性病管理的本地可部署对话式健康助手。ECHO整合了在共享监督下开发的三个互补软件模块，形成一个统一系统。核心模块是基于ReAct循环构建的智能聊天机器人，通过LangGraph进行协调，配备17个临床工具和一个用于持久跨会话记忆的时间知识图谱；在与GPT-5 Mini的59个场景基准测试中，其工具执行通过率达到94.9%。一个两阶段混合安全层拦截所有传入查询：基于规则的层在1毫秒内处理明确的危机信号和越狱尝试，而带有APPNP风格传播的签名图神经网络（GNN）通过临床意图对边界案例进行分类，准确率达到88.8%，在2,537个注释的土耳其健康数据集上不安全召回率为90.6%，并超越了包括Llama 3.3 70B在内的零-shot LLM基线。一个多模态语音评估模块结合了Whisper声学编码和BERT文本编码，通过交叉注意力融合来估计情感、抑郁和疼痛，达到平均宏F1值0.652。整个系统作为一个Web应用程序实现，能够完全在消费者硬件上运行，且不向外部服务传输患者数据，支持遵守GDPR和KVKK的要求。
cs.AI / 79 / 2608.06112
From Siloed Algorithms to Compliance-First Agentic Platforms: A Multi-Layered Architecture for Hospital AI Systems
从孤立算法到合规优先的自主平台：医院人工智能系统的多层架构
Abstract
Hospitals are rapidly adopting artificial intelligence for triage, imaging, scheduling etc., yet most deployments remain isolated point solutions locked inside departmental silos, resulting in duplicated effort, hidden risks, and unrealized enterprise value. Despite explosive growth of AI in healthcare market and accelerating investment, an estimated 70-80% of healthcare AI pilots fail to scale, largely due to governance gaps, fragmented data, and missing integration blueprints. This research proposes a hospital-specific, compliance-first, Agentic AI architecture with multiple interoperable layers, extending existing hospital AI platform models with: (i) an Agent Orchestration Layer for multi-agent workflows across clinical, operational, and financial domains, (ii) a Compliance and Policy Layer that centralizes policy-as-code for HIPAA, GDPR, the EU AI Act, DISHA Act, India's DPDP Act, and ISO/IEC security and safety standards, and (iii) a Privacy-Preserving Data Fabric that plugs federated learning, differential privacy, and secure enclaves into real-world Hospital Information Management System (HIMS) flows. Using a synthetic but structurally realistic hospital dataset and an open, ready-to-deploy prototype implementation, this study demonstrates the end-to-end orchestration of triage risk prediction, workflow optimization, and compliance logging, achieving substantial simulated reductions in task turnaround times and manual documentation effort while maintaining policy-guarded data access. The resulting architecture offers hospital leaders a pragmatic blueprint to move from ad hoc tools to a governed, globally compliant, ROI-focused AI platform that can be tailored to on-premise, hybrid and cloud-native deployments.
Chinese Translation
医院正在迅速采用人工智能进行分诊、影像学、排班等工作，但大多数部署仍然是被锁定在部门孤岛中的孤立点解决方案，导致重复工作、隐藏风险和未实现的企业价值。尽管医疗市场中的人工智能迅速增长，投资加速，但估计有70-80%的医疗人工智能试点项目未能扩展，主要是由于治理缺口、数据碎片化和缺乏集成蓝图。本研究提出了一种医院特定的、合规优先的自主人工智能架构，具有多个可互操作的层级，扩展了现有医院人工智能平台模型，包括：(i) 一个用于临床、运营和财务领域多智能体工作流的智能体协调层；(ii) 一个集中管理HIPAA、GDPR、欧盟人工智能法案、DISHA法案、印度DPDP法案以及ISO/IEC安全和安全标准的合规与政策层；(iii) 一个隐私保护的数据结构，将联邦学习、差分隐私和安全区块集成到实际的医院信息管理系统（HIMS）流程中。通过使用一个合成但结构上真实的医院数据集和一个开放的、可立即部署的原型实现，本研究展示了分诊风险预测、工作流优化和合规日志记录的端到端协调，实现了任务周转时间和人工文档工作量的显著模拟减少，同时保持了政策保护的数据访问。最终形成的架构为医院领导提供了一个务实的蓝图，以便从临时工具转向一个受治理的、全球合规的、以投资回报为重点的人工智能平台，能够根据本地、混合和云原生部署进行定制。
cs.AI / 80 / 2608.06115
Mind the Gaps: Mixture-of-Minds for Human Simulation
关注差距：人类模拟的思维混合模型
Abstract
Predicting how a population will answer a new question is a long-standing goal. Statistical methods succeed at the level of the mass but falter at the level of the individual. Large language model simulators inherit this gap. They recover a population's central tendencies while flattening its heterogeneity, and they carry social biases and prompt brittleness that distort individual predictions. This paper introduces Anacreon, an audience simulation model that targets the individual level within a narrow, well-specified domain. Anacreon learns an authorship embedding that separates individuals, clusters a real qualitative corpus around seed people, and trains a dedicated adapter for each cluster, a mixture of minds, on a Gemma~4 12B base. It harvests demographics, psychological traits, and survey responses from public text, and augments each record with a chain-of-emotion. It reduces prompt brittleness by shuffling response options and reduces positive bias by balancing the training distribution. On a large, externally sourced survey, Anacreon reaches a state-of-the-art ordinal alignment of 0.775, the individual-level accuracy measure on which the field has converged, with a small residual bias. The work is a step toward drawing aggregate insight from faithfully simulated individuals.
Chinese Translation
预测一个群体如何回答一个新问题是一个长期以来的目标。统计方法在整体层面上取得成功，但在个体层面上却表现不佳。大型语言模型模拟器继承了这一差距。它们恢复了群体的中心倾向，却扁平化了其异质性，并携带社会偏见和提示脆弱性，从而扭曲了个体预测。本文介绍了Anacreon，一个针对特定狭窄领域内个体层面的受众模拟模型。Anacreon学习一种作者嵌入，区分个体，将真实的定性语料围绕种子人物进行聚类，并为每个聚类训练一个专用适配器，即思维的混合，基于Gemma~4 12B模型。它从公共文本中收集人口统计信息、心理特征和调查响应，并用情感链增强每条记录。通过打乱响应选项，它减少了提示脆弱性，并通过平衡训练分布来降低积极偏见。在一个大型外部来源的调查中，Anacreon达到了0.775的最先进序数对齐，这是该领域所趋向的个体层面准确性度量，且残余偏见较小。这项工作是从真实模拟的个体中提取聚合洞察的一个重要步骤。
cs.AI / 81 / 2608.06123
Poli-Bias: Understanding and Measuring Large Language Model Biases in International Political Conflicts
Poli-Bias：理解和测量国际政治冲突中大型语言模型的偏见
Abstract
Measuring political bias in large language models (LLMs) remains challenging as it can manifest through subtle differences in framing, argumentation, and legal reasoning that are difficult to capture with a single metric. In this work, we introduce Poli-Bias, a counterfactual framework for measuring whether LLMs treat legally equivalent conflict scenarios differently depending on the countries involved. Poli-Bias compares responses to paired prompts in which country identities are systematically swapped across diverse geopolitical relationships, legal violations, and reasoning tasks. Rather than reducing bias to a single judgment, our framework decomposes response disparities into five interpretable dimensions, revealing how and where unequal treatment manifests. Across 13 contemporary LLMs spanning diverse model families and sizes, we find that country identities and user affiliations can systematically affect how equivalent actions are described, evaluated, and defended under international law. Our results thus establish Poli-Bias as a fine-grained framework for auditing political even-handedness and sycophancy in LLMs.
Chinese Translation
在大型语言模型（LLMs）中测量政治偏见仍然具有挑战性，因为这种偏见可能通过框架、论证和法律推理中的微妙差异表现出来，这些差异难以用单一指标捕捉。在本研究中，我们引入了Poli-Bias，一个用于测量LLMs是否根据涉及的国家以不同方式对待法律上等同的冲突场景的反事实框架。Poli-Bias比较了在不同地缘政治关系、法律违规和推理任务中，国家身份系统性交换的成对提示的响应。我们的框架并不是将偏见简化为单一判断，而是将响应差异分解为五个可解释的维度，揭示了不平等待遇的表现方式和位置。在涵盖多种模型家族和规模的13个当代LLMs中，我们发现国家身份和用户隶属关系可以系统性地影响如何描述、评估和辩护等价行为在国际法下的表现。因此，我们的结果确立了Poli-Bias作为审计LLMs政治公正性和谄媚行为的细致框架。
cs.AI / 82 / 2608.06128
Contextual Information Policy Optimization for Search Agents
搜索代理的上下文信息策略优化
Abstract
Search agents extend large language models beyond static parametric memory by enabling them to acquire and use ex ternal evidence during multi-step reasoning. For knowledge intensive tasks involving complex or evolving information, their reliability depends not only on retrieving relevant ev idence but also on using it to guide subsequent reasoning. However, existing methods primarily reward final-answer cor rectness or intermediate progress, without directly assessing whether post-retrieval actions are grounded in the retrieved evidence. This misalignment encourages prior-driven reason ing: agents form conclusions based on internal knowledge and use retrieval mainly to confirm them, resulting in confirma tion bias and inefficient evidenceuse.Toaddressthisissue, we propose Contextual Information Policy Optimization (CIPO), an evidence-oriented reinforcement learning framework that explicitly aligns policy optimization with external evidence use. CIPO assigns dense, turn-level credit to reasoning ac tions influenced by retrieved information, while combining this evidence-use signal with a global outcome reward to pre serveanswercorrectness.Withthismanner,CIPOdiscourages evidence-detached guesses and promotes reasoning trajecto ries in which retrieved facts can guide or revise subsequent reasoning. Importantly, CIPO requires neither human process annotations nor an additional reward model. Extensive exper iments on seven in-domain and out-of-domain benchmarks show that CIPO reduces the prevalence of prior-driven rea soning and achieves excellent performance on most tasks.
Chinese Translation
搜索代理通过使大型语言模型能够在多步骤推理过程中获取和使用外部证据，扩展了其静态参数记忆的能力。对于涉及复杂或不断变化信息的知识密集型任务，其可靠性不仅依赖于检索相关证据，还依赖于利用这些证据来指导后续推理。然而，现有方法主要奖励最终答案的正确性或中间进展，而没有直接评估后检索行为是否基于所检索的证据。这种不一致性促使了以先验知识驱动的推理：代理基于内部知识形成结论，并主要利用检索来确认这些结论，导致确认偏差和低效的证据使用。为了解决这个问题，我们提出了上下文信息策略优化（Contextual Information Policy Optimization, CIPO），这是一种以证据为导向的强化学习框架，明确将策略优化与外部证据的使用对齐。CIPO为受检索信息影响的推理行为分配密集的回报，同时将这种证据使用信号与全局结果奖励相结合，以保持答案的正确性。通过这种方式，CIPO抑制了与证据无关的猜测，并促进了检索事实可以指导或修正后续推理的推理轨迹。重要的是，CIPO既不需要人工过程注释，也不需要额外的奖励模型。在七个领域内和领域外的基准测试中进行的广泛实验表明，CIPO减少了以先验知识驱动的推理的普遍性，并在大多数任务上实现了优异的表现。
cs.AI / 83 / 2608.06144
FinEvo-Bench: A Longitudinal Benchmark for Self-Evolving Agents in Professional Financial Workflows
FinEvo-Bench：专业金融工作流程中自我进化代理的纵向基准
Abstract
Most agent benchmarks evaluate tasks independently and cannot measure whether experience from one task helps with later tasks. Existing self-evolution benchmarks do not jointly cover professional workflows, open-ended deliverables, and multi-aspect evaluation. We introduce FinEvo-Bench, a longitudinal benchmark with 120 real-case-grounded tasks, 20 business scenes across six financial domains. Institution-provided professional procedures define the required operations and constraints. Eligible institution-provided and publicly documented cases supply the task facts. Each scene contains six related but substantively distinct cases that share a professional procedure and a manually reviewed rubric for task quality and financial compliance. We compare four self-evolving agent scaffolds using the same Qwen3.7-Max backbone and three independently shuffled, globally interleaved task streams. Paired non-evolving controls estimate each scaffold's self-evolution gain from retained experience, while an independent Claude Code scoring agent backed by Claude Opus 4.6 evaluates all outputs. Letta achieves the highest evolved score (91.65) and fewest compliance issues (0.09 per task); Codex achieves the largest self-evolution gain (+19.37). Across scaffolds, the evolving condition raises scores by 9.33-19.37 points and reduces compliance issues by 0.12-0.44 per task. Paired score gains at within-scene ranks 4-6 exceed those at ranks 1-3 by 6.10-8.70 points. In Claude Code, skill-only evolution produces higher task quality and fewer compliance issues than memory-only and combined memory-skill evolution. Across all four scaffolds, rubric feedback also yields higher scores and fewer compliance issues than reference-answer feedback. FinEvo-Bench measures both professional performance and self-evolution ability: how effectively an agent turns prior experience into later improvement.
Chinese Translation
大多数代理基准独立评估任务，无法衡量一个任务的经验是否对后续任务有所帮助。现有的自我进化基准未能共同覆盖专业工作流程、开放式交付物和多方面评估。我们引入了FinEvo-Bench，这是一个纵向基准，包含120个基于真实案例的任务，涵盖六个金融领域的20个商业场景。机构提供的专业程序定义了所需的操作和约束条件。符合条件的机构提供和公开记录的案例提供了任务事实。每个场景包含六个相关但实质上不同的案例，这些案例共享一个专业程序和一个经过人工审核的任务质量及财务合规性评分标准。我们使用相同的Qwen3.7-Max骨干网络和三个独立洗牌、全局交错的任务流比较四个自我进化代理框架。配对的非进化控制估计每个框架从保留经验中获得的自我进化收益，同时由Claude Opus 4.6支持的独立Claude Code评分代理评估所有输出。Letta获得了最高的进化得分（91.65）和最少的合规问题（每个任务0.09）；Codex获得了最大的自我进化收益（+19.37）。在所有框架中，进化条件使得得分提高了9.33-19.37分，并减少了每个任务的合规问题0.12-0.44。在Claude Code中，仅技能的进化产生的任务质量高于仅记忆和结合记忆-技能进化的任务质量，且合规问题更少。在所有四个框架中，评分标准反馈也比参考答案反馈产生更高的得分和更少的合规问题。FinEvo-Bench同时衡量专业表现和自我进化能力：代理如何有效地将先前的经验转化为后续的改进。
cs.AI / 84 / 2608.06146
PaDoc: Layout-Grounded Parallel Decoding for Document Parsing
PaDoc：基于布局的文档解析并行解码
Abstract
End-to-end document parsers provide a unified interface, but serialize page layouts and regional contents into one autoregressive sequence. This formulation forces independent regions onto a decoding path whose length grows with the total content, whereas crop-based two-stage parsers expose region-level parallelism at the cost of repeated visual prefills and fragmented page context. To retain full-page context while removing dependencies, we propose PaDoc, a layout-grounded parser that treats the predicted layout as a branching structure over a shared page representation. Under a region-sufficiency assumption, we derive a prefix-conditioned factorization in which the layout stream and regional content branches advance concurrently, reducing the decoding depth to the longest layout-content path. We realize this factorization within a single MLLM: packed variable-length ancestor attention preserves the visibility under standard next-token training, while masked parallel decoding creates branches that the evaluated vLLM backend serves as concurrent requests with cache-resident shared-prefix reuse. On OmniDocBench Full, PaDoc attains an Overall layout F1 of 91.1 and, among end-to-end parsers, a top-tier Overall score of 94.24 together with the best Text Edit (0.038) and Formula CDM (95.59). On a 384-page subset and one A800 GPU, it is the fastest end-to-end parser at five concurrency levels, improving valid-page throughput by 67.4-118% and reducing P95 latency by 39.2-54.9% relative to a same-backbone Sequential SFT baseline. Code is available at https://github.com/Longin-Yu/Padoc
Chinese Translation
端到端文档解析器提供了统一的接口，但将页面布局和区域内容序列化为一个自回归序列。这种形式迫使独立区域沿着一个解码路径进行解码，该路径的长度随着总内容的增加而增长，而基于裁剪的两阶段解析器则以重复的视觉预填充和碎片化的页面上下文为代价，暴露了区域级的并行性。为了保留完整的页面上下文并消除依赖关系，我们提出了PaDoc，一种基于布局的解析器，将预测的布局视为共享页面表示上的分支结构。在区域充足性假设下，我们推导出一种前缀条件因式分解，其中布局流和区域内容分支并行推进，从而将解码深度减少到最长的布局-内容路径。我们在单个多模态大语言模型（MLLM）中实现了这种因式分解：打包的可变长度祖先注意力在标准的下一个标记训练下保持了可见性，而掩蔽的并行解码创建了分支，经过评估的可变长度语言模型（vLLM）后端作为并发请求提供服务，并重用缓存中的共享前缀。在OmniDocBench Full上，PaDoc达到了91.1的整体布局F1，并在端到端解析器中获得了94.24的顶级整体得分，以及最佳的文本编辑（0.038）和公式CDM（95.59）。在一个384页的子集和一台A800 GPU上，它是五个并发级别中速度最快的端到端解析器，相较于同骨干的顺序SFT基线，提高了有效页面吞吐量67.4%-118%，并将P95延迟降低了39.2%-54.9%。代码可在https://github.com/Longin-Yu/Padoc获取。
cs.AI / 85 / 2608.06150
CogVis: Must Open-Vocabulary Change Detection Perceive the Scene Anew for Every Query?
CogVis：开放词汇变化检测是否需要为每个查询重新感知场景？
Abstract
Earth-surface monitoring requires change detection models capable of recognizing arbitrary semantic categories. Open-Vocabulary Change Detection (OVCD) addresses this need. However, existing methods often entangle temporal perception, semantic discrimination, and region verification, causing unstable results and redundant computation. Inspired by human visual change perception, we propose CogVis, a cognitive memory-guided framework that reformulates OVCD as a perception-memory-verification paradigm. CogVis first employs a Scene Change Perceptron (SCP) to extract a reusable, category-agnostic change prior from frozen bi-temporal features, thereby decoupling temporal evidence from semantic category decisions. A Semantic Memory Calibrator (SMC) then compensates for category-dependent score shifts by dynamically estimating an image-query-specific decision threshold. Finally, an Adaptive Region Filter (ARF) filters connected candidates using learned semantic, temporal, and structural reliability. Experiments on seven benchmarks spanning semantic change detection, binary change localization, and building-damage assessment show that CogVis achieves state-of-the-art performance across all evaluated datasets. By sharing scene-level change perception, CogVis further avoids repeating category-agnostic temporal perception across queries and improves inference throughput by 28.50%.
Chinese Translation
地表监测需要能够识别任意语义类别的变化检测模型。开放词汇变化检测（Open-Vocabulary Change Detection, OVCD）正是为满足这一需求而提出。然而，现有方法往往将时间感知、语义区分和区域验证交织在一起，导致结果不稳定和冗余计算。受到人类视觉变化感知的启发，我们提出了CogVis，一个认知记忆引导的框架，将OVCD重新构建为感知-记忆-验证范式。CogVis首先采用场景变化感知器（Scene Change Perceptron, SCP）从冻结的双时间特征中提取可重用的、类别无关的变化先验，从而将时间证据与语义类别决策解耦。接着，语义记忆校准器（Semantic Memory Calibrator, SMC）通过动态估计图像查询特定的决策阈值来补偿依赖类别的得分偏移。最后，自适应区域过滤器（Adaptive Region Filter, ARF）利用学习到的语义、时间和结构可靠性过滤连接的候选区域。在涵盖语义变化检测、二元变化定位和建筑损坏评估的七个基准测试中的实验表明，CogVis在所有评估的数据集上均实现了最先进的性能。通过共享场景级变化感知，CogVis进一步避免了在查询之间重复类别无关的时间感知，并将推理吞吐量提高了28.50%。
cs.AI / 86 / 2608.06161
iARCS: Iterative Agentic RL for Controllable 3D Scene Generation
iARCS：用于可控3D场景生成的迭代智能强化学习
Abstract
Synthetic 3D scene generation is increasingly used as a data source for computer vision and embodied AI, but existing generators often optimize perceptual realism without reliably satisfying task-critical functional constraints. This mismatch limits the usefulness of synthetic data for downstream training, where accessibility, traversability, and spatial rule compliance are often essential. We present iARCS, an iterative agentic reinforcement learning framework that adapts a pretrained scene generator to natural-language task requirements. iARCS uses a two-stage strategy: universal-reward pretraining to improve physical plausibility and layout quality, followed by task-specific fine-tuning with LLM-generated reward programs that are iteratively refined from training feedback. Experiments show improved constraint fidelity on walkability, reachability, and clearance-focused tasks, effective task-specific constraint optimization, and competitive scene diversity. We further show that data generated by iARCS improves a base generator, supporting its value as a practical synthetic data generation tool rather than only a controllable scene editing method.
Chinese Translation
合成3D场景生成越来越多地被用作计算机视觉和具身人工智能的数据源，但现有的生成器通常优化感知现实性，而未能可靠地满足任务关键的功能约束。这种不匹配限制了合成数据在下游训练中的有效性，其中可达性、可行性和空间规则合规性通常是至关重要的。我们提出了iARCS，一个迭代智能强化学习框架，旨在将预训练的场景生成器适应于自然语言任务需求。iARCS采用两阶段策略：首先进行通用奖励预训练，以提高物理合理性和布局质量；然后通过使用从训练反馈中迭代精炼的LLM生成的奖励程序进行任务特定的微调。实验表明，在可步行性、可达性和间隙关注任务上，约束保真度得到了改善，任务特定的约束优化有效，场景多样性具有竞争力。我们进一步展示了iARCS生成的数据改善了基础生成器，支持其作为一种实用的合成数据生成工具的价值，而不仅仅是可控场景编辑方法。
cs.AI / 87 / 2608.06167
Schema-Guided Hierarchical Information Extraction and Semantic Evaluation Using Generative AI
基于模式的分层信息提取与生成性人工智能的语义评估
Abstract
We present a schema-based framework for extracting complex, structured information from unstructured text documents using generative AI, followed by automated semantic evaluation of the extracted information against a gold standard. The schema, serving as an information model encoding domain knowledge, provides a unified, systematic, and consistent framework for extraction of hierarchical, nested information, with attributes of variable cardinality, and subsequent evaluation of the results. Information extraction from a document is performed in a single call to the model, in zero-shot mode. In the evaluation step, we introduce a path-based semantic matching algorithm to align the nested, variable-cardinality attributes in the extracted results with those in the gold standard. We use generative AI for semantic comparison of the extracted and gold standard values of an attribute, and introduce a rubric to classify the result of the comparison, according to domain-specific considerations, as an exact, semantic, useful, or non-match. We were able to extract 12 out of 14 attributes with an F1 score of $>$90\% from documents published by the health technology assessment organisation NICE, using the generative AI model Claude Opus 3. The time needed to extract the attributes from a document was $\sim$30 times lower than the time taken by a human domain expert. We further demonstrate generalisability of this framework across different generative AI models and transferability across different HTA organisations and languages.
Chinese Translation
我们提出了一种基于模式的框架，利用生成性人工智能从非结构化文本文档中提取复杂的结构化信息，并对提取的信息进行自动化的语义评估，以与标准答案进行对比。该模式作为信息模型，编码了领域知识，为提取具有可变基数属性的分层嵌套信息提供了统一、系统和一致的框架，并对结果进行后续评估。信息提取通过对模型的单次调用在零样本模式下完成。在评估步骤中，我们引入了一种基于路径的语义匹配算法，将提取结果中的嵌套可变基数属性与标准答案中的属性进行对齐。我们使用生成性人工智能对提取值与标准答案值的语义进行比较，并根据特定领域的考虑引入一个评分标准，将比较结果分类为精确匹配、语义匹配、有用匹配或不匹配。我们能够从健康技术评估组织NICE发布的文档中提取14个属性中的12个，F1得分超过90%。从文档中提取属性所需的时间约为人类领域专家所需时间的30倍。我们进一步展示了该框架在不同生成性人工智能模型之间的通用性，以及在不同健康技术评估组织和语言之间的可转移性。
cs.AI / 88 / 2608.06183
MicroEvo: Knowledge-Guided LLM Sampling for Efficient Microarchitecture Design Space Exploration
MicroEvo：基于知识引导的LLM采样用于高效微架构设计空间探索
Abstract
Microarchitecture design space exploration suffers from expansive search spaces and expensive PPA evaluation, leaving only a small simulation budget for design decision-making. Existing methods perform blind search without considering microarchitectural dependencies and fail to learn from the iterative search effectively, leading to wasted evaluations and weak Pareto convergence. In this paper, we propose MicroEvo, a knowledge-guided framework that couples off-the-shelf LLMs with Monte Carlo Tree Search (MCTS) for multi-objective microarchitecture optimization. MicroEvo combines LLM-driven evolutionary operators, a Pareto-aware tree policy that balances Pareto contribution and diversity, an active knowledge accumulation mechanism that extracts and reuses optimization insights, and state-aware directives that adapt the search behavior online. Experiments show that MicroEvo improves Pareto-front quality by up to 36.2% over NSGA-II and achieves 10.6x higher search efficiency, and also demonstrates strong scalability to a complex industrial-scale core. The code repository is available at: https://github.com/GEAR-SEU/MicroEvo-ICCAD-26.
