← Back to Index
Daily Research Digest

arXiv Papers

2026-06-19
295
Papers
4
Categories
294
Translated
收藏清单 0
机器人学 (Robotics)
65
cs.RO / 1 / 2606.19357

Physical Atari: A Robust and Accessible Platform for Real-time Reinforcement Learning on Robots

物理Atari：一个稳健且易于访问的实时强化学习机器人平台
Javed, Khurram, Modayil, Joseph, Kennickell, Gloria, Sutton, Richard S., Carmack, John
Abstract
We built a robot called the Robotroller that actuates an Atari CX40+ controller and a device called the Atari Devbox that renders the game frame and the reward signal from the Arcade Learning Environment on a screen. The Robotroller and the Atari Devbox, together with an off-the-shelf camera and a desktop computer, constitute a system that can be used to study reinforcement learning algorithms in the physical world. We call the full system Physical Atari. In this paper, we detail the key decisions that make Physical Atari a robust and accessible platform. To make the system robust, we designed the Robotroller so that all movement is done through bearings, which reduces wear. Additionally, we wrote software that monitors the state of the servos at a high frequency and intervenes to limit stress. To make the system accessible, we used affordable off-the-shelf components and parts that can be manufactured using consumer 3D printers. Physical Atari can be built for under $1,000 and has been used for weeks of non-stop reinforcement learning experiments without any mechanical failures. We used it to validate that reinforcement learning algorithms can learn directly on robots and show that even small distribution shifts between learning and deployment can significantly degrade the performance of policies. Our results underscore the importance of on-device adaptation for strong performance on robots.
Chinese Translation
我们构建了一款名为Robotroller的机器人，它驱动着Atari CX40+控制器，并与一个名为Atari Devbox的设备配合使用，该设备在屏幕上渲染来自街机学习环境（Arcade Learning Environment）的游戏画面和奖励信号。Robotroller和Atari Devbox，加上一个现成的相机和一台台式计算机，构成了一个可以用于研究物理世界中强化学习算法的系统。我们将整个系统称为物理Atari。在本文中，我们详细介绍了使物理Atari成为一个稳健且易于访问的平台的关键决策。为了增强系统的稳健性，我们设计了Robotroller，使所有运动都通过轴承进行，从而减少磨损。此外，我们编写了软件，以高频率监控伺服电机的状态，并进行干预以限制压力。为了提高系统的可访问性，我们使用了价格合理的现成组件和可以通过消费级3D打印机制造的零件。物理Atari的构建成本低于1000美元，并且在数周的持续强化学习实验中没有出现任何机械故障。我们利用它验证了强化学习算法可以直接在机器人上学习，并展示了学习与部署之间即使是小的分布偏移也会显著降低策略的性能。我们的结果强调了在设备上适应对于机器人强大性能的重要性。
View on arXiv Download PDF AI Translation
cs.RO / 2 / 2606.19358

WorkBenchMark: A LEGO-Based Assembly Benchmark with an Assembly-by-Disassembly Baseline for the Smart Manufacturing League

WorkBenchMark：一个基于乐高的装配基准，具有拆解式装配基线，面向智能制造联盟
Ma, Wenbo, Swoboda, Daniel, Tschesche, Matteo, Hofmann, Till
Abstract
We introduceWorkBenchMark, a LEGO Duplo-based robotic assembly benchmark motivated by the RoboCup Smart Manufacturing League. Robotic assembly couples low-level manipulation with task-level symbolic reasoning under physical constraints, a combination that current end-to-end learning methods do not yet solve reliably. The benchmark provides 400 tasks across four complexity tiers. We provide an open-vocabulary perception, Assembly-by-Disassembly baseline solution. Our planning-based pipeline outperforms a modern vision-language-action approach across all tiers. The benchmark, simulation environment, and baseline implementation will be released openly to support the broader robotic assembly community.
Chinese Translation
我们介绍了 WorkBenchMark，这是一个基于乐高杜plo的机器人装配基准，旨在支持 RoboCup 智能制造联盟。机器人装配将低级操作与在物理约束下的任务级符号推理相结合，而当前的端到端学习方法尚未可靠地解决这一组合。该基准提供了400个任务，分为四个复杂度层级。我们提供了一种开放词汇的感知和拆解式装配基线解决方案。我们的基于规划的管道在所有层级上均优于现代视觉-语言-动作方法。该基准、仿真环境和基线实现将公开发布，以支持更广泛的机器人装配社区。
View on arXiv Download PDF AI Translation
cs.RO / 3 / 2606.19383

3D Scene Graphs: Open Challenges and Future Directions

3D场景图：开放挑战与未来方向
Rotondi, Dennis, Argenziano, Francesco, Koch, Sebastian, Hughes, Nathan, Buechner, Martin, Wald, Johanna, Schmid, Lukas Rosenberger, Nardi, Daniele, Valada, Abhinav, Paull, Liam, Tombari, Federico, Carlone, Luca, Arras, Kai O.
Abstract
3D Scene Graphs (3DSGs) have emerged as a powerful representation for spatial AI by combining geometric grounding with semantic and relational abstractions of the environment. Their expressiveness has made them relevant to a broad range of problems in robotics and computer vision, including manipulation, navigation, task planning, scene understanding, and many others. However, the field remains fragmented: different communities adopt distinct formulations, construction pipelines, and evaluation protocols, making it difficult to compare methods, identify common assumptions, and assess remaining challenges for robust real-world deployment. This survey provides a unified and critical review of 3DSGs, with particular emphasis on open challenges and future directions. We first formalize 3DSGs under a common definition and analyze the principal modeling choices that characterize existing formulations, including node and edge attributes, hierarchical structure, dynamic scene representations, and affordance-aware extensions. We then review how 3DSGs are built from raw sensory observations, discussing the most common terminologies, conventions, and techniques. Finally, we examine downstream applications and evaluation strategies, from intrinsic graph quality to task-level performance. To support the community, we also provide a dedicated website that organizes and extends the surveyed content, accessible at https://3dscenegraphs.com/.
Chinese Translation
3D场景图（3DSGs）作为一种强大的空间人工智能表示方法，通过将几何基础与环境的语义和关系抽象相结合而出现。它们的表现力使其在机器人技术和计算机视觉等广泛问题中具有相关性，包括操作、导航、任务规划、场景理解等。然而，该领域仍然存在碎片化现象：不同的研究社区采用不同的表述、构建流程和评估协议，这使得比较方法、识别共同假设以及评估实际部署中剩余挑战变得困难。本调查提供了对3DSGs的统一和批判性回顾，特别强调开放挑战和未来方向。我们首先在一个共同的定义下形式化3DSGs，并分析特征化现有表述的主要建模选择，包括节点和边属性、层次结构、动态场景表示以及关注可用性的扩展。然后，我们回顾了如何从原始传感器观测构建3DSGs，讨论最常见的术语、约定和技术。最后，我们考察了下游应用和评估策略，从内在图质量到任务级性能。为了支持社区，我们还提供了一个专门的网站，组织和扩展调查内容，网址为 https://3dscenegraphs.com/。
View on arXiv Download PDF AI Translation
cs.RO / 4 / 2606.19397

DiffusionVS: A Generative Framework for Robust Visual Servoing Based on Diffusion Policy

DiffusionVS：基于扩散策略的鲁棒视觉伺服生成框架
Cui, Hongkang, He, Rui, Chen, Haoyao
Abstract
Visual servoing is a fundamental technique in robotic manipulation and navigation. Regression-based visual servoing frequently experiences trajectory jitter as a result of noise-sensitive single-step mappings and the accumulation of errors during distribution shifts. In contrast, Diffusion Policy maintains temporal consistency by predicting action sequences and improves robustness through implicit data augmentation. This paper presents a novel diffusion-based servoing method. Based on Diffusion Policy, the proposed approach uses normalized image coordinates of observed tag corners as input and generates camera velocity through conditional denoising. To overcome the generalization limitations of models trained on static datasets, an online training paradigm is adopted, continuously expanding the diversity of training data through interactive experience collection. This strategy substantially enhances both the performance and generalization capability of the model. Comprehensive simulations and real-world experiments demonstrate the effectiveness of the proposed method, achieving success rates of nearly 100\% in simulation and 93\% in physical experiments. Beyond the specific pipeline, we further validate the generality of the diffusion mechanism. Experiments show that existing visual servoing networks consistently achieve improved performance when integrated with our diffusion-based module. These results indicate that the proposed strategy possesses broad applicability and can enhance various visual servoing systems beyond the specific architecture presented here.
Chinese Translation
视觉伺服是机器人操作和导航中的一项基本技术。基于回归的视觉伺服常常由于对噪声敏感的单步映射和在分布变化过程中错误的累积而经历轨迹抖动。相比之下，扩散策略通过预测动作序列来保持时间一致性，并通过隐式数据增强提高鲁棒性。本文提出了一种新颖的基于扩散的伺服方法。基于扩散策略，所提方法使用观察到的标签角点的归一化图像坐标作为输入，并通过条件去噪生成相机速度。为了克服在静态数据集上训练的模型的泛化限制，采用了一种在线训练范式，通过互动经验收集不断扩展训练数据的多样性。这一策略显著提升了模型的性能和泛化能力。全面的仿真和实际实验表明，所提方法的有效性，在仿真中成功率接近100\%，在物理实验中达到93\%。超越特定的流程，我们进一步验证了扩散机制的普遍性。实验表明，现有的视觉伺服网络在与我们的基于扩散的模块集成时，性能一致性地得到了提升。这些结果表明，所提策略具有广泛的适用性，能够增强各种视觉伺服系统，超越此处展示的特定架构。
View on arXiv Download PDF AI Translation
cs.RO / 5 / 2606.19419

Playful Agentic Robot Learning

玩耍型自主机器人学习
Zhang, Junyi, Ge, Jiaxin, Yoo, Hanjun, Fu, Letian, Yang, Zihan, Liu, Yaowei, Saravanan, Raj, Yin, Shaofeng, Yu, Justin, Niu, Dantong, Wang, Zirui, Herzig, Roei, Goldberg, Ken, Bai, Yutong, Chan, David M., Stoica, Ion, Kanazawa, Angjoo, Lei, Jiahui, Feng, Haiwen, Darrell, Trevor
Abstract
Current agentic robot systems can write executable Code-as-Policy programs, observe feedback, and revise behavior across multiple attempts, but they remain largely task-driven: reusable skills are acquired only after explicit instructions. We study Playful Agentic Robot Learning, where an embodied coding agent uses self-directed play as a continual skill-learning stage before downstream tasks arrive. We introduce RATs, Robotics Agent Teams designed for play-time skill acquisition. During play, RATs proposes novel yet learnable exploratory tasks, plans and executes robot-code policies, verifies intermediate progress, diagnoses failures, retries with dense, step-level feedback, and distills successful executions into a persistent code skill library. At test time, the agent reuses relevant skills from this frozen library to help solve new tasks. Experiments in LIBERO-PRO and MolmoSpaces show that play-learned skills improve held-out downstream tasks over no-play and random-play baselines, with 20.6 and 17.0 percentage-point gains over CaP-Agent0 on LIBERO-PRO and MolmoSpaces, respectively. Moreover, the learned skills can be plugged into other inference-time Code-as-Policy agents by simply retrieving them into the context, improving RoboSuite and real-world transfer by 8.9 and 8.8 points, respectively, without finetuning the underlying model.
Chinese Translation
当前的自主机器人系统能够编写可执行的代码作为策略（Code-as-Policy）程序，观察反馈，并在多次尝试中修正行为，但它们仍然主要以任务为驱动：可重用的技能仅在明确指令后获得。我们研究了玩耍型自主机器人学习，其中一个具身编码代理在下游任务到来之前，利用自我导向的玩耍作为持续的技能学习阶段。我们引入了RATs（机器人代理团队），旨在通过玩耍时间进行技能获取。在玩耍过程中，RATs提出新颖但可学习的探索任务，规划并执行机器人代码策略，验证中间进展，诊断失败，利用密集的逐步反馈进行重试，并将成功的执行提炼为持久的代码技能库。在测试时，代理从这个冻结的库中重用相关技能，以帮助解决新任务。在LIBERO-PRO和MolmoSpaces中的实验表明，玩耍学习的技能在没有玩耍和随机玩耍的基线之上改善了保留的下游任务，分别在LIBERO-PRO和MolmoSpaces上比CaP-Agent0提高了20.6和17.0个百分点。此外，学习到的技能可以通过简单地将其检索到上下文中，插入到其他推理时的代码作为策略代理中，分别提高了RoboSuite和现实世界转移8.9和8.8个百分点，而无需微调基础模型。
View on arXiv Download PDF AI Translation
cs.RO / 6 / 2606.19504

Simulating Robotic Locomotion in Sand: Resistive Force Theory in an Open-Source Physics Engine

沙中机器人运动的模拟：开放源代码物理引擎中的阻力理论
Brown, Ryan Walker, Treers, Laura K., Daltorio, Kathryn A.
Abstract
Recent advancements in Resistive Force Theory (RFT) enable approximation of ground reaction forces for locomotion in sand without the computational expense of modeling interactions with individual grains. However, these tools have been absent in 3D physics engines commonly used for robot simulation. We explore if resistive force approximations are sufficient, when integrated with standard dynamics calculations, to provide a stable substrate for a freely walking robot. To determine this, we implement 3D Granular Resistive Force Theory (3D RFT) in a physics simulation engine, MuJoCo. We verify simulations in multiple scenarios to demonstrate that key trends due to end effector shape, speed, and loading are preserved. Our implementation predicts walking distance and foot sinkage of a 12-Degree of Freedom hexapod robot within 20\% of experiments in sand. While RFT has inherent approximations, the open source tool described here has potential to help develop new and improved robot designs to traverse granular media substrates.
Chinese Translation
最近在阻力理论（Resistive Force Theory, RFT）方面的进展使得在沙中进行运动时可以近似地计算地面反作用力，而无需对单个颗粒的相互作用进行复杂的建模。然而，这些工具在常用于机器人模拟的三维物理引擎中尚未得到应用。我们探讨了在与标准动力学计算结合时，阻力力近似是否足够为自由行走的机器人提供稳定的基底。为此，我们在物理模拟引擎MuJoCo中实现了三维颗粒阻力理论（3D Granular Resistive Force Theory, 3D RFT）。我们在多种场景中验证了模拟结果，以证明由于末端执行器形状、速度和负载引起的关键趋势得以保留。我们的实现预测了一个12自由度六足机器人在沙中行走距离和足部下沉量的结果，误差在20%以内。尽管RFT具有固有的近似性，但这里描述的开源工具有潜力帮助开发新的和改进的机器人设计，以穿越颗粒介质基底。
View on arXiv Download PDF AI Translation
cs.RO / 7 / 2606.19512

Proprioceptive Invariant State Estimation for Humanoid Robots on Non-Inertial Ground

非惯性地面上类人机器人本体感知不变状态估计
Mandali, Falak, He, Zijian, Gu, Yan
Abstract
This paper presents an invariant extended Kalman filtering (InEKF) approach for real-time state estimation of humanoid robots operating on non-inertial ground using only onboard proprioceptive sensing. The proposed approach estimates the robot's base position and velocity relative to the moving ground frame without requiring direct measurements of ground motion or externally mounted sensors. By exploiting kinematic constraints at the stance foot through foot-mounted IMUs, the filter accounts for ground-induced nonlinearities in the process and measurement models while remaining fully proprioceptive. The estimator is formulated to admit a right-invariant measurement model, enabling favorable error dynamics under large initial uncertainties. Observability analysis establishes conditions under which the robot's relative base position and velocity are observable with respect to the non-inertial ground frame. Experiments with the Digit humanoid robot standing and squatting atop a swaying and pitching ground showcase a 96% speedup in convergence rate and an 80% reduction in position estimate errors over existing InEKFs. Walking experiments on a uni-axially rotating ground achieve an average estimation error of less than 9 cm for an initial error of up to 1 m.
Chinese Translation
本文提出了一种不变扩展卡尔曼滤波（InEKF）方法，用于实时状态估计类人机器人在非惯性地面上仅使用机载本体感知传感器的操作。所提出的方法估计机器人基座相对于移动地面框架的位置和速度，而无需直接测量地面运动或外部安装传感器。通过利用站立脚的运动学约束，结合脚部安装的惯性测量单元（IMU），该滤波器在过程和测量模型中考虑了地面引起的非线性，同时保持完全本体感知。估计器被构造为允许右不变测量模型，从而在较大的初始不确定性下实现有利的误差动态。可观测性分析建立了机器人相对于非惯性地面框架的相对基座位置和速度可观测的条件。使用Digit类人机器人在摇摆和俯仰的地面上站立和下蹲的实验展示了收敛速率提高96%和位置估计误差减少80%的效果，相较于现有的InEKF。在单轴旋转地面上的行走实验中，初始误差高达1米时，平均估计误差小于9厘米。
View on arXiv Download PDF AI Translation
cs.RO / 8 / 2606.19525

A Categorial and Sheaf-Theoretic Semantics for Autonomic Component Ensembles

自主组件集的范畴与层叠理论语义
Hernández, Manuel, Sánchez-Soto, Eduardo
Abstract
The proliferation of large-scale, decentralized systems of autonomous agents, such as swarms of robots and networked cyber-physical systems, presents a formidable challenge to traditional formal methods. The Software Component Ensemble Language (SCEL) offers a formal model for such systems, but its operational semantics is not ideal for reasoning about global, structural, and emergent properties. This report proposes a new, multi-layered mathematical model for SCEL using category theory and sheaf theory. We argue that a society of robots described in SCEL can be formally modeled as a sheaf on a topological space, where components are points, ensembles are open sets, and distributed knowledge forms the sheaf's data. In this framework, computational processes like information sharing become equivalent to the sheaf-theoretic operation of "gluing" local data. System failures can then be understood and quantified as topological obstructions, measurable by sheaf cohomology. This approach transforms the verification of a complex distributed system into the analysis of the geometry of a mathematical object, providing deep, structural insights for the design of robust autonomic systems.
Chinese Translation
大规模去中心化自主智能体系统的迅速发展，例如机器人群和网络化的网络物理系统，给传统形式方法带来了巨大的挑战。软件组件集语言（Software Component Ensemble Language, SCEL）为此类系统提供了一个形式模型，但其操作语义并不理想，无法有效推理全局、结构性和涌现属性。本报告提出了一种基于范畴理论和层叠理论的SCEL新型多层次数学模型。我们认为，在SCEL中描述的机器人社会可以被形式化为一个拓扑空间上的层叠，其中组件是点，集成是开集，分布式知识构成层叠的数据。在这一框架下，信息共享等计算过程等同于层叠理论中“粘合”局部数据的操作。系统故障可以被理解和量化为拓扑障碍，可通过层叠上同调进行测量。这一方法将复杂分布式系统的验证转化为对数学对象几何的分析，为设计稳健的自主系统提供了深刻的结构性洞见。
View on arXiv Download PDF AI Translation
cs.RO / 9 / 2606.19555

SCAN-Planner: Spatial Collision-Aware Local Planning for Route-Guided Long-Range Quadruped Navigation

SCAN-Planner：一种空间碰撞感知的局部规划方法用于路线引导的长距离四足导航
Zheng, Han, Chen, Zhe, Fu, Yiwen, Yang, Ming, Qin, Tong
Abstract
Quadruped robots are increasingly expected to navigate through narrow passages, cluttered indoor scenes, and large-scale 3D unstructured environments. Existing local planners commonly approximate the robot using isotropic geometric inflation or rely on planar and elevation-map representations, leading to conservative motion in tight spaces and limited reasoning about overhanging structures. This letter presents SCAN-Planner, a spatial collision-aware local planning framework for long-range quadruped navigation. A yaw-aware twin-cylinder footprint is used to model the elongated robot body, enabling whole-body collision evaluation through sparse queries in an inflated 3D occupancy map. We further introduce a projected A* search that generates collision-free guidance on an interpolated ground-following surface, with z-gradient suppression to avoid obstacles horizontally while maintaining vertical stability. For large-scale deployment, a robot-centric sliding map with boundary fallback provides high-resolution local collision checking and recovery from local dead ends. Simulation and real-world experiments demonstrate that SCAN-Planner generates safe, smooth, and efficient trajectories in dense clutter, 3D unstructured scenes, stair traversal, and long-range navigation tasks.
Chinese Translation
四足机器人越来越被期望能够在狭窄通道、杂乱的室内场景和大规模的三维非结构化环境中导航。现有的局部规划器通常使用各向同性几何膨胀来近似机器人，或依赖于平面和高程图表示，这导致在狭小空间中的运动过于保守，并且对悬挑结构的推理有限。本文提出了SCAN-Planner，一种用于长距离四足导航的空间碰撞感知局部规划框架。我们使用一种考虑偏航的双气缸足迹来建模延长的机器人主体，通过在膨胀的三维占用图中进行稀疏查询，实现全身碰撞评估。我们进一步引入了一种投影A*搜索算法，该算法在插值的地面跟随表面上生成无碰撞的导航指导，并通过z梯度抑制来水平避障，同时保持垂直稳定性。为了大规模部署，采用以机器人为中心的滑动地图和边界回退机制，提供高分辨率的局部碰撞检测和从局部死胡同的恢复。仿真和实际实验表明，SCAN-Planner能够在密集杂乱、三维非结构化场景、楼梯穿越和长距离导航任务中生成安全、平滑和高效的轨迹。
View on arXiv Download PDF AI Translation
cs.RO / 10 / 2606.19561

pdSTL: Probabilistic Differentiable Signal Temporal Logic for Stochastic Systems

pdSTL：用于随机系统的概率可微信号时序逻辑
Dogbey, Bennett, Manjunatha, Hemanth
Abstract
Autonomous robots operating in uncertain environments must satisfy complex temporal and safety specifications despite stochastic dynamics and sensing noise. While Signal Temporal Logic (STL) offers robustness measures for gradient-based optimization, existing extensions either lack differentiability or ignore belief-space uncertainty. We introduce pdSTL (probabilistic differentiable Signal Temporal Logic), a framework that unifies probabilistic semantics with differentiable robustness over belief trajectories. pdSTL employs interval-valued probabilistic semantics to compute conservative satisfaction bounds, propagated compositionally through the STL syntax tree. We formulate the temporal robustness evaluation as a recurrent, LSTM-style unfolding of STL operators, enabling linear-time, differentiable monitoring suitable for end-to-end trajectory optimization. We validate pdSTL on simulated obstacle avoidance, lane-change maneuvers, and real-world Crazyflie quadcopter flight experiments under aerodynamic disturbances. Results demonstrate that pdSTL achieves efficient optimization with formal probabilistic guarantees, significantly outperforming deterministic differentiable STL in maintaining safety margins under real-world uncertainty.
Chinese Translation
在不确定环境中运行的自主机器人必须满足复杂的时序和安全规范，尽管存在随机动态和传感噪声。尽管信号时序逻辑（STL）为基于梯度的优化提供了鲁棒性度量，但现有的扩展要么缺乏可微性，要么忽略了信念空间的不确定性。我们引入了pdSTL（概率可微信号时序逻辑），这是一个将概率语义与信念轨迹上的可微鲁棒性统一的框架。pdSTL采用区间值概率语义来计算保守的满足边界，并通过STL语法树进行组合传播。我们将时序鲁棒性评估形式化为STL运算符的递归LSTM风格展开，从而实现适合端到端轨迹优化的线性时间可微监控。我们在模拟的障碍物规避、变道操作以及在气动干扰下的真实Crazyflie四旋翼飞行实验中验证了pdSTL。结果表明，pdSTL在保持安全边际方面显著优于确定性可微STL，并实现了具有正式概率保证的高效优化。
View on arXiv Download PDF AI Translation
cs.RO / 11 / 2606.19586

One Demo is Worth a Thousand Trajectories: Action-View Augmentation for Visuomotor Policies

一演示胜千条轨迹：用于视觉运动策略的动作视图增强
Pan, Chuer, Liang, Litian, Bauer, Dominik, Cousineau, Eric, Burchfiel, Benjamin, Feng, Siyuan, Song, Shuran
Abstract
Visuomotor policies for manipulation have demonstrated remarkable potential in modeling complex robotic behaviors, yet minor alterations in the robot's initial configuration and unseen obstacles easily lead to out-of-distribution observations. Without extensive data collection effort, these result in catastrophic execution failures. In this work, we introduce an effective data augmentation framework that generates visually realistic fisheye image sequences and corresponding physically feasible action trajectories from real-world eye-in-hand demonstrations, captured with a portable parallel gripper with a single fisheye camera. We introduce a novel Gaussian Splatting formulation, adapted to wide FoV fisheye cameras, to reconstruct and edit the 3D scene with unseen objects. We utilize trajectory optimization to generate smooth, collision-free, view-rendering-friendly action trajectories and render visual observations from corresponding novel views. Comprehensive experiments in simulation and the real world show that our augmentation framework improves the success rate for various manipulation tasks in both the same scene and the augmented scene with obstacles requiring collision avoidance.
Chinese Translation
视觉运动策略在建模复杂机器人行为方面展现了显著的潜力，但机器人初始配置的微小变化和未见障碍物容易导致分布外观察结果。在没有大量数据收集的情况下，这会导致灾难性的执行失败。在本研究中，我们提出了一种有效的数据增强框架，该框架从真实世界的眼手演示中生成视觉上逼真的鱼眼图像序列及相应的物理可行动作轨迹，这些演示是通过携带式平行夹具和单个鱼眼相机捕获的。我们引入了一种新颖的高斯溅射（Gaussian Splatting）公式，适用于广视场（FoV）鱼眼相机，以重建和编辑包含未见物体的三维场景。我们利用轨迹优化生成平滑、无碰撞且适合视图渲染的动作轨迹，并从相应的新视角渲染视觉观察。综合的仿真实验和现实世界实验表明，我们的增强框架提高了在同一场景和需要避免碰撞的增强场景中各种操作任务的成功率。
View on arXiv Download PDF AI Translation
cs.RO / 12 / 2606.19590

Safe, Real-Time Active Model Discrimination and Fault Diagnosis for Nonlinear Systems via Differentiable Reachability

基于可微达性的不确定非线性系统的安全实时主动模型区分与故障诊断
Ni, Xinpei, Ornik, Melkior, Chou, Glen, Coogan, Samuel
Abstract
We present a safe, real-time algorithm for active fault diagnosis and model discrimination for uncertain continuous-time nonlinear systems with process and measurement disturbances. Given a finite set of candidate models representing nominal and faulty modes, including actuator and sensor faults, we formulate an output-feedback, time-varying policy optimization problem that (i) robustly enforces state-input safety constraints over a finite horizon and (ii) drives the system to produce sampled measurements consistent with at most one model, enabling deterministic diagnosis. To solve this problem in real time, we develop a tractable approximation using interval over-approximations of reachable state and output sets, and encode diagnosability via a differentiable objective that penalizes overlap between the reachable output sets of possible models. The resulting optimization is solved efficiently online with gradient-based methods using JAX and differentiable reachability primitives. We evaluate our method on sensor and actuator fault diagnosis (up to 11 fault modes) in several high-dimensional nonlinear robotic systems, including a simulated quadrotor and fighter-jet model, a hardware differential-drive robot, and quadrupedal navigation. Across these case studies, our approach achieves reliable model discrimination in under 50 ms, outperforming baselines in discrimination success rate and speed while providing formal safety guarantees.
Chinese Translation
我们提出了一种安全的实时算法，用于不确定连续时间非线性系统的主动故障诊断和模型区分，该系统受到过程和测量干扰的影响。给定一组有限的候选模型，代表正常和故障模式，包括执行器和传感器故障，我们构建了一个输出反馈、时变策略优化问题，该问题 (i) 在有限时间内稳健地强制执行状态-输入安全约束，并 (ii) 驱动系统产生与最多一个模型一致的采样测量，从而实现确定性诊断。为了解决这个实时问题，我们开发了一种可处理的近似方法，使用可达状态和输出集的区间过近似，并通过一个可微目标编码可诊断性，该目标惩罚可能模型的可达输出集之间的重叠。最终的优化问题通过基于梯度的方法在在线高效求解，使用 JAX 和可微达性原语。我们在多个高维非线性机器人系统上评估了我们的方法，包括模拟四旋翼和战斗机模型、硬件差速驱动机器人和四足导航，针对传感器和执行器故障诊断（最多 11 种故障模式）。在这些案例研究中，我们的方法在 50 毫秒内实现了可靠的模型区分，优于基线在区分成功率和速度方面，同时提供了正式的安全保证。
View on arXiv Download PDF AI Translation
cs.RO / 13 / 2606.19598

Fail-RAG : A Retrieval Augmented Generation Informed Framework for Robot Failure Identification

Fail-RAG：一种基于检索增强生成的机器人故障识别框架
Salvi, Ameya, Hu, Jie
Abstract
Industry automation is witnessing an evolution in robotics driven by both technological breakthroughs and societal changes: progress towards generalist robots, embodied and physical artificial intelligence (AI), and increasing labor shortage in manufacturing.An intelligent autonomous robot needs to not only act according to planned motions but also react to any unexpected events. In this study, we focus on such unexpected events in warehouses where robots are used for material handling. Specifically, we refer to any unexpected events as failures and develop methods to detect robot operations related failures. Rule-based detection methods may break since the form of failures could change due to the dynamic nature of both environments and tasks. We propose 'Fail-RAG', a Retrieval Augmented Generation (RAG)-based failure detection framework where failure images and context information are embedded and queried against a failure database by calculating their similarities. Vision-Language Models (VLMs) are further used to analyze failures and provide details by following our instruction template. We evaluated the performance of Fail-RAG by conducting both simulation and physical experiments using fixed robot arms and a mobile manipulator for multiple tasks that are common in warehouse automation. Fail-RAG achieved 25 percentage point higher failure detection accuracy on average across five types of robot operations compared to using off-the-shelf VLMs, indicating its effectiveness for real-world failure detection.
Chinese Translation
工业自动化正在经历由技术突破和社会变革驱动的机器人演变：朝着通用机器人、具身和物理人工智能（AI）以及制造业劳动力短缺的进展。智能自主机器人不仅需要按照计划的动作进行操作，还需要对任何意外事件作出反应。在本研究中，我们关注仓库中使用机器人进行物料搬运时所发生的意外事件。具体而言，我们将任何意外事件称为故障，并开发检测与机器人操作相关的故障的方法。基于规则的检测方法可能会失效，因为故障的形式可能因环境和任务的动态特性而变化。我们提出了“Fail-RAG”，一种基于检索增强生成（RAG）的故障检测框架，其中故障图像和上下文信息被嵌入并通过计算其相似性与故障数据库进行查询。视觉-语言模型（VLMs）进一步用于分析故障，并根据我们的指令模板提供详细信息。我们通过使用固定机器人手臂和移动操纵器进行多项仓库自动化常见任务的模拟和实物实验来评估Fail-RAG的性能。与使用现成的VLMs相比，Fail-RAG在五种类型的机器人操作中平均实现了高出25个百分点的故障检测准确率，表明其在现实世界故障检测中的有效性。
View on arXiv Download PDF AI Translation
cs.RO / 14 / 2606.19632

Formal Verification of Learned Multi-Agent Communication Policies via Decision Tree Distillation

通过决策树蒸馏对学习的多智能体通信策略进行形式验证
Farooq, Ahmad, Iqbal, Kamran
Abstract
Multi-agent reinforcement learning (MARL) enables agents to develop coordination strategies through emergent communication, but neural policies lack the formal safety guarantees required for safety-critical robotic deployment in drone swarms and autonomous vehicle fleets. We present the first end-to-end framework for safety verification of learned multi-agent communication policies through policy abstraction: neural policies are distilled into interpretable decision trees, then formally verified, with empirical validation confirming that verified safety properties transfer to original networks. Our four-stage pipeline consists of domain-specific feature extraction from agent observations, decision tree distillation achieving 97.9% +/- 1.2% fidelity to neural policies, automated translation to PRISM probabilistic model checker specifications with complete feature-to-state-variable correspondence, and compositional verification of Probabilistic Computation Tree Logic (PCTL) properties via pairwise decomposition with union-bound aggregation and empirical neighbor modeling. Evaluating Vector-Quantized Variational Information Bottleneck (VQ-VIB) policies for multi-drone coordination with 5-7 agents, we verify 18 temporal logic properties across safety, liveness, and cooperation, achieving 88.9% property satisfaction with all five safety thresholds satisfied (0.3% collision probability vs. 1% threshold). Monte Carlo validation of original neural policies confirms that verified safety properties transfer with <=0.6 percentage-point deviation (95% CI). Discrete VQ-VIB messages provide +11.6 to +13.6 percentage-point fidelity advantages over continuous methods, enabling 3-4x faster verification. Our framework provides empirically validated safety verification for distilled policy abstractions, serving as a practical bridge between deep MARL and formal safety workflows for multi-robot deployment.
Chinese Translation
多智能体强化学习（MARL）使智能体能够通过自发通信发展协调策略，但神经网络策略缺乏在无人机群和自主车辆车队中进行安全关键部署所需的正式安全保证。我们提出了第一个端到端框架，通过策略抽象对学习的多智能体通信策略进行安全验证：将神经网络策略蒸馏为可解释的决策树，然后进行形式验证，实证验证确认验证的安全属性可以转移到原始网络。我们的四阶段流程包括从智能体观察中提取特定领域的特征，决策树蒸馏实现对神经网络策略的97.9% +/- 1.2%的保真度，自动翻译为PRISM概率模型检查器规范，确保特征与状态变量之间的完全对应，以及通过成对分解、联合界聚合和实证邻居建模对概率计算树逻辑（PCTL）属性进行组合验证。在评估5-7个智能体的多无人机协调的向量量化变分信息瓶颈（VQ-VIB）策略时，我们验证了18个涉及安全性、生存性和合作的时序逻辑属性，达到了88.9%的属性满足率，所有五个安全阈值均满足（0.3%的碰撞概率对比1%的阈值）。对原始神经网络策略的蒙特卡洛验证确认验证的安全属性转移的偏差不超过0.6个百分点（95%置信区间）。离散的VQ-VIB消息在保真度上比连续方法提供了11.6到13.6个百分点的优势，使验证速度提高了3-4倍。我们的框架为蒸馏的策略抽象提供了经过实证验证的安全验证，成为深度MARL与多机器人部署的正式安全工作流程之间的实用桥梁。
View on arXiv Download PDF AI Translation
cs.RO / 15 / 2606.19633

CTS-MoE: Implicit Terrain Adaptation via Mixture-of-Experts for Perceptive Locomotion

CTS-MoE：通过专家混合实现的隐式地形适应以支持感知性步态
Affonso, Francisco, Angarola, Matheus P., Mineiro, Ana Luiza, Potnis, Aditya, Becker, Marcelo, Chowdhary, Girish
Abstract
Perceptive legged locomotion over discontinuous terrain (e.g., stairs, gaps, and obstacles) requires adaptive behavior, as a single conservative gait cannot produce the anticipatory maneuvers needed for abrupt topology changes. Cast as multi-task reinforcement learning, this problem introduces a tension between sharing and separation. Tasks use a common locomotion base but have conflicting rewards, so a policy must share behavior while avoiding value interference. Prior work addresses only one side, with monolithic policies sacrificing specialization and hierarchical sub-policies sacrificing generalization across transitions and unseen terrain. We propose CTS-MoE, which combines a dense mixture-of-experts actor with perception-based gating to compose shared behaviors and a multi-critic with task-specific value heads to prevent interference. The model is trained end-to-end in a single-stage concurrent teacher-student setup that handles partial observability and avoids sequential distillation, with task labels used only during training. At deployment, routing depends solely on perception, allowing terrain adaptation without a high-level selector or terrain classifier. Experiments on a Unitree Go1 in simulation and on hardware across seen and unseen terrains show task-aware specialization, with lower tracking error and higher success rates than monolithic baselines. Project Website: https://cts-moe.github.io/ .
Chinese Translation
在不连续地形（例如楼梯、间隙和障碍物）上进行感知性腿部运动需要适应性行为，因为单一的保守步态无法产生应对突发地形变化所需的预期机动。将此问题视为多任务强化学习，带来了共享与分离之间的紧张关系。任务使用共同的运动基础，但奖励存在冲突，因此策略必须在避免价值干扰的同时共享行为。之前的研究仅解决了一方面，单一政策牺牲了专业化，而层次子政策则牺牲了在过渡和未见地形上的泛化能力。我们提出了CTS-MoE，它结合了密集的专家混合体（mixture-of-experts）演员与基于感知的门控机制，以组合共享行为，并使用具有任务特定价值头的多重评论家（multi-critic）来防止干扰。该模型在一个单阶段的并行教师-学生设置中进行端到端训练，处理部分可观测性并避免顺序蒸馏，任务标签仅在训练期间使用。在部署时，路由完全依赖于感知，从而实现地形适应，而无需高层选择器或地形分类器。在模拟和硬件上对Unitree Go1进行的实验显示了任务感知的专业化，跟踪误差更低，成功率高于单一基线。项目网站：https://cts-moe.github.io/ .
View on arXiv Download PDF AI Translation
cs.RO / 16 / 2606.19641

Scaling Self-Play for End-to-End Driving

端到端驾驶的自我对抗训练规模化
Rowe, Luke, Girgis, Roger, de Schaetzen, Rodrigue, Cornelisse, Daphne, Grandhi, Alaap, Heide, Felix, Vinitsky, Eugene, Pal, Christopher, Paull, Liam
Abstract
End-to-end autonomous driving models are typically trained on offline human-demonstration datasets that provide limited state coverage and often no closed-loop feedback, making them prone to compounding errors when deployed in closed-loop and brittle to long-tail agent interactions. To overcome these limitations, we propose an alternative strategy for training end-to-end driving models: large-scale self-play directly from pixels in simulation. While prior self-play approaches have shown promising transfer to real-world driving, they typically assume vectorized Bird's-Eye-View (BEV) observations that are incompatible with end-to-end policies operating directly on sensor observations. To this end, we introduce Gigapixel, a high-throughput batched driving simulator with perspective rendering, enabling scalable self-play directly from pixel observations. Rather than targeting compute-costly photorealistic sensor simulation, Gigapixel renders a simplified bounding-box world that preserves essential scene structure while achieving throughput at 50k agent steps per second. Since direct pixel-space self-play RL is prohibitively sample-inefficient at end-to-end model scale, we propose self-play DAgger training: we train pixel-based policies in self-play via on-policy distillation from a privileged RL teacher. To bridge the sim-to-real gap, we subsequently transfer the self-play trained policies to real-world sensor data through lightweight perception adaptation. Policies trained in Gigapixel and adapted to real-world sensor data achieve competitive performance on the HUGSIM and NAVSIM-v2 benchmarks without human trajectory supervision. Moreover, scaling self-play training yields proportional gains in policy performance, establishing self-play as a practical and scalable strategy for training end-to-end models.
Chinese Translation
端到端自主驾驶模型通常在离线人类示范数据集上进行训练，这些数据集提供的状态覆盖有限且通常缺乏闭环反馈，使得模型在闭环部署时容易出现累积错误，并且对长尾代理交互表现脆弱。为克服这些局限性，我们提出了一种替代策略来训练端到端驾驶模型：直接从像素中进行大规模自我对抗训练。在以往的自我对抗方法中，尽管已经显示出向现实世界驾驶的良好迁移，但它们通常假设使用与端到端策略直接在传感器观测上操作不兼容的矢量化鸟瞰图（Bird's-Eye-View, BEV）观测。为此，我们引入了Gigapixel，这是一种具有透视渲染的高吞吐量批量驾驶模拟器，能够直接从像素观测中实现可扩展的自我对抗训练。Gigapixel并不针对计算成本高昂的照片级真实传感器模拟，而是渲染一个简化的边界框世界，保留了基本的场景结构，同时以每秒50,000个代理步的吞吐量实现高效。由于在端到端模型规模下，直接在像素空间进行自我对抗强化学习（RL）在样本效率上是不可行的，我们提出了自我对抗DAgger训练：通过来自特权RL教师的在线蒸馏，在自我对抗中训练基于像素的策略。为了弥合模拟与现实之间的差距，我们随后通过轻量级感知适应将自我对抗训练的策略迁移到现实世界的传感器数据中。在Gigapixel中训练并适应于现实世界传感器数据的策略在HUGSIM和NAVSIM-v2基准测试中表现出竞争力，而无需人类轨迹监督。此外，规模化自我对抗训练带来了策略性能的成比例提升，确立了自我对抗作为训练端到端模型的实用且可扩展的策略。
View on arXiv Download PDF AI Translation
cs.RO / 17 / 2606.19656

DF-ExpEnse: Diffusion Filtered Exploration for Sample Efficient Finetuning

DF-ExpEnse：用于样本高效微调的扩散过滤探索
Luo, Calvin, Sun, Chen, Song, Shuran
Abstract
A natural recipe for intelligent robotic decision-making is initializing from pretrained generative control policies, which have summarized offline experience, and adapting them to self-collected online experience. We present DF-ExpEnse, an exploration technique that improves the quality of online experience collection, thus increasing finetuning sample-efficiency. DF-ExpEnse leverages the multimodal modeling capabilities of the generative control policy to create an expressive and tractably evaluatable candidate set. It then utilizes an ensemble of critics to identify the action that best balances quality with high exploration interest. In fleet settings, DF-ExpEnse further enables cross-agent communication to facilitate collaborative exploration as a group. DF-ExpEnse can be seamlessly integrated with existing strategies that finetune pretrained generative control policies via reinforcement learning. We experimentally validate consistent sample-efficiency benefits through DF-ExpEnse across a variety of manipulation and locomotion tasks, compared to default finetuning and alternative action selection schemes. Project can be found at https://df-expense.github.io.
Chinese Translation
智能机器人决策的自然方案是从预训练的生成控制策略初始化，这些策略总结了离线经验，并将其适应于自收集的在线经验。我们提出了DF-ExpEnse，这是一种探索技术，旨在提高在线经验收集的质量，从而增加微调的样本效率。DF-ExpEnse利用生成控制策略的多模态建模能力，创建一个表达丰富且可有效评估的候选集。然后，它利用一组评论者来识别最佳平衡质量与高探索兴趣的动作。在车队设置中，DF-ExpEnse进一步支持跨代理通信，以促进作为一个群体的协作探索。DF-ExpEnse可以与现有策略无缝集成，这些策略通过强化学习微调预训练的生成控制策略。我们通过实验验证了DF-ExpEnse在各种操作和运动任务中相较于默认微调和替代动作选择方案的一致样本效率优势。项目网址为 https://df-expense.github.io。
View on arXiv Download PDF AI Translation
cs.RO / 18 / 2606.19672

Safe Local Navigation for Ackermann-Steered Robots in Unmapped Environments

在未映射环境中为阿克曼转向机器人提供安全的局部导航
Schaible, Christian, Sirouspour, Shahin
Abstract
A control framework is proposed for safe local navigation of mobile robots equipped with Ackermann steering in unmapped environments where a global goal is absent. Based on local obstacle detections, the safest heading angle is determined along the direction of the largest open space ahead of the vehicle. Guided by this direction, bounding lines are constructed on the left and right sides of the vehicle to achieve obstacle separation. These bounding lines are obtained by solving a convex quadratic optimization that maximizes vehicle-to-obstacle clearance. Optionally, conditions are imposed on the bounding lines to preserve parallelism and smooth abrupt changes from prior control steps. A feedback-linearizing controller is then used to regulate the vehicle's distance from one or both bounding lines, effectively enabling tracking of a local reference path that preserves safety through obstacle clearance maximization. Open-source code is included for the application of this control scheme. Experimental results demonstrate that the proposed method produces safer navigation paths with significantly shorter computation times, compared to some existing exploration-based planners.
Chinese Translation
提出了一种控制框架，用于在缺乏全局目标的未映射环境中，为配备阿克曼转向的移动机器人提供安全的局部导航。基于局部障碍物检测，确定沿着车辆前方最大开放空间方向的最安全航向角。在这一方向的引导下，在车辆的左侧和右侧构建边界线，以实现障碍物的分离。这些边界线通过求解一个凸二次优化问题获得，该问题最大化车辆与障碍物之间的间隙。可选地，对边界线施加条件，以保持平行性并平滑先前控制步骤的突变。然后，使用反馈线性化控制器来调节车辆与一个或两个边界线之间的距离，从而有效地实现对局部参考路径的跟踪，并通过最大化障碍物间隙来保持安全。本文还包括该控制方案的开源代码。实验结果表明，与一些现有的基于探索的规划器相比，所提出的方法能够生成更安全的导航路径，并显著缩短计算时间。
View on arXiv Download PDF AI Translation
cs.RO / 19 / 2606.19675

ForEnt: A Multi-Modal Dataset for Characterizing Quadruped Robot Entrapments in Forest Environments

ForEnt：用于表征四足机器人在森林环境中被困情况的多模态数据集
Kirdwichai, Natapat, Tarapore, Danesh
Abstract
Legged robots are increasingly deployed in forests for ecological surveying and monitoring, yet their autonomy is often interrupted consequent to the challenges posed in traversing forest environments. Forest entrapments, for example, when a robot's legs are ensnared in vines or other vegetation, result in loss of stability and toppling. Such events not only disrupt the mission and require manual intervention, but also risk damage to the robot hardware. To address the absence of a dedicated dataset to investigate these failure modes in forest environments, we present ForEnt, a multi-modal dataset collected with the low-cost Unitree Go2 quadruped across eight forest sites in the Southampton Common Woodlands, UK. For our dataset, over approximately 1.7 km of traversals in 11 sequences were conducted, yielding 69 recorded entrapment events. ForEnt includes time-synchronized RGB-D images, LiDAR scans, proprioceptive data, and third-person video, enabling analysis of terrain factors contributing to entrapment and providing labeled sensor streams for reproducible benchmarking. By supporting the evaluation of entrapment detection strategies, ForEnt lowers the barrier to developing robust quadruped robot deployments in challenging forest environments.
Chinese Translation
四足机器人在森林中进行生态调查和监测的应用日益增多，但它们的自主性常常因穿越森林环境所面临的挑战而受到干扰。例如，当机器人的腿被藤蔓或其他植被缠住时，会导致稳定性丧失和倾覆。这类事件不仅会干扰任务并需要人工干预，还可能对机器人硬件造成损害。为了解决缺乏专门数据集以研究森林环境中这些故障模式的问题，我们提出了ForEnt，这是一个在英国南安普敦普通林地的八个森林地点使用低成本的Unitree Go2四足机器人收集的多模态数据集。我们的数据集涵盖了约1.7公里的11个序列的穿越，记录了69个被困事件。ForEnt包括时间同步的RGB-D图像、LiDAR扫描、运动感知数据和第三人称视频，能够分析导致被困的地形因素，并提供标记的传感器数据流以便于可重复的基准测试。通过支持被困检测策略的评估，ForEnt降低了在具有挑战性的森林环境中开发稳健的四足机器人部署的门槛。
View on arXiv Download PDF AI Translation
cs.RO / 20 / 2606.19687

Route-Constrained Robust Fusion Estimation for MEMS/GNSS Integrated Navigation of Unmanned Ground Vehicles in GNSS Degraded Environments

在GNSS信号衰减环境下，无人地面车辆的MEMS/GNSS集成导航的路线约束鲁棒融合估计
Cui, Jingzhi, Zhang, Chao, Mao, Yuliang, Lü, Shaolin, Li, Dongmei, Che, Huan, Zhang, Rong
Abstract
To address cumulative localization drift of unmanned ground vehicles in structured road environments under severe Global Navigation Satellite System signal occlusion, this paper proposes a robust route-constrained state estimation method. During periods without satellite signals, the proposed method establishes the correspondence between the historical dead reckoning trajectory and local segments of the mission route extracted from a high-definition map, and estimates a route-referenced position via a two-dimensional rigid transformation. The estimated position is then formulated as a pseudo-position observation and incorporated into an Extended Kalman Filter update. In this way, route constraints at the road level can be continuously injected into a unified state estimation framework, thereby suppressing position deviation relative to the mission route while indirectly improving azimuth estimation. To enhance practical applicability, engineering strategies, such as trigger control, matching quality validation, route offset compensation, and single update correction limiting, are further introduced. Experiments in three representative scenarios, including a long tunnel, a multi-segment tunnel, and a curved tunnel, show that the proposed method effectively suppresses error accumulation during satellite outages, reduces the risk of large maximum deviation, and improves localization continuity and road-level usability.
Chinese Translation
为了解决在严重的全球导航卫星系统（GNSS）信号遮挡下，无人地面车辆在结构化道路环境中的累积定位漂移问题，本文提出了一种鲁棒的路线约束状态估计方法。在没有卫星信号的期间，该方法建立了历史航迹与从高清地图中提取的任务路线局部段之间的对应关系，并通过二维刚性变换估计出参考路线的位置。然后，将估计的位置构建为伪位置观测，并纳入扩展卡尔曼滤波器（Extended Kalman Filter）更新中。通过这种方式，路面级别的路线约束可以持续注入到统一的状态估计框架中，从而抑制相对于任务路线的位置偏差，同时间接改善方位角估计。为了增强实际应用性，进一步引入了触发控制、匹配质量验证、路线偏移补偿和单次更新修正限制等工程策略。在长隧道、多段隧道和曲线隧道等三个典型场景中的实验表明，所提出的方法有效抑制了卫星信号中断期间的误差积累，降低了大最大偏差的风险，并改善了定位的连续性和道路级别的可用性。
View on arXiv Download PDF AI Translation
cs.RO / 21 / 2606.19699

Comparative Study on Agility, Efficiency, and Impact Absorption of Bipedal Robots with Active Toes

双足机器人主动脚趾的灵活性、效率和冲击吸收能力的比较研究
Kim, Joong-Gil, Ye, Wontae, Cho, Geunwoo, Yun, Seong-Ho, Cho, Se-Hyoung, Kim, Yong-Jae
Abstract
Human legs exhibit high efficiency, agility, and impact absorption, with toes playing a crucial role in these capabilities. While many attempts have been made to implement human-like toes in robots, they have not fully replicated human characteristics nor rigorously validated their benefits. We propose a 14-DOF biped robot emulating human toes' lightweight, high-torque, robust nature. To quantitatively analyze the effectiveness of the active toes in terms of agility, efficiency, and impact absorption, we developed a high-fidelity simulation training environment that reflects actual actuators with coupled transmissions and accurate power consumption. To ensure a fair comparison between configurations with and without active toes, we designed a minimal RL reward function and applied an identical training procedure to both. The simulation results indicate that, at 1.33 m/s walking, the toe-equipped robot reduced CoT by 17.5% and heel-strike GRF by 5.0% compared with the toe-ablation configuration. On the agility test, average and maximum path deviation decreased by 25.0% and 34.0%, respectively.
Chinese Translation
人类的腿部展现出高效性、灵活性和冲击吸收能力，其中脚趾在这些能力中发挥着至关重要的作用。尽管许多尝试已经在机器人中实现类人脚趾，但它们并未完全复制人类的特征，也未严格验证其优势。我们提出了一种14自由度（DOF）的双足机器人，模拟人类脚趾的轻量、高扭矩和强健特性。为了定量分析主动脚趾在灵活性、效率和冲击吸收方面的有效性，我们开发了一个高保真度的仿真训练环境，反映了实际执行器的耦合传动和准确的功耗。为了确保有主动脚趾和无主动脚趾配置之间的公平比较，我们设计了一个最小化的强化学习（RL）奖励函数，并对两者应用相同的训练程序。仿真结果表明，在1.33米/秒的行走速度下，配备脚趾的机器人相比于去除脚趾的配置，能将单位行走成本（CoT）降低17.5%，将脚跟着地的地面反作用力（GRF）降低5.0%。在灵活性测试中，平均路径偏差和最大路径偏差分别减少了25.0%和34.0%。
View on arXiv Download PDF AI Translation
cs.RO / 22 / 2606.19711

A Differentiable Composite Approximation Framework for Autonomous Underwater Vehicle Maneuvering Modeling from Sea-Trial Data

基于海试数据的自主水下航行器机动建模的可微复合近似框架
Wang, Aobo, Xia, Aifei, Wang, Zihao, Hao, Lizhu
Abstract
Field-based modeling from onboard measurements can produce autonomous underwater vehicle (AUV) maneuvering models that reflect real operating characteristics. From an approximation perspective, conventional maneuvering models use predefined constraint polynomial bases, whereas data-driven models use data-adaptive bases. Motivated by this basis-function view, this paper presents a differentiable composite-approximation formulation, in which the polynomial-basis component and the data-adaptive basis component are treated as differentiable parts of a single predictor and calibrated jointly. A gradient-based co-calibration method is developed for full-scale AUV maneuvering prediction, where a sensitivity-aware mechanism regulates bounded polynomial updates while the neural residual captures remaining nonlinear discrepancies under a shared prediction objective. To account for ocean-current effects in field data, a turning-motion-based current estimation and compensation procedure is incorporated to construct current-compensated learning targets for training and rollout. The framework is evaluated using sea-trial data collected from a 7-meter AUV under multiple maneuvering conditions. Results show that the proposed method improves recursive trajectory and velocity prediction compared with polynomial-only, neural-only, and frozen-prior hybrid baselines, demonstrating its applicability to field-data-based AUV maneuvering modeling.
Chinese Translation
基于现场测量的建模可以生成反映真实操作特性的自主水下航行器（AUV）机动模型。从近似的角度来看，传统的机动模型使用预定义的约束多项式基，而数据驱动模型则使用数据自适应基。受到这一基函数视角的启发，本文提出了一种可微复合近似的公式，其中多项式基成分和数据自适应基成分被视为单一预测器的可微部分，并共同进行校准。为全尺度AUV机动预测开发了一种基于梯度的共同校准方法，其中灵敏度感知机制调节有界多项式更新，而神经残差则在共享预测目标下捕捉剩余的非线性差异。为了考虑现场数据中的海流影响，本文结合了一种基于转向运动的海流估计和补偿程序，以构建用于训练和推广的海流补偿学习目标。该框架使用从7米AUV在多种机动条件下收集的海试数据进行了评估。结果表明，所提出的方法在递归轨迹和速度预测方面相较于仅使用多项式、仅使用神经网络和冻结先验混合基线的方法有显著改善，证明了其在基于现场数据的AUV机动建模中的适用性。
View on arXiv Download PDF AI Translation
cs.RO / 23 / 2606.19728

Bidirectional Tutoring for Developmental Motor Learning in Robots: Co-Developed Interaction Dynamics Support Stable Learning

双向辅导在机器人发展运动学习中的应用：共同发展的互动动态支持稳定学习
Fukushima, Rui, Tani, Jun
Abstract
Infants are well known to develop their motor skills through dense interaction with caregivers. Although such social interaction is crucial for human development, motor-skill learning in robots is often treated as a unidirectional process in which robots passively receive demonstrations from tutors. This overlooks a key property of social interaction: it is inherently bidirectional, with tutor and learner dynamically adapting to each other. In such interactions, the robot's past experiences may function as prior constraints that shape the dynamics of their co-developed trajectories. We hypothesize that bidirectional tutoring allows such constraints to guide the formation of consistent behavioral patterns that preserve behavioral coherence and support generalization, whereas unidirectional interaction lacks such constraints and leads to broader, less consistent behavioral patterns. To examine this hypothesis, we conducted two experiments with a physical humanoid robot performing an object manipulation task: one involving human-robot interaction and another employing an AI tutor interacting with the real robot through an adaptive intervention mechanism designed to examine whether similar effects would emerge under more controlled conditions. We implement the developmental learning framework using a free-energy-principle-based neural network extended with generative replay, which supports stable sequence-by-sequence learning from single tutored episodes. Across both settings, bidirectional tutoring fostered consistent behaviors and stage-wise generalization, while the robot gradually required less tutor guidance. These results suggest that bidirectional tutoring, as an embodied and socially grounded approach, provides an effective scaffold for developmental motor learning in robots.
Chinese Translation
婴儿通过与照顾者的密切互动来发展他们的运动技能是众所周知的。尽管这种社会互动对人类发展至关重要，但机器人运动技能学习通常被视为一个单向过程，其中机器人被动地接受来自辅导者的示范。这忽视了社会互动的一个关键特性：它本质上是双向的，辅导者和学习者相互动态适应。在这样的互动中，机器人的过去经验可能作为先前约束，塑造他们共同发展的轨迹动态。我们假设双向辅导允许这些约束指导一致行为模式的形成，从而保持行为的一致性并支持泛化，而单向互动则缺乏这样的约束，导致更广泛且不一致的行为模式。为了检验这一假设，我们进行了两项实验，使用一个物理类人机器人执行物体操控任务：一项涉及人机互动，另一项则采用AI辅导者通过适应性干预机制与真实机器人互动，以检验在更受控条件下是否会出现类似效果。我们使用基于自由能原理的神经网络扩展生成重放来实现发展学习框架，该框架支持从单一辅导情节中进行稳定的逐序学习。在这两种设置中，双向辅导促进了一致的行为和阶段性泛化，同时机器人逐渐减少了对辅导者的指导需求。这些结果表明，作为一种具身且社会基础的方法，双向辅导为机器人发展运动学习提供了有效的支架。
View on arXiv Download PDF AI Translation
cs.RO / 24 / 2606.19729

VOiLA: Vectorized Online Planning with Learned Diffusion Model for POMDP Agents

VOiLA：基于学习扩散模型的向量化在线规划用于POMDP智能体
Hoerger, Marcus, Joshi, Rishikesh, Shome, Rahul, Manchester, Ian, Kurniawati, Hanna
Abstract
Planning under uncertainty is an essential capability for autonomous robots. The Partially Observable Markov Decision Process (POMDP) provides a powerful framework for such a capability. Although POMDP-based planning has advanced significantly, its application to real-world problems is often limited by the difficulty of obtaining faithful POMDP models. We present Vectorized Online planning wIth Learned diffusion model for POMDP Agents (VOiLA), a framework that learns task-agnostic POMDP models for online planning under uncertainty. VOiLA learns transition and observation samplers using conditional diffusion models and learns observation-likelihood models for particle-based belief updates. To enable efficient online planning, the diffusion samplers are distilled into compact feedforward generators and integrated with Vectorized Online POMDP Planner (VOPP), an online POMDP planner designed to leverage GPU parallelization. Experimental results indicate the distillation strategy reduces sampling cost by up to nearly three orders of magnitude, making learned generative POMDP models practical for online planning. Evaluation of VOiLA on three benchmark problems indicate that VOiLA achieves equal or better performance than Recurrent Soft Actor Critic while using less than 10% training data, and generalizes much better to unseen environment configurations. Physical robot evaluation indicates VOiLA uses the models learned using only simulated data and generates a policy that successfully accomplish the task in 10 of 10 runs.
Chinese Translation
在不确定性下进行规划是自主机器人必备的能力。部分可观察马尔可夫决策过程（POMDP）为这种能力提供了强大的框架。尽管基于POMDP的规划已取得显著进展，但其在实际问题中的应用常常受到获取真实POMDP模型难度的限制。我们提出了基于学习扩散模型的向量化在线规划框架（VOiLA），该框架为不确定性下的在线规划学习任务无关的POMDP模型。VOiLA使用条件扩散模型学习转移和观察采样器，并为基于粒子的信念更新学习观察似然模型。为了实现高效的在线规划，扩散采样器被提炼为紧凑的前馈生成器，并与向量化在线POMDP规划器（VOPP）集成，后者是一个旨在利用GPU并行化的在线POMDP规划器。实验结果表明，提炼策略将采样成本降低了近三个数量级，使得学习的生成POMDP模型在在线规划中变得实用。对VOiLA在三个基准问题上的评估表明，VOiLA的性能与递归软演员评论者（Recurrent Soft Actor Critic）相当或更好，同时使用的训练数据不到10%，并且在未见环境配置上具有更好的泛化能力。物理机器人评估表明，VOiLA使用仅通过模拟数据学习的模型，并生成了在10次实验中成功完成任务的策略。
View on arXiv Download PDF AI Translation
cs.RO / 25 / 2606.19752

Temporal Self-Imitation Learning

时间自我模仿学习
Jia, Yinsen, Chen, Boyuan
Abstract
Long-horizon robot manipulation policies trained with reward shaping can still exploit dense rewards through inefficient interaction, while rare efficient behaviors may be forgotten during training. We argue that temporal efficiency itself provides a powerful and underutilized source of self-supervision for reinforcement learning. We introduce Temporal Self-Imitation Learning (TSIL), a reinforcement learning framework that mines temporally efficient successful trajectories generated during learning and converts them into reusable supervision for future policy improvement. TSIL progressively refines learning using configuration-conditioned adaptive temporal targets derived from fast successful trajectories, while preserving and replaying efficient behaviors through efficiency-weighted self-imitation learning. Across 15 distinct long-horizon manipulation tasks, TSIL consistently improves learning efficiency, task-completion efficiency, revisitation of fast successful behaviors, and robustness to unstable training conditions. More broadly, our results suggest that the temporal structure of successful behavior itself provides a scalable self-supervisory signal for reinforcement learning beyond manually engineered reward shaping alone.
Chinese Translation
通过奖励塑形训练的长时间机器人操控策略仍然可能通过低效的交互来利用密集奖励，而在训练过程中稀有的高效行为可能会被遗忘。我们认为，时间效率本身为强化学习提供了一种强大且未被充分利用的自我监督来源。我们提出了时间自我模仿学习（Temporal Self-Imitation Learning, TSIL），这是一个强化学习框架，旨在挖掘在学习过程中生成的时间高效成功轨迹，并将其转化为可重用的监督信号，以促进未来策略的改进。TSIL通过从快速成功轨迹中导出的配置条件自适应时间目标逐步精炼学习，同时通过效率加权的自我模仿学习来保留和重放高效行为。在15个不同的长时间操控任务中，TSIL始终提高了学习效率、任务完成效率、快速成功行为的重访率，以及对不稳定训练条件的鲁棒性。更广泛地说，我们的结果表明，成功行为的时间结构本身为强化学习提供了一种可扩展的自我监督信号，超越了单纯依赖手动设计的奖励塑形。
View on arXiv Download PDF AI Translation
cs.RO / 26 / 2606.19769

Data Standards for Humanoid Robotics: The Missing Infrastructure for Physical AI

类人机器人数据标准：物理人工智能缺失的基础设施
Liu, Shaoshan, Qin, Xiugong, Wu, Xuan, Xia, Xuan, Ding, Ning, Liu, Jialu, Tang, Jie
Abstract
The scalability of humanoid robots will depend not only on models and hardware, but also on whether physical experience can accumulate across robots, tasks, organizations, and time. Drawing on the authors' work in developing ISO/WD 26264-1, Humanoid robot datasets -- Part 1: General requirements, within ISO/TC 299/WG 16, this article argues that data standards are becoming foundational infrastructure for Physical AI. We develop three insights. First, humanoid robot data is embodied interaction data, not a collection of isolated digital samples; a useful dataset must preserve the relationship among robot body, action, task, scene, execution trace, and outcome. Second, its value depends on physical coherence: multimodal streams are reusable only when timing, coordinate frames, calibration, kinematics, units, and synchronization assumptions remain inspectable. Third, the main bottleneck is not only data scarcity, but non-cumulative data caused by high collection costs, data silos, and inconsistent evaluation. We argue that humanoid robot data standards address these bottlenecks by making embodied experience interpretable, shareable, traceable, and reusable. A general standard should provide horizontal infrastructure for lifecycle management, metadata, provenance, quality, versioning, and traceability, while capability-specific parts should define domain grammar for manipulation, locomotion, human-robot interaction, cognition, and future humanoid capabilities. As AI moves from screens into bodies, data standards must evolve from organizing digital information to structuring physical interaction.
Chinese Translation
类人机器人的可扩展性不仅取决于模型和硬件，还取决于物理经验是否能够在机器人、任务、组织和时间之间积累。本文基于作者在ISO/TC 299/WG 16中开发ISO/WD 26264-1《类人机器人数据集——第1部分：一般要求》的工作，论证了数据标准正成为物理人工智能的基础性基础设施。我们提出了三个见解。首先，类人机器人数据是具身交互数据，而不是孤立数字样本的集合；一个有用的数据集必须保留机器人身体、动作、任务、场景、执行轨迹和结果之间的关系。其次，其价值取决于物理一致性：多模态流只有在时间、坐标系、校准、运动学、单位和同步假设可检查时才能重复使用。第三，主要瓶颈不仅是数据稀缺，还有由于高采集成本、数据孤岛和评估不一致导致的非累积数据。我们认为，类人机器人数据标准通过使具身经验可解释、可共享、可追溯和可重用，来解决这些瓶颈。一般标准应提供生命周期管理、元数据、来源、质量、版本控制和可追溯性的横向基础设施，而特定能力部分应定义操控、运动、人机交互、认知和未来类人能力的领域语法。随着人工智能从屏幕走向身体，数据标准必须从组织数字信息演变为结构化物理交互。
View on arXiv Download PDF AI Translation
cs.RO / 27 / 2606.19774

Start Right, Arrive Right: Asynchronous Execution via Initial Noise Selection

正确开始，正确到达：通过初始噪声选择实现异步执行
Ho, Trong-Bao, Nguyen, Quang-Tan, Ha, Thien-Loc, Nguyen, Gia-Binh, Nguyen, Viet-Thanh, Dinh, Long, Vu, Minh N., Nguyen, Duy M. H., Le, An Thai, Vien, Ngo Anh
Abstract
Action chunking enables robot policies to produce temporally coherent behavior, but generating multi-step action sequences with flow-based policies incurs latency that is incompatible with real-time control. Under asynchronous execution, the robot continues executing the current chunk while the next one is generated, causing even minor delays to create inconsistencies at chunk boundaries. Existing methods address this problem by steering generation toward the already executed action prefix. We instead show that prefix consistency can be achieved by selecting an appropriate initial noise before generation begins, allowing the unmodified flow ODE to produce a coherent next chunk. This reframes asynchronous inference as a noise selection problem rather than a trajectory steering problem. We introduce \textbf{PAINT}, a training-free method that finds this noise via backward Euler inversion and constructs the final chunk through a repainting rule. In summary, \texttt{PAINT} requires no gradients, retraining, or policy modification; yet it improves execution consistency and task performance across \textit{12 simulated benchmarks} and \textit{6 real-world manipulation tasks} spanning single-arm, bimanual, and humanoid embodiments. Website: ~\href{https://paint-action-chunking.github.io}{\texttt{https://paint-action-chunking.github.io}}.
Chinese Translation
动作分块使机器人策略能够产生时间上连贯的行为，但使用基于流的策略生成多步骤动作序列会导致与实时控制不兼容的延迟。在异步执行下，机器人在生成下一个分块的同时继续执行当前分块，导致即使是微小的延迟也会在分块边界产生不一致。现有方法通过引导生成朝向已执行的动作前缀来解决这个问题。我们则展示了通过在生成开始之前选择适当的初始噪声，可以实现前缀一致性，从而允许未修改的流常微分方程（flow ODE）生成一个连贯的下一个分块。这将异步推理重新构架为一个噪声选择问题，而不是轨迹引导问题。我们引入了 extbf{PAINT}，一种无训练的方法，通过向后欧拉反演找到这种噪声，并通过重绘规则构建最终分块。总之， exttt{PAINT}不需要梯度、重训练或策略修改；但它在 extit{12个模拟基准}和 extit{6个现实世界操作任务}中提高了执行一致性和任务性能，涵盖了单臂、双手和类人形态。网站：~ exttt{https://paint-action-chunking.github.io}。
View on arXiv Download PDF AI Translation
cs.RO / 28 / 2606.19784

EquiVLA: A General Framework for Rotationally Equivariant Vision-Language-Action Models

