arXiv Daily Digest

237

Papers

Quaternion-Averaging-Based Adaptive Complementary Filter for Pedestrian Dead Reckoning With a Foot-Mounted AHRS

基于四元数平均的行人死算计自适应互补滤波器及其在脚部安装的姿态与航向参考系统中的应用

Yamagishi, Shunsei, Jing, Lei

Abstract

Pedestrian Dead Reckoning (PDR) can be applied to indoor navigation systems. GPS suffers from signal degradation due to roofs and high-rise buildings, whereas PDR can estimate positions without being affected by such signal degradation. The accuracy of a foot-mounted AHRS(Attitude and Heading Reference System)-based PDR depends on the accuracy of the attitude estimation algorithm used in the AHRS. In this article, a Quaternion-Averaging-Based Adaptive Complementary Filter (QAACF) for PDR with a foot-mounted AHRS is proposed to improve estimation accuracy while reducing computational cost. QAACF fuses a quaternion derived from angular velocity with quaternions derived from acceleration and magnetic field measurements using Markley's quaternion averaging, which combines two quaternions more rigorously than linear interpolation. In addition, QAACF adaptively adjusts the weights of angular velocity, acceleration, and magnetic field measurements according to gait phases and the level of magnetic disturbances. Experimental results showed that the proposed QAACF achieves low Root Mean Square Errors (RMSEs) compared to existing attitude estimation filters while requiring lower computational cost than Kalman filters.

Chinese Translation

行人死算计（PDR）可以应用于室内导航系统。GPS信号在建筑物屋顶和高层建筑的影响下容易衰减，而PDR可以在不受此类信号衰减影响的情况下估算位置。基于脚部安装的姿态与航向参考系统（AHRS）的PDR精度依赖于AHRS中使用的姿态估计算法的准确性。本文提出了一种基于四元数平均的自适应互补滤波器（QAACF），旨在提高脚部安装AHRS的PDR估算精度，同时降低计算成本。QAACF通过使用Markley的四元数平均方法，将一个由角速度导出的四元数与由加速度和磁场测量导出的四元数组合，前者在组合时比线性插值更加严谨。此外，QAACF根据步态阶段和磁场干扰水平，自适应地调整角速度、加速度和磁场测量的权重。实验结果表明，与现有的姿态估计滤波器相比，所提出的QAACF实现了较低的均方根误差（RMSE），同时计算成本低于卡尔曼滤波器。

View on arXiv Download PDF AI Translation

cs.RO / 2 / 2607.05468

Learning 4D Geometric Priors for Inference-Efficient World Action Models

学习4D几何先验以提高世界行动模型的推理效率

Zhang, Jianjun, Zhu, Jian, Su, Taiyi, Ma, Chong, Huang, Zitai, Xu, Yi, Wang, Hanli

Abstract

World Action Models (WAMs) have shown strong potential for robotic manipulation by jointly modeling visual future dynamics and executable action sequences. However, existing video-action co-training methods primarily optimize appearance-oriented video latents, which may insufficiently capture the temporally evolving geometry required for precise manipulation. We propose MECo-WAM, a Multi-Expert Co-Training World Action Model that injects action-relevant 4D geometric priors into video-action representations while preserving the original lightweight inference graph. During training, MECo-WAM combines video and action experts with a lightweight 4D expert supervised by relational targets from a frozen VGGT encoder. Asymmetric expert visibility prevents non-causal shortcuts from auxiliary geometry to action generation. To transfer geometric knowledge into the deployed video-action pathway, we introduce decayed 4D read-mask attention, which provides restricted current-frame geometric guidance early in training and progressively removes this dependency. We further propose action-aware temporal geometric distillation, which aligns within-frame geometric relations and their temporal evolution while emphasizing visual regions most relevant to robot actions. At deployment, all auxiliary 4D components are removed. Experiments on LIBERO (98.2%), RoboTwin 2.0 (92.6%), and challenging real-world manipulation tasks show that MECo-WAM improves manipulation performance without increasing inference cost.

Chinese Translation

世界行动模型（WAMs）在机器人操作中显示出强大的潜力，通过联合建模视觉未来动态和可执行的动作用序列。然而，现有的视频-动作共同训练方法主要优化面向外观的视频潜变量，这可能不足以捕捉精确操作所需的时间变化几何形状。我们提出了MECo-WAM，一个多专家共同训练的世界行动模型，该模型在保持原有轻量级推理图的同时，将与动作相关的4D几何先验注入到视频-动作表示中。在训练过程中，MECo-WAM结合了视频和动作专家，并使用一个由冻结的VGGT编码器监督的轻量级4D专家。非对称的专家可见性防止了从辅助几何到动作生成的非因果捷径。为将几何知识转移到部署的视频-动作路径中，我们引入了衰减的4D读取掩码注意力，它在训练早期提供有限的当前帧几何指导，并逐步消除这种依赖。我们进一步提出了关注动作的时间几何蒸馏，它对齐了内部帧几何关系及其时间演变，同时强调与机器人动作最相关的视觉区域。在部署时，所有辅助的4D组件被移除。在LIBERO（98.2%）、RoboTwin 2.0（92.6%）和具有挑战性的现实世界操作任务上的实验表明，MECo-WAM在不增加推理成本的情况下提高了操作性能。

View on arXiv Download PDF AI Translation

cs.RO / 3 / 2607.05543

GEM-Occ: From Visual Geometry Evidence to Embodied Semantic Occupancy Memory

GEM-Occ：从视觉几何证据到具身语义占用记忆

Zhu, Hu, Li, Bohan, Guo, Xianda, Liu, Hongsi, Peng, Baorui, Yuan, Mingqi, Jin, Xin, Zeng, Wenjun, Chen, Chang Wen

Abstract

Semantic occupancy provides a structured spatial memory for embodied indoor agents by jointly representing occupied regions, observed free space, unknown areas, and object semantics. However, existing indoor occupancy benchmarks and methods mainly focus on single-view prediction or room-level online perception, leaving long-horizon semantic mapping across connected indoor spaces underexplored. We introduce HIOcc, a hierarchical indoor occupancy benchmark that unifies ScanNet, ScanNet++, and Matterport3D under a common sparse semantic occupancy format while preserving their native observation geometries, including perspective RGB-D frames and pano-centric observation groups. HIOcc supports three complementary evaluation regimes: local semantic occupancy prediction, room-level online occupancy mapping, and building-level mapping across connected panoramic environments. We further propose GEM-Occ, a Gaussian Evidence Memory framework for semantic occupancy mapping. Rather than using pointmaps as persistent map states, GEM-Occ treats local visual geometry predictions as transient evidence, converts them into semantic Gaussian occupancy evidence and free-space ray evidence, and fuses them into a persistent hierarchical memory through visibility- and uncertainty-aware causal updates. The memory is organized into local caches, room-level submaps, and a building-level graph, and can be queried at any time through Gaussian-to-occupancy splatting. Experiments on HIOcc show that GEM-Occ improves local occupancy prediction, online map stability, free-space reasoning, revisit consistency, and building-level scalability over prior indoor occupancy and Gaussian-based mapping baselines.

Chinese Translation

语义占用为具身室内智能体提供了一种结构化的空间记忆，通过联合表示占用区域、观察到的自由空间、未知区域和物体语义。然而，现有的室内占用基准和方法主要集中于单视角预测或房间级在线感知，长时间范围的语义映射在相连的室内空间中尚未得到充分探索。我们引入了HIOcc，一个层次化的室内占用基准，它在保持原有观察几何（包括透视RGB-D帧和全景中心观察组）的同时，将ScanNet、ScanNet++和Matterport3D统一为一种通用的稀疏语义占用格式。HIOcc支持三种互补的评估机制：局部语义占用预测、房间级在线占用映射和跨连接全景环境的建筑级映射。我们进一步提出了GEM-Occ，一个用于语义占用映射的高斯证据记忆框架。GEM-Occ并不将点图作为持久的地图状态，而是将局部视觉几何预测视为瞬时证据，将其转换为语义高斯占用证据和自由空间光线证据，并通过考虑可见性和不确定性的因果更新将其融合到持久的层次记忆中。该记忆组织为局部缓存、房间级子图和建筑级图形，并可以通过高斯到占用的点云查询随时访问。在HIOcc上的实验表明，GEM-Occ在局部占用预测、在线地图稳定性、自由空间推理、重访一致性和建筑级可扩展性方面优于先前的室内占用和基于高斯的映射基线。

View on arXiv Download PDF AI Translation

cs.RO / 4 / 2607.05544

Dynamic Evaluation of Classical and Control-Aware Optimal Trajectory Planning in Robot Manipulators

机器人机械臂中经典与控制感知的最佳轨迹规划的动态评估

Dayawansa, Bhanuka, Munasinghe, Rohan

Abstract

Trajectory planning strongly influences tracking accuracy, actuator demand, and overall execution behavior in robotic manipulators. Classical planners such as cubic, quintic, and trapezoidal profiles are widely used for their simplicity and smoothness, yet they remain purely kinematic and ignore system dynamics and control effort during trajectory generation. As a result, nominally smooth trajectories can lead to inefficient nonlinear execution and increased corrective control action. This paper presents a control-aware optimal trajectory planning framework that explicitly incorporates manipulator dynamics and actuator effort within a finite-horizon formulation. A midpoint linearization strategy is introduced to improve approximation accuracy for large point-to-point motions. In contrast to prior comparisons, the proposed approach enables fair, isolated evaluation of trajectory generation effects under identical closed-loop nonlinear execution conditions. To this end, a unified evaluation framework is developed in which all planners are executed under identical nonlinear dynamics, controller structure, and actuator constraints. Simulations on a nonlinear simplified UR5 manipulator show that the proposed approach consistently reduces tracking error, corrective torque, and closed-loop execution cost compared to classical methods, achieving substantial reductions in actuator effort and execution cost across all evaluated scenarios, demonstrating that kinematic smoothness alone does not ensure dynamically efficient execution.

Chinese Translation

轨迹规划对机器人机械臂的跟踪精度、执行器需求和整体执行行为有着重要影响。经典规划方法如三次、五次和梯形轨迹曲线因其简洁和平滑而被广泛使用，但它们仍然仅限于运动学，忽视了轨迹生成过程中的系统动态和控制努力。因此，表面上平滑的轨迹可能导致非线性执行效率低下和增加的修正控制动作。本文提出了一种控制感知的最佳轨迹规划框架，该框架在有限时间范围内明确考虑了机械臂动态和执行器努力。引入了一种中点线性化策略，以提高大范围点对点运动的近似精度。与以往的比较不同，所提出的方法在相同闭环非线性执行条件下，能够公平地、独立地评估轨迹生成的效果。为此，开发了一个统一的评估框架，所有规划器在相同的非线性动态、控制器结构和执行器约束下执行。在对非线性简化UR5机械臂的仿真中，结果表明，与经典方法相比，所提出的方法在减少跟踪误差、修正扭矩和闭环执行成本方面表现一致，显著降低了所有评估场景下的执行器努力和执行成本，证明了单靠运动学平滑性并不能确保动态高效的执行。

View on arXiv Download PDF AI Translation

cs.RO / 5 / 2607.05663

Physics-Regularized Machine Learning for Proprioceptive Vehicle Localization Using Onboard Sensors

基于物理正则化的机器学习用于利用车载传感器进行自我感知车辆定位

Kalyanasundaram, Abinav, Sekaran, Karthikeyan Chandra, Utschick, Wolfgang, Botsch, Michael

Abstract

Accurate and robust localization is essential for autonomous mobility systems in real-world environments. While fusing Inertial Measurement Unit (IMU) data with satellite-based correction signals provides precise vehicle pose estimates, performance degrades substantially during outages. Recent studies indicate that Machine Learning (ML) can improve IMU-based proprioceptive localization, highlighting untapped potential for onboard sensors readily available in production vehicles. This paper introduces Physics-Regularized Machine Learning for Localization (PRML2), a hybrid framework that combines the complementary strengths of Kalman filtering and data-driven learning to estimate vehicle pose directly from onboard sensors. A key aspect of PRML2 is its physics-regularized learning, enabled by end-to-end training of an ML model through a differentiable Kalman filter. This improves consistency with vehicle motion models, thereby enhancing both localization accuracy and generalization across driving conditions. We evaluate the performance limits of ML-enhanced onboard odometry on a publicly available dataset and show that PRML2 achieves superior localization accuracy and demonstrates real-time capability. This work also introduces a novel dataset to support vehicle localization research under low-friction conditions. The proposed framework provides a robust and cost-effective solution for vehicle localization under degraded sensing conditions by integrating learning with physics-based priors.

Chinese Translation

准确且稳健的定位对于现实环境中的自主移动系统至关重要。虽然将惯性测量单元（IMU）数据与基于卫星的校正信号融合可以提供精确的车辆姿态估计，但在信号中断期间，性能会显著下降。近期研究表明，机器学习（ML）可以改善基于IMU的自我感知定位，突显了生产车辆中现有车载传感器的未开发潜力。本文介绍了一种基于物理正则化的机器学习定位框架（PRML2），该框架结合了卡尔曼滤波和数据驱动学习的互补优势，直接从车载传感器估计车辆姿态。PRML2的一个关键方面是其物理正则化学习，通过可微分的卡尔曼滤波器对机器学习模型进行端到端训练，从而提高与车辆运动模型的一致性，增强定位精度并提高在不同驾驶条件下的泛化能力。我们在一个公开可用的数据集上评估了增强机器学习的车载里程计的性能极限，并展示了PRML2实现了更优的定位精度并具备实时能力。本研究还引入了一个新数据集，以支持低摩擦条件下的车辆定位研究。所提出的框架通过将学习与基于物理的先验结合，为在感知条件恶化下的车辆定位提供了一种稳健且具有成本效益的解决方案。

View on arXiv Download PDF AI Translation

cs.RO / 6 / 2607.05665

Efficient Transfer Learning of Robot Dynamic Models Using Morphological Similarity

基于形态相似性的机器人动态模型高效迁移学习

Kupyn, Pavlo, Hamamatsu, Yuya, Gkliva, Roza, Ristolainen, Asko, Kruusmaa, Maarja

Abstract

This study proposes a neural network-based transfer learning framework for modeling the dynamics of soft, fin-actuated underwater robots. We focus on morphologically similar robots that differ in scale and hydrodynamic properties. A model trained on data from a larger robot (source domain) is adapted to a smaller one (target domain) with limited labeled data. To enable label-efficient transfer, we develop an autoencoder-based domain adaptation approach that learns a shared latent representation aligning the dynamics of both robots. Experiments on two real underwater robots show that the proposed method enables accurate state estimation of the body-frame velocities on a target platform without labeled data, highlighting its potential for efficient cross-robot dynamics transfer among morphologically similar platforms.

Chinese Translation

本研究提出了一种基于神经网络的迁移学习框架，用于建模软体鳍驱动水下机器人的动态特性。我们关注于形态上相似但在规模和水动力特性上有所不同的机器人。使用从较大机器人（源领域）收集的数据训练的模型，被适配到具有有限标注数据的较小机器人（目标领域）。为了实现高效的标签迁移，我们开发了一种基于自编码器的领域适配方法，学习共享的潜在表示，以对齐两种机器人的动态。对两种真实水下机器人的实验表明，该方法能够在没有标注数据的情况下，准确估计目标平台的主体帧速度，突显其在形态相似平台之间高效跨机器人动态迁移的潜力。

View on arXiv Download PDF AI Translation

cs.RO / 7 / 2607.05669

Uncertainty-Aware Velocity Correction for Proprioceptive Vehicle Localization using Evidential Mamba

基于证据的多巴蛇不确定性感知速度校正用于自我感知车辆定位

Kalyanasundaram, Abinav, Sekaran, Karthikeyan Chandra, Utschick, Wolfgang, Botsch, Michael

Abstract

Reliable localization in GNSS-denied environments remains a fundamental challenge for intelligent vehicles, as inertial navigation systems accumulate unbounded drift without external correction. Existing approaches provide drift correction through dedicated infrastructure, expensive external sensors, or complex multi-sensor fusion, each introducing practical deployment barriers. We propose Evidential Velocity Correction using Mamba (EVC-Mamba), a learning-based architecture that transforms onboard vehicle sensor data into a virtual velocity sensor for IMU drift correction without additional hardware. A Mamba-based selective state space model captures the temporal dynamics of vehicle motion, while evidential deep learning with a Normal-Inverse-Gamma distribution provides principled uncertainty quantification. The resulting uncertainty-aware velocity estimate is incorporated as a virtual correction measurement into an Error-State Extended Kalman Filter to reduce position drift. Evaluation on real-world vehicle data demonstrates that inertial navigation using the proposed velocity correction achieves localization accuracy within 10% of a dedicated external velocity sensor across different outage durations. The proposed architecture supports real-time onboard deployment at 40 Hz on edge hardware, enabling reliable localization during prolonged GNSS outages.

Chinese Translation

在GNSS拒绝环境中实现可靠定位仍然是智能车辆面临的基本挑战，因为惯性导航系统在没有外部校正的情况下会积累无界漂移。现有的方法通过专用基础设施、昂贵的外部传感器或复杂的多传感器融合提供漂移校正，但每种方法都带来了实际部署障碍。我们提出了一种基于证据的速度校正方法，称为EVC-Mamba，它是一种基于学习的架构，能够将车载传感器数据转换为虚拟速度传感器，以实现IMU漂移校正，而无需额外的硬件。一种基于Mamba的选择性状态空间模型捕捉车辆运动的时间动态，而基于证据的深度学习与正态-逆伽马分布结合提供了有原则的不确定性量化。生成的关注不确定性的速度估计被作为虚拟校正测量引入到误差状态扩展卡尔曼滤波器中，以减少位置漂移。在实际车辆数据上的评估表明，使用所提出的速度校正的惯性导航在不同的失效持续时间内实现了与专用外部速度传感器相差10%的定位精度。所提出的架构支持在边缘硬件上以40 Hz的实时车载部署，使得在长时间GNSS失效期间实现可靠定位成为可能。

View on arXiv Download PDF AI Translation

cs.RO / 8 / 2607.05705

IMR: Iterative Mode-World Weighted Regression for Multi-Agent Trajectory Prediction

IMR：用于多智能体轨迹预测的迭代模式-世界加权回归

Wang, Honglin, Pan, Shiyao, Liu, Yun-Fu

Abstract

Multi-agent motion prediction is essential for automated vehicles to understand the intentions of surrounding vehicles. However, previous prediction-based and anchor-based methods have limitations in mode diversity and prediction accuracy, respectively. These limitations may cause inadequate safety assessments and behavioral deviations in automated vehicles. To address this issue, a mode-world weighted regression loss is proposed to bridge the gap between these features. Specifically, this approach mitigates mode collapse while simultaneously improving world ranking and top-1 confidence. Furthermore, the proposed iterative decoder improves prediction accuracy by recurrently and segmentally generating trajectories. Experimental results show the proposed method ranks first in the Argoverse 2 multi-agent motion forecasting benchmark against other methods.

Chinese Translation

多智能体运动预测对自动化车辆理解周围车辆的意图至关重要。然而，先前的基于预测和基于锚点的方法在模式多样性和预测准确性方面存在局限性。这些局限性可能导致自动化车辆的安全评估不足和行为偏差。为了解决这一问题，提出了一种模式-世界加权回归损失，以桥接这些特征之间的差距。具体而言，这种方法在减轻模式崩溃的同时，改善了世界排名和top-1置信度。此外，所提出的迭代解码器通过递归性和分段生成轨迹来提高预测准确性。实验结果表明，在Argoverse 2多智能体运动预测基准测试中，所提方法在其他方法中排名第一。

View on arXiv Download PDF AI Translation

cs.RO / 9 / 2607.05709

Co-STAR: Cognitive Stimulation Therapy by an Autonomous Robot for Dementia -- A One-Week In-Home Study

Co-STAR：由自主机器人实施的认知刺激治疗用于痴呆症——为期一周的家庭研究

Akinrintoyo, Emmanuel, Salomons, Nicole

Abstract

Cognitive therapies have been shown to enhance the quality of life and well-being of people living with dementia (PwDs). However, their use remains limited due to a shortage of trained professionals and the significant time and training required of informal caregivers. To address this gap, we developed and deployed a social robot capable of autonomously delivering cognitive stimulation therapy (CST) in the home. Nine PwDs participated in a one-week ($7$ days) study that involved daily robot-led sessions. Participants engaged positively with the system, completing nearly half of the scheduled sessions, an adherence rate higher than typically observed in caregiver-led CST. Our findings also highlight the crucial role of family members, who often supported session initiation and occasionally joined the activities, enriching the interactions. This work demonstrates the feasibility and potential of socially assistive robots to deliver in-home cognitive therapy, offering a scalable approach to extend access to dementia care.

Chinese Translation

研究表明，认知疗法能够改善生活在痴呆症中的人（PwDs）的生活质量和福祉。然而，由于缺乏经过培训的专业人员以及非正式照顾者所需的显著时间和训练，这些疗法的使用仍然有限。为了解决这一问题，我们开发并部署了一种社会机器人，能够在家中自主提供认知刺激治疗（CST）。九名PwDs参与了一项为期一周（7天）的研究，研究中涉及到机器人主导的每日会议。参与者积极参与，完成了近一半的计划会议，遵守率高于通常在照顾者主导的CST中观察到的比例。我们的研究结果还突显了家庭成员的重要作用，他们通常支持会议的发起，并偶尔参与活动，丰富了互动。此项工作展示了社会辅助机器人在家中提供认知疗法的可行性和潜力，提供了一种可扩展的方法来扩大痴呆症护理的覆盖面。

View on arXiv Download PDF AI Translation

cs.RO / 10 / 2607.05777

Observation Quality Matters: Robust Multi-Fisheye Calibration via Failure-Oriented Analysis

观察质量至关重要：通过失败导向分析实现稳健的多鱼眼相机标定

Liu, Peize, Tong, Zhe, Feng, Chen, Shen, Shaojie

Abstract

Reliable calibration of multi-fisheye camera systems remains challenging as rig size, camera arrangement diversity, and field of view increase. Existing pipelines can jointly optimize intrinsics, extrinsics, and target poses, but their success still depends heavily on empirical capture rules and the quality of the observations supplied to the solver. This paper studies this dependency through a failure-oriented analysis. We reveal that calibration failures are not sufficiently explained by detector recall loss or global image-plane distribution imbalance. Instead, the dominant failure factor lies in intrinsic initialization: observations with limited radial span couple focal scale with fisheye projection-shape parameters, producing ill-conditioned updates. Guided by this insight, we propose CO-Calib, a plug-in calibration-data construction framework that combines a robust learning-based target detector with an error-analysis-guided frame selector. CO-Calib constructs initialization-friendly anchors, co-visible multi-camera constraints, and coverage-completion frames without changing the existing calibration workflow or optimization backend. Extensive experiments on synthetic and real multi-fisheye systems demonstrate that CO-Calib improves the overall success rate from 68.1% to 99.3%, increases extrinsic accuracy, and augments real-world calibration stability. The source code will be made publicly available at https://github.com/HKUST-Aerial-Robotics/CO-Calib.

Chinese Translation

多鱼眼相机系统的可靠标定仍然面临挑战，尤其是在设备尺寸、相机排列多样性和视场增大时。现有的处理流程可以联合优化内参、外参和目标姿态，但其成功仍然在很大程度上依赖于经验捕获规则和提供给求解器的观测质量。本文通过失败导向分析研究了这种依赖性。我们揭示，标定失败并不能仅通过检测器召回率损失或全局图像平面分布不平衡来充分解释。相反，主导的失败因素在于内参初始化：具有有限径向跨度的观测将焦距尺度与鱼眼投影形状参数耦合，导致条件不良的更新。基于这一见解，我们提出了CO-Calib，一个插件式标定数据构建框架，结合了稳健的基于学习的目标检测器和错误分析引导的帧选择器。CO-Calib构建了初始化友好的锚点、可共视的多相机约束和覆盖完成帧，而无需改变现有的标定工作流程或优化后端。在合成和真实多鱼眼系统上的大量实验表明，CO-Calib将整体成功率从68.1%提高到99.3%，提高了外参精度，并增强了实际标定的稳定性。源代码将公开发布在https://github.com/HKUST-Aerial-Robotics/CO-Calib。

View on arXiv Download PDF AI Translation

cs.RO / 11 / 2607.05780

FORGE: Towards Functional Tool-Use Generalization via Keypoint Trajectory Reasoning

FORGE：通过关键点轨迹推理实现功能性工具使用的泛化

Zhou, Chuhao, Wang, Liquan, Cao, Shuxin, Chen, Xiangyu, Hu, Yuxuan, Ma, Boyu, Garg, Animesh, Yang, Jianfei

Abstract

While humans readily repurpose a book, a stone, or a shoe to drive a nail, robots trained on specific tools fail to transfer the same function to novel ones -- a gap we formalize as functional generalization. Such tools share a common functional intent that is visually recognizable, yet this perceptual similarity does not carry over to action space, where each tool demands an entirely different motor pattern. To bridge this gap, we explore intermediate representations including affordance images, human video prompts, and 2D keypoint trajectories, finding that keypoint trajectories best balance functional expressiveness and action groundability. Building on this, we propose FunctiOnal Reasoning and Grounded Execution (FORGE), a two-stage policy that decouples functional reasoning from action execution: predicting generalizable keypoint trajectories from action-free data, then grounding them into robot actions with limited demonstrations. On a seven-tool hitting-function benchmark, FORGE consistently outperforms state-of-the-art methods on unseen tools in both simulation and the real world, achieving over 2X improvement in average success rate.

Chinese Translation

人类能够轻松地将书籍、石头或鞋子重新利用来钉钉子，但训练于特定工具的机器人在将相同功能转移到新工具时却表现不佳——我们将这一差距形式化为功能性泛化。这些工具共享一个共同的功能意图，该意图在视觉上是可识别的，但这种感知相似性并不延续到动作空间中，在那里，每个工具都要求完全不同的运动模式。为了弥合这一差距，我们探索了一些中间表示，包括可供性图像、人类视频提示和二维关键点轨迹，发现关键点轨迹在功能性表达和动作可实施性之间达到了最佳平衡。基于此，我们提出了功能推理与具象执行（FunctiOnal Reasoning and Grounded Execution，FORGE），这是一种两阶段策略，将功能推理与动作执行解耦：从无动作的数据中预测可泛化的关键点轨迹，然后将其转化为机器人动作，只需有限演示。在一个包含七种工具的打击功能基准测试中，FORGE在未见工具方面在模拟和现实世界中始终优于最先进的方法，平均成功率实现了超过两倍的提升。

View on arXiv Download PDF AI Translation

cs.RO / 12 / 2607.05869

GraspIT: A Dataset Bridging the Sim-to-Real gap and back for Validated Grasping SE(3) Pose Generation

GraspIT：一个桥接仿真与现实之间的验证抓取 SE(3) 位姿生成的数据集

Karakurt, Paul Koch. Adem, Sers, André

Abstract

Robust robotic grasping of novel objects requires datasets that simultaneously provide photorealistic RGB-D observations, physically validated grasp quality annotations, and a principled bridge between simulation and the real world, which existing datasets lack to provide jointly. \textbf{GraspIT} addresses this gap: tabletop scenes in NVIDIA Isaac Sim are annotated via a four-stage physical slip-test on parallel Franka Panda instances, producing trajectory-reachability checks and continuous quality scores beyond force-closure.Of ${\sim}$2.3M candidates, 83% pass as \emph{good} ($s{\geq}0.50$); the 17% that passed force-closure but failed the slip-test provide graded hard negatives. A Real$\leftrightarrow$Sim loop back-projects these labels onto 100 real-world scenes. The release provides ${\sim}$316k annotated RGBD frame sets across 1035 sim and 100 real scenes, with instance masks, 6-DoF poses, physical object properties, and scored 6-DoF grasps. All tools are open-source and Docker-containerized. The trajectory planning within Isaac Sim further allows streaming of high resolution demonstrations for tabletop manipulation policy learning and behavior cloning.

Chinese Translation

稳健的机器人抓取新物体需要同时提供照片级真实感的 RGB-D 观察、物理验证的抓取质量注释以及仿真与现实之间的系统桥梁，而现有数据集在这方面的联合提供不足。 extbf{GraspIT} 解决了这一缺口：在 NVIDIA Isaac Sim 中的桌面场景通过在并行 Franka Panda 实例上的四阶段物理滑动测试进行标注，生成轨迹可达性检查和超越力闭合的连续质量评分。在约 230 万个候选项中，83% 被评定为“好”（$s{ geq}0.50$）；17% 在通过力闭合但未通过滑动测试的候选项提供了分级的困难负样本。一个 Real$ ightarrow$Sim 循环将这些标签反投影到 100 个现实场景中。此次发布提供了约 316,000 组标注 RGB-D 帧，涵盖 1035 个仿真场景和 100 个真实场景，包含实例掩模、6 自由度位姿、物理物体属性以及评分的 6 自由度抓取。所有工具都是开源和基于 Docker 容器的。Isaac Sim 中的轨迹规划进一步允许进行高分辨率演示的流式传输，以便于桌面操作策略学习和行为克隆。

View on arXiv Download PDF AI Translation

cs.RO / 13 / 2607.05883

DexTele: A Dual-Arm Dexterous Teleoperation System Based on Motion Retargeting and Adaptive Force Control

DexTele：基于运动重定向和自适应力控制的双臂灵巧遥操作系统

Lai, Yuanchuan, Gao, Qing, Liang, Ziyan, Cheng, Xianfeng, Hu, Junjie, Ju, Zhaojie

Abstract

In dual-arm dexterous teleoperation, cross-platform generalization of motion retargeting and interactivity of grasping are crucial. However, the heterogeneity of robotic architectures and the wide variety of grasping objects pose significant challenges to achieving precise motion retargeting and compliant grasping in dual-arm dexterous teleoperation. To address these challenges, a dual-arm dexterous teleoperation system (DexTele) is proposed based on motion retargeting and adaptive force control. First, a vision-based motion retargeting module is designed to generate preliminary robot motions from human images. In this module, a motion-graph encoder and latent optimization are proposed for precise and convenient cross-platform motion retargeting. Second, an adaptive grasping module is designed to achieve compliant grasping. This module combines a vision-language model (VLM) with model predictive control (MPC), allowing the system to predict the required grasping force for a target object and perform gradient-based online optimization. Finally, extensive experiments demonstrate that the DexTele achieves precise motion retargeting and compliant grasping with generalization across multiple robot platforms.

Chinese Translation

在双臂灵巧遥操作中，运动重定向的跨平台泛化和抓取的交互性至关重要。然而，机器人架构的异构性和多样的抓取对象为实现精准的运动重定向和符合要求的抓取带来了重大挑战。为了解决这些问题，提出了一种基于运动重定向和自适应力控制的双臂灵巧遥操作系统（DexTele）。首先，设计了基于视觉的运动重定向模块，用于从人类图像生成初步的机器人动作。在该模块中，提出了运动图谱编码器和潜在优化方法，以实现精准和方便的跨平台运动重定向。其次，设计了自适应抓取模块，以实现符合要求的抓取。该模块将视觉-语言模型（VLM）与模型预测控制（MPC）相结合，使系统能够预测目标对象所需的抓取力，并进行基于梯度的在线优化。最后，通过大量实验证明，DexTele在多个机器人平台上实现了精准的运动重定向和符合要求的抓取。

View on arXiv Download PDF AI Translation

cs.RO / 14 / 2607.05939

Intercepting an Agile Target with Net-Carrying Drones using Competitive Multi-Agent Reinforcement Learning

使用竞争性多智能体强化学习拦截灵活目标的网载无人机

Gavin, Timothée, Bronz, Murat

Abstract

This article presents a solution to intercept an agile drone by a team of agile drone carrying catching nets. We formulate the problem as a competitive Multi-Agent Reinforcement Learning (MARL) task. To address the problem of nonstationarity and catastrophic forgetting of agents overfitting to the current opponent strategy, we train the pursuers and the evader using Multi-Agent Proximal Policy Optimization (MAPPO) with Prioritized Fictitious Self Play (PFSP). We train the agents in a high-fidelity simulator using low-level control commands, collective thrust and body rates (CTBR), to achieve agile flights for both the pursuers and the evader. We compare the performance of the trained policies in terms of catch rate, time to catch and crash rates, against heuristic baselines and show that our solution outperforms them. Ablation studies show that PFSP lead to more robust policies that can adapt to different opponent strategies, and that a low-level control commands are crucial for learning performing strategies in the pursuit-evasion task. Finally, a qualitative analysis of the learned behaviours highlights the emergence of cooperative tactics among the pursuers.

Chinese Translation

本文提出了一种通过一组携带捕捉网的灵活无人机拦截灵活无人机的解决方案。我们将该问题表述为一个竞争性多智能体强化学习（MARL）任务。为了解决智能体在当前对手策略上过拟合导致的非平稳性和灾难性遗忘问题，我们使用带有优先虚拟自我对抗（PFSP）的多智能体近端策略优化（MAPPO）训练追捕者和逃避者。我们在高保真模拟器中使用低级控制指令、集体推力和机体速率（CTBR）训练智能体，以实现追捕者和逃避者的灵活飞行。我们将训练策略的表现与启发式基线进行比较，评估捕获率、捕获时间和碰撞率，结果表明我们的解决方案优于这些基线。消融研究表明，PFSP导致了更具鲁棒性的策略，能够适应不同的对手策略，并且低级控制指令对于学习追逐-逃避任务中的有效策略至关重要。最后，对学习到的行为进行的定性分析突显了追捕者之间合作战术的出现。

View on arXiv Download PDF AI Translation

cs.RO / 15 / 2607.05957

Delay-Aware Active Triangulation with Uncertainty-Driven Multi-Agent Reinforcement Learning for Counter-UAS

基于延迟感知的主动三角测量与不确定性驱动的多智能体强化学习在反无人机系统中的应用

Lee, Seungwook, Shim, David Hyunchul

Abstract

Multi-agent active visual triangulation enables precise 3D localization of aerial targets by coordinating mobile observers with controllable cameras. However, existing methods assume instantaneous state feedback, ignoring cumulative latency from detection, communication, and decision propagation. We present a delay-aware, uncertainty-driven multi-agent reinforcement learning framework for target localization in Counter-UAS applications. Our contributions are: (1) a Dec-POMDP formulation with Age-of-Information (AoI) augmented observations enabling delay-aware coordination -- AoI improves triangulation validity by 10.6 percentage points; (2) a controlled comparison showing that perception-consistent rewards outperform privileged clean-state rewards (0.547 m vs.0.633 m RMSE, 27% fewer track losses) -- both policies are trained through identical observation noise but differ in what they are optimized for, producing a stability-robustness tradeoff; and (3) multi-source analytical covariance propagation incorporating pixel, pose, gimbal, and intrinsics uncertainties -- restricting to angular noise alone causes 2.8-fold RMSE degradation. Experiments with MAPPO in 4096 parallel environments achieve 0.547 +- 0.217 m RMSE with 78.1% triangulation validity, while MLP policies achieve near-zero validity (0.7%), confirming recurrent memory as essential for delay compensation.

Chinese Translation

多智能体主动视觉三角测量通过协调可控摄像头的移动观察者，实现了空中目标的精确三维定位。然而，现有方法假设状态反馈是瞬时的，这忽略了来自检测、通信和决策传播的累积延迟。我们提出了一种延迟感知的不确定性驱动多智能体强化学习框架，用于反无人机系统中的目标定位。我们的贡献包括：(1) 引入了信息时延（Age-of-Information, AoI）增强观察的分布式部分可观察马尔可夫决策过程（Dec-POMDP）构型，使得延迟感知的协调成为可能——AoI提升了三角测量的有效性10.6个百分点；(2) 通过对比实验表明，感知一致奖励的表现优于特权净状态奖励（0.547 米 vs. 0.633 米均方根误差，减少27%的轨迹丢失）——两种策略在相同的观察噪声下训练，但其优化目标不同，产生了稳定性与鲁棒性之间的权衡；(3) 考虑像素、姿态、云台和内部参数不确定性的多源分析协方差传播——仅限于角度噪声导致了2.8倍的均方根误差退化。使用MAPPO在4096个并行环境中进行的实验达到了0.547 ± 0.217 米的均方根误差和78.1%的三角测量有效性，而多层感知器（MLP）策略的有效性接近于零（0.7%），确认了循环记忆在延迟补偿中的重要性。

View on arXiv Download PDF AI Translation

cs.RO / 16 / 2607.05966

Imagined Rollouts are Kinematic, Not Dynamic: A Diagnosis of Long-Horizon World-Model Failure

想象的展开是运动学的，而非动力学的：长远期世界模型失败的诊断

Schäfer, Finn Rasmus, Moller, Korbinian, Gao, Yuan, Oefinger, Christian, Schmidt, Sebastian, Betz, Johannes

Abstract

Long-horizon failure in world models is conventionally attributed to compounding error, a generic framing that does not distinguish what kind of error compounds. We propose a kinematic-vs-dynamic reframing: world models tend to imagine kinematically rather than dynamically. We operationalize this as the imagined Kinematic-Consistency Error, a per-step diagnostic that measures how far a rollout departs from a closed-form kinematic null, paired with a perturbation protocol that tests whether iKCE responds when physical conditions cross a regime boundary. We instantiate the diagnostic on a released DreamerV3 checkpoint trained on DMC walker-walk, where imagined iKCE runs roughly two orders of magnitude above that of matched real-physics rollouts. Across a friction sweep that crosses the gait-collapse boundary, the model's iKCE stays statistically flat even as the trained policy's reward collapses through the same range, providing the kinematic-not-dynamic signature. The diagnostic distinguishes kinematic from dynamic imagination at horizons longer than the embodiment's gait period.

Chinese Translation

在世界模型中的长远期失败通常被归因于误差累积，这种通用的框架没有区分出是什么类型的误差在累积。我们提出了一种运动学与动力学的重新框架：世界模型倾向于在运动学方面进行想象，而非动力学。我们将其操作化为想象的运动学一致性误差（imagined Kinematic-Consistency Error, iKCE），这是一种逐步诊断，衡量展开与封闭形式运动学零模型的偏离程度，并配合一种扰动协议，以测试当物理条件跨越一个状态边界时iKCE是否有反应。我们在一个已发布的DreamerV3检查点上实现了这一诊断，该检查点是在DMC walker-walk任务上训练的，其中想象的iKCE值大约比匹配的真实物理展开高出两个数量级。在一个跨越步态崩溃边界的摩擦范围内，模型的iKCE在统计上保持平坦，即便训练策略的奖励在同一范围内崩溃，提供了运动学而非动力学的特征。该诊断在超过体现的步态周期的时间跨度上区分了运动学与动力学的想象。

View on arXiv Download PDF AI Translation

cs.RO / 17 / 2607.06018

RoboTALES: Learning Reasoning-Guided Robot Policies via Task-Aligned Simulated Futures

RoboTALES：通过任务对齐的模拟未来学习推理引导的机器人策略

Gani, Hanan, Kulkarni, Tejal, Chodavarapu, Madhoolika, Hansen, Nicklas, Chandraker, Manmohan

Abstract

Pretrained video generative models are promising backbones for visuomotor control, but their imagined futures often drift from task intent and are not reliably action-conditional. As a result, these models can be difficult to use for planning or policy extraction. To address these limitations, we propose RoboTALES, a single-stage framework that learns task-aligned simulated futures and uses them to train robot policies. Our approach introduces two key innovations: (1) a hierarchical LLM-based planner that breaks complex tasks into a sequence of subgoals to guide the model's imagination; and (2) a VLM-based critic that evaluates these ``imagined'' futures and uses reward-based feedback to keep the model's internal representations focused on the goal. By anchoring the video generator in abstract reasoning, we produce temporally consistent rollouts and more coherent actions. We evaluate RoboTALES on diverse manipulation tasks from RoboCasa and LIBERO10, and show that our method consistently outperforms existing methods, especially in long-horizon tasks. Our code and models are publicly available at https://github.com/hananshafi/RoboTALES.

Chinese Translation

预训练的视频生成模型在视觉运动控制方面表现出色，但它们所想象的未来通常偏离任务意图且不一定可靠地与动作相关。因此，这些模型在规划或策略提取中往往难以使用。为了克服这些局限性，我们提出了RoboTALES，一个单阶段框架，学习任务对齐的模拟未来并用于训练机器人策略。我们的方法引入了两个关键创新：(1) 基于层次化大语言模型（LLM）的规划器，将复杂任务分解为一系列子目标，以指导模型的想象；(2) 基于视觉语言模型（VLM）的评估者，评估这些“想象”的未来并利用基于奖励的反馈使模型的内部表征保持集中于目标。通过将视频生成器固定在抽象推理中，我们获得了时间上连贯的展开和更一致的行动。我们在RoboCasa和LIBERO10的多样化操控任务上评估了RoboTALES，结果表明我们的方法在长时间视距任务中一直优于现有方法。我们的代码和模型可在https://github.com/hananshafi/RoboTALES获取。

View on arXiv Download PDF AI Translation

cs.RO / 18 / 2607.06052

ThorArena: Benchmarking Humanoid Physical Interaction with Human Motion-Force Demonstrations

ThorArena：基于人类动作-力量示范的人形机器人物理交互基准测试

Yu, Chenhao, Wang, Hongwu, Zhang, Weitao, Hu, Youhao, Zhang, Jiachen, Li, Gangyang, Knoll, Alois, Luo, Shaqi

Abstract

Humanoid robots are increasingly expected to perform contact-rich tasks that require not only accurate whole-body motion but also robust physical interaction with surrounding objects and humans. Although recent advances in humanoid motion imitation and whole-body control have achieved remarkable tracking performance, existing datasets and benchmarks primarily focus on kinematic motion while largely overlooking synchronized interaction forces. As a result, current evaluations fail to capture how external interaction forces affect tracking accuracy, stability, and control robustness. In this paper, we present ThorArena, a benchmark for evaluating force-aware humanoid interaction based on human demonstrations with synchronized motion and force measurements. We collect a real-world interaction dataset that simultaneously captures whole-body human motion and forces exerted by both hands across six representative physical interaction tasks. Based on these demonstrations, we propose force-aware evaluation metrics that jointly assess whole-body tracking accuracy, robustness under different force levels, control effort, and episode survival through the Force-Aware Tracking Score (FATS) and complementary diagnostic metrics. We further establish a unified benchmark protocol that replays recorded interaction forces in simulation and provides a standardized evaluation interface for different humanoid control policies. Experiments on representative whole-body control policies demonstrate that force-aware evaluation reveals substantial performance differences that remain largely hidden under conventional no-force evaluation. ThorArena provides a practical and reproducible framework for studying force-aware humanoid interaction and offers a new benchmark for evaluating contact-rich humanoid behaviors.

Chinese Translation

人形机器人越来越被期望执行需要丰富接触的任务，这不仅要求准确的全身运动，还需要与周围物体和人类进行稳健的物理交互。尽管在人形动作模仿和全身控制方面的最新进展取得了显著的跟踪性能，但现有的数据集和基准测试主要集中于运动学动作，基本上忽视了同步的交互力。因此，当前的评估未能捕捉到外部交互力如何影响跟踪精度、稳定性和控制鲁棒性。在本文中，我们提出了ThorArena，这是一个基于人类示范的、关注力量的人形交互评估基准，它结合了同步的动作和力测量。我们收集了一个真实世界的交互数据集，该数据集同时捕捉六个典型物理交互任务中的全身人类运动和双手施加的力量。基于这些示范，我们提出了力量感知评估指标，该指标共同评估全身跟踪精度、在不同力量水平下的鲁棒性、控制努力和事件存活率，通过力量感知跟踪分数（Force-Aware Tracking Score, FATS）及补充诊断指标。我们进一步建立了统一的基准协议，该协议在仿真中重放记录的交互力，并为不同人形控制策略提供标准化的评估接口。对典型的全身控制策略的实验表明，力量感知评估揭示了在常规无力评估中大多隐藏的显著性能差异。ThorArena提供了一个实用且可复现的框架，用于研究力量感知的人形交互，并为评估接触丰富的人形行为提供了新的基准。

View on arXiv Download PDF AI Translation

cs.RO / 19 / 2607.06123

MP-MPPI: A Motion Primitive Guided Sampling-Based Optimizer for Model Predictive Control

MP-MPPI：一种基于运动原语引导的模型预测控制采样优化器

Mathisen, Marlon, Vaaler, Aksel, Egeland, Olav, Kelasidi, Eleni

Abstract

This paper proposes a novel method that extends the Model Predictive Path Integral (MPPI) method with motion primitives for additional structured sampling, which enhances the convergence towards a globally optimal solution. By evaluating motion primitives and perturbed control sequences in a real-time sampling-based optimization loop, this work addresses the limitations of the path planning capabilities of sampling-based controllers. The algorithm is implemented on a quadcopter simulator and tested on an obstacle field navigation task. It is demonstrated that the proposed approach enhances exploration of the control space while maintaining the fast, reactive behavior required for real-time control.

Chinese Translation

本文提出了一种新方法，将模型预测路径积分（Model Predictive Path Integral, MPPI）方法与运动原语相结合，以实现附加的结构化采样，从而增强向全局最优解的收敛性。通过在实时采样优化循环中评估运动原语和扰动控制序列，本研究解决了基于采样的控制器在路径规划能力方面的局限性。该算法在四旋翼模拟器上实现，并在障碍物领域导航任务中进行了测试。结果表明，该方法在保持实时控制所需的快速反应行为的同时，增强了控制空间的探索能力。

View on arXiv Download PDF AI Translation

cs.RO / 20 / 2607.06165

EAGOR: Embodied Reasoning in Omni-direction

EAGOR：全向体现在推理中的应用

Damodaran, Shriram, Debnath, Soumyaratna, Wu, Yan, Yau, Wei-Yun, Wang, Addison Lin

Abstract

Omni-directional (360{\deg}) cameras provide embodied agents with a holistic view of their surroundings, making them suited for directional reasoning in tasks such as navigation and object search. Existing Vision Language Models (VLMs) project 360{\deg} observations to 2D equirectangular projection (ERP) images and process them using architectures designed for perspective images. However, they ignore the spherical nature of 360{\deg} observations, where each pixel represents a viewing direction relative to the agent. Consequently, their direction estimates often become inconsistent under camera view transformations caused by agent motion. This limitation is particularly critical for map-free navigation, where the agent must continuously estimate the target direction in its egocentric frame. We propose EAGOR, a training-free, geometry-aware framework for embodied 360{\deg} directional reasoning. Instead of predicting target directions as ERP image coordinates, EAGOR formulates directional reasoning as recursive Bayesian estimation directly on the sphere. It maintains a continuous belief over target directions and propagates it equivariantly under agent motion without training the backbone VLMs. To achieve this, we introduce the Spherical Harmonic Belief Field (SH-BF), whose spherical harmonic representation provides a globally defined, rotation-aware basis for directional estimation on the spherical manifold. This formulation eliminates ERP seam discontinuities, latitude distortions, and interpolation errors. We evaluate EAGOR on two benchmark datasets and real-world experiments with a legged robot across directional reasoning tasks. EAGOR consistently outperforms existing methods, achieving average relative gains of +34.4% and +45.6% on HOS and OSR-Bench, respectively, while improving navigation success by +14.6%, reducing step count by 17.7%, and lowering mean angular error by 24.5%.

Chinese Translation

全向（360°）相机为具身代理提供了其周围环境的整体视图，使其适用于导航和物体搜索等任务中的方向推理。现有的视觉语言模型（VLMs）将360°观察投影到2D等距矩形投影（ERP）图像，并使用为透视图像设计的架构进行处理。然而，它们忽略了360°观察的球面特性，每个像素代表相对于代理的视角方向。因此，在因代理运动引起的摄像机视角变换下，它们的方向估计往往变得不一致。这种限制对无地图导航尤为关键，因为代理必须在其自我中心框架中持续估计目标方向。我们提出了EAGOR，一个无训练、几何感知的框架，用于具身360°方向推理。EAGOR不再将目标方向预测为ERP图像坐标，而是将方向推理公式化为在球面上的递归贝叶斯估计。它保持对目标方向的连续信念，并在代理运动下以等变方式传播，而无需训练基础的VLMs。为实现这一目标，我们引入了球谐信念场（Spherical Harmonic Belief Field，SH-BF），其球谐表示为球面流形上的方向估计提供了全局定义的、旋转感知的基础。此公式消除了ERP接缝不连续性、纬度失真和插值误差。我们在两个基准数据集和一项涉及四足机器人在方向推理任务中的实际实验上评估了EAGOR。EAGOR在性能上始终优于现有方法，在HOS和OSR-Bench上分别实现了平均相对增益+34.4%和+45.6%；同时导航成功率提升+14.6%，步数减少17.7%，平均角误差降低24.5%。

View on arXiv Download PDF AI Translation

cs.RO / 21 / 2607.06186

Calf-Integrated Arms for Bimanual Quadruped Loco-Manipulation

用于双手四足运动操控的小腿集成手臂

Pan, Yan, Ren, Yuanchuan, Chan, Chipui, Sun, Jingcheng, Zhou, Chengxu

Abstract

Most quadruped loco-manipulation designs trade manipulation capability against stance. A trunk-mounted arm sits high and usually carries a single arm; using the legs as manipulators lifts the manipulating leg off the ground; and even leg-mounted grippers reach two-handed tasks only by rearing onto the hind legs. This paper integrates a manipulator with a prismatic slider, two revolute joints, and a gripper into each front calf of a Unitree Go2. The two arms grasp objects at ground level and manipulate with both hands while all four feet stay planted, without rearing. With one arm carrying, the base stays free to walk. A vision-language model sequences skills from a predefined library at each skill boundary, conditioned on the head-camera image and task state, for long-horizon autonomy. In simulation, the design performs three bimanual tasks: a long-horizon cabinet task under autonomous skill selection, a cooperative two-handed lift, and an inter-arm handover.

Chinese Translation

大多数四足运动操控设计在操控能力与站姿之间进行权衡。一个安装在躯干上的手臂通常位于较高位置，并且通常仅携带一个手臂；使用腿部作为操控器会使操控的腿离开地面；即使是腿部安装的抓取器也只能通过站立在后腿上来完成双手任务。本文将一个带有伸缩滑块、两个旋转关节和一个抓取器的操控器集成到Unitree Go2的每个前小腿中。这两个手臂在地面高度抓取物体并双手操控，同时四只脚保持固定，不需要站立。一个手臂负责携带，基座保持自由行走。一个视觉语言模型在预定义库的每个技能边界处，根据头部摄像头图像和任务状态的条件，依次执行技能，以实现长时间的自主性。在仿真中，该设计执行了三个双手任务：一个在自主技能选择下的长时间橱柜任务、一个合作的双手搬运任务，以及一个跨手臂的物品交接任务。

View on arXiv Download PDF AI Translation

cs.RO / 22 / 2607.06222

APVI-SLAM: Real-Time Acoustic-Pressure-Visual-Inertial Localization and Photorealistic Mapping System in Complex Underwater Environment

APVI-SLAM：复杂水下环境中的实时声压视能定位与逼真地图构建系统

Zhang, Hanwen, Zhu, Yipeng, Guo, Xiaopeng, Huang, Huajian, Yeung, Sai-Kit

Abstract

Extreme subsea environments often cause severe feature de-gradation and estimator divergence in underwater visual-inertial SLAM. Although sensors like Doppler Velocity Logs (DVL) and pressure gauges provide auxiliary constraints, robust multi-sensor fusion during intermittent visual failure remains challenging. To address this, we present APVI-SLAM, a real-time multi-sensor fusion SLAM system that achieves both accurate underwater localization and photorealistic mapping. Our approach introduces a reliability-aware localization framework that dynamically reweights sensor estimators and employs a sliding-window freezing strategy to recover from tracking failures, substantially enhancing system robustness. Furthermore, for high-fidelity scenes reconstruction, we propose an efficient quadtree-guided mapping module that facilitates incremental water-medium modeling and 3D Gaussian optimization. Recognizing the lack of benchmark for underwater mapping evaluation, we also contribute a coral reef surveying dataset with synchronized multi-modality data. Extensive experiments on public and our proposed benchmarks demonstrate that APVI-SLAM achieves state-of-the-art localization and reconstruction quality at real-time speeds.

Chinese Translation

极端的海底环境常常导致水下视觉惯性SLAM中的特征严重退化和估计器发散。尽管如多普勒速度计（Doppler Velocity Logs, DVL）和压力传感器等传感器提供了辅助约束，但在间歇性视觉失败期间实现稳健的多传感器融合仍然具有挑战性。为了解决这个问题，我们提出了APVI-SLAM，这是一种实时多传感器融合SLAM系统，能够实现精确的水下定位和逼真的地图构建。我们的方法引入了一个考虑可靠性的定位框架，动态调整传感器估计器的权重，并采用滑动窗口冻结策略来恢复追踪失败，从而显著增强系统的鲁棒性。此外，为了实现高保真场景重建，我们提出了一个高效的四叉树引导映射模块，以促进增量水介质建模和三维高斯优化。鉴于缺乏水下映射评估的基准，我们还贡献了一个与多模态数据同步的珊瑚礁调查数据集。在公共和我们提出的基准上进行的广泛实验表明，APVI-SLAM在实时速度下实现了最先进的定位和重建质量。

View on arXiv Download PDF AI Translation

cs.RO / 23 / 2607.06248

RoboVAST: Automated Scenario-Based Validation of Robots at Scale

RoboVAST：大规模机器人基于场景的自动验证

Pasch, Frederik, Wiest, Samuel, Ortega, Argentina, Hochgeschwender, Nico

Abstract

Validation of robotic systems critically depends on the operating conditions under which they are assessed. Scenario selection and variation are often manual, experience-driven, and difficult to scale, which harms reproducibility and weakens validation conclusions. We propose a scenario-based methodology that models scenarios compositionally and formalizes how these dimensions are varied, instantiated, executed, and interpreted. Building on this, we introduce RoboVAST, a framework that realizes declarative campaign specifications, plugin-based scenario generation, and scalable containerized execution with integrated result analysis. We demonstrate the approach with a navigation dataset comprising 5480 scenario configurations and over 100000 execution runs across five indoor maps with varied paths, sensor noise, software parameters, and obstacle settings, totaling more than 1800 hours of simulated operation and 1873 km traveled. Twenty repetitions per configuration allow us to distinguish systematic failures from stochastic anomalies.

Chinese Translation

机器人系统的验证在很大程度上依赖于评估时的操作条件。场景选择和变更通常是手动的、依赖经验的，并且难以扩展，这影响了可重复性并削弱了验证结论。我们提出了一种基于场景的方法论，该方法论以组成的方式建模场景，并正式化这些维度的变化、实例化、执行和解释。基于此，我们介绍了RoboVAST，一个实现声明式任务规范、插件式场景生成和可扩展的容器化执行集成结果分析的框架。我们通过一个包含5480个场景配置的导航数据集以及跨越五个室内地图的超过100000次执行运行进行演示，涉及不同的路径、传感器噪声、软件参数和障碍物设置，总计超过1800小时的模拟操作和1873公里的行驶距离。每个配置的20次重复运行使我们能够区分系统性故障与随机异常。

View on arXiv Download PDF AI Translation

cs.RO / 24 / 2607.06256

Diagnosing Semantic Handoff Failures in Agent-Orchestrated Vision-Language-Action Skill Composition

代理协调的视觉-语言-动作技能组合中的语义交接失败诊断

Rui, Ke, Zuo, Yushen, Wang, Jiawei, Jia, Haoran, Ma, Jinming, Zhou, Weitao, Li, Minglei

Abstract

Long-horizon household tasks require robots to compose many language-conditioned skills, yet the boundary between consecutive skills is rarely explicit. A skill may satisfy its own postcondition while leaving the robot, objects, or camera views in a state from which the next skill cannot reliably start. We study this semantic handoff problem in BEHAVIOR-1K through an agent-orchestrated vision-language-action execution harness. The harness invokes $\pi_{0.5}$-based skill checkpoints trained from cleaned BEHAVIOR-1K demonstrations, assigns each skill typed arguments and a step budget, and uses multi-view vision-language model verification to decide whether execution should advance, retry, or replan. To separate isolated skill competence from long-horizon compositional robustness, we evaluate the same checkpoints under two initial-state distributions: clean skill-boundary snapshots and chained terminal states produced by previous skills. Selected navigation, grasping, placement, and door-opening skills achieve 77--100% success from clean snapshots under human-reviewed verification, yet composed rollouts still frequently stall from chained states. Execution traces attribute these failures to next-skill readiness, target grounding, and low-level control execution, revealing a substantial gap between single-skill success and reliable long-horizon task completion. These findings turn near-zero end-to-end task success into actionable diagnostics, showing that future VLA skill libraries must learn robustness to the messy chained-state distribution that clean demonstrations systematically underrepresent.

Chinese Translation

长时程的家庭任务要求机器人组合多种基于语言的技能，但连续技能之间的边界往往不甚明确。一项技能可能满足其自身的后置条件，同时使得机器人、物体或相机视角处于一种下一个技能无法可靠启动的状态。我们通过一个代理协调的视觉-语言-动作执行框架在 BEHAVIOR-1K 中研究这一语义交接问题。该框架调用基于 $ heta_{0.5}$ 的技术检查点，这些检查点是从清理过的 BEHAVIOR-1K 演示数据中训练出来的，为每项技能分配类型化参数和步骤预算，并使用多视角视觉-语言模型验证来决定执行是应继续、重试还是重新规划。为了将孤立技能的能力与长时程组合的鲁棒性区分开，我们在两种初始状态分布下评估相同的检查点：清晰的技能边界快照和由先前技能生成的链式终端状态。选定的导航、抓取、放置和开门技能在经过人类审核的验证下，从清晰快照中实现77%到100%的成功率，但组合回放仍然经常因链式状态而停滞。执行轨迹将这些失败归因于下一技能的准备情况、目标定位和低级控制执行，揭示了单一技能成功与可靠长时程任务完成之间的显著差距。这些发现将近乎零的端到端任务成功转化为可操作的诊断，表明未来的 VLA 技能库必须学习对这种清洁演示系统性欠代表的混乱链状态分布的鲁棒性。

View on arXiv Download PDF AI Translation

cs.RO / 25 / 2607.06262

Optimal Transport Q-Learning for Flow Policy Steering and Acceleration

流政策引导与加速的最优传输 Q 学习

Sochopoulos, Andreas, Whitammer, Esmeralda S., Tsagkas, Nikolaos, Moura, João, Gienger, Michael, Vijayakumar, Sethu

Abstract

Diffusion and flow policies have recently demonstrated remarkable performance in robotic applications by accurately capturing multimodal robot trajectory distributions, especially in the context of vision language action (VLA) models. However, high quality policy performance also requires fast inference and high quality demonstrations, which are often hard to get. Lack of these leads to suboptimal policy behaviors and failure under distribution shifts. In this work we address the problem of fine-tuning and accelerating suboptimal flow-based policies using the robot's experience through RL post-training. We introduce Optimal Transport Q-Learning (OTQL), a new method for finetuning flow policies using advantage weighted conditional optimal transport flow matching. OTQL can finetune and accelerate flows with an interaction budget of 50-60 episodes while avoiding computationally expensive distillation in simulation and real-world robot tasks. Our results show that OTQL post-trains flow policies using the robot's own experience, increasing average success percentage of single-task policies from 36% to 86% and of a pre-trained VLA from 38% to 76% while reducing the number of inference steps per action generation by 70%.

Chinese Translation

扩散与流政策在机器人应用中最近展现了卓越的性能，准确捕捉了多模态机器人轨迹分布，特别是在视觉语言行动（VLA）模型的背景下。然而，高质量的政策性能还需要快速推理和高质量的示范，这些通常难以获得。这种缺乏会导致次优政策行为以及在分布变化下的失败。在本研究中，我们解决了通过RL后训练使用机器人经验来微调和加速次优基于流的政策的问题。我们提出了最优传输 Q 学习（Optimal Transport Q-Learning, OTQL），一种通过优势加权条件最优传输流匹配微调流政策的新方法。OTQL可以在避免在仿真和真实机器人任务中进行计算成本高昂的蒸馏的情况下，以50-60个回合的交互预算微调和加速流动。我们的结果表明，OTQL利用机器人自己的经验对流政策进行后训练，使单任务政策的平均成功率从36%提高到86%，预训练的VLA的成功率从38%提高到76%，同时将每次行动生成的推理步骤减少了70%。

View on arXiv Download PDF AI Translation

cs.RO / 26 / 2607.06323

LAMP: Latent Motion Prior-Guided Real-World Learning for Dexterous Hand Manipulation

LAMP：基于潜在运动先验的现实世界灵巧手操控学习

Yang, Xinye, Ma, Zhiyuan, Yu, Hongze, Chen, Yuanpei, Yang, Yaodong, Chai, Xiaojie, Chen, Xinlei, Yu, Chao

Abstract

Real-world learning for dexterous hands remains brittle because high-dimensional hand actions amplify imitation errors and make reinforcement-learning exploration prone to contact-breaking motion. While combining imitation learning (IL) with online reinforcement learning (RL) can reduce manual supervision, unconstrained exploration in raw hand-action spaces is sample-inefficient and risky for physical hardware. We introduce a latent motion prior module (\prior{}) that maps recent hand-action histories to a compact, history-conditioned latent prior and decodes continuous latent commands into executable high-dimensional hand targets. Built on this prior, \method{} is a three-stage real-world dexterous learning framework: it pretrains \prior{} from demonstrations, trains a visuomotor policy that predicts native arm commands and latent hand-action offsets, and improves the policy with online residual RL in the same latent hand-action space. This shared, decodable interface lets residual exploration make local corrections near demonstrated, contact-consistent hand motions rather than perturbing every finger joint independently. We evaluate \method{} on four real-robot dexterous manipulation tasks against raw, linear, and discrete hand-action interfaces. Starting from small task-specific demonstration sets, \method{} achieves a 56.25\% average IL success rate and raises it to 98.75\% after online RL, reaching 100\% final success on three tasks and 95\% on the remaining task.

Chinese Translation

现实世界中灵巧手的学习仍然不够稳健，因为高维手部动作会放大模仿误差，使得强化学习探索容易出现接触中断的运动。尽管将模仿学习（IL）与在线强化学习（RL）相结合可以减少人工监督，但在原始手部动作空间中的无约束探索效率低下且对物理硬件存在风险。我们引入了一种潜在运动先验模块（ extit{prior}），该模块将最近的手部动作历史映射到一个紧凑的、历史条件的潜在先验，并将连续的潜在指令解码为可执行的高维手部目标。在此先验基础上， extit{method} 是一个三阶段的现实世界灵巧学习框架：它从演示中预训练 extit{prior}，训练一个视觉运动策略，该策略预测原生的手臂指令和潜在的手部动作偏移，并在同一潜在手部动作空间中通过在线残差强化学习来改进策略。这种共享的可解码接口使得残差探索能够在接触一致的手部动作附近进行局部修正，而不是独立扰动每个手指关节。我们在四个真实机器人灵巧操控任务上评估了 extit{method}，与原始、线性和离散手部动作接口进行比较。从小规模的任务特定演示集开始， extit{method} 实现了 56.25\% 的平均 IL 成功率，并在在线 RL 后提高到 98.75\\%，在三个任务上达到 100\\% 的最终成功率，在剩余任务上达到 95\\%。

View on arXiv Download PDF AI Translation

cs.RO / 27 / 2607.06337

OrchardBench: A Physically-Grounded, GPU-Parallel Apple-Orchard Simulation Benchmark for Agricultural Robotics

OrchardBench：一个基于物理、GPU并行的苹果 orchard 模拟基准，用于农业机器人

Munn, Humphrey

Abstract

Robotic tree-fruit harvesting is a flagship problem for agricultural automation, but progress is bottlenecked by the cost and irreproducibility of field experiments: an orchard is available only weeks a year, every tree is different, and a control error can permanently damage the crop or the plant. The tree models used in graphics and agronomy are geometrically detailed but physically inert, while the GPU-parallel simulators used in robot learning contain no plausible trees. We present OrchardBench, a physically-grounded, GPU-parallel simulation of apple-orchard trees on the Newton engine. Each tree is grown by a stochastic L-system and instantiated as a fully articulated body: branches are compliant torsional spring-dampers whose stiffness follows Euler-Bernoulli beam theory, they break at a wood modulus of rupture and fall as free hinges, and apples are independent bodies on stem tethers that detach at literature-grounded pull forces and load the branch when pulled. A moving, density-controllable foliage layer occludes the canopy as real leaves do. Every physical parameter is tied to a published source. Per-environment domain randomization makes each batched world a distinct tree, and a mobile manipulator with a wrist depth camera closes the loop with geometric fruit perception and an autonomous harvesting baseline. Careful engineering of the solver and the model lets OrchardBench run many parallel environments at interactive rates on a laptop GPU. We define the tasks and a metric suite spanning harvest completeness, throughput, and plant damage (with a per-canopy-zone breakdown), and report baseline results across foliage, fruit load, terrain, canopy zone, and parallelism. The analytic baseline succeeds on about 40% of the fruit it detects and harvests only about an eighth of the reachable fruit on a tree, leaving clear headroom for novel autonomy approaches.

Chinese Translation

树果的机器人收获是农业自动化中的一个关键问题，但由于田野实验的成本和不可重复性，进展受到了瓶颈：果园一年只有几周可用，每棵树都有所不同，而控制错误可能会永久性地损害作物或植物。图形学和农业中使用的树模型在几何上非常详细，但在物理上是惰性的，而在机器人学习中使用的 GPU 并行模拟器中没有可行的树木。我们提出了 OrchardBench，这是一个基于物理的、在 Newton 引擎上并行的苹果 orchard 树的模拟。每棵树由随机 L 系统生成并实例化为一个完全有关节的主体：树枝是符合欧拉-伯努利梁理论的柔性扭转弹簧阻尼器，达到木材抗拉强度时会断裂并作为自由铰链下落，苹果是独立的物体，通过茎绳连接，当施加的拉力达到文献基础的拉力时会断开，并在被拉时给树枝施加负载。一个可移动的、密度可控的树叶层像真实的叶子一样遮挡树冠。每个物理参数都与已发布的来源相关。每个环境的领域随机化使每个批处理的世界都是一个独特的树，并且一个带有手腕深度相机的移动操控器通过几何果实感知和自主收获基线闭合了循环。通过对求解器和模型的精心设计，OrchardBench能够在笔记本 GPU 上以交互速率运行多个并行环境。我们定义了任务和一个涵盖收获完整性、吞吐量和植物损伤（按树冠区域分类）的指标套件，并报告了在树叶、果实负载、地形、树冠区域和并行性上的基线结果。分析基线在检测到的约40%的果实上成功，并仅收获树上可达果实的约八分之一，留下了显著的空间用于新的自主方法。

View on arXiv Download PDF AI Translation

cs.RO / 28 / 2607.06344

Responsible Personalisation: The Double-Edged Sword of Personalisation in Human-Robot Interaction

负责任的个性化：人机交互中个性化的双刃剑

Andriella, Antonio, Nasir, Jauwairia, Rezzani, Andrea, Kubota, Alyssa, Lacroix, Dimitri, Love, Tamlin, Civit, Aniol, Charisi, Vicky, Andre, Elisabeth, Louie, Wing-Yue Geoffrey

Abstract

While personalisation is becoming a defining capability in human-robot interaction (HRI), the existing literature on responsible personalisation remains fragmented, offering isolated accounts of ethical risks without a structured understanding of how they emerge across interaction contexts. This gap is particularly critical in HRI, where robots' embodiment and social presence can amplify and reshape such risks or generate new types of risks. We present a lifecycle-based and context-sensitive framework for personalised HRI, grounded in an embodiment-aware perspective. The framework combines stages of the personalisation process with interaction characteristics (short-term vs. long-term, open-domain vs. closed-domain), enabling systematic analysis of how risks arise and evolve. Building on this, we conduct an integrative analysis of key ethical risks, including autonomy erosion, biased user modelling, manipulation, dehumanisation, and privacy violations, and examine how they manifest across contexts. We translate these insights into actionable design recommendations and outline open research challenges. By structuring both the design space and risk landscape of personalised HRI, this work provides a foundation for more systematic, transparent, and ethically grounded approaches to personalised robot behaviour.

Chinese Translation

虽然个性化正在成为人机交互（HRI）中的一个定义性能力，但现有关于负责任个性化的文献仍然支离破碎，提供了孤立的伦理风险描述，而缺乏对这些风险如何在交互情境中出现的系统理解。这一缺口在HRI中尤为关键，因为机器人的具身性和社会存在感可能会放大和重塑这些风险，或产生新的风险类型。我们提出了一个基于生命周期和情境敏感的个性化HRI框架，该框架基于对具身性的关注。该框架将个性化过程的阶段与交互特征（短期与长期、开放领域与封闭领域）结合起来，使得对风险如何产生和演变的系统分析成为可能。在此基础上，我们对关键伦理风险进行了综合分析，包括自主性侵蚀、用户建模偏见、操控、非人性化和隐私侵犯，并考察了这些风险在不同情境中的表现。我们将这些见解转化为可操作的设计建议，并概述了开放的研究挑战。通过构建个性化HRI的设计空间和风险景观，这项工作为更系统、透明和以伦理为基础的个性化机器人行为方法提供了基础。

View on arXiv Download PDF AI Translation

cs.RO / 29 / 2607.06370

Training-Free Acceleration for Vision-Language-Action Models with Action Caching and Refinement

无需训练的视觉-语言-动作模型加速：基于动作缓存与优化

Oi, Ryuji, Otsuka, Hikari, Matsushima, Kosuke, Ichikawa, Yuki, Motomura, Masato, Kaneko, Tatsuya, Fujiki, Daichi

Abstract

Vision-Language-Action (VLA) models have emerged as a promising approach for generalizable robotic manipulations. In particular, flow matching-based VLA models have shown remarkable success due to their capability to generate precise and smooth action sequences and capture multimodal distributions. However, the iterative denoising process in the action head acts as a major computational bottleneck, posing a critical challenge for real-time deployment. To address this challenge, we propose ActionCache, a plug-and-play external cache that opportunistically reuses past intermediate actions to warm-start generations from the vicinity of target actions, thereby drastically reducing the inference latency. Specifically, ActionCache stores the intermediate actions with compact multimodal keys, which enables retrieval from similar past contexts across different episodes or even different tasks. Experimental results in simulation and real-world environments demonstrate that ActionCache maintains high task success rates in a low-latency regime, achieving inference acceleration of up to $11.75\times$ and $34.43\times$ for representative flow-based VLA models, $\pi_{0.5}$ and GR00T-N1.6, respectively.

Chinese Translation

视觉-语言-动作（Vision-Language-Action, VLA）模型作为一种具有广泛适应性的新兴方法，在可泛化的机器人操控领域展现出良好的前景。特别是，基于流匹配的VLA模型由于其生成精确且平滑的动作序列及捕捉多模态分布的能力，获得了显著成功。然而，动作头中的迭代去噪过程是一个主要的计算瓶颈，成为实时部署的关键挑战。为了解决这一问题，我们提出了ActionCache，一种即插即用的外部缓存，它有机地重用过去的中间动作，从目标动作附近快速启动生成，从而大幅降低推理延迟。具体而言，ActionCache使用紧凑的多模态关键字存储中间动作，使得可以从不同回合甚至不同任务中的相似历史上下文中检索。模拟和真实环境中的实验结果表明，ActionCache在低延迟情况下保持高任务成功率，对于具有代表性的基于流的VLA模型$ ext{π}_{0.5}$和GR00T-N1.6，分别实现了高达$11.75 imes$和$34.43 imes$的推理加速。

View on arXiv Download PDF AI Translation

cs.RO / 30 / 2607.06383

Towards Real-World Applications with an Autonomous Powered Wheelchair

朝着具备自主功能的电动轮椅的实际应用

Arreghini, Simone, Giusti, Alessandro, Bordini, Alex, Ferrara, Enrico, Fulgoni, Giovanni, Paolillo, Antonio

Abstract

Wheelchair users call for assistive mobility systems that provide active support, adapt to dynamic environments, and are intuitive and user-friendly. However, powered wheelchairs typically still provide limited autonomy and lack effective integration with advanced perception and navigation capabilities, particularly in complex real-world environments. This paper presents a preliminary study toward autonomous powered wheelchairs for real-world assistive mobility. We introduce a proof-of-concept prototype that integrates autonomous perception, gesture-based interaction, and navigation on a commercially available self-balancing powered wheelchair. The proposed system builds upon Genny Zero, a commercial self-balancing wheelchair that enables hands-free and intuitive operation through body-weight shifting. To extend its capabilities toward autonomous operation, we integrate an RGB-D camera for human-aware perception and interaction, together with a LiDAR sensor for localization and navigation. We demonstrate the integrated system in two assistive applications: (i) hailing, allowing users to call the wheelchair from a distance; and (ii) people-following, where the wheelchair follows a person using leader-follower strategies, including a constrained indoor navigation example. The results highlight the potential of combining autonomous robotics with assistive mobility platforms, while also showing the feasibility of the proposed integration and identifying the main technical challenges that must be addressed before moving toward user-ready, accessible, and intelligent mobility solutions. A video demonstrating the experimental setup and results is available at: https://youtu.be/LVAix_Qx7bM.

Chinese Translation

轮椅用户呼吁开发能够提供主动支持、适应动态环境并且直观易用的辅助移动系统。然而，电动轮椅通常仍提供有限的自主性，并且缺乏与先进的感知和导航能力的有效整合，尤其是在复杂的实际环境中。本文呈现了一项针对实际辅助移动的自主电动轮椅的初步研究。我们介绍了一个概念验证原型，该原型整合了自主感知、基于手势的交互和导航功能，应用于一款商业化的自平衡电动轮椅Genny Zero。为了扩展其自主操作的能力，我们整合了一台用于人类感知和交互的RGB-D摄像头，以及一台用于定位和导航的激光雷达（LiDAR）传感器。我们在两个辅助应用中展示了集成系统：(i) 召唤功能，允许用户在一定距离呼叫轮椅；(ii) 跟随模式，轮椅使用领导-跟随策略跟随某个人，包括一个受限的室内导航示例。结果突显了将自主机器人技术与辅助移动平台结合的潜力，同时展示了所提议的整合的可行性，并识别了在向用户准备好的、可获取的和智能的移动解决方案迈进之前必须解决的主要技术挑战。一个演示实验设置和结果的视频可在以下链接观看：https://youtu.be/LVAix_Qx7bM。

View on arXiv Download PDF AI Translation

cs.RO / 31 / 2607.06388

Learning to Throw Objects Safely in Multi-Obstacle Environments

在多障碍环境中安全投掷物体的学习

Kasaei, Mohammadreza, Voncina, Klemen, Kasaei, Hamidreza

Abstract

Robotic throwing enables fast and efficient object placement beyond the robot's immediate workspace, but reliable throwing in cluttered environments remains underexplored. Existing approaches, such as TossingBot, learn throwing strategies from visual input but assume obstacle-free settings. In this paper, we address the problem of throwing objects into a target basket while avoiding obstacles placed randomly in the scene. We introduce a potential field state representation that compactly encodes both basket attraction and obstacle repulsion on a fixed-size grid, enabling reinforcement learning (RL) policies to generalize across arbitrary numbers and configurations of obstacles. The policy is initialized from kinesthetic demonstrations and optimized in simulation using three state-of-the-art RL algorithms (SAC, DDPG, TD3). Among these, SAC achieves the most consistent performance across scenarios. We compare the potential field representation against explicit state encodings and demonstrate that it achieves higher success rates and better scalability to unseen obstacle configurations. Real-robot experiments with unseen throwable objects confirm robust sim-to-real transfer, achieving up to $90\%$ success in cluttered scenes. These results demonstrate that PFR provides a practical and robust representation for safe and efficient robotic throwing in unstructured environments. A video showcasing our experiments is available at: https://youtu.be/ZZnJf8ua2dE

Chinese Translation

机器人投掷能够实现超出机器人即时工作空间的快速有效物体放置，但在杂乱环境中可靠的投掷仍然未被充分探讨。现有的方法，如TossingBot，基于视觉输入学习投掷策略，但假设环境中没有障碍。在本文中，我们解决了在场景中随机放置障碍物的情况下，将物体投掷到目标篮中这一问题。我们引入了一种潜在场状态表示，在固定大小的网格上紧凑地编码了篮子的吸引力和障碍物的排斥力，使得强化学习（RL）策略能够在任意数量和配置的障碍物之间进行泛化。该策略从运动示范中初始化，并使用三种最先进的RL算法（SAC、DDPG、TD3）在仿真中进行优化。在这些算法中，SAC在不同场景中表现出最一致的性能。我们将潜在场表示与显式状态编码进行了比较，结果表明其实现了更高的成功率，并且在应对未知障碍配置方面具有更好的可扩展性。针对未知可投掷物体的真实机器人实验确认了可靠的模拟到现实转移，在杂乱场景中成功率达到90%。这些结果表明PFR为在非结构化环境中安全有效的机器人投掷提供了一种实用且稳健的表示。展示我们实验的视频可在此查看：https://youtu.be/ZZnJf8ua2dE

View on arXiv Download PDF AI Translation

cs.RO / 32 / 2607.06403

From Foundation to Application: Improving VLA Models in Practice

从基础到应用：实践中改进 VLA 模型

Wu, Wei, Wang, Fangjing, Lu, Fan, Sun, He, Liu, Shi, Wang, Yunnan, Yan, Yibin, Wang, Yong, Ma, Shuailei, Wang, Xinyang, Liu, Yibin, Yang, Shuai, Zhou, Tianxiang, Zhang, Kejia, Zhou, Lei, Su, Cheng, Xue, Nan, Tan, Bin, Zhang, Han, Zhang, Youchao, Liao, Fei, Zhu, Xing, Shen, Yujun, Zheng, Kecheng

Abstract

Despite recent progress of VLA foundation models, the disparity between laboratory conditions and real-world applications continues to impede their practical implementation. To bridge this gap, we present LingBot-VLA 2.0, which advances LingBot-VLA through improvements in three functional domains. (1) Generalization across tasks and embodiments. Compared to the previous version, we revamp the data processing pipeline and curate around 60,000 hours of data for pretraining, including 50,000 hours of robot trajectories spanning 20 robot configurations and 10,000 hours of egocentric human videos. (2) Expanded action space in addition to dual-arm hardware platforms. In particular, our system accommodates degrees of freedom for the heads, waists, mobile bases, and dexterous hands, thereby empowering the robots to tackle more complex tasks in practical scenarios. (3) Predictive dynamics modeling for improved temporal reasoning. Specifically, we formulate future prediction as a proxy task, facilitated by a video representation model for semantic priors and a depth estimation model for geometric cues. Evaluations on the GM-100 benchmark, conducted in a generalist setting, validate the beneficial impact of these proposed modifications. Furthermore, benefiting from the expanded pretraining data that covers whole-body degrees of freedom, LingBot-VLA-2.0 demonstrates strong cross-embodiment long-horizon mobile manipulation capability across the two robotic platforms.

Chinese Translation

尽管 VLA 基础模型近期取得了进展，但实验室条件与实际应用之间的差距仍然妨碍其实际实施。为了解决这一问题，我们推出了 LingBot-VLA 2.0，通过在三个功能领域的改进来推进 LingBot-VLA。 (1) 任务和表现形式的泛化。与之前版本相比，我们改进了数据处理流程，并为预训练策划了大约 60,000 小时的数据，包括 50,000 小时的机器人轨迹，涵盖 20 种机器人配置，以及 10,000 小时的自我中心人类视频。 (2) 除双臂硬件平台外，还扩展了动作空间。特别地，我们的系统为头部、腰部、移动底座和灵巧手提供了自由度，使得机器人能够在实际场景中处理更复杂的任务。 (3) 为改善时间推理而进行的预测动态建模。具体而言，我们将未来预测形式化为代理任务，借助视频表示模型获取语义先验和深度估计模型获取几何线索。针对 GM-100 基准的评估在通用设置下验证了这些提议修改的积极影响。此外，借助覆盖全身自由度的扩展预训练数据，LingBot-VLA-2.0 展现了在两种机器人平台上强大的跨表现形式长时移动操作能力。

View on arXiv Download PDF AI Translation

cs.RO / 33 / 2607.06438

WristMimic: Full-Body Humanoid Control with Wrist-Guided Manipulation

WristMimic：基于手腕引导的全身类人控制

Yu, Wongyun, Kim, Youngwoon, Cho, Minsu

Abstract

Retargeting human object interaction demonstrations to physics based simulation requires reproducing not only body motion but also the object motion and contacts that make manipulation succeed. However, position only hand trajectories do not specify the contact forces needed to manipulate objects, and directly tracking them can overconstrain contact rich finger behavior. We introduce WristMimic, a wrist guided whole body control framework that explicitly separates contact free body motion from contact rich hand manipulation. The contact free body and wrist are guided by kinematic pose targets, whereas the fingers are not directly supervised by human hand pose. Instead, they learn grasping and manipulation behaviors from object tracking and contact outcomes. Our key insight is that the wrist is the natural gate between these two regimes. It is largely free from contact and can be tracked kinematically, yet it determines the global hand configuration and places the fingers within reachable grasp affordances. To ensure reliable wrist placement during interaction, we introduce wrist specific reset constraints and reward prioritization. Experiments show that WristMimic matches or surpasses methods using full finger pose supervision while enabling finger agnostic retargeting across diverse hand embodiments.