Chinese Translation
微架构设计空间探索面临着庞大的搜索空间和昂贵的PPA评估，仅留有少量的仿真预算用于设计决策。现有方法在盲目搜索中未考虑微架构依赖关系，且未能有效从迭代搜索中学习，导致评估浪费和较弱的Pareto收敛。在本文中，我们提出了MicroEvo，一个知识引导的框架，将现成的LLM与蒙特卡罗树搜索（Monte Carlo Tree Search, MCTS）结合，用于多目标微架构优化。MicroEvo结合了LLM驱动的进化算子、一个平衡Pareto贡献和多样性的Pareto感知树策略、一个提取和重用优化洞察的主动知识积累机制，以及在线适应搜索行为的状态感知指令。实验表明，MicroEvo在Pareto前沿质量上比NSGA-II提高了最高36.2%，并实现了10.6倍的搜索效率提升，同时在复杂的工业规模核心上也展现了强大的可扩展性。代码库可在以下地址获取：https://github.com/GEAR-SEU/MicroEvo-ICCAD-26。
cs.AI / 89 / 2608.06196
Comparative Approaches to Agent Retrieval over Large Skill Libraries
大规模技能库中代理检索的比较方法
Abstract
Agents backed by large skill libraries must decide which skills to load and in what order. Loading the entire library into context is expensive and provides no structure for autonomous sequencing. We study two systems for this problem over a corpus of 690 skills: a hybrid ranker combining lexical and dense-embedding retrieval for sparse, on-demand loading, and a typed knowledge graph encoding workflow relations such as prerequisites, data flow, and ordering. On a set of 117 realistic, non-echoing queries, the hybrid ranker retrieves the correct skill within the top five in 73.5% +/- 8.0 of cases, leaving roughly a quarter of queries unserved. When used as the design intended (substituting graph neighbours for additional ranked results at matched token budget), the graph is significantly worse (-11.2 points, p = 0.0007). Its LLM-generated edge layer adds nothing over neighbours obtained free from a local embedding pass, and 73% of the queries the ranker misses are not reachable through the graph at all. We attribute this to a pre-filter topology bound. Because the graph's candidate edges are drawn from the same embedding neighbourhood the ranker already searches, 98.6% of typed edges connect skills the ranker had already surfaced together. The graph can enrich relation semantics but cannot extend retrieval reach. We further show that evaluating on author-written queries overstates hit@5 by up to 44 points, which would have hidden these results entirely. Our contribution is a mechanistic account of why added structure does not improve retrieval over a strong ranker, and identify the conditions under which adding structural interdependence into the retrieval is optimal.
Chinese Translation
依托于大型技能库的代理必须决定加载哪些技能以及加载的顺序。将整个库加载到上下文中是昂贵的，并且无法为自主排序提供结构。我们研究了针对690个技能的两个系统：一种混合排序器，结合了词汇检索和稠密嵌入检索，以实现稀疏的按需加载；另一种是编码工作流关系（如前提条件、数据流和排序）的类型知识图谱。在117个现实的非重复查询集上，混合排序器在73.5% +/- 8.0的情况下能够在前五名中检索到正确的技能，约有四分之一的查询未得到满足。当按照设计意图使用（用图邻居替代匹配令牌预算下的额外排序结果）时，图的表现显著更差（-11.2分，p = 0.0007）。其LLM生成的边层在从本地嵌入传递中获得的邻居上没有任何增益，而排序器错过的73%的查询根本无法通过图访问。我们将此归因于预过滤拓扑的限制。由于图的候选边来自排序器已经搜索的相同嵌入邻域，98.6%的类型边连接了排序器已经一起呈现的技能。图可以丰富关系语义，但无法扩展检索范围。我们进一步表明，在作者编写的查询上进行评估会将hit@5高估多达44分，这将完全掩盖这些结果。我们的贡献在于提供了一个机制性解释，说明为何增加结构并未改善强排序器的检索效果，并识别出在何种条件下将结构性相互依赖性引入检索是最优的。
cs.AI / 90 / 2608.06197
EnvACE: Internalizing Environment Dynamics via World Rehearsal for Agentic Reinforcement Learning
EnvACE：通过世界排练内化环境动态以实现自主强化学习
Abstract
Training large language model agents for long-horizon tool use typically relies on interactions with real or synthesized executable environments, whose construction and verification are costly, or on external simulators that are difficult to ground. We introduce EnvACE, an agentic reinforcement learning method that replaces external environment interaction during training with world rehearsal. The policy alternates between acting and rehearsal: it first generates a tool call, then plays the role of the environment to produce the response induced by that action, and conditions subsequent decisions on the rehearsed response. Both roles are jointly optimized end-to-end using task-success rewards. Through world rehearsal, the policy internalizes the relationship between actions and their environment responses in its parameters, yielding an agent world model that directly supports decision making. Across BFCL-v4, tau^2-Bench, VitaBench, and FinMCP-Bench, EnvACE achieves strong and transferable performance, outperforming environment-scaling baselines in the overall evaluation. Controlled studies further show that world rehearsal consistently improves policy learning across model scales. At test time, the internalized world model enables private rehearsal before committed execution, yielding further gains under a moderate rehearsal budget without additional external interaction. Our findings establish world rehearsal as a new path toward scaling LLM agent training beyond the constraints of external environments. Our code is publicly available at https://github.com/Within-yao/EnvACE.
Chinese Translation
训练大型语言模型代理以进行长期工具使用通常依赖于与真实或合成可执行环境的交互，这些环境的构建和验证成本高昂，或者依赖于难以扎根的外部模拟器。我们提出了EnvACE，一种自主强化学习方法，它在训练过程中用世界排练替代了外部环境交互。该策略在行动和排练之间交替进行：首先生成一个工具调用，然后扮演环境的角色以产生该动作引发的响应，并根据排练的响应调整后续决策。这两个角色通过任务成功奖励进行端到端的联合优化。通过世界排练，该策略在其参数中内化了动作与环境响应之间的关系，从而产生一个直接支持决策的代理世界模型。在BFCL-v4、tau^2-Bench、VitaBench和FinMCP-Bench上，EnvACE实现了强大且可转移的性能，在整体评估中超越了环境扩展基线。控制研究进一步表明，世界排练在不同模型规模中始终改善策略学习。在测试时，内化的世界模型使得在执行前能够进行私密排练，在适度的排练预算下实现了进一步的收益，而无需额外的外部交互。我们的研究结果确立了世界排练作为超越外部环境约束的LLM代理训练的新路径。我们的代码已公开发布在 https://github.com/Within-yao/EnvACE。
cs.AI / 91 / 2608.06223
TS-RAG: Retrieval Augmented Generation for Time Series Forecasting
TS-RAG：用于时间序列预测的检索增强生成
Abstract
While deep learning models, particularly transformer-based architectures, have shown impressive performance in time series forecasting, the application of retrieval-augmented generation (RAG) in this domain remains limited. Since RAG has proven effective in enhancing the capabilities of large language models by incorporating relevant external information, retrieving similar time series sequences as references might also improve accuracy in time series forecasting tasks. However, most time series models are constrained by limited training data, smaller parameter scales, and a lack of the extensive generative capabilities found in large language models. Simply concatenating reference sequences into the prompt, as done in language models, may not yield the expected results. To address these challenges, we propose a novel approach, TS-RAG, which leverages RAG to enhance forecasting performance. The framework introduces specially designed reference tokens to effectively fuse information from the input sequence with that from retrieved similar sequences, enabling a more robust capture of complex temporal dynamics. Experimental results demonstrate that TS-RAG achieves consistent state-of-the-art performance across several real-world forecasting benchmarks.
Chinese Translation
尽管深度学习模型，特别是基于变换器的架构，在时间序列预测中表现出色，但检索增强生成（RAG）在该领域的应用仍然有限。由于RAG已被证明能够通过整合相关的外部信息来增强大型语言模型的能力，因此检索相似的时间序列作为参考也可能提高时间序列预测任务的准确性。然而，大多数时间序列模型受到训练数据有限、参数规模较小以及缺乏大型语言模型所具备的广泛生成能力的限制。简单地将参考序列连接到提示中，如同在语言模型中所做的，可能无法产生预期的效果。为了解决这些挑战，我们提出了一种新颖的方法，TS-RAG，利用RAG来增强预测性能。该框架引入了特别设计的参考标记，以有效地融合输入序列与检索到的相似序列的信息，从而更稳健地捕捉复杂的时间动态。实验结果表明，TS-RAG在多个真实世界预测基准测试中实现了一致的最先进性能。
cs.AI / 92 / 2608.06243
DASH: Divergence-Adaptive Supervision Horizons for On-Policy Self-Distillation of Reasoning Models
DASH：用于策略自蒸馏推理模型的发散自适应监督视野
Abstract
Reinforcement learning with verifiable rewards (RLVR) improves the reasoning capabilities of large language models using automatically verifiable outcome signals, but these signals are typically sparse and at the sequence-level. On-policy self-distillation (OPSD) mitigates this sparsity by querying a privileged teacher at student-visited prefixes and providing dense token-level distributional supervision. Although this dense supervision alleviates signal sparsity, we find that standard OPSD still underexploits the temporal structure of the rollout. It assigns every local divergence the same coefficient, regardless of its position or the divergence sequence in which it occurs. In on-policy autoregressive generation, the same divergence magnitude can follow different discrepancy histories, reflecting different evolutions of the mismatch between the teacher and student. Since the local scalar alone cannot distinguish these temporal contexts, standard OPSD cannot adapt its token-level weights to the realized discrepancy sequence. To address this limitation, we propose Divergence-Adaptive Supervision Horizons (DASH). DASH maps the gap between each local distillation signal and the sequence-level mean to an adaptive propagation gate and then uses these gates to control backward multi-step aggregation. By doing so, DASH adjusts token-level supervision weights according to how local divergences evolve during generation. Experiments on three mathematical reasoning benchmarks across three model scales show that DASH improves over our matched vanilla OPSD reruns on every benchmark at all three scales. DASH reuses the teacher and student distributions that OPSD already computes, so the gains require no additional teacher or student forward pass. Code: https://github.com/DBtxy/DASH-OPSD
Chinese Translation
具有可验证奖励的强化学习（RLVR）利用自动可验证的结果信号提升大型语言模型的推理能力，但这些信号通常稀疏且处于序列级别。策略自蒸馏（OPSD）通过在学生访问的前缀处查询特权教师并提供密集的标记级分布监督来缓解这种稀疏性。尽管这种密集监督减轻了信号稀疏性，但我们发现标准OPSD仍然未能充分利用回滚的时间结构。它对每个局部发散分配相同的系数，而不考虑其位置或发生的发散序列。在策略自回归生成中，相同的发散幅度可以跟随不同的差异历史，反映教师与学生之间不匹配的不同演变。由于局部标量无法单独区分这些时间上下文，标准OPSD无法根据实现的差异序列调整其标记级权重。为了解决这一限制，我们提出了发散自适应监督视野（DASH）。DASH将每个局部蒸馏信号与序列级均值之间的差距映射到一个自适应传播门，然后使用这些门控制向后多步聚合。通过这样做，DASH根据局部发散在生成过程中的演变调整标记级监督权重。在三个数学推理基准测试和三个模型规模上的实验表明，DASH在所有三个规模的每个基准测试中均优于我们匹配的普通OPSD重跑。DASH重用了OPSD已经计算的教师和学生分布，因此这些增益不需要额外的教师或学生前向传播。代码：https://github.com/DBtxy/DASH-OPSD
cs.AI / 93 / 2608.06265
Improving the Realism of Synthetic Clinical Benchmarks Under Utility Constraints
在效用约束下提高合成临床基准的真实性
Abstract
Synthetic clinical benchmarks for enterprise AI agents can pass existing utility checks and still remain structurally unrealistic, especially in privacy-sensitive healthcare settings where operational data are hard to access. We study how to improve such benchmarks without breaking the downstream utility checks already used in practice. We formulate benchmark revision as utility-constrained realism improvement: dataset changes should increase realism while staying above an operational utility floor. We instantiate this idea on a care-gap benchmark derived from Synthea-generated patients exercised through demonstration electronic health record workflows and then processed by the same downstream pipeline as operational data. Realism is measured through missingness structure, simplicity, structural plausibility, and population alignment. The baseline benchmark is extremely thin: sampled-pair missingness is 79.44%, only 12.75% of rows are actionable, 38.94% of patients have zero actionable measures, and top-three token concentration reaches 100.0%. Two deterministic revisions improve these panels while remaining above the current utility floor, whereas a naive densification control preserves unrealistic templating. We further show that internal benchmark realism and source fidelity to an aggregate operational reference are related but distinct objectives. These results suggest that synthetic benchmark quality should be optimized explicitly, with utility treated as one constraint rather than as sufficient evidence of realism.
Chinese Translation
企业人工智能代理的合成临床基准可以通过现有的效用检查，但在结构上仍然不够真实，尤其是在隐私敏感的医疗环境中，操作数据难以获取。我们研究如何在不破坏实践中已使用的下游效用检查的情况下改善这些基准。我们将基准修订表述为效用约束下的真实性提升：数据集的变化应在保持高于操作效用底线的同时提高真实性。我们在一个基于Synthea生成的患者的护理差距基准上实现了这一理念，该基准通过演示电子健康记录工作流程进行处理，并与操作数据通过相同的下游流程进行处理。真实性通过缺失结构、简单性、结构合理性和人群一致性进行测量。基线基准极其稀薄：样本对缺失率为79.44%，仅有12.75%的行是可操作的，38.94%的患者没有可操作的措施，前三个标记的集中度达到100.0%。两个确定性的修订在保持高于当前效用底线的同时改善了这些面板，而简单的稠密化控制则保留了不现实的模板。我们进一步表明，内部基准的真实性和对聚合操作参考的源保真度是相关但不同的目标。这些结果表明，合成基准的质量应明确优化，效用应视为一个约束，而不是现实性的充分证据。
cs.AI / 94 / 2608.06270
The Illusion of Visual Tool-Use: A Causal Audit of Thinking with Images
视觉工具使用的幻觉：图像思维的因果审计
Abstract
The "thinking-with-images" paradigm equips multimodal LLMs with active visual operations such as crop-and-zoom. However, models using these operations often achieve only marginal or negative gains over direct inference at substantially higher token cost. They may also repeatedly crop irrelevant regions and fail on questions that direct inference answers correctly. We ask whether the returned visual evidence causally affects the answer. To answer this question, we formulate visual tool-use as a causal graph that separates observation-mediated paths from action-induced shortcuts. We then audit it through interventions at the three levels: policy (comparing tool-use with direct inference), trajectory (corrupting all observations during rollout), and step (counterfactually replacing one individual observation under a fixed prefix). Our step-level estimand, Visual Evidence Gain, isolates the contribution of each returned observation. Across six representative models and five fine-grained perception benchmarks, we uncover policy miscalibration with two failure modes. In Calling Without Looking, returned observations have no causal effect on the answer. In Looking Without Planning, observations are informative but the call schedule is incoherent. A trajectory-level diagnostic decomposes the policy-level accuracy gain and shows that the gain is concentrated in a Calibrated minority. We term this discrepancy the illusion of visual tool-use: despite aggregate accuracy gains, visual tool-use is not causally effective across a broad range of rollouts. The code is available at https://github.com/OpenCausaLab/CauAudit.
Chinese Translation
“图像思维”范式为多模态大语言模型（LLMs）提供了主动的视觉操作，如裁剪和缩放。然而，使用这些操作的模型通常在显著更高的标记成本下，仅获得边际或负面的收益，且可能反复裁剪无关区域，对直接推理能够正确回答的问题失败。我们探讨返回的视觉证据是否对答案产生因果影响。为了解答这个问题，我们将视觉工具使用表述为一个因果图，区分观察介导路径与行动诱导捷径。然后，我们通过三个层次的干预进行审计：政策层（比较工具使用与直接推理）、轨迹层（在回滚过程中破坏所有观察）和步骤层（在固定前缀下反事实替换一个单独观察）。我们的步骤级估计量，视觉证据增益，隔离了每个返回观察的贡献。在六个代表性模型和五个细粒度感知基准上，我们揭示了政策失调的两个失败模式。在“无视而呼叫”中，返回的观察对答案没有因果影响。在“无计划而观察”中，观察是信息性的，但调用顺序不连贯。轨迹级诊断分解了政策级准确性增益，显示增益集中在一个经过校准的少数群体中。我们将这种差异称为视觉工具使用的幻觉：尽管总体准确性有所提高，视觉工具使用在广泛的回滚范围内并不具备因果有效性。代码可在 https://github.com/OpenCausaLab/CauAudit 获取。
cs.AI / 95 / 2608.06294
QuanTiMedAI: Quantum-Enhanced Time-Series Model guided by Agentic AI for Cardiac Arrest Mortality Prediction
QuanTiMedAI：基于代理人工智能的量子增强时间序列模型用于心脏骤停死亡率预测
Abstract
Cardiac arrest remains one of the most lethal conditions encountered in intensive care units. Despite the growing availability of electronic health record data, existing mortality prediction studies in this population largely depend on static summaries derived from early admission. Such approaches ignore the temporal progression of physiological deterioration and recovery that unfolds throughout a patient's ICU stay. To address this limitation, we introduce QuanTiMedAI, a quantum-agentic framework developed for cardiac arrest mortality prediction using agentic AI guided quantum enhancement time series model. The proposed system combines an agentic large language model (LLM) for clinically informed feature discovery with a compact quantum recurrent network for temporality aware mortality prediction. Our findings demonstrate that agentic LLM-guided feature selection consistently outperforms conventional feature selection approaches, and the proposed quantum architecture achieves competitive predictive performance through nonlinear feature enhancement while keeping the number of parameters very low. Through extensive experimentation on a MIMIC-IV cohort of cardiac arrest patients, QuanTiMedAI's quantum-enhanced architecture attains an AUROC of 0.852 using only 605 parameters, an improvement of approximately 2.9\% over a current state-of-the-art baseline for this task. A structured ablation study systematically validates the contribution of each architectural design choice. These results show that quantum-enhanced sequential modeling can exceed classical recurrent networks while using substantially fewer parameters.
Chinese Translation
心脏骤停仍然是重症监护病房中最致命的病症之一。尽管电子健康记录数据的可用性日益增加，但现有的针对该人群的死亡率预测研究在很大程度上依赖于早期入院所获得的静态摘要。这些方法忽视了患者在重症监护期间生理恶化和恢复的时间进程。为了解决这一局限性，我们提出了QuanTiMedAI，这是一个为心脏骤停死亡率预测而开发的量子代理框架，采用由代理人工智能引导的量子增强时间序列模型。该系统结合了一个用于临床特征发现的代理大型语言模型（LLM）和一个用于时间感知死亡率预测的紧凑型量子递归网络。我们的研究结果表明，代理LLM引导的特征选择始终优于传统的特征选择方法，而所提出的量子架构通过非线性特征增强实现了具有竞争力的预测性能，同时保持了非常低的参数数量。在对MIMIC-IV心脏骤停患者队列进行广泛实验后，QuanTiMedAI的量子增强架构在仅使用605个参数的情况下达到了0.852的AUROC，相较于当前该任务的最先进基线提高了约2.9%。结构化的消融研究系统地验证了每个架构设计选择的贡献。这些结果表明，量子增强的序列建模可以超越经典的递归网络，同时使用显著更少的参数。
cs.AI / 96 / 2608.06300
Bias Analysis of L2 Speaking Assessment Systems Using Concept Activation Vectors
使用概念激活向量的第二语言口语评估系统偏差分析
Abstract
Automatic speaking assessment systems are increasingly deployed in high-stakes settings to mark second language (L2) learners' speaking tests, making it critical to show that their scores depend on speaking proficiency rather than irrelevant speaker attributes such as first language (L1) or age. Transformer-based foundation models have improved the accuracy of these L2 speaking graders, but their black-box representations make fairness and interpretability analysis more difficult. Building on prior work that used Concept Activation Vectors (CAVs) to detect bias towards unwanted attributes (`concepts') in feature-based graders, we extend CAV-based analysis to two neural speaking assessment systems: a text-based BERT grader and a speech-and-text multimodal grader based on Whisper. CAVs represent human-interpretable concepts as directions in a model's activation space, allowing us to distinguish between whether a concept is encoded in a model's internal representations and whether it influences the predicted score, the latter quantified using a gradient-based sensitivity metric. Since CAVs rely on linear separability, which is less likely in complex neural embedding spaces, we also investigate whether sparse autoencoders (SAEs) provide cleaner concept directions by learning CAVs in a sparse latent space and mapping them back to activation space. Our analysis shows that concept recoverability depends strongly on the representation and architecture being probed, rather than on the concept alone. Sensitivity to concepts is also architecture-dependent. SAEs make concepts more linearly recoverable, but attenuate the original activation-space sensitivity, especially in low-dimensional layers. These findings highlight the need to distinguish concept recoverability from concept influence when auditing bias in speaking assessment systems.
Chinese Translation
自动口语评估系统越来越多地在高风险环境中用于评分第二语言（L2）学习者的口语测试，因此证明其评分依赖于口语能力而非无关的说话者属性（如第一语言（L1）或年龄）变得至关重要。基于变换器的基础模型提高了这些L2口语评分器的准确性，但其黑箱表示使得公平性和可解释性分析变得更加困难。在之前的研究基础上，我们使用概念激活向量（CAVs）来检测特征基础评分器中对不希望属性（“概念”）的偏差，我们将CAV基础分析扩展到两个神经口语评估系统：一个基于文本的BERT评分器和一个基于Whisper的语音与文本多模态评分器。CAVs将人类可解释的概念表示为模型激活空间中的方向，使我们能够区分一个概念是否被编码在模型的内部表示中，以及它是否影响预测分数，后者通过基于梯度的敏感度指标进行量化。由于CAVs依赖于线性可分性，而这在复杂的神经嵌入空间中不太可能，因此我们还研究稀疏自编码器（SAEs）是否通过在稀疏潜在空间中学习CAVs并将其映射回激活空间来提供更清晰的概念方向。我们的分析表明，概念可恢复性在很大程度上取决于被探测的表示和架构，而不仅仅是概念本身。对概念的敏感性也依赖于架构。SAEs使得概念更具线性可恢复性，但减弱了原始激活空间的敏感性，尤其是在低维层中。这些发现强调了在审计口语评估系统中的偏差时，需要区分概念可恢复性和概念影响。
cs.AI / 97 / 2608.06301
HarnessOpt-Bench: Evaluating LLMs at Harness Optimization
HarnessOpt-Bench：评估大型语言模型在工具优化中的表现
Abstract
As LLMs are increasingly deployed within agentic systems, their capabilities depend not only on the model weights but also on the harness: the prompts, tools, control flow, memory, and orchestration code surrounding them. This makes automated harness optimization -- the iterative and evaluation-guided improvement of a harness by an AI system -- both an important route to improving AI systems and a demanding capability for AI systems themselves. Yet the community lacks a common protocol for measuring how well frontier LLMs perform at this task. We introduce HarnessOpt-Bench, a benchmark for end-to-end harness optimization under expensive and stochastic evaluation. An optimizer, an LLM paired with a coding harness, receives a target agent's seed harness, graded evaluation feedback, and a fixed target-evaluation budget. It edits the harness and nominates a final candidate, which is scored by its normalized gain over the seed on a held-out test partition that remains inaccessible throughout search. A trusted execution environment enforces the evaluation boundary, meters target-agent resource use, and preserves candidate versions for audit. We evaluate 5 frontier LLMs as optimizers both under a shared coding harness and under their native harnesses across 4 downstream tasks, over 111 scored runs. Experiment results show that optimizer models separate more than the coding harnesses they act through, native harnesses are not consistently superior, and gains vary substantially across tasks and seed regimes. These results establish harness optimization as a measurable and discriminative capability with large space for improvement.
Chinese Translation
随着大型语言模型（LLMs）在自主系统中的广泛应用，其能力不仅依赖于模型权重，还依赖于工具的设计：包括提示、工具、控制流程、记忆和围绕它们的调度代码。这使得自动化工具优化——通过人工智能系统对工具进行迭代和评估指导的改进——成为提升人工智能系统的重要途径，同时也是对人工智能系统自身的一项严峻挑战。然而，当前社区缺乏一个共同的协议来衡量前沿大型语言模型在此任务中的表现。我们提出了HarnessOpt-Bench，这是一个在高成本和随机评估下进行端到端工具优化的基准。优化器是一个与编码工具配对的大型语言模型，它接收目标代理的种子工具、评分评估反馈和固定的目标评估预算。优化器编辑工具并提名最终候选者，最终候选者的得分基于其在一个保持不可访问的测试分区上相对于种子的标准化增益。一个受信执行环境强制执行评估边界，控制目标代理的资源使用，并保留候选版本以供审计。我们在4个下游任务中评估了5个前沿大型语言模型作为优化器，使用共享编码工具和它们各自的原生工具，共进行了111次评分实验。实验结果表明，优化器模型的表现与它们所使用的编码工具有显著区别，原生工具并不总是优于共享工具，并且增益在不同任务和种子条件下变化显著。这些结果确立了工具优化作为一种可测量和具有区分能力的能力，并且在此领域有很大的改进空间。
cs.AI / 98 / 2608.06305
Beyond Top-K: Replacing Black-Box Retrieval with Interpretable Agentic Operations
超越Top-K：用可解释的代理操作替代黑箱检索
Abstract
Retrieval-augmented generation over long documents is dominated by one design: chunk the text, embed the chunks, and surface the top-k nearest neighbours of the query. We argue that for an important class of documents -- financial statements, audit reports, regulatory returns -- this design is structurally unsound, and we make the argument measurable. On a 780-page government financial report, 86.8% of content lines are table rows, thousands of near-identical figures compete in one embedding space, and a figure inherits its unit from a header a median of 13 lines above it -- so a chunk boundary routinely separates a number from whether it is in lakh or crore, an error of two orders of magnitude. A table-aware chunker built as a steelman fixes the unit problem but leaves 27-30% of numeric chunks with no fiscal-year header at every chunk size we tried. We propose READ (Reliable Embedding-free Agentic Document-search), in which an agent reads the raw document through three deterministic operations -- normalized lexical search, structural navigation, and bounded span reads -- exposed over the Model Context Protocol, so a trajectory is a replayable audit trail, not an opaque similarity score. On 51 verified questions READ answers 58.8% against dense retrieval's 15.7% (p_Holm = 2 x 10^-5) -- or 35.3% tuned, which READ still leads by 23.5 points (p_Holm = 0.017). An agent given the same loop but a top-k tool reaches only 27.5%, locating the gain in the interface rather than in iteration. We also report what the evidence does not support: BM25 is statistically indistinguishable from READ, so our result separates embedding-based from embedding-free retrieval, not agentic from lexical search.