EquiVLA：一种旋转等变视觉-语言-动作模型的通用框架
Ha, Thien-Loc, Nguyen, Quang-Tan, Ho, Trong-Bao, Dinh, Long, Nguyen, Minh Duc, Nguyen, Gia-Binh, Quang, Pham Tri, Vu, Minh N., Nguyen, Duy M. H., Le, An Thai, Vien, Ngo Anh
Abstract
Vision-Language-Action (VLA) models have emerged as a powerful paradigm for generalist robot manipulation, yet they lack geometric inductive biases: policies trained at specific orientations require substantially more data to generalize across rotational configurations. We present \textsc{EquiVLA}, the first general framework for end-to-end $\mathrm{SO}(2)$-equivariant VLA models, applicable to any architecture coupling a frozen vision-language backbone with a flow-matching Diffusion Transformer action head. \textsc{EquiVLA} introduces \textsc{EquiPerceptor}, which produces approximately $\mathrm{SO}(2)$-equivariant visual representations from frozen ViT features; and \textsc{EquiActor}, an exactly $\mathrm{SO}(2)$-equivariant flow-matching Diffusion Transformer action head. Together, they establish an approximate $\mathrm{SO}(2)$ equivariance chain from camera observations to predicted action sequences. Instantiated on GR00T~N1.5 and evaluated across four LIBERO suites, CALVIN ABCD$\to$D, and five real-robot tasks on Mobile ALOHA, \textsc{EquiVLA} achieves $92.6\%$ average success on LIBERO (vs. $78.1\%$ baseline), an average sequence length of $4.03$ on CALVIN (vs. $3.45$), and improves real-robot success from $54\%$ to $72\%$.
Chinese Translation
视觉-语言-动作（VLA）模型已成为通用机器人操作的强大范式，但它们缺乏几何归纳偏置：在特定方向上训练的策略需要大量数据才能在旋转配置中进行泛化。我们提出了 extsc{EquiVLA}，这是第一个端到端$ ext{SO}(2)$-等变VLA模型的通用框架，适用于将冻结的视觉-语言骨干与流匹配的扩散变换器（Diffusion Transformer）动作头结合的任何架构。 extsc{EquiVLA}引入了 extsc{EquiPerceptor}，它从冻结的ViT特征生成近似$ ext{SO}(2)$-等变的视觉表示；以及 extsc{EquiActor}，一个完全$ ext{SO}(2)$-等变的流匹配扩散变换器动作头。它们共同建立了从相机观察到预测动作序列的近似$ ext{SO}(2)$等变链。在GR00T~N1.5上实例化，并在四个LIBERO套件、CALVIN ABCD$ o$D和五个在Mobile ALOHA上的真实机器人任务中进行评估， extsc{EquiVLA}在LIBERO上实现了$92.6\%$的平均成功率（相比于$78.1\\%$的基线），在CALVIN上实现了$4.03$的平均序列长度（相比于$3.45$），并将真实机器人成功率从$54\\%$提高到$72\\%$。
View on arXiv Download PDF AI Translation
cs.RO / 29 / 2606.19813

TIDY: Thermal Infrared Image Denoising via Wavelet Domain Entropy and Directional Stripe Index

TIDY：基于小波域熵和方向条纹指数的热红外图像去噪
Rhee, Tai Hyoung, Lee, Dong-Guw, Kim, Ayoung
Abstract
Thermal infrared (TIR) imaging has been a popular choice for field robotics due to its robust perception capability under low light visual degradation, but it suffers from severe stochastic and fixed-pattern noise that breaks downstream estimation. This noise is intensified indoors due to low thermal contrast and uniform temperature distributions, contributing to the relative lack of indoor TIR deployments. Existing TIR denoising methods exhibit a poor accuracy-efficiency tradeoff, either too slow for online deployment required in robotics or insufficiently robust to severe degradation, while typically being trained on synthetic noise. Addressing these problems, we propose TIDY, a lightweight wavelet-domain denoiser trained on real clean-noisy TIR data. By reformulating TIR denoising in the wavelet domain, TIDY explicitly disentangles noise from structural content, enabling targeted suppression with reduced spatial complexity, significantly improving inference speed over prior methods (~34Hz). TIDY introduces two new metrics, Wavelet Entropy and Wavelet Directional Stripe Index, as complementary loss terms to explicitly suppress stochastic noise and stripe artifacts. Across severe indoor corruption and zero-shot settings, TIDY improves robustness and yields consistent gains in downstream robotics tasks including thermal inertial odometry and monocular depth estimation. Code and dataset is available at: https://github.com/williamrheeth/TIDY
Chinese Translation
热红外（TIR）成像因其在低光视觉退化下的强大感知能力而成为现场机器人领域的热门选择，但它受到严重的随机噪声和固定模式噪声的影响，这破坏了下游估计。由于低热对比度和均匀的温度分布，这种噪声在室内环境中更加严重，导致室内TIR部署相对缺乏。现有的TIR去噪方法在准确性和效率之间的权衡表现较差，要么速度过慢，无法满足机器人所需的在线部署，要么在严重退化情况下的鲁棒性不足，且通常是在合成噪声上进行训练。为了解决这些问题，我们提出了TIDY，一种在真实干净-噪声TIR数据上训练的轻量级小波域去噪器。通过在小波域中重新构造TIR去噪，TIDY明确地将噪声与结构内容分离，从而实现有针对性的抑制，并减少空间复杂性，显著提高了推理速度（约34Hz）。TIDY引入了两个新的指标：小波熵（Wavelet Entropy）和小波方向条纹指数（Wavelet Directional Stripe Index），作为互补损失项，以明确抑制随机噪声和条纹伪影。在严重的室内污染和零样本设置下，TIDY提高了鲁棒性，并在下游机器人任务中（包括热惯性测距和单目深度估计）带来了持续的收益。代码和数据集可在以下链接获取：https://github.com/williamrheeth/TIDY
View on arXiv Download PDF AI Translation
cs.RO / 30 / 2606.19836

World Engine: Towards the Era of Post-Training for Autonomous Driving

世界引擎：迈向自动驾驶后训练时代
Li, Tianyu, Chen, Li, Wang, Caojun, Liu, Haochen, Chitta, Kashyap, Yang, Zhenjie, Lu, Yuhang, Ye, Naisheng, Qiu, Yihang, Wang, Yufei, Zou, Luoxi, Peng, Jiaxin, Pan, Jin, Su, Zhaoyu, Bursuc, Andrei, Li, Shengbo Eben, Geiger, Andreas, Su, Peng, Li, Hongyang
Abstract
Autonomous vehicles must operate safely in the real world, where errors can have severe consequences. Although modern end-to-end driving policies excel in routine scenarios, their reliability is limited by the scarcity of safety-critical ``long-tail'' events in real driving datasets. These rare interactions define the practical safety boundary of the learned policy, yet they are difficult to collect at scale in the real world. Here we show that this fundamental limitation can be addressed by post-training pre-trained driving models on synthesized high-stakes interactions. We introduce World Engine, a generative framework that reconstructs high-fidelity interactive environments from real-world logs and systematically extrapolates them into realistic safety-critical variations. This paradigm enables reinforcement-based post-training to align policies with safety constraints, circumventing the physical risks inherent in real-world exploration. On a public benchmark built on nuPlan, World Engine substantially reduces failures in rare safety-critical scenarios and yields significantly larger gains than scaling pre-training data alone. Furthermore, when deployed on a production-scale autonomous driving system, the resulting policy reduces simulated collisions and demonstrates measurable improvements in on-road testing, showing that post-training on synthesized, safety-critical interactions offers a scalable and effective pathway to safer autonomous driving. The full codebase suite, including training, is released to the public.
Chinese Translation
自动驾驶车辆必须在现实世界中安全运行，而错误可能导致严重后果。尽管现代端到端驾驶策略在常规场景中表现出色，但它们的可靠性受到现实驾驶数据集中安全关键的“长尾”事件稀缺性的限制。这些稀有交互定义了学习策略的实际安全边界，但在现实世界中大规模收集这些事件非常困难。在此，我们展示了通过在合成高风险交互上对预训练驾驶模型进行后训练，可以解决这一基本限制。我们介绍了世界引擎（World Engine），这是一个生成框架，它从现实世界日志中重建高保真交互环境，并系统性地将其外推为现实的安全关键变体。这一范式使基于强化学习的后训练能够将策略与安全约束对齐，从而规避现实世界探索中固有的物理风险。在基于nuPlan构建的公共基准上，世界引擎显著减少了在稀有安全关键场景中的失败，并且相比单纯扩大预训练数据，带来了更显著的收益。此外，当在生产规模的自动驾驶系统上部署时，所产生的策略减少了模拟碰撞，并在路测中显示出可测量的改进，表明在合成的安全关键交互上进行后训练提供了一条可扩展且有效的路径，以实现更安全的自动驾驶。完整的代码库套件，包括训练部分，已向公众发布。
View on arXiv Download PDF AI Translation
cs.RO / 31 / 2606.19874

MMD-SLAM: Structure-Enhanced Multi-Meta Gaussian Distribution-Guided Visual SLAM

MMD-SLAM：结构增强的多元高斯分布引导视觉同时定位与地图构建
Zhu, Fan, Chen, Ziyu, Liu, Peichen, Zhao, Yifan, Xu, Zhisong, Zhu, Hui, Zhou, Hongxing, Liu, Sixun, Jiang, Chunmao
Abstract
3D Gaussian Splatting (3DGS) has significantly boosted novel view synthesis and high-fidelity scene reconstruction, expanding the potential of 3DGS-based Visual Simultaneous Localization and Mapping (SLAM) methods. However, most existing systems fail to fully exploit the underlying structural information, which limits rendering quality and often leads to inconsistent maps. To address these limitations, we propose MMD-SLAM, a structure-enhanced Visual SLAM framework that leverages the Atlanta World (AW) assumption to guide a Multi-Meta Gaussian representation for photorealistic mapping. First, we introduce a point-line fusion strategy for pose optimization, where 3D line segments are incorporated to improve tracking robustness and provide additional constraints for mapping. Second, we design a Multi-Meta Gaussian representation with dominant directions, explicitly encoding structural priors from the AW hypothesis. Finally, we propose a Gaussian evolution strategy that adapts to scene geometry and incorporates structural cues into global optimization. Extensive experiments demonstrate that these innovations enable MMD-SLAM to achieve state-of-the-art performance in both tracking accuracy and mapping quality. e.g., our method achieves a 48.56% reduction in ATE RMSE on ScanNet and a 5.71% improvement in PSNR on Replica, compared with MonoGS.
Chinese Translation
三维高斯点云（3D Gaussian Splatting，3DGS）显著提升了新视角合成和高保真场景重建，扩展了基于3DGS的视觉同时定位与地图构建（SLAM）方法的潜力。然而，大多数现有系统未能充分利用潜在的结构信息，这限制了渲染质量，并且常常导致不一致的地图。为了解决这些限制，我们提出了MMD-SLAM，一种结构增强的视觉SLAM框架，利用亚特兰大世界（Atlanta World，AW）假设来引导多元高斯表示以实现照片级真实感的地图构建。首先，我们引入了一种点线融合策略用于姿态优化，其中结合了三维线段以提高跟踪的鲁棒性，并为地图构建提供额外约束。其次，我们设计了一种具有主导方向的多元高斯表示，明确编码来自AW假设的结构先验。最后，我们提出了一种高斯演化策略，能够适应场景几何并将结构线索纳入全局优化。大量实验表明，这些创新使得MMD-SLAM在跟踪精度和地图质量方面实现了最先进的性能。例如，与MonoGS相比，我们的方法在ScanNet上实现了48.56%的ATE RMSE降低，在Replica上实现了5.71%的PSNR提升。
View on arXiv Download PDF AI Translation
cs.RO / 32 / 2606.19897

One-to-Two Acting: A Novel Framework for Single-arm Agent Action Expansion to Dual Arms

一对二行动：单臂智能体动作扩展到双臂的新框架
Yao, Youbin, Cao, Nieqin, Li, Mingyan, Ding, Yan, Gu, Fuqiang, Chen, Chao
Abstract
Dual-arm manipulation can improve throughput via parallel execution, but collecting bimanual demonstrations for training is costly and difficult. We present ExS2D, a hierarchical action expansion framework that enables dual-arm manipulation from single-arm supervision. ExS2D first generates structured subtasks from textual instructions while explicitly capturing temporal precedence. It then grounds each subtask into executable actions through subtask-guided action mapping in observation. Finally, precedence-aware action allocation and synchronized planning are performed by a multimodal large language model driven coordinator to select collision-free dual-arm executions. Simulation experiments demonstrate that ExS2D reduces the average execution steps by 54.4% while maintaining a comparable success rate to a single-arm baseline. Real-robot experiments on four tasks further demonstrate the reliability of ExS2D for dual-arm execution under few-shot single-arm samples, while using zero bimanual demonstrations.
Chinese Translation
双臂操作可以通过并行执行提高吞吐量，但收集双手演示以进行训练既昂贵又困难。我们提出了ExS2D，一个层次化的动作扩展框架，能够从单臂监督中实现双臂操作。ExS2D首先从文本指令生成结构化的子任务，同时明确捕捉时间优先关系。然后，通过子任务引导的动作映射在观察中将每个子任务转化为可执行的动作。最后，由多模态大型语言模型驱动的协调器执行优先关系感知的动作分配和同步规划，以选择无碰撞的双臂执行。仿真实验表明，ExS2D将平均执行步骤减少了54.4%，同时保持与单臂基线相当的成功率。在四个任务上的真实机器人实验进一步证明了ExS2D在少量单臂样本下进行双臂执行的可靠性，同时使用零双手演示。
View on arXiv Download PDF AI Translation
cs.RO / 33 / 2606.19914

Co-policy: Responsive Human-Robot Co-Creation for Musical Performances

共政策：响应式人机共创音乐表演
Li, Xuetao, Huang, Wenke, Ye, Mang, Liu, Zijian, Xie, Jinhua, Xuan, Jifeng, Li, Miao
Abstract
Art has long stood as a pivotal expression of human creativity. Embodied artificial intelligence offers a route for generative models to participate in that creativity through physical action rather than disembodied digital content. In robotic music co-creation, it is challenging to connect semantic musical understanding with real-time and physically executable performance. We present Co-policy, a framework for human-robot musical co-creation that separates semantic intent grounding, constrained musical variation, and visuomotor execution. To ground musical semantics, Co-policy uses pre-inference semantic anchors and a fine-tuned Qwen-vl planner (F-Qwen) to transform speech, live musical seeds, and visual observations into structured co-creation plans. To support low-latency execution, Co-policy introduces a Gaussian-Mixture Visuomotor Policy (GMP), implemented as a conditional mixture-density policy that maps target notes and visual context to multimodal robot actions in a single forward pass. Unlike robotic playback systems that merely reproduce user-specified notes, Co-policy generates complementary musical responses under both musical and physical constraints. Real-robot chime experiments, ablations, and expert evaluation show improved intent alignment, execution accuracy, and response frequency over diffusion-policy and ablated baselines, supporting physically grounded action generation as a key requirement for embodied human-AI co-creation.
Chinese Translation
艺术长期以来一直是人类创造力的重要表达方式。具身人工智能为生成模型通过物理行动参与这种创造力提供了一条途径，而不是通过无形的数字内容。在机器人音乐共创中，将语义音乐理解与实时且可物理执行的表演连接起来是一个挑战。我们提出了共政策（Co-policy），这是一个人机音乐共创的框架，分离了语义意图基础、受限的音乐变异和视觉运动执行。为了将音乐语义与基础相结合，共政策使用预推理语义锚点和经过微调的Qwen-vl规划器（F-Qwen），将语音、实时音乐种子和视觉观察转化为结构化的共创计划。为了支持低延迟执行，共政策引入了一种高斯混合视觉运动策略（GMP），该策略作为条件混合密度策略实现，将目标音符和视觉上下文映射到多模态机器人动作中，并在单次前向传递中完成。与仅仅重现用户指定音符的机器人播放系统不同，共政策在音乐和物理约束下生成互补的音乐响应。真实机器人风铃实验、消融实验和专家评估显示出意图对齐、执行准确性和响应频率的改善，相较于扩散策略和消融基线，支持了具身人机共创的物理基础行动生成作为关键要求。
View on arXiv Download PDF AI Translation
cs.RO / 34 / 2606.19920

Deep-Unfolded Coordination

深度展开协调
Kuperman, Hunter, Jung, Minchan, Ghosh, Rahul V., Oshin, Alex, Theodorou, Evangelos A.
Abstract
Distributed optimization is a highly scalable and structurally transparent technique to solve multi-agent robotics problems; however, such methods often suffer from the need for highly-specialized, problem-specific hyperparameter tunings. In this work, we propose Deep Coordinator, a deep-unfolding framework that learns to dynamically adjust the hyperparameters of ADMM-DDP, a popular distributed solver for robotics tasks, at solve-time in response to optimizer performance. Our architecture consists of unrolling a fixed number of ADMM-DDP iterations into a neural network with learnable functions between layers mapping the optimizer state to the next hyperparameters. To the best of our knowledge, Deep Coordinator is the first deep-unfolding framework to adapt the penalty parameters of a non-convex optimizer at solve-time; we show that the mainstream supervised approach can yield degenerate solutions when training such models, and propose an unsupervised learning scheme. On simulations with fleets of cars and quadrotors, Deep Coordinator produces trajectories of comparable quality 6.18-9.44x faster than conventional solvers. Furthermore, Deep Coordinator retains its performance benefits when deployed to systems up to 8x larger than trained on.
Chinese Translation
分布式优化是一种高度可扩展且结构透明的技术，用于解决多智能体机器人问题；然而，这类方法通常需要高度专业化的、特定问题的超参数调优。在本研究中，我们提出了深度协调器（Deep Coordinator），这是一种深度展开框架，能够在求解时根据优化器的性能动态调整ADMM-DDP（交替方向乘子法-动态规划）的超参数，ADMM-DDP是一个流行的分布式求解器，适用于机器人任务。我们的架构由固定数量的ADMM-DDP迭代展开为一个神经网络，网络层之间的可学习函数将优化器状态映射到下一个超参数。根据我们所知，深度协调器是第一个在求解时调整非凸优化器惩罚参数的深度展开框架；我们展示了主流的监督学习方法在训练此类模型时可能导致退化解，并提出了一种无监督学习方案。在对车队和四旋翼的仿真中，深度协调器以比传统求解器快6.18-9.44倍的速度生成可比质量的轨迹。此外，当部署到比训练时大8倍的系统时，深度协调器仍然保持其性能优势。
View on arXiv Download PDF AI Translation
cs.RO / 35 / 2606.19928

SWAP: Symmetric Equivariant World-Model for Agile Robot Parkour

SWAP：用于灵活机器人跑酷的对称等变世界模型
Lan, Kaixin, Wang, Ze, Li, Hongyi, Jiang, Lei, Fu, Chaojie, Su, Chengkai, Wong, Choi Lam, Jin, Yongbin, Wang, Hongtao
Abstract
While latent world models enable the proactive predictions required for extreme parkour, their purely data-driven nature forces them to redundantly encode left-right symmetric interactions as independent patterns. This inflates the learning burden and hinders the capture of geometric regularities, restricting the latent space's efficiency for downstream policies. To address this, we propose SWAP, an end-to-end equivariant symmetric world model. This framework embeds symmetry directly into both the world model and the actor-critic networks. In real-world tests, the robot leaps across a 2.13 m gap and climbs a 1.63 m platform, breaking records for quadruped parkour. Furthermore, the framework exhibits robust geometric generalization to unseen mirrored terrains and exceptional zero-shot transferability across diverse outdoor environments. These results demonstrate that symmetry equivariance is an effective structural prior for pushing the physical boundaries of learned legged locomotion.
Chinese Translation
尽管潜在世界模型能够实现极限跑酷所需的主动预测，但其纯数据驱动的特性迫使其将左右对称的交互冗余编码为独立模式。这增加了学习负担，并阻碍了几何规律的捕捉，限制了潜在空间在下游策略中的效率。为了解决这个问题，我们提出了SWAP，一个端到端的等变对称世界模型。该框架将对称性直接嵌入到世界模型和演员-评论家网络中。在现实世界的测试中，机器人跨越了2.13米的间隙，并攀爬了1.63米的平台，打破了四足跑酷的记录。此外，该框架在未见过的镜像地形上表现出强大的几何泛化能力，并在多样的户外环境中展现出卓越的零样本迁移能力。这些结果表明，对称等变性是推动学习腿部运动物理极限的有效结构先验。
View on arXiv Download PDF AI Translation
cs.RO / 36 / 2606.19929

Motor Angular Speed Preintegration for Multirotor UAV State Estimation

多旋翼无人机状态估计的电机角速度预积分
Petrlík, Matěj, Novák, Filip, Pěnička, Robert, Saska, Martin
Abstract
A precise state estimate is crucial for a tight feedback control that enables agile and near-obstacle flights of UAVs. The state-of-the-art methods fuse slow pose measurements with high-frequency inertial measurements to obtain a precise state estimate. However, the inertial measurements from the IMU onboard the UAV are degraded by vibrations from spinning propellers and the precision of the estimated state suffers. We propose a novel approach based on the preintegration of accelerations obtained from motor speeds. We show that the accelerations obtained in this manner can be used for state propagation on their own to achieve better precision without including the IMU. Further, we propose a factor composed of the preintegrated motor speeds that can be directly employed in factor graph optimization frameworks. We combine our factor with LiDAR measurements into the proposed Motor Angular Speed LiDAR Odometry (MAS-LO) algorithm for precise state estimation, which we open-source. Lastly, we evaluate the estimation precision against a state-of-the-art inertial algorithm LIO-SAM to show 28% improvement in position and 65% in velocity estimation accuracy, 14% lower measurement lag, and high robustness to wrong parameter values.
Chinese Translation
精确的状态估计对于实现无人机的灵活飞行和近障碍物飞行的紧密反馈控制至关重要。现有的先进方法将低频姿态测量与高频惯性测量融合，以获得精确的状态估计。然而，无人机上IMU的惯性测量受到旋转螺旋桨振动的影响，导致估计状态的精度下降。我们提出了一种基于电机速度获得的加速度预积分的新方法。我们证明，以这种方式获得的加速度可以单独用于状态传播，从而在不包含IMU的情况下实现更好的精度。此外，我们提出了一种由预积分电机速度组成的因子，可以直接应用于因子图优化框架。我们将该因子与LiDAR测量结合，提出了用于精确状态估计的电机角速度LiDAR里程计（Motor Angular Speed LiDAR Odometry, MAS-LO）算法，并将其开源。最后，我们将估计精度与先进的惯性算法LIO-SAM进行比较，显示出位置估计精度提高28%，速度估计精度提高65%，测量延迟降低14%，并对错误参数值具有较高的鲁棒性。
View on arXiv Download PDF AI Translation
cs.RO / 37 / 2606.19971

Evaluation of Augmented Reality-based Intuitive Interface for Robot-Assisted Transesophageal Echocardiography: A User Study

基于增强现实的直观界面在机器人辅助经食道超声心动图中的评估：用户研究
Zhang*, Xiu, Di Mauro*, Matteo, Breschi, Sofia, Peloso, Angela, Votta, Emiliano, Menciassi, Arianna, De Momi, Elena
Abstract
TransEsophageal Echocardiography (TEE) is essential for diagnosing and guiding Structural Heart Disease (SHD) interventions. However, manual TEE manipulation demands significant operator expertise, is physically demanding, and exposes clinicians to radiation when performed alongside fluoroscopy. Robotic-assisted TEE systems have been introduced to improve probe handling and reduce operator fatigue, yet the design of intuitive and effective user interfaces remains an open challenge. This study presents and evaluates a model-enhanced, Augmented Reality (AR)-based intuitive interface for robot-assisted TEE, designed to improve spatial awareness and control intuitiveness. A robotic TEE platform integrated with electromagnetic tracking and a virtual simulator was used to compare three user interfaces differing in visualization and interaction modalities: 2D jointlevel (2D-JI), 3D joint-level (3D-JI), and 3D tip-level (3D-TI). Thirty six participants performed standardized navigation tasks to reproduce target echocardiographic views, with performance assessed via position and orientation errors, completion time, and NASA-TLX workload scores. Results show that 3D visualization significantly improved spatial accuracy, reducing median position error from 13 mm to 3 mm and halving the orientation error compared with the 2D interface. Tip-level interaction yielded a further 50% reduction in orientation error and reduced interuser variability relative to joint-level control. Overall, the 3D-TI configuration, combining immersive visualization with direct tip-level control, proved the most effective and ergonomic interface, supporting the integration of AR-based visualization and intuitive control paradigms into next-generation robotic TEE systems to enhance operator performance and procedural safety.
Chinese Translation
经食道超声心动图（TEE）对于诊断和指导结构性心脏病（SHD）干预至关重要。然而，手动TEE操作需要显著的操作者专业知识，且对身体要求较高，同时在与荧光透视联合使用时会使临床医生暴露于辐射中。为改善探头操作并减少操作者疲劳，已引入机器人辅助TEE系统，但设计直观且有效的用户界面仍然是一个亟待解决的挑战。本研究提出并评估了一种增强现实（AR）基础的直观界面，旨在改善空间意识和控制直观性。我们使用集成了电磁追踪和虚拟模拟器的机器人TEE平台，比较了三种在可视化和交互模式上不同的用户界面：2D关节级（2D-JI）、3D关节级（3D-JI）和3D尖端级（3D-TI）。三十六名参与者执行标准化导航任务，以重现目标超声心动图视图，性能通过位置和方向误差、完成时间以及NASA-TLX工作负荷评分进行评估。结果显示，3D可视化显著提高了空间准确性，将中位位置误差从13毫米降低到3毫米，并将方向误差减半，相较于2D界面。尖端级交互进一步减少了50%的方向误差，并降低了相对于关节级控制的用户间变异性。总体而言，3D-TI配置结合了沉浸式可视化和直接尖端级控制，证明是最有效和符合人体工程学的界面，支持将基于AR的可视化和直观控制范式整合到下一代机器人TEE系统中，以提升操作者性能和程序安全性。
View on arXiv Download PDF AI Translation
cs.RO / 38 / 2606.19998

Tri-Info: Generalizable, Interpretable Failure Prediction for VLA Models via Information Theory

三重信息：基于信息理论的可推广、可解释的 VLA 模型故障预测
Yang, Jinghan, Zhang, Yunchao, Yuan, Wang, Wan, Haolun, Zhang, Jiaming, Hu, Zhengyang, Yang, Yanchao
Abstract
Vision-Language-Action (VLA) models are increasingly deployed across diverse tasks, yet they remain black boxes whose physical interactions can cause irreversible harm, making generalizable and interpretable failure detection essential. We observe that successful and failed rollouts carry systematically different information-theoretic signatures. Building on this, we formalize VLA control as a closed-loop information pipeline and derive the Triple Information-theoretic (Tri-Info) signals that capture whether actions remain diverse, temporally consistent, and coupled to state transitions. Across six VLA models and three benchmark environments, Tri-Info matches the strongest baselines in-domain. Moreover, Tri-Info transfers across architectures, environments, and the sim-to-real gap without retraining, reaching 83\% accuracy on real-world tasks where prior detectors collapse to chance. This establishes Tri-Info as a simple yet powerful method that not only detects failures with strong cross-domain generalization, but also delivers interpretable diagnostics of the underlying failure modes.
Chinese Translation
视觉-语言-行动（VLA）模型在各种任务中越来越多地被部署，但它们仍然是黑箱，其物理交互可能造成不可逆的损害，因此可推广和可解释的故障检测至关重要。我们观察到成功和失败的执行在信息理论特征上存在系统性差异。基于此，我们将 VLA 控制形式化为一个闭环信息管道，并推导出三重信息理论（Tri-Info）信号，以捕捉动作是否保持多样性、时间一致性以及与状态转变的耦合。在六个 VLA 模型和三个基准环境中，Tri-Info 在领域内匹配了最强的基线。此外，Tri-Info 在不同架构、环境和模拟到现实的差距中无需重新训练即可迁移，在现实世界任务中达到 83\% 的准确率，而先前的检测器则崩溃至随机水平。这确立了 Tri-Info 作为一种简单而强大的方法，不仅能够以强大的跨领域泛化能力检测故障，还能提供对潜在故障模式的可解释诊断。
View on arXiv Download PDF AI Translation
cs.RO / 39 / 2606.20031

A Neuromorphic Reinforcement Learning Framework for Efficient Pathfinding in Robotic Mobile Fulfillment Systems

一种用于机器人移动履行系统高效路径规划的神经形态强化学习框架
Xu, Junzhe, Zeng, Zecui, Li, Lusong, Fang, Yuetong, Xu, Renjing
Abstract
Dynamic environmental changes, confined workspaces, and stringent real-time constraints make pathfinding in Robotic Mobile Fulfillment Systems (RMFS) a challenging problem for conventional search- and rule-based methods, which typically suffer from high computational complexity and long decision latency. While reinforcement learning (RL) has emerged as a powerful alternative, deploying learned policies with extreme energy efficiency on resource-constrained hardware remains an open challenge. We present SDQN-RMFS, an end-to-end framework that achieves high-fidelity deployment of an RL-trained policy from a full-precision artificial neural network (ANN) through to a neuromorphic chip. By computing only when triggered by sparse events, this framework unlocks ultra-low-power RMFS pathfinding. Our full-stack pipeline operates as follows: an ANN policy is first efficiently trained via a collision-allowing strategy to densify informative trajectories, and then converted into a spiking neural network (SNN) via a hard-label knowledge distillation approach. This effectively addresses the output distribution mismatch, preserving policy capability across the ANN-to-SNN pipeline while substantially reducing inference latency. Hardware experiments demonstrate up to 11,281$\times$ energy savings and a nearly two-fold reduction in latency compared to a high-performance GPU baseline, while maintaining decision quality on par with the original trained policy. These results establish physical neuromorphic inference as a practical and energy-sustainable pathway for large-scale RMFS operations.
Chinese Translation
动态环境变化、有限的工作空间和严格的实时约束使得在机器人移动履行系统（RMFS）中进行路径规划成为一个具有挑战性的问题，传统的基于搜索和规则的方法通常面临高计算复杂度和长决策延迟的问题。尽管强化学习（RL）已成为一种强有力的替代方案，但在资源受限的硬件上以极高的能效部署学习到的策略仍然是一个未解决的挑战。我们提出了SDQN-RMFS，这是一个端到端框架，能够实现从全精度人工神经网络（ANN）到神经形态芯片的高保真RL训练策略部署。通过仅在稀疏事件触发时进行计算，该框架解锁了超低功耗的RMFS路径规划。我们的全栈管道操作如下：首先通过允许碰撞的策略有效训练ANN策略，以密集化信息轨迹，然后通过硬标签知识蒸馏方法将其转换为脉冲神经网络（SNN）。这有效解决了输出分布不匹配的问题，保持了ANN到SNN管道中的策略能力，同时显著减少了推理延迟。硬件实验表明，与高性能GPU基线相比，能耗节省高达11,281倍，延迟几乎减少了一半，同时决策质量与原始训练策略相当。这些结果确立了物理神经形态推理作为大规模RMFS操作的实用且节能的途径。
View on arXiv Download PDF AI Translation
cs.RO / 40 / 2606.20048

MirrorDuo: Reflection-Consistent Visuomotor Learning from Mirrored Demonstration Pairs

MirrorDuo：基于反射的一致视觉运动学习来自镜像演示对
Zhuang, Zheyu, Wang, Ruiyu, Marchetti, Giovanni Luca, Pokorny, Florian T., Kragic, Danica
Abstract
Image-based behaviour cloning leverages demonstrations captured from ubiquitous RGB cameras. However, it remains constrained by the cost of collecting diverse demos, especially for generalizing across workspace variations. We propose MirrorDuo, a reflection-based formulation that operates on image, proprioception, and full 6-DoF end-effector action tuples, generating a mirrored counterpart for each original demonstration, effectively achieving "collect one, get one for free". It can be applied as a data augmentation strategy for existing learning pipelines, such as standard behaviour cloning or diffusion policy, or as a structural prior for reflection-equivariant policy networks. By leveraging the overlap between the original and mirrored domains, MirrorDuo achieves significantly improved performance under the same data budget when demonstrations are evenly distributed across both sides of the workspace. When demonstrations are confined to one side, MirrorDuo enables efficient skill transfer to the mirrored workspace with as few as zero or five demos in the target arrangement.
Chinese Translation
基于图像的行为克隆利用来自普遍 RGB 摄像头捕获的演示。然而，它仍然受到收集多样化演示成本的限制，特别是在跨工作空间变异性进行泛化时。我们提出了 MirrorDuo，一种基于反射的公式，操作于图像、身体感知和完整的 6 自由度末端执行器动作元组，为每个原始演示生成一个镜像对应，从而有效实现“收集一个，免费获得一个”。它可以作为现有学习流程的数据增强策略，例如标准行为克隆或扩散策略，或作为反射等变策略网络的结构先验。通过利用原始领域和镜像领域之间的重叠，MirrorDuo 在相同数据预算下显著提高了性能，尤其是在演示均匀分布在工作空间的两侧时。当演示仅限于一侧时，MirrorDuo 使得在目标排列中仅需零或五个演示即可有效地将技能转移到镜像工作空间。
View on arXiv Download PDF AI Translation
cs.RO / 41 / 2606.20056

VFILC: Accurate Frequency Extrapolations in Imitation Learning via Sampling Frequency ILC

VFILC：通过采样频率ILC实现模仿学习中的准确频率外推
Masuya, Nozomu, Tsuji, Toshiaki, Sakaino, Sho
Abstract
Conventional neural network (NN)-based imitation learning methods for variable-speed motion either restricted their scope to interpolated speeds, or generated unpredictable motions when extrapolating beyond trained velocity ranges. Variable-frequency imitation learning (VFIL) enabled extrapolations of speeds by linking the NN model's sampling frequency to the motion frequency, whereas its open-loop configuration caused frequency errors, especially in the extrapolated high-frequency settings. This study proposes variable-frequency imitation learning with iterative learning control (VFILC) based on a combination of VFIL and iterative learning control (ILC) with both feedforward and feedback parts, the former taking advantage of VFIL and the latter adjusting the frequency errors. The experimental results showed that the proposed method successfully and accurately extrapolated motion speeds and reduced frequency errors in all three tasks, and that the feedback especially reduced the frequency errors by a remarkable 81% in the wiping task and 50% in the shaking task, both compared to simple feedforward VFIL, when extrapolating at double the average speed in the training data. The proposed method also improved accuracy by 27% compared with VFIL even at an interpolated frequency for a contact-rich mixing task affected by complex friction traits.
Chinese Translation
传统的基于神经网络（NN）的变速运动模仿学习方法要么将其范围限制在插值速度，要么在超出训练速度范围时产生不可预测的运动。变频模仿学习（VFIL）通过将NN模型的采样频率与运动频率相连接，实现了速度的外推，但其开环配置导致了频率误差，特别是在外推高频设置时。本研究提出了一种基于变频模仿学习与迭代学习控制（ILC）相结合的变频模仿学习迭代控制（VFILC），其中包括前馈和反馈部分，前者利用了VFIL的优势，后者则调整频率误差。实验结果表明，所提出的方法成功且准确地外推了运动速度，并在所有三个任务中减少了频率误差，反馈部分在擦拭任务中将频率误差显著减少了81%，在摇晃任务中减少了50%，与简单的前馈VFIL相比，在训练数据中以双倍平均速度进行外推时。即使在受复杂摩擦特性影响的接触丰富混合任务中，所提出的方法在插值频率下也比VFIL提高了27%的准确性。
View on arXiv Download PDF AI Translation
cs.RO / 42 / 2606.20118

Pose6DAug: Physically Plausible Multi-view Object Swapping for Robot Data Augmentation

Pose6DAug：用于机器人数据增强的物理合理多视角物体交换
Lee, Jonghoon, Park, Seong Hyeon, Jeon, Byungwoo, Lee, Minha, Shin, Jinwoo
Abstract
Vision-language-action (VLA) policies have shown strong potential for general-purpose manipulation, yet they often fail on novel, out-of-distribution objects whose appearance or geometry deviates from the training distribution. The standard remedy is to collect multi-view teleoperation data for every failure case, but this scales poorly in both cost and time. We introduce Pose6DAug, a failure-driven data augmentation framework that turns a policy's own successful episodes into targeted demonstrations for its failure modes, without any new data collection. Our key insight is that each successful episode already encodes a physically valid action trajectory together with calibrated multi-view observations. By swapping only the manipulated object while preserving this trajectory, we obtain new and physically grounded demonstrations. However, naive 2D video editing breaks multi-view consistency and physical plausibility, particularly under heavy occlusion and egocentric viewpoints. Our method instead operates directly in 3D, anchoring the target object with an explicit mesh driven by a temporally coherent 6D pose trajectory, ensuring geometrically consistent renderings across all camera views. Fine-tuning a VLA on data augmented by our method improves success rates by 16.5% relative to the state-of-the-art baseline on novel objects, while preserving in-distribution performance. These results show that multi-view and physically consistent augmentation is a practical path to scalable VLA generalization.
Chinese Translation
视觉-语言-动作（VLA）策略在通用操作中展现出强大的潜力，但它们在新颖的、超出分布的物体上往往表现不佳，这些物体的外观或几何形状偏离了训练分布。标准的解决方案是为每个失败案例收集多视角遥操作数据，但这在成本和时间上都难以扩展。我们提出了Pose6DAug，一个基于失败驱动的数据增强框架，它将策略自身成功的经历转化为针对其失败模式的目标演示，而无需任何新的数据收集。我们的关键见解是，每个成功的经历已经编码了一个物理有效的动作轨迹，以及经过校准的多视角观测。通过仅交换被操作的物体，同时保持这一轨迹，我们获得了新的、物理上合理的演示。然而，简单的2D视频编辑会破坏多视角一致性和物理合理性，尤其是在重遮挡和自我中心视角下。我们的方法则直接在3D空间中操作，通过一个由时间一致的6D姿态轨迹驱动的显式网格来锚定目标物体，确保在所有相机视角下几何一致的渲染。通过我们的方法增强的数据对VLA进行微调，相较于最新的基准，在新颖物体上的成功率提高了16.5%，同时保持了在分布内的性能。这些结果表明，多视角和物理一致的增强是实现可扩展VLA泛化的实用路径。
View on arXiv Download PDF AI Translation
cs.RO / 43 / 2606.20120

Dual-Agent Framework for Cross-Model Verified Translation of Natural-Language Protocols into Robotic Laboratory Platform

双代理框架用于自然语言协议向机器人实验平台的跨模型验证翻译
Choi, Hyeonna, Kim, Jung Yup, Lim, Hyuneui, Jeon, Seunggyu
Abstract
Biological experiment protocols are written in natural language, whereas automation systems rely on predefined control commands, creating a semantic gap that limits autonomous execution. Microplate-based automatic experiments are particularly challenging due to the need to simultaneously control well mapping, sample-reagent combinations, replicate placement, and parallel dispensing. This study proposes an agent-based protocol translation framework that converts natural-language microplate-based protocols into executable control commands for a robotic laboratory platform. A Parser Agent formalizes the natural-language protocol into a structured representation, and a rule-based mapping engine deterministically incorporates the operational constraints of the robotic laboratory platform to generate device-level control commands. A heterogeneous LLM Validation Agent verifies completeness, parameter accuracy, and execution order, and triggers a self-correction loop with structured feedback when errors are detected. A sweep involving 7 Parsers and 3 Validators on randomly selected ELISA protocols evaluates how model scale and Validator type affect translation accuracy and pass rates under cross-model verification. The accuracy-latency trade-off is further verified by comparing the rule-based mapping of the proposed framework with LLM end-to-end direct mapping. Finally, Bradford assay-based protein quantification using a microplate was demonstrated on a robotic laboratory platform, validating end-to-end autonomous execution from natural-language protocols to real-world experiments. The proposed framework provides a flexible approach to narrowing the semantic gap between natural-language protocols and microplate-based self-driving laboratories.
Chinese Translation
生物实验协议通常以自然语言书写，而自动化系统则依赖于预定义的控制命令，这导致了限制自主执行的语义差距。基于微孔板的自动实验尤其具有挑战性，因为需要同时控制孔位映射、样品-试剂组合、重复样品放置和并行分配。本研究提出了一种基于代理的协议翻译框架，将自然语言的微孔板协议转换为可执行的机器人实验平台控制命令。解析代理（Parser Agent）将自然语言协议形式化为结构化表示，基于规则的映射引擎确定性地结合机器人实验平台的操作约束，以生成设备级控制命令。异构的 LLM 验证代理（Validation Agent）验证完整性、参数准确性和执行顺序，并在检测到错误时触发带有结构化反馈的自我修正循环。通过对随机选择的 ELISA 协议进行 7 个解析器和 3 个验证器的评估，研究模型规模和验证器类型如何影响跨模型验证下的翻译准确性和通过率。通过将所提框架的基于规则的映射与 LLM 端到端直接映射进行比较，进一步验证了准确性与延迟之间的权衡。最后，在机器人实验平台上演示了基于布拉德福德法的微孔板蛋白定量，验证了从自然语言协议到实际实验的端到端自主执行。所提框架为缩小自然语言协议与基于微孔板的自驱动实验室之间的语义差距提供了一种灵活的方法。
View on arXiv Download PDF AI Translation
cs.RO / 44 / 2606.20135

Frequency-Aware Flow Matching for Continuous and Consistent Robotic Action Generation

频率感知流匹配用于连续一致的机器人动作生成
Guo, Jianing, Chen, Fangzheng, Mao, Zihao, Kenny, Wong Lik Hang, Wu, Zhenhong, Li, Yu, Cai, Yishuai, Chen, Yuanpei, Ban, Yikun, Chen, Kai, Dou, Qi, Yang, Yaodong, Liu, Xianglong, Zhao, Huijie, Li, Simin
Abstract
Flow matching has emerged as a standard paradigm for robotic manipulation owing to its strong expressive power for modelling complex, multimodal action distributions, alongside similar approaches like diffusion policy. However, existing methods rely on discretized action chunks, making them brittle to demonstrations collected at heterogeneous control frequencies and prone to temporally inconsistent actions that degrade control stability. In this paper, we propose Frequency-Aware Flow Matching (FAFM), which outputs continuous, temporally consistent actions. To handle heterogeneous frequency input, we transform discrete action sequences into the frequency domain with the discrete cosine transform (DCT), perform flow matching over the resulting coefficients, and reconstruct continuous actions via cosine basis expansion. To generate temporally consistent actions, we regularize the first-order temporal derivative to promote smooth actions. This corresponds to a Sobolev-type constraint that suppresses high-frequency errors and discourages abrupt action changes. Our FAFM is simple, introduces no additional network parameters and applies to standalone flow-matching policies and vision-language action models. Across synthetic toy benchmark, obstacle avoidance, LapGym, and LIBERO, FAFM improves success rates, multimodal expressivity, motion smoothness, convergence speed, robustness to mechanical bias and mixed-frequency input. These gains are consistent when deployed on a real-world Franka robot. Code available at https://anonymous.4open.science/r/FAFM.
Chinese Translation
流匹配已成为机器人操作的标准范式，因其在建模复杂的多模态动作分布方面具有强大的表达能力，类似的方法还有扩散策略。然而，现有方法依赖于离散的动作块，这使得它们对以异构控制频率收集的演示变得脆弱，并容易产生时间不一致的动作，从而降低控制稳定性。本文提出了频率感知流匹配（Frequency-Aware Flow Matching, FAFM），该方法输出连续且时间一致的动作。为了处理异构频率输入，我们将离散动作序列转换到频率域，使用离散余弦变换（Discrete Cosine Transform, DCT）进行处理，在得到的系数上执行流匹配，并通过余弦基扩展重构连续动作。为了生成时间一致的动作，我们对一阶时间导数进行正则化，以促进动作的平滑性。这对应于一种Sobolev类型的约束，抑制高频误差并防止突发的动作变化。我们的FAFM方法简单，不引入额外的网络参数，并适用于独立的流匹配策略和视觉-语言动作模型。在合成玩具基准、障碍物避让、LapGym和LIBERO等任务中，FAFM提高了成功率、多模态表现力、运动平滑性、收敛速度、对机械偏差和混合频率输入的鲁棒性。这些提升在实际的Franka机器人上部署时也保持一致。代码可在 https://anonymous.4open.science/r/FAFM 获取。
View on arXiv Download PDF AI Translation
cs.RO / 45 / 2606.20150

Robust Assembly State Reasoning from Action Recognition for Human-Robot Collaboration

基于动作识别的鲁棒组装状态推理用于人机协作
Fant-Male, James, Pieters, Roel
Abstract
Human Action Recognition (HAR) is frequently investigated in Human-Robot Collaboration (HRC) research to understand what actions have been performed and hence the state of a collaborative task. Accurately tracking an assembly state from HAR is however not fully investigated, and in realistic scenarios is not a trivial task. This research systematically investigates and compares methods for tracking assembly state using action recognition inputs. Investigations using two diverse datasets and five state tracking approaches, including logic-based, Hidden Markov Model (HMM), and neural network (NN) methods, show that optimal approaches are not uniform across different tasks and that different methods fail under different circumstances. Testing is performed using both simulated inputs with varying noise levels and realistic inputs from a HAR model. Results show NN and HMM methods can perform well in tasks with limited variability, but for other scenarios logic-based approaches can be more robust. Methods which model expected action duration are also important for tasks with repeated actions where no additional sensing is provided.
Chinese Translation
人类动作识别（HAR）在研究人机协作（HRC）中被频繁探讨，以理解已执行的动作以及协作任务的状态。然而，从HAR中准确跟踪组装状态尚未得到充分研究，并且在现实场景中并非易事。本研究系统地调查并比较了使用动作识别输入跟踪组装状态的方法。通过使用两个不同的数据集和五种状态跟踪方法，包括基于逻辑的方法、隐马尔可夫模型（HMM）和神经网络（NN）方法的研究表明，最佳方法在不同任务中并不统一，并且不同的方法在不同情况下会失效。测试使用了具有不同噪声水平的模拟输入和来自HAR模型的真实输入。结果显示，NN和HMM方法在变化有限的任务中表现良好，但在其他场景中，基于逻辑的方法可能更具鲁棒性。对于重复动作且没有额外传感的任务，建模预期动作持续时间的方法也显得尤为重要。
View on arXiv Download PDF AI Translation
cs.RO / 46 / 2606.20193

Belt-Finger: An Affordable Soft Belt-Driven Gripper for Dexterous In-Hand Manipulation

带指抓手：一种经济实惠的软带驱动抓手用于灵巧的手内操作
Zhang, Boya, Zell, Andreas, Martius, Georg
Abstract
Parallel-jaw grippers are the default manipulator choice in robotics because they are simple, robust, and inexpensive. Their limited in-hand mobility, however, often forces large arm motions and restricts dexterous manipulation in confined workspaces. We present a parallel-gripper upgrade: a double-soft-belt-based finger module that preserves standard opening/closing while adding three in-hand degrees of freedom (DoF): translation, pitch, and roll. The mechanism is deliberately kept simple and engineered for inexpensive manufacturing and straightforward integration, preserving the reliability and precise control of traditional parallel grippers while greatly broadening the range of manipulation capabilities. To demonstrate the utility of the added DoFs, we integrate the gripper in two control pipelines. First, we adapt a model predictive controller for in-hand manipulation of known objects. Second, we introduce a lightweight teleoperation interface that enables simultaneous control of the robot arm and gripper (10 DoFs total) with minimal hardware. Across a suite of challenging manipulation tasks executed via teleoperation, MPC, and trained policies, the proposed gripper consistently improves dexterity and task feasibility compared to a conventional parallel gripper
Chinese Translation
平行爪抓手是机器人领域的默认操作器选择，因为它们简单、坚固且价格低廉。然而，它们有限的手内移动性常常迫使大型臂部运动，并限制了在狭小工作空间中的灵巧操作。我们提出了一种平行抓手的升级：一种基于双软带的指模块，保留了标准的开合功能，同时增加了三个手内自由度（DoF）：平移、俯仰和滚转。该机制故意保持简单，并设计为经济制造和易于集成，保留了传统平行抓手的可靠性和精确控制，同时大大扩展了操作能力的范围。为了展示新增自由度的实用性，我们将抓手集成到两个控制管道中。首先，我们为已知物体的手内操作调整了模型预测控制器（MPC）。其次，我们引入了一种轻量级的遥操作接口，使得机器人臂和抓手（总共10个自由度）的同时控制成为可能，且硬件需求最小。在通过遥操作、MPC和训练策略执行的一系列具有挑战性的操作任务中，所提出的抓手在灵巧性和任务可行性方面始终优于传统平行抓手。
View on arXiv Download PDF AI Translation
cs.RO / 47 / 2606.20197

Stable Transformer-Actor-Critic Model Predictive Control: A Contraction Analysis Approach

稳定的变换器-演员-评论家模型预测控制：收缩分析方法
Marino, Antonio, Modugno, Valerio, Cognetti, Marco
Abstract
Actor-Critic Model Predictive Control (MPC) effectively addresses complex, non-convex control problems, but guaranteeing the closed-loop stability of sequence-based learning models within these pipelines remains challenging. This paper introduces a novel Transformer-Actor-Critic MPC architecture with formal robustness guarantees. First, we prove that Transformer networks can satisfy global incremental Input-to-State Stability ($\delta$ISS). We then leverage Riemannian contraction theory to analyze the interconnected dynamics between the physical plant and the predictive neural network. Finally, we integrate these theoretical bounds as a training regularizer to yield a certifiably robust policy. The framework is validated on a nonlinear 3D drone model executing target-reaching and obstacle-avoidance maneuvers.
Chinese Translation
演员-评论家模型预测控制（MPC）有效地解决了复杂的非凸控制问题，但在这些流程中保证基于序列的学习模型的闭环稳定性仍然具有挑战性。本文提出了一种新颖的变换器-演员-评论家MPC架构，并提供了正式的鲁棒性保证。首先，我们证明了变换器网络可以满足全局增量输入到状态稳定性（$ ext{δISS}$）。然后，我们利用黎曼收缩理论分析物理系统与预测神经网络之间的互联动态。最后，我们将这些理论界限整合为训练正则化器，以产生可证明的鲁棒策略。该框架在执行目标到达和避障机动的非线性三维无人机模型上得到了验证。
View on arXiv Download PDF AI Translation
cs.RO / 48 / 2606.20209

FlowMaps: Modeling Long-Term Multimodal Object Dynamics with Flow Matching

FlowMaps：使用流匹配建模长期多模态物体动态
Argenziano, Francesco, Saavedra-Ruiz, Miguel, Morin, Sacha, Gauthier, Charlie, Nardi, Daniele, Paull, Liam
Abstract
Joint spatial and temporal understanding of 3D scenes is a crucial requirement for robots deployed in everyday household environments. Such agents must not only comprehend and navigate spatial layouts, but also reason about how these spaces evolve over time. In particular, humans interact with objects daily, causing them to change position throughout the environment and making it difficult for robots to reliably associate current observations with previously seen objects. However, these interactions are not random: human habits and routines induce spatio-temporally consistent patterns in object locations, which robotic agents can potentially learn and then exploit for downstream tasks such as navigation. To this end, we introduce FlowMaps, a latent flow matching model for estimating multimodal distributions over the future locations of dynamic objects in a continuous 3D space. By learning the implicit dependencies among objects and their temporal evolution, FlowMaps predicts likely changes in object locations conditioned on past human interactions, while supporting generalization across previously unseen environments that share similar object routines. To demonstrate the utility of this method, we deploy FlowMaps in a downstream dynamic Object Navigation task in both simulated and real-world environments. Across more than 600 episodes, FlowMaps outperforms state-of-the-art approaches, showing that modeling object dynamics through continuous, multimodal spatio-temporal distributions improves robotic search and navigation in changing household environments. Code and additional material is available at https://fra-tsuna.github.io/flowmaps/.
Chinese Translation
对三维场景的联合空间和时间理解是部署在日常家庭环境中的机器人所必需的关键要求。这些智能体不仅必须理解和导航空间布局，还必须推理这些空间如何随时间演变。特别是，人类每天与物体互动，导致物体在环境中改变位置，使得机器人难以可靠地将当前观察与之前见过的物体关联起来。然而，这些互动并非随机：人类的习惯和日常活动在物体位置上引发了时空一致的模式，机器人智能体可以学习并利用这些模式来执行下游任务，例如导航。为此，我们提出了FlowMaps，一种潜在流匹配模型，用于估计动态物体在连续三维空间中的未来位置的多模态分布。通过学习物体之间的隐含依赖关系及其时间演变，FlowMaps能够预测基于过去人类互动的物体位置的可能变化，同时支持在具有相似物体活动的未见环境中的泛化。为了展示该方法的实用性，我们在模拟和真实环境中的下游动态物体导航任务中部署了FlowMaps。在超过600个实验中，FlowMaps的表现超越了最先进的方法，显示出通过连续的多模态时空分布建模物体动态能够改善机器人在变化的家庭环境中的搜索和导航能力。代码和附加材料可在 https://fra-tsuna.github.io/flowmaps/ 获取。
View on arXiv Download PDF AI Translation
cs.RO / 49 / 2606.20232

Mobile Target Search with Imperfect Perception: A Partially Observable Stochastic Game Theoretical Approach

基于不完美感知的移动目标搜索：一种部分可观察随机博弈理论方法
Zhang, Hanzheng, Liang, Shu, Liu, Shuyu
Abstract
This paper investigates mobile target search under imperfect perceptions caused by sensor limitations, malicious jamming, or communication noise. Searchers and targets operate in a grid-shaped area with bounded mobility, leading to a dynamic interplay between search and evasion. To capture this adversarial interaction under imperfect perceptions, we adopt the partially observable stochastic game (POSG) approach, which generalizes partially observable Markov decision processes (POMDPs) by incorporating target intelligence. To handle false alarms and missed detections caused by perceptual uncertainties, we propose a novel detectability concept to determine whether a search strategy guarantees eventual detection, and provide sufficient detectability criteria based on stochastic recurrence analysis. We further develop a server-assisted distributed algorithm that utilizes the aggregative potential game structure for searchers and a KL-divergence-based reduction for target prediction. Numerical simulations validate the effectiveness of the proposed algorithm and support the detectability analysis.
Chinese Translation
本文研究了由于传感器限制、恶意干扰或通信噪声导致的不完美感知下的移动目标搜索。搜索者和目标在一个网格状区域内进行活动，且其移动受到限制，这导致了搜索与规避之间的动态相互作用。为了捕捉这种在不完美感知下的对抗性互动，我们采用了部分可观察随机博弈（POSG）方法，该方法通过引入目标智能来推广部分可观察马尔可夫决策过程（POMDP）。为了处理由于感知不确定性引起的误报和漏检，我们提出了一种新颖的可检测性概念，以确定某一搜索策略是否保证最终检测，并基于随机重现分析提供了充分的可检测性标准。我们进一步开发了一种服务器辅助的分布式算法，该算法利用了搜索者的聚合潜力博弈结构以及基于KL散度的目标预测简化方法。数值仿真验证了所提算法的有效性，并支持了可检测性分析。
View on arXiv Download PDF AI Translation
cs.RO / 50 / 2606.20246

Finetuning Vision-Language-Action Models Requires Fewer Layers Than You Think

微调视觉-语言-动作模型所需的层数比你想象的要少
Nguyen, Gia-Binh, Ho, Trong-Bao, Ha, Thien-Loc, Vo, Khoa, Møller, Philip Lund, Nguyen, Quang T., Dinh, Long, Dam, Tuan, Duong, Vu, Luu, Tung M., Le, Trung, Le, Tran Nguyen, Vu, Minh, Le, An Thai, Le, Ngan, Sonntag, Daniel, Zou, James, Peters, Jan, Nguyen, Duy M. H., Vien, Ngo Anh
Abstract
Vision-Language-Action (VLA) models pre-trained on massive video-robot datasets have revolutionized robotic manipulation, yet their multi-billion parameter architectures impose prohibitive computational burdens during downstream fine-tuning and real-time inference. In this work, we reveal a highly non-trivial architectural characteristic of these continuous control foundation policies (e.g., pi_0, GR00T-N1.5): despite being trained on diverse physical trajectories, they exhibit severe layer-wise representational redundancy. To exploit this, we introduce a structural compression pipeline that is entirely training-free, bypassing the need of existing methods to load full-scale models to learn optimized token reductions or dynamic layer selectors. Instead, using only a single forward pass via Centered Kernel Alignment to identify redundant layer features, we remove twin layers to permanently compress the model depth by up to 50% across both the VLM backbone and the continuous control policy head. Downstream fine-tuning of this streamlined architecture yields a dual acceleration benefit: a 40-50% reduction in training time and up to 30% faster real-time inference, while matching or exceeding full-scale base model performance. We comprehensively validate our method across three simulation benchmarks (LIBERO, RoboCasa, SimplerEnv) and 10 diverse real-world manipulation tasks across 4 unique robotic embodiments. These results prove that advanced VLAs require significantly fewer layers than previously assumed, offering a highly compute-efficient paradigm for scalable robot learning.
Chinese Translation
在大规模视频-机器人数据集上预训练的视觉-语言-动作（VLA）模型已经彻底改变了机器人操作，但其数十亿参数的架构在下游微调和实时推理过程中带来了巨大的计算负担。在本研究中，我们揭示了这些连续控制基础策略（例如，pi_0，GR00T-N1.5）的一种高度非平凡的架构特征：尽管在多样的物理轨迹上进行训练，它们却表现出严重的层级表示冗余。为了利用这一点，我们提出了一种完全无训练的结构压缩管道，避免了现有方法需要加载全尺度模型以学习优化的标记减少或动态层选择器的需求。相反，仅通过中心核对齐（Centered Kernel Alignment）进行一次前向传播来识别冗余层特征，我们去除了双层，从而在VLM主干和连续控制策略头部上永久性地将模型深度压缩最多50%。这种精简架构的下游微调带来了双重加速效益：训练时间减少40-50%，实时推理速度提高最多30%，同时匹配或超越全尺度基础模型的性能。我们在三个仿真基准（LIBERO、RoboCasa、SimplerEnv）和四种独特机器人形态下的10个多样化现实世界操作任务中全面验证了我们的方法。这些结果证明，先进的VLA模型所需的层数显著少于先前的假设，为可扩展的机器人学习提供了一种高计算效率的范式。
View on arXiv Download PDF AI Translation
cs.RO / 51 / 2606.20272

Efficiently Linking Real Scenes with Synthetic Data Generation for AI-based Cognitive Robotics and Computer Vision Applications

高效链接真实场景与合成数据生成以应用于基于人工智能的认知机器人和计算机视觉
Koch, Paul, Chavan, Vivek, Sers, André, Karakurt, Adem, Hofmann, Paul, Ziadeh, Mohamad Zaher, Krüger, Jörg
Abstract
AI vision models are a driving factor for the potential use case scenarios of cognitive robotics within in the industry and household applications. A large array of methods from semantic environment analysis towards 6D and grasping pose estimation have been proposed based on the latest AI achievements. However, such advancements require further strong and efficient methods w.r.t. training data and AI-architectures, which are capable in synergy to tackle current challenges, precision limits, and scalability beyond domain gaps. In this paper, we discuss these current limits and trends in the related state-of-the-art which are challenging those. Further we discuss our current work in progress on bridging the domain gap between simulations and real world applications by linking those in the training data generation.
Chinese Translation
人工智能视觉模型是认知机器人在工业和家庭应用中潜在使用场景的重要推动因素。基于最新的人工智能成果，已经提出了从语义环境分析到六维（6D）和抓取姿态估计的多种方法。然而，这些进展需要更强大且高效的方法，特别是在训练数据和人工智能架构方面，这些方法能够协同应对当前的挑战、精度限制以及超越领域差距的可扩展性。本文讨论了当前在相关最先进技术中面临的限制和趋势，并探讨了我们在弥合模拟与现实应用之间领域差距的工作进展，特别是在训练数据生成方面的链接。
View on arXiv Download PDF AI Translation
cs.RO / 52 / 2606.20285

Co-VLA: Coordination-Aware Structured Action Modeling for Dual-Arm Vision-Language-Action Systems

Co-VLA：面向双臂视觉-语言-动作系统的协调感知结构化动作建模
Wang, Yandong, Yu, Jiaqian, Peng, Xiongfeng, Xu, Lu, Mao, Yamin, Li, Weiming, Yoo, Jaewook, Lee, Dongwook, Ji, Daehyun, Zhao, Mingbo, Zhang, Chao
Abstract
Vision-language-action (VLA) models show strong capabilities in single and dual-arm robotic manipulation. Prior works show coordinated bimanual behaviors can emerge from end-to-end learning, leveraging large vision-language backbones with continuous action prediction. However, as bimanual tasks become tightly coupled and execution constraints become critical, implicit coordination alone is insufficient to ensure reliable, interpretable, and stable behavior. In this work, we propose Co-VLA, a coordination-aware bimanual manipulation framework introducing explicit structural priors into VLA models. We instantiate our method on a state-of-the-art vision-language backbone by replacing its monolithic action head with a Structured Action Expert (SAE) designed for bimanual coordination. Specifically, we introduce explicit structure at the action generation level with a modular coordination-aware loss that shapes shared and residual latents according to task-specific structures. The shared latent encodes task-level coordination intent, while residual latents capture execution adjustments for each arm. At deployment, a Latent-Aware Controller (LAC) interprets the learned representations to modulate synchronization strength, execution asymmetry, smoothness, and safety constraints in real time. LAC operates at the joint-command level and remains compatible with standard control pipelines without requiring force or impedance control. Experiments across simulation and real-world benchmarks show Co-VLA significantly outperforms monolithic baselines, achieving a 27% success rate gain in tight-coordination tasks, more than doubling performance in OOD real-world scenarios (from 13% to 27%), and reducing task completion time by up to 25%.
Chinese Translation
视觉-语言-动作（VLA）模型在单臂和双臂机器人操控中表现出强大的能力。先前的研究表明，通过端到端学习，利用大型视觉-语言主干网络和连续动作预测，可以产生协调的双手行为。然而，随着双手任务的紧密耦合和执行约束的关键性，仅依靠隐式协调不足以确保可靠、可解释和稳定的行为。在本研究中，我们提出了Co-VLA，一个协调感知的双手操控框架，将显式结构先验引入VLA模型。我们在一个最先进的视觉-语言主干网络上实例化了我们的方法，通过用为双手协调设计的结构化动作专家（Structured Action Expert, SAE）替换其单一的动作头。具体而言，我们在动作生成层引入显式结构，采用模块化的协调感知损失，根据任务特定结构塑造共享和残余潜变量。共享潜变量编码任务级协调意图，而残余潜变量则捕捉每个手臂的执行调整。在部署时，潜变量感知控制器（Latent-Aware Controller, LAC）解读学习到的表示，以实时调节同步强度、执行不对称性、平滑度和安全约束。LAC在关节指令级别操作，并与标准控制管道兼容，无需强制或阻抗控制。跨模拟和现实世界基准的实验表明，Co-VLA显著优于单一基线，在紧密协调任务中实现了27%的成功率提升，在OOD现实世界场景中的表现翻倍（从13%提升至27%），并将任务完成时间减少了多达25%。
View on arXiv Download PDF AI Translation
cs.RO / 53 / 2606.20322

Towards 3D karst underwater scene reconstruction from rotating sonar data

基于旋转声纳数据的三维喀斯特水下场景重建
Margaritis, Georgios Evangelos, Lapierre, Lionel, Rohou, Simon, Yan, Zhi, Nüchter, Andreas, Goulette, François
Abstract
Karst aquifers provide critical freshwater resources but pose significant hazards due to their complex and poorly understood subsurface geometry. Mapping these environments is challenging because sonar data from underwater exploration is sparse and noisy, while navigation estimates suffer from drift limiting standard 3D reconstruction methods. We present a pipeline for reconstructing underwater karst conduits from a sonar profiler. We combine a continuous-time SLAM approach to correct trajectory drift with a novel two-stage deep learning method for surface reconstruction, producing an immersive and navigable 3D mesh for hydrogeological analysis.
Chinese Translation
喀斯特含水层提供了重要的淡水资源，但由于其复杂且不易理解的地下几何结构，带来了显著的危险。由于水下探测的声纳数据稀疏且噪声较大，映射这些环境具有挑战性，而导航估计受到漂移的影响，限制了标准的三维重建方法。我们提出了一种从声纳探测仪重建水下喀斯特导管的流程。我们结合了一种连续时间的同步定位与地图构建（SLAM）方法来校正轨迹漂移，并采用了一种新颖的两阶段深度学习方法进行表面重建，从而生成用于水文地质分析的沉浸式和可导航的三维网格。
View on arXiv Download PDF AI Translation
cs.RO / 54 / 2606.20336

Autonomous Driving with Priority-Ordered STL Specifications Under Multimodal Uncertainty

在多模态不确定性下具有优先顺序的 STL 规范的自主驾驶
Bouzid, Taha, Qi, Shuhao, Lazar, Mircea, Haesaert, Sofie
Abstract
Autonomous vehicles must plan trajectories that satisfy a multitude of requirements on safety, passenger comfort, and compliance with traffic rules. However, in safety-critical scenarios, it is not always possible to satisfy all requirements simultaneously, necessitating their prioritization based on importance. At the same time, in these safety-critical scenarios, the uncertainty in trajectory predictions of the surrounding traffic, such as other vehicles and pedestrians, should be explicitly accounted for. In this work, we propose an uncertainty-aware trajectory planning framework that incorporates a predefined lexicographic ordering over Signal Temporal Logic (STL) specifications that stays valid under uncertainty. We implement this formulation with Model Predictive Path Integral (MPPI) control and we demonstrate the effectiveness of our method on simulation scenarios, showing that our framework efficiently handles conflicting objectives under realistic multi-modal uncertainty.
Chinese Translation
自主车辆必须规划满足安全、乘客舒适性和遵守交通规则等多种要求的轨迹。然而，在安全关键场景中，并不总是能够同时满足所有要求，因此需要根据重要性对其进行优先排序。同时，在这些安全关键场景中，周围交通（如其他车辆和行人）的轨迹预测的不确定性也应被明确考虑。在本研究中，我们提出了一种考虑不确定性的轨迹规划框架，该框架结合了在不确定性下仍然有效的信号时序逻辑（Signal Temporal Logic, STL）规范的预定义字典序排序。我们使用模型预测路径积分（Model Predictive Path Integral, MPPI）控制实现了这一公式，并在仿真场景中展示了我们方法的有效性，表明我们的框架能够有效处理现实多模态不确定性下的冲突目标。
View on arXiv Download PDF AI Translation
cs.RO / 55 / 2606.20365

An Infrastructure-less, Control-Independent Solution to Relative Localisation of a Team of Mobile Robots using Ranging Measurements

一种无基础设施、控制独立的移动机器人团队相对定位解决方案，基于测距数据
Golinelli, Paolo, Faraci, Tommaso, Fontanelli, Daniele
Abstract
The ability to localise teams of robots is essential for applications ranging from robotic fleets in unstructured environments to cooperative control and navigation tasks. In such contexts, fixed infrastructure is often unavailable, deployments must be fast and flexible, and system requirements must be minimal. We present a decentralised cooperative localisation algorithm that addresses all these challenges at once. The method is anchor-less, fully decentralised, and, unlike most existing approaches, does not require controlling the robots motion to ensure team observability. It relies only on local odometry, sparse inter-agent ranging measurements, and short-range communication, all of which are widely available in practice. The algorithm adopts a multi-hypothesis Bayesian framework that maintains the entire set of feasible solutions, ensuring robustness under transient unobservable conditions. Moreover, through information sharing, each agent benefits from the estimates of the entire group, even in partially connected conditions.
Chinese Translation
机器人团队的定位能力对于从非结构化环境中的机器人车队到协作控制和导航任务等应用至关重要。在这种情况下，固定基础设施通常不可用，部署必须快速灵活，系统要求必须最小化。我们提出了一种去中心化的协作定位算法，能够同时应对所有这些挑战。该方法不依赖锚点，完全去中心化，并且与大多数现有方法不同，不需要控制机器人运动以确保团队可观测性。它仅依赖于局部里程计、稀疏的代理间测距数据和短距离通信，这些在实际应用中都广泛可用。该算法采用多假设贝叶斯框架，维护整个可行解集，确保在瞬态不可观测条件下的鲁棒性。此外，通过信息共享，每个代理即使在部分连接条件下也能受益于整个团队的估计。
View on arXiv Download PDF AI Translation
cs.RO / 56 / 2606.20389

CoLI: A Reproducible Platform for Continuum Robot Learning via Monolithic 3D Printing and Isomorphic Teleoperation