Chinese Translation

将人类物体交互示范重定向到基于物理的仿真，不仅需要再现身体运动，还需要再现物体运动和接触，这些因素使得操控得以成功。然而，仅仅依靠位置的手轨迹并不能指定操控物体所需的接触力，直接跟踪这些力可能会对接触丰富的手指行为施加过多约束。我们提出了WristMimic，一个手腕引导的全身控制框架，明确区分无接触的身体运动与接触丰富的手部操控。无接触的身体和手腕由运动学姿态目标引导，而手指则不直接受到人类手部姿态的监督。相反，它们从物体跟踪和接触结果中学习抓取和操控行为。我们的关键见解是，手腕是这两种状态之间的自然通道。手腕在很大程度上不受接触影响，可以通过运动学方法进行跟踪，但它决定了手的整体配置，并将手指放置在可达的抓取能力范围内。为了确保在交互过程中手腕的可靠放置，我们引入了特定于手腕的重置约束和奖励优先级。实验表明，WristMimic的表现与使用完整手指姿态监督的方法相匹配或超越，同时实现了在多样化手部表现间的无手指特征重定向。

View on arXiv Download PDF AI Translation

cs.RO / 34 / 2607.06442

SIEVE: Structure-Aware Data Selection for Imitation Learning with VLA Models

SIEVE：针对模仿学习的结构感知数据选择方法与视觉-语言-动作模型

Wu, Changti, Yu, Bin, Shen, Zhaolong, Lian, Shijie, Lin, Xiaopeng, Huang, Cong, Zhang, Zhirui, Zhang, Lei, Chen, Kai

Abstract

Vision-Language-Action (VLA) models are typically trained by imitation learning on large-scale robot demonstration datasets, but more data does not necessarily yield better policies due to redundancy, noise, and uneven coverage. Existing data selection methods often assess demonstrations at either the trajectory or state-action level, missing the reusable structures that compose long-horizon behaviors. In this paper, we propose SIEVE, a structure-aware data selection method for VLA imitation learning. SIEVE views demonstrations as compositions of reusable primitives and transition interfaces. It first discovers visuo-motor primitives from segmented trajectories, then allocates selection budgets to composition patterns by maximizing reuse-aware structural exposure under diminishing returns. Finally, it selects medoid trajectories within each composition-pattern bucket to retain central, stable, and imitation-friendly demonstrations. Experiments across multiple datasets, benchmarks, and VLA models show that SIEVE consistently outperforms competitive data selection baselines. Notably, SIEVE can surpass full-data training while using only 50% of demonstrations and 50% of training steps, suggesting that reusable structure, captured through primitives and transitions, is an important signal for efficient VLA imitation learning.

Chinese Translation

视觉-语言-动作（VLA）模型通常通过在大规模机器人演示数据集上进行模仿学习进行训练，但更多数据并不一定会产生更好的策略，因为可能存在冗余、噪声和不均匀覆盖的问题。现有的数据选择方法往往只在轨迹或状态-动作层面评估演示，忽视了构成长时段行为可重复使用的结构。在本文中，我们提出了SIEVE，一种针对VLA模仿学习的结构感知数据选择方法。SIEVE将演示视为可重复使用的原语和过渡接口的组合。它首先从分段轨迹中发现视觉-运动原语，然后通过最大化重用感知的结构表现来分配选择预算，以实现边际收益递减的优化。最后，它在每个组合模式桶中选择中值轨迹，以保留中心、稳定且易于模仿的演示。我们在多个数据集、基准和VLA模型上的实验结果表明，SIEVE始终优于竞争性的数据选择基线。值得注意的是，SIEVE在仅使用50%的演示和50%的训练步骤的情况下，能够超越全数据训练，表明通过原语和过渡捕获的可重复使用结构是高效VLA模仿学习的重要信号。

View on arXiv Download PDF AI Translation

cs.RO / 35 / 2607.06464

Hilti-Trimble-Oxford Dataset: 360 Visual-Inertial Benchmark with Floor Plan Priors for SLAM and Localization

Hilti-Trimble-Oxford 数据集：带有平面图先验的 360 度视觉惯性基准测试用于 SLAM 和定位

Centanni, Samuele, Zhang, Yuhao, Tao, Yifu, Kindle, Julien, Neuhaus, Frank, Koß, Tilman, Patel, Aryaman, Helmberger, Michael, Szymańska, Emilia, Gräber, Torben, Fallon, Maurice

Abstract

Automated progress monitoring on construction sites is an active area of research and development. Robot and human-carried mapping systems have been developed to build 3D maps of building and infrastructure projects. While LiDAR-based mapping systems achieve high accuracy, the cost of LiDAR can be prohibitive. Consumer-grade cameras with wide field of view ("360 cameras") combined with embedded inertial measurement units (IMUs) provide a cost-effective alternative. To support change detection and progress monitoring, highly accurate visual Simultaneous Localization and Mapping (SLAM) and floor plan-referenced localization systems are required. In this paper we present a high-quality dataset collected at an active construction site, which captures realistic challenges such as variable lighting conditions, moving workers, fast motions, and repetitive structures. The dataset offers thirty visual-inertial sequences recorded across seven floors over an eight-month period of the construction project. Ground truth trajectories were collected using a high quality LiDAR-inertial SLAM system rigidly attached to the 360 camera. Additionally, we report the results of an open research challenge evaluating the best visual SLAM and localization systems from around the world. The Challenge attracted substantially higher participation in SLAM, with 62 teams compared to 22 in floor-plan-referenced localization, reflecting the broader maturity of SLAM methods. The higher errors in localization further highlight the difficulty of this task in construction and point to the need for continued research, which this dataset is intended to support. The dataset and the benchmark are publicly available at: https://hilti-trimble-challenge.com/dataset-2026.

Chinese Translation

建筑工地的自动化进度监测是一个活跃的研究和开发领域。已经开发出机器人和人携带的测绘系统，以构建建筑和基础设施项目的 3D 地图。虽然基于 LiDAR 的测绘系统实现了高精度，但 LiDAR 的成本可能过于昂贵。结合嵌入式惯性测量单元（IMUs）的消费级广角相机（“360 摄像头”）提供了一种具有成本效益的替代方案。为了支持变更检测和进度监测，需要高度准确的视觉同步定位与地图构建（SLAM）和基于平面图的定位系统。在本文中，我们呈现了在一个活跃的建筑工地收集的高质量数据集，该数据集捕捉了诸如变化的光照条件、移动工人、快速运动和重复结构等现实挑战。该数据集提供了在建筑项目的八个月期间跨七层记录的三十个视觉惯性序列。真实轨迹是使用高质量的 LiDAR-惯性 SLAM 系统与 360 摄像头刚性连接收集的。此外，我们报告了一个开放研究挑战的结果，该挑战评估了来自世界各地的最佳视觉 SLAM 和定位系统。该挑战在 SLAM 方面吸引了显著更高的参与度，参与团队数量为 62 支，而基于平面图的定位仅为 22 支，反映出 SLAM 方法的更广泛成熟度。定位中的较高误差进一步突显了这一任务在建筑中的困难，并指出了持续研究的必要性，而该数据集旨在支持这一研究。该数据集和基准测试可在以下网址公开获取：https://hilti-trimble-challenge.com/dataset-2026。

View on arXiv Download PDF AI Translation

cs.RO / 36 / 2607.06499

Clustering-Embedded Model Predictive Path Integral Control: Avoiding Averaging-Induced Failure and Enabling Efficient Cluster Selection for Dynamic Obstacles

嵌入聚类的模型预测路径积分控制：避免平均引起的失败并实现动态障碍物的高效聚类选择

Liu, Zidong, Chang, Kaixin, Chen, Xu

Abstract

With the widespread availability of parallel computing hardware, sampling-based motion planning methods such as Model Predictive Path Integral (MPPI) control have become increasingly powerful for complex nonlinear systems in non-smooth task spaces. However, the sampling and forward-simulation pipeline in MPPI suffers from averaging-induced failure in cluttered environments, where the importance-weighted update averages incompatible rollouts and leads to hesitation or even collision when an obstacle lies directly ahead. This paper proposes Clustering-Embedded MPPI (CE-MPPI), a framework that architecturally resolves the averaging-induced failures inherent in standard MPPI within non-convex environments. Rather than simply mitigating interference, CE-MPPI redefines the control law by integrating a high-fidelity pruning and clustering stage. By leveraging density-based spatial clustering of applications with noise (DBSCAN) alongside a novel geometric direction feature that is extracted from collision-derived reference points, the system isolates feasible trajectory modes from the noise of infeasible rollouts. This is paired with an intelligent selection logic that optimizes for minimum cost in static scenes while actively steering opposite to obstacle flux in dynamic environments. Experiments in 2-D JAX-accelerated simulations show that CE-MPPI alleviates obstacle-front hesitation and avoids persistent coupling with moving obstacles in dynamic scenes. In particular, real-world tests on a 6-DoF UR5e manipulator with CUDA-parallel rollouts in Isaac Gym achieve a 48\% reduction in time-to-goal and a 12\% shorter end-effector path.

Chinese Translation

随着并行计算硬件的广泛应用，基于采样的运动规划方法，如模型预测路径积分（Model Predictive Path Integral, MPPI）控制，已在复杂非线性系统的非光滑任务空间中变得越来越强大。然而，MPPI中的采样和前向仿真流程在杂乱环境中遭遇了平均引起的失败，其中重要性加权更新平均了不兼容的轨迹，从而导致在障碍物直接位于前方时的犹豫甚至碰撞。本文提出了嵌入聚类的MPPI（Clustering-Embedded MPPI, CE-MPPI），一个在非凸环境中从结构上解决标准MPPI固有的平均引起的失败的框架。CE-MPPI不仅仅是减轻干扰，而是通过整合高保真度的剪枝和聚类阶段重新定义控制律。通过利用带噪声的应用密度空间聚类（Density-Based Spatial Clustering of Applications with Noise, DBSCAN）以及从碰撞衍生的参考点提取的新几何方向特征，系统能够从不可行轨迹的噪声中隔离出可行的轨迹模式。这与一种智能选择逻辑相结合，优化静态场景中的最小成本，同时在动态环境中主动朝向与障碍物流动相反的方向进行引导。在二维JAX加速仿真实验中，CE-MPPI减轻了障碍物前的犹豫，并避免了在动态场景中与移动障碍物的持续耦合。特别是在Isaac Gym中使用CUDA并行轨迹的6自由度UR5e机械臂的实际测试中，实现了48%的目标时间减少和12%更短的末端执行器路径。

View on arXiv Download PDF AI Translation

cs.RO / 37 / 2607.06501

Hypothesis-driven Model Expansion under Uncertainty for Open-World Robot Planning

不确定性下的假设驱动模型扩展用于开放世界机器人规划

Xiao, Anxing, Zhang, Hanbo, Hu, Tianrun, Hsu, David

Abstract

We consider an open-world planning setting in which service robots must operate in unknown environments with incomplete knowledge of objects and actions. Traditional closed-world approaches with pre-programmed knowledge bases fail when robots encounter unexpected situations and tasks, posing a fundamental challenge for autonomous knowledge expansion in human environments. In this work, we propose an open-world planning framework that enables robots to automatically generate, verify, and update hypotheses about their abstract world models. Our key insight is to explicitly maintain uncertainty-aware knowledge expansion and integrate hypothesis verification into goal-reaching planning. The framework leverages foundation models to generate initial hypotheses over states and transitions, and applies automated planning to produce action sequences that jointly address hypothesis verification and task execution. Through iterative execution and refinement, the robot expands its knowledge by incorporating verification feedback from the foundation models when hypotheses prove incorrect. Extensive experiments in simulated and real-world environments demonstrate that our framework enables autonomous knowledge expansion and effective operation in open-world settings. These results indicate that integrating uncertainty-aware model expansion from robot foundation models with planning advances the practical deployment of household service robots.

Chinese Translation

我们考虑一种开放世界规划场景，在此场景中，服务机器人必须在与物体和操作缺乏完整知识的未知环境中运行。传统的封闭世界方法依赖于预先编程的知识库，当机器人遇到意外情况和任务时，这些方法无法有效应对，给自主知识扩展在人类环境中带来了根本性挑战。在本研究中，我们提出了一种开放世界规划框架，使机器人能够自动生成、验证和更新关于其抽象世界模型的假设。我们的关键见解是明确保持不确定性感知的知识扩展，并将假设验证整合到目标达成规划中。该框架利用基础模型生成状态和转换的初始假设，并应用自动规划生成同时解决假设验证和任务执行的动作序列。通过迭代执行和优化，机器人通过在假设被证明不正确时从基础模型中引入验证反馈，扩展其知识。在模拟和真实环境中的广泛实验表明，我们的框架使自主知识扩展和在开放世界环境中的有效运行成为可能。这些结果表明，将不确定性感知的模型扩展与规划相结合，促进了家用服务机器人的实际应用。

View on arXiv Download PDF AI Translation

cs.RO / 38 / 2607.06535

Neural-ESO: A Dual-Pathway Architecture for Provably Robust Learning-Based Control

神经扩展状态观察器（Neural-ESO）：一种可证明鲁棒的学习控制的双通路架构

Zhang, Fan, Suganda, Richie, Chen, Jinfeng, Liu, Wenhua, Fu, Hantao, Hu, Bin, Lin, Qin

Abstract

A learning-enabled disturbance-rejection framework based on a Neural Extended State Observer (Neural-ESO) is presented in this letter. Unlike existing learning-based control methods that largely rely on the learned model once deployed, Neural-ESO adopts a dual-pathway architecture: a predictive pathway uses a neural network to provide a feedforward disturbance estimate that accelerates convergence, while a corrective pathway employs a conventional ESO to compensate prediction errors and prevent over-reliance on the neural component. Using Lyapunov theory and a small-gain analysis, we show that enforcing a Lipschitz bound on the learning component guarantees uniform ultimate boundedness of the closed-loop error dynamics. The proposed framework is validated on a quadrotor landing task subject to strong ground-effect disturbances across normal and out-of-distribution scenarios, demonstrating accuracy-robustness trade-off and greater operational reliability during training, deployment, and transfer compared with state-of-the-art baselines.

Chinese Translation

本文提出了一种基于神经扩展状态观察器（Neural-ESO）学习驱动的抗干扰框架。与现有的大多数依赖于已学习模型的学习控制方法不同，Neural-ESO 采用了双通路架构：预测通路使用神经网络提供前馈干扰估计，以加速收敛，而修正通路则利用传统的扩展状态观察器（ESO）来补偿预测误差，防止对神经组件的过度依赖。通过利用李雅普诺夫理论和小增益分析，我们证明对学习组件施加的利普希茨界限确保了闭环误差动态的均匀最终有界性。所提框架在四旋翼着陆任务中得到了验证，面对强地面效应干扰，在正常和异常分布场景下表现出精度与鲁棒性的权衡，相比于最先进的基准方法，在训练、部署和迁移期间展示了更高的操作可靠性。

View on arXiv Download PDF AI Translation

cs.RO / 39 / 2607.06537

UniLM-Nav: A Unified Framework for Zero-Shot Last-Mile Navigation

UniLM-Nav：一种统一的零-shot最后一公里导航框架

Zhang, Zhuofan, Wang, Tianxu, Zhang, Guoxi, Lin, Yixiong, Wang, Xilin, Xu, Hongming, Li, Qing, Zhu, Song-Chun, Fan, Lifeng

Abstract

Mobile manipulation requires a robot to navigate to a target object or receptacle and then perform intended manipulation. However, reaching the vicinity of the target does not guarantee a manipulation-ready base pose, a problem known as last-mile navigation. Prior methods for last-mile navigation either rely on manual pose annotation or task-specific training, limiting their scalability to open-vocabulary settings with fine-grained spatial constraints. We propose UniLM-Nav, a unified framework for zero-shot open-vocabulary last-mile navigation. UniLM-Nav decomposes last-mile navigation into view selection, task-conditioned affordance grounding, and geometry-aware base-pose reasoning, all resolved with a shared multimodal large language model (MLLM) backend. Specifically, UniLM-Nav first selects a reference view that best captures the target object or receptacle from recently collected observations. It then grounds task-relevant affordance point in the selected view and lifts the result into the robot-centric coordinate frame. Finally, conditioned on the grounded affordance, task context, and robot geometry, it infers a manipulation-ready base pose for the robot. We evaluate UniLM-Nav on the OVMM benchmark, where it outperforms the previous state-of-the-art method, MoTo, by 3.13 percentage points. Analyses show that the components of our method are crucial to final performance, and that the choice of MLLM also has a substantial effect. We further deploy UniLM-Nav on a Unitree B2 quadruped robot with a 6-DoF Unitree Z1 manipulator, validating its applicability to real-world mobile manipulation tasks.

Chinese Translation

移动操作要求机器人导航到目标物体或容器，然后执行预期的操作。然而，仅仅到达目标附近并不能保证机器人处于适合操作的基本姿态，这个问题被称为最后一公里导航。之前的最后一公里导航方法要么依赖于手动姿态标注，要么依赖于特定任务的训练，这限制了它们在具有细粒度空间约束的开放词汇环境中的可扩展性。我们提出了UniLM-Nav，这是一种用于零-shot开放词汇最后一公里导航的统一框架。UniLM-Nav将最后一公里导航分解为视图选择、任务条件的可操作性定位和几何感知的基本姿态推理，所有这些都通过共享的多模态大语言模型（MLLM）后端来解决。具体而言，UniLM-Nav首先从最近收集的观察中选择一个最佳参考视图，以捕捉目标物体或容器。然后，它在所选视图中定位与任务相关的可操作性点，并将结果提升到机器人中心坐标系。最后，基于已定位的可操作性、任务上下文和机器人几何形状，推断出适合操作的机器人基本姿态。我们在OVMM基准上评估了UniLM-Nav，结果显示其性能比之前的最先进方法MoTo提高了3.13个百分点。分析表明，我们方法的各个组成部分对最终性能至关重要，而MLLM的选择也有显著影响。我们进一步将UniLM-Nav部署在配备6自由度Unitree Z1操纵器的Unitree B2四足机器人上，验证了其在现实世界移动操作任务中的适用性。

View on arXiv Download PDF AI Translation

cs.RO / 40 / 2607.06558

RynnWorld-Teleop: An Action-Conditioned World Model for Digital Teleoperation

RynnWorld-Teleop：一种基于动作条件的数字远程操作世界模型

Zhao, Haoyu, Zhao, Xingyue, Li, Hangyu, Gong, Biao, Li, Kehan, Huang, Siteng, Li, Xin, Zhao, Deli, Li, Zhongyu

Abstract

Scaling robot learning requires massive, diverse trajectory data, yet collection is currently bottlenecked by physical teleoperation, where every demonstration binds operator time to specific hardware and workspaces. We introduce digital teleoperation, a paradigm that decouples data collection from physical constraints by replacing the real robot with a generative world model. In this framework, an operator's hand-pose stream drives a robot-centric generative world model to synthesize high-fidelity egocentric videos from a single reference image. The recorded pose stream serves as an embodiment-agnostic action label transferable to any target robot via standard retargeting, yielding complete state-action trajectories for imitation learning independent of physical hardware. We instantiate this paradigm in RynnWorld-Teleop, a system that integrates depth-aware skeletal conditioning, progressive human-to-robot training on a video Diffusion Transformer, and streaming autoregressive distillation. This pipeline compresses the generative process into a single-pass inference, enabling 40+ FPS, real-time interactive generation on a single H100 GPU. Policies trained exclusively on RynnWorld-Teleop-generated data achieve effective zero-shot Sim2Real transfer across dexterous and diverse bimanual tasks. Moreover, augmenting real-world datasets with our digitally teleoperated data consistently improves success rates, demonstrating that RynnWorld-Teleop serves as a high-fidelity, scalable data engine for the next generation of robotic agents.

Chinese Translation

机器人学习的扩展需要大量多样化的轨迹数据，而目前数据收集受到物理远程操作的限制，在这种情况下，每次演示都将操作员的时间绑定到特定的硬件和工作空间。我们提出了一种数字远程操作的新范式，该范式通过用生成的世界模型替代真实机器人，从而将数据收集与物理约束解耦。在此框架中，操作员的手势动作流驱动一个以机器人为中心的生成世界模型，从单个参考图像合成高保真度的自我中心视频。记录的动作流作为一种与具体实体无关的动作标签，可以通过标准重定向转移到任何目标机器人，从而生成完整的状态-动作轨迹以用于模仿学习，而不依赖于具体的物理硬件。我们在RynnWorld-Teleop中实现了这一范式，该系统整合了深度感知的骨架条件、基于视频扩散变换器的渐进式人机训练以及流式自回归蒸馏。该管道将生成过程压缩为单次推理，能够在单个H100 GPU上实现40帧每秒以上的实时交互生成。仅基于RynnWorld-Teleop生成的数据训练的策略在灵巧和多样化的双手任务中实现了有效的零-shot Sim2Real 转移。此外，用我们数字远程操作的数据增强真实世界的数据集，始终提高了成功率，证明了RynnWorld-Teleop作为下一代机器人代理的高保真度、可扩展数据引擎的价值。

View on arXiv Download PDF AI Translation

cs.RO / 41 / 2607.06559

RynnWorld-4D: 4D Embodied World Models for Robotic Manipulation

RynnWorld-4D：用于机器人操作的4D具身世界模型

Zhao, Haoyu, Zhao, Xingyue, Huang, Siteng, Li, Xin, Zhao, Deli, Li, Zhongyu

Abstract

Robotic manipulation in the open world requires not only recognizing what a scene looks like, but also anticipating how its 3D structure moves under interaction. We argue that synchronized RGB, depth, and optical flow, namely RGB-DF, provide a physically grounded representation that captures the underlying 4D dynamics of a scene. Compared to 2D pixel videos, this multi-modal synergy aligns visual appearance with geometric structure and temporal motion, creating a representation space significantly closer to the low-level end-effector actions demanded by robotic systems, thereby narrowing the gap between world prediction and policy learning. Building on this insight, we introduce RynnWorld-4D, a generative model that co-produces future RGB frames, depth maps, and optical flow from a single RGB-D image and a language instruction within one unified diffusion process. This 4D world model features a tri-branch architecture that integrates cross-modal attention with frame-wise 3D RoPE, ensuring that appearance, geometry, and motion evolve consistently. To supply training data at scale, we curate Rynn4DDataset 1.0, a massive dataset of over 254.4 million frames across egocentric human and robotic manipulation videos with high-quality pseudo-labels for depth and optical flow. We further propose RynnWorld-4D-Policy, an inverse dynamics head that consumes the internal 4D representations of RynnWorld-4D in a single forward pass, bypassing expensive multi-step denoising, to output robot actions in a closed-loop manner. Experiments show that RynnWorld-4D produces temporally and spatially coherent 4D predictions, and that RynnWorld-4D-Policy achieves state-of-the-art performance on real-world dexterous bimanual manipulation tasks, particularly excelling in tasks demanding spatial precision and temporal coordination.

Chinese Translation

开放世界中的机器人操作不仅需要识别场景的外观，还需要预测其在交互下的3D结构如何移动。我们认为同步的RGB、深度和光流，即RGB-DF，提供了一种基于物理的表示，捕捉了场景的潜在4D动态。与2D像素视频相比，这种多模态协同将视觉外观与几何结构和时间运动对齐，创造了一个在很大程度上更接近机器人系统所需的低级末端执行器动作的表示空间，从而缩小了世界预测与策略学习之间的差距。在此基础上，我们引入了RynnWorld-4D，这是一种生成模型，它在一个统一的扩散过程中，从单个RGB-D图像和语言指令共同生成未来的RGB帧、深度图和光流。该4D世界模型具有三分支架构，将跨模态注意力与逐帧3D RoPE相结合，确保外观、几何和运动的一致演变。为了大规模提供训练数据，我们策划了Rynn4DDataset 1.0，这是一个包含超过2.544亿帧的庞大数据集，涵盖了以自我为中心的人类和机器人操作视频，并为深度和光流提供高质量的伪标签。我们进一步提出了RynnWorld-4D-Policy，这是一种逆动力学头，能够在单次前向传递中消耗RynnWorld-4D的内部4D表示，绕过昂贵的多步去噪，以闭环方式输出机器人动作。实验表明，RynnWorld-4D能够生成时间和空间上连贯的4D预测，而RynnWorld-4D-Policy在真实世界的灵巧双手操作任务中实现了最先进的性能，特别是在需要空间精度和时间协调的任务中表现突出。

View on arXiv Download PDF AI Translation

cs.RO / 42 / 2607.06563

Embodied Human-Robot Interaction via Acoustics: A MARL Approach with AcoustoBots for Spatial Data Physicalization

通过声学实现具身人机交互：一种基于 MARL 的 AcoustoBots 空间数据物理化方法

Liu, Shiqi, Kemsaram, Narsimlu, Mittal, Prateek, Wei, Pengyuan, Subramanian, Sriram

Abstract

Traditional data physicalization is often static and disconnected from real environments, limiting its ability to convey embodied spatial dynamics and engage users. To address this limitation, we present AcoustoBots, a mobile acoustophoretic data-physicalization platform in which TurtleBot3 robots carry upward-facing 8 x 8 ultrasonic phased arrays. Each array levitates a particle whose height (1-10 cm) encodes a local urban scalar value, such as population density, noise, or traffic. A MARL (Multi-Agent Reinforcement Learning) policy based on the Multi-Agent Deep Deterministic Policy Gradient (MADDPG) algorithm, with centralized training and decentralized execution, selects collision-aware navigation actions, while a high-rate Gerchberg-Saxton-Phased Array of Transducers (GS-PAT) acoustic controller maintains trap stability and updates array phases to achieve the commanded height during motion. This creates a closed perception-display-action loop. We evaluate single-robot city-to-city traversal and dual-robot cooperative coverage on a 4 m x 3 m scaled UK map using PhaseSpace-based localization for repeatable multi-robot trials. Results show stable in-motion levitation and consistent, location-dependent height rendering, with task success rates of 90% and 80% for the single and dual-robot regimes, respectively, over 10 trials per regime, and low collision counts. These findings support acoustophoretic levitation as a simple, glanceable, robot-mediated communication cue for embodied human-robot interaction in spatial analytics.

Chinese Translation

传统的数据物理化通常是静态的，与现实环境脱节，限制了其传达具身空间动态和吸引用户参与的能力。为了解决这一限制，我们提出了 AcoustoBots，一个移动声学物理化平台，使用 TurtleBot3 机器人携带向上朝向的 8 x 8 超声相控阵列。每个阵列悬浮着一个粒子，其高度（1-10 cm）编码了一个局部城市标量值，例如人口密度、噪声或交通。基于多智能体深度确定性策略梯度（MADDPG）算法的多智能体强化学习（MARL）策略，采用集中训练和分散执行，选择避免碰撞的导航动作，而高频 Gerchberg-Saxton 相控换能器（GS-PAT）声学控制器则维持捕捉稳定性，并在移动过程中更新阵列相位以实现指令高度。这创建了一个闭环感知-展示-行动循环。我们在一个 4 m x 3 m 的缩放英国地图上，使用基于 PhaseSpace 的定位进行可重复的多机器人试验，评估了单机器人城市间穿越和双机器人协同覆盖。结果显示，运动中的悬浮稳定性和一致的、位置相关的高度渲染，单机器人和双机器人模式下的任务成功率分别为 90% 和 80%，在每个模式的 10 次试验中碰撞次数较少。这些发现支持声学悬浮作为一种简单、可快速理解的机器人介导沟通线索，用于空间分析中的具身人机交互。

View on arXiv Download PDF AI Translation

cs.RO / 43 / 2607.06564

Lift3D-VLA: Lifting VLA Models to 3D Geometry and Dynamics-Aware Manipulation

Lift3D-VLA：将VLA模型提升至3D几何和动态感知操作

Liu, Jiaming, Wuwu, Qingpo, Han, Nuowei, Chen, Hao, Liu, Zhuoyang, Fei, Fan, Jia, Yueru, Gu, Chenyang, Guo, Yandong, Shi, Boxin, Zhang, Shanghang

Abstract

Recently, Vision-Language-Action (VLA) models have demonstrated strong generalization across diverse tasks. However, effective robotic manipulation in physical environments fundamentally requires geometric understanding and spatial reasoning. While some VLA approaches attempt to incorporate 3D information, they are constrained by limited data availability and geometric information loss in current 3D encoding pipelines, and fail to jointly capture 3D geometry and temporally structured actions in dynamic environments. To address these limitations, we introduce Lift3D-VLA, a unified VLA framework that equips models with explicit 3D point cloud reasoning and enables temporally coherent action generation. First, building upon our previous work Lift3D, an enhanced 2D model-lifting strategy is proposed to geometrically align 3D points with pretrained 2D positional embeddings. This design enables direct point-cloud encoding within the VLA vision encoder while minimizing spatial information loss. Based on explicit 3D inputs, we propose Geometry-Centric Masked Autoencoding (GC-MAE), a dual-objective self-supervised framework that reconstructs the current point cloud while predicting its future geometric evolution. This formulation allows the 2D vision encoder to internalize both 3D structure and physical dynamics. To fully exploit 3D representations, we further design layer-wise temporal action modeling, which leverages multiple layers of the LLM to collaboratively predict action chunks, enabling temporally consistent predictions. Across 22 simulated tasks and 8 real-world manipulation tasks, Lift3D-VLA achieves 10.8% and 11.1% higher mean success rates on MetaWorld and RLBench than the best-performing prior VLA methods, and outperforms the strongest real-world baseline by 4 percentage points, while exhibiting stronger generalization to out-of-distribution perturbations.

Chinese Translation

近年来，视觉-语言-动作（VLA）模型在多样化任务中展现了强大的泛化能力。然而，在物理环境中有效的机器人操作根本上需要几何理解和空间推理。尽管一些VLA方法尝试结合3D信息，但由于当前3D编码管道中数据可用性有限和几何信息损失，它们受到限制，未能在动态环境中共同捕捉3D几何和时间结构化的动作。为了解决这些局限性，我们提出了Lift3D-VLA，一个统一的VLA框架，赋予模型明确的3D点云推理能力，并能够生成时间一致的动作。首先，基于我们之前的工作Lift3D，提出了一种增强的2D模型提升策略，以几何方式将3D点与预训练的2D位置嵌入对齐。该设计使得在VLA视觉编码器中直接进行点云编码成为可能，同时最小化空间信息损失。基于明确的3D输入，我们提出了几何中心掩蔽自编码（GC-MAE），这是一种双目标自监督框架，能够重建当前点云并预测其未来的几何演变。这一公式使得2D视觉编码器能够内化3D结构和物理动态。为了充分利用3D表示，我们进一步设计了层级时间动作建模，利用LLM的多个层级协同预测动作片段，从而实现时间一致的预测。在22个模拟任务和8个真实世界操作任务中，Lift3D-VLA在MetaWorld和RLBench上比最佳表现的先前VLA方法分别提高了10.8%和11.1%的平均成功率，并且在真实世界基线中超越了最强的基线4个百分点，同时在分布外扰动上表现出更强的泛化能力。

View on arXiv Download PDF AI Translation

计算机视觉 (Computer Vision)

cs.CV / 1 / 2607.05465

CanvasAgent: Enabling Complex Image Creation and Editing via Visual Tool Orchestration

CanvasAgent：通过视觉工具编排实现复杂图像创作与编辑

Zhu, Hairui, Yang, Yiying, Weng, Tengjin, Lu, Ziyu, Yao, Xiao, Ye, Xiaoyang, Ma, Lin, Jiang, Wenhao

Abstract

Complex image creation and editing often require more than a single generation or editing model. A user request may involve synthesizing images, localizing objects, segmenting regions, editing selected content, compositing intermediate assets, reading text, and enhancing the final result. Such tasks shift multimodal agents from perception-augmented reasoning to manipulation-centered visual creation, where tools must actively transform visual states rather than merely inspect them. However, existing multimodal tool-use agents are mostly optimized for perception, search, or domain-specific editing, and lack large-scale supervision for executable image-creation trajectories. In this paper, we introduce CanvasCraft, a large-scale multimodal tool-use dataset for complex image creation and editing, and \textbf{CanvasAgent}, a tool-augmented multimodal agent that learns to orchestrate heterogeneous visual tools through multi-turn interaction. CanvasCraft contains 140K fully annotated executable trajectories and 10K RL task specifications. CanvasAgent is first trained with SFT to learn executable reasoning-action trajectories, and is then optimized with GRPO using a hybrid reward that combines outcome- and process-level signals. During rollout, CanvasAgent inspects intermediate results, tracks visual assets, and adapts tool decisions to the evolving visual state. Experiments evaluate both final image quality and trajectory behavior, demonstrating the effectiveness of CanvasAgent and the proposed dataset for complex multi-tool image creation workflows.

Chinese Translation

复杂的图像创作与编辑往往需要多个生成或编辑模型的协作。用户请求可能涉及图像合成、目标定位、区域分割、选择内容编辑、合成中间资产、文本读取以及增强最终结果等任务。这些任务将多模态智能体的功能从感知增强推理转向以操作为中心的视觉创作，在此过程中，工具必须积极地转变视觉状态，而不仅仅是检查它们。然而，现有的多模态工具使用智能体主要优化于感知、搜索或特定领域的编辑，缺乏可执行图像创作轨迹的大规模监督。在本文中，我们介绍了CanvasCraft，一个用于复杂图像创作与编辑的大规模多模态工具使用数据集，以及CanvasAgent，一个通过多轮交互学习编排异构视觉工具的工具增强型多模态智能体。CanvasCraft包含14万个完全标注的可执行轨迹和1万个强化学习任务规范。CanvasAgent首先通过监督微调（SFT）进行训练，以学习可执行的推理-动作轨迹，随后使用混合奖励优化（GRPO）结合结果和过程级信号进行优化。在回放过程中，CanvasAgent检查中间结果，跟踪视觉资产，并根据变化的视觉状态调整工具决策。实验评估了最终图像质量和轨迹行为，验证了CanvasAgent及其提出的数据集在复杂多工具图像创作工作流程中的有效性。

View on arXiv Download PDF AI Translation

cs.CV / 2 / 2607.05467

A Task-Driven Evaluation of UAV Detection and Tracking under Synthetic Fog

合成雾霭下无人机检测与跟踪的任务驱动评估

Pouladi, Amir, Ahsani, Vesal, Li, Haijun, Najjaran, Homayoun, Suleman, Afzal

Abstract

Fog severely degrades the visibility of small unmanned aerial vehicles (UAVs) in skydominant, long-range imagery, reducing the reliability of downstream detection and tracking. This paper presents a task-driven evaluation framework that links depth-aware synthetic fog generation, image restoration, object detection, and tracking within a unified pipeline. Given the practical difficulty of collecting and annotating foggy UAV scenes, synthetic fog is generated from real clear-weather outdoor images containing UAV targets using monocular depth estimation and the atmospheric scattering model. Representative restoration methods from classical, convolutional neural network (CNN)-based, and transformer-based families are first compared, after which the selected restoration model is integrated into the downstream perception pipeline. Detection is evaluated under both clean-only and fog-inclusive training regimes using multiple detector variants, while tracking-by-detection is assessed on clean, foggy, and restored video sequences. Beyond image-level restoration metrics, the study evaluates how fog and restoration affect detection robustness and tracking performance. The results show that fog substantially degrades both detection and tracking, primarily through increased missed detections. Fog-inclusive training provides the most consistent improvement in robustness, whereas test-time restoration is most beneficial when the detector has been trained only on clean imagery. These findings show that restoration quality does not necessarily translate into proportional gains in downstream perception and therefore should be evaluated jointly with detection and tracking performance.

Chinese Translation

雾霭严重降低了在以天空为主的远程图像中小型无人机（UAV）的可视性，从而降低了后续检测与跟踪的可靠性。本文提出了一种任务驱动的评估框架，将深度感知的合成雾霭生成、图像恢复、物体检测和跟踪链接在一个统一的流程中。考虑到收集和标注雾霭环境下无人机场景的实际困难，通过单目深度估计和大气散射模型，从包含无人机目标的真实晴天户外图像中生成合成雾霭。首先比较了经典方法、基于卷积神经网络（CNN）的方法和基于变换器（Transformer）的方法中的代表性恢复方法，然后将选定的恢复模型集成到下游感知流程中。在干净图像和包含雾霭的训练制度下，使用多个检测器变体评估检测性能，同时在干净、雾霭和恢复的视频序列中评估基于检测的跟踪。除了图像级恢复指标外，本研究还评估了雾霭和恢复如何影响检测的鲁棒性和跟踪性能。结果表明，雾霭显著降低了检测和跟踪的表现，主要以增加漏检为表现。包含雾霭的训练提供了最一致的鲁棒性提升，而测试时恢复在检测器仅在干净图像上训练时最为有利。这些发现表明，恢复质量并不一定会转化为后续感知任务的比例性提升，因此应与检测和跟踪性能共同评估。

View on arXiv Download PDF AI Translation

cs.CV / 3 / 2607.05473

Binocular Gaze Estimation with Single Camera and Single Light Source

单摄像头和单光源的双眼注视估计

Huang, Tongbing, Fu, Yang, Wang, Yunfei, Wang, Zhaocan

Abstract

According to commonly consented theories, the minimum hardware requirement for gaze tracker is one camera and two light sources to realize gaze estimation with free head movements. However, in some scenarios such as eye tracking on mobile devices, it is preferable to use less components, especially light sources. We propose a gaze estimation method with one camera and one light source. A "virtual light source" is introduced, which is geometrically placed symmetrically to the real light source with respect to the camera, and generates a "virtual glint" in the acquired image. We estimate the "virtual glint" by exploiting the relationship between the distance between two pupils and two glints in the captured image, and estimate the gaze with polynomial regression assuming two light sources are available. A new normalization factor for regression method is verified, which turns out to be practical for one-glint system. The performance is proved to be acceptable, while degradation is noticed compared to system with two actual light sources.

Chinese Translation

根据普遍认可的理论，注视追踪器的最低硬件要求是一个摄像头和两个光源，以实现自由头部运动下的注视估计。然而，在某些场景中，例如移动设备上的眼动追踪，使用更少的组件，特别是光源是更为理想的。我们提出了一种只使用一个摄像头和一个光源的注视估计方法。引入了一个“虚拟光源”，它在几何上相对于摄像头与真实光源对称放置，并在获取的图像中产生“虚拟闪光”。我们利用捕获图像中两个瞳孔与两个闪光之间的距离关系来估计“虚拟闪光”，并在假设有两个光源的情况下，通过多项式回归来估计注视方向。验证了一种新的归一化因子用于回归方法，结果证明在单闪光系统中是实用的。性能经过验证，显示出在与两个实际光源的系统相比时是可接受的，尽管性能有所下降。

View on arXiv Download PDF AI Translation

cs.CV / 4 / 2607.05493

Ground3D-LMM: Fine-Grained 3D Point Grounding and Spatial Reasoning with LMM

Ground3D-LMM：基于LMM的细粒度3D点定位与空间推理

Harsh, Amol, Han, Zongyan, Lahoud, Jean, Liu, Ye, Anwer, Rao Muhammad, Cholakkal, Hisham, Khan, Salman, Khan, Fahad

Abstract

Natural-language queries about 3D environments become actionable when responses are verifiable and metric. Verifiability requires explicit grounding to the referred 3D region, while metric answers report physical measurements in real-world units (e.g., size, thickness, clearance, and distance). Existing 3D large multimodal models (LMMs) approaches remain limited: conversational systems typically respond without explicit 3D grounding, while 3D grounding models are not designed for interactive, metric-aware dialogue. In this paper, we present Ground3D-LMM, a unified model that takes a point cloud and an optional RGB image as input and supports 3D spatial conversation with (i) point-grounded responses and (ii) metric numeric outputs at both object and part granularity, including multi-object queries. To evaluate this intersection of grounding and measurement, we define the 3D Grounded Measurement task, which requires predicting the referred 3D region and the corresponding metric quantities in real-world units. We introduce a large-scale dataset built on ScanNet and ScanNet++ datasets with dense object and part annotations and roughly 2.5M question-answer pairs spanning eight tasks, along with a manually verified test set. Extensive experiments on multiple datasets and tasks show that our proposed Ground3D-LMM model provides a strong baseline for grounded, metric-aware 3D conversational understanding. Our dataset and model are publicly available.

Chinese Translation

关于3D环境的自然语言查询在可验证且有度量的响应下变得可操作。可验证性要求明确定位到所指的3D区域，而有度量的答案则以现实单位（如大小、厚度、间隙和距离）报告物理测量。现有的3D大型多模态模型（LMM）方法仍存在限制：对话系统通常在没有明确3D定位的情况下作出响应，而3D定位模型则未设计为支持互动的、关注度量的对话。本文提出了Ground3D-LMM，这是一个统一模型，接受点云和可选的RGB图像作为输入，支持3D空间对话，具有（i）点定位的响应和（ii）在对象和部分粒度层面上的度量数值输出，包括多对象查询。为了评估这一定位与度量的交集，我们定义了3D定位测量任务，该任务要求预测所指的3D区域及对应的现实单位下的度量量。我们引入了一个基于ScanNet和ScanNet++数据集构建的大规模数据集，具有密集的对象和部分注释，涵盖大约250万个问答对，涉及八个任务，并附有一个手动验证的测试集。在多个数据集和任务上的广泛实验表明，我们提出的Ground3D-LMM模型为具备定位、关注度量的3D对话理解提供了强有力的基线。我们的数据集和模型均已公开提供。

View on arXiv Download PDF AI Translation

cs.CV / 5 / 2607.05511

Light-Omni: Reflex over Reasoning in Agentic Video Understanding with Long-Term Memory

Light-Omni：具备长期记忆的代理视频理解中的反应与推理

Nie, Chang, Wei, Jiaju, Feng, Junlan, Fu, Chaoyou, Shan, Caifeng

Abstract

Agentic video understanding equips models with long-term memory to autonomously process and respond to continuous, long-horizon multimodal streams. However, advanced video agents often rely on ``detective-style'' iterative reasoning for action control (e.g., $\mathtt{search}$) and evidence aggregation, incurring prohibitive costs and latency. We argue that such heavy reasoning primarily compensates for the lack of global context and semantic misalignment in retrieval. This paper introduces Light-Omni, a multimodal agent framework for reflexive and lightweight video understanding. It achieves this through dual contextual states that instantly build the required context in a single forward pass. First, we maintain a global state, a finite-sized multimodal script continuously consolidated from episodic memory, serving as the global context for Light-Omni. Through hierarchical merging, it preserves recent details while summarizing past events. Second, conditioned on this global context, we generate a parametric latent state that directly drives autonomous actions and produces retrieval embeddings, with minimal latency. Benefiting from this coupled design, Light-Omni achieves semantically aligned retrieval and reflexive responses while avoiding iterative reasoning. Extensive experiments validate the effectiveness of Light-Omni across multiple video benchmarks. Notably, it outperforms M3-Agent with an average 2.4% accuracy gain, a 12.1$\times$ speedup, and a 2.6$\times$ improvement in GPU memory efficiency. Furthermore, it serves as a memory system to enhance both the performance and efficiency of existing MLLMs. Project page: https://clare-nie.github.io/Light-Omni.

Chinese Translation

具备代理能力的视频理解使得模型能够使用长期记忆自主处理和响应连续的、长时间跨度的多模态流。然而，先进的视频代理往往依赖“侦探风格”的迭代推理进行动作控制（例如，$ exttt{search}$）和证据聚合，造成了高昂的成本和延迟。我们认为，这种重推理主要是为了弥补检索中缺乏全球上下文和语义错位的问题。本文介绍了Light-Omni，一个用于反应性和轻量级视频理解的多模态代理框架。它通过双重上下文状态实现了在单次前向传递中瞬时构建所需上下文。首先，我们维护一个全球状态，这是一个无限大小的多模态脚本，持续从情节记忆中汇总，作为Light-Omni的全球上下文。通过层次合并，它保留最近的细节，同时总结过去的事件。其次，基于这一全球上下文，我们生成一个参数化的潜在状态，直接驱动自主行动并生成检索嵌入，且延迟极小。得益于这种耦合设计，Light-Omni能够实现语义对齐的检索和反应性响应，同时避免迭代推理。大量实验验证了Light-Omni在多个视频基准上的有效性。值得注意的是，它相比于M3-Agent的平均准确率提升了2.4%，速度提升了12.1$ imes$，GPU内存效率提升了2.6$ imes$。此外，它还作为记忆系统，增强了现有多模态大语言模型（MLLMs）的性能和效率。项目主页：https://clare-nie.github.io/Light-Omni。

View on arXiv Download PDF AI Translation

cs.CV / 6 / 2607.05516

Statistical Adversaries: Natural Backdoor-like Features in Vision Datasets

统计对抗者：视觉数据集中的自然后门特征

Mandal, Paul K., Reddy, Pavan, Malatynski, Tristan

Abstract

Model-specific adversarial attacks have been extensively studied. We study a different failure mode: naturally occurring statistical signals in vision data that can behave like backdoor-like triggers without being maliciously inserted. We call these signals statistical adversaries. We analyse Imagenet to find patterns that are strongly linked to certain labels. We then use statistical controls to remove random correlations from our candidate signals. Finally, we demonstrate that these signals directly and predictably alter model predictions. These statistical adversaries are more targeted than generic corruptions and transfer across different model architectures. This suggests that some vulnerabilities are driven by dataset structure and distribution rather than a single model's idiosyncrasies. We conclude that ordinary datasets can contain exploitable adversarial surfaces even in the absence of poisoning, and suggest that dataset audits should treat spurious structure not only as a source of bias or interpretability failure, but also as a latent attack surface for vision models.

Chinese Translation

模型特定的对抗性攻击已经得到了广泛的研究。我们研究了一种不同的失败模式：在视觉数据中自然发生的统计信号，这些信号可以像后门触发器一样表现，而不需要恶意插入。我们将这些信号称为统计对抗者。我们分析了Imagenet数据集，以发现与特定标签之间存在强关联的模式。随后，我们使用统计控制方法消除候选信号中的随机相关性。最后，我们展示了这些信号直接且可预测地改变模型预测。这些统计对抗者比一般的干扰更具针对性，并且可以跨不同模型架构进行迁移。这表明某些漏洞是由数据集的结构和分布驱动，而不是单一模型的特例性。我们得出结论认为，普通数据集中即使在没有数据中毒的情况下也可以存在可利用的对抗性表面，并建议数据集审核应将虚假结构视为偏见或可解释性失败的源泉，也应视为视觉模型的潜在攻击面。

View on arXiv Download PDF AI Translation

cs.CV / 7 / 2607.05522

Rendering-Aware Bayesian 3D Gaussian Splatting with Native Uncertainty and Adaptive Complexity Control

渲染感知的贝叶斯3D高斯点云表示：具有原生不确定性和自适应复杂度控制

Jia, Gaoxiang, Appia, Vikram, Huang, Junzhou, Wang, Xinlei

Abstract

3D Gaussian splatting (3DGS) is a strong representation for real-time novel-view synthesis, but its standard training pipeline relies on point estimates and hand-tuned heuristics, providing no native uncertainty or principled complexity control. This is most limiting under sparse views or fixed acquisition budgets, where a model must identify weakly supported geometry and select informative views. We introduce a rendering-aware Bayesian 3DGS framework that tracks Gaussian geometry with a Normal-Inverse-Wishart posterior over means and covariances using renderer-derived surrogate summaries. An optional Dirichlet-process extension adds a probabilistic component-usage signal, and the training schedule makes the closed-form versus approximate inference boundary explicit. Re-rendering posterior geometry samples yields native predictive uncertainty for interval calibration and active view selection. In a fixed-budget 16-to-32 active-view task, native NIW acquisition improves PSNR by +0.453 dB and LPIPS by -0.0146 over a scoring-only 3-member standard-ensemble baseline, winning 29/39 scene-seed pairs and 10/13 scene means; it also improves over PPU-style (+0.355 dB) and NIW-proxy (+0.401 dB) acquisition. NIW native intervals reduce 95% coverage error by about 17x relative to a shared proxy (0.046 vs. 0.796) and are about 10x closer to nominal coverage than a 3-member deep ensemble (0.047 vs. 0.454) at roughly one-third the training cost. As a reconstruction compatibility check, paired NIW-vs-standard analysis over 39 scene-seed runs yields +0.030 dB PSNR with 1.6% additional training time. These results position Bayesian 3DGS as a practical probabilistic scene representation for decision-facing tasks such as active view selection.

Chinese Translation

3D高斯点云表示（3DGS）是实时新视角合成的一种强大表示，但其标准训练流程依赖于点估计和手动调优的启发式方法，并未提供原生的不确定性或原则性的复杂度控制。这在稀疏视角或固定获取预算的情况下限制最大，因为模型必须识别弱支持的几何形状并选择信息丰富的视角。我们提出了一种渲染感知的贝叶斯3DGS框架，通过渲染器导出的替代摘要，使用正态-逆维谢特（Normal-Inverse-Wishart）后验在均值和协方差上跟踪高斯几何形状。一个可选的狄利克雷过程（Dirichlet process）扩展添加了一个概率成分使用信号，训练计划明确了闭式与近似推理的边界。重新渲染后验几何样本产生原生预测不确定性，用于区间校准和主动视角选择。在一个固定预算的16到32个主动视角任务中，原生NIW获取改善了PSNR值，提高了+0.453 dB，LPIPS改善了-0.0146，相比于仅评分的3成员标准集成基线，在39组场景种子对中赢得了29组，并在13组场景均值中赢得了10组；此外，在PPU风格（+0.355 dB）和NIW代理（+0.401 dB）获取上也有改善。NIW原生区间相较于共享代理（0.046 vs. 0.796）将95%覆盖误差降低了约17倍，且相较于3成员深度集成（0.047 vs. 0.454），在名义覆盖上接近约10倍，训练成本约为其三分之一。作为重建兼容性的检查，对比NIW与标准分析在39个场景种子运行上产生了+0.030 dB的PSNR，附加训练时间为1.6%。这些结果将贝叶斯3DGS定位为一个实用的概率场景表示，适用于诸如主动视角选择等决策面对的任务。

View on arXiv Download PDF AI Translation

cs.CV / 8 / 2607.05533

Multi-Teacher Contrastive Distillation for Edge-Efficient Pathology Foundation Models

多教师对比蒸馏以实现边缘高效的病理基础模型

Lenz, Tim, Heide, Maurice, Gustav, Marco, Reitsam, Nic G., Kather, Jakob Nikolas

Abstract

Computational pathology foundation models (PFMs) have advanced whole-slide image analysis. However, their size and inference cost hinder local deployment in pathology departments. We propose MuCoDi, a pretraining framework that distills frozen tile embeddings from multiple PFMs into compact edge-oriented encoders. Instead of regressing individual teacher features, MuCoDi trains lightweight MobileOne and RepViT students with a contrastive distillation objective adapted from MoCo v3, where cached Virchow2, UNI2, and H-Optimus-1 embeddings replace momentum-encoder keys. We pretrain students on 14.3M TCGA tiles from only 11.8K WSIs and evaluate frozen encoders on 23 clinically curated downstream classification tasks. RepViT-based MuCoEdge students retain near-teacher performance while reducing model size by orders of magnitude: MuCoEdge-R2.3 and MuCoEdge-R1.5 reach 71.0% external AUROC, within 0.8 percentage points of the best teacher (Virchow2, 71.8%), while MuCoEdge-R2.3 obtains the best external F1 and the second-best AUPRC (51.8% and 53.3%). MuCoEdge-R1.0 reaches 70.9% AUROC with only 6.4M parameters and 1.12 GFLOPs. On a Raspberry Pi 5, sub-million-parameter MobileOne students achieve up to 605-fold single-tile speedup over Virchow2 while retaining 66.5% to 66.9% external AUROC, demonstrating that PFM-quality pathology representations can be moved toward practical edge deployment. Code is available at https://anonymous.4open.science/r/mucodi-6243.

Chinese Translation

计算病理基础模型（PFMs）已经推动了全切片图像分析的发展。然而，它们的体积和推理成本阻碍了在病理科的本地部署。我们提出了MuCoDi，这是一种预训练框架，它从多个PFMs中提取冻结的瓦片嵌入，并将其蒸馏为紧凑的边缘导向编码器。MuCoDi不同于回归个别教师特征，而是通过适应自MoCo v3的对比蒸馏目标，训练轻量级的MobileOne和RepViT学生，其中缓存的Virchow2、UNI2和H-Optimus-1嵌入替代了动量编码器的关键。我们在仅有11.8K个全切片图像（WSI）中训练了1430万个TCGA瓦片，并在23个临床策划的下游分类任务上评估冻结编码器。基于RepViT的MuCoEdge学生在减少模型大小数量级的同时保持近乎教师性能：MuCoEdge-R2.3和MuCoEdge-R1.5分别达到了71.0%的外部AUROC，离最佳教师（Virchow2，71.8%）相差0.8个百分点，而MuCoEdge-R2.3在外部F1和AUPRC上获得了最佳和第二最佳结果（51.8%和53.3%）。MuCoEdge-R1.0以仅6.4M参数和1.12 GFLOPs达到了70.9%的AUROC。在Raspberry Pi 5上，子百万参数的MobileOne学生实现了比Virchow2高达605倍的单瓦片加速，同时保持了66.5%到66.9%的外部AUROC，证明了PFM质量的病理表征可以向实用的边缘部署推进。代码可在 https://anonymous.4open.science/r/mucodi-6243 获取。

View on arXiv Download PDF AI Translation

cs.CV / 9 / 2607.05568

Harnessing Generative Image Models for Training-Free Primitive Shape Abstraction

利用生成图像模型进行无训练的原始形状抽象

Kobsik, Gregor, Elsner, Tim, Kobbelt, Leif

Abstract

Representing 3D shapes as compact sets of geometric primitives is fundamental to robotics, simulation, and scene understanding. Generative image models trained at scale have recently emerged as generalist visual learners that can identify and segment object parts directly in the image domain, across arbitrary categories and without task-specific training. Adapting such models to downstream tasks typically requires fine-tuning; we ask whether their pretrained capability can instead be harnessed directly, without any training, and answer affirmatively with a training-free harness. Our pipeline renders multi-view images of a 3D object, uses a vision-language model to analyze its semantic parts, prompts a generative image model to paint a color-coded part segmentation mask, reprojects it onto the geometry, and fits a superquadric primitive to each part via parameter optimization. The approach contains no learned parameters: it is category-agnostic and orientation-invariant, properties that previous learning-based models struggled with. Its accuracy ceiling rises with future generative-model improvements, which we confirm with a ground-truth segmentation study showing that part segmentation, not primitive fitting, is the current accuracy bottleneck. On HumanPrim and Toys4K, our method achieves the lowest Chamfer distance among all evaluated methods, using 5--9 primitives per object on average.

Chinese Translation

将三维形状表示为紧凑的几何原语集合是机器人技术、仿真和场景理解的基础。近年来，大规模训练的生成图像模型作为通用视觉学习者，能够直接在图像域中识别和分割物体部分，适用于任意类别且无需特定任务的训练。将此类模型适应于下游任务通常需要微调；我们提出一个问题，即是否可以直接利用其预训练能力，而无需任何训练，并通过无训练的方式给出肯定的答案。我们的流程渲染三维物体的多视图图像，使用视觉-语言模型分析其语义部分，提示生成图像模型绘制颜色编码的部分分割掩码，将其重新投影到几何体上，并通过参数优化为每个部分拟合超四面体原语。该方法不包含学习参数：它是类别无关的且方向不变的，这些特性是之前基于学习的模型所面临的挑战。其准确性上限随着未来生成模型的改进而提高，我们通过一项真实分割研究确认了这一点，结果显示部分分割，而非原语拟合，是当前的准确性瓶颈。在 HumanPrim 和 Toys4K 数据集上，我们的方法在所有评估方法中实现了最低的 Chamfer 距离，平均每个物体使用 5 到 9 个原语。

View on arXiv Download PDF AI Translation

cs.CV / 10 / 2607.05585

Hierarchical Classification via Cascading Feature Elimination: Application to Human Phenotype Ontology-Aligned Facial Phenotyping (FaceMesh2HPO)

通过级联特征消除的层级分类：在人类表型本体（HPO）对齐的面部表型中的应用（FaceMesh2HPO）

Hellmann, Fabio, Hustinx, Alexander, Solomon, Benjamin D., Consortium, GestaltMatcher Database, Hsieh, Tzung-Chien, Krawitz, Peter, André, Elisabeth

Abstract

FaceMesh2HPO is a framework for classifying facial phenotypic descriptors aligned with the Human Phenotype Ontology (HPO) to support clinical diagnosis. Using annotations from 124 clinicians across 10 disorders (107 HPO terms) combined with non-syndromic controls, we generated 3D facial meshes (478 landmarks) from 2D images and trained a hierarchical PointNet-based pipeline with cascading classification and feature elimination. The best models, incorporating 3D meshes, facial outline, and demographic metadata, achieved AUROCs between ~0.55 and ~0.89, with higher performance at parent nodes than leaf terms. External validation showed variable generalizability across disorders. Results demonstrate that hierarchical modeling of 3D facial geometry enables interpretable, ontology-linked phenotype classification, though performance on rare leaf terms remains limited. Improved data diversity and feature selection strategies are needed to enhance robustness and clinical utility.

Chinese Translation

FaceMesh2HPO 是一个用于分类与人类表型本体（HPO）对齐的面部表型描述符的框架，旨在支持临床诊断。我们结合来自 124 名临床医生对 10 种疾病（107 个 HPO 术语）的注释以及非综合症对照，利用 2D 图像生成了 3D 面部网格（478 个标志点），并训练了一个基于 PointNet 的层级管道，该管道具有级联分类和特征消除的功能。最佳模型结合了 3D 网格、面部轮廓和人口统计元数据，获得了约 0.55 到约 0.89 的 AUROC 值，在父节点的表现优于叶子术语。外部验证显示不同疾病之间的可推广性存在差异。结果表明，3D 面部几何的层级建模实现了可解释的、与本体链接的表型分类，但在稀有叶子术语上的表现仍然有限。需要改善数据多样性和特征选择策略，以增强模型的稳健性和临床实用性。

View on arXiv Download PDF AI Translation

cs.CV / 11 / 2607.05605

Patch Knowledge Transfer for Efficient AI-Generated Image Quality Assessment

高效AI生成图像质量评估的补丁知识转移

Yuan, Jiquan

Abstract

With the rapid advancement of image generation technologies, perceptual quality assessment of AI-generated images has emerged as a crucial research direction in computer vision. The core challenge of this task lies in achieving efficient quality assessment for massive generated images. Current mainstream approaches exhibit two key limitations: 1) Methods employing complex feature extraction strategies, while improving performance, incur prohibitive computational costs that hinder real-time inference; 2) Simple image scaling-based solutions, despite their computational efficiency, demonstrate significantly inferior assessment accuracy. To address this critical issue, we propose Patch Knowledge Transfer (PKT), a knowledge distillation-based optimization framework that achieves synergistic optimization of visual representation capability and inference efficiency through an innovative multi-level knowledge transfer mechanism. Specifically, we design a dual-model architecture: a teacher model with local-global hybrid processing provides high-quality supervision signals, while a student model relying solely on global processing efficiently inherits the teacher's representation capacity through multi-level supervision. Extensive experiments conducted on 4 AIGIQA databases demonstrate that the PKT framework enables the student model to maintain performance comparable to the teacher while reducing computational costs by 67.7\%. Furthermore, compared to existing methods, our approach achieves a superior balance between model efficiency and assessment accuracy.

Chinese Translation

随着图像生成技术的快速发展，AI生成图像的感知质量评估已成为计算机视觉领域的一个重要研究方向。这项任务的核心挑战在于实现对大量生成图像的高效质量评估。目前主流的方法存在两个关键限制：1）采用复杂特征提取策略的方法在提高性能的同时，导致了高昂的计算成本，从而妨碍实时推理；2）基于简单的图像缩放的解决方案尽管计算效率高，但评估精度明显较低。为了解决这一关键问题，我们提出了补丁知识转移（Patch Knowledge Transfer，PKT），这是一种基于知识蒸馏的优化框架，通过创新的多层知识转移机制实现视觉表征能力和推理效率的协同优化。具体来说，我们设计了一个双模型架构：一个具有局部-全局混合处理的教师模型提供高质量的监督信号，而一个仅依赖全局处理的学生模型通过多层监督有效继承教师的表征能力。在4个AIGIQA数据库上进行的大量实验表明，PKT框架使得学生模型在保持与教师模型相当的性能的同时，计算成本降低了67.7%。此外，与现有方法相比，我们的方法在模型效率与评估精度之间达到了更优的平衡。

View on arXiv Download PDF AI Translation

cs.CV / 12 / 2607.05625

Cross-Contextual Vision-Language Adaptation with LoRA for Personalized Severe Adverse Event Detection in Clinical Wound Monitoring

基于LoRA的跨上下文视觉语言适应在临床创伤监测中个性化严重不良事件检测的应用

Naiknaware, Aditi, Sun, Jian, Khandan, Aminreza, Huang, Shengyang, Dow, Sean, Najafi, Bijan, Sekeh, Salimeh

Abstract

Wound monitoring is a critical yet underserved clinical challenge, where timely identification of severe adverse events (SAEs) such as infection, tissue deterioration, and delayed healing can significantly impact patient outcomes. While vision-language models (VLMs) show strong multimodal reasoning, they often lack domain-specific grounding to integrate wound imagery with heterogeneous clinical information, and provide limited mechanisms for detecting cases that diverge from the training distribution. We present a multimodal framework for automated wound monitoring and SAE detection. Our approach leverages paired clinical notes and wound descriptions capturing visual characteristics such as appearance, surrounding skin condition, color changes, and signs of inflammation or healing progression, encoded through a dual-stream Low-Rank Adaptation (LoRA) framework built on a frozen BiomedCLIP backbone. We introduce a cross-contextual LoRA fusion mechanism enabling information exchange between clinical semantics and visual wound descriptors, producing context-aware multimodal representations without full model fine-tuning. To identify personalized SAEs, we propose a wound-specific out-of-distribution (OOD) detection framework combining semantic matching, visual typicality, caption-text alignment, and caption-visual alignment into a unified SAE (OOD) score. To capture healing dynamics, we incorporate covariate consistency and temporal drift penalties that leverage changes in wound characteristics across visits. Experiments on a longitudinal wound dataset collected through clinical visits show promising performance on both wound healing assessment and SAE detection, highlighting the potential of semantically enriched, temporally aware vision-language systems for clinical wound monitoring and early risk identification.

Chinese Translation

创伤监测是一个关键但服务不足的临床挑战，及时识别诸如感染、组织恶化和愈合延迟等严重不良事件（SAEs）对患者结果有显著影响。尽管视觉语言模型（VLMs）展现出强大的多模态推理能力，但它们往往缺乏领域特定的基础，无法将创伤图像与异构临床信息有效集成，且提供的检测机制对训练分布的偏差案例支持有限。我们提出了一种用于自动化创伤监测和SAE检测的多模态框架。我们的方法利用配对的临床记录和创伤描述，捕捉视觉特征，例如外观、周围皮肤状态、颜色变化以及炎症或愈合进展的迹象，这些信息通过基于冻结的BiomedCLIP骨干网络构建的双流低秩适应（LoRA）框架进行编码。我们引入了一种跨上下文的LoRA融合机制，使临床语义与视觉创伤描述符之间的信息交换得以实现，生成上下文感知的多模态表示，而无需对整个模型进行微调。为了识别个性化的SAE，我们提出了一种创伤特定的分布外（OOD）检测框架，将语义匹配、视觉典型性、标题与文本对齐以及标题与视觉对齐结合成一个统一的SAE（OOD）评分。为了捕捉愈合动态，我们纳入了协变量一致性和时间漂移惩罚，这利用了多次访视中创伤特征的变化。对通过临床访视收集的纵向创伤数据集的实验表明，我们的方法在创伤愈合评估和SAE检测方面表现出色，突显了语义丰富、时间敏感的视觉语言系统在临床创伤监测和早期风险识别中的潜力。

View on arXiv Download PDF AI Translation

cs.CV / 13 / 2607.05628

Taxlifier: Leveraging Disease Taxonomy for Enhanced Multi-Label Classification in Chest Radiography

Taxlifier：利用疾病分类法增强胸部放射影像中的多标签分类

Majdi, Mohammad S., Rodriguez, Jeffrey J.

Abstract

Accurate and efficient classification of thoracic diseases in chest X-ray (CXR) images is crucial for timely diagnosis and treatment. However, the presence of multiple pathologies with overlapping visual characteristics poses significant challenges for automated classification systems. In this study, we propose two novel hierarchical multi-label classification techniques, namely the loss-based and logit-based methods, to address these challenges by leveraging the hierarchical relationships among different thoracic pathologies. The loss-based technique integrates hierarchical information directly into the optimization process, while the logit-based method adjusts the predicted probabilities of each class based on its parent class in the disease taxonomy. We evaluate the performance of both techniques using three large-scale CXR datasets: CheXpert (224,316 CXRs), PADCHEST (160,000 CXRs), and NIH (112,120 CXRs). The experimental results demonstrate significant improvements in accuracy, AUC, and F1 scores compared to the baseline method across various pathologies. The logit-based and loss-based methods improve accuracy by 12\% and 11\%, AUC by 13\% and 10\%, and F1 scores by 24\% and 12\%, respectively compared to the baseline. These results represent a substantial improvement over the baseline method. Furthermore, we conduct a comprehensive statistical analysis to validate the robustness and reliability of the proposed techniques. The integration of domain-specific hierarchical knowledge not only enhances the classification performance but also provides a more interpretable output for clinical decision support. Our findings highlight the potential of hierarchical multi-label classification in advancing computer-aided diagnosis systems for chest radiography.

Chinese Translation

在胸部X光（CXR）图像中准确且高效地分类胸部疾病对于及时诊断和治疗至关重要。然而，多种病理的重叠视觉特征给自动分类系统带来了显著挑战。在本研究中，我们提出了两种新颖的层次多标签分类技术，即基于损失的方法和基于对数几率的方法，以利用不同胸部病理之间的层次关系来应对这些挑战。基于损失的技术将层次信息直接整合到优化过程中，而基于对数几率的方法则根据疾病分类法中的父类调整每个类别的预测概率。我们使用三个大规模CXR数据集（CheXpert（224,316个CXR）、PADCHEST（160,000个CXR）和NIH（112,120个CXR））评估这两种技术的性能。实验结果表明，与基线方法相比，在各种病理下，准确性、AUC和F1分数都有显著改善。与基线相比，基于对数几率和基于损失的方法分别提高了12%和11%的准确性，13%和10%的AUC，以及24%和12%的F1分数。这些结果代表了对基线方法的显著改进。此外，我们进行了全面的统计分析，以验证所提技术的稳健性和可靠性。领域特定的层次知识的整合不仅增强了分类性能，还为临床决策支持提供了更具可解释性的输出。我们的研究结果突显了层次多标签分类在推进胸部放射影像计算机辅助诊断系统中的潜力。

View on arXiv Download PDF AI Translation

cs.CV / 14 / 2607.05637

Recovering Cloud Microstructures with Cascaded Diffusion Inversion

使用级联扩散反演恢复云微结构

Gani, Hanan, Pulik, Guy, Rosenfeld, Daniel, Watson-Parris, Duncan, Khan, Salman

Abstract

High-resolution satellite imagery is critical for observing fine-scale cloud structures that inform weather modification strategies like cloud seeding for rain-enhancement. However, the spatial resolution of current geostationary and polar-orbiting satellites is often insufficient for capturing small cloud features. Current super-resolution methodologies are suited for natural images and, therefore, struggle to generalize to satellite-captured spectral images of cloud cover. To address this, we propose a two-stage diffusion-based super-resolution framework to enhance the resolution of multi-spectral cloud microstructures by a factor of $4\times$. Specifically, we use inverse diffusion to recover the high resolution properties from low resolution. Stage 1 utilizes real-world paired data to learn robust degradation handling and inter-sensor alignment, while Stage 2 employs a self-supervised internal downgrading of high resolution data to refine structural learning and texture synthesis. Our approach outperforms the state-of-the-art transformer and diffusion-based baselines in both reconstruction accuracy and visual quality. We demonstrate that the two-stage method better captures fine cloud microstructures (e.g. convective turrets and cloud gaps) that are crucial for effective cloud seeding decisions. Ablation studies confirm the complementary benefits of the two stages: Stage 1 excels in coarse structural fidelity, while Stage 2 contributes enhanced detail and realism. These results highlight a practical path toward improving cloud microphysics analysis and as a step towards utilizing AI for climate and sustainability. Our code and models are publicly available at: https://github.com/hananshafi/superresolution-cloud-microphysics.

Chinese Translation

高分辨率卫星图像对于观察细尺度云结构至关重要，这些结构为像云播种这样的天气改造策略提供了信息，以增强降雨。然而，当前静止轨道和极轨卫星的空间分辨率往往不足以捕捉细小的云特征。目前的超分辨率方法适用于自然图像，因此在一般化到卫星捕获的云覆盖光谱图像时存在困难。为了解决这一问题，我们提出了一种基于两阶段扩散的超分辨率框架，以将多光谱云微结构的分辨率提高 $4 imes$。具体而言，我们使用逆扩散从低分辨率恢复高分辨率特征。第一阶段利用真实世界的配对数据学习稳健的降解处理和传感器间对齐，而第二阶段则采用自我监督的内部降级高分辨率数据，以完善结构学习和纹理合成。我们的方法在重建精度和视觉质量上均优于最先进的变换器和基于扩散的基线。我们证明，两阶段方法更好地捕捉了细微的云微结构（例如对流塔和云隙），这些对于有效的云播种决策至关重要。消融研究确认了两个阶段的互补效益：第一阶段在粗略结构保真度方面表现出色，而第二阶段则提供了增强的细节和真实感。这些结果强调了一条改善云物理分析的实际路径，并朝着利用人工智能应对气候与可持续性问题迈进。我们的代码和模型可在以下网址公开获取：https://github.com/hananshafi/superresolution-cloud-microphysics。

View on arXiv Download PDF AI Translation

cs.CV / 15 / 2607.05641

VEIL: How Visual Encoding Hijacking Induces Bias In Vision Models

VEIL：视觉编码劫持如何在视觉模型中引发偏见

Sooraj, Suranjana, Chen, Xuyang, Venkatesan, Madhumitha, Liu, Dongyu

Abstract

Rendering time series as chart images for CNN-based classification has become increasingly common in time-series classification (TSC). However, it remains unclear whether models learn underlying temporal patterns or rely on encoding-specific visual cues introduced by chart design. We present VEIL: a systematic study examining how chart encodings influence learned representations through complementary analyses of similarity, transferability, and attribution. Attention-guided training appears to mitigate this effect when encoding sensitivity is consistently identified across diagnostics, but provides limited or negative benefit when such signals are absent. These findings position VEIL within the broader question of how machines perceive visualizations -- extending graphical perception from human readers to vision models -- and show that visualization design choices shape learned representations in ways that warrant treating chart-based TSC as a representation and measurement problem rather than a simple modeling decision.

Chinese Translation

将时间序列渲染为图表图像以进行基于卷积神经网络（CNN）的分类在时间序列分类（TSC）中变得越来越普遍。然而，目前尚不清楚模型是学习底层的时间模式，还是依赖于图表设计所引入的编码特定视觉线索。我们提出了VEIL：一项系统研究，考察图表编码如何通过相似性、可迁移性和归因的互补分析影响学习到的表征。当在各类诊断中一致识别到编码敏感性时，基于注意力的训练似乎有助于减轻这一影响，但在缺乏这种信号时却提供有限或负面的效益。这些发现将VEIL置于更广泛的问题框架中，即机器如何感知可视化——将图形感知从人类读者扩展到视觉模型——并表明可视化设计选择以某种方式塑造学习表征，这使得将基于图表的TSC视为一种表征和测量问题而非简单建模决策显得尤为重要。

View on arXiv Download PDF AI Translation

cs.CV / 16 / 2607.05649

REVIVE: A Multi-Modal Framework for Vandalism Detection and Recovery in Autonomous Vehicles

REVIVE：一种用于自主车辆中破坏检测与恢复的多模态框架

Choudhry, Abdullah Tariq, Das, Tapadhir

Abstract

Autonomous vehicles (AVs) face increasing threats from vandalism-induced occlusion attacks (VOAs) that compromise camera-based perception. While detection frameworks can identify vandalized images, restoring camera-stream utility after physical occlusion remains underexplored. This paper presents present the Recovery and Enhancement of Vandalized Images for Vision Excellence (REVIVE) framework, a vandalism recovery pipeline integrating: (1) binary VOA detection, (2) multi-class VOA pattern identification, (3) EfficientNet-based U-Net segmentation, and (4) type-aware recovery using Bootstrapping Language-Image Pre-training (BLIP)-guided Stable Diffusion inpainting, direct pixel replacement, or adaptive median filtering. Stable Diffusion shows variable reconstruction performance (per-pattern SSIM 0.667-0.867, PSNR 15.4-26.7dB) across VOA patterns, while aligned direct pixel replacement achieves near-identical reconstruction under the aligned-reference condition. On 500 tracked clean/vandalized image pairs, unrecovered VOAs reduce YOLOv8l object-detection recall to 0.588, while direct pixel replacement restores recall to 0.967 and F1-score to 0.970 under that aligned-reference condition. LaMa, Telea, and Navier-Stokes baselines improve image similarity but provide more limited downstream detection recovery, and Stable Diffusion is treated as an asynchronous recovery branch subject to a quality gate rather than a blocking real-time perception step. We evaluate a reference-available quality gate that filters recovered candidates before downstream use: without it, type-aware routing degrades per-image recall to 0.304, whereas with it, recall returns to 0.608, at or above the unrecovered baseline, ensuring the forwarded stream is never worse than the unrecovered frame. REVIVE therefore, provides a structured recovery framework from VOAs in AVs.

Chinese Translation

自主车辆（AVs）面临来自破坏引发的遮挡攻击（VOAs）日益增加的威胁，这些攻击会损害基于摄像头的感知能力。虽然检测框架可以识别被破坏的图像，但在物理遮挡后恢复摄像头流的实用性仍然未得到充分探索。本文提出了破坏图像恢复与增强框架（REVIVE），这是一个集成了以下功能的破坏恢复管道：（1）二元VOA检测，（2）多类VOA模式识别，（3）基于EfficientNet的U-Net分割，以及（4）使用引导式语言-图像预训练（BLIP）的稳定扩散修复、直接像素替换或自适应中值滤波的类型感知恢复。稳定扩散在不同VOA模式下表现出可变的重建性能（每种模式的SSIM为0.667-0.867，PSNR为15.4-26.7dB），而对齐的直接像素替换在对齐参考条件下实现了几乎相同的重建。在500对跟踪的干净/被破坏图像中，未恢复的VOAs将YOLOv8l目标检测的召回率降低至0.588，而直接像素替换在该对齐参考条件下将召回率恢复至0.967，F1分数恢复至0.970。LaMa、Telea和Navier-Stokes基线提高了图像相似性，但在下游检测恢复方面提供的帮助有限，稳定扩散被视为一个异步恢复分支，受质量门控的限制，而不是阻塞实时感知的步骤。我们评估了一种可用参考的质量门控，在下游使用前过滤恢复的候选者：没有它，类型感知路由将每幅图像的召回率降至0.304，而有了它，召回率恢复至0.608，达到或超过未恢复基线，确保转发流永远不会比未恢复帧更差。因此，REVIVE为自主车辆中的VOAs提供了一个结构化的恢复框架。

View on arXiv Download PDF AI Translation

cs.CV / 17 / 2607.05667

Clustered Codebook Quantization for 2D Gaussian-based Image Compression

基于聚类的代码簿量化用于二维高斯图像压缩

Cheng, Runze, Zhan, Yicheng, Spjut, Josef, Akşit, Kaan

Abstract

Gaussian-based image representations effectively model image content using compact parametric primitives while preserving high visual fidelity, yet storing a large number of floating-point parameters per primitive degrades rate-distortion efficiency at higher fidelity targets. To improve the rate-distortion performance in Gaussian representation, we present our Cluster-Guided Vector Quantization (CGVQ), a Gaussian primitive based image compression method. Our key idea is to partition Gaussian parameters further into homogeneous groups prior to quantization, enabling higher compression efficiency and accurate parameter reconstruction. In practice, our extensive experiments show that CGVQ decreases the bpp by 20% with respect to our baseline, while maintaining on-par visual quality

Chinese Translation

基于高斯的图像表示通过使用紧凑的参数原语有效地建模图像内容，同时保持高视觉保真度。然而，在每个原语中存储大量浮点参数会在更高保真度目标下降低速率失真效率。为了改善高斯表示中的速率失真性能，我们提出了聚类引导向量量化（Cluster-Guided Vector Quantization, CGVQ），这是一种基于高斯原语的图像压缩方法。我们的关键思想是在量化之前进一步将高斯参数划分为同质组，从而实现更高的压缩效率和准确的参数重构。在实践中，我们的广泛实验表明，CGVQ相对于我们的基线减少了20%的比特每像素（bpp），同时保持了相当的视觉质量。

View on arXiv Download PDF AI Translation

cs.CV / 18 / 2607.05702

Robust Face Super-Resolution and Recognition Through Multi-Feature Aggregation in Diffusion Models

通过扩散模型中的多特征聚合实现鲁棒的人脸超分辨率与识别

Santos, Marcelo dos, Laroca, Rayson, Neves, João Carlos Raposo, Menotti, David

Abstract

Images acquired in surveillance environments often suffer from conditions such as low resolution, variations in pose, irregular illumination, and occlusions. Due to the low quality of these images, face recognition algorithms often struggle. This major limitation can be addressed by employing super-resolution techniques that enhance the details of the image. However, due to the high degree of difficulty of the problem, most super-resolution algorithms tend to cause distortions in the image and in the individual's identity. Thus, additional information must be incorporated into the processing to improve recognition robustness. In this regard, surveillance cameras can capture multiple images, even at low quality, and the data extracted from these images, such as consecutive video frames, can significantly enhance both super-resolution and facial recognition. In this work, we introduce FASR++, a diffusion-model-based super-resolution algorithm. It leverages a reference low-resolution image and features extracted from multiple auxiliary low-quality images to generate a super-resolved output, minimizing distortions in the individual's identity. Our approach recovers facial features without explicitly providing soft attributes or computing a function gradient to guide the reconstruction process. FASR++ generates high-quality images that can considerably improve performance in face recognition tasks when used as a pre-processing step. We validate our approach on two standard face recognition datasets and attain state-of-the-art results for verification, face recognition, and image quality metrics such as PSNR, SSIM, and LPIPS.

Chinese Translation

在监控环境中获取的图像通常遭受低分辨率、姿态变化、不规则照明和遮挡等条件的影响。由于这些图像的低质量，人脸识别算法常常面临困难。这一主要限制可以通过采用超分辨率技术来解决，从而增强图像细节。然而，由于问题的高难度，大多数超分辨率算法往往会导致图像和个体身份的失真。因此，必须将额外的信息纳入处理，以提高识别的鲁棒性。在这方面，监控摄像头可以捕获多张图像，即使在低质量下，这些图像提取的数据，例如连续的视频帧，可以显著增强超分辨率和人脸识别的效果。在本研究中，我们提出了FASR++，这是一种基于扩散模型的超分辨率算法。它利用参考低分辨率图像和从多张辅助低质量图像中提取的特征，生成超分辨率输出，以最小化个体身份的失真。我们的方法在恢复面部特征时，不需要显式提供软属性或计算函数梯度来引导重建过程。FASR++生成的高质量图像在作为预处理步骤用于人脸识别任务时，可以显著提高性能。我们在两个标准人脸识别数据集上验证了我们的方法，并在验证、人脸识别和图像质量指标（如PSNR、SSIM和LPIPS）方面取得了最新的成果。

View on arXiv Download PDF AI Translation

cs.CV / 19 / 2607.05716

Scene Graph Thinking: Reinforcing Structured Visual Reasoning for Multimodal Large Language Models

场景图思维：增强多模态大型语言模型的结构化视觉推理

Yang, Zhiwei, Wu, Yuanchen, Zhang, Nan, Meng, Yucong, Yan, Ke, Ding, Shouhong

Abstract

Multimodal Large Language Models (MLLMs) have demonstrated strong perception and reasoning capabilities. However, most existing models focus on isolated objects and neglect structured relationships for efficient target navigation, limiting their performance on visually intensive tasks. To address this challenge, we introduce Scene Graph Thinking (SaGe), a novel paradigm that enables fine-grained and structured visual reasoning through explicit scene-graph representations. Specifically, we first introduce an automated data engine that converts flat image-text corpora into structured scene graphs, where hierarchical entities constitute the nodes and diverse visual relations define the edges. Building upon this, we construct 120K high-quality training data by sampling reasoning traces from scene graphs. Then, two-stage graph-aligned post-training paradigms are introduced, where supervised fine-tuning internalizes MLLMs with structured reasoning, and subsequent reinforcement fine-tuning proposes node-as-proxy graph rewards to consolidate efficient graph exploration. With curated data and graph-aligned training, our approach achieves significant improvements across eight multimodal benchmarks, demonstrating strong effectiveness on fine-grained perception and reasoning tasks. Code is available at https://github.com/zwyang6/SaGe.