Chinese Translation
基于检索增强的长文档生成主要依赖一种设计：将文本分块、嵌入这些块，并提取与查询最接近的前k个邻居。我们认为，对于一类重要的文档——财务报表、审计报告、监管回报——这种设计在结构上是不合理的，我们将这一论点进行了量化。在一份780页的政府财务报告中，86.8%的内容行是表格行，成千上万个几乎相同的数字在一个嵌入空间中竞争，而一个数字的单位通常来自其上方中位数为13行的标题——因此，块边界常常将一个数字与其是否为lakh或crore区分开来，这导致了两个数量级的错误。一个基于表格的分块器作为钢人论证修复了单位问题，但在我们尝试的每种块大小中，仍有27-30%的数字块没有财年标题。我们提出了READ（可靠的无嵌入代理文档检索），其中代理通过三种确定性操作——标准化词汇搜索、结构导航和有限跨度读取——读取原始文档，这些操作通过模型上下文协议暴露，因此轨迹是可重放的审计轨迹，而不是不透明的相似性评分。在51个经过验证的问题中，READ的回答率为58.8%，而密集检索为15.7%（p_Holm = 2 x 10^-5）——或者经过调优的35.3%，READ仍领先23.5个百分点（p_Holm = 0.017）。在相同的循环中，使用top-k工具的代理仅达到27.5%，其增益体现在界面而非迭代中。我们还报告了证据不支持的内容：BM25在统计上与READ无显著区别，因此我们的结果将基于嵌入的检索与无嵌入的检索区分开来，而不是代理与词汇搜索。
cs.AI / 99 / 2608.06346
TRAJDEBUG: Tracing Error Lifecycle to Identify Critical Failures in Long-Horizon Agent Trajectories
TRAJDEBUG：追踪错误生命周期以识别长时间跨度代理轨迹中的关键失败
Abstract
LLM-based agentic systems have shown remarkable capabilities in complex domains, while suffering from cascading errors and difficulty in debugging. Critical error detection aims to locate the earliest error step in a failed trajectory that is responsible for the final failure. However, progress faces two main challenges. First, long trajectories make it difficult to identify individual errors, since the evidence for judging a step may be scattered across distant instructions, observations, and prior context. Second, failed trajectories often contain multiple local errors with different downstream effects, only some of which remain responsible for the final failure. In this work, we propose TrajDebug, an error-lifecycle tracing framework that addresses long-trajectory error discovery with multi-granularity history compression and evidence-based error identification, and supports critical attribution by tracing each error's resolution status and terminal impact. We further construct TrajErrBench, a benchmark of 486 manually annotated failed trajectories from Tau2Bench and SWE-Bench Pro, covering realistic tool-use and coding scenarios. Experiments across diverse agent benchmarks show that TrajDebug achieves the best overall performance over existing baselines, and application studies further demonstrate that its diagnoses provide actionable feedback for improving downstream agent success. We will release the codes and data to facilitate further research.
Chinese Translation
基于大型语言模型（LLM）的代理系统在复杂领域展现出了显著的能力，但也面临着级联错误和调试困难的问题。关键错误检测旨在定位导致最终失败的失败轨迹中的最早错误步骤。然而，进展面临两个主要挑战。首先，长轨迹使得识别单个错误变得困难，因为判断一个步骤的证据可能分散在遥远的指令、观察和先前的上下文中。其次，失败的轨迹通常包含多个具有不同下游影响的局部错误，其中只有部分错误仍然对最终失败负责。在本研究中，我们提出了TrajDebug，一个错误生命周期追踪框架，旨在通过多粒度历史压缩和基于证据的错误识别来解决长轨迹错误发现问题，并通过追踪每个错误的解决状态和终极影响来支持关键归因。我们进一步构建了TrajErrBench，这是一个包含486个手动标注失败轨迹的基准数据集，来源于Tau2Bench和SWE-Bench Pro，涵盖了现实的工具使用和编码场景。针对多样化代理基准的实验表明，TrajDebug在现有基准上实现了最佳的整体性能，应用研究进一步表明其诊断提供了可操作的反馈，以改善下游代理的成功率。我们将发布代码和数据以促进进一步的研究。
cs.AI / 100 / 2608.06351
Challenges in Evaluating Explanation Methods for Static and Evolving Data
评估静态和动态数据解释方法的挑战
Abstract
This paper addresses the limitations of Explainable Artificial Intelligence (XAI) with respect to insufficient evaluation. They are illustrated through the DetoxAI image recognition system for bias detection and concept unlearning. Then, an example of a human-grounded evaluation of methods for explaining image classification is presented. The paper further explores methods for adapting explanations to evolving data streams with concept drift. Experiences with adapting counterfactuals for this problem are discussed. Finally it is related to the challenges of tracking the co-evolution of data, models, and explanations.\footnote{This paper has been accepted for a publication in J.Nalepa (ed) Explainable AI in Space. Proceedings of EASi 2026 Workshop at IJCAI-ECAI 2026 Bremen, Springer CCIS vol 3107 (2016).}
Chinese Translation
本文讨论了可解释人工智能（XAI）在评估不足方面的局限性。通过DetoxAI图像识别系统在偏见检测和概念遗忘中的应用进行了说明。接着，提供了一个基于人类的图像分类解释方法评估的示例。本文进一步探讨了如何将解释方法适应于具有概念漂移的动态数据流。讨论了在这一问题上调整反事实的经验。最后，本文将其与跟踪数据、模型和解释的共同演化的挑战联系起来。
cs.AI / 101 / 2608.06361
The Low Frequency Trap: Video Language Models Fail at Simple Event Bookkeeping
低频陷阱：视频语言模型在简单事件记账中失败
Abstract
Real-world video benchmarks provide broad coverage, but their fixed clips entangle event count, rate, duration, and visual complexity, making failure modes hard to isolate. While existing programmatic benchmarks offer better control, they score only the final answer rather than auditing reported events against executable ground truth. To bridge this gap, we introduce trace-grounded parametric profiling for event counting in three controlled video tasks: bouncing-ball wall contacts, visual blinks, and categorical state transitions. Across 2,190 videos, we vary event count N and frequency F while holding rendering fixed. Each video includes an executable event trace for capability-surface estimation and timestamp-level evaluation. Our results reveal a staged temporal failure. At an 80% reliability threshold, Gemini 3.6 Flash reliably counts persistent state transitions up to 12 events at 0.5 and 1.0 Hz, yet demonstrates no reliable positive-count region for transient blinking events. Thus, event representation dictates whether a model initially accesses evidence -- a limitation that compounds as count and frequency increase. In the high-count, high-frequency regime, only 0.2% of final counts are correct and the model recovers just 18.1% of true events. To test if visual access is the primary bottleneck, we increase sampling rate. Although this boosts Bounce Ball accuracy from 19.6% to 29.3%, the reported sequence agrees with ground truth only 3.7% of the time. Extra frames can therefore inflate final scores without producing faithful event recovery. Different prompting strategies yield similarly limited gains, and real-world video evaluations show the same concentration of success at low event counts. Ultimately, trace-grounded profiling shifts video evaluation from aggregate accuracy metrics to a detailed diagnostic of where temporal reasoning fails.
Chinese Translation
现实世界的视频基准测试提供了广泛的覆盖，但其固定片段将事件计数、速率、持续时间和视觉复杂性纠缠在一起，使得失败模式难以隔离。虽然现有的程序化基准测试提供了更好的控制，但它们仅对最终答案进行评分，而不是对报告的事件与可执行的真实数据进行审计。为了解决这一问题，我们引入了基于追踪的参数化事件计数分析，用于三个受控视频任务：弹跳球与墙壁接触、视觉眨眼和类别状态转换。在2190个视频中，我们在保持渲染固定的情况下，改变事件计数N和频率F。每个视频都包含一个可执行的事件追踪，用于能力表面估计和时间戳级评估。我们的结果揭示了一个分阶段的时间失败。在80%的可靠性阈值下，Gemini 3.6 Flash能够可靠地计数持久状态转换，最多可达12个事件，频率为0.5和1.0 Hz，但对于瞬态眨眼事件却没有可靠的正计数区域。因此，事件表示决定了模型是否最初能够获取证据——这一限制在计数和频率增加时加剧。在高计数、高频率的情况下，最终计数中只有0.2%是正确的，模型仅恢复了18.1%的真实事件。为了测试视觉访问是否是主要瓶颈，我们提高了采样率。尽管这将弹跳球的准确率从19.6%提升至29.3%，但报告的序列与真实数据一致的比例仅为3.7%。因此，额外的帧可能会在不产生真实事件恢复的情况下夸大最终得分。不同的提示策略也产生了类似有限的收益，而现实世界的视频评估显示在低事件计数下成功的集中度相同。最终，基于追踪的分析将视频评估从总体准确性指标转向对时间推理失败的详细诊断。
cs.AI / 102 / 2608.06366
Tracing the Heart: An Evidence-Linked Pipeline for Heart-Failure Feature Engineering
追踪心脏：一种基于证据的心力衰竭特征工程管道
Abstract
Electronic health record (EHR) feature engineering is a major bottleneck in clinical research and AI, accounting for 39-45% of data scientists' workload. This is especially pronounced in heart failure, which affects an estimated 6.7 million U.S. adults and requires integrating fragmented EHR data with disease-specific, guideline-based clinical reasoning. Existing rule-based and large language model (LLM)-based approaches offer only partial automation with limited maintainability and evidence traceability. We developed the Nimblemind Multi-Agent System (nMAS), an evidence-linked, rubric-grounded pipeline for automated heart-failure feature engineering, and evaluated it on 500 dummy patient records from nine EHR source tables. nMAS generated 132 structured and 70 rubric-scored aggregated features, verified for structural integrity, rubric compliance, and provenance, and audited by a restricted LLM. Adding the aggregated features improved held-out AUROC from 0.895 to 0.963 for HFrEF and 0.870 to 0.910 for HFpEF phenotyping, and an independent LLM-based rubric assessment of evidence support and methodological soundness scored the features at 81.5% of maximum points. These results demonstrate the feasibility of automated, auditable feature engineering for complex cardiovascular EHR data, though evaluation was limited to a single-institution cohort and external validation is needed.
Chinese Translation
电子健康记录（EHR）特征工程是临床研究和人工智能中的一个主要瓶颈，占数据科学家工作量的39-45%。这一问题在心力衰竭中尤为明显，估计有670万美国成年人受到影响，且需要将碎片化的EHR数据与特定疾病的、基于指南的临床推理相结合。现有的基于规则和大型语言模型（LLM）的方法仅提供有限的自动化，且维护性和证据可追溯性不足。我们开发了Nimblemind多智能体系统（nMAS），这是一个基于证据、以评分标准为基础的自动化心力衰竭特征工程管道，并在来自九个EHR源表的500个虚拟患者记录上进行了评估。nMAS生成了132个结构化特征和70个评分标准评分的聚合特征，经过结构完整性、评分标准合规性和来源的验证，并由限制性LLM进行了审计。添加聚合特征后，HFrEF的持出AUROC从0.895提高到0.963，HFpEF表型从0.870提高到0.910，且独立的基于LLM的评分标准评估了证据支持和方法论的合理性，特征得分为最大分数的81.5%。这些结果表明，对于复杂的心血管EHR数据，自动化、可审计的特征工程是可行的，尽管评估仅限于单一机构的队列，仍需进行外部验证。
cs.CL / 1 / 2608.05151
Simulator-Grounded Large Language Models for Industrial Causal Reasoning: Tool-Use, Structured Injection, and Plant-Portable Retrieval for Wastewater Treatment Decision Support
基于模拟器的大型语言模型用于工业因果推理：工具使用、结构化注入和植物可移植检索在废水处理决策支持中的应用
Abstract
Wastewater operators need answers grounded in how their plant's variables interact and how fast effects propagate, not in generic pretraining text, when asking causal questions such as "why is N2O rising?" or "what happens if I cut aeration by 20%?". We compare three concrete ways to ground a frozen Qwen2.5-32B-Instruct model in an architecturally interpretable wastewater simulator (CCSS-IX): a live simulator oracle (Method 1), structured parameter injection (Method 2), and a Decoupled Recall-Reasoning (DRR) retriever (Method 3). On a 198-question causal benchmark the three reach 99.5%, 79%, and 75.8%, forming a deployment ladder above the strongest retrieval-augmented baseline at 48%. The DRR retriever has 110M parameters and trains per plant in ~17 seconds; after cross-plant transfer to a biologically distinct plant it still reaches 88%, while Method 2's static table cannot transfer. On a 60-question counterfactual benchmark only Method 3 handles queries about what happens after an intervention: +16.3 pp over Method 2, paired 95% CI [+7.1, +26.4] pp, with 100% on the timescale and operating-regime categories. On the AI2 Reasoning Challenge (ARC) with an OpenBookQA fact corpus, the same selective-retrieval mechanism reaches 79% versus unconstrained Llama-3.1-8B 76% and full-injection 74%, a +3 pp out-of-domain replication that argues against a result specific to wastewater treatment. We provide the first single-simulator comparison of live tool-use, static parameter injection, and learned numerical-parameter retrieval for industrial causal question answering.
Chinese Translation
废水操作员在询问因果问题时，例如“为什么N2O上升？”或“如果我将曝气减少20%会发生什么？”时，需要基于其工厂变量之间的相互作用及效果传播速度的答案，而不是基于通用的预训练文本。我们比较了三种具体方法，将冻结的Qwen2.5-32B-Instruct模型与一个在架构上可解释的废水模拟器（CCSS-IX）结合：实时模拟器神谕（方法1）、结构化参数注入（方法2）和解耦回忆推理（Decoupled Recall-Reasoning, DRR）检索器（方法3）。在198个因果基准问题上，三种方法的准确率分别为99.5%、79%和75.8%，形成了高于最强检索增强基线（48%）的部署梯度。DRR检索器具有1.1亿个参数，每个工厂训练约17秒；在跨工厂转移到生物学上不同的工厂后，仍能达到88%的准确率，而方法2的静态表无法转移。在60个反事实基准问题上，只有方法3能够处理关于干预后发生情况的查询：比方法2高出16.3个百分点，配对95%置信区间为[+7.1, +26.4]个百分点，在时间尺度和操作模式类别上均达到100%。在AI2推理挑战（ARC）中，使用OpenBookQA事实语料库的相同选择性检索机制达到了79%的准确率，而不受限制的Llama-3.1-8B为76%，全注入为74%，在域外复制中提高了3个百分点，这表明结果并非特定于废水处理。我们提供了首次针对工业因果问题回答的实时工具使用、静态参数注入和学习的数值参数检索的单一模拟器比较。
cs.CL / 2 / 2608.05152
Mean-Field Dynamics of Chain-of-Thought Reasoning in Large Language Models
大语言模型中链式思维推理的均场动力学
Abstract
Large language models (LLMs) with chain-of-thought reasoning have been widely applied in recent years, and theoretical explanations of their behavior may help deepen our understanding and guide model optimization. In this study, we introduce a framework that seeks statistical regularities and theoretical interpretations in LLM reasoning without simplifying the model architecture or making analogies to existing physical systems. We formulate LLM reasoning as a guided discovery process on a clue graph, and derive a one-dimensional ordinary differential equation for the fraction of discovered clues using the mean-field approximation. Experimentally, clue tokens are identified using the normalized surprisal of a student LLM on the outputs of a teacher LLM, and statistical regularities are obtained by averaging over many reasoning chains of thought. Our experiments show that the resulting statistical regularities are reproducible within the same dataset and can be fitted by the solving the proposed theoretical equation.
Chinese Translation
近年来，具有链式思维推理的大语言模型（LLMs）得到了广泛应用，对其行为的理论解释可能有助于加深我们的理解并指导模型优化。在本研究中，我们引入了一个框架，旨在寻找LLM推理中的统计规律和理论解释，而不简化模型架构或类比现有物理系统。我们将LLM推理形式化为在线索图上的引导发现过程，并利用均场近似推导出发现线索的比例的一维常微分方程。实验中，使用学生LLM对教师LLM输出的标准化惊讶度来识别线索标记，并通过对多个思维推理链的平均获得统计规律。我们的实验表明，得到的统计规律在同一数据集内是可重复的，并且可以通过求解所提出的理论方程进行拟合。
cs.CL / 3 / 2608.05153
Universal Pathologies, Conditional Consequences: A Triple-Robustness Analysis of RAG for Multi-Hop Traceability
普遍病理学、条件后果：多跳可追溯性中 RAG 的三重稳健性分析
Abstract
GraphRAG underperforms vector RAG on citation precision in many reports, but where and why have remained corpus-bound. We present a triple-robustness analysis that holds the retrieval architecture fixed and varies three orthogonal axes embedder (local e5-small -> Azure text-embedding-3-small), corpus (DO-178C typed-edge requirements -> Wikipedia paragraph chains via MuSiQue), and judge (paired GPT-5.4 x GPT-4.1) across 4,440 main-matrix runs, 600 cross-corpus runs, and 1,200 paired faithfulness judgments. (C2a) Over-citation is architecturally universal: GraphRAG emits 11-15 IDs per answer at citation precision 0.12-0.23 and retrieval recall 0.68-0.87 across all three settings. (C2b) Its faithfulness consequence is corpus-conditional: in typed-edge DO-178C, GraphRAG faithfulness collapses 74%->40% across hops; on Wikipedia chains the same pipeline rises 42%->58% because over-cited paragraphs remain topically supporting. (C1) Stratum-conditional winners are corpus-conditional but embedder-robust: vanilla wins 2-hop on DO-178C, GraphRAG wins 2-hop on MuSiQue, identical under either embedder. (C3) Single-judge LLM faithfulness is fragile to retrieval state: same-judge self-kappa across embedders is 0.137 for GPT-5.4 (verdict change on 41% of items). A learned router on dense embeddings alone reaches macro-F1 0.86 on hop classification (C4). We argue triple-robustness is the minimum bar for trustworthy RAG architecture claims.
Chinese Translation
在许多报告中，GraphRAG 在引用精度上表现不如向量 RAG，但其原因和具体情况仍然受限于语料库。我们提出了一种三重稳健性分析，固定检索架构并变化三个正交轴：嵌入器（local e5-small -> Azure text-embedding-3-small）、语料库（DO-178C 类型边缘需求 -> 通过 MuSiQue 的维基百科段落链）和评判者（配对的 GPT-5.4 x GPT-4.1），在 4,440 次主矩阵运行、600 次跨语料库运行和 1,200 次配对可信度判断中进行分析。(C2a) 过度引用在架构上是普遍存在的：GraphRAG 在所有三种设置下每个答案发出 11-15 个 ID，引用精度为 0.12-0.23，检索召回率为 0.68-0.87。(C2b) 其可信度后果是语料库条件的：在类型边缘的 DO-178C 中，GraphRAG 的可信度在跳跃中从 74% 降至 40%；而在维基百科链上，同一管道的可信度则从 42% 上升至 58%，因为过度引用的段落仍然在主题上提供支持。(C1) 分层条件的赢家是语料库条件的，但嵌入器稳健：在 DO-178C 上，vanilla 在 2 跳中获胜，而 GraphRAG 在 MuSiQue 上的 2 跳中获胜，两者在任一嵌入器下表现相同。(C3) 单评判者 LLM 的可信度对检索状态非常脆弱：不同嵌入器下的同评判者自我 Kappa 为 0.137（41% 项目的裁决发生变化）。仅在密集嵌入上学习的路由器在跳跃分类上达到宏观 F1 0.86 (C4)。我们认为三重稳健性是可信 RAG 架构声明的最低标准。
cs.CL / 4 / 2608.05154
RIG-RoPE: Relation- and Instance-Gated Rotary Positional Encoding with Duration-Aware Temporal Coordinates
RIG-RoPE：具有持续时间感知时间坐标的关系和实例门控旋转位置编码
Abstract
Rotary positional encoding (RoPE) is a core component of modern language models and has been extended to multimodal LLMs through multidimensional variants such as multimodal RoPE (M-RoPE), which split positional channels into temporal, height, and width subspaces. This report identifies two limitations of static multidimensional position assignment in interleaved multimodal contexts. First, height/width rotations may be applied to token pairs whose spatial displacement is not a well-defined geometric object, producing cross-modal and inter-instance spatial interference. Second, temporal coordinates are often treated as equal-step counters, so a text token, an image block, and a video segment can advance the temporal phase by comparable amounts despite different information density. We propose RIG-RoPE, a relation- and instance-gated RoPE mechanism with duration-aware temporal coordinates. RIG-RoPE augments each token with a modality indicator, a visual instance identifier, and a scalar information-duration coordinate. It enables H/W rotations only for query-key pairs from the same visual instance; otherwise the unknown spatial displacement is marginalized rather than set to zero. Temporal rotations use interpolated cumulative block durations: text tokens consume unit duration, images use a dimension-aware logarithmic spatial scale, and videos further apply a logarithmic temporal extension over effective frames. We provide a gauge-invariance argument for avoiding ordinary cross-instance spatial rotation, an impossibility result for static IDs under shared H/W subspaces, and a duration-consistency argument against equal-step multimodal time. RIG-RoPE adds no learned parameters and can be implemented inside tiled attention kernels with constant additional metadata per token. This preliminary report establishes the formulation and validation path without claiming empirical superiority.
Chinese Translation
旋转位置编码（RoPE）是现代语言模型的核心组成部分，并通过多维变体扩展到多模态大语言模型（LLMs），例如多模态RoPE（M-RoPE），该变体将位置通道分割为时间、高度和宽度子空间。本报告识别了在交错多模态上下文中静态多维位置分配的两个局限性。首先，高度/宽度旋转可能应用于空间位移不是明确定义几何对象的令牌对，从而产生跨模态和实例间的空间干扰。其次，时间坐标通常被视为等步计数器，因此文本令牌、图像块和视频片段可以在信息密度不同的情况下以相似的量推进时间相位。我们提出了RIG-RoPE，这是一种具有持续时间感知时间坐标的关系和实例门控RoPE机制。RIG-RoPE为每个令牌增强了模态指示符、视觉实例标识符和标量信息持续时间坐标。它仅对来自同一视觉实例的查询-键对启用高度/宽度旋转；否则，未知的空间位移被边缘化而不是设为零。时间旋转使用插值的累积块持续时间：文本令牌消耗单位持续时间，图像使用维度感知的对数空间尺度，视频进一步对有效帧应用对数时间扩展。我们提供了一个避免普通跨实例空间旋转的测量不变性论证，一个关于共享高度/宽度子空间下静态ID的不可能性结果，以及一个反对等步多模态时间的持续时间一致性论证。RIG-RoPE不增加任何学习参数，并且可以在每个令牌的常量附加元数据中实现于平铺注意力核内。本初步报告确立了公式和验证路径，但不声称经验上的优越性。
cs.CL / 5 / 2608.05155
Beyond Sentiment: Comparing Traditional NLP and LLM-Based Multi-Dimensional Analysis for Political News Evaluation
超越情感：比较传统自然语言处理与基于大型语言模型的多维度政治新闻评估分析
Abstract
Traditional sentiment analysis (SA) models, while effective for polarity classification, provide limited insight into the rhetorical, ideological, and framing dimensions of political discourse -- dimensions that are central to research in the social sciences and humanities (SSH). In this paper, we present a comparative study of RoBERTa-based sentiment analysis and an LLM-based multi-dimensional framing analysis platform applied to a corpus of 50 political news articles from 17 international media outlets. The results reveal a critical limitation we term neutral collapse: RoBERTa classifies 70% of articles as neutral, effectively flattening substantively rich political content into an analytically uninformative category. We find that 23% of neutral-classified articles exhibit negative probability scores above 0.30. By contrast, the LLM-based approach captures political bias direction and intensity, sensationalism, emotional appeal, and political framing -- yielding multi-dimensional analytical outputs aligned with SSH epistemologies. We argue that for political media analysis, traditional SA alone is insufficient, and that LLM-based multi-dimensional frameworks offer a more epistemologically adequate computational lens for SSH research needs.