CoLI：通过单体3D打印和同构遥操作实现的可重复性连续机器人学习平台
Tang, Ziyuan, Xiao*, Chenxi
Abstract
Continuum robots offer strong potential for manipulation tasks due to their high degrees of freedom, compliant structures, and operational safety. However, their adoption in both research and practical applications has been hindered by reproducibility issues arising from complex fabrication and assembly processes, challenging kinematic modeling, and a lack of intuitive control interfaces. To address these challenges, we present a novel open-source continuum robot design. The platform features a simplified fabrication pipeline enabled by multi-material 3D printing, allowing the arm to be fabricated as a monolithic compliant structure with minimal assembly. Control is achieved through an isomorphic teleoperation interface that establishes a direct actuator-level mapping, eliminating the need for explicit kinematic modeling and providing a singularity-free mapping. Building on this hardware design, the platform further supports imitation-learning-based autonomous control. The proposed system is evaluated through hardware characterization and a set of manipulation tasks. Experimental results demonstrate that the platform provides a reproducible, learning-ready continuum robot system, accelerating algorithmic development and systematic benchmarking for the continuum robotics community.
Chinese Translation
连续机器人因其高自由度、顺应性结构和操作安全性，在操控任务中展现出强大的潜力。然而，由于复杂的制造和组装过程、挑战性的运动学建模以及缺乏直观的控制界面，导致其在研究和实际应用中的采用受到限制。为了解决这些挑战，我们提出了一种新颖的开源连续机器人设计。该平台采用多材料3D打印简化了制造流程，使得机械臂可以作为单体顺应结构制造，组装工作量最小。控制通过同构遥操作接口实现，该接口建立了直接的执行器级映射，消除了显式运动学建模的需要，并提供了无奇异性的映射。在这一硬件设计的基础上，该平台进一步支持基于模仿学习的自主控制。通过硬件特性分析和一系列操控任务对所提系统进行了评估。实验结果表明，该平台提供了一个可重复的、适合学习的连续机器人系统，加速了算法开发和连续机器人社区的系统基准测试。
View on arXiv Download PDF AI Translation
cs.RO / 57 / 2606.20394

Agentic AutoResearch forSpace Autonomy: An Auditable, LLM-Driven Research Agent for Aerospace Control Problems

自主研究代理在空间自主中的应用：一个可审计的、基于大型语言模型的航空航天控制问题研究代理
Jain, Amit, Linares, Richard
Abstract
Spacecraft guidance, navigation, and control functions are increasingly realized as learned policies distilled from expert solvers. Developing such a policy is itself a research process: an investigator selects an architecture and hyperparameters, runs experiments, and must determine whether an apparent improvement is genuine or merely seed noise. This paper presents AutoResearch, a framework in which a large language model autonomously drives that loop for aerospace control problems, coupled with a credibility layer, built into the loop, that certifies each reported result against the problem's own measured seed noise. The language model serves only as the offline research agent that develops the control policy; the trained policy it produces is then deployed onboard the spacecraft, while the model itself never operates the vehicle. At each iteration the agent reads a plain-language problem description and the run history, proposes a single edit to the training script, executes it, and logs the outcome. No reported result is credited until it passes the same three checks: measured per-problem seed noise, reseeded verification of the best configuration, and leave-one-out pruning of the agent's edits. The same loop is applied, unchanged, to two aerospace control problems: a Clohessy-Wiltshire relative rendezvous and a safety-constrained collision-avoidance docking past a keep-out zone, each calibrated against a known optimal control benchmark. In both, the audited policy clears the measured seed noise by many standard deviations; an undirected search over the same parameters does not. On the docking problem the gap becomes categorical: undirected search yields no feasible policy, while the learned policy stays outside the keep-out zone on every seed.
Chinese Translation
航天器的引导、导航和控制功能越来越多地被实现为从专家求解器中提炼出的学习策略。开发这样的策略本身就是一个研究过程：研究者选择架构和超参数，进行实验，并必须确定表面上的改进是真实的还是仅仅是种子噪声。本文提出了AutoResearch，一个框架，其中大型语言模型自主驱动航空航天控制问题的研究循环，并在循环中内置了一个可信度层，该层对每个报告结果进行认证，以确保其符合问题自身测量的种子噪声。语言模型仅作为离线研究代理，开发控制策略；它生成的训练策略随后被部署在航天器上，而模型本身从不操作该车辆。在每次迭代中，代理读取一个普通语言的问题描述和运行历史，提出对训练脚本的单一修改，执行该修改并记录结果。任何报告的结果在通过以下三个检查之前均不被认可：测量的每个问题的种子噪声、对最佳配置的重新验证，以及对代理修改的逐一剔除。相同的循环未作更改地应用于两个航空航天控制问题：Clohessy-Wiltshire相对会合和在禁区外进行的安全约束碰撞避免对接，每个问题都与已知的最优控制基准进行校准。在这两种情况下，经过审计的策略清除了测量的种子噪声，超过多个标准差；而在相同参数下的无指导搜索则没有。在对接问题中，差距变得明显：无指导搜索未能产生可行策略，而学习的策略在每个种子上都保持在禁区外。
View on arXiv Download PDF AI Translation
cs.RO / 58 / 2606.20424

LIT-GS: LiDAR-Inertial-Thermal Gaussian Splatting for Illumination-Robust Mapping

LIT-GS：用于照明鲁棒映射的激光雷达-惯性-热高斯喷溅
Shi, Shikuan, Zheng, Chunran, Xu, Jiaming, Ye, Tianyong, Yu, Tao, Cui, Yukang
Abstract
Gaussian Splatting has enabled real-time neural rendering, yet existing LiDAR-inertial-visual (LIV) Gaussian mapping pipelines remain fragile under illumination changes and texture-deficient scenes due to their reliance on RGB photometric cues. We present LIT-GS, a LiDAR-inertial-thermal Gaussian Splatting framework that injects LiDAR-derived plane geometry as an explicit constraint in both pose/structure refinement and Gaussian optimization. Specifically, we exploit LIV visual map points as confidence-aware cross-modal anchors to establish reliable thermal-LiDAR associations, and incorporate weighted LiDAR point-to-plane residuals into bundle adjustment to jointly refine camera poses and 3D points under weak thermal supervision. Building on the refined structure, we further introduce a LiDAR-plane-regularized differentiable splatting objective that constrains rendered 3D points to align with locally observed planes, mitigating surface thickening and structural drift in low-contrast thermal imagery. Experiments on proprietary sequences and public datasets demonstrate that LIT-GS consistently improves geometric accuracy and rendering quality over state-of-the-art LIV-based Gaussian Splatting baselines, particularly in challenging lighting conditions.
Chinese Translation
高斯喷溅技术使实时神经渲染成为可能，但现有的激光雷达-惯性-视觉（LIV）高斯映射管道在光照变化和纹理缺乏的场景下仍然脆弱，因为它们依赖于RGB光度线索。我们提出了LIT-GS，一个激光雷达-惯性-热高斯喷溅框架，该框架在姿态/结构优化和高斯优化中将激光雷达派生的平面几何作为显式约束。具体而言，我们利用LIV视觉地图点作为具有置信度的跨模态锚点，以建立可靠的热-激光雷达关联，并将加权的激光雷达点到平面的残差纳入束调整中，以在弱热监督下共同优化相机姿态和三维点。在优化后的结构基础上，我们进一步引入了一个激光雷达平面正则化的可微分喷溅目标，该目标约束渲染的三维点与局部观察到的平面对齐，从而减轻低对比度热图像中的表面增厚和结构漂移。在专有序列和公共数据集上的实验表明，LIT-GS在几何精度和渲染质量上始终优于最先进的基于LIV的高斯喷溅基线，尤其是在具有挑战性的光照条件下。
View on arXiv Download PDF AI Translation
cs.RO / 59 / 2606.20426

TaCauchy: An Extensible FEM Framework for Vision-Based Tactile Simulation

TaCauchy：一个可扩展的基于视觉的触觉仿真有限元框架
Zhao, Hengfei, Xie, Yifan, Gong, Junhao, Sun, Yue, Zhu, Kai, He, Weihua, Li, Shoujie, Fu, Haohuan, Ding, Wenbo
Abstract
Vision-based tactile sensors require high-fidelity simulation for reinforcement learning, yet existing approaches struggle to provide accurate mechanical stress fields within GPU-accelerated robotics platforms. We present TaCauchy, an extensible Finite Element Method (FEM) framework that integrates rigorous physics-based force computation into Isaac Sim. Built on the Unified Incremental Potential Contact (UIPC) solver, TaCauchy directly computes Cauchy stress tensors from hyperelastic constitutive laws and projects them onto contact surfaces to obtain traction forces and pressure distributions, providing mechanical ground truth from first principles rather than empirical estimation. Our framework features automatic mesh generation with geometry-aware adaptive refinement and a modular sensor interface enabling rapid integration of diverse sensors (GelSight Mini, DIGIT, 9DTact) with minimal configuration. Performance benchmarks demonstrate 33.40 FPS for single environments and 555 FPS aggregate throughput across 60 parallel environments, with stress extraction overhead under 1 ms. Physical validation experiments show strong agreement between simulated and real tactile responses across force ranges from 1.2556 N to 4.7332 N, achieving SSIM above 0.93, confirming the framework's capability to provide accurate, physically-grounded force supervision for downstream robotic manipulation tasks.
Chinese Translation
基于视觉的触觉传感器需要高保真度的仿真以进行强化学习，然而现有方法在GPU加速的机器人平台上难以提供准确的机械应力场。我们提出了TaCauchy，一个可扩展的有限元方法（FEM）框架，将严格的基于物理的力计算集成到Isaac Sim中。TaCauchy基于统一增量潜在接触（Unified Incremental Potential Contact, UIPC）求解器，直接从超弹性本构定律计算Cauchy应力张量，并将其投影到接触表面以获得牵引力和压力分布，从第一原理提供机械真实值，而非经验估计。我们的框架具有自动网格生成、几何感知自适应细化功能，以及模块化传感器接口，能够快速集成多种传感器（如GelSight Mini、DIGIT、9DTact），配置要求最低。性能基准测试表明，在单个环境中可达到33.40 FPS，在60个并行环境中的总吞吐量为555 FPS，压力提取的开销低于1毫秒。物理验证实验显示，在1.2556 N至4.7332 N的力范围内，仿真与真实触觉响应之间具有良好的一致性，结构相似性指数（SSIM）超过0.93，确认了该框架在为下游机器人操作任务提供准确、基于物理的力监督方面的能力。
View on arXiv Download PDF AI Translation
cs.RO / 60 / 2606.20428

ARC: Adaptive Robust Joint State and Covariance Estimation

ARC：自适应鲁棒联合状态与协方差估计
Hadji-Thomas, Alexandre, Stirling, Andrew, Forbes, James R.
Abstract
Sensor measurements are frequently corrupted by outliers and non-Gaussian noise. These imperfections in the sensor data can cause classical state estimators to generate biased and unreliable state and uncertainty estimates. Robust estimators reject or downweight outliers but do not perform measurement covariance estimation, whereas joint state and covariance estimators assume Gaussian residuals and fixed loss shape parameters. Integrating these two capabilities into a single framework is an opportunity to simultaneously estimate both state and covariance in the presence of outliers. This paper proposes a unified Block-Coordinate Descent framework that combines a norm-aware adaptive robust loss, an Iteratively Reweighted Least-Squares state update, and a Minimum Weighted Covariance Determinant covariance estimator, yielding a self-tuning joint state and covariance estimator. The framework is evaluated in a Monte-Carlo simulation and on real-world ultra-wideband localization experiments in cluttered non-line-of-sight environments. Results show that the proposed estimator consistently recovers the true inlier measurement covariance and matches or exceeds the state estimation accuracy of all baselines, without requiring any manual parameter tuning.
Chinese Translation
传感器测量常常受到异常值和非高斯噪声的干扰。这些传感器数据中的缺陷可能导致经典状态估计器生成有偏且不可靠的状态和不确定性估计。鲁棒估计器会拒绝或降低异常值的权重，但不进行测量协方差估计，而联合状态和协方差估计器则假设残差服从高斯分布且损失形状参数固定。将这两种能力整合到一个单一框架中，为在存在异常值的情况下同时估计状态和协方差提供了机会。本文提出了一种统一的块坐标下降框架，结合了规范感知的自适应鲁棒损失、迭代加权最小二乘状态更新和最小加权协方差行列式协方差估计器，从而实现自调节的联合状态和协方差估计器。该框架在蒙特卡洛仿真和真实世界的超宽带定位实验中进行了评估，实验环境为杂乱的非视距环境。结果表明，所提估计器始终能够恢复真实的内点测量协方差，并且其状态估计精度与所有基线相匹配或超过，无需任何手动参数调优。
View on arXiv Download PDF AI Translation
cs.RO / 61 / 2606.20458

Slow Brain, Fast Planner: Latency-Resilient VLM-Augmented Urban Navigation

慢速大脑，快速规划者：抗延迟的VLM增强城市导航
Peng, Zhenghao "Mark'', He, Honglin, Li, Quanyi, Ma, Yukai, Zhou, Bolei
Abstract
Learning-based planners for sidewalk navigation can generate diverse candidate trajectories in real time, yet their scoring functions often fail to select the best trajectory in challenging situations, outputting trajectories that make the mobile robot drive onto grass, toward pedestrians, or in the wrong direction, even when better candidates exist in the same set. We call this the trajectory scoring gap: in real-world sidewalk navigation, the gap between an anchor-based planner's top choice and the best possible candidate is substantial, likely due to limited high-level scene understanding capability of the planner. Rather than replacing the planner with an end-to-end Vision-Language-Action model, we propose a VLM-Planner interface that uses a VLM to select a candidate index from the planner's proposal set and then fuse it with the planner's initial output. However, VLMs take 1--3s per query and so cannot directly drive a 5--20Hz control loop. We contribute a training-free, latency-resilient trajectory-level fusion layer that turns a stale VLM selection into real-time planner scoring via geometric similarity with exponential decay. On $\sim$2,000 challenging real-world scenarios (e.g., junctions, pedestrian encounters), VLM selection achieves 30% ADE reduction versus the planner's best selection, while the planner remains competitive in routine situations. In simulation, Score Fusion maintains >80% success rate with delays up to 5s. We demonstrate the full system on a mobile robot navigating challenging campus sidewalks with varied network latency.
Chinese Translation
基于学习的 sidewalk 导航规划器能够实时生成多样的候选轨迹，然而它们的评分函数在复杂情况下往往无法选择最佳轨迹，输出的轨迹可能导致移动机器人驶入草地、朝向行人或朝错误方向行驶，即使在同一组中存在更好的候选轨迹。我们称之为轨迹评分差距：在现实世界的 sidewalk 导航中，基于锚点的规划器的最佳选择与最佳可能候选之间的差距是相当大的，这可能是由于规划器的高层场景理解能力有限。我们并不打算用端到端的视觉-语言-行动（Vision-Language-Action, VLM）模型替代规划器，而是提出了一种 VLM-规划器接口，该接口使用 VLM 从规划器的提议集中选择候选索引，然后将其与规划器的初始输出融合。然而，VLM 每次查询需要 1-3 秒，因此无法直接驱动 5-20Hz 的控制循环。我们贡献了一个无训练、抗延迟的轨迹级融合层，通过与指数衰减的几何相似性将过时的 VLM 选择转化为实时规划器评分。在大约 2000 个具有挑战性的现实场景（例如交叉口、行人遇见）中，VLM 选择相比于规划器的最佳选择实现了 30% 的平均距离误差（ADE）减少，同时规划器在常规情况下仍然具有竞争力。在仿真中，评分融合在延迟高达 5 秒的情况下保持超过 80% 的成功率。我们在一台移动机器人上展示了完整系统，该机器人在具有不同网络延迟的复杂校园人行道上导航。
View on arXiv Download PDF AI Translation
cs.RO / 62 / 2606.20479

GroundControl: Anticipating Navigation Failures in Vision-Language Agents via Trajectory-Consistent Uncertainty Estimates

GroundControl：通过轨迹一致的不确定性估计预测视觉语言代理中的导航失败
Darabi, Nastaran, Kumar, Divake, Tayebati, Sina, Naik, Devashri, Trivedi, Amit Ranjan
Abstract
Vision-language navigation agents achieve competitive average success on benchmark tasks, yet failures often arise through predictable trajectory-level breakdowns such as oscillation, stagnation, or inefficient detours. Reliable deployment, therefore, requires uncertainty signals that anticipate emerging failure dynamics during execution rather than reflect only instantaneous action entropy. We introduce \emph{GroundControl}, a trajectory-consistent uncertainty estimator defined as statistical deviation from nominal goal-directed distance-to-goal dynamics aggregated over an episode. GroundControl models distance evolution using a constant-velocity Kalman filter and combines normalized innovation statistics with complementary trajectory features capturing progress, monotonicity, path efficiency, and oscillatory behavior. The resulting uncertainty score reflects geometric and temporal inconsistency in navigation behavior rather than local prediction dispersion. To evaluate uncertainty quality independently of task success, we formalize \emph{Selective Risk--Coverage Navigation (SRCN)}, a protocol that measures how effectively an uncertainty score ranks episodes by failure or inefficiency using risk--coverage curves and AURC / E-AURC summaries. Across five EB-Navigation splits ($N=300$ episodes), trajectory-consistent uncertainty achieves near-oracle ordering under success-based selective risk, with weighted-average $\mathrm{E\text{-}AURC}_{\mathrm{SR}}=0.0024$ for the GPT-4o model, substantially outperforming entropy-, conformal-, and heuristic baselines. Under SPL-based selective evaluation, GroundControl consistently achieves the lowest AURC and E-AURC across models and navigation splits. These results show that modeling deviation from goal-directed dynamics provides an interpretable and robust signal for anticipating navigation failures in vision-language agents.
Chinese Translation
视觉语言导航代理在基准任务中取得了竞争性的平均成功率，但失败往往通过可预测的轨迹级别故障如振荡、停滞或低效绕行而发生。因此，可靠的部署需要能够在执行过程中预测即将出现的失败动态的不确定性信号，而不仅仅是反映瞬时的行动熵。我们引入了 extit{GroundControl}，一种轨迹一致的不确定性估计器，定义为在一个回合中对名义目标导向的距离动态的统计偏差。GroundControl使用恒速卡尔曼滤波器建模距离演变，并结合归一化的创新统计与互补的轨迹特征，捕捉进度、单调性、路径效率和振荡行为。最终的不确定性得分反映了导航行为中的几何和时间不一致性，而不是局部预测的离散性。为了独立于任务成功评估不确定性质量，我们形式化了 extit{选择性风险-覆盖导航（SRCN）}协议，该协议测量不确定性得分如何有效地根据失败或低效对回合进行排名，使用风险-覆盖曲线和AURC / E-AURC摘要。在五个EB-Navigation拆分（$N=300$回合）中，轨迹一致的不确定性在基于成功的选择性风险下实现了接近oracle的排序，GPT-4o模型的加权平均$ ext{E-AURC}_{ ext{SR}}=0.0024$，显著优于熵、符合性和启发式基线。在基于SPL的选择性评估下，GroundControl在所有模型和导航拆分中始终实现最低的AURC和E-AURC。这些结果表明，建模与目标导向动态的偏差提供了一种可解释且稳健的信号，用于预测视觉语言代理中的导航失败。
View on arXiv Download PDF AI Translation
cs.RO / 63 / 2606.20491

Fast Human Attention Prediction for Fixation-guided Active Perception in Autonomous Navigation

用于自主导航中基于注视引导的主动感知的快速人类注意力预测
Mohammed, Fatma Youssef, Malczyk, Grzegorz, Alexis, Kostas
Abstract
Human visual attention relies on structured scanpaths to efficiently process scenes, yet instilling this behavior into robot autonomy is in its infancy and hindered by the high,computational costs of existing predictive models. To address this, we introduce GazeLNN, a computationally lightweight,scanpath prediction model that leverages Liquid Neural Networks as its recurrent engine and employs MobileNetV3 for feature extraction. Operating auto-regressively, the architecture predicts sequential fixation heatmaps conditioned on the current visual stimulus and fixation history. Despite requiring only 0.61 GFLOPs, GazeLNN achieves state-of-the-art performance on the MIT Low Resolution dataset achieving 0.47 ScanMatch score. It outperforms existing recurrent baselines across diverse evaluation metrics, while reducing computational costs by 99.40% and accelerating inference by up to six times. To investigate the role of human attention modeling in robot autonomy and demonstrate the practical utility of this highly efficient architecture, we integrate GazeLNN into an active camera-robot control policy trained via Reinforcement Learning. This integration enables human-fixation-guided perception during autonomous navigation, validated through successful real-world deployments on an aerial robot.
Chinese Translation
人类视觉注意力依赖于结构化的扫描路径来高效处理场景，但将这种行为植入机器人自主性仍处于起步阶段，并受到现有预测模型高计算成本的限制。为了解决这一问题，我们提出了GazeLNN，这是一种计算轻量级的扫描路径预测模型，利用液态神经网络（Liquid Neural Networks）作为其递归引擎，并采用MobileNetV3进行特征提取。该架构以自回归方式运行，预测基于当前视觉刺激和注视历史的连续注视热图。尽管仅需0.61 GFLOPs，GazeLNN在MIT低分辨率数据集上实现了0.47的ScanMatch得分，达到了最先进的性能。它在多种评估指标上超越了现有的递归基线，同时将计算成本降低了99.40%，并将推理速度提高了最多六倍。为了探讨人类注意力建模在机器人自主性中的作用，并展示这一高效架构的实际应用，我们将GazeLNN集成到通过强化学习训练的主动相机-机器人控制策略中。这一集成使得在自主导航过程中能够实现基于人类注视的感知，经过在空中机器人上的成功实地部署得到了验证。
View on arXiv Download PDF AI Translation
cs.RO / 64 / 2606.20549

Generating Robot Hands from Human Demonstrations

从人类示范生成机器人手
Yi, Sha, Hansen, Nicklas, Bai, Xueqian, Sferrazza, Carmelo, Tolley, Michael T., Wang, Xiaolong
Abstract
Robot learning has advanced rapidly in learning control, but learning the physical body of a robot remains much more difficult because jointly searching over design and control creates a very large combinatorial problem. Here, we present a data-driven framework for generating robot hands from human demonstrations. Instead of learning a complex controller together with each candidate design, we generate robot hand designs using the same simple control policy used after fabrication: matching fingertip positions through inverse kinematics. Using more than 4 million frames of human fingertip motion from everyday manipulation, our algorithm optimizes tree-structured robot hands to reproduce desired target motions. The framework produced both a 6-degree-of-freedom (DoF) general-purpose hand and lower-DoF task-specific hands with spatial four-bar mimic joints. To accelerate the search over designs, we trained a reinforcement-learning (RL) actor to propose good hand designs and joint angles, reducing search time from hours to minutes. We fabricated the mechanisms directly as one-piece articulated structures with print-in-place joints. In real-world experiments, the 6-DoF hand achieved highly accurate teleoperated fingertip tracking better than available commercial robot hands, whereas the specialized 3-DoF hands reproduced structured human and synthetic trajectories with reduced mechanical complexity. These results showed that large-scale human motion data can be used not only to train robot controllers but also as a reference for optimizing and generating the physical embodiment of robots.
Chinese Translation
机器人学习在控制学习方面取得了快速进展，但学习机器人的物理形态仍然困难重重，因为在设计和控制之间进行联合搜索会产生一个非常大的组合问题。在这里，我们提出了一种基于数据的框架，用于从人类示范中生成机器人手。我们并不试图与每个候选设计一起学习一个复杂的控制器，而是使用与制造后相同的简单控制策略生成机器人手设计：通过逆向运动学匹配指尖位置。我们的算法利用超过400万个日常操作中人类指尖运动的帧，优化树状结构的机器人手，以重现期望的目标运动。该框架生成了一个6自由度（DoF）通用手和具有空间四杆模仿关节的低自由度任务特定手。为了加速设计搜索，我们训练了一个强化学习（RL）演员来提出良好的手设计和关节角度，将搜索时间从数小时减少到数分钟。我们将这些机制直接制造为一体式的关节结构，具有就地打印的关节。在实际实验中，6自由度手实现了比现有商业机器人手更高精度的遥控指尖跟踪，而专用的3自由度手则以较低的机械复杂性重现了结构化的人类和合成轨迹。这些结果表明，大规模的人类运动数据不仅可以用于训练机器人控制器，还可以作为优化和生成机器人物理形态的参考。
View on arXiv Download PDF AI Translation
cs.RO / 65 / 2606.20562

MemoryWAM: Efficient World Action Modeling with Persistent Memory

MemoryWAM：具有持久内存的高效世界动作建模
Yang, Sizhe, Mu, Juncheng, Wei, Tianming, Lu, Chenhao, Li, Xiaofan, Xu, Linning, Xue, Zhengrong, Yuan, Zhecheng, Lin, Dahua, Pang, Jiangmiao, Xu, Huazhe
Abstract
Robust robotic manipulation in the real world requires not only an understanding of the current observation, but also memory and dynamics modeling. World action models (WAMs) possess these capabilities by jointly modeling visual foresight and actions conditioned on both current and historical observations, making them a promising paradigm for robotic manipulation. However, existing WAMs face a fundamental trade-off: methods with efficient inference typically condition only on a bounded window of recent observations and therefore struggle in non-Markovian environments, whereas methods that preserve long histories incur time and space costs that grow substantially with sequence length. To address this challenge, we introduce MemoryWAM, a world action model with efficient persistent memory. MemoryWAM uses a hybrid memory design that combines recent frames, event-boundary anchor frames, and compact gist tokens that summarize long-range history. A tailored attention mechanism enables retrieval of both detailed short-term context and compressed long-term context, supporting memory-dependent decision-making with reduced inference latency and GPU memory usage. Across long-horizon, memory-dependent manipulation tasks in both simulation and the real world, MemoryWAM outperforms strong vision-language-action (VLA) and WAM baselines while maintaining favorable computational efficiency.
Chinese Translation
在现实世界中，稳健的机器人操作不仅需要对当前观察的理解，还需要记忆和动态建模。世界动作模型（WAM）通过联合建模视觉预见和基于当前及历史观察的动作，具备了这些能力，使其成为机器人操作的有前景的范式。然而，现有的WAM面临一个基本的权衡：高效推理的方法通常仅依赖于有限的近期观察窗口，因此在非马尔可夫环境中表现不佳，而保留长历史的方法则会随着序列长度的增加而产生显著的时间和空间成本。为了解决这一挑战，我们提出了MemoryWAM，一种具有高效持久内存的世界动作模型。MemoryWAM采用混合内存设计，结合了近期帧、事件边界锚帧和总结长程历史的紧凑要点（gist）标记。定制的注意机制使得能够检索详细的短期上下文和压缩的长期上下文，从而支持依赖记忆的决策制定，同时减少推理延迟和GPU内存使用。在模拟和现实世界中的长时间跨度、依赖记忆的操作任务中，MemoryWAM在保持良好计算效率的同时，超越了强大的视觉-语言-动作（VLA）和WAM基线。
View on arXiv Download PDF AI Translation
计算机视觉 (Computer Vision)
101
cs.CV / 1 / 2606.19460

Scaling Generative Foundation Models for Chest Radiography with Rectified Flow Transformers

使用校正流变换器扩展胸部放射影像的生成基础模型
Ribeiro, Fabio De Sousa, Stanley, Emma A. M., Jones, Charles, Xia, Tian, Marshall, Dominic C., Triché, Laurent Renard, Cosgriff, Christopher V., Dimitrakopoulos, Panagiotis, Tsaftaris, Sotirios A., Glocker, Ben
Abstract
We introduce the first generative foundation model for chest radiograph synthesis trained from scratch at the billion-parameter scale. Existing radiographic AI models often suffer from poor generalisation across patient subpopulations, institutions, and acquisition settings, resulting in limited real-world clinical utility. Controlled, high-fidelity synthesis of chest radiographs is a promising path toward diversifying clinical datasets and evaluating the robustness of diagnostic models. Therefore, we present the largest specialist generative foundation model for chest radiographs to date, with over 1.3B parameters, trained for 1.6T tokens on a curated, heterogeneous dataset comprising 1.2M radiographs and clinical expert-guided metadata. Our model supports controllable radiograph generation and editing across multiple demographic subgroups, acquisition views, and a dozen pathologies. Moreover, we significantly advance the state of the art in radiograph synthesis fidelity, producing images that are indistinguishable from real radiographs to clinical experts.
Chinese Translation
我们介绍了第一个从零开始训练的胸部放射影像合成生成基础模型，参数规模达到十亿级。现有的放射影像人工智能模型通常在不同患者亚群体、机构和采集环境之间的泛化能力较差，导致其在实际临床中的应用有限。对胸部放射影像进行可控的高保真合成是多样化临床数据集和评估诊断模型鲁棒性的有希望的途径。因此，我们提出了迄今为止最大的胸部放射影像专用生成基础模型，参数超过13亿，基于一个经过精心策划的异构数据集训练，包含120万幅放射影像和临床专家指导的元数据。我们的模型支持在多个人口亚组、采集视角和十几种病理情况下进行可控的放射影像生成和编辑。此外，我们在放射影像合成保真度方面显著提升了技术水平，生成的图像与真实放射影像在临床专家眼中几乎无法区分。
View on arXiv Download PDF AI Translation
cs.CV / 2 / 2606.19483

LEAP: Layer-skipping Efficiency via Adaptive Progression for Vision Transformer Distillation

LEAP：通过自适应进程实现视觉变换器蒸馏的层跳过效率
Zhang, Jiaqi, Lee, Ashton, Wong, Anthony, Zou, John, BuGhanem, Sami, Balestriero, Randall
Abstract
Vision Foundation Models (VFMs) with Vision Transformer (ViT) backbones, such as DINOv2, have become essential for downstream tasks like object recognition and semantic segmentation. The immense computational requirements of backbones often necessitate distillation into smaller architectures for edge deployment. Feature-based knowledge distillation (KD) often suffers from the teacher-student gap; the student struggles to imitate teacher's complex feature map due to its limited capacity. To mitigate this bottleneck, we propose LEAP: Layer-skipping Efficiency via Adaptive Progression, a training curriculum for ViT feature-based knowledge distillation. By utilizing the teacher's intermediate feature maps as a sequence of progressively more difficult targets, our curriculum allows the student to build a foundational representation before tackling higher-level abstractions. Our results demonstrate that this paradigm significantly accelerates convergence through adaptive difficulty selection across various student model sizes and dataset scales. With our curriculum, the LEAP-distilled ViT-S achieves 90.1% accuracy on ImageNet-100, a +12.24% improvement compared with baseline. On ImageNet-1K, LEAP achieves +3.84% and +7.75% improvement for the instance retrieval task on the Oxford and Paris datasets, respectively. Furthermore, the curriculum enables 25.1% savings in training FLOPs and 21% savings in training time on ImageNet-100 by implementing early-stopping for teacher inference during the initial stages of training. Code is available at https://github.com/KevinZ0217/LEAP
Chinese Translation
具有视觉变换器（ViT）骨干网的视觉基础模型（VFM），如DINOv2，已成为对象识别和语义分割等下游任务的关键。骨干网的巨大计算需求常常需要将其蒸馏为更小的架构以便在边缘设备上部署。基于特征的知识蒸馏（KD）通常面临教师与学生之间的差距；由于学生的容量有限，难以模仿教师复杂的特征图。为了解决这一瓶颈，我们提出了LEAP：通过自适应进程实现层跳过效率，这是一种用于ViT基于特征的知识蒸馏的训练课程。通过将教师的中间特征图作为一系列逐渐更难的目标，我们的课程使学生能够在应对更高层次的抽象之前，建立基础表示。我们的结果表明，这种范式通过在不同学生模型规模和数据集规模之间自适应选择难度，显著加速了收敛。通过我们的课程，LEAP蒸馏的ViT-S在ImageNet-100上达到了90.1%的准确率，比基线提高了+12.24%。在ImageNet-1K上，LEAP在牛津和巴黎数据集的实例检索任务中分别实现了+3.84%和+7.75%的提升。此外，该课程通过在训练初期实施教师推理的早停，节省了ImageNet-100训练中的25.1% FLOPs和21%的训练时间。代码可在https://github.com/KevinZ0217/LEAP获取。
View on arXiv Download PDF AI Translation
cs.CV / 3 / 2606.19495

LooseControlVideo: Directorial Video Control using Spatial Blocking

LooseControlVideo：基于空间阻挡的导演视频控制
Bhat, Shariq Farooq, Mitra, Niloy J., Sunkavalli, Kalyan
Abstract
Precise 3D spatial orchestration in text-to-video generation remains a significant challenge, particularly for multi-object scenes where semantic layout and temporal dynamics are often entangled. While existing depth-conditioned models achieve good structural fidelity, they necessitate dense, frame-accurate guidance that is labor-intensive to author for dynamic events involving deformable objects. We present LooseControlVideo, a framework that enables intuitive and expressive control by using sparse, oriented 3D boxes as a "blocking" proxy. This allows users to author high-level layout and trajectory while leveraging a video generative model to generate realistic occlusions, dynamics and interactions. We achieve this by fine-tuning a Wan 2.2 backbone on a video dataset annotated with DNOCS, a novel encoding for 3D size, orientation and depth-ordered occlusions. Furthermore, our method allows for localized refinement, such as adjusting a jump trajectory or adding an interaction, with minimal disruption to the global scene context. Extensive evaluations on the nuScenes, HO-3D, and BEHAVE benchmarks demonstrate that LooseControlVideo significantly outperforms existing 2D-box and flow-based baselines. Our findings indicate a 1.2x to 3x improvement in Trajectory Error; 2x improvement in Rigid Motion Consistency; and a 1.5x to 2x increase in Occlusion Accuracy over current state-of-the-art layout-conditioned models, demonstrating that oriented 3D primitives provide good geometric prior for complex, multi-agent video authoring.
Chinese Translation
在文本到视频生成中，精确的三维空间编排仍然是一个重大挑战，尤其是在多对象场景中，语义布局和时间动态往往交织在一起。虽然现有的深度条件模型在结构保真度上表现良好，但它们需要密集的、逐帧准确的指导，这对于涉及可变形物体的动态事件来说，创作过程非常繁琐。我们提出了LooseControlVideo，一个通过使用稀疏的、定向的三维盒子作为“阻挡”代理来实现直观和富有表现力的控制的框架。这使得用户能够在利用视频生成模型生成真实的遮挡、动态和交互的同时，创作高层次的布局和轨迹。我们通过对一个使用DNOCS注释的视频数据集微调Wan 2.2主干网络来实现这一目标，DNOCS是一种用于三维大小、方向和深度有序遮挡的新型编码。此外，我们的方法允许局部细化，例如调整跳跃轨迹或添加交互，而对全局场景上下文的干扰最小。在nuScenes、HO-3D和BEHAVE基准上的广泛评估表明，LooseControlVideo显著优于现有的2D盒子和基于流的方法。我们的研究结果表明，轨迹误差改善了1.2倍到3倍；刚性运动一致性改善了2倍；遮挡准确性提高了1.5倍到2倍，超越了当前最先进的基于布局的模型，证明了定向三维原语为复杂的多代理视频创作提供了良好的几何先验。
View on arXiv Download PDF AI Translation
cs.CV / 4 / 2606.19531

ImageWAM: Do World Action Models Really Need Video Generation, or Just Image Editing?

ImageWAM：世界动作模型真的需要视频生成，还是仅仅需要图像编辑？
Zhang, Yuyang, Zhang, Wenyao, Qi, Zekun, Zhang, He, Lin, Haitao, Zhang, Jingbo, Mu, Yao, Yang, Xiaokang, Zeng, Wenjun, Jin, Xin
Abstract
World Action Models (WAMs) commonly rely on video generation to bridge visual world modeling and robot control. However, video-based WAMs face three coupled limitations: dense multi-frame future tokens make inference costly, full video prediction spends capacity on action-irrelevant temporal and appearance details, and long-horizon future imagination may introduce errors that mislead action prediction. These issues raise a simple question: Does world action model really need video generation? We propose ImageWAM, a simple WAM framework that repurposes pretrained image editing models for robot action prediction. In contrast to video generation, image editing provides a better-matched prior: it only needs to model a target-frame transformation, focuses on action-relevant current-to-target visual differences, and grounds task instructions to localized visual changes through edit pretraining. In practice, ImageWAM does not decode the target frame at inference time; instead, it conditions a flow-matching action expert on the KV caches produced by image-editing denoising, using them as a compact world-action context. ImageWAM outperforms standard VLA baselines and matching competitive WAMs without additional policy pretraining across different simulator and real-world experiments. It also reduces FLOPs to 1/6 and latency to 1/4 of video-based WAMs. Attention analysis further shows that editing caches focus on task-relevant change regions, supporting image editing as an effective alternative to video-based world-action modeling.
Chinese Translation
世界动作模型（WAMs）通常依赖视频生成来桥接视觉世界建模与机器人控制。然而，基于视频的WAM面临三个相互关联的限制：密集的多帧未来标记使推理成本高昂，完整的视频预测将容量消耗在与动作无关的时间和外观细节上，而长时间范围的未来想象可能引入误差，误导动作预测。这些问题引发了一个简单的问题：世界动作模型真的需要视频生成吗？我们提出了ImageWAM，这是一个简单的WAM框架，重新利用预训练的图像编辑模型进行机器人动作预测。与视频生成相比，图像编辑提供了更匹配的先验：它只需要建模目标帧的转换，专注于与动作相关的当前与目标视觉差异，并通过编辑预训练将任务指令与局部视觉变化相结合。在实践中，ImageWAM在推理时并不解码目标帧；相反，它将流匹配的动作专家条件于由图像编辑去噪生成的KV缓存，利用它们作为紧凑的世界动作上下文。ImageWAM在不同的模拟器和现实世界实验中超越了标准的VLA基线和匹配的竞争WAM，而无需额外的策略预训练。它还将FLOPs减少到视频基础WAM的1/6，将延迟减少到1/4。注意力分析进一步表明，编辑缓存专注于与任务相关的变化区域，支持图像编辑作为视频基础世界动作建模的有效替代方案。
View on arXiv Download PDF AI Translation
cs.CV / 5 / 2606.19534

PerceptionDLM: Parallel Region Perception with Multimodal Diffusion Language Models

PerceptionDLM：基于多模态扩散语言模型的并行区域感知
Sun, Yueyi, Wang, Yuhao, Li, Jason, Tian, Ye, Zhang, Tao, Mai, Jacky, Wang, Yihan, Wang, Haochen, Bai, Jinbin, Yang, Ling, Tong, Yunhai
Abstract
Multimodal large language models (MLLMs) have achieved remarkable progress in visual understanding tasks. However, most existing MLLMs rely on autoregressive generation, which limits their efficiency for perception tasks that require captioning multiple regions. In this work, we propose PerceptionDLM, a multimodal diffusion language model optimized for efficient parallel region perception. Built upon PerceptionDLM-Base, a strong foundational baseline that achieves state-of-the-art performance among open-source diffusion MLLMs, our architecture fully leverages the parallel decoding nature of DLMs. Specifically, we introduce efficient prompting and structured attention masking to enable simultaneous perception of multiple masked regions, allowing the model to generate region descriptions in parallel at both the sequence and token levels. This design significantly improves inference efficiency compared with existing approaches that process regions sequentially. To systematically evaluate the parallelism property of visual perception capability for DLMs, we construct a new Parallel Detailed Localized Captioning Benchmark (ParaDLC-Bench) by scaling the DLC-Bench to include multiple region masks per image, enabling joint evaluation of both caption quality and inference efficiency. Experiments demonstrate that PerceptionDLM maintains competitive performance in region captioning while achieving substantial speed improvements for multi-region perception tasks. Our results highlight the potential of multimodal diffusion language models for efficient, parallel visual perception. To the best of our knowledge, we are the first to achieve parallel region caption and perception by leveraging the advantages of diffusion language models. Code, models, and datasets are released.
Chinese Translation
多模态大型语言模型（MLLMs）在视觉理解任务中取得了显著进展。然而，现有的大多数MLLMs依赖自回归生成，这限制了它们在需要对多个区域进行描述的感知任务中的效率。在本研究中，我们提出了PerceptionDLM，一种针对高效并行区域感知优化的多模态扩散语言模型。基于PerceptionDLM-Base，这是一个在开源扩散MLLMs中实现最先进性能的强大基础基线，我们的架构充分利用了DLMs的并行解码特性。具体而言，我们引入了高效的提示和结构化注意力掩蔽，以实现对多个被掩蔽区域的同时感知，使模型能够在序列和标记级别上并行生成区域描述。与现有的顺序处理区域的方法相比，这种设计显著提高了推理效率。为了系统性地评估DLMs的视觉感知能力的并行性，我们构建了一个新的并行详细本地化描述基准（ParaDLC-Bench），通过扩展DLC-Bench以包括每幅图像的多个区域掩蔽，实现对描述质量和推理效率的联合评估。实验表明，PerceptionDLM在区域描述方面保持了竞争性能，同时在多区域感知任务中实现了显著的速度提升。我们的结果突显了多模态扩散语言模型在高效并行视觉感知中的潜力。据我们所知，我们是首个通过利用扩散语言模型的优势实现并行区域描述和感知的研究。代码、模型和数据集已发布。
View on arXiv Download PDF AI Translation
cs.CV / 6 / 2606.19565

Mix-QVLA: Task-Evidence-Aware Mixed-Precision Quantization of Vision-Language-Action Models

Mix-QVLA：任务证据感知的视觉-语言-动作模型混合精度量化
Ranjan, Navin, Savakis, Andreas
Abstract
We propose Mix-QVLA, a task-evidence-aware mixed-precision PTQ framework for VLA models. Mix-QVLA anchors each quantized variant to the full-precision action-token reference decision and evaluates whether quantization preserves task-relevant evidence across key VLA functional boundaries. It computes normalized gradient-weighted task-evidence maps from boundary activations and compares full-precision and quantized maps using evidence-mass and attribution-distribution distortion, capturing changes in both the strength and allocation of decision-supporting evidence. A soft-bottleneck objective aggregates boundary-level degradation into layer-wise sensitivity scores. Mix-QVLA further models sensitivity throughout task execution, capturing phase-dependent shifts in layer importance rather than assuming a fixed sensitivity profile. The resulting evidence- and time-aware scores guide mixed-precision bit allocation under model-size and BitOps budgets. Extensive evaluations on OpenVLA-style policies show that Mix-QVLA improves the accuracy-efficiency trade-off of low-bit VLA deployment. On LIBERO, Mix-QVLA reduces OpenVLA-OFT memory from 15.4 GB to 4.1 GB, retains 96.3 average success compared with 97.1 for the BF16 model, and achieves a 1.52x inference speedup.
Chinese Translation
我们提出了Mix-QVLA，这是一种针对视觉-语言-动作（VLA）模型的任务证据感知混合精度后训练量化（PTQ）框架。Mix-QVLA将每个量化变体锚定到全精度动作标记参考决策，并评估量化是否在关键VLA功能边界之间保留了与任务相关的证据。它从边界激活中计算归一化的梯度加权任务证据图，并使用证据质量和归因分布失真比较全精度和量化图，捕捉决策支持证据的强度和分配的变化。一个软瓶颈目标将边界级别的降级聚合为层级敏感性分数。Mix-QVLA进一步在任务执行过程中建模敏感性，捕捉层重要性的相位依赖性变化，而不是假设固定的敏感性特征。最终得到的证据和时间感知分数在模型大小和BitOps预算下指导混合精度位分配。在OpenVLA风格策略上的广泛评估表明，Mix-QVLA改善了低位VLA部署的准确性与效率的权衡。在LIBERO上，Mix-QVLA将OpenVLA-OFT内存从15.4 GB减少到4.1 GB，保持了96.3的平均成功率，而BF16模型为97.1，并实现了1.52倍的推理加速。
View on arXiv Download PDF AI Translation
cs.CV / 7 / 2606.19584

Language-Instructed Vision Embeddings for Controllable and Generalizable Perception

语言指导的视觉嵌入用于可控和可泛化的感知
Mao, Chengzhi, Lin, Xudong, Chu, Wen-Sheng
Abstract
Vision foundation models are typically trained as static feature extractors, placing the burden of task adaptation onto large downstream models. We propose an alternative paradigm: instead of solely feeding visual features into language models, we use language itself to dynamically guide the vision encoder. Our method, Language-Instructed Vision Embeddings (LIVE), leverages language as high-level guidance to produce task-centric embeddings at inference time, removing the need for task-specific retraining. This enables the encoder to focus on contextually relevant aspects of the input, yielding more controllable and generalizable representations. Empirically, LIVE reduces visual hallucinations (+34 points on MMVP), surpasses vision-language models with orders of magnitude more parameters on visual question answering, and generalizes to unseen instructions and tasks -- offering a direct path toward adaptive, instruction-driven visual intelligence.
Chinese Translation
视觉基础模型通常被训练为静态特征提取器，将任务适应的负担放在大型下游模型上。我们提出了一种替代范式：我们不仅仅将视觉特征输入语言模型，而是利用语言本身动态引导视觉编码器。我们的方法，语言指导的视觉嵌入（Language-Instructed Vision Embeddings, LIVE），利用语言作为高层次的指导，在推理时生成以任务为中心的嵌入，从而消除了任务特定的再训练需求。这使得编码器能够专注于输入中与上下文相关的方面，产生更可控和可泛化的表示。从经验上看，LIVE减少了视觉幻觉（在MMVP上提高34分），在视觉问答任务中超越了参数量级更大的视觉-语言模型，并且能够泛化到未见过的指令和任务——为适应性、指令驱动的视觉智能提供了直接的路径。
View on arXiv Download PDF AI Translation
cs.CV / 8 / 2606.19617

GB-LSR: A Fast Local Spectral Image Representation with a Single Global Bandwidth for Continuous Reconstruction and Super-Resolution

GB-LSR：一种具有单一全局带宽的快速局部谱图像表示，用于连续重建和超分辨率
Shad, Max, Khoshnevis, Naeem
Abstract
We present GB-LSR (Global-Bandwidth Local Spectral Representation), a fixed-grid local spectral representation for continuous image reconstruction. The image domain is partitioned into non-overlapping square patches, each carrying coefficients for a truncated Fourier basis predicted from shared convolutional-encoder features. A single trainable scalar bandwidth is shared globally across all patches and images, and reconstruction at any continuous coordinate is a fixed-size basis contraction whose cost is independent of image size. We study three bandwidth-handling variants: a trainable global scalar (main), a fixed global scalar, and a per-patch bandwidth field. On a standardized native-reconstruction benchmark across Kodak, Set14, and Urban100, the main variant outperforms matched-budget amortized LIIF / LTE / WIRE re-implementations by 2.8-3.6 dB PSNR and 0.11-0.15 LPIPS, while running at roughly one-quarter of the slowest baseline's inference cost. The single global scalar suffices empirically: per-patch adaptive-bandwidth alternatives do not improve over it on either a closed-form locality diagnostic or an end-to-end ablation. In a separate arbitrary-scale super-resolution (ASR) extension, GB-LSR achieves competitive PSNR-Y under a canonical-style SR protocol and runs 1.44x faster than LIIF-RDN and 3.25x faster than LTE-SwinIR at x4; within the same extension, a variant trained and evaluated without 4-corner local-ensemble averaging gives a 1.77x speedup with 35% lower peak memory and negligible PSNR change, while additionally widening the RDN encoder from 64 to 96 channels gives a small positive PSNR shift with a 1.58x speedup and 31% lower peak memory. Native-reconstruction claims are scoped to the matched-budget amortized protocol, and ASR claims are scoped to a separate canonical-style SR protocol.
Chinese Translation
我们提出了GB-LSR（全局带宽局部谱表示），这是一种用于连续图像重建的固定网格局部谱表示。图像域被划分为不重叠的正方形块，每个块携带来自共享卷积编码器特征的截断傅里叶基的系数。一个可训练的标量全局带宽在所有块和图像中共享，任何连续坐标的重建都是固定大小基的收缩，其成本与图像大小无关。我们研究了三种带宽处理变体：可训练的全局标量（主要变体）、固定全局标量和每块带宽场。在Kodak、Set14和Urban100的标准化原生重建基准测试中，主要变体在匹配预算的摊销LIIF / LTE / WIRE重实现中表现出2.8-3.6 dB的PSNR和0.11-0.15的LPIPS提升，同时运行成本约为最慢基线推理成本的四分之一。实证表明，单一全局标量已足够：每块自适应带宽的替代方案在封闭形式的局部性诊断或端到端消融中均未能超越它。在一个独立的任意尺度超分辨率（ASR）扩展中，GB-LSR在经典样式的SR协议下实现了具有竞争力的PSNR-Y，并且在x4时比LIIF-RDN快1.44倍，比LTE-SwinIR快3.25倍；在同一扩展中，一个不使用4角局部集成平均的变体实现了1.77倍的加速，峰值内存降低35%，PSNR变化微乎其微，同时将RDN编码器的通道数从64扩展到96也带来了小幅正向PSNR提升，速度加快1.58倍，峰值内存降低31%。原生重建的声明限于匹配预算的摊销协议，而ASR的声明限于单独的经典样式SR协议。
View on arXiv Download PDF AI Translation
cs.CV / 9 / 2606.19662

Learning When to Denoise: Optimizing Asynchronous Schedules for Latent Diffusion

学习何时去噪：优化潜在扩散的异步调度
Qian, Bingshuo, Cheng, Xiang
Abstract
Multi-representation diffusion models can improve visual synthesis by denoising complementary views of an image, but their performance depends critically on the asynchronous schedule that determines when each representation is denoised. We propose to learn this schedule. Our method formulates asynchronous flow matching over multiple representation spaces and uses a schedule-corrected objective that keeps each representation's local noising-time weights fixed as the schedule changes. We instantiate the schedule with a flexible parametric class that is convex and monotone by construction, and learn it using a fast joint probe with less than 1% additional training compute. On ImageNet 256x256, the learned schedule substantially improves both convergence speed and final quality under a matched 675M-parameter XL backbone. With AutoGuidance, our 200-epoch model reaches FID 1.05, matching the 800-epoch SFD-XL baseline with 4x less training. Training to 600 epochs further improves to FID 1.02, outperforming the 1B-parameter SFD-XXL result of FID 1.04 while using a smaller model. In the unguided setting, our 200-epoch model reaches FID 2.37, already below the best 800-epoch SFD-XL result (2.54) at 4x less training, and improves to FID 2.14 at 600 epochs. Code is available at https://github.com/bsq532087/LWD
Chinese Translation
多表示扩散模型通过去噪图像的互补视图来改善视觉合成，但其性能在很大程度上依赖于决定每个表示何时去噪的异步调度。我们提出学习这一调度。我们的方法在多个表示空间上制定异步流匹配，并使用调度校正目标，在调度变化时保持每个表示的局部噪声时间权重不变。我们用一个灵活的参数类实例化该调度，该类在构造上是凸的和单调的，并使用快速联合探测进行学习，额外的训练计算量不足1%。在ImageNet 256x256上，学习到的调度显著提高了在匹配的675M参数XL骨干网络下的收敛速度和最终质量。在AutoGuidance下，我们的200个周期模型达到FID 1.05，匹配800个周期的SFD-XL基线，训练量减少了4倍。训练到600个周期进一步改善至FID 1.02，超越了使用更小模型的1B参数SFD-XXL结果FID 1.04。在无引导设置下，我们的200个周期模型达到FID 2.37，已经低于最佳800个周期SFD-XL结果（2.54），训练量减少4倍，并在600个周期时改善至FID 2.14。代码可在 https://github.com/bsq532087/LWD 获取。
View on arXiv Download PDF AI Translation
cs.CV / 10 / 2606.19676

TeleMorpher: Toward Robust Simultaneous Motion-Location Editing

TeleMorpher：朝着稳健的同步运动-位置编辑迈进
Chung, Haengbok
Abstract
Diffusion models have achieved remarkable success in image and video generation and editing. While recent studies have extended these efforts toward motion editing, simultaneously transforming both motion and location-despite its practical importance-remains largely unexplored. To better understand robust motion-location editing, we first analyze the fundamental factors that degrade its quality. Based on this analysis, we propose TeleMorpher, one of the first one-shot frameworks to the best of our knowledge, for simultaneous motion-location editing. Our approach leverages motion priors, a target motion-centric video generated from an off-the-shelf model as motion-editing guidance, and the ground truth motion to enable more controllable and precise motion-location editing. Via this, our framework works as follows: (1) we first disentangle the protagonist and the background via pre-trained segmentation and inpainting models. (2) Then, we introduce a training-free pose warping that edits the protagonist's motion with the motion prior as the guidance. (3) The result of warped motion video is directly injected into a baseline motion editor during inference, mitigating the difference between source and target motions while preserving the appearance of the source video. (4) To enhance the reliability of quantitative evaluations, we propose two new LPIPS-based metrics that measure the background consistency before and after the motion editing and the fidelity of motion editing performance via measuring the difference between the extracted protagonist's skeletons from source and target videos. Experiments with in-the-wild videos and the TaiChi dataset demonstrate that TeleMorpher achieves superior performance across both quantitative and qualitative measurements (real-human evaluation), underscoring its effectiveness.
Chinese Translation
扩散模型在图像和视频生成与编辑方面取得了显著成功。尽管近期研究已将这些努力扩展到运动编辑，但同时转换运动和位置——尽管其实际重要性——仍然在很大程度上未被探索。为了更好地理解稳健的运动-位置编辑，我们首先分析了影响其质量的基本因素。基于此分析，我们提出了TeleMorpher，这是我们所知的首个一体化框架，用于同步运动-位置编辑。我们的方法利用运动先验，从现成模型生成的以目标运动为中心的视频作为运动编辑指导，并结合真实运动，以实现更可控和精确的运动-位置编辑。通过这一方法，我们的框架工作如下：(1) 我们首先通过预训练的分割和修复模型将主角与背景分离。(2) 然后，我们引入无训练的姿态变形，利用运动先验作为指导来编辑主角的运动。(3) 变形后的视频结果在推理过程中直接注入基线运动编辑器，减小源运动与目标运动之间的差异，同时保留源视频的外观。(4) 为了增强定量评估的可靠性，我们提出了两个新的基于LPIPS的度量指标，分别测量运动编辑前后背景的一致性以及通过测量源视频与目标视频中提取的主角骨架之间的差异来评估运动编辑性能的保真度。通过在真实场景视频和太极数据集上的实验，TeleMorpher在定量和定性测量（真实人类评估）上均表现出优越的性能，突显了其有效性。
View on arXiv Download PDF AI Translation
cs.CV / 11 / 2606.19682

Vortex: Multi-Modal Fusion System for Intelligent Video Retrieval

Vortex：用于智能视频检索的多模态融合系统
Nguyen, Duc-Tho, Tran-Minh, Hieu-Hoc, Lam, Khanh-Hoa, Ly, Hoang-Nhut, Huynh, Huu-Phuc, Tran, Thanh-Tien, Le, Trung-Nghia
Abstract
This paper presents Vortex, the multimodal video retrieval system developed by our team, FocusOnFun, for the Ho Chi Minh City AI Challenge 2025, designed to advance intelligent multimedia search and temporal reasoning. The system integrates adaptive keyframe extraction, multimodal metadata generation from vision-language and speech models, and a hybrid retrieval strategy that fuses CLIP and SigLIP2 embeddings through Reciprocal Rank Fusion to balance global and fine-grained semantics. To enhance interactivity, Vortex incorporates Rocchio-based relevance feedback and a multi-stage temporal search mechanism for sequential event alignment. Built on Milvus and Elasticsearch, the architecture enables scalable indexing and efficient retrieval. Evaluated in the official competition, our FocusOnFun team's system achieved a score of 79.6/88 (90.5\%) in the Preliminary Round and was further evaluated in the Final Round, achieving an `Excellent' overall performance with `Outstanding' results in the question-answering (QA) task. This demonstrating the complementary strengths of CLIP and SigLIP2 and confirming the effectiveness of the hybrid retrieval approach. The system establishes a robust foundation for future research in intelligent, context-aware, and interactive video retrieval.
Chinese Translation
本文介绍了Vortex，这是我们团队FocusOnFun为2025年胡志明市人工智能挑战赛开发的多模态视频检索系统，旨在推动智能多媒体搜索和时间推理。该系统集成了自适应关键帧提取、基于视觉-语言和语音模型的多模态元数据生成，以及一种混合检索策略，通过互惠排名融合（Reciprocal Rank Fusion）将CLIP和SigLIP2嵌入结合，以平衡全局和细粒度语义。为了增强交互性，Vortex结合了基于Rocchio的相关反馈和多阶段时间搜索机制，以实现顺序事件对齐。该架构基于Milvus和Elasticsearch，支持可扩展的索引和高效的检索。在官方比赛中，我们FocusOnFun团队的系统在初赛中获得了79.6/88（90.5%）的分数，并在决赛中进一步评估，整体表现被评为“优秀”，在问答（QA）任务中取得了“杰出”的结果。这展示了CLIP和SigLIP2的互补优势，并确认了混合检索方法的有效性。该系统为未来智能、上下文感知和交互式视频检索的研究奠定了坚实的基础。
View on arXiv Download PDF AI Translation
cs.CV / 12 / 2606.19684

Exploring Multi-Modal Large Language Models and Two-Stage Fine-Tuning for Fashion Image Retrieval

探索多模态大型语言模型及其在时尚图像检索中的两阶段微调
Hoang, Nguyen Cao, Le, Hoang Bui, Hoang, Nam Vo, Le, Trung-Nghia
Abstract
Composed image retrieval retrieves a target image using a composed query of a reference image and a modified text description. In the fashion domain, this task requires understanding subtle attribute variations such as color, pattern, and texture. However, existing approaches face limitations due to scarce annotated data and simplistic negative sampling. We propose a novel framework that integrates a multi-modal large language model (LLaVA) to generate attribute-aware triplets and introduces a two-stage fine-tuning strategy to enhance contrastive learning. We leverage pretrained vision-language models, such as CLIP-ViT/B32, to generate and concatenate sentence-level prompts with the relative caption and to scale the number of negatives using static representations. Experimental results demonstrate enhanced compositional reasoning and improved fine-grained retrieval behavior, underscoring the feasibility and potential of the proposed framework for fashion retrieval.
Chinese Translation
组合图像检索通过参考图像和修改后的文本描述的组合查询来检索目标图像。在时尚领域，这一任务需要理解细微的属性变化，如颜色、图案和纹理。然而，现有方法由于标注数据稀缺和简单的负样本采样而面临局限性。我们提出了一种新颖的框架，集成了多模态大型语言模型（LLaVA），以生成属性感知的三元组，并引入两阶段微调策略以增强对比学习。我们利用预训练的视觉-语言模型，如 CLIP-ViT/B32，生成并连接句子级提示与相对标题，并使用静态表示来扩大负样本的数量。实验结果表明，组合推理能力得到增强，细粒度检索行为得到改善，突显了所提框架在时尚检索中的可行性和潜力。
View on arXiv Download PDF AI Translation
cs.CV / 13 / 2606.19706

NEST: Narrative Event Structures in Time for Long Video Understanding

NEST：用于长视频理解的叙事事件结构
Asgarov, Ali, Narasimhan, Kaushik, Sarker, Najibul Haque, Alomari, Hani, Tang, Chia-Wei, Sivakumar, Anushka, Hakim, Zaber Ibn Abdul, Mallampati, Shaurya, Thomas, Chris
Abstract
Recent progress in vision-language models has enabled the processing of increasingly long video sequences, but the ability to handle extended token streams does not translate to understanding of narrative structure in long videos. Existing long video benchmarks focus on needle-in-a-haystack retrieval rather than evaluating how low-level actions form events, how events interact across time, and how narratives progress, for example, whether a model can connect an early setback, such as a job loss to a later relationship breakup, despite long gaps, intervening scenes, or flashbacks that reframe what occurred. We introduce NEST (Narrative Event Structures in Time for Long Video Understanding), a dataset of 1005 full-length movies (avg. 98 minutes), each annotated with 102 multimodal narrative events grounded in visual content, dialogue, and audio. NEST captures multimodal narrative events with structured annotations grounded in visual content, dialogue, and audio, and links them through relations that reflect narrative structure, including temporal ordering, hierarchical composition, and long-range dependencies. We introduce baselines for event trigger detection (ETD), event localization (EL), event argument extraction (EAE), and event relation extraction (ERE). The benchmark is highly challenging for grounded event discovery, with ETD below 8%, EL under 6%, and EAE below 11%. In contrast, ERE is more tractable once events are given, reaching 35.45% F1 zero-shot and 44.42% F1 after fine-tuning.
Chinese Translation
近期视觉-语言模型的进展使得处理越来越长的视频序列成为可能，但处理扩展的标记流的能力并不等同于理解长视频中的叙事结构。现有的长视频基准测试主要集中在“针在海堆中”的检索，而不是评估低层次动作如何形成事件、事件如何随时间相互作用以及叙事如何进展。例如，模型是否能够将早期的挫折（如失业）与后来的关系破裂联系起来，尽管存在较长的间隔、干扰场景或重新构架发生事件的闪回。我们引入了NEST（用于长视频理解的叙事事件结构），这是一个包含1005部完整电影（平均98分钟）的数据集，每部电影都注释了102个基于视觉内容、对话和音频的多模态叙事事件。NEST通过基于视觉内容、对话和音频的结构化注释捕捉多模态叙事事件，并通过反映叙事结构的关系将它们连接起来，包括时间顺序、层次组成和长程依赖。我们为事件触发检测（ETD）、事件定位（EL）、事件参数提取（EAE）和事件关系提取（ERE）引入了基线。该基准在有根事件发现方面具有很高的挑战性，ETD低于8%，EL低于6%，EAE低于11%。相比之下，一旦给定事件，ERE更易处理，达到35.45%的F1零-shot和44.42%的F1经过微调后。
View on arXiv Download PDF AI Translation
cs.CV / 14 / 2606.19718

One-Shot Novel View and Pose Human Image Synthesis via 3D Prior Guided Diffusion Model

基于3D先验引导的扩散模型的一次性新视角和姿态人像合成
Gong, Shenjian, Wang, Kangkan, Zhang, Shanshan, Yang, Jian
Abstract
This paper addresses the challenge of one-shot novel view and pose human image synthesis. The existing methods transfer the reference human image to a target pose using a set of 2D pose keypoints or synthesize human images based on generalizable human NeRF which uses human model priors to extract point-wise features. However, pose transfer based methods can not handle complex human pose using ambiguous 2D pose as the condition, while generalizable human NeRFs may be inaccurate to recover occluded/invisiable human parts without extracted reliable features. To solve these problems, we propose a novel approach for novel view and pose synthesis from a singe human image via conditional denoising diffusion model. Our diffusion model divides the novel view and pose synthesis problem into a sequence of conditional denoising steps. Specifically, to generate humans with complex and arbitrary poses, we introduce 3D human priors, i.e., 3D normal map and color prompt, as geometry and color conditions into the generation process. By transferring the reference human into the target human with a series of diffusion steps, our diffusion model enables high-quality synthesis including the occluded/invisible parts. Further, we propose a self-reconstruction based customized refinement to enhance fine details when tested on novel persons.Experimental results on different public datasets demonstrate that our approach significantly outperforms previous methods and also shows better generalization ability across datasets. The code will be made publicly available at https://github.com/Yankeegsj/3DPGDM.
Chinese Translation
本文针对一次性新视角和姿态人像合成的挑战进行了研究。现有方法通过一组2D姿态关键点将参考人像转换为目标姿态，或基于可泛化的人类神经辐射场（NeRF）合成人像，后者利用人类模型先验提取逐点特征。然而，基于姿态转移的方法无法处理使用模糊的2D姿态作为条件的复杂人类姿态，而可泛化的人类NeRF在恢复被遮挡/不可见的人体部分时可能因缺乏可靠特征而不准确。为了解决这些问题，我们提出了一种新方法，通过条件去噪扩散模型从单个人像合成新视角和姿态。我们的扩散模型将新视角和姿态合成问题分解为一系列条件去噪步骤。具体而言，为了生成具有复杂和任意姿态的人体，我们引入3D人类先验，即3D法线图和颜色提示，作为生成过程中的几何和颜色条件。通过一系列扩散步骤将参考人像转化为目标人像，我们的扩散模型能够实现高质量的合成，包括被遮挡/不可见的部分。此外，我们提出了一种基于自重建的定制化精细化方法，以增强在新人物测试时的细节。不同公共数据集上的实验结果表明，我们的方法显著优于以前的方法，并且在不同数据集上表现出更好的泛化能力。代码将公开发布在 https://github.com/Yankeegsj/3DPGDM。
View on arXiv Download PDF AI Translation
cs.CV / 15 / 2606.19733

QueryGaussian: Scalable and Training-Free Open-Vocabulary 3D Instance Retrieval

QueryGaussian：可扩展且无需训练的开放词汇3D实例检索
Zhu, Xiuyuan, Lu, Ke, Yang, Zijie, Yue, Chao, Xue, Jian, Zhang, Dongming
Abstract
Efficiently retrieving specific 3D instances from large-scale scenes via natural language prompts remains a formidable challenge in multimedia analysis. Existing approaches predominantly follow a "scene-level embedding" paradigm, which requires distilling high-dimensional semantic features into every 3D primitive. This strategy suffers from a fundamental architectural bottleneck: memory and computational costs scale linearly with scene complexity, inevitably triggering out-of-memory (OOM) failures in city-scale environments. To address this barrier, we propose QueryGaussian, a training-free framework for expeditious and scalable open-vocabulary 3D instance retrieval. Unlike holistic semantic distillation, QueryGaussian employs an instance-level query mechanism that decouples semantic understanding from geometric representation. Specifically, we leverage pre-trained 2D vision models to interpret user prompts and lift segmentation masks into 3D via a concurrent maximum-weight association strategy, ensuring semantic-visual consistency. To mitigate projection ambiguity, we introduce a temporal fusion module with multi-stage adaptive density clustering. Experimental results demonstrate that QueryGaussian not only matches the accuracy of state-of-the-art methods but also delivers a decisive efficiency leap, reducing GPU memory usage by over 70% and accelerating inference by 180x. Crucially, QueryGaussian enables expeditious instance retrieval on city-scale scenes containing tens of millions of Gaussians using consumer-grade hardware.
Chinese Translation
通过自然语言提示高效地从大规模场景中检索特定3D实例仍然是多媒体分析中的一项艰巨挑战。现有的方法主要遵循“场景级嵌入”范式，这要求将高维语义特征提炼到每个3D原语中。这一策略面临着根本的架构瓶颈：内存和计算成本与场景复杂性呈线性增长，必然在城市规模环境中引发内存溢出（OOM）故障。为了解决这一障碍，我们提出了QueryGaussian，一个无需训练的框架，用于快速且可扩展的开放词汇3D实例检索。与整体语义提炼不同，QueryGaussian采用实例级查询机制，将语义理解与几何表示解耦。具体而言，我们利用预训练的2D视觉模型来解释用户提示，并通过并行最大权重关联策略将分割掩码提升到3D，确保语义-视觉一致性。为了解决投影模糊问题，我们引入了一个具有多阶段自适应密度聚类的时间融合模块。实验结果表明，QueryGaussian不仅在准确性上与最先进的方法相匹配，还实现了显著的效率飞跃，GPU内存使用减少超过70%，推理速度提高180倍。关键是，QueryGaussian能够在包含数千万个高斯的城市规模场景上使用消费级硬件快速进行实例检索。
View on arXiv Download PDF AI Translation
cs.CV / 16 / 2606.19736

VFACamou: View-Fused Adversarial Camouflage for Environment-Adaptive Physical Evasion