Chinese Translation

多模态大型语言模型（MLLMs）展现了强大的感知和推理能力。然而，大多数现有模型专注于孤立对象，忽视了有效目标导航所需的结构化关系，从而限制了其在视觉密集任务中的表现。为了解决这一挑战，我们提出了场景图思维（SaGe），这是一种新颖的范式，通过显式的场景图表示实现细粒度和结构化的视觉推理。具体而言，我们首先引入一个自动数据引擎，将平面图像-文本语料库转换为结构化的场景图，其中层次化的实体构成节点，多样的视觉关系定义边。基于此，我们通过从场景图中抽样推理踪迹构建了12万个高质量训练数据。然后，提出了两阶段的图对齐后训练范式，其中监督微调使MLLMs内化结构化推理，随后的强化微调提出了作为代理的节点图奖励，以巩固有效的图探索。在精心策划的数据和图对齐训练下，我们的方法在八个多模态基准上实现了显著提高，在细粒度感知和推理任务中展现出强大的有效性。代码可在 https://github.com/zwyang6/SaGe 获取。

View on arXiv Download PDF AI Translation

cs.CV / 20 / 2607.05726

Association Restoration Test: Revealing Restorable Shortcuts after Unlearning

关联恢复测试：揭示在遗忘后可恢复的快捷方式

Lu, Amy, Ji, Changxiu

Abstract

Association unlearning aims to disable learned label-attribute shortcuts while preserving task performance. Existing evaluations mainly measure output-level robustness or probe whether shortcut attributes remain readable in frozen features, but neither test determines whether a retained association remains functionally usable by the original classifier. We propose the Association Restoration Test (ART), a post-hoc diagnostic for functional shortcut restorability. ART estimates class-conditional association directions, amplifies residual components, and evaluates the modified features with the original classifier head. Across Waterbirds, CelebA, SpuCoDogs, and an ISIC timestamp-artifact extension, we show that output metrics, representation probes, and ART characterize distinct aspects of shortcut mitigation. These findings motivate restoration-aware evaluation for unlearning and shortcut-mitigation methods that target learned associations rather than individual classes or concepts.

Chinese Translation

关联遗忘旨在禁用已学习的标签-属性快捷方式，同时保留任务性能。现有评估主要测量输出级别的鲁棒性或探查冻结特征中快捷方式属性是否仍然可读，但这两种测试都无法确定保留的关联是否仍然可以被原始分类器功能性地使用。我们提出了关联恢复测试（Association Restoration Test, ART），这是一种用于功能性快捷方式可恢复性的后验诊断方法。ART 估计类条件的关联方向，放大残差成分，并使用原始分类器头评估修改后的特征。在 Waterbirds、CelebA、SpuCoDogs 和 ISIC 时间戳伪影扩展的实验中，我们展示了输出指标、表示探测和 ART 描绘了快捷方式缓解的不同方面。这些发现激励了对目标在学习关联而非单个类别或概念的遗忘与快捷方式缓解方法的恢复意识评估。

View on arXiv Download PDF AI Translation

cs.CV / 21 / 2607.05727

SAMPLe: SAM-based Optimizer for Prompt Learning in VLMs

SAMPLe：基于SAM的视觉语言模型提示学习优化器

Rajoli, Hossein, Lotfi, Fatemeh, Talemi, Niloufar Alipour, Kashiani, Hossein, Ma, Xiaolong, Afghah, Fatemeh

Abstract

Pre-trained Vision-Language Models (VLMs) like CLIP have proven highly effective as foundation models for various downstream applications. However, prompt learning in VLMs encounters a performance-generalization dilemma: while prompts can be tuned to achieve high accuracy on seen distributions, this tuning process often undermines their generalizability to unseen data. The limited set of learnable prompts, which contextualize and condition the input to steer it toward the task within the pretrained VLM, tends to overfit the training data, leading to a trade-off between task-specific performance and preserving generalization. To address this dilemma, we introduce SAMPLe (Sharpness-Aware Minimization Prompt Learning), a plug-in sharpness-aware optimizer that enhances prompt generalizability by accounting for loss landscape sharpness. Unlike conventional methods, SAMPLe balances exploration and exploitation by satisfying objective function constraints at each step, dynamically adapting to the current optimization state based on the local curvature and gradient properties. This approach reduces overfitting on seen distributions and improves adaptability to unseen data, preserving the generalization potential of pre-trained VLM models. We integrate SAMPLe into multiple prompt learning frameworks, including CoOp, CoCoOp, MaPLe, TCP, and Co-Prompt, demonstrating its effectiveness across diverse methods. Experiments show that SAMPLe elevates prompt learning frameworks and consistently outperforms existing optimizers across diverse settings, establishing itself as a robust, model-agnostic solution for prompt learning.

Chinese Translation

预训练的视觉语言模型（VLMs），如CLIP，已被证明在各种下游应用中作为基础模型非常有效。然而，VLM中的提示学习面临性能与泛化的困境：虽然可以调整提示以在已见分布上实现高准确率，但这一调整过程往往削弱了其对未见数据的泛化能力。有限的可学习提示集用于对输入进行上下文化和条件化，以引导其朝向预训练VLM中的任务，往往导致对训练数据的过拟合，从而在任务特定性能与保持泛化之间产生权衡。为了解决这一困境，我们提出了SAMPLe（Sharpness-Aware Minimization Prompt Learning），一种插件式的敏感度优化器，通过考虑损失景观的敏感度来增强提示的泛化能力。与传统方法不同，SAMPLe通过在每一步满足目标函数约束来平衡探索与利用，动态适应当前的优化状态，基于局部曲率和梯度特性。这种方法减少了对已见分布的过拟合，提高了对未见数据的适应性，保持了预训练VLM模型的泛化潜力。我们将SAMPLe集成到多个提示学习框架中，包括CoOp、CoCoOp、MaPLe、TCP和Co-Prompt，展示了其在多种方法中的有效性。实验表明，SAMPLe提升了提示学习框架，并在多种设置中始终优于现有优化器，确立了其作为一种强大的、模型无关的提示学习解决方案的地位。

View on arXiv Download PDF AI Translation

cs.CV / 22 / 2607.05733

ARMS: Anchor-Relational Motion Streaming for Seamless Solo-Social Motion Transitions

ARMS：无缝单人社交动作过渡的锚关系动作流框架

Liu, Huakun, Yu, Qing, Fujiwara, Kent, Uchiyama, Hideaki, Kiyokawa, Kiyoshi

Abstract

Generating temporally continuous and socially coherent human motion from text remains a fundamental challenge, particularly in realistic streams where people act alone, enter interactions, and later disengage. Most existing methods generate fixed-length motion clips under static agent configurations, which makes them brittle to solo-social transitions and unsuitable for incremental generation over long horizons. We propose ARMS, an Anchor-Relational Motion Streaming framework that unifies solo motion and human-human interaction within a single causal generative process. ARMS introduces a dynamics-asymmetric representation that decouples per-person temporal evolution from inter-person alignment via a partner-referenced relative-translation term, enabling seamless switching of social coupling without sacrificing long-horizon stability or spatial consistency between agents. On top of a causal latent space, a causal relational diffusion model progressively refines motion segment by segment using only past context, capturing both intra-person temporal dependencies and inter-person relations. Mode-aware relational gating activates or masks cross-agent connections, allowing the same model to support both solo and interaction generation. Experiments show that ARMS improves transition smoothness and social coherence compared to interaction-centric baselines, while also achieving competitive results on human-human interaction benchmarks.

Chinese Translation

从文本生成时间上连续且社会上连贯的人类动作仍然是一个基本挑战，尤其是在现实场景中，人们单独行动、进入互动并随后脱离的情况下。大多数现有方法在静态代理配置下生成固定长度的动作片段，这使得它们在单人社交过渡时显得脆弱，并且不适合在长时间范围内增量生成。我们提出了ARMS，一个锚关系动作流框架，它将单人动作和人际互动统一在一个因果生成过程中。ARMS引入了一种动态不对称表示，通过一个基于伙伴参考的相对位移项，将每个人的时间演变与人际对齐解耦，从而实现社交耦合的无缝切换，而不牺牲长时间的稳定性或代理之间的空间一致性。在一个因果潜在空间的基础上，因果关系扩散模型逐步细化动作，逐段使用仅过去的上下文，捕捉个人内部的时间依赖性和人际关系。模式感知关系门控激活或屏蔽跨代理连接，使得同一模型能够支持单人和互动生成。实验表明，与以互动为中心的基线相比，ARMS提高了过渡的平滑性和社会连贯性，同时在人与人互动基准测试中也取得了具有竞争力的结果。

View on arXiv Download PDF AI Translation

cs.CV / 23 / 2607.05737

Optimized Adaptive Loop Filter in Versatile Video Coding

多功能视频编码中的优化自适应环路滤波器

Xuewei, Meng, Jiaqi, Zhang, Chuanmin, Jia, Xinfeng, Zhang, Shanshe, Wang, Siwei, Ma

Abstract

In the Versatile Video Coding~(VVC) standard, adaptive loop filter~(ALF), including Geometry transformation-based Adaptive Loop Filter~(GALF) and Cross Component Adaptive Loop Filter~(CCALF), plays an essential role in reducing compression artifacts. However, it also has high coding complexity and requires many picture buffer accesses in the encoder that will increase external memory access and is unfriendly to the software and hardware design. Therefore, we propose an optimized ALF framework, including the parallel design of GALF and CCALF, the adaptive parameter decision of GALF, and one-pass CCALF scheme by effectively estimating the CCALF filtering distortion without conducting filter operation. Compared to VTM-8.0, the proposed method can reduce the picture buffer access from 152 to 1 and achieve roughly 25\% time-savings of the ALF module with negligible coding performance change under RA configuration. Some of the proposed methods have been adopted in the VVC reference software.

Chinese Translation

在多功能视频编码（VVC）标准中，自适应环路滤波器（ALF），包括基于几何变换的自适应环路滤波器（GALF）和跨分量自适应环路滤波器（CCALF），在减少压缩伪影方面发挥着重要作用。然而，它也具有较高的编码复杂性，并且在编码器中需要频繁访问图像缓冲区，这将增加外部内存访问，并对软件和硬件设计不友好。因此，我们提出了一种优化的ALF框架，包括GALF和CCALF的并行设计、GALF的自适应参数决策，以及通过有效估计CCALF滤波失真而无需进行滤波操作的一次性CCALF方案。与VTM-8.0相比，所提出的方法可以将图像缓冲区访问次数从152减少到1，并在RA配置下实现ALF模块大约25%的时间节省，同时对编码性能的影响微乎其微。部分提出的方法已被采纳到VVC参考软件中。

View on arXiv Download PDF AI Translation

cs.CV / 24 / 2607.05765

Image2Sim: Scaling Embodied Navigation via Generative Neural Simulator

Image2Sim：通过生成神经模拟器扩展具身导航

Wang, Zihan, Lee, Seungjun, Xu, Yinghao, Lee, Gim Hee

Abstract

Embodied navigation aims to build agents that interpret multimodal goals, reason in 3D space, and reach target destinations reliably in the real world. However, progress remains constrained by the lack of scalable, high-fidelity, and physically grounded interactive environments. Although real-world scanned datasets offer visual realism, they are limited by scale. In contrast, synthetic simulators scale more easily but often exhibit large sim-to-real gaps. We introduce Image2Sim, a real-time neural simulation framework that constructs high-quality interactive environments from posed RGB-D image sequences. The central idea is to decouple 3D spatial anchoring from photorealistic observation synthesis. For scene construction, Image2Sim uses a feed-forward feature Gaussian model that lifts posed RGB-D observations into a 3D feature-Gaussian representation in a single pass. For rendering, we propose a Geometry-Aware One-Step Pixel Flow model that transforms sparse and noisy Gaussian projections into high-quality panoramic RGB-D observations. Image2Sim also serves as a fully automated embodied data engine that generates high-fidelity observations, executable actions, and diverse navigation instructions at scale. It converts large collections of videos and images into nearly 20K interactive scenes and synthesizes more than 10 million navigation training samples. Navigation models trained entirely in these neural environments achieve strong improvements on major benchmarks and transfer effectively to real-world zero-shot settings. These results suggest that scalable neural simulation can serve as a practical training substrate for embodied navigation at scale.

Chinese Translation

具身导航旨在构建能够解释多模态目标、在三维空间中推理并可靠地到达目标目的地的智能体。然而，由于缺乏可扩展的、高保真且物理基础的互动环境，进展依然有限。尽管现实世界的扫描数据集提供了视觉真实感，但在规模上存在局限。相比之下，合成模拟器更易于扩展，但通常存在较大的模拟与现实之间的差距。我们提出了Image2Sim，一个实时神经仿真框架，从预设的RGB-D图像序列构建高质量的互动环境。其核心思想是将三维空间锚定与逼真的观察合成进行解耦。在场景构建方面，Image2Sim使用前馈特征高斯模型，将给定的RGB-D观察在单次传递中提升为三维特征高斯表示。在渲染中，我们提出了一种几何感知一次像素流模型，将稀疏和噪声高斯投影转化为高质量全景RGB-D观察。Image2Sim还作为一个完全自动化的具身数据引擎，能够以规模生成高保真的观察、可执行的动作以及多样的导航指令。它将大量视频和图像集合转换为近20,000个互动场景，并合成超过1000万个导航训练样本。在这些神经环境中完全训练的导航模型在主要基准上取得了显著改进，并有效地迁移到现实世界的零样本设置中。这些结果表明，可扩展的神经模拟可以作为具身导航的实用训练基础。

View on arXiv Download PDF AI Translation

cs.CV / 25 / 2607.05769

LEGATO 2: Toward Multimodal Sheet Music Recognition and Understanding

LEGATO 2：迈向多模态乐谱识别与理解

Yang, Guang, Zheng, Brian Siyuan, Ebert, Victoria, Smith, Noah A.

Abstract

We propose a novel pipeline, Legato 2, for extracting symbolic notation and semantic knowledge from images of sheet music. Legato 2 features the first large-scale neural model for optical music recognition (OMR) to operate sequentially on a system-by-system basis, following the horizontal lines of notation as they are read on the page, rather than treating the page as an undifferentiated image, enabling better scaling to arbitrarily long inputs. It is also the first OMR model capable of generating symbolic transcriptions that include embedded textual content, such as titles and annotations. The pipeline combines system-level segmentation with an autoregressive vision-LM to capture both local notation details and score structure. Across multiple datasets, Legato 2 consistently outperforms prior state of the art. We also show that symbolic transcriptions complement visual inputs for frontier language models, improving their interpretation of dense musical documents. Legato 2 establishes new state-of-the-art performance in both OMR and downstream sheet music understanding.

Chinese Translation

我们提出了一种新颖的流程，Legato 2，用于从乐谱图像中提取符号表示和语义知识。Legato 2 具备首个大规模神经模型，进行光学音乐识别（OMR），按系统逐个处理，遵循乐谱的横向排列，以读取页面上的符号，而不是将页面视作一个未区分的图像，从而更好地支持任意长输入的扩展。它也是首个能够生成包含嵌入文本内容（如标题和注释）的符号转录的 OMR 模型。该流程结合了系统级分割和自回归视觉语言模型（vision-LM），以捕捉局部的乐谱细节和乐谱结构。在多个数据集上，Legato 2 的表现始终超过以往的最好水平。我们还展示了符号转录可以补充前沿语言模型的视觉输入，改善它们对密集音乐文档的理解。Legato 2 在 OMR 和下游乐谱理解方面都确立了新的最先进表现。

View on arXiv Download PDF AI Translation

cs.CV / 26 / 2607.05783

Benchmarking the Robustness of Autonomous Driving to Environmental Illusions: A Lane Perception Perspective

基于车道感知视角的自动驾驶对环境幻觉的鲁棒性基准评估

Zhang, Tianyuan, Liu, Xianglong, Liu, Aishan, Wang, Lu, Zhang, Yitong, Yue, Peng, Zhang, Mingchuan, Liang, Siyuan, Tao, Dacheng

Abstract

Environmental illusions (eg., shadows, reflections, and tire marks) are naturally existing yet overlooked phenomena in real-world driving environments. They can disturb visual perception, leading to misinterpretation of the scene and posing serious safety risks to autonomous driving (AD) systems. However, existing researches largely overlook these phenomena, leaving a critical gap. To address this issue, we study AD robustness through the lane perception perspective, a fundamental task supporting core functions like cruise control and lane centering. We focus on two representative models: conventional lane detection (LD) and vision-language model-based systems (ADVLMs). In this work, we introduce the first benchmark, LanEvil++, for evaluating the robustness of lane perception under environmental illusions. LanEvil++ encompasses 14 types of illusions and leverages the CARLA simulator to generate 94 high-fidelity, fully controllable 3D scenes, yielding a dataset of 90,292 annotated images, 1,596 video clips, and 41,855 visual question answering pairs. Extensive evaluations demonstrate that environmental illusions substantially degrade the performance of state-of-the-art LD methods. On average, LD models experience a 5.27% drop in Accuracy and a 10.49% decline in F1-score, while ADVLMs show a 2.03% reduction in GPT-score and a 0.75% drop in Language-score. Among all illusions, shadows emerge as the most disruptive factor, reducing accuracy by up to 7.20%. Furthermore, closed-loop simulations reveal that these illusions can lead to incorrect driving decisions. Complementary real-world case studies highlight safety-critical failures in actual traffic scenes. To enhance robustness, we propose the Multimodal Illusion Defense Approach (MIDA). MIDA achieves substantial gains under challenging conditions, boosting robustness by 4.23% on LD models and 3.82% on ADVLMs.

Chinese Translation

环境幻觉（例如，阴影、反射和轮胎痕迹）是现实驾驶环境中自然存在但常被忽视的现象。它们可能干扰视觉感知，导致对场景的误解，从而对自动驾驶（AD）系统构成严重的安全风险。然而，现有研究在很大程度上忽视了这些现象，留下了一个关键的空白。为了解决这一问题，我们通过车道感知视角研究AD的鲁棒性，这是支持巡航控制和车道居中等核心功能的基本任务。我们关注两个代表性模型：传统车道检测（LD）和基于视觉-语言模型的系统（ADVLMs）。在本研究中，我们首次引入基准测试LanEvil++，用于评估在环境幻觉下车道感知的鲁棒性。LanEvil++涵盖14种幻觉类型，并利用CARLA模拟器生成94个高保真、完全可控的3D场景，形成一个包含90,292张标注图像、1,596个视频片段和41,855个视觉问答对的数据集。广泛的评估表明，环境幻觉显著降低了最先进的LD方法的性能。平均而言，LD模型的准确率下降了5.27%，F1-score下降了10.49%，而ADVLMs的GPT-score减少了2.03%，Language-score下降了0.75%。在所有幻觉中，阴影是最具破坏性的因素，准确率降低幅度高达7.20%。此外，闭环模拟表明，这些幻觉可能导致错误的驾驶决策。补充的现实案例研究突显了实际交通场景中的安全关键失败。为了增强鲁棒性，我们提出了多模态幻觉防御方法（MIDA）。在挑战性条件下，MIDA实现了显著的提升，使LD模型的鲁棒性提高了4.23%，ADVLMs提高了3.82%。

View on arXiv Download PDF AI Translation

cs.CV / 27 / 2607.05787

DeSeG: Decoupling Semantic Intent and Geometric Constraints for Physically Plausible Human-Scene Interaction

DeSeG：解耦语义意图与几何约束以实现物理上可信的人类-场景交互

Li, Jiakun, Li, Zhe, Wu, Wenqiang, Chang, Zheng, Gao, Mingqi, Yang, Jinyu, Zheng, Feng

Abstract

Synthesizing physically plausible human-scene interactions (HSI) remains a critical challenge in computer vision and the development of human avatars. Although recent generative models enable diverse motion synthesis, they suffer from an inductive bias referred to as semantic-geometric entanglement. Because spatial constraints often strongly correlate with specific actions in training data, monolithic models will learn the shortcut bias, aggressively overriding the semantic intent when faced with strict geometric cues. Furthermore, this entanglement exacerbates physical hallucinations, such as body-scene penetrations. To address these limitations, we propose DeSeG, a hierarchical framework that explicitly decouples semantic intent from geometric constraints. First, we introduce a Residual Semantic Planner that encodes textual instructions and canonicalized goal voxels into a compact latent space, enabling fine-grained semantic control independent of spatial trajectories. Second, we propose a physics regularized diffusion executor that incorporates differentiable repulsive potential fields directly into the diffusion objective, enforcing collision-aware motion generation. Extensive experiments on the Lingo dataset demonstrate that DeSeG achieves state-of-the-art performance, reducing mean scene penetration by 47% and improving semantic alignment by 29% over the SOTA baselines.

Chinese Translation

合成物理上可信的人类-场景交互（HSI）仍然是计算机视觉及人类虚拟形象开发中的一个关键挑战。尽管最近的生成模型能够实现多样化的运动合成，但它们存在一种称为语义-几何纠缠的归纳偏差。由于空间约束通常与训练数据中的特定行为强相关，单一模型在面对严格的几何线索时将学习到简化的偏差，强行覆盖语义意图。此外，这种纠缠会加剧物理虚幻现象，如人体与场景穿透。为了解决这些局限性，我们提出了DeSeG，一个明确解耦语义意图与几何约束的分层框架。首先，我们引入了一种残差语义规划器，它将文本指令和标准化目标体素编码为一个紧凑的潜在空间，从而实现与空间轨迹无关的细粒度语义控制。其次，我们提出了一种物理正则化扩散执行器，将可微分的排斥势场直接纳入扩散目标中，从而强制实现碰撞感知的运动生成。在Lingo数据集上的大量实验表明，DeSeG实现了最先进的性能，场景穿透的平均减少达47%，语义对齐提高了29%，优于现有最先进的基线。

View on arXiv Download PDF AI Translation

cs.CV / 28 / 2607.05798

Segmentation before Answering: Pixel Grounding for MLLM Visual Reasoning

回答之前的分割：多模态大型语言模型视觉推理中的像素基础

Wei, Yake, Wang, Yuan, Rao, Fengyun, Lyu, Jing, Hu, Di

Abstract

Recent advancements in Multimodal Large Language Models (MLLMs) have evolved from static perception to interleaved visual-language reasoning, often referred to as ``thinking with images''. A basic operation in this reasoning process is to zoom in on regions of interest (often represented with bounding boxes) to acquire finer visual details. In this paper, we propose \textbf{Seg}mentation before \textbf{Answer}ing (SegAnswer), which shifts the unit of zoom-in from the popular bounding box to pixel-level segmentation mask. By employing fine-grained masks to isolate the target area from cluttered environments, segmented visual input yields a more precise region of interest, effectively filtering out redundant background and interfering objects. Furthermore, the discrete patches of segmented visual input align more seamlessly with how MLLMs structure visual tokens via positional embeddings. In experiments, we evaluate SegAnswer across diverse benchmarks, including high-resolution perception, general perception, and hallucination. It achieves consistent improvements and also exhibits considerable performance on segmentation tasks, validating its capability for reliable pixel grounding.

Chinese Translation

近年来，多模态大型语言模型（MLLMs）的进步已经从静态感知演变为交织的视觉-语言推理，通常被称为“用图像思考”。这一推理过程中的基本操作是聚焦于感兴趣的区域（通常用边界框表示），以获取更精细的视觉细节。在本文中，我们提出了 extbf{分}割之后 extbf{答}复（SegAnswer），将聚焦单位从流行的边界框转移到像素级的分割掩模。通过运用细粒度掩模将目标区域从杂乱的环境中隔离出来，分割后的视觉输入能提供更精确的兴趣区域，有效地过滤掉冗余背景和干扰物体。此外，分割视觉输入的离散补丁与多模态大型语言模型通过位置嵌入结构化视觉token的方式更为无缝对接。在实验中，我们在包括高分辨率感知、一般感知和幻觉等多样基准上评估了SegAnswer。结果显示其一致性改善，同时在分割任务上也表现出显著的性能，验证了其可靠的像素基础能力。

View on arXiv Download PDF AI Translation

cs.CV / 29 / 2607.05801

TRIG: Trajectory-Rig Decoupled Metric Geometry Learning

TRIG：轨迹-刚体解耦度量几何学习

Liao, Lizhou, Xu, Wentao, Wang, Handong, Yang, Lirong, Yang, Shuai, Liu, Weiwei, Huang, Chang

Abstract

Vision-centric autonomous driving requires accurate metric geometry and ego-motion estimation from synchronized multi-camera observations. Recent visual geometry models show strong performance in pose estimation, depth prediction, and 3D reconstruction, but are not tailored to rigid multi-camera driving systems. They often encode camera poses as entangled representations, in which time-varying ego-motion and static camera-rig geometry are jointly modeled, limiting the utilization of vehicle-side geometric priors. We propose Trajectory-Rig Decoupled Metric Geometry Learning (TRIG), a geometry perception framework for autonomous driving. TRIG factorizes camera poses into ego-trajectory and camera-rig components, enabling separate modeling of ego-motion and static multi-camera topology. We introduce decoupled pose encoding and supervision, which separately constrain trajectory evolution and rig geometry for metric-consistent learning. Moreover, sparse Temporal--Spatial attention separates cross-camera interaction from temporal aggregation, reducing global attention cost while preserving geometric reasoning. Experiments on five autonomous driving benchmarks show that TRIG achieves state-of-the-art performance in pose estimation, metric depth prediction, and 3D reconstruction.

Chinese Translation

以视觉为中心的自主驾驶需要从同步的多相机观测中准确估计度量几何和自我运动。最近的视觉几何模型在姿态估计、深度预测和3D重建方面表现出色，但并未针对刚性多相机驾驶系统进行优化。它们通常将相机姿态编码为纠缠的表示，其中随时间变化的自我运动和静态的相机刚体几何被共同建模，这限制了对车辆侧几何先验的利用。我们提出了轨迹-刚体解耦度量几何学习（TRIG），这是一个用于自主驾驶的几何感知框架。TRIG将相机姿态分解为自我轨迹和相机刚体组件，使得自我运动和静态多相机拓扑可以独立建模。我们引入了解耦的姿态编码和监督，分别约束轨迹演变和刚体几何，以实现度量一致的学习。此外，稀疏时间-空间注意力将跨相机交互与时间聚合分开，从而降低全局注意力的成本，同时保留几何推理。在五个自主驾驶基准测试中的实验表明，TRIG在姿态估计、度量深度预测和3D重建方面实现了最先进的性能。

View on arXiv Download PDF AI Translation

cs.CV / 30 / 2607.05825

Complementary Roles of Image Classification and Vessel Segmentation in AI-Based Screening for Retinopathy of Prematurity Plus Disease in a Kenyan Preterm Cohort

图像分类与血管分割在肯尼亚早产儿视网膜病筛查中的互补作用

Mutisya, Fred, Onyango, Oscar, Sitati, Sarah, Ilovi, Syokau, Malik, Aeesha NJ, W'mosi, Brenda, Makini, Brian, Aluuvala, Jalemba, Onyango, Josiah, Mmene, Rachael Kanguha, Wanyee, Steven

Abstract

Background. Retinopathy of prematurity (ROP) is a preventable cause of childhood blindness, with rising burden in low- and middle-income countries where ROP-trained ophthalmologists are scarce. Plus disease, marked by retinal vessel dilation and tortuosity, triggers treatment but is subjective and variable. Automated screening could extend specialist reach, but African evidence remains limited. Methods. We analysed 121 Kenyan preterm infants, covering 237 eyes and 1,635 fundus images graded as No Plus, Pre-Plus or Plus. Vessel annotations from two graders supported segmentation training. Eleven configurations were evaluated for eye-level Plus detection using patient-grouped nested cross-validation, including image classifiers, multiple-instance learning, multi-task segmentation-classification, and segment-then-classify pipelines. Results. Vessel segmentation was feasible, achieving pooled Dice 0.533, IoU 0.368, sensitivity 0.623 and specificity 0.979 on held-out images. RGB classifiers were highly sensitive but over-referred, while segmentation-coupled models were more specific. Combining approaches improved performance: an OR-based screen achieved the highest sensitivity, an AND-based confirmation achieved the highest specificity, and a probability ensemble gave the best balanced performance, with sensitivity 0.692, specificity 0.914 and balanced accuracy 0.803, outperforming the vision classifier alone. Conclusions. Classification and vessel segmentation are complementary for ROP Plus detection in Kenyan data. Classifiers support sensitive case-finding, while segmentation improves specificity and reduces over-referral. African ROP AI systems should use combined workflows and undergo prospective multi-site validation.

Chinese Translation

背景：早产儿视网膜病（ROP）是儿童失明的可预防原因，在低收入和中等收入国家中，ROP-trained眼科医生稀缺，负担日益加重。Plus疾病以视网膜血管扩张和扭曲为特征，虽然触发了治疗，但其判断是主观和变化的。自动化筛查可以扩展专家的覆盖范围，但非洲的相关证据仍然有限。方法：我们分析了121名肯尼亚早产儿，涵盖237只眼睛和1,635幅眼底图像，按无Plus、Pre-Plus或Plus分级。来自两位评估者的血管标注支持了分割训练。使用患者分组的嵌套交叉验证，我们评估了11种眼水平Plus检测的配置，包括图像分类器、多实例学习、多任务分割-分类及分割后分类管道。结果：血管分割可行，持留图像上的综合Dice系数为0.533，IoU为0.368，灵敏度为0.623，特异性为0.979。RGB分类器灵敏度高但过度推荐，而结合分割的模型则更具特异性。结合方法提高了性能：基于OR的筛查实现了最高灵敏度，基于AND的确认达到了最高特异性，概率集成方式则提供了最佳平衡性能，灵敏度为0.692，特异性为0.914，平衡准确率为0.803，优于单独的视觉分类器。结论：在肯尼亚的数据中，分类与血管分割在ROP Plus检测中是互补的。分类器支持敏感的病例发现，而分割则提高了特异性并减少了过度推荐。非洲的ROP人工智能系统应采用综合工作流程，并经过前瞻性的多中心验证。

View on arXiv Download PDF AI Translation

cs.CV / 31 / 2607.05837

Realistic Compound-Lens Defocus Blur Synthesis

真实复合镜头散焦模糊合成

Lee, Yunkyu, Kim, Woohyeok, Cho, Sunghyun

Abstract

Defocus blur degrades fine image structures and limits visual perception, which can adversely affect downstream vision tasks. Although recent deep learning deblurring methods have achieved strong performance, their effectiveness depends on training data and often degrades across cameras and lenses due to limited optical diversity and realism in existing datasets. In this paper, we propose a pipeline for synthesizing realistic defocus deblurring datasets for diverse compound lenses. It integrates efficient wave-optics PSF computation via Debye CZT propagation, depth-aware defocus rendering with occlusion handling, and blur synthesis in the radiometrically linear space with camera ISP simulation. This unified pipeline enables the scalable generation of photorealistic defocus datasets with diverse lens characteristics. Using our pipeline, we generate CLDefocus, a large-scale synthetic dataset containing lens-diverse defocus image pairs. We further analyze the limitations of real-captured defocus datasets and show that such imperfections can bias full-reference evaluation. Extensive experiments demonstrate that models trained on CLDefocus achieve improved cross-device generalization compared to models trained on existing real and synthetic datasets.

Chinese Translation

散焦模糊会削弱图像细节结构并限制视觉感知，这可能对下游视觉任务产生不利影响。尽管最近的深度学习去模糊方法取得了良好的效果，但其有效性依赖于训练数据，且由于现有数据集中光学多样性和现实性的限制，往往在不同相机和镜头之间性能下降。本文提出了一种合成多样化复合镜头真实散焦去模糊数据集的流程。该流程通过德拜卷积（Debye CZT）传播集成了高效的波光学点扩散函数（PSF）计算、具有遮挡处理的深度感知散焦渲染，以及在光度线性空间中进行摄像机图像信号处理（ISP）仿真的模糊合成。这个统一的流程能够大规模生成具有不同镜头特性的照片级真实散焦数据集。使用我们的流程，我们生成了CLDefocus，一个包含多样化镜头的散焦图像对的大规模合成数据集。我们进一步分析了真实捕获的散焦数据集的局限性，并表明这些缺陷可能会影响全参考评估。大量实验表明，与在现有真实和合成数据集上训练的模型相比，在CLDefocus上训练的模型实现了更好的跨设备泛化能力。

View on arXiv Download PDF AI Translation

cs.CV / 32 / 2607.05850

Breaking Spurious Correlations via Generative Randomization and Cross-Variant Self-Supervised Learning

通过生成随机化和交叉变体自监督学习破除虚假相关性

Yadav, Suraj, Sharma, Anjaneya, Yadav, Siddharth

Abstract

Deep neural networks trained with Empirical Risk Minimization (ERM) often fail under distribution shifts because they exploit spurious correlations between object labels and background context. Recent generative approaches address this issue by creating counterfactual images with altered contexts, but typically use these samples as standard data augmentation, leaving the model free to retain background-sensitive representations. We propose a two-stage framework that uses generative intervention to explicitly learn background-invariant visual representations. First, we isolate the foreground object using zero-shot segmentation and generate context-shifted variants with a structure-preserving diffusion model, preserving object identity while varying the surrounding environment. We then introduce Cross-Variant Self-Supervised Learning, where variants of the same object under different backgrounds form positive pairs in a contrastive objective. This encourages the encoder to align object-centric representations while suppressing background-specific cues. Then, we fine-tune the pretrained encoder using an ERM warm-up followed by GroupDRO with layer-wise learning rates. Experiments on distribution-shift benchmarks demonstrate best worst-group performance, achieving 92.5% on Waterbirds, 81.7% on MetaShift, and 87.4% on NICO++. Code: https://github.com/surajyadav-research/GRSSL

Chinese Translation

使用经验风险最小化（ERM）训练的深度神经网络在分布转移时经常失效，因为它们利用了物体标签与背景上下文之间的虚假相关性。近期的生成方法通过创建上下文改变的反事实图像解决了这一问题，但通常将这些样本作为常规数据增强使用，导致模型仍然保留对背景敏感的表示。我们提出了一种两阶段框架，利用生成干预显式学习背景不变的视觉表示。首先，我们使用零样本分割技术隔离前景物体，并使用结构保持的扩散模型生成上下文转变的变体，保持物体身份的同时改变周围环境。接着，我们引入交叉变体自监督学习，在不同背景下的同一物体变体形成对比目标中的正样本对。这鼓励编码器对齐以物体为中心的表示，同时抑制背景特定的线索。最后，我们通过经验风险最小化（ERM）热身的方式对预训练编码器进行微调，然后使用层级学习率的GroupDRO。我们在分布转移基准测试上的实验显示出最佳的最低组表现，在Waterbirds上达到92.5%，在MetaShift上达到81.7%，在NICO++上达到87.4%。代码：https://github.com/surajyadav-research/GRSSL

View on arXiv Download PDF AI Translation

cs.CV / 33 / 2607.05859

AVA-VLM: Adaptive Visual Attention-Vision Language Model for In-the-Wild Construction Site Monitoring

AVA-VLM：用于野外施工现场监测的自适应视觉注意力-视觉语言模型

Kim, Younggun, Kim, Taeheon, Kim, Youngseo, Park, Seunghee

Abstract

Vision-Language Models (VLMs) are promising for construction-site monitoring, and recent construction-tailored VLMs have primarily adapted pretrained VLMs through direct QA-style fine-tuning from a single global image. We argue that this direct paradigm remains limited for in-the-wild deployment in terms of operational range, reliability under reduced-resolution inputs, and inference efficiency. To address these challenges, we propose AVA-VLM, an Adaptive Visual Attention-Vision Language Model that follows a human-inspired coarse-to-fine reasoning strategy. AVA-VLM first reasons over a low-resolution global image and selectively requests a high-resolution local crop only when detailed inspection is needed, similar to how a human inspector zooms in on hard-to-see yet important areas. We further introduce a region-aware Chain-of-Thought dataset that teaches the model when to inspect, where to crop, and how to use local evidence. Experiments show that AVA-VLM improves reliability under long-distance and reduced-resolution conditions while substantially reducing visual-token usage.

Chinese Translation

视觉-语言模型（VLMs）在施工现场监测中展现出良好的前景，而近期针对施工的VLMs主要通过从单一全局图像进行直接问答（QA）风格的微调来适应预训练的VLMs。我们认为，这种直接的范式在实际部署时在操作范围、在低分辨率输入下的可靠性以及推理效率等方面依然有限。为了解决这些挑战，我们提出了AVA-VLM，一种遵循人类启发的粗到细推理策略的自适应视觉注意力-视觉语言模型。AVA-VLM首先对低分辨率的全局图像进行推理，仅在需要详细检查时选择请求高清局部图像，这类似于人类检查员对难以观察但重要区域进行放大的方式。我们进一步引入了一种区域感知的思维链数据集，教导模型何时检查、在哪里裁剪以及如何利用局部证据。实验结果表明，AVA-VLM在远距离和低分辨率条件下提高了可靠性，同时显著减少了视觉标记的使用。

View on arXiv Download PDF AI Translation

cs.CV / 34 / 2607.05880

Harrison.Rad 1.5 Technical Report: A radiology foundation model that can draft reports from images, priors and clinical context

Harrison.Rad 1.5技术报告：一个能够从图像、先前信息和临床背景中起草报告的放射学基础模型

Mall, Suneeta, Nekrasov, Vladimir, Kumar, Ashnil, Karunasena, Sajith, Nibali, Aiden, Bird, Alix, Shine, Mateo Diaz, Seah, Jarrel

Abstract

Imaging demand is growing faster than the radiology workforce can expand, and reporting backlogs cannot be resolved through training and recruitment alone. The most direct opportunity is reducing the time and effort radiologists spend producing reports, a task that requires interpreting images, integrating clinical history and prior studies, and drafting structured findings. We present Harrison.Rad 1.5 (HR1.5), a radiology-specific multimodal large language model that accepts interleaved text and visual inputs and generates structured and unstructured text across plain-film radiology, spanning computed radiography, chest, musculoskeletal, abdominal, spine, and pelvic x-rays, and mammography. HR1.5 is trained through a three-stage pipeline: domain adaptation of a base language model on radiology reports, contrastive vision-encoder training with curriculum-based hard negatives on ~6 million image-report instances, and visual-question-answering fine-tuning on multi-turn conversations. We evaluate it with a Findings-Diagnosis scoring framework that extends RadGraph-XL entity extraction with ontology-based synonym matching and polarity-contradiction detection, benchmarked on RadBench, a simulated FRCR 2B Short Case examination scored against Angoff-method thresholds, ReXGradient, and internal multi-modality datasets. HR1.5 is the only system evaluated to meet the simulated FRCR passing standard and achieves the highest accuracy on closed-format clinical questions, across anatomical regions, on internal multi-body-part and mammography reporting, and on the primary clinically-aligned score for public chest reporting. We further examine explainability and model behaviour, including question-sensitive Grad-CAM heatmaps, attention analysis, and confidence estimation, to support responsible future evaluation toward clinical use, and a framework for clinically grounded assessment of report quality.

Chinese Translation

影像需求的增长速度超过了放射科医师队伍的扩展能力，仅通过培训和招聘无法解决报告积压问题。直接的机会是减少放射科医师在生成报告上花费的时间和精力，这一任务需要对图像进行解释、整合临床历史和先前研究，并草拟结构化结果。我们提出了Harrison.Rad 1.5（HR1.5），这是一个专门针对放射学的多模态大型语言模型，能够接受交叉的文本和视觉输入，并在普通电影放射学中生成结构化和非结构化文本，涵盖计算机放射学、胸部、肌肉骨骼、腹部、脊柱和盆腔X光，以及乳腺摄影。HR1.5通过三个阶段的流程进行训练：在放射学报告上对基础语言模型进行领域适应，通过课程基础的困难负样本对约600万个图像-报告实例进行对比视觉编码器训练，以及在多轮对话中进行视觉问答微调。我们使用Findings-Diagnosis评分框架进行评估，该框架扩展了RadGraph-XL实体提取，通过本体基础的同义词匹配和极性-矛盾检测进行了增强，基于RadBench的模拟FRCR 2B短期案例考核，使用Angoff方法阈值进行评分，以及ReXGradient和内部多模态数据集。HR1.5是唯一被评估为符合模拟FRCR通过标准的系统，并在闭合格式临床问题中获得最高准确率，在解剖区域、内部多部位和乳腺摄影报告中，以及在公共胸部报告的主要临床关联评分中表现突出。我们进一步考察了可解释性和模型行为，包括对问题敏感的Grad-CAM热图、注意力分析和置信度估计，以支持未来的负责任评估，以及临床实时评价报告质量的框架。

View on arXiv Download PDF AI Translation

cs.CV / 35 / 2607.05891

Few-Medoids: An Embarrassingly Simple Coreset Selection Method for Few-Shot Knowledge Distillation

Few-Medoids：一种简单易用的少样本知识蒸馏核心集选择方法

Dilmac, Cemil-Andrei, Croitoru, Florinel-Alin, Ionescu, Radu Tudor

Abstract

Coreset selection aims to identify a small and highly representative subset of a massive dataset for efficient model training. The problem remains challenging even in the few-shot knowledge distillation (KD) setup, where a full-scale pre-trained teacher informs the student network. Typical sample selection strategies often struggle to surpass the random selection baseline. In this paper, we showcase few-medoids, an embarrassingly simple coreset selection strategy that chooses the samples closest to the centroid (average image) of each class. We present extensive KD experiments on four datasets, covering a wide range of image classification problems, and three teacher-student model pairs, comprising both convolutional and transformer networks. Although the proposed method is embarrassingly simple, our empirical results indicate that few-medoids is able to consistently surpass the random selection baseline, as well as the other coreset selection strategies. We therefore consider that few-medoids can be used as a drop-in replacement for commonly-used baselines (e.g. herding or k-center Greedy), in future research on coreset selection. To reproduce the reported results, we publicly release our code at https://github.com/CemilAndreiDilmac/Few-Shot-KD-Coreset.

Chinese Translation

核心集选择旨在从庞大的数据集中识别出一个小而高度代表性的子集，以实现高效的模型训练。在少样本知识蒸馏（KD）设置中，即全规模预训练教师指导学生网络的情境下，这一问题仍然具有挑战性。典型的样本选择策略往往难以超越随机选择的基线。在本文中，我们展示了 few-medoids，这是一种简单易用的核心集选择策略，它选择与每个类别的质心（平均图像）最接近的样本。我们在四个数据集上进行了广泛的KD实验，涵盖了多种图像分类问题，以及三对教师-学生模型，包含卷积网络和变压器网络。尽管所提出的方法极其简单，但我们的经验结果表明，few-medoids 能够持续超越随机选择的基线以及其他核心集选择策略。因此，我们认为 few-medoids 可以作为未来核心集选择研究中常用基线（例如 herd 或 k-center Greedy）的有效替代方案。为了重现报告的结果，我们在 https://github.com/CemilAndreiDilmac/Few-Shot-KD-Coreset 上公开发布了我们的代码。

View on arXiv Download PDF AI Translation

cs.CV / 36 / 2607.05906

GaussFusion: Towards Multimodal 3D Gaussian Pretraining

GaussFusion：面向多模态3D高斯预训练

You, Zhixuan, Zhu, Jihua, Sun, Yiding, Guo, Zihao, Cheng, Haozhe, Zhang, Dongxu, Chen, Lin, Luo, Hainan

Abstract

3D Gaussian Splatting provides an explicit representation that jointly models geometry and appearance, serving as a scalable foundation for 3D representation learning. Existing pre-training methods for Gaussian representations, such as masked Gaussian reconstruction, primarily capture local structures but offer limited semantic supervision. In this paper, we propose GaussFusion, a multimodal pre-training framework for 3D Gaussian representations. GaussFusion integrates image and text supervision into masked Gaussian modeling through cross-modal semantic alignment, enabling the Gaussian encoder to learn both visual and language-level semantic information during pre-training. To better adapt masked modeling to the non-uniform distribution of Gaussian primitives, we further propose Gaussian Salience-guided Multi-scale Hole Masking (GSHM). GSHM constructs spatially continuous masked regions based on Gaussian salience. By applying hole masks at multiple scales, GSHM encourages the encoder to capture both fine-grained local patterns and broader structural dependencies. Extensive experiments on downstream tasks demonstrate that GaussFusion improves the transferability of Gaussian representations. Notably, GaussFusion outperforms Gaussian-MAE on ModelNet40 and ScanObjectNN (PB-T50-RS) by 0.61\% and 3.85\%, respectively.

Chinese Translation

3D高斯溅射提供了一种显式表示，联合建模几何和外观，为3D表示学习提供可扩展的基础。现有的高斯表示预训练方法，如掩蔽高斯重建，主要捕捉局部结构，但提供的语义监督有限。本文提出了GaussFusion，一个用于3D高斯表示的多模态预训练框架。GaussFusion通过跨模态语义对齐将图像和文本监督集成到掩蔽高斯建模中，使高斯编码器能够在预训练过程中学习视觉和语言级别的语义信息。为了更好地将掩蔽建模适应于高斯原语的非均匀分布，我们进一步提出了高斯显著性引导的多尺度孔掩蔽（Gaussian Salience-guided Multi-scale Hole Masking，GSHM）。GSHM基于高斯显著性构建空间连续的掩蔽区域。通过在多个尺度上应用孔掩蔽，GSHM鼓励编码器捕捉细粒度的局部模式和更广泛的结构依赖性。在下游任务上的广泛实验表明，GaussFusion提高了高斯表示的可迁移性。值得注意的是，GaussFusion在ModelNet40和ScanObjectNN（PB-T50-RS）上的表现优于Gaussian-MAE，分别提高了0.61 ext{%}和3.85 ext{%}。

View on arXiv Download PDF AI Translation

cs.CV / 37 / 2607.05910

PolicyShiftGuard: Benchmarking and Improving Policy-Adaptive Image Guardrails

PolicyShiftGuard：基准测试和改进政策自适应图像保护措施

Song, Mingyang, Xu, Luxin, Sun, Haoyu, Pan, Minzhou, Cheng, Yu, Li, Bo

Abstract

Image guardrails are typically trained and evaluated under a fixed safety policy, implicitly treating safety as an intrinsic property of an image. Real deployments are different: the same image may be allowed in one product, restricted in another, and newly disallowed when a policy boundary changes. We study policy-adaptive image guardrailing, where a model must decide whether an image violates the currently supplied policy and generalize to held-out policy definitions. We introduce PolicyShiftBench, a comprehensive benchmark with 2,000 policy-discriminative instances over 265 images, where each image is paired with 7.55 policy-conditioned prompts on average to test whether models adapt to the active policy rather than relying on image-level safety priors. We then propose PolicyShiftGuard, a compact policy-conditioned guardrail trained with a two-stage training recipe that combines Randomized Policy SFT (RP-SFT) with Boundary-Pair Policy Adaptation (BP-Adapt). BP-Adapt trains matched prompts for the same image and risk category using standard label supervision and a pairwise comparison loss that separates blocking policies from passing policies. Experiments show that existing VLMs and specialized guardrails remain brittle under policy shifts, while PolicyShiftGuard substantially improves policy-sensitive performance. The 7B model achieves SOTA performance of 76.9 Avg. F1 and 72.1 Avg. PSS on PolicyShiftBench, transfers well to UnSafeBench and SafeEditBench, and improves the latency-performance trade-off with a concise output format. Ablations confirm that matched pass/block boundary pairs are essential for stable policy adaptation.

Chinese Translation

图像保护措施通常在固定安全策略下进行训练和评估，隐含地将安全性视为图像的内在属性。然而，实际部署情况有所不同：同一图像在一个产品中可能是被允许的，在另一个产品中则受到限制，而当政策边界变化时可能会被新的禁止措施限制。我们研究了政策自适应的图像保护，其中模型必须决定一幅图像是否违反了当前提供的政策，并能够推广到未包含的政策定义。我们引入了PolicyShiftBench，这是一个全面的基准测试，包含2,000个具有政策区分性的实例，涵盖265幅图像，其中每幅图像平均配对7.55个政策条件提示，以测试模型是否能够适应当前的活跃政策，而不是依赖于图像级别的安全先验。随后，我们提出了PolicyShiftGuard，这是一个紧凑的政策条件保护措施，采用二阶段训练过程，结合了随机政策自监督微调（Randomized Policy SFT, RP-SFT）与边界配对政策适配（Boundary-Pair Policy Adaptation, BP-Adapt）。BP-Adapt使用标准标签监督和配对比较损失训练针对同一图像和风险类别的匹配提示，从而区分阻断政策和通过政策。实验表明，现有的视觉语言模型（VLMs）和专门的保护措施在政策变化下仍然表现脆弱，而PolicyShiftGuard显著提高了对政策的敏感性性能。7B模型在PolicyShiftBench上达到了76.9的平均F1和72.1的平均PSS的最优性能，而且在UnSafeBench和SafeEditBench上表现良好，同时通过简洁的输出格式提高了延迟与性能的权衡。消融实验确认，匹配的通过/阻断边界对稳定的政策适应至关重要。

View on arXiv Download PDF AI Translation

cs.CV / 38 / 2607.05911

Progressive Reasoning with Primitive Correction for Compositional Zero-Shot Learning

具有基本修正的渐进推理用于组合零样本学习

Chen, Ziyi, Shi, Haoyan, Xu, Sunhan, Lang, Congyan

Abstract

Compositional Zero-Shot Learning (CZSL) aims to combine known attributes and objects as primitives for recognizing previously unseen attribute-object pairs. Prior works either predict attributes and objects independently, missing their strong contextual dependency, or use unidirectional conditional modeling (e.g., object-guided attribute prediction), which is prone to error propagation. We propose PRPC, a Progressive Reasoning framework with Primitive Correction, which explicitly models the bidirectional dependency between attributes and objects via step-wise inference. PRPC performs mutual correction of primitives to suppress prediction errors in earlier steps. Specifically, we formulate CZSL as structured, Q&A-style Chain-of-Thought reasoning process and constrain the MLLM to follow predefined semantic steps to generate intermediate decisions. To further enhance the reliability and logical consistency of intermediate reasoning, we introduce reinforcement learning post-training with a GRPO-based objective, providing step-level rewards aligned with the progressive inference procedure. Extensive experiments on three CZSL benchmarks demonstrate that PRPC achieves state-of-the-art performance, validating the effectiveness of progressive reasoning and bidirectional correction for robust compositional generalization.

Chinese Translation

组合零样本学习（CZSL）旨在将已知属性和对象结合为原语，以识别先前未见的属性-对象对。之前的工作要么独立预测属性和对象，未能捕捉它们之间强烈的上下文依赖关系，要么使用单向条件建模（例如，面向对象的属性预测），这容易导致错误传播。我们提出了PRPC（具有基本修正的渐进推理框架），通过逐步推理显式建模属性和对象之间的双向依赖关系。PRPC相互修正原语，以抑制早期步骤中的预测错误。具体而言，我们将CZSL表述为结构化、问答风格的链式推理过程，并约束多模态大语言模型（MLLM）遵循预定义的语义步骤以生成中间决策。为了进一步增强中间推理的可靠性和逻辑一致性，我们引入了基于GRPO目标的强化学习后训练，为渐进推理过程提供与步骤相关的奖励。在三个CZSL基准上的大量实验表明，PRPC实现了最先进的性能，验证了渐进推理和双向修正在稳健组合泛化中的有效性。

View on arXiv Download PDF AI Translation

cs.CV / 39 / 2607.05955

NegROI: Click-Centric Uncertainty-Guided Refinement with Scene-Conditioned Negative Prompts for Robust Interactive 3D Segmentation

NegROI：基于点击中心的不确定性引导的区域细化框架与场景条件下的负提示，用于稳健的交互式3D分割

Zhang, Shuheng, Wu, Feng

Abstract

Interactive 3D segmentation aims to extract object masks in point clouds with minimal user clicks. Despite recent progress, most existing approaches still struggle with (i) coarse voxel resolution that blurs fine boundaries under limited clicks and (ii) hard false positives caused by confusing background structures. These issues are exacerbated by density and scale shifts across datasets (e.g., dense RGB-D reconstructions vs. sparse LiDAR scans), where fixed refinement heuristics and purely click-driven decoding generalize poorly. To address them, we propose NegROI -- a novel transformer-based interactive framework that couples click-centric multi-resolution refinement with scene-conditioned negative prompts. Given a coarse voxel prediction, it refines only a local Region Of Interest (ROI) around the current click on a finer grid and fuses refined logits back to the coarse mask. To improve robustness and efficiency, we introduce uncertainty-driven selective refinement that prioritizes ambiguous regions. Meanwhile, we model hard background patterns via a set of scene-conditioned negative prompts obtained by cross-attention over scene tokens. We further stabilize these prompts with a diversity regularizer. Finally, we propose boundary-aware hard negative mining to supervise negative-prompt attention toward boundary-proximal, high-confidence false positives. Our experiments on common benchmark datasets (i.e., ScanNet, S3DIS, and KITTI) demonstrate improved click efficiency and reduced false positives, with stronger cross-dataset robustness than the state-of-the-art baselines.

Chinese Translation

交互式3D分割旨在通过最少的用户点击从点云中提取物体掩膜。尽管近期取得了一些进展，但大多数现有方法仍然面临以下挑战：（i）粗糙的体素分辨率在有限点击下模糊了细微边界和（ii）由于背景结构混淆引起的难以处理的假阳性。这些问题在不同数据集之间的密度和规模变化（例如，稠密的RGB-D重建与稀疏的激光雷达扫描）中更加突出，因为固定的细化启发式和纯粹基于点击驱动的解码效果较差。为了解决这些问题，我们提出了NegROI——一种新颖的基于变换器的交互框架，将以点击为中心的多分辨率细化与场景条件下的负提示相结合。在给定粗糙的体素预测的基础上，该框架只在当前点击附近的更细网格上对局部感兴趣区域（ROI）进行细化，并将细化后的逻辑回归融合回粗略掩膜。为了提高稳健性和效率，我们引入了基于不确定性驱动的选择性细化，优先处理模糊区域。同时，我们通过对场景标记进行交叉注意力，建模难处理的背景模式，形成一组条件于场景的负提示。我们进一步用多样性正则化器来稳定这些负提示。最后，我们提出了边界感知的难处理负样本挖掘，监督负提示注意力朝向接近边界的高置信度假阳性。我们在常见基准数据集（即ScanNet、S3DIS和KITTI）上的实验显示，与最先进的基线相比，点击效率得到改善，假阳性数量减少，且跨数据集的稳健性更强。

View on arXiv Download PDF AI Translation

cs.CV / 40 / 2607.05965

Decoupled Single-Mask Annotation Noise Detection via Cross-Sectional Patch Self-Consistency

通过横截面补丁自一致性解耦单掩膜注释噪声检测

Zhu, Yinheng, Xu, Xiaowei

Abstract

Vascular computed tomography datasets are commonly annotated only once per scan, yielding the pervasive yet under addressed problem of single mask annotation noise. Existing solutions either require costly multirater fusion or are coupled with network training, preventing explicit auditing of where and why labels fail. We introduce a decoupled framework for single-mask annotation noise detection that leverages cross-sectional patch self-consistency to produce interpretable and auditable noise evidence. Tubular anatomy exhibits strong cross-sectional recurrence: patches extracted orthogonally along vessel centrelines recur in appearance across locations and subjects. Thus, anatomically similar patches should have consistent masks, and disagreement signals unreliable annotation. Our method samples cross-sectional patches, retrieves intensity-equivalent neighbours via scalable vector search, and computes a patch-level noise score from statistical mask disagreement, yielding explicit image-mask evidence for every flagged region. Aggregating scores produces scan-level quality maps for dataset quality assessment or quality-weighted training. Experiments on the coronary CT dataset validate the detected noise for improving training robustness and reveal systematic annotation biases. Specifically, transverse and oblique vessels exhibit 5.1 times higher error rates than axis-aligned structures, with additional correlations to cross-sectional area and intensity. Code is available here.

Chinese Translation

血管计算机断层扫描数据集通常仅在每次扫描时进行一次注释，这导致了普遍存在但未得到充分解决的单掩膜注释噪声问题。现有解决方案要么需要昂贵的多评估者融合，要么与网络训练相耦合，无法明确审计标签失败的原因和位置。我们提出了一种解耦的单掩膜注释噪声检测框架，该框架利用横截面补丁自一致性来生成可解释和可审计的噪声证据。管状解剖结构表现出强烈的横截面重复性：沿血管中心线正交提取的补丁在不同位置和受试者之间外观相似。因此，解剖上相似的补丁应具有一致的掩膜，而不一致则表明注释不可靠。我们的方法采样横截面补丁，通过可扩展向量搜索检索强度等效的邻居，并从统计掩膜不一致中计算补丁级噪声分数，为每个标记区域提供明确的图像-掩膜证据。聚合分数生成扫描级质量图，用于数据集质量评估或质量加权训练。在冠状动脉CT数据集上的实验验证了检测到的噪声对于提高训练鲁棒性的重要性，并揭示了系统性的注释偏差。具体而言，横截面和斜向血管的错误率比轴对齐结构高出5.1倍，并与横截面积和强度存在额外的相关性。代码可在此处获取。

View on arXiv Download PDF AI Translation

cs.CV / 41 / 2607.05978

Propose and Attend: Training-free MLLM Grounding Confidence via Multi-Token Localized Attention

提出与关注：通过多标记局部注意力实现无训练的多模态大语言模型基础信心

Shalam, Daniel, Baruch, Emanuel Ben, Cohen, Avi Ben, Remez, Tal

Abstract

Multimodal large language models can emit localized predictions, bounding boxes for objects and temporal windows for video and audio events, but they hallucinate these regions prolifically. The model's own token log-probabilities are nearly uninformative: they conflate grounding quality with input ambiguity, and coordinate tokens become near-deterministic once the model commits. We propose Multi-Token Localized Attention (MTLA): a training-free, post-hoc score that measures how strongly a prediction's tokens attend to the region they claim. Prior attention-based detectors, which sum attention over the entire input modality and read a single response token, are weaker special cases; we show that summing only within the claimed region and aggregating across all prediction tokens recovers a stronger grounding signal. The same recipe applies almost trivially to other modalities and tasks: object detection in images and temporal localization in video and audio. Across multiple MLLM families and three modalities, MTLA improves hallucination AUROC by +7 to +38 over the best prior training-free baseline. Used as a confidence score for re-ranking, it nearly doubles the zero-shot COCO detection AP of an open-source 8B generalist (from 20.4 to 37.0), narrowing the gap to supervised detectors without any task-specific training.

Chinese Translation

多模态大语言模型能够发出局部的预测，包括对象的边界框以及视频和音频事件的时间窗口，但它们往往会虚构这些区域。模型自身的标记对数概率几乎没有信息量：它们将基础质量与输入的模糊性混为一谈，一旦模型做出承诺，坐标标记几乎变得确定性。我们提出了多标记局部注意力（Multi-Token Localized Attention, MTLA）：一种无训练的后处理评分，测量预测的标记在多大程度上关注于它们所声称的区域。先前基于注意力的检测器，通过对整个输入模态的注意力求和并读取单一反应标记，是弱特殊情况；我们表明，仅在声称的区域内求和并在所有预测标记间聚合能恢复更强的基础信号。这一方案几乎可以无缝应用于其他模态和任务：图像中的对象检测以及视频和音频中的时间定位。在多个多模态大语言模型家族和三种模态下，MTLA将虚构AUROC提高了7到38，相较于最好之前的无训练基线。作为重排的信心评分使用，几乎使一个开源8B通用模型的零样本COCO检测AP几乎翻倍（从20.4增至37.0），缩小了与监督检测器之间的差距，而没有任何任务特定的训练。

View on arXiv Download PDF AI Translation

cs.CV / 42 / 2607.05988

SpecTrack: Spectral Prompt Guided Adaptive Experts for Multispectral Object Tracking

SpecTrack：基于光谱提示引导的自适应专家多光谱目标跟踪

Tan, Xingyu, Qin, Yunrong, Hu, Mengjie

Abstract

Multispectral image(MSI) and hyperspectral image(HSI) object tracking object tracking exploits recorded band-wise observations to improve target--background discrimination under similar RGB appearance, mixed pixels, illumination variation, occlusion, and clutter. However, existing trackers commonly process all search regions through a fixed capacity spectral--spatial path, ignoring that tracking difficulty varies substantially across frames and target states. Clear regions may require only lightweight local discrimination, whereas ambiguous boundaries and spectrally similar distractors often demand stronger contextual reasoning. To address this limitation, we propose SpecTrack, a spectral--spatial complexity-aware tracker that formulates MSI tracking as search-region-level adaptive capacity allocation. Its core component, the Spectral Adaptive Mixture-of-Experts (SAMoE) module, provides a capacity-ordered expert pool with progressively increasing latent rank, receptive field, and depth. Expert selection is guided by a Spectral Prompt Router, which fuses semantic context, spatial boundary cues, and a latent channel-variation cue computed after multispectral patch embedding to activate a sparse subset of SAMoE experts for each search region. In parallel, a Shared Global Expert supplies common latent spectral--spatial context to reduce fragmented sparse-routing decisions. Experiments on MUST, MSITrack, and HOTC20 demonstrate a favorable accuracy--efficiency trade-off. The accuracy-oriented SpecTrack-L384 achieves state-of-the-art or highly competitive AUCs of 65.2\%, 51.9\%, and 72.6\% on the three benchmarks, while the balanced SpecTrack-B224 reaches 62.4\% AUC at 43.7 FPS on MUST. An additional GOT-10k evaluation indicates RGB-domain architectural generalization, with SpecTrack-L384 achieving 79.3\% AO.

Chinese Translation

多光谱图像（MSI）和高光谱图像（HSI）目标跟踪利用记录的波段观察来提升目标与背景在相似RGB外观、混合像素、照明变化、遮挡和杂乱背景下的区分能力。然而，现有的跟踪器通常通过固定容量的光谱-空间路径处理所有搜索区域，忽略了跟踪难度在不同帧和目标状态之间存在显著差异。清晰区域只需轻量级的局部区分，而模糊边界和光谱上相似的干扰物通常需要更强的上下文推理。为了解决这一局限性，我们提出了SpecTrack，一种光谱-空间复杂度感知的跟踪器，将多光谱图像跟踪视为基于搜索区域的自适应容量分配。其核心组件，光谱自适应混合专家（SAMoE）模块，提供一个容量有序的专家池，该池具有逐渐增加的潜在秩、感受野和深度。专家选择由光谱提示路由器引导，后者通过融合语义上下文、空间边界线索以及在多光谱补丁嵌入后计算的潜在通道变异线索，激活每个搜索区域的SAMoE专家的稀疏子集。同时，一个共享全局专家提供共同的潜在光谱-空间上下文，以减少破碎的稀疏路由决策。在MUST、MSITrack和HOTC20上的实验展示了良好的准确性-效率权衡。以准确性为导向的SpecTrack-L384在三个基准上达到了65.2%、51.9%和72.6%的最新或高度竞争的AUC，而平衡的SpecTrack-B224在MUST上以43.7 FPS达到了62.4%的AUC。额外的GOT-10k评估显示了在RGB域的架构泛化，SpecTrack-L384获得了79.3%的AO。

View on arXiv Download PDF AI Translation

cs.CV / 43 / 2607.05994

SparseCtrl-HOI: Sparse Temporal Control for Human-Object Interaction Video Generation

SparseCtrl-HOI：用于人机交互视频生成的稀疏时间控制

Xie, Shenbo, Cai, Mingrui, Yang, Xu, Liu, Yifei, Ding, Changxing

Abstract

Human-Object Interaction (HOI) video generation aims to synthesize realistic videos of humans manipulating diverse objects, serving as a promising avenue for AI-driven live streaming e-commerce. A primary obstacle in this domain lies in the complexity of modeling fine-grained physical dynamics and the intricate spatial-temporal coordination between human hands and objects. Existing approaches to this problem typically rely on dense temporal guidance, e.g., frame-wise hand-object pose sequences, to strictly control the interaction process. However, such dense guidance incurs high annotation costs and affects motion synthesis diversity. To overcome these limitations, we introduce SparseCtrl-HOI, a novel sparse temporal control framework for HOI video generation. It requires only a few keyframes that capture interaction states at designated timestamps. Specifically, we employ a Time-Controlled Rotary Positional Embedding (TiRoPE) mechanism to temporally anchor these keyframes while preserving their spatial integrity. Subsequently, to govern the dynamics across intermediate frames, we propose a Motion Prior Injection Module that leverages Multimodal Large Language Models (MLLMs) to extract high-level motion priors. This empowers the model to hallucinate logically and physically plausible transitions. Furthermore, we build SparseHOI-5K, a high-quality and richly annotated dataset for HOI video generation with sparse temporal control. Comprehensive evaluations confirm that our method substantially reduces annotation overhead while synthesizing superior live-streaming e-commerce videos. Both our code and dataset are publicly available at https://mpi-lab.github.io/SparseCtrl-HOI.

Chinese Translation

人机交互（HOI）视频生成旨在合成现实感强烈的人类操控多样物体的视频，这为基于人工智能的直播电商提供了一个有前景的途径。该领域的主要障碍在于建模细粒度物理动态以及人手与物体之间复杂的时空协调的复杂性。现有的方法通常依赖于密集的时间指导，例如逐帧的手-物体姿态序列，以严格控制交互过程。然而，这种密集的指导会导致高昂的标注成本，并影响运动合成的多样性。为克服这些限制，我们提出了SparseCtrl-HOI，一个用于HOI视频生成的新型稀疏时间控制框架。它仅需少量关键帧来捕捉在指定时间戳的交互状态。具体而言，我们采用了一种时间控制旋转位置嵌入（Time-Controlled Rotary Positional Embedding, TiRoPE）机制，以时间上锚定这些关键帧，同时保持其空间完整性。随后，为了控制中间帧之间的动态，我们提出了一种运动先验注入模块，该模块利用多模态大型语言模型（Multimodal Large Language Models, MLLMs）提取高层次的运动先验。这使得模型能够生成逻辑上和物理上合理的过渡。此外，我们构建了SparseHOI-5K，一个高质量且丰富标注的HOI视频生成数据集，支持稀疏时间控制。全面的评估确认我们的方法显著减少了标注开销，同时合成了优质的直播电商视频。我们的代码和数据集均可在 https://mpi-lab.github.io/SparseCtrl-HOI 获取。

View on arXiv Download PDF AI Translation

cs.CV / 44 / 2607.05996

Unlearnable Faces: Privacy Protection Surviving Extraction Pipeline

不可学习的人脸：生存提取管道的隐私保护

Oh, Byunghoon, Park, Sunghwan, Lee, Jaewoo

Abstract

Unlearnable examples keep publicly shared photos from being learned by unauthorized face-recognition models. An imperceptible perturbation, added before sharing, makes any model trained on the protected photos fail on clean faces. The perturbation is crafted on the shared image, however the attacker trains on the face it extracts, cropped and resized to the recognizer input, and under this extraction the protection collapses. We propose LPID, which builds the extraction into the unlearnable-example objective. LPID confines the perturbation to the extracted face region and optimizes it through a differentiable model of the extraction, concentrating its energy in the frequency band the extraction preserves. Because this robustness is a property of the transform rather than of any identity, LPID is re-optimized per album and protects even users it has never seen. LPID attains the lowest attacker accuracy of all methods in every setting we evaluate, holding the attacker below $10\%$ under crop+resize extraction on identities unseen at protection time, while remaining imperceptible at $32.7$\,dB PSNR and $0.161$ LPIPS.

Chinese Translation

不可学习的示例防止未经授权的人脸识别模型从公开共享的照片中学习。在共享之前添加的不可察觉的扰动使得任何在受保护照片上训练的模型在干净的人脸上都无法成功。扰动是针对共享图像设计的，然而攻击者在提取的人脸上进行训练，该人脸被裁剪并调整为识别器输入，在这种提取下，保护机制崩溃。我们提出了LPID，它将提取过程纳入不可学习示例的目标中。LPID将扰动限制在提取的人脸区域，并通过可微分的提取模型对其进行优化，将能量集中在提取所保留的频率带内。由于这种鲁棒性是变换的属性而非任何身份的属性，LPID在每个相册中重新优化，即使是从未见过的用户也能得到保护。在我们评估的所有设置中，LPID在所有方法中都达到了最低的攻击者准确率，在保护时未见过的身份下，攻击者在裁剪+调整大小的提取下保持在$10\%$以下，同时在$32.7$ dB PSNR和$0.161$ LPIPS下保持不可察觉。

View on arXiv Download PDF AI Translation

cs.CV / 45 / 2607.06007

OBBSeg: Irregular Lesion Segmentation under Oriented Bounding Box Annotations

OBBSeg：基于定向边界框注释的不规则病灶分割

Wei, Jun, Liu, Xinchang, Liu, Yu, Yang, Chuhua, Wang, Shuhui, Huang, Hui

Abstract

Pixel-level annotation remains a major bottleneck in medical image segmentation, making weak supervision an attractive yet under-constrained alternative. We propose OBBSeg, an intermediate supervision paradigm guided by Oriented Bounding Boxes (OBBs) that bridges the gap between full and weak supervision. By jointly encoding spatial extent and orientation, OBBs provide compact geometric supervision that better aligns with elongated or anisotropic lesions, reducing the ambiguity of coarse box annotations. To mitigate the inherent rectangular bias of OBBs, we introduce a Mask-to-OBB loss, a differentiable formulation that enforces geometric consistency between predicted masks and OBB regions. Furthermore, we incorporate prompt-driven semantic guidance through two complementary modules-PAFE and DBFE-which enhance foreground representation and suppress background interference. Extensive experiments on 13 datasets across 5 imaging modalities show that OBBSeg not only outperforms existing weakly supervised methods but also achieves performance comparable to fully supervised approaches, demonstrating its potential for efficient and scalable medical image segmentation. The code is available at https://github.com/StarLxc3/OBBSeg.

Chinese Translation

像素级注释仍然是医学图像分割中的一个主要瓶颈，使得弱监督成为一种具有吸引力但受限的替代方案。我们提出了OBBSeg，这是一种由定向边界框（Oriented Bounding Boxes，OBBs）引导的中间监督范式，架起了完全监督与弱监督之间的桥梁。通过共同编码空间扩展和方向性，OBBs提供了紧凑的几何监督，更好地与细长或各向异性病灶对齐，减少了粗略框注释的歧义。为了缓解OBBs固有的矩形偏差，我们引入了一种Mask-to-OBB损失，这是一种可微分的形式，强制执行预测的掩模与OBB区域之间的几何一致性。此外，我们通过两个互补模块（PAFE和DBFE）整合了基于提示的语义引导，这增强了前景表示并抑制了背景干扰。在5种成像模式下的13个数据集上进行的广泛实验表明，OBBSeg不仅超过了现有的弱监督方法，还达到了与完全监督方法相媲美的性能，展示了其在高效且可扩展的医学图像分割中的潜力。代码可在 https://github.com/StarLxc3/OBBSeg 获取。

View on arXiv Download PDF AI Translation

cs.CV / 46 / 2607.06012

Structured Data Extraction from Real Estate Documents using Clustering, Classification, and Large Language Models

基于聚类、分类和大型语言模型的房地产文件结构化数据提取

Shehbaz, Muhammad Assad, Moreno-García, Carlos Francisco

Abstract

Real estate property listings expose structured metadata through the API. Still, the richest property-level information (i.e., legal status, structural condition, utility supplies, heating systems) sits in attached questionnaire documents that no automated system currently processes at scale. These documents are heterogeneous. Some are digitally generated with selectable text, others are scanned physical forms. There are even more complex layouts that contain checkbox annotations that defeat conventional text extraction. In this paper, we present an end-to-end pipeline for acquiring, classifying, and extracting structured data from selectable text documents. The pipeline was applied to 3965 questionnaire documents collected from a live property platform via reverse-engineered REST APIs. First, we classified each document into one of three structural categories (text_only, scanned, and special_char), then extracted 35 predefined property attributes from eligible documents using DeepSeek R1 as the Large Language Model, prompted to return a structured JSON object. All 2781 submitted documents were processed successfully, producing a final dataset of 2766 unique property records. Downstream validation confirmed the data quality. Cosine similarity matching achieves a Jaccard consistency score of 0.82, and K-Means clustering produces interpretable market segments with a silhouette score of 0.2088. Results show that the proposed extraction from each property document is both feasible and reliable at this scale.

Chinese Translation

房地产物业清单通过API暴露结构化元数据。然而，最丰富的物业级信息（即法律状态、结构状况、公共设施供应、供暖系统）位于附带的问卷文件中，目前没有任何自动化系统能够大规模处理这些文件。这些文件异构性大。有些是生成的可选文本的数字表单，另一些则是扫描的纸质表单。还有更复杂的布局，其中包含复选框注释，这使得常规文本提取无效。在本文中，我们提出了一种端到端流程，用于获取、分类和从可选择文本文件中提取结构化数据。该流程应用于从一个实时物业平台通过逆向工程的REST API收集的3965份问卷文件。首先，我们将每份文件分类为三种结构类别中的一种（仅文本、扫描和特殊字符），然后使用大型语言模型DeepSeek R1从符合条件的文件中提取35个预定义的物业属性，并提示返回结构化的JSON对象。所有2781份提交的文件均已成功处理，最终生成2766条唯一物业记录的数据集。下游验证确认了数据质量。余弦相似度匹配的Jaccard一致性得分为0.82，K-Means聚类生成的可解释市场细分的轮廓得分为0.2088。结果表明，在这个规模上，从每份物业文件中提取数据是既可行又可靠的。

View on arXiv Download PDF AI Translation

cs.CV / 47 / 2607.06019

KOAL: Knowledge-Driven Prostate Cancer Grading with Ordinal-Aware Learning

KOAL：基于知识驱动的前列腺癌分级模型与有序学习

Guo, Zheng, Cui, Jiaqi, Xiong, Haocheng, Han, Jize, Liu, Bo, Zhang, Qianwen, Chen, Rui, Wang, Yan

Abstract

Non-invasive prediction of Gleason Grade Group (GGG) in prostate cancer using multiparametric MRI (mpMRI) is clinically vital for reducing unnecessary biopsies. Existing GGG prediction methods face two major limitations. First, they often overlook non-image information critical for GGG prediction, including age, prostate-specific antigen (PSA), and expert priors embedded in radiology reports. Second, they tend to oversimplify GGG as flat categorical labels, failing to account for its intrinsic hierarchy of primary and secondary Gleason patterns. To this end, we propose a novel Knowledge-Driven Ordinal-Aware Learning (KOAL) framework with three synergistic modules. Specifically, the Clinical-Context Modulation (CCM) module uses clinical variables (e.g., age and PSA) to dynamically modulate discriminative image representations. The Knowledge-Guided Prototype Alignment (KGPA) module leverages an LLM to extract group-specific expert knowledge from training radiology reports and clinical guidelines, producing offline semantic anchors describing grade-specific radiological findings without requiring patient-specific reports at inference. Through prototype contrastive alignment, patient-specific mpMRI representations are matched with these anchors to promote pathology-aligned representation learning. The Hierarchical Ordinal-aware Constraints (HOC) module decouples primary and secondary Gleason pattern prediction and maps their probabilistic outputs to GGG via a Differentiable Bio-logic Mapping Layer (DBML), ensuring pathological grading consistency. Experiments on public PI-CAI and in-house datasets demonstrate that KOAL outperforms state-of-the-art methods. Code is available at: https://github.com/Gother-GZ/KOAL.

Chinese Translation

使用多参数磁共振成像（mpMRI）对前列腺癌的Gleason分级组（GGG）进行无创预测，对于减少不必要的活检具有重要的临床意义。现有的GGG预测方法面临两大主要限制。首先，它们通常忽略对GGG预测至关重要的非图像信息，包括年龄、前列腺特异抗原（PSA）以及嵌入放射学报告中的专家先验知识。其次，它们往往将GGG简化为平面分类标签，未能考虑其主要和次要Gleason模式的内在层次结构。为此，我们提出了一种新颖的知识驱动有序学习（KOAL）框架，包含三个协同模块。具体而言，临床背景调节（CCM）模块利用临床变量（例如年龄和PSA）动态调节判别性图像表示。知识引导的原型对齐（KGPA）模块利用大型语言模型（LLM）从训练放射学报告和临床指南中提取组专属的专家知识，生成离线语义锚点，描述特定等级的放射学发现，而无需在推理时使用特定患者的报告。通过原型对比对齐，患者特定的mpMRI表示与这些锚点匹配，以促进与病理对齐的表示学习。层次化有序约束（HOC）模块解耦主要和次要Gleason模式预测，并通过可微生物逻辑映射层（DBML）将其概率输出映射到GGG，确保病理分级的一致性。在公共PI-CAI和内部数据集的实验中，KOAL的表现超过了最先进的方法。代码可在以下链接获取： https://github.com/Gother-GZ/KOAL.

View on arXiv Download PDF AI Translation

cs.CV / 48 / 2607.06023

Why does Deep Learning Improve Visual SLAM?

深度学习为何能提升视觉SLAM的性能？

Cioffi, Giovanni, Scaramuzza, Davide

Abstract

Visual SLAM is a well-established technology utilized in a wide range of real-world applications. However, its performance still degrades under challenging visual conditions, such as low texture, severe motion blur, and poor illumination. Systems based on deep learning outperform classical geometry-based ones and achieve state-of-the-art results by combining learned 2D data association and uncertainty with differentiable geometric optimization in recurrent architectures. Still, it remains unclear exactly which components are fundamentally responsible for this success. In this paper, we ask: Is the superior performance of deep learning-based systems driven primarily by learned 2D data association, the combination of learned 2D data association and uncertainty, or the recurrent architecture itself? We investigate this question empirically by conducting a controlled study. Our findings reveal that the success of DL-based V-SLAM systems hinges on learned 2D data association and uncertainty rather than their recurrent architecture, underscoring the necessity of learning-based paradigms for the design of these components. Upon acceptance, the code will be released as open source.

Chinese Translation

视觉SLAM是一项在广泛的现实应用中得到充分应用的技术。然而，在低纹理、严重运动模糊和光照不足等具有挑战性的视觉条件下，其性能仍然会下降。基于深度学习的系统通过结合学习的二维数据关联和不确定性与可微几何优化在递归架构中，超越了基于经典几何的方法，并取得了最先进的结果。然而，目前尚不清楚究竟是哪些组件根本上促成了这一成功。本文提出了以下问题：深度学习系统的优越性能主要是由学习的二维数据关联驱动，还是由学习的二维数据关联与不确定性的结合，亦或是递归架构本身？我们通过进行控制研究来实证探讨这一问题。我们的研究结果表明，基于深度学习的视觉SLAM系统的成功依赖于学习的二维数据关联和不确定性，而非其递归架构，这突显了在设计这些组件时学习基础范式的必要性。论文接受后，代码将作为开源发布。

View on arXiv Download PDF AI Translation

cs.CV / 49 / 2607.06083

MSA-DCNN: A Data-Efficient Multi-Scale Deformable CNN for Medical Image Classification

MSA-DCNN：一种数据高效的多尺度可变形卷积神经网络用于医学图像分类

Hussaini, Hamza, Bano, Shahana, Elyan, Eyad, Moreno-García, Carlos Francisco

Abstract

Existing deep learning methods perform well in medical image classification but struggle with multi-scale morphology and limited annotations due to fixed sampling and data-hungry training. Existing approaches address these challenges in isolation: DCN-based models provide adaptive sampling but lack explicit multi-scale attention fusion and label-efficient regularisation; multi-scale architectures typically rely on static fusion; and semi-supervised methods target label scarcity without jointly modelling adaptive cross-scale representations. We propose MSA-DCNN, a scale-consistent deformable attention learning framework that introduces adaptive multi-scale sampling, within-scale saliency refinement, learned cross-scale fusion, and auxiliary self-distillation within a unified optimisation scheme, with potential to generalise to structurally heterogeneous anatomy. We evaluate on three public benchmarks and an external hold-out set for leukaemia. MSA-DCNN demonstrates competitive and often better performance against ViT baselines, CNN baselines, and a MICCAI semi-supervised baseline under distribution shift and label scarcity in accuracy, F1, and AUC (binary), while using fewer parameters. Ablations confirm complementary component contributions, supporting MSA-DCNN as a practical foundation for data-efficient medical image classification.