Chinese Translation
传统情感分析（SA）模型在极性分类方面虽然有效，但对政治话语的修辞、意识形态和框架维度提供的洞察有限，这些维度在社会科学和人文学科（SSH）研究中至关重要。本文呈现了一项比较研究，比较了基于RoBERTa的情感分析与基于大型语言模型（LLM）的多维度框架分析平台，后者应用于来自17家国际媒体的50篇政治新闻文章的语料库。结果揭示了一个我们称之为中立崩溃的关键限制：RoBERTa将70%的文章分类为中立，有效地将实质丰富的政治内容压缩为一个分析上不具信息性的类别。我们发现，23%的中立分类文章的负概率分数超过0.30。相比之下，基于LLM的方法捕捉了政治偏见的方向和强度、耸人听闻的表现、情感诉求和政治框架，从而产生与SSH认识论相一致的多维度分析输出。我们认为，对于政治媒体分析，仅依赖传统的SA是不够的，而基于LLM的多维度框架为SSH研究需求提供了更为认识论上充分的计算视角。
cs.CL / 6 / 2608.05156
Scaffold-Mediated Post-Training: Co-Evolving Model Parameters and Procedural Scaffold Graphs
支架介导的后训练：共同演化的模型参数与过程支架图
Abstract
Post-training of large language models optimizes only parameters, while inference-time procedural scaffolds are typically designed independently of parameter training. This disconnect makes it difficult to automatically acquire and internalize complex strategies. We propose scaffold-mediated post-training: procedural scaffolds are organized into an evolvable graph structure that co-evolves with model parameters through discovery, distillation, and dynamic recompilation. We instantiate this paradigm as Skill Training. On FeatureBench, automatically discovered skills improve the passed rate by 8.1pp, and after progressive distillation the model still achieves a 27.7% passed rate without any external scaffold (distillation retention rate 85.2%, defined as post-distillation / with-skill passed rate), significantly outperforming standard SFT on the same data.
Chinese Translation
大型语言模型的后训练仅优化参数，而推理时的过程支架通常独立于参数训练进行设计。这种脱节使得自动获取和内化复杂策略变得困难。我们提出了支架介导的后训练：过程支架被组织成一个可演化的图结构，通过发现、蒸馏和动态重编译与模型参数共同演化。我们将这一范式实例化为技能训练。在 FeatureBench 上，自动发现的技能使通过率提高了 8.1 个百分点，并且在逐步蒸馏后，模型在没有任何外部支架的情况下仍然达到了 27.7% 的通过率（蒸馏保留率为 85.2%，定义为蒸馏后/有技能的通过率），显著优于在相同数据上标准的 SFT。
cs.CL / 7 / 2608.05157
Large Language Models Threaten Double-blind Review
大型语言模型威胁双盲评审
Abstract
Double blind peer review serves as the scientific community primary defense against status and affiliation bias. Its effectiveness rests on the assumption that anonymized manuscripts convey scientific merit without revealing their authors. While authorship can often be recovered using citation networks or stylistic markers, we show that this assumption is increasingly fragile in the presence of large language models (LLMs). Using only titles and abstracts from papers published after model training, we find that LLMs collapse anonymity more efficiently than humans, with belief concentrating onto a small subset of plausible authors drawn from pools of five domain expert candidates. This vulnerability persists even when stylistic and bibliographic cues are excluded, indicating that stable patterns in problem framing and research focus function as latent conceptual signatures of authorship. Together, these findings indicate that double blind review is vulnerable to automated semantic inference, necessitating a revaluation of how anonymity and fairness are maintained in an AI augmented research ecosystem.
Chinese Translation
双盲同行评审是科学界抵御地位和隶属关系偏见的主要防线。其有效性基于一个假设，即匿名的手稿能够传达科学价值而不揭示其作者身份。尽管作者身份通常可以通过引用网络或风格特征恢复，但我们表明，在大型语言模型（LLMs）存在的情况下，这一假设正变得越来越脆弱。仅使用模型训练后发表论文的标题和摘要，我们发现LLMs在消除匿名性方面的效率超过人类，信念集中在从五位领域专家候选人中提取的小范围可信作者上。这一脆弱性在排除风格和书目线索时依然存在，表明问题框架和研究重点中的稳定模式作为作者身份的潜在概念特征发挥作用。这些发现共同表明，双盲评审易受到自动语义推理的影响，迫切需要重新评估在人工智能增强的研究生态系统中如何维护匿名性和公平性。
cs.CL / 8 / 2608.05158
Safe Evolution with Circuit Anchors
带电路锚的安全进化
Abstract
In biological evolution, unconstrained mutation can lead to catastrophic outcomes: organisms may evolve enhanced capabilities while losing essential functions for survival. Nature's solution is \textit{developmental constraints}, where core regulatory genes remain anchored while peripheral genes adapt freely. We observe that current self-evolution algorithms for large language models lack analogous constraints. They optimize purely for capability, implicitly assuming safety will be preserved. Our experiments reveal this assumption to be dangerously wrong: models can \textit{misevolve} into powerful yet dangerous entities. Inspired by how Hox genes anchor body structure across $500$ million years of evolution, we propose \textbf{Circuit-Anchored Evolution (CAE)}. Using mechanistic interpretability, we identify a tiny \textit{safety circuit}, comprising less than $2$\% of model features, that causally mediates safety behaviors. We anchor this circuit during evolution, constraining it within a small displacement bound while allowing the remaining features to evolve freely. This mirrors the biological principle of \textit{evolvability with constraint}: preserving what is essential while adapting what is peripheral. Experiments across $3$ model families and two evolution algorithms demonstrate that CAE achieves superior safety preservation with minimal capability loss, substantially outperforming explicit reward-based constraints in both effectiveness and efficiency. Just as developmental constraints prevent biological evolution from producing nonviable organisms, circuit anchoring prevents model evolution from producing capable but dangerous systems.
Chinese Translation
在生物进化中，无约束的突变可能导致灾难性的结果：生物体可能在增强能力的同时失去生存所需的基本功能。自然的解决方案是 extit{发育约束}，其中核心调控基因保持固定，而外围基因则可以自由适应。我们观察到，目前针对大型语言模型的自我进化算法缺乏类似的约束。它们纯粹优化能力，隐含假设安全性将得到保留。我们的实验揭示了这一假设的危险性：模型可能 extit{错误进化}为强大但危险的实体。受到Hox基因在5亿年进化中锚定身体结构的启发，我们提出了 extbf{电路锚定进化（CAE）}。通过机制可解释性，我们识别出一个微小的 extit{安全电路}，其包含的模型特征不足2\%，并且因果性地介导安全行为。在进化过程中，我们锚定这个电路，将其限制在一个小的位移范围内，同时允许其余特征自由进化。这反映了生物学原理 extit{在约束下的进化能力}：保留必要的部分，同时适应外围部分。在3个模型家族和两种进化算法的实验中，CAE在保持安全性方面表现优越，且能力损失最小，显著优于基于显性奖励的约束，无论在有效性还是效率上。正如发育约束防止生物进化产生不可生存的生物体，电路锚定则防止模型进化产生有能力但危险的系统。
cs.CL / 9 / 2608.05161
SemiAdapt-Instruct: Extensible Instruction Tuning via Latent Domain-Specialised Adapters
SemiAdapt-Instruct：通过潜在领域专用适配器的可扩展指令调优
Abstract
Instruction-tuned LLMs are deployed into environments where domains evolve, yet extending a fine-tuned model's capabilities without full retraining remains an unsolved practical challenge. We present SemiAdapt-Instruct, a modular framework that discovers latent instruction domains, trains per-domain LoRA adapters in parallel, and performs parameter-free routing, incorporating new domains via single-adapter training without modifying existing components. SemiAdapt-Instruct outperforms full model fine-tuning across all configurations on both ROUGE-L and LLM-as-a-judge evaluation, while matching single LoRA fine-tuning and delivering extensibility that monolithic approaches cannot provide. We empirically demonstrate this extensibility by showing that updating a single adapter with new domain data outperforms all monolithic baselines. Our study also finds that independent discovery methods converge on the same specialisation-friendly domains. These findings demonstrate that decomposing heterogeneous instruction data into latent domains enables extensible NLP systems where evolving domains require only targeted single-adapter updates, eliminating the need for full model retraining.
Chinese Translation
指令调优的大型语言模型（LLMs）被部署到领域不断演变的环境中，但在不进行全面重新训练的情况下扩展微调模型的能力仍然是一个未解决的实际挑战。我们提出了SemiAdapt-Instruct，一个模块化框架，能够发现潜在的指令领域，平行训练每个领域的LoRA适配器，并执行无参数路由，通过单适配器训练将新领域纳入，而无需修改现有组件。SemiAdapt-Instruct在所有配置下的ROUGE-L和LLM-as-a-judge评估中均优于全面模型微调，同时与单个LoRA微调相匹配，并提供单体方法无法提供的可扩展性。我们通过实证证明这种可扩展性，显示用新领域数据更新单个适配器的效果优于所有单体基线。我们的研究还发现，独立发现方法收敛于相同的有利于专业化的领域。这些发现表明，将异构指令数据分解为潜在领域使得可扩展的自然语言处理系统成为可能，在不断演变的领域中只需针对性的单适配器更新，消除了全面模型重新训练的需要。
cs.CL / 10 / 2608.05162
PoolBench: A Benchmark for Pooling Strategies in Concept Representation Evaluation for Decoder-Only LLMs
PoolBench：解码器仅模型中概念表示评估的池化策略基准
Abstract
Pooling is a consequential but under-examined design choice in decoder-only concept representation work: practitioners must collapse token-level hidden states into a passage-level vector, yet no shared protocol exists for comparing this choice across concepts, models, and tasks. Reported gains are confounded by simultaneous changes in dataset, layer, construction method, and pooling rule, making principled decisions impossible. We introduce PoolBench, a benchmark that isolates pooling as the experimental variable under a fixed evaluation protocol. PoolBench covers 17 concepts, 19 pooling strategies, and 3 open-weight decoder-only models (Llama-3.1-8B, Gemma-2-9B, Mistral-7B), evaluated on a single audited corpus of 37,693 real-text passages. The primary axis is linear separability (D1/AUROC); steered concept prevalence (D2/SCP) and output-level disentanglement (D3) serve as diagnostic axes. The primary finding is decisive: W4_hierarchical reaches a cross-model mean AUROC of 0.7799, while the widely adopted P1_last_token baseline reaches only 0.7640 and is statistically significantly worse (Friedman+Nemenyi, p = 2.0e-36; 77 significant pairs among 18 effective strategies). Rankings are stable across layers (rho = 0.961--0.990). A key negative result: strong detection does not imply strong steering -- D2 and D3 are substantially weaker than D1 for most concepts, indicating a fundamental representational limit rather than a pooling failure. On mid-difficulty concepts, W4_hierarchical outperforms P1_last_token by 0.042--0.113 AUROC; construction method choice (DiffMean vs. REPE) has a larger effect (delta AUROC 0.15) than pooling (delta AUROC 0.016), establishing the correct practical hierarchy. We release the corpus, pre-extracted activations, scorer models, steering vectors, and evaluation code as a reusable protocol for pooling research.
Chinese Translation
池化是解码器仅概念表示工作中一个重要但未被充分研究的设计选择：实践者必须将令牌级隐藏状态压缩为段落级向量，但目前没有共享的协议用于跨概念、模型和任务比较这一选择。报告的增益受到数据集、层次、构建方法和池化规则的同时变化的影响，使得做出原则性的决策变得不可能。我们引入了PoolBench，一个在固定评估协议下将池化作为实验变量的基准。PoolBench覆盖17个概念、19种池化策略和3个开放权重的解码器仅模型（Llama-3.1-8B、Gemma-2-9B、Mistral-7B），在一个经过审计的37,693个真实文本段落的语料库上进行评估。主要轴线是线性可分性（D1/AUROC）；引导概念的普遍性（D2/SCP）和输出级别的解耦（D3）作为诊断轴线。主要发现是决定性的：W4_hierarchical的跨模型平均AUROC达到0.7799，而广泛采用的P1_last_token基线仅达到0.7640，并且在统计上显著更差（Friedman+Nemenyi，p = 2.0e-36；18种有效策略中有77对显著差异）。排名在各层之间是稳定的（rho = 0.961--0.990）。一个关键的负结果是：强检测并不意味着强引导——对于大多数概念，D2和D3显著弱于D1，这表明存在基本的表示限制而非池化失败。在中等难度的概念上，W4_hierarchical的表现比P1_last_token高出0.042--0.113 AUROC；构建方法选择（DiffMean与REPE）对结果的影响（增量AUROC 0.15）大于池化（增量AUROC 0.016），确立了正确的实际层次。我们发布了语料库、预提取的激活、评分模型、引导向量和评估代码，作为池化研究的可重用协议。
cs.CL / 11 / 2608.05163
Where Privacy Risk Lives in English-Source Multilingual RAG: A Stage-Decomposed Audit Across Five Query Languages
英语来源多语言RAG中的隐私风险：跨五种查询语言的阶段分解审计
Abstract
A common assumption holds that switching to a non-English language makes a multilingual RAG system easier to attack for personal information. We test this on an English-source synthetic-PII corpus with five query languages and a two-stage defence (LLM input judge + regex output filter), in a pipeline whose translator, judge, back-translator, and generator are all Qwen2.5-7B -- so every finding below is pipeline-conditional, not a causal ranking of language-inherent risk. Under output-only filtering, English has the highest observed unstructured-PII leak rate; only English-vs-Swahili separates cleanly under document-level bootstrap intervals. Once the input judge is added, residual leaks remain on Arabic and Swahili, and back-translating the query does not close the gap (an ablation we report but cannot use as a causal diagnostic, since the back-translator is also Qwen). On a separate n=17 multilingual-prompted-judge residual corner, attaching the gold corpus document to the input judge blocks 15/17 residual cells. We frame this last result as a mechanism diagnostic, not a deployable defence: it uses oracle retrieval, BLOCK/ALLOW rates are measured on adversarial queries only, and we measure no benign-query false-positive rate and no answer-utility cost. The supplementary material contains code, corpora, queries, and per-trial JSONLs; the priority follow-up is an independent-MT plus non-Qwen-judge replication with a native-speaker query set, scoped in the Limitations section.
Chinese Translation
一个普遍的假设认为，切换到非英语语言会使多语言RAG系统更容易受到个人信息的攻击。我们在一个英语来源的合成个人可识别信息（PII）语料库上进行了测试，该语料库包含五种查询语言，并采用了两阶段防御（大型语言模型输入判断 + 正则输出过滤），在一个所有翻译器、判断器、反向翻译器和生成器均为Qwen2.5-7B的流程中进行，因此以下每个发现都是条件于该流程的，而不是语言固有风险的因果排名。在仅输出过滤的情况下，英语的非结构化PII泄露率最高；只有英语与斯瓦希里语在文档级自助引导区间下清晰分离。一旦添加了输入判断器，阿拉伯语和斯瓦希里语仍然存在残余泄露，而反向翻译查询并未缩小这一差距（这是我们报告的消融实验，但由于反向翻译器也是Qwen，因此无法将其用作因果诊断）。在一个独立的n=17多语言提示判断器的残余角落中，将金标准语料库文档附加到输入判断器可以阻止15/17个残余单元。我们将这一最后结果框架视为机制诊断，而非可部署的防御：它使用了神谕检索，BLOCK/ALLOW率仅在对抗性查询上进行测量，我们未测量到良性查询的误报率，也没有答案效用成本。补充材料包含代码、语料库、查询和每次试验的JSONL；优先后续工作是使用独立机器翻译和非Qwen判断器的复制研究，采用母语者查询集，具体范围在局限性部分中说明。
cs.CL / 12 / 2608.05164
Cross-Architecture Steering Transfer in Language Models: A Systematic Empirical Study
语言模型中的跨架构引导迁移：系统实证研究
Abstract
Independently trained large language models may develop shared internal representations of semantic concepts despite architectural differences -- but whether this geometric similarity has functional consequences for cross-model behavioural control remains untested. We present the first systematic evaluation of cross-model steering transfer and show that shared LLM geometry is functionally exploitable, conditionally: concept directions from one model can steer a different independently trained model when sufficient representational capacity exists. We study five open-weight models spanning three parameter scales (0.8B--8B) and two architectural lineages, training one Sparse Autoencoder per model across 15 semantic domains and testing alignment across all 20 directed model pairs. We observe a suggestive discontinuity near 1.7B parameters: at >= 1.7B scale, 47--49% of cross-model feature pairs validate (Pearson r >= 0.60, Procrustes cosines 0.895--0.956), while alignment degrades sharply below 0.8B. Cross-model steering vectors (B3-TI) achieve a 71.0% win rate across 15 supervised concepts versus 68.0% for same-model native vectors; a single universal vector achieves 67.3% in 4 of 5 models without any per-model supervision. Transfer degrades for models below 1.7B and for one model with generation instability, confirming that functional exploitability requires sufficient representational capacity. Our findings underscore the importance of scale thresholds in mechanistic interpretability: tools validated at 7B scale may not transfer to smaller models without revalidation. We provide the first functional complement to the Platonic Representation Hypothesis -- geometric convergence across independently trained LLMs supports cross-model behavioural control without fine-tuning, under the identified scale conditions.
Chinese Translation
独立训练的大型语言模型尽管在架构上存在差异，但可能会发展出共享的语义概念内部表征——然而，这种几何相似性是否对跨模型行为控制具有功能性影响尚未得到验证。我们首次系统评估了跨模型引导迁移，并表明共享的LLM几何形状在功能上是可利用的，条件是：来自一个模型的概念方向可以引导另一个独立训练的模型，只要其具有足够的表征能力。我们研究了五个开放权重模型，涵盖三个参数规模（0.8B--8B）和两个架构谱系，为每个模型训练一个稀疏自编码器，跨越15个语义领域，并测试所有20对定向模型之间的对齐情况。我们观察到在1.7B参数附近存在显著的不连续性：在>= 1.7B规模下，47--49%的跨模型特征对验证有效（Pearson r >= 0.60，Procrustes余弦0.895--0.956），而在0.8B以下对齐显著下降。跨模型引导向量（B3-TI）在15个监督概念中实现了71.0%的胜率，而同模型本地向量为68.0%；在5个模型中，一个单一的通用向量在没有任何模型特定监督的情况下达到了67.3%。对于低于1.7B的模型以及一个生成不稳定的模型，迁移效果下降，确认功能可利用性需要足够的表征能力。我们的研究结果强调了机制可解释性中规模阈值的重要性：在7B规模下验证的工具可能无法在更小的模型中转移而无需重新验证。我们提供了对柏拉图表征假说的首个功能补充——在识别的规模条件下，独立训练的LLM之间的几何收敛支持跨模型行为控制，而无需微调。
cs.CL / 13 / 2608.05165
A Study of ASR Adaptation and Representation Dimensionality Reduction in Persian Speech Emotion Recognition Using Whisper
使用 Whisper 进行波斯语语音情感识别的 ASR 适应性与表示维度降低研究
Abstract
Speech Emotion Recognition (SER) in low-resource languages remains a challenging problem due to limited labeled data. In this work, we study the use of Whisper for Persian SER with a particular focus on representation dimensionality reduction and language-specific model adaptation. We propose a SER framework in which frame-level embeddings extracted from the Whisper encoder are reduced in dimensionality using PCA, eliminating the need for learned projection layers and substantially reducing the number of trainable parameters. The reduced representations are aggregated using an attention-based pooling mechanism and classified with a lightweight prediction head. In addition, we investigate whether fine-tuning Whisper on a Persian automatic speech recognition (ASR) task improves downstream SER performance. Experiments conducted on the ShEMO dataset under a speaker-independent evaluation protocol show that PCA-based dimensionality reduction consistently improves emotion recognition performance while reducing training latency and memory usage. ASR fine-tuning yields only modest gains for SER, suggesting limited transfer from language adaptation to emotion-related representations under the evaluated conditions. These findings provide practical insights into the efficient use of large pretrained speech models for emotion recognition in low-resource languages.
Chinese Translation
由于标注数据有限，低资源语言中的语音情感识别（SER）仍然是一个具有挑战性的问题。在本研究中，我们探讨了使用 Whisper 进行波斯语 SER，特别关注表示维度降低和特定语言模型的适应性。我们提出了一个 SER 框架，其中从 Whisper 编码器提取的帧级嵌入通过主成分分析（PCA）降低维度，消除了对学习投影层的需求，并显著减少了可训练参数的数量。降低维度后的表示通过基于注意力的池化机制进行聚合，并使用轻量级预测头进行分类。此外，我们还研究了在波斯语自动语音识别（ASR）任务上微调 Whisper 是否能提高下游 SER 性能。在无关说话者的评估协议下，对 ShEMO 数据集进行的实验表明，基于 PCA 的维度降低在提高情感识别性能的同时，减少了训练延迟和内存使用。ASR 微调对 SER 的提升仅为适度，表明在评估条件下，语言适应对情感相关表示的转移有限。这些发现为在低资源语言中有效利用大型预训练语音模型进行情感识别提供了实用的见解。
cs.CL / 14 / 2608.05166
Conditional Cognitive Biases in LLMs: How Biased User Turns Modulate In-Context Reasoning
大型语言模型中的条件认知偏见：偏见用户轮次如何调节上下文推理
Abstract
We present an evaluation of cognitive bias expression in state-of-the-art instruction-tuned LLMs under realistic multi-turn interaction settings. Our work introduces a novel three-condition experimental framework that disentangles the effect of exposure to a biased user turn from the effect of the turn's semantic content, alongside a benchmark of 24,300 jury-validated user prompts spanning all 81 cells of a 9x9 target-human bias interaction matrix. Across eight frontier LLMs, we find that biased conversational context systematically increases bias expression relative to zero-shot baselines in 6 of 8 models. We identify two competing behavioral dynamics underlying this effect: conversational exposure to biased reasoning generally amplifies downstream bias tendencies, while explicitly stated bias cues often trigger alignment-related suppression behaviors that reduce overt bias expression. We release our framework, codebase, and dataset to support future research on context-conditioned cognitive biases and behavioral adaptation in LLMs.
Chinese Translation
我们对在现实多轮交互环境下，最先进的指令调优大型语言模型（LLMs）中认知偏见的表现进行了评估。我们的工作引入了一种新颖的三条件实验框架，旨在将偏见用户轮次的暴露效应与轮次语义内容的效应进行区分，同时基准测试了24,300个经过陪审团验证的用户提示，涵盖了9x9目标人类偏见交互矩阵的所有81个单元。在八个前沿大型语言模型中，我们发现偏见的对话上下文系统性地增加了相对于零-shot 基线的偏见表现，在8个模型中的6个模型中均有体现。我们识别出两种相互竞争的行为动态：对偏见推理的对话暴露通常会增强下游偏见倾向，而明确陈述的偏见线索则常常触发与一致性相关的抑制行为，从而减少明显的偏见表现。我们发布了我们的框架、代码库和数据集，以支持未来关于上下文条件下认知偏见和大型语言模型行为适应的研究。
cs.CL / 15 / 2608.05167
CNM-BERT: A Drop-In Structural Embedding for Chinese Characters via Ideographic Description Sequences
CNM-BERT：通过表意描述序列实现的中文字符插入式结构嵌入
Abstract
Token-based encoders like BERT treat Chinese characters as atomic identifiers, ignoring their recursive orthographic structure. Consequently, models rely on contextual co-occurrence, degrading performance on rare and out-of-vocabulary (OOV) characters. We propose the Compositional Network Model (CNM), a lightweight augmentation that injects discrete compositional structure into Transformer encoders. CNM parses Ideographic Description Sequences (IDS) into trees, encodes them via a recursive Tree-MLP, and fuses the structural embeddings into BERT without modifying the backbone. Evaluated on the Wu et al. (2025) structural-probing benchmark, CNM-BERT outperforms the strongest baseline (ChineseBERT) on long-tail and OOV characters by +9.8 Structure accuracy and +7.7 Radical F1. Furthermore, CNM-BERT achieves consistent gains across CLUE, MRC, and NER tasks at both base and large scales, demonstrating that explicit structural injection delivers both robust OOV understanding and tangible downstream value.
Chinese Translation
基于令牌的编码器如BERT将中文字符视为原子标识符，忽略了它们递归的正字法结构。因此，模型依赖于上下文共现，导致在稀有和超出词汇表（OOV）字符上的性能下降。我们提出了组合网络模型（Compositional Network Model, CNM），这是一种轻量级增强方法，将离散的组合结构注入到Transformer编码器中。CNM将表意描述序列（Ideographic Description Sequences, IDS）解析为树结构，通过递归树多层感知器（Tree-MLP）进行编码，并将结构嵌入融合到BERT中，而无需修改其主干。在Wu等人（2025）的结构探测基准测试中，CNM-BERT在长尾和OOV字符上以+9.8的结构准确率和+7.7的部首F1超越了最强基线（ChineseBERT）。此外，CNM-BERT在CLUE、MRC和NER任务中在基础和大型规模上均取得了一致的提升，证明了显式结构注入不仅提供了强大的OOV理解能力，还带来了实质性的下游价值。
cs.CL / 16 / 2608.05169
ConWriter: Transition-Constrained Stateful Long-Form Story Generation with Lightweight Neuro-Symbolic Consistency Control
ConWriter：具有轻量级神经符号一致性控制的过渡约束状态长篇故事生成
Abstract
Long-form story generation requires models to preserve narrative consistency across extended contexts, yet existing prompting-based methods often accumulate temporal, factual, character, commonsense, and stylistic errors as the story grows. We propose ConWriter, a training-free framework for consistency-aware long story generation. ConWriter writes stories incrementally at the scene level, guided by static story requirements, dynamic narrative memory, symbolic state reasoning, and uncertainty-aware risk signals. Rather than treating long-story generation as a single free-form decoding process, ConWriter maintains evolving story states, checks whether new scenes satisfy required narrative transitions, and uses uncertainty-aware risk signals to prioritize validation and localized repair. This enables consistency control during generation, before local errors propagate into later scenes. We evaluate ConWriter on ConStory-Bench, covering four long-story tasks: continuation, generation, expansion, and completion. Due to the high cost of long-form generation and evaluation, we use the first five cases from each task and test 3k, 6k, and 12k target lengths across Qwen3.5-Plus, DeepSeek-V4-Flash, and GPT-5 series. Experiments follow the official ConStory-Bench evaluation protocol.