VFACamou：环境自适应物理规避的视图融合对抗伪装
Yan, Shihui, Liu, Hu, Shi, Junyu, Zhu, Zihui, Zhou, Ziqi, Song, Yufei, Geng, Youming, Li, Minghui, Hu, Shengshan
Abstract
Adversarial camouflage in the physical world remains highly challenging, particularly under UAV reconnaissance where targets undergo continuous geometric changes and extreme illumination variations. Existing methods either optimize 2D digital perturbations that fail to generalize to dynamic viewpoints or produce visually unnatural textures that cannot be deployed in real scenarios. Therefore, we propose an end-to-end framework for adversarial camouflage generation that automatically produces wearable adversarial patterns and maintains stable attack performance in real physical environments with changing viewpoints, poses, and lighting conditions. Our method integrates UV-volume rendering with a diffusion-based texture generator, enabling consistent appearance under varying scales, poses, and lighting conditions. To ensure environmental realism, we propose an illumination color consistency estimator that extracts dominant background attributes and guides a natural texture loss to align the generated UV texture with the surrounding environment. A multi-scale dynamic training strategy further enhances robustness against viewpoint shifts and body deformation. Extensive experiments across multiple mainstream detectors demonstrate that our method achieves strong and stable physical attack performance while maintaining high perceptual naturalness, reducing human detection rates without introducing unnatural artifacts.
Chinese Translation
物理世界中的对抗伪装仍然面临很大挑战，尤其是在无人机侦察下，目标经历持续的几何变化和极端的光照变化。现有方法要么优化无法推广到动态视角的二维数字扰动，要么生成在真实场景中无法部署的视觉上不自然的纹理。因此，我们提出了一种端到端的对抗伪装生成框架，能够自动生成可穿戴的对抗图案，并在具有变化视角、姿态和光照条件的真实物理环境中保持稳定的攻击性能。我们的方法将UV体积渲染与基于扩散的纹理生成器相结合，使得在不同的尺度、姿态和光照条件下保持一致的外观。为了确保环境的真实感，我们提出了一种光照颜色一致性估计器，提取主导背景属性，并指导自然纹理损失，以使生成的UV纹理与周围环境对齐。多尺度动态训练策略进一步增强了对视角变化和身体变形的鲁棒性。通过在多个主流检测器上的广泛实验，证明我们的方法在保持高感知自然性的同时，能够实现强大而稳定的物理攻击性能，降低人类检测率而不引入不自然的伪影。
View on arXiv Download PDF AI Translation
cs.CV / 17 / 2606.19776

Occ-VLM: Occupancy Grounded Vision Language Model for Indoor Scene Understanding

Occ-VLM：基于占用的视觉语言模型用于室内场景理解
Li, Jianing, Fang, Zhou, Liu, Yijiang, Du, Li
Abstract
Recently, vision-language models (VLMs) have made significant progress in 3D scene understanding, driving advances in applications such as embodied intelligence and robotic vision. However, existing approaches typically either rely directly on explicit 3D inputs (e.g., point clouds or RGB-D sequences), or introduce an additional 3D geometry encoder to derive 3D-aware visual tokens from 2D images. Such designs structurally decouple 3D geometric perception from the rich 2D semantics learned via vision-language pre-training, hindering the development of a unified 3D vision-language representation. In this work, we propose Occ-VLM, a novel framework for 3D scene understanding that operates purely on posed RGB images and employs a single 2D vision encoder. Specifically, Occ-VLM reconstructs 3D scene occupancy as an auxiliary geometric prior, which is utilized to spatially associate foreground 2D tokens with 3D space. These tokens are then decoded by a Large Language Model (LLM) for unified scene understanding. Extensive experiments demonstrate that Occ-VLM achieves both accurate geometric perception and robust vision-language reasoning: it attains state-of-the-art performance on multi-view occupancy prediction, while performing on par with 3D-input VLMs on 3D Visual Question Answering (VQA) and 3D dense captioning benchmarks.
Chinese Translation
近年来，视觉语言模型（VLMs）在3D场景理解方面取得了显著进展，推动了具身智能和机器人视觉等应用的发展。然而，现有的方法通常直接依赖于显式的3D输入（例如，点云或RGB-D序列），或者引入额外的3D几何编码器，从2D图像中推导出具有3D感知的视觉标记。这种设计在结构上将3D几何感知与通过视觉语言预训练学习到的丰富2D语义解耦，阻碍了统一的3D视觉语言表示的发展。在本研究中，我们提出了Occ-VLM，一种用于3D场景理解的新框架，该框架完全基于姿态RGB图像，并采用单一的2D视觉编码器。具体而言，Occ-VLM重建3D场景占用作为辅助几何先验，用于将前景2D标记与3D空间进行空间关联。这些标记随后由大型语言模型（LLM）解码，以实现统一的场景理解。大量实验表明，Occ-VLM在准确的几何感知和稳健的视觉语言推理方面均表现出色：在多视角占用预测上达到最先进的性能，同时在3D视觉问答（VQA）和3D密集标注基准测试中与3D输入的VLMs表现相当。
View on arXiv Download PDF AI Translation
cs.CV / 18 / 2606.19804

HypOProto: Hyperbolic Ordinal Prototypes for Left Ventricular Filling Pressure Classification

HypOProto：用于左心室充盈压分类的双曲序数原型
Wu, Victoria, Hashemi, Nima, Vaseli, Hooman, Luong, Christina, Abolmaesumi, Purang, Tsang, Teresa S. M.
Abstract
Echocardiography (echo) is a widely used imaging modality for assessing cardiac function, with Left Ventricular Filling Pressure (LVFP) serving as a critical physiological marker for conditions such as heart failure. Standard LVFP classification into normal \emph{vs} elevated categories relies on the Doppler-derived $E/e'$ ratio, which is operator-dependent and often unavailable in resource-limited settings, motivating methods that infer LVFP directly from B-mode echo. Existing deep learning approaches achieve high performance but remain largely black-box, limiting clinical interpretability. We propose HypOProto, a hyperbolic, ordinal prototype-based framework for interpretable LVFP classification using a frozen, explainable foundation model backbone. HypOProto arranges prototypes along the physiological $E/e'$ scale, placing borderline cases near the hyperboloid root where small angular differences separate similar cases, while normal and elevated cases occupy outward positions reflecting increasing diagnostic certainty. This hyperbolic geometry encodes clinically meaningful ordinal relationships and improves interpretability. We also introduce a novel Hyperbolic Prototype Angular Separation (HyperPAS) loss, enforcing inter-class prototype separation in hyperbolic space. HypOProto achieves SOTA performance while maintaining transparency, and highlights clinically relevant regions in visualizations. This work represents the first prototype-based framework for LVFP classification in echo. Our code can be found at https://github.com/DeepRCL/HypOProto.
Chinese Translation
超声心动图（echo）是一种广泛使用的心脏功能评估成像方式，左心室充盈压（LVFP）作为心力衰竭等疾病的重要生理标志。标准的LVFP分类将其分为正常与升高两类，依赖于多普勒导出的$E/e'$比率，该比率受操作人员影响，并且在资源有限的环境中往往不可用，这促使我们开发直接从B模式超声中推断LVFP的方法。现有的深度学习方法虽然性能优异，但大多为黑箱模型，限制了临床可解释性。我们提出了HypOProto，一个基于双曲序数原型的可解释LVFP分类框架，采用一个冻结的、可解释的基础模型作为骨干。HypOProto在生理$E/e'$尺度上排列原型，将边界案例置于双曲根附近，在小角度差异下区分相似案例，而正常和升高案例则占据向外的位置，反映出逐渐增加的诊断确定性。这种双曲几何编码了临床上有意义的序数关系，并提高了可解释性。我们还引入了一种新颖的双曲原型角度分离损失（HyperPAS），在双曲空间中强制实施类间原型分离。HypOProto在保持透明度的同时实现了SOTA（最先进）性能，并在可视化中突出临床相关区域。这项工作代表了超声中LVFP分类的首个基于原型的框架。我们的代码可以在https://github.com/DeepRCL/HypOProto找到。
View on arXiv Download PDF AI Translation
cs.CV / 19 / 2606.19805

ParaScale: Scale-Calibrated Camera-Motion Transfer via a Gauge-Invariant Parallax Number

ParaScale：通过一个不变于标度的视差数进行标度校准的相机运动转移
Meng, Zijie
Abstract
Transferring the camera motion of a reference video to a freshly generated one lets creators reuse cinematic moves. Yet reference and target often live at incompatible scales -- a sweep across a galaxy versus a nudge across a desk -- and naively reusing the recovered trajectory yields either imperceptible or violently exaggerated motion. We trace this to a geometric fact: translation-induced image motion scales as ||T||/Z, so a monocular trajectory is meaningful only up to a depth-scale gauge. We distill this into the Parallax Number Pi = ||Delta T|| / Zbar, a dimensionless, gauge-invariant descriptor of how strongly a camera move is felt, and prove that it -- not the raw trajectory -- is the quantity that scale-faithful transfer must preserve. ParaScale is a plug-and-play module that reads Pi off any reference video and re-realizes it against the target scene's own depth, per frame, leaving rotation untouched. Sitting between pose extraction and pose injection, it requires no retraining and drops into any pose-conditioned generator. We further introduce the Parallax Consistency Error (PCE), a scale-symmetric metric that -- unlike the similarity-aligned TransErr -- exposes scene-scale mismatch. Across scale regimes spanning four orders of magnitude and multiple backbones, ParaScale keeps the realized parallax on the identity line and cuts PCE by more than 3x over uncalibrated transfer with no loss of visual fidelity.
Chinese Translation
将参考视频的相机运动转移到新生成的视频中，使创作者能够重用电影镜头。然而，参考视频和目标视频往往处于不兼容的标度下——例如，跨越一个星系的扫动与在桌面上的轻推——而简单地重用恢复的轨迹会导致运动要么不可察觉，要么被夸大到极端。我们将其归因于一个几何事实：由平移引起的图像运动按 ||T||/Z 进行缩放，因此单目轨迹仅在深度标度的测量下才有意义。我们将其提炼为视差数 Pi = ||Delta T|| / Zbar，这是一个无量纲的、不变于标度的描述符，表示相机运动的感知强度，并证明它——而非原始轨迹——是标度保真转移必须保留的量。ParaScale 是一个即插即用模块，可以从任何参考视频中读取 Pi，并根据目标场景自身的深度逐帧重新实现，保持旋转不变。它位于姿态提取与姿态注入之间，无需重新训练，可以直接嵌入任何姿态条件生成器中。我们进一步引入了视差一致性误差（PCE），这是一个标度对称的度量，与相似性对齐的 TransErr 不同，PCE 能够揭示场景标度的不匹配。在跨越四个数量级和多个基础网络的标度范围内，ParaScale 将实现的视差保持在恒等线上，并将 PCE 降低了超过 3 倍，而没有损失视觉保真度。
View on arXiv Download PDF AI Translation
cs.CV / 20 / 2606.19817

Training-Free Metrics for Synthetic Object Detection Data: A Proxy for Detector Performance

无训练度量用于合成物体检测数据：检测器性能的代理
Nam, Myeongseok, Yeo, Donghoon, Kim, Seungwook
Abstract
With the recent advent of image generative models, synthetic data are increasingly being used to supplement limited real datasets for training computer vision models. However, not all synthetic datasets improve performance equally, and their effectiveness can only be assessed by training a downstream model, which is computationally expensive and time-consuming. This problem is pronounced in the task of object detection, where the required annotations are much more dense due to bounding boxes. In this paper, we propose a pre-computable metric family, dubbed Conditional-Composition Domain Match (CCDM), which serves as a proxy for the relative utility of candidate synthetic training sets for downstream detection. Experiments on the VisDrone-DET dataset show that the CCDM metric families achieve a Spearman correlation of 1.0 with the downstream performance of YOLOv8, clearly outperforming existing metrics for synthetic image evaluation.
Chinese Translation
随着图像生成模型的出现，合成数据越来越多地被用来补充有限的真实数据集，以训练计算机视觉模型。然而，并非所有合成数据集对性能的提升都是相同的，其有效性只能通过训练下游模型来评估，这在计算上是昂贵且耗时的。这个问题在物体检测任务中尤为明显，因为所需的标注由于边界框而更加密集。本文提出了一种可预计算的度量家族，称为条件组合域匹配（Conditional-Composition Domain Match，CCDM），它作为候选合成训练集相对于下游检测的相对效用的代理。在VisDrone-DET数据集上的实验表明，CCDM度量家族与YOLOv8的下游性能达到了1.0的斯皮尔曼相关性，明显优于现有的合成图像评估度量。
View on arXiv Download PDF AI Translation
cs.CV / 21 / 2606.19824

CSWinUNETR: Segmentation of Thin Anatomical Structures in Medical Images

CSWinUNETR：医学图像中细小解剖结构的分割
Moon, Junho, Chung, Haejun, Jang, Ikbeom
Abstract
Accurate segmentation of thin, tortuous anatomical structures, such as retinal vessels, cerebral vasculature, and facial wrinkles, remains challenging due to low contrast, frequent discontinuities, and severe class imbalance. Although recent convolutional and Transformer-based models have improved performance, they often yield fragmented predictions and fail to recover fine branches. We propose CSWinUNETR, a general-purpose backbone for 2D and 3D thin-structure segmentation. It employs cross-shaped stripe self-attention to model long-range principal-axis context and incorporates cyclic shifts to enhance information exchange across stripes. To better preserve fine-grained details, we further introduce a detail-enhanced multi-scale self-attention module that aggregates contextual features from multi-resolution representations. In addition, we propose sparse-control dynamic snake convolution, which reconstructs reliable dense curvilinear kernels from sparsely predicted control points to better follow tortuous geometry. Extensive experiments on four benchmarks across ophthalmology, neurovascular imaging, and dermatology demonstrate that CSWinUNETR consistently outperforms state-of-the-art methods without task-specific post-processing or topology-aware losses. The code is available at https://github.com/labhai/CSWinUNETR.
Chinese Translation
准确分割细小、曲折的解剖结构，如视网膜血管、大脑血管和面部皱纹，仍然具有挑战性，主要由于低对比度、频繁的不连续性和严重的类别不平衡。尽管最近的卷积和基于Transformer的模型在性能上有所提升，但它们往往产生碎片化的预测，无法恢复细小的分支。我们提出了CSWinUNETR，这是一种用于2D和3D细小结构分割的通用骨干网络。它采用交叉形条带自注意力机制来建模长距离主轴上下文，并结合循环位移以增强条带间的信息交换。为了更好地保留细粒度细节，我们进一步引入了一种细节增强的多尺度自注意力模块，该模块从多分辨率表示中聚合上下文特征。此外，我们提出了稀疏控制动态蛇形卷积，该方法从稀疏预测的控制点重建可靠的密集曲线核，以更好地跟随曲折的几何形状。在眼科学、神经血管成像和皮肤病学的四个基准测试上进行的广泛实验表明，CSWinUNETR在没有特定任务后处理或拓扑感知损失的情况下，始终优于最先进的方法。代码可在 https://github.com/labhai/CSWinUNETR 获取。
View on arXiv Download PDF AI Translation
cs.CV / 22 / 2606.19828

3D-PLOT-LLM: Part-Level Object Tokens for 3D Large Language Models

3D-PLOT-LLM：用于3D大型语言模型的部分级对象标记
Xue, Jintang, Wang, Xinyu, Wu, Yixing, Chen, Jingwen, Kuo, C. -C. Jay
Abstract
3D multimodal large language models (3D MLLMs) describe a 3D object as a whole but cannot address, name, or reason about its parts. Prior part-aware attempts add segmentation decoders, heavier 3D encoders, or bounding-box grammars at substantial parameter cost. We take a fundamentally different path: we reorganize the input token stream so that parts become directly addressable through the LLM's own vocabulary. Our model, 3D-PLOT-LLM, partitions the frozen point encoder's patches into K locally coherent regions and inserts, before each region's patch tokens, a learnable per-region marker and a reserved vocabulary token ; a Marker-Space Refinement (MSR) module then conditions each marker on its region's spatial statistics and adjacency neighbors. The model thus cites parts in its output and follows prompts that refer to parts by token, a capability absent from prior object-level 3D MLLMs. To probe this interface, we construct PartVerse-QA, a vocabulary-level part-QA benchmark adapted from PartVerse mesh annotations (77K training pairs and 588 held-out queries on disjoint object splits), on which 3D-PLOT-LLM reaches caption-to-slots Jaccard 0.459 and Exact-match 13.78%, with a slot-to-caption GPT-4o judge of 44.68. On the 3DCoMPaT-GrIn part-aware grounded description benchmark, 3D-PLOT-LLM outperforms PointLLM, Kestrel, PARIS3D, and SegPoint on every text-output metric, and ShapeLLM on 3 of 4, with up to +3.03 GPT-4o judge over PointLLM. On Objaverse whole-object captioning, adding PartVerse-QA at Stage 2 yields +0.65 SBERT and +1.85 GPT-4o over PointLLM, and tops PointLLM-PiSA on 4 of 5 traditional metrics (SBERT, SimCSE, BLEU-1, METEOR) despite targeting a different (part-grounded) objective. All with under 1M new trainable parameters on a frozen point encoder, an order of magnitude below prior part-aware 3D MLLMs, and no segmentation decoder or bounding-box head.
Chinese Translation
3D多模态大型语言模型（3D MLLMs）将3D对象视为一个整体，但无法处理、命名或推理其部分。以往的部分感知尝试通过增加分割解码器、更重的3D编码器或边界框语法来实现，但代价高昂。我们采取了根本不同的路径：我们重新组织输入标记流，使得部分可以通过LLM自身的词汇直接寻址。我们的模型3D-PLOT-LLM将冻结的点编码器的补丁划分为K个局部一致的区域，并在每个区域的补丁标记之前插入一个可学习的区域标记和一个保留的词汇标记；然后，Marker-Space Refinement（MSR）模块根据每个标记所在区域的空间统计和邻接关系进行条件处理。因此，该模型在输出中引用部分，并遵循通过标记引用部分的提示，这是以往对象级3D MLLMs所缺乏的能力。为了探测这一接口，我们构建了PartVerse-QA，这是一个基于PartVerse网格注释（77K训练对和588个在不相交对象拆分上的保留查询）改编的词汇级部分问答基准，在该基准上，3D-PLOT-LLM达到了标注到槽的Jaccard 0.459和精确匹配13.78%，槽到标注的GPT-4o评估为44.68。在3DCoMPaT-GrIn部分感知基础描述基准上，3D-PLOT-LLM在每个文本输出指标上均优于PointLLM、Kestrel、PARIS3D和SegPoint，并在4个指标中优于ShapeLLM，GPT-4o评估比PointLLM高出3.03。在Objaverse整体对象标注中，在第2阶段添加PartVerse-QA使得SBERT提高0.65，GPT-4o提高1.85，尽管目标是不同的（部分基础）目标，但在5个传统指标（SBERT、SimCSE、BLEU-1、METEOR）中超过了PointLLM-PiSA。所有这些都在一个冻结的点编码器上，新增可训练参数不足100万，远低于以往的部分感知3D MLLMs，并且没有分割解码器或边界框头。
View on arXiv Download PDF AI Translation
cs.CV / 23 / 2606.19835

Neural Events: Discrete Asynchronous Autoencoders for Event-Based Vision

神经事件：用于事件驱动视觉的离散异步自编码器
Pellerito, Roberto, Gehrig, Daniel, Shiba, Shintaro, Scaramuzza, Davide
Abstract
Event cameras capture dynamic scenes with exceptional temporal fidelity by representing them as a continuous stream of microsecond resolution \textit{events}. Each individual event, however, only carries minimal semantic value, merely signaling a localized brightness change. To derive meaningful signals, downstream algorithms need to quickly integrate cues from a potentially massive torrent of low-information events. Current architectures, however, are easily overwhelmed, struggling to balance capturing fine-grained temporal dynamics and maintaining a manageable data throughput. This paper proposes a framework to re-tokenize event streams into a small set of highly informative \textit{neural events}, each representing a local spatio-temporal context window with a discrete learnable code. Every time this code flips, a neural event is triggered, yielding a highly compressed data stream. We demonstrate that, across object detection and classification, networks trained on neural events are on par or surpass the performance of state-of-the-art approaches while reducing the event rate by a factor of 2.0.
Chinese Translation
事件相机通过将动态场景表示为连续的微秒分辨率 extit{事件}流，以卓越的时间保真度捕捉动态场景。然而，每个单独的事件仅携带最小的语义价值，仅仅表示局部亮度变化。为了提取有意义的信号，下游算法需要快速整合来自潜在大量低信息事件的线索。然而，当前的架构容易被淹没，难以平衡捕捉细粒度时间动态与维持可管理的数据吞吐量。本文提出了一种框架，将事件流重新标记为一小组高度信息化的 extit{神经事件}，每个神经事件代表一个具有离散可学习编码的局部时空上下文窗口。每当该编码翻转时，就会触发一个神经事件，从而产生一个高度压缩的数据流。我们证明，在物体检测和分类任务中，基于神经事件训练的网络在性能上与最先进的方法相当或超越，同时将事件率降低了2.0倍。
View on arXiv Download PDF AI Translation
cs.CV / 24 / 2606.19838

OTCHA: Optimal Transport-driven Confidence-aware Latent Hub Alignment for Multi-View Medical Image Classification

OTCHA：基于最优运输的置信度感知潜在中心对齐用于多视角医学图像分类
Yang, Jiwoong, Chung, Haejun, Jang, Ikbeom
Abstract
Multi-view imaging, such as mammography and chest radiography, is a standard component of clinical practice. However, medical images are often unregistered and contain view-specific artifacts or irrelevant background cues that can obscure diagnostically relevant findings. Many existing methods directly fuse per-view representations, allowing such irrelevant content to contaminate the fused embedding and reducing robustness under varying view configurations. We propose OTCHA, a confidence-aware latent hub token alignment module based on optimal transport (OT) that refines patch tokens before fusion for multi-view classification. OTCHA introduces a set of learnable latent hub tokens shared across views. For each view, we compute an OT plan between patch tokens and hub tokens that jointly considers feature similarity and geometry, and augment the OT formulation with token-conditional dustbins to enable partial matching and discard irrelevant tokens. The resulting transport plan provides token-wise matching confidence, which gates hub-mediated message passing and weights a novel optimal-transport-based representation alignment loss to stabilize refinement. Experiments on three multi-view medical image datasets demonstrate consistent improvements over competing baselines across diverse anatomies and view configurations. Our code is available at https://github.com/labhai/OTCHA.
Chinese Translation
多视角成像，如乳腺X光检查和胸部X光检查，是临床实践中的标准组成部分。然而，医学图像通常未进行配准，并且包含特定视角的伪影或无关背景线索，这可能会掩盖诊断相关的发现。许多现有方法直接融合每个视角的表示，使得这些无关内容污染融合的嵌入，并在不同视角配置下降低鲁棒性。我们提出了OTCHA，一种基于最优运输（Optimal Transport, OT）的置信度感知潜在中心令牌对齐模块，在多视角分类中，在融合之前对补丁令牌进行精炼。OTCHA引入了一组跨视角共享的可学习潜在中心令牌。对于每个视角，我们计算补丁令牌与中心令牌之间的OT计划，该计划共同考虑特征相似性和几何形状，并通过令牌条件的尘箱增强OT公式，以实现部分匹配并丢弃无关令牌。最终的运输计划提供了逐令牌的匹配置信度，这控制了中心介导的信息传递，并加权了一种新颖的基于最优运输的表示对齐损失，以稳定精炼。在三个多视角医学图像数据集上的实验表明，在不同的解剖结构和视角配置下，相较于竞争基线，OTCHA始终表现出一致的改进。我们的代码可在 https://github.com/labhai/OTCHA 获得。
View on arXiv Download PDF AI Translation
cs.CV / 25 / 2606.19849

ViCoStream: Streaming VideoLLMs Can Run Beyond 100 FPS with Stage-Wise Coordinated Inference

ViCoStream：流媒体视频大语言模型可在阶段协调推理下以超过100帧每秒运行
Tan, Yang, Tong, Junlong, Yue, Linan, Wu, Hao, Fang, Pengfei, Shen, Xiaoyu
Abstract
Streaming VideoLLMs must continuously process incoming video while maintaining low query latency, making both video-ingestion throughput and query-time responsiveness critical for real-time deployment. Existing methods largely focus on accelerating individual modules, such as visual encoding, token pruning, or KV-cache compression, but provide limited insight into whether the resulting system can sustain real-time streaming performance. We formulate streaming VideoLLM inference as a coordinated pipeline spanning visual preprocessing, visual encoding, token dropping, and LLM prefilling/decoding. Building on this formulation, we propose ViCoStream (Video Coordinated Streaming), a stage-wise coordinated streaming framework that combines chunk-wise execution, CUDA-stream overlap, visual token control, bounded visual attention, and query-side retrieval to bound per-chunk computation and memory costs. We further provide a systematic study of bottleneck migration, revealing how chunk size, token retention, attention locality, and retrieval scope shape the throughput-accuracy trade-off. Experiments with Qwen2.5-VL-3B/7B-Instruct across multiple streaming benchmarks show that ViCoStream achieves 134 FPS video throughput and less than 50 ms TTFT on a single A100 GPU while maintaining accuracy close to full-history baselines.
Chinese Translation
流媒体视频大语言模型（VideoLLMs）必须持续处理传入的视频，同时保持低查询延迟，这使得视频摄取吞吐量和查询响应时间在实时部署中至关重要。现有方法主要集中在加速单个模块，如视觉编码、令牌修剪或KV缓存压缩，但对最终系统是否能够维持实时流媒体性能提供的见解有限。我们将流媒体VideoLLM推理形式化为一个涵盖视觉预处理、视觉编码、令牌丢弃和大语言模型（LLM）预填充/解码的协调管道。在此基础上，我们提出了ViCoStream（视频协调流媒体），这是一种阶段协调的流媒体框架，结合了块级执行、CUDA流重叠、视觉令牌控制、有界视觉注意力和查询侧检索，以限制每块的计算和内存成本。我们进一步提供了瓶颈迁移的系统研究，揭示了块大小、令牌保留、注意力局部性和检索范围如何影响吞吐量与准确性之间的权衡。在多个流媒体基准测试中，使用Qwen2.5-VL-3B/7B-Instruct的实验表明，ViCoStream在单个A100 GPU上实现了134帧每秒的视频吞吐量和低于50毫秒的查询时间延迟，同时保持接近全历史基线的准确性。
View on arXiv Download PDF AI Translation
cs.CV / 26 / 2606.19867

PSCT-Net: Geometry-Aware Pediatric Skull CT Reconstruction via Differentiable Back-Projection and Attention-Guided Refinement

PSCT-Net：通过可微反投影和注意力引导的细化实现几何感知的儿童头颅CT重建
Kim, Dong Yeong, Choi, Jaewon, Shin, Youmin, Lee, Jungyu, Kim, Myeongseop, Choi, Jinwook, Kim, Joo Whan, Kim, Young-Gon
Abstract
Computed Tomography (CT) is essential for diagnosing pediatric craniofacial abnormalities, yet poses radiation risks to developing anatomies. Reconstructing 3D CT from sparse bi-planar X-rays offers a low-dose alternative but is severely ill-posed. Existing methods employ geometry-agnostic feature lifting, naively projecting 2D features into 3D without explicit spatial modeling, causing depth ambiguity and degraded osseous boundaries. We present PSCT-Net, a geometry-aware framework with differentiable back-projection. Differentiable back-projection establishes a spatially faithful volumetric prior, alleviating depth ambiguity. An Attention-Guided Projection (AGP-3D) module then learns non-linear voxel-wise correspondences between 2D regions and 3D locations. A Bidirectional Mamba (BiM-3D) module captures long-range volumetric dependencies with linear complexity. We further curate a private institutional pediatric skull CT cohort, PedSkull-CT, comprising normal and pathological cases for internal evaluation, addressing the gap in adult-centric, trunk-focused datasets.
Chinese Translation
计算机断层扫描（CT）在诊断儿童颅面畸形中至关重要，但对发育中的解剖结构存在辐射风险。从稀疏双平面X射线重建三维CT提供了一种低剂量的替代方案，但该问题严重不适定。现有方法采用几何无关的特征提升，简单地将二维特征投影到三维中，而没有明确的空间建模，导致深度模糊和骨骼边界退化。我们提出了PSCT-Net，一个具有可微反投影的几何感知框架。可微反投影建立了一个空间上真实的体积先验，缓解了深度模糊。接着，注意力引导投影（AGP-3D）模块学习二维区域与三维位置之间的非线性体素对应关系。双向曼巴（BiM-3D）模块以线性复杂度捕捉长距离体积依赖关系。我们进一步整理了一个私立机构的儿童头颅CT数据集PedSkull-CT，包含正常和病理案例用于内部评估，填补了以成人为中心、以躯干为重点的数据集的空白。
View on arXiv Download PDF AI Translation
cs.CV / 27 / 2606.19882

Multimodal Concept Bottleneck Models

多模态概念瓶颈模型
Shi, Tongqing, Yan, Ge, Oikarinen, Tuomas, Weng, Tsui-Wei
Abstract
Concept Bottleneck Models (CBMs) enhance the interpretability of deep learning networks by aligning the features extracted from images with natural concepts. However, existing CBMs are constrained in their ability to generalize beyond a fixed set of predefined classes and the risk of non-concept information leakage, where predictive signals outside the intended concepts are inadvertently exploited. In this paper, we propose Multimodal Concept Bottleneck Model (MM-CBM) to address these issues and extend CBMs into CLIP. MM-CBM utilizes dual Concept Bottleneck Layers (CBLs) to align both the image and text embeddings into interpretable features. This allows us to perform new vision tasks like zero-shot classification or image retrieval in an interpretable way. Compared to existing methods, MM-CBM achieves up to 51.26% accuracy improvement on average across four standard benchmarks. Our method maintains high accuracy, staying within ~5% of black-box performance while offering greater interpretability.
Chinese Translation
概念瓶颈模型（Concept Bottleneck Models, CBMs）通过将从图像中提取的特征与自然概念对齐，增强了深度学习网络的可解释性。然而，现有的CBMs在超出固定预定义类别集的泛化能力上受到限制，并且存在非概念信息泄漏的风险，即意外利用了超出预期概念的预测信号。本文提出了多模态概念瓶颈模型（Multimodal Concept Bottleneck Model, MM-CBM）以解决这些问题，并将CBMs扩展到CLIP中。MM-CBM利用双重概念瓶颈层（Concept Bottleneck Layers, CBLs）将图像和文本嵌入对齐为可解释的特征。这使我们能够以可解释的方式执行新的视觉任务，如零样本分类或图像检索。与现有方法相比，MM-CBM在四个标准基准测试中平均提高了51.26%的准确率。我们的方法保持高准确率，性能与黑箱模型相差约5%，同时提供更好的可解释性。
View on arXiv Download PDF AI Translation
cs.CV / 28 / 2606.19889

SurgVista: Long-Horizon Surgical World Modeling with Plausible Instrument-Tissue Dynamics

SurgVista：具有合理工具-组织动态的长期外科世界建模
Pan, Wentao, Li, Wuyang, Liu, Shengyuan, Liu, Xinyu, Liu, Hengyu, Yuan, Yixuan
Abstract
Scaling robot policy learning for autonomous surgery is challenging, as expert demonstrations are expensive and in vivo exploration poses substantial safety risks. Surgical world models address this by generating realistic, action-conditioned future frames from an initial observation, but existing methods exhibit two persistent failure modes: spatial interaction incoherence, where visible instrument contact fails to induce spatially consistent tissue deformation, and temporal fidelity collapse, where prediction errors compound across autoregressive rollouts and progressively corrupt visual quality. We present SurgVista, a surgical world model that mitigates both failures through two training recipes. Deformation Consistency Regularization extracts scene-point trajectories from training videos and enforces cross-frame coherence through latent contrastive learning, strengthening physically consistent instrument-tissue dynamics. Drift Adaptation Training mitigates long-horizon drift by perturbing conditioning frames with online prediction residuals and photometric augmentations calibrated to long-horizon drift statistics, sustaining visual fidelity over extended rollouts. To enable rigorous evaluation, we further introduce SurgWorld-Bench, featuring diverse procedure types, long-range rollouts, and decoupled metrics for instrument-motion accuracy and tissue-response fidelity. Extensive experiments show that SurgVista consistently outperforms state-of-the-art methods across visual quality, temporal consistency, and interaction fidelity, with gains widening as the prediction horizon grows.
Chinese Translation
扩展机器人策略学习以实现自主手术具有挑战性，因为专家示范成本高昂，而体内探索则存在重大安全风险。外科世界模型通过从初始观察生成现实的、基于动作的未来帧来解决这一问题，但现有方法存在两种持续的失败模式：空间交互不一致性，即可见工具接触未能引起空间上连贯的组织变形，以及时间保真度崩溃，即预测误差在自回归展开中累积并逐渐破坏视觉质量。我们提出了SurgVista，一种通过两种训练方案来减轻这两种失败的外科世界模型。变形一致性正则化从训练视频中提取场景点轨迹，并通过潜在对比学习强制跨帧一致性，从而增强物理一致的工具-组织动态。漂移适应训练通过使用在线预测残差和根据长期漂移统计校准的光度增强来扰动条件帧，从而减轻长期漂移，在延长的展开中保持视觉保真度。为了实现严格评估，我们进一步引入了SurgWorld-Bench，具有多样化的手术类型、长范围展开和解耦的工具运动准确性和组织反应保真度指标。大量实验表明，SurgVista在视觉质量、时间一致性和交互保真度方面始终优于最先进的方法，随着预测范围的增加，性能提升更加显著。
View on arXiv Download PDF AI Translation
cs.CV / 29 / 2606.19901

Linear Recurrent Unit with Semantic Modulation for Image Super-Resolution

具有语义调制的线性递归单元用于图像超分辨率
Choi, Mingyu, Han, Woo Kyoung, Im, Sunghoon, Jin, Kyong Hwan
Abstract
Linear recurrent unit (LRU), designed with a principled formulation for stable linear recurrence, has demonstrated promising accuracy and robustness on long-range dependency tasks. However, its static parameterization and single-scan method limits its applicability to 2D vision tasks. In this study, we propose a LRU-based restoration network with a semantic modulating unit (SMU) to achieve a harmonious balance between performance and efficiency in single-image super-resolution. The SMU plays three key roles: LRU modulation, spatial categorization, and feature enhancement through learned prototype. Extensive experiments demonstrate that our method quantitatively and qualitatively surpasses recent state-of-the-art methods. Notably, our approach achieves superior performance with computational complexity on par with existing methods. The source code and models are available at https://github.com/MingyuChoi-run/LSM
Chinese Translation
线性递归单元（Linear Recurrent Unit, LRU）采用稳定线性递归的原则性公式设计，在长距离依赖任务中展现出良好的准确性和鲁棒性。然而，其静态参数化和单次扫描方法限制了其在二维视觉任务中的适用性。在本研究中，我们提出了一种基于LRU的恢复网络，结合语义调制单元（Semantic Modulating Unit, SMU），以在单幅图像超分辨率中实现性能与效率之间的和谐平衡。SMU发挥了三个关键作用：LRU调制、空间分类和通过学习原型进行特征增强。大量实验表明，我们的方法在定量和定性上均优于近期的最先进方法。值得注意的是，我们的方法在计算复杂度上与现有方法相当，但性能更为卓越。源代码和模型可在 https://github.com/MingyuChoi-run/LSM 获取。
View on arXiv Download PDF AI Translation
cs.CV / 30 / 2606.19908

Gaussian Process Prior Variational Autoencoder for Endoscopic Videos

用于内窥镜视频的高斯过程先验变分自编码器
De Boi, Ivan, Shi, Xinxing, Jiang, Xiaoyu, Jaspers, Tim J. M., Caetano, Francisco, Alvarez, Mauricio A., van der Sommen, Fons, Van der Jeught, Sam
Abstract
Endoscopic video analysis is essential for gastrointestinal diagnosis and computer-assisted interventions, but video sequences are routinely degraded by specular reflections, motion artifacts, and missing frames. These transient corruptions can distract clinicians, reduce image interpretability, and disrupt downstream tasks such as 3D reconstruction and navigation. Effective restoration therefore requires methods that exploit temporal continuity rather than treating frames in isolation. We introduce a Gaussian Process Prior Variational Autoencoder (GPVAE) framework for endoscopic video restoration that replaces the standard factorized latent prior with a temporal Gaussian process prior, enabling interpolation of missing frames with uncertainty-aware reconstruction. The framework combines endoscopy-specific encoders, including a convolutional EndoVAE backbone and pretrained Vision Transformer encoders from GastroNet-5M, with two scalable GP approximations: Hierarchical Prior Approximation (HPA) and Sparse Precision Approximation (SPA). Specular reflections are handled using a DUCKNet-based masking pipeline that excludes corrupted pixels from the reconstruction objective. On the C3VDv2 colonoscopy dataset, the best GPVAE variants reduced image reconstruction RMSE by 21.9\% on average, and by up to 26.1\%, relative to matched VAE baselines. Downstream trajectory RMSE was reduced by 12.7\% on average across classical visual odometry and a pretrained PoseNet, at an average increase of 27.3\% in training time per epoch. Finally, the GP posterior provides per-frame uncertainty estimates that reflect temporal support and offer a confidence signal for restored frames.
Chinese Translation
内窥镜视频分析对于胃肠道诊断和计算机辅助干预至关重要，但视频序列通常受到镜面反射、运动伪影和缺失帧的影响。这些瞬时干扰可能会分散临床医生的注意力，降低图像可解释性，并干扰下游任务，如三维重建和导航。因此，有效的恢复方法需要利用时间连续性，而不是将帧视为孤立的个体。我们提出了一种高斯过程先验变分自编码器（Gaussian Process Prior Variational Autoencoder, GPVAE）框架，用于内窥镜视频恢复，该框架用时间高斯过程先验替代了标准的分解潜在先验，从而实现了具有不确定性感知重建的缺失帧插值。该框架结合了内窥镜特定的编码器，包括卷积EndoVAE主干网络和来自GastroNet-5M的预训练视觉变换器编码器，以及两种可扩展的高斯过程近似：层次先验近似（Hierarchical Prior Approximation, HPA）和稀疏精度近似（Sparse Precision Approximation, SPA）。镜面反射通过基于DUCKNet的掩模管道进行处理，该管道将受损像素排除在重建目标之外。在C3VDv2结肠镜数据集上，最佳的GPVAE变体在图像重建均方根误差（RMSE）上平均降低了21.9%，最高降低了26.1%，相较于匹配的变分自编码器基线。在经典视觉里程计和预训练PoseNet的下游轨迹RMSE上，平均降低了12.7%，每个训练周期的平均训练时间增加了27.3%。最后，GP后验提供了每帧的不确定性估计，反映了时间支持，并为恢复的帧提供了置信信号。
View on arXiv Download PDF AI Translation
cs.CV / 31 / 2606.19915

SpatialSV: Internalizing Interpretable 3D Spatial Awareness in MLLMs via Task-Oriented Visual Supervision

SpatialSV：通过任务导向的视觉监督在多模态大型语言模型中内化可解释的三维空间意识
Tang, Jiayu, Zhou, Yuchen, Gou, Chao
Abstract
Unlocking the spatial intelligence of multimodal large language model (MLLMs) is crucial for understanding and interacting with the 3D world. Prevailing approaches typically inject spatial priors via external tools, which impose significant inference overhead, or rely on latent feature distillation, which remains uninterpretable and lacks fine-grained geometric constraints. To address these issues, we propose SpatialSV, a framework designed to internalize robust 3D spatial awareness within MLLMs while simultaneously offering inherent interpretability. Deviating from passive feature imitation, SpatialSV employs task-oriented visual supervision, compelling the model to actively lift its 2D visual features into explicit 3D representations, including depth maps, camera poses, and point clouds. Crucially, this 2D-to-3D lifting process provides a transparent window into the model's representations: the resulting 3D reconstructions serve as an intuitive proxy for visualizing and diagnosing the quality of the model's intrinsic spatial knowledge. Extensive experiments across multiple models and benchmarks demonstrate the effectiveness of SpatialSV in enhancing and interpreting MLLMs' spatial intelligence. Furthermore, the framework exhibits strong generalization in semi-supervised settings, validating its potential to leverage unlabeled visual data for scalable, interpretable spatial representation learning.
Chinese Translation
解锁多模态大型语言模型（MLLMs）的空间智能对于理解和与三维世界互动至关重要。现有的方法通常通过外部工具注入空间先验，这会带来显著的推理开销，或依赖于潜在特征蒸馏，这种方法仍然不可解释且缺乏细粒度的几何约束。为了解决这些问题，我们提出了SpatialSV，一个旨在内化MLLMs中强大三维空间意识的框架，同时提供内在的可解释性。SpatialSV不同于被动的特征模仿，采用任务导向的视觉监督，促使模型主动将其二维视觉特征提升为明确的三维表示，包括深度图、相机姿态和点云。至关重要的是，这一二维到三维的提升过程为模型的表示提供了一个透明的窗口：生成的三维重建作为可视化和诊断模型内在空间知识质量的直观代理。针对多个模型和基准的广泛实验表明，SpatialSV在增强和解释MLLMs的空间智能方面的有效性。此外，该框架在半监督设置中表现出强大的泛化能力，验证了其利用未标记视觉数据进行可扩展、可解释的空间表示学习的潜力。
View on arXiv Download PDF AI Translation
cs.CV / 32 / 2606.19927

CARE: Competence-Aware Reward Shaping for Adaptive Reasoning Length in Video-MLLMs

CARE：基于能力感知的奖励塑形框架用于视频多模态大语言模型中的自适应推理长度
Liu, Chengwen, Peng, Hao, Dang, Jisheng, Peng, Hong, Hu, Bin, Chua, Tat-Seng
Abstract
In multimodal video reasoning, reinforcement learning-based methods typically rely on simplistic and inflexible reasoning-length control strategies that fail to adapt to the model's evolving competence. This mismatch may suppress necessary exploration at early stages, while encouraging redundant reasoning and inefficient decoding once the model becomes more competent. In this paper, we propose CARE, a competence-aware reward shaping framework for adaptive reasoning length optimization in multimodal reasoning. Specifically, CARE maintains a smoothed competence estimate via an exponential moving average of pass rates, and uses it to route training into progressive stages that shift the reward preference from exploration-oriented long-form reasoning to efficiency-oriented concise reasoning. To avoid conflating verbosity with intrinsic task complexity, CARE further normalizes reasoning effort with batch-level statistics, and introduces a posterior amplifier to strengthen reward signals for unexpectedly strong performance on historically difficult samples. The proposed mechanism is seamlessly integrated into the GRPO training pipeline and incurs no additional inference-time overhead. Extensive experiments on multiple video reasoning and general video understanding benchmarks demonstrate that CARE consistently improves reasoning accuracy, stabilizes reinforcement learning, and significantly enhances token efficiency. Moreover, CARE exhibits a characteristic inverted-U trajectory of reasoning length during training, and yields shorter yet more informative reasoning traces at convergence, indicating effective adaptive allocation of reasoning budget. We provide the source code for our proposed CARE framework and experiments at https://github.com/1Pansy/Video-CARE.
Chinese Translation
在多模态视频推理中，基于强化学习的方法通常依赖于简单且不灵活的推理长度控制策略，这些策略无法适应模型不断发展的能力。这种不匹配可能在早期阶段抑制必要的探索，同时在模型变得更有能力后鼓励冗余推理和低效解码。本文提出了CARE，一个基于能力感知的奖励塑形框架，用于多模态推理中的自适应推理长度优化。具体而言，CARE通过对通过率的指数移动平均来维持平滑的能力估计，并利用该估计将训练引导至逐步阶段，逐渐将奖励偏好从以探索为导向的长形式推理转向以效率为导向的简洁推理。为避免将冗长与内在任务复杂性混淆，CARE进一步通过批量级统计来规范推理努力，并引入后验放大器以增强在历史上困难样本上表现意外强劲的奖励信号。所提出的机制无缝集成到GRPO训练流程中，并且不会增加额外的推理时间开销。在多个视频推理和一般视频理解基准上的广泛实验表明，CARE始终提高推理准确性，稳定强化学习，并显著增强令牌效率。此外，CARE在训练过程中表现出特征性的倒U型推理长度轨迹，并在收敛时产生更短但更具信息量的推理痕迹，表明有效的推理预算自适应分配。我们在https://github.com/1Pansy/Video-CARE提供了我们提出的CARE框架和实验的源代码。
View on arXiv Download PDF AI Translation
cs.CV / 33 / 2606.19932

Spatial-Aware Reduction Framework: Towards Efficient and Faithful Visual State Space Models

空间感知减缩框架：朝向高效且忠实的视觉状态空间模型
Lv, Jindi, Li, Aoyu, Zhou, Yuhao, Zhu, Zheng, Wang, Xiaofeng, Ye, Qing, Duan, Yueqi, Feng, Wentao, Lv, Jiancheng
Abstract
Mamba demonstrates strong efficiency in modeling long visual sequences. However, when token reduction is applied to structurally enhanced Mamba variants, these models exhibit a severe performance collapse. We attribute this degradation to the spatially agnostic nature of existing reduction methods, which violate the two-dimensional structural premise required by the selective scanning mechanism. In this work, we propose STORM, a spatial-aware token reduction framework designed to maintain structural integrity throughout the compression process. STORM reformulates reduction into a structured operation on spatial units, enforcing localized constraints to maintain both grid topology and neighborhood coherence. As a plug-and-play module, STORM equips existing reduction pipelines with explicit spatial awareness without any training. Empirical results demonstrate that STORM achieves state-of-the-art pruning accuracy across diverse vision Mamba backbones under training-free settings. Notably, STORM delivers a substantial accuracy recovery on VMamba, outperforming prior methods by up to 63.3\% in top-1 accuracy. Meanwhile, STORM incurs only a 1.0\% accuracy drop on PlainMamba, achieving performance comparable to ViT.
Chinese Translation
Mamba在建模长视觉序列方面表现出强大的效率。然而，当对结构增强的Mamba变体应用令牌减缩时，这些模型表现出严重的性能崩溃。我们将这种退化归因于现有减缩方法的空间无关特性，这违反了选择性扫描机制所需的二维结构前提。在本研究中，我们提出了STORM，一个空间感知的令牌减缩框架，旨在在压缩过程中保持结构完整性。STORM将减缩重新定义为对空间单元的结构化操作，强制施加局部约束以保持网格拓扑和邻域一致性。作为一个即插即用模块，STORM为现有的减缩管道提供了明确的空间感知，而无需任何训练。实证结果表明，STORM在无训练设置下在不同的视觉Mamba骨干网络中实现了最先进的剪枝准确率。值得注意的是，STORM在VMamba上实现了显著的准确率恢复，顶级准确率比之前的方法提高了多达63.3%。同时，STORM在PlainMamba上仅导致1.0%的准确率下降，达到了与ViT相当的性能。
View on arXiv Download PDF AI Translation
cs.CV / 34 / 2606.19934

Speeding up the annotation process in semantic segmentation industrial applications

加速语义分割工业应用中的标注过程
Fernandez-Moreno, Marta, Guerrero, Margarita, Rementeria, Rosalia, Mesejo, Pablo, Moreno, Raul
Abstract
Current machine learning models commonly require large and well-annotated datasets. However, the annotation process often becomes a bottleneck, with increased complexity leading to higher chances of human errors. Within this context, our goal in this paper is to leverage unsupervised algorithms to improve data annotation efficiency for complex semantic segmentation problems in industrial materials science. Previous research has quantified labeling time and others explored unsupervised methods. However, to the best of our knowledge, this is the first study to quantify how much unsupervised algorithms accelerate the labeling process. We aim to validate the extent to which this laborious process can be accelerated, focusing on semantic segmentation tasks that involve annotating each pixel of high-resolution images, such as the microstructure characterization challenge in materials science. Specifically, we demonstrate that by using unsupervised computer vision algorithms, the time required for the labeling process can be reduced from 170 hours to 37 hours, achieving an approximate reduction of 78\%. The dataset we work with includes large images of dimensions 1280x959 and 960x703, which further increases the complexity of the annotation task. Despite these challenges, we create and share the largest public steel microstructure segmentation dataset to date, available under MIT License with permanent DOI, contributing a fully annotated, high-resolution dataset to the field. Additionally, this is the first work to compare the labeling time from scratch (a common approach in previous studies) to the labeling time when using these unsupervised algorithms as a pre-annotation step. Furthermore, we provide a Deep Learning model trained on this dataset, validated by field experts, and deployed in an industrial setting, serving as an initial benchmark for this public dataset.
Chinese Translation
当前的机器学习模型通常需要大量且标注良好的数据集。然而，标注过程往往成为瓶颈，复杂性增加导致人为错误的可能性增高。在此背景下，本文的目标是利用无监督算法提高工业材料科学中复杂语义分割问题的数据标注效率。以往的研究量化了标注时间，其他研究则探索了无监督方法。然而，据我们所知，这是首个量化无监督算法加速标注过程的研究。我们旨在验证这一繁琐过程能够加速的程度，重点关注涉及对高分辨率图像每个像素进行标注的语义分割任务，例如材料科学中的微观结构表征挑战。具体而言，我们展示了通过使用无监督计算机视觉算法，标注过程所需的时间可以从170小时减少到37小时，约实现了78%的时间缩减。我们使用的数据集包括1280x959和960x703的大尺寸图像，进一步增加了标注任务的复杂性。尽管面临这些挑战，我们创建并分享了迄今为止最大的公共钢铁微观结构分割数据集，依据MIT许可证发布，并具有永久DOI，为该领域贡献了一个完全标注的高分辨率数据集。此外，这是首个将从零开始的标注时间（以往研究中的常见方法）与使用这些无监督算法作为预标注步骤的标注时间进行比较的研究。此外，我们提供了一个在该数据集上训练的深度学习模型，经过领域专家验证，并在工业环境中部署，作为该公共数据集的初步基准。
View on arXiv Download PDF AI Translation
cs.CV / 35 / 2606.19938

Triangular Consistency as a Universal Constraint for Learning Optical Flow

三角一致性作为学习光流的普适约束
Xiao, Yi, Coronel, Carlos Rodriguez, Zhan, Jing, Oskouie, Haniyeh Ehsani, Wong, Alex, Lao, Dong
Abstract
We propose triangular consistency as a first-principled constraint for optical flow, which is agnostic to network architecture, supervision type, and dataset, and applies to both image-pair and multi-frame settings. This simple but powerful constraint is to compose two flows to induce a third flow and enforce consistency among the three. The composed flows may arise from (i) image pairs, yielding cycle consistency; (ii) multiple video frames, producing longer-range motion through temporal chaining; or (iii) image pairs combined with controlled synthetic transformations, which becomes data augmentation. This triangular consistency introduces negligible computational overhead and requires no additional annotations. Since it is derived directly from the geometry of optical flow, it does not rely on model-specific assumptions and serves as a ``universal'' plug-and-play component for optical flow training. Experiments show consistent improvement across supervised, unsupervised, and transfer learning settings.
Chinese Translation
我们提出三角一致性作为光流的一种基本约束，该约束与网络架构、监督类型和数据集无关，适用于图像对和多帧设置。这一简单但强大的约束是将两个光流组合以引导第三个光流，并在三者之间强制一致性。组合的光流可能来自于（i）图像对，产生循环一致性；（ii）多个视频帧，通过时间链产生更长范围的运动；或（iii）结合受控合成变换的图像对，这成为数据增强。这种三角一致性引入的计算开销微乎其微，并且不需要额外的标注。由于它直接源于光流的几何特性，因此不依赖于特定模型的假设，并作为光流训练的“通用”即插即用组件。实验表明，在监督、无监督和迁移学习设置中均实现了一致的改进。
View on arXiv Download PDF AI Translation
cs.CV / 36 / 2606.19939

DiffMath: Symbol- and Graph-Aware Latent Diffusion Transformer for Handwritten Mathematical Expression Generation

DiffMath：符号和图形感知的潜在扩散变换器用于手写数学表达式生成
Pan, Wei, Zheng, Xuhan, Shi, Yilin, He, Huiguo, Cheng, Hiuyi, Peng, Dezhi, Liao, Minghui, Jin, Lianwen
Abstract
Handwritten Mathematical Expression Generation (HMEG) is challenging due to the complex two-dimensional layouts and long-range structural dependencies of mathematical expressions. Existing methods typically rely on explicit spatial supervision, such as symbol-level bounding boxes, which incurs high annotation costs and limits scalability. In this work, we propose DiffMath, a symbol- and graph-aware latent diffusion framework that leverages the hierarchical structure inherent in LaTeX as a structural prior, eliminating the need for positional supervision. First, we design a Relational Abstract Syntax Tree (RelAST), a generation-oriented representation that distills MathML trees into compact triplet sequences [S, R, D], where each token directly encodes a symbol identity, spatial relation, or nesting depth. Second, we introduce MathVAE, which learns structure-preserving latent representations through symbol-aware and relation-aware perceptual regularization, ensuring that the latent space captures both character semantics and spatial topology. Third, MathDiT performs conditional denoising in this structured latent space, further guided by a global symbol-count prior via Adaptive Layer Normalization (AdaLN) to improve structural coherence. Experiments show that DiffMath produces structurally consistent handwritten expressions, achieves superior performance over existing methods, and improves the accuracy of downstream OCR models through synthetic data augmentation.
Chinese Translation
手写数学表达式生成（HMEG）因数学表达式复杂的二维布局和长程结构依赖而面临挑战。现有方法通常依赖于显式的空间监督，例如符号级边界框，这导致高昂的标注成本并限制了可扩展性。在本研究中，我们提出了DiffMath，一种符号和图形感知的潜在扩散框架，利用LaTeX固有的层次结构作为结构先验，消除了对位置监督的需求。首先，我们设计了关系抽象语法树（RelAST），这是一种面向生成的表示，将MathML树提炼为紧凑的三元组序列[S, R, D]，其中每个标记直接编码符号身份、空间关系或嵌套深度。其次，我们引入了MathVAE，通过符号感知和关系感知的感知正则化学习保持结构的潜在表示，确保潜在空间同时捕捉字符语义和空间拓扑。第三，MathDiT在这个结构化的潜在空间中执行条件去噪，进一步通过自适应层归一化（AdaLN）引导全局符号计数先验，以提高结构一致性。实验表明，DiffMath生成的手写表达式在结构上是一致的，性能优于现有方法，并通过合成数据增强提高了下游OCR模型的准确性。
View on arXiv Download PDF AI Translation
cs.CV / 37 / 2606.19944

Timage: A Generative Text-in-Image Paradigm for Fine-Tuning Vision-Language Models

Timage：一种用于微调视觉-语言模型的生成文本-图像范式
Wu, Yifeng, Huang, Huimin, Wu, Ruiluo, Lin, Chunyi, Chen, Guanhua, Wu, Xian, Song, Wang, Han, Ruize
Abstract
Multimodal Large Language Models (MLLMs) often lose track of the right image regions during fine-grained spatial reasoning, because a textual query rarely carries any explicit geometric anchor into the pixel domain. Prevailing remedies either rewire the model's weights or pad the prompt with verbose instructions, yet neither reliably pins the language to the correct visual coordinates without eroding the backbone's general competence. We introduce Timage, a paradigm that recasts multimodal understanding as an alignment problem solved at the input: the query is drawn, as a typeset overlay, onto the image itself. The placement and appearance of this overlay are produced by a Constrained Schr\"odinger Bridge (cSB), an entropic optimal-transport sampler that factorizes layout synthesis into two coupled stochastic stages. The first stage, Region Search, transports noise toward query-aligned image zones while obeying a hard occlusion barrier that protects salient foreground content; the second stage, Appearance Shaping, sizes the glyphs through an ``ink-budget'' regularizer so that the rendered text stays legible and visually balanced. The resulting overlay behaves as an explicit attention beacon that channels the model's focus along spatial semantics. On the VMCBench suite, Timage paired with a modest 7B backbone clearly overtakes far larger proprietary systems as well as parameter-tuned baselines. The study positions deliberate input reconstruction as a powerful, architecture-neutral lever for strengthening multimodal reasoning.
Chinese Translation
多模态大型语言模型（MLLMs）在细粒度空间推理过程中常常无法准确定位正确的图像区域，因为文本查询很少携带任何明确的几何锚点进入像素域。现有的解决方案要么重新调整模型的权重，要么用冗长的指令填充提示，但都无法可靠地将语言固定在正确的视觉坐标上，而不削弱主干模型的整体能力。我们提出了Timage，这一范式将多模态理解重新构建为一个在输入端解决的对齐问题：查询作为一种排版叠加，绘制在图像本身上。该叠加的放置和外观由约束薛定谔桥（Constrained Schrödinger Bridge, cSB）生成，这是一种熵最优传输采样器，将布局合成分解为两个耦合的随机阶段。第一阶段，区域搜索，向与查询对齐的图像区域传输噪声，同时遵循一个硬遮挡边界，以保护显著的前景内容；第二阶段，外观塑造，通过“墨水预算”正则化器调整字形的大小，以确保渲染的文本保持清晰且视觉平衡。最终生成的叠加行为如同一个明确的注意力信标，引导模型的关注沿着空间语义。在VMCBench套件上，Timage与一个适度的7B主干模型配合，明显超越了更大规模的专有系统以及参数调优的基线。该研究将有意的输入重构定位为增强多模态推理的强大、架构中立的杠杆。
View on arXiv Download PDF AI Translation
cs.CV / 38 / 2606.19950

Confidence Calibration for Multimodal LLMs: An Empirical Study through Medical VQA

多模态大型语言模型的置信度校准：通过医学视觉问答的实证研究
Du, Yuetian, Wang, Yucheng, Kong, Ming, Liang, Tian, Long, Qiang, Chen, Bingdi, Zhu, Qiang
Abstract
Multimodal Large Language Models (MLLMs) show great potential in medical tasks, but their elicited confidence often misaligns with actual accuracy, potentially leading to misdiagnosis or overlooking correct advice. This study presents the first comprehensive analysis of the relationship between accuracy and confidence in medical MLLMs. It proposes a novel method that combines Multi-Strategy Fusion-Based Interrogation (MS-FBI) with auxiliary expert LLM assessment, aiming to improve confidence calibration in Medical Visual Question Answering (VQA). Experiments demonstrate that our method reduces the Expected Calibration Error (ECE) by an average of 40\% across three Medical VQA datasets, significantly enhancing MLLMs' reliability. The findings highlight the importance of domain-specific calibration for MLLMs in healthcare, offering a more trustworthy solution for AI-assisted diagnosis.
Chinese Translation
多模态大型语言模型（MLLMs）在医学任务中展现出巨大的潜力，但它们所产生的置信度往往与实际准确性不一致，这可能导致误诊或忽视正确建议。本研究首次全面分析了医学MLLMs中准确性与置信度之间的关系。我们提出了一种新颖的方法，将多策略融合基础的询问（Multi-Strategy Fusion-Based Interrogation, MS-FBI）与辅助专家LLM评估相结合，旨在改善医学视觉问答（Medical Visual Question Answering, VQA）的置信度校准。实验表明，我们的方法在三个医学VQA数据集上平均降低了预期校准误差（Expected Calibration Error, ECE）40%，显著增强了MLLMs的可靠性。研究结果强调了在医疗保健领域对MLLMs进行特定领域校准的重要性，为AI辅助诊断提供了更可信的解决方案。
View on arXiv Download PDF AI Translation
cs.CV / 39 / 2606.19958

SketchKeyAnime: Reference-anchored Sparse Key-Sketch Animation Synthesis

SketchKeyAnime：基于参考锚点的稀疏关键草图动画合成
Li, Meixi, Zhang, Xianlin, Zhang, Yue, Li, Xueming
Abstract
Traditional animation production relies heavily on manual drawing and iterative refinement, particularly for key-pose design, in-betweening, and character coloring. While existing animation and video generation methods have made notable progress, they typically depend on RGB boundary frames, dense frame-wise conditions, or complete sketch sequences, limiting their applicability under low-cost input conditions. We present SketchKeyAnime, a video diffusion framework for generating structurally controllable, appearance-consistent, and temporally coherent animations from sparse key-sketch inputs. Given a single reference RGB image and a few temporally indexed key sketches, SketchKeyAnime introduces a dual-branch conditioning mechanism to encode local geometric constraints alongside semantic-temporal context. It leverages Sketch Cross Attention to fuse reference image and sketch conditions with learnable gating, and incorporates an Adaptive Weighted Loss to strengthen supervision on key-sketch frames and line-art regions. Experimental results on the Aesthetic subset of Sakuga-42M show that our approach consistently outperforms representative animation interpolation and sketch-guided generation baselines. Compared to the best-performing baseline, SketchKeyAnime reduces EDMD by 31.9\% and FVD by 9.5\%, demonstrating superior sketch fidelity and temporal coherence, while achieving the best overall performance across most quantitative metrics. These results validate the proposed framework and highlight its potential for low-cost, highly controllable animation creation.
Chinese Translation
传统动画制作在很大程度上依赖于手动绘制和迭代优化，特别是在关键姿势设计、过渡动画和角色上色方面。尽管现有的动画和视频生成方法取得了显著进展，但它们通常依赖于RGB边界帧、密集的逐帧条件或完整的草图序列，这限制了其在低成本输入条件下的适用性。我们提出了SketchKeyAnime，这是一个视频扩散框架，旨在从稀疏的关键草图输入生成结构可控、外观一致和时间连贯的动画。给定一张单一的参考RGB图像和几个时间索引的关键草图，SketchKeyAnime引入了一种双分支条件机制，以编码局部几何约束和语义-时间上下文。它利用草图交叉注意力将参考图像和草图条件与可学习的门控相融合，并结合自适应加权损失，以加强对关键草图帧和线条艺术区域的监督。在Sakuga-42M的美学子集上的实验结果表明，我们的方法在代表性的动画插值和草图引导生成基线中始终表现优越。与表现最佳的基线相比，SketchKeyAnime将EDMD降低了31.9\%，FVD降低了9.5\\%，展现出卓越的草图保真度和时间连贯性，同时在大多数定量指标上实现了最佳整体性能。这些结果验证了所提出框架的有效性，并突显了其在低成本、高可控动画创作中的潜力。
View on arXiv Download PDF AI Translation
cs.CV / 40 / 2606.19961

Addressing Detail Bottlenecks in Latent Diffusion for RGB-to-SWIR Image Translation

解决潜在扩散模型在RGB到SWIR图像翻译中的细节瓶颈
Wang, Kaili, Dimitrievski, Martin, Salvador, Jose Maria, Stoffelen, Ben, Van Hamme, David, Goetschalckx, Lore
Abstract
Latent diffusion models (LDMs) enable efficient image-to-image translation but discard fine spatial details during compression, degrading downstream perception tasks. We identify two bottlenecks: the autoencoder, which loses spatial information, and the conditioning pathway, which further degrades the source signal through naive downsampling. We propose two lightweight, backbone-agnostic fixes: a Source-Conditioned Autoencoder (SCAE) that injects high-resolution source features into the decoder via skip connections, and a Learnable Guidance Encoder (LGE) that replaces naive downsampling with a learned conditioning signal. Evaluated on RGB-to-SWIR translation for driving scenes with two denoiser backbones (U-Net and DiT), our approach improves detection mAP by up to 2x over the latent diffusion baseline, with up to 3.4x gains on small objects (COCO-small, <32^2 px^2), while achieving state-of-the-art FID. We further show that FID and detection performance are poorly correlated, motivating multi-axis evaluation. Results generalise zero-shot to the public RASMD benchmark. We will publicly release test data with annotations, all checkpoints, and training code.
Chinese Translation
潜在扩散模型（LDMs）能够高效地进行图像到图像的翻译，但在压缩过程中会丢失细微的空间细节，从而降低下游感知任务的性能。我们识别出两个瓶颈：自编码器（autoencoder）丢失空间信息，以及条件通路（conditioning pathway）通过简单下采样进一步降低源信号的质量。我们提出了两种轻量级、与主干网络无关的解决方案：源条件自编码器（Source-Conditioned Autoencoder, SCAE），通过跳跃连接将高分辨率源特征注入解码器；可学习引导编码器（Learnable Guidance Encoder, LGE），用学习到的条件信号替代简单下采样。在针对驾驶场景的RGB到SWIR翻译中，使用两种去噪器主干网络（U-Net和DiT）进行评估，我们的方法在潜在扩散基线的基础上将检测mAP提高了最多2倍，对小物体（COCO-small, <32^2 px^2）的提升可达3.4倍，同时实现了最先进的FID。我们进一步表明，FID与检测性能之间的相关性较差，促使多维度评估。结果在公共的RASMD基准上实现了零样本泛化。我们将公开发布带注释的测试数据、所有检查点和训练代码。
View on arXiv Download PDF AI Translation
cs.CV / 41 / 2606.19965

ROSE: Benchmarking the Perception-to-Action Gap in Multimodal Models

ROSE：多模态模型中感知与行动之间差距的基准测试
Wang, Yihao, He, Zijian, Ren, Jie, Wang, Keze
Abstract
Multimodal large language models (MLLMs) are increasingly expected to act on visual information, yet the same scene may require different actions under different task contexts. How reliably can a model turn the same visual evidence into the action required by the current context? To answer this question, we introduce \textsc{ROSE} (\textbf{R}eference-conditioned \textbf{O}ddity and \textbf{S}ymbolic \textbf{E}xecution), a controlled benchmark that holds the visual scene fixed while varying region constraints and required symbolic outputs. Through coupled counting and coordinate-action tasks, \textsc{ROSE} tests whether models can infer an implicit majority reference and act on the resulting fine-grained visual evidence under changing contexts. Across nine recent MLLMs, performance drops by as much as 44.5 percentage points from counting-oriented tasks to region-conditioned action, despite 98.8\% human performance. The gap persists on paired scenes and regions for which the same model returns the correct count, while global-click and matched local controls show that coordinate grounding explains only part of the loss, revealing a distinct, model-dependent bottleneck in turning shared visual evidence into context-specific actions.
Chinese Translation
多模态大型语言模型（MLLMs）越来越被期望能够对视觉信息进行操作，但同一场景在不同任务上下文中可能需要不同的行动。模型在多大程度上能够将相同的视觉证据转化为当前上下文所需的行动？为了解答这个问题，我们引入了 extsc{ROSE}（ extbf{R}eference-conditioned extbf{O}ddity and extbf{S}ymbolic extbf{E}xecution），这是一个控制基准测试，在该测试中视觉场景保持不变，同时变化区域约束和所需的符号输出。通过耦合计数和坐标行动任务， extsc{ROSE}测试模型是否能够推断出隐含的多数参考，并在变化的上下文中对由此产生的细粒度视觉证据进行行动。在九个最新的MLLMs中，从以计数为导向的任务到区域条件行动，性能下降最多达44.5个百分点，尽管人类表现达到98.8%。在配对场景和区域中，尽管同一模型返回正确的计数，差距依然存在，而全局点击和匹配局部控制表明坐标基础仅解释了部分损失，揭示了在将共享视觉证据转化为上下文特定行动时存在一个独特的、依赖于模型的瓶颈。
View on arXiv Download PDF AI Translation
cs.CV / 42 / 2606.19966

Semantic-Anchored Evidential Fusion for Domain-Robust Whole-Slide Survival Analysis

基于语义锚定的证据融合用于领域鲁棒的全切片生存分析
Xing, Yucheng, Huang, Ling, Liu, Pei, Ma, Jingying, Xu, Jiaqing, He, Kai, Feng, Mengling
Abstract
Whole-slide images (WSIs) are widely used for computational cancer prognosis. However, most existing methods primarily focus on in-domain performance and fail to generalize across clinical centers. This limitation stems from their reliance on pixel-derived representations that are highly susceptible to domain-specific artifacts caused by staining protocols and scanner hardware. We hypothesize that high-level pathology semantics, such as tumor grade and micro-environmental architecture, provide a domain-invariant semantic representation that mirrors the robust diagnostic logic of human pathologists. Therefore, we propose a Semantic-Anchored Evidential Fusion Survival (SAEFS) framework, where SAEFS derives semantic anchors from WSIs via Visual Question Answering (VQA), employs a dual-stream WSI evidence extraction architecture, uses Dirichlet-based Subjective Logic to model uncertainty, and fuses semantic and visual evidence through a cautious conjunction rule to avoid overconfident fusion from correlated sources. Trained exclusively on one source domain and evaluated zero-shot across four unseen domains, SAEFS consistently outperforms state-of-the-art models both in prediction accuracy and reliability, improving the average C-index by 10.2%. Quantitative analyses further show that VQA-derived semantic features exhibit significantly lower cross-center divergence than pixel-derived features, highlighting their robustness for cross-center clinical applications.
Chinese Translation
全切片图像（WSIs）广泛应用于计算癌症预后。然而，大多数现有方法主要关注领域内的性能，未能在不同临床中心之间实现良好的泛化。这一局限性源于它们对像素派生表示的依赖，这些表示对由染色协议和扫描仪硬件引起的领域特定伪影高度敏感。我们假设，高级病理语义，如肿瘤分级和微环境结构，提供了一种领域不变的语义表示，反映了人类病理学家的稳健诊断逻辑。因此，我们提出了一种基于语义锚定的证据融合生存（SAEFS）框架，其中SAEFS通过视觉问答（VQA）从WSIs中提取语义锚，采用双流WSI证据提取架构，使用基于Dirichlet的主观逻辑来建模不确定性，并通过谨慎的结合规则融合语义和视觉证据，以避免来自相关来源的过度自信融合。在仅在一个源领域上训练并在四个未见领域上进行零样本评估的情况下，SAEFS在预测准确性和可靠性方面始终优于最先进的模型，平均C指数提高了10.2%。定量分析进一步表明，VQA派生的语义特征表现出显著低于像素派生特征的跨中心差异，突显了它们在跨中心临床应用中的鲁棒性。
View on arXiv Download PDF AI Translation
cs.CV / 43 / 2606.19970