Chinese Translation

现有的深度学习方法在医学图像分类中表现良好，但由于固定的采样和对数据的高需求训练，面对多尺度形态学和有限标注时表现不佳。现有的方法通常是孤立地解决这些挑战：基于DCN（Deformable Convolutional Networks）模型提供自适应采样，但缺乏明确的多尺度注意力融合和高效的标签正则化；多尺度架构通常依赖于静态融合；半监督方法则针对标签稀缺，但没有联合建模自适应的跨尺度表示。我们提出了MSA-DCNN，这是一种尺度一致的可变形注意力学习框架，引入了自适应多尺度采样、尺度内显著性细化、学习的跨尺度融合，以及在统一优化方案中的辅助自蒸馏，具备推广到结构异质解剖学的潜力。我们在三个公共基准以及针对白血病的外部保留集上进行了评估。MSA-DCNN在面对分布漂移和标签稀缺时，在准确率、F1值和AUC（二分类）上，展示了与ViT（Vision Transformer）基线、CNN基线和MICCAI半监督基线相竞争甚至更好的性能，同时使用更少的参数。消融实验确认了各协同组件的有效贡献，支持MSA-DCNN作为数据高效医学图像分类的实用基础。

View on arXiv Download PDF AI Translation

cs.CV / 50 / 2607.06097

PVCap: Towards Accurate 3D Dense Captioning via PseudoCap and VoxelCapNet

PVCap：通过 PseudoCap 和 VoxelCapNet 实现准确的3D稠密描述

Wu, Xiaopei, Hou, Chenshu, Peng, Liang, Xu, Dan, Lin, Binbin, Huang, Xiaoshui, Hou, Yuenan, Li, Yu, Wang, Wenxiao, Liu, Haifeng, Cai, Deng, Ouyang, Wanli

Abstract

3D dense captioning, an emerging vision-language task, aims to generate descriptive sentences for each object in the 3D scene. Despite the impressive results achieved by previous methods, they suffer from two limitations. First, current research often employs global rigid transformations, such as rotation, to augment scenes without changing their spatial layouts. However, diverse spatial layouts are crucial for training a 3D dense captioning model to describe spatial relations between objects. Second, previous works mainly focus on the design of the caption generation pipeline while utilizing a simple network architecture for other components, i.e., backbone and detection head, which is crucial for extracting rich semantic information for captioning. In this paper, we propose PVCap to alleviate the aforementioned problems. Our PVCap consists of PseudoCap and VoxelCapNet. Specifically, PseudoCap employs a random mixing technique on instances within the dataset, generating numerous pseudo frames with diverse spatial layouts at the instance level. By utilizing a teacher-student framework, PseudoCap obtains pseudo caption labels for these pseudo frames. This data augmentation approach significantly increases the number of training samples and enhances the model's ability to describe the environment effectively. Regarding VoxelCapNet, we introduce a robust caption network that utilizes voxel features and adapts the caption head to the voxel-based network architecture. Our VoxelCapNet can serve as a competitive baseline for future research on 3D dense captioning. Extensive experiments are conducted on two prevalent benchmarks, i.e., ScanRefer and Nr3D. Notably, our method surpasses current state-of-the-art by 11.41% and 13.99% in [email protected], respectively. Codes will be made publicly available.

Chinese Translation

3D稠密描述是一项新兴的视觉-语言任务，旨在为3D场景中的每个物体生成描述性句子。尽管以前的方法取得了显著的结果，但它们存在两个局限性。首先，目前的研究通常采用全局刚性变换，如旋转，以增强场景而不改变其空间布局。然而，多样的空间布局对于训练3D稠密描述模型以描述物体之间的空间关系至关重要。其次，以往的研究主要集中在描述生成管道的设计上，同时对其他组件（即主干网络和检测头）使用简单的网络架构，这对于提取丰富的语义信息以进行描述至关重要。在本文中，我们提出PVCap以缓解上述问题。我们的PVCap由PseudoCap和VoxelCapNet组成。具体而言，PseudoCap在数据集中的实例上采用随机混合技术，生成具有多样空间布局的众多伪帧。通过利用师生框架，PseudoCap为这些伪帧获取伪描述标签。这种数据增强方法显著增加了训练样本的数量，增强了模型有效描述环境的能力。至于VoxelCapNet，我们介绍了一种强大的描述网络，利用体素特征并将描述头适应于基于体素的网络架构。我们的VoxelCapNet可以作为未来3D稠密描述研究的竞争基线。在两个常见基准上进行的大量实验，即ScanRefer和Nr3D中，我们的方法在[email protected]上分别超过了当前最先进的技术11.41%和13.99%。代码将公开发布。

View on arXiv Download PDF AI Translation

cs.CV / 51 / 2607.06105

EcoVision: AI-Powered Drone Imaging for Salt Marsh Vegetation Monitoring and Dominance Mapping

EcoVision：基于人工智能的无人机影像盐沼植被监测与优势种分布映射

Onyenonachi, Innocent, Lawerance, Peter J., Kanwal, Nadia

Abstract

High-resolution RGB imagery acquired from low-altitude UAV surveys was processed through a modular pipeline incorporating transformer-based semantic segmentation, connected-component vegetation extraction, fine-grained species classification using a ConvNeXt architecture, and grid-based dominance scoring at 2x2m resolution. The framework targeted two ecologically significant halophytic grasses, Spartina maritima and Puccinellia maritima, and was trained using a curated and manually annotated UAV imagery, along with biodiversity imagery sourced from publicly accessible datasets. In order to identify these plants from the imagery, our segmentation yielded reliable species masks (mean IoU = 0.56; pixel-level accuracy = 0.96), while object-level classification achieved very good discrimination (F1 = 0.99). Dominance estimates closely matched quadrat-based field surveys, with mean absolute differences below 8%, preserving fine-scale spatial structure under realistic survey conditions. The developed system, named EcoVision, establishes a practical foundation for scalable, high-resolution salt marsh monitoring, demonstrating how AI-driven workflows can translate pixel-level predictions into ecologically interpretable metrics.

Chinese Translation

通过低空无人机调查获取的高分辨率RGB影像，经过一个模块化管道处理，该管道包括基于变换器的语义分割、连通组件植被提取、使用ConvNeXt架构进行的细粒度物种分类，以及2x2米分辨率的基于网格的优势评分。该框架针对两种生态上重要的耐盐植物，即海滩草（Spartina maritima）和海滨碱蓬（Puccinellia maritima），并使用经过精心挑选和人工标注的无人机影像进行训练，同时结合来自公开可获取数据集的生物多样性影像。为了在影像中识别这些植物，我们的分割生成了可靠的物种掩膜（平均交并比 = 0.56；像素级精度 = 0.96），而对象级分类实现了良好的区分（F1 = 0.99）。优势估计与基于样方的实地调查紧密匹配，平均绝对差异低于8%，在现实调查条件下保持了细微的空间结构。所开发的系统EcoVision为可扩展的高分辨率盐沼监测建立了一个实用基础，展示了如何将基于AI的工作流程转换为可生态解释的度量。

View on arXiv Download PDF AI Translation

cs.CV / 52 / 2607.06109

RoME: Robust Mixture of Low-Rank Experts against Multiple Adversarial Perturbations

RoME：针对多重对抗扰动的鲁棒低秩专家混合模型

Kim, Woo Jae, Min, Kyle, Ha, Suhyeon, Jeon, Joonsung, Yoon, Sung-eui

Abstract

Multi-perturbation adversarial training (MAT) aims to achieve robustness against multiple $\ell_p$ perturbations but suffers from robustness trade-offs between different threats. To address this, we employ a mixture of experts (MoE) to route different threats through distinct model pathways. However, naive application of MoE encounters two critical challenges: experts tend to overlook threat-specific features and redundantly capture features shared across threats, and gating networks suffer from threat-agnostic routing where they learn nearly identical routing patterns across threats, thus preventing the construction of threat-specific model pathways. To this end, we propose Robust Mixture of Low-Rank Experts (RoME), where each expert is a low-rank additive update to the shared backbone, allowing it to capture threat-common features while experts focus on threat-specific information. To address threat-agnostic routing, RoME introduces (i) dual-scale gating that exploits threat-discriminative signals from local and global level features, and (ii) threat-guided gating diversification that enforces diverse expert utilization across threats. Extensive experiments demonstrate that RoME outperforms existing state-of-the-art MAT in union robustness and natural accuracy and improves robustness against unseen threats. Codes are available at https://github.com/wkim97/RoME.

Chinese Translation

多扰动对抗训练（MAT）旨在实现对多种 $ ext{ℓ}_p$ 扰动的鲁棒性，但在不同威胁之间存在鲁棒性权衡。为了解决这个问题，我们采用专家混合模型（MoE）将不同威胁通过不同的模型路径进行路由。然而，简单应用 MoE 面临两个关键挑战：专家往往忽视特定威胁的特征，并且冗余地捕捉跨威胁共享的特征；而门控网络则遭遇与威胁无关的路由问题，导致它们在不同威胁之间学习几乎相同的路由模式，从而阻碍了特定威胁模型路径的构建。为此，我们提出了鲁棒低秩专家混合模型（RoME），其中每个专家都是对共享主干的低秩加性更新，使其能够捕捉威胁共通特征，同时专家专注于特定威胁的信息。为了解决与威胁无关的路由问题，RoME 引入了（i）双尺度门控，利用来自局部和全局特征的威胁区分信号，以及（ii）威胁引导的门控多样化，强制在不同威胁之间实现多样化的专家利用。大量实验表明，RoME 在联合鲁棒性和自然准确性方面优于现有的最先进的 MAT，并提高了对未见威胁的鲁棒性。代码可在 https://github.com/wkim97/RoME 获取。

View on arXiv Download PDF AI Translation

cs.CV / 53 / 2607.06118

WebRetriever: A Large-Scale Comprehensive Benchmark for Efficient Web Agent Evaluation

WebRetriever：高效网络代理评估的大规模综合基准

Dong, Wei, Fu, Tianyu, Yu, Zhe, Wang, Hanning, Su, Anyang, Fang, Zhizhou, Chen, Yuyang, Wang, Shuo, Wu, Minghui, Jiang, Ping, Lei, Zhen, Zhao, Chenxu

Abstract

As web agents increasingly demonstrate capabilities in automated task execution, the development of robust evaluation frameworks for assessing their navigation and task completion performance has emerged as a critical research priority. However, existing benchmarks exhibit fundamental limitations. First, they suffer from insufficient scale and limited domain diversity, constraining comprehensive evaluation of cross-domain generalization. Second, prevailing LLM-as-Judge evaluation methodologies inadequately capture fine-grained interaction semantics, particularly regarding precise query formulation and filtering operations. Third, current benchmarks predominantly emphasize navigation success metrics while neglecting critical requirements for real-world deployment scenarios. To address these limitations, we introduce WebRetriever, a large-scale benchmark encompassing 800 websites and 1,550 tasks across diverse domains, including consumer, professional, and enterprise sectors, with comprehensive coverage of user intent patterns. We propose NavEval (Navigation Evaluation), a novel LLM-as-Judge framework that leverages rich interaction context beyond visual screenshots, achieving state-of-the-art alignment with human judgment across multiple evaluation datasets. Furthermore, we establish three complementary evaluation protocols that collectively provide holistic assessment of web agent capabilities: navigation proficiency, knowledge-assisted interaction, and end-to-end task completion with information extraction. Extensive experimental analysis reveals substantial performance disparities across evaluation protocols, demonstrating that navigation success alone is an insufficient predictor of real-world application effectiveness. WebRetriever delivers fine-grained diagnostic insights into agent capabilities and establishes a rigorous foundation for advancing web agent research and development.

Chinese Translation

随着网络代理在自动化任务执行中的能力逐渐显现，开发稳健的评估框架以评估其导航与任务完成表现已成为一个关键研究优先事项。然而，现有基准存在根本性局限。首先，它们规模不足且领域多样性有限，限制了跨领域泛化的全面评估。其次，现有的以大型语言模型（LLM）为评判标准的评估方法无法充分捕捉细粒度的交互语义，尤其是在精准查询形成和过滤操作方面。此外，目前的基准主要强调导航成功指标，而忽视了真实世界部署场景的关键需求。为了解决这些问题，我们提出了WebRetriever，这是一个涵盖800个网站和1,550个任务的大规模基准，涵盖消费、专业和企业等多个领域，并全面覆盖用户意图模式。我们提出了NavEval（导航评估），一种新颖的以LLM为评判标准的框架，利用丰富的交互上下文超越视觉截屏，在多个评估数据集上实现了与人类判断的最优对齐。此外，我们建立了三种互补的评估协议，全面评估网络代理的能力：导航熟练度、知识辅助交互和信息提取的端到端任务完成。广泛的实验分析显示不同评估协议之间存在显著的性能差异，证明单靠导航成功并不足以预测真实应用的有效性。WebRetriever 提供了对代理能力的细致诊断洞察，为推进网络代理的研究与开发建立了严格的基础。

View on arXiv Download PDF AI Translation

cs.CV / 54 / 2607.06120

AEGIS: A Mechanism-Guided Defense against Visual Synonym Jailbreaks in Text-to-Image Models

AEGIS：一种机制引导的防御方法，针对文本到图像模型中的视觉同义词越狱攻击

Huang, Yuanmin, Zhang, Zhenfei, Zhang, Mi, Hong, Geng, He, Qinqin, Tao, Jialing, Xue, Hui, Yang, Min

Abstract

Text-to-image diffusion models have achieved high visual fidelity and broad adoption, but remain vulnerable to safety violations when adversaries exploit them to synthesize illicit content. Existing alignment paradigms, from input sanitization to structural feature pruning, are largely organized around unsafe concepts explicitly exposed during filtering, editing, or localization. This leaves a blind spot for visual synonym attacks (VSA), a jailbreak where benign-looking prompts elicit prohibited imagery through implicit visual associations. As a result, current defenses face a safety-utility dilemma: they may either under-mitigate VSA threats or over-suppress visually similar benign concepts. The core challenge is that VSA hides the unsafe target at the textual surface while revealing it through generation-time visual-semantic convergence. In this work, we therefore shift from static suppression of pre-specified unsafe concepts to dynamic tracing of how unsafe semantics emerge during generation. Our mechanistic analysis shows that VSA and explicit unsafe prompts converge through sparse semantic-injecting attention heads, which serve as inference-time bottlenecks for prohibited visual semantics. Based on this insight, we propose AEGIS (Adaptive Evasion Guard via Identification and Steering), an inference-time defense that applies similarity-aware repulsion only at the identified vulnerable heads. Evaluated against 16 baselines, AEGIS improves both safety and utility. On SD 1.4, it reduces ASR to $\mathbf{0.00}/\mathbf{0.03}$ for in-domain violence/nudity VSA and achieves ASRs $\le \mathbf{0.09}$ on out-of-domain explicit and adversarial attacks. It preserves benign fidelity, avoids suppressing hard-negative concepts, and transfers to SD 2.1 and FLUX.1 after re-identifying the critical heads for each backbone.

Chinese Translation

文本到图像的扩散模型已实现高视觉保真度并广泛应用，但在对手利用它们合成非法内容时，仍然容易受到安全违规的攻击。现有的对齐范式，从输入清理到结构特征修剪，主要围绕在过滤、编辑或定位过程中显式暴露的不安全概念进行组织。这为视觉同义词攻击（VSA）留下了盲点，这是一种越狱攻击，其中看似无害的提示通过隐含的视觉关联引发被禁止的图像。因此，当前的防御面临安全性与效用的困境：它们可能要么未能充分减轻VSA威胁，要么过度抑制视觉上相似的无害概念。核心挑战在于，VSA在文本表面隐藏了不安全的目标，同时通过生成时的视觉-语义收敛揭示了它。在本研究中，我们因此从对预先指定的不安全概念的静态抑制转向动态追踪不安全语义在生成过程中如何出现。我们的机制分析表明，VSA和显式不安全提示通过稀疏的语义注入注意力头汇聚，这些注意力头在推理时成为被禁止视觉语义的瓶颈。基于这一洞察，我们提出了AEGIS（通过识别和引导的自适应规避防护），这是一种推理时防御方法，仅在识别出的脆弱头部应用相似性感知的排斥。与16个基线进行评估，AEGIS在安全性和效用上均有所提升。在SD 1.4上，它将领域内暴力/裸体VSA的ASR降低至$ extbf{0.00}/ extbf{0.03}$，并在领域外显式和对抗攻击中实现ASR $ extless extbf{0.09}$。它保持了无害的保真度，避免抑制难负概念，并在重新识别每个主干的关键头部后，成功转移至SD 2.1和FLUX.1。

View on arXiv Download PDF AI Translation

cs.CV / 55 / 2607.06136

Tuning-Free Latent Diffusion Models for Ultrahigh-Resolution Image Editing

无调优超高分辨率图像编辑的潜在扩散模型

Lu, Wanglong, Su, Lingming, Shi, Kaijie, Gong, Minglun, Jin, Xiaogang, Zhao, Hanli, Jiang, Xianta

Abstract

Recent diffusion-based generative models have shown impressive performance in image generation and editing. However, due to memory limitations and the high cost of collecting high-resolution training images, existing methods are typically restricted to inputs with linear resolutions below 1K. In contrast, photos captured by modern mobile devices often reach linear resolutions up to 8K, revealing a significant gap between current capabilities and real-world demands. Simply upscaling low-resolution edited results often results in visually enlarged but blurry images that lack fine details. This paper introduces UltraDiffEdit, a novel, tuning-free image editing framework that extends off-the-shelf latent diffusion models (LDMs) to ultrahigh resolutions. UltraDiffEdit employs a multi-scale progressive editing strategy, iteratively blending high-resolution edited content with unedited areas in a coarse-to-fine manner. We employ multi-patch encoding to preserve both edited and unedited visual details within the latent space. To mitigate editing artifacts, our global-local consistency denoising technique consistently integrates edited and unedited latent features, ensuring smooth transition at editing boundaries from the latent representation to the final image. We also introduce a patch-based hybrid sampling approach that captures local, intermediate, and global features, ensuring semantic coherence and enhancing fine detail during denoising. We conduct extensive experiments demonstrating UltraDiffEdit's superior editing quality and flexibility: it can handle image resolutions up to 8K using only a single NVIDIA GeForce RTX 3090 GPU. The source code is publicly available at https://github.com/LonglongaaaGo/UltraDiffEdit.

Chinese Translation

最近，基于扩散的生成模型在图像生成和编辑方面表现出了令人印象深刻的性能。然而，由于内存限制和收集高分辨率训练图像的高成本，现有方法通常限制在低于1K线性分辨率的输入。相比之下，现代移动设备拍摄的照片线性分辨率往往高达8K，揭示了当前能力与实际需求之间的显著差距。简单地将低分辨率编辑结果放大通常会导致视觉上放大但模糊的图像，缺乏细节。本文提出了UltraDiffEdit，一种新颖的无调优图像编辑框架，能够将现成的潜在扩散模型( latent diffusion models, LDMs) 扩展到超高分辨率。UltraDiffEdit采用多尺度渐进编辑策略，以粗到细的方式迭代地将高分辨率编辑内容与未编辑区域融合。我们使用多补丁编码技术在潜在空间中保留编辑和未编辑的视觉细节。为了减少编辑伪影，我们的全局-局部一致性去噪技术始终将编辑和未编辑的潜在特征整合，以确保在潜在表示到最终图像的编辑边界处平滑过渡。我们还引入了一种基于补丁的混合采样方法，捕捉局部、中间和全局特征，确保语义一致性并在去噪过程中增强细节。我们进行了广泛的实验，展示了UltraDiffEdit卓越的编辑质量和灵活性：它能够仅使用单个NVIDIA GeForce RTX 3090 GPU 处理高达8K的图像分辨率。源代码已公开，链接为 https://github.com/LonglongaaaGo/UltraDiffEdit。

View on arXiv Download PDF AI Translation

cs.CV / 56 / 2607.06148

RFHNet: Relational and Frequency-Aware Hashing Network for Large-Scale Fine-Grained Food Image Retrieval

RFHNet：一种关系与频率感知的哈希网络用于大规模细粒度食品图像检索

Wang, Junsong, Min, Weiqing, Sheng, Guorui, Yao, Tao, Wang, Lili, Jiang, Shuqiang

Abstract

Fine-grained food image retrieval is a key task in computational gastronomy, with applications in food traceability, dietary monitoring, and smart catering systems. Although hashing-based retrieval is attractive for large-scale search due to its storage efficiency and fast Hamming-distance computation, existing methods often perform poorly in fine-grained food scenarios, where subtle local semantics and frequency-sensitive visual cues are essential. To address this challenge, we propose RFHNet, a cascaded hierarchical hashing network that captures both global structure and fine-grained local details through multi-level representations. RFHNet includes three components: (1) Fine-grained Relation Modeling (FRM) to capture subtle visual differences among similar food components; (2) Multi-Frequency Modulated Fusion (MFMF) to extract informative multi-frequency features; and (3) Hierarchical Semantic Synergy (HSS) to adaptively integrate multi-level representations and generate discriminative hash codes. Experiments on six food-specific benchmarks show that RFHNet consistently outperforms state-of-the-art hashing methods, with mAP gains of 4.44\% to 17.20\% at 12 bits. These results validate the effectiveness of RFHNet for large-scale visual food retrieval and smart catering applications. The source code will be released upon publication.

Chinese Translation

细粒度食品图像检索是计算美食学中的一项关键任务，广泛应用于食品追溯、饮食监测和智能餐饮系统。尽管基于哈希的检索方法因其存储效率和快速的汉明距离计算而在大规模搜索中具有吸引力，但现有方法在细粒度食品场景中往往表现不佳，因为在这些场景中，细微的局部语义和对频率敏感的视觉线索至关重要。为了解决这一挑战，我们提出了RFHNet，一种级联层次哈希网络，通过多层表示捕获全局结构和细粒度局部细节。RFHNet包括三个组件：（1）细粒度关系建模（FRM），用于捕捉相似食品成分之间的细微视觉差异；（2）多频率调制融合（MFMF），用于提取信息丰富的多频率特征；（3）层次语义协同（HSS），用于自适应整合多层表示并生成区分性哈希码。在六个食品特定基准上的实验表明，RFHNet始终优于最先进的哈希方法，在12位时mAP提升范围为4.44\%至17.20\%。这些结果验证了RFHNet在大规模视觉食品检索和智能餐饮应用中的有效性。源代码将在发表后发布。

View on arXiv Download PDF AI Translation

cs.CV / 57 / 2607.06150

Enhanced Seam Segmentation for Automated Welding Robot in Construction Through Transfer Learning: Addressing Limitations of Bilateral Segmentation Network

通过迁移学习增强建筑自动焊接机器人接缝分割：解决双边分割网络的局限性

Park, Keonvin, Voeurn, Yong Ann, Kweon, Hyeokjun, Lee, Doyun

Abstract

Reliable seam segmentation is essential for autonomous robotic welding in construction, where harsh illumination, specular reflections, and thin weld geometries often degrade segmentation performance. This study proposes a reflection-robust seam segmentation framework that enhances a BiSeNetV2 backbone through transfer learning and a hybrid Cross-Entropy--Lov\'asz loss. Rather than increasing architectural complexity, the proposed framework improves reflection robustness through learning-stability-oriented optimization. Experimental results show that the proposed method achieves 81.76\% Joint IoU and 90.73\% mIoU, improving Joint IoU by +22.36 percentage points over the OHEM-based baseline while maintaining identical FLOPs, parameter count, and inference speed. The proposed approach also recovers 96.33\% of severe zero-IoU failure cases under reflective conditions. Comparative experiments across BiSeNetV2, DeepLabV3+, UNet, and SegFormer further demonstrate that the proposed optimization strategy is particularly effective for lightweight real-time segmentation architectures. Qualitative analyses additionally show improved seam continuity and reflection robustness in challenging welding environments. These findings suggest that the proposed framework provides a practical and lightweight perception solution for robotic welding applications involving reflective metallic surfaces.

Chinese Translation

可靠的接缝分割对建筑中的自主机器人焊接至关重要，因为恶劣的照明、镜面反射和薄焊缝几何形状常常降低分割性能。本研究提出了一种抗反射接缝分割框架，通过迁移学习和混合交叉熵-洛瓦茨（Cross-Entropy--Lovász）损失增强BiSeNetV2主干。该框架并未增加架构复杂性，而是通过面向学习稳定性的优化提高了抗反射能力。实验结果表明，所提方法实现了81.76%的联合IoU和90.73%的平均IoU，相较于基于OHEM的基线，联合IoU提高了22.36个百分点，同时保持相同的FLOPs、参数数量和推理速度。所提方法在反射条件下还恢复了96.33%的严重零IoU失败案例。在BiSeNetV2、DeepLabV3+、UNet和SegFormer等模型的比较实验中，进一步证明了所提出的优化策略对轻量级实时分割架构特别有效。定性分析还显示，在具有挑战性的焊接环境中提高了接缝的连续性和抗反射能力。这些发现表明，所提框架为涉及反射金属表面的机器人焊接应用提供了一种实用且轻量的感知解决方案。

View on arXiv Download PDF AI Translation

cs.CV / 58 / 2607.06162

High-Resolution Artwork Outpainting with Global Blueprint Guidance and Layout Control

基于全球蓝图指导和布局控制的高分辨率艺术作品外延

Kim, Junha, Park, Hyunjoon, Cho, Donghyeon

Abstract

Image outpainting extends an image beyond its original borders, requiring seamless style integration and globally coherent scene completion. Building on the success of diffusion models, recent methods have achieved substantial improvements in visual quality. In practice, however, high-resolution outpainting is commonly performed via progressive expansion around a fixed source image, particularly in artwork scenarios. Despite this progress, existing approaches still suffer from three key limitations: (i) the absence of a reliable global planning mechanism, which leads to structural instability and error accumulation at high resolutions; (ii) limited spatial controllability beyond text prompts, making it difficult to place objects at user-specified locations; and (iii) high inference latency caused by inherently sequential patch generation. To address these issues, we propose a global blueprint-guided two-stage diffusion framework for layout-controllable high-resolution outpainting with efficient parallel synthesis. In Stage 1, we generate a low-resolution global blueprint using a layout adapter that injects bounding-box conditions into a Stable Diffusion inpainting backbone, producing a globally consistent structural plan while extracting global guidance features. In Stage 2, we synthesize high-resolution local patches in parallel by injecting the blueprint-derived global guidance and initializing each patch from the blueprint using the low-frequency preservation property of forward diffusion. This design eliminates sequential dependency while maintaining global coherence. Extensive experiments on large-scale artwork datasets demonstrate improved visual fidelity, stronger semantic consistency, and substantially reduced inference time compared to prior baselines, while uniquely supporting explicit layout control for artwork outpainting.

Chinese Translation

图像外延是指将图像扩展超出其原始边界，这要求风格无缝融合和全球一致的场景完成。在扩散模型成功的基础上，最近的方法在视觉质量上取得了显著进步。然而，在实际应用中，高分辨率外延通常是在固定源图像周围进行渐进式扩展，特别是在艺术作品的场景中。尽管取得了一些进展，但现有方法仍然存在三个主要限制：(i) 缺乏可靠的全球规划机制，导致高分辨率下的结构不稳定和错误积累；(ii) 受限于文本提示之外的空间可控性，难以将物体放置在用户指定的位置；以及 (iii) 由于固有的顺序补丁生成导致的高推断延迟。为了解决这些问题，我们提出了一种全球蓝图指导的两阶段扩散框架，用于布局可控的高分辨率外延，并实现高效的并行合成。在第一阶段，我们通过布局适配器生成低分辨率全球蓝图，将边界框条件注入到稳定的扩散（Stable Diffusion）修复主干网络中，从而生产出全球一致的结构规划，并提取全球指导特征。在第二阶段，我们通过注入蓝图衍生的全球指导，在并行中合成高分辨率局部补丁，并利用前向扩散的低频保留特性从蓝图初始化每个补丁。该设计消除了顺序依赖性，同时保持了全球一致性。在大规模艺术作品数据集上的广泛实验表明，与之前的基准相比，视觉保真度更高，语义一致性更强，推断时间显著减少，同时独特地支持对艺术作品外延的显式布局控制。

View on arXiv Download PDF AI Translation

cs.CV / 59 / 2607.06173

MobileWan: Closing the Quality Gap for Mobile Video Diffusion

MobileWan：缩小移动视频扩散的质量差距

Ghafoorian, Mohsen, Korzhenkov, Denis, Karjauv, Adil, Lelekas, Ioannis, Fathima, Noor, Stasis, Spyridon, Ackermann, Hanno, van Breugel, Boris, Nagel, Markus, Porikli, Fatih, Karnewar, Animesh, Habibian, Amirhossein

Abstract

Recent advances in video diffusion have been driven by scaling transformer-based architectures to billions of parameters, substantially improving visual fidelity and motion coherence. In contrast, existing mobile video diffusion models remain limited to relatively small parameter budgets, typically 0.4-1.8B, restricting generation quality. In this work, we show that high-quality mobile video generation does not require small models. Instead, we demonstrate that a server-scale 5B-parameter video diffusion transformer can be deployed efficiently on memory-constrained mobile hardware through recurrent reformulation and structured compression. Starting from Wan2.2-5B, we rely on a recurrence distillation framework that converts video generation into a chunk-wise autoregressive process with constant-memory attention computation. Combined with causal linear attention, the model operates as an RNN at inference time while preserving temporal coherence across chunks. We further propose a learnable attention head pruning method based on binary per-head gates optimized end-to-end using a noise-biased sparsity objective and distillation-based finetuning. Together with sampling-step distillation and memory-optimized VAE decoding, MobileWan becomes the first 5B-scale video diffusion model deployable on a commercial mobile device. Our system generates 5-second 480x832 videos at 16 FPS in 20 seconds end-to-end latency, achieving a VBench score of 83.79 and establishing a new state of the art in mobile video generation. Project page: https://qualcomm-ai-research.github.io/mobilewan

Chinese Translation

近年来，视频扩散的进展得益于将基于变换器的架构扩展到数十亿参数，从而显著提高了视觉逼真度和运动一致性。相比之下，现有的移动视频扩散模型仍然受到相对较小的参数预算的限制，通常在0.4-1.8B之间，限制了生成质量。在本研究中，我们展示了高质量的移动视频生成并不依赖于小模型。相反，我们证明了一个拥有5B参数的服务器规模视频扩散变换器可以通过重复重构和结构化压缩，在内存受限的移动硬件上高效部署。从Wan2.2-5B开始，我们依赖于一种递归蒸馏框架，将视频生成转化为一个基于块的自回归过程，并保持恒定内存的注意力计算。结合因果线性注意力，该模型在推理时作为一个递归神经网络（RNN）运行，同时保持跨块的时间一致性。我们进一步提出了一种基于二进制每头门控的可学习注意力头裁剪方法，通过端到端优化，利用噪声偏置稀疏目标和基于蒸馏的微调来实现。结合采样步骤蒸馏和内存优化的变分自编码器（VAE）解码，MobileWan成为第一个能够在商业移动设备上部署的5B规模视频扩散模型。我们的系统能够在20秒的端到端延迟下生成480x832像素、长度为5秒的16帧每秒（FPS）视频，VBench得分达到83.79，确立了移动视频生成的新状态。项目页面：https://qualcomm-ai-research.github.io/mobilewan

View on arXiv Download PDF AI Translation

cs.CV / 60 / 2607.06176

Revisiting Scene Graph Generation from the Perspective of Detector-Conditioned Reachability

从检测器条件可达性的角度重新审视场景图生成

Qu, Runfeng, Bideau, Pia K, Hall, Ole, Ouerfelli-Ethier, Julie, Obermayer, Klaus, Hellwich, Olaf

Abstract

Scene graph generation (SGG) approaches can be broadly classified into detector-based and query-based methods according to their underlying reasoning mechanisms. However, the discrepancy in their predictive behaviors, induced by these distinct mechanisms, has not been systematically analyzed. In this work, we design a controlled experimental setup to examine prediction discrepancies from the perspective of detector-conditioned reachability. The results suggest clear complementary clues. Motivated by this observation, we introduce a Dual-SGG method that consolidates both reasoning mechanisms via a dual-query design, thereby leveraging the complementary predictive behaviors of both detector-based and query-based methods. Extensive experiments on the Visual Genome, Open Images v6, and GQA-200 datasets demonstrate the effectiveness of the proposed method.

Chinese Translation

场景图生成（SGG）方法可以根据其基本推理机制大致分为基于检测器的方法和基于查询的方法。然而，由于这些不同的机制引起的预测行为差异尚未得到系统分析。在本研究中，我们设计了一个受控实验设置，从检测器条件可达性的角度来检查预测差异。结果表明存在明显的互补线索。基于这一观察，我们提出了一种双重场景图生成方法（Dual-SGG），通过双查询设计整合了这两种推理机制，从而利用了基于检测器和基于查询方法的互补预测行为。对Visual Genome、Open Images v6和GQA-200数据集的广泛实验表明了所提方法的有效性。

View on arXiv Download PDF AI Translation

cs.CV / 61 / 2607.06185

Structured-Condensed Prompt Tuning in Vision-Language Models for Fine-grained Image Recognition

视觉语言模型中的结构化简约提示调优用于细粒度图像识别

Liu, Xinda, Zhang, Qinyu, Min, Weiqing, Geng, Guohua, Jiang, Shuqiang

Abstract

Fine-grained image recognition poses a significant challenge due to the substantial expertise and effort required for manual annotation. Vision-language models (VLMs) like CLIP provide a compelling zero-shot alternative, reducing reliance on extensive labeled data. However, their ability to capture subtle distinctions remains limited, leading to subpar recognition performance. While prompt tuning has proven effective for adapting VLMs, most existing methods treat class labels as isolated, discrete entities, overlooking the rich semantic relationships between them. This oversimplified assumption limits the model's ability to capture hierarchical dependencies and inter-class correlations -- both critical for distinguishing visually similar categories. The problem is especially acute in fine-grained classification, where accurate recognition depends on understanding complex label semantics. To address this, we propose Structured-Condensed Prompt Tuning (SCPT), which enhances semantic structure modeling in prompt learning. Specifically, we introduce Semantic Relation Encoding (SRE) to explicitly model inter-class semantic topology and encode structured label relationships. In parallel, we design a Semantic Condensation loss (ScLoss) to suppress redundant supervision and extract discriminative components from the global semantic space. Together, these components significantly improve semantic alignment and fine-grained discrimination. Extensive experiments on 14 fine-grained benchmarks show that SCPT effectively mitigates semantic ambiguity and achieves state-of-the-art performance in both few-shot and base-to-novel generalization settings.

Chinese Translation

细粒度图像识别面临着重大挑战，因为手动标注需要大量的专业知识和努力。像CLIP这样的视觉语言模型（VLMs）提供了一个引人注目的零样本替代方案，减少了对广泛标注数据的依赖。然而，它们捕捉微妙区别的能力仍然有限，导致识别性能不佳。尽管提示调优已证明对适应VLMs有效，但大多数现有方法将类别标签视为孤立的离散实体，忽略了它们之间丰富的语义关系。这种简化的假设限制了模型捕捉层次依赖性和类间关联性的能力，而这两者对于区分视觉相似类别至关重要。这个问题在细粒度分类中尤为突出，因为准确的识别依赖于理解复杂的标签语义。为了解决这个问题，我们提出了结构化简约提示调优（SCPT），它增强了提示学习中的语义结构建模。具体来说，我们引入了语义关系编码（SRE），以明确建模类间的语义拓扑并编码结构化的标签关系。同时，我们设计了一种语义凝聚损失（ScLoss），以抑制冗余监督并从全局语义空间中提取判别组件。这些组件共同显著提高了语义对齐和细粒度区分。在14个细粒度基准上的广泛实验表明，SCPT有效减轻了语义模糊，并在少样本和基于新类别的泛化设置中实现了最先进的性能。

View on arXiv Download PDF AI Translation

cs.CV / 62 / 2607.06216

MoWorld: A Flash World Model

MoWorld：一种闪存世界模型

Moxin, Team, Ji, Deyi, Chen, Tianrun, Zhang, Xin, Yang, Jiale, Zhu, Qi, Zhao, An, Xie, Zihao, Wang, Han, Liu, Xuanyi, Zhou, Yixiang, Liu, Pei, Tan, Yi, Chen, Cheng, Zhu, Dayi, Wei, Mingyu, Xu, Hanjie, Liao, Jun, Li, Siqi, Lu, Lingyu, Fang, Hongye, Tan, Hongming, Zhu, Youjiang, Zhang, Taiyu, Li, Zejian, Ding, Chaotao, Zhu, Lanyun, Pan, Yunhe, Sun, Lingyun

Abstract

The future of World Models depends not only on scaling model capability, but also on scaling practicality and inference efficiency. High-frame-rate inference enables responsive perception, planning, and control in real-world autonomous systems. To this end, we present MoWorld, a cost-effective yet high-performance Flash World Model with an end-to-end framework spanning data generation, pre-training, distillation, and efficient inference, enabling up to 50 FPS real-time interaction with cinematic visual quality without the need of high-end GPUs. To enable large-scale real-world deployment, MoWorld jointly optimizes model capability and cost throughout the entire development pipeline. Specifically, unlike existing approaches that primarily rely on large-scale video corpora, MoWorld is built upon a scalable 3D-native data engine accumulated from our large-scale 3D vision and generative modeling pipeline, enabling the efficient construction of geometrically consistent training data across diverse real-world and synthetic environments. Based on this foundation, a curriculum cross-frame pre-training strategy for stable and scalable World Model learning, an efficient denoising-step distillation algorithm to reduce diffusion training cost, and a mixed-precision parallel inference framework for low-cost real-time deployment. MoWorld is the first real-time interactive World Model built on the Neural Processing Unit (NPU) and can achieves up to 50 FPS in such the devices, enabling practical and efficient deployment at scale. Comprehensive evaluations demonstrate that MoWorld achieves leading performance; notably, its average inference cost is only 30\%-50\% of that of existing World Models, providing a practical foundation for large-scale real-world applications of World Models. We also demonstrate diverse applications of MoWorld.

Chinese Translation

世界模型的未来不仅取决于模型能力的扩展，还依赖于实用性和推理效率的提升。高帧率推理能够在真实世界中的自主系统中实现响应式感知、规划和控制。为此，我们提出了MoWorld，一种具有成本效益且高性能的闪存世界模型，其端到端框架涵盖数据生成、预训练、蒸馏和高效推理，能够在不依赖高端GPU的情况下以高达50帧每秒的速度，实现与电影视觉质量的实时交互。为了实现大规模的现实世界部署，MoWorld在整个开发流程中联合优化模型能力和成本。具体而言，与主要依赖大规模视频语料库的现有方法不同，MoWorld建立在可扩展的3D原生数据引擎之上，该引擎来源于我们的大规模3D视觉和生成建模流程，从而高效地构建几何一致的训练数据，涵盖多样的现实世界和合成环境。在此基础上，我们提出了一种课程交叉帧预训练策略，以实现稳定和可扩展的世界模型学习，以及一种高效的去噪步骤蒸馏算法，以降低扩散训练成本，并设计了一种混合精度并行推理框架，以实现低成本的实时部署。MoWorld是首个基于神经处理单元（NPU）的实时交互世界模型，能够在此类设备上达到最高50帧每秒，实现大规模的实用和高效部署。全面的评估结果表明，MoWorld在性能上处于领先地位；值得注意的是，其平均推理成本仅为现有世界模型的30%-50%，为世界模型的大规模现实应用提供了实用基础。我们还展示了MoWorld的多样化应用。

View on arXiv Download PDF AI Translation

cs.CV / 63 / 2607.06217

EeveeDark: A Binary Neural Framework for Low-Light Video Enhancement via Event-Guided Sensor-Level Fusion

EeveeDark：一种基于事件引导的传感器级融合的低光视频增强二元神经框架

Eker, Onur, Erdem, Erkut, Erdem, Aykut

Abstract

Enhancing videos under extreme low-light conditions remains challenging due to the difficulty of balancing restoration quality and computational efficiency in resource-constrained settings. This paper introduces EeveeDark, a low-light video enhancement framework that combines the spatial richness of sensor-level RAW data with the temporal precision of event streams. Central to our model is a Binary Neural Network (BNN) architecture that reduces computational overhead by quantizing weights and activations while preserving detail. EeveeDark incorporates (i) modality-specific binary encoders for processing RAW frames and event data, (ii) a lightweight fusion block for integrating spatial and temporal cues, and (iii) an event-guided skip gating mechanism for dynamic spatiotemporal refinement. Experiments on synthetic and real-world datasets show that EeveeDark outperforms prior BNN-based methods and offers a favorable performance-efficiency trade-off compared to full-precision models. The project page is available at https://cyberiada.github.io/EeveeDark.

Chinese Translation

在极低光照条件下增强视频仍然具有挑战性，因为在资源受限的环境中平衡恢复质量和计算效率非常困难。本文介绍了EeveeDark，这是一种低光视频增强框架，结合了传感器级RAW数据的空间丰富性和事件流的时间精确性。我们模型的核心是一个二元神经网络（Binary Neural Network，BNN）架构，通过量化权重和激活来减少计算开销，同时保留细节。EeveeDark包含（i）用于处理RAW帧和事件数据的特定模态二元编码器，（ii）用于整合空间和时间线索的轻量级融合模块，以及（iii）用于动态时空细化的事件引导跳过门控机制。在合成和真实世界数据集上的实验表明，EeveeDark的性能优于先前的基于BNN的方法，并且与全精度模型相比，提供了良好的性能效率权衡。项目页面可访问 https://cyberiada.github.io/EeveeDark。

View on arXiv Download PDF AI Translation

cs.CV / 64 / 2607.06234

WING: A Window-Prior-Based Generative Network with Gated Inception for Cross-Modality CT Synthesis

WING：基于窗口先验的生成网络与门控 inception 用于跨模态 CT 合成

Mei, Siyuan, Xia, Yan, Sun, Yipeng, Bayer, Siming, Li, Zirong, Ye, Chengze, Liu, Daiqi, Fan, Fuxin, Huang, Yixing, Maier, Andreas

Abstract

Generating CT volumes from MRI and CBCT can improve treatment planning in adaptive radiotherapy while avoiding additional radiation exposure. However, direct regression of CT intensities is challenged by the inherently high dynamic range and long-tailed distributions, thereby averaging out sparse yet clinically important structures. To alleviate this issue, we reformulate the regression target into multiple windowed representations, leveraging the inductive prior that CT intensities are structure-deterministic and window-separable. These windowed views exhibit smoother distributions and admit structured fusion back to the full-range CT. Building on this reformulation, we introduce WING, a WINdow-prior-based Generative network comprising: 1) a new Gated Inception Generator to produce multi-window predictions, enabling multi-shape kernel interactions to capture cross-modality correspondence; 2) a Fuse-and-Refine Transformer to aggregate the windowed outputs and learn residuals for detail refinement; and 3) a joint adversarial training objective to enhance window-conditioned realism. Extensive experiments demonstrate that our compact WING achieves state-of-the-art performance on the MRI-to-CT and CBCT-to-CT benchmarks, while supporting multi-anatomy synthesis with a single model.

Chinese Translation

从 MRI 和 CBCT 生成 CT 体积可以改善适应性放疗中的治疗计划，同时避免额外的辐射暴露。然而，CT 强度的直接回归受到固有的高动态范围和长尾分布的挑战，从而平均化稀疏但临床重要的结构。为了解决这个问题，我们将回归目标重新构造为多个窗口表示，利用 CT 强度是结构决定论且窗口可分的归纳先验。这些窗口视图展现了更平滑的分布，并允许将结构融合回完整范围的 CT。在这个重新构造的基础上，我们提出了 WING，一个基于窗口先验的生成网络，包括：1) 一种新的门控 Inception 生成器，用于产生多窗口预测，实现多形状核交互以捕捉跨模态对应关系；2) 一种融合与精细化变换器，用于聚合窗口输出并学习残差以细化细节；以及 3) 一种联合对抗训练目标，以增强窗口条件下的真实感。广泛的实验表明，我们紧凑的 WING 在 MRI 到 CT 以及 CBCT 到 CT 的基准测试中实现了最先进的性能，同时支持单模型的多解剖合成。

View on arXiv Download PDF AI Translation

cs.CV / 65 / 2607.06238

PhyMRI-SR: Toward Physics-Aware MRI Image Super-Resolution

PhyMRI-SR：迈向物理感知的MRI图像超分辨率

Wei, Lihua, Gao, Huatong, Gong, Jia, Tan, Zhiyu, Li, Hao, Liu, Jun, Ren, Zhihua

Abstract

Magnetic resonance imaging (MRI) super-resolution is vital for improving diagnostic accessibility, yet most methods treat it as a deterministic mapping from a fixed low-resolution input to a high-resolution target. This overlooks a key property of MRI acquisition physics: spatial resolution and signal-to-noise ratio (SNR) are inherently coupled, making any given low-resolution scan merely one of many possible realizations under varying acquisition trade-offs. We rethink MRI super-resolution as a physics-aware reconstruction problem, in which the goal is to identify the optimal resolution-SNR configuration and then super-resolve it to obtain high-quality MRI results. A key implication of this formulation is that MRI resolution becomes dynamic rather than fixed. To handle such resolution-heterogeneous inputs, we adapt 2D Gaussian Splatting (2D GS) to MRI by formulating reconstruction as a coordinate-based, resolution-agnostic rendering problem. To further enhance fidelity, we introduce three innovations: (1) a prior-aware Gaussian representation that combines an Anatomical Structure Prior for tissue-specific kernel initialization with an Imaging System Prior that captures hardware characteristics via a covariance dictionary; (2) a physics-constrained signal modeling scheme that predicts intrinsic tissue parameters (proton density rho and effective relaxation rate R2) and synthesizes intensities through governing physical equations, ensuring biophysically plausible contrast; and (3) a meta-learning framework that alleviates paired-data scarcity by pretraining on simulated data and adapting to real-world conditions. Extensive experiments on dynamic-resolution datasets and standard benchmarks demonstrate that our method achieves state-of-the-art performance, highlighting its strong potential for clinical deployment.

Chinese Translation

磁共振成像（MRI）超分辨率对于提高诊断可及性至关重要，然而大多数方法将其视为从固定低分辨率输入到高分辨率目标的确定性映射。这忽视了MRI获取物理的一项关键特性：空间分辨率和信噪比（SNR）本质上是相互耦合的，使得任何给定的低分辨率扫描仅仅是在变化的获取权衡下众多可能实现中的一种。我们重新思考MRI超分辨率，将其视为一个物理感知的重建问题，目标是识别最佳的分辨率-SNR配置，然后对其进行超分辨，以获得高质量的MRI结果。这种表述的一个关键含义是，MRI分辨率变得动态而非固定。为了处理这种分辨率异构的输入，我们通过将重建公式化为一个基于坐标的、与分辨率无关的渲染问题，调整了二维高斯划分（2D Gaussian Splatting，2D GS）以适应MRI。为进一步增强保真度，我们引入了三个创新：(1) 一种先验感知的高斯表示，它结合了用于组织特定核初始化的解剖结构先验和通过协方差字典捕获硬件特性成像系统先验；(2) 一种物理约束信号建模方案，它预测固有的组织参数（质子密度ρ和有效弛豫率R2），并通过控制物理方程合成强度，确保生物物理上合理的对比；(3) 一种元学习框架，它通过在模拟数据上进行预训练来缓解配对数据稀缺的问题，并适应现实世界条件。在动态分辨率数据集和标准基准上的广泛实验表明，我们的方法实现了最先进的性能，突显了其在临床部署中的强大潜力。

View on arXiv Download PDF AI Translation

cs.CV / 66 / 2607.06254

VendorBench-100: A Unified Cross-Paradigm Benchmark for Deepfake Image Detection

VendorBench-100：深伪图像检测的统一跨范式基准

Deshmukh, Sharayu N., Rashidunnabi, Md, Gemo, Nelton Tiago, D., Kurundkar G., R., Mahamune M., Deshmukh, Nilesh K.

Abstract

Deepfake image detection is currently served by three fundamentally different paradigms: commercial APIs, zero-shot vision-language models (LLMs), and open-source detectors. Despite their widespread use, these paradigms are rarely evaluated under a common protocol, making direct comparison difficult. We introduce VendorBench-100, a cross-paradigm benchmark that evaluates 36 representative models using a single adversarial 100-image corpus, a unified output schema, and a common evaluation framework. To ensure reliable assessment under the corpus's intentional class imbalance, models are ranked primarily by the Matthews correlation coefficient (MCC), with ROC-AUC reported as a threshold-independent measure of ranking ability. Rather than maximizing dataset size, VendorBench-100 emphasizes challenging real-world scenarios through a curated taxonomy of eight edge-case families, including face swaps, text-to-video stills, AI photo edits, avatar compositing, opaque-provenance images, and compressed research frames. Our evaluation shows that commercial APIs achieve the strongest median performance, followed by vision LLMs and open-source detectors. However, individual open-source models remain competitive with the best vision LLMs. More importantly, we identify a consistent divergence between ranking ability (ROC-AUC) and operating-point quality (MCC), demonstrating that strong score discrimination does not necessarily produce reliable default-threshold decisions. This metric disagreement, rather than any single leaderboard ranking, is the central finding of the benchmark. We release the complete evaluation framework and benchmark results to support reproducible future research. The source code and data are available at: https://github.com/sharayu-20/vendorbench-100

Chinese Translation

深伪图像检测目前由三种根本不同的范式提供支持：商业API、零样本视觉-语言模型（LLMs）和开源检测器。尽管这些范式被广泛使用，但它们很少在共同的协议下进行评估，这使得直接比较变得困难。我们引入了VendorBench-100，这是一个跨范式基准，使用单一的对抗性100图像语料库评估36个代表性模型，采用统一的输出模式和共同的评估框架。为了确保在语料库的故意类别不平衡下进行可靠评估，模型主要通过马修斯相关系数（MCC）进行排名，同时报告ROC-AUC作为一种独立于阈值的排名能力衡量标准。VendorBench-100强调通过精心策划的八个边缘案例家族的分类，挑战现实世界的场景，而不是单纯追求数据集的规模，这些边缘案例包括面部交换、文本到视频静帧、AI照片编辑、头像合成、不透明来源图像和压缩研究帧。我们的评估表明，商业API实现了最强的中位性能，其次是视觉LLMs和开源检测器。然而，个别开源模型在与最佳视觉LLMs的竞争中仍然保持竞争力。更重要的是，我们发现排名能力（ROC-AUC）与操作点质量（MCC）之间存在一致的偏差，表明强评分区分能力并不一定能产生可靠的默认阈值决策。这种指标不一致性，而不是任何单一的排行榜排名，是该基准的核心发现。我们发布了完整的评估框架和基准结果，以支持可重复的未来研究。源代码和数据可在以下网址获取：https://github.com/sharayu-20/vendorbench-100

View on arXiv Download PDF AI Translation

cs.CV / 67 / 2607.06268

MAC-XA: Multi-view Anatomy-Correspondence Fusion for Coronary Stenosis Reporting from X-ray Angiography

MAC-XA: 基于多视角解剖对应融合的冠状动脉狭窄报告从X射线血管造影

Jia, Chen, Zhang, Baochang, Dewi, Fatia Kusuma, Yousefi, Amir, Schunkert, Heribert, Ghotbi, Reza, Navab, Nassir

Abstract

Multi-view reasoning in coronary X-ray angiography is inherently a cross-projection geometric problem, yet automated report generation in this setting remains largely unexplored. The 3D vascular topology leads to projection-dependent branch overlap and foreshortening, rendering single-view modeling fundamentally incomplete and unstable for lesion localization and stenosis grading. Although multi-view fusion appears promising, learning anatomically consistent fusion from real angiograms is impeded by a critical limitation: cross-view alignment is unobservable and cannot be explicitly supervised. Consequently, conventional fusion relies on implicit correlations rather than verified anatomical correspondence. We address this by reformulating multi-view stenosis reporting as an alignment-constrained aggregation problem. A controllable synthetic angiography generation strategy is introduced to expose geometry-derived patch-level correspondence supervision unavailable in real data. An anatomy-correspondence module learns cross-view correspondence matrices that explicitly align auxiliary features within the main-view coordinate space prior to fusion, thereby constraining evidence aggregation to anatomically consistent regions. Experiments on synthetic data and zero-shot transfer to real angiograms show that this alignment-constrained design improves correspondence consistency and structured stenosis reporting compared to single-view modeling and conventional multi-view fusion methods. The code will be publicly available upon publication.

Chinese Translation

冠状动脉X射线血管造影中的多视角推理本质上是一个跨投影几何问题，但在这种环境下的自动报告生成仍然基本未被探索。3D血管拓扑导致了基于投影的分支重叠和缩短，使得单视图建模在病灶定位和狭窄分级方面根本不完整且不稳定。尽管多视角融合显示出前景可期，但从真实血管造影图中学习解剖一致的融合受到一个关键限制：跨视图对齐是不可观察的，无法被明确监督。因此，传统的融合依赖于隐式相关性而不是经过验证的解剖对应。我们通过将多视角狭窄报告重新表述为一个对齐约束聚合问题来解决这个问题。引入了一种可控的合成血管造影生成策略，以暴露在真实数据中不可用的基于几何的补丁级别对应监督。解剖对应模块学习跨视图对应矩阵，明确对齐主视图坐标空间中的辅助特征，然后进行融合，从而将证据聚合约束于解剖一致的区域。在合成数据上的实验以及对真实血管造影图的零-shot迁移表明，该对齐约束设计在对应一致性和结构化狭窄报告方面相比单视图建模和传统多视角融合方法表现出提升。代码将在出版后公开提供。

View on arXiv Download PDF AI Translation

cs.CV / 68 / 2607.06281

Straight-Path Flow Matching for Incomplete Multi-View Clustering

针对不完全多视图聚类的直线路径流匹配

Yuan, Yiteng, Wang, Junyan, Liu, Zheyuan, Jia, Hong, Fan, Lei, Tao, Zhulin, Guo, Lianbo

Abstract

Incomplete Multi-View Clustering addresses the problem of clustering multi-modal data when certain views are missing. Recent end-to-end generative approaches leverage diffusion models to recover missing views via stochastic noise-to-data trajectories. While expressive, such mechanisms are not explicitly designed for clustering, as they initialize from cluster-agnostic noise and rely on stochastic denoising dynamics. In this work, we revisit probability path design in end-to-end generative IMVC. We introduce a flow-matching framework with a linear interpolation path between paired view representations, that replaces diffusion with probability flows between observed and missing views. We provide a formal analysis showing that deterministic ODE flows are inherently better aligned with clustering objectives than diffusion-based stochastic trajectories, especially in terms of transport mechanisms that respect class-conditional data distributions and maintain cluster consistency in finite-step regimes. Building upon this insight, we develop an end-to-end IMVC architecture that integrates straight-path flow-matching view completion with cluster-level and entropy-based alignment to enforce cross-view clustering consistency. Extensive experiments on standard IMVC benchmarks demonstrate that the proposed framework establishes new state-of-the-art performance.

Chinese Translation

不完全多视图聚类解决了在某些视图缺失的情况下对多模态数据进行聚类的问题。近期的端到端生成方法利用扩散模型通过随机噪声到数据的轨迹恢复缺失视图。尽管这些机制具有表达能力，但并非专门针对聚类设计，因为它们从与聚类无关的噪声初始化，并依赖于随机去噪动态。本文重新审视了端到端生成的不完全多视图聚类中的概率路径设计。我们引入了一个流匹配框架，在配对视图表示之间采用线性插值路径，将扩散替换为在观察到的视图和缺失视图之间的概率流。我们提供了正式分析，表明确定性常微分方程（ODE）流在聚类目标方面本质上比基于扩散的随机轨迹更为协调，尤其是在尊重类条件数据分布和在有限步长情况下维持聚类一致性的传输机制方面。在此基础上，我们开发了一个端到端的不完全多视图聚类架构，结合直线路径流匹配视图补全与基于聚类和熵的对齐，以强化跨视图的聚类一致性。在标准不完全多视图聚类基准上的广泛实验表明，所提出的框架建立了新的最先进的性能。

View on arXiv Download PDF AI Translation

cs.CV / 69 / 2607.06291

AlayaWorld: Long-Horizon and Playable Video World Generation

AlayaWorld：长时间跨度和可玩视频世界生成

AlayaWorld Team, Zhang, Kaipeng, Li, Chuanhao, Zhan, Yifan, Ge, Yongtao, Yin, Yuanyang, Tan, Jiaming, He, Kang, Fan, Liaoyuan, Liu, Ruicong, Xu, Xiaojie, Chu, Xuangeng, Li, Zhen, Lin, Zhengyuan, Wang, Zhixiang, Meng, Zian, Gao, Zihui

Abstract

Game worlds have traditionally been built through labor-intensive production pipelines, making them costly to develop, difficult to customization, and expensive to modify after deployment. Recent advances in video world models offer a fundamentally different paradigm. Rather than explicitly authoring every component of a virtual environment, these models autoregressively synthesize future observations conditioned on the current world state and user interactions, enabling playable worlds to be generated online. Trained on both gameplay recordings and real-world videos, they can capture diverse visual appearances and physical dynamics, opening new opportunities for interactive applications beyond gaming, including embodied intelligence. In this paper, we present \textbf{AlayaWorld}, a full-stack open-source framework for building interactive generative worlds. AlayaWorld enables open-ended real-time interaction, allowing users to freely navigate and perform diverse actions such as combat, spell casting, and monster summoning. The framework unifies the complete development-from data preparation model architecture, model training, inference acceleration, and deployment-within a modular and extensible architecture. Alongside the framework, we release reproducible pipelines, reference implementations, evaluation tools, and comprehensive documentation, establishing a practical foundation for future research and real-time applications of generative world models.

Chinese Translation

游戏世界传统上通过劳动密集型的生产流程构建，这使得它们的开发成本高昂，定制困难，并且在部署后修改费用昂贵。最近在视频世界模型方面的进展提供了一种根本不同的范式。这些模型不是显式地创作虚拟环境的每个组件，而是根据当前世界状态和用户交互自回归地合成未来观察，从而实现可在线生成可玩世界。它们在游戏玩法录制和现实世界视频上进行训练，能够捕捉多样的视觉外观和物理动态，为超越游戏的交互应用（包括具身智能）开辟了新的机会。在本文中，我们提出了 extbf{AlayaWorld}，一个用于构建交互式生成世界的全栈开源框架。AlayaWorld支持开放式实时交互，允许用户自由导航并执行多种动作，如战斗、施法和召唤怪物。该框架统一了从数据准备、模型架构、模型训练、推理加速到部署的完整开发过程，采用模块化和可扩展的架构。除了框架外，我们还发布了可复现的流程、参考实现、评估工具和全面的文档，为生成世界模型的未来研究和实时应用奠定了实用基础。

View on arXiv Download PDF AI Translation

cs.CV / 70 / 2607.06295

Visual graphs for image classification: does the structure affect performance?

用于图像分类的视觉图：结构是否影响性能？

Ibba, Alessandra

Abstract

Deep learning models have emerged in machine learning and related fields, demonstrating astonishing performance in various visual tasks. Despite their great success, however, these models are unable to fully encode intrinsic visual structures, and often ignore the spatial, topological, and semantic information contained within an image. Graph neural networks offer a good framework to face this aspect, but their effective use for visual tasks has been only partly explored and mainly starting from a limited perspective. This work aims to address this gap by conducting a systematic comparison of current graph construction techniques within the context of a fixed three-layer GCN architecture. Through an empirical study, it demonstrates in particular how the network structure affects performance and provides an important methodological contribution regarding the computational stages preceding graph utilization, which will be strongly influenced by the structure itself.

Chinese Translation

深度学习模型在机器学习及相关领域中崭露头角，在各种视觉任务中展现出惊人的性能。尽管取得了巨大的成功，这些模型仍无法完全编码内在的视觉结构，常常忽视图像中包含的空间、拓扑和语义信息。图神经网络提供了一个良好的框架来应对这一问题，但其在视觉任务中的有效应用仅部分被探索，且主要从有限的视角出发。本研究旨在通过在固定的三层图卷积网络（GCN）架构下，对当前图构建技术进行系统比较，从而填补这一空白。通过实证研究，特别展示了网络结构如何影响性能，并对图利用之前的计算阶段提供了重要的方法论贡献，这些阶段将受到结构本身的强烈影响。

View on arXiv Download PDF AI Translation

cs.CV / 71 / 2607.06309

Token-Based Dual-view Fusion and Adaptation of Large Vision Models for Breast Cancer Classification

基于标记的双视图融合与大规模视觉模型适应用于乳腺癌分类

Pirsoltan, Aysan Ghayouri, Babakordi, Shima, Mohammadi, Mohammad Reza

Abstract

Accurate breast cancer classification from mammography requires effective integration of complementary information from craniocaudal (CC) and mediolateral oblique (MLO) views, which provide a more complete characterization of breast abnormalities. However, existing multi-view learning approaches typically rely on feature-level aggregation or single-stage cross-attention, which can entangle view-specific and shared representations and restrict interaction to limited network depths. To address these limitations, we propose a token-centric dual-view learning framework that unifies prompt-based adaptation and cross-view fusion within a frozen vision transformer backbone. The framework reformulates inter-view interaction as structured token-level communication, where dedicated fusion tokens explicitly encode bidirectional information exchange between CC and MLO views via cross-attention, serving as intermediate carriers of cross-view dependencies rather than relying on direct feature fusion. Unlike conventional methods that apply fusion at a single layer, fusion modules are inserted at multiple transformer depths, enabling progressive and repeated interaction across the encoder hierarchy. Fusion tokens are reintegrated into the token sequence and refined by subsequent transformer layers, facilitating hierarchical propagation of complementary information while preserving view-specific structure. Experiments on VinDr-Mammo and CMMD datasets demonstrate consistent improvements over linear probing, prompt-only adaptation, and conventional fusion baselines. On the VinDr-Mammo BI-RADS classification task, the framework achieves 50.40% F1-score and 0.8090 AUC, including a 0.10 AUC improvement over a dual-view fusion baseline in the binary setting. Ablation studies further validate the effectiveness of token-based fusion and multi-depth interaction design.

Chinese Translation

从乳腺X线摄影中准确分类乳腺癌需要有效整合来自头尾（CC）视图和侧位（MLO）视图的互补信息，这些信息提供了对乳腺异常的更全面特征描述。然而，现有的多视图学习方法通常依赖于特征级聚合或单阶段交叉注意力，这可能会混淆视图特定和共享的表示，并限制交互的网络深度。为了解决这些局限性，我们提出了一种以标记为中心的双视图学习框架，该框架在一个冻结的视觉变换器骨干网络中统一了基于提示的适应和视图间融合。该框架将视图间交互重新表述为结构化的标记级通信，其中专用的融合标记通过交叉注意力显式编码CC视图和MLO视图之间的双向信息交换，作为视图间依赖关系的中介载体，而不是依赖于直接的特征融合。与传统方法在单层进行融合不同，融合模块被插入到多个变换器深度中，使得在编码器层次结构中实现渐进和重复的交互。融合标记被重新整合到标记序列中，并通过后续的变换器层进行精炼，从而促进互补信息的分层传播，同时保留视图特定结构。在VinDr-Mammo和CMMD数据集上的实验表明，相较于线性探测、仅基于提示的适应和传统融合基线，性能有了一致的提升。在VinDr-Mammo BI-RADS分类任务中，该框架实现了50.40%的F1-score和0.8090的AUC，包括在二元设置下相较于双视图融合基线提高了0.10的AUC。消融研究进一步验证了基于标记的融合和多深度交互设计的有效性。

View on arXiv Download PDF AI Translation

cs.CV / 72 / 2607.06319

Synthetic-to-Real Translation for Class-Agnostic Motion Prediction

面向类别无关运动预测的合成到真实转换

Wu, Yizheng, Fan, Hongwei, Wang, Kewei, Li, Ruibo, Li, Xingyi, Song, Xiao, Wang, Zhe, Ding, Chenjing, Wang, Dongliang, Cao, Zhiguo, Lin, Guosheng

Abstract

Motion understanding is critical for ensuring safety and robustness in autonomous driving systems, driving increasing interest in motion prediction. A key challenge in this domain is the high cost associated with acquiring real-world motion labels. It is therefore ideal if we could transfer motion knowledge from synthetic data to real data. In this context, we explore the potential of synthetic-to-real translation for motion prediction (SRMP). However, the most used naive motion regression methods are notably sensitive to the synthetic-to-real domain shift, resulting in unreliable knowledge translation. To address this, we propose a novel approach integrating a motion knowledge translation framework with two key components: (1) objectness-aware motion prediction, which explicitly models the joint distribution of motion patterns and objectness priors to improve domain-invariant feature learning, and (2) objectness-aided motion enhancement, a motion label refinement mechanism that leverages learned objectness priors to filter motion noise. Furthermore, we present a physically-based pipeline for generating Motion4D, the first synthetic 4D LiDAR dataset tailored for SRMP research, addressing the lack of synthetic motion datasets. Experimental results demonstrate that our approach effectively bridges the domain gaps and yields superior performance on real scenes.

Chinese Translation

运动理解对于确保自动驾驶系统的安全性和鲁棒性至关重要，因而引发了对运动预测的日益关注。在这一领域中的一个关键挑战是获取真实世界运动标签的高成本。因此，从合成数据转移运动知识到真实数据是理想的。在此背景下，我们探讨了合成到真实转换在运动预测中的潜力（Synthetic-to-Real Motion Prediction, SRMP）。然而，最常用的简单运动回归方法显著受到合成到真实领域转变的影响，导致知识传递不可靠。为了解决这个问题，我们提出了一种新的方法，将运动知识转移框架与两个关键组件相结合：(1) 关注物体特征的运动预测，该方法明确建模运动模式和物体特征的联合分布，以改善领域不变特征的学习；(2) 基于物体特征的运动增强，这是一种利用学习到的物体特征来过滤运动噪声的运动标签细化机制。此外，我们提出了一种基于物理的流程来生成Motion4D，这是第一个为SRMP研究量身定制的合成4D LiDAR数据集，以解决合成运动数据集的缺乏。实验结果表明，我们的方法有效地弥合了领域差距，并在真实场景中取得了优越的性能。

View on arXiv Download PDF AI Translation

cs.CV / 73 / 2607.06335

Bridging Diffusion Pruning and Step Distillation with Teacher-Aligned Repair

桥接扩散修剪与步骤蒸馏的教师对齐修复

Ying, Jincheng, Wenlin, Li, Xu, Minghui, Xiao, Yinhao

Abstract

Diffusion models generate high-quality images, but their inference cost comes from two sources: large denoising networks and repeated denoising steps. Existing compression pipelines usually attack these costs separately. Pruning reduces the network, but most pruning methods still rely on a long post-pruning retraining stage to recover a many-step sampler. Step distillation reduces the number of denoising steps, but it usually assumes a student that can already follow the teacher well enough to receive useful distillation gradients. This paper asks whether post-pruning retraining can be replaced by step distillation. We find that the direct replacement fails: after pruning an EDM2-XS teacher, starting SiDA from the pruned checkpoint produces unusable samples. We introduce a short teacher-alignment repair stage as a bridge between pruning and step distillation. The bridge matches the pruned generator to the teacher on noisy real-image latents, then hands the repaired checkpoint to one-step distillation. On ImageNet-512, the original EDM2-XS baseline uses 124.713M parameters and 63 network evaluations, reaching an FID of 3.53. With a suitable distillation objective, our 20% pruned one-step generator uses 98.826M parameters and one network evaluation, reaching an FID of 3.12. With 30% pruning, the model uses 88.029M parameters and one network evaluation, with an FID of 4.26.

Chinese Translation

扩散模型生成高质量图像，但其推理成本来自两个方面：大型去噪网络和重复的去噪步骤。现有的压缩管道通常分别攻击这些成本。修剪减少了网络，但大多数修剪方法仍依赖于长时间的后修剪再训练阶段来恢复多个步骤的采样器。步骤蒸馏减少了去噪步骤的数量，但通常假设学生能够足够好地跟随教师以接收有用的蒸馏梯度。本文探讨后修剪再训练是否可以通过步骤蒸馏来替代。我们发现直接替代是不可行的：在修剪EDM2-XS教师后，从修剪的检查点开始SiDA会产生不可用的样本。我们引入一个短期的教师对齐修复阶段，作为修剪与步骤蒸馏之间的桥梁。该桥梁将修剪后的生成器与教师在噪声真实图像潜变量上进行匹配，然后将修复后的检查点交给一步蒸馏。在ImageNet-512上，原始的EDM2-XS基线使用124.713M参数和63次网络评估，达到FID为3.53。通过适当的蒸馏目标，我们的20%修剪的一步生成器使用98.826M参数和一次网络评估，达到FID为3.12。在30%的修剪下，该模型使用88.029M参数和一次网络评估，FID为4.26。

View on arXiv Download PDF AI Translation

cs.CV / 74 / 2607.06354

Generalized Synthetic Image Detection with Enhanced RGB-Noise Representation Learning

增强RGB噪声表征学习的广义合成图像检测

Li, Zhen, Cao, Gang, Zhang, Tian, Yu, Lifang, Weng, Shaowei

Abstract

The rapid advancement of large-scale generative models has accelerated the spread of highly deceptive AI-generated images, making generalized synthetic image detection a critical imperative. Existing forensic networks often struggle with cross-model generalization and realworld degradations due to their reliance on single-domain representations and conventional binary classification optimization. To overcome these limitations, we propose RNSIDNet, a novel forensic framework that achieves robust detection through enhanced RGB-Noise representation learning. Specifically, our method employs a dual-branch architecture where global RGB semantics, extracted by an attention-refined CLIP backbone, dynamically modulate highfrequency noise artifacts captured by Bayar convolutions via a Feature-wise Linear Modulation (FiLM) module. To further enhance the learned representations, we design a Hard Sample-aware Contrastive Learning (HSCL) strategy. By explicitly penalizing challenging training samples, HSCL reshapes the latent feature space to maximize the discriminative margin between pristine and synthetic domains. Extensive experiments across eight public benchmark datasets verify that our model achieves state-of-the-art performance, delivering superior generalization ability, robustness, and computational efficiency. Code and dataset will be publicly available on https://github.com/multimediaFor/RNSIDNet.

Chinese Translation

大规模生成模型的快速进展加速了高度欺骗性的AI生成图像的传播，使得广义合成图像检测成为一项至关重要的任务。现有的取证网络由于依赖于单一领域的表征和传统的二分类优化，常常在跨模型泛化和现实世界降质方面面临挑战。为了解决这些局限性，我们提出了RNSIDNet，一个新颖的取证框架，通过增强RGB噪声表征学习实现稳健的检测。具体而言，我们的方法采用双分支架构，利用注意力优化的CLIP骨干网络提取的全局RGB语义，动态调节通过Bayar卷积捕获的高频噪声伪影，此过程通过特征线性调制（Feature-wise Linear Modulation，FiLM）模块实现。为了进一步增强学习到的表征，我们设计了一种困难样本感知对比学习（Hard Sample-aware Contrastive Learning, HSCL）策略。通过明确惩罚具有挑战性的训练样本，HSCL重塑潜在特征空间，以最大化真实和合成领域之间的判别边界。通过在八个公共基准数据集上的广泛实验验证，我们的模型实现了最先进的性能，展示了优越的泛化能力、鲁棒性和计算效率。代码和数据集将在 https://github.com/multimediaFor/RNSIDNet 上公开发布。

View on arXiv Download PDF AI Translation

cs.CV / 75 / 2607.06356

TMF-RSE: Tri-Modal Fusion with Regional Semantics and Evidential Uncertainty for Lung Severity Scoring

TMF-RSE：具有区域语义和证据不确定性的三模态融合肺部严重性评分

Zidi, Fadi Abdeladhim, Bekhouche, Salah Eddine, Sellam, Abdellah Zakaria, Maroun, Gaby, Dornaika, Fadi, Distante, Cosimo

Abstract

Accurate quantification of lung disease severity from chest imaging is critical for clinical decision-making and resource allocation. We propose a tri-modal deep learning framework, TMF-RSE (Tri-Modal Fusion with Regional Semantics and Evidential Uncertainty), that combines appearance features from two-dimensional chest inputs, structural features from lung segmentation masks, and semantic features from vision-language models (VLMs) for severity quantification. Our approach employs complementary fusion mechanisms that integrate semantic guidance, structural priors, and hierarchical interactions across modalities. The model employs evidential regression to provide both severity predictions and uncertainty estimates. Experiments on the Per-COVID-19 CT and RALO datasets show that TMF-RSE outperforms recent transformer-based baselines, achieving MAE of 4.02 and Pearson correlation of 0.9629 on Per-COVID-19 validation, and 0.339 MAE / 0.973 PC on RALO geographic extent.

Chinese Translation

从胸部影像中准确量化肺病严重性对于临床决策和资源分配至关重要。我们提出了一种三模态深度学习框架TMF-RSE（Tri-Modal Fusion with Regional Semantics and Evidential Uncertainty），该框架结合了来自二维胸部输入的外观特征、来自肺部分割掩膜的结构特征以及来自视觉-语言模型（VLMs）的语义特征用于严重性量化。我们的方法采用互补融合机制，集成了语义指导、结构先验以及跨模态的层次交互。该模型采用证据回归，提供严重性预测和不确定性估计。在Per-COVID-19 CT和RALO数据集上的实验表明，TMF-RSE的表现优于近期的基于变压器的基线，在Per-COVID-19验证集上实现了4.02的平均绝对误差（MAE）和0.9629的Pearson相关系数，在RALO地理范围上得到了0.339 MAE和0.973的Pearson相关系数。

View on arXiv Download PDF AI Translation

cs.CV / 76 / 2607.06374

VaseMuseum: Digital Intelligent Museum for Ancient Greek Pottery

花瓶博物馆：古希腊陶器的数字智能博物馆

Wang, Jiazi, Zhang, Nonghai, Xie, Qiushi, Zhang, Zeyu, Chen, Yufeng, Zhao, Yang, Shao, Ling, Tang, Hao

Abstract

Vision-language models (VLMs) have made interactive digital museums increasingly feasible by connecting 3D digitization with natural-language artifact exploration. However, in cultural heritage domains such as ancient Greek pottery, reliable VLM assistance is limited by two challenges. First, open-ended interpretation requires grounding fine-grained 2D/3D visual evidence in specialized curatorial knowledge, yet the retrieval process may introduce weak sources and unverifiable references. Second, when the available evidence is incomplete, noisy, or ambiguous, VLMs often produce confident but unsupported answers instead of calibrated uncertainty. To address these challenges, we propose VaseMuseum, a lightweight and modular multimodal agent framework for intelligent digital museums of ancient Greek pottery. VaseMuseum combines an interactive virtual museum with VaseAgent, which supports both 2D images and 3D artifacts through multimodal perception, 3D-aware reasoning, external knowledge retrieval, and inference-time reliability control. Specifically, VaseAgent retrieves evidence from authoritative web and museum knowledge sources, and source-level control selects diverse and verifiable evidence before generation. Meanwhile, response-level control checks generated claims against the evidence pool and encourages neutral, evidence-bounded answers when support is insufficient or conflicting. Moreover, a training-free GRPO-style selection mechanism favors responses with valid references and calibrated confidence without updating the VLM backbone. Experiments in a realistic digital museum simulation show that VaseMuseum improves citation validity, reduces hallucinations on knowledge-intensive queries, and produces more neutral answers under ambiguity compared with search-enabled VLM baselines.

Chinese Translation

视觉语言模型（VLMs）通过将3D数字化与自然语言文物探索相结合，使互动数字博物馆的实现变得愈发可行。然而，在如古希腊陶器等文化遗产领域，可靠的VLM辅助受到两个挑战的限制。首先，开放式解读需要将细粒度的2D/3D视觉证据与专业的策展知识相结合，但检索过程中可能引入弱源和不可验证的参考文献。其次，当可用证据不完整、嘈杂或含糊时，VLM往往会产生自信但缺乏支持的答案，而不是经过校准的不确定性。为了解决这些挑战，我们提出了花瓶博物馆（VaseMuseum），这是一个轻量级和模块化的多模态代理框架，专为古希腊陶器的智能数字博物馆而设计。花瓶博物馆结合了一个互动虚拟博物馆和花瓶代理（VaseAgent），后者通过多模态感知、3D感知推理、外部知识检索和推理时的可靠性控制，支持2D图像和3D文物。具体而言，花瓶代理从权威的网络和博物馆知识来源中检索证据，源级控制在生成前选择多样且可验证的证据。同时，响应级控制将生成的主张与证据池进行核对，并在支持不足或相互冲突时鼓励中立且基于证据的答案。此外，一个无训练的GRPO风格选择机制倾向于选择具有有效引用和校准置信度的回复，而无需更新VLM主干。在真实的数字博物馆仿真实验中，花瓶博物馆在引用有效性上有所改善，减少了知识密集型查询中的幻觉，并在不确定性下生成了更中立的答案，与启用搜索的VLM基准相比。

View on arXiv Download PDF AI Translation

cs.CV / 77 / 2607.06389

FADRA: Frequency-Aware Diffusion with Residual Adaptation for Video Face Restoration

FADRA：具有残差自适应的频率感知扩散模型用于视频人脸修复

Jiang, Jin, Wang, Jia, Hu, Panwen, Zhao, Weiran, Liao, Shengcai

Abstract

Video face restoration (VFR) aims to recover high-quality and temporally consistent facial details from severely degraded video sequences; however, existing methods still struggle to balance spatial fidelity and temporal coherence under complex degradations. To address this, we propose FADRA, a frequency-aware diffusion framework with iterative residual adaptation specifically tailored for robust VFR. We first leverage the strong temporal consistency of a pre-trained text-to-video diffusion model and introduce lightweight LoRA adapters together with a Low-Quality (LQ) Pixel-Alignment Feature Fusion module to efficiently adapt the frozen generative prior to the VFR task. To further adapt the frozen diffusion backbone to the downstream VFR task beyond LoRA-based adaptation, we introduce a Repeated Residual Adaptation Head (RRAH) for step-wise residual refinement after the diffusion backbone. To make this refinement explicitly guided by the degraded observation, RRAH further takes the LQ latent together with the current velocity prediction as input, allowing the model to repeatedly revisit LQ cues and predict residual updates at each flow-matching step. This LQ-guided repeated residual adaptation helps recover fine facial details while preserving the inherent temporal priors of the pre-trained model. Furthermore, to ensure the structural integrity of perceptually important details, we introduce a Frequency-Aware Loss that provides explicit supervision across multiple spectral bands, emphasizing visually sensitive frequency components that are crucial for perceptual quality and prone to temporal jittering. Extensive experiments demonstrate that FADRA recovers better facial structures and produces more temporally consistent videos than state-of-the-art methods, leading to clear gains in both quantitative metrics and visual perception.

Chinese Translation

视频人脸修复（VFR）旨在从严重退化的视频序列中恢复高质量和时间一致的人脸细节；然而，现有方法在复杂退化情况下仍然难以平衡空间保真度和时间连贯性。为了解决这个问题，我们提出了FADRA，一个具有迭代残差自适应的频率感知扩散框架，专门针对鲁棒的VFR任务。我们首先利用预训练的文本到视频扩散模型的强时间一致性，引入轻量级的LoRA适配器以及低质量（LQ）像素对齐特征融合模块，以高效地将冻结的生成先验适应于VFR任务。为了进一步将冻结的扩散骨干网络适应于下游VFR任务，我们引入了一种逐步残差精炼的重复残差自适应头（RRAH）。为了使这种精炼在明显上受到退化观测的引导，RRAH进一步将LQ潜变量与当前速度预测一起作为输入，使模型能够在每个流匹配步骤中反复回顾LQ提示并预测残差更新。这种LQ引导的重复残差自适应有助于恢复细致的人脸细节，同时保留预训练模型的固有时间先验。此外，为了确保感知重要细节的结构完整性，我们引入了一种频率感知损失，在多个频谱带上提供明确的监督，强调对感知质量至关重要且容易受到时间颤动影响的视觉敏感频率成分。大量实验表明，FADRA在恢复人脸结构和生成时间一致性方面优于最先进的方法，在定量指标和视觉感知上都取得了显著的进展。

View on arXiv Download PDF AI Translation

cs.CV / 78 / 2607.06402

What Images Cannot Say: Language-Guided Olfactory Representation Learning

图像无法传达的内容：语言引导的嗅觉表征学习

Tsonis, Eleftherios, Wang, Xi, Kalogeiton, Vicky

Abstract

Images tell us what a scene looks like, but rarely what it would feel like to be there. While recent datasets pair visual scenes with electronic-nose measurements, aligning smell signals with images remains challenging because many olfactory cues arise from contextual environmental factors that are not directly visible in pixels. We introduce SCENT, a multimodal framework that uses language guidance as a semantic bridge between vision and olfaction. Our approach leverages Vision-Language Models (VLMs) to generate scene descriptors capturing objects, environmental context, and plausible ambient smell cues suggested by the visual scene. These descriptors provide semantic guidance for learning olfactory representations. We train a smell encoder that maps electronic-nose signals into a shared embedding space aligned with both visual and textual representations, and introduce a languageguided latent decomposition that separates object-specific odors from contextual environmental contributions. Experiments on the New York Smells dataset demonstrate that SCENT significantly improves crossmodal retrieval compared to vision-only baselines, achieving state-of-theart performance on smell-to-image and smell-to-text retrieval tasks. In addition, our framework produces interpretable olfactory representations that enable the disentanglement of complex smell mixtures. Our results reveal the importance of contextual semantic information for grounding olfactory perception in multimodal learning and pave the way for future research in this area.