Chinese Translation
长篇故事生成要求模型在扩展上下文中保持叙事一致性，但现有的基于提示的方法在故事发展过程中往往会积累时间、事实、角色、常识和风格等方面的错误。我们提出了ConWriter，这是一个无需训练的一致性感知长篇故事生成框架。ConWriter在场景层面逐步撰写故事，受到静态故事要求、动态叙事记忆、符号状态推理和不确定性感知风险信号的指导。ConWriter并不将长篇故事生成视为单一的自由形式解码过程，而是维护不断演变的故事状态，检查新场景是否满足所需的叙事过渡，并利用不确定性感知风险信号来优先进行验证和局部修复。这使得在生成过程中能够进行一致性控制，防止局部错误传播到后续场景。我们在ConStory-Bench上评估ConWriter，涵盖四个长篇故事任务：续写、生成、扩展和完成。由于长篇生成和评估的高成本，我们使用每个任务的前五个案例，并在Qwen3.5-Plus、DeepSeek-V4-Flash和GPT-5系列上测试3k、6k和12k的目标长度。实验遵循官方的ConStory-Bench评估协议。
cs.CL / 17 / 2608.05170
DREAM: LLM-based Dynamic Role-playing via Event-Aware Memory Graph
DREAM：基于大型语言模型的动态角色扮演通过事件感知记忆图
Abstract
Role-playing agents (RPAs) have emerged as a key application of large language models, enabling immersive and high-fidelity character simulation. Accurate role-playing of established characters requires not only stylistic imitation but also temporally consistent and causally grounded behavioral reasoning. However, existing RPAs primarily rely on static character descriptions and unstructured memory, limiting their ability to maintain long-term narrative and personality coherence. We introduce DREAM, a structured memory framework for role-playing agents inspired by the Activating Event-Belief-Consequence (ABC) cognitive model. DREAM transforms unstructured literary text into an Event-aware Memory Graph (EMG) that organizes character experiences into temporally ordered and causally linked event graph. This representation enables the construction of dynamic, dual-granularity character profiles that capture both stable personality traits and event-driven behavioral evolution. We further propose the Temporal Causal Memory (TCM) benchmark to evaluate temporal consistency and long-range causal narrative coherence. DREAM achieves state-of-the-art performance across CoSER, LIFECHOICE, and TCM, outperforming multiple strong baselines. Our approach demonstrates the effectiveness of structured memory in enhancing the interpretability and consistency of role-playing agents.
Chinese Translation
角色扮演代理（RPA）作为大型语言模型的一个关键应用，能够实现沉浸式和高保真的角色模拟。准确的角色扮演不仅需要风格上的模仿，还需要时间上一致且因果基础的行为推理。然而，现有的RPA主要依赖于静态角色描述和非结构化记忆，这限制了它们维持长期叙事和个性一致性的能力。我们提出了DREAM，一个受激活事件-信念-后果（ABC）认知模型启发的角色扮演代理结构化记忆框架。DREAM将非结构化的文学文本转化为事件感知记忆图（EMG），将角色经历组织成时间顺序和因果关联的事件图。这种表示方式使得构建动态的、双粒度的角色档案成为可能，能够捕捉到稳定的个性特征和事件驱动的行为演变。我们进一步提出了时间因果记忆（TCM）基准，以评估时间一致性和长程因果叙事的一致性。DREAM在CoSER、LIFECHOICE和TCM上实现了最先进的性能，超越了多个强基线。我们的方法展示了结构化记忆在增强角色扮演代理的可解释性和一致性方面的有效性。
cs.CL / 18 / 2608.05188
Position: It's Time to Optimize LLMs for Self-Consistency
立场：是时候优化大型语言模型以实现自我一致性了
Abstract
Despite ever-increasing sophistication in language model (LM) pre- and post-training pipelines, many important failures persist: models overcondition on user framing ("sycophancy"), exhibit incomplete logical generalization, and produce confident but incorrect responses. We argue that these failures arise from a modeling assumption permeating all aspects of the pipeline: that behavior can be specified and evaluated independently on single-output pairs. Many model failures are difficult, if not impossible, to detect without reasoning about relationships between a model's responses across inputs. In this position paper, we propose self-consistency as a framework for understanding these failures. We first observe that a wide variety of techniques designed to improve specific aspects of LM behavior-targeting properties as diverse as adversarial robustness and factual coherence-can be understood as special cases of a common "consistency optimization" procedure and addressed with a standard set of optimization tools. We next outline a set of new model properties that could be achieved by optimizing for consistency, and conclude with a discussion of what it would mean to develop generally consistent LMs, including the capabilities they would enable and the objections they raise.
Chinese Translation
尽管语言模型（LM）的预训练和后训练流程日益复杂，但许多重要的失败依然存在：模型过度依赖用户的表述（“谄媚”），表现出不完整的逻辑泛化，并产生自信但错误的回答。我们认为，这些失败源于贯穿整个流程的建模假设：行为可以在单一输出对上独立指定和评估。许多模型的失败在没有推理模型在不同输入间响应关系的情况下是难以发现的，甚至是不可能的。在这篇立场论文中，我们提出自我一致性作为理解这些失败的框架。我们首先观察到，旨在改善LM行为特定方面的各种技术——针对对抗鲁棒性和事实一致性等多样属性——可以被理解为一种共同的“一致性优化”过程的特例，并可以通过一套标准的优化工具进行处理。接下来，我们概述了一组通过优化一致性可以实现的新模型属性，并以讨论开发普遍一致的语言模型的意义作为结尾，包括它们所能实现的能力及其引发的争议。
cs.CL / 19 / 2608.05232
Analysis of Numerical Localisation in LLM Translations
大型语言模型翻译中的数值本地化分析
Abstract
The work of Tang et. al. (2025) on numerical translation is extended by analysing the capability of five large language models (LLMs) for the localisation of times, numbers, and dates instead of translation. Models were selected that could be loaded onto and run on commodity hardware and a baseline quality for each mode is computed, then three different strategies to improve on that accuracy were tested. In contrast to Tang et. al., it was discovered that on the tested LLMs, embedding the localisation principles into the prompt context provided a statistically significant improvement in accuracy compared to direct translation or the alternative strategies.
Chinese Translation
Tang 等人（2025）关于数值翻译的研究通过分析五种大型语言模型（LLMs）在时间、数字和日期的本地化能力得到了扩展，而非仅仅是翻译。所选模型能够在普通硬件上加载和运行，并计算出每种模式的基线质量，然后测试了三种不同的策略以提高准确性。与 Tang 等人不同的是，研究发现，在测试的 LLMs 中，将本地化原则嵌入提示上下文中相比于直接翻译或其他替代策略，提供了统计上显著的准确性提升。
cs.CL / 20 / 2608.05254
Constraint-First Reasoning: A Training-Free Protocol for Exploiting Answer-Space Constraints in Mathematical Problem Solving
约束优先推理：一种无训练协议，用于在数学问题求解中利用答案空间约束
Abstract
Large language models can derive a plausible mathematical object yet still violate explicit requirements--for example, by omitting a modular reduction, returning a non-integer, or using the wrong encoded answer form. We introduce Constraint-First Reasoning (CFR), a training-free two-stage prompting protocol: Stage 1 extracts and summarizes constraints entailed by the problem, and Stage 2 solves while checking intermediate and final results against that summary. Routed-CFR activates the two-stage protocol only when a text-only regex router detects restrictive cues; otherwise it uses direct chain-of-thought (CoT). Across AIME, CMIMC, BRUMO, and AIMO_AMC, the method improves direct CoT on multiple backbones. We further report convention-controlled routing experiments, matched prompting baselines, problem-level paired tests, decoding robustness, constraint-quality audits, total-token accounting, and an OlympiadBench evaluation. These analyses position CFR as a targeted test-time intervention whose benefit depends on recoverable constraints and reliable Stage 1 extraction, rather than as a general-purpose replacement for mathematical reasoning.
Chinese Translation
大型语言模型可以推导出一个合理的数学对象，但仍可能违反明确的要求——例如，遗漏模数约简、返回非整数或使用错误的编码答案形式。我们提出了约束优先推理（Constraint-First Reasoning, CFR），这是一种无训练的两阶段提示协议：第一阶段提取并总结问题所涉及的约束，第二阶段在解决问题时检查中间和最终结果是否符合该总结。Routed-CFR 仅在文本正则表达式路由器检测到限制性线索时激活两阶段协议；否则，它使用直接的思维链（chain-of-thought, CoT）。在 AIME、CMIMC、BRUMO 和 AIMO_AMC 上，该方法在多个基础模型上改善了直接 CoT 的表现。我们进一步报告了控制约定的路由实验、匹配的提示基线、问题级配对测试、解码鲁棒性、约束质量审计、总令牌计数和 OlympiadBench 评估。这些分析将 CFR 定位为一种针对性的测试时干预，其效益依赖于可恢复的约束和可靠的第一阶段提取，而不是作为数学推理的一种通用替代方案。
cs.CL / 21 / 2608.05353
Evidence Lock Before Commitment: A Frozen Interface Degrades LLM-as-Judge Evaluation
承诺前的证据锁定：冻结接口降低了大型语言模型作为评判者的评估效果
Abstract
LLM judges are often asked to extract criteria and evidence before choosing between candidate answers. This workflow assumes that the intermediate record preserves the information needed for a later verdict. For reasoning-capable models, visible field order does not reveal internal decision order, so we test an observable alternative: persist the evidence in one call and make it the exclusive input to the next. Across 24,000 judgments over HelpSteer3, FeedbackQA, and CoVal, we compare standard pairwise judging, structured one-call judging, two-call evidence locking, and three-call pointwise locking with Claude Sonnet 4.5 and GPT-5. Evidence locking reduces agreement with released human preferences by 4 to 6 percentage points and increases answer-order inconsistency by 8 to 10 points relative to structured one-call judging. Pointwise locking is also harmful, while structured evidence elicitation remains close to standard judging. The result holds for both judges and all three datasets. Persisted evidence can support auditability, but it should not replace the source answers at decision time.
Chinese Translation
大型语言模型（LLM）评判者通常被要求在选择候选答案之前提取标准和证据。这一工作流程假设中间记录保留了后续裁决所需的信息。对于具备推理能力的模型，显式字段顺序并不能揭示内部决策顺序，因此我们测试了一种可观察的替代方案：在一次调用中持久化证据，并将其作为下一次调用的唯一输入。在对 HelpSteer3、FeedbackQA 和 CoVal 的 24,000 次判断中，我们比较了标准的成对判断、结构化的一次调用判断、两次调用的证据锁定和三次调用的逐点锁定，使用了 Claude Sonnet 4.5 和 GPT-5。证据锁定使得与发布的人类偏好的一致性降低了 4 到 6 个百分点，并且相较于结构化的一次调用判断，答案顺序不一致性增加了 8 到 10 个百分点。逐点锁定同样有害，而结构化的证据引导仍然接近于标准判断。该结果适用于所有评判者和所有三个数据集。持久化的证据可以支持可审计性，但在决策时不应替代源答案。
cs.CL / 22 / 2608.05364
The interface of intonation and lexical tone: Boundary phenomena in Mandarin varieties
语调与声调的界面：普通话方言中的边界现象
Abstract
This chapter explores the intricate interplay between intonation and tone in Mandarin Chinese varieties, focusing on f0, the primary acoustic cue for both intonation and tone. The main empirical base is intonation boundary phenomena, where intonation and tone intersect and influence each other in conveying a range of sentence-level linguistic functions -- such as question vs. statement -- and a rich array of speakers' attitudinal information. Theoretical models and emerging techniques are also discussed to account for the observed interactions of tonal aspects and boundary phenomena to convey multiple levels of communicative meanings.
Chinese Translation
本章探讨了普通话方言中语调与声调之间复杂的相互作用，重点关注 f0，这是语调和声调的主要声学线索。主要的实证基础是语调边界现象，在这些现象中，语调与声调相互交织并相互影响，从而传达一系列句子层面的语言功能——例如疑问句与陈述句的区别——以及丰富的说话者态度信息。本文还讨论了理论模型和新兴技术，以解释声调特征与边界现象的观察到的相互作用，从而传达多层次的交际意义。
cs.CL / 23 / 2608.05409
Mood Matters: How Syntactic Sensitivity Undermines Safety Alignment
情绪重要：句法敏感性如何削弱安全对齐
Abstract
Large language models typically undergo post-training to align them with safety policies but there exist many sophisticated jailbreaks that sidestep established safeguards. For instance, prior work by Andriushchenko et al. (2025) has found that changing the grammatical tense from present to past can be enough to elicit harmful responses. In this work, we uncover a more general failure of non-imperative syntactic forms. We demonstrate that this syntactic vulnerability exists in 16 models up to 70B parameters, using behavioral evaluation. To investigate the root cause, we apply causal mediation analysis, finding that refusal is partially conditioned on upstream syntactic features. By steering these purely syntactic features we are able to trigger and suppress refusal. Finally, we trace this ill-conditioning to linguistically biased post-training data of open-source models and show that increasing syntactic diversity can mitigate the issue. Our findings suggest that current alignment approaches introduce confounders that prevent a pure semantic grounding of the refusal decision.
Chinese Translation
大型语言模型通常经过后期训练以使其与安全政策对齐，但存在许多复杂的越狱方法可以规避既定的安全措施。例如，Andriushchenko 等人（2025）的先前研究发现，将语法时态从现在时改为过去时可能足以引发有害反应。在本研究中，我们揭示了一种更普遍的非命令性句法形式的失败。我们通过行为评估展示了这种句法脆弱性在多达 70B 参数的 16 个模型中存在。为了调查根本原因，我们应用因果中介分析，发现拒绝反应在一定程度上受上游句法特征的影响。通过引导这些纯粹的句法特征，我们能够触发和抑制拒绝反应。最后，我们追溯到开源模型的语言偏见后期训练数据，并表明增加句法多样性可以缓解这一问题。我们的发现表明，当前的对齐方法引入了混杂因素，阻碍了拒绝决策的纯语义基础。
cs.CL / 24 / 2608.05447
Example-Guided Prompting for Document-Level Text Simplification
基于示例引导的文档级文本简化
Abstract
Document-level text simplification requires large language models (LLMs) to rewrite complex documents while preserving meaning, readability, and discourse coherence. Although prompt-based LLMs have shown promising performance, they often produce inconsistent simplifications because textual instructions alone provide limited guidance for complex document-level transformations. We investigate whether retrieved document-simplification examples can improve document-level generation by augmenting prompts with examples selected from a parallel simplification corpus. This example-guided prompting approach enables LLMs to exploit relevant simplification patterns without task-specific fine-tuning. Experiments on the OneStopEnglish corpus using multiple state-of-the-art LLMs show that incorporating retrieved examples consistently improves simplification quality over prompt-only generation and achieves competitive or superior performance compared with representative supervised (T5) and planning-based (PlanSimp) document simplification systems. Furthermore, we find that the benefits of example-guided prompting vary across LLMs, suggesting that effective use of retrieved examples depends on a model's ability to integrate contextual information during generation.
Chinese Translation
文档级文本简化要求大型语言模型（LLMs）在保留意义、可读性和话语连贯性的同时重写复杂文档。尽管基于提示的LLMs表现出良好的性能，但它们往往会产生不一致的简化，因为仅靠文本指令对复杂的文档级转换提供的指导有限。我们研究了检索的文档简化示例是否可以通过从平行简化语料库中选择示例来增强提示，从而改善文档级生成。这种基于示例引导的提示方法使LLMs能够利用相关的简化模式，而无需特定任务的微调。在使用多种最先进的LLMs对OneStopEnglish语料库进行的实验中，结果表明，纳入检索示例的生成在简化质量上始终优于仅基于提示的生成，并且与代表性的监督（T5）和基于规划（PlanSimp）的文档简化系统相比，表现出竞争力或更优的性能。此外，我们发现示例引导的提示在不同LLMs之间的效果有所不同，这表明有效利用检索示例依赖于模型在生成过程中整合上下文信息的能力。
cs.CL / 25 / 2608.05448
DBLAST: Dependent Block Drafting for Stochastic Speculative Decoding
DBLAST：用于随机推测解码的依赖块草拟
Abstract
Speculative decoding accelerates large language models' inference by using a lightweight drafter to propose multiple future tokens and a target model to verify them. While recent block and diffusion-style drafters can predict several positions in a single pass, their training and sampling procedures are typically optimized for greedy decoding or assume that positions in the draft block are conditionally independent. This assumption becomes brittle in non-greedy speculative decoding, where the target distribution is deliberately stochastic and multiple continuations become plausible. We study this mismatch for block diffusion drafters and show that the accepted draft length degrades as the entropy of the target sampling distribution increases. We propose a dependent block drafter based on a low-rank latent mixture over token positions, complemented by an acceptance-oriented training objective that directly targets the expected verified length. Experiments with Qwen3-4B and Qwen3-8B on GSM8K, MT-Bench, HumanEval, and creative-writing benchmarks show that our approach, namely DBLast, consistently improves accepted length over independent block sampling, especially in higher-entropy decoding regimes.
Chinese Translation
推测解码通过使用轻量级草拟器提出多个未来标记，并利用目标模型对其进行验证，从而加速大型语言模型的推理。尽管最近的块式和扩散式草拟器能够在一次传递中预测多个位置，但它们的训练和采样过程通常是针对贪婪解码进行优化的，或者假设草拟块中的位置是条件独立的。这一假设在非贪婪推测解码中变得脆弱，因为目标分布故意是随机的，多个延续变得合理。我们研究了块扩散草拟器的这种不匹配，并表明随着目标采样分布的熵增加，接受的草拟长度会降低。我们提出了一种基于标记位置的低秩潜在混合的依赖块草拟器，并辅以一个接受导向的训练目标，直接针对期望的验证长度。对GSM8K、MT-Bench、HumanEval和创意写作基准上的Qwen3-4B和Qwen3-8B的实验表明，我们的方法，即DBLast，始终在接受长度上优于独立块采样，尤其是在高熵解码环境中。
cs.CL / 26 / 2608.05510
Different Perturbations, Different Mechanisms: Understanding Continued Pre-training for Zero-Shot Dialect Robustness
不同的扰动，不同的机制：理解零-shot 方言鲁棒性的持续预训练
Abstract
Dialectal variation remains a major challenge for multilingual language models. Perturbation-based continued pre-training (CPT) has emerged as a promising approach to improving robustness, yet existing work largely evaluates individual perturbation strategies in isolation and provides limited insight into why they work. We present a systematic study of perturbation-based CPT for multilingual dialect robustness in LLMs, comparing six training conditions across nine German, Italian, and Arabic dialect tasks. Perturbation-based CPT, especially character-noised CPT, consistently improves zero-shot dialect robustness while largely preserving standard variety performance. More importantly, we show that methods with similar downstream performance induce distinct mechanisms of robustness, exhibiting different patterns of language model adaptation, representational alignment, and prediction repair. Our results provide a more complete understanding of how synthetic surface variation improves robustness and offer practical guidance for selecting CPT strategies in multilingual and dialectal settings.
Chinese Translation
方言变异仍然是多语言语言模型面临的主要挑战。基于扰动的持续预训练（CPT）已成为提高鲁棒性的有希望的方法，然而现有研究主要孤立地评估个别扰动策略，且对其有效性背后的原因提供的见解有限。我们对多语言方言鲁棒性的基于扰动的CPT进行了系统研究，比较了九个德语、意大利语和阿拉伯语方言任务中的六种训练条件。基于扰动的CPT，尤其是字符噪声CPT，始终提高了零-shot 方言鲁棒性，同时在很大程度上保持了标准变体的性能。更重要的是，我们展示了具有相似下游性能的方法引发了不同的鲁棒性机制，表现出语言模型适应、表示对齐和预测修复的不同模式。我们的结果提供了对合成表面变异如何提高鲁棒性的更全面理解，并为在多语言和方言环境中选择CPT策略提供了实用指导。
cs.CL / 27 / 2608.05576
Where Models Converge and Humans Diverge: A Coverage Framework for Distributional Pluralism in Open-Ended Generation
模型收敛与人类分歧：开放式生成中的分布多元化覆盖框架
Abstract
When a large language model (LLM) writes Harry Potter fanfiction, it reliably produces fundamental elements of the Hogwarts universe, such as recognizable places and characters. Human-written Harry Potter fanfictions, however, typically include these fundamentals and much more, incorporating stylistically irregular content and relationship-diverse plotlines. This gap between LLM and human writing has been noted across a variety of domains. LLMs tend to produce "average" writing, while human writing contains more diverse content that covers a broader distribution. Existing work has shown the existence of this distributional "gap", but no work has proposed a systematic way to measure it. Our paper proposes a human-grounded framework that uses the empirical distribution of human writing on a topic to measure the distributional breadth of LLM-generated content on that same topic. We propose two metrics, LLM Coverage (LLM-Cov) and In-Boundary Rate (IBR), that separate the plausibility of LLM content from its distributional breadth. Across ideation and narrative tasks, we find that current LLMs produce plausible but narrow content that concentrates near the center of the human response space. Our framework can enable researchers to better assess the distributional breadth of LLM-authored content, which we term its "cultural reach".
Chinese Translation
当大型语言模型（LLM）创作哈利·波特的同人小说时，它可靠地生成霍格沃茨宇宙的基本元素，如可识别的地点和角色。然而，人类创作的哈利·波特同人小说通常不仅包括这些基本元素，还融入了风格上不规则的内容和关系多样的情节线。这种LLM与人类写作之间的差距在多个领域中都得到了注意。LLM倾向于生成“平均”的写作，而人类写作则包含更多样化的内容，覆盖更广泛的分布。现有研究已表明这种分布“差距”的存在，但尚无研究提出系统性的方法来衡量它。我们的论文提出了一种以人类为基础的框架，利用人类在某一主题上的写作经验分布来衡量LLM生成内容在该主题上的分布广度。我们提出了两个指标，LLM覆盖率（LLM Coverage, LLM-Cov）和边界内比率（In-Boundary Rate, IBR），用于将LLM内容的可信度与其分布广度区分开。在创意生成和叙事任务中，我们发现当前的LLM生成了可信但狭窄的内容，集中在与人类响应空间中心附近。我们的框架可以帮助研究人员更好地评估LLM创作内容的分布广度，我们称之为其“文化覆盖范围”。
cs.CL / 28 / 2608.05604
SkillZip: Contract-Preserving Graph Compression for Scalable Agent Skill Libraries
SkillZip：用于可扩展代理技能库的合同保持图压缩
Abstract
Large Language Models (LLMs) increasingly act as agents whose procedural knowledge is stored in reusable skill packages and loaded at inference time. As skill libraries grow, a central challenge is to expose the smallest sufficient executable context under a limited context budget. Existing systems struggle to reuse routines below the whole-skill level, preserve procedural contracts during compression, keep compressed routines executable and expandable, and update the compressed library as skills evolve. These challenges reveal a unit mismatch: skills are retrieved as packages, compressed as text, and converted into execution graphs only after retrieval, whereas reliable reuse requires a contract-bearing procedural unit. We propose SkillZip, an execution-aware procedural abstraction framework that performs contract-preserving compression over section-level graphs. SkillZip rewrites recurring contract-valid motifs into reversible ported macros while preserving boundary signatures, dependency closure, verifier reachability, and source-level expansion. At inference time, it hydrates a compact, dependency-closed context and expands macros only when required. ReZip further integrates new skills and revises risky macros using execution evidence. Comprehensive experiments1 on technical and embodied agent benchmarks show SkillZip consistently outperforms the strongest baseline by up to 12.2 points, while achieving a 3.46x compression ratio with 99.2% dependency preservation and 98.7% verifier reachability. Scaling analyses further confirm robust retrieval across skill libraries ranging from 200 to 100K skills.
Chinese Translation
大型语言模型（LLMs）越来越多地作为代理，其程序知识存储在可重用的技能包中，并在推理时加载。随着技能库的增长，一个核心挑战是如何在有限的上下文预算下暴露出最小的足够可执行上下文。现有系统在技能级别以下重用例程、在压缩过程中保持程序合同、保持压缩例程的可执行性和可扩展性，以及随着技能演变更新压缩库方面面临困难。这些挑战揭示了一个单位不匹配的问题：技能作为包被检索，以文本形式被压缩，并且仅在检索后转换为执行图，而可靠的重用需要一个承载合同的程序单元。我们提出了SkillZip，一个执行感知的程序抽象框架，它对节段级图执行合同保持压缩。SkillZip将重复的合同有效模式重写为可逆的移植宏，同时保持边界签名、依赖闭合、验证器可达性和源级扩展。在推理时，它充实一个紧凑的、依赖闭合的上下文，并仅在需要时扩展宏。ReZip进一步整合新技能，并使用执行证据修订风险宏。在技术和具身代理基准上的全面实验表明，SkillZip始终比最强基线高出最多12.2分，同时实现了3.46倍的压缩比，99.2%的依赖保持率和98.7%的验证器可达性。规模分析进一步确认了在200到10万技能范围内的技能库中稳健的检索。
cs.CL / 29 / 2608.05611
FOCUS: Decoupling Expert Personas in LLMs to Enhance Domain Expert Capabilities
FOCUS：解耦大型语言模型中的专家角色以增强领域专家能力
Abstract
Large Language Models (LLMs) can exhibit diverse personas, and activating expert personas has been shown to improve domain expertise and task accuracy. However, existing persona control methods often suffer from cross-domain coupling, which may lead to overly aggressive behavior in high-caution domains such as healthcare, or excessive conservatism in risk-sensitive domains such as financial trading. To address this issue, we propose FOCUS (\textbf{\underline{F}}ine-tuning with \textbf{\underline{O}}rthogonal \textbf{\underline{C}}ontrol for \textbf{\underline{U}}ncoupled persona\textbf{\underline{S}}). FOCUS first automatically extracts expert persona vectors from LLMs, then applies orthogonal decomposition to decouple domain-specific expert personas, and finally introduces an expert gating module to adaptively control persona activation according to task contexts. With a two-stage training strategy and a gated selection regularizer, the model learns to activate appropriate personas for both single-domain and cross-domain tasks. Experiments on financial, legal, medical, and cross-domain benchmarks show that FOCUS improves task accuracy and outperforms existing persona control methods. Our code is available at \href{https://anonymous.4open.science/r/openpersona-48F4}{this url}.