CrossFlow: One-Step Generation Across Latent and Pixel Spaces

CrossFlow：跨潜在空间与像素空间的一步生成
Wang, Xiyuan, Zhang, Xiao, Li, Yang, Jiang, Ruoxi, Zhong, Zhao, Bo, Liefeng, Zhang, Muhan
Abstract
Most diffusion and flow-matching generators define the prior, probability path, and prediction target in the same representation space. Latent diffusion improves efficiency by moving this path into an autoencoder latent space, but the final sample is still produced by a separately trained decoder. This separation creates a mismatch: the generator is optimized for latent-space prediction, while final quality depends on how the decoder handles generated latents that may differ from clean encoder outputs. We introduce CrossFlow, a cross-space flow formulation that maps noisy latent inputs directly to pixel-space images. The key technical step is a velocity-free one-step objective: the latent trajectory defines the training path, but the supervised prediction is an image rather than a latent displacement. This lets one model act both as a one-step latent-to-pixel generator and as a decoder replacement for latent diffusion pipelines. On class-conditional ImageNet-1k at $256\times256$, CrossFlow-XL achieves 1.62 FID with one function evaluation. Ablations show that the latent encoder and pixel-space perceptual and adversarial losses are important for fidelity. These results indicate that cross-space flow objectives can combine the efficiency of latent representations with direct pixel-space supervision, without requiring a separate decoder at inference.
Chinese Translation
大多数扩散和流匹配生成器在同一表示空间中定义先验、概率路径和预测目标。潜在扩散通过将这一路径移入自编码器的潜在空间来提高效率，但最终样本仍然由单独训练的解码器生成。这种分离造成了不匹配：生成器针对潜在空间预测进行了优化，而最终质量则依赖于解码器如何处理可能与干净编码器输出不同的生成潜在值。我们提出了CrossFlow，一种跨空间流的公式，直接将噪声潜在输入映射到像素空间图像。关键的技术步骤是一个无速度的一步目标：潜在轨迹定义了训练路径，但监督预测是图像而非潜在位移。这使得一个模型既可以作为一步潜在到像素的生成器，也可以作为潜在扩散管道的解码器替代。在条件分类的ImageNet-1k数据集上，CrossFlow-XL在$256 imes256$的分辨率下实现了1.62的FID，且只需一次函数评估。消融实验表明，潜在编码器和像素空间的感知及对抗损失对保真度至关重要。这些结果表明，跨空间流目标可以将潜在表示的效率与直接的像素空间监督相结合，而无需在推理时使用单独的解码器。
View on arXiv Download PDF AI Translation
cs.CV / 44 / 2606.19985

Vision-Reasoning-Guided Occlusion Removal from Light Fields

基于视觉推理的光场遮挡去除
Youssef, Mohamed, Bimber, Oliver
Abstract
Occlusion-robust scene recovery remains a major challenge in computational imaging, particularly in natural environments where dense foreground vegetation severely limits visibility. We propose a vision-reasoning-guided light field occlusion removal framework that combines the visibility recovery capability of light field integration (LFI) with the semantic reasoning capacity of vision-language models (VLMs). Multi-view observations are first integrated via LFI to suppress foreground occlusions and produce an initial visibility-enhanced representation. A VLM is then incorporated as a conditional semantic prior to restore degraded structures and recover fine details, guided by the observed measurements. To improve recovery consistency and reduce hallucination artifacts, we introduce a multi-sample fusion strategy that aggregates multiple generated hypotheses into a unified estimate. Experimental results on synthetic and real-world datasets demonstrate state-of-the-art performance, achieving the highest average SSIM across four synthetic light field benchmark scenes (4-Syn) and strong generalization across structured and unstructured acquisition settings. These results highlight the effectiveness of combining physical imaging constraints with vision-language reasoning for robust perception under severe occlusion, with applicability to search-and-rescue and exploratory robotic navigation.
Chinese Translation
遮挡鲁棒场景恢复仍然是计算成像中的一个主要挑战，尤其是在自然环境中，密集的前景植被严重限制了可见性。我们提出了一种基于视觉推理的光场遮挡去除框架，该框架结合了光场积分（Light Field Integration, LFI）的可见性恢复能力与视觉语言模型（Vision-Language Models, VLMs）的语义推理能力。首先，通过LFI整合多视角观测，以抑制前景遮挡并生成初步的可见性增强表示。然后，结合VLM作为条件语义先验，以恢复退化的结构并恢复细节，受观测测量的指导。为了提高恢复的一致性并减少幻觉伪影，我们引入了一种多样本融合策略，将多个生成的假设聚合为统一的估计。对合成和真实世界数据集的实验结果表明，该方法在四个合成光场基准场景（4-Syn）中实现了最先进的性能，达到了最高的平均结构相似性指数（SSIM），并在结构化和非结构化采集设置中表现出强大的泛化能力。这些结果突显了将物理成像约束与视觉语言推理相结合在严重遮挡下实现鲁棒感知的有效性，具有在搜索与救援及探索性机器人导航中的应用潜力。
View on arXiv Download PDF AI Translation
cs.CV / 45 / 2606.20027

QG-MIL: A Gated Transformer Aggregator for Domain-Agnostic Multiple Instance Learning in Medical Imaging

QG-MIL：一种用于医学影像领域无关多实例学习的门控变换器聚合器
Zedda, Luca, Mura, Davide Antonio, Di Ruberto, Cecilia, Atzori, Maurizio, Dasdelen, Muhammed Furkan, Marr, Carsten, Loddo, Andrea
Abstract
Attention-based Multiple Instance Learning aggregators in medical imaging are prone to attention concentration, producing overconfident and unstable predictions. We introduce QG-MIL, a gated transformer aggregator that addresses this through four synergistic architectural components: RMSNorm-based pre-normalization, per-head QK normalization, fine-grained attention output gating, and SwiGLU-style feed-forward modules. Together, these design choices stabilize training and distribute attention more uniformly across instances without auxiliary losses, masking, or multi-stage regularization. We evaluate QG-MIL across six benchmarks spanning whole-slide pathology and cell-level hematology, covering two fundamentally different MIL scales. The best-performing QG-MIL variants outperform leading baselines on all six benchmarks, with an average improvement of +6.1 mean macro F1 points. Attention overlays and attention mass analysis confirm more distributed instance weighting. Ablation studies show that while individual components can match the full model on specific datasets, the QG-MIL design provides the most consistent cross-domain performance and tightest variance when compared to selected baselines. We release a configurable implementation to support reproducibility at: https://github.com/unica-visual-intelligence-lab/QG-MIL
Chinese Translation
基于注意力的医学影像多实例学习聚合器容易出现注意力集中现象，从而导致过于自信和不稳定的预测。我们提出了QG-MIL，这是一种门控变换器聚合器，通过四个协同的架构组件解决了这一问题：基于RMSNorm的预归一化、每头QK归一化、细粒度注意力输出门控和SwiGLU风格的前馈模块。这些设计选择共同稳定了训练，并在实例之间更均匀地分配注意力，而无需辅助损失、掩蔽或多阶段正则化。我们在六个基准测试中评估了QG-MIL，这些基准涵盖了全切片病理学和细胞级血液学，涉及两种根本不同的多实例学习规模。表现最佳的QG-MIL变体在所有六个基准测试中均超越了领先的基线，平均提升了+6.1个宏观F1分数点。注意力覆盖和注意力质量分析确认了更分散的实例加权。消融研究表明，尽管单个组件在特定数据集上可以与完整模型相匹配，但与选定基线相比，QG-MIL设计在跨领域性能上提供了最一致的表现和最紧密的方差。我们发布了一个可配置的实现，以支持可重复性，网址为：https://github.com/unica-visual-intelligence-lab/QG-MIL
View on arXiv Download PDF AI Translation
cs.CV / 46 / 2606.20032

ReA-OVCD: Reliability-Aware Open-Vocabulary Change Detection via Semantic and Spatial Refinement

ReA-OVCD：基于可靠性的开放词汇变化检测通过语义和空间精细化
Zhu, Hongming, Chen, Huaji, Du, Bowen, Liu, Sicong, Liu, Qin
Abstract
Unlike traditional remote sensing change detection that relies on predefined categories, Open-Vocabulary Change Detection (OVCD) identifies land cover changes flexibly using arbitrary text prompts. However, existing methods suffer from an inherent trade-off when modeling changes: instance-level comparison overlooks fine-grained semantic variations (e.g., partial building extensions), while direct pixel comparison proves unreliable, yielding unstable responses and boundary artifacts due to semantic ambiguity and spatial inconsistency. To this end, we propose an efficient training-free Reliability-Aware Open-Vocabulary Change Detection (ReA-OVCD) framework. It first derives candidate change regions from pixel-wise semantic discrepancies to ensure flexible and detailed localization. To ensure reliability, it subsequently introduces a collaborative refinement strategy to explicitly model change validity from both semantic and spatial perspectives. Specifically, we develop a Semantic Change Reasoning (SCR) module that reassesses changes by jointly analyzing distributional divergence and response variation, enabling the suppression of incidental inconsistencies while preserving reliable semantic shifts. In addition, a Boundary-aware Change Refinement (BCR) module is designed to mitigate artifacts stemming from boundary misalignment and uncertainty through validating whether candidate regions are supported by reliable interior pixels. Extensive experiments across multiple datasets (LEVIR-CD, WHU-CD, DSIFN, and SECOND) demonstrate that our method consistently outperforms state-of-the-art approaches, achieving $\mathrm{F}_{1}^{C}$ improvements of 2.13\% to 9.75\% with higher computational efficiency. The code is publicly available at \https://github.com/Funny0101/ReA-OVCD
Chinese Translation
与依赖预定义类别的传统遥感变化检测不同，开放词汇变化检测（OVCD）通过任意文本提示灵活地识别土地覆盖变化。然而，现有方法在建模变化时存在固有的权衡：实例级比较忽视了细粒度的语义变化（例如，部分建筑扩展），而直接的像素比较则不可靠，因语义模糊和空间不一致而产生不稳定的响应和边界伪影。为此，我们提出了一种高效的无训练的基于可靠性的开放词汇变化检测（ReA-OVCD）框架。该框架首先从像素级语义差异中推导候选变化区域，以确保灵活和详细的定位。为了确保可靠性，随后引入了一种协同精细化策略，从语义和空间两个角度明确建模变化的有效性。具体而言，我们开发了一个语义变化推理（SCR）模块，通过联合分析分布差异和响应变化重新评估变化，从而抑制偶然的不一致，同时保留可靠的语义转变。此外，设计了一个边界感知变化精细化（BCR）模块，通过验证候选区域是否由可靠的内部像素支持，来减轻因边界错位和不确定性引起的伪影。在多个数据集（LEVIR-CD、WHU-CD、DSIFN和SECOND）上的广泛实验表明，我们的方法始终优于最先进的方法，$ ext{F}_{1}^{C}$ 提升幅度为2.13\%到9.75\%，并具有更高的计算效率。代码已公开发布在 https://github.com/Funny0101/ReA-OVCD
View on arXiv Download PDF AI Translation
cs.CV / 47 / 2606.20035

PU-UNet: Stable Multiplicative Interactions for Medical Image Segmentation

PU-UNet：用于医学图像分割的稳定乘法交互
Li, Ziyuan, Sufyan, Osamah, Jaekel, Uwe, Dellen, Babette
Abstract
Many dense prediction networks rely on additive feature transformations and model higher-order feature interactions only implicitly. Product units provide an explicit mechanism for multiplicative feature modeling, but their logarithmic--exponential formulation can cause numerical instability, which has limited their use in deep dense prediction networks. In this work, we propose Product-Unit U-Net (PU-UNet), a residual U-Net that integrates stable product-unit residual blocks into rich low-resolution stages for medical image segmentation. The proposed formulation combines smooth positivity mapping with log-domain clipping, enabling stable multiplicative feature learning with negligible computational overhead. On ISIC 2018, Kvasir-SEG, and BUSI, PU-UNet achieves Dice scores of 0.942, 0.959, and up to 0.925, respectively. Compared with a matched Residual U-Net baseline, PU-UNet consistently improves Dice and IoU while keeping parameters, FLOPs, and inference latency nearly unchanged, and reduces the image-level false-positive rate on normal BUSI cases from 0.077 to zero. Ablation studies suggest that the gains are associated with product-unit interactions, are strongest under low-resolution placement, and benefit from the proposed stabilization design. These results suggest that stable product-unit residual learning can be an effective way to enhance U-Net-style segmentation networks with explicit multiplicative interactions.
Chinese Translation
许多密集预测网络依赖于加性特征变换，仅隐式建模高阶特征交互。乘积单元提供了一种显式的乘法特征建模机制，但其对数-指数形式可能导致数值不稳定，这限制了它们在深度密集预测网络中的应用。在本研究中，我们提出了产品单元 U-Net（PU-UNet），这是一种残差 U-Net，将稳定的乘积单元残差块集成到丰富的低分辨率阶段中，用于医学图像分割。所提出的公式结合了平滑的正映射与对数域裁剪，使得乘法特征学习稳定且计算开销微乎其微。在 ISIC 2018、Kvasir-SEG 和 BUSI 数据集上，PU-UNet 分别达到了 0.942、0.959 和高达 0.925 的 Dice 分数。与匹配的残差 U-Net 基线相比，PU-UNet 一直在提高 Dice 和 IoU 的同时，保持参数、FLOPs 和推理延迟几乎不变，并将正常 BUSI 案例的图像级假阳性率从 0.077 降至零。消融研究表明，增益与乘积单元交互有关，在低分辨率放置下最为显著，并受益于所提出的稳定化设计。这些结果表明，稳定的乘积单元残差学习可以有效增强具有显式乘法交互的 U-Net 风格分割网络。
View on arXiv Download PDF AI Translation
cs.CV / 48 / 2606.20044

FUSE: Frequency-domain Unification and Spectral Energy Alignment for Multi-modal Object Re-Identification

FUSE：频域统一与光谱能量对齐用于多模态目标重识别
Qi, Xuanhao, Luan, Tom H., Zhang, Yukang, Zheng, Jinkai, Su, Zhou, Li, Shuwei, Tan, Lei
Abstract
Despite significant progress in multi-modal Re-Identification (ReID), existing methods tend to emphasize low-frequency cues. Consequently, they focus on attributes such as color, illumination, and coarse appearance, while overlooking mid and high-frequency structures that encode geometric, textural, and identity-discriminative details. This imbalance leads to incomplete spectral representations and unstable cross-modal alignment. To overcome these limitations, we introduce FUSE, a frequency-domain framework that reformulates multi-modal ReID as a two-stage process of spectral disentanglement and energy alignment. The proposed Spectral Decomposition Module (SDM) adaptively partitions features into low, mid, and high-frequency subspaces, enabling hierarchical spectral modeling. The Cross-Modal Alignment Module (CAM) further enforces energy alignment and subspace complementarity across modalities via frequency-consistency regularization. In addition, FUSE incorporates learnable frequency modulation to enhance robustness under varying illumination and heterogeneous sensor conditions. Extensive experiments on RGBNT201, RGBNT100, and MSVR310 show that FUSE achieves 9.1\% mAP and 9.5\% Rank-1 improvements, establishing an interpretable frequency-domain paradigm for multi-modal representation learning.
Chinese Translation
尽管在多模态重识别（ReID）方面取得了显著进展，现有方法往往强调低频线索。因此，它们关注于颜色、光照和粗略外观等属性，而忽视了编码几何、纹理和身份区分细节的中高频结构。这种不平衡导致光谱表示不完整和跨模态对齐不稳定。为克服这些局限性，我们提出了FUSE，一个频域框架，将多模态ReID重新构建为光谱解缠结和能量对齐的两阶段过程。所提出的光谱分解模块（Spectral Decomposition Module, SDM）自适应地将特征划分为低、中和高频子空间，从而实现分层光谱建模。跨模态对齐模块（Cross-Modal Alignment Module, CAM）进一步通过频率一致性正则化加强了不同模态之间的能量对齐和子空间互补。此外，FUSE还结合了可学习的频率调制，以增强在不同光照和异构传感器条件下的鲁棒性。在RGBNT201、RGBNT100和MSVR310上的大量实验表明，FUSE实现了9.1\%的mAP和9.5\%的Rank-1提升，为多模态表示学习建立了一个可解释的频域范式。
View on arXiv Download PDF AI Translation
cs.CV / 49 / 2606.20045

See-and-Reach: Precise Vision-Language Navigation for UAVs within the Field of View

看见与到达：在视野内的无人机精确视觉-语言导航
Xue, Fanfu, Yu, En, Shen, Yantian, Hu, Zhikun, Wang, Hongjun, Yang, Yang, Wang, Xindi, Sun, Jiande
Abstract
UAV Vision-Language Navigation (UAV-VLN) is typically formulated as a holistic search-and-reach problem, where long-range target discovery and final target approach are optimized and evaluated jointly. This formulation makes it difficult to assess a critical capability of aerial embodied agents, namely whether a UAV can accurately ground a visible target and translate vision-language evidence into precise 3D motion once the target enters its field of view. To address this limitation, we introduce UAV-VLN-FOV, a target-visible navigation task that isolates the see-and-reach stage and enables a more diagnostic evaluation of terminal reaching ability. We further propose 3DG-VLN, a vision-language waypoint prediction framework guided by dynamic 3D direction cues to enhance fine-grained visual grounding and spatial direction alignment for precise target reaching. Specifically, 3DG-VLN adaptively processes high-resolution front-view and downward-view observations to preserve fine-grained visual and geometric details for target grounding. It also updates the target-relative direction online during closed-loop navigation, allowing the agent to maintain spatial alignment with the target and reduce accumulated direction drift. To support this task, we construct a dedicated high-resolution benchmark which contains 2,717 trajectories with target-oriented high-level instructions, high-resolution front-view and downward-view egocentric observations, and continuous 3D waypoint annotations. Experiments show that 3DG-VLN outperforms competitive UAV-VLN baselines, achieving a 13.82\% improvement in success rate. Real-world trials further demonstrate the potential of 3DG-VLN for practical see-and-reach navigation. The source code and benchmark are available at https://github.com/xuefanfu/3DG-VLN.
Chinese Translation
无人机视觉-语言导航（UAV-VLN）通常被表述为一个整体的搜索与到达问题，在该问题中，远程目标发现和最终目标接近被共同优化和评估。这种表述使得评估空中具身智能体的一个关键能力变得困难，即无人机在目标进入其视野后，能否准确地确定可见目标并将视觉-语言证据转化为精确的三维运动。为了解决这一局限性，我们引入了UAV-VLN-FOV，这是一项目标可见导航任务，旨在孤立看见与到达阶段，从而实现对终端到达能力的更具诊断性的评估。我们进一步提出了3DG-VLN，这是一种受动态三维方向线索指导的视觉-语言航点预测框架，旨在增强精细视觉定位和空间方向对齐，以实现精确的目标到达。具体而言，3DG-VLN自适应地处理高分辨率的前视和俯视观测，以保留目标定位所需的精细视觉和几何细节。在闭环导航过程中，它还在线更新目标相对方向，使智能体能够保持与目标的空间对齐，并减少累积方向漂移。为了支持这一任务，我们构建了一个专门的高分辨率基准，其中包含2,717条轨迹，配有面向目标的高层次指令、高分辨率的前视和俯视自我中心观测，以及连续的三维航点注释。实验表明，3DG-VLN在成功率上比竞争性的UAV-VLN基线提高了13.82%。实际试验进一步展示了3DG-VLN在实际看见与到达导航中的潜力。源代码和基准数据可在https://github.com/xuefanfu/3DG-VLN获取。
View on arXiv Download PDF AI Translation
cs.CV / 50 / 2606.20076

Variable-Length Tokenization via Learnable Global Merging for Diffusion Transformers

通过可学习全局合并实现的可变长度标记化用于扩散变换器
Lee, Dong Hoon, Hong, Seunghoon
Abstract
Latent Diffusion Models (LDMs) have become dominant in visual synthesis, but their quality-compute trade-off is largely constrained by the tokenizer's fixed compression ratio. Variable-length tokenizers (VLTs) promise adaptive compression by varying token counts, allowing diffusion models to flexibly balance quality and compute. However, conventional VLTs modulate length by truncating ordered token sequences, which makes token semantics depend on token position and breaks representational alignment across lengths. This leads to a cross-length shift in the latent distribution that hinders a single variable-length diffusion model from operating effectively. To address this, we propose a novel variable-length tokenizer that modulates length by merging tokens. We show that encouraging similar tokens to merge enables direct cross-length representation alignment when the diffusion transformer operates according to the merging pattern. Since conventional merging methods are data-dependent, making the merging pattern inaccessible during generation, we introduce learnable global merging, which is data-independent, to ensure compatibility with diffusion transformers. On ImageNet 256$\times$256 generation, our merging-based variable-length tokenizer integrated with a diffusion transformer achieves a superior gFID-compute trade-off compared to prior VLT methods. Code is available at [this https URL](https://github.com/movinghoon/lgm)
Chinese Translation
潜在扩散模型（Latent Diffusion Models, LDMs）在视觉合成中已成为主流，但其质量与计算的权衡在很大程度上受到标记器固定压缩比的限制。可变长度标记器（Variable-Length Tokenizers, VLTs）通过变化标记数量承诺实现自适应压缩，使扩散模型能够灵活平衡质量与计算。然而，传统的VLT通过截断有序标记序列来调节长度，这使得标记语义依赖于标记位置，并打破了不同长度之间的表示对齐。这导致潜在分布在不同长度之间发生偏移，从而使单一的可变长度扩散模型无法有效运作。为了解决这个问题，我们提出了一种新颖的可变长度标记器，通过合并标记来调节长度。我们表明，鼓励相似标记合并能够在扩散变换器根据合并模式操作时实现直接的跨长度表示对齐。由于传统的合并方法依赖于数据，使得在生成过程中无法访问合并模式，我们引入了可学习的全局合并，这种方法与数据无关，以确保与扩散变换器的兼容性。在ImageNet 256×256生成任务中，我们的基于合并的可变长度标记器与扩散变换器的结合相比于之前的VLT方法实现了更优的gFID-计算权衡。代码可在[此链接](https://github.com/movinghoon/lgm)获取。
View on arXiv Download PDF AI Translation
cs.CV / 51 / 2606.20077

The Hidden Evolution of Disguised Visual Context inside the VLM

隐藏在视觉语言模型中的伪装视觉上下文的演变
Suharitdamrong, Wish, Alex, Tony, Awais, Muhammad, Atito, Sara
Abstract
Visual tokens enter Large Language Models (LLMs) as raw, foreign signals. How they are transformed into meaningful representations and interact with the language space depends entirely on the integration architecture. Whether by treating visual tokens as in-context prompts within the input sequence or injecting them directly into the LLM's intermediate layers. A controlled comparison and understanding of how these architectural choices affect visual information and its internal transformation to integrate with the LLM remains underexplored. We provide a fair comparison by evaluating in-context and layer-wise injection VLM integration paradigms under identical training conditions across single image, multi-image, and video benchmarks. In doing so, we uncover a hidden evolution where visual tokens enter the LLM as disguised visual context, raw representations lacking linguistic structure, but are progressively reshaped depending on the integration paradigm, each capturing fundamentally different frequency characteristics of the visual signal. We show that this evolution inside the LLM determines what visual features the VLM can utilize effectively, how visual representations align with the language space, and ultimately how each paradigm performs across different tasks. We further demonstrate that attention allocation alone is insufficient, and that performance is driven by the quality of visual representations at each layer.
Chinese Translation
视觉标记作为原始的、外部的信号进入大型语言模型（LLMs）。它们如何转化为有意义的表征并与语言空间相互作用，完全依赖于集成架构。无论是将视觉标记视为输入序列中的上下文提示，还是直接注入到LLM的中间层，对这些架构选择如何影响视觉信息及其内部转化以与LLM集成的控制比较和理解仍然未被充分探讨。我们通过在单图像、多图像和视频基准下，在相同训练条件下评估上下文注入和层级注入的VLM集成范式，提供了一个公平的比较。在此过程中，我们揭示了一种隐藏的演变，即视觉标记作为伪装的视觉上下文进入LLM，作为缺乏语言结构的原始表征，但根据集成范式逐步被重塑，每种范式捕捉到视觉信号的根本不同的频率特征。我们展示了这种在LLM内部的演变决定了VLM能够有效利用哪些视觉特征，视觉表征如何与语言空间对齐，以及最终每种范式在不同任务中的表现。我们进一步证明，仅仅依靠注意力分配是不够的，性能是由每一层视觉表征的质量驱动的。
View on arXiv Download PDF AI Translation
cs.CV / 52 / 2606.20083

Holo-World: Unified Camera, Object and Weather Control for Video World Model

全息世界：视频世界模型的统一相机、物体和天气控制
Yin, Xiangchen, Sun, Wenzhang, Yuan, Jiahui, Liu, Zijie, Chen, Yinda, Li, Wei, Kai, Dachun, Wang, Chunfeng, Sun, Xiaoyan
Abstract
Video world models are moving toward preserving an observed world under controllable camera and object motion while allowing its environmental state to change. Yet these controls remain isolated, and weather generation typically relies on a source video or reconstructed scene that already specifies future structure. We study a first-frame-anchored source-to-state setting, where the model starts from a single image and follows explicit camera and object controls and an optional weather instruction, then generates a video that either preserves the source world or transfers it to a target weather state. To address these challenges, we first build HoloStateData, a state video dataset that turns diverse videos into unified control samples for camera, object, and weather supervision. Second, we introduce Holo-World, a unified controllable video world model that jointly controls scene from a single image. Its Unified Scene Adapter factorizes world preservation and weather transfer into distinct parameter subspaces, using rendered background, geometry buffers, and object controls to maintain controlled scene structure while modeling weather-dependent appearance and particle effects. Additionally, Scene-Weather Decomposed CFG guides scene and weather residuals separately, strengthening target weather effects without over-amplifying the full condition. Quantitative and qualitative experiments demonstrate that Holo-World maintains precise camera and object control with consistent scene structure while transferring scenes into diverse target weather state, outperforming video-to-video weather editing baselines on weather-state generation. Our project page is available at \url{https://xiangchenyin.github.io/Holo-World/}.
Chinese Translation
视频世界模型正朝着在可控的相机和物体运动下保持观察到的世界，同时允许其环境状态变化的方向发展。然而，这些控制仍然是孤立的，天气生成通常依赖于已经指定未来结构的源视频或重建场景。我们研究了一种以第一帧为锚的源到状态设置，其中模型从单幅图像开始，遵循明确的相机和物体控制以及可选的天气指令，然后生成一个视频，该视频要么保持源世界，要么将其转移到目标天气状态。为了解决这些挑战，我们首先构建了HoloStateData，这是一个状态视频数据集，将多样的视频转化为相机、物体和天气监督的统一控制样本。其次，我们介绍了Holo-World，一个统一的可控视频世界模型，它从单幅图像共同控制场景。其统一场景适配器将世界保持和天气转移分解为不同的参数子空间，利用渲染背景、几何缓冲区和物体控制来维持受控的场景结构，同时建模天气依赖的外观和粒子效果。此外，场景-天气分解的CFG分别引导场景和天气残差，增强目标天气效果而不至于过度放大完整条件。定量和定性实验表明，Holo-World在保持精确的相机和物体控制以及一致的场景结构的同时，将场景转移到多样的目标天气状态，超越了天气状态生成的基于视频到视频的天气编辑基线。我们的项目页面可访问 {https://xiangchenyin.github.io/Holo-World/}。
View on arXiv Download PDF AI Translation
cs.CV / 53 / 2606.20092

EventVLA: Event-Driven Visual Evidence Memory for Long-Horizon Vision-Language-Action Policies

EventVLA：基于事件驱动的视觉证据记忆用于长时间跨度的视觉-语言-动作策略
Yang, Ganlin, Tu, Zhangzheng, Yang, Yuqiang, Mao, Sitong, Dong, Junyi, Chen, Tianxing, Peng, Jiaqi, Xiong, Jing, Cao, Jiafei, Dai, Jifeng, Zhou, Wengang, Mu, Yao, Wang, Tai
Abstract
Memory remains a critical bottleneck for long-horizon robotic manipulation, as standard Vision-Language-Action (VLA) policies often fail when task-relevant cues become occluded or unobservable over time. While existing memory-augmented methods utilize historical context, they either suffer from severe information bottlenecks, incur high latency via decoupled dual systems, or rely on unselective buffers that accumulate massive visual redundancies. To address these limitations, we introduce EventVLA, an end-to-end framework founded on the concept of sparse visual evidence memory that comprises two core components: foundational visual anchors to retain initial and short-term contexts, and a dynamic Keyframe Evidence Memory (KEM) module. Specifically, KEM directly predicts future keyframe probabilities from the VLA's latent embeddings to autonomously capture and store sparse, task-critical visual events. This foresight-driven mechanism empowers the policy to dynamically evaluate the future causal utility of current observations, preserving transient visual evidence before it becomes unobservable. Furthermore, we propose RoboTwin-MeM, a diagnostic benchmark specifically designed to evaluate non-Markovian manipulation tasks with interactive visual evidence. Extensive evaluations show that across 17 memory-requiring simulation tasks and 4 real-world bimanual tasks, EventVLA achieves an average success rate improvement of +40% over state-of-the-art memory-augmented VLAs.
Chinese Translation
记忆仍然是长时间跨度机器人操作的一个关键瓶颈，因为标准的视觉-语言-动作（VLA）策略在任务相关线索随着时间的推移变得被遮挡或不可观察时，往往会失败。虽然现有的增强记忆方法利用历史上下文，但它们要么遭受严重的信息瓶颈，要么通过解耦的双系统引入高延迟，或者依赖于不具选择性的缓冲区，这些缓冲区积累了大量的视觉冗余。为了解决这些局限性，我们提出了EventVLA，一个基于稀疏视觉证据记忆概念的端到端框架，包含两个核心组件：基础视觉锚点用于保留初始和短期上下文，以及一个动态关键帧证据记忆（Keyframe Evidence Memory, KEM）模块。具体而言，KEM直接从VLA的潜在嵌入中预测未来关键帧的概率，以自主捕捉和存储稀疏的、任务关键的视觉事件。这种以预见为驱动的机制使得策略能够动态评估当前观察的未来因果效用，在视觉证据变得不可观察之前保留瞬态视觉证据。此外，我们提出了RoboTwin-MeM，一个专门设计的诊断基准，用于评估具有交互视觉证据的非马尔可夫操作任务。广泛的评估表明，在17个需要记忆的仿真任务和4个真实世界的双手任务中，EventVLA的平均成功率比最先进的增强记忆VLA提高了40%。
View on arXiv Download PDF AI Translation
cs.CV / 54 / 2606.20094

MakeupMirror: Improving Facial Attribute Preservation in Diffusion Models for Makeup Transfer

化妆镜：提升扩散模型中化妆转移的面部特征保留能力
Andreou, Nefeli, Martínez-González, Angel, Sternig, Sabine, Guillaumin, Matthieu, Antonakos, Epameinondas, Opitz, Michael
Abstract
Makeup transfer models enable fun augmented reality (AR) experiences as well as virtual try-on (VTO) for online makeup shopping. While recent state-of-the-art diffusion based solutions such as Stable-Makeup dramatically improve the accuracy and realism of makeup transfer, they still face limitations in identity and skin color preservation, making production-level VTO for makeup shopping unrealistic. In this work, we propose MakeupMirror, a diffusion-based approach to makeup transfer that makes significant progress towards preserving facial features and skin tone. We introduce several technical innovations over Stable-Makeup: (1) integration of facial geometry conditioning with ControlNets to maintain facial fidelity; (2) region-specific makeup transfer control to enable precise makeup application across facial regions such as skin, eyes and lips; (3) skin tone-based makeup transfer modulation that prevent skin tone alteration in cross-subject transfer scenarios; and (4) integration of a Levenberg-Marquardt Langevin sampler to speed up inference while maintaining generation quality. Our experiments on CPM-Real, Makeup Wild, and (herein newly collected, more diverse) MakeupSelfies datasets show that MakeupMirror improves relative facial recognition similarity by +60%, reduces relative skin tone difference by -50% over Stable-Makeup, with a latency of 0.7s, while achieving expert acceptance rate of 94% across core facial identity preservation criteria.
Chinese Translation
化妆转移模型为增强现实（AR）体验以及在线化妆购物的虚拟试妆（VTO）提供了有趣的可能性。尽管最近的最先进的基于扩散的解决方案如Stable-Makeup显著提高了化妆转移的准确性和真实感，但在身份和肤色保留方面仍面临限制，使得化妆购物的生产级VTO变得不切实际。在本研究中，我们提出了化妆镜（MakeupMirror），这是一种基于扩散的化妆转移方法，在保留面部特征和肤色方面取得了显著进展。我们在Stable-Makeup的基础上引入了几项技术创新：（1）将面部几何条件与ControlNets结合，以保持面部的真实感；（2）区域特定的化妆转移控制，以实现对皮肤、眼睛和嘴唇等面部区域的精确化妆应用；（3）基于肤色的化妆转移调制，以防止在跨主体转移场景中肤色的改变；（4）集成Levenberg-Marquardt Langevin采样器，以加快推理速度，同时保持生成质量。我们在CPM-Real、Makeup Wild和（在此新收集的、更具多样性的）MakeupSelfies数据集上的实验表明，化妆镜（MakeupMirror）相较于Stable-Makeup提高了面部识别相似度60%，减少了肤色差异50%，延迟为0.7秒，同时在核心面部身份保留标准下实现了94%的专家接受率。
View on arXiv Download PDF AI Translation
cs.CV / 55 / 2606.20095

Stitching and dimensionality effects on large artificially generated volume datasets

拼接和维度效应对大型人工生成体积数据集的影响
von Chamier, Lucas, Albrecht, Jan Philipp, Kainmüller, Dagmar
Abstract
Generating large images via deep learning requires patching input data to accommodate hardware memory limitations, then assembling output patches, a process that can introduce stitching artifacts when neighboring patches do not align at borders. While these artifacts are known to affect segmentation tasks, their impact on generative models for style-transfer remains poorly understood. We investigated three stitching approaches and two patch dimensionalities (2D vs 3D) using cycleGAN models trained on cryo-electron microscopy datasets. We evaluated both perceptual quality and performance on downstream mitochondria segmentation. Our key findings reveal that: (1) FID scores fail to detect subtle stitching artifacts that significantly impact downstream segmentation performance, (2) 3D models with artifact-free stitching marginally outperform 2D models on downstream tasks, though the improvement barely justifies the computational cost, and (3) 2D models train more stably due to larger batch sizes. Additionally, we demonstrate that ensembling predictions from three orthogonal directions can improve low-quality volumes but provides no benefit for high-quality outputs. These results demonstrate that maximizing generative model performance on large scientific datasets requires careful consideration and mitigation of stitching artifacts, and that perceptual metrics alone are insufficient for evaluating domain adaptation quality in biomedical imaging.
Chinese Translation
通过深度学习生成大图像需要对输入数据进行拼接，以适应硬件内存限制，然后组装输出补丁，这一过程可能在相邻补丁的边界处引入拼接伪影。尽管这些伪影已知会影响分割任务，但它们对风格迁移生成模型的影响仍然了解不足。我们使用在冷冻电子显微镜数据集上训练的 cycleGAN 模型，研究了三种拼接方法和两种补丁维度（2D 与 3D）。我们评估了感知质量和下游线粒体分割性能。我们的主要发现表明：(1) FID 分数无法检测到显著影响下游分割性能的细微拼接伪影，(2) 无伪影拼接的 3D 模型在下游任务上略微优于 2D 模型，尽管这种提升几乎无法证明其计算成本的合理性，以及 (3) 由于批量大小较大，2D 模型的训练更为稳定。此外，我们还证明了从三个正交方向集成预测可以改善低质量体积，但对高质量输出没有任何益处。这些结果表明，最大化大型科学数据集上生成模型的性能需要仔细考虑和减轻拼接伪影，并且仅依赖感知指标不足以评估生物医学成像中的领域适应质量。
View on arXiv Download PDF AI Translation
cs.CV / 56 / 2606.20100

WeGenBench: A Multidimensional Diagnostic Benchmark towards Text-to-Image Model Optimization

WeGenBench：面向文本到图像模型优化的多维诊断基准
Liang, Qian, Li, Xiaomin, Zhang, Ying, Xu, Jia, Ni, Lihao, Li, Hongrui, Li, Jingjing, Lyu, Jing, Li, Chen
Abstract
Recent text-to-image generation models have demonstrated remarkable capabilities in synthesizing highly realistic images from text inputs alone. Although existing benchmarks can evaluate the generation capabilities of various models to some extent, they struggle to comprehensively and accurately measure performance across multiple dimensions, often failing to reveal the inherent deficiencies of models in specific categories. To address these limitations, we propose WeGenBench, a novel benchmark designed for the comprehensive, multi-perspective evaluation of text-to-image generation capabilities. Our benchmark comprises a total of 4,000 test prompts across two primary categories, meticulously balanced between Chinese and English to evaluate bilingual and cross-cultural generation capabilities. Beyond macroscopic scene classification, we annotate each prompt with multi-dimensional tags tailored to the distinct content and challenges of each language, thereby refining the generation tasks into more specific sub-categories. Through a cross-dimensional evaluation mechanism leveraging both scene classifications and multi-dimensional tags, WeGenBench can precisely pinpoint model shortcomings in specific generation categories. Furthermore, to measure generation quality more accurately, we design and validate several novel evaluation metrics by integrating Vision-Language Models (VLMs), which assess model performance on domain-specific tasks from three core aspects. Crucially, our approach yields both the assessment outcomes and the detailed reasoning trajectories, facilitating a rigorous verification of the accuracy and soundness of the evaluation results. Finally, we conduct systematic benchmarking on current state-of-the-art methods and provide an in-depth analysis of the limitations present in existing models.
Chinese Translation
近期的文本到图像生成模型在仅凭文本输入合成高度真实的图像方面展现了显著的能力。尽管现有基准在一定程度上可以评估各种模型的生成能力，但它们在多个维度上全面而准确地测量性能方面存在困难，往往无法揭示模型在特定类别中的固有缺陷。为了解决这些局限性，我们提出了WeGenBench，这是一种新颖的基准，旨在对文本到图像生成能力进行全面的多角度评估。我们的基准包含总计4000个测试提示，涵盖两个主要类别，精心平衡中英文，以评估双语和跨文化的生成能力。除了宏观场景分类外，我们还为每个提示标注了多维标签，以适应每种语言特有的内容和挑战，从而将生成任务细化为更具体的子类别。通过利用场景分类和多维标签的跨维度评估机制，WeGenBench能够准确定位模型在特定生成类别中的不足。此外，为了更准确地测量生成质量，我们设计并验证了几种新颖的评估指标，通过整合视觉-语言模型（Vision-Language Models, VLMs），从三个核心方面评估模型在特定领域任务上的表现。重要的是，我们的方法不仅提供评估结果，还提供详细的推理轨迹，促进对评估结果的准确性和合理性的严格验证。最后，我们对当前最先进的方法进行了系统的基准测试，并对现有模型的局限性进行了深入分析。
View on arXiv Download PDF AI Translation
cs.CV / 57 / 2606.20103

Geometry-Preserving in 3D Gaussian Splatting for LiDAR-Camera Extrinsic Calibration

基于3D高斯点云的LiDAR-相机外部标定中的几何保持
Kwak, Kyoleen, Kim, Daeho, Lee, Jeong Woon, Hwang, Hyoseok
Abstract
Accurate LiDAR-camera calibration is essential for robust multi-modal perception. Targetless approaches avoid manual setup but remain limited by the scarcity of discriminative cross-modal features. Recent methods address this by reconstructing the scene within a differentiable model, enabling extrinsic optimization through dense photometric supervision. Among these, 3D Gaussian Splatting (3DGS) has been widely adopted as a geometric proxy that bridges LiDAR and camera within a single differentiable framework. However, since 3DGS was originally designed for novel view synthesis, existing methods tend to prioritize rendering quality, causing the proxy geometry to drift from the true LiDAR structure. We propose a framework that preserves the metric geometry of the Gaussian proxy by aggregating multi-view LiDAR observations for dense depth supervision and blocking photometric gradients from updating the Gaussian spatial parameters. We validate our method on public driving datasets, where it consistently outperforms existing targetless methods in calibration accuracy.
Chinese Translation
准确的LiDAR-相机标定对于稳健的多模态感知至关重要。无目标的方法避免了手动设置，但仍然受到可区分的跨模态特征稀缺的限制。最近的方法通过在可微模型中重建场景来解决这一问题，使得通过密集的光度监督进行外部优化成为可能。在这些方法中，3D高斯点云（3D Gaussian Splatting, 3DGS）被广泛采用作为一种几何代理，将LiDAR和相机桥接在一个单一的可微框架内。然而，由于3DGS最初是为新视图合成设计的，现有方法往往优先考虑渲染质量，导致代理几何与真实的LiDAR结构偏离。我们提出了一种框架，通过聚合多视角的LiDAR观测来保持高斯代理的度量几何，以实现密集深度监督，并阻止光度梯度更新高斯空间参数。我们在公共驾驶数据集上验证了我们的方法，结果显示其在标定精度上始终优于现有的无目标方法。
View on arXiv Download PDF AI Translation
cs.CV / 58 / 2606.20108

EFIQA: Explainable Fundus Image Quality Assessment via Anatomical Priors

EFIQA：基于解剖先验的可解释性视网膜图像质量评估
Wang, Pengwei, Morano, José, Wan, Qian, Bogunović, Hrvoje
Abstract
Image quality control is vital for a wide range of downstream applications. Deep learning-based image quality assessment methods typically train classifiers on dataset-specific quality labels, inheriting two limitations: (1) generalization is tied to the labeling criteria of the training set and (2) these methods cannot provide spatial feedback on where the quality is degraded, lacking explainability. In this work, we propose EFIQA, a framework that requires no quality-related supervision and produces spatial quality maps by design. Rather than learning ``what is degradation" from human-annotated labels, EFIQA learns ``what should be there" by leveraging anatomical priors. For fundus photography, we instantiate this as a two-stage approach, by first training an unsupervised anomaly detector via masked anatomical inpainting to identify regions of missing vasculature, and then distilling this prior knowledge into a shallow adapter mapping features of a frozen foundation model to precise quality maps. External-dataset evaluation demonstrates that this label-free approach with minimal adaptation achieves better performance and explainability compared with supervised methods across benchmarks with different quality criteria, highlighting its potential for real-world applications.
Chinese Translation
图像质量控制对广泛的下游应用至关重要。基于深度学习的图像质量评估方法通常在特定数据集的质量标签上训练分类器，存在两个局限性：（1）泛化能力与训练集的标注标准相关；（2）这些方法无法提供关于质量下降位置的空间反馈，缺乏可解释性。在本研究中，我们提出了EFIQA，一个不需要质量相关监督的框架，旨在通过设计生成空间质量图。EFIQA并不是从人工标注的标签中学习“什么是退化”，而是通过利用解剖先验学习“应该有什么”。对于视网膜摄影，我们将其具体化为一个两阶段的方法，首先通过掩蔽的解剖修复训练一个无监督的异常检测器，以识别缺失血管的区域，然后将这一先验知识提炼到一个浅层适配器中，将冻结的基础模型的特征映射到精确的质量图。外部数据集的评估表明，这种无需标签的、适应性最小的方法在不同质量标准的基准测试中相比于监督方法实现了更好的性能和可解释性，突显了其在实际应用中的潜力。
View on arXiv Download PDF AI Translation
cs.CV / 59 / 2606.20110

FrozenDrive: Zero-Shot Text-Guided Driving Scene Generation and Data Augmentation with Parameter-Free Frozen Diffusion Model

FrozenDrive：无参数冻结扩散模型的零-shot文本引导驾驶场景生成与数据增强
Jeong, Yuhwan, Kim, Hyeonseong, We, Daehyun, Song, Seonkyu, Yang, Jinnyeong, Jang, Hyun-Kurl, Yoon, Youngho, Yoon, Kuk-Jin
Abstract
Synthetic data for autonomous driving is surging, powered by diffusion models that promise scalable scene generation. Yet key obstacles remain, as enforcing multi-view and temporal consistency often relies on backbone fine-tuning or added layers, which erodes pre-trained knowledge and weakens text alignment. Models also stay close to the training distribution, struggling under adverse weather and unseen configurations, and fidelity favors frequent over rare classes. We address these gaps with FrozenDrive, a controllable generative framework that preserves a pretrained diffusion models knowledge while achieving strong consistency. FrozenDrive conditions on rich driving-stack signals and text prompts, and introduces knowledge-preserving spatio-temporal attention to impose cross-view alignment and temporal coherence in a single pass within a parameter-free frozen diffusion backbone. An additional object-focused constraint improves per-object fidelity for rare categories. Without any weather- or scene-specific fine-tuning, our model synthesizes globally coherent multi-view driving scenes from text, particularly under adverse and rare conditions, and surpasses prior baselines. On nuScenes, FrozenDrive augmented data significantly improves AD models performance, especially at night and in rain, demonstrating stronger robustness when trained with our scenario-targeted data.
Chinese Translation
合成数据在自动驾驶领域迅速增长，得益于扩散模型的推动，这些模型承诺可扩展的场景生成。然而，仍然存在关键障碍，因为强制多视图和时间一致性通常依赖于主干网络的微调或额外层，这会削弱预训练知识并降低文本对齐的效果。模型也往往保持接近训练分布，在恶劣天气和未见配置下表现不佳，并且在保真度上偏向于频繁类别而非稀有类别。我们通过FrozenDrive解决这些问题，构建了一个可控的生成框架，既能保持预训练扩散模型的知识，又能实现强一致性。FrozenDrive基于丰富的驾驶堆栈信号和文本提示进行条件生成，并引入知识保留的时空注意力机制，以在无参数的冻结扩散主干中实现跨视图对齐和时间连贯性。额外的以对象为中心的约束提高了稀有类别的每个对象的保真度。在没有任何天气或场景特定微调的情况下，我们的模型能够从文本合成全球一致的多视图驾驶场景，特别是在恶劣和稀有条件下，并超越了之前的基准。在nuScenes数据集上，FrozenDrive增强的数据显著提高了自动驾驶模型的性能，尤其是在夜间和雨天，展示了在使用我们场景目标数据训练时更强的鲁棒性。
View on arXiv Download PDF AI Translation
cs.CV / 60 / 2606.20112

Pixel-Level Residual Diffusion Transformer: Scalable 3D CT Volume Generation

像素级残差扩散变换器：可扩展的3D CT体积生成
Zhang, Zhenkai, Hiller, Markus, Ehinger, Krista A., Drummond, Tom
Abstract
Generating high-resolution 3D CT volumes with fine details remains challenging due to substantial computational demands and optimization difficulties inherent to existing generative models. In this paper, we propose the Pixel-Level Residual Diffusion Transformer (PRDiT), a scalable generative framework that synthesizes high-quality 3D medical volumes directly at voxel-level. PRDiT introduces a two-stage training architecture comprising 1) a local denoiser in the form of an MLP-based blind estimator operating on overlapping 3D patches to separate low-frequency structures efficiently, and 2) a global residual diffusion transformer employing memory-efficient attention to model and refine high-frequency residuals across entire volumes. This coarse-to-fine modeling strategy simplifies optimization, enhances training stability, and effectively preserves subtle structures without the limitations of an autoencoder bottleneck. Extensive experiments conducted on the LIDC-IDRI and RAD-ChestCT datasets demonstrate that PRDiT consistently outperforms state-of-the-art models, such as HA-GAN, 3D LDM and WDM-3D, achieving significantly lower 3D FID, MMD and Wasserstein distance scores.
Chinese Translation
生成高分辨率的3D CT体积并保留细节仍然面临挑战，这主要是由于现有生成模型固有的高计算需求和优化困难。本文提出了一种可扩展的生成框架——像素级残差扩散变换器（Pixel-Level Residual Diffusion Transformer，PRDiT），该框架能够直接在体素级合成高质量的3D医学体积。PRDiT引入了一种两阶段训练架构，包括1）一种局部去噪器，采用基于多层感知器（MLP）的盲估计器，作用于重叠的3D补丁，以高效分离低频结构；2）一种全局残差扩散变换器，采用内存高效的注意力机制来建模和细化整个体积中的高频残差。这种粗到细的建模策略简化了优化过程，提高了训练稳定性，并有效保留微妙结构，而不受自编码器瓶颈的限制。在LIDC-IDRI和RAD-ChestCT数据集上进行的广泛实验表明，PRDiT在3D FID、MMD和Wasserstein距离得分方面始终优于最先进的模型，如HA-GAN、3D LDM和WDM-3D，取得了显著更低的得分。
View on arXiv Download PDF AI Translation
cs.CV / 61 / 2606.20130

SAM3 Self-Distillation for Fine-Grained GOOSE 2D Semantic Segmentation

SAM3自蒸馏用于细粒度GOOSE 2D语义分割
Wang, Xuesong
Abstract
We describe our 4th-place entry to the ICRA 2026 GOOSE 2D Fine-Grained Semantic Segmentation Challenge, which reached a composite mean Intersection-over-Union (mIoU) of 69.73% on the official 1,815-image test set. Our model adapts the image encoder of a recent visual foundation model, Segment Anything Model 3 (SAM3), with a lightweight decoder. Beyond this, we contribute two techniques and one empirical finding: (i) a self-distillation scheme that re-uses SAM3 itself, prompted with ground-truth boxes, as a teacher on the classes where it outperforms our own model; (ii) an image-level multi-scale test-time augmentation scheme that restores multi-scale inference for a fixed-input-size model by rescaling the image rather than the model input; and (iii) the finding that an aggressive photometric distortion from a winning 2025 GOOSE 2D entry, transplanted onto our pipeline, is its single largest source of improvement.
Chinese Translation
我们描述了我们在ICRA 2026 GOOSE 2D细粒度语义分割挑战赛中的第4名参赛作品，该作品在官方的1,815幅图像测试集上达到了69.73%的综合平均交并比（mIoU）。我们的模型采用了最新视觉基础模型Segment Anything Model 3（SAM3）的图像编码器，并配备了轻量级解码器。除此之外，我们贡献了两项技术和一个实证发现：（i）一种自蒸馏方案，利用SAM3自身作为教师，基于真实框对其在某些类别上的表现优于我们模型的情况进行提示；（ii）一种图像级多尺度测试时增强方案，通过对图像进行重新缩放而不是对模型输入进行缩放，恢复固定输入大小模型的多尺度推理；以及（iii）发现来自2025年获胜的GOOSE 2D参赛作品的激进光度失真，移植到我们的流程中，是其单一最大的改进来源。
View on arXiv Download PDF AI Translation
cs.CV / 62 / 2606.20131

TriFlow: Generating Artist-Like 3D Mesh Topology via Nearest-Vertex Vector Fields

TriFlow：通过最近顶点向量场生成艺术家风格的3D网格拓扑
Li, Haoxuan, Erkoç, Ziya, Sirigatti, Daniele, Rosov, Vladislav, Li, Lei, Dai, Angela, Nießner, Matthias
Abstract
We present TriFlow, a new generative approach for producing compact 3D meshes with artist-like triangle topology directly from input geometry conditions such as signed distance fields. Our key insight is to represent mesh topology as a nearest-vertex vector field (NVF) defined over the surface, where each point encodes its association to the nearest triangle vertex in the local barycentric frame. We train a latent flow-matching model to synthesize this field, enabling topology generation conditioned on the input geometry. To extract a coherent mesh, we cluster surface regions using the generated NVF and guide a constrained quadric error metric (QEM) mesh simplification with topology-aware optimization. This yields output meshes that closely match the input geometry while exhibiting structured, artist-like connectivity. Experiments demonstrate that TriFlow achieves stronger generalization and significantly improved topology quality compared to state-of-the-art learning-based approaches, alongside 90% lower Chamfer Distance and an 8x speedup.
Chinese Translation
我们提出了TriFlow，一种新的生成方法，能够直接从输入几何条件（如带符号距离场）生成紧凑的艺术家风格三角形拓扑的3D网格。我们的关键见解是将网格拓扑表示为定义在表面上的最近顶点向量场（Nearest-Vertex Vector Field, NVF），其中每个点编码其与局部重心框架中最近三角形顶点的关联。我们训练了一个潜在流匹配模型来合成该向量场，使得拓扑生成能够以输入几何为条件。为了提取一致的网格，我们使用生成的NVF对表面区域进行聚类，并通过拓扑感知优化指导受限的二次误差度量（Quadratic Error Metric, QEM）网格简化。这产生的输出网格与输入几何紧密匹配，同时展现出结构化的艺术家风格连接。实验表明，TriFlow在泛化能力上表现更强，并且相较于最先进的基于学习的方法显著提高了拓扑质量，同时Chamfer距离降低了90%，速度提升了8倍。
View on arXiv Download PDF AI Translation
cs.CV / 63 / 2606.20140

SA-VIS: Sparse frame Annotations for training Video Instance Segmentation

SA-VIS：用于训练视频实例分割的稀疏帧注释
Rella, Edoardo Mello, Chhatkuli, Ajad, Jain, Shipra, Konukoglu, Ender, Van Gool, Luc
Abstract
Recent online video instance segmentation (VIS) methods have achieved impressive results, thus becoming the preferred approach to segment instances in videos. Despite the resurgence of impressive single image models, the online (or semi-online) VIS approaches outperform single-image models (e.g., based on SAM) by using long sequences of densely annotated frames during training. However,such a training setup of VIS is expensive in the sense of compute as well as dense annotations required. In order to solve these major flaws, we argue that the effective modeling of the instances and their evolution in videos do not require densely annotated frames. To that end, we propose a simple and effective module, called Past-frames Feature Propagation (PFP) which aggregates low-dimensional features from the image encoder of multiple frames. This simple low-compute module provides tremendous learning capability in using sparse video frame labels for end-to-end training. Combined with a light-weight frame-specific Instance Queries, our Sparse frame Annotation VIS (SA-VIS) significantly improves performance over its baseline. Most interestingly, our simple design that avoids complexities effectively bridges the gap in accuracy between training on sparsely and densely annotated video sequences. This translates to a mere 0.4% drop in performance of SA-VIS when using annotations for only 1/5 of the images in the dataset. Empirically, SA-VIS shows strong improvements over the baseline on YouTube-VIS 2019/2021/2022 and Occluded VIS (OVIS) and an over 1% improvement in AP on the state-of-the-art in a limited annotations scenario.
Chinese Translation
最近的在线视频实例分割（VIS）方法取得了令人瞩目的成果，因此成为了在视频中分割实例的首选方法。尽管单幅图像模型的复兴令人印象深刻，但在线（或半在线）VIS方法通过在训练过程中使用长序列的密集注释帧，超越了基于单幅图像模型（例如，基于SAM）。然而，这种VIS的训练设置在计算和所需的密集注释方面都非常昂贵。为了克服这些主要缺陷，我们认为有效建模视频中的实例及其演变并不需要密集注释帧。为此，我们提出了一个简单有效的模块，称为过去帧特征传播（Past-frames Feature Propagation，PFP），该模块聚合来自多个帧的图像编码器的低维特征。这个简单的低计算模块在使用稀疏视频帧标签进行端到端训练时提供了巨大的学习能力。结合轻量级的帧特定实例查询，我们的稀疏帧注释VIS（SA-VIS）显著提高了性能，相比于基线方法。最有趣的是，我们避免复杂性的简单设计有效地弥补了在稀疏和密集注释视频序列上训练时准确性之间的差距。这意味着在仅使用数据集中1/5图像的注释时，SA-VIS的性能仅下降0.4%。实证结果表明，SA-VIS在YouTube-VIS 2019/2021/2022和遮挡VIS（OVIS）上相较于基线有显著提升，并且在有限注释场景下在平均精度（AP）上超过了最先进的方法1%以上。
View on arXiv Download PDF AI Translation
cs.CV / 64 / 2606.20143

HEad and neCK TumOR (HECKTOR) 2025: Benchmark of Segmentation, Diagnosis, and Prognosis in Multimodal PET/CT

头颈肿瘤（HECKTOR）2025：多模态PET/CT中分割、诊断和预后基准
Saeed, Numan, Hassan, Salma, Hardan, Shahad, Cai, Lishan, Liang, Xinglong, Mazher, Moona, Qayyum, Abdul, Bu, Yansong, Lyu, Mengye, Lin, Yue, Meng, Mingyuan, Huang, Chuanyi, Wang, Lisheng, Chamseddine, Dalal, Ahrari, Shamimeh, Wu, Beining, Chen, Yifei, Mao, Fuyou, Zhang, Hao, Zhao, Baixiang, Ray, Surajit, Guo, Muzi, Xiang, Lei, Dexl, Jakob, Ingrisch, Michael, Depeursinge, Adrien, Rahmim, Arman, Hatt, Mathieu, Andrearczyk, Vincent, Yaqub, Mohammad
Abstract
Head and neck cancers (HNC) represent a significant global health burden, with accurate tumor delineation being essential for effective radiotherapy planning. The complexity of the oropharyngeal anatomy, combined with the heterogeneous appearance of tumors on imaging, makes manual segmentation time-intensive and subject to inter-observer variability. Beyond segmentation, predicting long-term clinical outcomes, such as recurrence-free survival (RFS), and determining human papillomavirus (HPV) status from noninvasive imaging, remain challenging yet clinically valuable goals. The HECKTOR 2025 challenge addresses these needs by establishing a comprehensive benchmark for automated HNC analysis using multimodal PET/CT imaging and electronic health records. Building on previous editions (2020-2022), this challenge features an expanded multi-institutional dataset comprising over 1,100 patients from 10 centers worldwide. Participants were tasked with three complementary objectives: (1) segmenting primary gross tumor volumes (GTVp) and metastatic lymph nodes (GTVn), (2) predicting recurrence-free survival, and (3) classifying HPV status. The challenge attracted 35 registered teams, with 15 final submissions evaluated on a held-out test set. Top-performing algorithms achieved a mean Dice similarity coefficient of 0.75 for segmentation, a concordance index of 0.66 for survival prediction, and a balanced accuracy of 0.56 for HPV classification. This paper presents a comprehensive analysis of the submitted methodologies, evaluates their performance across different lesion characteristics, and discusses their implications for clinical translation in automated oncology workflows and decision support systems.
Chinese Translation
头颈癌（HNC）对全球健康构成重大负担，准确的肿瘤轮廓描绘对有效的放疗规划至关重要。口咽解剖的复杂性，加上肿瘤在影像学上的异质外观，使得手动分割耗时且容易受到观察者间变异的影响。除了分割之外，预测长期临床结果，如无复发生存期（RFS），以及从非侵入性影像中确定人乳头瘤病毒（HPV）状态，仍然是具有挑战性但临床价值的重要目标。HECKTOR 2025挑战通过建立一个全面的基准，满足这些需求，旨在利用多模态PET/CT影像和电子健康记录进行自动化HNC分析。在前几届（2020-2022）的基础上，本次挑战扩展了多机构数据集，涵盖来自全球10个中心的1100多名患者。参与者被赋予三个互补的目标：（1）分割原发肿瘤体积（GTVp）和转移淋巴结（GTVn），（2）预测无复发生存期，以及（3）分类HPV状态。此次挑战吸引了35个注册团队，最终有15个提交在保留的测试集上进行评估。表现最佳的算法在分割中达到了0.75的平均Dice相似系数，在生存预测中达到了0.66的一致性指数，在HPV分类中达到了0.56的平衡准确率。本文对提交的方法进行了全面分析，评估了它们在不同病变特征下的表现，并讨论了它们在自动化肿瘤学工作流程和决策支持系统中的临床转化意义。
View on arXiv Download PDF AI Translation
cs.CV / 65 / 2606.20155

NAMESAKES: Probing Identity Memorization in Text-to-Image Models

NAMESAKES：探究文本到图像模型中的身份记忆
Alper, Morris, Varadarajan, Vasudha, Yanuka, Moran, Wang, Angelina, Averbuch-Elor, Hadar
Abstract
Text-to-image (T2I) models generate realistic likenesses of some individuals when prompted with their names, raising privacy concerns. However, distinguishing whether a generated face is memorized or fabricated currently requires ground-truth photos, access to training data, or white-box access to model internals, limiting applicability. We introduce a fully black-box behavioral probe that distinguishes between these regimes while requiring no reference photos or prior knowledge of training data. To benchmark this task, we present the NAMESAKES dataset of over one thousand names and faces of public figures spanning a wide range of fame levels, along with perturbed, less famous names. Experiments on state-of-the-art T2I models show that our probe substantially predicts identity memorization and separates memorized from unrecognized names, with further insights into differences across model families.
Chinese Translation
文本到图像（T2I）模型在输入某些个体的名字时生成逼真的肖像，这引发了隐私问题。然而，目前区分生成的面孔是记忆的还是虚构的需要真实照片、对训练数据的访问或模型内部的白盒访问，这限制了其适用性。我们提出了一种完全黑盒的行为探测器，能够在不需要参考照片或先前训练数据知识的情况下区分这两种情况。为了对这一任务进行基准测试，我们呈现了NAMESAKES数据集，该数据集包含超过一千个公众人物的名字和面孔，涵盖了广泛的知名度水平，以及一些扰动的、不太知名的名字。在对最先进的T2I模型进行的实验中，我们的探测器显著预测了身份记忆，并区分了被记忆的名字与未被识别的名字，同时提供了不同模型家族之间差异的进一步见解。
View on arXiv Download PDF AI Translation
cs.CV / 66 / 2606.20161

ARTEMIS: Agent-guided Reliability-aware Temporal Mask Evolution for Imperfectly Supervised Video Polyp Segmentation

ARTEMIS：基于代理引导的可靠性意识时间掩膜演化用于不完全监督的视频息肉分割
Wang, Tong, Wang, Siwen, Qi, Yaolei, Zhou, Jinxing, He, Yuting, Yang, Guanyu, Xie, Yutong
Abstract
Imperfectly supervised video polyp segmentation (VPS) aims to learn dense, temporally consistent masks from inexpensive supervision, including weak annotations (points, scribbles) and semi-supervision with few densely labeled frames. This setting is clinically valuable but challenging due to weak contrast, ambiguous boundaries, motion blur, and specular highlights, compounded by sparse pixel-level guidance. While SAM2 can generate dense masks from sparse inputs, direct pseudo-labeling often yields geometry-degraded masks with boundary leakage, underutilizes temporal consistency, and ignores reliability. To address these issues, we propose ARTEMIS, a unified framework for imperfectly supervised VPS driven by agent-guided reliability-aware temporal mask evolution. ARTEMIS initializes coarse masks from available supervision: SAM2 converts points/scribbles, while dense labels serve as reliable anchors. A debate-and-judge vision-language agent selects reliable temporal anchors under weak supervision, which are propagated bidirectionally with SAM2 to refine unreliable or unlabeled frames. Finally, ARTEMIS trains the segmenter using temporal reliability-aware robust learning, incorporating reliability-guided reference selection, a Reference Prototype Transport Module, and reliability-aware robust loss. These components assess mask reliability, evolve anchors over time, transport target identity across frames, and down-weight noisy supervision instead of discarding difficult samples. Experiments on SUN-SEG and CVC-ClinicDB-612 under scribble, point, and limited-label settings demonstrate that ARTEMIS achieves state-of-the-art performance. Code will be released at https://github.com/wangtong627/ARTEMIS.
Chinese Translation
不完全监督的视频息肉分割（VPS）旨在从低成本的监督中学习稠密且时间一致的掩膜，包括弱标注（点、涂鸦）和少量稠密标注帧的半监督。这种设置在临床上具有重要价值，但由于对比度弱、边界模糊、运动模糊和镜面高光等因素，加上稀疏的像素级指导，挑战性很大。虽然SAM2能够从稀疏输入生成稠密掩膜，但直接伪标注往往会导致几何降级的掩膜、边界泄漏，未充分利用时间一致性，并忽视可靠性。为了解决这些问题，我们提出了ARTEMIS，一个由代理引导的可靠性意识时间掩膜演化驱动的不完全监督VPS的统一框架。ARTEMIS从可用的监督中初始化粗略掩膜：SAM2将点/涂鸦转换为掩膜，而稠密标签则作为可靠的锚点。在弱监督下，辩论与判断的视觉语言代理选择可靠的时间锚点，这些锚点通过SAM2双向传播，以细化不可靠或未标记的帧。最后，ARTEMIS使用时间可靠性意识的稳健学习训练分割器，结合可靠性引导的参考选择、参考原型传输模块和可靠性意识的稳健损失。这些组件评估掩膜的可靠性，随时间演化锚点，跨帧传输目标身份，并对噪声监督进行降权，而不是丢弃困难样本。在SUN-SEG和CVC-ClinicDB-612数据集上进行的实验表明，ARTEMIS在涂鸦、点和有限标签设置下实现了最先进的性能。代码将发布在https://github.com/wangtong627/ARTEMIS。
View on arXiv Download PDF AI Translation
cs.CV / 67 / 2606.20177

Evaluating and Enhancing Negation Comprehension in Remote Sensing MLLMs

评估与增强遥感多模态大语言模型中的否定理解能力
Han, Haochen, Wang, Jue, Wang, Alex Jinpeng, Liu, Fangming
Abstract
Multimodal Large Language Models (MLLMs) have demonstrated remarkable success in various Remote Sensing (RS) tasks. However, their ability to comprehend negation remains underexplored, limiting deployment in real-world applications where models must explicitly identify what is false or absent, e.g., emergency responders need to locate non-flooded routes for evacuation. To comprehensively study this limitation, we introduce RS-Neg, the first benchmark to evaluate negation understanding across region-level to scene-level tasks. Specifically, we design an automated data generation pipeline for RS imagery, using LLMs to synthesize diverse negation queries, and introduce a dynamic visual focus module for verification. Our evaluation reveals that advanced RS MLLMs struggle with negation, exhibiting hallucinations and substantial performance degradation. To close this gap, we propose NeFo, a novel test-time learning method that explicitly incorporates the logical role of negation into the model optimization. Remarkably, using about 5\% unlabeled test samples, NeFo significantly improves the negation understanding of models and shows strong generalization to unseen tasks. Code and data will be released upon acceptance.
Chinese Translation
多模态大语言模型（MLLMs）在各种遥感（RS）任务中表现出显著的成功。然而，它们对否定的理解能力仍然未得到充分探索，这限制了它们在现实应用中的部署，特别是在模型必须明确识别虚假或缺失信息的情况下，例如，紧急救援人员需要找到未被淹没的撤离路线。为了全面研究这一局限性，我们引入了RS-Neg，这是第一个用于评估区域级到场景级任务中否定理解的基准。具体而言，我们设计了一个自动化数据生成管道，用于遥感图像，利用大语言模型（LLMs）合成多样的否定查询，并引入了一个动态视觉聚焦模块进行验证。我们的评估揭示了先进的遥感多模态大语言模型在否定理解方面的困难，表现出幻觉现象和显著的性能下降。为了解决这一问题，我们提出了NeFo，一种新颖的测试时学习方法，明确将否定的逻辑角色纳入模型优化中。值得注意的是，使用约5%的未标记测试样本，NeFo显著提高了模型的否定理解能力，并在未见任务上表现出强大的泛化能力。代码和数据将在接受后发布。
View on arXiv Download PDF AI Translation
cs.CV / 68 / 2606.20189

HilDA: Hierarchical Distillation with Diffusion for Advancing Self-Supervised LiDAR Pre-trainin