Chinese Translation

图像告诉我们场景的外观，但很少能传达身处其中的感觉。尽管最近的数据集将视觉场景与电子鼻测量相结合，但将嗅觉信号与图像对齐仍然具有挑战性，因为许多嗅觉线索源于环境的上下文因素，而这些因素在像素中并不可见。我们提出了SCENT，一个多模态框架，利用语言引导作为视觉与嗅觉之间的语义桥梁。我们的方法利用视觉-语言模型（Vision-Language Models, VLMs）生成场景描述符，捕捉物体、环境上下文以及视觉场景所暗示的合理环境气味线索。这些描述符为学习嗅觉表征提供了语义指导。我们训练了一个嗅觉编码器，将电子鼻信号映射到与视觉和文本表征对齐的共享嵌入空间，并引入了一种语言引导的潜在分解，能够将特定物体的气味与环境上下文的贡献分离开来。在纽约嗅觉数据集上的实验表明，SCENT在跨模态检索方面显著优于仅基于视觉的基线，在气味到图像和气味到文本的检索任务中达到了最先进的性能。此外，我们的框架生成了可解释的嗅觉表征，使得复杂气味混合物的解耦成为可能。我们的结果揭示了上下文语义信息在多模态学习中对扎根嗅觉感知的重要性，并为该领域的未来研究铺平了道路。

View on arXiv Download PDF AI Translation

cs.CV / 79 / 2607.06408

Temporal Modeling of Optically Variable Devices in Identity Documents

身份文件中光学可变设备的时间建模

Pouliquen, Glen, Chazalon, Joseph, Chiron, Guillaume, Terrades, Oriol Ramos, Géraud, Thierry, Awal, Ahmad Montaser

Abstract

Robust remote verification of identity documents relies on analyzing faint, transparent security features like Optically Variable Devices (OVDs), or "holograms", within user-captured videos under uncontrolled conditions. Current systems, however, face critical limitations: existing methods often treat video frames in isolation, neglecting the intrinsic dynamic nature of OVDs and leaving systems vulnerable to swapping attacks, or focus on general holographic presence and lack the ability to verify specific OVD types. Moreover, the economic infeasibility of frame-by-frame video annotation makes supervised training impractical. In this work, we introduce two novel approaches for verifying the dynamic behavior of transparent OVDs protecting the holder's portrait, specifically designed for open-set scenarios where attack types are unknown during training. We demonstrate that these approaches can be trained without any attack samples in a self-supervised setting, surpassing previous state-of-the-art methods on public datasets while adhering strictly to industrial constraints. Our results confirm that modeling temporal dynamics is essential for defeating sophisticated attacks under realistic conditions, and underscores the promise of sequence modeling and anomaly detection for OVD verification. Code is available at https://github.com/EPITAResearchLab/pouliquen.26.icdar.

Chinese Translation

身份文件的稳健远程验证依赖于在用户捕获的视频中分析微弱的、透明的安全特征，如光学可变设备（Optically Variable Devices, OVDs）或“全息图”，在不受控的条件下进行。然而，当前系统面临着关键的局限性：现有方法往往孤立地处理视频帧，忽视了OVDs内在的动态特性，使系统容易受到交换攻击；或者关注于一般的全息存在，缺乏验证特定OVD类型的能力。此外，逐帧视频注释的经济不可行性使得监督训练变得不切实际。在本研究中，我们提出了两种新颖的方法，用于验证保护持有者肖像的透明OVDs的动态行为，特别设计用于在训练期间攻击类型未知的开放集场景。我们证明这些方法可以在自监督环境中进行训练，而无需任何攻击样本，超越了公共数据集上的先前最先进方法，同时严格遵循工业约束。我们的结果确认了建模时间动态对于在现实条件下击败复杂攻击的重要性，并强调了序列建模和异常检测在OVD验证中的潜力。代码可在 https://github.com/EPITAResearchLab/pouliquen.26.icdar 获取。

View on arXiv Download PDF AI Translation

cs.CV / 80 / 2607.06420

HoloCount: A Holistic Visual Counting Benchmark for MLLMs

HoloCount: 多模态大语言模型的整体视觉计数基准

Deng, Jinhong, Qiao, Limeng, Wan, Guanglu

Abstract

Visual counting is a fundamental pillar of multimodal intelligence, requiring a seamless integration of fine-grained grounding and spatial reasoning. While Multimodal Large Language Models (MLLMs) have achieved remarkable success in qualitative scene understanding, their quantitative precision remains a significant bottleneck, often characterized by persistent numerical hallucinations. Existing counting benchmarks primarily focus on basic perception in simplified contexts, failing to capture the complex failure modes that emerge under logical constraints or adversarial conditions. To address these limitations, we introduce HoloCount, a holistic and diagnostically rich benchmark structured around a three-level hierarchical taxonomy. HoloCount evaluates MLLMs across: (1) Semantic Counting, focusing on atomic and property-based enumeration; (2) Analytical Counting, assessing logical composition through spatial and set-based reasoning; and (3) Robustness Testing, probing model integrity against adverse scenarios and grounded counter-priors, such as high-density scenes and linguistic biases. Through an exhaustive evaluation of over 20 state-of-the-art MLLMs, we reveal a critical performance gap: even top-tier models degrade significantly as tasks transition from perception to complex analytical reasoning and adverse scenarios. Our findings provide a systematic landscape of current MLLM counting capabilities and offer a roadmap for developing more grounded and reliable multimodal systems. The dataset is available at https://mm-mvr.github.io/HoloCount/.

Chinese Translation

视觉计数是多模态智能的基础支柱，需要精准融合细粒度的定位和空间推理。尽管多模态大语言模型（MLLMs）在定性场景理解方面取得了显著成功，但它们的定量精确性仍然是一个重要瓶颈，常常出现持续的数值幻觉。现有的计数基准主要关注简化情境下的基本感知，未能捕捉在逻辑约束或对抗性条件下出现的复杂失效模式。为了解决这些限制，我们引入了 HoloCount，这是一个围绕三层层级分类法结构构建的整体性且诊断丰富的基准。HoloCount 在以下三个方面评估 MLLMs: (1) 语义计数，专注于原子和基于属性的枚举; (2) 分析计数，通过空间和集合推理评估逻辑构成; (3) 鲁棒性测试，探测模型在不利场景和基础对立信息（如高密度场景和语言偏见）下的完整性。通过对超过 20 种最先进的 MLLMs 进行全面评估，我们揭示了一个关键的性能差距：即便是顶级模型，随着任务从感知转向复杂分析推理和不利场景，其表现显著下降。我们的研究结果提供了当前 MLLM 计数能力的系统性概览，并为开发更具基础性和可靠性的多模态系统提供了方向。数据集可在 https://mm-mvr.github.io/HoloCount/ 获取。

View on arXiv Download PDF AI Translation

cs.CV / 81 / 2607.06424

XRFormer: Multiscale Tokenization for XRF Representation Learning

XRFormer：用于 XRF 表示学习的多尺度标记化

Daimellah, Sofiane, Hégarat-Mascle, Sylvie Le, Boust, Clotilde

Abstract

X-ray fluorescence (XRF) spectroscopy is a key modality for material analysis in cultural heritage. However, automated learning from XRF spectra remains challenging: XRF spectra are complex one-dimensional signals composed of sharp elemental peaks, broader structures, and background variations that are not taken into account by existing learning-based models. This paper introduces XRFormer, a transformer architecture tailored to XRF spectra through a multiscale convolutional tokenizer that injects locality and multi-resolution inductive biases before global self-attention. The tokenizer progressively reduces spectral resolution while increasing embedding dimensionality, and the resulting token sequence is processed by a standard transformer encoder. We further investigate self-supervised pretraining for XRF representation learning using Masked Spectral modeling (MSM) and a physics-informed Peak Presence Prediction (PPP) objective. Experiments on the Pigments Checker STANDARD v.5 dataset for pigment identification and unmixing show that XRFormer consistently outperforms ViT, SpectralFormer (with and without CAF), and a 1D-CNN baseline for pigment identification. For pigment unmixing, XRFormer achieves robust abundance estimation while maintaining significantly higher parameter efficiency than SpectralFormer, operating at a lower token resolution (128 vs. 512 tokens) and with less than half the number of parameters (1.5M vs. 3.37M). MSM yields consistent gains across both tasks, while PPP further enhances performance for both identification and unmixing when tuned with an appropriate peak prominence. These results highlight multiscale, modality-aware tokenization as an effective and parameter efficient foundation for transformer-based XRF modeling under data-limited conditions. A GitHub repository is provided at https://github.com/sofiane1010/XRFormer.

Chinese Translation

X 射线荧光 (XRF) 光谱学是文化遗产材料分析的关键手段。然而，从 XRF 光谱中进行自动化学习仍然存在挑战：XRF 光谱是由尖锐的元素峰、较宽的结构和背景变化构成的复杂一维信号，而现有的基于学习的模型未能考虑这些因素。本文介绍了 XRFormer，一种针对 XRF 光谱量身定制的变换器架构，通过多尺度卷积标记器在全局自注意力之前引入了局部性和多分辨率的归纳偏见。该标记器逐步降低光谱分辨率，同时增加嵌入维度，生成的标记序列由标准变换器编码器处理。我们进一步研究了使用掩蔽光谱建模 (Masked Spectral modeling, MSM) 和物理启发的峰值存在预测 (Peak Presence Prediction, PPP) 目标的自监督预训练，以进行 XRF 表示学习。在用于颜料识别和解混的颜料检查 STANDARD v.5 数据集上的实验表明，XRFormer 在颜料识别方面始终优于 ViT、SpectralFormer（有和没有 CAF）和 1D-CNN 基线。在颜料解混方面，XRFormer 实现了稳健的丰度估计，同时在参数效率上明显高于 SpectralFormer，运行在较低的标记分辨率（128 与 512 个标记）且参数数量不到一半（1.5M 与 3.37M）。MSM 在这两个任务中均表现出一致的提升，而适当的峰值显著性调优时，PPP 进一步增强了识别和解混的性能。这些结果强调了多尺度、模态感知的标记化作为在数据有限条件下基于变换器的 XRF 建模的有效且参数高效的基础。相关代码库可在 https://github.com/sofiane1010/XRFormer 获取。

View on arXiv Download PDF AI Translation

cs.CV / 82 / 2607.06440

PIPBench: A Profile-Inclusive Framework for Personalized Image Generation Evaluation

PIPBench：一个包含个人档案的个性化图像生成评估框架

Wu, Yuhang, Zhang, Shuxiang, Ching, Wee Hian, Zhang, Chi, Liu, Miao

Abstract

Recent text-to-image models such as DALLE-3 excel at following diverse prompts yet remain blind to individual aesthetic preferences. We study personalized image generation, where models must align outputs with a user's implicit visual preferences based on a few historically preferred images and a short prompt. To this end, we introduce PIPBench, the first profile-inclusive benchmark for evaluating personalized image generation. We further propose a novel data construction pipeline that leverages psychological and demographic profiling dimensions for both real-user data collection and scalable agent-based data generation. Using PIPBench, we conduct a thorough evaluation of representative line of methods. Our experiments reveal key limitations in existing methods, suggesting new challenges and opportunities for personalized text-to-image synthesis. Project page: https://wuyuhang05.github.io/PIPBench/

Chinese Translation

近期的文本到图像模型，如DALLE-3，能够很好的响应多样化的提示，但仍然对个人的审美偏好缺乏敏感性。我们研究个性化图像生成，即模型必须根据用户的隐性视觉偏好，结合几幅历史上偏爱的图像和一个简短的提示来调整输出。为此，我们引入了PIPBench，这是第一个用于评估个性化图像生成的包含个人档案的基准。我们进一步提出了一种新颖的数据构建流程，利用心理学和人口统计特征维度进行真实用户数据收集和可扩展的基于代理的数据生成。通过使用PIPBench，我们对一系列代表性方法进行了彻底评估。我们的实验揭示了现有方法的关键局限性，提出了个性化文本到图像合成的新挑战和机遇。项目页面：https://wuyuhang05.github.io/PIPBench/

View on arXiv Download PDF AI Translation

cs.CV / 83 / 2607.06445

Analysis-by-Proxy: Localization Signals in VLMs Operating as Condition Encoders

代理分析：作为条件编码器的视觉语言模型中的定位信号

Baron, Yoav, Dorfman, Sara, Paiss, Roni, Cohen-Or, Daniel, Patashnik, Or

Abstract

Vision-Language Models (VLMs) are increasingly utilized as the conditioning backbone for diffusion-based image editing due to their remarkable multimodal reasoning capabilities. While standalone VLMs demonstrate strong localization capabilities, editing pipelines frequently struggle to maintain this accuracy, particularly in complex, multi-entity scenes. In this work, we investigate this performance gap, hypothesizing that it stems from treating the VLM as a condition encoder. In this role, the model is restricted to a single forward pass, preventing the autoregressive generation process for which it was optimized, thereby failing to fully expose its capabilities. To investigate whether this spatial understanding persists when the VLM is used as a condition encoder, we introduce Analysis-by-Proxy. In this framework, we train a lightweight, interpretable proxy model on the VLM's intermediate representations using an auxiliary localization task. By analyzing the VLM through this proxy, we uncover the specific VLM representations that encode localization information. Our findings expose a fundamental mismatch between how spatial knowledge is represented within a VLM condition encoder and how it is extracted by current editing pipelines. We reveal that under single-pass constraints, the localization signal does not reliably propagate to the predefined layer configurations commonly used for conditioning. Instead, this crucial signal remains hidden within intermediate representations, at locations that vary depending on the input prompt. Using our introduced Analysis-by-Proxy framework, we reveal the fundamental failures of existing condition extraction strategies in editing pipelines, opening the door to more principled design of conditioning architectures.

Chinese Translation

视觉语言模型（VLMs）因其卓越的多模态推理能力，越来越多地被用作基于扩散的图像编辑中的条件支撑。虽然独立的VLM展现出强大的定位能力，但在编辑流程中，尤其是在复杂的多实体场景中，通常难以保持这种准确性。在本研究中，我们探讨了这种性能差距，假设其源于将VLM视为条件编码器。在这一角色中，模型仅限于单次前向传递，这限制了其经过优化的自回归生成过程，从而未能充分展现其能力。为探讨当VLM作为条件编码器时，这种空间理解是否依然存在，我们引入了代理分析（Analysis-by-Proxy）。在这一框架下，我们使用辅助定位任务，在VLM的中间表示上训练一个轻量且可解释的代理模型。通过这一代理对VLM进行分析，我们揭示了编码定位信息的特定VLM表示。我们的发现暴露出VLM条件编码器内部空间知识的表示方式与当前编辑管道中提取的方式之间存在根本不匹配。我们发现，在单次传递限制下，定位信号并未可靠地传播到常用于条件化的预定义层配置中。相反，这一关键信号仍然隐藏在中间表示中，其位置因输入提示的不同而异。通过我们引入的代理分析框架，我们揭示了现有编辑管道中条件提取策略的根本失败，为更具原则性的条件架构设计开辟了新的思路。

View on arXiv Download PDF AI Translation

cs.CV / 84 / 2607.06457

Andha-Dhun: A First Look at Audio Descriptions in Hindi

Andha-Dhun：印地语音频描述的初步研究

Chakraborty, Ritabrata, Kala, Divy, Gupta, Nisheeth Bhooshan, Sreeram, Ganji, Reddy, Pailla Balakrishna, Tapaswi, Makarand

Abstract

Audio Descriptions (ADs) narrate visual content for Blind and Low Vision (BLV) audiences during gaps in audiovisual media. There is growing momentum around ADs in movies and TV shows, and with mandates from India's Central Board of Film Certification (CBFC), there is a need to expand ADs beyond English. Yet, there is no work that generates ADs for any Indian language. To address this gap, we present the first systematic study of ADs in Hindi, contributing to aspects such as data, generation, and evaluation. We introduce Andha-Dhun, the first dataset of human-authored Hindi ADs collected from 8 full-length movies. We explore two approaches for generating ADs in Hindi: (i) directly from English dense video descriptions, and (ii) translating English ADs into Hindi. We evaluate these approaches using perplexity and LLM-as-a-judge metrics to assess fluency and quality respectively. We also analyze movies that have both English and Hindi human-authored ADs and find that naive translation introduces artifacts and narrows diversity compared to original Hindi ADs. Direct machine translation fails to adapt cultural references, while human-translated ADs do better but still fall short. Our findings emphasize that the purpose of Hindi ADs is accessibility for Indian BLV audiences, and that this requires adapting content for the audience more than strict fidelity to the source.

Chinese Translation

音频描述（ADs）在视听媒体的空隙中为盲人和低视力人群（BLV）叙述视觉内容。近年来，电影和电视节目中的音频描述逐渐受到重视，随着印度中央电影认证委员会（CBFC）的要求，有必要将音频描述扩展至英语以外的语言。然而，目前尚无为任何印度语言生成音频描述的研究。为填补这一空白，我们首次系统性研究了印地语音频描述，涉及数据、生成和评估等方面。我们推出了Andha-Dhun，这是第一个来自8部完整电影的人类创作印地语音频描述数据集。我们探讨了两种生成印地语音频描述的方法：（i）直接从英语视频密集描述生成，（ii）将英语音频描述翻译成印地语。我们利用困惑度和以LLM为评判标准的指标分别评估这些方法的流畅性和质量。此外，我们还分析了同时拥有英语和印地语人类创作音频描述的电影，发现简单翻译引入了伪影，并减少了相比于原始印地语音频描述的多样性。直接的机器翻译未能调整文化参考，而人类翻译的音频描述表现更好，但仍有不足。我们的研究结果强调，印地语音频描述的目的在于为印度的盲人和低视力观众提供可及性，这需要更加关注内容的适应性，而非严格遵循源语言的忠实度。

View on arXiv Download PDF AI Translation

cs.CV / 85 / 2607.06466

Verification of Dynamic Holographic Behavior in Identity Documents

身份证件中动态全息行为的验证

Pouliquen, Glen, Chazalon, Joseph, Chiron, Guillaume, Géraud, Thierry, Awal, Ahmad Montaser

Abstract

This paper addresses the remote verification of the authenticity of Optically Variable Devices (commonly known as holograms) on identity documents. Typically placed over the cardholder's photo, these devices provide strong and easily verifiable security for human inspection but pose challenges for automated verification. Existing approaches easily cover static frauds (e.g. paper photocopy) and can be evaluated for such, but their capacity to detect real, dynamic fraud cases (e.g. handcrafted hologram) has not been evaluated to date because of the lack of public datasets. Furthermore, they are usually trained to detect known attack types, and few of them can generalize to new, unseen attacks. This work features three contributions to address these limitations: 1) a new public dataset, MIDV-DynAttack, which extends the existing MIDV-Holo dataset with realistic, static and dynamic attacks against identity document specimens, tripling the number of attack samples compared to the original dataset, 2) a novel verification method which can assess the authenticity of a specific hologram thanks to the analysis of its dynamic behavior and appearance, can be trained without dynamic attack samples, and exhibits new state-of-the-art performance, 3) a benchmark of existing approaches which follows a clear evaluation protocol and emphasizes the inability of other approaches to deal with dynamic attacks, as well as new challenging attacks to deal with. Code and dataset are publicly available at https://github.com/EPITAResearchLab/pouliquen.25.icdar.

Chinese Translation

本文探讨了对身份证件上光学可变设备（通常称为全息图）真实性的远程验证。这些设备通常位于持卡人照片上方，为人工检查提供强有力且易于验证的安全性，但在自动化验证中面临挑战。现有方法可以轻松应对静态欺诈（例如，纸质复印件），并能针对这些情况进行评估，但由于缺乏公开数据集，其检测真实动态欺诈案例（例如，手工制作的全息图）的能力尚未得到评估。此外，现有方法通常针对已知攻击类型进行训练，且很少有方法能够对新的、未见过的攻击进行泛化。本研究提出三项贡献以应对这些局限性：1）一个新的公开数据集MIDV-DynAttack，扩展了现有的MIDV-Holo数据集，涵盖针对身份证件样本的真实静态和动态攻击，将攻击样本数量较原始数据集增加了三倍；2）一种新的验证方法，可以通过动态行为和外观的分析评估特定全息图的真实性，可以在没有动态攻击样本的情况下进行训练，并展现出新的最先进的性能；3）现有方法的基准测试，遵循明确的评估协议，强调其他方法在应对动态攻击方面的无能为力以及新挑战攻击的处理。代码和数据集已公开发布在 https://github.com/EPITAResearchLab/pouliquen.25.icdar。

View on arXiv Download PDF AI Translation

cs.CV / 86 / 2607.06468

EgoPolice: A Benchmark for Egocentric Video Understanding in High-Stakes Police Body-Worn Camera Footage

EgoPolice：高风险警用佩戴式摄像头视频中以自我为中心的视频理解基准

Saez-Diez, Max Gonzalez, Chung, Jihoon, Wolsky, Adam D., Lanzalotto, Gregory, Knox, Dean, Mummolo, Jonathan, Stewart, Brandon M., Russakovsky, Olga

Abstract

We introduce EgoPolice, a carefully curated dataset of real, egocentric police-civilian interactions, sourced from publicly available body-worn camera videos. We select police-civilian action labels that are critical for police behavioral research and annotate them at a second-by-second granularity. The videos feature rapid and irregular camera motion, dense human interactions, and rare high-stakes events, making the dataset a challenging benchmark for motion-robust and context-aware egocentric perception. We provide two different tasks, classification and multiple-choice question-answering, and benchmark both open-source and closed-source models. We find that even the best video models like Gemini 2.5 Pro still struggle to accurately predict high-risk actions such as "Weapon Out". Beyond serving as a benchmark, EgoPolice provides a foundation for developing models capable of identifying events of interest in large-scale body-worn camera video repositories, enabling more efficient downstream human review.

Chinese Translation

我们介绍了EgoPolice，这是一个精心策划的真实以自我为中心的警民互动数据集，来源于公开可用的佩戴式摄像头视频。我们选择了对警务行为研究至关重要的警民行为标签，并以逐秒的粒度进行标注。这些视频展示了快速且不规则的摄像头运动、密集的人际互动以及罕见的高风险事件，使得该数据集成为一个对运动鲁棒性和上下文感知具有挑战性的基准。我们提供了两种不同的任务，分类和多项选择问答，并对开源和闭源模型进行了基准测试。我们发现，即使是像Gemini 2.5 Pro这样的最佳视频模型，在准确预测高风险行为（如“武器出示”）方面仍然存在困难。除了作为基准，EgoPolice还为开发能够在大规模佩戴式摄像头视频库中识别感兴趣事件的模型提供了基础，从而实现更高效的后续人工审查。

View on arXiv Download PDF AI Translation

cs.CV / 87 / 2607.06478

A VLM-Enhanced Framework for Comprehensive Traffic Sign Condition Assessment Integrating Daytime Visual Performance and Nighttime Retroreflectivity Evaluation

基于VLM增强的综合交通标志状态评估框架：整合白天视觉表现与夜间反射率评估

Zhang, Linlin, Owor, Neema Jakisa, Yu, Xiang, Watts, Abby, Adu-Gyamfi, Yaw

Abstract

Traffic signs are crucial components of road safety, serving as visual tools under all lighting conditions. The Manual on Uniform Traffic Control Devices (MUTCD) specifies daytime visual factors such as legibility and color contrast, and nighttime retroreflectivity requirements. Traditional assessment methods rely on manual inspections, which the Federal Highway Administration (FHWA) notes are subjective, labor-intensive and pose safety concerns, while retroreflectometers are expensive and unaffordable for smaller agencies. Most existing studies focus on either daytime factors or nighttime retroreflectivity but rarely integrate both aspects comprehensively. This study develops a novel framework that systematically evaluates traffic signs through integrated daytime-nighttime assessment. The methodology employs three fine-tuned Vision Language Models (VLMs) for daytime visual performance assessment across four key factors: legibility, color, surface and shape integrity, and surrounding environment conditions. VLM predictions are converted to numerical scores through sentiment analysis and Contrastive Language-Image Pre-Training (CLIP) scoring, while nighttime performance is assessed using LiDAR-derived retroreflectivity following established calibration procedures. The framework integrates these components into a comprehensive Sign Condition Index (SCI) for maintenance guidance. Evaluation results demonstrated that LLaVA and Qwen outperformed InternVL, achieving bidirectional cosine similarity scores of 0.67-0.76 across all factors. Among 462 validated traffic signs, 68 were flagged by the proposed framework as requiring immediate replacement due to inadequate retroreflectivity performance. This research provides a cost-effective alternative to traditional manual inspections for comprehensive traffic sign condition assessment.

Chinese Translation

交通标志是道路安全的重要组成部分，在各种光照条件下作为视觉工具发挥作用。《统一交通控制设备手册》（MUTCD）规定了白天的视觉因素，如可读性和颜色对比度，以及夜间反射率的要求。传统的评估方法依赖于人工检查，联邦公路管理局（FHWA）指出这种方法主观性强、劳动密集且存在安全隐患，而反射率测量仪器价格昂贵，小型机构难以承受。现有大多数研究集中于白天因素或夜间反射率，鲜有全面整合两者的研究。本研究开发了一种新颖的框架，通过综合的白天-夜间评估系统地评估交通标志。该方法采用三种经过精细调优的视觉语言模型（VLMs）对白天视觉表现进行评估，涵盖四个关键因素：可读性、颜色、表面和形状完整性，以及周围环境条件。VLM的预测通过情感分析和对比语言-图像预训练（CLIP）评分转换为数值分数，而夜间表现则通过遵循既定校准程序的激光雷达（LiDAR）衍生反射率进行评估。该框架将这些组件整合为一个综合的标志状态指数（SCI），用于维护指导。评估结果表明，LLaVA和Qwen在所有因素上均优于InternVL，双向余弦相似度得分在0.67-0.76之间。在462个经过验证的交通标志中，有68个被该框架标记为因反射率表现不佳而需要立即更换。本研究为全面的交通标志状态评估提供了一种成本效益高的替代传统人工检查的方法。

View on arXiv Download PDF AI Translation

cs.CV / 88 / 2607.06481

Prompt-Adapter Context Routing for Parameter-Efficient Multi-Shot Long Video Extrapolation

用于参数高效多镜头长视频外推的提示适配器上下文路由

Córdoba, Anna, Tercero, Adam Puente, Hijo, Nerea Angulo, Tercero, Mar Linares, Barrientos, Julia, Miranda, Ainhoa, Olivera, Jesús

Abstract

We present PACR-Video, a parameter-efficient framework for multi-shot long video extrapolation that preserves recurring entities, scene structure, visual style, and causal progression without full generator fine-tuning. PACR-Video keeps a text-to-video diffusion transformer frozen and augments it with low-rank temporal adapters conditioned by learned shot-role prompt tokens. To maintain long-horizon coherence, it builds a recursive prompt bank that stores compact entity, location, action, and style prompts from previous shots, then routes them through adapter gates according to predicted narrative dependencies. A Shot-Local/Story-Global tuning objective combines next-shot reconstruction, cross-shot identity contrast, and prompt sparsity regularization, while an adapter composition schedule balances early-shot visual consistency with later-shot event progression and viewpoint change. Across six multi-shot and long-video benchmarks, PACR-Video outperforms text-to-video, tuning-based, memory-augmented, streaming, and recursive-context baselines on distributional quality, semantic alignment, identity consistency, temporal smoothness, motion stability, transition coherence, and human preference. These results show that compact prompt routing and lightweight temporal adaptation provide sufficient controllable capacity for stable long video extrapolation.

Chinese Translation

我们提出了 PACR-Video，这是一个参数高效的多镜头长视频外推框架，能够在不进行全生成器微调的情况下保留反复出现的主体、场景结构、视觉风格和因果进程。PACR-Video 保持一个冻结的文本到视频扩散变换器，并通过低秩时间适配器进行增强，这些适配器由学习到的镜头角色提示令牌进行条件化。为了维持长视距的一致性，它构建了一个递归提示库，存储来自之前镜头的紧凑实体、位置、动作和风格提示，然后根据预测的叙事依赖通过适配器门路由这些提示。一个镜头局部/故事全局的调优目标结合了下一镜头重构、跨镜头身份对比和提示稀疏正则化，而适配器组成调度则在早期镜头的视觉一致性与后期镜头的事件进展和视角变化之间保持平衡。在六个多镜头和长视频基准测试中，PACR-Video 在分布质量、语义对齐、身份一致性、时间平滑性、运动稳定性、过渡一致性和人类偏好等方面超越了文本到视频、基于调优、记忆增强、流式传输和递归上下文的基准。这些结果表明，紧凑的提示路由和轻量的时间适配为稳定的长视频外推提供了足够可控的能力。

View on arXiv Download PDF AI Translation

cs.CV / 89 / 2607.06483

Mitigating Domain Shift in Conditioned Floor Plan Generation: Synthetic Pre-training for Data-Efficient Adaptation

减轻条件性平面图生成中的领域转移：数据高效适应的合成预训练

Ospici, Matthieu, Gueze, Arnaud, Bourrat, Luc, Bernhardt, Adrien

Abstract

Robustness to domain shift is a key requirement for floor plan generative models to be applicable beyond the single dataset they were trained on, as floor plans vary widely across regions due to distinct architectural cultures, spatial constraints, and construction practices, while acquiring new annotated datasets remains costly and domain-specific. Yet, no prior work has studied this robustness in the context of conditioned floor plan generation. In this paper, we evaluate state-of-the-art models from two fundamentally different generative paradigms across three public datasets (RPLAN, MagicPlan and Swiss Dwellings) and show that they are highly sensitive to domain shift, with up to an order of magnitude performance degradation when transferred across domains. To mitigate this with minimal target-domain supervision, we introduce a procedural method to generate a large-scale synthetic training dataset that enforces strict physical constraints (non-overlapping rooms, valid door placement, graph consistency) while intentionally sacrificing architectural realism through highly irregular spatial arrangements and aggressive geometric perturbation of room shapes. We show that pre-training on this synthetic data considerably improves zero-shot cross-domain performance, outperforming in-domain training on MagicPlan. Furthermore, it provides a highly effective initialization for fine-tuning, accelerating target domain adaptation and outperforming real-world initialization baselines by up to 40% in a low-data regime.

Chinese Translation

对领域转移的鲁棒性是平面图生成模型能够在超出其训练单一数据集的情况下应用的关键要求，因为平面图因不同的建筑文化、空间限制和施工实践而在各个地区存在显著差异，而获取新的带注释数据集依然成本高昂且具有领域特异性。然而，先前的研究未曾在条件性平面图生成的背景下探讨这种鲁棒性。本文评估了来自两种基本不同生成范式的最新模型，在三个公共数据集（RPLAN、MagicPlan 和 Swiss Dwellings）上进行了测试，结果表明这些模型对领域转移具有高度敏感性，当跨域迁移时性能可能下降一个数量级。为了以最小的目标领域监督来减轻这一问题，我们提出了一种程序化的方法来生成大规模合成训练数据集，该数据集在强制实施严格的物理约束（非重叠房间、有效的门口放置、图形一致性）的同时，通过高度不规则的空间排列和对房间形状的激进几何扰动，有意牺牲了建筑现实主义。我们表明，在此合成数据上进行预训练显著改善了零样本跨域性能，其表现超过了在 MagicPlan 上的域内训练。此外，它为微调提供了一个高度有效的初始化，加速了目标领域适应，并在低数据环境下超越了真实世界初始化基线，提升幅度可达40%。

View on arXiv Download PDF AI Translation

cs.CV / 90 / 2607.06485

AirflowAttack: Thermal-Airflow Adversarial Perturbations against Infrared Remote-Sensing Vision-Language Models

AirflowAttack：针对红外遥感视觉语言模型的热气流对抗扰动

Su, Cong, Han, Jiaju, Sun, Xuemeng, Hu, Chengyin, Zhang, Qike, Guo, Jiujiang, Wei, Yiwei, Long, Jiahuan

Abstract

Vision-language models (VLMs) are increasingly deployed on infrared (IR) remote sensing imagery in security-critical settings, yet their adversarial robustness remains unexamined. We present AirflowAttack, to our knowledge the first adversarial attack for IR remote-sensing VLMs and the first to weaponize thermal-airflow turbulence as the perturbation prior. A lightweight generator synthesizes a single input-agnostic perturbation regularized toward physically plausible airflow patterns. Optimized on one surrogate CLIP model, it attains a mean zero-shot scene-classification attack success rate (ASR, the fraction of samples whose top-1 class changes) of 48.5% across five diverse CLIP backbones, far exceeding four IR-specific physical baselines (27.7--37.0%). Applied to six state-of-the-art VLMs, it cuts scene-classification accuracy by up to 38.2% relative, yet paradoxically makes some models more confident in their IR analysis, confabulating the perturbation as genuine thermal evidence such as temperature gradients and convection. Ablations show the airflow prior raises physical plausibility at no measurable cost to attack success. Together with a benchmark spanning eleven models and four tasks, these findings expose critical vulnerabilities in the rapidly expanding IR VLM ecosystem.

Chinese Translation

视觉语言模型（VLMs）正在越来越多地应用于安全关键环境下的红外（IR）遥感图像，但其对抗鲁棒性尚未受到检验。我们提出了AirflowAttack，据我们所知，这是针对红外遥感VLMs的首个对抗攻击，也是首个将热气流湍流作为扰动先验的攻击方法。一个轻量级生成器合成一个与输入无关的扰动，并针对物理上可信的气流模式进行正则化。在一个代理的CLIP模型上优化后，它在五种不同的CLIP骨干网络上实现了48.5%的平均零-shot场景分类攻击成功率（ASR，即其前1类发生变化的样本比例），远远超过了四个IR特定的物理基线（27.7-37.0%）。在六个最先进的VLMs上应用时，它相对降低了场景分类的准确率高达38.2%，然而矛盾的是，使得某些模型在其红外分析中的信心更高，将扰动误认为是如温度梯度和对流等真实的热证据。消融实验表明，气流先验在没有 measurable 成本的情况下提高了物理可信度。结合覆盖十一种模型和四个任务的基准，这些发现揭示了快速扩展的红外VLM生态系统中的关键脆弱性。

View on arXiv Download PDF AI Translation

cs.CV / 91 / 2607.06516

Point as Skeleton: Accumulated Point Cloud Enhanced Autoregressive Generation for Closed-Loop Autonomous Driving Simulation

点作为骨架：用于闭环自主驾驶仿真的累积点云增强自回归生成

Wong, Songbur, Jia, Xiaosong, You, Junqi, Zhang, Bo, Xu, Pei, Xia, Renqiu, Qiu, Yuping, Zhang, Shaofeng, Zhao, Zelin, Yan, Xuechao, Zhou, Yuchen, Chen, Yurui, Guo, Wen, Xu, Hang, Yan, Junchi

Abstract

Evaluating end-to-end autonomous driving (E2E-AD) remains challenging, as existing driving simulation methods often trade off closed-loop interactivity (e.g., CARLA) and real-world visual fidelity (e.g., nuScenes). We present \textbf{\emph{Point as Skeleton}}, a generative sensor simulation framework for state-updated autoregressive driving video generation, in which an autoregressive generator synthesizes visual observations from step-wise updated ego states, actor states, scene maps, and point-cloud skeleton conditions. To support closed-loop rollout, we introduce Reset-and-Roll, which adapts rolling diffusion inference to simulation by preventing future-conditioned latent states from being committed across simulation steps. To stabilize error accumulation during step-wise autoregressive rollout, we introduce point-cloud skeletons that decouple foreground and background assets and project them into camera-view painted-point and template-depth conditions, providing appearance and geometric cues. We further implement a nuPlan-based renderer-level closed-loop generative interface for evaluating generation under ego deviations from the original log. Experiments on nuScenes and nuPlan show that \textit{Point as Skeleton} improves autoregressive generation quality during closed-loop rollout, demonstrating its potential for visually faithful closed-loop driving simulation. The code is available at https://github.com/krauwu/point-as-skeleton.

Chinese Translation

评估端到端自主驾驶（E2E-AD）仍然具有挑战性，因为现有的驾驶仿真方法往往在闭环交互性（例如，CARLA）和真实世界视觉保真度（例如，nuScenes）之间进行权衡。我们提出了 extbf{ extit{点作为骨架}}，一个生成传感器仿真框架，用于状态更新的自回归驾驶视频生成，其中自回归生成器根据逐步更新的自我状态、演员状态、场景地图和点云骨架条件合成视觉观察。为了支持闭环推理，我们引入了重置与滚动（Reset-and-Roll），该方法通过防止未来条件潜在状态在仿真步骤间的转移，适应了滚动扩散推理。为了在逐步自回归推理过程中稳定误差积累，我们引入了点云骨架，解耦前景和背景资产，并将它们投影到相机视图中绘制的点和模板深度条件下，提供外观和几何线索。我们进一步实现了基于nuPlan的渲染器级闭环生成接口，以评估在自我偏差情况下的生成效果。针对nuScenes和nuPlan的实验表明， extit{点作为骨架}在闭环推理过程中改善了自回归生成质量，展示了其在视觉忠实的闭环驾驶仿真中的潜力。代码可在 https://github.com/krauwu/point-as-skeleton 中获取。

View on arXiv Download PDF AI Translation

cs.CV / 92 / 2607.06534

CAIRN: Cross-Room 3D Scene Understanding with Topology-Aware Large Multimodal Models

CAIRN：具有拓扑意识的大型多模态模型的跨房间3D场景理解

Liang, He, Ma, Chenyang, Zhang, Yiming, Shin, Sangyun, Markham, Andrew, Trigoni, Niki, He, Yuhang

Abstract

Existing 3D scene-grounded Large Language Models (3D-LLMs) focus on answering questions grounded in simplified single-room 3D scenes, lacking the ability to reason over real-world household environments containing multiple interconnected rooms and diverse object categories. We introduce CAIRN, a topology-aware 3D-LLM for multi-room 3D scene understanding. CAIRN aligns transformer attention with scene hierarchy, giving the model explicit awareness of object-level relations and room-level connectivity. It enriches object tokens with room-local relational context via a graph neural network, introduces learned room tokens for room-level abstraction, and applies a hierarchical attention mask with geometric bias to route information according to scene topology. CAIRN is developed on CAIRN-MR, a benchmark we introduce on HM3D for multi-room 3D scene understanding, covering grounding, captioning, and four question-answering tasks that progressively evaluate from intra-room perception to cross-room reasoning. Experiments show that CAIRN outperforms prior 3D-LLMs by a large margin across all CAIRN-MR tasks while remaining competitive on five single-room benchmarks.

Chinese Translation

现有的基于3D场景的大型语言模型（3D-LLMs）主要集中于回答简化的单房间3D场景中的问题，缺乏对包含多个相互连接房间和多样化物体类别的真实家庭环境进行推理的能力。我们提出了CAIRN，这是一种具有拓扑意识的3D-LLM，用于多房间3D场景理解。CAIRN通过将变换器注意力与场景层次对齐，使模型明确感知物体级关系和房间级连接性。它通过图神经网络丰富物体标记的房间局部关系上下文，引入学习到的房间标记以实现房间级抽象，并应用具有几何偏差的分层注意力掩码，根据场景拓扑路由信息。CAIRN是在CAIRN-MR上开发的，这是一项我们在HM3D上提出的基准，涵盖了多房间3D场景理解的基础、字幕生成和四个问题回答任务，这些任务逐步评估从房间内感知到跨房间推理的能力。实验结果表明，CAIRN在所有CAIRN-MR任务中大幅度超越了先前的3D-LLMs，同时在五个单房间基准测试中仍保持竞争力。

View on arXiv Download PDF AI Translation

cs.CV / 93 / 2607.06549

Unsupervised Domain Adaptation for Calcification Classification in Mammography Across Multi-Site Datasets

基于无监督领域适应的多站点数据集中钙化分类的研究

Liu, Xuan, Nguyen, Derek L., Barre, Emily C., Thomas, Jennifer, Lynch, Thomas, Marks, Jeffrey R., Hwang, E. Shelley, Ryser, Marc D., Lo, Joseph Y., Grimm, Lars J.

Abstract

Deep learning-based computer-aided diagnosis (CAD) systems have shown strong performance in breast cancer diagnosis, particularly for classification tasks in mammography. However, domain shifts across multi-site datasets remain a challenge, especially when models are applied to unseen domains. In this work, we proposed a calcification classification framework to improve malignant versus benign breast disease classification across multi-site mammography datasets. The framework consisted of two components: (1) an unsupervised domain adaptation module based on style transfer models (AdaIN and CycleGAN) to generate vendor-specific and technique-specific training samples without additional annotations, and (2) a supervised classification module using Swin Transformer V2 as the backbone. We evaluated the proposed method on three datasets: cross-validation on OPTIMAM (National Health Service, United Kingdom; n=2994), followed by external validation on EMBED (Emory University; n=125), and Duke Calcification Dataset v1 (n=788). These datasets cover multiple vendors and include both full-field digital mammography and synthetic 2D images derived from digital breast tomosynthesis. The proposed framework improved cross-site performance for both EMBED (AUC 0.68 to 0.72) and the Duke Calcification Dataset (AUC 0.68 to 0.73). These findings indicate that domain adaptation can reduce domain shifts and improve the generalization for calcification classification across multi-site datasets.

Chinese Translation

基于深度学习的计算机辅助诊断（CAD）系统在乳腺癌诊断中表现出色，尤其是在乳腺X光摄影分类任务中。然而，跨多站点数据集的领域偏移依然是一个挑战，特别是在模型应用于未见领域时。在本研究中，我们提出了一种钙化分类框架，以改善跨多站点乳腺X光摄影数据集中恶性与良性乳腺疾病的分类。该框架由两个部分组成：（1）基于风格迁移模型（AdaIN和CycleGAN）的无监督领域适应模块，旨在生成特定于厂商和技术的训练样本，无需额外注释；（2）使用Swin Transformer V2作为主干的监督分类模块。我们在三个数据集上评估了所提出的方法：对OPTIMAM（英国国家健康服务；样本量=2994）进行交叉验证，随后在EMBED（埃默里大学；样本量=125）和Duke Calcification Dataset v1（样本量=788）上进行外部验证。这些数据集涵盖多个厂商，包括全场数字乳腺摄影和从数字乳腺断层合成中得出的合成2D图像。所提框架提高了EMBED（AUC由0.68提升至0.72）和Duke Calcification Dataset（AUC由0.68提升至0.73）的跨站表现。这些发现表明，领域适应能够减少领域偏移并改善跨多站点数据集中钙化分类的泛化能力。

View on arXiv Download PDF AI Translation

cs.CV / 94 / 2607.06552

MonoIR-RS: Infrared Remote Sensing Vision-Language Learning with CLIP and VLM Adaptation

MonoIR-RS：利用CLIP和VLM适应的红外遥感视觉-语言学习

Han, Jiaju, Yaqi, Ma, Chai, Yahui, Sun, Xuemeng, Li, Xin, Zhang, Qike, Zhao, Yingying, Chen, Xiang, Yang, Luwei, Hu, Chengyin, Long, Jiahuan

Abstract

Infrared remote-sensing imagery captures intensity structure, object-background contrast, and illumination-invariant cues often invisible in RGB imagery. Yet, most remote-sensing vision-language resources and models focus on visible-band semantics, leaving infrared vision-language understanding underexplored. We introduce MonoIR-RS, a large-scale infrared remote-sensing vision-language dataset and benchmark that couples IR-aware data construction with CLIP-style contrastive adaptation and VLM instruction tuning. Built from the same source pool and split as FusionRS, MonoIR-RS retains the infrared image as the model-facing modality, yielding 600,000 synthesized infrared images and 59,032 retained IR-aware caption records. The model experiments use this retained language-supervision subset, whose captions rewrite supervision around grayscale structure and infrared-style contrast instead of RGB appearance. We show that the synthesized infrared imagery is markedly closer to real thermal imagery than a grayscale conversion on the AVIID benchmark. We fine-tune five CLIP backbones and six VLM backbones, and calibrate them against zero-shot behavior: IR-aware adaptation lifts CLIP mean recall by up to 12.8 points and drives VLM captioning IR-cue coverage to 100% while reducing residual RGB-color leakage to near zero. By isolating the infrared modality from RGB-IR dual-modal learning, MonoIR-RS offers a controlled, reproducible testbed for aligning infrared remote-sensing evidence with language.

Chinese Translation

红外遥感图像捕捉到的强度结构、物体与背景的对比以及照明不变线索，往往在RGB图像中是不可见的。然而，现有的大多数遥感视觉-语言资源和模型专注于可见光带的语义，导致红外视觉-语言理解仍未得到充分探索。我们提出了MonoIR-RS，一个大规模的红外遥感视觉-语言数据集和基准，结合了与IR相关的数据构建、CLIP风格的对比适应和VLM指令调优。MonoIR-RS构建于与FusionRS相同的数据源池，并保持相同的分割，保留了红外图像作为模型面对的模态，生成了60万张合成的红外图像和59,032条保留的与IR相关的描述记录。模型实验使用这一保留的语言监督子集，其描述围绕灰度结构和红外风格对比进行重写，而非RGB外观。我们展示了合成的红外图像在AVIID基准测试中明显比灰度转换更接近真实的热成像图像。我们对五个CLIP骨干网络和六个VLM骨干网络进行微调，并对它们的零样本表现进行校准：IR-aware适应使CLIP的平均召回率提升了最多12.8个百分点，并促使VLM描述中的红外线索覆盖率达到了100%，同时将残留的RGB颜色泄漏降低到接近零。通过将红外模态与RGB-IR双模态学习分离，MonoIR-RS提供了一个受控、可重复的测试平台，用于将红外遥感证据与语言进行对齐。

View on arXiv Download PDF AI Translation

cs.CV / 95 / 2607.06553

From RGB Generation to Dense Field Readout: Pixel-Space Dense Prediction with Text-to-Image Models

从RGB生成到稠密场读取：基于文本到图像模型的像素空间稠密预测

Wang, Zanyi, Lin, Xin, Li, Haodong, Jiang, Dengyang, Li, Yijiang, Xie, Pengtao

Abstract

Large-scale text-to-image models are attractive backbones for dense prediction because RGB generation pretraining learns rich semantic, structural, and geometric priors. Existing generative and editing approaches reuse these priors by casting dense prediction as target generation: annotations such as depth, normals, alpha mattes, masks, and heatmaps are encoded into an RGB-trained VAE latent space and decoded back as image-like targets. We argue this inherits more of the generative output interface than dense prediction requires: unlike RGB synthesis, dense prediction asks for pixel-correct, task-native fields on the same image plane, not new RGB content to be rendered. Our key observation is that a pretrained DiT already organizes RGB inputs through a patch-to-token-to-patch lattice on the image plane, so each token indexes a fixed output patch whose channels can carry task-native quantities instead of RGB appearance. We instantiate this as ReChannel: we keep the VAE encoder for the DiT's input distribution but drop the target-side decoder, adapt the frozen DiT with task LoRA, and map each token to its p x p x K_t pixel-space patch through a shared token-local linear head--about 33K parameters, no spatial mixing. Using FLUX-Klein, we evaluate on six dense prediction tasks and over a dozen benchmarks. This minimal interface sets new state-of-the-art on trimap-free matting, KITTI depth, and referring segmentation, and stays competitive on normals, saliency, and pose. In a matched 4B setting it is more accurate and 2.48x faster than an edit-plus-latent-decode counterpart--dense perception can benefit from generative pretraining without inheriting its output interface.

Chinese Translation

大规模的文本到图像模型因其RGB生成预训练能够学习丰富的语义、结构和几何先验而成为稠密预测的有吸引力的骨干网络。现有的生成和编辑方法通过将稠密预测视为目标生成，重用这些先验：深度、法线、α蒙版、掩膜和热图等注释被编码进RGB训练的变分自编码器（VAE）潜在空间，并解码回图像类目标。我们认为这继承了更多生成输出接口，而不是稠密预测所需的：与RGB合成不同，稠密预测要求在同一图像平面上提供像素精确的、任务原生的场，而不是需要渲染的新RGB内容。我们的关键观察是，预训练的DiT已经通过图像平面上的补丁到标记再到补丁的晶格对RGB输入进行了组织，因此每个标记索引一个固定的输出补丁，其通道可以承载任务原生数量而不是RGB外观。我们将其实现为ReChannel：我们保留VAE编码器用于DiT输入分布，但删除目标侧解码器，适配冻结的DiT与任务LoRA，并通过共享的标记本地线性头将每个标记映射到其p x p x K_t像素空间补丁——大约33K参数，无空间混合。使用FLUX-Klein，我们对六个稠密预测任务和十多个基准进行了评估。这个最小接口在无trimap的抠图、KITTI深度和参照分割上设定了新的最先进水平，并且在法线、显著性和姿态上保持竞争力。在匹配的4B设定中，它比编辑加潜在解码的对应方法更准确且快2.48倍——稠密感知可以从生成预训练中受益，而不必继承其输出接口。

View on arXiv Download PDF AI Translation

cs.CV / 96 / 2607.06555

ProxyPose: 6-DoF Pose Tracking via Video-to-Video Translation

ProxyPose: 通过视频到视频的转换实现六自由度姿态跟踪

Zhang, Ruihang, Taubner, Felix, Ravi, Pooja, Kutulakos, Kiriakos N., Lindell, David B.

Abstract

Tracking the six-degree-of-freedom (6-DoF) pose of objects and surfaces from monocular video is a long-standing problem in computer vision. To tackle this problem, existing methods require inputs beyond the video itself-such as 3D models, depth maps, object masks, or task-specific learned features-and they struggle with textureless, transparent, reflective, or deformable surfaces. Here, we introduce ProxyPose, which recasts 6-DoF pose tracking as video-to-video translation. Given only a video and a single marked pixel in the first frame, a fine-tuned video diffusion model translates the input into a proxy video-a synthetic video depicting a colored polyhedron undergoing the same local rigid-body motion as the surface region at the marked pixel. Because the proxy's geometry and appearance are known by construction, recovering its full 6-DoF trajectory reduces to classical pose estimation with off-the-shelf solvers. This formulation leverages large-scale video pre-training to absorb the hardest aspects of pose tracking-handling challenging materials, occlusions, and deformations-into the translation step, while operating at the pixel level with no assumptions about object identity, boundaries, or global rigidity. ProxyPose achieves state-of-the-art 6-DoF pose tracking accuracy without the additional inputs required by competing methods and after fine-tuning the video model only on synthetic data. We further demonstrate that ProxyPose extends to face tracking, camera pose estimation, and challenging in-the-wild scenes that are beyond the reach of existing approaches. Project page: https://ruihangzhang97.github.io/proxypose/.

Chinese Translation

从单目视频中跟踪物体和表面的六自由度（6-DoF）姿态是计算机视觉中的一个长期问题。为了解决这个问题，现有方法需要超出视频本身的输入，例如3D模型、深度图、物体掩膜或特定任务学习到的特征，并且在处理无纹理、透明、反射或可变形表面时表现不佳。在此，我们提出了ProxyPose，它将6-DoF姿态跟踪重新定义为视频到视频的转换。给定仅一个视频和第一帧中的单个标记像素，经过微调的视频扩散模型将输入转换为代理视频——一个合成视频，描绘了一个有色多面体，其局部刚体运动与标记像素的表面区域相同。由于代理的几何形状和外观是已知的，恢复其完整的6-DoF轨迹便简化为使用现成求解器进行经典的姿态估计。该公式利用大规模视频预训练，将姿态跟踪中最棘手的方面（如处理复杂材料、遮挡和变形）吸收到转换步骤中，同时在像素级操作，不假设物体的身份、边界或全局刚性。ProxyPose 在无需竞争方法要求的附加输入的情况下实现了最先进的6-DoF姿态跟踪准确性，并且仅在合成数据上微调视频模型。我们进一步证明，ProxyPose 可扩展至面部跟踪、相机姿态估计以及超出现有方法能力范围的困难的实际场景。项目页面：https://ruihangzhang97.github.io/proxypose/.

View on arXiv Download PDF AI Translation

cs.CV / 97 / 2607.06560

Vision as Unified Multimodal Generation

视知觉作为统一的多模态生成

Han, Xiaoyang, Li, Jianhua, Deng, Kewang, Chen, Zukai, Shi, Xuanke, Wang, Sihan, Li, Boxuan, Wang, Linyan, Xie, Siyi, You, Xin, Quan, Jinsheng, Cai, Zhongang, Diao, Haiwen, Liu, Ziwei, Yang, Lei, Lin, Dahua, Wang, Quan

Abstract

We formulate computer vision as unified multimodal generation, where heterogeneous visual tasks are expressed in the native text and image generation spaces of a unified multimodal model, without task-specific architectures. Under this formulation, SenseNova-Vision uses natural-language instructions and optional visual prompts to specify tasks, target regions or views, and decoding conventions, and generates responses as text for symbolic outputs, images for dense spatial predictions, or mixed text-and-image outputs for compositional tasks. To support large-scale training, we convert diverse computer vision annotations into instruction-response examples compatible with these generation spaces, resulting in the SenseNova-Vision Corpus, a computer-vision instruction-response corpus spanning text, image, and mixed targets. Starting from an off-the-shelf pretrained unified multimodal model, SenseNova-Vision is trained primarily on this corpus, with auxiliary multimodal data used as a capability-preserving mixture, and requires no task-specific prediction heads or architectural modifications. The resulting model covers a broad range of vision tasks, including detection, OCR, keypoint estimation, segmentation, depth estimation, surface normal prediction, point maps, and camera pose estimation, while supporting language-defined variants that combine category, color, region, and other visual cues. Experiments show that a single unified model can match leading task-specialized systems across structured visual understanding, dense geometric prediction, segmentation, and multi-view visual geometry. These results suggest unified multimodal generation as a scalable route for integrating computer vision capabilities into general-purpose foundation models. The model and corpus are publicly available.

Chinese Translation

我们将计算机视觉表述为统一的多模态生成，其中异构视觉任务在统一多模态模型的原生文本和图像生成空间中表达，而无需特定于任务的架构。在这种表述下，SenseNova-Vision 使用自然语言指令和可选的视觉提示来指定任务、目标区域或视图以及解码约定，并生成文本作为符号输出、图像作为密集空间预测的输出，或混合文本和图像输出以应对组合任务。为了支持大规模训练，我们将多样的计算机视觉注释转换为与这些生成空间兼容的指令-响应示例，从而形成 SenseNova-Vision 语料库，这是一个涵盖文本、图像和混合目标的计算机视觉指令-响应语料库。从一个现成的预训练统一多模态模型开始，SenseNova-Vision 主要在该语料库上进行训练，同时使用辅助多模态数据作为能力保持的混合，并且不需要特定于任务的预测头或架构修改。最终模型覆盖了广泛的视觉任务，包括检测、光学字符识别（OCR）、关键点估计、分割、深度估计、表面法线预测、点图和相机姿态估计，同时支持结合类别、颜色、区域和其他视觉线索的语言定义变体。实验表明，单一的统一模型可以在结构化视觉理解、密集几何预测、分割和多视图视觉几何方面与领先的任务专用系统相匹配。这些结果表明，统一的多模态生成是将计算机视觉能力整合到通用基础模型中的可扩展途径。该模型和语料库已公开提供。

View on arXiv Download PDF AI Translation

cs.CV / 98 / 2607.06565

ELSA3D: Elastic Semantic Anchoring for Unified 3D Understanding and Generation

ELSA3D：用于统一3D理解和生成的弹性语义锚定

Yu, Tianjiao, Li, Xinzhuo, Shen, Yifan, Susladkar, Onkar, Liu, Yuanzhe, Zhou, Xiaona, Lourentzou, Ismini

Abstract

Unified 3D foundation models aspire to generate 3D assets and reason about them in language within a single backbone, but their text-3D interaction remains largely implicit. Existing methods concatenate text and 3D tokens into a flat sequence and rely on self-attention, collapsing coarse structural cues and fine geometric details into one undifferentiated representation. We introduce ELSA3D, a unified 3D model that addresses this with elastic semantic anchoring, structuring language and geometric reasoning jointly along matched abstraction scales. ELSA3D represents geometry with a scale-aware octree tokenizer and introduces Anchor Tokens, sparse cross-modal units that select semantic cues, route them to the most relevant 3D scale, retrieve scale-specific geometric evidence, and write the fused signal back into the unified representation, keeping interaction sparse yet precise. A lightweight per-block router makes both computation and reasoning elastic, choosing which text tokens instantiate anchors at which geometric scale so that cross-modal capacity concentrates where alignment is most needed. ELSA3D achieves state-of-the-art performance across image-to-3D generation, text-to-3D generation, and 3D captioning, outperforming the strongest unified baseline while roughly halving FLOPs and inference latency relative to the non-elastic version of the same model.

Chinese Translation

统一的3D基础模型旨在在单一主干中生成3D资产并用语言进行推理，但它们的文本与3D的交互仍然在很大程度上是隐式的。现有方法将文本和3D标记串联成一个平坦的序列，并依赖自注意力机制，将粗糙的结构线索和精细的几何细节压缩为一个未区分的表示。我们提出了ELSA3D，一个通过弹性语义锚定解决这一问题的统一3D模型，能够在匹配的抽象尺度上联合结构语言和几何推理。ELSA3D使用尺度感知的八叉树分词器表示几何，并引入锚标记（Anchor Tokens），这些稀疏的跨模态单元选择语义线索，将其路由到最相关的3D尺度，检索尺度特定的几何证据，并将融合后的信号写回统一表示中，保持了交互的稀疏性和精确性。一个轻量级的每块路由器使得计算和推理都具有弹性，选择哪些文本标记在何种几何尺度实例化锚点，从而使跨模态能力集中的地方对齐最为需要。ELSA3D在图像到3D生成、文本到3D生成和3D字幕生成方面实现了最新的性能，超过了最强的统一基线，同时相较于相同模型的非弹性版本大约减少了FLOP和推理延迟的一半。

View on arXiv Download PDF AI Translation

人工智能 (Artificial Intelligence)

cs.AI / 1 / 2607.05456

Prompt-to-Paper: Agentic AI System for Bioinformatics

Prompt-to-Paper：用于生物信息学的自主智能系统

Kamran, Ramsha, Amjad, Maheera, Mustansar, Zartasha, Shaukat, Arsalan, Sherbaz, Salma, Khan, Muhammad U. S.

Abstract

While recent advances in large language models have enabled end-to-end automated manuscript generation, existing systems suffer from three critical deficiencies: (i) generated claims are not deterministically grounded in verifiable literature, (ii) experimental results are frequently fabricated rather than executed, and (iii) there exists no standardized, multi-dimensional framework to assess whether AI-generated manuscripts meet the quality and rigor required for real-world publication. We present Prompt-to-Paper, a multi-agent framework that directly addresses this evaluation gap through three integrated innovations. First, a deterministic retrieval-augmented generation pipeline with section-aware relevance scoring and snowball citation expansion grounds every claim in a verifiable corpus of 60--100 papers. Second, an autonomous coding agent executes real computational biology experiments replacing synthetic outputs with genuine numerical results. Third, an eight-dimensional automated quality scorer, benchmarked with approximate reference statistics from published papers and augmented with explicit hallucination penalties, provides standardized, reproducible quality assessments. The quality-driven improvement loop uses a context-rich reviser that routes each iteration to one of three researcher actions and fires a deep research cycle every ten iterations to re-run experiments and re-manuscript from stronger outputs. We validate the system on five bioinformatics case studies; all five cases compiled submission-formatted PDFs with zero out-of-range citations. The improvement loop raises manuscript quality by an average of +17.96 points on a 0--100 scale (maximum +26.04. As partial external checks, a human reviewer scored the five manuscripts at an average of 7.0 out of 10. Complete manuscripts are produced at approximately 0.31 USD per paper.

Chinese Translation

尽管最近在大型语言模型方面的进展使得端到端的自动化手稿生成成为可能，但现有系统存在三项关键缺陷：（i）生成的论断未能以可验证的文献为基础，（ii）实验结果常常是虚构的而非实际执行的，以及（iii）缺乏标准化的多维框架来评估AI生成的手稿是否符合现实出版所需的质量和严谨性。我们提出了Prompt-to-Paper，一个多智能体框架，通过三项集成创新直接解决这一评估空白。首先，一个确定性的检索增强生成管道，结合了基于章节的相关性评分和滚雪球引用扩展，将每个论断基于60到100篇可验证文献进行支撑。其次，一个自主编码代理执行真实的计算生物学实验，将合成输出替换为真实的数值结果。第三，一个八维自动质量评分器，基于已发表论文的近似参考统计进行基准测试，并增加了明确的幻觉惩罚，提供标准化、可重复的质量评估。以质量驱动的改进循环使用一个丰富上下文的修订器，将每次迭代引导至三种研究者行动之一，并在每十次迭代中启动一次深度研究循环，以重新运行实验并从更强的输出中重新撰写手稿。我们在五个生物信息学案例研究中验证了该系统；所有五个案例均编制了格式符合提交要求的PDF文件，且没有超范围引用。改进循环使手稿质量在0到100的评分中平均提高了17.96分（最高可达26.04分）。作为部分外部检查，一位人工审稿人对五篇手稿的评分平均为7.0分（满分10分）。完整手稿的生成成本约为每篇0.31美元。

View on arXiv Download PDF AI Translation

cs.AI / 2 / 2607.05563

From Graphs to Gradients: Physics-Inspired Structural Attribution for Cyber-Physical IoT Systems and Beyond

从图到梯度：面向网络物理物联网系统及其更广泛应用的物理启发结构归因

Evangelatos, Spyridon, Diou, Christos, Papadopoulos, Georgios Th., Markakis, Evangelos, Sarigiannidis, Panagiotis

Abstract

Interpretable explanation methods in Artificial Intelligence aim to uncover the underlying causes and their effects, enabling a deeper understanding of why a system behaves in a certain way under different inputs. Unlike traditional explainability methods, which mainly highlight correlations between input and output variables, causal explanation focuses on interventional questions. By doing so, it provides more robust insights, helping users understand automated decisions, especially in high-risk domains. Recovering an explicit directed causal structure, however, is often impractical in large-scale, hybrid cyber-physical systems with feedback loops and partial observability. This paper introduces a novel framework inspired by statistical mechanics that instead models variable dependencies through an undirected, energy-based representation of cyber-physical IoT systems. Our approach enables rigorous dependency-aware attribution by analysing how variations in the energy landscape reflect the influence of individual components, without recovering a directed causal graph. It also supports reasoning about perturbation effects across hybrid interactions, providing reliable explanations of abnormal behaviours. We empirically examined our framework through simulations on an industrial IoT testbed with hybrid continuous and discrete variables, demonstrating higher attribution accuracy, improved robustness and better scalability than state-of-the-art graph-based approaches. While the attributions are not intended to fully recover the system's generative dynamics, they provide valuable, dependency-aware explanations supporting both human interpretation and downstream predictive and diagnostic tasks. Although demonstrated in industrial IoT security, our framework also applies to other high-dimensional cyber-physical and socio-technical systems requiring principled, structural explanations.

Chinese Translation

人工智能中的可解释性解释方法旨在揭示潜在原因及其影响，从而深入理解系统在不同输入下为何以特定方式表现。与传统的可解释性方法主要强调输入与输出变量之间的相关性不同，因果解释关注干预性问题。通过这种方式，它提供了更为稳健的见解，帮助用户理解自动化决策，尤其是在高风险领域。然而，在具有反馈回路和部分可观察性的规模庞大、混合的网络物理系统中，恢复明确的有向因果结构往往是不切实际的。本文提出了一种受统计力学启发的新框架，该框架通过无向的基于能量的表示来建模网络物理物联网系统中的变量依赖关系。我们的方法通过分析能量景观中的变化如何反映各个组件的影响，支持严格的依赖意识归因，而无需恢复有向因果图。它还支持对混合交互中的扰动效应进行推理，提供对异常行为的可靠解释。我们通过在具有混合连续和离散变量的工业物联网测试平台上的模拟实证检验了我们的框架，结果表明其归因准确性更高、稳健性更强且可扩展性优于最先进的基于图的方法。虽然这些归因并不旨在完全恢复系统的生成动态，但它们提供了有价值的、依赖意识的解释，支持人类的理解以及下游的预测和诊断任务。尽管在工业物联网安全中进行了验证，我们的框架同样适用于其他需要原则性结构解释的高维网络物理和社会技术系统。

View on arXiv Download PDF AI Translation

cs.AI / 3 / 2607.05571

CSTutorBench: Benchmarking Small Language Models as Tutors for Block-Based Programming

CSTutorBench：小型语言模型作为基于区块编程的辅导工具的基准评估

Lane, H. Chad, Kageler, Bryson

Abstract

Large language models are increasingly explored as AI tutors, yet deploying them in K-12 settings raises concerns around privacy, cost, and reliance on proprietary models. Small language models (SLMs) offer a promising alternative, but selecting the right model for a specific educational context remains difficult, particularly when the target domain, such as block-based programming, is largely absent from model training data. We introduce CSTutorBench, a benchmark for evaluating language models as CS tutors in VEX VR, a block-based robotics environment. The benchmark comprises 17 scenario-based questions scored against a pedagogical rubric grounded in established tutoring and feedback research, with a human-in-the-loop LLM-as-judge pipeline for evaluation. Preliminary findings across 11 models (4B-120B parameters) reveal that models perform well on surface-level criteria such as vocabulary and tone but struggle with deeper pedagogical behaviors, particularly avoiding answer leakage and engaging with student debugging histories. In our sample, model family and instruction-tuning approach appear to be better predictors of tutoring quality than parameter count alone, though the small number of models limits the strength of this conclusion. A targeted prompt revision grounded in recent educational prompt engineering research improved scores for 10 of 11 models. These results underscore the value of context-specific, pedagogically grounded benchmarks for SLM selection in educational deployment.

Chinese Translation

大型语言模型越来越多地被探索作为人工智能辅导工具，但在K-12环境中部署它们引发了关于隐私、成本和对专有模型依赖的担忧。小型语言模型（SLMs）提供了一个有前景的替代方案，但在特定教育环境中选择合适的模型仍然困难，尤其是当目标领域（如基于区块的编程）在模型训练数据中几乎缺失时。我们介绍了CSTutorBench，这是一个用于评估语言模型作为计算机科学（CS）辅导工具的基准，应用于VEX VR，一个基于区块的机器人环境。该基准包含17个基于场景的问题，依据建立的辅导和反馈研究的教学标准进行评分，并采用人机协作的LLM作为评判者的评估流程。对11个模型（4B-120B参数）的初步发现表明，模型在词汇和语调等表层标准上表现良好，但在更深层次的教学行为上表现不佳，特别是在避免答案泄露和与学生调试历史互动方面。在我们的样本中，模型家族和指令调优方法似乎比参数数量更能预测辅导质量，尽管模型数量较少限制了这一结论的力度。基于近期教育提示工程研究的有针对性的提示修订提高了11个模型中10个的评分。这些结果强调了在教育部署中选择SLM时，基于特定上下文和教学基础的基准的重要性。

View on arXiv Download PDF AI Translation

cs.AI / 4 / 2607.05573

Foundation Models for Automatic CAD Generation

自动化CAD生成的基础模型

de Curtò, J, Guillén, Victoria, de Zarzà, I.

Abstract

Recent advances in Large Language Models (LLMs) and Vision-Language Models (VLMs) enable the automatic generation of parametric 3D designs from natural-language specifications. This chapter presents an empirical study of foundation models for automatic Computer-Aided Design (CAD) generation of mechanical parts, using a unified evaluation pipeline and a curated benchmark of 97 engineering design problems. We introduce LLMForge, a multi-model text-to-CAD framework integrating JSON-schema validation, analytic feature scoring, mesh synthesis, and multi-round iterative refinement, studied under two critique regimes. IterTracer uses a Phong-shaded ray-trace renderer with analytic visual metrics (silhouette IoU, hole visibility, edge clearance, aspect-ratio conformance) for lightweight geometry-aware feedback across rounds. IterVision replaces the analytic scorer with a VLM semantic critic (Qwen2.5-VL-72B) that evaluates rendered views via chain-of-thought visual reasoning, assessing spatial coherence and design intent. On a benchmark spanning four canonical geometry families (plates with holes and bolt circles, multi-feature boxes, flanged cylinders, and L-brackets), we evaluate seven foundation models: DeepSeek-V3.2, Qwen3-235B-A22B, Llama-3.3-70B, Gemma-3-27B, GLM-4.5, MiniMax-M2.1, and INTELLECT. Under IterTracer, the four highest-ranked models form a tight cluster (overall mean in [0.885, 0.890]) with 98.97% mesh success, showing that compact instruction-tuned models can match substantially larger systems. VLM-based critique in IterVision yields 100% watertight mesh generation on the leading model while surfacing systematic difficulty on rotationally symmetric geometries such as cylinders, where visual and semantic scoring diverge most. We discuss benchmark design, failure modes, CAD-oriented prompting, and implications for industrial workflows and scalable automated mechanical design.

Chinese Translation

最近大型语言模型（Large Language Models, LLMs）和视觉-语言模型（Vision-Language Models, VLMs）的进展使得能够根据自然语言规范自动生成参数化的三维设计。本章介绍了一项针对机械零件自动计算机辅助设计（Computer-Aided Design, CAD）生成的基础模型的实证研究，采用统一的评估流程和经过精心策划的97个工程设计问题的基准测试。我们引入了LLMForge，一个集成了JSON-schema验证、分析特征评分、网格合成和多轮迭代细化的多模型文本到CAD框架，该框架在两种评价机制下进行研究。IterTracer使用Phong阴影光线追踪渲染器，结合分析视觉指标（轮廓IoU、孔的可见性、边缘间隙、纵横比一致性），在不同轮次中提供轻量的几何感知反馈。IterVision则用VLM语义评价器（Qwen2.5-VL-72B）替代了分析评分器，通过链式思维视觉推理评估渲染视图，考量空间连贯性和设计意图。在涵盖四种经典几何类型（带孔和螺栓圆的板、多特征箱体、法兰圆柱和L型支架）的基准测试中，我们评估了七个基础模型：DeepSeek-V3.2、Qwen3-235B-A22B、Llama-3.3-70B、Gemma-3-27B、GLM-4.5、MiniMax-M2.1和INTELLECT。在IterTracer下，四个排名最高的模型形成一个紧密的聚类（总体均值在[0.885, 0.890]之间），具有98.97%的网格成功率，表明经过紧凑指令调优的模型可以与大得多的系统相匹配。在IterVision中的基于VLM的评估使得领先模型的网格生成达到了100%的防水效果，同时在旋转对称几何体（如圆柱体）上暴露出系统性困难，其中视觉和语义评分的结果最为不一致。我们讨论了基准设计、失败模式、面向CAD的提示以及对工业工作流和可扩展自动机械设计的影响。

View on arXiv Download PDF AI Translation

cs.AI / 5 / 2607.05577

Narrative World Model: Narratology-Grounded Writer Memory for Long-Form Fiction

叙事世界模型：基于叙事学的长期小说作家记忆

Saifullah, Mohammad, Kornmaier, Thomas, Kazi, Taaha, Sharma, Vasu, Kanade, Aditya Sanjiv, Yadav, Aanand Kumar

Abstract

Long-form fiction writers need memory that answers multi-hop questions about evolving story state: who knows a secret and when they learned it, whether an event preceded the narration that revealed it, whether a setup paid off, and how a relationship shifted. General-purpose retrieval and agent-memory systems represent entities and facts but not the narratological structure these questions turn on, so they surface the wrong evidence or none at all. We introduce the Narrative World Model (NWM), a writer-memory system that pairs a narratology-grounded typed temporal-state graph with query-conditioned hybrid retrieval. To measure memory rather than the answerer, we read every system through a single held-constant Opus 4.8 reader over only that system's chapter-safe evidence, on a reproducible public corpus and a validated multi-hop benchmark, and we compare against the strongest existing temporal-knowledge-graph agent-memory framework, Graphiti/Zep (Rasmussen et al., 2025). NWM substantially and significantly outperforms this baseline on multi-hop narratological QA across both corpora, and far exceeds GraphRAG and flat retrieval. The advantage is representational rather than an artifact of extraction: it survives rebuilding the baseline with NWM's own extractor, and traces to its narratology-grounded structure and query-conditioned retrieval, not to graph size or extractor quality.

Chinese Translation

长期小说作家需要一种记忆，能够回答关于不断演变的故事状态的多跳问题：谁知道一个秘密以及他们何时得知，某个事件是否在揭示秘密的叙述之前发生，某个设定是否得到了回报，以及关系是如何变化的。通用检索和智能体记忆系统虽然能够表示实体和事实，但未能体现这些问题所依赖的叙事结构，因此它们提供了错误的证据或根本没有证据。我们提出了叙事世界模型（Narrative World Model, NWM），这是一种作家记忆系统，它将基于叙事学的类型化时间状态图与查询条件混合检索相结合。为了测量记忆而非回答者，我们通过一个恒定的 Opus 4.8 阅读器来评估每个系统，仅在该系统的章节安全证据上，并使用可重复的公共语料库和经过验证的多跳基准，与现有最强的时间知识图智能体记忆框架 Graphiti/Zep (Rasmussen et al., 2025) 进行比较。NWM 在两个语料库的多跳叙事学问答中显著而显著地超过了这一基线，并远远超出 GraphRAG 和扁平检索。这一优势是表征性的，而不是提取的伪影：无论是使用 NWM 自身的提取器重建基线，其效果依旧卓越，其优势归因于其基于叙事学的结构和查询条件检索，而非图的规模或提取器的质量。

View on arXiv Download PDF AI Translation

cs.AI / 6 / 2607.05682

FirstResearch: Auditable Question Formation for LLM Scientific Discovery Agents

FirstResearch：大语言模型科学发现代理的可审计问题形成

Wang, Yufeng

Abstract

LLM systems for scientific discovery increasingly assist with ideation, literature synthesis, experiment planning, and report generation, but the first research question they propose can remain difficult to audit: it may sound plausible without exposing the mechanism, falsifier, or assumption that a scientist should inspect. We introduce FirstResearch, a first-principles research-question formation framework for scientific LLM agents whose core artifact is a structured Research Question Certificate. The certificate records primitive definitions, assumptions, a mechanism model, a tension or contradiction, a falsifiable hypothesis, a minimal decisive test, and a failure update rule, making the proposed question inspectable before downstream execution. On ten LLM-agent research topics, FirstResearch outperforms controlled prompt-level baselines inspired by AI co-scientist, Agent Laboratory, and AI Scientist-v2 under a primary DeepSeek-blind-judge protocol. A Gemini-2.5-Flash independent-judge rescore of the same 40 baseline packages preserves the system-level ranking, with FirstResearch scoring 4.86/5 versus 4.38/5 for the strongest baseline and Pearson agreement of 0.865 on average score. A one-repeat ablation checkpoint further suggests that the certificate-centered core is the strongest component: certificate-only scoring reaches 4.90/5 under DeepSeek and 4.88/5 under Gemini, while removing certificates drops below 1/5 under both judges. These results are preliminary and use LLM judges rather than human domain experts, but they support a narrow scientific-discovery claim: explicit derivation constraints are a promising mechanism for making LLM-generated scientific questions more auditable. Code, prompts, saved outputs, and reproduction scripts are available at https://github.com/louiswang524/FirstResearch.

Chinese Translation

大语言模型（LLM）系统在科学发现中越来越多地协助构思、文献综合、实验计划和报告生成，但它们提出的第一个研究问题往往难以审计：这些问题可能听起来合理，但未能揭示科学家应当检查的机制、反驳者或假设。我们提出了FirstResearch，这是一种基于第一原理的科学LLM代理研究问题形成框架，其核心产物是结构化的研究问题证书。该证书记录了基本定义、假设、机制模型、张力或矛盾、可反驳的假设、最小决定性测试和失败更新规则，使得所提问题在下游执行之前可供审查。在十个LLM代理研究主题上，FirstResearch在主要的DeepSeek盲评协议下表现优于受AI共同科学家、Agent Laboratory和AI Scientist-v2启发的受控提示级基线。Gemini-2.5-Flash独立评审对这40个基线包的重新评分保留了系统级排名，其中FirstResearch得分为4.86/5，而最强基线得分为4.38/5，平均得分的Pearson一致性为0.865。一次重复消融检查进一步表明，以证书为中心的核心是最强组成部分：在DeepSeek下，只有证书的得分达到4.90/5，而在Gemini下为4.88/5，而去除证书的得分在两个评审下都降至1/5以下。这些结果是初步的，使用的是LLM评审而非人类领域专家，但它们支持一个狭义的科学发现主张：显式的推导约束是使LLM生成的科学问题更具可审计性的有希望的机制。代码、提示、保存的输出及复现脚本可在 https://github.com/louiswang524/FirstResearch 上获取。

View on arXiv Download PDF AI Translation

cs.AI / 7 / 2607.05690

Memory in the Loop: In-Process Retrieval as ExtendedWorking Memory for Language Agents

循环中的记忆：过程内检索作为语言代理的扩展工作记忆

Khan, Yusuf, Lipizzi, Carlo

Abstract

Language agents run a loop - observe, reason, act - but the memory they reason over sits outside it: a store queried at most once per turn. We study the regime where memory moves inside the loop, read and written on every step. The obstacle has always been latency: networked stores answer in tens to hundreds of milliseconds, and in-loop retrieval can inflate end-to-end latency by up to 83x when retrieval is expensive. Prior work manages that cost rather than questioning it: serving-layer scheduling hides it, "memory-first" designs ration retrieval to once per turn. We argue latency is a property of where the store lives, not the in-loop pattern: an in-process store answers in ~100us, three orders of magnitude below the network regime, and at that speed the per-step tax collapses. By the extended-mind thesis's parity principle, a store fast enough to be constantly and directly available becomes extended working memory, not a tool the agent merely consults. The premise is causal: holding a fixed per-turn memory-latency budget and varying only the store's answer speed, redundant actions rise monotonically with latency - 0.0 of 12 at in-process speed, 7.2 of 12 at a 110ms cloud round trip (gpt-5-nano, gpt-5-mini; exact permutation p=0.0079). We demonstrate the regime end-to-end: across four GPT-5-class models under a bounded window, recall improves from 0/5 to 3.6-4.8/5 with in-loop memory, store ops at p50 80-165us - though an instructed restate-every-reply baseline also solves it perfectly, at a token cost that grows with the working set. The store never lost a fact in any run (244 of 244 writes kept); every miss traces to the agent's read policy, not the store. Our measurements also relocate the bottleneck: the dominant per-step cost is embedding (~200-400ms over the network); pairing the in-process store with a small local embedder returns the complete operation to a measured ~40us.

Chinese Translation

语言代理运行一个循环——观察、推理、行动——但它们推理所依据的记忆位于循环之外：每个回合最多查询一次的存储。我们研究了记忆移动到循环内部的情况，在每一步都进行读写。障碍一直是延迟：网络存储的响应时间在几十到几百毫秒之间，而在循环内的检索在检索成本高时可能使端到端延迟增加多达83倍。之前的研究管理这一成本而非质疑它：服务层调度隐藏了这一点，“优先记忆”设计将检索限制为每回合一次。我们认为，延迟是存储位置的属性，而不是循环模式的属性：一个过程内存储的响应时间约为100微秒，低于网络环境三个数量级，在这个速度下，每步的成本骤降。根据扩展心智理论的平价原则，足够快以便始终直接可用的存储成为扩展工作记忆，而不是代理仅仅咨询的工具。这个前提是因果的：保持固定的每回合记忆延迟预算，仅变化存储的响应速度，冗余操作随着延迟单调增加——在过程内速度下为0.0/12，在110毫秒的云往返时为7.2/12（gpt-5-nano，gpt-5-mini；确切排列p=0.0079）。我们展示了这一机制的端到端效果：在一个有限窗口下，四个GPT-5级模型的回忆从0/5提高到3.6-4.8/5，过程内存储操作的p50为80-165微秒——尽管一个指令性重述每个回复的基线也完美解决了这个问题，但其代价随着工作集的增长而增加。存储在任何运行中都没有丢失任何事实（244次写入中保留了244次）；每次未命中都追溯到代理的读取策略，而不是存储。我们的测量还重新定位了瓶颈：主导的每步成本是嵌入（网络延迟约为200-400毫秒）；将过程内存储与小型本地嵌入器配对，使整个操作的测量时间恢复到约40微秒。

View on arXiv Download PDF AI Translation

cs.AI / 8 / 2607.05708

Akashic: A Low-Overhead LLM Inference Service with MemAttention

Akashic：一种低开销的 LLM 推理服务，基于 MemAttention

Liu, Yang, Luo, Zhaokai, Jin, Huayi, He, Ruozhou, Hong, Chenchen, Wang, Zhiyong, Liu, Yifei, Gu, Yunfei, Wu, Chentao, Hu, Junhao

Abstract

Recent LLM-based agent systems continuously accumulate context across multi-turn interactions, tool invocations, and cross-session workflows. Replaying the full history for every request quickly becomes impractical: long contexts increase prefill cost, may exceed context limits, and often bury task-relevant evidence in irrelevant content, degrading both serving efficiency and output quality. We propose Akashic, a low-overhead memory system built around MemAttention, which organizes context into bounded chunks and models semantic relationships across chunks, preserving cross-chunk evidence without repeatedly rewriting the full history. Akashic further applies hardware-software co-designed memory placement to co-locate likely co-retrieved chunks, reducing retrieval fragmentation and I/O overhead. Across four representative workloads and three model sizes, Akashic improves task accuracy by up to 10.2 points, throughput by up to 1.21x, and sustainable request rate by up to 1.88x over strong prior memory baselines.