Chinese Translation
大型语言模型（LLMs）可以展现出多样化的角色，激活专家角色已被证明能够提高领域专业知识和任务准确性。然而，现有的角色控制方法往往受到跨领域耦合的影响，这可能导致在高谨慎领域（如医疗保健）中表现出过于激进的行为，或在风险敏感领域（如金融交易）中表现出过于保守的行为。为了解决这一问题，我们提出了FOCUS（ extbf{ extit{F}}ine-tuning with extbf{ extit{O}}rthogonal extbf{ extit{C}}ontrol for extbf{ extit{U}}ncoupled persona extbf{ extit{S}}）。FOCUS首先从LLMs中自动提取专家角色向量，然后应用正交分解以解耦领域特定的专家角色，最后引入专家门控模块，根据任务上下文自适应地控制角色激活。通过两阶段训练策略和门控选择正则化器，模型学习在单领域和跨领域任务中激活适当的角色。在金融、法律、医疗和跨领域基准测试中的实验表明，FOCUS提高了任务准确性，并且优于现有的角色控制方法。我们的代码可在 exttt{https://anonymous.4open.science/r/openpersona-48F4}找到。
cs.CL / 30 / 2608.05630
Human-Like Anaphor Resolution in Large Language Models
大型语言模型中的类人指代消解
Abstract
Anaphors are expressions that refer to other expressions, called antecedents. The process of connecting the two is called resolution. Cognitive science has identified multiple factors that affect the speed and success of anaphor resolution, including discourse structure, situation-model properties, and semantic factors. Here, we investigate whether these factors also affect anaphor resolution in five Large Language Models (LLMs) with open weights: GPT-2-XL, Llama-3.1-8B, Pythia-12B, Mistral-7B, and Mistral-24B. To model processing difficulty, we adopt the standard linking hypothesis that relates human reading times to model surprisal at the anaphor. As a second behavioral measure, we compare model accuracy to human accuracy on comprehension questions probing the antecedents of anaphors. The results show selective cognitive alignment: some LLMs exhibit human-like sensitivity to discourse prominence and distance-based factors in anaphor resolution, while showing weaker or absent sensitivity to semantic interference effects. These findings delimit the conditions under which LLMs approximate human anaphor resolution.
Chinese Translation
指代是指向其他表达的表达，称为前指。将两者连接的过程称为消解。认知科学已确定多个影响指代消解速度和成功率的因素，包括话语结构、情境模型特性和语义因素。在此，我们研究这些因素是否也会影响五种具有开放权重的大型语言模型（LLMs）的指代消解：GPT-2-XL、Llama-3.1-8B、Pythia-12B、Mistral-7B 和 Mistral-24B。为了建模处理难度，我们采用标准的链接假设，将人类阅读时间与模型在指代处的惊讶度相关联。作为第二个行为测量，我们比较模型在探测指代前指的理解问题上的准确性与人类的准确性。结果显示出选择性的认知一致性：一些LLMs在指代消解中对话语突出性和基于距离的因素表现出类人敏感性，而对语义干扰效应的敏感性则较弱或缺失。这些发现界定了LLMs近似人类指代消解的条件。
cs.CL / 31 / 2608.05651
Relay, Don't Route: Adaptive Population Handoff for Cost-Efficient LLM-Driven Evolution
中继，而非路由：成本高效的 LLM 驱动演化的自适应种群切换
Abstract
Large language model (LLM)-driven evolution has shown promise for program search and algorithm discovery, but relying on strong models throughout long evolutionary runs is costly. A natural alternative is to combine cheap and strong models under a fixed inference budget. However, existing approaches typically allocate models at the level of individual queries or mutation steps, overlooking that evolutionary search is \textit{stateful}: each generated candidate changes the population from which subsequent mutations are produced. We empirically analyze LLM-driven evolutionary trajectories and find that search progress is strongly front-loaded, early trajectory performance is informative but noisy, and cheap models recover much of the early progress achieved by strong models at lower cost. Motivated by these findings, we propose \textbf{\model}, a training-free framework that shifts budget allocation from individual calls to evolving populations through adaptive \textit{population handoff}. A cheap model explores multiple trajectories in short blocks allocated by a bandit scheduler. Relay Gain, defined as the marginal improvement of a compact, quality-diverse candidate bank constructed for handoff, serves as the scheduler reward and determines when to hand off. The curated candidates initialize a shared strong model population for refinement. Across four benchmarks and three budgets, \model achieves the highest mean score in 11 of 12 settings, outperforming competitive baselines. Our results suggest that in stateful search, budget allocation should be organized around the population, not the individual call.
Chinese Translation
大型语言模型（LLM）驱动的演化在程序搜索和算法发现方面展现了潜力，但在长时间的演化过程中依赖强模型的成本较高。一种自然的替代方案是结合廉价和强大的模型，在固定的推理预算下进行操作。然而，现有的方法通常在单个查询或变异步骤的层面上分配模型，忽视了演化搜索是 extit{有状态的}这一事实：每个生成的候选者都会改变后续变异所产生的种群。我们对 LLM 驱动的演化轨迹进行了实证分析，发现搜索进展在初期阶段显著，早期轨迹的表现具有信息性但噪声较大，而廉价模型在较低成本下恢复了强模型所取得的大部分早期进展。基于这些发现，我们提出了 extbf{ extit{model}}，一个无训练框架，通过自适应 extit{种群切换}将预算分配从单个调用转移到演化种群。廉价模型在由强盗调度器分配的短时间块中探索多个轨迹。Relay Gain 被定义为为切换构建的紧凑且质量多样的候选库的边际改进，作为调度器的奖励并决定何时进行切换。经过筛选的候选者初始化一个共享的强模型种群以进行优化。在四个基准和三个预算下， extit{model} 在 12 种设置中取得了 11 种的最高平均得分，超越了竞争基线。我们的结果表明，在有状态的搜索中，预算分配应围绕种群进行，而非单个调用。
cs.CL / 32 / 2608.05687
Answer First, Reason Later: Commitment Order in Diffusion LLMs
先回答，再推理：扩散大语言模型中的承诺顺序
Abstract
Masked diffusion language models (dLLMs) can commit tokens in any order -- a freedom marketed as their core advantage over autoregressive decoding. We show that on reasoning tasks this freedom is instead the axis of failure. Logging every commitment during decoding of LLaDA-8B on GSM8K, we find that unconstrained (pure) decoding commits the final answer at 15-24% of the trajectory while half the reasoning region is still masked, and collapses to answer-only outputs on up to 90% of problems as the canvas grows. The cause is not the model's termination beliefs -- EOS "pressure" is nearly identical across decoders -- but reachability: whether the sampler may act on those beliefs at distant positions. A 2x2 prompt-decoder design shows that chain-of-thought helps only under ordered commitment (interaction +34.8 percentage points, 95% CI [26.8, 42.8]; without reasoning text the decoders are indistinguishable), an interaction we decompose into a collapse channel and an order channel and replicate on Dream-7B and MATH-500. A single-knob intervention -- frontier-gated commitment -- causally recovers the full gap (0.528 to 0.852) while preserving up to 4x parallel decoding, along a measured frontier whose optimal window flips from w=1 at full refinement to unconstrained at 8 tokens/step. Our results reframe existing window-style samplers, previously motivated by efficiency, as the minimal fix for a reasoning pathology they were never designed to address.
Chinese Translation
掩蔽扩散语言模型（dLLMs）可以以任意顺序承诺标记——这种自由被视为其相对于自回归解码的核心优势。我们发现，在推理任务中，这种自由反而成为失败的轴心。在对LLaDA-8B在GSM8K上的解码过程中记录每次承诺时，我们发现无约束（纯）解码在轨迹的15-24%处承诺最终答案，而一半的推理区域仍然被掩蔽，并且在画布扩展时，答案输出在多达90%的问题上崩溃。造成这一现象的原因并非模型的终止信念——EOS“压力”在不同解码器之间几乎相同——而是可达性：采样器是否可以在远处位置对这些信念进行操作。一个2x2的提示-解码器设计显示，思维链仅在有序承诺下才有帮助（交互提升34.8个百分点，95%置信区间[26.8, 42.8]；在没有推理文本的情况下，解码器无法区分），我们将这一交互分解为崩溃通道和顺序通道，并在Dream-7B和MATH-500上进行了复制。一个单一的干预——前沿门控承诺——因果性地恢复了完整的差距（从0.528到0.852），同时保持了高达4倍的并行解码，沿着一个测量的前沿，其最优窗口在完全细化时从w=1翻转到在8个标记/步时无约束。我们的结果重新构架了现有的窗口式采样器，这些采样器之前是出于效率动机而设计的，作为解决其从未设计来应对的推理病理的最小修复。
cs.CL / 33 / 2608.05724
Sparse Mutual Information Graph Averaging for Improving Random Indexing Embeddings
稀疏互信息图平均化以改善随机索引嵌入
Abstract
Sparse word embedding pipelines can avoid dense co-occurrence matrix materialization, dense factorization, and gradient training while still relying on sparse global corpus statistics. This paper studies Random Indexing (RI) vectors refined by weighted averaging on a sparse Positive Pointwise Mutual Information (PPMI) graph. On a fairytales corpus, the covered semantic analogy set consists of 272 Google family- category questions. On this family subset, PPMI top-K graph averaging repairs a weak RI initialization, improving accuracy from 19.4+-0.7% to 30.7+-2.9% across five seeds. Under the single tested runs, the same neighborhood averaging reduces family- subset analogy accuracy for PPMI+SVD (singular value decom- position), Binary+SVD, CBOW, and Skip-gram. Thus the method is not competitive with neural baselines on text8 and gives near- zero strict similarity correlation on SimLex-999. While Bloom filter sketches underperform RI in the tested configuration, we find that PPMI graph averaging with top-K pruning is a useful non-gradient repair for weak RI embeddings. On the fairytales dataset, PPMI top-K=50 graph averaging improves RI with accuracy going from 19.4+-0.7% to 30.7+-2.9%, and performing best with a seed42 of 34.6%.
Chinese Translation
稀疏词嵌入管道可以避免密集共现矩阵的实现、密集分解和梯度训练，同时仍依赖于稀疏的全局语料库统计。本文研究了通过在稀疏正点互信息（PPMI）图上进行加权平均来优化的随机索引（RI）向量。在一个童话语料库上，涵盖的语义类比集包含272个谷歌家庭类别问题。在这个家庭子集上，PPMI top-K图平均化修复了一个较弱的RI初始化，使准确率从19.4±0.7%提高到30.7±2.9%，在五个种子下均如此。在单次测试运行中，相同的邻域平均化降低了PPMI+SVD（奇异值分解）、Binary+SVD、CBOW和Skip-gram的家庭子集类比准确率。因此，该方法在text8上的表现不及神经基线，并且在SimLex-999上几乎没有严格的相似性相关性。尽管布隆过滤器草图在测试配置下表现不及RI，我们发现PPMI图平均化与top-K剪枝是对弱RI嵌入的有用非梯度修复。在童话数据集上，PPMI top-K=50图平均化使RI的准确率从19.4±0.7%提高到30.7±2.9%，并在种子42下表现最佳，达到34.6%。
cs.CL / 34 / 2608.05726
Mitigating Scoring Bias in LLM-as-a-Judge via Random Number Generation
通过随机数生成减轻 LLM 作为评判者的评分偏差
Abstract
Large Language Models (LLMs) are often used as evaluators of text quality, known as LLM-as-a-Judge, which can outperform conventional automatic evaluation metrics that rely on reference texts. However, LLM evaluators tend to generate particular scores regardless of the context of the evaluated text, which is known as scoring bias. This study proposes a novel method to mitigate this scoring bias. An LLM is instructed to randomly generate number tokens, and the latent numerical bias of the LLM is identified by measuring the deviation of the observed distribution of numbers from the uniform distribution. A definition of a downstream task, for which an LLM evaluator is used, is added to the prompts for random number generation to measure task-specific latent number bias. In the evaluation by an LLM, the token generation probabilities for a given input are rectified considering the LLM's latent number bias. Results of the experiment on four different tasks, evaluation of LLM alignment, evaluation of summarization, Semantic Textual Similarity, and Semantic Textual Relatedness, demonstrate that our proposed method outperforms the baselines, including an LLM without debiasing and previous calibration methods. In addition, it is confirmed that scoring bias varies across LLMs, tasks, and score ranges, indicating the importance of measuring latent number bias as the case may be.
Chinese Translation
大型语言模型（LLMs）常被用作文本质量的评估者，称为 LLM 作为评判者，其表现优于依赖参考文本的传统自动评估指标。然而，LLM 评估者往往会生成特定的分数，而不考虑被评估文本的上下文，这被称为评分偏差。本研究提出了一种新颖的方法来减轻这种评分偏差。我们指示 LLM 随机生成数字标记，并通过测量观察到的数字分布与均匀分布的偏差来识别 LLM 的潜在数字偏差。为了测量特定任务的潜在数字偏差，我们在随机数生成的提示中添加了一个下游任务的定义，该任务使用 LLM 评估者。在 LLM 的评估中，考虑到 LLM 的潜在数字偏差，对给定输入的标记生成概率进行了修正。在四个不同任务上的实验结果，包括 LLM 对齐评估、摘要评估、语义文本相似性和语义文本相关性，证明我们提出的方法优于基线，包括未去偏差的 LLM 和以前的校准方法。此外，研究确认评分偏差在不同的 LLM、任务和分数范围之间存在差异，这表明在特定情况下测量潜在数字偏差的重要性。
cs.CL / 35 / 2608.05741
Once a Response, Always a Response: Detecting LLM-generated Text via Latent Prompt Restoration
一旦响应，便永远响应：通过潜在提示恢复检测LLM生成的文本
Abstract
Large language models (LLMs) can generate fluent and convincing text at scale, creating growing risks for misinformation dissemination, educational misuse, and platform governance. These concerns make robust detection of machine-generated text increasingly necessary. Recent zero-shot detectors mainly exploit probability-based statistical discrepancies, but they do not explicitly account for the training process of LLMs, which leaves a distinct generation mechanism insufficiently modeled and limits detection robustness. To address this issue, we propose EchoPrompt, a training-free detector based on latent prompt restoration. Our key intuition is that machine-generated text is typically produced conditioned on an upstream prompt, and this hidden dependency can be partially reactivated by prepending a unified generic prefix. Specifically, EchoPrompt restores a generic assistant-response context, measures the induced likelihood gain with an instruction-tuned model, calibrates it against the corresponding base model, and aggregates the resulting differences into a score that quantifies latent prompt dependency. Extensive experiments show that EchoPrompt achieves state-of-the-art performance among zero-shot detectors while maintaining strong robustness across challenging evaluation settings.
Chinese Translation
大型语言模型（LLMs）能够大规模生成流畅且令人信服的文本，这带来了日益增长的信息传播风险、教育误用和平台治理问题。这些担忧使得机器生成文本的稳健检测变得愈发必要。近期的零-shot 检测器主要利用基于概率的统计差异，但它们并未明确考虑LLMs的训练过程，这导致了生成机制的独特性未被充分建模，从而限制了检测的稳健性。为了解决这一问题，我们提出了EchoPrompt，一种基于潜在提示恢复的无训练检测器。我们的核心直觉是，机器生成的文本通常是在上游提示的条件下产生的，而这种隐藏的依赖关系可以通过在前面添加统一的通用前缀部分地重新激活。具体而言，EchoPrompt恢复了通用的助手响应上下文，利用指令调优模型测量诱导的可能性增益，并将其与相应的基础模型进行校准，最后将产生的差异汇总为一个量化潜在提示依赖性的分数。大量实验表明，EchoPrompt在零-shot 检测器中实现了最先进的性能，同时在具有挑战性的评估设置中保持了强大的稳健性。
cs.CL / 36 / 2608.05759
How to Recognize New Words: A Comparison Between Context Biasing Methods and Speech LLMs
如何识别新词：上下文偏置方法与语音大型语言模型的比较
Abstract
Recognizing new and rare words - named entities, acronyms, domain specific special words, and other items scarce in training data - remains a key challenge for automatic speech recognition (ASR). We compare two strategies for this: context biasing methods, where an ASR model is extended such that during inference a word list can be supplied, and speech large language models (LLMs) prompted with context directly. We evaluate two context biasing methods based on Whisper against three speech LLMs across read and non-read speech, reporting biased, unbiased, and overall word error rate (WER). The context biasing methods cut biased WER by up to 88% relative while leaving other words largely unaffected. Speech LLMs excel on read speech but generalize less well to non-read speech, and prove sensitive to distractor count and prompt word order. We characterize the resulting trade-offs to guide method selection.
Chinese Translation
识别新词和稀有词汇——命名实体、缩略词、特定领域的专业词汇以及其他在训练数据中稀缺的项目——仍然是自动语音识别（ASR）的一项关键挑战。我们比较了两种策略：上下文偏置方法，其中扩展了ASR模型，使得在推理过程中可以提供一个词汇列表，以及直接用上下文提示的语音大型语言模型（LLMs）。我们评估了基于Whisper的两种上下文偏置方法与三种语音LLMs在朗读和非朗读语音中的表现，报告了偏置、非偏置和整体词错误率（WER）。上下文偏置方法在偏置WER方面相对降低了多达88%，而对其他词汇的影响则相对较小。语音LLMs在朗读语音中表现优异，但在非朗读语音中的泛化能力较差，并且对干扰词数量和提示词顺序敏感。我们描述了由此产生的权衡，以指导方法选择。
cs.CL / 37 / 2608.05785
Task-Conditional Flow Matching for Balanced Multilingual Text Embedding Adaptation
任务条件流匹配用于平衡多语言文本嵌入适应
Abstract
Multilingual text embedding models are commonly adapted using a single training objective across diverse tasks, despite different tasks requiring fundamentally different optimization strategies. We introduce Task-Conditional Flow Matching (TCFM), a multilingual embedding adaptation framework that selectively applies Flow Matching to translation tasks while optimizing retrieval, classification, and pair-classification tasks with objectives better aligned to their learning dynamics. TCFM further combines teacher-guided representation preservation with a three-stage curriculum to enable stable adaptation. Evaluated on the Indic Massive Text Embedding Benchmark, TCFM establishes a new state-of-the-art, consistently improving embedding quality across a diverse set of multilingual tasks and generalizing across embedding model families. We will publicly release the codebase and datasets upon acceptance of the paper.
Chinese Translation
多语言文本嵌入模型通常通过在不同任务中使用单一训练目标进行适应，尽管不同任务在本质上需要不同的优化策略。我们提出了任务条件流匹配（Task-Conditional Flow Matching, TCFM），这是一个多语言嵌入适应框架，它选择性地将流匹配应用于翻译任务，同时对检索、分类和配对分类任务进行优化，使用与其学习动态更为一致的目标。TCFM 进一步结合了教师引导的表示保留和三阶段课程，以实现稳定的适应。在印度大规模文本嵌入基准测试中评估，TCFM 建立了新的最先进水平，在多种多语言任务中持续提高嵌入质量，并在嵌入模型系列中实现泛化。我们将在论文接受后公开发布代码库和数据集。
cs.CL / 38 / 2608.05802
On-Policy Delta Distillation for Multilingual Math Reasoning
用于多语言数学推理的在线策略增量蒸馏
Abstract
On-Policy Distillation (OPD) is emerging as a promising alternative to reinforcement learning for LLM post-training, yet its effectiveness in multilingual settings remains underexplored. We study OPD and its advanced variant, On-Policy Delta Distillation (OPD$^2$), for mathematical reasoning in English, Korean, and Japanese. OPD$^2$ improves OPD by using the probability gap between a post-trained teacher and its base model as the learning signal. Experiments with Qwen3 show that OPD$^2$ consistently outperforms the original OPD, with particularly strong improvements in Korean and Japanese, and generally narrows the English-Korean performance gap. We further find that English-only OPD can also increase performance for Korean and Japanese, but often shifts the responses toward English, highlighting the importance of multilingual data to preserving target-language responses.
Chinese Translation
在线策略蒸馏（On-Policy Distillation, OPD）作为强化学习在大型语言模型（LLM）后训练中的一种有前景的替代方案，然而其在多语言环境中的有效性仍然未被充分探索。我们研究了OPD及其高级变体在线策略增量蒸馏（On-Policy Delta Distillation, OPD$^2$）在英语、韩语和日语的数学推理中的应用。OPD$^2$通过利用后训练教师模型与其基础模型之间的概率差距作为学习信号，从而改进了OPD。与Qwen3的实验结果表明，OPD$^2$始终优于原始的OPD，尤其在韩语和日语方面表现出显著的提升，并且通常缩小了英语与韩语之间的性能差距。我们进一步发现，仅使用英语的OPD也可以提高韩语和日语的性能，但往往会使响应向英语倾斜，强调了多语言数据在保留目标语言响应中的重要性。
cs.CL / 39 / 2608.05806
Hierarchical Latent Prediction for Language Models
语言模型的层次潜在预测
Abstract
While standard Next-Token Prediction (NTP) lays the foundation of language model pre- training, its teacher-forced training paradigm may not be optimal for long-horizon reasoning and planning. Recent works such as Multi-Token Prediction (MTP) and Next-Latent prediction (NextLat) try to mitigate the problem through predicting multiple future tokens and self-supervised prediction in the latent space. However, those auxiliary objectives either have a limited horizon or suffer from compounding error from multi-step rollout. We introduce Hierarchical Latent Prediction (HiLP), which introduces an auxiliary higher-level abstract latent to help reduce the error accumulation effect in latent-space rollouts. Experiments show that HiLP can lead to longer-horizon coherent belief state representation and demonstrate the effectiveness of our method across coding and multi-step reasoning benchmarks, and offers more speculative decoding efficiency.
Chinese Translation
虽然标准的下一个标记预测（Next-Token Prediction, NTP）为语言模型的预训练奠定了基础，但其教师强制训练范式可能并不适合长时间范围的推理和规划。近期的研究如多标记预测（Multi-Token Prediction, MTP）和下一个潜在预测（Next-Latent Prediction, NextLat）试图通过预测多个未来标记和在潜在空间中的自监督预测来缓解这一问题。然而，这些辅助目标要么具有有限的时间范围，要么受到多步骤展开带来的累积误差的影响。我们提出了层次潜在预测（Hierarchical Latent Prediction, HiLP），引入了一个辅助的更高层次抽象潜在变量，以帮助减少潜在空间展开中的误差累积效应。实验表明，HiLP能够实现更长时间范围内一致的信念状态表示，并在编码和多步骤推理基准测试中展示了我们方法的有效性，同时提供了更高效的推测解码能力。
cs.CL / 40 / 2608.05817
M$^3$R-Bench: A Unified Benchmark for Evidence-Grounded Multimodal Metaphor Understanding
M$^3$R-Bench：一个统一的基于证据的多模态隐喻理解基准
Abstract
Metaphor enables the understanding of abstract concepts through cross-domain mappings while conveying affective attitudes. In multimodal scenarios, visual and textual information jointly construct Target--Source mappings, requiring both conceptual understanding and cross-modal reasoning. However, existing benchmarks mainly evaluate metaphor understanding through isolated subtasks and lack evidence-grounded explanations, making it difficult to assess whether models establish mappings grounded in visual and textual cues.To address these limitations, we introduce M$^3$R-Bench, a unified and evidence-grounded benchmark containing 1,000 image--text instances with human-verified annotations. Guided by Conceptual Metaphor Theory and theories of nonliteral language understanding, M$^3$R-Bench provides joint annotations for metaphor occurrence, Target--Source mapping, sentiment, and stage-wise explanations following ``evidence identification--mapping establishment--sentiment inference.''Evaluations on M$^3$R-Bench reveal that existing models often overlook visual evidence, rely on superficial textual cues, and produce inaccurate Target--Source mappings, exposing a cross-modal evidence--mapping mismatch. To address this mismatch, we propose M$^3$R-Reasoner, which combines curriculum-based reasoning supervision with task-aware reinforcement learning to align model reasoning with metaphor interpretation. Experiments show that, with only an 8B-parameter backbone, M$^3$R-Reasoner outperforms larger proprietary MLLMs across four unified-task metrics and improves Visual Evidence and Sentiment Justification scores over GPT-5.5 by 28.45 and 30.11 points, respectively, while surpassing Claude-Sonnet-4.6 by 8.00 points in mean rubric score. The dataset and code are available at https://github.com/hongshi4/M3R-Bench.