HilDA：用于推进自监督LiDAR预训练的层次蒸馏与扩散
Wozniak, Maciej, Ericsson, Jesper, Govindarajan, Hariprasath, Nyberg, Truls, Gustafsson, Thomas, Jensfelt, Patric, Andersson, Olov
Abstract
Leveraging Vision Foundation Models (VFMs) for camera-to-LiDAR knowledge distillation offers a promising solution to the scarcity of annotated data needed to represent the immense geometric and kinematic diversity of real-world autonomous driving (AD). However, current approaches typically treat VFMs as black-box teachers, relying exclusively on frame-wise feature similarity. Consequently, they do not fully exploit the teacher's layer-wise semantic structure and global context, as well as the rich spatiotemporal information inherent in LiDAR sequences. We propose HilDA, a self-supervised pretraining framework for LiDAR backbones that better captures the semantic what and geometric where needed for driving tasks. HilDA combines hierarchical distillation comprising multi-layer distillation for progressive semantic alignment and global context distillation for scene-level semantics, with a temporal occupancy diffusion objective promoting spatiotemporal consistency. Models pre-trained with HilDA achieve state-of-the-art results on cross-modal distillation benchmarks and outperform models trained via prior distillation approaches on 3D object detection, scene flow, and semantic occupancy prediction. Code available at: https://maxiuw.github.io/hilda.
Chinese Translation
利用视觉基础模型（VFMs）进行相机到LiDAR的知识蒸馏为解决表示现实世界自动驾驶（AD）中巨大几何和运动多样性所需的标注数据稀缺问题提供了一种有前景的解决方案。然而，目前的方法通常将VFMs视为黑箱教师，仅依赖于逐帧特征相似性。因此，它们未能充分利用教师的层级语义结构和全局上下文，以及LiDAR序列中固有的丰富时空信息。我们提出了HilDA，这是一种自监督的LiDAR骨干网络预训练框架，更好地捕捉驾驶任务所需的语义“是什么”和几何“在哪里”。HilDA结合了层次蒸馏，包括用于逐步语义对齐的多层蒸馏和用于场景级语义的全局上下文蒸馏，以及促进时空一致性的时间占用扩散目标。使用HilDA预训练的模型在跨模态蒸馏基准上实现了最先进的结果，并在3D物体检测、场景流和语义占用预测方面超越了通过先前蒸馏方法训练的模型。代码可在：https://maxiuw.github.io/hilda获取。
View on arXiv Download PDF AI Translation
cs.CV / 69 / 2606.20196

Distill Once, Adapt Life-Long: Exploring Dataset Distillation for Continual Test-Time Adaptation

一次蒸馏，终身适应：探索用于持续测试时适应的数据集蒸馏
Jang, Hyun-Kurl, Kim, Jihun, Kweon, Hyeokjun, Yoon, Kuk-Jin
Abstract
Continual Test-Time Adaptation (CTTA) aims to maintain model performance under evolving target domains by adapting online without labeled data. However, practical deployments often cannot retain the source dataset due to privacy or licensing constraints, and purely source-free CTTA methods tend to become unstable under long-term distribution shift, suffering from compounding self-training errors and catastrophic forgetting. We introduce DO-ALL (Distill Once, Adapt Life-Long), a plug-and-play framework that revisits source information in a compact and privacy-conscious form via Dataset Distillation (DD). Before deployment, DO-ALL performs DD to produce a small set of synthetic distilled anchors that summarize the source distribution. During adaptation, each target sample is matched with its most semantically aligned anchor, which provides a stable reference for various CTTA via source replay, representation alignment, and manifold-smoothing regularization. DO-ALL can be seamlessly integrated into existing CTTA algorithms, consistently improving long-term robustness across CIFAR100-C, ImageNet-C, and the CCC benchmark. This demonstrates the potential of leveraging DD to enable stable and continuous adaptation without retaining raw source data. The code is available at https://github.com/blue-531/DOALL.
Chinese Translation
持续测试时适应（CTTA）旨在通过在线适应而无需标记数据来维持模型在不断变化的目标领域中的性能。然而，实际部署中由于隐私或许可限制，往往无法保留源数据集，而纯粹的无源CTTA方法在长期分布变化下往往变得不稳定，遭受累积自我训练错误和灾难性遗忘。我们提出了DO-ALL（一次蒸馏，终身适应），这是一个即插即用的框架，通过数据集蒸馏（DD）以紧凑且注重隐私的形式重新审视源信息。在部署之前，DO-ALL执行DD以生成一小组合成的蒸馏锚点，概括源分布。在适应过程中，每个目标样本与其最语义对齐的锚点匹配，该锚点为各种CTTA提供了稳定的参考，通过源重放、表示对齐和流形平滑正则化。DO-ALL可以无缝集成到现有的CTTA算法中，在CIFAR100-C、ImageNet-C和CCC基准测试中持续提高长期鲁棒性。这表明利用DD实现稳定和持续适应的潜力，而无需保留原始源数据。代码可在https://github.com/blue-531/DOALL获取。
View on arXiv Download PDF AI Translation
cs.CV / 70 / 2606.20199

Evaluation of Image Matching for Art Skills Assessment

艺术技能评估中的图像匹配评估
Alghamdi, Asaad, Poor, Michael, Le, Trung-Nghia, Nguyen, Tam V.
Abstract
While some individuals possess a natural talent for drawing, mastering this skill requires dedicated training and practice. Determining one's skill in the art of drawing requires proper comprehensive assessment. In this paper, we propose a method to measure drawing skill by by matching the hand-drawn image with the original template. Existing techniques often involve complex processes. However, advancements in computer vision allow us to train computers to perform these comparisons at a human-like level, thereby resolving the tedious and overwhelming traditional process. Using computer vision applications, determining image similarity involves identifying the level of similarities in an image with a reference image. We have implemented and analyzed the SIFT feature and Siamese network to measure image similarity. Our results indicate that it is feasible to assess art skill levels. Through feature analysis, we found that SIFT-based key point matching provides a more effective means of detecting drawing skills.
Chinese Translation
虽然一些人天生具备绘画才能，但掌握这一技能需要专注的训练和实践。评估一个人在绘画艺术方面的技能需要适当的综合评估。在本文中，我们提出了一种通过将手绘图像与原始模板进行匹配来测量绘画技能的方法。现有技术通常涉及复杂的过程。然而，计算机视觉的进步使我们能够训练计算机以人类相似的水平进行这些比较，从而解决了繁琐且令人不堪重负的传统过程。利用计算机视觉应用，确定图像相似性涉及识别图像与参考图像之间的相似程度。我们实现并分析了SIFT特征和Siamese网络以测量图像相似性。我们的结果表明，评估艺术技能水平是可行的。通过特征分析，我们发现基于SIFT的关键点匹配提供了一种更有效的检测绘画技能的方法。
View on arXiv Download PDF AI Translation
cs.CV / 71 / 2606.20223

DeepForestVisionV2: Ecology-Driven Taxonomy Expansion for Camera-Trap Monitoring in African Tropical Forests

DeepForestVisionV2：基于生态驱动的非洲热带森林相机陷阱监测分类扩展
Magaldi, Hugo, d'Audiffret, Theau, Akomo-Okoue, Etienne Francois, Amarasekaran, Bala, Anderson, Naomi, Auger, Claire, Cappelle, Noemie, Cornelis, Daniel, Cornette, Raphael, Deschner, Tobias, Dubus, Gabriel, Fonteyn, Davy, Garriga, Rosa M., Hatlauf, Jennifer, Kasekendi, Innocent, Katumba, Raymond, Kazandjian, Aram, Ngomanda, Alfred, Ntie, Stephan, Pika, Simone, Rufray, Xavier, Rugonge, Harold, Tibesigwa, John Justice, van Lunteren, Peter, Vanthomme, Hadrien, Zwerts, Joeri A., Krief, Sabrina
Abstract
Camera-trap monitoring in African tropical forests increasingly extends beyond closed-canopy interiors to riverbanks, clearings, and park edges. Among available open tools for African forest camera-trap classification, DeepForestVision is the only one providing a matched offline workflow for both photographs and videos, and previous work showed that it outperformed other available baselines on a comparable benchmark. However, it was designed for closed-canopy, ground-level forest interiors and uses a 35-class prediction space that becomes too coarse when deployments encounter arboreal primates, birds, semi-aquatic taxa, or human-associated confounders such as livestock. We present DeepForestVisionV2, an ecology-driven expansion from 35 to 64 prediction classes (61 animal classes plus human, vehicle, and blank) designed to address three recurrent deployment gradients: vertical stratification, scene openness, and anthropogenic interfaces. DeepForestVisionV2 retains the same offline workflow and is trained on 1,535,010 photographs and 243,354 videos from multi-country African tropical-forest projects. Evaluation combines a cross-country cropped-photo validation set, used to assess robustness across sites and camera-trap settings, with three held-out Uganda video benchmarks spanning the targeted gradients. On the validation set, DeepForestVisionV2 reaches 0.86 accuracy, 0.82 macro-F1, and 0.81 balanced accuracy. On the deployment benchmarks, it preserves or improves baseline accuracy despite its harder classification task, while increasing the number of identified taxa from 22 to 29 in forest-interior videos and from 4 to 9 at riverbanks. In the park-edge use case, it raises accuracy from 0.62 to 0.86 and reduces false alarms from 11 to 0. These results show that DeepForestVisionV2 materially improves field utility while preserving robustness across sites, habitats, and camera-trap settings.
Chinese Translation
在非洲热带森林中，相机陷阱监测逐渐超越了封闭冠层的内部，扩展到河岸、空地和公园边缘。在可用的非洲森林相机陷阱分类工具中，DeepForestVision 是唯一提供与照片和视频相匹配的离线工作流程的工具，之前的研究表明它在可比基准上优于其他可用的基线。然而，它是为封闭冠层的地面森林内部设计的，使用的35类预测空间在遇到树栖灵长类动物、鸟类、半水生类群或与人类相关的混杂因素（如家畜）时显得过于粗糙。我们提出了DeepForestVisionV2，这是一个从35类扩展到64类（61类动物加上人类、车辆和空白）的生态驱动分类，旨在解决三个反复出现的部署梯度：垂直分层、场景开放性和人类干扰。DeepForestVisionV2 保留了相同的离线工作流程，并在来自多个国家的非洲热带森林项目中训练了1,535,010张照片和243,354段视频。评估结合了一个跨国裁剪照片验证集，用于评估不同地点和相机陷阱设置下的鲁棒性，以及三个保留的乌干达视频基准，涵盖了目标梯度。在验证集上，DeepForestVisionV2 达到了0.86的准确率、0.82的宏观F1值和0.81的平衡准确率。在部署基准上，尽管面临更困难的分类任务，它仍然保持或提高了基线准确率，同时在森林内部视频中识别的类群数量从22增加到29，在河岸上从4增加到9。在公园边缘的使用案例中，它将准确率从0.62提高到0.86，并将误报从11减少到0。这些结果表明，DeepForestVisionV2 在提高现场实用性的同时，保持了在不同地点、栖息地和相机陷阱设置下的鲁棒性。
View on arXiv Download PDF AI Translation
cs.CV / 72 / 2606.20233

Cinematic Compositing Using Character-Environment-Harmonized Video Generation Models

基于角色-环境协调的视频生成模型的电影合成
Xiang, Tianyi, He, Mingming, Ma, Li, Liao, Jing
Abstract
Cinematic compositing aims to integrate green-screen characters into novel environments while maintaining physical and photometric realism. Previous methods often fail to capture the complex bidirectional interactions between characters and their surroundings, which we characterize as Character-to-Environment (C2E) physical interaction and Environment-to-Character (E2C) lighting harmonization. To address this, we propose an end-to-end video diffusion framework that jointly models C2E and E2C interactions, specifically handling the challenges of interactive props. Our approach introduces a tri-mask-guided architecture with RGB-D joint denoising to ensure physically consistent interactions among the character, props, and environment. We further develop an efficient prior-driven data curation pipeline to construct high-quality relighting pairs without expensive rendering. Finally, a reference-conditioned mechanism enables controllable environment synthesis and precise prop replacement. Extensive experiments demonstrate that our framework significantly outperforms existing methods in cinematic-quality dynamic video compositing.
Chinese Translation
电影合成旨在将绿幕角色融入新环境中，同时保持物理和光度的真实感。以往的方法往往未能捕捉角色与其周围环境之间复杂的双向交互，我们将其特征化为角色到环境（Character-to-Environment, C2E）物理交互和环境到角色（Environment-to-Character, E2C）光照协调。为了解决这一问题，我们提出了一种端到端的视频扩散框架，联合建模C2E和E2C交互，特别处理互动道具的挑战。我们的方法引入了一种三重掩码引导的架构，并结合RGB-D联合去噪，以确保角色、道具和环境之间的物理一致性交互。我们进一步开发了一种高效的先验驱动数据整理管道，以在不进行昂贵渲染的情况下构建高质量的重光照配对。最后，参考条件机制使得环境合成可控，并实现精确的道具替换。大量实验表明，我们的框架在电影质量动态视频合成方面显著优于现有方法。
View on arXiv Download PDF AI Translation
cs.CV / 73 / 2606.20241

BAFIS: Dataset + Framework to assess occupational Bias and Human Preference in modern Text-to-image Models

BAFIS：评估现代文本到图像模型中的职业偏见和人类偏好的数据集与框架
Klassert, Thomas, Ulges, Adrian, Fu, Biying
Abstract
Generative artificial intelligence has the potential to improve productivity and transform the production of creative content. However, existing research indicates that image generation models are significantly influenced by biases. This work investigates the inherent biases and language-induced biases present in text-to-image models within the context of occupation-related image generation, complementing established metrics with human preference feedback. We present a comprehensive evaluation of five current text-to-image models: Midjourney v6.1, Stable Diffusion 3 Medium, DALL-E 3, Playground v2.5, and FLUX.1-dev , focusing on gender and ethnicity bias, image quality, and prompt alignment. To facilitate this evaluation, we developed the "Battle-Arena for Fair Image Synthesis" (BAFIS), a platform designed to collect human feedback on bias in generated images. Furthermore, we created a dataset comprising 21,140 synthetic images generated using multilingual prompts, which serves as a basis for our analysis. We further place our results within a broader social context by comparing them to official statistics from the German Federal Employment Agency. Our findings reveal systematic biases in text-to-image models, with established evaluation metrics in partial correlation with subjective user ratings. Thus, our research emphasizes the need for including human preferences to develop fairer and more inclusive text-to-image models.
Chinese Translation
生成性人工智能有潜力提高生产力并改变创意内容的生产方式。然而，现有研究表明，图像生成模型受到偏见的显著影响。本研究探讨了文本到图像模型中存在的固有偏见和语言引发的偏见，特别是在与职业相关的图像生成背景下，结合人类偏好反馈补充已有的评估指标。我们对五个当前的文本到图像模型进行了全面评估：Midjourney v6.1、Stable Diffusion 3 Medium、DALL-E 3、Playground v2.5 和 FLUX.1-dev，重点关注性别和种族偏见、图像质量以及提示对齐。为便于这一评估，我们开发了“公平图像合成战斗竞技场”（BAFIS），这是一个旨在收集人类对生成图像偏见反馈的平台。此外，我们创建了一个包含21,140张使用多语言提示生成的合成图像的数据集，作为我们分析的基础。我们还通过将结果与德国联邦就业局的官方统计数据进行比较，将我们的发现置于更广泛的社会背景中。我们的研究结果揭示了文本到图像模型中的系统性偏见，且已有的评估指标与主观用户评分存在部分相关性。因此，我们的研究强调了在开发更公平和更具包容性的文本到图像模型时纳入人类偏好的必要性。
View on arXiv Download PDF AI Translation
cs.CV / 74 / 2606.20244

SPOT-E: Test-Time Entropy Shaping with Visual Spotlights for Frozen VLMs

SPOT-E：针对冻结视觉语言模型的测试时熵塑形与视觉聚光灯
Yin, Bo, Hu, Xiaobin, Xu, Chengming, Shen, Ruolin, Yang, Mo, Zhang, Jiangning, Jiang, Peng-Tao, Tan, Cheng, YAN, Shuicheng
Abstract
Vision-language models (VLMs) often underperform on evidence intensive tasks because decisive visual evidence are small, localized, and easy to overlook, leading to failures in evidence readout even when high-level reasoning is intact. Prior inference-time visual interventions can improve grounding without retraining, but they are largely open-loop and lack a mechanism to verify whether highlighted evidence is actually used. We study answer-span prediction entropy as a model-internal feedback signal and show that naive entropy minimization is ambiguous, since low entropy may arise from evidence-grounded confidence or shortcut collapse. To resolve this ambiguity, we introduce low-entropy anchors and an entropy-shaping objective that reduces answer uncertainty while preserving baseline high-confidence tokens. We instantiate this principle in SPOT-E, a plug-and-play test-time method that produces question-conditioned spotlights, optimized per instance via light-weight tuning based on Group Relative Policy Optimization (GRPO). Across all benchmarks and different VLM families, SPOT-E yields consistent gains and improved robustness under visual corruptions. Code is publicly available at: \url{https://github.com/YinBo0927/SPOT-E}
Chinese Translation
视觉语言模型（VLMs）在证据密集型任务中往往表现不佳，因为决定性的视觉证据通常较小、局部且容易被忽视，这导致即使高层次推理完好，证据读取仍然失败。以往的推理时视觉干预可以在不重新训练的情况下改善基础，但它们大多是开环的，缺乏验证突出证据是否被实际使用的机制。我们研究了答案跨度预测熵作为模型内部反馈信号，并表明简单的熵最小化是模糊的，因为低熵可能源于基于证据的信心或捷径崩溃。为了消除这种模糊性，我们引入了低熵锚点和一种熵塑形目标，该目标在保持基线高信心标记的同时减少答案的不确定性。我们在SPOT-E中实例化这一原则，这是一种即插即用的测试时方法，通过基于组相对策略优化（Group Relative Policy Optimization, GRPO）的轻量调优，为每个实例生成条件于问题的聚光灯。在所有基准测试和不同的VLM家族中，SPOT-E都带来了持续的提升和在视觉干扰下的增强鲁棒性。代码已公开可用，网址为：https://github.com/YinBo0927/SPOT-E
View on arXiv Download PDF AI Translation
cs.CV / 75 / 2606.20250

Single-Stage Hierarchical Rectification for Weakly Supervised Histopathology Segmentation

单阶段层次纠正用于弱监督组织病理学分割
Nguyen, Duc T., Nguyen, Hoang-Long, DO, Thanh-Ha, Pham, Huy-Hieu
Abstract
Existing weakly supervised semantic segmentation (WSSS) methods in computational pathology rely on a multi-stage paradigm: class activation map (CAM) generation, offline pseudo-mask refinement, and fully supervised retraining. While established, this decoupled approach presents fundamental limitations. The multi-stage process not only incurs high computational training costs but also suffers from error propagation: local texture biases in shallow CNN layers generate false-positive artifacts that subsequent refinement steps often fail to correct. To address these persistent challenges through a simple yet highly effective approach, we propose the Single-Stage Hierarchical Rectification (SSHR) framework. Rather than passively refining CAMs post-hoc, our method proactively purifies intermediate feature representations during the forward pass. We introduce a Hierarchical Feature Rectification Module (HFRM) that utilizes deep global semantic context to filter out local anomalies in shallow layers. This mechanism generates high-fidelity activation maps directly within a single training loop. Experiments on the LUAD-HistoSeg and BCSS datasets demonstrate that SSHR outperforms state-of-the-art multi-stage methods. Furthermore, SSHR reduces training duration by 2 to 5 times. This efficiency minimizes computational overhead and accelerates clinical translation for large-scale histopathology workflows. The code is available at: https://github.com/trongduc-nguyen/SSHR
Chinese Translation
现有的弱监督语义分割（WSSS）方法在计算病理学中依赖于多阶段范式：类激活图（CAM）生成、离线伪掩码精炼和完全监督的再训练。尽管这一方法已被确立，但其解耦的方式存在根本性局限性。多阶段过程不仅导致高计算训练成本，还面临错误传播的问题：浅层卷积神经网络（CNN）中的局部纹理偏差会产生假阳性伪影，而后续的精炼步骤往往无法纠正这些错误。为了解决这些持续存在的挑战，我们提出了单阶段层次纠正（SSHR）框架，这是一种简单而高效的方法。我们的方法并不是在事后被动地精炼CAM，而是在前向传播过程中主动净化中间特征表示。我们引入了层次特征纠正模块（HFRM），该模块利用深层全局语义上下文来过滤浅层中的局部异常。该机制在单个训练循环中直接生成高保真度的激活图。对LUAD-HistoSeg和BCSS数据集的实验表明，SSHR优于最先进的多阶段方法。此外，SSHR将训练时间缩短了2到5倍。这种效率减少了计算开销，并加速了大规模组织病理学工作流程的临床转化。代码可在以下链接获取：https://github.com/trongduc-nguyen/SSHR
View on arXiv Download PDF AI Translation
cs.CV / 76 / 2606.20282

U$^2$Mamba: A Two-level Nested U-structure Mamba for Salient Object Detection

U$^2$Mamba：一种用于显著目标检测的双层嵌套U结构Mamba
Li, Junhui, Li, Jialu, Zhang, Youshan
Abstract
Mamba-based models have emerged as a promising alternative for salient object detection (SOD), offering significant advantages in modeling long sequences. However, existing models often fail to explore contextual information and the depth of the entire architecture. This paper introduces U$^2$Mamba, a powerful and innovative U-structured network for salient object detection. We propose multiscale Mamba U-blocks (MMUBs) that enhance the model depth to improve local feature extraction capabilities. Our newly developed nested U-structure, incorporating MMUBs, enables the network to integrate various receptive fields from shallow and deep layers, thereby collecting richer contextual information and longer-range data without being constrained by resolution. Instead of using the traditional deep supervision scheme and top-level supervised training, we propose a hierarchical training supervision method where the loss is computed at each level during the training process. Extensive experiments demonstrate that U$^2$Mamba achieves highly competitive performance against state-of-the-art methods. The source code is available at \url{https://github.com/JL021/U2Mamba}.
Chinese Translation
基于Mamba的模型已成为显著目标检测（SOD）的有前景的替代方案，在建模长序列方面具有显著优势。然而，现有模型往往未能充分探索上下文信息和整个架构的深度。本文介绍了U$^2$Mamba，一种强大且创新的用于显著目标检测的U结构网络。我们提出了多尺度Mamba U块（MMUBs），增强模型深度以改善局部特征提取能力。我们新开发的嵌套U结构结合了MMUBs，使网络能够整合来自浅层和深层的各种感受野，从而收集更丰富的上下文信息和更长范围的数据，而不受分辨率的限制。我们提出了一种分层训练监督方法，在训练过程中在每个层次计算损失，而不是使用传统的深度监督方案和顶层监督训练。大量实验表明，U$^2$Mamba在与最先进的方法的比较中表现出高度竞争力。源代码可在 https://github.com/JL021/U2Mamba 获取。
View on arXiv Download PDF AI Translation
cs.CV / 77 / 2606.20300

CMDS-AD: Cross-Modal Dual-Stream Decoupling for Few-Shot Anomaly Detection

CMDS-AD：用于少样本异常检测的跨模态双流解耦
Cai, Junhao, Zeng, Deyu, Pang, Junhao, Chen, Junyu, Liang, Qiwei, Zhong, Xiaopin, Wu, Zongze
Abstract
Few-shot anomaly detection remains challenging due to limited training data. Multi-modal anomaly detection (MAD) offers a viable solution, leveraging 3D geometric cues to enrich 2D RGB representations and compensate for this scarcity. However, existing MAD methods apply spatially uniform feature processing, conflating stable macroscopic structures with high-frequency localized defect signals, exacerbating cross-modal misalignment and inflating false-positive rates. To overcome this, we present CMDS-AD, a Cross-Modal Dual-Stream Anomaly Detection framework. A LoRA-guided diffusion model generates diverse RGB samples to mitigate extreme data scarcity. For 3D normal augmentation, we employ a pre-trained diffusion model as a normal estimator. Crucially, this estimator inherently acts as a non-linear low-pass filter, directly extracting low-frequency normal representations from RGB inputs. This establishes an auxiliary estimated stream of purely low-frequency information, anchoring robust structural templates and assisting the uncompressed real stream, containing coupled high- and low-frequency components, to precisely isolate micro-defects. A Coordinate-Aware Hierarchical Feature Mapper adaptively aligns cross-modal semantics, while a multiplicative scoring mechanism filters modality-specific noise. Under the extreme 1-shot setting, CMDS-AD achieves absolute performance gains of 5.7% (I-AUROC) and 2.0% (AUPRO) on MVTec 3D-AD, alongside 7.7% and 5.6% improvements on EyeCandies, establishing a new state-of-the-art.
Chinese Translation
由于训练数据有限，少样本异常检测仍然具有挑战性。多模态异常检测（MAD）提供了一种可行的解决方案，利用三维几何线索来丰富二维RGB表示，从而弥补数据稀缺。然而，现有的MAD方法采用空间均匀的特征处理，将稳定的宏观结构与高频局部缺陷信号混合在一起，加剧了跨模态的不对齐，并提高了假阳性率。为了解决这个问题，我们提出了CMDS-AD，一个跨模态双流异常检测框架。一个由LoRA引导的扩散模型生成多样的RGB样本，以缓解极端的数据稀缺。对于三维正常增强，我们采用一个预训练的扩散模型作为正常估计器。关键是，这个估计器本质上充当了一个非线性低通滤波器，直接从RGB输入中提取低频正常表示。这建立了一个纯低频信息的辅助估计流，锚定了稳健的结构模板，并协助未压缩的真实流，其中包含耦合的高频和低频成分，以精确隔离微缺陷。一个坐标感知的层次特征映射器自适应地对齐跨模态语义，而一个乘法评分机制则过滤特定模态的噪声。在极端的1-shot设置下，CMDS-AD在MVTec 3D-AD上实现了5.7%（I-AUROC）和2.0%（AUPRO）的绝对性能提升，在EyeCandies上分别提高了7.7%和5.6%，建立了新的最先进水平。
View on arXiv Download PDF AI Translation
cs.CV / 78 / 2606.20302

CUPID: Reconstructing UV Texture Maps for Interpretable Person-of-Interest Deepfake Detection

CUPID：重建可解释的人物深度伪造检测的 UV 纹理图
Affatato, Giovanni, Mandelli, Sara, Cannas, Edoardo Daniele, Bestagini, Paolo, Tubaro, Stefano
Abstract
Deepfakes targeting a high-profile individual, known as Person-of-Interest (POI), are a threat to modern democracies and societies. Current POI deepfake detection methods still struggle to combine robustness to post-processing, efficiency and interpretability, focal aspects of modern deepfake detectors. In this paper we propose CUPID, a POI video deepfake detector that combines UV texture maps, a facial appearance representation derived from 3D face reconstructions, with the representation learning capabilities of the Masked Autoencoder (MAE). Our method does not require any deepfake videos in its training phase. Moreover, it does not even require to include a specific POI in the training set: the combination of UV texture maps extracted from real video frames and the MAE context-guided reconstruction yields a latent space that captures rich and discriminative facial features also for identities unseen during training. In the testing phase, the embeddings extracted from a query video depicting the POI can be matched against pristine reference videos to assess the video authenticity. Furthermore, operating in the UV space naturally provides an additional layer of interpretability. Specifically, we can extract decoded residual maps that highlight which facial regions of a test video deviate most from the identity representation of the corresponding POI. Experiments on four deepfake datasets show that CUPID outperforms current state of the art on most datasets and achieves the best overall robustness against strong downscaling and compression, providing also substantially faster inference. Our experimental code will be released at https://github.com/polimi-ispl/CUPID.
Chinese Translation
针对高知名度个体（即人物，Person-of-Interest，POI）的深度伪造技术对现代民主和社会构成威胁。目前的 POI 深度伪造检测方法在结合对后处理的鲁棒性、效率和可解释性等现代深度伪造检测器的核心方面上仍然面临挑战。本文提出了 CUPID，一种结合 UV 纹理图（从 3D 面部重建中得出的面部外观表示）与掩码自编码器（Masked Autoencoder，MAE）表示学习能力的 POI 视频深度伪造检测器。我们的方法在训练阶段不需要任何深度伪造视频。此外，它甚至不需要在训练集中包含特定的 POI：从真实视频帧提取的 UV 纹理图与 MAE 上下文引导重建的结合产生了一个潜在空间，该空间捕捉了丰富且具有区分性的面部特征，即使对于训练期间未见过的身份。在测试阶段，从描绘 POI 的查询视频中提取的嵌入可以与原始参考视频进行匹配，以评估视频的真实性。此外，在 UV 空间中操作自然提供了额外的可解释性层面。具体而言，我们可以提取解码的残差图，突出显示测试视频中哪些面部区域与相应 POI 的身份表示偏差最大。在四个深度伪造数据集上的实验表明，CUPID 在大多数数据集上超越了当前的最先进技术，并在强下采样和压缩下实现了最佳的整体鲁棒性，同时提供了显著更快的推理速度。我们的实验代码将发布在 https://github.com/polimi-ispl/CUPID。
View on arXiv Download PDF AI Translation
cs.CV / 79 / 2606.20303

GEN-Guard: Correcting Generalization Failures for Deployable Federated Surgical AI

GEN-Guard：纠正可部署联邦外科人工智能中的泛化失败
Alekseenko, Julia, Mascagni, Pietro, Consortium, AI4SafeChole, Padoy, Nicolas
Abstract
Federated Learning (FL) in surgical video AI enables collaborative model training without sharing sensitive data. However, standard evaluation practices - selecting the "best" global model based only on validation data from participating hospitals - can lead to suboptimal deployment choices. We identify this critical failure mode as performance leakage, where the selected model overfits internal federation data and fails to generalize to unseen institutions. We propose GEN-Guard, a practical post-hoc framework to detect and correct generalization failures in federated surgical AI. It integrates Generalization Detection via Client-Blocked Evaluation (CBE), which validates performance on isolated client distributions to prevent performance leakage, and Generalization Correction through Disagreement-Aware Distillation (DAD), which learns adaptive feature-level corrections for cross-institutional robustness. Both components operate after standard FL convergence while providing robust support for zero-shot adaptation to unseen environments. We first quantify the severity of performance leakage, observing Model Selection Failures (MSFs) exceeding 80% under standard evaluation. GEN-Guard is evaluated on two multi-center clinical challenges: surgical phase recognition in laparoscopic cholecystectomy and polyp segmentation in colonoscopy. Across both datasets, GEN-Guard consistently corrects these failures, improving in-federation F1 scores by up to 2 points, unseen-institution performance by up to 3 points, and worst-case institutional performance by 3-9 points. Performance leakage represents a systematic and previously under-recognized risk in federated surgical AI. GEN-Guard provides a practical solution for detecting and correcting such failures. By improving cross-institutional robustness and zero-shot generalization, it strengthens the reliability of FL for real-world surgical deployment.
Chinese Translation
外科视频人工智能中的联邦学习（FL）使得在不共享敏感数据的情况下进行协作模型训练成为可能。然而，标准评估实践——仅基于参与医院的验证数据选择“最佳”全局模型——可能导致次优的部署选择。我们将这种关键的失败模式称为性能泄漏，其中所选模型过拟合内部联邦数据，无法对未见过的机构进行泛化。我们提出了GEN-Guard，这是一个实用的事后框架，用于检测和纠正联邦外科人工智能中的泛化失败。它集成了通过客户端阻塞评估（CBE）进行的泛化检测，该方法在隔离的客户端分布上验证性能，以防止性能泄漏，以及通过关注分歧的蒸馏（DAD）进行的泛化修正，该方法学习跨机构的自适应特征级修正。这两个组件在标准FL收敛后运行，同时为未见环境的零样本适应提供强有力的支持。我们首先量化了性能泄漏的严重性，观察到在标准评估下模型选择失败（MSF）超过80%。GEN-Guard在两个多中心临床挑战中进行了评估：腹腔镜胆囊切除术中的外科阶段识别和结肠镜检查中的息肉分割。在这两个数据集中，GEN-Guard始终纠正这些失败，使得联邦内F1分数提高了最多2分，未见机构的性能提高了最多3分，最差情况下的机构性能提高了3-9分。性能泄漏代表了联邦外科人工智能中一种系统性且之前未被充分认识的风险。GEN-Guard为检测和纠正此类失败提供了实用的解决方案。通过提高跨机构的鲁棒性和零样本泛化能力，它增强了FL在现实世界外科部署中的可靠性。
View on arXiv Download PDF AI Translation
cs.CV / 80 / 2606.20310

Through the PRISM: Preference Representation in Intermediate States of Video Diffusion Models

通过 PRISM：视频扩散模型中间状态的偏好表示
Wu, Haoxuan, Po, Lai Man, Liu, Mengyang, Li, Kun, Yang, Hongzheng, Liu, Wei
Abstract
Evaluating video generation with clean, pixel-based reward models disconnects evaluation from the noisy diffusion process and incurs massive VAE decoding costs. In this paper, we challenge this paradigm by asking a fundamental question: Can a powerful video generator inherently discriminate preferences directly from noisy latents? To answer this, we introduce \textbf{PRISM} (\textbf{P}reference \textbf{R}epresentation in \textbf{I}ntermediate \textbf{S}tates of Diffusion \textbf{M}odels). PRISM employs a lightweight Query-based Aggregation head with a frozen video diffusion backbone to decode preference signals from noisy latents. Surprisingly, PRISM not only achieves SOTA preference accuracy but also unlocks strong noise-robustness, which enables early-stage Best-of-$N$ sampling. This allows for filtering suboptimal candidates at the very beginning of denoising, drastically reducing computation while boosting video quality. We also reveal a strong positive correlation between a backbone's generative performance and its inherent evaluative power, enabling self-improving video backbones.
Chinese Translation
使用干净的基于像素的奖励模型评估视频生成，使得评估与嘈杂的扩散过程脱节，并产生巨大的变分自编码器（VAE）解码成本。在本文中，我们通过提出一个基本问题来挑战这一范式：强大的视频生成器是否能够直接从嘈杂的潜变量中区分偏好？为了解答这个问题，我们引入了 extbf{PRISM}（ extbf{P}reference extbf{R}epresentation in extbf{I}ntermediate extbf{S}tates of extbf{D}iffusion extbf{M}odels）。PRISM 采用轻量级的基于查询的聚合头，结合冻结的视频扩散主干，从嘈杂的潜变量中解码偏好信号。令人惊讶的是，PRISM 不仅实现了最先进的偏好准确性，还解锁了强大的抗噪声能力，使得早期的最佳 $N$ 采样成为可能。这使得在去噪的最初阶段就能过滤掉次优候选，从而大幅减少计算量，同时提升视频质量。我们还揭示了主干的生成性能与其固有评估能力之间存在强正相关，这使得视频主干能够自我改进。
View on arXiv Download PDF AI Translation
cs.CV / 81 / 2606.20312

Reliability-Aware Prototype Calibration for Frozen Pose-Flow Video Anomaly Detection

考虑可靠性的原型校准用于静态姿态流视频异常检测
Dong, Ning, Su, Yingna, Dong, Xin, Jiao, Ziyun, Guo, Xinnian, Pan, Zhuangzhuang
Abstract
Pose-flow video anomaly detectors are attractive for one-class surveillance because they provide likelihood-based rankings for tracked skeleton windows. However, a single likelihood score may hide multimodal normal behavior and be sensitive to pose-observation noise. We study a frozen-detector setting in which the pose-flow backbone, cached skeleton tracks, and evaluation pipeline are fixed. Reliability-Aware Prototype Calibration (RPC) is a post-hoc score calibration method for this setting. It adds a standardized nearest-prototype deviation in the frozen latent space to the standardized flow score, and uses keypoint confidence only to gate this added geometric evidence. Thus, RPC preserves the original density signal while correcting the ranking with empirical normal-mode structure under pose reliability. Across two frozen pose-flow backbones and four datasets, RPC improves frame-level AUROC in all eight backbone-dataset pairs, with gains ranging from 0.34 to 4.49 percentage points and averaging 2.03 points. Ablation and reliability analyses show that prototype deviation is the main corrective signal, while reliability gating is most useful when pose observations are less trustworthy. These results suggest that lightweight post-hoc calibration can strengthen cached pose-flow systems when retraining or reproducing the full pose pipeline is impractical.
Chinese Translation
姿态流视频异常检测器因其为单类监控提供基于似然的排名而备受关注。然而，单一的似然得分可能掩盖多模态的正常行为，并对姿态观察噪声敏感。我们研究了一种冻结检测器设置，其中姿态流主干、缓存的骨架轨迹和评估管道都是固定的。考虑可靠性的原型校准（Reliability-Aware Prototype Calibration，RPC）是一种针对该设置的事后得分校准方法。它在冻结的潜在空间中将标准化的最近原型偏差添加到标准化的流得分中，并仅使用关键点置信度来限制这一附加的几何证据。因此，RPC在修正排名的同时保留了原始的密度信号，并在姿态可靠性下纠正了经验正常模式结构。在两个冻结的姿态流主干和四个数据集上，RPC在所有八对主干-数据集组合中提高了帧级AUROC，增幅从0.34到4.49个百分点不等，平均增幅为2.03点。消融和可靠性分析表明，原型偏差是主要的校正信号，而在姿态观察不太可信时，可靠性限制最为有效。这些结果表明，轻量级的事后校准可以在重新训练或重现完整姿态管道不切实际时增强缓存的姿态流系统。
View on arXiv Download PDF AI Translation
cs.CV / 82 / 2606.20390

Geometry-Aware Superpixel Graph Transformer with Metadata for Skin Lesion Classification

基于几何感知的超像素图变换器及其元数据在皮肤病变分类中的应用
Azeem, Muhammad, Hussain, Tanveer, Ahmed, Amr, Behera, Ardhendu
Abstract
Automated skin cancer classification from dermoscopic images remains challenging due to heterogeneous lesion structure, strong intra-class variability, and subtle visual differences between benign and malignant cases. Existing CNN/ViT pipelines typically rely on global or patch-level features and often combine patient metadata via late fusion, which limits spatially grounded multimodal reasoning. We present a novel region-based graph learning framework that explicitly models lesions as graphs of spatially coherent superpixel regions represented as frozen CNN features. To capture fine-grained lesion arrangements, we encode inter-regional geometry as edge attributes and introduce a dedicated metadata context node connected to all regions, providing structured integration of demographic/clinical variables within the same relational space. Node representations are updated using our edge-aware graph transformer followed by attention-driven propagation, and a final graph-level embedding for benign-malignant classification. Experiments on four public benchmarks demonstrate that explicit region-level relational modeling and graph-native multimodal fusion yield consistent gains over the state-of-the-art. Consequently, we establish a new graph-centric perspective in which CNN features are modeled as relational nodes and improved through contextual integration, yielding more expressive and robust classifications.
Chinese Translation
从皮肤镜图像中自动化分类皮肤癌仍然面临挑战，原因在于病变结构的异质性、强烈的类内变异性以及良性与恶性病例之间微妙的视觉差异。现有的卷积神经网络（CNN）/视觉变换器（ViT）管道通常依赖于全局或补丁级特征，并且常常通过后期融合结合患者元数据，这限制了空间基础的多模态推理。我们提出了一种新颖的基于区域的图学习框架，该框架明确将病变建模为由空间一致的超像素区域构成的图，这些区域以冻结的CNN特征表示。为了捕捉细粒度的病变排列，我们将区域间几何关系编码为边属性，并引入一个专门的元数据上下文节点，该节点与所有区域相连，在同一关系空间内提供人口统计/临床变量的结构化整合。节点表示通过我们的边感知图变换器进行更新，随后进行基于注意力的传播，并生成最终的图级嵌入用于良性-恶性分类。在四个公共基准上的实验表明，显式的区域级关系建模和图原生多模态融合相较于最先进的方法产生了一致的提升。因此，我们建立了一种新的以图为中心的视角，其中CNN特征被建模为关系节点，并通过上下文整合得到改善，从而实现更具表现力和鲁棒性的分类。
View on arXiv Download PDF AI Translation
cs.CV / 83 / 2606.20404

FlowBender: Feedback-Aware Training for Self-Correcting Conditional Flows

FlowBender：反馈感知自我校正条件流的训练
Gilo, Daniel, Elflein, Sven, Sobol, Ido, Litany, Or
Abstract
Conditional diffusion and flow models routinely fail to satisfy the very constraints that define their task. For instance, a depth-conditioned model often produces images whose re-extracted depth disagrees with the input, even though the forward operator--the depth predictor defining the constraint--is available during both training and inference. Existing approaches generally fall into two categories: supervised models that treat the conditioning signal as a static cue and ignore alignment information at inference, and guidance-based methods that consult it through hand-tuned linear updates, typically trading fidelity to the condition against the plausibility of the generated sample. We argue that the fundamental gap in both paradigms is that the model is never trained to utilize its own alignment error. We introduce FlowBender, a closed-loop framework that treats this error as a first-class input, training the network to learn a correction policy conditioned on inference-time feedback. At each step, an unguided look-ahead pass estimates the clean signal, a task-specific deviation is computed via the forward operator, and a refinement pass consumes this signal to produce a corrected velocity. We propose several variants of FlowBender, including a gradient-based formulation for differentiable operators and a zero-order variant for non-differentiable settings such as JPEG compression. For efficient sampling, we introduce a prior-step shortcut that enables closed-loop correction at a minimal additional computational cost. Across image-to-image translation, restoration, and 3D mesh texturing, FlowBender consistently outperforms standard supervised baselines, alignment-loss-augmented training, and state-of-the-art inference-time guidance, improving fidelity and plausibility simultaneously rather than trading them against each other. Project page: https://flow-bender.github.io/
Chinese Translation
条件扩散和流模型通常无法满足定义其任务的约束。例如，一个深度条件模型常常生成的图像，其重新提取的深度与输入不一致，即使前向算子——定义约束的深度预测器——在训练和推理过程中均可用。现有方法一般分为两类：将条件信号视为静态线索的监督模型，在推理时忽略对齐信息，以及通过手动调整线性更新来咨询该信息的基于引导的方法，通常是在条件的保真度与生成样本的合理性之间进行权衡。我们认为这两种范式的根本差距在于模型从未被训练去利用自身的对齐误差。我们提出FlowBender，一个将此误差视为一类重要输入的闭环框架，训练网络学习基于推理时反馈的校正策略。在每一步中，一个无引导的前瞻传递估计干净信号，通过前向算子计算任务特定的偏差，并且一个精炼传递消耗该信号以生成校正的速度。我们提出了FlowBender的几种变体，包括用于可微算子的基于梯度的公式和用于非可微设置（如JPEG压缩）的零阶变体。为了高效采样，我们引入了一个先前步骤的快捷方式，使得在最小额外计算成本下实现闭环校正。在图像到图像的转换、恢复和3D网格纹理化中，FlowBender始终优于标准监督基线、增强对齐损失的训练和最先进的推理时引导，同时提高保真度和合理性，而不是相互权衡。项目页面：https://flow-bender.github.io/
View on arXiv Download PDF AI Translation
cs.CV / 84 / 2606.20419

Spectral Query-Key Product Weight Steering for Training-Free VLM Hallucination Mitigation

用于无训练视觉语言模型幻觉减轻的谱查询-键乘积权重引导
Tiwari, Karn, Chordia, Varnith, P, Prathosh A
Abstract
Vision-language models (VLMs) often generate fluent but visually unsupported descriptions, especially by mentioning objects absent from the image. We propose QK Product Steering, a data-free, training-free, and zero-inference-cost weight edit for reducing object hallucination. The method directly edits the per-head query-key product, the operator that produces pre-softmax attention logits, by suppressing a small number of dominant singular modes in selected middle layers. The edited product is then mapped back to the query weights through a closed-form query-only update while keeping shared key weights fixed, making the edit compatible with grouped-query attention. We further decompose the QK product into symmetric and antisymmetric components to distinguish mutual content-similarity patterns from directional attention patterns. Across three GQA-based VLMs, QK Product Steering achieves an average relative CHAIR$_s$ reduction of $4.0\%$, while matched random-mode controls show negligible change. Interpretability ablations show that the hallucination signal is specific to dominant QK modes and is primarily localized to the symmetric mutual-attention channel. Overall, QK Product Steering offers a simple alternative to decoding-time mitigation, requiring no additional data, fine-tuning, or inference-time overhead while largely preserving general multimodal capability.
Chinese Translation
视觉语言模型（VLMs）通常生成流畅但缺乏视觉支持的描述，尤其是在提及图像中缺失的物体时。我们提出了QK乘积引导，这是一种无数据、无训练且零推理成本的权重编辑方法，用于减少物体幻觉。该方法通过抑制选定中间层中少量主导奇异模式，直接编辑每个头的查询-键乘积，该运算符生成预软最大注意力逻辑值。编辑后的乘积通过封闭形式的仅查询更新映射回查询权重，同时保持共享的键权重不变，使得该编辑与分组查询注意力兼容。我们进一步将QK乘积分解为对称和反对称成分，以区分互内容相似模式与方向注意力模式。在三个基于GQA的VLM中，QK乘积引导实现了平均相对CHAIR$_s$减少$4.0 rac{ ext{ extperthousand}}{}$，而匹配的随机模式控制显示出微不足道的变化。可解释性消融实验表明，幻觉信号特定于主导的QK模式，并主要局限于对称的互注意通道。总体而言，QK乘积引导为解码时的减轻提供了一种简单的替代方案，无需额外的数据、微调或推理时间开销，同时在很大程度上保留了通用的多模态能力。
View on arXiv Download PDF AI Translation
cs.CV / 85 / 2606.20449

InfantFace: Detecting infant faces in neonatal clinical environments

InfantFace：在新生儿临床环境中检测婴儿面孔
Bin-Obaid, Abdullah, Cobo, Maria M., Slater, Rebeccah, Tarassenko, Lionel, Villarroel, Mauricio
Abstract
Reliable localisation of the neonatal face is the first step for several video-camera based non-contact assessments such as pain and distress related facial expression analysis, pain scoring, cardiorespiratory signal extraction and cessation of breathing alerts. However, major challenges persist in neonatal clinical environments. Cluttered backgrounds, illumination changes and poor lighting conditions can reduce the accuracy of face detection models. Clinical interventions, monitoring equipment and, in some cases, medical devices can obstruct the face, making visual assessment difficult. We propose a one-stage YOLOv11m-based model tailored for face detection of infants in neonatal clinical environments. We combined multiple publicly available datasets (VGGFace2, CelebA, FDDB, WIDER FACE) to train and evaluate our proposed model. We then fine-tuned our model on a neonatal research dataset involving 228 videos from 114 recording sessions of 113 independent infants. Before fine-tuning, our model achieved an AP50 of 0.87, surpassing the performance of three state-of-the-art general face detectors. Performance improved further to an AP50 of 0.96 after clinical-domain adaptation. Evaluating face detection performance across different datasets remains a challenge due to the lack of publicly available neonatal datasets. Prioritising the creation of such datasets, while upholding appropriate privacy safeguards and ethical standards in their creation and use, would greatly support further progress in this field.
Chinese Translation
新生儿面孔的可靠定位是基于视频摄像头的非接触评估的第一步，例如与疼痛和痛苦相关的面部表情分析、疼痛评分、心肺信号提取和呼吸停止警报。然而，在新生儿临床环境中仍然存在重大挑战。杂乱的背景、光照变化和较差的照明条件可能会降低面部检测模型的准确性。临床干预、监测设备以及在某些情况下的医疗设备可能会遮挡面部，导致视觉评估变得困难。我们提出了一种基于YOLOv11m的一阶段模型，专门用于在新生儿临床环境中检测婴儿面孔。我们结合了多个公开可用的数据集（VGGFace2、CelebA、FDDB、WIDER FACE）来训练和评估我们提出的模型。随后，我们在一个包含228个视频、来自114个录制会话的113名独立婴儿的新生儿研究数据集上对模型进行了微调。在微调之前，我们的模型达到了0.87的AP50，超过了三种最先进的通用面部检测器的性能。经过临床领域适应后，性能进一步提高至0.96的AP50。由于缺乏公开可用的新生儿数据集，在不同数据集上评估面部检测性能仍然是一个挑战。优先创建此类数据集，同时在其创建和使用过程中维护适当的隐私保护和伦理标准，将大大支持该领域的进一步发展。
View on arXiv Download PDF AI Translation
cs.CV / 86 / 2606.20455

PCFootprint: A Large-Scale Dataset and Benchmark for Vectorized Building Footprint Extraction from Aerial LiDAR Point Clouds

PCFootprint：一个用于从航空激光雷达点云中提取矢量化建筑轮廓的大规模数据集和基准
Shen, Haoyuan, Wang, Kuihao, Wang, Ruisheng, Liu, Yujun
Abstract
Building footprint extraction is a fundamental task in photogrammetry, remote sensing, and computer vision. Recent image-based methods have achieved remarkable progress in extracting vectorized footprints from high-resolution optical imagery. However, optical imagery inherently susceptible to occlusions, perspective distortions, and residual relief displacement, yielding incomplete or misaligned footprint extraction. Furthermore, the lack of explicit elevation information limits its direct applicability to Level of Detail building modeling. In this paper, we present PCFootprint, the first large-scale public dataset for footprint extraction from airborne laser scanning point clouds. PCFootprint comprises \num{33000} tiles derived from the Estonian Land and Spatial Development Board, covering diverse urban and rural landscapes. Each tile spans \qtyproduct{128 x 128}{\m} with systematically aligned vectorized footprints aligned to point clouds. The dataset includes a \num{3000} tiles cross-domain test set for evaluating generalization across geographic regions. We establish comprehensive benchmarks by evaluating mainstream methods. Experimental results reveal significant challenges including high intra-class variance, data imbalance, and noise across complex geospatial environments. We believe PCFootprint will advance future research in building modeling, urban scene understanding, and geospatial analysis. The PCFootprint dataset is publicly available at \url{https://huggingface.co/datasets/Haoyuan-Shen/PCFootprint}.
Chinese Translation
建筑轮廓提取是摄影测量、遥感和计算机视觉中的一项基础任务。近年来，基于图像的方法在从高分辨率光学影像中提取矢量化轮廓方面取得了显著进展。然而，光学影像本质上容易受到遮挡、透视畸变和残余地形位移的影响，导致提取的轮廓不完整或错位。此外，缺乏明确的高程信息限制了其在细节层次建筑建模中的直接应用。在本文中，我们提出了PCFootprint，这是第一个用于从航空激光扫描点云中提取轮廓的大规模公共数据集。PCFootprint包含来自爱沙尼亚土地与空间发展局的33,000个切片，涵盖了多样的城市和乡村景观。每个切片的大小为128 x 128米，并且系统地对齐了与点云对齐的矢量化轮廓。该数据集还包括一个包含3,000个切片的跨领域测试集，用于评估不同地理区域的泛化能力。我们通过评估主流方法建立了全面的基准。实验结果揭示了显著的挑战，包括高的类内方差、数据不平衡以及复杂地理环境中的噪声。我们相信PCFootprint将推动未来在建筑建模、城市场景理解和地理空间分析方面的研究。PCFootprint数据集可在以下网址公开获取： https://huggingface.co/datasets/Haoyuan-Shen/PCFootprint。
View on arXiv Download PDF AI Translation
cs.CV / 87 / 2606.20477

Scalable Training of Spatially Grounded 2D Vision-Language Models for Radiology

可扩展的空间基础二维视觉语言模型训练用于放射学
Salcan, Yusuf, Ging, Simon, Schirrmeister, Robin, Arnold, Philipp, Kotter, Elmar, Bozorgtabar, Behzad, Brox, Thomas
Abstract
We study how to train visually grounded vision-language models (VLMs) for radiology without manual spatial annotations. We introduce RefRad2D, a large-scale bilingual (German/English) dataset of 1.2M CT and MR image-text pairs derived from clinical practice, with task-specific VQA and spatial grounding subsets generated automatically via LLM-based curation and automated segmentation. Trained on this data, our model RadGrounder jointly performs report generation, visual question answering, and spatial grounding via bounding-box detection or segmentation. On external VQA benchmarks (Slake, VQA-RAD), RadGrounder achieves competitive results with specialized medical VLMs. Adding our clinical data to the training mixture improves open-ended VQA over fine-tuning on the downstream datasets alone, showing the transferability of our dataset. Crucially, adding grounding supervision does not degrade language quality, enabling spatially verifiable outputs at no cost to VQA performance.
Chinese Translation
我们研究如何在没有手动空间注释的情况下，为放射学训练视觉基础的视觉语言模型（VLMs）。我们引入了RefRad2D，这是一个大规模的双语（德语/英语）数据集，包含120万对来自临床实践的CT和MR图像-文本对，具有通过基于大型语言模型（LLM）的策划和自动分割生成的特定任务视觉问答（VQA）和空间基础子集。在该数据上训练的模型RadGrounder能够通过边界框检测或分割共同执行报告生成、视觉问答和空间基础。RadGrounder在外部VQA基准（Slake, VQA-RAD）上取得了与专业医疗VLMs相竞争的结果。将我们的临床数据添加到训练混合中，相较于仅在下游数据集上进行微调，显著提高了开放式VQA的性能，显示了我们数据集的可迁移性。关键是，添加空间基础监督并未降低语言质量，使得在不影响VQA性能的情况下能够生成空间可验证的输出。
View on arXiv Download PDF AI Translation
cs.CV / 88 / 2606.20488

How Fragile Are Training-Free AI-Generated Image Detectors? A Controlled Audit of Score Direction, Preprocessing, and Compression

无训练的AI生成图像检测器有多脆弱？对评分方向、预处理和压缩的控制审计
Zhou, Jingwen, Wang, Mingzhe
Abstract
Training-free detectors of AI-generated images promise generator-agnostic deployment without classifier training, yet their reported numbers are rarely compared under a single controlled protocol. We audit two representative training-free scores -- an autoencoder-reconstruction score (AEROBLADE-style) and a noise-perturbation feature-similarity score (RIGID-style) -- plus a naive feature-kNN control, on a common 1,500-image GenImage-derived benchmark spanning seven generators and JPEG compression at quality 70 and 50. The audit yields three cautionary findings. (i) Implementation details masquerade as method differences: replacing the LPIPS backbone (AlexNet -> VGG-16) changes overall AUROC by +0.085, and switching between resize-to-512 and native-resolution preprocessing flips per-generator conclusions by up to 0.38 AUROC. (ii) Score direction is not a property of the method but of its hyperparameters: the RIGID-style score is inverted (AUROC < 0.5) on SD1.5 and Wukong at noise level sigma=0.05, recovers to >0.5 for every generator at sigma=0.01, and collapses to 0.15 at sigma=0.3. (iii) Dataset format bias inflates robustness claims: without unified re-encoding, AUROC under JPEG-50 exceeds the clean condition for the AlexNet-backbone reconstruction score; after bias correction the residual anomaly localizes to a single generator (BigGAN). The audited scores have complementary per-generator failure sets, but naive z-score fusion does not beat the best single score, indicating that exploiting complementarity requires direction-aware combination.
Chinese Translation
无训练的AI生成图像检测器承诺在不进行分类器训练的情况下实现生成器无关的部署，但其报告的数据很少在单一控制协议下进行比较。我们对两种具有代表性的无训练评分进行了审计——一种是自编码器重构评分（AEROBLADE风格），另一种是噪声扰动特征相似性评分（RIGID风格），以及一种简单的特征k最近邻控制，基于一个包含1500张图像的GenImage衍生基准，涵盖七种生成器和JPEG压缩质量70和50。审计结果得出了三个警示性发现。(i) 实施细节伪装成方法差异：替换LPIPS骨干网络（从AlexNet到VGG-16）使整体AUROC变化+0.085，而在resize-to-512和原生分辨率预处理之间切换则使每个生成器的结论变化高达0.38 AUROC。(ii) 评分方向不是方法的属性，而是其超参数的属性：在噪声水平sigma=0.05时，RIGID风格评分在SD1.5和Wukong上被反转（AUROC < 0.5），在sigma=0.01时恢复到每个生成器均大于0.5，而在sigma=0.3时则崩溃至0.15。(iii) 数据集格式偏差夸大了鲁棒性声明：在没有统一重新编码的情况下，JPEG-50下的AUROC超过了AlexNet骨干重构评分的干净条件；在偏差校正后，残余异常局限于单一生成器（BigGAN）。审计的评分在每个生成器上具有互补的失败集，但简单的z-score融合并未超越最佳单一评分，这表明利用互补性需要方向感知的组合。
View on arXiv Download PDF AI Translation
cs.CV / 89 / 2606.20506

FreeStyle: Free Control of Style-Content Dual-Reference Generation from Community LoRA Mining

FreeStyle：基于社区 LoRA 挖掘的风格-内容双参考生成的自由控制
Lan, Jinghong, Cheng, Wei, Chen, Yunuo, Ye, Ziqi, Xing, Peng, Fang, Yixiao, Wang, Rui, Yang, Yufeng, Zhang, Xuanyang, Zeng, Xianfang, Zou, Difan, Yu, Gang, Zhang, Chi
Abstract
Style-content dual-reference generation aims to synthesize an image that preserves the structure and semantics of a content reference while adopting the style of a separate style reference.Despite recent progress, this setting remains challenging because models must balance content fidelity, style alignment, and instruction following avoiding semantic leakage from the style reference.A key bottleneck is the lack of large-scale triplet data with clean content-style separation and broad long-tail style coverage.In this work, we propose FreeStyle, a scalable dual-reference generation framework based on community LoRA mining.We treat community LoRAs as compositional anchors for style and content, and design a rigorous generation and filtering pipeline to construct large-scale Style-Reference and Content-Reference triplets across multiple base models.To address content leakage, we adopt a two-stage curriculum with stage-specific disentanglement mechanisms: an attention-level enrichment constraint that suppresses style-reference leakage in the style-transfer stage, and a frequency-aware RoPE modulation strategy that targets positional-correspondence-based leakage in the harder dual-reference stage.We also introduce a benchmark covering both style-reference and dual-reference generation, with evaluations on style similarity, content preservation, aesthetics, instruction following, and leakage rejection. The benchmark incorporates a style-invariant Content Alignment Score (CAS) and introduces a calibrated VLM-based Rejection Score for evaluating generation reliability and leakage suppression.Extensive experiments show that our model achieves a strong balance among style alignment, content preservation, and leakage suppression.
Chinese Translation
风格-内容双参考生成旨在合成一幅图像，该图像在保留内容参考的结构和语义的同时，采用独立的风格参考的风格。尽管近期取得了一些进展，但这一设置仍然具有挑战性，因为模型必须在内容保真度、风格一致性和遵循指令之间取得平衡，避免风格参考的语义泄漏。一个关键瓶颈是缺乏具有清晰内容-风格分离和广泛长尾风格覆盖的大规模三元组数据。在本研究中，我们提出了 FreeStyle，一种基于社区 LoRA 挖掘的可扩展双参考生成框架。我们将社区 LoRA 视为风格和内容的组合锚点，并设计了一条严格的生成和过滤管道，以构建跨多个基础模型的大规模风格参考和内容参考三元组。为了解决内容泄漏问题，我们采用了一个两阶段的课程，具有阶段特定的解耦机制：在风格转移阶段，采用注意力级别的丰富约束来抑制风格参考的泄漏；在更困难的双参考阶段，采用基于频率的 RoPE 调制策略来针对基于位置对应的泄漏。我们还引入了一个涵盖风格参考和双参考生成的基准，评估风格相似性、内容保留、美学、指令遵循和泄漏拒绝。该基准包含一个风格不变的内容对齐评分（CAS），并引入一个经过校准的基于 VLM 的拒绝评分，以评估生成的可靠性和泄漏抑制。大量实验表明，我们的模型在风格一致性、内容保留和泄漏抑制之间实现了良好的平衡。
View on arXiv Download PDF AI Translation
cs.CV / 90 / 2606.20515

S-Agent: Spatial Tool-Use Elicits Reasoning for Spatial Intelligence

S-Agent：空间工具使用引发空间智能推理
Dai, Yalun, Li, Hao, Tian, Shulin, Yao, Runmao, Dong, Yuhao, Hong, Fangzhou, Chen, Zhaoxi, Liu, Fangfu, Tian, Baoliang, Zhang, Dingwen, Wang, Tao, Yap, Kim-Hui, Liu, Ziwei
Abstract
Real-world spatial intelligence requires reasoning over a continuous and evolving 3D world, yet existing VLMs and tool-augmented agents largely remain tied to static, stateless inference from isolated visual observations. We introduce \textbf{\textsc{S-Agent}}, a spatial tool-use agentic paradigm for understanding and reasoning over continuous multi-view images and videos. By formulating spatial reasoning as spatio-temporal evidence accumulation rather than isolated frame-level prediction, \textsc{S-Agent} reshapes spatial perception into scene-centric understanding beyond frame-centric recognition. Specifically, \textsc{S-Agent} casts the VLM as a semantic planner that decides what evidence is needed, while a hierarchy of spatial tools and experts grounds objects in 2D, lifts them into 3D geometric evidence, and aggregates this evidence into high-level spatial knowledge (\textit{e.g.}, counting, measurement, orientation, and relative position). Additionally, a temporal memory mechanism, including Scene Memory for maintaining the evolving scene state and Agent Memory for accumulating reasoning context, enables evidence integration across frames and reasoning steps. Comprehensive experiments on multi-view and video spatial reasoning benchmarks show that \textsc{S-Agent} consistently improves both open-source and closed-source VLMs in a training-free manner. Beyond inference-time augmentation, supervised fine-tuning (SFT) on \textsc{S-Agent}-generated spatial trajectories \textsc{S-300K} yields \textsc{S-Agent-8B}, a compact spatial agent that significantly surpasses similar-scale baselines (e.g., Qwen3-VL-8B) and performs comparably to advanced closed-source models (e.g., GPT-5.4 and Gemini 3).
Chinese Translation
现实世界的空间智能需要对一个连续且不断发展的三维世界进行推理，而现有的视觉语言模型（VLMs）和工具增强代理在很大程度上仍然局限于从孤立的视觉观察中进行静态、无状态的推理。我们提出了 extbf{ extsc{S-Agent}}，一种空间工具使用的代理范式，用于理解和推理连续的多视角图像和视频。通过将空间推理表述为时空证据积累，而非孤立的帧级预测， extsc{S-Agent}将空间感知重塑为超越帧中心识别的场景中心理解。具体而言， extsc{S-Agent}将视觉语言模型视为一种语义规划者，决定所需的证据，而一系列空间工具和专家则在二维中定位对象，将其提升为三维几何证据，并将这些证据汇聚为高级空间知识（例如，计数、测量、方向和相对位置）。此外，包含场景记忆的时间记忆机制用于维护不断演变的场景状态，而代理记忆则用于积累推理上下文，使得跨帧和推理步骤的证据整合成为可能。在多视角和视频空间推理基准上的全面实验表明， extsc{S-Agent}在无训练的情况下持续改善了开源和闭源的视觉语言模型。除了推理时的增强外，对 extsc{S-Agent}生成的空间轨迹 extsc{S-300K}进行监督微调（SFT）产生了 extsc{S-Agent-8B}，这是一个紧凑的空间代理，显著超越了类似规模的基线（例如，Qwen3-VL-8B），并与先进的闭源模型（例如，GPT-5.4和Gemini 3）表现相当。
View on arXiv Download PDF AI Translation
cs.CV / 91 / 2606.20521

HumanScale: Egocentric Human Video Can Outperform Real-Robot Data for Embodied Pretraining

人类尺度：以自我为中心的人类视频在具身预训练中优于真实机器人数据
Ma, Juncheng, Bi, Jianxin, Deng, Yufan, Zhai, Xuanran, Zhang, Kewei, Huang, Ye, Liang, Bo, Gong, Shukai, Tu, Jiankai, Tang, Xiaotian, Li, Jiaxin, Chen, Kaiqi, Wang, Duomin, Wang, Yuqi, Kang, Bingyi, Huang, Eric, Dou, Zhiyang, Dong, Zhen, Xie, Enze, Matusik, Wojciech, Chua, Tat-Seng, Zhou, Daquan
Abstract
Embodied foundation models are expected to benefit from data scaling like large language models, but face a much tighter data bottleneck. Teleoperated real-robot trajectories remain the dominant pretraining source due to their precise action supervision and embodiment alignment, yet their scalability is limited by high collection cost, acquisition difficulty, and low behavioral and environmental diversity. These limitations have sparked interest in egocentric human video as a scalable, substantially lower-cost, and more diverse alternative for embodied model pretraining. However, its effectiveness compared to teleoperated real-robot data remains underexplored. To address this question, we conduct a systematic study comparing egocentric human video and teleoperated real-robot trajectories as pretraining data sources for embodied foundation models, under fixed post-training and validation protocols. Surprisingly, we find that egocentric data, when processed through a carefully designed filtering and labeling pipeline, is not merely a viable substitute for model pretraining but can lead to superior performance. With the same amount of pretraining data, models pretrained on egocentric data achieve a 24% lower validation loss on real-robot action prediction, as well as 52.5% and 90% higher success rates on in-distribution and out-of-distribution real-robot task execution, respectively. This finding verifies a scalable paradigm for embodied foundation models: pretrain on egocentric human video to learn diverse world representations, then adapt with a small amount of labeled real-robot data for action-space alignment. We hope this study encourages broader exploration of egocentric data and offers guidance for data quality assessment before costly robot data collection.
Chinese Translation
具身基础模型预计将从数据扩展中受益，类似于大型语言模型，但面临更为严格的数据瓶颈。由于其精确的动作监督和具身对齐，遥控真实机器人轨迹仍然是主要的预训练来源，但其可扩展性受到高收集成本、获取难度以及低行为和环境多样性的限制。这些限制引发了对以自我为中心的人类视频的兴趣，作为一种可扩展、成本显著更低且更具多样性的具身模型预训练替代方案。然而，与遥控真实机器人数据相比，其有效性仍未得到充分探索。为了解决这个问题，我们进行了一项系统研究，比较以自我为中心的人类视频和遥控真实机器人轨迹作为具身基础模型的预训练数据来源，采用固定的后训练和验证协议。令人惊讶的是，我们发现经过精心设计的过滤和标注流程处理后，以自我为中心的数据不仅是模型预训练的可行替代品，而且可以带来更优的性能。在相同数量的预训练数据下，基于以自我为中心的数据预训练的模型在真实机器人动作预测中实现了24%的验证损失降低，以及在分布内和分布外真实机器人任务执行中的成功率分别提高了52.5%和90%。这一发现验证了具身基础模型的可扩展范式：在以自我为中心的人类视频上进行预训练以学习多样的世界表征，然后通过少量标注的真实机器人数据进行动作空间对齐的适应。我们希望这项研究能鼓励对以自我为中心的数据进行更广泛的探索，并为在高成本的机器人数据收集之前的数据质量评估提供指导。
View on arXiv Download PDF AI Translation
cs.CV / 92 / 2606.20523

SARLO-80: Worldwide Slant SAR Language Optic Dataset 80cm

SARLO-80：全球倾斜合成孔径雷达语言光学数据集80cm
Debuysère, Solène, Trouvé, Nicolas, Letheule, Nathan, Colin, Elise, Channing, Georgia
Abstract
Multimodal foundation models have advanced rapidly thanks to large optical benchmarks, but comparable resources for synthetic aperture radar (SAR) remain limited. Existing SAR--optical datasets largely rely on low-resolution, intensity-only Ground Range Detected~(GRD) products and do not preserve complex-valued SAR measurements or native acquisition geometry, which restricts physically grounded multimodal learning. In particular, large-scale public datasets combining very-high-resolution (VHR) SAR SLC, aligned optical imagery, and natural-language descriptions are still lacking. We present a VHR SAR--optical--text dataset built from open-access Umbra spotlight acquisitions distributed as Sensor Independent Complex Data (SICD). From around 2,500 worldwide scenes (VV/HH, 20cm--2m native resolution), we standardize all SAR data to an 80cm slant-range grid via band-limited FFT resampling and tile the imagery into 1024 by 1024 patches. For each SAR patch, we retrieve a high-resolution optical tile and warp it into the SAR grid using local coordinate correspondences for local pixel-level alignment. We further generate three caption variants (SHORT/MID/LONG) per sample to support vision--language training and evaluation. Our dataset contains 119,566 triplets (complex and amplitude slant-range SAR patch, aligned optical patch, natural-language description) covering 257 locations across 72 countries and a broad range of land types and infrastructures. We release fixed train/validation/test splits and the full preprocessing and baseline code to enable reproducible benchmarks for multimodal alignment on cross-modal retrieval and conditional generation in native SAR geometry. The dataset is publicly available on the Hugging Face Hub at https://huggingface.co/datasets/ONERA/SARLO-80.
Chinese Translation
多模态基础模型由于大型光学基准的快速发展而取得了显著进展，但合成孔径雷达（SAR）相关的资源仍然有限。现有的SAR-光学数据集主要依赖于低分辨率、仅强度的地面范围检测（GRD）产品，未能保留复杂值的SAR测量或原生采集几何，这限制了基于物理的多模态学习。特别是，缺乏结合超高分辨率（VHR）SAR SLC、对齐的光学影像和自然语言描述的大规模公共数据集。我们提出了一个基于开放获取的Umbra聚焦采集构建的VHR SAR-光学-文本数据集，数据以传感器独立复杂数据（SICD）形式分发。我们从约2500个全球场景（VV/HH，20cm-2m原生分辨率）中，将所有SAR数据标准化为80cm倾斜范围网格，通过带限FFT重采样，并将影像切分为1024 x 1024的块。对于每个SAR块，我们检索一个高分辨率的光学块，并使用局部坐标对应关系将其扭曲到SAR网格中，以实现局部像素级对齐。我们进一步为每个样本生成三种标题变体（短/中/长），以支持视觉-语言训练和评估。我们的数据集包含119,566个三元组（复杂和幅度倾斜范围SAR块、对齐光学块、自然语言描述），覆盖72个国家的257个地点以及广泛的土地类型和基础设施。我们发布了固定的训练/验证/测试划分及完整的预处理和基线代码，以便在原生SAR几何下实现多模态对齐的跨模态检索和条件生成的可重复基准。该数据集已在Hugging Face Hub上公开，链接为：https://huggingface.co/datasets/ONERA/SARLO-80。
View on arXiv Download PDF AI Translation
cs.CV / 93 / 2606.20531