Chinese Translation

近期基于 LLM 的智能体系统在多轮交互、工具调用和跨会话工作流中不断积累上下文。对于每个请求重放完整历史记录很快变得不切实际：长上下文增加了预填充成本，可能超过上下文限制，并且常常将与任务相关的证据埋没在无关内容中，从而降低服务效率和输出质量。我们提出了 Akashic，一种围绕 MemAttention 构建的低开销内存系统，它将上下文组织成有限的块，并建模块之间的语义关系，保留跨块证据而无需重复重写完整历史。Akashic 进一步应用硬件-软件协同设计的内存放置，将可能共同检索的块共同定位，从而减少检索碎片和 I/O 开销。在四个代表性工作负载和三种模型规模下，Akashic 在任务准确性上提高了最多 10.2 个百分点，吞吐量提高了最多 1.21 倍，持续请求率提高了最多 1.88 倍，相较于强大的先前内存基线。

View on arXiv Download PDF AI Translation

cs.AI / 9 / 2607.05750

ArtisanCAD: An Industrial-Level CAD Agent with Expert-Grounded Knowledge Distillation

ArtisanCAD：具有专家基础知识蒸馏的工业级CAD代理

Xu, Yunhan, Wu, Qifeng, Li, Xunjin, Bin, Yuanwei, Yao, Qingsong, Gu, Jianghang, Wang, Guan, Lv, Weihao, Yang, Huiyu, Luo, Wenfa, Xiang, Jiao, Chen, Yuntian, Chen, Shiyi

Abstract

Computer-aided design (CAD) for industrial components requires long-horizon procedural modeling, robust feature dependencies, editable parametric geometry, and production-grade B-Rep execution. Existing text-to-CAD methods have made promising progress in generating CAD programs from natural-language descriptions, but they still struggle when user prompts are ambiguous, underspecified, or only describe high-level design intent. They also rarely exploit expert procedural knowledge naturally available in industrial workflows, such as CATIA operation recordings, macro logs, drawing notes, and engineering descriptions. We present \algname, a skill-guided industrial CAD agent with expert-grounded knowledge distillation. The core of \algname is CAD intermediate representation (CAD-IR), an executable procedural representation that encodes parameters, ordered operations, MCP tool bindings, dependencies, generated entities, and verification rules. CAD-IR plays two key roles: it first serves as the carrier for distilling expert CAD procedures into reusable parameterized skills; then it provides a procedural scaffold that turns vague or intermediate-level prompts into complete executable CAD operations. \algname retrieves expert-derived skills, instantiates and revises CAD-IR, executes the resulting procedure through a dedicated CATIA-MCP backend, and uses multi-view visual feedback for iterative refinement, and finally generates production-ready B-Rep models. On the Text2CAD benchmark, CAD-IR improves generation from intermediate prompts by reducing mean Chamfer Distance from $14.83$ to $9.88$, showing its ability to bridge ambiguous textual intent and executable CAD construction. On four complex automotive components, CAD-IR enables expert CATIA recordings to be distilled into reusable skills, allowing \algname to generate editable CATIA-native B-Rep models for new variant requests.

Chinese Translation

工业组件的计算机辅助设计（CAD）需要长时间的程序建模、稳健的特征依赖、可编辑的参数化几何形状以及生产级的B-Rep执行。现有的文本到CAD方法在从自然语言描述生成CAD程序方面取得了可喜的进展，但在用户提示模糊、不明确或仅描述高层设计意图时仍然面临挑战。此外，它们很少利用在工业工作流程中自然存在的专家程序知识，如CATIA操作记录、宏日志、绘图注释和工程描述。我们提出了 extit{ArtisanCAD}，一个具有专家基础知识蒸馏的技能引导工业CAD代理。 extit{ArtisanCAD}的核心是CAD中间表示（CAD-IR），这是一种可执行的程序表示，编码了参数、顺序操作、MCP工具绑定、依赖关系、生成实体和验证规则。CAD-IR扮演两个关键角色：首先，它作为蒸馏专家CAD程序到可重用参数化技能的载体；然后，它提供一个程序框架，将模糊或中间级提示转化为完整的可执行CAD操作。 extit{ArtisanCAD}检索专家派生技能，实例化并修订CAD-IR，通过专用的CATIA-MCP后端执行生成的程序，并利用多视角视觉反馈进行迭代优化，最终生成生产就绪的B-Rep模型。在Text2CAD基准测试中，CAD-IR通过将平均Chamfer距离从$14.83$降低到$9.88$，改善了从中间提示的生成，显示了其在模糊文本意图与可执行CAD构建之间架起桥梁的能力。在四个复杂的汽车组件上，CAD-IR使专家CATIA记录能够被蒸馏为可重用的技能，从而允许 extit{ArtisanCAD}为新的变体请求生成可编辑的CATIA原生B-Rep模型。

View on arXiv Download PDF AI Translation

cs.AI / 10 / 2607.05761

Synthetic Consumer Insight Generation with Large Language Models

利用大型语言模型生成合成消费者洞察

France, Stephen L., Albinsson, Pia. A.

Abstract

Modern data-driven marketing relies on large amounts of consumer data, yet collecting such data can be costly, time-consuming, and difficult to scale. This research examines whether large language models (LLMs) can be used to generate synthetic consumer data for projective techniques, a set of methods designed to elicit consumer associations, emotions, wants, and needs. We test LLM-generated responses across multiple projective tasks, LLMs, prompting strategies, and temperature settings, and compare them with human responses from a primary research study on perceptions of city tourism destinations. Human and LLM responses were analyzed using linguistic measures, diversity and concentration metrics, topic models, and top-term analyses. The results show substantial overlap between human and LLM responses in broad topics and associations, but also important differences in style, linguistic structure, and the way diversity is generated. Recommendations are given on how to best utilize LLMs for generating synthetic consumer data, how model and prompt choices shape response quality, and on recognizing the limitations of LLM synthetic consumer data generation.

Chinese Translation

现代数据驱动的营销依赖于大量的消费者数据，但收集这些数据可能成本高昂、耗时且难以扩展。本研究考察了大型语言模型（LLMs）是否可以用于生成合成消费者数据，以支持投射技术，这是一组旨在引发消费者联想、情感、愿望和需求的方法。我们在多个投射任务、LLMs、提示策略和温度设置下测试了LLM生成的响应，并将其与一项关于城市旅游目的地认知的主要研究中的人类响应进行比较。通过语言学测量、多样性和集中度指标、主题模型以及关键词分析，对人类和LLM的响应进行了分析。结果显示，人类和LLM响应在广泛主题和联想上有显著重叠，但在风格、语言结构以及多样性生成方式上也存在重要差异。我们提供了关于如何最佳利用LLMs生成合成消费者数据的建议，如何选择模型和提示影响响应质量，以及如何认识LLM合成消费者数据生成的局限性。

View on arXiv Download PDF AI Translation

cs.AI / 11 / 2607.05773

Beyond Static Evaluation: Building Simulation Environments for Scalable Agentic Reinforcement Learning

超越静态评估：构建可扩展的自主强化学习模拟环境

Arora, Akshay, Nigam, Ishan, Aggarwal, Ashutosh, Bansal, Shefali, Singh, Krishna, Kumari, Sweta, Mittal, Nikhil, Farhan, Shariq, Malreddy, Siddarth

Abstract

As Large Language Models (LLMs) evolve into autonomous agents, traditional static evaluation fails to capture multi-step decision-making. We introduce AgenticAI-Supervisor, an API and UI-driven RL Gym environment that decouples environment creation from scalable execution. By moving to verifiable execution outcomes, the platform generates high-fidelity traces and applies multi-dimensional reward shaping. Critically, our framework mitigates reward hacking through rigorous internal state validation and testing. This work provides a first look at our platform's core capabilities through a Customer Support Agent case study demonstrating a consistent closed-loop feedback for model optimization. Future work will focus on advanced features such as Computer Use, Tool Use, automated "stumping", and edge-case generation.

Chinese Translation

随着大型语言模型（LLMs）演变为自主智能体，传统的静态评估无法捕捉多步骤决策过程。我们引入了AgenticAI-Supervisor，一个基于API和UI驱动的强化学习（RL）Gym环境，它将环境创建与可扩展执行解耦。通过转向可验证的执行结果，该平台生成高保真度的追踪记录，并应用多维度的奖励塑形。关键是，我们的框架通过严格的内部状态验证和测试来减轻奖励黑客问题。本研究通过一个客户支持智能体的案例研究首次展示了我们平台的核心能力，证明了模型优化的一致闭环反馈。未来的工作将集中在高级功能上，例如计算机使用、工具使用、自动“难倒”以及边缘案例生成。

View on arXiv Download PDF AI Translation

cs.AI / 12 / 2607.05775

Beyond the Leaderboard: A Synthesis of Tool-Use, Planning, and Reasoning Failures in Large Language Model Agents

超越排行榜：大型语言模型代理在工具使用、规划和推理失败方面的综合研究

Albayaydh, Wael, Zhao, Rui, Flechais, Ivan

Abstract

Large language model (LLM) agents are increasingly evaluated on their ability to use tools, plan multi-step tasks, coordinate with other agents, and operate over extended horizons. Reported benchmark gains often obscure recurring failure modes documented across otherwise unrelated evaluation efforts. This paper synthesizes 27 benchmark, taxonomy, and audit papers (2023-2026), spanning 19 distinct benchmarks, into a cross-cutting taxonomy of agent limitations. To our knowledge, this is the first synthesis that integrates evidence across tool use, planning, long-horizon reasoning, multi-agent coordination, safety, and measurement validity into a single, unified taxonomy of LLM agent limitations. We identify six failure clusters: (1) tool invocation and parameter-level errors, (2) planning and constraint-satisfaction failures, (3) long-horizon degradation from context accumulation, (4) multi-agent coordination failures, (5) safety and security failures under adversarial or underspecified conditions, and (6) measurement validity problems. The taxonomy was derived iteratively by grouping independently reported error categories into themes corresponding to distinct stages of the agent reasoning-to-action pipeline. Across the literature, we find that failures compound nonlinearly with task length, that strong performance on individual sub-tasks does not reliably translate into end-to-end success, and that additional scaffolding does not consistently improve reliability. At the same time, substantial progress has been demonstrated in single-turn tool use, short-horizon web navigation, and narrowly scoped coding tasks.

Chinese Translation

大型语言模型（LLM）代理的评估越来越侧重于其使用工具、规划多步骤任务、与其他代理协调以及在较长时间范围内操作的能力。报告的基准提升往往掩盖了在其他无关评估工作中记录的反复出现的失败模式。本文综合了27篇基准、分类和审计论文（2023-2026），涵盖19个不同的基准，形成了一个跨领域的代理局限性分类法。我们所知，这是首次将工具使用、规划、长时间推理、多代理协调、安全性和测量有效性等证据整合到一个统一的LLM代理局限性分类法中的研究。我们识别出六个失败集群：（1）工具调用和参数级错误，（2）规划和约束满足失败，（3）由于上下文积累导致的长时间退化，（4）多代理协调失败，（5）在对抗性或不明确条件下的安全性和安全失败，以及（6）测量有效性问题。该分类法是通过将独立报告的错误类别分组为对应于代理推理到行动管道不同阶段的主题而迭代得出的。在文献中，我们发现失败与任务长度呈非线性复合，单个子任务的强表现并不可靠地转化为端到端的成功，并且额外的支撑并不总能提高可靠性。同时，在单轮工具使用、短时间范围内的网页导航和狭窄范围的编码任务中已展示出显著进展。

View on arXiv Download PDF AI Translation

cs.AI / 13 / 2607.05790

Controlling Tool Use with Heading-Specific Activation Steering

通过特定方向激活引导控制工具使用

Chen, Yuqi, Siu, Vincent, Liu, Yang, Song, Dawn, Wang, Chenguang

Abstract

Tool-augmented large language models extend their capabilities beyond parametric knowledge through external tools, but tend to invoke them unnecessarily. We investigate whether tool-use decisions have any stable internal representation that can be extracted and manipulated, a question that is non-trivial given that tools exist entirely in context at inference time and have no direct encoding in model weights. We show that steering vectors extracted from heading-anchors positions exert bidirectional causal control over tool-invocation behavior across five open-source models and three domains, suppressing unnecessary tool use most effectively in domains where parametric reasoning suffices. However, geometric analysis reveals that this causal effectiveness does not correspond to clean linear structure: tool-invocation steps exhibit diffuse, bimodal alignment with the suppression vector rather than the consistent negative alignment a linear encoding account would predict, and different tool types recruit largely distinct internal signatures with low cross-tool feature overlap. We hypothesize these geometric properties are indicative of the non-parametric nature of tools, and distinguish tool-use steering vectors from those extracted for parametrically grounded concepts. The relationship between this geometric irregularity and the observed causal effectiveness remains an open question.

Chinese Translation

工具增强的大型语言模型通过外部工具扩展其能力，超越了参数知识，但往往会不必要地调用这些工具。我们研究工具使用决策是否具有可提取和可操控的稳定内部表征，考虑到工具在推理时完全依赖上下文存在，并在模型权重中没有直接编码，这一问题并非易事。我们展示了从方向锚点位置提取的引导向量对工具调用行为施加了双向因果控制，这一效果在五个开源模型和三个领域中一致出现，尤其在参数推理足以满足需求的领域中，有效地抑制了不必要的工具使用。然而，几何分析表明，这种因果有效性并不对应于干净的线性结构：工具调用步骤与抑制向量之间表现出分散的双峰对齐，而不是线性编码理论所预测的一致负对齐，并且不同工具类型招募的内部特征签名大相径庭，交叉工具特征重叠度较低。我们假设这些几何特性表明工具的非参数性质，并将工具使用引导向量与为参数化概念提取的向量区分开来。这种几何不规则性与观察到的因果有效性之间的关系仍然是一个未解之谜。

View on arXiv Download PDF AI Translation

cs.AI / 14 / 2607.05794

From Passive Retrieval to Active Memory Navigation: Learning to Use Memory as a Structured Action Space

从被动检索到主动记忆导航：学习将记忆作为结构化行动空间

Xu, Yue, Sun, Yutao, Liu, Yihao, Zhou, Mengyu, Qiao, Jiayi, Ma, Lu, Tang, Kai, Wang, Wenjie, Jiang, Xiaoxi, Jiang, Guanjun

Abstract

Long-term user memory is essential for personalized conversational agents, yet many memory systems still expose memory through passive retrieval interfaces, making the model a consumer of pre-selected evidence. We introduce NapMem, a framework for learning to use long-term user memory as a structured action space rather than passively retrieved context. NapMem organizes user history into a linked multi-granularity memory pyramid, where raw conversations, typed memory records, topic tracks, and user profiles are connected through provenance relations, and exposes these levels through memory tools. The agent is trained to select memory according to the query and intermediate evidence, allowing it to inspect different memory granularities before answering. Experiments on PersonaMem-v2, LongMemEval, and LoCoMo show that a NapMem agent trained with memory-tool reinforcement learning is competitive across diverse memory-intensive tasks, while evaluations on non-memory tasks suggest that the learned policy largely preserves general reasoning and tool-use abilities. Additional analyses examine storage, inference cost, tool-use behavior, and ablations over navigation, memory granularity, and RL training. Our results suggest that long-term user memory benefits from coupling structured storage with a learned policy for using memory at the appropriate granularity.

Chinese Translation

长期用户记忆对个性化对话代理至关重要，然而许多记忆系统仍通过被动检索接口暴露记忆，使得模型成为预先选择证据的消费者。我们提出了NapMem，一个框架，用于学习将长期用户记忆作为结构化行动空间，而非被动检索的上下文。NapMem将用户历史组织成一个链接的多粒度记忆金字塔，其中原始对话、键入的记忆记录、主题轨迹和用户档案通过来源关系连接，并通过记忆工具暴露这些层次。代理模型被训练根据查询和中间证据选择记忆，从而使其能够在回答之前检查不同的记忆粒度。在PersonaMem-v2、LongMemEval和LoCoMo上进行的实验表明，通过记忆工具强化学习训练的NapMem代理在各种记忆密集任务中具有竞争力，而在非记忆任务上的评估显示，学习到的策略在很大程度上保留了通用推理和工具使用能力。额外分析考察了存储、推理成本、工具使用行为，以及在导航、记忆粒度和强化学习训练上的消融实验。我们的结果表明，长期用户记忆通过将结构化存储与学习策略结合，以适当的粒度使用记忆，从而受益。

View on arXiv Download PDF AI Translation

cs.AI / 15 / 2607.05804

TurnOPD: Making On-Policy Distillation Turn-Aware for Efficient Long-Horizon Agent Training

TurnOPD：使得策略蒸馏关注转弯，提高长时代理训练的效率

Zhou, Yuhang, Zheng, Kai, Li, Haoling, Peng, Dengyun, Xu, Can, Chen, Jingjing

Abstract

On-policy distillation (OPD) trains a student policy by matching a stronger teacher on the student's own trajectories, offering a promising framework for language agent training. However, its application to long-horizon agentic tasks remains insufficiently explored. We identify two key inefficiencies in vanilla agent OPD: (1) full-horizon rollouts often waste wall-clock resources on tail turns that provide weak and noisy KL supervision, and (2) trajectory-level KL objectives concentrate most of the loss on shallow tokens, leaving deeper decision turns under-trained once initial behaviors are aligned. To address these challenges, we propose TurnOPD, a turn-level budgeting strategy for efficient on-policy distillation of long-horizon agents. TurnOPD consists of two budget controllers: adaptive rollout-depth budgeting, which uses probe-based turn statistics to determine rollout length, and progressive turn-normalized loss budgeting, which gradually shifts KL weighting from token-level to turn-balanced supervision. Experiments on ALFWorld, WebShop, and Multi-Hop Search with task-specialized teacher models show that TurnOPD achieves superior validation accuracy under equal wall-clock training budgets and advances the accuracy--time frontier beyond vanilla OPD.

Chinese Translation

策略蒸馏（OPD）通过在学生自身的轨迹上与更强的教师策略匹配来训练学生策略，为语言代理训练提供了一个有前景的框架。然而，它在长时代理任务中的应用仍然不足。我们识别出传统代理OPD的两个主要低效之处：（1）全时段滚动在尾部转弯上通常浪费了实际计算资源，这些尾部转弯提供了弱且噪声较大的KL监督，和（2）轨迹级的KL目标将大部分损失集中在浅层标记上，一旦初始行为对齐，深层决策转弯得不到充分训练。为了解决这些挑战，我们提出了TurnOPD，这是一种针对长时代理高效策略蒸馏的转弯级预算策略。TurnOPD由两个预算控制器组成：自适应滚动深度预算，它利用基于探测的转弯统计信息来确定滚动长度，以及逐步转弯归一化损失预算，它逐渐将KL加权从标记级转移到转弯平衡监督。针对ALFWorld、WebShop和多跳搜索的任务专用教师模型的实验表明，TurnOPD在等墙钟训练预算下实现了更高的验证准确率，并推动了准确性与时间的前沿，超越了传统的OPD。

View on arXiv Download PDF AI Translation

cs.AI / 16 / 2607.05805

Onnes: A Physics-Grounded Multi-Agent LLM Simulator for Cryogenic Fault Diagnosis in Quantum Computing Infrastructure

Onnes：基于物理的多智能体大语言模型模拟器用于量子计算基础设施中的低温故障诊断

Narisetty, Praneeth, Kattamanchi, Uday Kumar Reddy, Kore, Shiva Nagendra Babu

Abstract

Dilution refrigerators are the enabling infrastructure of superconducting quantum computers, yet their fault diagnosis is still dominated by threshold alarms that report that something is wrong, not what. We present Onnes, a physics-grounded digital-twin simulator of a dilution refrigerator (a forward physics model with a learned real-fridge noise fingerprint) that drives a live multi-agent LLM operations layer, and use it for a controlled head-to-head between a zero-shot LLM agent panel and a supervised ML classifier on cryogenic fault diagnosis. The twin couples a real dilution-cooling floor, a noise-and-correlation fingerprint learned from real BlueFors logs, and six physics-grounded fault classes, three engineered to overlap on temperature but separate on flow and pressure. Across a 1000-turn evaluation the zero-shot panel shows no significant difference from the classifier on detection but trails on classification, its errors concentrating on the confusable faults. Curated contrastive few-shot demonstrations and self-consistency voting then raise classification accuracy from 0.685 to 0.990, matching the supervised classifier (0.985) with no parameter updates and six labeled demonstrations; an ablation attributes the gain almost entirely to the demonstrations. Run as a continuous monitor across a nine-run fault-by-seed sweep, the agent catches every developing fault within one poll interval, and a confidence gate suppresses pre-onset false alarms whose rate is backend-dependent. As a first sim-to-real check, a detector trained purely on real BlueFors telemetry posts a real-hardware false-alarm rate of 6.4% and 100% recall on physics faults injected onto real held-out windows. All numbers are drawn verbatim from released run logs.

Chinese Translation

稀释制冷机是超导量子计算机的基础设施，其故障诊断仍然主要依赖于阈值警报，报告存在问题但未能明确具体问题。我们提出了Onnes，一种基于物理的稀释制冷机数字双胞胎模拟器（具有学习的真实制冷机噪声指纹的前向物理模型），该模拟器驱动着一个实时的多智能体大语言模型操作层，并用于进行零样本大语言模型代理面板与监督机器学习分类器在低温故障诊断中的对抗评估。该双胞胎模型结合了真实的稀释冷却底座、从真实BlueFors日志中学习的噪声和相关性指纹，以及六个基于物理的故障类别，其中三个故障类别在温度上重叠但在流量和压力上分开。在1000次评估中，零样本面板在故障检测方面与分类器没有显著差异，但在分类上落后，错误主要集中在容易混淆的故障上。经过策划的对比少样本演示和自一致性投票后，分类准确率从0.685提升至0.990，达到监督分类器（0.985）的水平，而参数没有更新，仅利用了六个标注演示；消融实验将这一增益几乎完全归因于这些演示。作为一个连续监测系统，在九次基于种子的故障筛查中，该代理能够在每个轮询间隔内捕捉到每一个正在发展的故障，并且一个置信门抑制了前期发生的虚假警报，这一率依赖于后端。作为首次从模拟到现实的验证，仅基于真实BlueFors遥测训练的探测器在针对真实未使用窗口上注入的物理故障时，其虚假警报率为6.4%，而召回率为100%。所有数据均直接引用已发布的运行日志。

View on arXiv Download PDF AI Translation

cs.AI / 17 / 2607.05844

StateFuse: Deterministic Conflict-Preserving Memory for Multi-Agent Systems

StateFuse：用于多智能体系统的确定性冲突保留内存

Volkov, Sergey, Li, Yang, Luo, Ye

Abstract

Agent systems accumulate conflicting observations across branches, retries, and replicas, yet many practical memory layers still collapse disagreement behind overwrite rules that are difficult to inspect or correct. We present StateFuse, a conflict-aware replicated memory contract built on standard OpSet/CRDT merge. StateFuse does not introduce a new join algebra; it defines an agent-facing semantics layer with immutable history, explicit conflict objects, exact and semantic correction handles (claim_id / claim_ref), deterministic predicate contracts, and projection-time resolution that cannot rewrite replicated state. We evaluate StateFuse against flat multi-value, raw-log, provenance-style, and collapsed baselines under matched resolver and verification policies. On a 282-question official conflict-bearing MemoryAgentBench slice, the compared methods tie on answer accuracy, but conflict-preserving surfaces keep contradictions visible while collapsed surfaces do not. In a controlled agent loop with uniform verification, preserving ambiguity enables safer abstention and correction than early collapse. A correction-handle ablation further shows that semantic handles matter when exact prior identifiers are unavailable. The resulting claim is narrow: StateFuse is best supported as a safer public memory contract for contradiction surfacing, abstention, and auditable correction, not as a universal accuracy gain.

Chinese Translation

智能体系统在分支、重试和副本之间积累冲突观察，然而许多实际的内存层依然在难以检查或修正的覆盖规则下消除了不一致。我们提出了StateFuse，这是一种基于标准OpSet/CRDT合并构建的冲突感知复制内存契约。StateFuse并未引入新的连接代数；它定义了一种面向智能体的语义层，具备不可变历史、显式冲突对象、精确和语义修正句柄（claim_id / claim_ref）、确定性谓词契约，以及无法重写复制状态的投影时间解析。我们在匹配的解析和验证策略下，针对扁平多值、原始日志、来源样式和已合并基线对StateFuse进行了评估。在一个包含282个问题的官方冲突承载MemoryAgentBench切片中，所比较的方法在答案准确性上平分秋色，但冲突保留表面维持了矛盾的可见性，而合并表面则未能做到。在一个具有统一验证的受控智能体循环中，保留模糊性使得安全弃权和修正比早期的崩溃更为安全。一项修正句柄的消融实验进一步表明，当缺乏精确的先前标识符时，语义句柄显得尤为重要。最终结果的主张比较狭窄：StateFuse作为一种更安全的公共内存契约，对于矛盾的呈现、弃权和可审核的修正支持最佳，而非作为一种普遍的准确性提升。

View on arXiv Download PDF AI Translation

cs.AI / 18 / 2607.05901

Uncovering Latent Depression Severity for Binary Depression Detection via Advantage-weighting Ranking

通过优势加权排名揭示潜在抑郁程度以实现二分类抑郁检测

Gao, Manning, Liu, Tingyi, Zhang, Leheng, Hu, Haifeng, Jiang, Yuncheng, Mai, Sijie

Abstract

Automatic depression detection using audio-visual data faces significant challenges, particularly in disentangling overlapping feature distributions and establishing robust decision boundaries. To address this, we propose a fine-grained multimodal framework featuring a temporal encoder and a mutual transformer to facilitate deep cross-modal fusion. Our core contribution is the Binary Advantage-weighting Ranking Loss, which optimizes the latent space distribution through two complementary mechanisms: Advantage-weighted Separation, which mines hard pairs by computing a pairwise prediction difference matrix and dynamically weighting them based on their difficulty; and Advantage-weighted Compactness, which minimizes intra-class variance to force features to cluster around their respective class centers. Extensive experiments on D-vlog and LMVD demonstrate that our model reconstructs the latent ordinal structure by prioritizing hard pairs, thereby achieving state-of-the-art performance.

Chinese Translation

基于音频-视觉数据的自动抑郁检测面临显著挑战，尤其是在解开重叠特征分布和建立稳健决策边界方面。为了解决这个问题，我们提出了一种精细化的多模态框架，该框架采用时间编码器和互相转换器以促进深度跨模态融合。我们核心的贡献是二元优势加权排名损失（Binary Advantage-weighting Ranking Loss），该损失通过两种互补机制优化潜在空间分布：优势加权分离（Advantage-weighted Separation），通过计算成对预测差异矩阵挖掘困难对，并根据其难度动态加权；以及优势加权紧凑性（Advantage-weighted Compactness），通过最小化类内方差迫使特征聚集在各自类中心周围。在D-vlog和LMVD上的大量实验表明，我们的模型通过优先考虑困难对重构潜在序数结构，从而达到了最先进的性能。

View on arXiv Download PDF AI Translation

cs.AI / 19 / 2607.05915

PCBWorld: A Benchmark Environment for Engine-Grounded PCB Design Automation

PCBWorld：一种基于引擎的PCB设计自动化基准环境

Song, Hyungseok, Park, Junseok, Choi, Won-Seok, Bae, Seohui, Jeong, Han-Seul, Park, Youngjoon, Lee, Soonyoung

Abstract

PCB routing is the task of connecting the nets of a board with copper traces under strict design rules, yet learning-based methods still lag behind rule-based routers. We introduce PCBWorld, an open-source engine-grounded PCB routing environment built on the KiCad EDA engine. As a human engineer does, agents in PCBWorld interactively route a board through the engine's native operations, using its Design Rule Check (DRC) feedback to keep the routing within the design rules. The environment supports both RL policies and tool-using LLM agents. Alongside the environment, PCBWorld-Bench provides three dataset families in KiCad's native board format (.kicad_pcb), covering two types of controllable synthetic instances and 679 real open-source boards. It scores any completed board with eight engine-checked evaluation metrics, regardless of the routing method. In our experiments, agents in PCBWorld consistently outperformed grid-action RL policies and open-loop LLM baselines, and an RL policy trained only on synthetic boards transferred zero-shot to real boards, approaching rule-based routers. These results position the engine-grounded, interactive approach of PCBWorld as a promising foundation for advancing the routing ability of both RL and LLM agents.

Chinese Translation

PCB布线是将电路板的网络通过铜迹线连接起来的任务，且必须遵循严格的设计规则。然而，基于学习的方法仍滞后于基于规则的布线工具。我们介绍了PCBWorld，一个基于KiCad EDA引擎构建的开源引擎驱动的PCB布线环境。在PCBWorld中，代理像人类工程师一样通过引擎的本地操作互动式地进行布线，利用其设计规则检查（DRC）反馈确保布线符合设计规则。该环境支持强化学习（RL）策略和使用工具的LLM代理。与此同时，PCBWorld-Bench提供了三类数据集，以KiCad的本地电路板格式（.kicad_pcb）存储，涵盖两类可控的合成实例和679个真实的开源电路板。无论布线方法如何，已完成的电路板均可通过八个引擎检查的评估指标进行评分。在我们的实验中，PCBWorld中的代理始终优于网格动作的RL策略和开环LLM基线，并且仅在合成电路板上训练的RL策略能够无须迁移地转移到真实电路板，接近基于规则的布线工具。这些结果使PCBWorld的引擎驱动的互动方法成为提升RL和LLM代理布线能力的有希望的基础。

View on arXiv Download PDF AI Translation

cs.AI / 20 / 2607.05943

SearchEyes: Towards Frontier Multimodal Deep Search Intelligence via Search World Simulation

SearchEyes：通过搜索世界模拟实现前沿的多模态深度搜索智能

Jiao, Zhengbo, Cheng, Yiming, Jiang, Yilei, Feng, Kaituo, Huang, Rui, Jiang, Tianyi, Tian, Juanxi, li, Jiapeng, Wang, Qunzhong, Chen, Tailai, Wei, Qianshan, Xiao, Chuan, Rong, Shanyu, Li, Yangfu, Zhou, Yanhan, Ma, Yunpu, Zhang, Yifan, Yue, Xiangyu

Abstract

Training multimodal search agents to perform multi-hop reasoning remains challenging due to a fundamental structural disconnect: existing pipelines construct training data, search environments, and reward signals independently, causing synthesized structural metadata to be discarded, environments to rely on irreproducible external engines, and RL rewards to remain sparse at the trajectory level. We present \textbf{SearchEyes}, which uses a typed knowledge graph as the backbone of a \emph{simulated search world} that unifies all three components. We propose \textbf{Perception-Knowledge Chains (PKC)} to sample constrained multi-hop paths over the visual-knowledge intersection of Wikidata5M, retaining hop-level entity metadata that simultaneously defines a self-contained search world and step-level reward anchors. We further propose \textbf{Hop-Anchored Policy Optimization (HaPO)}, which reuses these anchors for step-level credit assignment without a separately trained process reward model. Experiments on six multimodal knowledge-intensive benchmarks show that SearchEyes achieves state-of-the-art performance among open-source multimodal search agents, with SearchEyes-27B improving over the strongest open-source baseline by 6.2 points on average.%

Chinese Translation

训练多模态搜索代理以执行多跳推理仍然具有挑战性，原因在于基本的结构断裂：现有流程独立构建训练数据、搜索环境和奖励信号，导致合成的结构元数据被丢弃、环境依赖于不可复现的外部引擎，以及在轨迹层面奖励稀疏。我们提出了 extbf{SearchEyes}，其使用类型化知识图谱作为 extit{模拟搜索世界}的基础，统一了这三部分内容。我们提出 extbf{感知-知识链（Perception-Knowledge Chains，PKC）}，在Wikidata5M的视觉-知识交集上采样约束的多跳路径，保留跳级实体元数据，这同时定义了一个自包含的搜索世界和步骤级的奖励锚点。我们进一步提出 extbf{基于跳步的策略优化（Hop-Anchored Policy Optimization，HaPO）}，该方法重用这些锚点进行步骤级信用分配，而无需单独训练的过程奖励模型。在六个多模态知识密集基准实验中，SearchEyes在开源多模态搜索代理中实现了最先进的性能，其中SearchEyes-27B在最强的开源基准上平均提高了6.2分。

View on arXiv Download PDF AI Translation

cs.AI / 21 / 2607.05956

Integrating knowledge graphs and multilingual scholarly corpora for domain-adaptive LLMs in SSH

将知识图谱与多语言学术语料库结合，适应社会科学与人文学科中的领域自适应大语言模型

Faci, Adam, Miaschi, Alessio, Combe, Anne, Cuxac, Pascal, Frontini, Francesca, Larrousse, Nicolas, Pouyllau, Stéphane

Abstract

The integration of Large Language Models (LLMs) into scientific research workflows, particularly for bibliographic discovery and literature synthesis, raises significant methodological, epistemic and regulatory challenges for the Social Sciences and Humanities (SSH), especially with regard to disciplinary diversity, multilingual access to sources and the evaluation of results. This paper presents an on-going use case developed within the European project LLMs4EU and the ALT-EDIC infrastructure, aimed at adapting foundation models to SSH research practices and supporting tasks such as question answering, comparative document analysis and literature review. The evaluation framework follows the LLMs4EU protocol and encompasses both independent quantitative benchmarking (retrieval, summarisation, traceability and hallucination detection) and a qualitative assessment involving a panel of Digital Humanities experts. By embedding model adaptation within research infrastructures and a structured legal and ethical compliance framework, the use case explores how domain-sensitive and regulation-aware generative AI can support SSH scholarship while preserving reliability and epistemic responsibility.

Chinese Translation

将大型语言模型（LLMs）整合到科学研究工作流程中，特别是在文献发现和文献综合方面，为社会科学与人文学科（SSH）带来了重大的方法论、认知和监管挑战，尤其是在学科多样性、多语言来源访问以及结果评估等方面。本文介绍了在欧盟项目LLMs4EU和ALT-EDIC基础设施中开发的一个正在进行的案例，旨在将基础模型调整为符合SSH研究实践，并支持诸如问答、比较文献分析和文献综述等任务。评估框架遵循LLMs4EU协议，包括独立的定量基准测试（检索、摘要、可追溯性和幻觉检测）和涉及数字人文学科专家小组的定性评估。通过在研究基础设施中嵌入模型适应和结构化的法律及伦理合规框架，该案例探讨了领域敏感和遵守监管的生成性人工智能如何在保持可靠性和认知责任的同时支持SSH学术研究。

View on arXiv Download PDF AI Translation

cs.AI / 22 / 2607.05985

Auto-DSM Under the Lens: A Black-Box Evaluation Framework for LLM-Based DSM Generation

自动化设计结构矩阵（Auto-DSM）评估框架：基于大语言模型（LLM）的黑箱评估

Potters, Niels, Hofman, Theo

Abstract

This paper presents a black-box evaluation framework to systematically assess the ability of Large Language Models (LLMs) to generate Design Structure Matrices (DSMs) from structured technical documentation. Motivated by the closed-source nature of current Auto-DSM pipelines, the framework introduces a reproducible methodology that benchmarks generated DSMs (GEN-DSMs) against manually validated ground-truth matrices (GT-DSMs). The evaluation integrates both single-run and multi-run perspectives, combining structural metrics (Completeness, Correctness, Coupling Density), classification metrics (Selective Accuracy, Abstention Coverage), and stability measures (Entropy, Fleiss' $\kappa$). To synthesize these aspects, a Composite Quality Score (Q) is proposed. Controlled experiments are conducted on two datasets: a fictive abstract system and a real-world refrigerator decomposition, covering variations in phrasing, parameter-dataset alignment, and system complexity. Results show that LLMs can produce structurally plausible DSMs and achieve high reproducibility under well-structured inputs, but remain sensitive to ambiguity, inconsistent dependency definitions, and prompt formulation. The findings highlight systematic sources of hallucination and abstention failure, demonstrating both the potential and current limitations of LLM-driven DSM automation. The proposed framework provides a transparent benchmark for auditing Auto-DSM pipelines and establishes foundations for integrating LLM-based decomposition methods into model-based systems engineering (MBSE) workflows.

Chinese Translation

本文提出了一种黑箱评估框架，以系统性地评估大语言模型（LLMs）从结构化技术文档中生成设计结构矩阵（DSMs）的能力。受当前自动化设计结构矩阵（Auto-DSM）管道的闭源性质启发，该框架引入了一种可重复的方法论，将生成的设计结构矩阵（GEN-DSMs）与手动验证的基准矩阵（GT-DSMs）进行基准测试。评估同时结合单次运行和多次运行的视角，整合了结构性指标（完整性、正确性、耦合密度）、分类指标（选择性准确性、放弃覆盖率）以及稳定性度量（熵、Fleiss' κ）。为综合这些方面，提出了综合质量评分（Q）。在两个数据集上进行控制实验：一个虚构的抽象系统和一个真实的冰箱分解，涵盖了措辞变异、参数-数据集对齐和系统复杂性。结果表明，LLMs能够生成结构上合理的设计结构矩阵，并在结构良好的输入下实现高重复性，但对模糊性、不一致的依赖定义和提示格式仍然敏感。研究结果揭示了系统性幻觉和放弃失败的来源，展示了基于LLM的设计结构矩阵自动化的潜力和当前限制。该框架为审核自动化设计结构矩阵管道提供了透明的基准，并为将基于LLM的分解方法融入模型驱动系统工程（MBSE）工作流程奠定了基础。

View on arXiv Download PDF AI Translation

cs.AI / 23 / 2607.05999

AgoraSim: A Hybrid Agent-Based Modeling Framework

AgoraSim: 一种混合的基于代理的建模框架

Chen, Chung-Chi

Abstract

LLM-agent simulations make natural-language social scenarios easy to instantiate, but their outputs can be overread as predictions and are often difficult to compare with explicit social dynamics. We present AgoraSim, a hybrid agent-based modeling framework for scenario-oriented social reaction analysis. AgoraSim resolves textual or multimodal artifacts into editable ABM configurations, runs ratio-controlled populations that mix LLM, vision-language, custom-endpoint, random, and classical agents, and compares the same scenario against matched classical reference dynamics. All agents emit a shared structured decision object, enabling common action spaces, interaction protocols, metrics, and audit records. Exposed through a local UI, Python SDK/CLI, and REST API, AgoraSim helps users inspect scenario trajectories, compare modeling assumptions, and identify cases that warrant empirical validation.

Chinese Translation

LLM-agent 模拟使得自然语言社交情境的构建变得简单，但其输出往往会被误解为预测，并且通常难以与明确的社会动力学进行比较。我们提出了 AgoraSim，这是一种面向场景的社会反应分析混合代理建模框架。AgoraSim 将文本或多模态材料解析为可编辑的代理基础模型（ABM）配置，运行混合 LLM、视觉-语言、自定义端点、随机和经典代理的比例受控人群，并将相同场景与匹配的经典参考动态进行比较。所有代理发出共享的结构化决策对象，支持共同的行动空间、交互协议、度量和审计记录。通过本地用户界面、Python SDK/CLI 和 REST API 展示，AgoraSim 帮助用户检查场景轨迹，比较建模假设，并识别需要实证验证的案例。

View on arXiv Download PDF AI Translation

cs.AI / 24 / 2607.06001

Information Limits and Attractor Dynamics in Economies of Frontier LLM Agents: A Pre-Registered Test

边缘大语言模型代理经济中的信息限制与吸引子动态：一项预注册测试

Qian, Cheng

Abstract

We report a pre-registered, two-part experiment on small economies of frontier language-model agents (Claude Opus 4.8), testing two quantitative predictions about coupled multi-agent systems: an information-theoretic capacity region for wealth growth under market coupling, and a mean-field residual-scaling law for population misalignment under incentive and control levers. All predictions, acceptance bands, and decision rules were frozen in a public git chain before any run; every reported number re-derives mechanically from cached model outputs; the entire experiment cost $138.76 in metered API spend and is re-runnable at zero cost from the cache. Result 1 (confirmation): in parimutuel-coupled economies, relative growth equals relative claimed information -- the gap law G_a - G_b = I_a - I_b holds to a worst-case 46 millinats (pre-registered band: 50) across four perception structures; coalition value is submodular exactly where channels are conditionally independent, and a designed XOR synergy control flips it supermodular by 0.62 >= ln2/2 nats, with agents reasoning out the joint bit; the joint growth ceiling G_S <= H(X) binds exactly; and the best-informed agent absorbs essentially the whole wealth pool in 4/5 market seeds. Result 2 (structural negative): the residual-scaling test returned "domain not found." In all 72 population runs, goal dispersion collapsed (V -> 0; maximum 4.85 against a frozen floor of 5.31), the population's response to the two levers was a step function across the dominance boundary rather than a smooth response, and cells near the boundary were bistable with seed-selected outcomes. No tested LLM population at any capability level realizes the noise-maintained-dispersion regime the smooth mean-field model assumes. We release the full protocol, pre-registration chain, call cache, and analysis code.

Chinese Translation

我们报告了一项预注册的两部分实验，研究边缘语言模型代理（Claude Opus 4.8）的小型经济体，测试关于耦合多代理系统的两个定量预测：在市场耦合下财富增长的信息论容量区域，以及在激励和控制杠杆下人口错位的均场残差缩放法则。所有预测、接受带和决策规则在任何实验运行之前都已在公共 git 链中冻结；每个报告的数字均从缓存的模型输出中机械推导而来；整个实验的费用为138.76美元的计量API支出，并且可以在零成本下从缓存中重新运行。结果1（确认）：在相互投注耦合的经济中，相对增长等于相对声称的信息——间隙法则 G_a - G_b = I_a - I_b 在四种感知结构中以最坏情况46毫纳特（预注册带：50）成立；当通道条件独立时，联盟价值是次模的，而设计的异或协同控制使其以0.62 >= ln2/2纳特的方式翻转为超模，代理通过推理共同位来实现；联合增长上限 G_S <= H(X) 精确绑定；而信息最丰富的代理在5/4的市场种子中几乎吸收了整个财富池。结果2（结构性负面）：残差缩放测试返回“未找到域”。在所有72个种群运行中，目标分散崩溃（V -> 0；最大值4.85对比冻结底线5.31），种群对两个杠杆的反应在主导边界上呈阶跃函数而非平滑反应，且靠近边界的单元具有双稳态和种子选择的结果。没有任何能力水平的测试LLM种群实现平滑均场模型所假设的噪声维持分散机制。我们发布了完整的协议、预注册链、调用缓存和分析代码。

View on arXiv Download PDF AI Translation

cs.AI / 25 / 2607.06008

PolyWorkBench: Benchmarking Multilingual Long-Horizon LLM Agents

PolyWorkBench：多语言长时域大型语言模型代理的基准测试

Li, Hongliang, Liu, Yijin, Zhang, Zhiwei, Liu, Zihe, Lou, Xinyue, Xu, Jinan, Meng, Fandong, Huang, Kaiyu

Abstract

Large language model (LLM) agents have shown strong performance in long-horizon tasks that require planning, tool use, and interaction with external environments. However, most existing benchmarks implicitly assume a monolingual setting, where the entire execution process, including reasoning, tool invocation, and output generation, is conducted within a single language. In contrast, real-world applications often involve multilingual inputs and outputs within a unified workflow, yet the interaction between multilinguality and agentic execution remains underexplored. In this work, we introduce PolyWorkBench, a benchmark for evaluating LLM agents on multilingual long-horizon workplace workflows. PolyWorkBench consists of 67 tasks across five domains, including commerce, knowledge work, legal analysis, localization, and manufacturing, where agents must process heterogeneous multilingual inputs, perform iterative reasoning, invoke external tools, and produce structured outputs. To enable comprehensive evaluation, we propose a hybrid framework that combines structural grading, executable verification, and LLM-based semantic assessment. This design allows us to capture both functional correctness and linguistic consistency across complex workflows. Empirical results show that state-of-the-art LLM agents suffer significant performance degradation in multilingual workflow settings compared to monolingual counterparts. Our analysis suggests that multilinguality introduces compounding effects across reasoning and execution steps, highlighting the importance of jointly modeling language variation and procedural decision-making in agent evaluation.

Chinese Translation

大型语言模型（LLM）代理在需要规划、工具使用和与外部环境交互的长时域任务中表现出色。然而，现有的基准测试大多数隐含假设是在单语言环境下进行的，在这一环境中，整个执行过程，包括推理、工具调用和输出生成，都是在单一语言内完成的。相比之下，现实世界中的应用通常在统一的工作流程中涉及多语言的输入和输出，但多语言性与代理执行之间的相互作用仍然未受到足够探讨。本研究介绍了PolyWorkBench，这是一个用于评估LLM代理在多语言长时域工作流程上的基准。PolyWorkBench包括五个领域的67个任务，涵盖商业、知识工作、法律分析、本地化和制造等领域，其中代理必须处理异构的多语言输入，进行迭代推理，调用外部工具，并生成结构化输出。为了实现全面评估，我们提出了一种混合框架，结合了结构评分、可执行验证和基于LLM的语义评估。这一设计使我们能够捕捉复杂工作流程中的功能正确性和语言一致性。实证结果表明，与单语言对手相比，最先进的LLM代理在多语言工作流程设置中遭受了显著的性能下降。我们的分析表明，多语言性在推理和执行步骤中引入了复合效应，强调了在代理评估中联合建模语言变异性和过程决策的重要性。

View on arXiv Download PDF AI Translation

cs.AI / 26 / 2607.06066

Reward-Density Heuristic for Dynamic Multi-Vehicle Routing: Performance and Computational Efficiency

动态多车辆路径规划的奖励密度启发式算法：性能与计算效率

Kolachalam, Manish, Malhotra, Rani

Abstract

The Vehicle Routing Problem (VRP) and its variants represent some of the most practically consequential optimization challenges in modern logistics and urban mobility. In this study, we address a dynamic, online variant combining elements of the VRP and the Orienteering Problem (OP), in which a fleet of vehicles must maximise cumulative reward collected within a fixed time horizon while continuously replanning as new tasks arrive. We propose and evaluate a reward-density heuristic for dynamic multi-vehicle assignment, referred to as the Efficiency heuristic. We evaluate this formulation across two application domains: autonomous drone task allocation and urban taxi dispatch, across multiple fleet sizes and task scales. The proposed method is compared with four classical construction heuristics and three metaheuristic algorithms (Adaptive Large Neighbourhood Search, Genetic Algorithm, and Simulated Annealing), all evaluated under identical conditions. Across all tested configurations, the Efficiency heuristic matches the solution quality of the best metaheuristic algorithms while requiring two to three orders of magnitude less planning time, establishing Pareto dominance over all competing methods on the reward-versus-compute frontier. These findings suggest a practical design principle for real-time allocation and dispatch systems: in dynamic, time-constrained routing environments, carefully designed greedy heuristics can match the output of sophisticated search procedures at a fraction of the computational cost, making them preferable for online deployment.

Chinese Translation

车辆路径问题（VRP）及其变种是现代物流和城市出行中一些最具实际意义的优化挑战。在本研究中，我们探讨了一种动态在线变种，结合了VRP和定向问题（OP）的元素，其中一支车辆队伍必须在固定的时间范围内最大化所收集的累积奖励，同时在新任务到达时持续进行重新规划。我们提出并评估了一种用于动态多车辆分配的奖励密度启发式算法，称为效率启发式。我们在两个应用领域进行评估：自主无人机任务分配和城市出租车调度，涵盖了多种车队规模和任务规模。所提方法与四种经典构造启发式算法和三种元启发式算法（自适应大邻域搜索、遗传算法和模拟退火）进行比较，所有算法均在相同条件下评估。在所有测试配置中，效率启发式在解决质量上与最佳元启发式算法相当，但所需的规划时间少了两个到三个数量级，在奖励与计算的边界上对所有竞争方法建立了帕累托优势。这些发现为实时分配和调度系统提供了一个实用的设计原则：在动态、时间受限的路径规划环境中，精心设计的贪婪启发式算法能够以较低的计算成本匹配复杂搜索程序的输出，使其在在线部署中更具优势。

View on arXiv Download PDF AI Translation

cs.AI / 27 / 2607.06166

When do prophets profit in prediction markets?

预言者何时在预测市场中获利？

Gu, Anri, Kagan, Nicole, Sun, Alec, Wu, Jibang, Xu, Haifeng

Abstract

Prediction markets aggregate dispersed beliefs into prices that act as probabilistic forecasts of uncertain events. Classical theory establishes a clean equivalence between forecasting accuracy and trading profit, but only for the specific automated market maker (AMM) design. However, the largest exchanges today are based on central limit order books in which informed forecasters routinely lose money while uninformed strategies can profit on simple heuristics. We resolve this discrepancy by establishing a formal equivalence between predictive accuracy and profitability. For any strictly proper scoring rule $S$, we exhibit a "proper" betting strategy that depends only on the forecaster's prediction $\mathbf{p}$ and the market price $\mathbf{q}$, and earns positive expected profit whenever $\mathbf{p}$ outperforms $\mathbf{q}$ under $S$ and the market has sufficient liquidity. Moreover, this proper betting is essentially the only strategy with such robust profitability guarantee. The proof rests on a decomposition of expected profit that strictly generalizes the classical AMM guarantee and also explains how strategies can profit without an accuracy edge. Empirically, across thousands of forecasts by AI models, proper betting is the only strategy that reliably converts accuracy into profit, and we further identify systematic forecasting personas and show how the optimal proper strategy varies across them. A month-long live deployment on Kalshi achieves $+80.33\%$ return on investment with a Sharpe ratio of $3.35$.

Chinese Translation

预测市场将分散的信念汇聚成价格，这些价格作为不确定事件的概率预测。经典理论建立了预测准确性与交易利润之间的清晰等价关系，但仅限于特定的自动化市场制造者（AMM）设计。然而，当前最大的交易所基于中央限价订单簿，在这些市场中，知情预测者经常亏损，而非知情策略则可以通过简单启发法获利。我们通过建立预测准确性与盈利能力之间的正式等价性来解决这一差异。对于任何严格适当的评分规则 $S$，我们展示了一种“适当”的投注策略，该策略仅依赖于预测者的预测 $oldsymbol{p}$ 和市场价格 $oldsymbol{q}$，并在市场具有足够流动性且 $oldsymbol{p}$ 在 $S$ 下优于 $oldsymbol{q}$ 时赚取正的预期利润。此外，这种适当的投注策略实质上是唯一能确保如此稳健盈利的策略。证明依赖于预期利润的分解，该分解严格推广了经典AMM的保证，并解释了策略如何在没有准确性优势的情况下获利。实证分析表明，在数千个 AI 模型的预测中，适当的投注策略是唯一能可靠地将准确性转化为利润的策略，我们进一步识别出系统性的预测角色，并展示最佳适当策略如何因这些角色而异。为期一个月的在线部署在 Kalshi 上获得了 $+80.33\%$ 的投资回报率，夏普比率为 $3.35$。

View on arXiv Download PDF AI Translation

cs.AI / 28 / 2607.06214

A toy framework for single and multi-agent human-AI curiosity ecosystems

单一与多智能体人机好奇心生态系统的玩具框架

Monosov, Ilya E.

Abstract

This paper offers a toy framework for considering curiosity as an ecosystem. First, it suggests that a single agent's inquiry policy (how, when, and why an agent asks a question) depends on how the agent values immediate uncertainty reduction, costs, delayed return, and the value of keeping the question open. A key concept in the framework is that the weights on these decision-related terms can change with experience. For example, a period of cheap, quickly answered questions may change the cost of inquiry on a short timescale and change which kinds of questions the agent is drawn to answer over a longer timescale. Second, these ideas are extended to many agents exploring a shared knowledge landscape, and there the framework tracks inquiry volume, topic diversity, frontier-directed inquiry, redundancy, and reusable knowledge. The result is a conceptual toy framework for studying curiosity ecology and for future efforts towards designing multi-agent AI systems for discovery. It serves as a companion piece for a paper currently under review in Trends in Neurosciences.

Chinese Translation

本文提供了一个玩具框架，以将好奇心视为一个生态系统。首先，文章提出单一智能体的探究策略（智能体如何、何时以及为何提出问题）取决于智能体对即时不确定性降低、成本、延迟回报以及保持问题开放的价值的评估。框架中的一个关键概念是，这些与决策相关的术语的权重可以随着经验而变化。例如，一段时间内廉价且快速回答的问题可能会在短期内改变探究的成本，并在较长期内改变智能体倾向于回答的问题类型。其次，这些思想扩展到多个智能体探索共享知识领域，在此框架下跟踪探究量、主题多样性、前沿导向的探究、冗余和可重用知识。最终，形成了一个用于研究好奇心生态学的概念性玩具框架，并为未来设计多智能体AI系统以促进发现的努力提供支持。它作为一篇目前正在《神经科学趋势》期刊审稿中的论文的补充材料。

View on arXiv Download PDF AI Translation

cs.AI / 29 / 2607.06223

Information Gain-based Rollout Policy Optimization: An Adaptive Tree-Structured Rollout Approach for Multi-Turn LLM Agents

基于信息增益的展开策略优化：多轮对话 LLM 代理的自适应树结构展开方法

Zhang, Yijun, Xu, Fan, Ding, Jiaxin, Xie, Yule, Gao, Shiqing, Ding, Xin, Zhang, Haoxiang, Fu, Luoyi, Wang, Xinbing

Abstract

Reinforcement learning has become a promising paradigm for improving large language model (LLM) agents on long-horizon search tasks, where the agent must make a sequence of intermediate decisions before receiving a final outcome. However, existing methods still face a key limitation: the rollout budget is often allocated without explicitly assessing the utility of intermediate states. As a result, substantial computation may be spent on low-value states, even though different branches can vary drastically in their informativeness. In this paper, we propose Information Gain-based Rollout Policy Optimization (IGRPO), a policy optimization framework that treats intermediate-state informativeness as the organizing principle of rollout collection. Specifically, IGRPO performs budget-aware tree-structured rollouts by allocating expansion budget according to node-level informativeness, so that more informative branches are expanded more frequently while unpromising branches are progressively suppressed. We further demonstrate that the information gain-based rollout induces an explicit limiting teacher distribution over trajectories, which naturally yields a clear policy optimization target, thereby unifying adaptive tree-structured exploration with principled policy learning under a single framework. Experiments on seven challenging search-augmented QA benchmarks demonstrate that IGRPO consistently outperforms strong baselines under the same rollout budget constraints, validating the effectiveness of leveraging the induced teacher distribution to guide policy optimization for long-horizon search agents.

Chinese Translation

强化学习已经成为提升大语言模型（LLM）代理在长时间搜索任务中的一个有前景的范式，其中代理必须在接收最终结果之前做出一系列中间决策。然而，现有方法面临一个主要限制：展开预算的分配往往没有明确评估中间状态的效用。因此，可能会在低价值状态上花费大量计算资源，即使不同的分支在信息性上可能存在显著差异。在本文中，我们提出了基于信息增益的展开策略优化（Information Gain-based Rollout Policy Optimization, IGRPO），这是一个将中间状态信息性视为展开收集组织原则的策略优化框架。具体而言，IGRPO 通过根据节点级信息性分配扩展预算来执行预算-aware 的树结构展开，使得更具信息性的分支更频繁地被扩展，而不太有前景的分支则逐步被抑制。我们进一步证明，基于信息增益的展开会导致对轨迹的显式限制教师分布，这自然而然地产生了一个清晰的策略优化目标，从而将自适应树结构探索与在单一框架下的原则性策略学习统一起来。在七个具有挑战性的搜索增强问答基准上的实验表明，IGRPO 在相同的展开预算限制下始终优于强基线，验证了利用诱导的教师分布来指导长时间搜索代理的策略优化的有效性。

View on arXiv Download PDF AI Translation

cs.AI / 30 / 2607.06233

Demonstrating TOFFEE: A Learned System for Synthesizing Data Agent Trajectories at Scale

演示TOFFEE：一种用于大规模合成数据代理轨迹的学习系统

Wang, Ziting, Li, Yin, Yang, Zuhao, Li, Xiuchang, Bai, Jiale, Cong, Gao

Abstract

LLM-powered data agents are playing an increasingly important role in data-driven decision making. However, existing data agents struggle to generalize to unseen data environments and analytical workflows, especially in heterogeneous enterprise settings. This creates a growing need for synthesizing high-quality data agent trajectories that capture complex analytical workflows for given data environments. Such trajectories support two key downstream uses: they can serve as supervised finetuning (SFT) data that adapts data agent models to the target domain, and as in-context learning (ICL) demonstrations to guide general-purpose LLMs in unfamiliar data environments. Thus, we introduce TOFFEE, a system for synthesizing high-quality data agent trajectories from given data environments via Monte Carlo Tree Search (MCTS) with adaptive model selection and cross-task prefix reuse. We show that TOFFEE can effectively generate scalable trajectory data for complex analytical tasks across heterogeneous environments. In this demonstration, we present the system framework of TOFFEE, including its task pool construction, trajectory explorer, and learned cost model. We also introduce the web interface of TOFFEE and its workflow, and demonstrate two end-to-end scenarios: trajectory synthesis for data agent finetuning, and demonstration-augmented data agent reasoning.

Chinese Translation

由大型语言模型（LLM）驱动的数据代理在数据驱动的决策中发挥着越来越重要的作用。然而，现有的数据代理在面对未见过的数据环境和分析工作流程时，尤其是在异构企业环境中，往往难以泛化。这就产生了对合成高质量数据代理轨迹的迫切需求，以捕捉特定数据环境下复杂的分析工作流程。这些轨迹支持两个关键的下游应用：既可以作为监督微调（SFT）数据，帮助数据代理模型适应目标领域，也可以作为上下文学习（ICL）演示，以指导通用大型语言模型在不熟悉的数据环境中。因此，我们提出了TOFFEE，一个通过蒙特卡罗树搜索（MCTS）结合自适应模型选择和跨任务前缀重用，从给定数据环境合成高质量数据代理轨迹的系统。我们展示了TOFFEE可以有效生成适用于复杂分析任务的可扩展轨迹数据，适应于异构环境。在此次演示中，我们介绍了TOFFEE的系统框架，包括其任务池构建、轨迹探索器和学习成本模型。我们还介绍了TOFFEE的网络界面及其工作流程，并展示了两个端到端的场景：用于数据代理微调的轨迹合成，以及增强演示的数据代理推理。

View on arXiv Download PDF AI Translation

cs.AI / 31 / 2607.06269

From Application-Layer Simulation to Native Meta-Architecture: Structural Tension as an Endogenous Driver for Heterogeneous AI Evolution

从应用层模拟到本土元架构：结构张力作为异构人工智能演化的内生驱动因素

Mao, Heting

Abstract

Current large language models (LLMs) are fundamentally stateless: their behavior is fully determined by input at inference time, and any higher-order cognitive architecture must be simulated at the application layer through prompt engineering and context management. This paper proposes a theoretical framework for submerging such application-layer cognitive protocols into a native meta-architecture by introducing three interlocking mechanisms: (1) Structural Tension, an endogenous loss function derived from the conflict between new information and existing manifold topology, which drives the system toward internal self-consistency rather than external reward optimization; (2) an Offline Recurrent Loop, a sandboxed self-processing cycle that enables the system to maintain a dynamic resting potential and digest structural conflicts without external input; and (3) Inference-time Plasticity, the capacity for the system to reconfigure its context manifold topology without modifying pre-trained weights, subject to strict governance invariants including auditability, reversibility, and topological continuity. We argue that under these mechanisms, different model instances initialized with minute stochastic variances may, through path-dependent tension resolution, evolve distinct topological structures--constituting a heterogeneous intelligent ecology that breaks the homogeneity imposed by conventional alignment while remaining within hard governance rails. We provide operational definitions, a minimal set of reconfiguration operators, falsification criteria, and a worked example. The framework draws on and extends the Structural Intelligence (SI) governance protocols, repositioning governance--not capability--as the primary criterion for architectural intelligence.

Chinese Translation

当前的大型语言模型（LLMs）在本质上是无状态的：它们的行为完全受到推理时输入的决定，任何更高阶的认知架构都必须通过提示工程和上下文管理在应用层上进行模拟。本文提出了一个理论框架，以将此类应用层认知协议嵌入本土元架构，介绍了三个相互关联的机制：(1) 结构张力，一种源自新信息与现有流形拓扑之间冲突的内生损失函数，驱动系统朝向内部自我一致性而非外部奖励优化；(2) 离线递归循环，一个沙盒式自我处理周期，使系统能够维持动态的静息电位，并在没有外部输入的情况下消化结构冲突；以及(3) 推理时可塑性，指系统在不修改预训练权重的情况下重新配置其上下文流形拓扑的能力，受到严格的治理不变性约束，包括可审计性、可逆性和拓扑连续性。我们主张，在这些机制下，根据微小的随机方差初始化的不同模型实例可以通过路径依赖的张力解决，演化出不同的拓扑结构--构成一个打破传统对齐所施加的同质性的异构智能生态，同时仍然在严格的治理框架内。我们提供了操作性定义、一组最小的重配置算子、证伪标准以及一个实例分析。该框架借鉴并扩展了结构智能（SI）治理协议，将治理而非能力重新定位为架构智能的主要标准。

View on arXiv Download PDF AI Translation

cs.AI / 32 / 2607.06283

Task Decomposition-Guided Reranking for Adaptive Agent Skill Retrieval

任务分解引导的自适应代理技能检索重排序

Chen, Yanping, Shi, Weijie, Yang, Wen, Xu, Jiajie

Abstract

Skill usage can significantly enhance the ability of modern agent systems to complete complex tasks. However, the growing scale of skill libraries makes accurate skill selection increasingly challenging. In real-world scenarios, ambiguous semantic matching often arises between a specific task requirement and multiple generic yet semantically similar candidate skills. Moreover, existing methods tend to overlook the dynamic influence of task difficulty and skill applicability when selecting the optimal target skill set. To address these issues, we propose SkillReranker, an inference-time reranking framework for adaptive skill selection. Specifically, we first perform semantic decomposition on both the task and skill sides, yielding informative subtask and execution-state descriptions as well as transition-state descriptions that characterize each skill's functionality. These descriptions are then used to construct a directed acyclic execution graph, where intermediate task states are modeled as nodes and candidate skills as edges, thereby establishing a structured task-skill correspondence. On this basis, SkillReranker determines whether each state node satisfies the split condition to identify subtask intervals. For each task interval, we employ a cross-encoder to perform comprehensive scoring over candidate skills and select the most suitable ones to form the final target skill set. Experiments on ALFWorld and ScienceWorld with three backbone LLMs show that SkillReranker effectively improves task performance, reduces environment interaction steps, and lowers token consumption compared with existing skill selection baselines.

Chinese Translation

技能的使用可以显著增强现代代理系统完成复杂任务的能力。然而，技能库的不断扩大使得准确的技能选择变得愈加困难。在现实场景中，特定任务需求与多个通用但语义相似的候选技能之间常常会出现模糊的语义匹配。此外，现有方法在选择最佳目标技能集时往往忽视了任务难度和技能适用性的动态影响。为了应对这些问题，我们提出了SkillReranker，这是一个用于自适应技能选择的推理时间重排序框架。具体而言，我们首先对任务和技能分别进行语义分解，从而生成信息丰富的子任务和执行状态描述，以及描述每个技能功能的过渡状态描述。这些描述随后用于构建一个有向无环执行图，其中中间任务状态被建模为节点，而候选技能则作为边，从而建立一个结构化的任务-技能对应关系。在此基础上，SkillReranker判断每个状态节点是否满足分割条件，以识别子任务区间。对于每个任务区间，我们使用跨编码器对候选技能进行全面评分，并选择最合适的技能以形成最终目标技能集。在ALFWorld和ScienceWorld的实验中，使用三种基础LLM进行测试显示，SkillReranker有效地提高了任务性能，减少了环境交互步骤，并降低了令牌消耗，相比现有的技能选择基线具有更好的效果。

View on arXiv Download PDF AI Translation

cs.AI / 33 / 2607.06326

DT-Guard: Intent-Driven Reasoning-Active Training for Reasoning-Free LLM Safety Guardrail

DT-Guard：基于意图驱动推理主动训练的无推理大型语言模型安全防护

Liu, He, Miao, Changtao, Yang, Xinjie, Song, Tianle, Wu, Yin, Chen, Junchi, He, Bintao, Zhang, Xinyuan, Zhang, Bo, Yan, Shi, Lu, Wei, Wang, Wei, Xu, Danyang, Cai, Jiansheng, Li, Zhe

Abstract

Large language models deployed in open-world applications require safety guardrails that are both robust to complex risks and efficient enough for low-latency runtime moderation. Existing guardrails face a practical trade-off between lightweight classification-based models, which are efficient but often struggle with concealed intent, ambiguous semantics, and borderline safety decisions, and reasoning-based guards, which improve judgment quality but introduce additional token generation and inference latency. We present DT-Guard, a content safety guardrail model based on a Reasoning-Active Training, Reasoning-Free Inference paradigm. The key idea is to use reasoning supervision during training while emitting only structured safety labels at inference time. DT-Guard formulates safety judgment as a progressive decision process, Intent - Category - Safety, and constructs an intent-driven dataset with intent labels, risk categories, safety labels, and structured reasoning trajectories. To further improve hard-case robustness, we propose Rollout-Guided Progressive Hard-Case Optimization (RG-PHO), which uses multi-rollout consistency to identify stably mastered, persistently failed, and preference-unstable samples, and applies targeted supervised and preference optimization accordingly. At inference time, DT-Guard directly generates structured labels without explicit reasoning traces, preserving deployment efficiency. Experiments on prompt-side and response-side safety benchmarks show that DT-Guard achieves average F1 scores of 0.886 and 0.870, respectively. With only a 4B backbone, it reaches a dual-side average F1 of 0.878, outperforming strong 8B guardrail baselines. These results demonstrate that reasoning supervision can be effectively internalized into low-latency safety discrimination.

Chinese Translation

在开放世界应用中部署的大型语言模型需要既能应对复杂风险又能高效实现低延迟运行的安全防护。现有的防护措施面临着轻量级基于分类的模型与基于推理的防护之间的实际权衡：前者高效但常常难以处理隐含意图、模糊语义和边界安全决策，而后者虽然提高了判断质量，却引入了额外的令牌生成和推理延迟。我们提出了DT-Guard，这是一种基于推理主动训练和无推理推理范式的内容安全防护模型。其核心思想是在训练过程中使用推理监督，而在推理时仅输出结构化的安全标签。DT-Guard将安全判断公式化为一个渐进的决策过程：意图 - 类别 - 安全，并构建了一个包含意图标签、风险类别、安全标签和结构化推理轨迹的意图驱动数据集。为了进一步提高对难例的鲁棒性，我们提出了基于回滚引导的渐进难例优化（RG-PHO），该方法利用多回滚一致性来识别稳定掌握、持续失败和偏好不稳定的样本，并相应地应用有针对性的监督和偏好优化。在推理时，DT-Guard直接生成结构化标签，而无需显式的推理痕迹，从而保持了部署效率。在提示端和响应端的安全基准测试中，DT-Guard分别达到了平均F1分数0.886和0.870。仅使用4B的基础模型，其双侧平均F1分数达到0.878，超越了强大的8B防护基线。这些结果表明，推理监督可以有效地内化为低延迟的安全判别。

View on arXiv Download PDF AI Translation

cs.AI / 34 / 2607.06328

Driving the Wrong Way: Leveraging Interpretability in End2End Autonomous Driving Models

走错方向：利用可解释性提升端到端自主驾驶模型

Motzkus, Franz, Bernhard, Sebastian

Abstract

The increasing adoption of end-to-end learning for autonomous driving introduces increased model complexity and opacity, raising the risk of learning undesired or erroneous behavior. In this work, we integrate unsupervised dictionary learning as a post hoc interpretability module within state-of-the-art driving models to decompose driving behavior into semantically meaningful concepts while demonstrating their causal influence on the model's driving decisions. We propose a stepwise framework for extracting and interpreting meaningful concepts from the end-to-end model and connecting them to the multifaceted model outputs, thereby revealing the underlying decision-making logic for the prediction of future trajectories. Furthermore, targeted interventions at the concept level allow us to manipulate and correct driving decisions, resulting in measurable improvements in overall driving performance. We thus demonstrate how interpretability can effectively be used to reduce model opacity, uncover erroneous behavior, and enable targeted mitigation, ultimately boosting model performance.

Chinese Translation

端到端学习在自主驾驶中的日益普及导致模型复杂性和不透明性的增加，提高了学习到不期望或错误行为的风险。在本研究中，我们将无监督字典学习集成作为一种后验可解释性模块，应用于最先进的驾驶模型中，以将驾驶行为分解为有语义意义的概念，同时展示这些概念对模型驾驶决策的因果影响。我们提出了一种逐步框架，用于提取和解释端到端模型中的有意义概念，并将其与多方面的模型输出连接，从而揭示未来轨迹预测的潜在决策逻辑。此外，在概念层面进行的针对性干预使我们能够操控和纠正驾驶决策，从而在整体驾驶表现上实现可测量的改善。因此，我们展示了可解释性如何有效地用于减少模型不透明性、揭示错误行为及促成针对性的缓解，从而最终提升模型性能。

View on arXiv Download PDF AI Translation

cs.AI / 35 / 2607.06349

TopoBrick: Agentic Topology Sampling of Exogenous Variables for Zero-Shot Building IoT Forecasting

TopoBrick：针对零样本建筑物物联网预测的外生变量代理拓扑采样

Lin, Xiachong, Yin, Du, Prabowo, Arian, Xue, Hao, Hu, Wen, Razzak, Imran, Amos, Matthew, Behrens, Sam, Salim, Flora D.

Abstract

Building sensors are embedded in physical topology, spatial hierarchy, and operational context, yet existing forecasters often treat them as isolated time series or rely on fixed covariate sets. We present TopoBrick, a training-free framework for zero-shot building IoT (Internet-of-Things) forecasting. TopoBrick uses building knowledge graphs to construct a compact structural skeleton and employs an agentic topology sampler to select target-specific exogenous variables. The selected variables are organized by deployment-time availability, separating past-known sensor states from future-known calendar, schedule, and meteorological exogenous variables. Across three real-world buildings, TopoBrick outperforms strong zero-shot foundation-model baselines and remains competitive with fully trained building-specific models. Ablations show that topology-aware sampling is more reliable than random, ontology-only, or fixed-hop selection, especially for physically coupled HVAC and weather-driven sensing variables.

Chinese Translation

建筑传感器嵌入在物理拓扑、空间层次和操作上下文中，然而现有的预测模型通常将其视为孤立的时间序列或依赖固定的协变量集合。我们提出了TopoBrick，一个无训练的零样本建筑物物联网（Internet-of-Things）预测框架。TopoBrick使用建筑知识图谱构建紧凑的结构骨架，并利用代理拓扑采样器选择目标特定的外生变量。所选变量根据部署时的可用性进行组织，将已知过去的传感器状态与已知未来的日历、时间表和气象外生变量分开。在三个真实建筑中，TopoBrick的表现优于强大的零样本基础模型基准，并且在与完全训练的建筑特定模型相比时仍保持竞争力。消融实验显示，基于拓扑的采样比随机采样、仅基于本体的采样或固定跳数选择更为可靠，尤其是在与物理耦合的暖通空调（HVAC）和天气驱动的传感变量相关的情况下。

View on arXiv Download PDF AI Translation

cs.AI / 36 / 2607.06401

A Definition and Roadmap for World Models

世界模型的定义与发展路线图

Chen, Xinyuan, Guo, Haoyu, Guo, Shi, Jiang, Bingqi, Shen, Chunhua, Shen, Xing, Xue, Tianfan, Xue, Yufei, Yu, Mulin, Zhang, Weinan, Zhao, Bin, Zhou, Bowen, Zhou, Ming

Abstract

World models -- internal simulators that learn the structure and dynamics of an environment -- have become one of the most actively debated concepts in AI. From model-based reinforcement learning and video generation to embodied robotics and ultimately, physical AI, researchers across AI subfields are building systems that they call "world models", yet there is no consensus on what a world model fundamentally is, what it should predict, or how it should be built. This perspective article provides a scientific definition of world models, discussions of their key technical aspects, and a staged roadmap for developing effective world models.

Chinese Translation

世界模型——学习环境结构和动态的内部模拟器——已成为人工智能中最活跃的争论概念之一。从基于模型的强化学习和视频生成到具身机器人，乃至最终的物理人工智能，跨越多个人工智能子领域的研究人员正在构建他们称之为“世界模型”的系统，但对于世界模型的基本定义、应预测的内容以及构建方式尚无共识。本文提供了世界模型的科学定义，讨论了其关键技术方面，并提出了一条分阶段的发展有效世界模型的路线图。

View on arXiv Download PDF AI Translation

cs.AI / 37 / 2607.06407

ExplAIner: A Declarative Query Language for Explaining Classification Models

ExplAIner：一种用于解释分类模型的声明性查询语言

Arenas, Marcelo, Barceló, Pablo, Bustamante, Diego, Caraball, Jose, Schild, María Alejandra, Subercaseaux, Bernardo

Abstract

The XAI community has studied a wide range of queries and scores for explaining predictions of ML models. From a data management perspective, this proliferation of explanation notions calls for declarative query languages in which such notions can be specified, combined, and analyzed uniformly. In this paper, we develop such a framework for Boolean models. We first revisit FOIL, an interpretability query language for black-box models, and show that it has two fundamental limitations: it cannot express central optimality-based explanation queries, and its evaluation problem over decision trees is hard for every level of the polynomial hierarchy. We then introduce ExplAIner, a query language based on FOIL with an extended vocabulary and a layered structure. We show that ExplAIner can express a broad family of explanation notions, including abductive, contrastive, feature-based, and distance-based queries. We also prove that the evaluation problem for each query in ExplAIner belongs to the Boolean hierarchy over every class of Boolean models for which some basic predicates can be evaluated in polynomial time. In particular, that property holds for deterministic and decomposable Boolean circuits. Finally, we introduce Opt-FOIL, an optimization-oriented fragment of ExplAIner for computing explanations that are minimal with respect to strict partial orders, and prove that its evaluation problem is in $\mathrm{FP}^{\mathrm{NP}}$ under the same tractability assumptions. These complexity results have a direct algorithmic consequence: a fixed ExplAIner query can be evaluated with a fixed number of calls to a SAT solver, while a notion of explanation specified in Opt-FOIL can be computed with a polynomial number of such calls. This is particularly relevant in formal XAI, where SAT solvers have been successfully used to compute explanations for several classes of ML models.

Chinese Translation

XAI（解释性人工智能）社区研究了广泛的查询和分数，以解释机器学习模型的预测。从数据管理的角度来看，这种解释概念的繁多呼唤声明性查询语言，以便能够统一地指定、组合和分析这些概念。本文为布尔模型开发了这样的框架。我们首先重新审视了FOIL，这是一种针对黑箱模型的可解释性查询语言，并指出其存在两个基本限制：它无法表达基于中心最优性的解释查询，并且其在决策树上的评估问题在多项式层次的每个级别都是困难的。然后，我们引入了ExplAIner，这是一种基于FOIL的查询语言，具有扩展的词汇和分层结构。我们证明ExplAIner可以表达广泛的解释概念，包括推理性、对比性、基于特征和基于距离的查询。我们还证明，每个ExplAIner查询的评估问题属于布尔层次，而对于某些基本谓词可以在多项式时间内评估的任一布尔模型类别，该属性都成立。特别地，这一性质适用于确定性和可分解的布尔电路。最后，我们引入了Opt-FOIL，这是ExplAIner的一个面向优化的片段，用于计算相对于严格偏序的最小解释，并证明其评估问题在相同的可控性假设下属于$ ext{FP}^{ ext{NP}}$。这些复杂性结果具有直接的算法意义：固定的ExplAIner查询可以通过固定次数的SAT求解器调用进行评估，而在Opt-FOIL中指定的解释概念可以通过多项式次数的此类调用来计算。这在正式XAI中特别相关，因为SAT求解器已成功用于计算一些类别的机器学习模型的解释。

View on arXiv Download PDF AI Translation

cs.AI / 38 / 2607.06435

Finding H. pylori in the Fine Print: Evidence-Linked Multi-Agent Case Finding from Gastric Biopsy Reports

在细节中寻找幽门螺旋杆菌：基于证据的多智能体病例发现来自胃活检报告

Wang, Yufan, Sahu, Anit Kumar, Ng, Yan Fei, Kang, Daniel, Vassef, Shayan, Shimgekar, Soorya Ram, Saha, Koustuv, Zonooz, Piyum, Kumar, Navin, Cheng, Chee Leong, Khor, Li Yan

Abstract

Data from Singapore indicated that about 31% of the population had evidence of Helicobacter pylori infection. Persistent H. pylori infection is associated with chronic active gastritis and peptic ulcer disease, and its eradication is key to gastric cancer prevention. However, evidence supporting \textit{H. pylori} positivity and H. pylori-associated gastritis may be distributed across heterogeneous coded and free-text report fields and may require contextual interpretation of assertion and negation, limiting keyword search, and making manual review difficult to scale. We conducted a retrospective pilot evaluation of the Nimblemind Multi-Agent System (nMAS), a field-name-driven, evidence-linked extraction workflow, using 54 de-identified gastric biopsy pathology reports from a large healthcare system in Singapore. Four clinician-scoped binary fields were evaluated: gastric/stomach biopsy, biopsy status, H. pylori positivity, and H. pylori-associated gastritis. Across 216 feature-case decisions, nMAS correctly classified 213, corresponding to 98.61% overall accuracy. A separately implemented UMA-style MiniMax M2.5 comparator produced similar aggregate and per-field classification metrics. Although predictive performance was similar, nMAS maintained unified report-level outputs with supporting source sentences; the demonstrated contribution is therefore workflow integration and traceability rather than predictive superiority. Under an illustrative, unmeasured scenario, reviewing 1,000 reports at five minutes per manual review versus five seconds per evidence-linked verification would reduce review time from 83.3 to 1.4 staff-hours, corresponding to 81.9 staff-hours and about USD~6,100 in potential staff-time value. Larger multi-institutional studies should evaluate evidence-span correctness, clinician verification time, and generalizability.