Chinese Translation
隐喻通过跨领域映射使抽象概念的理解成为可能，同时传达情感态度。在多模态场景中，视觉和文本信息共同构建目标-源映射，这需要概念理解和跨模态推理。然而，现有的基准主要通过孤立的子任务评估隐喻理解，缺乏基于证据的解释，使得评估模型是否建立了基于视觉和文本线索的映射变得困难。为了解决这些局限性，我们提出了M$^3$R-Bench，这是一个统一的、基于证据的基准，包含1,000个经过人工验证注释的图像-文本实例。在概念隐喻理论和非字面语言理解理论的指导下，M$^3$R-Bench提供了隐喻出现、目标-源映射、情感以及遵循“证据识别-映射建立-情感推断”的阶段性解释的联合注释。在M$^3$R-Bench上的评估显示，现有模型常常忽视视觉证据，依赖表面的文本线索，并产生不准确的目标-源映射，暴露了跨模态证据-映射的不匹配。为了解决这一不匹配，我们提出了M$^3$R-Reasoner，它结合了基于课程的推理监督和任务感知的强化学习，以使模型推理与隐喻解释对齐。实验表明，仅使用一个8B参数的主干，M$^3$R-Reasoner在四个统一任务指标上超越了更大的专有多语言大模型（MLLM），并在视觉证据和情感合理性评分上分别提高了28.45和30.11分，相比之下，在平均评分上超越Claude-Sonnet-4.6达8.00分。数据集和代码可在 https://github.com/hongshi4/M3R-Bench 获取。
cs.CL / 41 / 2608.05823
Decomposed Entailment for Factuality Checking and Hallucination Detection
分解蕴涵用于事实性检查和幻觉检测
Abstract
The reliability of Large Language Models (LLMs) is often compromised by factual inconsistencies, including hallucinations---cases where generated content is not supported by the underlying source. We present HallDetect, a lightweight, reference-free, and black-box framework for hallucination detection that we evaluate not only on summarization but across a broader range of source-grounded generation settings. HallDetect builds on decomposition-based factuality evaluation: generated content is decomposed into atomic claims, each verified by a compact encoder-based entailment model through a contrastive formulation over a multi-scale library of source chunks, and aggregated with an asymmetric score in which a single confidently contradicted claim flags the response. Under a controlled protocol in which all methods share the same 4-bit quantized backbones and consumer-grade hardware budget, HallDetect outperforms comparably resourced generative and embedding-based baselines on three of four benchmarks while remaining stable across backbone families, and yields a claim-to-span audit trail that localizes each error.
Chinese Translation
大型语言模型（LLMs）的可靠性常常受到事实不一致性的影响，包括幻觉——即生成内容未得到基础来源支持的情况。我们提出了HallDetect，这是一种轻量级、无参考和黑箱的幻觉检测框架，我们不仅在摘要生成上进行了评估，还在更广泛的基于来源的生成设置中进行了测试。HallDetect基于分解的事实性评估：生成的内容被分解为原子声明，每个声明通过基于紧凑编码器的蕴涵模型进行验证，该模型通过对多尺度来源片段库的对比形式进行处理，并通过不对称评分进行聚合，其中单个自信地被否定的声明会标记该响应。在一个控制协议下，所有方法共享相同的4位量化骨干网络和消费级硬件预算，HallDetect在四个基准中的三个上超越了资源相当的生成和嵌入基线，同时在骨干网络系列中保持稳定，并生成一个声明到跨度的审计轨迹，以定位每个错误。
cs.CL / 42 / 2608.05825
MoCA: Implicit Social Context Analysis
MoCA：隐性社会语境分析
Abstract
Human social communication, such as affection and intent, is often conveyed in highly implicit ways, where underlying meanings are expressed through indirect, socially and culturally grounded signals rather than explicit statements. Such implicit social contexts are pervasive in real-world interactions, yet there remains a lack of a formal and systematic framework for studying them. In this paper, we introduce Implicit Social Context Analysis (MoCA), a novel task that systematically models implicit social scenarios along three key dimensions: affection, intent, and stance. We construct a high-quality benchmark containing 3,108 multimodal instances collected from real-world sources, with fine-grained cognitive annotations revealing who expresses what toward whom, as well as how and why it is conveyed. Using the MoCA dataset, we show that state-of-the-art multimodal large language models struggle significantly with this task because of their reliance on explicit cues and limited ability to reason over latent social contexts. To address this challenge, we propose Conflict-Driven Abductive Reasoning (CoDAR), a novel framework that models the discrepancy between observed expressions and expected truthful behavior as cognitive conflict, thereby enabling the inference of hidden mental states. Extensive experiments demonstrate that CoDAR substantially improves model performance. Nevertheless, a large gap from human reasoning remains, highlighting the fundamental difficulty of implicit social understanding.
Chinese Translation
人类社会交流，如情感和意图，通常以高度隐性的方式传达，其中潜在的意义通过间接的、社会和文化基础的信号而非明确的陈述来表达。这种隐性社会语境在现实世界的互动中普遍存在，但目前仍缺乏一个正式且系统的框架来研究它们。在本文中，我们介绍了隐性社会语境分析（MoCA），这是一项新颖的任务，系统地沿着三个关键维度建模隐性社会场景：情感、意图和立场。我们构建了一个高质量的基准数据集，包含3,108个来自现实世界来源的多模态实例，配有细粒度的认知注释，揭示了谁对谁表达了什么，以及如何和为什么传达这些信息。利用MoCA数据集，我们展示了最先进的多模态大型语言模型在此任务上面临显著挑战，因为它们依赖于明确的线索，并且在推理潜在社会语境方面能力有限。为了解决这一挑战，我们提出了基于冲突驱动的溯因推理（CoDAR）框架，该框架将观察到的表达与预期的真实行为之间的差异建模为认知冲突，从而使得推断隐藏的心理状态成为可能。大量实验表明，CoDAR显著提高了模型的性能。然而，与人类推理之间仍存在较大差距，突显了隐性社会理解的根本困难。
cs.CL / 43 / 2608.05832
Enhancing Social Intelligence in LLMs with Hierarchical Reasoning and Utterance-Level Goal Rewarding
通过层次推理和发话级目标奖励增强大型语言模型的社会智能
Abstract
Large language models (LLMs) excel in structured tasks but struggle with dynamic social interactions, where success requires long-term goal coordination and rapid adaptation. Current methods often apply uniform goal-based rewards to every utterance, overlooking the specificity of objectives at each dialogue turn and failing to account for the rationale of potential strategies. Inspired by the Theory of Planned Behavior, we propose the Think-Strategy-Response (TSR) framework, which decomposes social dialogue into two hierarchical stages: high-level strategic planning and low-level linguistic execution. To optimize TSR, we introduce Linearized Hierarchical Reinforcement Learning with Variance-Gated Rewards (LHRL-VGR), a novel algorithm that dynamically routes rewards - balancing goal completion and strategy adherence - based on the variance of goal achievement scores. Experiments on the SOTOPIA benchmark show that our approach fine-tunes a Qwen2.5-7B agent to surpass the GPT-4o baseline by 7.32% in goal completion success, demonstrating state-of-the-art performance in multi-agent social negotiation tasks.
Chinese Translation
大型语言模型（LLMs）在结构化任务中表现出色，但在动态社交互动中却面临挑战，成功需要长期目标协调和快速适应。目前的方法通常对每个发话应用统一的基于目标的奖励，忽视了每个对话轮次中目标的特异性，并未考虑潜在策略的合理性。受到计划行为理论的启发，我们提出了思考-策略-反应（Think-Strategy-Response, TSR）框架，该框架将社会对话分解为两个层次的阶段：高层次的战略规划和低层次的语言执行。为了优化TSR，我们引入了一种新算法——线性化层次强化学习与方差门控奖励（Linearized Hierarchical Reinforcement Learning with Variance-Gated Rewards, LHRL-VGR），该算法根据目标实现分数的方差动态调整奖励，平衡目标完成和策略遵循。在SOTOPIA基准上的实验表明，我们的方法使Qwen2.5-7B代理的目标完成成功率超越了GPT-4o基线7.32%，在多智能体社会协商任务中展示了最先进的性能。
cs.CL / 44 / 2608.05850
MameLoshnLM: Yiddish Language Model and Evaluation Benchmark
MameLoshnLM：意第绪语言模型及评估基准
Abstract
We present MameLoshnLM, the first open-source 8B-parameter language model built specifically for Yiddish. Despite Yiddish's rich textual tradition, its limited digital presence and the scarcity of reliable evaluation resources have constrained progress in Yiddish language modeling. Existing multilingual corpora and benchmarks are often poor proxies for the language, containing substantial amounts of noisy, machine-translated, and misclassified text. We address these gaps by introducing Oytser, a high-quality Yiddish pretraining corpus that combines contemporary web-native sources with literary materials, and Kashes, a multi-task benchmark spanning translation, linguistic analysis, information extraction, and language understanding. Using these resources, we continue pretraining Llama 3.1 8B to obtain MameLoshnLM. Across the tasks in the benchmark, MameLoshnLM outperforms open baselines of similar scale. Our analyses show that these gains are not only quantitative: relative to general-purpose multilingual models, MameLoshnLM better captures language-defining lexical and morphological patterns, pointing to a broader failure mode of noisy web-scale multilingual data for low-resource languages. Our results provide both a foundation for Yiddish NLP and a practical template for language model development in historically rich but digitally underrepresented languages.
Chinese Translation
我们提出了MameLoshnLM，这是第一个专门为意第绪语构建的开源8B参数语言模型。尽管意第绪语拥有丰富的文本传统，但其有限的数字化存在和可靠评估资源的匮乏限制了意第绪语言建模的进展。现有的多语言语料库和基准往往无法有效代表该语言，包含大量嘈杂的、机器翻译的和错误分类的文本。我们通过引入Oytser，一个高质量的意第绪语预训练语料库，结合当代网络原生来源与文学材料，以及Kashes，一个涵盖翻译、语言分析、信息提取和语言理解的多任务基准，来解决这些问题。利用这些资源，我们继续对Llama 3.1 8B进行预训练，以获得MameLoshnLM。在基准中的各项任务中，MameLoshnLM的表现超越了相似规模的开放基线。我们的分析表明，这些提升不仅是定量的：与通用多语言模型相比，MameLoshnLM更好地捕捉了定义语言的词汇和形态模式，指向了嘈杂的网络规模多语言数据在低资源语言中的更广泛失效模式。我们的结果为意第绪自然语言处理提供了基础，并为在历史丰富但数字化代表性不足的语言中进行语言模型开发提供了实用模板。
cs.CL / 45 / 2608.05857
Mapping Similarity Spaces across Embedding Models with Synthetic Query Probing
通过合成查询探测在嵌入模型之间映射相似性空间
Abstract
Retrieval-Augmented Generation systems rely on similarity scores to retrieve relevant content, yet scores are not directly comparable across embedding models due to differing geometric properties, complicating model migration and limiting threshold reuse. We study how similarity scores can be related by learning mappings between score distributions rather than embeddings. We introduce Synthetic Query Probing, generating queries from documents to create controlled query-chunk pairs, enabling large-scale, reference-free analysis of cross-model similarity behavior. We evaluate the approach on multiple embedding configurations and learn score conversion functions using linear, isotonic, and quantile mappings. Experiments on SciFact and a proprietary corpus show that while models largely agree on rankings, their absolute scores exhibit systematic distortions. Learned mappings partially align these spaces and improve threshold portability, with isotonic regression performing best. Our results highlight the need for cross-model calibration and position Synthetic Query Probing as a scalable framework for analyzing embedding comparability.
Chinese Translation
检索增强生成系统依赖于相似性分数来检索相关内容，但由于嵌入模型之间几何属性的不同，这些分数无法直接比较，从而使模型迁移变得复杂并限制了阈值的重用。我们研究了如何通过学习分数分布之间的映射而不是嵌入来关联相似性分数。我们引入了合成查询探测（Synthetic Query Probing），从文档生成查询以创建受控的查询-块对，从而实现对跨模型相似性行为的大规模、无参考分析。我们在多个嵌入配置上评估该方法，并使用线性、同调（isotonic）和分位数（quantile）映射学习分数转换函数。在SciFact和一个专有语料库上的实验表明，尽管模型在排名上大体一致，但它们的绝对分数表现出系统性的扭曲。学习到的映射部分对齐了这些空间，并改善了阈值的可移植性，其中同调回归表现最佳。我们的结果强调了跨模型校准的必要性，并将合成查询探测定位为分析嵌入可比性的可扩展框架。
cs.CL / 46 / 2608.05872
MACRO: Markov Chain Routing of Transformer Layers
MACRO：变压器层的马尔可夫链路由
Abstract
Standard Large Language Models (LLMs) execute layers sequentially. Dynamic layer routing, i.e. search for a different execution path through layers involving layer repetitions, skips and other moves, can improve performance. Existing routing approaches often require updating model weights, running expensive search loops per test instance, or demand ground-truth labels during inference. In this work, we propose Markov Chain Routing of Transformer Layers (MACRO), a framework that learns task-specific routes over LLM architectures without modifying underlying parameters. MACRO models layer routing as a context-dependent Markov policy conditioned on layer indices, computation budget phases, directional displacements, and operator context, supporting skip, repeat, and residual hidden-state addition operations. The Markov route distribution is updated via feedback on training data and decoded using a top-k Viterbi algorithm to isolate high-probability candidate programs. We evaluate MACRO across diverse reasoning and knowledge benchmarks on multiple open-weight LLMs. MACRO achieves a +5.0% average accuracy improvement over the unrouted baselines, with largest gains on small models. We outperform the best dynamic routing approach Dr. LLM by +7.2%, while reducing route-search time 9.4x (from 14.8 to 1.6 hours). Our code is publicly available at https://github.com/Batorskq/MACRO.
Chinese Translation
标准的大型语言模型（LLMs）按顺序执行层。动态层路由，即通过层的不同执行路径进行搜索，包括层的重复、跳过和其他移动，可以提高性能。现有的路由方法通常需要更新模型权重，在每个测试实例上运行昂贵的搜索循环，或在推理过程中要求真实标签。在本研究中，我们提出了变压器层的马尔可夫链路由（MACRO），这是一个在不修改底层参数的情况下学习特定任务路由的框架。MACRO将层路由建模为依赖上下文的马尔可夫策略，该策略以层索引、计算预算阶段、方向位移和操作上下文为条件，支持跳过、重复和残差隐藏状态加法操作。马尔可夫路径分布通过对训练数据的反馈进行更新，并使用top-k Viterbi算法进行解码，以隔离高概率候选程序。我们在多个开放权重的LLMs上评估MACRO，涵盖多种推理和知识基准。MACRO在未路由基线的基础上实现了+5.0%的平均准确率提升，尤其在小模型上获得了最大的增益。我们在动态路由方法Dr. LLM的基础上超越了+7.2%，同时将路由搜索时间减少了9.4倍（从14.8小时降至1.6小时）。我们的代码已公开可用，网址为https://github.com/Batorskq/MACRO。
cs.CL / 47 / 2608.05906
Causal Episodic Memory for Feedback-Driven Agent Repair
基于反馈驱动的代理修复的因果情节记忆
Abstract
LLM agents that repair failures often discard successful corrections, forcing later episodes to rediscover similar solutions. We study whether finalized repair outcomes can improve subsequent Text-to-SQL episodes without parameter updates. We introduce MERIT, a training-free agent that maintains an online dual-polarity memory of oracle-verified corrections and observed unsuccessful directions. Under oracle-assisted benchmark feedback, only memories from earlier finalized episodes are eligible for retrieval. A deterministic classifier assigns a coarse failure type, which conditions a hybrid lexical-dense retriever before the frozen model generates each revision. Using Qwen2.5-7B-Instruct with identical initial predictions and repair budgets, MERIT improves execution accuracy over stateless iterative repair from \(66.34\%\) to \(69.79\%\) on Spider and from \(47.35\%\) to \(48.44\%\) on BIRD. Paired analyses provide clear evidence for the Spider gain but weaker evidence on BIRD. MERIT is not reliably separated from untyped dynamic retrieval on either benchmark, while Reflexion-style memory reaches \(51.24\%\) on BIRD at substantially higher inference cost. Ablations show that negative memory contributes modestly, the value of type conditioning and lexical--dense ranking is dataset dependent, and schema-local experience provides the most consistent benefit. These results clarify when causal cross-query memory improves repair and when broader memory representations remain preferable.
Chinese Translation
修复故障的LLM代理通常会丢弃成功的修正，迫使后续情节重新发现类似的解决方案。我们研究了最终修复结果是否能够在不更新参数的情况下改善后续的文本到SQL（Text-to-SQL）情节。我们引入了MERIT，一个无需训练的代理，维护一个在线的双极性记忆，记录经过oracle验证的修正和观察到的不成功方向。在oracle辅助的基准反馈下，只有早期最终情节的记忆可以被检索。一个确定性的分类器分配粗略的故障类型，这为混合词汇-密集检索器提供条件，然后在冻结模型生成每次修订之前进行处理。使用具有相同初始预测和修复预算的Qwen2.5-7B-Instruct，MERIT在Spider上的执行准确率从66.34%提高到69.79%，在BIRD上的准确率从47.35%提高到48.44%。配对分析为Spider的提升提供了明确证据，但在BIRD上证据较弱。MERIT在这两个基准上与未分类的动态检索没有显著区别，而Reflexion风格的记忆在BIRD上达到了51.24%的准确率，但推理成本显著更高。消融实验表明，负记忆的贡献适度，类型条件和词汇-密集排名的价值依赖于数据集，而模式局部经验提供了最一致的收益。这些结果阐明了何时因果交叉查询记忆改善修复，以及何时更广泛的记忆表示仍然更可取。
cs.CL / 48 / 2608.05993
Clinical Communication Processing with Models Trained on LLM-Generated Synthetic Data: A Structured Survey and Novel Application Case Studies
基于大语言模型生成的合成数据训练的临床沟通处理：结构化调查与新应用案例研究
Abstract
Much clinical value is conveyed not through structured records but through communication: exchanges in which patients describe symptoms, clinicians reason and give instructions, ambulances hand over to emergency departments, and nurses pass on a shift. Such language differs from tabular data because meaning depends on speaker role, intent, causality, uncertainty, omission, and channel noise. Healthcare natural language processing must therefore interpret information as conveyed rather than coded. This requires well-annotated corpora, which are scarce because authentic exchanges are private, fragmented, and costly to annotate. Large language models offer a way forward by transforming clinical sources, such as records, diagnostic labels, symptom lists, or care plans, into written and transcribed communication for downstream models. We present a structured narrative survey organized by source representation, communication form and participants, generation method, and downstream task, complemented by thirteen novel case studies. These build clinical NLP systems for communication channels and languages without labeled real-world data, including EMS pre-arrival reports, field-radio casualty documentation, nurse handoffs, patient-portal triage, and low-resource discharge communication. They show that synthetic communication can bootstrap such systems. Findings include the competitiveness of fine-tuned encoder models over evaluated zero-shot baselines and the value of deliberately degraded communication for robustness. The main limitation is that most studies evaluate on held-out synthetic communication, while train-on-synthetic, test-on-authentic evidence remains limited. We conclude that syn-thetic clinical communication is becoming a practical research resource; establishing it as reusable clinical infrastructure will require authentic-data transfer, safety and external validation.
Chinese Translation
许多临床价值并非通过结构化记录传达，而是通过沟通实现：患者描述症状、临床医生推理并给出指示、救护车向急救部门交接、护士进行班次交接等交流。这种语言与表格数据不同，因为其意义依赖于说话者角色、意图、因果关系、不确定性、遗漏和渠道噪声。因此，医疗自然语言处理必须解释传达的信息，而非编码的信息。这需要良好标注的语料库，但由于真实交流是私密的、碎片化的，并且标注成本高，相关语料库十分稀缺。大型语言模型为此提供了一种解决方案，通过将临床来源（如记录、诊断标签、症状列表或护理计划）转化为书面和转录的沟通，以供下游模型使用。我们呈现了一项结构化叙述调查，按来源表示、沟通形式和参与者、生成方法及下游任务进行组织，并补充了十三个新案例研究。这些案例构建了用于沟通渠道和语言的临床自然语言处理系统，而无需标注的真实世界数据，包括EMS到达前报告、现场无线电伤员文档、护士交接、患者门户分诊和低资源出院沟通。结果表明，合成沟通可以为这些系统提供启动支持。研究发现包括微调的编码器模型在评估的零样本基线上的竞争力，以及故意降级沟通对鲁棒性的价值。主要限制在于大多数研究在保留的合成沟通上进行评估，而在合成数据上训练、在真实证据上测试的研究仍然有限。我们得出结论，合成临床沟通正成为一种实用的研究资源；将其建立为可重用的临床基础设施将需要真实数据的转移、安全性和外部验证。
cs.CL / 49 / 2608.06022
EpiBench: Can LLMs Understand Epitopes for Antibody Drug Discovery?
EpiBench：大型语言模型能理解抗体药物发现中的表位吗？
Abstract
Epitopes determine where antibodies bind antigens and shape downstream therapeutic properties such as functional blockade and escape resistance, making epitope understanding central to antibody drug discovery. Although large language models (LLMs) have shown strong biomedical reasoning ability, it remains unclear whether they can infer epitope information directly from antigen and antibody sequences. Existing epitope resources typically focus on isolated prediction tasks or rely on specialized structural settings, while general protein benchmarks do not evaluate epitope-centered decisions across the antibody development workflow. To address this gap, we introduce EpiBench, a closed-book, sequence-based, and automatically scorable benchmark for evaluating epitope reasoning in LLMs. EpiBench contains 1,609 curated samples grounded in structural antibody--antigen contacts, curated functional B-cell assays, and deep mutational scanning escape measurements. It covers five connected tasks: targetable region discovery, antibody-conditioned epitope identification, epitope binning, functional epitope assessment, and antibody escape assessment, with controlled sampling to reduce shortcut-based evaluation artifacts. We evaluate nine general-purpose LLMs and analyze their behavior through task-specific baselines, antigen length stratification, explicit-reasoning comparison, and failure-mode inspection. The results show that current LLMs capture partial epitope-related signals but remain limited in antibody-specific sequence grounding, long-context residue localization, and biologically grounded reasoning. Therefore, EpiBench provides a diagnostic testbed for measuring and improving sequence-aware biomedical LLMs toward reliable LLM-assisted antibody discovery.
Chinese Translation
表位决定了抗体与抗原结合的位置，并影响下游的治疗特性，如功能阻断和逃逸抗性，因此理解表位对于抗体药物发现至关重要。尽管大型语言模型（LLMs）在生物医学推理方面表现出强大的能力，但尚不清楚它们是否能够直接从抗原和抗体序列中推断表位信息。现有的表位资源通常专注于孤立的预测任务或依赖于专业的结构设置，而一般的蛋白质基准测试并未评估抗体开发工作流程中的表位中心决策。为了解决这一空白，我们引入了EpiBench，这是一个封闭书籍、基于序列且可自动评分的基准，用于评估LLMs在表位推理方面的能力。EpiBench包含1,609个基于结构抗体-抗原接触、策划的功能性B细胞测定和深度突变扫描逃逸测量的样本。它涵盖了五个相关任务：可靶向区域发现、抗体条件下的表位识别、表位分组、功能性表位评估和抗体逃逸评估，并通过控制采样减少基于捷径的评估伪影。我们评估了九个通用LLMs，并通过任务特定基线、抗原长度分层、显式推理比较和失败模式检查分析它们的行为。结果表明，当前的LLMs捕获了部分与表位相关的信号，但在抗体特定序列的基础、长上下文残基定位和生物学基础推理方面仍然有限。因此，EpiBench为测量和改进序列感知的生物医学LLMs提供了一个诊断测试平台，以助力可靠的LLM辅助抗体发现。
cs.CL / 50 / 2608.06027
FormBharo: Designing and Evaluating a Voice Agent for Conversational Form Filling in Rural India
FormBharo：为印度农村设计和评估一个用于对话式表单填写的语音代理
Abstract
In India, almost every social benefit starts with a form, yet the people who need these benefits most are often unable to read or write. Reaching them requires a spoken conversation. Today that work falls to frontline health workers who enroll beneficiaries one at a time, a poor use of stretched capacity. We built FormBharo ("fill the form" in Hindi), a voice agent that fills a structured form over a phone call under tight latency and cost budgets by pairing Large Language Models (LLMs) with deterministic, rule-based validation and flow control. It is being piloted with ARMMAN, an NGO running large-scale maternal and child mobile-health programs in India, to enroll low-income, Hindi-speaking mothers in antenatal and postnatal care. To our knowledge, it is the first voice agent piloted to fill an enrollment form for this population. We openly release FormVoiceAgentBench, a benchmark pairing human-recorded Hindi audio with 3,760 multi-turn conversation tests across 960 simulated calls, to evaluate our agent's components (transcription, extraction, reply generation) and end-to-end form completion under real acoustic variations. Form completion drops by up to ~41 points when LLMs receive error-prone real-speech transcripts instead of reference ones. The rule-based controls recover many turn-level extraction errors, helping smaller, cheaper models match or surpass frontier models on form completion. Component performance does not predict end-to-end performance: GPT-5.5 leads turn-level extraction accuracy on reference transcripts (99.8%) but ranks lower on form completion. Since errors both propagate and cancel across the pipeline, the optimal model choice of models emerges only through end-to-end evaluation. Finally, no single model is best across accuracy, cost, and latency at once, so we use a Pareto-based weighted-sum scalarization to select a deployable configuration balancing the three.