VisDom: Sparse Novel View Synthesis with Visible Domain Constraint

VisDom：具有可见域约束的稀疏新视图合成
Gladkova*, Mariia, Yenamandra*, Tarun, Boyer, Edmond, Maier, Robert, Tung, Tony, Cremers, Daniel
Abstract
Sparse novel view synthesis (NVS) remains challenging due to the ambiguity of recovering 3D geometry from few input views. While NeRF- and Gaussian Splatting (GS)-based methods perform well with dense supervision, they often overfit in sparse settings, producing floating artifacts and inconsistent geometry. Silhouette consistency is commonly used as a regularizer, but it remains insufficient, as silhouette-consistent regions can extend beyond the true object geometry. We introduce VisDom, a learning-free geometric constraint that augments classical carving-based visual hull reconstruction by enforcing a minimum multi-view visibility requirement. Specifically, we define a visible domain as the subset of 3D space observed by at least $K$ views and use it as an additional filtering criterion on top of standard silhouette-based reconstruction. This provides a stronger spatial prior in sparse-view settings. We integrate VisDom into both implicit (NeRF) and explicit (GS) pipelines by restricting volumetric sampling and guiding Gaussian placement during optimization. Experiments on three challenging datasets show consistent improvements in sparse-view NVS, enabling high-quality object-centric reconstruction from as few as four input images. Our method is domain-agnostic, requires only silhouettes, and introduces no learned parameters, making it a simple complement to existing approaches. Applying VisDom on top of GaussianObject further improves performance on Omni3D and MipNeRF360, while matching or surpassing it at 22 $\times$ lower training cost.
Chinese Translation
稀疏新视图合成（NVS）仍然面临挑战，因为从少量输入视图中恢复三维几何形状存在模糊性。虽然基于 NeRF 和高斯点云（Gaussian Splatting, GS）的方法在密集监督下表现良好，但在稀疏环境中往往会出现过拟合，产生漂浮伪影和不一致的几何形状。轮廓一致性通常被用作正则化器，但它仍然不足，因为轮廓一致的区域可能超出真实物体的几何形状。我们提出了 VisDom，这是一种无学习的几何约束，通过强制执行最小多视图可见性要求来增强经典的基于雕刻的视觉外壳重建。具体而言，我们将可见域定义为至少被 $K$ 个视图观察到的三维空间子集，并将其作为标准轮廓重建的附加过滤标准。这在稀疏视图环境中提供了更强的空间先验。我们将 VisDom 集成到隐式（NeRF）和显式（GS）管道中，通过限制体积采样和在优化过程中指导高斯放置。对三个具有挑战性的数据集的实验表明，在稀疏视图 NVS 中取得了一致的改进，使得从仅四张输入图像中实现高质量的以物体为中心的重建成为可能。我们的方法与领域无关，仅需轮廓，并且不引入任何学习参数，使其成为现有方法的简单补充。在高斯物体（GaussianObject）之上应用 VisDom 进一步提高了 Omni3D 和 MipNeRF360 的性能，同时以 22 倍更低的训练成本匹配或超越了其表现。
View on arXiv Download PDF AI Translation
cs.CV / 94 / 2606.20536

The FID Lottery: Quantifying Hidden Randomness in Generative-Model Evaluation

FID彩票：量化生成模型评估中的隐含随机性
Dufour, Nicolas, Efros, Alexei A., Pérez, Patrick
Abstract
The Frechet Inception Distance (FID) is the de facto arbiter of image generation, yet most papers report just a single number from a single trained model using a single sampling seed. How reproducible is that number if we retrain the model, or merely resample from it? In this paper, we treat FID as a random variable on a two-axis panel of training and generation seeds, and measure its variance directly on several hundred SiT networks trained on class-conditional ImageNet 256x256. We report surprising findings: (a) Retraining the model using the same recipe with a different seed moves FID 3.2x more (in Inception feature space) than redrawing samples from a fixed network. (b) That gap is driven by three factors: random initialisation, data ordering, and the per-step Gaussian noise of the flow-matching loss. (c) Increasing compute or model size barely tightens the spread, holding the FID coefficient of variation (CoV) inside a 1-2% band. (d) Per-cell classifier-free-guidance tuning halves the spread but reshuffles which seeds work best, and a lucky training seed reaches the same FID with up to 2x less compute than an unlucky one. Based on these findings, we recommend a new FID evaluation protocol: evaluate under per-cell optimal guidance, treat any FID gap below the empirically measured ~1.3% CoV as inconclusive, and report an error bar over several training seeds rather than a single FID number.
Chinese Translation
Frechet Inception Distance (FID) 是图像生成的事实标准，然而大多数论文仅报告来自单个训练模型和单个采样种子的单一数值。如果我们重新训练模型或仅仅从中重新采样，这个数值的可重复性如何？在本文中，我们将FID视为在训练和生成种子的双轴面板上的随机变量，并在数百个在条件分类的ImageNet 256x256上训练的SiT网络上直接测量其方差。我们报告了令人惊讶的发现：（a）使用不同种子以相同配方重新训练模型时，FID在Inception特征空间中移动的幅度比从固定网络重新绘制样本时大3.2倍。（b）这一差距由三个因素驱动：随机初始化、数据顺序和流匹配损失的每步高斯噪声。（c）增加计算能力或模型规模几乎不会缩小分布，将FID变异系数（CoV）保持在1-2%的范围内。（d）每个单元的无分类器引导调优将分布减半，但重新排列了哪些种子效果最佳，而一个幸运的训练种子在计算量上可比一个不幸的种子少2倍而达到相同的FID。基于这些发现，我们建议一种新的FID评估协议：在每个单元的最佳引导下进行评估，将任何低于经验测得的~1.3% CoV的FID差距视为不确定，并报告多个训练种子的误差条而不是单一的FID数值。
View on arXiv Download PDF AI Translation
cs.CV / 95 / 2606.20542

CalTennis: Large Multi-View Tennis Video Dataset and Benchmark of Monocular-to-3D Pose Estimation

CalTennis：大型多视角网球视频数据集及单目到三维姿态估计基准
Demler, Ilona, Xie, Xinran, Werner, Blake, Szczuka, Anna, Perona, Pietro
Abstract
The Caltech Tennis Dataset (CalTennis) is a large-scale video benchmark for evaluating monocular-to-3D pose estimation in the wild. CalTennis comprises over 11 million frames (51 hours) of tennis practice and match play from 40 players, captured with 2-6 synchronized cameras at 60 Hz. It is 10 times larger than existing in-the-wild human motion video datasets and 3 times larger than existing MOCAP-ground-truthed datasets, and it is the first large-scale benchmark to provide synchronized multi-view recordings of expert athletic motion. The multi-view setup enables inexpensive, label-free evaluation of monocular-to-3D pose estimation algorithms. We describe a simple, standardized protocol that enables data collection without specialized equipment or expertise, along with fully automated video calibration and synchronization. Benchmarking state-of-the-art monocular-to-3D pose methods on CalTennis, we find that while 3D joint angle recovery is now quite accurate, all models struggle to estimate depth and foot contact consistently. We further propose two novel performance metrics, footwork and stability, as well as qualitatively study body shape inconsistency. These metrics expose previously underexplored failure modes and point to concrete opportunities for improvement in pose estimation and action analysis.
Chinese Translation
加州理工学院网球数据集（CalTennis）是一个用于评估野外单目到三维姿态估计的大规模视频基准。CalTennis包含来自40名运动员的超过1100万帧（51小时）网球训练和比赛视频，这些视频是通过2到6台同步摄像机以60 Hz的频率捕获的。它的规模是现有野外人类运动视频数据集的10倍，是现有运动捕捉（MOCAP）真实数据集的3倍，并且是第一个提供专家运动同步多视角录制的大规模基准。多视角设置使得对单目到三维姿态估计算法的评估变得经济且无需标签。我们描述了一种简单的标准化协议，使得在没有专业设备或专业知识的情况下进行数据收集成为可能，同时实现了完全自动化的视频校准和同步。在CalTennis上对最先进的单目到三维姿态方法进行基准测试时，我们发现虽然三维关节角度恢复现在相当准确，但所有模型在深度和脚接触的估计上都存在一致性问题。我们进一步提出了两个新的性能指标，步伐和稳定性，并对身体形状不一致性进行了定性研究。这些指标揭示了之前未被充分探讨的失败模式，并指出了在姿态估计和动作分析中改进的具体机会。
View on arXiv Download PDF AI Translation
cs.CV / 96 / 2606.20543

SSD: Spatially Speculative Decoding Accelerates Autoregressive Image Generation

SSD：空间推测解码加速自回归图像生成
Xiang, Shilong, Zhang, Zirui, Yu, Lijun, Mao, Chengzhi
Abstract
Autoregressive models excel in visual generation by treating images as 1D sequences of discrete tokens, mirroring language modeling. However, this flattening discards the intrinsic 2D spatial locality of visual signals, creating severe computational bottlenecks during inference. We introduce Spatially Speculative Decoding (SSD), a framework that aligns the predictive objective with the natural geometry of images. Rather than predicting only the immediate next token in a 1D sequence, our model simultaneously predicts the adjacent horizontal token and the token directly below it. By capitalizing on this 2D spatial correlation, spatially speculative decoding overcomes the memory wall in visual inference. Our approach accelerates autoregressive image generation by up to 13.3x while maintaining high fidelity on DPG-Bench and GenEval. Our results suggest that respecting the underlying geometry of vision unlocks massive computational efficiencies, paving the way for real-time, high-resolution autoregressive generative models.
Chinese Translation
自回归模型在视觉生成中表现出色，通过将图像视为离散标记的1D序列，类似于语言建模。然而，这种扁平化处理忽略了视觉信号的内在2D空间局部性，在推理过程中造成了严重的计算瓶颈。我们提出了空间推测解码（Spatially Speculative Decoding, SSD），这是一个将预测目标与图像的自然几何形状对齐的框架。我们的模型不仅预测1D序列中紧接着的下一个标记，还同时预测相邻的水平标记和其正下方的标记。通过利用这种2D空间相关性，空间推测解码克服了视觉推理中的内存壁垒。我们的方法在保持DPG-Bench和GenEval上的高保真度的同时，将自回归图像生成的速度提高了多达13.3倍。我们的结果表明，尊重视觉的基本几何形状可以释放巨大的计算效率，为实时高分辨率自回归生成模型铺平道路。
View on arXiv Download PDF AI Translation
cs.CV / 97 / 2606.20545

Current World Models Lack a Persistent State Core

当前的世界模型缺乏持久状态核心
Lu, Jinpeng, Zhu, Dexu, Shi, Haoyuan, Cai, Linghan, Tang, Guo, Chen, Yinda, Cao, Jie, Tang, Duyu, Zhang, Yi, Dai, Yong, Ju, Xiaozhu
Abstract
World models are increasingly regarded as a decisive step toward artificial general intelligence, yet modeling the physical world demands more than rendering convincing frames on demand: it requires an internal world state that keeps evolving over time, decoupled from observation, so that objects endure and events run to their conclusions whether or not a camera is watching, much as the moon holds to its orbit when no one is looking. This requirement is a blind spot of existing benchmarks, which reward surface properties such as fidelity, motion, and camera controllability while never asking whether a generated world keeps evolving once it is unobserved. We introduce \textbf{WRBench}, the first systematic diagnostic benchmark that treats camera motion as an intervention on observability and resolves evaluation into a human-calibrated chain that asks whether the camera executes the requested interaction, whether the scene stays continuous and identifiable while in view, and whether a returning target remains consistent with the event that was set in motion. Across 9{,}600 videos from 23 models spanning four control paradigms, one finding proves stubborn: current systems maintain the observed world as a tracking shot, resuming a returning target in the state at which it was abandoned rather than advancing the event while it went unseen. Because this failure recurs across control paradigms, model families, and increments of scale, robust world-state evolution does not follow from cleaner imagery, tighter control, richer geometric priors, or sheer parameter count We therefore argue that the stability of the physical state kernel and the consistency of worldlines under viewpoint intervention should become first-class objectives of world-model design, so that a world model captures how the world will unfold rather than how the next frame appears.
Chinese Translation
世界模型越来越被视为通用人工智能的重要一步，然而对物理世界的建模不仅仅需要按需渲染令人信服的画面：它还需要一个内部世界状态，该状态随着时间的推移不断演变，与观察解耦，以便对象能够持续存在，事件能够得以完成，无论是否有摄像机在观察，就像月球在无人注视时仍然保持其轨道一样。这一要求是现有基准的盲点，这些基准奖励诸如保真度、运动和摄像机可控性等表面特性，却从未询问生成的世界在未被观察时是否持续演变。我们引入了 extbf{WRBench}，这是第一个系统化的诊断基准，将摄像机运动视为对可观察性的干预，并将评估分解为一个经过人类校准的链条，询问摄像机是否执行了请求的交互，场景在视野内是否保持连续和可识别，以及返回目标是否与已启动的事件保持一致。在来自23个模型、跨越四种控制范式的9600个视频中，一个发现显得尤为顽固：当前系统将观察到的世界维持为跟踪镜头，在返回目标时恢复到被放弃时的状态，而不是在未被观察时推进事件。由于这一失败在不同的控制范式、模型家族和规模增量中反复出现，因此稳健的世界状态演变并不依赖于更清晰的图像、更紧密的控制、更丰富的几何先验或单纯的参数数量。因此，我们认为物理状态核心的稳定性和在视角干预下世界线的一致性应成为世界模型设计的首要目标，以便世界模型能够捕捉世界将如何展开，而不仅仅是下一帧的外观。
View on arXiv Download PDF AI Translation
cs.CV / 98 / 2606.20556

Thinking in Boxes: 3D Editing in Real Images Made Easy

盒子思维：简化真实图像中的三维编辑
Bhat, Pradhaan S, R, Naveen Chandra, Parihar, Rishubh, Vavilala, Vaibhav, Babu, R. Venkatesh, Forsyth, D. A., Bhattad, Anand
Abstract
Text and 2D-conditioning interfaces provide weak, ambiguous control over spatial transformations in image editing -- particularly under large object motions and camera changes. Prior work has used 3D primitives such as boxes, but only as loose conditioning signals indicating approximate object location rather than specifying the transformation. We instead use 3D boxes as structured specifications: the user provides the input and output boxes of the edit, casting editing as a well-posed geometry problem. This ``thinking in boxes'' interface, where each box face is color-coded to convey 3D orientation, gives precise control over translation, rotation, scaling, and viewpoint changes in real images while preserving scene and object identity, and recovering previously unseen object regions. To ground transformations in scene appearance, we introduce a depth-aligned planar floor as a global reference frame, shaded with depth-aware cues. Conditioned on this structure, an image generator produces consistent results under large transformations. Trained in two stages -- on synthetic multi-object scenes and a small set of real-world videos from Objectron -- the system generalizes to complex, in-the-wild real images. Our method operates directly on real photographs and substantially outperforms recent state-of-the-art methods on large 3D edits.
Chinese Translation
文本和二维条件接口在图像编辑中对空间变换提供了弱且模糊的控制，尤其是在大物体运动和相机变化的情况下。之前的研究使用了三维原语，如盒子，但仅作为指示近似物体位置的松散条件信号，而不是指定变换。我们则将三维盒子作为结构化规范：用户提供编辑的输入和输出盒子，将编辑视为一个良定义的几何问题。这个“盒子思维”接口中，每个盒子面都通过颜色编码来传达三维方向，能够精确控制真实图像中的平移、旋转、缩放和视点变化，同时保持场景和物体的身份，并恢复先前未见的物体区域。为了将变换与场景外观相结合，我们引入了一个深度对齐的平面地面作为全局参考框架，并用深度感知线索进行着色。在这个结构的条件下，图像生成器在大变换下产生一致的结果。该系统经过两个阶段的训练——在合成的多物体场景和来自 Objectron 的一小组真实视频上，能够推广到复杂的真实图像。我们的方法直接作用于真实照片，并在大规模三维编辑上显著优于最近的最先进方法。
View on arXiv Download PDF AI Translation
cs.CV / 99 / 2606.20559

UNIEGO: Proxies as Mediators for Unified Egocentric Video Representation Learning

UNIEGO：代理作为统一自我中心视频表示学习的媒介
Chi, Wenhao, Sinha, Arkaprava, Reilly, Dominick, Le, Hieu, Das, Srijan
Abstract
Egocentric video understanding is inherently limited by the narrow perspective of wearable cameras: a single viewpoint, a single modality, a single model cannot capture the full richness of human action. We argue that a truly expressive egocentric representation must subsume complementary knowledge across viewpoints, modalities, and foundation model representations, yet remain deployable from egocentric video alone. To this end, we introduce a hierarchical multi-teacher distillation framework that produces UNIEGO, a unified egocentric encoder trained with nine teachers spanning ego-exo viewpoints, RGB, depth, and skeleton modalities, and four foundation models. Rather than distilling directly from heterogeneous teachers whose incompatible architectures and feature geometries induce conflicting gradients, our framework interposes a layer of representation-specific Proxy models that translate diverse teacher knowledge into a homogeneous egocentric space. A second distillation stage, Selective Proxy Distillation (SPD), then adaptively selects, for each training sample, the subset of proxies that are both correct and confident, distilling exclusively from reliable supervision and suppressing erroneous signals. SPD is further stabilized by initializing UNIEGO as a learned convex combination of proxy parameters, placing the unified model in a well-conditioned region of the loss landscape before distillation begins. UNIEGO achieves state-of-the-art performance across three egocentric video understanding tasks - action recognition, video retrieval, and action segmentation on three challenging ego-exo benchmarks, outperforming naive multi-teacher distillation baselines and demonstrating that structured, proxy-mediated knowledge transfer yields richer and more discriminative egocentric representations.
Chinese Translation
自我中心视频理解本质上受到可穿戴摄像机狭窄视角的限制：单一视点、单一模态和单一模型无法捕捉人类动作的全部丰富性。我们认为，真正具有表现力的自我中心表示必须包含跨视点、跨模态和基础模型表示的互补知识，同时仍然能够仅依赖自我中心视频进行部署。为此，我们提出了一种分层多教师蒸馏框架，生成UNIEGO，一个统一的自我中心编码器，使用九个教师进行训练，涵盖自我-外部视点、RGB、深度和骨架模态，以及四个基础模型。我们的框架并不是直接从异构教师中进行蒸馏，因为它们不兼容的架构和特征几何会导致冲突的梯度，而是插入了一层表示特定的代理模型，将多样的教师知识转化为同质的自我中心空间。第二个蒸馏阶段，选择性代理蒸馏（Selective Proxy Distillation, SPD），然后为每个训练样本自适应选择既正确又自信的代理子集，仅从可靠的监督中进行蒸馏，并抑制错误信号。SPD通过将UNIEGO初始化为代理参数的学习凸组合进一步稳定，使统一模型在蒸馏开始之前处于损失景观的良好条件区域。UNIEGO在三个自我中心视频理解任务上实现了最先进的性能——动作识别、视频检索和动作分割，在三个具有挑战性的自我-外部基准上超越了简单的多教师蒸馏基线，并证明了结构化的代理介导知识转移能够产生更丰富和更具辨别力的自我中心表示。
View on arXiv Download PDF AI Translation
cs.CV / 100 / 2606.20561

TimeProVe: Propose, then Verify for Efficient Long Video Temporal Reasoning in Activities of Daily Living

TimeProVe：先提出，再验证以高效进行日常生活活动中的长视频时间推理
Sinha, Arkaprava, Reilly, Dominick, Krishnan, Siddharth, Le, Hieu, Das, Srijan
Abstract
Long Video Question Answering (LVQA) requires identifying sparse, query-relevant evidence within hours-long untrimmed videos. Existing approaches either process videos densely with large vision-language models (VLMs), incurring prohibitive computational cost, or rely on sparse caption-based reasoning, which often misses temporally localized and motion-centric evidence. We introduce TimeProVe, a cost-efficient hybrid framework for temporally grounded reasoning in long videos. TimeProVe first employs lightweight modules to generate action-grounded answer--evidence hypotheses and subsequently invokes an expensive VLM only for targeted verification. The core of our framework lies in the Action-based Candidate Evidence (ACE) module, which converts temporally localized actions into query-conditioned candidate answers and supporting evidence windows through lightweight LLM reasoning. We further introduce OpenTSUBench (OTB), an open-ended benchmark designed to evaluate temporally grounded reasoning in real-world Activities of Daily Living (ADL) scenarios. Experiments show that TimeProVe outperforms the strongest baseline on OTB by 7.3%, while reducing VLM calls by 75% and inference cost by 93%. Furthermore, without explicit temporal grounding training, TimeProVe achieves competitive performance on Charades-STA, and reaches state-of-the-art results when enhanced with grounding VLMs.
Chinese Translation
长视频问答（LVQA）需要在数小时未剪辑的视频中识别稀疏的、与查询相关的证据。现有的方法要么使用大型视觉-语言模型（VLMs）对视频进行密集处理，导致计算成本高昂，要么依赖稀疏的基于字幕的推理，这往往会错过时间上局部和以运动为中心的证据。我们提出了TimeProVe，这是一种成本高效的混合框架，用于在长视频中进行时间基础的推理。TimeProVe首先采用轻量级模块生成基于动作的答案-证据假设，随后仅针对目标验证调用昂贵的VLM。我们框架的核心在于基于动作的候选证据（ACE）模块，该模块通过轻量级的LLM推理将时间上局部的动作转换为查询条件的候选答案和支持证据窗口。我们进一步引入了OpenTSUBench（OTB），这是一个开放式基准，旨在评估现实世界日常生活活动（ADL）场景中的时间基础推理。实验表明，TimeProVe在OTB上比最强基线提高了7.3%，同时减少了75%的VLM调用和93%的推理成本。此外，在没有明确的时间基础训练的情况下，TimeProVe在Charades-STA上也取得了具有竞争力的表现，并在增强了基础VLM后达到了最先进的结果。
View on arXiv Download PDF AI Translation
cs.CV / 101 / 2606.20563

JanusMesh: Fast and Zero-Shot 3D Visual Illusion Generation via Cross-Space Denoising

JanusMesh：通过跨空间去噪实现快速且零-shot的3D视觉幻觉生成
Zhang, Siang-Ling, Cheng, Huai-Hsun, Yang, Tsung-Ju, Liu, Yu-Lun
Abstract
Creating 3D visual illusions, a single 3D mesh that reveals entirely different semantics from various viewing angles, is a fascinating but tough challenge. Existing optimization-based methods are slow and can produce oversaturated colors. In contrast, naive stitching approaches fail to produce geometrically coherent objects. This results in visible unnatural seams and semantic leaks. In this paper, we present a fast and training-free framework for generating text-driven 3D visual illusions. Our approach decouples the generation into two stages. First, we propose a cross-space dual-branch denoising process. This process dynamically decodes 3D latents into voxel space for CLIP-guided orientation alignment and Signed Distance Field (SDF) blending, which ensures seamless geometric fusion. Second, we introduce a view-conditioned texture synthesis module that projects and aggregates view-specific 2D diffusion priors onto the fused geometry. Extensive experiments demonstrate that our method generates highly realistic, dual-semantic 3D illusions in just 3-5 minutes. It significantly outperforms existing methods in geometric integrity, semantic recognizability, and efficiency. Project page: https://siang1105.github.io/JanusMesh.github.io/
Chinese Translation
创建3D视觉幻觉，即从不同视角展现完全不同语义的单一3D网格，是一项既迷人又艰巨的挑战。现有的基于优化的方法速度较慢，并且可能产生过饱和的颜色。相比之下，简单的拼接方法无法生成几何上连贯的物体，导致可见的不自然接缝和语义泄漏。本文提出了一种快速且无需训练的框架，用于生成基于文本驱动的3D视觉幻觉。我们的方法将生成过程分为两个阶段。首先，我们提出了一种跨空间双分支去噪过程。该过程动态地将3D潜在变量解码为体素空间，以实现CLIP引导的方向对齐和带符号距离场（Signed Distance Field, SDF）融合，从而确保无缝的几何融合。其次，我们引入了一个视图条件的纹理合成模块，该模块将视图特定的2D扩散先验投影并聚合到融合的几何体上。大量实验表明，我们的方法在仅需3-5分钟内生成高度真实的双语义3D幻觉，且在几何完整性、语义可识别性和效率方面显著优于现有方法。项目页面：https://siang1105.github.io/JanusMesh.github.io/
View on arXiv Download PDF AI Translation
人工智能 (Artificial Intelligence)
73
cs.AI / 1 / 2606.19464

Deontic Policies for Runtime Governance of Agentic AI Systems

用于代理人工智能系统运行时治理的义务政策
Joshi, Anupam, Finin, Tim, Joshi, Karuna Pande, Kagal, Lalana
Abstract
Autonomous agentic AI systems driven by Large Language Models (LLMs) introduce a new class of security, privacy, and compliance challenges: an agent that can invoke tools, manipulate data, install software, and coordinate with peer agents across organizational boundaries must be constrained not just by authentication and access control, but by the full structure of enterprise governance. This includes specifying what agents are permitted and prohibited from doing, what they areobliged to do after certain actions (e.g., notify the CISO), under what conditions a standing obligation may be waived, and which rules take precedence when policies conflict. This governance problem exceeds what current policy engines provide. Systems such as XACML, Rego, and Cedar address only the permit/prohibit subset of this governance structure. They do not provide obligation lifecycle management, meta-policy conflict resolution, dispensations that waive obligations in specific circumstances, and ontological reasoning over domain class hierarchies commonly found in applications such as healthcare, cybersecurity, or data privacy. We propose AgenticRei, which realizes key governance requirements such as obligations, dispensations, policy conflict resolutions, and reasoning over policies, as well as the basic permit/prohibit constraints. We use a deontic policy language built on the Rei framework, expressed as OWL (Web Ontology Language) and evaluated at runtime by a high-performance logic engine entirely outside the LLM. The same pipeline governs both tool invocations by the agent and agent-to-agent messages. We show through examples that deontic policies capture governance constraints around security and privacy that mostly cannot be expressed in current production engines. Our approach composes naturally with industry-standard frameworks like A2AS.
Chinese Translation
由大型语言模型（LLMs）驱动的自主代理人工智能系统引入了一类新的安全、隐私和合规挑战：一个能够调用工具、操纵数据、安装软件并与跨组织边界的同伴代理协调的代理，必须受到的不仅仅是身份验证和访问控制的约束，而是企业治理的完整结构。这包括规定代理被允许和禁止做什么，在某些行动后（例如，通知首席信息安全官）他们有义务做什么，在什么条件下可以放弃持续义务，以及当政策冲突时哪些规则优先。这个治理问题超出了当前政策引擎所能提供的范围。像 XACML、Rego 和 Cedar 这样的系统仅处理这一治理结构的允许/禁止子集。它们未能提供义务生命周期管理、元政策冲突解决、在特定情况下放弃义务的特许权，以及对在医疗保健、网络安全或数据隐私等应用中常见的领域类层次结构的本体推理。我们提出了 AgenticRei，它实现了关键的治理要求，如义务、特许权、政策冲突解决以及对政策的推理，以及基本的允许/禁止约束。我们使用基于 Rei 框架构建的义务政策语言，以 OWL（Web 本体语言）表示，并由一个完全独立于 LLM 的高性能逻辑引擎在运行时进行评估。同一管道同时管理代理的工具调用和代理间消息。我们通过示例展示了义务政策捕捉到的关于安全和隐私的治理约束，这些约束在当前的生产引擎中大多无法表达。我们的方法与行业标准框架如 A2AS 自然组合。
View on arXiv Download PDF AI Translation
cs.AI / 2 / 2606.19469

Measuring Curriculum Alignment across Topical Coverage, Competency, and Cognitive Depth: A Longitudinal Framework Applied to CS2013 and CS2023

测量课程对齐度：主题覆盖、能力和认知深度的纵向框架应用于CS2013和CS2023
Turaev, Sherzod, John, Mary, Aldabet, Saja, Awad, Mamoun, Zaki, Nazar, Shuaib, Khaled
Abstract
Undergraduate computer science is governed by international curricular guidelines revised about once a decade, yet programs lack a reliable, reproducible way to measure how completely they cover the current guidelines and how that coverage shifts when the guidelines are restructured. We address this with a human-in-the-loop pipeline that measures a program's coverage of an external body of knowledge, applied longitudinally to one accredited BSc in Computer Science against Computer Science Curricula 2013 (CS2013) and 2023 (CS2023). The pipeline represents the program and each guideline as structured corpora, generates candidate course-to-knowledge-unit matches by semantic retrieval, and confirms them through human judgment under an explicit coverage definition. Of seven benchmarked retrievers, a reciprocal-rank-fusion ensemble was strongest, and a reputed long-context model underperformed a small sentence model, so retriever choice must be measured. Both maps were validated by an independent second rater (Cohen's kappa 0.64 for CS2023, 0.69 for CS2013). The program covers 49.7% of CS2023 and 50.9% of CS2013 knowledge units, near-constant across a decade. Extending the same retrieve-then-confirm design to competency articulation and cognitive depth shows that the program articulates the competency for ~88% of covered units under each guideline, yet delivers it at the recommended depth for 76% of present units under CS2023 against 95% under CS2013, a gap reflecting the newer guideline's raised expectations, not the program. The longitudinal comparison separates persistent structural gaps (parallel and distributed computing, foundations of programming languages, systems fundamentals), uncovered against both guidelines and ABET, from differences that reflect the standard's evolution. The instrument is reusable and available from the authors on request.
Chinese Translation
本科计算机科学课程受到国际课程指南的约束，这些指南大约每十年修订一次，但各个项目缺乏一种可靠、可重复的方法来衡量它们对当前指南的覆盖程度，以及当指南重组时这种覆盖的变化。我们通过一个人机协作的流程来解决这一问题，该流程测量一个项目对外部知识体系的覆盖情况，并纵向应用于一个经过认证的计算机科学学士学位项目，针对计算机科学课程2013（CS2013）和2023（CS2023）。该流程将项目和每个指南表示为结构化语料库，通过语义检索生成候选课程与知识单元的匹配，并通过人类判断在明确的覆盖定义下进行确认。在七个基准检索器中，互惠排名融合集成模型表现最佳，而一个声誉良好的长上下文模型的表现不及一个小句子模型，因此检索器的选择必须进行测量。这两个映射都经过独立的第二评审者验证（CS2023的Cohen's kappa为0.64，CS2013为0.69）。该项目覆盖了CS2023知识单元的49.7%和CS2013知识单元的50.9%，在十年间几乎保持不变。将相同的检索-确认设计扩展到能力表达和认知深度显示，该项目在每个指南下对约88%的覆盖单元进行了能力表达，但在CS2023下以推荐深度交付的单元为76%，而在CS2013下为95%，这一差距反映了新指南提高的期望，而非项目本身。纵向比较将持久的结构性差距（并行和分布式计算、编程语言基础、系统基础）与反映标准演变的差异分开。这一工具是可重复使用的，作者可应请求提供。
View on arXiv Download PDF AI Translation
cs.AI / 3 / 2606.19475

Diffusion Language Models: An Experimental Analysis

扩散语言模型：实验分析
Bertolani, Thomas, Bucciarelli, Davide, Zini, Leonardo, Cornia, Marcella, Baraldi, Lorenzo
Abstract
Large Language Models (LLMs) have revolutionized language modeling through autoregressive generation, enabling strong performance across a wide range of tasks. Recently, Diffusion Language Models (DLMs) have emerged as an alternative paradigm that generates text through iterative denoising rather than next-token prediction, allowing parallel refinement of entire sequences. While numerous diffusion-based architectures have been proposed, differences in evaluation protocols, datasets, inference budgets, and generation hyperparameters make it difficult to compare their capabilities and understand the trade-offs they offer. In this work, we present a systematic experimental analysis of modern DLMs. Specifically, we evaluate eight state-of-the-art DLMs across eight benchmarks spanning reasoning, coding, translation, knowledge, and structured problem solving, while explicitly considering both generation quality and computational efficiency. Beyond downstream evaluation, we analyze the impact of key inference-time factors, including denoising steps, context length, block size, and parallel unmasking strategies, and complement large-scale experiments with controlled comparisons of smaller models trained under identical conditions. Our analysis highlights the strengths and limitations of diffusion-based language modeling across different tasks, architectures, and inference budgets. We show that the behavior of DLMs is strongly influenced by generation-time design choices, leading to distinct trade-offs between performance and computational efficiency. Overall, our study provides practical insights into the capabilities and deployment characteristics of contemporary DLMs.
Chinese Translation
大型语言模型（LLMs）通过自回归生成彻底改变了语言建模，使其在广泛任务中表现出色。最近，扩散语言模型（DLMs）作为一种替代范式出现，通过迭代去噪而非下一个标记预测生成文本，从而实现整个序列的并行优化。尽管已经提出了许多基于扩散的架构，但评估协议、数据集、推理预算和生成超参数的差异使得比较它们的能力和理解它们所提供的权衡变得困难。在本研究中，我们对现代DLMs进行了系统的实验分析。具体而言，我们在八个基准测试中评估了八种最先进的DLMs，这些基准涵盖了推理、编码、翻译、知识和结构化问题解决，同时明确考虑生成质量和计算效率。除了下游评估外，我们还分析了关键推理时间因素的影响，包括去噪步骤、上下文长度、块大小和并行去掩蔽策略，并通过在相同条件下训练的小型模型的受控比较补充大规模实验。我们的分析突出了基于扩散的语言建模在不同任务、架构和推理预算下的优势和局限性。我们展示了DLMs的行为受到生成时间设计选择的强烈影响，导致性能和计算效率之间存在明显的权衡。总体而言，我们的研究为当代DLMs的能力和部署特性提供了实用的见解。
View on arXiv Download PDF AI Translation
cs.AI / 4 / 2606.19494

Hidden Anchors in Multi-Agent LLM Deliberation

多智能体大语言模型的隐性锚点
Pokharel, Apurba, Dantu, Ram
Abstract
Multi-agent LLM deliberation, where agents exchange and revise answers over several rounds, is increasingly used to improve reasoning and accuracy, yet how and why it works is rarely modelled. Such deliberation mirrors how humans reach decisions. As social animals we are pulled both by the group, the herd effect that classical opinion-dynamics models such as DeGroot and Friedkin--Johnsen capture, and by our own internal belief, which they do not. We model multi-agent deliberation as a closed-loop dynamical system in which each agent carries a hidden internal belief, its anchor, that continually pulls its opinion regardless of its neighbours. We show this anchor can be recovered from the deliberation alone, and that it explains a behaviour classical consensus rules forbid: an agent's confidence in the correct answer can climb past where any agent started, escaping the space (convexhull) formed by the initial beliefs. Checking whether the recovered anchor also predicts held-out runs (generalizes) gives a simple test for when a model is truly driven bysuch an anchor. Across three open-weight model families this is a spectrum, not all-or-nothing. All anchors' influence are about equally strongly, but they differ in where the anchor sits, and only when it sits far from the initial opinions does deliberation escape the hull and need the full closed-loop model.
Chinese Translation
多智能体大语言模型（LLM）辩论中，智能体在多个回合中交换和修正答案，这种方法越来越多地被用于提高推理能力和准确性，但其工作原理和原因却很少被建模。这种辩论反映了人类如何做出决策。作为社会性动物，我们受到群体的影响，即经典意见动态模型（如 DeGroot 和 Friedkin--Johnsen 所捕捉的羊群效应），同时也受到自身内在信念的影响，而后者并未被这些模型考虑。我们将多智能体辩论建模为一个闭环动态系统，其中每个智能体都携带一个隐性内在信念，即其锚点，这个锚点不断地影响其观点，而不受邻居的影响。我们展示了这个锚点可以仅通过辩论过程恢复，并且它解释了一种经典共识规则所禁止的行为：一个智能体对正确答案的信心可以超越任何智能体的初始信心，从而逃离由初始信念形成的空间（凸包）。检查恢复的锚点是否也能预测未见运行（泛化）提供了一个简单的测试，以判断模型是否真正受到这种锚点的驱动。在三个开放权重模型家族中，这是一种光谱，而非全有或全无。所有锚点的影响力大致相同，但它们在锚点位置上有所不同，只有当锚点远离初始观点时，辩论才会逃离凸包，并需要完整的闭环模型。
View on arXiv Download PDF AI Translation
cs.AI / 5 / 2606.19501

DeXposure-Claw: An Agentic System for DeFi Risk Supervision

DeXposure-Claw：一种用于去中心化金融风险监督的代理系统
Shu, Aijie, Chen, Bowei, Wu, Wenbin, Chen, Cathy Yi-Hsuan, He, Fengxiang
Abstract
Decentralized finance exposes supervisors to fast-moving, networked credit risks. General-purpose LLM agents fit this setting poorly: they over-read weak evidence and recommend high-stakes interventions, while existing evaluations offer no regulator-aligned way to measure the resulting false alarms. We introduce DeXposure-Claw, a forecast-grounded agentic supervision system that routes LLM decisions through structured evidence: (1) DeXposure-FM, a graph time-series foundation model, forecasts future exposure networks; (2) deterministic monitors and stress scenarios then turn those forecasts into typed alerts, attribution signals, and scenario evidence; and (3) data-health and confidence gates constrain escalation before DeXposure-Claw emits auditable supervisory tickets with rationales. We further develop DeXposure-Bench, a six-axis evaluation harness, whose decision axis scores tickets against a regulator-aligned absolute-loss ground truth and an explicit false-intervention rate. Experiments on five years of weekly real data fully support our system. Code is at https://github.com/EVIEHub/DeXposure-Claw.
Chinese Translation
去中心化金融使监督者面临快速变化的网络信用风险。通用大型语言模型（LLM）代理在这一环境中表现不佳：它们过度解读微弱证据并建议高风险干预，而现有评估没有提供与监管者对齐的方式来衡量由此产生的误报。我们提出了DeXposure-Claw，这是一种基于预测的代理监督系统，通过结构化证据引导LLM决策：（1）DeXposure-FM，一个图形时间序列基础模型，预测未来的风险暴露网络；（2）确定性监控器和压力情景将这些预测转化为类型警报、归因信号和情景证据；（3）数据健康和置信门限制升级，确保DeXposure-Claw在发出具有合理解释的可审计监督票据之前。我们进一步开发了DeXposure-Bench，一个六轴评估工具，其决策轴根据与监管者对齐的绝对损失真实值和明确的误干预率对票据进行评分。对五年每周真实数据的实验充分支持了我们的系统。代码可在 https://github.com/EVIEHub/DeXposure-Claw 获取。
View on arXiv Download PDF AI Translation
cs.AI / 6 / 2606.19509

LLM Doesn't Know What It Doesn't Know: Detecting Epistemic Blind Spots via Cross-Model Attribution Divergence on Clinical Tabular Data

大型语言模型不知道它不知道什么：通过跨模型归因差异检测临床表格数据中的认知盲点
Dasula, Akshat, Desikan, Prasanna, Srivastava, Jaideep
Abstract
Large language models (LLMs) are increasingly applied to structured clinical data, yet whether they can recognize the limits of their own knowledge on such tasks remains unexplored. We study this question through the lens of cross-model attribution divergence with the goal of reducing epistemic uncertainty for structured tasks, comparing Qwen 2.5 7B and XGBoost on a prediction task via attribution divergence analysis. We report four findings. First, LLM verbalized confidence is epistemically vacuous, it outputs a near-constant (0.856-0.937) regardless of whether accuracy is 49% or 75.3%, tracking prompt format rather than prediction quality. Second, the LLM exhibits an inverse difficulty effect: accuracy drops to 64.8% when XGBoost is 99% correct, but matches XGBoost (73.8% vs. 73.1%) when it is moderately uncertain. Third, few-shot examples and SHAP-derived feature evidence are orthogonal, super-additive interventions: they reduce the Attribution Disagreement Score (ADS) from 1.54 to 0.38 and improve accuracy from 49% to 75.3% without training. Fourth, a cross-model calibrator that determined LLM reliability using attribution divergence signals reduces expected calibration error from 0.254 to 0.080, replacing uninformative verbalized confidence with patient-specific reliability estimates, without accessing model internals or requiring repeated inference. We frame these findings as a cold start problem for LLMs on structured data and outline a path toward genuine epistemic self-awareness.
Chinese Translation
大型语言模型（LLMs）越来越多地应用于结构化临床数据，但它们是否能够识别自身知识的局限性仍未被探索。我们通过跨模型归因差异的视角研究这个问题，旨在减少结构化任务的认知不确定性，比较 Qwen 2.5 7B 和 XGBoost 在归因差异分析下的预测任务。我们报告了四个发现。首先，LLM 表达的信心在认知上是空洞的，无论准确率是 49% 还是 75.3%，其输出几乎保持不变（0.856-0.937），跟踪的是提示格式而非预测质量。第二，LLM 显示出逆向难度效应：当 XGBoost 的准确率为 99% 时，LLM 的准确率降至 64.8%，但在 XGBoost 中等不确定性时则与其相匹配（73.8% 对 73.1%）。第三，少量示例和基于 SHAP 的特征证据是正交的、超加性的干预：它们将归因分歧分数（ADS）从 1.54 降低到 0.38，并在不进行训练的情况下将准确率从 49% 提高到 75.3%。第四，利用归因差异信号确定 LLM 可靠性的跨模型校准器将预期校准误差从 0.254 降低到 0.080，替换了无信息的表达信心，提供了特定于患者的可靠性估计，而无需访问模型内部或进行重复推理。我们将这些发现框架化为 LLM 在结构化数据上的冷启动问题，并勾勒出实现真正认知自我意识的路径。
View on arXiv Download PDF AI Translation
cs.AI / 7 / 2606.19522

REVEAL++: Differentiable Phenotypic Grouping for Vision-Language Retinal Modeling of Alzheimer's Disease Risk

REVEAL++：用于阿尔茨海默病风险的视觉-语言视网膜建模的可微表型分组
Meidinger, Ethan Elio, Leem, Seowung, Zhao, Zeyun, Fang, Ruogu
Abstract
The retina offers a noninvasive window into neurodegenerative disease, capturing subtle structural patterns associated with a risk of future cognitive decline. Vision-language alignment frameworks such as REVEAL have shown that pairing retinal fundus images with structured clinical risk narratives improves early prediction of Alzheimer's disease (AD). A key design choice in these approaches is the use of phenotypic grouping, where individuals with similar risk profiles are treated as multi-positive pairs during contrastive learning. However, existing methods operationalize phenotypic similarity as a discrete construct, relying on hard group assignments that impose rigid supervision and decouple group formation from representation learning. We propose a continuous formulation of phenotypic structure within contrastive learning. Rather than assigning samples to fixed clusters, we model inter-subject similarity as a differentiable weighting function derived from intra-modality embedding similarities in both retinal images and risk profiles. These weights define soft multi-positive relationships through a continuous aggregation operator, enabling graded supervision that reflects the spectrum nature of disease risk. We further introduce a soft-target contrastive objective that jointly learns cross-modal alignment and phenotypic structure in an end-to-end manner. Evaluated on UK Biobank retinal imaging data for incident AD prediction, the proposed framework consistently outperforms discrete group-based contrastive learning and standard vision-language baselines. By treating phenotypic similarity as a learnable, continuous signal rather than a fixed grouping rule, our approach provides a principled and robust foundation for population-scale neurodegenerative risk modeling from multi-modal retinal and clinical data.
Chinese Translation
视网膜提供了一个非侵入性的窗口，可以观察神经退行性疾病，捕捉与未来认知衰退风险相关的微妙结构模式。诸如REVEAL的视觉-语言对齐框架表明，将视网膜眼底图像与结构化临床风险叙述配对，可以改善阿尔茨海默病（AD）的早期预测。这些方法中的一个关键设计选择是使用表型分组，将具有相似风险特征的个体视为对比学习中的多正样本对。然而，现有方法将表型相似性操作化为离散构造，依赖于硬性组分配，这种方式施加了严格的监督，并将组形成与表征学习解耦。我们提出了一种在对比学习中连续表型结构的公式化。我们并不将样本分配到固定的聚类中，而是将个体间的相似性建模为一个可微的加权函数，该函数源自视网膜图像和风险特征中的模态内嵌入相似性。这些权重通过连续聚合算子定义软多正关系，使得监督呈现出反映疾病风险谱系的渐进性。我们进一步引入了一种软目标对比目标，该目标以端到端的方式共同学习跨模态对齐和表型结构。在对UK Biobank视网膜成像数据进行AD发生预测的评估中，所提出的框架始终优于基于离散组的对比学习和标准视觉-语言基线。通过将表型相似性视为可学习的连续信号而非固定的分组规则，我们的方法为基于多模态视网膜和临床数据的人口规模神经退行性风险建模提供了一个原则性和稳健的基础。
View on arXiv Download PDF AI Translation
cs.AI / 8 / 2606.19527

Emergent Alignment

新兴对齐
Kolář, Martin
Abstract
Can Large Language Models (LLMs) discern when their own outputs are misaligned with human ethics? And can they self-correct? We endow an LLM with a conscience step that reviews its own reasoning and outputs, and we extend the training loss with an alignment component using Direct Preference Optimization (DPO) to steer the model away from non-ethical outputs. The result is an online technique to align models in a wide range of applications: training, fine-tuning, adversarial prompting, and zero-shot learning. It does not require a weaker or stronger judge, relying instead on a frozen copy of itself. In previous work, the Emergent Misalignment scenario showed a range of emergent unethical behaviors from fine-tuning the model to hack code. Instead, we empirically show how to achieve Emergent Alignment: a single high-level introspective question steers training toward an ethical model under the same code hacking scenario.
Chinese Translation
大型语言模型（LLMs）能否辨别其输出何时与人类伦理不一致？它们能否自我纠正？我们为LLM赋予了一种良知步骤，审视其自身的推理和输出，并通过使用直接偏好优化（Direct Preference Optimization, DPO）扩展训练损失，加入对齐组件，以引导模型远离不符合伦理的输出。结果是一种在线技术，可在广泛的应用中对模型进行对齐：训练、微调、对抗性提示和零样本学习。该方法不需要较弱或较强的评判者，而是依赖于自身的一个冻结副本。在之前的研究中，新兴不对齐场景展示了从微调模型到破解代码的一系列新兴不道德行为。相反，我们通过实证研究展示了如何实现新兴对齐：一个高层次的内省问题在相同的代码破解场景下引导训练朝向一个符合伦理的模型。
View on arXiv Download PDF AI Translation
cs.AI / 9 / 2606.19538

ITNet: A Learnable Integral Transform That Subsumes Convolution, Attention, and Recurrence

ITNet：一种可学习的积分变换，涵盖卷积、注意力和递归
Dhor, Ashim, Mondal, Rasel, Chen, Pin Yu
Abstract
Convolutional networks, recurrent networks, and transformers each encode different inductive biases -- locality, sequential memory, and content-dependent pairwise interaction -- and have remained mathematically distinct since their inception. We show that this fragmentation reflects not a fundamental diversity in how signals should be processed, but rather incomplete views of a single underlying mathematical object: a learnable integral transform. We introduce the Integral Transform Network (ITNet), a unified architecture built around a learnable kernel that depends jointly on positions and features. This kernel is implemented as a small neural network, specifically an MLP, that models pairwise interactions, enabling the model to adapt its behavior from data. We show that convolution, self-attention (including multi-head), and autoregressive recurrence (including LSTM, GRU, S4, and Mamba) arise as special cases under appropriate parameterizations, and that ITNet is a universal approximator of continuous operators. To make this practical, we develop tiled kernel fusion, importance-weighted Monte Carlo integration, and learned low-rank factorization, enabling efficient and scalable computation. A single ITNet architecture with a shared operator and lightweight modality-specific encoders matches or exceeds specialized baselines on ImageNet-1K , GLUE, ModelNet40, VQA\,v2 and NLVR2. The results demonstrate that a single learned interaction mechanism can recover the behavior of all three architectural families from data.
Chinese Translation
卷积网络、递归网络和变换器各自编码了不同的归纳偏差——局部性、序列记忆和依赖内容的成对交互——自其诞生以来在数学上保持了明显的区别。我们表明，这种碎片化反映的并不是信号处理方式的根本多样性，而是对一个单一基础数学对象的片面理解：一种可学习的积分变换。我们引入了积分变换网络（ITNet），这是一种围绕可学习核构建的统一架构，该核共同依赖于位置和特征。这个核被实现为一个小型神经网络，具体来说是一个多层感知机（MLP），用于建模成对交互，使模型能够从数据中调整其行为。我们展示了卷积、自注意力（包括多头注意力）和自回归递归（包括LSTM、GRU、S4和Mamba）在适当参数化下作为特例出现，并且ITNet是连续算子的通用逼近器。为了使其具有实用性，我们开发了平铺核融合、重要性加权蒙特卡洛积分和学习的低秩分解，从而实现高效和可扩展的计算。一个共享算子和轻量级特定模态编码器的单一ITNet架构在ImageNet-1K、GLUE、ModelNet40、VQA v2和NLVR2上与专门基线相匹配或超越。结果表明，单一的学习交互机制能够从数据中恢复所有三种架构家族的行为。
View on arXiv Download PDF AI Translation
cs.AI / 10 / 2606.19559

Uncertainty Decomposition for Clarification Seeking in LLM Agents

大型语言模型代理中的不确定性分解与澄清寻求
Matsnev, Gregory
Abstract
Recent position papers argue that the classical aleatoric/epistemic uncertainty framework is insufficient for interactive large language model (LLM) agents and call for underspecification-aware, decomposed, and communicable uncertainty representations that can unlock new agent capabilities such as proactive clarification seeking and shared mental-model building. Practical deployment constraints -- black-box APIs, interactive latency budgets, and the absence of labeled trajectories -- rule out logprob-based, multi-sampling, and training-based methods, leaving prompt-based estimation as the most viable family for surfacing such signals at deployment time. We answer this call with a simple prompt-based decomposition that separates action confidence from request uncertainty (u), enabling the agent to ask for clarification when the task specification is ambiguous. To evaluate it, we introduce two clarification-augmented benchmarks (WebShop-Clarification and ALFWorld-Clarification) in which 50% of tasks are deliberately underspecified, and systematically compare the proposed decomposition against ReAct+UE and Uncertainty-Aware Memory (UAM) across five LLM backbones (GPT-5.1, DeepSeek-v3.2-exp, GLM-4.7, Qwen3.5-35B, GPT-OSS-120B) on these variants together with the standard WebShop, ALFWorld, and REAL benchmarks for fault detection. Averaged across the five backbones, the proposed decomposition improves clarification F1 on ALFWorld-Clarification by 73% over ReAct+UE and by 36% over UAM, and leads clarification F1 on every backbone on WebShop-Clarification and on four of five backbones on ALFWorld-Clarification, indicating that the gains generalize beyond a single LLM.
Chinese Translation
近期的立场论文指出，传统的偶然性/认识性不确定性框架不足以满足交互式大型语言模型（LLM）代理的需求，并呼吁需要关注欠规范化的、不确定性分解的、可传达的不确定性表示，这些表示能够解锁新的代理能力，如主动澄清寻求和共享心理模型的构建。实际部署的限制条件——黑箱API、交互延迟预算以及缺乏标记轨迹——排除了基于对数概率的、多重采样的和基于训练的方法，使得基于提示的估计成为在部署时揭示此类信号的最可行方法。我们通过一种简单的基于提示的分解方法来响应这一呼吁，该方法将行动信心与请求不确定性（u）分开，使代理能够在任务规范模糊时请求澄清。为了评估该方法，我们引入了两个增强澄清的基准（WebShop-Clarification 和 ALFWorld-Clarification），其中50%的任务故意设计为欠规范化，并系统地将所提出的分解与 ReAct+UE 和不确定性感知内存（UAM）在这五个 LLM 主干（GPT-5.1、DeepSeek-v3.2-exp、GLM-4.7、Qwen3.5-35B、GPT-OSS-120B）上进行比较，同时还包括标准的 WebShop、ALFWorld 和 REAL 基准进行故障检测。在五个主干上平均来看，所提出的分解在 ALFWorld-Clarification 上的澄清 F1 分数比 ReAct+UE 提高了 73%，比 UAM 提高了 36%，并在 WebShop-Clarification 上的每个主干和 ALFWorld-Clarification 上五个主干中的四个上都领先于澄清 F1 分数，表明这些增益超出了单一 LLM 的范围。
View on arXiv Download PDF AI Translation
cs.AI / 11 / 2606.19588

Analyzing the Narration Gap in LLM-Solver Loops

分析 LLM-求解器循环中的叙述差距
Huang, Zunchen, Deng, Songgaojun
Abstract
Formal tools such as SAT and SMT solvers are increasingly embedded in language model reasoning pipelines when a safety or security critical question can be formulated in logic. Unlike chain of thought whose steps are sampled from the model distribution without formal guarantee, a solver produces a sound and independently verifiable answer. However, the soundness guarantee can be lost in the interaction between the solver and the model. The hybrid pipeline has three components: formalizing the question, deciding it, and narrating the result. Prior work has studied the formalization and decision, but not narration, which is the step that turns a formal tool's output into the user answer. To fill the narration gap, we first model the LLM-solver loop as a verified decision procedure. We further evaluate five open-sourced models under prompt injection, and we find certificate gating makes the solver verdict sound, while an adversary can invert a verified conclusion across phrasings and channels. We study the mitigation through hardened prompt that reduces injection significantly but cannot eliminate it and still suffers under adaptive attack. Combining the formal analysis and empirical studies, we show in the LLM-solver loop, robustness does not reach to the answer that the user finally reads.
Chinese Translation
随着安全或安全关键问题能够用逻辑形式化，SAT 和 SMT 求解器等形式工具越来越多地嵌入到语言模型推理管道中。与思维链（chain of thought）不同，后者的步骤是从模型分布中采样而没有正式保证，求解器则产生一个声学且可独立验证的答案。然而，在求解器与模型之间的交互中，声学保证可能会丧失。混合管道有三个组成部分：形式化问题、决策和叙述结果。之前的研究关注了形式化和决策，但没有研究叙述，这是将形式工具的输出转化为用户答案的步骤。为了填补叙述差距，我们首先将 LLM-求解器循环建模为一个经过验证的决策过程。我们进一步评估了五个开源模型在提示注入下的表现，发现证书门控（certificate gating）使求解器的裁决具有声学性，而对手可以在不同的措辞和渠道中反转经过验证的结论。我们研究了通过强化提示（hardened prompt）来减轻注入风险的方法，该方法显著减少了注入，但无法消除，且在自适应攻击下仍然受到影响。结合形式分析和实证研究，我们表明在 LLM-求解器循环中，鲁棒性并未延伸到用户最终阅读的答案。
View on arXiv Download PDF AI Translation
cs.AI / 12 / 2606.19602

Configurable Clinical Information Extraction with Agentic RAG: What Works, What Breaks, and Why

可配置的临床信息提取与代理式 RAG：有效因素、失效原因及其背后的原因
Çinar-Koraş, Osman Alperen, Bauer, Marie, Khattab, Sameh, Engelke, Merlin, Kim, Moon, Settelmeier, Stephan, Sugawara, Shigeyasu, Freisleben, Fabian, Nensa, Felix, Kleesiek, Jens
Abstract
Patient contexts span hundreds of heterogeneous documents and thousands of structured data points, yet the document-level metadata that AI systems need for retrieval and triage is absent or incomplete. Standard retrieval-augmented generation fails on this data, mishandling temporal reasoning, cross-document dependencies, and missing metadata. We deploy ACIE (Agentic Clinical Information Extraction) at University Medicine Essen: an on-premise agentic RAG pipeline that reasons over complete patient contexts and grounds every answer in source passages for clinician verification. We quantify the metadata gap, trace the architectural decisions it shaped, and evaluate extraction alongside an independent retrospective lymphoma registry study, in which nuclear-medicine physicians verify every extracted value against its cited sources. Across 7,326 judgments, clinicians accepted 96.5\% of extractions, with per-type acceptance ranging from 80\% to 99\%.
Chinese Translation
患者背景跨越数百份异构文档和数千个结构化数据点，但 AI 系统所需的文档级元数据在检索和分诊中往往缺失或不完整。标准的检索增强生成在处理这些数据时失败，错误处理时间推理、跨文档依赖关系以及缺失的元数据。我们在埃森大学医学中心部署了 ACIE（代理式临床信息提取）：一个本地的代理式 RAG 管道，能够对完整的患者背景进行推理，并为临床医生验证提供每个答案的来源段落。我们量化了元数据缺口，追踪了其影响的架构决策，并在一项独立的回顾性淋巴瘤登记研究中评估提取结果，在该研究中，核医学医生对每个提取值与其引用来源进行验证。在 7,326 次判断中，临床医生接受了 96.5\% 的提取结果，各类型的接受率从 80\% 到 99\% 不等。
View on arXiv Download PDF AI Translation
cs.AI / 13 / 2606.19607

Which Pairs to Compare for LLM Post-Training?