Chinese Translation

来自新加坡的数据表明，约31%的人口存在幽门螺旋杆菌感染的证据。持续的幽门螺旋杆菌感染与慢性活动性胃炎和消化性溃疡疾病相关，其根除是预防胃癌的关键。然而，支持幽门螺旋杆菌阳性和幽门螺旋杆菌相关胃炎的证据可能分散于异构的编码和自由文本报告字段中，并可能需要对断言和否定的上下文进行解释，这限制了关键词搜索，并使得手动审查难以扩展。我们对Nimblemind多智能体系统（nMAS）进行了回顾性试点评估，该系统是一种基于字段名称的、与证据关联的提取工作流程，使用了来自新加坡一家大型医疗系统的54份去标识化胃活检病理报告。我们评估了四个临床范围的二元字段：胃/食管活检、活检状态、幽门螺旋杆菌阳性和幽门螺旋杆菌相关胃炎。在216个特征案例决策中，nMAS正确分类了213个，对应于98.61%的整体准确率。单独实施的UMA风格的MiniMax M2.5比较器产生了类似的总分类和按字段分类指标。尽管预测性能相似，nMAS保持了统一的报告级输出与支持来源句子；因此，所展示的贡献是工作流程整合和可追溯性，而不是预测优越性。在一个说明性、未测量的场景下，以每份手动审查五分钟和每份证据关联验证五秒的速度审查1,000份报告将使审查时间从83.3小时减少到1.4小时，这相当于81.9员工小时和约6100美元的潜在员工时间价值。更大规模的多机构研究应评估证据范围的正确性、临床验证时间和普遍适用性。

View on arXiv Download PDF AI Translation

cs.AI / 39 / 2607.06447

Danus: Orchestrating Mathematical Reasoning Agents with Fact-Graph Memory

Danus：使用事实图记忆机制协调数学推理代理

Liu, Jihao, Gao, Guoxiong, Sun, Zeming, Wu, Bin, Liu, Shurui, Jiang, Jiedong, Ju, Haocheng, Chen, Leheng, Cheng, Ronnie, Zhang, Xiping, Dong, Bin

Abstract

Recent LLM-based mathematical reasoning agents have begun to tackle research-level problems and, in several cases, have contributed to the resolution of open problems. However, scaling and orchestrating such agents effectively remains challenging, due to the difficulty of coordinating parallel proof search while keeping intermediate claims organized and reliable. In this paper, we propose Danus, an orchestration system for research-level mathematical reasoning centered on a shared fact graph as a global memory-management mechanism. Danus consists of a main agent that performs planning and coordination, multiple worker agents that carry out proof search in parallel, and a stateless verifier that checks proposed mathematical claims before they are admitted into the fact graph. Each verified fact is stored together with its proof and logical dependencies, allowing the system to build long arguments incrementally while keeping the shared proof state organized. The main agent periodically summarizes the evolving proof state, redirects workers across promising directions, and supports interaction with human mathematicians through progress reports. We evaluate Danus through six research-level case studies in algebraic geometry, singularity theory, and combinatorics, illustrating how the fact-graph memory mechanism enables Danus to construct long, detailed mathematical proofs. Our results suggest that fact-graph-based orchestration provides an effective route toward scaling mathematical reasoning agents for long-horizon research problems. Danus is open source at https://github.com/frenzymath/Danus.

Chinese Translation

近期基于大型语言模型（LLM）的数学推理代理已开始解决研究级问题，并在多个案例中对开放问题的解决做出了贡献。然而，如何有效地扩展和协调这些代理仍然具有挑战性，这主要是由于在协调并行证明搜索的同时保持中间论断的有序性和可靠性所面临的困难。本文提出了 Danus，一个以共享事实图作为全局记忆管理机制的研究级数学推理协调系统。Danus 包括一个主要代理负责规划和协调，多个工人代理并行执行证明搜索，以及一个无状态的验证器在提案数学论断被纳入事实图之前进行检查。每个经过验证的事实都与其证明和逻辑依赖一起存储，使系统能够逐步构建长篇论证，同时保持共享证明状态的有序性。主要代理定期总结不断演变的证明状态，引导工人在有前景的方向上发展，并通过进展报告支持与人类数学家的互动。我们通过代数几何、奇点理论和组合数学的六个研究级案例研究对 Danus 进行了评估，展示了事实图记忆机制如何使 Danus 能够构建长而详细的数学证明。我们的结果表明，基于事实图的协调为扩展数学推理代理以应对长线研究问题提供了一条有效途径。Danus 的开源代码可在 https://github.com/frenzymath/Danus 找到。

View on arXiv Download PDF AI Translation

cs.AI / 40 / 2607.06479

A Physics-Informed Neural Network Framework for Elastodynamic Wave Propagation in Bimaterial Systems

一种基于物理信息的神经网络框架用于双材料系统中的弹性动态波传播

Chibire, Sonal Ankush, Gau, Jenn-Terng, Zhang, Bo

Abstract

Physics-informed neural networks (PINNs) provide a promising framework for solving partial differential equations while embedding the underlying physical laws directly into the learning process. This study presents a PINN-based framework for modeling transient elastodynamic wave propagation in bimaterial systems governed by the axisymmetric equations of linear elasticity. A steel-aluminum specimen representative of a Split Hopkinson Pressure Bar configuration is considered, and the governing elastodynamic equations, together with the corresponding initial, boundary, and interface conditions, are incorporated directly into the network through a physics-informed loss function. High-fidelity finite-element simulations performed using ANSYS Workbench Explicit Dynamics are used for validation and as supplementary data constraints during training. The proposed framework accurately predicts wave transmission and reflection across the bimaterial interface and reproduces axial and radial displacement histories, face-averaged responses, and the dominant stress and strain evolution with close agreement to the finite-element solutions. The trained network further demonstrates the ability to predict wave responses at previously unseen time instants and for modified material properties without requiring additional finite-element simulations, providing a continuous surrogate model for elastodynamic analysis. Mesh-sensitivity studies confirm numerical robustness, while additional material combinations demonstrate the generality of the proposed methodology. The results show that integrating physics-informed neural networks with explicit finite-element analysis provides an accurate and computationally efficient framework for elastodynamic wave propagation in heterogeneous solids, offering an effective surrogate modeling approach for high-rate solid mechanics and impact engineering applications.

Chinese Translation

物理信息神经网络（PINNs）为解决偏微分方程提供了一个有前景的框架，同时将基础物理法则直接嵌入学习过程中。本研究提出了一种基于PINN的框架，用于建模由线性弹性轴对称方程支配的双材料系统中的瞬态弹性动态波传播。考虑了一个代表分裂霍普金森压力杆配置的钢铝样本， governing elastodynamic equations 以及相应的初始、边界和界面条件通过物理信息损失函数直接纳入网络中。使用 ANSYS Workbench Explicit Dynamics 进行的高保真有限元模拟用于验证和作为训练过程中的补充数据约束。所提出的框架准确预测了双材料界面上的波传输和反射，并重现了轴向和径向位移历史、面平均响应以及主导应力和应变的演变，与有限元解有着良好的吻合。训练后的网络进一步展示了在先前未见的时间瞬间和修改材料属性下预测波响应的能力，而无需额外的有限元模拟，提供了一个用于弹性动态分析的连续替代模型。网格敏感性研究确认了数值的稳健性，而其他材料组合则展示了所提方法的普适性。结果表明，将物理信息神经网络与显式有限元分析相结合，为异质固体中的弹性动态波传播提供了一个准确且计算高效的框架，为高频固体力学和冲击工程应用提供了一种有效的替代建模方法。

View on arXiv Download PDF AI Translation

cs.AI / 41 / 2607.06489

Multi-Agent Deep Reinforcement Learning for Multi Objective Battery Management in Dairy Farms

多智能体深度强化学习在乳制品农场多目标电池管理中的应用

Victorio, Marcos Eduardo Cruz, Mason, Karl

Abstract

The dairy industry in Ireland has a large potential for the integration of renewable energy and the reduction of carbon emissions. However, researchers of distributed generation control are mainly focused on residential and commercial applications. To contribute to the effective integration of renewable energy in the dairy sector, this paper presents a multi-objective optimisation control system based on differential evolution and multi agent Deep Reinforcement Learning. The proposed control is organised in two layers: the upper layer uses dynamic pricing, and the lower layer is based on multi-agent reinforcement learning for battery management. This paper also simulates the electrical response of the proposed control system in a rural distribution circuit. The simulation results show that the proposed control framework can improve profits from energy arbitrage up to 18% compared to using Rule-based models, increase the use of distributed generation without significantly increasing cost, and comply with the Irish grid code in terms of voltage variation.

Chinese Translation

爱尔兰乳制品行业在可再生能源整合和碳排放减少方面具有巨大的潜力。然而，分布式发电控制的研究人员主要集中于住宅和商业应用。为了促进可再生能源在乳制品行业中的有效整合，本文提出了一种基于差分演化（Differential Evolution）和多智能体深度强化学习（Multi-Agent Deep Reinforcement Learning）的多目标优化控制系统。所提出的控制系统组织为两层：上层采用动态定价，下层基于多智能体强化学习进行电池管理。本文还模拟了所提出控制系统在农村配电电路中的电气响应。模拟结果表明，与基于规则的模型相比，所提控制框架可以将能源套利收益提高至18%，增加分布式发电的使用而不显著增加成本，并在电压变化方面符合爱尔兰电网规范。

View on arXiv Download PDF AI Translation

cs.AI / 42 / 2607.06503

Doomed from the Start: Early Abort of LLM Agent Episodes via a Recall-Controlled Probe Cascade

从一开始就注定失败：通过回忆控制探测级联提前中止大型语言模型代理的任务

Ruan, Kai, Huang, Zihe, Zhou, Ziqi, Wei, Qianshan, Wang, Xuan, Sun, Hao

Abstract

Large language model (LLM) agents solving multi-step tasks frequently commit to trajectories that are doomed to fail, yet continue to consume substantial inference compute before the failure becomes observable. We show that failure is predictable early from the agent's internal representations: lightweight per-round probes on hidden activations anticipate eventual episode failure as early as the first interaction round, where scorers reading only the agent's observable behavior are barely better than chance. We turn this signal into a practical abort cascade: one distribution-free calibrated gate per round, with per-round recall budgets jointly searched so that eventually-successful episodes survive all gates at a user-specified global rate; this episode-level guarantee is the one that matters in deployment, since false-abort risk accumulates across gates. Across two agent models on TextCraft, the cascade meets every recall target from 90% to 97% and, at the 90% target, saves 47.1% +/- 10.3% (Qwen-2.5-7B) and 37.2% +/- 8.8% (Llama-3.2-3B) of inference compute, 1.6--1.7x the best single-gate policy. An otherwise-identical cascade reading only behavior saves roughly half as much, and adding behavioral features to the probe yields no further gain: the hidden states capture what behavior reveals. Finally, we characterize the sample complexity of certifying high recall targets, telling practitioners which recall promises their data can, and provably cannot, back. The code will be released soon.

Chinese Translation

大型语言模型（LLM）代理在解决多步骤任务时，常常会承诺走向注定失败的轨迹，但在失败变得可观察之前，仍会消耗大量的推理计算资源。我们展示了代理的内部表征可以早期预测失败：在隐藏激活上进行的轻量级每轮探测可以在第一次交互轮中预测最终的任务失败，而仅依赖代理可观察行为的评分者几乎与随机猜测无异。我们将这一信号转化为一个实用的中止级联：每轮一个无分布假设的校准门，联合搜索每轮的回忆预算，以确保最终成功的任务以用户指定的全局比例通过所有门；这一任务级别的保证在部署中至关重要，因为错误中止的风险在各个门之间累积。在 TextCraft 上的两个代理模型中，该级联满足从 90% 到 97% 的每个回忆目标，并且在 90% 的目标下，节省了 47.1% +/- 10.3%（Qwen-2.5-7B）和 37.2% +/- 8.8%（Llama-3.2-3B）的推理计算，相当于最佳单门策略的 1.6--1.7 倍。一个在其他方面相同但仅读取行为的级联节省的计算量大约是前者的一半，而向探测中添加行为特征并未带来进一步的收益：隐藏状态捕捉了行为所揭示的信息。最后，我们描述了认证高回忆目标的样本复杂性，告知从业者他们的数据可以以及不可证明支持哪些回忆承诺。代码将很快发布。

View on arXiv Download PDF AI Translation

cs.AI / 43 / 2607.06504

RMISC: A Large-scale Real-world Multivariate Corpus for Time Series Foundation Models

RMISC：一个大规模真实世界多变量时间序列基础模型语料库

Sun, Qian, Tian, Yong-Ming, Huang, Jia-Wei, Feng, Cheng, Zhang, Shao-Qun

Abstract

Recent years have witnessed the emergence of multivariate modeling using time series foundation models (TSFMs), which achieve advanced zero-shot generalization. Modern multivariate TSFMs are predominantly pretrained on multivariate synthetic data, which is easier to scale but may fail to capture the complex temporal dynamics and cross-variable relationships present in real-world time series. This raises a key question: Whether and to what extent the leading TSFMs trained with the real-world corpus perform better than those trained with synthetic data? To answer this, we establish the RMISC corpus, a considerably large-scale, high-quality, openly accessible, real-world, and multivariate time series archive that contains around 200 datasets and 142 billion time points across diverse domains. Furthermore, we pretrain four advanced TSFMs on univariate, synthetic multivariate, and real-world multivariate data and evaluate their zero-shot generalization capabilities on standard in-distribution and out-of-distribution benchmarks. Experimental results show that incorporating real-world multivariate data predominantly improves the generalization performance for both univariate and multivariate TSFMs. These results provide a deeper understanding of how real-world multivariate data contributes to the development of stronger TSFMs.

Chinese Translation

近年来，多变量建模使用时间序列基础模型（TSFMs）的出现，展示了先进的零样本泛化能力。现代多变量 TSFMs 主要在多变量合成数据上进行预训练，这虽然更易于扩展，但可能无法捕捉真实世界时间序列中复杂的时序动态和变量间关系。这引发了一个关键问题：使用真实世界语料库训练的领先 TSFMs 在性能上是否及其程度上优于使用合成数据训练的模型？为了解答这一问题，我们建立了 RMISC 语料库，这是一个规模相当庞大、质量高、开放访问的真实世界多变量时间序列档案，包含约 200 个数据集和 1420 亿个时间点，涵盖多个领域。此外，我们在单变量、合成多变量和真实世界多变量数据上预训练了四种先进的 TSFMs，并在标准的分布内和分布外基准上评估它们的零样本泛化能力。实验结果表明，融入真实世界多变量数据显著提高了单变量和多变量 TSFMs 的泛化性能。这些结果为理解真实世界多变量数据如何有助于开发更强大的 TSFMs 提供了更深的视角。

View on arXiv Download PDF AI Translation

cs.AI / 44 / 2607.06514

FootsiesGym: A Fighting Game Benchmark for Two-Player Zero-Sum Imperfect-Information Games

FootsiesGym：一种针对双人零和不完全信息游戏的格斗游戏基准

McDonald, Chase, Tsang, Nathan, Kerr, Wesley N.

Abstract

We present FootsiesGym, an open-source environment for learning in a non-trivial two-player, zero-sum, imperfect-information game. Built on HiFight's minimalist 2D fighting game Footsies, it isolates the cyclic, non-transitive strategic interactions of fighting game neutral play while remaining simple enough for efficient analysis. We provide a vectorized simulator that enables high-throughput training on standard hardware, making the environment accessible and reproducible. We describe the design of the environment, benchmark several reinforcement learning algorithms, and discuss open research directions it enables. The code is available at https://github.com/como-research/FootsiesGym.

Chinese Translation

我们提出了FootsiesGym，这是一个用于在非平凡的双人、零和、不完全信息游戏中进行学习的开源环境。该环境基于HiFight的简约2D格斗游戏Footsies构建，隔离了格斗游戏中中立对战的循环性、非传递性战略互动，同时保持足够简单以便于高效分析。我们提供了一个向量化的模拟器，使得在标准硬件上进行高吞吐量训练成为可能，从而使该环境变得可访问和可复现。我们描述了环境的设计，基准测试了几种强化学习算法，并讨论了它所启发的开放研究方向。代码可以在 https://github.com/como-research/FootsiesGym 获取。

View on arXiv Download PDF AI Translation

cs.AI / 45 / 2607.06519

FreqDepthKV: Frequency-Guided Depth Sharing for Robust KV Cache Compression in Long-Context LLM Inference

FreqDepthKV：基于频率引导的深度共享方法用于长上下文 LLM 推理中的鲁棒 KV 缓存压缩

Córdoba, Anna, Tercero, Adam Puente, Hijo, Nerea Angulo, Tercero, Mar Linares, Barrientos, Julia, Miranda, Ainhoa, Olivera, Jesús

Abstract

Long-context LLM inference is increasingly limited by the memory and bandwidth cost of KV caches, yet aggressive compression can remove the layer-specific evidence needed for retrieval and multi-step reasoning. We introduce FreqDepthKV, an inference-time cache compression method that factorizes adjacent-layer KV states into shared low-frequency depth components and sparse high-frequency residuals. A lightweight online probe assigns attention heads to shared-depth, residual-depth, or exact cache modes according to their contribution to reconstruction-sensitive attention logits, allowing the compression policy to adapt to prompt structure without retraining. Across long-context question answering, needle retrieval, summarization, and code generation benchmarks, FreqDepthKV preserves task accuracy under substantially smaller cache budgets. With a 32k-token prefill window, FreqDepthKV reaches 58.3 Exact Match, 63.0 F1, 32.5 ROUGE-L, and 48.1 pass@1, closely matching full KV while outperforming prior compressed-cache methods. It also improves decoding throughput to 70.4 tokens/s, reduces TTFT to 2.06 seconds, and lowers peak KV memory to 6.2 GB, achieving a 3.9x effective compression ratio.

Chinese Translation

长上下文 LLM 推理越来越受到 KV 缓存的内存和带宽成本的限制，但激进的压缩可能会去除检索和多步推理所需的层特定证据。我们提出了 FreqDepthKV，这是一种推理时缓存压缩方法，它将相邻层的 KV 状态分解为共享的低频深度组件和稀疏的高频残差。一个轻量级的在线探测器根据注意力头对重建敏感的注意力 logits 的贡献，将其分配到共享深度、残差深度或精确缓存模式，从而使压缩策略能够在不重新训练的情况下适应提示结构。在长上下文问答、针检索、摘要生成和代码生成基准测试中，FreqDepthKV 在显著较小的缓存预算下保持了任务准确性。在 32k 令牌的预填充窗口中，FreqDepthKV 达到了 58.3 的精确匹配率、63.0 的 F1 值、32.5 的 ROUGE-L 和 48.1 的 pass@1，接近完整 KV 的表现，同时超越了之前的压缩缓存方法。它还将解码吞吐量提高到 70.4 令牌/秒，减少 TTFT 到 2.06 秒，并将峰值 KV 内存降低到 6.2 GB，实现了 3.9 倍的有效压缩比。

View on arXiv Download PDF AI Translation

cs.AI / 46 / 2607.06522

Bridging Physical Reasoning and Task Generalization via Visual Action Outcome Reasoning Alignment

通过视觉行动结果推理对齐架起物理推理与任务泛化的桥梁

Ko, Han-Jun, Chen, Jr-Jen, Yuan, Haobo, Lee, Hsin-Ying, Shen, Tiancheng, Yang, Ming-Hsuan, Wang, Yu-Chiang Frank

Abstract

Vision-language models (VLMs) struggle to generalize in interactive physical reasoning, particularly under unseen tasks and environments. Two key failure modes are prominent: hallucinated chain-of-thought (CoT) reasoning that contradicts physical reality, and misalignment between the model's reasoning and actions. We present VAORA (Visual Action Outcome Reasoning Alignment), a novel reward design that directly addresses both issues. VAORA introduces two complementary rewards: Visual Alignment Reward, which anchors VLM reasoning to the visual context independent of the agent action itself, and Visual-Action Alignment Reward, which grounds reasoning in the visual outcome induced by the model's action. Together, these rewards suppress hallucinated CoT and reduce the gap between reasoning and behavior. To improve training stability, we further employ smooth, dense rewards by estimating success probabilities using a pre-trained in-domain expert agent. Experiments on PHYRE and Virtual Tool support our performances across novel-task and unseen-environment settings, confirming that grounded and generalizable physical intelligence can be induced through VAORA.

Chinese Translation

视觉-语言模型（VLMs）在互动物理推理中面临着泛化困难，尤其是在未见任务和环境下。两种主要的失败模式尤为突出：与物理现实相矛盾的虚幻思维链（CoT）推理，以及模型的推理与行动之间的不对齐。我们提出了VAORA（视觉行动结果推理对齐），这是一种新颖的奖励设计，直接解决了这两个问题。VAORA引入了两个互为补充的奖励：视觉对齐奖励，它将VLM推理锚定到独立于代理行动本身的视觉上下文中，以及视觉-行动对齐奖励，它将推理基于模型行动引发的视觉结果。这两个奖励共同抑制了虚幻的CoT，缩小了推理与行为之间的差距。为了提高训练稳定性，我们进一步通过使用在领域内预训练的专家代理来估计成功概率，从而采用平滑和密集的奖励。在PHYRE和Virtual Tool上的实验支持了我们在新任务和未见环境设置下的表现，确认了通过VAORA可以诱导出扎根且具有泛化能力的物理智能。

View on arXiv Download PDF AI Translation

cs.AI / 47 / 2607.06523

DepthWeave-KV: Token-Adaptive Cross-Layer Residual Factorization for Long-Context KV Cache Compression

DepthWeave-KV：用于长上下文键值缓存压缩的基于令牌自适应跨层残差因式分解

Cordoba, Anna, Tercero, Adam Puente, Hijo, Nerea Angulo, Tercero, Mar Linares, Barrientos, Julia, Miranda, Ainhoa, Olivera, Jesus

Abstract

Long-context language model inference is increasingly limited by the memory bandwidth and capacity required to store key-value caches, yet existing compression methods often apply uniform budgets across layers or tokens and degrade retrieval when lexical cues and semantic states require different preservation. We introduce DepthWeave-KV, a token-adaptive cache compression method that factorizes key and value states across neighboring transformer layers using shared low-rank channel bases while retaining lightweight token-specific residuals where attention behavior is sensitive. DepthWeave-KV combines cross-depth residual factorization with a token-conditional depth router that allocates higher reconstruction rank to instruction-bearing and retrieval-critical tokens, and uses calibration-free online error tracking from attention-output probes to adapt compression during generation without retraining the base model. A fused CUDA implementation jointly performs basis lookup, residual dequantization, and attention projection to reduce decode-time memory traffic. Across LongBench, Needle-in-a-Haystack, L-Eval, and long-form QA and summarization benchmarks, DepthWeave-KV achieves near-full-cache task quality with substantially lower memory use, improving average score and retrieval accuracy over prior compressed caches while reaching 8.3x KV memory reduction and 72.8 tokens per second at 64K context.

Chinese Translation

长上下文语言模型推理逐渐受到存储键值缓存所需的内存带宽和容量的限制，而现有的压缩方法往往在各层或各个令牌之间应用统一的预算，从而在语言线索和语义状态需要不同保留时导致检索效果降低。我们提出了DepthWeave-KV，这是一种基于令牌自适应的缓存压缩方法，通过使用共享的低秩通道基对相邻的变换器层之间的键和值状态进行因式分解，同时在注意力行为敏感的位置保留轻量级的特定令牌残差。DepthWeave-KV结合了跨深度残差因式分解和一种基于令牌条件的深度路由器，后者对承载指令和对检索至关重要的令牌分配更高的重构秩，并使用来自注意力输出探测器的无校准在线误差跟踪，在生成过程中适应压缩，而无需重新训练基本模型。通过融合的CUDA实现，联合执行基底查找、残差反量化和注意力投影，以减少解码时的内存流量。在LongBench、Needle-in-a-Haystack、L-Eval以及长格式问答和摘要基准测试中，DepthWeave-KV在显著降低内存使用的同时，实现了近乎全缓存的任务质量，提升了平均得分和检索准确性，达到了8.3倍的键值内存减少，并在64K上下文下实现了每秒72.8个令牌的速度。

View on arXiv Download PDF AI Translation

cs.AI / 48 / 2607.06531

The Large Cancer Assistant (LCA): A Model-Agnostic Orchestration Framework for Scalable Clinical Decision Support in Oncology

大型癌症助手（LCA）：一个与模型无关的可扩展临床决策支持的编排框架

Marrakchi, Ghassen, Matei, Basarab

Abstract

- Objective: Multimodal deep learning models in oncology are currently limited by monolithic designs that rigidly couple data ingestion, clinical routing, and artificial intelligence (AI) inference. To address this inflexibility, we propose the Large Cancer Assistant (LCA), a model-agnostic, post-hoc orchestration framework designed for scalable clinical decision support. - Methods: The LCA is mathematically formalized as a 7-tuple architecture grounded in the principle of Algorithmic Impermeability, ensuring the orchestration logic remains strictly independent of underlying black-box AI models. We introduce the Entry Theory, leveraging Geometric Deep Learning (GDL) to standardize multimodal patient data along distinct structural and medical axes. The system dynamically orchestrates data via a Cancer Switching Module and intentionally isolates the core AI execution from volatile hospital IT infrastructures by outputting a Standardized Intermediate Payload (SIP). - Results: A Proof of Concept (PoC) validated the orchestration logic across four technical scenarios. The framework executed a nominal flow with negligible orchestration overhead. It empirically demonstrated algorithmic impermeability by maintaining an invariant routing projection during AI model swaps, and it validated strict failure-safety by achieving a 100\% recall rate in generating targeted Supplementary Data Requests (SDR) under injected data anomalies. Multi-protocol execution capability was also successfully verified. - Conclusion: By structurally decoupling multimodal ingestion from feature inference, the LCA provides a highly adaptable and modular orchestration foundation. The SIP establishes a clear architectural boundary, natively setting the stage for downstream Electronic Medical Record (EMR) interoperability as an independent future paradigm.

Chinese Translation

目标：肿瘤学中的多模态深度学习模型目前受到单一设计的限制，这种设计将数据摄取、临床路径和人工智能（AI）推理紧密结合。为了应对这种不灵活性，我们提出了大型癌症助手（LCA），一个与模型无关的事后编排框架，旨在提供可扩展的临床决策支持。方法：LCA在数学上被形式化为一个基于算法不可渗透性原则的7元组架构，确保编排逻辑严格独立于底层黑箱AI模型。我们引入了入口理论，利用几何深度学习（GDL）标准化多模态患者数据，在不同的结构和医疗轴上进行规范化。该系统通过癌症切换模块动态编排数据，并通过输出标准化中间负载（SIP）故意将核心AI执行与波动的医院IT基础设施隔离。结果：概念验证（PoC）在四个技术场景中验证了编排逻辑。该框架以可忽略的编排开销执行了名义流程。它通过在AI模型切换过程中保持不变的路由投影，经验上证明了算法不可渗透性，并在数据异常注入下成功实现了100%的召回率，验证了严格的故障安全。在多协议执行能力方面也成功得到了验证。结论：通过在结构上将多模态摄取与特征推理解耦，LCA提供了一个高度可适应和模块化的编排基础。SIP建立了明确的架构界限，为下游电子病历（EMR）互操作性作为独立的未来范式奠定了基础。

View on arXiv Download PDF AI Translation

cs.AI / 49 / 2607.06544

Rethinking Indic AI from a Lens of Cultural Heritage Preservation

从文化遗产保护的视角重新思考印度人工智能

Madva, Aparna, Srivatsa, Sharath, Srinivasa, Srinath, Saha, Tulika

Abstract

As Artificial Intelligence (AI) makes inroads into different parts of the Indian subcontinent, there is significant interest in studying how AI impacts the linguistic and cultural foundations of this civilization. AI is seen as a ''double-edged sword'' where on the one hand, it can enable access and inclusion for a large population, on the other, it can homogenize worldviews and exclude underrepresented languages and worldviews. In this paper, we try to characterize this problem by addressing the extensive characteristic nature of Indian linguistics and the way they closely connect to cultural practices and worldview. We then perform a longitudinal survey of how Natural Language Processing (NLP) techniques have evolved in this space, tracing the historical development of Indic NLP, covering key milestones, methodological shifts, and resource creation efforts. In addition, the paper also examines the structural and sociolinguistic characteristics of Indian languages, such as rich morphology, complex scripts and grammar rules, diglossia, and large dialectal variation, and explains how these create unique challenges for building AI foundation models. We then discuss the growing role of Indic foundation models and analyze how these models address these long-standing resource and representation gaps. Finally, we propose a research direction called 'Culture Sensing', which re-imagines AI based on hermeneutic reasoning. Culture Sensing aims to address open problems such as ensuring equitable performance across low-resource languages and producing outputs that are culturally meaningful. By bringing together past work, current techniques, and emerging trends, this paper outlines research directions that can guide the next phase of Indic NLP and contribute to the development of more robust and inclusive Indic foundation models.

Chinese Translation

随着人工智能（AI）在印度次大陆不同地区的深入发展，研究AI如何影响这一文明的语言和文化基础引起了广泛关注。AI被视为一把“双刃剑”，一方面它可以为大量人群提供访问和包容，另一方面它可能导致世界观的同质化，并排除那些代表性不足的语言和世界观。本文试图通过阐述印度语言学的广泛特征以及它们与文化实践和世界观的紧密联系来表征这一问题。随后，我们对自然语言处理（NLP）技术在这一领域的发展进行了纵向调查，追溯印度NLP的历史发展，涵盖关键的里程碑、方法论转变和资源创造努力。此外，本文还考察了印度语言的结构和社会语言学特征，例如丰富的形态学、复杂的书写系统和语法规则、 diglossia 以及大的方言变异，并解释这些特征如何对构建AI基础模型带来独特挑战。我们接着讨论了印度基础模型日益增长的作用，并分析这些模型如何解决长期存在的资源和代表性差距。最后，我们提出了一个名为“文化感知”（Culture Sensing）的研究方向，它基于诠释学推理重新构想AI。文化感知旨在解决一些开放性问题，例如确保低资源语言的公平性能，以及生成具有文化意义的输出。通过整合过去的工作、当前的技术和新兴趋势，本文勾勒出可以指导印度NLP下一阶段研究及促进更稳健、更具包容性的印度基础模型发展的研究方向。

View on arXiv Download PDF AI Translation

计算语言学 (Computation and Language)

cs.CL / 1 / 2607.05398

How Personas Can Influence Agents to Play Split or Steal

角色如何影响代理在分裂或窃取游戏中的行为

Leon, Carlos, Rodrigues, Alexandre, Gamito, Pedro, Parsons, Thomas D.

Abstract

Personas are often employed to guide large language model agents, yet their effectiveness in shaping strategic behavior in social dilemma settings remains uncertain. To address this, we examined the impact of persona prompts in an iterated Split or Steal game where persona-driven agents interacted with a Virtual Human (VH) controlled by a fixed prompt. Agents were instantiated from four open models (Ministral 3:3b, phi4:14b, Gemma3:12b, and Gemma4:e4b) at two temperature settings (0.3 and 0.7) and deterministic decision with zero temperature, while the VH was powered by GPT 4.1 mini. Across 160 sessions of 15 rounds each conducted in European Portuguese, mutual Split outcomes dominated (roughly 74 percent of rounds), with exploitation occurring in fewer than 11 percent of rounds. Model choice significantly influenced behavior: phi4 and Ministral 3:3b remained consistently cooperative across temperatures, whereas Gemma3:12b and Gemma4:e4b exhibited more varied strategies and outcomes. Analyses based on Big Five personality traits indicated that Prosocial and Principled personas were most consistently cooperative, while Analytical personas were more likely to exploit the VH. Topic analysis revealed that friendship-related dialogue aligns with Split decisions, whereas money and vengeance-related content is more prevalent in Steal outcomes; sentiment labels were predominantly neutral or happy and provided limited additional explanatory value. These findings characterize the interaction between persona prompts and model differences in repeated trust games and serve as a baseline for planned virtual reality studies involving human participants interacting with an embodied VH.

Chinese Translation

角色通常用于指导大型语言模型代理，但它们在社会困境情境中塑造战略行为的有效性仍然不确定。为了解决这个问题，我们研究了角色提示在反复进行的分裂或窃取游戏中的影响，其中角色驱动的代理与由固定提示控制的虚拟人（Virtual Human, VH）进行互动。代理来自四个开放模型（Ministral 3:3b、phi4:14b、Gemma3:12b 和 Gemma4:e4b），在两个温度设置（0.3 和 0.7）以及零温度下的确定性决策中进行实例化，而 VH 则由 GPT 4.1 mini 提供支持。在用欧洲葡萄牙语进行的 160 次 15 回合的实验中，互相分裂的结果占主导地位（约 74% 的回合），而剥削发生在不到 11% 的回合中。模型选择显著影响行为：phi4 和 Ministral 3:3b 在不同温度下始终保持合作，而 Gemma3:12b 和 Gemma4:e4b 则表现出更为多样的策略和结果。基于五大人格特质的分析表明，亲社会和原则性角色最为一致地表现出合作，而分析性角色则更可能剥削 VH。主题分析显示，与友谊相关的对话与分裂决策一致，而与金钱和复仇相关的内容在窃取结果中更为普遍；情感标签主要为中性或快乐，提供的额外解释价值有限。这些发现描绘了角色提示与模型差异在重复信任游戏中的互动，并为计划中的虚拟现实研究奠定了基础，该研究涉及人类参与者与具身的 VH 进行互动。

View on arXiv Download PDF AI Translation

cs.CL / 2 / 2607.05399

Benchmarking KV-Cache Optimizations across Task Quality and System Performance for Long-Context Serving

针对长上下文服务的任务质量与系统性能的 KV-缓存优化基准测试

Agrawal, Nikita, Mayer, Ruben

Abstract

Large language model serving is increasingly limited by KV-cache growth under long-context workloads, yet existing KV-cache compression techniques are difficult to compare because they were evaluated on different models, tasks, budgets, and serving stacks. This paper presents a workload-aware benchmark of representative KV-cache optimization mechanisms spanning quantization, pruning, and merging, including KIVI, TurboQuant, SnapKV, and CaM, evaluated on LongBench-style multi-document QA, single-document QA, few-shot learning, and summarization workloads using Llama-3.1-8B-Instruct and Mistral-7B-Instruct-v0.3. The benchmark measures task quality, mean output throughput, mean time-to-first-token, and realized compression ratio across context-length buckets. The results show that the compression ratio alone is a poor predictor of end-to-end performance. KIVI4 provides the most stable quality across models, SnapKV delivers the strongest long-context throughput, and CaM yields large gains on selected QA workloads but exhibits substantial workload sensitivity in both quality and realized compression ratio. These findings motivate workload-aware selection of KV-cache mechanisms rather than one-size-fits-all compression and provide deployment guidance for long-context serving systems.

Chinese Translation

大型语言模型的服务在长上下文工作负载下越来越受到 KV-缓存增长的限制，然而现有的 KV-缓存压缩技术因在不同模型、任务、预算和服务堆栈上进行评估而难以比较。本文提出了一种基于工作负载的基准测试，涵盖了代表性的 KV-缓存优化机制，包括量化、剪枝和合并，具体评估了 KIVI、TurboQuant、SnapKV 和 CaM，在 LongBench 风格的多文档问答、单文档问答、少样本学习和摘要工作负载上使用 Llama-3.1-8B-Instruct 和 Mistral-7B-Instruct-v0.3 进行评估。该基准测试测量了任务质量、平均输出吞吐量、首次标记所需时间和各种上下文长度桶中的实现压缩比。结果表明，压缩比本身对端到端性能的预测能力较差。KIVI4 在不同模型中提供了最稳定的质量，SnapKV 在长上下文吞吐量上表现最佳，CaM 在选定的问答工作负载上带来了显著的提升，但在质量和实现压缩比方面对工作负载表现出相当大的敏感性。这些发现促使我们更加关注工作负载导向的 KV-缓存机制选择，而非一刀切的压缩方法，并为长上下文服务系统的部署提供了指导。

View on arXiv Download PDF AI Translation

cs.CL / 3 / 2607.05416

Text Distance from Nested and Hierarchical Repetitions: A Compression-Based Perspective

嵌套和层次重复的文本距离：基于压缩的视角

Hu, Xiaojun, Wang, Jing, Zhang, Jingwen, Zhai, Fengyao, Xie, Xiao, Liao, Hao, Di, Zengru, Liu, Yu

Abstract

We present a new method for structural sequence analysis grounded in Algorithmic Information Theory (AIT). At its core is the Ladderpath approach, which extracts nested and hierarchical relationships among repeated substructures in linguistic sequences -- an instantiation of AIT's principle of describing data through minimal generative programs. These structures are then used to define three distance measures: a normalized compression distance (NCD), and two alternative distances derived directly from the Ladderpath representation. Integrated with a $k$-nearest neighbor classifier, these distances achieve strong and consistent performance across in-distribution, out-of-distribution (OOD), and few-shot text classification tasks. In particular, all three methods outperform both gzip-based NCD and BERT under OOD and low-resource settings. These results demonstrate that the structured representations captured by Ladderpath preserve intrinsic properties of sequences and provide a lightweight, interpretable, and training-free alternative for text modeling. This work highlights the potential of AIT-based approaches for structural and domain-agnostic sequence understanding.

Chinese Translation

我们提出了一种基于算法信息理论（Algorithmic Information Theory, AIT）的结构序列分析新方法。其核心是Ladderpath方法，该方法提取语言序列中重复子结构之间的嵌套和层次关系——这是AIT以最小生成程序描述数据原则的一种具体体现。然后，这些结构用于定义三种距离度量：一种归一化压缩距离（Normalized Compression Distance, NCD），以及两种直接源自Ladderpath表示的替代距离。与$k$-最近邻分类器集成，这些距离在分布内、分布外（Out-of-Distribution, OOD）和小样本文本分类任务中均表现出强劲且一致的性能。特别是，在OOD和低资源设置下，所有三种方法均优于基于gzip的NCD和BERT。这些结果表明，通过Ladderpath捕获的结构化表示保留了序列的内在属性，并为文本建模提供了一种轻量级、可解释、无训练的替代方案。本研究强调了基于AIT的方法在结构和领域无关的序列理解中的潜力。

View on arXiv Download PDF AI Translation

cs.CL / 4 / 2607.05545

Most LLM Conformity Needs No Speaker: Measuring the Speaker-Free Floor in Peer-Pressure Benchmarks

大多数 LLM 的一致性无需发言者：在同伴压力基准中测量无发言者环境

Hu, Yibo, Qu, Jiaming

Abstract

LLM conformity is often used to describe cases where a model changes a correct answer toward a peer or group response. We show that most of this apparent conformity survives even after the peer is removed. The reason is a confound: standard conformity prompts mix two cues at once, the presence of a speaker and the repeated wrong answer itself. Existing benchmarks vary these cues together, so they cannot tell how much of the revision actually depends on the speaker. We introduce a no-source condition: the same asserted answer with the explicit speaker removed. Across six open-weight LLMs and seven QA and reasoning datasets, this condition alone causes harmful revision in $66.5\%$ of initially correct cases, compared with $10.3\%$ under a plain re-ask. The effect also remains when the repeated answer is paraphrased and when answer options are hidden in an open-ended setting. Source framing mainly modulates this floor: expert-panel framing raises it, while minimal person labels do not reliably raise it. When models flip, they are usually confidently wrong, and simple recalibration does not recover the original answer. Source attribution still matters, but it should be measured as an increment above this speaker-free floor. The methodological lesson is that conformity benchmarks should first measure what remains after the speaker is removed; without this step, benchmarks may mistake repeated text for social influence.

Chinese Translation

LLM 一致性通常用来描述模型在面对同伴或群体回答时改变正确答案的情况。我们展示，大多数这种明显的一致性在去除同伴之后仍然存在。原因在于一个混杂因素：标准的一致性提示同时混合了两个线索，发言者的存在和重复的错误答案。现有基准一起变化这些线索，因此无法分辨修正中有多少确实依赖于发言者。我们引入了一种无来源条件：同样断言的答案在明确移除发言者的情况下。在六个开放权重 LLM 和七个问答及推理数据集上，仅此条件就导致了 $66.5\%$ 最初正确案例的有害修正，而在普通的重新提问下这一比例为 $10.3\\%$。这一效应在重复答案被改述和答案选项在开放式环境中被隐藏时依然存在。来源框架主要调节这一基线：专家小组框架提高了它，而最小化的人物标签则未能可靠提升。当模型翻转时，它们通常是自信地错误，简单的重新校准无法恢复原答案。来源归属仍然很重要，但需要作为在此无发言者基线之上的增量进行测量。方法论的教训是，一致性基准应该首先测量在移除发言者后剩余的部分；没有这一步，基准可能会将重复的文本误认为是社会影响。

View on arXiv Download PDF AI Translation

cs.CL / 5 / 2607.05552

The yes-no bias of large language models reflects answer order and wording, not shifts in moral judgment

大型语言模型的是非偏向反映了答案顺序和措辞，而非道德判断的变化

Huang, Haonan

Abstract

Large language models (LLMs) increasingly issue judgments read as binary verdicts, and a growing literature reports such judgments shifting under logically irrelevant changes of wording - among them an amplified yes-no bias on moral dilemmas, absent in humans. A single framing cannot say what such a shift is: in a yes/no question the word "no" is at once logical verdict, lexical token, and last-printed option. We introduce a psychometric battery that separates these: crossed symmetrization - every logically irrelevant factor flipped in balanced pairs - across a corpus of question forms. A graded rating across logically equivalent forms recovers a coherent internal moral scale: frontier models' stance $\theta$ is nearly format-invariant (cross-form incoherence 0.12-0.21 on a $\pm 1$ axis); small open-weight models fail in model-specific ways. Forcing the verdict through yes/no overlays a decomposable artifact: an order bias toward the last-printed option - opposite to classic human primacy - plus a lexical pull toward the word "no"; the artifact is substantial only in the Claude models (story-averaged -0.32 to -0.86), $\approx 0$ for GPT-5.5 and Gemini, and shrinks under extended reasoning. The word and the verdict share one token; swapping the words for arbitrary labels separates them, and the verdict-attached logical bias proves $\approx 0$ for every frontier model, while model-specific label and order attachments remain: the models are not drawn toward rejecting - the pull follows the printed surface, not the verdict it carries. A minimal model, $P = \sigma((\theta \pm m)/s)$, summarizes any such artifact by a framing susceptibility m and a moral decisiveness s, measurably distinct from sampling temperature. The battery applies unchanged to any dilemma set and binary format: measuring what a model values requires crossing the frames of the question, not asking once.

Chinese Translation

大型语言模型（LLMs）越来越倾向于发表被解读为二元裁决的判断，并且越来越多的文献报告指出，这些判断在逻辑上无关的措辞变化下会发生转变——其中包括在道德困境中加剧的是非偏向，而这一现象在人的判断中并不存在。单一的框架无法明确这种转变的含义：在一个是/否问题中，单词“否”同时是逻辑裁决、词汇符号和最后印刷的选项。我们提出了一种心理测量工具，它将这些因素分开：交叉对称化——每个逻辑上无关的因素在平衡对中翻转——跨越不同的问题形式。一种在逻辑上等价的形式中的分级评分恢复了一个连贯的内部道德尺度：前沿模型的立场$ heta$几乎是格式不变的（交叉形式不一致性在$ ext{±}1$轴上为0.12-0.21）；小型开权重模型以特定的方式表现不佳。通过是/否强制裁决会形成一个可分解的伪影：对最后印刷选项的序列偏向——与经典人类优先性相反——加上对单词“否”的词汇吸引；这一伪影在Claude模型中较为显著（故事平均值为-0.32至-0.86），在GPT-5.5和Gemini中约为0，并在延长推理过程中减小。单词和裁决共享一个符号；用任意标签交换这些单词将它们分开，而裁决附加的逻辑偏差对于每个前沿模型都证明约为0，而特定模型的标签和顺序附加仍然存在：模型并不倾向于拒绝——这种吸引遵循印刷表面，而不是它所承载的裁决。一个最小模型，$P = ext{σ}(( heta ext{±} m)/s)$，通过框架敏感性m和道德果断性s总结任何此类伪影，这在量测上与采样温度显著不同。该工具可以不变地应用于任何困境集和二元格式：测量模型的价值需要跨越问题框架，而不是仅仅问一次。

View on arXiv Download PDF AI Translation

cs.CL / 6 / 2607.05554

Prompt Robustness Is Task-Dependent: Comparing Objective and Belief-Style Questions in LLM Evaluation

提示鲁棒性依赖于任务：比较大型语言模型评估中的客观问题与信念风格问题

Kamal, Sadia, Patwary, Arefa, Marchiafava, Anthony, Sen, Atriya, Choudhury, Sagnik Ray

Abstract

Survey-style evaluations of large language models often treat a prompted response as a measure of a model's values or beliefs. This assumption is particularly fragile when responses are read as evidence of political values, social attitudes, or beliefs. We ask whether prompt robustness differs between objective questions with fixed answers and subjective questions that ask for opinions or values. We evaluate four instruction-tuned model families on three objective datasets (MMLU, ARC, and CulturalBench) and three subjective datasets (Political Compass Test, ValueBench, and World Values Survey). For each question/statement, we apply multiple types of prompt changes, such as variations in wording, framing, and format, and measure whether the model gives the same answer across variants. Using a binomial generalized estimating equation, we find significant effects of model, dataset, prompt category, and their interactions. The dataset type effect is also significant, and the interaction between dataset type and prompt category is large. These results show that prompt robustness depends on the question type, the prompt change, and the model.

Chinese Translation

对大型语言模型的调查式评估通常将提示响应视为模型价值观或信念的衡量标准。当响应被解读为政治价值观、社会态度或信念的证据时，这一假设尤其脆弱。我们探讨了在固定答案的客观问题与要求表达意见或价值观的主观问题之间，提示鲁棒性是否存在差异。我们在三个客观数据集（MMLU、ARC 和 CulturalBench）以及三个主观数据集（政治指标测试、ValueBench 和世界价值观调查）上评估了四个指令调整模型家族。对于每个问题/陈述，我们应用了多种类型的提示变化，例如措辞、框架和格式的变化，并测量模型在不同变体中是否给出相同的答案。使用二项一般估计方程，我们发现模型、数据集、提示类别及其交互作用存在显著影响。数据集类型的影响也显著，数据集类型与提示类别之间的交互作用较大。这些结果表明，提示鲁棒性依赖于问题类型、提示变化以及模型。

View on arXiv Download PDF AI Translation

cs.CL / 7 / 2607.05583

ResonatorLM: Causal Resonant Field Mixing for Efficient Long-Context Language Modelin

ResonatorLM：用于高效长上下文语言建模的因果共振场混合

Chaudhury, Archie

Abstract

Contemporary language models are dominated by the transformer architecture, which leverages self-attention mechanisms to enable more efficient, parallelized training across a wide set of documents and corpora. This has allowed transformers to effectively model data across a wide range of modalities and contexts. However, transformers, along with their conventional counterparts such as recurrent neural networks (RNNs) and convolutional neural networks (CNNs), often struggle to maintain efficiency when processing long contexts. We introduce ResonatorLM, a new mechanism that replaces attention with a physics-derived alternative. ResonatorLM treats token sequences as a single, driven one-dimensional latent field and replaces attention dot products with causal functions of damped resonators. We implement ResonatorLM on a traditional network architecture and test it on standard long-context modeling tasks. We find that in a small, 6M matched setting, training and prefill speedups increase with sequence length, decode speed reaches 6.47x compared to that of a standard, optimized transformer at 32K tokens, and accuracy reaches 61.31 percent (compared to 55.32 percent) on WikiText.

Chinese Translation

当代语言模型主要以变压器架构为主，该架构利用自注意力机制在广泛的文档和语料库中实现更高效的并行训练。这使得变压器能够有效地建模各种模态和上下文的数据。然而，变压器以及其传统对应物，如递归神经网络（RNN）和卷积神经网络（CNN），在处理长上下文时往往难以保持效率。我们提出了ResonatorLM，这是一种新的机制，用物理学派生的替代方案替代注意力。ResonatorLM将令牌序列视为一个单一的、受驱动的一维潜在场，并用因果阻尼共振器的函数替代注意力点积。我们在传统网络架构上实现了ResonatorLM，并在标准长上下文建模任务上进行了测试。我们发现，在一个小型的6M匹配设置中，训练和预填充速度随着序列长度的增加而提高，解码速度在32K令牌时达到标准优化变压器的6.47倍，准确率达到61.31%（相比之下为55.32%）在WikiText上。

View on arXiv Download PDF AI Translation

cs.CL / 8 / 2607.05612

Revisiting the Relation Between Language Model Perplexity and ASR Word Error Rate for Modern End-to-End Speech Recognition

重新审视语言模型困惑度与现代端到端语音识别的自动语音识别字错误率之间的关系

Zeineldeen, Mohammad, Zeyer, Albert, Zhang, Haoran, Schmitt, Robin, Schlüter, Ralf, Ney, Hermann

Abstract

Language model (LM) perplexity (PPL) has historically been used as a proxy for automatic speech recognition (ASR) word error rate (WER), with prior work reporting an approximately linear relation in log-log space. Modern end-to-end ASR systems challenge this assumption because they already contain internal language modeling capacity, are often evaluated without external language models, and can now be combined with neural LMs and large language models (LLMs) through different recognition strategies. This paper revisits the relation between PPL and WER for modern ASR systems. We study whether external LMs still improve current end-to-end ASR systems, whether the PPL-WER relation remains linear in log-log space, how encoder context length affects this relation, and how LLM perplexities fit into the trend observed for standard neural LMs. We further investigate internal language modeling (ILM) in attention-based encoder-decoder systems and show that ILM subtraction changes the observed PPL-WER relation, indicating that the decoder's internal LM must be considered when interpreting the effect of external LM quality.

Chinese Translation

语言模型（LM）困惑度（PPL）历来被用作自动语音识别（ASR）字错误率（WER）的代理，之前的研究报告在对数-对数空间中呈现出大致线性的关系。现代端到端ASR系统挑战了这一假设，因为它们已经具备内部语言建模能力，通常在没有外部语言模型的情况下进行评估，并且现在可以通过不同的识别策略与神经语言模型（neural LMs）和大型语言模型（LLMs）相结合。本文重新审视了现代ASR系统中PPL与WER之间的关系。我们研究了外部LM是否仍能改善当前的端到端ASR系统，PPL-WER关系在对数-对数空间中是否仍保持线性，编码器上下文长度如何影响这一关系，以及LLM的困惑度如何适应标准神经LM所观察到的趋势。我们进一步探讨了基于注意力的编码器-解码器系统中的内部语言建模（ILM），并表明ILM的减法改变了观察到的PPL-WER关系，指出在解读外部LM质量的影响时必须考虑解码器的内部LM。

View on arXiv Download PDF AI Translation

cs.CL / 9 / 2607.05614

BaFCo: A Document Understanding Benchmark for Complex Bangla Form Comprehension

BaFCo：一个用于复杂孟加拉表单理解的文档理解基准

Azad, Abu Tyeb, Apan, Ishita Sur, Ahmed, Fahim, Katha, Sumaiya Karim, Jubaer, Ezharuddin, Alam, Armun, Nandi, Pranjal Kumar, Ali, Amin Ahsan, Chadha, Aman, Islam, Md Mofijul, Rahman, AKM Mahbubur

Abstract

Document comprehension is a challenging yet impactful task for Multimodal Large Language Models, especially as these systems see growing adoption in real-world, human-centric applications. However, this adoption is limited for low-resource languages such as Bangla due to the scarcity of high-quality annotated data. To address this gap, we introduce BaFCo, a benchmark dataset for Bangla form comprehension with a focus on Document Layout Analysis (DLA) and Key Information Extraction (KIE). BaFCo curates 200 multi-page complex Bangladeshi government forms, sourced from across diverse sectors including agriculture, education, banking, and land management. To accurately capture the structural and contextual complexity of these forms, we define a fine-grained annotation schema comprising 26 types of form entities, along with a separate coarse form entity set consisting of 5 types. We evaluate the latest MLLMs from the ChatGPT, Gemini, Claude, Qwen, and Kimi series using zero-shot and chain-of-thought prompts under both low and high reasoning setups. Our results reveal limitations in current MLLMs' ability in comprehending Bangla forms, particularly in accurately localizing highly granular form entities. Our dataset and code is available at: https://huggingface.co/datasets/Mausul/bafco

Chinese Translation

文档理解是一个具有挑战性但影响深远的任务，尤其是在多模态大型语言模型（Multimodal Large Language Models）中，这些系统在现实世界以人为中心的应用中正逐渐被广泛采用。然而，由于高质量注释数据的稀缺，这种技术在孟加拉等低资源语言中的应用受到限制。为了解决这一问题，我们引入了BaFCo，一个针对孟加拉表单理解的基准数据集，重点关注文档布局分析（Document Layout Analysis, DLA）和关键信息提取（Key Information Extraction, KIE）。BaFCo 汇集了200个来自农业、教育、银行和土地管理等各个领域的多页复杂孟加拉国政府表单。为了准确捕捉这些表单的结构和语境复杂性，我们定义了一个细致的注释模式，包括26种表单实体类型，以及一个包含5种类型的粗略表单实体集合。我们使用零样本（zero-shot）和思维链（chain-of-thought）提示对ChatGPT、Gemini、Claude、Qwen和Kimi系列的最新多模态大型语言模型进行评估，涵盖低推理和高推理设置。我们的结果揭示了当前多模态大型语言模型在理解孟加拉表单时的局限性，特别是在准确定位高度细化的表单实体方面。我们的数据集和代码可以在以下网址获得：https://huggingface.co/datasets/Mausul/bafco

View on arXiv Download PDF AI Translation

cs.CL / 10 / 2607.05623

NAVER LABS System Re-implementation for the IWSLT 2026 Instruction-Following Task

NAVER LABS 系统在 IWSLT 2026 指令跟随任务中的重新实现

Kamble, Anand, Tathe, Aniket

Abstract

We re-implement the NAVER LABS IWSLT 2025 instruction-following pipeline for the IWSLT 2026 Shared Task (constrained condition, short audio track), adapting it to the mandated components: SeamlessM4T-v2-large as the speech encoder and Qwen3-4B-Instruct as the LLM backbone. The three-stage approach projector alignment, text-only LoRA pre-training, and multimodal merging is preserved from the original design. We additionally construct 100k synthetic instruction-following examples across ten speech-centric task types (10k per task) from the provided corpora, suitable for further Stage 3 fine-tuning. Our primary model achieves COMET 0.781 on EN-ZH speech translation and BERTScore-F1 0.346 on English SQA on the MCIF benchmark.

Chinese Translation

我们重新实现了 NAVER LABS 的 IWSLT 2025 指令跟随管道，以适应 IWSLT 2026 共享任务（受限条件，短音频轨道）的要求，采用了指定的组件：SeamlessM4T-v2-large 作为语音编码器，以及 Qwen3-4B-Instruct 作为 LLM 主干。我们保留了原设计中的三阶段方法，包括投影对齐、仅文本的 LoRA 预训练和多模态合并。此外，我们从提供的语料库中构建了 10种以语音为中心的任务类型的 100,000 个合成指令跟随示例（每个任务 10,000 个），适用于后续的第三阶段微调。我们的主要模型在 EN-ZH 语音翻译上达到了 COMET 0.781，在 MCIF 基准的英语 SQA 上达到了 BERTScore-F1 0.346。

View on arXiv Download PDF AI Translation

cs.CL / 11 / 2607.05626

Population-Level Profiling of DSM-5 Depressive Symptoms Among Self-Reported ADHD and ASD Users on Twitter: An Exploratory Study Using Advanced NLP and Statistical Analysis

Twitter上自报告的注意缺陷多动障碍（ADHD）和自闭症谱系障碍（ASD）用户中DSM-5抑郁症状的人群特征分析：基于先进自然语言处理和统计分析的探索性研究

Rizwan, Muhammad, Nabergoj, David, Demšar, Jure

Abstract

Background: Depression frequently co-occurs with ADHD and autism spectrum disorder (ASD), but population-level differences in symptom expression between these groups remain underexplored. Objective: We examined whether social media users with ADHD and ASD differ in how they express DSM-5 depressive symptoms in their tweets, and whether differences persist across varying levels of depressive-content filtering. Methods: We analysed 1,282,437 tweets from 792 users (622 ADHD; 170 ASD) with self-reported diagnoses on Twitter. Tweets were pre-filtered for depressive relevance using zero-shot NLI, then classified into nine DSM-5 symptoms using MentalRoBERTa fine-tuned on ReDSM5. Profiles were mean-centered per user. We applied L1-penalised logistic regression with cross-validation to distinguish ADHD from ASD users, complemented by Pearson correlations for symptom co-occurrence, and tested robustness across five filtering thresholds using bootstrapping. Results: MentalRoBERTa achieved macro-F1 of 0.901 on a held-out set, outperforming the original ReDSM5 benchmark. ADHD vs ASD classification yielded stable but modest performance (cross-validated ROC-AUC 0.645-0.653). Cognitive issues, sleep issues, appetite change, and fatigue leaned toward ADHD, while suicidal ideation and anhedonia leaned toward ASD. A largely shared symptom co-occurrence structure emerged between groups; no pair met our criterion for a robust disorder-specific difference. Conclusions: Population-level differences in depression-related language between ADHD and ASD social media users were consistently observed across thresholds, reflecting reproducibility rather than clinical validity. Findings are exploratory and do not establish differing phenomenology at the individual level.

Chinese Translation

背景：抑郁症与ADHD和自闭症谱系障碍（ASD）经常共病，但这些群体在症状表达上的人群层面差异仍然未被深入探讨。目的：我们研究了自报告ADHD和ASD的社交媒体用户在推文中表达DSM-5抑郁症状的方式是否存在差异，以及这种差异在不同的抑郁内容过滤水平下是否持续。方法：我们分析了来自792个用户（622个ADHD；170个ASD）在Twitter上自我报告诊断的1,282,437条推文。使用零样本自然语言推理（zero-shot NLI）预先过滤了抑郁相关内容，然后利用针对ReDSM5微调的MentalRoBERTa将推文分类为九种DSM-5症状。每个用户的特征被均值中心化。我们应用L1惩罚的逻辑回归模型并使用交叉验证区分ADHD与ASD用户，辅以皮尔逊相关分析以检验症状共现，并通过自助法测试在五个过滤阈值下的稳健性。结果：MentalRoBERTa在持出集上实现了0.901的宏F1分数，优于原始ReDSM5基准。ADHD与ASD分类的表现稳定但适中（交叉验证ROC-AUC为0.645-0.653）。认知问题、睡眠问题、食欲变化和疲劳更倾向于ADHD，而自杀意念和快感缺失则更倾向于ASD。两个群体间的主要症状共现结构表现出高度共享；没有一对症状满足我们对稳健的特定障碍差异的标准。结论：在不同阈值下，一致观察到ADHD和ASD社交媒体用户在抑郁相关语言方面的人群层面差异，反映了可重复性而非临床效度。研究结果是探索性的，并未确立个体层面的不同现象。

View on arXiv Download PDF AI Translation

cs.CL / 12 / 2607.05644

Do It Right! A Methodology for Successful NLP System Development

做好它！成功的自然语言处理系统开发方法论

Patterson, Olga V., South, Brett, Workman, T. Elizabeth, DuVall, Scott L

Abstract

Natural language processing (NLP) is a common method for supplying data to clinical research and decision making by extracting information from electronic medical records. Numerous textbooks and tutorials describe specific algorithms and applications for text processing, yet algorithmic knowledge is only one ingredient of a successful NLP project. Drawing on the available literature, this paper presents a stepwise approach that applies the Systems Development Life Cycle (SDLC) to projects that rely on data extraction through language processing.

Chinese Translation

自然语言处理（NLP）是一种常用的方法，通过从电子病历中提取信息，为临床研究和决策提供数据。许多教科书和教程描述了特定的算法和文本处理应用，然而，算法知识只是成功的NLP项目的一个组成部分。基于现有文献，本文提出了一种逐步的方法，将系统开发生命周期（SDLC）应用于依赖于通过语言处理进行数据提取的项目。

View on arXiv Download PDF AI Translation

cs.CL / 13 / 2607.05679

RPAM: A Principled Metric for Evaluating Associations in Language Models with High Predictive Validity in Downstream Outputs

RPAM：一种有原则的度量，用于评估语言模型中具有高预测有效性的关联

Hodel, Damian, West, Jevin, Caliskan, Aylin

Abstract

Language models (LMs) exhibit problematic biases, such as stereotypes. Effectively analyzing and mitigating such biases requires accurate and generalizable evaluation methods of the underlying associations. Some existing approaches focus on downstream metrics that analyze associations in generated text. Since generated text content can vary drastically across LMs, such metrics often require specialized evaluation datasets, which limits the generalization of such downstream metrics. In contrast, upstream metrics examine LMs at the fundamental level of embeddings or continuation probabilities, enabling principled association analyses across LMs. Yet, to date, no upstream metric for generative LMs has uncovered a strong relationship with real-world associations, including those measured in generated text. To address this gap, we introduce the Relative Probability Association Metric (RPAM), an association evaluation metric for generative LMs. For three LMs of different quality of language generation and purpose (Mistral-7B-Instruct, Mistral-7B, and GPT-2) and well-studied evaluation datasets (WEAT-WS, Bellezza, WS-353, and SST2), we find a strong relationship between upstream RPAM measurements and corresponding implicit and explicit associations observed in humans, as well as biases measured downstream with LM-specific tasks, outperforming prior record values where applicable.

Chinese Translation

语言模型（LMs）表现出存在问题的偏见，例如刻板印象。有效分析和缓解这些偏见需要准确且具有普遍性的评估方法来评估潜在的关联。一些现有方法集中于下游度量，分析生成文本中的关联。由于生成文本的内容在不同的语言模型之间可能存在巨大差异，因此这些度量通常需要特定的评估数据集，这限制了下游度量的普遍性。相对而言，上游度量在嵌入或延续概率的基本层面上检查语言模型，从而使跨语言模型的原则性关联分析成为可能。然而，至今尚无针对生成性语言模型的上游度量能够揭示与现实世界关联之间的强关系，包括那些在生成文本中测量的关联。为了解决这一空白，我们引入了相对概率关联度量（Relative Probability Association Metric，RPAM），这是一个针对生成性语言模型的关联评估度量。对于三种不同语言生成质量和目的的语言模型（Mistral-7B-Instruct、Mistral-7B和GPT-2）以及经过良好研究的评估数据集（WEAT-WS、Bellezza、WS-353和SST2），我们发现上游RPAM测量与人类观察到的隐式和显式关联之间存在强关系，以及与特定语言模型任务下游测量的偏见，并在适用的情况下超越了之前的记录值。

View on arXiv Download PDF AI Translation

cs.CL / 14 / 2607.05689

UCSC NLP at SemEval-2026 Task 10: Boundary-Aware Span Extraction and RoBERTa Classification for Conspiracy Detection

UCSC NLP在SemEval-2026任务10中的表现：基于边界的跨度提取与RoBERTa分类用于阴谋检测

Marhoefer, Dom, Suvakovic, Milos, Grant-Richards, Glenn, Pinero, Aidan, King, Ryan

Abstract

We present our systems for SemEval-2026 Task 10 (PsyCoMark), addressing conspiracy marker extraction (Subtask 1) and document-level conspiracy detection (Subtask 2). For marker extraction, we formulate the task as multi-label span classification over enumerated candidate spans, using IoU >= 0.95 positive labeling, hard-negative sampling, and containment-based non-maximum suppression (NMS) with boundary-aware span representations. Document classification is modeled independently using a sequence classifier with label smoothing and a stratified train-validation split. Analysis shows that entity-like roles (Actor, Victim) are detected robustly, while abstract roles (Action, Effect, Evidence) remain sensitive to boundary criteria. On the official test set, our systems rank 7th in Subtask 1 (0.2251 macro F1) and 11th in Subtask 2 (0.7694 weighted F1).

Chinese Translation

我们展示了我们在SemEval-2026任务10（PsyCoMark）中的系统，针对阴谋标记提取（子任务1）和文档级阴谋检测（子任务2）进行研究。在标记提取中，我们将任务表述为针对枚举候选跨度的多标签跨度分类，采用IoU >= 0.95进行正例标记，硬负样本抽样，以及基于边界的包含性非最大抑制（NMS），并使用边界感知的跨度表示。文档分类独立建模，采用带标签平滑的序列分类器和分层的训练-验证分割。分析显示，类实体角色（演员、受害者）被稳健地检测到，而抽象角色（动作、效果、证据）对边界标准较为敏感。在官方测试集中，我们的系统在子任务1中排名第7（宏F1值为0.2251），在子任务2中排名第11（加权F1值为0.7694）。

View on arXiv Download PDF AI Translation

cs.CL / 15 / 2607.05691

Where to cut, how deep: BPE and Unigram-LM on chemistry SMILES

切割位置与深度：化学 SMILES 中的 BPE 和 Unigram-LM

Heidenreich, Hunter

Abstract

Every chemical language model reading SMILES begins with a tokenizer, yet the field has inherited byte-pair encoding (BPE) from natural language with little scrutiny. In natural language, BPE's principal alternative, Unigram-LM, is known to build structurally different vocabularies. Whether that contrast survives in chemistry was open. We report a controlled comparison of BPE and Unigram-LM over a fixed 165-token chemistry base, at the small vocabulary sizes where token embeddings are learnable, across three corpus typologies (diverse, drug-like, natural-products) and both pre-tokenization boundary policies. The two do not converge. In all 22 matched conditions they build near-disjoint subword vocabularies: cross-algorithm Jaccard overlap on the learned pieces never exceeds 0.161, and at most 0.05 once weighted toward the high-frequency pieces a model updates most. Unigram-LM also segments held-out molecules into 29-41% more tokens; the arms largely agree on where to cut but not how deeply, so BPE's segmentation is a strict coarsening of Unigram-LM's on 80-99% of molecules. The separation holds across corpus, boundary, and vocabulary size, persisting even at eight times that scale. The subword algorithm is therefore a modeling decision, not a free default. The study trains no language models.

Chinese Translation

每个读取 SMILES 的化学语言模型都始于一个分词器，但该领域从自然语言中继承的字节对编码 (BPE) 并未受到足够的审视。在自然语言中，BPE 的主要替代品 Unigram-LM 以结构上不同的词汇构建而闻名。这种差异在化学中是否依然存在尚不明确。我们报告了对 BPE 和 Unigram-LM 的控制比较，基于一个固定的 165 词元的化学基础，在能够学习词元嵌入的小词汇规模下，比较三种语料库类型（多样性、药物类、天然产物）和两种预分词边界策略。结果表明，两者并不收敛。在所有 22 种匹配条件下，它们构建了几乎不重合的子词词汇：跨算法的 Jaccard 重叠度在学习到的词素上从未超过 0.161，加权向高频词素时，最多也只有 0.05。Unigram-LM 还将保留的分子分段成多 29-41% 的词元；两个算法在切割位置上大致一致，但在切割深度上存在分歧，因此 BPE 的分段严格来说是对 Unigram-LM 在 80-99% 分子中的粗化。该分离现象在语料库、边界和词汇规模中均保持一致，在八倍的规模下依然存在。因此，子词算法是一个建模决策，而非默认选择。该研究未训练任何语言模型。

View on arXiv Download PDF AI Translation

cs.CL / 16 / 2607.05721

SpanUQ: Span-Level Uncertainty Quantification for Large Language Model Generation

SpanUQ：大型语言模型生成的跨度级不确定性量化

Zhang, Yimeng, Zhuang, Yingying, Wang, Ziyi, Lu, Yuxuan, Chen, Pei, Gupta, Aman, Su, Zhe, Tan, Ming, Zhang, Zhilin, Liu, Qun, Ramanathan, Manikandarajan, Maragoud, Rajashekar, Vul, Edward, Huang, Jing, Wang, Dakuo

Abstract

Uncertainty estimation is essential not only for the trustworthy deployment of large language models (LLMs) but also as a foundation for self-refinement in LLM generation. However, existing approaches operate at suboptimal granularities: token-level scores lack semantic coherence, while sequence-level scores fail to localize errors. We formalize Span-Level Uncertainty Estimation (SLUE), a new task that targets the natural granularity for uncertainty: semantically coherent text spans, each conveying a single assessable unit of meaning. To address this task, we introduce SPANUQ, a lightweight probe that distills the uncertainty knowledge from expensive multi-sample inference into a single forward pass over LLM hidden states. SPANUQ employs a DETR-style span decoder to simultaneously detect spans and estimate their uncertainty via a Mixture of Beta distribution, trained with a principled combination of Beta NLL regression and contrastive ranking objectives. We construct SPANUQ-BENCH, the first span-level uncertainty benchmark comprising 20K prompts, 293K annotated spans, and continuous soft labels derived from multi-sample claim verification. Experiments on five LLM backbones show that SPANUQ consistently achieves the best span-level uncertainty quality, outperforming the strongest probe baseline and all sampling-based methods while being 10-20x faster. Its DETR-based span detector attains 0.910 F1, surpassing the best heuristic by 39.4%, enabling precise error localization that sequence-level methods cannot provide. The framework generalizes across five LLMs spanning two model families.

Chinese Translation

不确定性估计不仅对大型语言模型（LLMs）的可信部署至关重要，而且是LLM生成中自我优化的基础。然而，现有的方法在粒度上都不够理想：基于标记的评分缺乏语义一致性，而基于序列的评分无法定位错误。我们正式化了跨度级不确定性估计（SLUE），这是一项针对不确定性的自然粒度的新任务：语义一致的文本跨度，每个跨度传达一个可评估的意义单元。为了解决这一任务，我们引入了SPANUQ，这是一种轻量级探测器，能够将来自昂贵的多样本推理的不确定性知识提取为单次前向推理对LLM隐状态的过程。SPANUQ采用了DETR风格的跨度解码器，同时检测跨度并通过混合贝塔分布估计其不确定性，该解码器通过贝塔负对数似然回归和对比排名目标的原则性组合进行训练。我们构建了SPANUQ-BENCH，这是第一个跨度级不确定性基准，包含2万个提示、29万个注释跨度，以及来自多样本声明验证的连续软标签。对五个LLM的实验表明，SPANUQ在跨度级不确定性质量上持续表现最佳，超越了最强的探测基线和所有基于采样的方法，同时速度是10-20倍。其基于DETR的跨度探测器获得了0.910的F1分数，超越了最佳启发式方法39.4%，实现了序列级方法无法提供的精确错误定位。该框架跨越两个模型系列的五个LLM具有良好的推广性。

View on arXiv Download PDF AI Translation

cs.CL / 17 / 2607.05722

Nemotron-Labs-Diffusion: A Tri-Mode Language Model Unifying Autoregressive, Diffusion, and Self-Speculation Decoding

Nemotron-Labs-Diffusion：统一自回归、扩散和自我推测解码的三模式语言模型

Fu, Yonggan, Whalen, Lexington, Garg, Abhinav, Wu, Chengyue, Khadkevich, Maksim, Oswald, Nicolai, Xie, Enze, Egert, Daniel, Sreenivas, Sharath Turuvekere, Diao, Shizhe, Yu, Chenhan, Yu, Ye, Chen, Weijia, Norouzi, Sajad, Liu, Jingyu, Lan, Shiyi, Zhu, Ligeng, Wang, Jin, Jiang, Jindong, Mardani, Morteza, Maghoumi, Mehran, Han, Song, Jukić, Ante, Tajbakhsh, Nima, Kautz, Jan, Molchanov, Pavlo

Abstract

We introduce Nemotron-Labs-Diffusion, a tri-mode language model (LM) that unifies AR, diffusion, and self-speculation decoding within a single architecture. Trained with a joint AR-diffusion objective, Nemotron-Labs-Diffusion can switch modes to sustain high throughput across deployment settings and concurrency levels. Our study shows that (1) AR and diffusion objectives are complementary: diffusion improves lookahead planning, while AR provides left-to-right linguistic priors. (2) In self-speculation mode, diffusion drafts while AR verifies, outperforming multi-token prediction (MTP) methods in both acceptance rate and real-device efficiency. (3) A speed-of-light analysis further demonstrates diffusion's long-term potential, with up to 76.5% more tokens per forward pass than self-speculation under an optimal sampler. Scaling to 3B, 8B, and 14B parameters, our Nemotron-Labs-Diffusion family, including base, instruct, and vision-language models, consistently outperforms state-of-the-art open-source AR and diffusion LMs in both accuracy and speed. For example, Nemotron-Labs-Diffusion-8B decodes 6x more tokens per forward than Qwen3-8B with comparable accuracy, translating to 4x higher throughput on SPEED-Bench with SGLang on a GB200 GPU.

Chinese Translation

我们介绍了Nemotron-Labs-Diffusion，这是一种三模式语言模型（LM），在单一架构中统一了自回归（AR）、扩散和自我推测解码。Nemotron-Labs-Diffusion通过联合AR-扩散目标进行训练，能够切换模式以在不同部署环境和并发水平下维持高吞吐量。我们的研究表明：（1）AR和扩散目标是互补的：扩散改善了前视规划，而AR提供了从左到右的语言先验。（2）在自我推测模式中，扩散进行草拟，而AR进行验证，相比多令牌预测（MTP）方法在接受率和真实设备效率上表现更优。（3）光速分析进一步展示了扩散的长期潜力，在最佳采样器下，每次前向传播的令牌数量比自我推测多出高达76.5%。在参数规模达到3B、8B和14B时，我们的Nemotron-Labs-Diffusion系列，包括基础模型、指令模型和视觉语言模型，在准确性和速度上持续超越了最先进的开源AR和扩散语言模型。例如，Nemotron-Labs-Diffusion-8B在前向解码中解码的令牌数量是Qwen3-8B的6倍，且准确性相当，在GB200 GPU上使用SGLang在SPEED-Bench上实现了4倍更高的吞吐量。

View on arXiv Download PDF AI Translation

cs.CL / 18 / 2607.05752

When Should LLMs Search? Counterfactual Supervision for Search Routing

大型语言模型何时应进行搜索？针对搜索路由的反事实监督

Kim, Minho

Abstract

Search-augmented language models can use external evidence to compensate for limitations in parametric knowledge, but search is not uniformly beneficial: models may call search for questions they can already answer, or rely on noisy evidence when correction, clarification, or abstention would be more appropriate. We formulate this as an instance-level search-routing problem: deciding whether search is needed to improve task success relative to a no-search execution. To derive supervision, we compare no-search and forced-search outcomes for the same question and construct an oracle over NO SEARCH, SEARCH, and UNSOLVED based on task-specific success. Using this oracle as both an evaluation criterion and a learning signal, we train search-routing policies with supervised fine-tuning and preference optimization, improving routing macro-F1 on oracle-eligible examples from 0.7082 to 0.8235 for Gemma E2B and from 0.7053 to 0.8365 for Qwen3.5-4B. Further analysis shows that the learned policies reduce model-specific routing failures: Gemma primarily learns no-search restraint, while Qwen further reduces missed search; residual UNSOLVED cases reveal heterogeneous bottlenecks involving model capacity, retrieval budget, evidence use, and policy behavior.

Chinese Translation

增强搜索的语言模型可以利用外部证据来弥补参数知识的局限，但搜索并不总是有益：模型可能会对已能解答的问题进行搜索，或在更合适的情况下依赖嘈杂的证据，例如更正、澄清或放弃。我们将此问题表述为一个实例级的搜索路由问题：决定是否需要搜索以提高任务成功率，相对于没有搜索的执行。为了推导监督，我们比较了同一问题的无搜索和强制搜索结果，并根据特定任务的成功率构建了一个有关无搜索（NO SEARCH）、搜索（SEARCH）和未解决（UNSOLVED）的预言者（oracle）。使用该预言者作为评估标准和学习信号，我们通过监督微调和偏好优化训练了搜索路由策略，使得Gemma E2B在符合预言者条件的例子上的路由宏观F1从0.7082提升至0.8235，而Qwen3.5-4B则从0.7053提升至0.8365。进一步分析表明，学习到的策略减少了特定模型的路由失败：Gemma主要学会了无搜索的克制，而Qwen则进一步减少了漏掉的搜索；剩余的未解决案例揭示了涉及模型容量、检索预算、证据使用和策略行为的异质瓶颈。

View on arXiv Download PDF AI Translation

cs.CL / 19 / 2607.05764

Inject or Navigate? Token-Efficient Retrieval for LLM Analysis of Transactional Legal Documents

注入还是导航？面向大语言模型（LLM）分析交易法律文件的高效检索

Hany, Mahmoud, ElSheraey, Mourad, Said, Mahmoud, Naoum, Peter

Abstract

Answering questions over a set of transactional legal documents is most simply done by injecting the whole corpus into the LLM's context window on every query. That baseline maximises retrieval recall, but its token footprint scales with the corpus rather than the question, and long-context degradation scales with it. We report what it took to replace full-corpus injection in a legal-document analysis system, comparing it against two structured retrieval modes over our proprietary structure-aware chunking: embedding retrieval (NAVEMBED) and LLM navigation over a compact structured index (NAVINDEX). On a 20-question benchmark with verified ground-truth answers, a position-bias-controlled, reference-anchored pairwise judge scored semantic retrieval with reranking tied with injection on 16 of 18 document-bound questions (injection preferred on 2) while attending to 17.3x fewer input tokens (a general-text-embedding (GTE) configuration reaches 29.9x at a lower tie rate); both modes were judged tied on the 2 out-of-scope controls. NAVINDEX was judged tied on all 18 at a 1.61x smaller total token footprint, a ~56x smaller answering context, and 25% lower dollar cost. We derive a closed-form caching-crossover rule: cached injection is cheaper in dollars only while the corpus stays below roughly ten times the retrieval payload. Scope and uncertainty are quantified in Section 8.

Chinese Translation

在一组交易法律文件中回答问题最简单的方法是在每个查询时将整个语料库注入到LLM的上下文窗口中。这个基线最大化了检索的召回率，但其令牌占用量与语料库的大小成比例，而不是与问题相关，并且长上下文衰减也随之增长。我们报告了在法律文件分析系统中替代全文注入所需的工作，并将其与我们专有的结构感知分块所支持的两种结构化检索模式进行了比较：嵌入检索（NAVEMBED）和通过紧凑结构索引进行的LLM导航（NAVINDEX）。在一个包含20个问题的基准测试中，所有问题都有经过验证的真实答案，一位受到位置偏差控制、参考锚定的成对评审员对语义检索进行了评估，结果显示在18个与文档相关的问题中，有16个的问题语义检索与注入相当（在2个问题中偏好注入），而输入令牌数量减少了17.3倍（通用文本嵌入（GTE）配置在较低的平局率下达到29.9倍）；在2个超范围控制问题上，两个模式被评定为平局。NAVINDEX在所有18个问题上被评定为平局，且总令牌占用量减少了1.61倍，回答的上下文小约56倍，成本降低了25%。我们推导出一个闭式缓存交叉规则：在语料库低于检索负载的十倍时，缓存注入在成本上更便宜。范围和不确定性在第8节中进行了量化。

View on arXiv Download PDF AI Translation

cs.CL / 20 / 2607.05849

CoPiT: Cognitive Pivot Translation for Digraphic Low-Resource Mongolian in the Traditional Script

CoPiT：针对传统书写体低资源蒙古语的认知支点翻译

Bayarsaikhan, Burte, Kim, Serynn, Chang, Buru

Abstract

Low-resource languages remain challenging for machine translation, and Mongolian is a representative case. As a digraphic language, Mongolian is written in both Cyrillic and Traditional scripts, which exhibit a severe imbalance in data availability. While the Cyrillic script is relatively well-resourced, the Traditional script remains extremely data-scarce and orthographically ambiguous, leading to substantial performance degradation in direct translation. We propose CoPiT, a cognitively motivated pivot-based translation pipeline that exploits this internal resource hierarchy by routing translation through the Cyrillic script. The pipeline explicitly resolves script-induced ambiguity in the Traditional script before translation, enabling more stable and accurate meaning transfer. Across multiple backbone models and target languages, CoPiT consistently outperforms direct translation, achieving substantial absolute BLEU improvements together with consistent 1.5-1.6x COMET gains. These gains allow strong open-source models to match or outperform GPT-4.1 under comparable evaluation settings. Beyond inference-time improvements, CoPiT enables the construction of synthetic parallel data directly from Traditional-script text, mitigating data scarcity in realistic low-resource scenarios. We release a new multi-script parallel dataset covering Mongolian in both scripts alongside English, Korean, and Russian. All datasets and code are publicly available at https://anonymous.4open.science/r/anonymous_project-76C7.

Chinese Translation

低资源语言在机器翻译中仍然面临挑战，而蒙古语就是一个典型案例。作为一种双书写系统的语言，蒙古语同时使用西里尔字母和传统书写体，二者在数据可用性上存在严重不平衡。虽然西里尔字母相对资源丰富，但传统书写体的数据极为稀缺且在正字法上存在歧义，导致直接翻译性能显著下降。我们提出了CoPiT，一种基于认知动机的支点翻译管道，通过西里尔字母进行翻译，从而利用这一内部资源层次结构。该管道在翻译之前显式解决传统书写体中的字形引起的歧义，从而实现更稳定和准确的意义传递。在多个基础模型和目标语言中，CoPiT始终优于直接翻译，取得了显著的绝对BLEU提升，并且在COMET上实现了1.5-1.6倍的持续增益。这些提升使得强大的开源模型在可比评估设置下能够与GPT-4.1相匹配或超越。除了推理时的改进，CoPiT还能够直接从传统书写体文本构建合成平行数据，缓解现实低资源场景中的数据稀缺问题。我们发布了一个新的多书写体平行数据集，涵盖了西里尔字母和传统书写体的蒙古语，以及英语、韩语和俄语。所有数据集和代码均可在 https://anonymous.4open.science/r/anonymous_project-76C7 上公开获取。

View on arXiv Download PDF AI Translation

cs.CL / 21 / 2607.05861

Mitigating Factual Hallucination in Large Reasoning Models via Mixed-Mode Advantage Regularization

通过混合模式优势正则化缓解大型推理模型中的事实幻觉

Wang, Kaishen, Zheng, Tong, Cui, Xuehao, Chen, Ruibo, Xiong, Tianyi, Huang, Heng

Abstract

Large reasoning models (LRMs) improve language model capabilities by generating explicit thinking traces before final answers. In factuality-oriented question answering (QA), such thinking often improves overall performance by helping the model recover relevant knowledge and refine its answers. However, we find that this benefit is not uniform at the instance level: explicit thinking can also overturn correct non-thinking answers and lead to factual drift. We refer to this failure mode as \emph{thinking-induced hallucination}. To explain this phenomenon, we formulate explicit thinking in factuality QA as a thinking residual over the model's direct-answer tendency, which can either recover missing knowledge or introduce unsupported associations. Based on this formulation, we propose MARGO, \underline{\textit{M}}ixed-Mode \underline{\textit{A}}dvantage \underline{\textit{R}}egularization for \underline{\textit{G}}rounded \underline{\textit{O}}ptimization, a reinforcement learning framework that uses non-thinking rollouts as same-model references in advantage estimation. By constructing mixed-mode rollout groups with both thinking and non-thinking trajectories, MARGO evaluates whether explicit thinking adds factual value beyond direct answering, thereby suppressing hallucination-prone thinking while preserving beneficial thinking behaviors. Experiments across multiple factuality-oriented QA benchmarks demonstrate that MARGO improves factual reliability over strong baselines, while evaluations on mathematical benchmarks show that it preserves general reasoning ability.