Chinese Translation
在印度，几乎所有的社会福利都始于一份表单，但最需要这些福利的人往往无法阅读或书写。要接触到他们，需要进行口头交流。如今，这项工作落在了前线卫生工作者的身上，他们一次只能登记一位受益人，这对紧张的工作能力来说是一种低效的利用。我们构建了FormBharo（在印地语中意为“填写表单”），这是一个在电话通话中填充结构化表单的语音代理，能够在严格的延迟和成本预算下，通过将大型语言模型（LLMs）与确定性、基于规则的验证和流程控制相结合来实现。该系统正在与ARMMAN合作试点，该非政府组织在印度开展大规模的母婴移动健康项目，旨在为低收入、讲印地语的母亲提供产前和产后护理。根据我们的了解，这是首个针对该人群试点的用于填写登记表的语音代理。我们公开发布了FormVoiceAgentBench，这是一个基准数据集，将人类录制的印地语音频与3,760个多轮对话测试相结合，涵盖960个模拟通话，以评估我们代理的各个组件（转录、提取、回复生成）以及在真实声学变化下的端到端表单完成情况。当LLMs接收到错误率较高的真实语音转录而非参考转录时，表单完成率下降了约41个百分点。基于规则的控制机制能够恢复许多轮级提取错误，帮助较小、成本更低的模型在表单完成上与前沿模型相匹配或超越。组件性能并不能预测端到端性能：GPT-5.5在参考转录上领先于轮级提取准确率（99.8%），但在表单完成上排名较低。由于错误在整个流程中既会传播又会相互抵消，最佳模型选择仅通过端到端评估才能显现。最后，没有单一模型能够在准确性、成本和延迟上同时表现最佳，因此我们使用基于帕累托的加权和标量化方法来选择一个可部署的配置，以平衡这三者。
cs.CL / 51 / 2608.06069
Training-Free Token-Level Steering for LLM Personalized Co-Writing
无训练的令牌级引导用于大型语言模型个性化协同写作
Abstract
While Large Language Models (LLMs) show great promise for personalization, they often lack specialized domain knowledge. Conventional solutions like fine-tuning struggle with high computational costs and rapid data updates, while Retrieval-Augmented Generation fails to provide fine-grained, token-level steering. Furthermore, chat-based interfaces remain dominant, whereas productive co-writing paradigms have not yet been well exploited beyond the coding domain. To this end, we introduce SteerWrite, a training-free framework designed for personalized co-writing. Our method effectively adapts the base model to specialized domains without gradient updates, with specific designs tailored to small datasets. Experiments demonstrate that SteerWrite achieves state-of-the-art performance across diverse datasets, metrics, and models, significantly reducing human editing effort.
Chinese Translation
尽管大型语言模型（LLMs）在个性化方面展现出巨大潜力，但它们往往缺乏专业领域知识。传统的解决方案如微调面临着高计算成本和快速数据更新的挑战，而检索增强生成（Retrieval-Augmented Generation）无法提供细粒度的令牌级引导。此外，基于聊天的界面仍然占主导地位，而在编码领域之外，富有成效的协同写作范式尚未得到充分利用。为此，我们提出了SteerWrite，一个旨在个性化协同写作的无训练框架。我们的方法有效地将基础模型适应于专业领域，而无需梯度更新，并针对小型数据集进行了特定设计。实验表明，SteerWrite在多样化的数据集、指标和模型上实现了最先进的性能，显著减少了人工编辑的工作量。
cs.CL / 52 / 2608.06111
Beyond Sequence Order: Syntax-Informed Positional Embeddings for Transformers
超越序列顺序：基于语法的Transformer位置嵌入
Abstract
Positional embeddings (PE) in Transformers encode token distance and order but are largely agnostic to \textit{syntactic structure}. We introduce \textbf{S}yntax-\textbf{i}nformed \textbf{P}ositional \textbf{E}mbeddings (\textbf{SiPE}), which learns a lightweight syntactic prior from dependency parses during pretraining and injects it across all three dominant PE families (absolute, relative, rotary), for both encoders and decoders, leaving self-attention and the rest of the architecture untouched. We isolate \emph{where} and \emph{how} the prior should enter the model, and find it depends on the architecture: for autoregressive decoders that use relative PE, the prior is strongest when coupled multiplicatively with the relative-position term of the attention score, outperforming injection into the input embeddings, into self-attention, or into the positional and attention terms jointly---while for encoders it is best added directly to the input embeddings, composing with each encoder's native positional mechanism. We find that models pre-trained with SiPE improve on the SyntaxGym benchmark by up to $10.3\%$ while simultaneously reducing perplexity by $9.0\%$ over a base model with no syntactic supervision---a metric nearly every existing syntax-injection method instead degrades. Crucially, these gains extend beyond syntactic generalization: SiPE also improves real-world language understanding, raising scores on the GLUE benchmark by up to $8.2\%$ over a model trained without it. Unlike existing syntactic language models that marginalize over many parses at inference or discard syntax at runtime, SiPE conditions on a single parse, establishing a new Pareto frontier between syntactic supervision and inference cost.
Chinese Translation
Transformer中的位置嵌入（PE）编码了标记的距离和顺序，但对 extit{句法结构}基本上是无关的。我们提出了 extbf{S}yntax- extbf{i}nformed extbf{P}ositional extbf{E}mbeddings（ extbf{SiPE}），该方法在预训练期间从依赖解析中学习轻量级的句法先验，并将其注入到所有三种主要的PE家族（绝对、相对、旋转）中，适用于编码器和解码器，同时保持自注意力机制和其余架构不变。我们明确了 extit{何处}和 extit{如何}将先验引入模型，并发现这取决于架构：对于使用相对PE的自回归解码器，当先验与注意力分数的相对位置项乘法结合时，其效果最强，优于注入到输入嵌入、自注意力或位置与注意力项的联合注入——而对于编码器，最好直接将其添加到输入嵌入中，与每个编码器的本地位置机制结合。我们发现，使用SiPE预训练的模型在SyntaxGym基准测试中提高了高达$10.3\%$的性能，同时在没有句法监督的基础模型上将困惑度降低了$9.0\\%$——几乎所有现有的句法注入方法在这一指标上都表现不佳。重要的是，这些提升超越了句法泛化：SiPE还改善了现实世界的语言理解，使得GLUE基准测试的得分提高了高达$8.2\\%$，相较于未使用该方法的模型。与现有的句法语言模型在推理时对多个解析进行边际化或在运行时丢弃句法不同，SiPE依赖于单一解析，从而在句法监督和推理成本之间建立了新的帕累托前沿。
cs.CL / 53 / 2608.06141
Decolonizing Linguistic Policies in Automated Speech Recognition: A Framework for Cross-Culturally Competent Speech AI
去殖民化自动语音识别中的语言政策：跨文化能力语音人工智能框架
Abstract
This paper focuses on automatic speech recognition (ASR) and ASR-mediated voice interfaces that shape access to public services, healthcare, and education. We argue that persistent failures for low-resource, Indigenous, and non-standard language varieties are not only technical errors, but also implicit linguistic policies that reproduce colonial language hierarchies. Drawing on linguistic capital, raciolinguistic ideology, language policy research, and decolonial computing, we show how data, metrics, and model priors determine whose voices become machine-legible. We introduce the Three Harms (3M) taxonomy---Misrecognition, Misalignment, and Mistrust---and a seven-layer situatedness model for linguistic diversity in ASR and ASR-mediated voice interfaces. We then propose a participatory framework and minimum audit protocol for culturally competent ASR, positioning affected communities as co-designers, evaluators, and governance partners.
Chinese Translation
本文聚焦于自动语音识别（ASR）及其介导的语音接口，这些接口影响着公共服务、医疗保健和教育的获取。我们认为，对于低资源、土著和非标准语言变体的持续失败不仅是技术错误，还隐含着再现殖民语言等级的语言政策。基于语言资本、种族语言意识形态、语言政策研究和去殖民计算，我们展示了数据、指标和模型先验如何决定哪些声音能够被机器识别。我们引入了三种伤害（Three Harms, 3M）分类法——误识别（Misrecognition）、误对齐（Misalignment）和不信任（Mistrust），以及一个七层次的ASR及其介导的语音接口中的语言多样性情境模型。随后，我们提出了一个参与式框架和最低审计协议，以实现文化能力的ASR，将受影响的社区定位为共同设计者、评估者和治理伙伴。
cs.CL / 54 / 2608.06171
Routing Is Least Learnable Where It Is Most Valuable: Bounds on Representation Routing for Web Agents
路由在最有价值的地方最难学习：网络代理的表示路由界限
Abstract
Web agents observe a browser through text, pixels, or both, and the choice is usually fixed once for all tasks. We measure six observation modes across eight site-model combinations (cells) on VisualWebArena and WebArena and ask what choosing per task would buy. The modes are complementary: each solves tasks the others miss, they fail in structurally different ways, and the best choice reverses between task sets. The obvious prize, an oracle that picks a winning mode for every task, looks large but is inflated by run-to-run noise: rerunning the same mode on the same tasks changes 12-14% of outcomes, so a second run of a mode already in hand gains about as much as adding a new one. What survives is a cost bound: sending only the tasks no mode solves to the cheapest mode cuts cost by 9.5-30.6% in 8 of 8 cells at unchanged success. We then test five routing policies (picking the mode, deciding when to spend on the strong mode, a zero-cost rule read off the task text, a confidence cascade, and pooled cost tiers), and none robustly beats simply fixing one well-chosen mode; the one exception is a fragile result in our sparsest cell. The central obstruction is that routing supervision is produced at the agent's success rate: the weaker the agent, the fewer labels a router gets, exactly where routing would be most valuable. This limit belongs to today's agents rather than to routing itself. Label supply and routing opportunity rise together (correlation 0.95 across cells), so a stronger agent can overturn the result, and we report the rerun noise bands and the full measurement protocol.
Chinese Translation
网络代理通过文本、像素或两者观察浏览器，且这种选择通常在所有任务中是固定的。我们在 VisualWebArena 和 WebArena 上测量了六种观察模式与八种站点模型组合（单元）的组合，并探讨了按任务选择所能带来的收益。这些模式是互补的：每种模式解决了其他模式无法解决的任务，它们在结构上以不同的方式失败，而最佳选择在任务集之间会发生反转。显而易见的奖赏是一个能够为每个任务选择获胜模式的神谕，这看起来很大，但被运行间噪声所膨胀：在相同任务上重新运行相同模式会改变 12-14% 的结果，因此对已掌握模式的第二次运行所获得的收益大约与添加新模式相当。最终留下的是一个成本界限：将仅未被任何模式解决的任务发送给最便宜的模式，可以在 8 个单元中以不变的成功率降低 9.5-30.6% 的成本。随后，我们测试了五种路由策略（选择模式、决定何时在强模式上花费、从任务文本中读取的零成本规则、置信级联和汇总成本层级），但没有一种策略能够稳健地优于简单地固定一个精心选择的模式；唯一的例外是在我们最稀疏的单元中得到的脆弱结果。主要障碍在于路由监督是根据代理的成功率产生的：代理越弱，路由器获得的标签就越少，这正是路由最有价值的地方。这个限制属于今天的代理，而不是路由本身。标签供应和路由机会是共同上升的（在单元之间的相关性为 0.95），因此更强的代理可以推翻这一结果，我们报告了重新运行噪声带和完整的测量协议。
cs.CL / 55 / 2608.06292
NeSy-RAG: Neuro-Symbolic RAG for Explainable Question Answering
NeSy-RAG：用于可解释问答的神经符号RAG
Abstract
Retrieval-augmented generation (RAG) improves question answering by grounding large language models (LLMs) in external knowledge such as text corpora. However, its reasoning process remains largely opaque: intermediate reasoning steps are difficult to verify and cannot be reliably attributed to specific evidence. Moreover, missing user-specific context is rarely detected systematically, often leading to incomplete or incorrect output. We propose NeSy-RAG, a modular neuro-symbolic RAG framework that synthesizes attributable Prolog modules from retrieved text chunks. For each chunk, the system generates semantically meaningful predicates that encode Boolean claims, which may depend on user facts. Using joint natural language-code embeddings, predicates are retrieved and composed into Prolog queries. To address incomplete user context, we introduce a symbolic knowledge-gap detection mechanism that identifies missing user facts whose truth values affect the query outcome and automatically triggers follow-up interactions. Executing the resulting Prolog queries yields deterministic answers together with transparent execution traces that link each reasoning step to its originating source. On the ShARC benchmark, without domain-specific training, NeSy-RAG achieves 61.1% accuracy, outperforming a same-model RAG baseline that achieves 42.8% accuracy.
Chinese Translation
检索增强生成（RAG）通过将大型语言模型（LLMs）与外部知识（如文本语料库）结合，提升了问答能力。然而，其推理过程仍然在很大程度上不透明：中间推理步骤难以验证，且无法可靠地归因于特定证据。此外，缺失的用户特定上下文往往未被系统检测到，常导致输出不完整或不正确。我们提出了NeSy-RAG，一个模块化的神经符号RAG框架，该框架从检索到的文本块中合成可归因的Prolog模块。对于每个文本块，系统生成语义上有意义的谓词，这些谓词编码布尔声明，可能依赖于用户事实。通过联合自然语言-代码嵌入，谓词被检索并组合成Prolog查询。为了解决用户上下文不完整的问题，我们引入了一种符号知识缺口检测机制，识别缺失的用户事实，其真值会影响查询结果，并自动触发后续交互。执行生成的Prolog查询可产生确定性的答案，并提供透明的执行轨迹，将每个推理步骤与其来源链接。在ShARC基准测试中，NeSy-RAG在没有领域特定训练的情况下达到了61.1%的准确率，优于同模型RAG基线的42.8%准确率。
cs.CL / 56 / 2608.06312
Benchmarking and Enhancing LLMs for Rule-Intensive Review of National Standard Documents
基准测试与增强大型语言模型在国家标准文件规则密集审查中的应用
Abstract
Large language models (LLMs) increasingly support complex professional tasks, yet their capabilities in rule-intensive document review remain insufficiently evaluated. National standard documents, such as China GB/T standards, offer a representative testbed: they are lengthy, highly structured, and governed by explicit rules for scope, terminology, normative wording, and cross-section consistency. Existing benchmarks focus on domain knowledge and question answering, largely overlooking intrinsic quality review for professional documents. Such reviews rely heavily on human experts, making them costly and difficult to scale. To bridge this gap, we introduce GB/T-Bench, the first benchmark for the structured review of national standard documents. Its GB/T Review Taxonomy is a hierarchical schema covering document structure, scope alignment, normative modality, terminology consistency, and normative references, with 25 diagnosable error types. A controllable counterexample generation mechanism combines deterministic rules and constrained LLM rewriting to process 488 documents into 7,306 traceable review error instances for evaluation. We also develop a diagnosis-oriented evaluation protocol requiring exact matches on error location, review dimension, and error type, plus document-level coverage metrics. We further propose GB/T-Reviewer, a multi-agent framework that converts review knowledge into specialized skills and coordinates global inspection, targeted diagnosis, rule scanning, and result verification. Experiments with 14 mainstream LLMs reveal a substantial human-LLM gap: the strongest model achieves only 0.3280 CMCS versus 0.6640 for experts. GB/T-Reviewer raises the best CMCS to 0.5094, showing the value of structured skill coordination for rule-intensive document review. This work paves the way for trustworthy AI in standardization and other high-stakes document domains.
Chinese Translation
大型语言模型（LLMs）越来越多地支持复杂的专业任务，但它们在规则密集型文件审查中的能力仍然未得到充分评估。国家标准文件，如中国GB/T标准，提供了一个具有代表性的测试平台：这些文件篇幅较长、结构高度规范，并受到范围、术语、规范性措辞和交叉一致性等明确规则的约束。现有的基准测试主要关注领域知识和问答，往往忽视了对专业文件的内在质量审查。这类审查严重依赖人类专家，导致成本高昂且难以扩展。为填补这一空白，我们提出了GB/T-Bench，这是第一个针对国家标准文件结构化审查的基准测试。其GB/T审查分类法是一个层次化的框架，涵盖文档结构、范围一致性、规范性方式、术语一致性和规范性引用，共有25种可诊断的错误类型。可控的反例生成机制结合了确定性规则和约束性LLM重写，将488份文件处理为7,306个可追溯的审查错误实例以供评估。我们还开发了一种以诊断为导向的评估协议，要求在错误位置、审查维度和错误类型上进行精确匹配，并提供文档级覆盖度指标。此外，我们进一步提出了GB/T-Reviewer，这是一个多智能体框架，将审查知识转化为专业技能，并协调全球检查、针对性诊断、规则扫描和结果验证。对14个主流LLM的实验显示出显著的人类与LLM之间的差距：最强模型的CMCS仅为0.3280，而专家的CMCS为0.6640。GB/T-Reviewer将最佳CMCS提高至0.5094，展示了结构化技能协调在规则密集型文件审查中的价值。这项工作为标准化及其他高风险文件领域的可信AI铺平了道路。
cs.CL / 57 / 2608.06329
Benchmarking the Benchmarks: Evaluating Benchmarks for Conversational Agents
基准的基准：评估对话代理的基准
Abstract
Task-oriented conversational agents are evaluated using curated or automatically generated benchmarks, yet benchmark quality is rarely assessed. Poor benchmarks may contain inconsistent tasks, simplistic scenarios, or limited policy coverage, leading to unreliable evaluations. We introduce a reference-free framework that uses LLM judges to assess benchmark consistency, complexity, and policy coverage, while providing actionable diagnostics of weaknesses. We validate the framework by demonstrating agreement with independent human annotations and by evaluating benchmarks generated by LLMs of varying capabilities, as well as benchmarks subjected to controlled quality-degrading perturbations. Across domains and judge models, the proposed metrics consistently distinguish between benchmark quality levels. We further demonstrate the framework's applicability to manually curated benchmarks. Our framework offers a practical approach for evaluating synthetic and manually curated conversational-agent benchmarks.
Chinese Translation
任务导向的对话代理通常使用精心策划或自动生成的基准进行评估，但基准质量很少被评估。劣质基准可能包含不一致的任务、过于简单的场景或有限的策略覆盖，导致评估结果不可靠。我们引入了一种无参考框架，利用大型语言模型（LLM）评估基准的一致性、复杂性和策略覆盖，同时提供可操作的弱点诊断。我们通过展示与独立人类注释的一致性以及评估不同能力的LLM生成的基准和经过控制质量降级扰动的基准来验证该框架。在不同领域和评估模型中，所提出的指标始终能够区分基准质量水平。我们进一步展示了该框架在手动策划基准中的适用性。我们的框架为评估合成和手动策划的对话代理基准提供了一种实用的方法。
cs.CL / 58 / 2608.06347
RP-OPSD: Reasoning-Pivot-Guided On-Policy Self-Distillation for Multilingual Reasoning Transfer
RP-OPSD：基于推理枢轴的在线自蒸馏方法用于多语言推理迁移
Abstract
Multilingual reasoning transfer is crucial for extending reasoning capabilities of large language models (LLMs) beyond high-resource languages. On-policy self-distillation (OPSD) and its variants have emerged as a promising paradigm, providing dense token-level supervision on student-generated rollouts, yet their objectives do not explicitly prioritize reasoning signals most critical to cross-lingual transfer. We characterize that target-language reasoning comprises the generation of both surface text and reasoning pivots, which are decisions that advance or redirect the reasoning process and shape subsequent inference. This motivates concentrating privileged distillation around such pivots. We therefore propose RP-OPSD, Reasoning-Pivot-guided On-Policy Self-Distillation, using the distributional shift between matched teacher views with and without an English reference solution as an operational proxy to guide privileged distillation and reference anchoring. Experiments on mathematical reasoning benchmarks covering 17 languages and multiple difficulty levels show that our method outperforms strong multilingual reasoning baselines and OPSD variants. Further analysis reveals that RP-OPSD concentrates privileged distillation on reasoning-control and problem-condistioned state-update tokens, while downweighting it for tokens that mainly support surface realization. Our code is available at https://github.com/NJUNLP/RP-OPSD.
Chinese Translation
多语言推理迁移对于扩展大型语言模型（LLMs）在高资源语言之外的推理能力至关重要。在线自蒸馏（OPSD）及其变体作为一种有前景的范式，提供了对学生生成的回滚的密集标记级监督，但其目标并未明确优先考虑对跨语言迁移至关重要的推理信号。我们认为目标语言的推理包括表面文本和推理枢轴的生成，推理枢轴是推动或重新引导推理过程并塑造后续推理的决策。这促使我们将特权蒸馏集中在这些枢轴周围。因此，我们提出了RP-OPSD，即基于推理枢轴的在线自蒸馏，利用匹配教师视图之间（有无英语参考解）的分布变化作为操作代理，以指导特权蒸馏和参考锚定。在涵盖17种语言和多个难度级别的数学推理基准上的实验表明，我们的方法在性能上超越了强大的多语言推理基线和OPSD变体。进一步的分析显示，RP-OPSD将特权蒸馏集中在推理控制和问题条件状态更新标记上，同时对主要支持表面实现的标记降低权重。我们的代码可在 https://github.com/NJUNLP/RP-OPSD 获取。
cs.CL / 59 / 2608.06370
The Bitter Lesson of Tool Calling
工具调用的苦涩教训
Abstract
Tool use transforms LLMs into agents that act beyond their training data, and for code-capable models, programmatic tool calling extends this further by replacing rigid JSON calls with scripts that chain and parallelize naturally. However, a systematic evaluation of tools as code on an established benchmark across current and prior model generations under real-world task conditions has not been conducted. In this work, we empirically compare programmatic tool calling (PTC) to native JSON tool calling across 14 language models on BFCL v4. In the programmatic tool calling paradigm, tools are exposed as typed Python stubs that the model invokes through code, with execution and results handled in a single agent turn. Programmatic tool calling matches or exceeds native JSON tool calling in 11 of 14 models on BFCL v4, with the GPT-5.6 family achieving a 10.6% improvement over the JSON tool calling baseline. Further, it matches or outperforms baseline in 13 of 14 models under parallel fan-out, and holds stable under context rot conditions where baseline degrades 2.3% on average. Our results demonstrate that programmatic tool calling is a viable and robust alternative to JSON tool calling, with performance tracking model capability across release generations.
Chinese Translation
工具使用将大型语言模型（LLMs）转变为能够超越其训练数据的智能体，对于具备编程能力的模型，程序化工具调用通过用自然链式和并行化的脚本替代僵化的JSON调用，进一步扩展了这一点。然而，尚未在真实世界任务条件下对当前和之前模型代际在既定基准上的工具作为代码进行系统评估。在本研究中，我们在BFCL v4基准上对14种语言模型的程序化工具调用（PTC）与原生JSON工具调用进行了实证比较。在程序化工具调用范式中，工具被暴露为类型化的Python存根，模型通过代码调用这些工具，执行和结果在单个智能体回合中处理。在BFCL v4上，程序化工具调用在14种模型中有11种的表现与原生JSON工具调用相当或更优，其中GPT-5.6系列在JSON工具调用基准上实现了10.6%的提升。此外，在并行分发的情况下，它在14种模型中有13种的表现与基准相匹配或超越，并且在上下文衰退条件下保持稳定，而基准平均下降2.3%。我们的结果表明，程序化工具调用是JSON工具调用的一个可行且稳健的替代方案，其性能能够跟踪模型在不同发布代际中的能力。
cs.CL / 60 / 2608.06377
Learning When to Trust via Selective Context Preference Optimization
通过选择性上下文偏好优化学习何时信任
Abstract
Language models increasingly condition their answers on external signals, and a single misleading one can turn a correct answer wrong. The obvious remedy, training models to resist such signals, hides a failure mode: a model that ignores all context looks robust yet is useless when the context is worth trusting. We recast the problem as selective trust and introduce MIST, a human-annotated benchmark that renders each reasoning item under four matched conditions (clean, misleading, correct-context, and irrelevant-context), together with SC2W, a paired metric counting how often a misleading signal flips a clean-correct answer to wrong. Across a comprehensive benchmark study, we observe that such a susceptibility is universal. We then propose SCOPE, which mines clean-correct/misleading-wrong failures and optimizes a standard Direct Preference Optimization (DPO) objective over matched preference pairs balanced equally across all four conditions, rather than over misleading items alone. Our approach substantially reduces SC2W on popular open-sourced models while preserving accuracy when the added context is clean, correct, or irrelevant. With this work, we argue that models should be judged on selective trust, not on resistance alone.
Chinese Translation
语言模型越来越多地依赖外部信号来决定其回答，而一个误导性的信号就可能使正确的答案变为错误。显而易见的解决办法是训练模型抵抗这些信号，但这隐藏着一种失效模式：一个忽视所有上下文的模型看似稳健，但在上下文值得信任时却毫无用处。我们将这个问题重新表述为选择性信任，并引入了MIST，这是一个经过人工标注的基准，针对每个推理项在四种匹配条件（干净、误导、正确上下文和无关上下文）下进行评估，同时引入SC2W，这是一个配对指标，用于计算误导信号将干净正确答案翻转为错误答案的频率。在全面的基准研究中，我们观察到这种易受影响性是普遍存在的。随后，我们提出了SCOPE，它挖掘干净正确/误导错误的失败案例，并在四种条件下平衡的匹配偏好对上优化标准的直接偏好优化（DPO）目标，而不是仅针对误导项。我们的方法在流行的开源模型上显著降低了SC2W，同时在添加的上下文为干净、正确或无关时保持了准确性。通过这项工作，我们认为模型应该根据选择性信任进行评估，而不仅仅是抵抗能力。