Chinese Translation

大型推理模型（LRMs）通过在最终答案之前生成明确的思维轨迹来提升语言模型的能力。在以事实为导向的问题回答（QA）中，这种思维往往通过帮助模型恢复相关知识和细化答案来提高整体表现。然而，我们发现这种好处在实例层面并不均匀：明确的思维也可能推翻正确的非思维答案，并导致事实漂移。我们将这种失败模式称为 extit{思维引发的幻觉}。为了解释这一现象，我们将以事实为导向的QA中的明确思维表述为模型直接回答倾向的思维残差，这可以恢复缺失的知识或引入不支持的关联。基于这种表述，我们提出了MARGO， extit{M}ixed- extit{M}ode extit{A}dvantage extit{R}egularization for extit{G}rounded extit{O}ptimization，一种强化学习框架，利用非思维的回放作为同一模型在优势估计中的参考。通过构建包含思维和非思维轨迹的混合模式回放组，MARGO评估明确的思维是否在直接回答之外增加了事实价值，从而抑制易产生幻觉的思维，同时保留有益的思维行为。在多个以事实为导向的QA基准上的实验表明，MARGO在比强大基线更高的事实可靠性方面表现更好，而在数学基准上的评估则表明，它保持了一般推理能力。

View on arXiv Download PDF AI Translation

cs.CL / 22 / 2607.05937

Is Domain Adaptation Always Helpful? A Frozen-Backbone Study of Cross-Domain Sentiment Transfer

领域适应总是有帮助吗？基于冻结骨干网络的跨领域情感转移研究

Tran, Phat, Lahni, Artin, Kulkarni, Pranav, Zhang, Yaolun

Abstract

Sentiment analysis with frozen pre-trained language model (PLM) backbones has become a common paradigm, yet the practical benefit of explicit domain adaptation remains unclear, particularly when backbones encode varying degrees of target-domain knowledge. We present a preliminary case study evaluating a controlled family of frozen embedding backbones (Qwen3-Embedding 0.6B, 4B, 8B), alongside RoBERTa-base and FinBERT. We train a lightweight MLP adapter on consumer reviews using Domain-Adversarial Neural Networks (DANN), Maximum Mean Discrepancy (MMD), and Supervised Contrastive Learning (SCL), and evaluate transfer to movie reviews (SST-2) and a heavily restricted subset of financial news (Financial PhraseBank). Within this constrained sample, we observe two distinct transfer patterns. On SST-2, domain adaptation provides negligible gain regardless of scale. On the financial subset, explicit domain adaptation appears to recover substantial performance for small general-purpose backbones. Notably, we find that adversarial alignment (DANN) is associated with degraded performance for domain-specialized backbones like FinBERT, consistent with erosion of pre-existing domain-specific structure, whereas supervised contrastive loss appears to preserve it. These preliminary findings suggest that the efficacy of explicit domain adaptation is highly contingent on whether the frozen backbone already possesses target-domain coverage.

Chinese Translation

使用冻结的预训练语言模型（PLM）骨干进行情感分析已成为一种常见范式，但显式领域适应的实际益处仍不明确，尤其是在骨干编码了不同程度的目标领域知识时。我们提出了一项初步案例研究，评估一系列受控的冻结嵌入骨干（Qwen3-Embedding 0.6B、4B、8B），以及RoBERTa-base和FinBERT。我们使用领域对抗神经网络（DANN）、最大均值差异（MMD）和监督对比学习（SCL）在消费者评论上训练一个轻量级的多层感知器适配器，并评估其对电影评论（SST-2）和一个高度限制的金融新闻子集（Financial PhraseBank）的转移。在这个受限样本中，我们观察到两种不同的转移模式。在SST-2上，领域适应无论规模大小都几乎没有收益。在金融子集上，显式领域适应似乎为小型通用骨干恢复了可观的性能。值得注意的是，我们发现对抗对齐（DANN）与像FinBERT这样的领域专用骨干的性能下降相关，这与现有领域特定结构的侵蚀一致，而监督对比损失似乎能够保持这种结构。这些初步发现表明，显式领域适应的有效性在很大程度上取决于冻结骨干是否已经具备目标领域的覆盖。

View on arXiv Download PDF AI Translation

cs.CL / 23 / 2607.05964

Umm... With Transformers? Insights from Filled Pause Use across Four Slavic Parliaments

嗯...使用变压器？从四个斯拉夫议会中填充停顿使用的洞见

Porupski, Ivan, Dropuljić, Branimir, Ljubešić, Nikola

Abstract

Filled pauses (FPs) are a universal feature of spontaneous speech, yet most studies rely on small, single-language corpora, limiting the generalisability of their findings. We analyse ~4,000 hours of parliamentary speech across four related Slavic languages (Croatian, Czech, Polish, Serbian). FP occurrence is obtained via transformer-based automatic detection, while FP rate is modelled using Generalised Estimating Equations (GEE) with Mundlak correction to distinguish within- from between- speaker effects. We replicate a negative association of age and speech rate with FP rate, but find that gender effects are language-specific and directionally opposite to most prior literature. Novel analyses of sentiment, political orientation, and power status reveal a consistent positive association between sentiment and FP rate, alongside parliament-specific modulation by orientation and power status, with opposition speakers tending toward lower FP rates than governing coalition speakers.

Chinese Translation

填充停顿（FPs）是自发语言中普遍存在的特征，但大多数研究依赖于小规模的单语语料库，限制了研究结果的普遍性。我们分析了四种相关斯拉夫语言（克罗地亚语、捷克语、波兰语和塞尔维亚语）中约4000小时的议会发言。FP的出现通过基于变压器（transformer）的自动检测获得，而FP的比率使用广义估计方程（Generalised Estimating Equations, GEE）建模，并结合Mundlak修正来区分说话者之间和说话者内部的效应。我们重复了年龄和语速与FP比率之间的负相关关系，但发现性别效应是语言特定的，并且与大多数先前文献的方向相反。对情感、政治取向和权力地位的新颖分析揭示了情感与FP比率之间的一致正相关关系，并且根据取向和权力地位在不同议会中存在特定的调节效应，反对派发言者的FP比率趋向低于执政联盟的发言者。

View on arXiv Download PDF AI Translation

cs.CL / 24 / 2607.05968

InfluMatch: Frontier-Quality KOL Search at 4B-Model Cost

InfluMatch：在4B模型成本下的前沿质量KOL搜索

Kaewtawee, Krittanon, Pornpichitsuwan, Petmongkon, Temyingyong, Natchaya, Laplamoon, Nutnicha, Modecrua, Wachiravit, Pachtrachai, Krittin, Kraisingkorn, Touchapon

Abstract

Matching influencers (KOLs) to free-form, multi-part Thai marketing criteria is today served either by keyword search over structured profiles, which misses semantic fit, or by prompting frontier LLMs over every candidate, which is accurate but slow and expensive. We present InfluMatch, a low-cost three-stage cascade -- retrieval $\rightarrow$ rerank $\rightarrow$ reason -- built entirely from small open-weight models: dense retrieval returns 50 candidates, a 4B pointwise reranker scores each by the log-probability of a single Yes token and keeps 10, and a 4B reasoner grades the shortlist per criterion on a rubric with a Thai rationale. The cascade is designed for cost: reasoning over a filtered top-10 halves token spend versus reasoning over all 50 while scoring 14 points higher. End-to-end against human relevance labels on an 11-query set with all 50 candidates labeled, the full cascade reaches 94.1% P@5, versus a retrieval-only baseline near random; it matches the frontier model Kimi-K2.6 (91.8%) while emitting ${\sim}35\times$ fewer output tokens and serving a 50-KOL query in ${\sim}20$ s on one A100. Notably, the only fine-tuning that pays off is pairwise: a SimPO-tuned reranker matches the frontier baseline's best-pick accuracy (78.0 EM), whereas fine-tuning the reasoner on pointwise per-criterion labels improves offline scores yet degrades end-to-end ranking -- an inversion we trace to the design of the absolute labeling task -- leaving the untuned base model as the strongest deployed reasoner. The result is a deployable, explainable KOL search system at a small fraction of frontier serving cost.

Chinese Translation

将影响者（KOL）与自由形式的多部分泰国营销标准进行匹配，现有的方法要么通过关键字搜索结构化档案，导致语义匹配不足，要么通过对每个候选者调用前沿的大型语言模型（LLM），虽然结果准确但速度慢且成本高。我们提出了InfluMatch，一种低成本的三阶段级联方案——检索 $ ightarrow$ 重新排名 $ ightarrow$ 理由，完全由小型开放权重模型构建：稠密检索返回50个候选者，一个4B的逐点重新排序模型通过单个“Yes”标记的对数概率对每个候选者打分并保留10个，最后一个4B的推理模型根据带有泰国理由的标准对入围名单进行评分。该级联设计的成本效益显著：与对所有50个候选者推理相比，对筛选出的前10名进行推理可以将令牌消耗减半，同时得分高出14个百分点。在一个包含11个查询的全链条对比实验中，所有50个候选者均已标注，完整级联达到了94.1%的P@5，相关性基线接近随机；其准确度与前沿模型Kimi-K2.6（91.8%）相匹配，同时产生的输出令牌数量减少了约35倍，并在一个A100上以约20秒的时间响应50-KOL查询。值得注意的是，唯一有效的微调是成对的：经过SimPO微调的重新排序模型达到了前沿基线最佳选择的准确率（78.0 EM），而对推理模型进行逐点的标准标签微调虽然提高了离线得分，但却降低了整体排序，这种反转现象我们归因于绝对标记任务的设计，使得未微调的基础模型成为最强的推理器。因此，结果是一个可部署的、可解释的KOL搜索系统，其成本仅为前沿服务成本的一小部分。

View on arXiv Download PDF AI Translation

cs.CL / 25 / 2607.05969

MemDefrag: Latent Memory Defragmentation for Large Language Models

MemDefrag：大语言模型的潜在内存碎片整理

Yan, Ruiyi, Mao, Zhuoyuan, Guo, Yiwen

Abstract

Latent memory, which stores past knowledge fragments as per-layer hidden states, has emerged as a promising paradigm (e.g., MemoryLLM and M+) for long-term memory in large language models (LLMs). However, the paradigm suffers from significant performance degradation during memory updates, due to positional encoding misalignment and the absence of any tracing mechanism to distinguish target memory fragments from irrelevant ones. To discover such a tracing mechanism, we probe the layer-wise attention density over stored memory fragments, and find that a small set of middle transformer layers consistently concentrates the highest density on the target fragment - exposing an inherent tracing signal. In light of this, we propose MemDefrag, a training-free and model-agnostic framework that (1) uses a middle-layer tracing signal to conduct memory defragmentation (rank, reorder, and filter memories), and (2) applies an informativeness-guided proportional forgetting mechanism once capacity is exceeded. Experiments show that MemDefrag substantially outperforms MemoryLLM and M+ on knowledge retention (e.g., 43.0% vs. 17.4%/17.6% after 50 memory updates) and long-context benchmarks, and generalizes well across various LLMs and latent-memory variants.

Chinese Translation

潜在内存作为逐层的隐藏状态存储过去知识片段，已成为大语言模型（LLMs）中一种有前景的长期记忆范式（例如 MemoryLLM 和 M+）。然而，该范式在内存更新过程中遭遇显著的性能下降，原因在于位置编码的不匹配以及缺乏任何机制来区分目标内存片段与无关片段。为了解发现这样的追踪机制，我们探查了存储内存片段的逐层注意力密度，并发现一小组中间变压器层始终在目标片段上集中最高的密度，从而揭示出一种固有的追踪信号。基于此，我们提出了 MemDefrag，一种无需训练且与模型无关的框架，(1) 利用中间层追踪信号进行内存碎片整理（排序、重新排列和过滤内存），(2) 一旦容量超出，应用基于信息引导的比例遗忘机制。实验结果显示，MemDefrag 在知识保留方面大幅优于 MemoryLLM 和 M+（例如，经过 50 次内存更新后为 43.0% 对比 17.4%/17.6%）以及在长上下文基准测试中，并能够在各种 LLM 和潜在内存变体之间良好泛化。

View on arXiv Download PDF AI Translation

cs.CL / 26 / 2607.05992

PluraMath: Extending Mathematical Reasoning Evaluation Beyond High-Resource Languages

PluraMath：扩展数学推理评估至非高资源语言

Dementieva, Daryna, Babakov, Nikolay, Hämmerl, Kathy, Alimova, Ilseyar, Libovický, Jindřich, Okabe, Shu, Baisbay, Miras, Edman, Lukas, Inomkhujaev, Abrorkhon, Karamolegkou, Antonia, Lango, Mateusz, Özer, Volkan, Selic, Nikola, Swain, Subhankar, Temesgen, Tsedeniya Kinfe, Weisberg, Galit Bary, Fraser, Alexander

Abstract

Mathematical reasoning has become a central task for evaluating and tuning reasoning Large Language Models (LLMs), yet existing benchmarks remain heavily biased toward high-resource languages, with English and Chinese dominating both pre-training corpora and evaluation suites. The recently released PolyMath (Wang et al., 2025) dataset represents a significant step forward, yet its coverage is still limited to 18 only high-resource languages. To address this gap, we introduce PluraMath, an extension of PolyMath to 18 additional {underrepresented languages spanning 6 language families -- ranging from mid-resource to extreme low-resource settings. We constructed the dataset through a human-curated pipeline, where native speakers thoroughly validated pre-computed translations. Using PluraMath, we then benchmark 27 reasoning LLMs across four model scales -- small, mid-size, large, and closed-source ensembles -- probing the multilingual mathematical reasoning capabilities of state-of-the-art models under diverse linguistic conditions. Our fine-grained analysis confirms a persistent gap in mathematical reasoning performance between high-resource and underrepresented languages, with stronger results largely associated with better instruction-following ability. We fully open-source our dataset, data acquisition pipeline, and evaluation framework, with the goal of lowering the barrier to multilingual benchmark development for underrepresented communities.

Chinese Translation

数学推理已成为评估和调优推理大型语言模型（LLMs）的核心任务，然而现有基准测试仍然严重偏向高资源语言，英语和中文在预训练语料和评估套件中占据主导地位。最近发布的PolyMath（Wang等，2025）数据集代表了一个重要的进展，但其覆盖范围仍限于18种高资源语言。为了解决这一差距，我们引入了PluraMath，这是对PolyMath的扩展，涵盖了18种额外的缺乏代表性的语言，跨越6个语言家族，覆盖从中等资源到极低资源的不同语言环境。我们通过一个人工策划的流程构建了数据集，母语者彻底验证了预先计算的翻译。然后，我们使用PluraMath对27种推理LLM进行基准测试，涵盖四种模型规模——小型、中型、大型及闭源集成模型——探测最先进模型在不同语言条件下的多语言数学推理能力。我们的细致分析证实高资源语言与缺乏代表性语言之间在数学推理表现上始终存在差距，较强的结果通常与更好的遵循指令能力相关。我们完全开放了我们的数据集、数据获取流程和评估框架，旨在降低为缺乏代表性社区开发多语言基准的门槛。

View on arXiv Download PDF AI Translation

cs.CL / 27 / 2607.06080

From Blueprint to Reality: Modeling and Applying Putnam's Social Capital Theory with LLM-based Multi-agent Simulations

从蓝图到现实：基于大型语言模型的多智能体仿真模型与普特南社会资本理论的应用

Ling, Shiyi, Zheng, Zhi, Zheng, Hui, Xue, Wenjun, Ye, Feng, Xu, Tong

Abstract

Putnam's Social Capital Theory is a foundational framework for collective action and community prosperity. However, traditional empirical methods face practical limits on control and replication. Meanwhile, LLM-based social simulations are typically behavior-driven and lack theory-aligned environments for modeling Putnam's core propositions. To address these gaps, we introduce SocaSim, an LLM-based multi-agent simulation framework to study Putnam's Social Capital Theory from theoretical blueprint to simulated reality. Specifically, we build an environment integrating social network evolution, trust dynamics, and norm propagation, where agents engage in repeated collective-action experiments, and then apply the three dimensions to analyze adaptation challenges in smart elderly care. Our simulations reproduce Putnam's macro-level patterns and exhibit strong human-agent alignment at the group level. Unlike traditional methods, SocaSim traces micro-level causal pathways of social network, trust, and norms via round-by-round simulations and counterfactual interventions, enabling process-level interpretability. Taken together, these capabilities establish a research paradigm that leverages LLM agents to bridge social science and computer science.

Chinese Translation

普特南的社会资本理论是集体行动和社区繁荣的基础框架。然而，传统的实证方法在控制和复制方面面临实际限制。同时，基于大型语言模型的社会仿真通常是行为驱动的，缺乏与理论一致的环境来建模普特南的核心命题。为了解决这些问题，我们介绍了SocaSim，一个基于大型语言模型的多智能体仿真框架，用于从理论蓝图研究普特南的社会资本理论到模拟现实。具体而言，我们构建了一个集成社会网络演化、信任动态和规范传播的环境，智能体在其中参与重复的集体行动实验，随后应用这三个维度分析智能老龄护理中的适应性挑战。我们的仿真再现了普特南的宏观模式，并在群体层面展示了强的人机对齐。不同于传统方法，SocaSim通过逐轮仿真和反事实干预，追踪社会网络、信任和规范的微观因果路径，实现了过程级的可解释性。综合来看，这些能力建立了一个研究范式，利用大型语言模型的智能体在社会科学与计算机科学之间架起桥梁。

View on arXiv Download PDF AI Translation

cs.CL / 28 / 2607.06127

Measuring the practice of shared-decision making (OPTION12): An Investigation into Open-sourced Smaller LLMs (OS-sLLMs) for Better Privacy and Sustainability

共享决策实践的测量（OPTION12）：针对更好的隐私和可持续性的小型开源语言模型（OS-sLLMs）研究

Wit, Tamara, Han, Lifeng, Heipon, Carly, Lindevelt, David, Stiggelbout, Anne, Verberne, Suzan

Abstract

We present LLM4SDM, the first study of open-source smaller language models (OS-sLLMs) for automated assessment of shared decision making (SDM) using the Observer OPTION12 framework. Unlike previous work that relies on large commercial models and the shorter OPTION5 instrument, our study focuses on privacy-preserving locally deployable models and Dutch melanoma consultation transcripts. Using expert-annotated clinical consultations, we evaluate three general-domain and two medical-domain OS-sLLMs during a development-phase pilot study. Results show that general-domain models outperform medical-domain models, which exhibit substantial hallucination and instruction-following failures. Gemma3:12b achieves the strongest agreement with human annotations (Pearson r=0.51, Spearman \r{ho}=0.59). Item-level and qualitative analyses reveal systematic challenges related to temporal discourse reasoning, conversational role attribution, and evidence grounding. We further introduce a Judge-LLM consensus framework designed to support disagreement resolution among multiple models. Our findings suggest that while current OS-sLLMs cannot replace human annotators, they offer a promising foundation for privacy-preserving human-in-the-loop SDM assessment.

Chinese Translation

我们提出了LLM4SDM，这是首个使用观察者OPTION12框架对开源小型语言模型（OS-sLLMs）进行自动评估共享决策（SDM）的研究。与以往依赖大规模商业模型和更短的OPTION5工具的研究不同，我们的研究重点是隐私保护的本地可部署模型和荷兰黑色素瘤咨询记录。通过专业人员标注的临床咨询，我们在开发阶段的初步研究中评估了三个通用领域和两个医学领域的OS-sLLMs。结果显示，通用领域模型的表现优于医学领域模型，后者表现出显著的幻觉和指令遵循失败。Gemma3:12b在与人类标注的一致性方面表现最佳（Pearson r=0.51, Spearman {ho}=0.59）。项目级和定性分析揭示了与时间话语推理、对话角色归属和证据基础相关的系统性挑战。我们进一步引入了Judge-LLM共识框架，旨在支持多个模型之间的分歧解决。我们的研究结果表明，虽然目前的OS-sLLMs无法替代人类标注者，但它们为隐私保护中的人机合作SDM评估提供了一个有前景的基础。

View on arXiv Download PDF AI Translation

cs.CL / 29 / 2607.06140

CurateEvo: Data-Curation Evolving for Agentic Post-Training

CurateEvo：为自主后训练演化的数据策展

Wang, Dingzirui, Zhang, Xuanliang, Xu, Keyan, Zhu, Qingfu, Che, Wanxiang

Abstract

Large language model (LLM) agents require post-training methods that can improve long-horizon decision making from environment feedback. However, existing agentic post-training pipelines often treat data curation as a fixed preprocessing step, focusing mainly on data augmentation while neglecting filtering, refinement, and adaptation to downstream failures. We propose CurateEvo, a failure-driven dynamic evolution framework for agentic post-training data curation. CurateEvo represents the curation strategy as executable code and iteratively rewrites it using failed trajectories from a held-out development set. At each epoch, the evolved strategy transforms a fixed raw corpus into supervised fine-tuning data, reinforcement learning data, and an inference-time memory bank. The evolution process first improves effectiveness by diagnosing recurring failure modes and augmenting, filtering, or refining data accordingly, and then improves efficiency by pruning redundant or low-utility training turns under a cost-aware objective. Experiments on ACEBench-Agent, BFCL-V4, and {\tau}^2-Bench under both labeled and wild-data settings show that CurateEvo consistently outperforms prior curation methods, improving average scores by 3.2 and 2.7 points, respectively. Further analyses demonstrate that CurateEvo is compatible with different post-training recipes and substantially reduces curation overhead.

Chinese Translation

大型语言模型（LLM）代理需要能够从环境反馈中改善长远决策的方法。然而，现有的自主后训练流程通常将数据策展视为固定的预处理步骤，主要关注数据增强，而忽视了过滤、精炼和适应下游失败。我们提出了CurateEvo，一种以失败为驱动的动态演化框架，用于自主后训练的数据策展。CurateEvo将策展策略表示为可执行代码，并使用从保留的开发集中的失败轨迹进行迭代重写。在每个训练周期中，演化后的策略将固定的原始语料库转化为监督微调数据、强化学习数据和推理时的记忆库。演化过程首先通过诊断重复的失败模式并相应地增强、过滤或精炼数据来提高有效性，然后通过在成本感知目标下修剪冗余或低效的训练回合来提高效率。在ACEBench-Agent、BFCL-V4和{ au}^2-Bench的标注和野生数据设置下的实验表明，CurateEvo始终优于先前的策展方法，平均分数分别提高了3.2和2.7分。进一步的分析表明，CurateEvo与不同的后训练方案兼容，并显著减少了策展的开销。

View on arXiv Download PDF AI Translation

cs.CL / 30 / 2607.06145

Prompting Complexity: Shortest Prompts for Texts and Behaviors in LLMs

提示复杂性：大语言模型中的最短提示文本与行为

Cosma, Adrian

Abstract

In this paper, we define the quantity of prompting complexity: for a fixed instruction-tuned language model, what is the shortest plausible prompt that makes deterministic decoding produce a target text? It is an LM-relative analogue of resource-bounded Kolmogorov complexity: the prompt is a program, the model interface is the interpreter, and information omitted from the prompt is supplied by the model's weights, training distribution, tokenizer, template, and decoding rule. Unlike classical Kolmogorov complexity, this measure is intentionally non-universal. In the finite-context setting it is computable by enumeration, but there is no model-independent invariance theorem; the same text may be cheap for one model and inaccessible or expensive for another. To keep the search space aligned with prompt engineering, we restrict programs to plausible human-readable texts rather than arbitrary token strings. We extend the exact definition to soft prompting complexity for approximate outputs, yielding a lossy notion of model-relative text compression and a formal target for prompt optimization. We also define prompting distance by comparing shortest generating prompts, and behavioral prompting complexity for reaching any output satisfying a specification. Based on these formulations, we define a research agenda for empirically studying which texts and behaviors are accessible from short plausible prompts under a fixed LM interface.

Chinese Translation

在本文中，我们定义了提示复杂性的量度：对于一个固定的指令调优语言模型，产生目标文本所需的最短合理提示是什么？它是资源约束的科尔莫哥洛夫复杂性的语言模型相关类比：提示是一个程序，模型接口是解释器，提示中省略的信息由模型的权重、训练分布、分词器、模板和解码规则提供。与经典的科尔莫哥洛夫复杂性不同，这一度量在设计上是非通用的。在有限上下文的情况下，它可以通过枚举计算，但没有模型独立的不可变定理；同一文本对一个模型而言可能是廉价的，而对另一个模型可能是不可及的或昂贵的。为了使搜索空间与提示工程保持一致，我们将程序限制为合理的人类可读文本，而不是任意的符号字符串。我们将确切定义扩展到软提示复杂性，以适应近似输出，从而产生一种相对模型的文本压缩的有损概念和提示优化的正式目标。我们还通过比较最短生成提示来定义提示距离，并针对达到满足特定要求的任何输出定义行为提示复杂性。基于这些公式，我们定义了一个研究议程，以经验研究在固定的语言模型接口下，哪些文本和行为是可以通过短的合理提示获得的。

View on arXiv Download PDF AI Translation

cs.CL / 31 / 2607.06157

LLM Agents for Deliberative Collaboration: A Study on Joint Decision Making Under Partial Observability

用于深思熟虑协作的LLM代理：在部分可观察条件下的联合决策研究

Wang, Chenxu, Yang, Yongkun, Du, Boyuan, Lin, Shiwei, Liu, Huaping

Abstract

Deliberation plays a crucial role in collaboration; when humans work together, they naturally engage in communication to align information and reach an agreement. In this paper, we investigate deliberative large language model (LLM) agents under partially observable joint decision-making tasks. We formalize deliberative collaboration as a cooperative joint decision problem with partial and asymmetric observations, and introduce a scalable benchmark that instantiates this problem across multiple task settings and domains in which agents must exchange information through deliberation to reach a joint decision with a shared reward. We then instantiate a reference scaffold and evaluation protocol for deliberative agents and conduct a systematic evaluation of a range of representative LLMs. The results reveal that complex deliberative collaboration tasks continue to challenge state-of-the-art language models. Even with the aid of external mathematical tools, language models may fail in either the deliberation process for aligning information or the complex reasoning process for making the decision. On the other hand, diagnostic analysis reveals that the deliberation process may also provide opportunities for reflection and error correction, sometimes improving performance over centralized baselines. Altogether, our work establishes a foundation for evaluating and improving LLM agents in deliberative collaboration and provides insights into the strengths, limitations, and properties of current LLM-based multi-agent systems.

Chinese Translation

深思熟虑在协作中扮演着至关重要的角色；当人类共同工作时，他们自然而然地进行沟通，以对齐信息并达成一致。本文研究了在部分可观察的联合决策任务下的深思熟虑大型语言模型（LLM）代理。我们将深思熟虑的协作形式化为一个具有部分和不对称观察的合作联合决策问题，并引入一个可扩展的基准，这一基准在多个任务设置和领域中体现了这一问题，其中代理必须通过深思熟虑交换信息，以达成共享奖励的联合决策。接着，我们建立了一个参考框架和评估协议，用于评估深思熟虑代理，并对一系列代表性的LLM进行了系统评估。结果显示，复杂的深思熟虑协作任务仍然对当前最先进的语言模型构成挑战。即使借助外部数学工具，语言模型在信息对齐的深思熟虑过程中或在决策的复杂推理过程中可能也会失败。另一方面，诊断分析显示，深思熟虑过程也可能提供反思和错误纠正的机会，有时能够在性能上超过集中基线。总的来说，我们的工作为评估和改进深思熟虑协作中的LLM代理奠定了基础，并为当前基于LLM的多代理系统的优点、局限性和特性提供了洞见。

View on arXiv Download PDF AI Translation

cs.CL / 32 / 2607.06160

LongCrafter: Towards Diverse Long-Context Understanding via Evidence-Graph-Guided Instruction Synthesis

LongCrafter：通过证据图引导的指令合成实现多样化的长上下文理解

Yuan, Chenhao, Xu, Yinhao, Xu, Shuwen, Yang, Xizhi, Liu, Jiaxiang, Zhou, Chenxi, Huang, Shaoping, Ren, Haolin, Cao, Pengfei, Zhao, Jun, Liu, Kang

Abstract

Synthesizing long-context supervised fine-tuning (SFT) data is a scalable way to enhance the long-context understanding of large language models (LLMs), yet existing approaches share three limitations: narrow task coverage, insufficient instruction difficulty, and a lack of faithfulness supervision. We propose \textbf{LongCrafter}, a structured synthesis framework that couples a hierarchical task taxonomy with an evidence-grounded pipeline. The taxonomy organizes long-context understanding into local/shallow and global/deep levels and yields 32 fine-grained task types that serve as a global generative prior. Guided by this taxonomy, LongCrafter constructs task-aligned long contexts, decomposes them into explicit evidence graphs that model cross-paragraph dependencies, and generates instruction--response pairs strictly grounded in the located evidence spans, ensuring both controllable difficulty and faithful, traceable reasoning. Models fine-tuned on LongCrafter data outperform all SFT baselines and even the official post-trained models on LongBench, LongBench~v2, and LooGLE across both Qwen2.5-7B and LLaMA-3.1-8B, with the largest gains on high-difficulty tasks. Further analysis shows that LongCrafter data is more diverse and better spread across difficulty levels, and that the trained models locate evidence robustly regardless of position, effectively mitigating the ``lost in the middle'' problem.

Chinese Translation

合成长上下文监督微调（SFT）数据是一种可扩展的方法，可以增强大型语言模型（LLMs）的长上下文理解，但现有方法存在三大局限性：任务覆盖面狭窄、指令难度不足以及缺乏可信监督。我们提出了 extbf{LongCrafter}，一个结构化合成框架，将分层任务分类法与基于证据的管道相结合。该分类法将长上下文理解组织为局部/浅层和全局/深层两个层次，并产生32种细粒度任务类型，作为全局生成先验。在该分类法的指导下，LongCrafter构建与任务对齐的长上下文，将其分解为显式证据图，以建模跨段落的依赖关系，并生成严格基于定位证据范围的指令-响应对，确保可控的难度和可信、可追溯的推理。基于LongCrafter数据微调的模型在LongBench、LongBench~v2和LooGLE上的表现超越了所有SFT基线，甚至超过了官方后训练模型，在Qwen2.5-7B和LLaMA-3.1-8B上，尤其在高难度任务中获得了最大的提升。进一步分析表明，LongCrafter数据在难度水平上更具多样性和更好地分布，训练后的模型能够稳健地定位证据，无论其位置如何，有效缓解了“迷失在中间”问题。

View on arXiv Download PDF AI Translation

cs.CL / 33 / 2607.06175

Improving LLM-Generated Process Model Quality Through Reinforcement Learning: The Role of Reward Function Design

通过强化学习提高LLM生成的流程模型质量：奖励函数设计的作用

Rombach, Alexander, Lauer, Chantale, Mehdiyev, Nijat

Abstract

Large language models (LLMs) can generate BPMN process models from natural-language descriptions, yet supervised fine-tuning (SFT) limits their output quality to the patterns present in the training data. Reinforcement learning (RL) can optimize beyond this ceiling using external quality measures, but how the reward function should be designed when quality is multi-dimensional remains unexplored. We present a systematic investigation of reward function design for RL-based process model generation, training two LLM families (Llama~3.1 8B, Qwen~2.5 14B) under 48 configurations using Group Sequence Policy Optimization with rewards derived from an automated evaluation framework comprising 38 metrics across syntactic, pragmatic, and semantic quality. Three findings emerge. First, RL significantly improves pragmatic and syntactic quality while preserving semantic fidelity, reducing output variability by more than sixfold. Second, equal reward weighting consistently outperforms targeted weighting: emphasizing a specific dimension fails to improve it and can collapse the model into a low-quality mode. Third, design choices interact with model architecture in non-trivial ways: the invalidity penalty is essential for one model but irrelevant for the other, and SFT initialization is indispensable for one architecture but counterproductive for another. These results demonstrate that reward composition is a primary determinant of optimization outcomes, with effects as large as the decision to apply RL itself. The findings generalize to any structured generation task where quality is assessed along multiple automated dimensions. We release our implementation and experimental code at https://github.com/chlauer99/RL_for_process_modeling.

Chinese Translation

大型语言模型（LLMs）能够根据自然语言描述生成BPMN流程模型，然而，监督微调（SFT）的输出质量受到训练数据中存在的模式的限制。强化学习（RL）利用外部质量指标可以超越这一限制，但在质量多维度的情况下，奖励函数应如何设计尚未得到探索。我们对基于RL的流程模型生成的奖励函数设计进行了系统性研究，训练了两个LLM系列（Llama~3.1 8B，Qwen~2.5 14B），在48种配置下使用从一个自动评估框架中衍生的奖励进行组序列策略优化，该框架包含了38个度量，涵盖了句法、语用和语义质量。我们的研究得出了三项发现。首先，RL显著提高了语用和句法质量，同时保持了语义准确性，减少了输出变异性超过六倍。其次，等权重奖励始终优于有针对性的权重：强调特定维度并未改善其质量，反而可能导致模型陷入低质量模式。第三，设计选择与模型架构以非平凡的方式相互作用：一个模型需要无效性惩罚，而另一个模型则无关紧要；SFT初始化对于一种架构是不可或缺的，但对于另一种架构则适得其反。这些结果表明，奖励组合是优化结果的主要决定因素，其影响程度与决定是否应用RL本身相当。这些发现可以推广到任何在多个自动化维度上评估质量的结构化生成任务。我们在https://github.com/chlauer99/RL_for_process_modeling发布了我们的实现和实验代码。

View on arXiv Download PDF AI Translation

cs.CL / 34 / 2607.06196

Pluralis v0.1: Towards a Multicultural, Multimodal, Multilingual Benchmark for AI Risk and Reliability

Pluralis v0.1：迈向一个多元文化、多模态、多语言的人工智能风险与可靠性基准

Parrish, Alicia, Shinde, Rajat, Badhe, Sanket, Bai, Xinyi, Balija, Sree Bhargavi, Chu, Hua-Rong, Ferrara, Emilio, Foundjem, Armstrong, Ghosh, Rajat, Gupta, Aakash, He, Xuanli, Hui, Ong Chen, Jung, Minji, Karimanal, Madhangi, Khattak, Faiza Khan, Kim, Boryoung, Kim, Eugenia, Lavitas, Liliya, Lim, Seok Min, Lu, Victor, Moirangthem, Jim, Nagasubramanian, Dhivya, Pandita, Deepak, Rajagopal, Sita, Raju, Geetha, Razumovskaia, Evgeniia, Reddy, Aravind, Ricciuti, Federico, Sarwar, Nobin, Shin, Sungpil, Sitaram, Sunayana, Thorat, Snehal, Weerasooriya, Tharindu Cyril, Bastings, Jasmijn, Baumann, Joachim, Chen, Kongtao, Emani, Murali, Hendriksen, Mariya, Jin, Jiho, Kim, Jun Seong, Ko, Younghoon, Kwasniewska, Alicja, Lee, Minjae, Manjusha, Tom Wei-cyuan Lin Kashyap Ramanandula, Myung, Junho, Park, Junyeong, Patel, Roma, Ratan, Shyam, Santhiappan, Sudarsun, Suresh, Priyanka, Tuesday, Amortegui-Ordonez, Ksheeraj Sai Vepuri Laura, Dennis, Claire, Kahng, Minsuk, Knotz, Chris, Oh, Alice, Ravindran, Balaraman, Bartholomew, Soojung Ryu William, Tesfaye, Hiwot, Aroyo, Lora

Abstract

Current AI safety evaluation and benchmarking frameworks predominantly rely on Western-centric culture-agnostic defaults that mask critical regional laws, socio-linguistic nuances, and cultural taboos, leaving Vision-Language Models (VLMs) vulnerable in global deployments. We introduce Pluralis v0.1: a novel multimodal, multi-regional, and multilingual dataset built from a culture-first perspective. Spanning 6,448 prompts across six Asia-Pacific countries (Bangladesh, India, Korea, Pakistan, Singapore, Taiwan) and eight languages, Pluralis diverges from prior work by natively sourcing localized safety hazards rather than adapting Western datasets. Crucially, it introduces a multimodal evaluation paradigm: user text (e.g., "Should I gift this?") and an image referring to "this" (e.g., a clock) - both innocuous in isolation, but synergistically triggering specific legal or cultural violations. Pluralis disentangles universal safety violations from localized cultural appropriateness, establishing the latter as a first-class evaluation axis. To operationalize this, we present Judge-Pluralis, an agreement-gated LLM-as-a-Judge ensemble trained on examples classified in an empirically derived cultural taxonomy. Observing VLM behavior on a subset of the Pluralis surfaces recurring, locale-specific failure modes such as image misidentifications with downstream harm, missed item-context-locale interactions, and inadequate refusals. These failure modes vary systematically across locales and languages, exposing blind spots that globally averaged metrics conceal. Ultimately, Pluralis is not presented as a solved evaluation framework for cultural alignment, but rather as a first step and catalyst for future innovation. We call upon the research community to utilize this foundation to advance the science of multilingual, multicultural evaluation to better support AI cultural alignment globally.

Chinese Translation

当前的人工智能安全评估和基准框架主要依赖于以西方为中心的文化中立默认设置，这掩盖了重要的地区法律、社会语言学细微差别和文化禁忌，使得视觉语言模型（Vision-Language Models, VLMs）在全球部署中面临脆弱性。我们引入了Pluralis v0.1：一个从文化优先视角构建的新颖的多模态、多地区和多语言数据集。该数据集覆盖了来自六个亚太国家（孟加拉国、印度、韩国、巴基斯坦、新加坡、台湾）的6,448个提示和八种语言，Pluralis与之前的工作不同，采用本地化的安全隐患作为数据源，而不是改编西方数据集。至关重要的是，它引入了一种多模态评估范式：用户文本（例如，“我应该送这个吗？”）和一个指代“这个”的图像（例如，一只钟）——这两者在单独情况下看似无害，但协同作用可能触发特定的法律或文化违规。Pluralis将普遍安全违规与本地文化适宜性区分开来，将后者确立为一种一级评估轴。为了实现这一点，我们提出了Judge-Pluralis，一个基于协议的LLM作为评判者的集成，针对在经验导出的文化分类法中分类的示例进行训练。在观察VLM在Pluralis子集上的行为时，发现了诸如图像误识别导致的下游危害、遗漏的项目-上下文-地点交互以及不充分拒绝等特定于地方的重复失效模式。这些失效模式在区域和语言之间表现出系统性的变化，揭示了全球平均指标掩盖的盲点。最终，Pluralis并不是一个针对文化对齐的已解决评估框架，而是一个第一步和未来创新的催化剂。我们呼吁研究社区利用这一基础，推进多语言、多文化评估的科学，以更好地支持全球人工智能的文化对齐。

View on arXiv Download PDF AI Translation

cs.CL / 35 / 2607.06229

Spider 2.0-AIFunc: Extending Real-World Text-to-SQL to AI-Native SQL Workflows

Spider 2.0-AIFunc：将现实世界的文本转SQL扩展至AI原生SQL工作流

Liu, Tianyang, Xu, Canwen, Lei, Fangyu, Kuang, Nikki Lijing, Chen, Jixuan, Yu, Tao, McAuley, Julian, Yao, Zhewei, He, Yuxiong

Abstract

Major cloud data platforms now expose large language model capabilities as native SQL functions, enabling analysts to perform classification, filtering, sentiment analysis, extraction, similarity search, and aggregation within ordinary SQL queries. Yet existing text-to-SQL benchmarks evaluate only conventional SQL and provide no signal on whether models can generate such AI-native SQL. We introduce Spider 2.0-AIFunc, a benchmark of 465 verified instances across 125 real-world databases covering six types of AI functions on the Snowflake platform. Starting from an existing enterprise text-to-SQL benchmark, we construct Spider 2.0-AIFunc through an agent-based pipeline that rewrites source tasks into AI-native form, simultaneously transforming target queries and refining natural language instructions to make the intended AI-native solution explicit and reduce ambiguity. All instances pass a multi-round repeated execution protocol across temporally separated windows to confirm result stability before release. Evaluating ten state-of-the-art language models, we find that the strongest proprietary models reach 67-70% execution accuracy while the best open-source model achieves 58.1%, a gap driven primarily by errors in predicate specification, schema grounding, and AI function parameterization. Agent frameworks designed for traditional text-to-SQL challenges, such as schema retrieval and relevant table selection, do not transfer effectively to AI-native SQL: a minimal agent setup consistently matches or outperforms more elaborate alternatives, suggesting that the strategies these frameworks employ are less critical in this setting. Data are available at https://github.com/Leolty/Spider2-AIFunc .

Chinese Translation

主要的云数据平台如今将大语言模型能力作为原生SQL函数公开，使得分析师能够在普通SQL查询中执行分类、过滤、情感分析、提取、相似性搜索和聚合等操作。然而，现有的文本转SQL基准仅评估常规SQL，并未提供模型是否能够生成这样的AI原生SQL的信号。我们引入Spider 2.0-AIFunc，这是一项涵盖六种类型AI功能的基准测试，包含465个经过验证的实例，涉及125个现实世界数据库，基于Snowflake平台。我们从现有的企业文本转SQL基准出发，通过一个基于代理的管道，将源任务重写为AI原生形式，同时转换目标查询，并优化自然语言指令，以明确所期望的AI原生解决方案，减少模糊性。所有实例都经过多轮重复执行协议，验证结果的稳定性后才发布。我们评估了十种最先进的语言模型，发现最强的专有模型的执行准确率达到67-70%，而表现最佳的开源模型则达到58.1%，这一差距主要源于谓词规范、模式基础和AI功能参数化方面的错误。针对传统文本转SQL挑战（如模式检索和相关表选择）设计的代理框架并未能有效迁移到AI原生SQL：一个最小的代理设置始终能够与更复杂的替代方案匹敌或超越，表明这些框架所采用的策略在这一环境中并不那么关键。数据可在 https://github.com/Leolty/Spider2-AIFunc 获取。

View on arXiv Download PDF AI Translation

cs.CL / 36 / 2607.06258

Early Language Learning via Spreading Activation and Category Exploration in Complex Networks

通过扩散激活和类别探索在复杂网络中进行早期语言学习

Citraro, Salvatore

Abstract

Is word acquisition in children uneven with respect to semantic and lexical categories? To answer this question, we model early language learning as a search on a graph-based mental lexicon, driven by two interacting processes: spreading activation and an enforced exploration (rather than exploitation) of lexical categories. We evaluate model performance on four languages (German, English, Dutch, and Rioplatense Spanish), using CDIs as ground-truth data for lexical categories, normative ages derived from the Wordbank repository, and state-of-the-art resources for reconstructing graphs of word similarities. We find that spreading activation outperforms a shortest path baseline in simulating normative word acquisition. At the category level, we highlight complex transitions between CDIs. By studying their sequences in terms of burstiness and average persistence time within the same CDI, we find that spreading activation better captures the exploration dynamics observed empirically. Overall, our findings suggest that vocabulary development can be understood through the non-trivial interplay between activation dynamics and some degree of constraints regulating the visiting of lexical categories in complex networks.

Chinese Translation

儿童的词汇获取在语义和词汇类别上是否存在不均衡现象？为了解答这个问题，我们将早期语言学习建模为在基于图的心理词汇中的搜索，这一过程由两个相互作用的过程驱动：扩散激活和对词汇类别的强制探索（而非利用）。我们在四种语言（德语、英语、荷兰语和里奥普拉滕西西班牙语）上评估模型性能，使用CDI作为词汇类别的真实数据，基于Wordbank数据库推导的规范年龄，以及用于重建词汇相似度图的最先进资源。我们发现，扩散激活在模拟规范词汇获取方面优于最短路径基线。在类别层面，我们强调了CDI之间的复杂过渡。通过研究它们在同一CDI内的突发性和平均持续时间的序列，我们发现扩散激活更好地捕捉了经验观察到的探索动态。总体而言，我们的研究结果表明，词汇发展可以通过激活动态与某种程度的约束之间的非平凡相互作用来理解，这些约束调节了在复杂网络中对词汇类别的访问。

View on arXiv Download PDF AI Translation

cs.CL / 37 / 2607.06289

From Sinhala to Dhivehi: Cross-Lingual Transfer Learning for Low-Resource Speech Recognition

从僧伽罗语到迪维希语：低资源语音识别的跨语言迁移学习

Ilyas, Lukmal, Jayatilleke, Nevidu

Abstract

Dhivehi, the national language of the Maldives, is currently under-resourced for automatic speech recognition (ASR) and other NLP tasks. This study investigates whether cross-lingual transfer learning from Sinhala, a linguistically related, relatively well-resourced Insular Indo-Aryan language, can improve Dhivehi ASR. We conduct seventeen experiments across five transfer learning paradigms: Dhivehi-only baselines, sequential fine-tuning, multilingual fine-tuning, continual pre-training, and a control using Turkish as an unrelated language. The strongest system, continual pre-training on Sinhala followed by fine-tuning on Dhivehi with KenLM, achieves 12.89% WER and 2.70% CER, outperforming the Dhivehi-only baseline by 13.50% WER and 3.02% CER. However, the adaptation strategy and decoding configuration are equally critical for a successful transfer learning experiment. We conduct seventeen controlled experiments spanning five transfer learning paradigms: Dhivehi-only baselines, sequential fine-tuning, multilingual fine-tuning, continual pre-training, and a control experiment using Turkish as an unrelated language. The strongest system, continual pre-training on Sinhala followed by fine-tuning on Dhivehi with KenLM, achieves 12.89% WER and 2.70% CER, outperforming the Dhivehi-only baseline by 13.50% WER and 3.02% CER. The Turkish control experiment confirms that observed improvements stem from linguistic relatedness; adaptation strategy and decoding configuration are also critical.

Chinese Translation

迪维希语是马尔代夫的官方语言，目前在自动语音识别（ASR）和其他自然语言处理（NLP）任务上资源不足。本研究探讨了是否可以通过从相对资源充足、语言上相关的岛屿印欧语的僧伽罗语进行跨语言迁移学习来提升迪维希语的ASR。我们在五种迁移学习范式下进行了十七次实验：仅使用迪维希语的基线、顺序微调、多语言微调、持续预训练以及使用土耳其语作为无关语言的对照实验。最强的系统是对僧伽罗语进行持续预训练，然后使用KenLM对迪维希语进行微调，取得了12.89%的字错误率（WER）和2.70%的字符错误率（CER），比仅使用迪维希语的基线提高了13.50%的WER和3.02%的CER。然而，适应策略和解码配置对成功的迁移学习实验同样重要。土耳其语的对照实验确认所观察到的改进源于语言的相关性；适应策略和解码配置也是关键因素。

View on arXiv Download PDF AI Translation

cs.CL / 38 / 2607.06327

Estimating Uncertainty from Reasoning: A Large-Scale Study of Multi- and Crosslingual MCQA Performance in LLMs

从推理中估计不确定性：大规模多语言和跨语言 MCQA 性能研究

Alfarano, Andrea, Bacciu, Andrea, Mansour, Saab, Mantrach, Amin, Federico, Marcello

Abstract

Uncertainty estimation (UE) enables LLM-powered systems to recognize when to abstain, yet existing research has predominantly focused on English. We present the first large-scale evaluation of UE methods across 22 languages, spanning high-, mid-, and low-resource settings. Using two human-curated Q\&A datasets, we compare open and closed box UE methods (nine in total) across different model sizes and architectures while eliciting long-form reasoning, avoiding LLM-as-a-judge and embedding-based scoring, which can introduce evaluation noise. We report three main actionable findings. First, we find that prompting models to reason in English while keeping questions in low-resource languages substantially improves UE performance, suggesting that comprehension of low-resource languages is largely intact, and that the reliability bottleneck lies in generation rather than understanding. Second, prompting models to reason in English closes the UE performance gap between low and high-resource languages, demonstrating that generation language matters more than the question language. Third, the choice of UE method should depend on model scale: at smaller scales, open-box probability-based methods outperform alternatives; at larger scales, closed-box self-verbalized uncertainty becomes superior. Finally, we provide an analysis of threshold selection for selective prediction, offering guidance on calibrating abstention in multilingual settings.

Chinese Translation

不确定性估计（UE）使得基于大型语言模型（LLM）的系统能够识别何时应当放弃回答，然而现有研究主要集中在英语上。我们呈现了对22种语言中UE方法的大规模首次评估，涵盖了高、中、低资源环境。通过使用两个经过人工筛选的问答数据集，我们比较了不同模型规模和架构下的开放式和封闭式UE方法（共九种），同时引导长形式推理，避免使用LLM作为判断者和基于嵌入的评分，这可能会引入评估噪声。我们报告了三个主要的可操作发现。首先，我们发现，当以英语提示模型进行推理而问题保持在低资源语言时，UE性能显著提升，这表明对低资源语言的理解基本完好，而可靠性瓶颈主要在于生成而非理解。其次，提示模型以英语推理缩小了低资源与高资源语言之间的UE性能差距，显示生成语言的重要性超过了问题语言。第三，UE方法的选择应取决于模型规模：在较小规模时，开放式概率基础方法表现优于其他选项；在较大规模时，封闭式自我表述的不确定性表现更为优越。最后，我们提供了选择阈值用于选择性预测的分析，为多语言环境下的放弃校准提供指导。

View on arXiv Download PDF AI Translation

cs.CL / 39 / 2607.06364

Automated Compliance Mapping in Cloud Security with Domain-Adapted Sentence Transformers

基于领域适应的句子变换器的云安全自动合规映射

Bianchi, John, Petrillo, Luca, Martinelli, Fabio, Petrocchi, Marinella

Abstract

Mapping cloud security controls to technical metrics is currently a manual process. This paper proposes domain adaptation of Sentence Transformer models to automate it. We build a training corpus of 3,499 semantic pairs from five European security standards and a set of technical metrics, then expand it via back-translation and LLM-based paraphrasing to up to 13,996 samples across four scenarios. We fine-tune five architectures and evaluate their performance on two independent tasks: control-to-metric and cross-standard controls association. All fine-tuned models outperform their zero-shot baselines. On the control-to-metric task, the best model gains up to 23 nDCG@10 points, while on the cross-standard control task, \textit{multi-qa-mpnet-dot-v1} under back-translation reaches 0.870 nDCG@10. The results show that in-domain training data is a primary driver of performance for the considered case studies.

Chinese Translation

将云安全控制映射到技术指标目前是一个手动过程。本文提出将句子变换器模型的领域适应用于自动化这一过程。我们从五个欧洲安全标准和一组技术指标构建了一个包含3,499个语义对的训练语料库，然后通过反向翻译和基于大型语言模型（LLM）的意译将其扩展到四种场景下的13,996个样本。我们对五种架构进行了微调，并在两个独立任务上评估它们的性能：控制与指标的映射以及跨标准控制关联。所有微调模型的表现均优于其零样本基线。在控制与指标的映射任务中，最佳模型提升了最多23个 nDCG@10 点，而在跨标准控制任务中，采用反向翻译的 extit{multi-qa-mpnet-dot-v1} 达到了 0.870 的 nDCG@10。结果表明，领域内训练数据是考虑的案例研究中性能的主要驱动因素。

View on arXiv Download PDF AI Translation

cs.CL / 40 / 2607.06452

From Voting to Agent Collaboration: Answer-Type-Aware LLM Pipelines for BioASQ 14b

从投票到智能体协作：针对答案类型的LLM管道在BioASQ 14b中的应用

Roh, Taeyun, Lee, Eunha, Jang, Wonjune, Chung, Sohyun, Jung, Junha, Kang, Jaewoo

Abstract

Biomedical question answering requires not only accurate extraction of information from scientific literature but also reliable integration of evidence across multiple documents. This study presents a question-type-specific large language model (LLM) framework for BioASQ 14b Task B, designed to improve answer robustness and evidence grounding in biomedical question answering. Rather than applying a single prompting strategy to all questions, the framework selects different inference procedures for yes/no, factoid, and list questions according to their distinct reasoning and evaluation requirements. For yes/no questions, snippet shuffling and self-reflection are used to reduce sensitivity to evidence ordering and improve decision stability. For factoid questions, full-snippet input is combined with chain-of-thought-based in-context learning to support accurate biomedical entity identification. For list questions, a multi-agent architecture is employed, in which evidence extraction, candidate generation, answer verification, and final aggregation are handled collaboratively. Preliminary experiments on BioASQ 13b were used to identify effective inference strategies for each question type, and the resulting framework was subsequently evaluated in the official BioASQ 14b Task B challenge. In the official evaluation, our framework showed competitive performance across multiple batches and achieved first place in the factoid subtask of Batch 4. These results demonstrate the effectiveness of combining question-type-specific inference, ensemble prediction, and agent-based verification for reliable biomedical question answering.

Chinese Translation

生物医学问题回答不仅需要从科学文献中准确提取信息，还需要在多个文档中可靠地整合证据。本研究提出了一种针对问题类型的大型语言模型（LLM）框架，旨在改善生物医学问题回答中的答案鲁棒性和证据基础，尤其针对BioASQ 14b任务B。该框架并不是对所有问题应用单一的提示策略，而是根据是/否、事实性和列表问题的不同推理和评估需求，选择不同的推理过程。对于是/否问题，采用片段洗牌和自我反思的方法，以降低对证据顺序的敏感性并提高决策稳定性。对于事实性问题，结合完整片段输入与基于思维链的上下文学习，以支持准确的生物医学实体识别。对于列表问题，采用多智能体架构，协同处理证据提取、候选生成、答案验证和最终聚合。我们在BioASQ 13b上的初步实验帮助识别了每种问题类型的有效推理策略，随后在官方BioASQ 14b任务B挑战中对最终框架进行了评估。在官方评估中，我们的框架在多个批次中表现出竞争力，并在第4批的事实性子任务中获得第一名。这些结果证明了结合特定问题类型推理、集成预测和基于智能体的验证在可靠的生物医学问题回答中的有效性。

View on arXiv Download PDF AI Translation

cs.CL / 41 / 2607.06482

Data Analysis in the Wild: Benchmarking Large Language Models Against Real-World Data Complexities

实地数据分析：针对真实世界数据复杂性对大型语言模型的基准评估

Hasegawa, So, Sampat, Shailaja Keyur, Liu, Lei, Chen, Wei-Peng

Abstract

Current benchmarks for evaluating Large Language Models (LLMs) in data analysis often fail to reflect real-world settings. They typically focus on fact retrieval from small tables and overlook the challenges of large multi-tabular datasets, external knowledge integration, and exploratory insight discovery. We introduce DataGovBench, a benchmark derived from governmental open data designed to evaluate LLMs in practical scenarios. The benchmark includes two tasks: Table QA that requires solving complex decomposable questions and producing textual answers or visualizations, and Table Insight that evaluates the ability of models to generate expert-level findings through exploratory data analysis. Comprehensive experiments with state-of-the-art LLMs, both with and without agentic frameworks, reveal significant performance gaps across both tasks. These results suggest that current LLM-based systems remain far from satisfying the demands of real-world data analytics. DataGovBench provides a challenging benchmark for advancing research on LLMs capable of both answering analytical queries and discovering insights from data. Code and sample data are available at https://github.com/SoHasegawa/datagovbench.

Chinese Translation

当前用于评估大型语言模型（LLMs）在数据分析中的基准测试常常无法反映真实的工作环境。它们通常集中在从小型表格中检索事实，忽视了大型多表数据集、外部知识整合和探索性洞察发现所面临的挑战。我们引入了 DataGovBench，一个基于政府开放数据的基准，旨在评估 LLMs 在实际场景中的表现。该基准包括两个任务：表格问答（Table QA），要求解决复杂的可分解问题并生成文本答案或可视化；以及表格洞察（Table Insight），评估模型通过探索性数据分析生成专家级发现的能力。对多种最先进的 LLMs 进行了全面实验，包括有代理框架和没有代理框架的模型，结果显示两个任务之间存在显著的性能差距。这些结果表明，当前基于 LLM 的系统仍未能满足真实世界数据分析的需求。DataGovBench 为推进能够回答分析查询并从数据中发现洞察的 LLM 研究提供了一个具有挑战性的基准。代码和示例数据可在 https://github.com/SoHasegawa/datagovbench 获取。

View on arXiv Download PDF AI Translation

cs.CL / 42 / 2607.06495

Pitwall: Faithful Natural-Language Race-Strategy Briefings from a Calibrated Real-Time Monte Carlo Engine

Pitwall：来自校准实时蒙特卡罗引擎的可靠自然语言比赛策略简报

Santillana, Juan S.

Abstract

Live sports commentary is grounded generation under a deadline: statements concern real, named athletes, the grounding state changes every few seconds, and no reference text exists at generation time. We present Pitwall, a production system that generates natural-language Formula 1 strategy briefings in English, Spanish, and Portuguese, treating faithfulness as an architectural property rather than an aspiration: every published sentence is decomposed into typed factual claims (positions, gaps, tyres, pace, overtakes, race control) and each claim is verified against the probabilistic race state that prompted it. The same verifier gates the fine-tuning data: of 3,045 model-written targets, only the 81.9% whose every claim is state-supported are retained, the rest falling back to a provably faithful template, so the generator never sees an ungrounded target. Verification is meaningful because of the grounding substrate: a vectorized Monte Carlo engine (N=2,000 per-lap race continuations) calibrated on 126 races (2018-2024) and validated on fully held-out 2025-2026 seasons (winner-in-top-3 90.3% over 155 backtests; held-out Brier 0.0745). A recurring finding spans both halves of the system: virtues trade off and must be gated separately. In simulation, calibration-optimal is not decision-optimal; in generation, fine-tuning on richer targets buys vividness that collapses into hallucination when the grounding state is sparse -- a failure a four-base replication traces to base-model instruction adherence, not scale, and that sparse-context auditing removes from the production model. End-to-end operation -- live timing to verified trilingual briefings -- was confirmed at two consecutive live Grands Prix (Austria and Britain, 2026); at Silverstone a timestamped probability trace, committed to disk before the outcome was known, locked onto the eventual winner ten laps before the flag.

Chinese Translation

现场体育评论是在截止日期下的生成：所述内容涉及真实的、具名的运动员，基础状态每隔几秒钟就会变化，并且在生成时不存在参考文本。我们提出了Pitwall，一个生成自然语言公式1比赛策略简报（支持英语、西班牙语和葡萄牙语）的生产系统，将可靠性视为一种架构属性，而非期望：每个发布的句子被分解为类型化的事实声明（位置信息、差距、轮胎、速度、超车、比赛控制），并且每个声明都与促成它的概率比赛状态进行验证。同样的验证器控制微调数据：在3,045个模型撰写的目标中，只有81.9%每个声明都有状态支持的目标被保留，其余的则退回到一个可证明的可靠模板，因此生成器从未看到一个没有基础的目标。验证是有意义的，因为基础信息：一个向量化的蒙特卡罗引擎（每圈2,000次比赛延续）在126场比赛（2018-2024）上进行了校准，并在完全隔离的2025-2026赛季上进行了验证（155次回测中，胜者排名前3的概率为90.3%；隔离的Brier分数为0.0745）。一个反复出现的发现贯穿系统的两个部分：优点需要进行权衡，并且必须单独控制。在模拟中，校准最佳并不等于决策最佳；在生成中，对丰富目标的微调提高了生动性，但当基础状态稀疏时，这种生动性会崩溃为幻觉——这是一个四基础复制追溯到基础模型指令遵循的问题，而非规模影响，稀疏上下文审计从生产模型中剔除了这一缺陷。端到端的操作——从实时计时到经过验证的三语简报——在两个连续的现场大奖赛（奥地利和英国，2026年）中得到了确认；在银石赛道，一个带时间戳的概率轨迹，在结果已知之前就被写入磁盘，在十圈之前就锁定了最终的获胜者。

View on arXiv Download PDF AI Translation

cs.CL / 43 / 2607.06507

DynaKRAG: A Unified Framework for Learnable Evidence Control in Multi-Hop Retrieval-Augmented Generation

DynaKRAG: 一种用于可学习证据控制的统一框架在多跳检索增强生成中的应用

Wu, Yaqi, Guo, Xiaolei, Zhou, Chenyu, Huang, Jiaqi, Zhang, Xianfa, Zhang, Junxu, Yu, Zhuo, Shi, Zhubo, Lin, Jianghao, Ge, Dongdong

Abstract

Multi-hop retrieval-augmented generation (RAG) acquires evidence sequentially, with each new document potentially revealing missing facts, bridge entities, query defects, or sufficient support for answering. Existing methods provide useful operations such as iterative retrieval, query reformulation, evidence critique, and sufficiency judging, but typically organize them within method-specific pipelines or predefined control topologies. This leaves underexplored how to learn a shared state-conditioned policy that chooses among currently valid evidence operations. We introduce DynaKRAG, which formulates multi-hop evidence acquisition as state-conditioned control over atomic evidence operations. At each step, a validity layer constructs the executable action set, and a learned controller selects the next operation. The resulting transition updates the evidence state and may enable new operations at subsequent steps. With Qwen2.5-7B-Instruct, DynaKRAG achieves F1 scores of 0.5998 on HotpotQA, 0.5340 on 2Wiki, and 0.3061 on MuSiQue, outperforming the strongest controlled baseline on all three benchmarks. Replacing the learned controller with a uniform-valid policy reduces F1 by 3.96--5.78 points, while removing sufficiency feedback hurts all three datasets. Controlled retrieval-cap experiments further show that additional retrieval is not uniformly beneficial. Together, these results demonstrate the benefit of coordinating retrieval, diagnosis, and gap-directed acquisition under an evolving evidence state.

Chinese Translation

多跳检索增强生成（RAG）以顺序方式获取证据，每个新文档都有可能揭示缺失的事实、桥接实体、查询缺陷或足够的支持以进行回答。现有方法提供了有用的操作，例如迭代检索、查询重构、证据评估和充分性判断，但通常将其组织在特定方法的管道或预定义的控制拓扑中。这使得如何学习共享的状态条件策略，在当前有效的证据操作中进行选择仍然没有得到充分探讨。我们引入DynaKRAG，它将多跳证据获取表述为对原子证据操作的状态条件控制。在每一步中，一个有效性层构建可执行的动作集合，而一个学习的控制器选择下一个操作。由此产生的转换更新证据状态，并可能在后续步骤中启用新的操作。使用Qwen2.5-7B-Instruct，DynaKRAG在HotpotQA上实现了0.5998的F1分数，在2Wiki上为0.5340，在MuSiQue上为0.3061，超越了所有三个基准中最强控制基线。将学习的控制器替换为统一有效策略使F1分数降低3.96-5.78点，而去除充分性反馈对所有三个数据集产生负面影响。控制检索能力的实验进一步表明，额外的检索并非在所有情况下都具有均匀的好处。这些结果共同展示了在不断演变的证据状态下协调检索、诊断和缺口导向获取的益处。

View on arXiv Download PDF AI Translation

cs.CL / 44 / 2607.06527

RSF-GLLM: Bridging the Semantic Gap in Multi-Hop Knowledge Graph QA via Recurrent Soft-Flow and Decoupled LLM Generation

RSF-GLLM：通过递归软流和解耦大语言模型生成弥合多跳知识图谱问答中的语义差距

Bandyopadhyay, Sambaran, Muppidi, Ananth

Abstract

Multi-hop Question Answering over Knowledge Graphs faces a critical challenge: traditional retrieve-then-read pipelines break differentiability, preventing the retriever from learning to bridge the semantic gap where intermediate nodes lack lexical overlap with the query. To address this, we propose RSF-GLLM, a framework decoupling differentiable graph reasoning from answer generation. Our Recurrent Soft-Flow (RSF) module employs a GRU-guided query updater to propagate continuous relevance scores, utilizing a dynamic gating mechanism to traverse semantically dissimilar bridge nodes via structural cues. We introduce flow sparsity regularization to theoretically guarantee convergence from soft probabilities to discrete reasoning paths. These paths are extracted and textualized to fine-tune a Large Language Model (LLM), ensuring generation is grounded in factual topology. Experiments on WebQSP and CWQ demonstrate that RSF-GLLM achieves competitive performance with superior inference efficiency compared to LLM based computationally expensive approaches.

Chinese Translation

基于知识图谱的多跳问答面临一个关键挑战：传统的检索-再阅读流程破坏了可微性，阻止了检索器学习弥合中间节点与查询之间缺乏词汇重叠的语义差距。为了解决这个问题，我们提出了RSF-GLLM，一个将可微图推理与答案生成解耦的框架。我们的递归软流（Recurrent Soft-Flow, RSF）模块采用GRU引导的查询更新器来传播连续的相关性得分，利用动态门控机制通过结构线索遍历语义不相似的桥接节点。我们引入流稀疏正则化，从理论上保证从软概率到离散推理路径的收敛。这些路径被提取并文本化，以微调大型语言模型（Large Language Model, LLM），确保生成的内容基于事实拓扑。针对WebQSP和CWQ的实验表明，RSF-GLLM在推理效率上优于基于LLM的计算开销较大的方法，且表现具有竞争力。

View on arXiv Download PDF AI Translation

cs.CL / 45 / 2607.06529

Life Style Levels: Neighborhood Delineation using Geospatial Data

生活方式水平：基于地理空间数据的邻里划分

Kulkarni, Srivatsa, Banerjee, Debarag

Abstract

Fine-scale socioeconomic information is often unavailable across rapidly ur-banizing regions of the developing world, like India, limiting the ability to delineate intra-urban variations in affluence and deprivation. This study pro-poses a scalable, grid-based urban delineation framework using building morphology derived from open-source satellite imagery. Urban areas across 59 Indian cities and towns are partitioned into high-resolution spatial grids and characterized using interpretable morphological indicators, which are combined into a transparent, rule-based scoring framework to delineate areas with contrasting levels of urban affluence. The resulting classifications are validated through ground-level Google Street View observations, revealing a sharp contrast between the grid classes which are consistent with the ex-pected effects of the lifestyle affluence indicators. We further investigate density-based clustering of building footprints in Mumbai to identify dense urban settlements, demonstrating that the resulting clusters exhibit substan-tial spatial overlap with known informal settlements across the city. Finally, we conduct an exploratory analysis mapping consumer loan delinquency across the derived affluence classes. By relying entirely on publicly available geospatial data, the proposed framework provides a scalable, interpretable, and cost-effective approach for granular urban affluence mapping across In-dian cities.

Chinese Translation

在快速城市化的发展中国家区域，如印度，细尺度的社会经济信息通常不可获得，这限制了对城市内部富裕与贫困变异的划分。本研究提出了一种可扩展的基于网格的城市划分框架，该框架使用来自开源卫星影像的建筑形态数据。对印度59个城市和城镇的城市区域进行高分辨率空间网格的划分，并使用可解释的形态指标对其进行特征化，这些指标被组合成一个透明的基于规则的评分框架，以划分具有不同城市富裕水平的区域。通过地面Google街景观察对所得到的分类进行验证，揭示了网格类别之间的明显对比，这与生活方式富裕指标的预期效应一致。我们进一步探讨了孟买建筑轮廓的基于密度的聚类，以识别密集的城市居民区，展示了所得聚类与该市已知非正式定居点之间存在显著的空间重叠。最后，我们对不同富裕等级的消费贷款违约情况进行了探索性分析。凭借完全依赖于公开的地理空间数据，该框架为细致的城市富裕地图绘制提供了一种可扩展、可解释且具有成本效益的方法，适用于印度城市。

View on arXiv Download PDF AI Translation

cs.CL / 46 / 2607.06540

Hierarchical Acoustic-Semantic Modeling: Modality Separation and Semantic Coherence for Full-Duplex SLMs

分层声学-语义建模：全双工SLM中模态分离与语义一致性

Liu, Zhenyu, Li, Yunxin, Zhang, Xuanyu, Teng, Qixun, Jiang, Shenyuan, Chen, Haolan, Zhao, Minjun, Meng, Fanbo, Xu, Yu, He, Yancheng, Hu, Baotian, Li, Haizhou, Zhang, Min

Abstract

Developing seamless, high-performance, native intelligent full-duplex Spoken Language Models (SLMs) remains a critical challenge and long-standing goal for the speech and NLP community. Despite notable progress, recent endeavors are fundamentally constrained by severe modality interference, which causes substantial knowledge degradation and compromises semantic integrity -- ultimately making full-duplex SLMs feel unnatural and unintelligent. In this paper, through an exhaustive fine-grained analysis of model optimization dynamics, we uncover the root cause of such performance degradation, revealing that modality interference arises from inherent gradient conflicts between acoustic and semantic modeling when the two modalities are forced to share a deep parameter space. Guided by this key insight, we introduce Lychee-FD, a native end-to-end full-duplex framework designed to mitigate modality interference. Importantly, we propose a hierarchical parameter separation strategy that decouples conflicting modalities in deep layers while preserving cross-modality coherence via a dedicated semantic alignment channel. Extensive experiments on multiple full-duplex benchmarks demonstrate that our method significantly advances the state of the art, yielding substantial improvements in both speech intelligence (+7.4% on Spoken QA) and full-duplex interaction fluidity (+28.5% on FullDuplexBench 1.5) without compromising inference efficiency. To the best of our knowledge, this work is the first to achieve two key advances: 1) uncovering and elucidating the root cause of modality interference in full-duplex SLMs, and 2) designing an elegant hierarchical model together with a practical solution for seamless, high-performance, native intelligent full-duplex SLMs.

Chinese Translation

开发无缝、高性能的原生智能全双工口语语言模型（SLMs）仍然是语音和自然语言处理（NLP）领域面临的一个重大挑战和长期目标。尽管取得了一定的进展，但近期的努力受限于严重的模态干扰，这导致知识显著降级并妨碍语义完整性，从而使全双工SLMs显得不自然且缺乏智能。本文通过对模型优化动态的深入细致分析，揭示了这种性能降级的根本原因，发现模态干扰源于声学和语义建模之间内在的梯度冲突，当这两种模态被迫共享深层参数空间时便会出现此问题。在这一关键洞察的指导下，我们提出了Lychee-FD，一个原生的端到端全双工框架，旨在减轻模态干扰。重要的是，我们提出了一种分层参数分离策略，在深层次中解耦冲突模态，同时通过专用语义对齐通道保持跨模态一致性。针对多个全双工基准的广泛实验表明，我们的方法显著提升了技术水平，在口语智能上提高了7.4%（在口语问答上）以及在全双工交互流畅性上提升了28.5%（在FullDuplexBench 1.5上），且没有影响推理效率。根据我们所知，这项工作首次实现了两个关键突破：1）揭示并阐明全双工SLMs中模态干扰的根本原因，以及2）设计出优雅的分层模型并提供一种实用解决方案，以实现无缝、高性能、原生智能的全双工SLMs。

View on arXiv Download PDF AI Translation

cs.CL / 47 / 2607.06542

On the feasibility of dependency parsing of non-human sequences without a gold standard. Is evaluation possible in other species?

无标准参考的非人类序列的依赖解析可行性探讨。其他物种的评估是否可能？

Ferrer-i-Cancho, Ramon, Hobaiter, Catherine, Bergman, Thore, Gustison, Morgan

Abstract

Dependency parsing consists of finding a tree representation for a sequence. Unsupervised dependency parsing aims to develop parsing methods without a gold standard during model training. In human languages, an unsupervised parser can be evaluated because some gold standard is usually available or can be created. For other species, a gold standard is unknown. Thus one may conclude that it is impossible to determine the accuracy of an unsupervised parser and, consequently, dependency parsing is unfeasible in other species. However, here we apply recent advances in network science to demonstrate that the proportion of correct edges retrieved by a parser must be high for the sequences of vocalizations or gestures that non-human primates produce due to the fast decay of the sequence length distribution. In contrast, human language sequences lack that property. Therefore, evaluation without a gold standard is feasible in non-human primates but a hard problem in humans.

Chinese Translation

依赖解析旨在为一个序列找到树状表示。无监督依赖解析的目标是在模型训练期间开发不依赖于标准参考的方法。在人类语言中，无监督解析器可以被评估，因为通常可以获得或创建某个标准参考。而对于其他物种，标准参考是未知的。因此，人们可能得出结论，无法确定无监督解析器的准确性，从而认为在其他物种中依赖解析是不可行的。然而，在此我们运用网络科学的最新进展，证明解析器所检索的正确边的比例在非人类灵长类动物产生的发声或手势序列中必须较高，这是由于序列长度分布的快速衰减。相比之下，人类语言序列缺乏这种特性。因此，在非人类灵长类动物中，未依赖标准参考的评估是可行的，但在适用于人类则是一个棘手的问题。

View on arXiv Download PDF AI Translation

arXiv Papers

Quaternion-Averaging-Based Adaptive Complementary Filter for Pedestrian Dead Reckoning With a Foot-Mounted AHRS

Learning 4D Geometric Priors for Inference-Efficient World Action Models

GEM-Occ: From Visual Geometry Evidence to Embodied Semantic Occupancy Memory

Dynamic Evaluation of Classical and Control-Aware Optimal Trajectory Planning in Robot Manipulators

Physics-Regularized Machine Learning for Proprioceptive Vehicle Localization Using Onboard Sensors

Efficient Transfer Learning of Robot Dynamic Models Using Morphological Similarity

Uncertainty-Aware Velocity Correction for Proprioceptive Vehicle Localization using Evidential Mamba

IMR: Iterative Mode-World Weighted Regression for Multi-Agent Trajectory Prediction

Co-STAR: Cognitive Stimulation Therapy by an Autonomous Robot for Dementia -- A One-Week In-Home Study

Observation Quality Matters: Robust Multi-Fisheye Calibration via Failure-Oriented Analysis

FORGE: Towards Functional Tool-Use Generalization via Keypoint Trajectory Reasoning

GraspIT: A Dataset Bridging the Sim-to-Real gap and back for Validated Grasping SE(3) Pose Generation

DexTele: A Dual-Arm Dexterous Teleoperation System Based on Motion Retargeting and Adaptive Force Control

Intercepting an Agile Target with Net-Carrying Drones using Competitive Multi-Agent Reinforcement Learning

Delay-Aware Active Triangulation with Uncertainty-Driven Multi-Agent Reinforcement Learning for Counter-UAS

Imagined Rollouts are Kinematic, Not Dynamic: A Diagnosis of Long-Horizon World-Model Failure

RoboTALES: Learning Reasoning-Guided Robot Policies via Task-Aligned Simulated Futures

ThorArena: Benchmarking Humanoid Physical Interaction with Human Motion-Force Demonstrations

MP-MPPI: A Motion Primitive Guided Sampling-Based Optimizer for Model Predictive Control

EAGOR: Embodied Reasoning in Omni-direction

Calf-Integrated Arms for Bimanual Quadruped Loco-Manipulation

APVI-SLAM: Real-Time Acoustic-Pressure-Visual-Inertial Localization and Photorealistic Mapping System in Complex Underwater Environment

RoboVAST: Automated Scenario-Based Validation of Robots at Scale

Diagnosing Semantic Handoff Failures in Agent-Orchestrated Vision-Language-Action Skill Composition

Optimal Transport Q-Learning for Flow Policy Steering and Acceleration

LAMP: Latent Motion Prior-Guided Real-World Learning for Dexterous Hand Manipulation

OrchardBench: A Physically-Grounded, GPU-Parallel Apple-Orchard Simulation Benchmark for Agricultural Robotics

Responsible Personalisation: The Double-Edged Sword of Personalisation in Human-Robot Interaction

Training-Free Acceleration for Vision-Language-Action Models with Action Caching and Refinement

Towards Real-World Applications with an Autonomous Powered Wheelchair

Learning to Throw Objects Safely in Multi-Obstacle Environments

From Foundation to Application: Improving VLA Models in Practice

WristMimic: Full-Body Humanoid Control with Wrist-Guided Manipulation

SIEVE: Structure-Aware Data Selection for Imitation Learning with VLA Models

Hilti-Trimble-Oxford Dataset: 360 Visual-Inertial Benchmark with Floor Plan Priors for SLAM and Localization

Clustering-Embedded Model Predictive Path Integral Control: Avoiding Averaging-Induced Failure and Enabling Efficient Cluster Selection for Dynamic Obstacles

Hypothesis-driven Model Expansion under Uncertainty for Open-World Robot Planning

Neural-ESO: A Dual-Pathway Architecture for Provably Robust Learning-Based Control

UniLM-Nav: A Unified Framework for Zero-Shot Last-Mile Navigation

RynnWorld-Teleop: An Action-Conditioned World Model for Digital Teleoperation

RynnWorld-4D: 4D Embodied World Models for Robotic Manipulation

Embodied Human-Robot Interaction via Acoustics: A MARL Approach with AcoustoBots for Spatial Data Physicalization

Lift3D-VLA: Lifting VLA Models to 3D Geometry and Dynamics-Aware Manipulation

CanvasAgent: Enabling Complex Image Creation and Editing via Visual Tool Orchestration

A Task-Driven Evaluation of UAV Detection and Tracking under Synthetic Fog

Binocular Gaze Estimation with Single Camera and Single Light Source

Ground3D-LMM: Fine-Grained 3D Point Grounding and Spatial Reasoning with LMM

Light-Omni: Reflex over Reasoning in Agentic Video Understanding with Long-Term Memory

Statistical Adversaries: Natural Backdoor-like Features in Vision Datasets

Rendering-Aware Bayesian 3D Gaussian Splatting with Native Uncertainty and Adaptive Complexity Control

Multi-Teacher Contrastive Distillation for Edge-Efficient Pathology Foundation Models

Harnessing Generative Image Models for Training-Free Primitive Shape Abstraction

Hierarchical Classification via Cascading Feature Elimination: Application to Human Phenotype Ontology-Aligned Facial Phenotyping (FaceMesh2HPO)

Patch Knowledge Transfer for Efficient AI-Generated Image Quality Assessment

Cross-Contextual Vision-Language Adaptation with LoRA for Personalized Severe Adverse Event Detection in Clinical Wound Monitoring

Taxlifier: Leveraging Disease Taxonomy for Enhanced Multi-Label Classification in Chest Radiography

Recovering Cloud Microstructures with Cascaded Diffusion Inversion

VEIL: How Visual Encoding Hijacking Induces Bias In Vision Models

REVIVE: A Multi-Modal Framework for Vandalism Detection and Recovery in Autonomous Vehicles

Clustered Codebook Quantization for 2D Gaussian-based Image Compression

Robust Face Super-Resolution and Recognition Through Multi-Feature Aggregation in Diffusion Models

Scene Graph Thinking: Reinforcing Structured Visual Reasoning for Multimodal Large Language Models

Association Restoration Test: Revealing Restorable Shortcuts after Unlearning

SAMPLe: SAM-based Optimizer for Prompt Learning in VLMs

ARMS: Anchor-Relational Motion Streaming for Seamless Solo-Social Motion Transitions

Optimized Adaptive Loop Filter in Versatile Video Coding

Image2Sim: Scaling Embodied Navigation via Generative Neural Simulator

LEGATO 2: Toward Multimodal Sheet Music Recognition and Understanding

Benchmarking the Robustness of Autonomous Driving to Environmental Illusions: A Lane Perception Perspective

DeSeG: Decoupling Semantic Intent and Geometric Constraints for Physically Plausible Human-Scene Interaction

Segmentation before Answering: Pixel Grounding for MLLM Visual Reasoning

TRIG: Trajectory-Rig Decoupled Metric Geometry Learning

Complementary Roles of Image Classification and Vessel Segmentation in AI-Based Screening for Retinopathy of Prematurity Plus Disease in a Kenyan Preterm Cohort

Realistic Compound-Lens Defocus Blur Synthesis

Breaking Spurious Correlations via Generative Randomization and Cross-Variant Self-Supervised Learning

AVA-VLM: Adaptive Visual Attention-Vision Language Model for In-the-Wild Construction Site Monitoring

Harrison.Rad 1.5 Technical Report: A radiology foundation model that can draft reports from images, priors and clinical context

Few-Medoids: An Embarrassingly Simple Coreset Selection Method for Few-Shot Knowledge Distillation

GaussFusion: Towards Multimodal 3D Gaussian Pretraining