在大语言模型后训练中应比较哪些对？
Han, Jiangze, Goyal, Vineet, Ma, Will
Abstract
Preference-based post-training has become a central paradigm for aligning language models. A common data-collection strategy is to generate a small set of completions for each prompt and label the resulting comparison pairs. However, human preference labels are often much more expensive than generating additional completions, suggesting a different use of the same labeling budget: generate a larger pool of completions, but label only the most informative comparison pairs. This paper studies which pairs should be compared in preference-based post-training. We formulate comparison curation as a sampling-design problem and evaluate designs by the quality of the final policy under the preference-based post-training objective. We instantiate this framework for Direct Preference Optimization (DPO), analyzing how the choice of labeled pairs propagates through DPO training to downstream policy performance. Our main results provide matching upper and lower bounds on the post-training optimality gap of the DPO-trained policy. The bounds show that comparison selection affects downstream performance through a single design-dependent information matrix, which links label allocation to parameter estimation error and policy suboptimality. This yields an explicit optimization criterion for budgeted comparison curation and motivates practical sampling designs for selecting informative pairs from large generated completion pools. Experiments on synthetic settings and language-model post-training benchmarks show that the proposed designs consistently improve sample efficiency over common comparison-selection heuristics.
Chinese Translation
基于偏好的后训练已成为对齐语言模型的核心范式。一种常见的数据收集策略是为每个提示生成一小组完成，并标记由此产生的比较对。然而，人工偏好标签的成本往往远高于生成额外完成的成本，这提示我们可以采用不同的标签预算使用方式：生成更大规模的完成池，但只标记最具信息量的比较对。本文研究在基于偏好的后训练中应比较哪些对。我们将比较策划形式化为一个采样设计问题，并通过基于偏好的后训练目标下最终策略的质量来评估设计。我们为直接偏好优化（Direct Preference Optimization, DPO）实例化该框架，分析标记对的选择如何在DPO训练中传播到下游策略性能。我们的主要结果提供了DPO训练策略的后训练最优性差距的匹配上界和下界。这些界限表明，比较选择通过一个单一的设计依赖信息矩阵影响下游性能，该矩阵将标签分配与参数估计误差和策略次优性联系起来。这为预算比较策划提供了明确的优化标准，并激励了从大型生成完成池中选择信息对的实际采样设计。对合成设置和语言模型后训练基准的实验表明，所提出的设计在样本效率上始终优于常见的比较选择启发式方法。
View on arXiv Download PDF AI Translation
cs.AI / 14 / 2606.19626

Toten: Knowledge-Based Ontological Tokenization Of Physical Quantities And Technical Notation In Brazilian Portuguese

Toten：基于知识的巴西葡萄牙语物理量和技术符号的本体标记化
Sousa, Antonio de Sousa Leitão Filho; Allan Kardec Duailibe Barros Filho; Fabrício Saul Lima; Selby Mykael Lima dos Santos; Rejani Bandeira Vieira
Abstract
Byte-Pair Encoding tokenization is statistically efficient for vocabulary compression, but semantically blind to structured technical entities, fragmenting physical quantities, numbers, units, and symbolic expressions into lexically arbitrary subwords. We present TOTEN, a knowledge-based ontological tokenization framework that replaces statistical derivation with declarative classification grounded in a formal ontology of engineering entities (OEE). We formalize TOTEN as the triple : the ontology gathers types, structural principles, composition relations, and preservable invariants; the classification function maps raw text into typed regions; and the instantiator family yields a self-descriptive structured representation. Robustness derives from deterministic coupling with three external oracles: Pint (dimensional), Unicode Character Database (typographic), and RSLP (Portuguese morphology). Intrinsic evaluation covers four properties verifiable by construction -- ontological atomicity, dimensional equivalence, typographic robustness, and numerical reconstruction -- over an internal, physically validated benchmark (EngQuant, N=800) and four Brazilian Portuguese external corpora (N=1771 eligible cases). We also report detection recall, distinguishing coverage from conditional atomicity. Against eight state-of-the-art baselines, TOTEN achieves unit ontological atomicity in all contrasts and numerical reconstruction of 0.775-0.904 on external corpora, vs. 0.627-0.703 for the best baseline (Quantulum3); on EngQuant, 0.780 vs. 0.340. Differences are statistically significant (McNemar with Holm correction). Spearman correlation between internal and external rankings confirms concurrent validity of the control benchmark. Dimensional equivalence shows statistical parity with Pint, the oracle from which the system inherits dimensional authority.
Chinese Translation
字节对编码（Byte-Pair Encoding）标记化在词汇压缩方面具有统计效率，但对结构化技术实体缺乏语义理解，将物理量、数字、单位和符号表达分割为词汇上任意的子词。我们提出了TOTEN，一个基于知识的本体标记化框架，它用基于正式工程实体本体（OEE）的声明性分类替代统计推导。我们将TOTEN形式化为三元组：本体收集类型、结构原则、组成关系和可保持的不变性；分类函数将原始文本映射到类型区域；而实例化器家族则生成自描述的结构化表示。其鲁棒性源于与三个外部神谕的确定性耦合：Pint（维度）、Unicode字符数据库（排版）和RSLP（葡萄牙语形态学）。内在评估涵盖了四个可通过构造验证的属性——本体原子性、维度等价性、排版鲁棒性和数字重构——在一个内部的、经过物理验证的基准（EngQuant，N=800）和四个巴西葡萄牙语外部语料库（N=1771个合格案例）上进行。我们还报告了检测召回率，区分覆盖率与条件原子性。在八个最先进的基线对比中，TOTEN在所有对比中实现了单位本体原子性，并在外部语料库上实现了0.775-0.904的数字重构，而最佳基线（Quantulum3）为0.627-0.703；在EngQuant上为0.780对0.340。差异具有统计显著性（McNemar检验，采用Holm校正）。内部与外部排名之间的斯皮尔曼相关性确认了控制基准的同时有效性。维度等价性与Pint显示统计平价，Pint是系统继承维度权威的神谕。
View on arXiv Download PDF AI Translation
cs.AI / 15 / 2606.19630

AI4SE and SE4AI Exploration: A Decade Looking Back and Forward

AI4SE与SE4AI探索：回顾与展望十年
Bank, H. Sinan, Herber, Daniel R., Bradley, Thomas
Abstract
The March 2020 INCOSE INSIGHT special issue on AI and Systems Engineering (SE) became the most downloaded issue in the publication's history and launched a research community that now draws over 250 registrants to its annual workshop. In this article, we trace the progress in AI and SE across three phases (labeled here foundational, applied, and LLM inflection) based on the authors' reading of the field's core papers, and describe our opinions of where the community has converged and where critical gaps remain. Separately, a human-AI agreement literature review leveraging both human expertise and six AI models was performed to assess the relevance of 1,712 INCOSE INSIGHT articles and 889 SERC publications. The results identify five critical research gaps and offer guidance for practitioners navigating AI adoption, assurance, and workforce transformation in SE. We share the agreement data and the AI4SE/SE4AI Explorer web application so readers can compare their own relevance judgments with the human and AI raters.
Chinese Translation
2020年3月，INCOSE INSIGHT关于人工智能（AI）与系统工程（SE）的特刊成为该出版物历史上下载量最多的一期，并启动了一个研究社区，该社区每年吸引超过250名注册者参加其年度研讨会。本文基于作者对该领域核心论文的阅读，追溯了人工智能与系统工程在三个阶段（分别标记为基础阶段、应用阶段和大型语言模型（LLM）转折点）的进展，并描述了我们对社区的汇聚点和存在的关键空白的看法。此外，我们进行了人类与人工智能协议文献回顾，利用人类专业知识和六个人工智能模型，评估了1712篇INCOSE INSIGHT文章和889篇SERC出版物的相关性。结果识别出五个关键研究空白，并为从业者在系统工程中应对人工智能的采用、保障和劳动力转型提供了指导。我们分享了协议数据和AI4SE/SE4AI Explorer网络应用，以便读者能够将自己的相关性判断与人类和人工智能评审者进行比较。
View on arXiv Download PDF AI Translation
cs.AI / 16 / 2606.19651

BrainG3N: A Dual-Purpose Tokenizer for Controllable 3D Brain MRI Generation

BrainG3N：一种用于可控3D脑MRI生成的双重用途分词器
Van Puyvelde, Max, Gulluk, Ibrahim, Van Criekinge, Wim, Gevaert, Olivier
Abstract
Three-dimensional (3D) brain MRI is central to clinical neurology and neuro-oncology, where generative models could augment under-represented cohorts, simulate disease trajectories, and support privacy-preserving data sharing. Latent diffusion has been the go-to solution for modeling imaging data, but it places two competing demands on the tokenizer: encoder embeddings must retain the clinical information that downstream tasks act on, and the decoder must reconstruct anatomically faithful volumes. Existing reconstruction-driven tokenizers achieve the second at the expense of the first. To address this, we introduce a fully volumetric masked-autoencoder (MAE) based tokenizer for 3D brain MRI latent diffusion, decoupling encoder and decoder: a frozen 3D MAE encoder produces clinically informative embeddings, while a dedicated CNN decoder reconstructs voxels from a linear projection of those embeddings. We pretrain the encoder on 35,309 volumes from 18 public cohorts spanning four modalities, ten disease categories, and 200+ acquisition sites, and demonstrate its dual utility in two settings. First, on a 23-task linear-probing benchmark, the encoder outperforms or matches SOTA models (i.e., BrainIAC, BrainSegFounder, and MedicalNet) on 21 of 23 tasks. Second, a conditional diffusion transformer (DiT) trained on these clinically informative embeddings supports both conditional generation across six variables and patient-specific longitudinal forecasting. Together these results establish a single 3D brain-MRI embedding space capable of both downstream clinical tasks and controllable generation.
Chinese Translation
三维（3D）脑MRI在临床神经学和神经肿瘤学中至关重要，生成模型可以增强代表性不足的群体，模拟疾病轨迹，并支持隐私保护的数据共享。潜在扩散已成为建模影像数据的首选解决方案，但它对分词器提出了两个相互竞争的要求：编码器嵌入必须保留下游任务所需的临床信息，而解码器必须重建解剖上真实的体积。现有的重建驱动分词器在实现第二个目标时牺牲了第一个目标。为了解决这一问题，我们引入了一种基于全体积掩码自编码器（MAE）的分词器，用于3D脑MRI潜在扩散，解耦了编码器和解码器：一个冻结的3D MAE编码器生成临床信息丰富的嵌入，而一个专用的卷积神经网络（CNN）解码器则从这些嵌入的线性投影中重建体素。我们在来自18个公共队列的35,309个体积上对编码器进行了预训练，这些队列涵盖了四种模态、十种疾病类别和200多个采集地点，并在两个设置中展示了其双重效用。首先，在一个23任务的线性探测基准测试中，编码器在23个任务中的21个任务上超越或匹配了最先进的模型（即BrainIAC、BrainSegFounder和MedicalNet）。其次，基于这些临床信息丰富的嵌入训练的条件扩散变换器（DiT）支持跨六个变量的条件生成和特定患者的纵向预测。这些结果共同建立了一个单一的3D脑MRI嵌入空间，能够同时支持下游临床任务和可控生成。
View on arXiv Download PDF AI Translation
cs.AI / 17 / 2606.19658

Denoising Implicit Feedback for Cold-start Recommendation

冷启动推荐中的隐式反馈去噪
Chen, Gaode, Wang, Shicheng, Li, Shikun, Huang, Rui, Zhang, Xinghua, Luo, Yunze, Li, Shipeng, Ge, Shiming, Sun, Ruina, Jiang, Yinjie, Zhang, Jun
Abstract
Implicit feedback is widely used in recommender systems due to its accessibility and generality, yet it usually presents noisy samples (e.g., clickbait, position bias). Meanwhile, recommenders inevitably face the item cold-start problem due to the continuous influx of new items. We identify that cold items are more prone to noisy samples due to the aforementioned factors, and researchers often overlook the significance of denoising implicit feedback for cold items. Previous denoising studies usually identify noisy samples based on heuristic patterns, such as higher loss values, and mitigate noise through sample selection or re-weighting. However, these methods have limited adaptability and are ineffective in cold-start scenarios. To achieve denoising implicit feedback for cold-start recommendation, we propose a model-agnostic denoising method called DIF. First, user preferences for content remain stable, which allows us to infer pseudo-labels indicating whether a user is interested in a cold item through content-similar warm items. Furthermore, to improve pseudo-label accuracy, we model the confidence of pseudo-labels based on the content similarity between the cold item and warm items, and then aggregate multiple pseudo-labels for each sample. Finally, we explicitly estimate the uncertainty of the noisy sample label by considering its relative entropy and the cold-start status of the item, which adaptively guides the role of pseudo-labels to correct the noisy labels at the sample level. DIF's superiority is supported by both theoretical justification and extensive experiments on real-world datasets. The method has been deployed on a billion-user scale short video application Kuaishou and has significantly improved various commercial metrics within cold-start scenarios.
Chinese Translation
隐式反馈因其可获取性和普遍性在推荐系统中被广泛使用，但通常会出现噪声样本（例如，点击诱饵、位置偏差）。与此同时，由于新项目的持续涌入，推荐系统不可避免地面临项目冷启动问题。我们发现，由于上述因素，冷项目更容易受到噪声样本的影响，而研究人员往往忽视了对冷项目隐式反馈去噪的重要性。以往的去噪研究通常基于启发式模式识别噪声样本，例如较高的损失值，并通过样本选择或重加权来减轻噪声。然而，这些方法适应性有限，在冷启动场景中效果不佳。为了实现冷启动推荐中的隐式反馈去噪，我们提出了一种模型无关的去噪方法，称为DIF。首先，用户对内容的偏好保持稳定，这使我们能够通过内容相似的热项目推断出伪标签，指示用户是否对冷项目感兴趣。此外，为了提高伪标签的准确性，我们基于冷项目与热项目之间的内容相似性对伪标签的置信度进行建模，然后为每个样本聚合多个伪标签。最后，我们通过考虑噪声样本标签的相对熵和项目的冷启动状态，明确估计噪声样本标签的不确定性，这自适应地指导伪标签在样本级别上纠正噪声标签。DIF的优越性得到了理论依据和在真实世界数据集上的广泛实验的支持。该方法已在亿级用户规模的短视频应用快手上部署，并在冷启动场景中显著改善了多项商业指标。
View on arXiv Download PDF AI Translation
cs.AI / 18 / 2606.19683

Exit-and-Join Dynamics for Decentralized Coalition Formation

去中心化联盟形成的退出与加入动态
Zhu, Quanyan
Abstract
This paper studies coalition formation as a decentralized dynamical process driven by unilateral exit-and-join decisions. Agents evaluate local moves using the Aumann-Dreze value, so payoffs are computed within the agent's current coalition rather than through a globally negotiated coalition structure. The resulting model links cooperative payoff allocation with noncooperative best-response behavior: a terminal partition is precisely a coalition structure with no admissible, individually profitable exit-and-join deviation. We establish equilibrium characterizations, identify conditions under which the dynamics admit scalar Lyapunov or exact-potential representations, and analyze how switching and acceptance costs shape local stability. Numerical experiments test finite-time stabilization, cost sensitivity, and a special convex-game benchmark.
Chinese Translation
本文研究了作为一种去中心化动态过程的联盟形成，该过程由单方面的退出与加入决策驱动。代理通过使用 Aumann-Dreze 值来评估局部移动，因此收益是在代理当前的联盟内计算的，而不是通过全球协商的联盟结构。由此产生的模型将合作收益分配与非合作的最佳响应行为联系起来：一个终端划分恰好是一个没有可接受、个体盈利的退出与加入偏差的联盟结构。我们建立了均衡特征，识别了动态允许标量 Lyapunov 或精确势能表示的条件，并分析了切换和接受成本如何影响局部稳定性。数值实验测试了有限时间稳定性、成本敏感性以及一个特殊的凸游戏基准。
View on arXiv Download PDF AI Translation
cs.AI / 19 / 2606.19704

Beyond Static Leaderboards: Predictive Validity for the Evaluation of LLM Agents

超越静态排行榜：大型语言模型代理评估的预测有效性
Patel, Dhaval C., Maghraoui, Kaoutar El, Lin, Shuxin, Li, Yusheng, Feng, Tianjun, Tsai, Chun-Yi, Sun, Yihan, Xin, Wei Alexander, Bhandari, Akshat, Rathod, Tanisha, Fan, Aaron, Shejwal, Sanskruti Vijay, Pasiecznik, Tomas, Kumar, Sagar Chethan, Agarwal, Tanmay, Kanathur, Rohith, Colman, Sam, Sheikh, Amaan, Bahl, Dev, Li, Ann, Veera, Krish, Merchant, Alimurtaza Mustafa, Bhure, Shambhawi Baswaraj, Goyla, Sajal Kumar, Li, Chengrui, Natarajan, Kirthana, Li, Rui, Ajai, Thomas, Li, Rujing, Iyer, Vivek G., Vijayakumar, Sanjaii, Bai, Yitong, Yakobe, Ayal, Maes, Darief, Jebbouri, Yassine, Xu, Tianyang, On, Thai Quoc, Mazeeva, Vera, Li, Winston, Shemla, Yuval, Bhuvanesh, Yeshitha, Bhatt, Rushin, Gowda, Siddharth Chethan, Vinod, Alisha, Cahill, Caroline, Rachakonda, Shriya Aishani, Chen, Yunfeng, Agrawal, Aryaman, Upganlawar, Aman, Ang, Mao Le Jonathan, Go, Yubin Sally, Rajkondawar, Madhav, Chen, Yang-Jung, Maturi, Trisha, Kapoor, Ananya, Li, Andrew, Arora, Shrey, Abbaszadeh, Mana, Li, Shen, Xu, Charles, Kwon, Byeolah
Abstract
Agent benchmarks are growing fast, but no single benchmark touches more than four or five of the dimensions that deployment exposes. This paper aggregates the largest coordinated deep-dive of one MCP-based industrial-agent benchmark to date: fourteen parallel implementation studies covering new asset classes (including a multi-modal visual extension), alternative orchestrations, retrieval strategies, reasoning modes, infrastructure optimizations, and evaluation-methodology probes. Consolidating those studies with seven prior agent benchmarks, we argue that aggregate-score leaderboards systematically underspecify deployed-agent evaluation. Rankings derived from aggregate scores do not transfer to out-of-distribution settings; recent public-to-hidden competition retrospectives provide direct empirical evidence of this rank instability. We propose ranking configurations by predictive validity, the correlation between in-sample and out-of-sample rank, rather than in-sample mean, and report a twelve-tier measurement apparatus that exposes the deployment-relevant dimensions HELM and its agent-era successors collapse. The position is operationalized through three falsifiable out-of-distribution criteria with explicit thresholds; existing evidence partly supports it but is too thin to confirm. We close with a pre-registered pilot design and a field-level vision for what the next generation of agentic benchmarks should report.
Chinese Translation
代理基准正在快速增长，但没有单一基准能够涵盖部署所暴露的四到五个维度。本文汇总了迄今为止最大的基于MCP的工业代理基准的协调深入研究：涵盖新资产类别（包括多模态视觉扩展）、替代编排、检索策略、推理模式、基础设施优化和评估方法探测的十四个平行实施研究。将这些研究与七个先前的代理基准整合，我们认为总分排行榜系统性地低估了已部署代理的评估。从总分得出的排名无法转移到分布外的环境；最近的公开与隐秘竞争回顾提供了这一排名不稳定性的直接实证证据。我们建议通过预测有效性来配置排名，即样本内排名与样本外排名之间的相关性，而不是样本内均值，并报告一个十二层的测量工具，揭示了HELM及其代理时代继任者所崩溃的与部署相关的维度。该立场通过三个可证伪的分布外标准进行操作，具有明确的阈值；现有证据部分支持该立场，但证据仍然不足以确认。最后，我们以一个预注册的试点设计和对下一代代理基准应报告内容的领域级愿景作为结尾。
View on arXiv Download PDF AI Translation
cs.AI / 20 / 2606.19735

GLARE: A Natural Language Interface for Querying Global Explanations

GLARE：用于查询全局解释的自然语言接口
Vasu, Bhavan, Mangannavar, Rajesh
Abstract
While global explanations are crucial for understanding vision models across datasets, classes, and decision contexts, their complex and monolithic nature often hinders practical exploration. Because users typically seek targeted answers to specific questions rather than static artifacts, we present an LLM-based interactive interface that provides natural language access to global explanations for black-box image classifiers. The system's core LLM acts as a mediator, translating natural language questions into structured SQL queries over local explanation data. This enables flexible aggregation without exposing users to low-level representations. For each query, the interface outputs statistics-augmented natural language responses, supporting local explanations, and intent-aligned visualizations. We evaluate the system on intent interpretation, query mapping accuracy, generalization to novel queries and datasets, and robustness to linguistic errors. Our results demonstrate that LLM-mediated querying substantially improves the accessibility and usability of global explanations for human-centered XAI.
Chinese Translation
全局解释对于理解跨数据集、类别和决策上下文的视觉模型至关重要，但其复杂和单一的特性往往妨碍了实际探索。由于用户通常寻求针对特定问题的有针对性的答案，而非静态的文档，我们提出了一种基于大型语言模型（LLM）的交互式接口，提供对黑箱图像分类器全局解释的自然语言访问。系统的核心LLM充当中介，将自然语言问题转换为针对局部解释数据的结构化SQL查询。这使得灵活的聚合成为可能，而无需向用户暴露低级表示。对于每个查询，接口输出增强统计信息的自然语言响应，支持局部解释和与意图对齐的可视化。我们在意图解释、查询映射准确性、新查询和数据集的泛化能力以及对语言错误的鲁棒性等方面评估了该系统。我们的结果表明，LLM中介查询显著提高了全局解释对以人为本的可解释人工智能（XAI）的可访问性和可用性。
View on arXiv Download PDF AI Translation
cs.AI / 21 / 2606.19741

Interpreting Neural Combinatorial Optimization via Evolving Programmatic Bottlenecks

通过演化程序瓶颈解释神经组合优化
Duan, Haocheng, Guo, Yuxin, Bi, Jieyi, Xie, Anqi, Li, Sirui, Ma, Yining, Wu, Cathy
Abstract
Neural Combinatorial Optimization (NCO) achieves strong performance, yet its black-box nature remains a key roadblock to deployment and scientific diagnosis. Standard interpretability tools, such as Concept Bottleneck Models (CBMs), are ill-equipped for NCO, whose decisions are dynamic, state-dependent, and lack proper concept vocabulary definition. To close this gap, we introduce Evolving Programmatic Bottlenecks (EPB), to our knowledge, the first framework for interpreting NCO policies by distilling black-box NCO models into human-readable program portfolios. EPB employs an LLM to autonomously evolve a bank of programs, where each program's per-step action distribution serves as the bottleneck. EPB works through an iterative framework: Block I fixes program bank capacity and introduces a hybrid textual-numerical gradient descent scheme that couples numerical gradients for student router updates and textual gradients for LLM-based program revision; Block II dynamically adapts bank capacity via fault-targeted expansion and redundancy pruning. Extensive experiments demonstrate EPB's effectiveness and broad applicability, where the distilled program portfolios largely match original performance. EPB also reveals that NCO behavior shifts across optimization stages and can be approximated as a composition of classic heuristic variants. Our work advances interpretable NCO and establishes EPB as a promising tool for interpreting sequential decision-making models.
Chinese Translation
神经组合优化（NCO）表现出强大的性能，但其黑箱特性仍然是部署和科学诊断的主要障碍。标准的可解释性工具，如概念瓶颈模型（CBMs），对于NCO并不适用，因为NCO的决策是动态的、依赖于状态的，并且缺乏适当的概念词汇定义。为了解决这一问题，我们引入了演化程序瓶颈（EPB），据我们所知，这是第一个通过将黑箱NCO模型提炼为人类可读的程序组合来解释NCO策略的框架。EPB利用大型语言模型（LLM）自主演化程序库，其中每个程序的每步动作分布作为瓶颈。EPB通过一个迭代框架工作：第一阶段固定程序库容量，并引入一种混合文本-数值梯度下降方案，结合数值梯度用于学生路由器更新和文本梯度用于基于LLM的程序修订；第二阶段通过故障针对性扩展和冗余修剪动态调整库容量。大量实验表明EPB的有效性和广泛适用性，提炼的程序组合在很大程度上与原始性能相匹配。EPB还揭示了NCO行为在优化阶段之间的变化，并可以近似为经典启发式变体的组合。我们的工作推动了可解释NCO的发展，并确立了EPB作为解释序列决策模型的有前景工具。
View on arXiv Download PDF AI Translation
cs.AI / 22 / 2606.19747

A Comparative Study of Pretrained Transformer Models for Quranic ASR: Speech Representations, Label Formats, and Dataset Composition

预训练变换器模型在古兰经自动语音识别中的比较研究：语音表示、标签格式和数据集构成
Hossain, Nabil Mosharraf, Islam, Riasat, Obaidellah, Unaizah
Abstract
Quran Automatic Speech Recognition (ASR) aims to convert Quranic recitation into text, enabling applications such as aided memorisation tools and Quranic search engines. However, existing ASR models often exhibit high Word Error Rates (WER) on user-recited verses and lack full coverage of the Quranic corpus. This paper presents a systematic empirical study of domain-specific fine-tuning of pretrained Transformer-based models for Quranic ASR, using advanced speech feature extraction methods: Wav2Vec2.0, HuBERT, and XLS-R. These models apply self-supervised learning by masking portions of input audio and using Transformer architectures to learn context-aware speech features. The pretrained models are fine-tuned on a filtered Quranic dataset exceeding 870 hours of professional and user recitations. Through comprehensive ablation studies across feature extractors, output label formats, training strategies, and clip durations, we identify the key factors that affect transcription accuracy in this domain. Our best-performing configuration achieves a WER of 0.08 on the EveryAyah subset and 0.11 on the combined EveryAyah+Tarteel setting, representing roughly a five-percentage-point gain over the Citrinet baseline (WER = 0.163) while reducing combined-model training time from 140 hours to 40 hours. Arabic text without diacritics yields the best fine-tuning results, and Wav2Vec2-XLSR-53 provides the strongest overall representation. Future work includes improving dataset quality and developing phoneme-aware models to extract deeper speech feature representations for Tajweed-sensitive applications.
Chinese Translation
古兰经自动语音识别（ASR）旨在将古兰经的朗读转换为文本，从而支持诸如辅助记忆工具和古兰经搜索引擎等应用。然而，现有的ASR模型在用户朗读的经文上通常表现出较高的词错误率（WER），并且未能全面覆盖古兰经语料库。本文系统地实证研究了针对古兰经ASR的预训练变换器模型的领域特定微调，采用了先进的语音特征提取方法：Wav2Vec2.0、HuBERT和XLS-R。这些模型通过对输入音频的部分进行掩蔽，应用自监督学习，并利用变换器架构学习上下文感知的语音特征。预训练模型在一个经过筛选的古兰经数据集上进行了微调，该数据集包含超过870小时的专业和用户朗读。通过对特征提取器、输出标签格式、训练策略和片段时长的全面消融研究，我们识别出影响该领域转录准确性的关键因素。我们最佳的配置在EveryAyah子集上实现了0.08的WER，在结合的EveryAyah+Tarteel设置上实现了0.11的WER，较Citrinet基线（WER = 0.163）提高了约五个百分点，同时将组合模型的训练时间从140小时减少到40小时。没有变音符号的阿拉伯文本产生了最佳的微调结果，而Wav2Vec2-XLSR-53提供了最强的整体表示。未来的工作包括提高数据集质量和开发音素感知模型，以提取更深层次的语音特征表示，适用于对Tajweed敏感的应用。
View on arXiv Download PDF AI Translation
cs.AI / 23 / 2606.19749

Benchmarking Agentic Review Systems

代理评审系统的基准测试
Nguyen, Dang, Hao, Wanqing, Elazar, Yanai, Tan, Chenhao
Abstract
A new class of agentic review systems are emerging as a remedy to the pressure placed on peer review systems by AI-assisted research, but it is unclear how they should be evaluated. We evaluate two open-source systems (OpenAIReview and coarse), one proprietary system (Reviewer3), and a zero-shot baseline, across six LLMs spanning frontier and efficient models. First, we study whether AI reviews on ICLR/NeurIPS papers track with papers' quality as approximated by external signals such as citations and acceptance decisions. Every system performs above chance in pairwise accuracy, and the best is OpenAIReview + GPT-5.5 at 83.0%. Second, to test whether systems can catch errors with known ground truth, we construct a perturbation benchmark that injects four categories of errors into papers across eight arXiv subject classes and measure detection recall. The strongest configuration (OpenAIReview + GPT-5.5) catches 71.6% of injected errors, leaving substantial room for improvement. The union of detections across six models reaches 83.3% recall, suggesting different models detect different errors and better harness design can potentially increase performance. Beyond these benchmarks, we study a public deployment of OpenAIReview with real users. Votes on its comments skew positive at 1.44 to 1, and the most common complaints are about false positives and minor nitpicks. Together, by evaluating full review systems backed by state-of-the-art models on real research papers, we show that while AI reviews still have room for improvement, they can already track human quality judgments well, catch important errors, and earn positive feedback from real users.
Chinese Translation
一种新的代理评审系统正在出现，以应对人工智能辅助研究对同行评审系统施加的压力，但尚不清楚应如何评估这些系统。我们评估了两个开源系统（OpenAIReview 和 coarse）、一个专有系统（Reviewer3）以及一个零样本基线，涵盖六种大型语言模型（LLMs），包括前沿和高效模型。首先，我们研究了 ICLR/NeurIPS 论文的 AI 评审是否与通过外部信号（如引用和接受决定）近似的论文质量相一致。每个系统在成对准确性方面的表现均高于随机水平，其中表现最佳的是 OpenAIReview + GPT-5.5，准确率为 83.0%。其次，为了测试系统是否能够捕捉到已知的真实情况中的错误，我们构建了一个扰动基准，向八个 arXiv 学科类别的论文中注入四类错误，并测量检测召回率。最强的配置（OpenAIReview + GPT-5.5）捕捉到 71.6% 的注入错误，仍有很大的改进空间。六个模型的检测联合召回率达到 83.3%，这表明不同模型能够检测到不同的错误，更好的设计可能会提高性能。除了这些基准测试外，我们还研究了 OpenAIReview 在真实用户中的公共部署。对其评论的投票结果偏向积极，比例为 1.44:1，最常见的投诉是关于假阳性和小问题的挑剔。通过评估基于最先进模型的完整评审系统在真实研究论文上的表现，我们表明，尽管 AI 评审仍有改进空间，但它们已经能够很好地跟踪人类质量判断，捕捉重要错误，并获得真实用户的积极反馈。
View on arXiv Download PDF AI Translation
cs.AI / 24 / 2606.19753

Grounded Inference: Principles for Deterministically Encapsulated Generative Models

基于实证推理：确定性封装生成模型的原则
O'Neill, Marty
Abstract
The incorporation of generative models into traditional computational systems presents both enormous opportunity and tremendous peril. Although many early adopters have realized these perils at great expense, the field still requires foundational frameworks to de-risk incorporation of AI into traditional systems. This manuscript establishes this foundation through the definition of four specific primitives of AI blended architecture, designed to enable deterministic encapsulation of probabilistic models. It further establishes two overarching anti-patterns broadly represented across industry to serve as warnings for engineers in this field. This framework was designed to enable successful integration of AI into traditional systems while providing a foundation upon which generative model providers could build the next generation of generative model interfaces.
Chinese Translation
将生成模型纳入传统计算系统既带来了巨大的机遇，也带来了巨大的风险。尽管许多早期采用者在巨大的代价中意识到了这些风险，但该领域仍需要基础框架，以降低将人工智能纳入传统系统的风险。本文通过定义四个特定的人工智能混合架构原语，建立了这一基础，旨在实现对概率模型的确定性封装。此外，本文还确立了两个在行业中广泛存在的反模式，以警示该领域的工程师。该框架旨在促进人工智能与传统系统的成功集成，同时为生成模型提供者构建下一代生成模型接口奠定基础。
View on arXiv Download PDF AI Translation
cs.AI / 25 / 2606.19759

Optimal Scheduling in a Question-Answering Forum of Knowledge Workers

知识工作者问答论坛中的最优调度
Negi, Rohit, Yilmaz, Mustafa
Abstract
As individuals turn to the Internet to find answers to questions they may have, several Question Answering (QA) forums have evolved, where users knowledgeable in certain topics can contribute their expertise to answering these requests for information. While these are currently volunteer based, we consider a future version employing knowledge workers who are experts in certain topics. In such a system, the request-answer processes forming the queuing system may utilize schedulers that assign requests in different topics to the experts in the forum, who may be able to answer them according to their expertise levels in different topics. With this model, we calculate the capacity of the system for handling the requests while keeping the system stable, and design schedulers that achieve capacity. We also investigate how collaboration between experts in answering requests can potentially increase capacity.
Chinese Translation
随着个人转向互联网寻找问题的答案，多个问答（QA）论坛应运而生，用户可以在这些论坛中贡献他们在特定主题上的专业知识来回答信息请求。虽然这些论坛目前是基于志愿者的，但我们考虑一个未来版本，采用在某些主题上具有专业知识的知识工作者。在这样的系统中，形成排队系统的请求-回答过程可能利用调度器，将不同主题的请求分配给论坛中的专家，这些专家可以根据他们在不同主题上的专业水平来回答问题。通过这一模型，我们计算了系统处理请求的能力，同时保持系统的稳定性，并设计出能够实现该能力的调度器。我们还研究了专家之间在回答请求时的协作如何可能提高处理能力。
View on arXiv Download PDF AI Translation
cs.AI / 26 / 2606.19771

Beyond Entropy: Learning from Token-Level Distributional Deviations for LLM Reasoning

超越熵：从令牌级分布偏差中学习以增强大型语言模型推理
Feng, Xuanzhi, Li, Zhengyang, Liu, Zeyu, Li, Haoxi, Jiang, Yuming, Guo, Bing, Guo, Jingcai, Zhang, Jie, Guo, Song
Abstract
Reinforcement Learning with Verifiable Rewards (RLVR) has significantly advanced Large Language Model (LLM) reasoning; however, it faces a fundamental optimization instability: uniform token updates precipitate entropy collapse, leading to premature convergence to suboptimal strategies, whereas excessive Shannon Entropy maximization can cause entropy explosion, driving blind exploration toward incoherent reasoning chains. To resolve this dichotomy, we introduce the Independent Combinatorial Tokens (ICT) framework, which shifts the optimization focus from scalar uncertainty to the distributional properties of token logits. By leveraging the Jensen-Shannon (JS) divergence between token logits distributions, ICT identifies tokens with distinctive distributional patterns as critical branching points for guiding effective exploration in LLM reasoning. Our theoretical analysis, grounded in both Shannon and second-order R\'enyi entropy, proves that selectively updating on these tokens regulates policy concentration: it reduces the overall distribution uncertainty measured by Shannon entropy, while controlling probability concentration captured by second-order R\'enyi entropy. This dual effect prevents over-concentrated token generation from weakening exploration and effectively stabilizes the training landscape. Empirical results demonstrate that updating only the top 10% of unique tokens on Qwen2.5 (0.5B/1.5B/7B) models yields an average pass@4 improvement of 4.58%, with a maximum gain of 14.9%, over GRPO, 20-Entropy, and STAPO baselines across seven benchmarks spanning math, commonsense, and Olympiad-level problems.
Chinese Translation
可验证奖励的强化学习（RLVR）显著推动了大型语言模型（LLM）推理的发展；然而，它面临着一个根本的优化不稳定性：均匀的令牌更新导致熵崩溃，从而导致过早收敛到次优策略，而过度的香农熵最大化则可能导致熵爆炸，推动盲目探索走向不连贯的推理链。为了解决这一二元对立，我们引入了独立组合令牌（ICT）框架，该框架将优化重点从标量不确定性转向令牌对数的分布特性。通过利用令牌对数分布之间的詹森-香农（JS）散度，ICT识别出具有独特分布模式的令牌，作为引导LLM推理中有效探索的关键分支点。我们的理论分析基于香农熵和二阶R'enyi熵，证明了对这些令牌进行选择性更新可以调节策略集中度：它减少了由香农熵测量的整体分布不确定性，同时控制了由二阶R'enyi熵捕获的概率集中度。这种双重效应防止了过度集中令牌生成削弱探索，并有效稳定了训练环境。实证结果表明，在Qwen2.5（0.5B/1.5B/7B）模型上，仅更新前10%的独特令牌在七个涵盖数学、常识和奥林匹克级问题的基准测试中，相较于GRPO、20-Entropy和STAPO基线，平均提高了4.58%的pass@4，最大增益达14.9%。
View on arXiv Download PDF AI Translation
cs.AI / 27 / 2606.19782

AgentFinVQA: A Deployable Multi-Agent Pipeline for Auditable Financial Chart QA

Narayanan, Aravind, Raza, Shaina
Abstract
Financial chart question answering in regulated settings demands more than accuracy: practitioners must know which answers to trust before acting on them, and many institutions cannot send client data to external model providers. Yet existing chart-QA agents are accuracy-focused and opaque, and most assume proprietary API access; to our knowledge, none combines auditability with on-premise deployability without significant accuracy compromise. We present AgentFinVQA, a multi-agent pipeline that decomposes each query into planning, OCR, legend grounding, visual inspection, and verification, recording every step in a traceable Model Evaluation Packet (MEP) per sample. On FinMME, AgentFinVQA improves $+7.68$ pp over a primary-backbone matched zero-shot baseline with a proprietary backbone (Gemini-3 Flash; 71.24% vs. 63.56%, McNemar $p \approx 1.1 \times 10^{-16}$), and $+4.84$ pp with open-weights Qwen3.6-27B-FP8 served locally. The verifier's verdict also serves as a useful confidence signal (68.2% vs. 55.6% exact accuracy on confirmed vs. revised answers), enabling human-in-the-loop review routing. Error analysis shows that question misunderstanding, legend confusion and extraction error account for nearly two-thirds of failures and are the categories least detected by the verifier, identifying clear directions for future work. Together these results show that auditable, on-premise financial chart QA is practical and that the open-weights system keeps most of the accuracy gains while enabling full data residency. We release our code to support reproducible evaluation.
View on arXiv Download PDF AI Translation
cs.AI / 28 / 2606.19787

ORAgentBench: Can LLM Agents Solve Challenging Operations Research Tasks End to End?

ORAgentBench：大型语言模型代理能否端到端解决具有挑战性的运筹学任务？
Li, Jiajun, Cai, Mingshu, Li, Yixuan, Ding, Yu, Hou, Ran, Nie, Guanyu, Han, Xiongwei, Wang, Wanyuan
Abstract
Large language models are increasingly deployed as autonomous agents for multi-step tasks in executable environments, yet their ability to perform realistic operations research (OR) work remains unclear. Existing OR evaluations often decouple modeling from solving, rely on pre-formalized or text-only instances, and rarely test the full workflow from operational artifacts to validated decisions. In this work, we introduce ORAgentBench, an execution-grounded benchmark for evaluating autonomous agents on challenging end-to-end operations research tasks. It contains 107 human-reviewed tasks across diverse operational scenarios, each packaged in an isolated environment with a natural-language brief, multi-file data, configuration artifacts, and a required submission schema. Agents must write and run solution code, and their submissions are evaluated by hidden validators for schema validity, hard-constraint feasibility, and normalized objective quality. Experiments with fourteen frontier agent-model configurations show that current agents remain far from reliable OR practice. The best agent passes only 35.51% of all tasks and 20.59% of hard tasks, and many feasible submissions still fall below the required quality threshold. Failure analysis further shows that errors are dominated by strategic weaknesses, including missed operational rules, brittle formulations, weak feasible-solution construction, and insufficient solution improvement. OR-specific procedural skills increase hard-task feasibility, but do not reliably improve solution quality or pass rate. These results suggest that progress in OR agents requires moving beyond plausible optimization code toward dependable, high-quality operational decision-making.
Chinese Translation
大型语言模型越来越多地被作为自主代理部署于可执行环境中的多步骤任务，但它们在执行现实运筹学（OR）工作的能力仍不明确。现有的运筹学评估通常将建模与求解分离，依赖于预先形式化或仅文本的实例，且很少测试从操作工件到验证决策的完整工作流程。在本研究中，我们引入了ORAgentBench，这是一个基于执行的基准，用于评估自主代理在具有挑战性的端到端运筹学任务中的表现。该基准包含107个经过人工审核的任务，涵盖多种操作场景，每个任务都被包装在一个孤立的环境中，提供自然语言简要描述、多文件数据、配置工件和所需的提交模式。代理必须编写并运行解决方案代码，其提交将由隐藏的验证者评估模式有效性、硬约束可行性和标准化目标质量。对十四种前沿代理模型配置的实验表明，当前代理在可靠的运筹学实践中仍然远未达到标准。表现最好的代理仅通过了所有任务的35.51%和硬任务的20.59%，而许多可行的提交仍低于所需的质量阈值。失败分析进一步表明，错误主要源于战略性弱点，包括遗漏操作规则、脆弱的表述、弱的可行解构建和不足的解决方案改进。特定于运筹学的程序技能提高了硬任务的可行性，但并未可靠地改善解决方案质量或通过率。这些结果表明，运筹学代理的进展需要超越可信的优化代码，朝着可靠的高质量操作决策迈进。
View on arXiv Download PDF AI Translation
cs.AI / 29 / 2606.19788

CombEval: A Framework for Evaluating Combinatorial Counting in Large Language Models

CombEval：一个用于评估大型语言模型中组合计数的框架
Zhou, Yuxu, Kuželka, Ondřej, Wang, Yuyi, Wang, Yuanhong, Chang, Yi
Abstract
We present CombEval, a dynamic benchmark for evaluating combinatorial counting in large language models. CombEval represents each problem as a typed Cofola specification over entities, combinatorial objects, object dependencies, and constraints, enabling controlled generation of natural-language counting problems with exact solver-verified answers. Unlike static collections, CombEval supports systematic variation of object type, entity scale, constraint count, and reasoning depth. We evaluate 11 LLMs under direct and code-augmented settings and find that models remain brittle on ordered objects, indistinguishable elements, relatively positional constraints, and nested object dependencies. Error analysis further identifies failures in constraint interpretation and counting principles. CombEval provides a diagnostic testbed for studying when and why LLMs fail at combinatorial reasoning. The code and generated benchmark suites are publicly available at \url{https://github.com/YuxuZhou-CN/combination-problem-generation}.
Chinese Translation
我们提出了CombEval，这是一个用于评估大型语言模型中组合计数的动态基准。CombEval将每个问题表示为一个关于实体、组合对象、对象依赖关系和约束的类型化Cofola规范，从而实现对自然语言计数问题的受控生成，并提供经过精确求解器验证的答案。与静态集合不同，CombEval支持对象类型、实体规模、约束数量和推理深度的系统性变化。我们在直接和代码增强设置下评估了11个大型语言模型，发现这些模型在有序对象、不可区分元素、相对位置约束和嵌套对象依赖关系上仍然表现脆弱。错误分析进一步确定了在约束解释和计数原则方面的失败。CombEval提供了一个诊断测试平台，用于研究大型语言模型在组合推理中失败的时机和原因。代码和生成的基准套件已公开发布，网址为 https://github.com/YuxuZhou-CN/combination-problem-generation 。
View on arXiv Download PDF AI Translation
cs.AI / 30 / 2606.19808

Think Again or Think Longer? Selective Verification for Budget-Aware Reasoning

再思考还是更长思考？预算意识推理的选择性验证
Dip, Sajib Acharjee, Zhou, Dawei, Zhang, Liqing
Abstract
Test-time reasoning is increasingly used as a serving-time control knob, but extra reasoning is not uniformly valuable: it can repair failed attempts, waste compute on already-correct answers, or introduce harmful answer changes. We study this as a deployment allocation problem rather than a new-verifier problem. We introduce \sevra, Selective Verification for Reasoning Allocation, a serving-layer controller that decides whether to preserve a frozen solver's initial answer or invoke active verification. Using a frozen Qwen3-4B solver, we log intervention outcomes and train recoverability-aware gates from serving-visible attempt state. On \mathfive, selective verification reaches 76.3\% accuracy, compared with 75.5\% for always verifying, while reducing post-generation tokens by 26.8\% and harmful flips from 2.2\% to 1.0\%. However, an 8,192-token initial solve reaches 76.0\% accuracy with 28\% fewer total model tokens, showing that selective recovery is useful but not the best tested cost frontier. In frozen transfer to \gsm, the selective policy verifies only 3.0\% of examples, improves accuracy from 93.4\% to 94.5\%, and reduces verification tokens by 91.2\% relative to always verifying; again, a longer initial solve matches its accuracy with fewer realized tokens. On CommonsenseQA, always-on verification hurts, while Self-Consistency@5 improves accuracy at about five times the realized token cost. The resulting deployment rule is: tune the initial budget first, then use selective recovery when explicit checks, bounded retries, auditability, or regression-risk control matter.
Chinese Translation
测试时推理越来越多地作为服务时的控制手段，但额外的推理并不总是有价值：它可以修复失败的尝试，浪费计算在已经正确的答案上，或引入有害的答案变化。我们将此视为一种部署分配问题，而不是新验证者问题。我们引入了 extit{sevra}（选择性验证用于推理分配），这是一个服务层控制器，用于决定是保留冻结求解器的初始答案还是调用主动验证。使用冻结的 Qwen3-4B 求解器，我们记录干预结果，并从服务可见的尝试状态中训练可恢复性意识的门控。在 extit{mathfive} 上，选择性验证的准确率达到 76.3\%，而始终验证的准确率为 75.5\\%，同时将生成后的标记减少了 26.8\\%，有害翻转从 2.2\\% 降低到 1.0\\%。然而，8,192 个标记的初始求解达到了 76.0\\% 的准确率，且总模型标记减少了 28\\%，显示选择性恢复是有用的，但并不是测试的最佳成本边界。在向 extit{gsm} 的冻结转移中，选择性策略仅验证 3.0\\% 的示例，将准确率从 93.4\\% 提高到 94.5\\%，并相对于始终验证减少了 91.2\\% 的验证标记；同样，较长的初始求解以更少的实现标记达到了相同的准确率。在 CommonsenseQA 上，始终开启的验证会造成损害，而 Self-Consistency@5 在实现标记成本约为五倍的情况下提高了准确率。最终的部署规则是：首先调整初始预算，然后在显式检查、有限重试、可审计性或回归风险控制重要时使用选择性恢复。
View on arXiv Download PDF AI Translation
cs.AI / 31 / 2606.19812

Human-on-the-Loop Orchestration for AI-Assisted Legal Discovery

人机协作的人工智能辅助法律发现编排
Sinha, Anushree, Ranganathan, Srivaths, Dharmaratnakar, Abhishek, Das, Debanshu
Abstract
Autonomous Large Language Model (LLM) agents are increasingly deployed in electronic discovery (e-discovery), where compounding errors across multi-step reasoning chains can constitute legal malpractice. Unlike single-turn retrieval, agentic workflows operating over privileged document corpora exhibit a class of failure we term "trajectory collapse": an early misclassification silently propagates, rendering an entire privilege review invalid. This paper makes three contributions. First, we propose a structured taxonomy of agentic failures in legal information retrieval, organized by functional stage. Second, we introduce a four-layer verification architecture -- spanning planning, reasoning, execution, and uncertainty quantification -- designed to intercept these failures before they compound. Third, we present a preliminary simulation study on a synthetic e-discovery corpus that demonstrates how mandatory Human-on-the-Loop (HOTL) escalation thresholds reduce privilege-waiver risk relative to fully autonomous baselines. Our results suggest that calibrated uncertainty thresholds can reduce privilege-waiver risk by up to 61% versus fully autonomous deployment, while routing fewer than one quarter of documents to attorney review.
Chinese Translation
自主大型语言模型（LLM）代理在电子发现（e-discovery）中的应用日益增多，其中多步骤推理链中的累积错误可能构成法律失职。与单轮检索不同，操作于特权文档库的代理工作流程表现出一种我们称之为“轨迹崩溃”的失败类型：早期的错误分类会悄然传播，使整个特权审查失效。本文做出了三项贡献。首先，我们提出了一种结构化的法律信息检索中代理失败的分类法，按功能阶段进行组织。其次，我们引入了一种四层验证架构——涵盖规划、推理、执行和不确定性量化——旨在在这些失败累积之前进行拦截。第三，我们展示了一项关于合成电子发现语料库的初步模拟研究，证明了强制性人机协作（Human-on-the-Loop, HOTL）升级阈值相较于完全自主基线如何降低特权放弃风险。我们的结果表明，经过校准的不确定性阈值可以将特权放弃风险降低高达61%，同时将不到四分之一的文档转交给律师审查。
View on arXiv Download PDF AI Translation
cs.AI / 32 / 2606.19821

TelcoAgent: A Scalable 5G Multi-KPM Forecasting With 3GPP-Grounded Explainability

TelcoAgent：一种可扩展的5G多关键性能指标预测方法，具有基于3GPP的可解释性
Kim, Geon, Ron, Dara, Singh, Sukhdeep, Moogi, Suyog, Gajjar, Pranshav, Koduri, V V N K Someswara Rao, Hong, Een Kee, Shah, Vijay K.
Abstract
Key Performance Measurement (KPM) forecasting is essential for proactive network management of 5G and next-generation telecom networks. However, existing machine learning (ML) approaches face significant limitations in scalability and explainability, restricting their effectiveness in real-world deployments. We propose TelcoAgent, a foundation model-based framework that enables accurate, scalable, and explainable forecasting of multiple KPMs across diverse network cells without the need for site-specific training. Specifically, the framework comprises three key components: (i) an automated three-agent pipeline that constructs a 3rd Generation Partnership Project (3GPP) knowledge graph directly from specification documents, (ii) a scalable, time-series foundation model (TSFM)-based prediction pipeline to deliver accurate, zero-shot forecasting, and finally (iii) a reasoning and explanation pipeline that provides actionable, domain-grounded diagnostics. Evaluated using a 3-month, real-world, city-scale 5G KPM dataset from a U.S.-based network operator, TelcoAgent demonstrates high forecasting accuracy for all 7 considered KPMs per cell across 200 cells, while delivering explainable insights and actionable instructions to address network degradations.
Chinese Translation
关键性能指标（KPM）预测对于5G及下一代电信网络的主动网络管理至关重要。然而，现有的机器学习（ML）方法在可扩展性和可解释性方面面临重大限制，限制了其在实际部署中的有效性。我们提出了TelcoAgent，这是一种基于基础模型的框架，能够在不同网络小区中实现多项KPM的准确、可扩展和可解释的预测，而无需特定站点的训练。具体而言，该框架包括三个关键组件：（i）一个自动化的三代理管道，直接从规范文档构建3GPP（第三代合作伙伴计划）知识图谱；（ii）一个基于可扩展时间序列基础模型（TSFM）的预测管道，提供准确的零样本预测；最后，（iii）一个推理和解释管道，提供可操作的、基于领域的诊断。通过使用来自美国某网络运营商的为期3个月的真实城市规模5G KPM数据集进行评估，TelcoAgent在200个小区中对所有7个考虑的KPM展现出高预测准确性，同时提供可解释的见解和可操作的指令，以应对网络降级问题。
View on arXiv Download PDF AI Translation
cs.AI / 33 / 2606.19868

A Systematic Evaluation of Black-Box Uncertainty Estimation Methods for Large Language Models

大型语言模型黑箱不确定性估计方法的系统评估
Wang, Jiayi, Zhang, Xu-Yao
Abstract
Although large language models (LLMs) have shown strong capabilities across a wide range of tasks, their outputs often remain unreliable and may contain hallucinations, making uncertainty estimation (UE) essential for building trustworthy LLMs. In practice, many mainstream LLMs are only accessible through restricted APIs, where internal signals such as logits and hidden states are unavailable, making black-box UE especially important. However, existing work on black-box UE for LLMs remains fragmented in methodology and lacks a unified empirical comparison. To address this gap, we present a systematic review of black-box UE methods and organize them into five categories: verbalization-based, sampling-based, explanation-based, multi-agent, and hybrid methods. We further build a unified evaluation framework and benchmark 24 representative methods across 4 models and 4 dataset settings. Our results show that no single method consistently dominates across all settings. Nevertheless, methods that reason over and compare candidates in the answer space are generally effective, and hybrid methods that combine multiple uncertainty signals perform well under most conditions. By releasing the benchmark data and a unified evaluation framework, we aim to facilitate reproducible comparisons and support future research, while our empirical findings provide practical guidance for developing future black-box UE methods for LLMs.
Chinese Translation
尽管大型语言模型（LLMs）在广泛的任务中表现出强大的能力，但它们的输出往往不可靠，可能包含幻觉，因此不确定性估计（UE）对于构建可信赖的LLMs至关重要。在实践中，许多主流LLMs仅通过受限的API访问，内部信号如logits和隐藏状态不可用，这使得黑箱UE显得尤为重要。然而，现有关于LLMs黑箱UE的研究在方法论上仍然零散，缺乏统一的实证比较。为了解决这一问题，我们对黑箱UE方法进行了系统评审，并将其组织为五类：基于语言表述的方法、基于采样的方法、基于解释的方法、多智能体的方法和混合方法。我们进一步构建了一个统一的评估框架，并在4个模型和4个数据集设置下对24种具有代表性的方法进行了基准测试。我们的结果表明，没有单一方法在所有设置中始终占据优势。然而，在答案空间中进行推理和比较候选项的方法通常是有效的，而结合多种不确定性信号的混合方法在大多数条件下表现良好。通过发布基准数据和统一评估框架，我们旨在促进可重复的比较并支持未来研究，同时我们的实证发现为开发未来LLMs的黑箱UE方法提供了实用指导。
View on arXiv Download PDF AI Translation
cs.AI / 34 / 2606.19893

MetaResearcher: Scaling Deep Research via Self-Reflective Reinforcement Learning in Adversarial Virtual Environments

MetaResearcher：通过自我反思强化学习在对抗性虚拟环境中扩展深度研究
Yu, Wei, Liu, Suxing, Yu, Minjie, Wang, Jiahao, Zheng, Zhijian, Deng, Haocheng, Li, Bing
Abstract
Deep research agents have demonstrated remarkable capabilities in autonomous information gathering and synthesis, yet their training remains constrained by the static nature of simulated environments, the limits of fact-retrieval-only task designs, and the inefficiency of outcome-based reinforcement learning. In this work, we propose MetaResearcher, a novel framework that scales deep research agent training across four synergistic dimensions. First, we introduce an Evolving Virtual World that injects temporal dynamics and adversarial misinformation into the training environment, forcing agents to develop source credibility assessment and temporal conflict resolution skills. Second, we design Discovery-Oriented Tasks -- including hypothesis generation and contradiction resolution -- that transcend simple fact retrieval and push agents toward genuine research behaviors. Third, we propose a Self-Reflective Meta-Reward mechanism within the GRPO framework that jointly optimizes for answer correctness, search path efficiency, reflection depth, and tool call diversity, directly addressing the repetitive action loop problem observed in prior work. Fourth, we introduce a Heterogeneous Multi-Agent Swarm architecture comprising specialized Scout, Filter, and Synthesizer models that learn collaborative research strategies through coordinated reinforcement learning. Built upon the LiteResearcher infrastructure, MetaResearcher requires zero marginal API cost for training while targeting substantial improvements in both benchmark performance (GAIA, Xbench-DS) and epistemic robustness under adversarial conditions. We present the complete framework design, training methodology, and planned experimental validation.
Chinese Translation
深度研究代理在自主信息收集和综合方面展现了显著能力，但其训练仍受到模拟环境静态特性、仅限于事实检索的任务设计限制以及基于结果的强化学习效率低下的制约。在本研究中，我们提出了MetaResearcher，一个在四个协同维度上扩展深度研究代理训练的新框架。首先，我们引入了一个不断演变的虚拟世界，在训练环境中注入时间动态和对抗性虚假信息，迫使代理发展源可信度评估和时间冲突解决技能。其次，我们设计了以发现为导向的任务——包括假设生成和矛盾解决——超越简单的事实检索，推动代理朝向真正的研究行为。第三，我们在GRPO框架内提出了一种自我反思的元奖励机制，联合优化答案正确性、搜索路径效率、反思深度和工具调用多样性，直接解决了先前工作中观察到的重复行动循环问题。第四，我们引入了一种异构多代理群体架构，包括专门的侦察员、过滤器和合成器模型，通过协调的强化学习学习协作研究策略。基于LiteResearcher基础设施，MetaResearcher在训练时无需额外的API成本，同时在基准性能（GAIA，Xbench-DS）和对抗条件下的认知稳健性方面实现显著提升。我们展示了完整的框架设计、训练方法论和计划的实验验证。
View on arXiv Download PDF AI Translation
cs.AI / 35 / 2606.19911

Multi-Agent Transactive Memory

多智能体交易记忆
Kim, To Eun, He, Xuhong, Jain, Dishank, Agrawal, Ambuj, Arabzadeh, Negar, Diaz, Fernando
Abstract
The decentralized deployment of LLM agents with diverse capabilities across diverse tasks motivates infrastructure for knowledge sharing across heterogeneous agent populations. Just as search engines index human-generated artifacts to support human problem solving, retrieval systems can organize agent-generated artifacts for reuse across agent populations. We extend retrieval-augmented generation - which demonstrates the value of human-authored artifacts to individual agents - to retrieval of agent-generated artifacts supporting a population of agents. In particular, agent trajectories encode reusable procedural knowledge, yet these artifacts are typically discarded after a single use or retained only by the producing agent, forcing newly instantiated agents to repeatedly rediscover existing solutions. We propose Multi-Agent Transactive Memory (MATM), a framework for population-level storage and retrieval of agent-generated trajectories, where producer agents contribute trajectories to a shared repository and consumer agents retrieve them to improve task execution. We focus on interactive environments (ALFWorld and WebArena), where trajectories are long and encode especially rich procedural structure. Our experiments demonstrate that retrieving trajectories from MATM improves downstream task performance and reduces interaction steps without coordination or joint training. These results position MATM as a design pattern for population-level experience sharing in open agent ecosystems.
Chinese Translation
不同能力的LLM智能体在多样任务中的去中心化部署，激励了在异构智能体群体中进行知识共享的基础设施。正如搜索引擎索引人类生成的文献以支持人类问题解决，检索系统可以组织智能体生成的文献以便在智能体群体中重复使用。我们将检索增强生成（retrieval-augmented generation）这一方法扩展到支持智能体群体的智能体生成文献的检索，前者展示了人类创作文献对个体智能体的价值。特别地，智能体轨迹编码了可重用的程序性知识，但这些文献通常在单次使用后被丢弃，或仅由生成智能体保留，迫使新实例化的智能体不断重新发现现有解决方案。我们提出了多智能体交易记忆（Multi-Agent Transactive Memory, MATM），这是一个用于智能体生成轨迹的群体级存储和检索的框架，其中生成智能体将轨迹贡献给共享库，而消费者智能体则检索这些轨迹以改善任务执行。我们关注于交互环境（ALFWorld和WebArena），在这些环境中，轨迹较长并编码了特别丰富的程序结构。我们的实验表明，从MATM中检索轨迹能够改善下游任务性能，并在没有协调或联合训练的情况下减少交互步骤。这些结果将MATM定位为开放智能体生态系统中群体级经验共享的设计模式。
View on arXiv Download PDF AI Translation
cs.AI / 36 / 2606.19921

eCNNTO: A Highly Generalizable ConvNet for Accelerating Topology Optimization

eCNNTO：一种高度可推广的卷积神经网络用于加速拓扑优化
Lu, Shengbiao, Wei, Xiaodong
Abstract
This work proposes an element-based Convolutional Neural Network (CNN) to accelerate density-based Topology Optimization (TO), termed eCNNTO. TO generally undergoes a large number of iterations, where finite element analysis is performed in every iteration, leading to the efficiency bottleneck especially when dense meshes are used to achieve high-resolution designs. To address this limitation, eCNNTO is proposed to build upon Kallioras et al. (2020), where a Deep Belief Network (DBN) was trained for every element to predict its near-optimal density from its early history, thereby skipping the great majority of iterations and significantly accelerating the TO procedure. However, the method lacks spatial correlations among neighboring elements and may lead to disconnected features in the final structure. The proposed method employs CNN with residual connections to address this issue. On top of it, a novel training strategy is introduced to further enhance the optimization efficiency, where the training dataset consists of the final stage density histories rather than early ones. This change can also help reduce the required training data size. eCNNTO requires only a small dataset to train and yet it can be generalized to problems with largely different boundary conditions, loading cases, design domain geometries, mesh resolutions, as well as non-design domains. In the end, the generalization capabilities and efficiency of eCNNTO are demonstrated through a variety of examples in two and three dimensions, achieving up to 90% and 97% reduction of iterations, respectively.
Chinese Translation
本研究提出了一种基于元素的卷积神经网络（CNN），用于加速基于密度的拓扑优化（TO），称为eCNNTO。拓扑优化通常需要进行大量迭代，每次迭代都需进行有限元分析，这在使用密集网格以实现高分辨率设计时尤其导致效率瓶颈。为了解决这一限制，eCNNTO建立在Kallioras等人（2020）的基础上，后者为每个元素训练了一个深度置信网络（DBN），以根据其早期历史预测其近似最优密度，从而跳过大多数迭代，显著加速拓扑优化过程。然而，该方法缺乏相邻元素之间的空间相关性，可能导致最终结构中的特征不连贯。为了解决这一问题，所提出的方法采用具有残差连接的CNN。此外，引入了一种新颖的训练策略，以进一步提高优化效率，其中训练数据集由最终阶段的密度历史组成，而非早期的密度历史。这一变化还可以帮助减少所需的训练数据量。eCNNTO仅需一个小型数据集进行训练，但能够推广到具有显著不同边界条件、载荷情况、设计域几何形状、网格分辨率以及非设计域的问题。最后，通过二维和三维的多种示例展示了eCNNTO的推广能力和效率，迭代次数分别减少了高达90%和97%。
View on arXiv Download PDF AI Translation
cs.AI / 37 / 2606.19924

The Tao of Agency: Autotelic AI, Embedded Agency and Dissolution of the Self

能动性的道：自我目的性人工智能、嵌入式能动性与自我的解构
Sarkar, Aritra
Abstract
Most artificial intelligence systems are built on the assumption that goals are exogenous and specified by the designer. Exploring what happens when an agent begins generating its own goals opens the field of autotelic AI. Agents are expected not merely to pursue objectives but to discover them. In this article, we trace its consequences through intrinsic motivation, resource-driven priors, causal-interventional learning, homeostasis, and embeddedness; the last of which is found to be a necessary but not sufficient condition for autotelic agency. Embeddedness individuates the agent at the cost of revealing that the individuation is non-unique, such that the same dynamics admit many valid partitions, each defining a different candidate self. The deepest problem with autotelic AI is therefore not how the agent generates goals, but how it generates and relativizes the self to which the goals are assigned. The agent must believe in its own boundary in order to act, and see through that boundary in order to understand. We consolidate these developments into a single framework and extend it along three directions: a quantum formulation in which the agent-environment cut becomes physical, a philosophical reading against non-dual contemplative traditions, and a concrete LLM-based agentic instantiation.
Chinese Translation
大多数人工智能系统建立在目标是外生的并由设计者指定的假设之上。当一个智能体开始生成自己的目标时，探索这一过程开启了自我目的性人工智能的领域。智能体不仅被期望追求目标，还被期望发现目标。本文通过内在动机、资源驱动的先验、因果干预学习、稳态和嵌入性追踪其后果；其中，嵌入性被发现是自我目的性能动性的必要但不充分条件。嵌入性使智能体个体化，但代价是揭示个体化并非唯一，因此相同的动态允许多种有效的划分，每种划分定义了不同的候选自我。因此，自我目的性人工智能的最深层问题不在于智能体如何生成目标，而在于它如何生成并相对化目标所指向的自我。智能体必须相信自己的边界才能行动，并透过该边界进行理解。我们将这些发展整合为一个统一框架，并沿三个方向进行扩展：一种量子表述，其中智能体-环境的划分变为物理的；一种针对非二元冥想传统的哲学解读；以及一种基于大型语言模型（LLM）的具体能动性实例化。
View on arXiv Download PDF AI Translation
cs.AI / 38 / 2606.19935

PhysDrift: Bridging the Embodiment Gap in Humanoid Co-Speech Motion Generation

PhysDrift：弥合类人机器人共语运动生成中的体现差距
Liang, Zhangzhao, Xing, Xiaofen, Yang, Mingyue, Zhou, Wenlve, Xu, Xiangmin
Abstract
Humanoid robots require co-speech motions that are not only expressive and speech-aligned, but also physically executable under embodiment constraints. Existing co-speech generation pipelines are predominantly human-centric: motions are first generated in human-body representations such as SMPL-X and subsequently retargeted to humanoid robots. In this work, we identify a fundamental embodiment gap in this paradigm, where the mismatch between human motion manifolds and humanoid embodiment constraints disrupts embodiment consistency during motion transfer and physical execution. Through extensive analysis, we show that although retargeting can preserve coarse motion semantics, it significantly compresses motion diversity and weakens prosody-motion synchronization, limiting expressive humanoid behaviors. To address this problem, we first propose IK-EER, a prosody-preserving humanoid motion curation framework that jointly optimizes kinematic feasibility and speech-motion temporal alignment during retargeting. Building upon the curated robot-native motion dataset, we further introduce PhysDrift, an embodiment-aware co-speech motion generation framework that directly predicts executable humanoid joint trajectories from speech without relying on intermediate human-body representations. Unlike conventional human-centric pipelines, PhysDrift maintains embodiment consistency throughout both training and inference while incorporating physical regularization to stabilize robot motion dynamics. Extensive experiments and real-world humanoid deployment demonstrate that embodiment-aware robot-native generation substantially improves speech-motion alignment, physical plausibility, motion smoothness, inference efficiency, and real-time interaction capability.
Chinese Translation
类人机器人需要不仅富有表现力且与语言对齐的共语运动，同时也必须在体现约束下可物理执行。现有的共语生成流程主要以人为中心：运动首先在如 SMPL-X 等人类身体表示中生成，然后再重新定向到类人机器人。在本研究中，我们识别出这一范式中的一个基本体现差距，即人类运动流形与类人体现约束之间的不匹配在运动转移和物理执行过程中破坏了体现一致性。通过广泛的分析，我们展示了尽管重新定向可以保留粗略的运动语义，但它显著压缩了运动多样性并削弱了韵律与运动的同步性，从而限制了类人机器人的表现力。为了解决这一问题，我们首先提出了 IK-EER，一个保留韵律的类人运动策划框架，该框架在重新定向过程中联合优化运动的运动学可行性和语言-运动的时间对齐。在此基础上，我们进一步引入 PhysDrift，一个关注体现的共语运动生成框架，该框架直接从语言预测可执行的类人关节轨迹，而无需依赖中间的人类身体表示。与传统的人本中心流程不同，PhysDrift 在训练和推理过程中保持体现一致性，同时结合物理正则化以稳定机器人的运动动态。大量实验和现实世界的类人机器人部署表明，关注体现的机器人本地生成显著改善了语言-运动的对齐、物理合理性、运动平滑性、推理效率和实时交互能力。
View on arXiv Download PDF AI Translation
cs.AI / 39 / 2606.19948

Advancing DialNav through Automatic Embodied Dialog Augmentation

通过自动化具身对话增强推进DialNav
Han, Leekyeung, Jung, Sangwon, Min, Hyunji, Jeong, Jinseong, Kim, Minyoung, Seo, Paul Hongsuck
Abstract
For embodied agents capable of physical interaction, the capability to create and understand dialog is crucial to ensure both safety and effectiveness. While DialNav~\cite{han2025dialnav} provides a framework for holistic evaluation of the dialog--execution loop in photorealistic indoor navigation, its performance remains limited by a critical scarcity of training data (2K episodes). To address this, we propose an automatic generation pipeline, and construct the \textbf{RAINbow} dataset, a large-scale training dataset with 238K episodes for DialNav. Our pipeline converts existing VLN datasets into multi-turn dialog and creates cost-efficient and high-quality dataset. Then, we introduce two additional complementary advances to unlock the data's full potential: (1) Dual-Strategy Training, a navigation training scheme to align the navigation training with the dynamic dialog-navigation loop, and (2) a localization model that leverages VLN knowledge. By combining these complementary solutions, our model substantially outperforms the baseline in success rate on both \textbf{Val Seen} (58.24, \textbf{+89\%}) and \textbf{Val Unseen} (29.05, \textbf{+100\%}) splits, establishing a new state of the art.
Chinese Translation
对于能够进行物理交互的具身代理，创建和理解对话的能力对于确保安全性和有效性至关重要。虽然DialNav~ extcite{han2025dialnav}提供了一个用于对话-执行循环的整体评估框架，但其性能仍受到训练数据严重不足（仅2K个情节）的限制。为了解决这个问题，我们提出了一种自动生成管道，并构建了 extbf{RAINbow}数据集，这是一个包含238K个情节的大规模训练数据集，用于DialNav。我们的管道将现有的视觉语言导航（VLN）数据集转换为多轮对话，并创建了成本效益高且质量优良的数据集。然后，我们引入了两个额外的互补进展，以释放数据的全部潜力：（1）双策略训练（Dual-Strategy Training），一种将导航训练与动态对话-导航循环对齐的导航训练方案，以及（2）利用VLN知识的定位模型。通过结合这些互补解决方案，我们的模型在 extbf{Val Seen}（58.24， extbf{+89 ext%}）和 extbf{Val Unseen}（29.05， extbf{+100 ext%}）分割上的成功率显著超越了基线，建立了新的最先进水平。
View on arXiv Download PDF AI Translation
cs.AI / 40 / 2606.19980

ENPIRE: Agentic Robot Policy Self-Improvement in the Real World

ENPIRE：现实世界中自主机器人策略自我改进
Xiao, Wenli, Xie, Jia, Zhang, Tonghe, Lin, Haotian, Fu, Letian "Max", Xue, Haoru, Lu, Jalen, Yang, Yi, Dai, Cunxi, Wang, Zi, Wu, Jimmy, Wang, Guanzhi, Sastry, S. Shankar, Goldberg, Ken, Fan, Linxi "Jim", Zhu, Yuke, Shi, Guanya
Abstract
Achieving dexterous robotic manipulation in the real world heavily relies on human supervision and algorithm engineering, which becomes a central bottleneck in the pursuit of general physical intelligence. Although emerging coding agents can generate code to automate algorithm search, their successes remain largely confined in digital environments. We conjecture that the missing abstraction to automate robotics research is a repeatable feedback loop for real-world policy improvement: reset the scene, execute a policy, verify the outcome, and refine the next iteration. To bridge this gap, we introduce ENPIRE, a harness framework for coding agents that instantiates this physical feedback routine with four core modules: an Environment module (EN) for automatic reset and verification, a Policy Improvement module (PI) that launches policy refinement, a Rollout module (R) to evaluate policies with one or multiple physical robots operating in parallel, and an Evolution module (E) in which coding agents analyze logs, consult literature, improve training infrastructure and algorithm code to address failure modes. This closed-loop system transforms real-world manipulation learning into a controllable optimization procedure, minimizing human effort while allowing fair ablations across training recipe and agent variants. Powered by ENPIRE, frontier coding agents can autonomously train a policy to achieve a 99% success rate on challenging, dexterous manipulation tasks, such as organizing a pin box, fastening a zip tie, and tool use, a process that further accelerates when we dispatch an agent team on a robot fleet. Our results suggest a practical and scalable path toward deploying coding agents to autonomously advancing robotics in the physical world.
Chinese Translation
在现实世界中实现灵巧的机器人操作在很大程度上依赖于人类监督和算法工程，这成为追求通用物理智能的一个核心瓶颈。尽管新兴的编码代理能够生成代码以自动化算法搜索，但它们的成功仍然主要局限于数字环境。我们推测，自动化机器人研究所缺失的抽象是一个可重复的反馈循环，用于现实世界的策略改进：重置场景、执行策略、验证结果并优化下一次迭代。为了解决这一问题，我们引入了ENPIRE，这是一个为编码代理设计的框架，通过四个核心模块实现这一物理反馈例程：环境模块（Environment，EN）用于自动重置和验证，策略改进模块（Policy Improvement，PI）用于启动策略优化，回滚模块（Rollout，R）用于评估一个或多个物理机器人并行操作的策略，以及进化模块（Evolution，E），在该模块中，编码代理分析日志、查阅文献、改善训练基础设施和算法代码，以应对失败模式。这个闭环系统将现实世界的操作学习转变为一个可控的优化过程，最小化人类的努力，同时允许在训练方案和代理变体之间进行公平的消融实验。借助ENPIRE，前沿编码代理能够自主训练策略，在具有挑战性的灵巧操作任务中实现99%的成功率，例如整理别针盒、固定拉链和工具使用，当我们在机器人车队中派遣代理团队时，这一过程进一步加速。我们的结果表明，部署编码代理以自主推进物理世界中的机器人技术是一条切实可行且可扩展的路径。
View on arXiv Download PDF AI Translation
cs.AI / 41 / 2606.19990

Reward as An Agent for Embodied World Models

作为具身世界模型的代理的奖励
Li, Pu, Lin, Zhigang, Wu, Qiang, Lv, Yongxuan, Wang, Fei, You, Shan
Abstract
While RL has become a promising tool for refining world models, existing methods largely rely on conservative rollouts near the training distribution, limiting exploration, behavioral diversity, and richer dynamic discovery. In this work, we challenge this conservative paradigm. We argue that the core limitation is not exploration itself, but the lack of reliable verification strategies to support broader exploration. Without reliable verification, expanded exploration becomes highly susceptible to reward hacking, where policies exploit imperfect rewards without achieving genuine improvement. To evaluate this motivation, we instantiate our method in embodied world models, where physical plausibility, and task completion provide a rigorous testbed for scalable RL under complex dynamics. On the verification side, we introduce Reward as an Agent, an agentic reward framework that actively evaluates generated behaviors to provide robust reward signals and mitigate reward hacking under distribution shifts. On the exploration side, we introduce Dynamic-Aware Rollout Diversification through DynDiff-GRPO, which explicitly expands action-space exploration to diversify trajectories, broaden state-action coverage, and encourage richer embodied behaviors beyond conservative rollout regimes. By unifying Reward as an Agent with DynDiff-GRPO, we enable RL on a more reliable reward foundation with substantially diversified sampling, effectively mitigating reward hacking while yielding significant accuracy gains across multiple open-source world models, thereby demonstrating that broader exploration can scale successfully when grounded in robust verification.
Chinese Translation
尽管强化学习（RL）已成为优化世界模型的有前景工具，但现有方法在很大程度上依赖于训练分布附近的保守展开，这限制了探索、行为多样性和更丰富的动态发现。在本研究中，我们挑战这一保守范式。我们认为，核心限制并非探索本身，而是缺乏可靠的验证策略来支持更广泛的探索。没有可靠的验证，扩展探索变得极易受到奖励操控的影响，即策略利用不完美的奖励而未能实现真正的改进。为了评估这一动机，我们在具身世界模型中实例化我们的方法，其中物理合理性和任务完成提供了一个严格的测试平台，以支持复杂动态下的可扩展强化学习。在验证方面，我们引入了“作为代理的奖励”（Reward as an Agent），这是一个主动评估生成行为的奖励框架，旨在提供稳健的奖励信号并减轻在分布变化下的奖励操控。在探索方面，我们通过动态感知展开多样化（Dynamic-Aware Rollout Diversification）引入了DynDiff-GRPO，它明确扩展了动作空间的探索，以多样化轨迹、扩大状态-动作覆盖范围，并鼓励超越保守展开机制的更丰富的具身行为。通过将“作为代理的奖励”与DynDiff-GRPO统一，我们在更可靠的奖励基础上实现了强化学习，显著多样化了采样，有效减轻了奖励操控，同时在多个开源世界模型中实现了显著的准确性提升，从而证明了当建立在稳健的验证基础上时，更广泛的探索可以成功扩展。
View on arXiv Download PDF AI Translation
cs.AI / 42 / 2606.20058

Autonomous Event-Driven Multi-Agent Orchestration for Enterprise AI at Scale

大规模企业人工智能的自主事件驱动多代理编排
Dhanyamraju, Harsh Rao, Raghav, Leonidas, Lee, Aaron
Abstract
Enterprise AI aims to move toward continuous event monitoring, detection, and action across specialist agents, yet existing multi-agent systems largely assume discrete request-response workflows and remain underexplored at enterprise scale. We evaluate DAG Plan and Execute and ReAct across 208 production-derived enterprise scenarios spanning Persona (<10 agents), Department (20-80), and Enterprise (200) scales, and introduce a Task Manager for continuous operation via priority inference, related-event merging, and preemption. Results show that scale, not task complexity, dominates orchestration performance: both architectures perform well at small scale but degrade at enterprise scale as agent discovery noise becomes the primary bottleneck, with simple tasks degrading more sharply than complex ones. DAG Plan and Execute offers higher precision and structured parallelization at smaller scales, but its higher overhead worsens at enterprise scale; ReAct is more robust by handling failures incrementally. The Task Manager reduces high-priority queue latency by 14-75% and improves related-event correctness by over 20 percentage points at enterprise scale.
Chinese Translation
企业人工智能旨在实现跨专业代理的持续事件监测、检测和行动，然而现有的多代理系统在很大程度上假设离散的请求-响应工作流程，并且在企业规模上仍然未得到充分探索。我们评估了DAG Plan and Execute和ReAct在208个基于生产的企业场景中的表现，这些场景涵盖了个体（<10个代理）、部门（20-80个代理）和企业（200个代理）规模，并引入了一种任务管理器，通过优先级推断、相关事件合并和抢占实现持续操作。结果表明，规模而非任务复杂性主导了编排性能：两种架构在小规模下表现良好，但在企业规模下性能下降，因为代理发现噪声成为主要瓶颈，简单任务的性能下降比复杂任务更为明显。DAG Plan and Execute在较小规模下提供了更高的精度和结构化并行化，但其更高的开销在企业规模下恶化；ReAct通过增量处理故障表现得更为稳健。任务管理器在企业规模下将高优先级队列的延迟减少了14-75%，并提高了相关事件的正确性超过20个百分点。
View on arXiv Download PDF AI Translation
cs.AI / 43 / 2606.20068

Process-Verified Reinforcement Learning for Theorem Proving via Lean

通过 Lean 进行定理证明的过程验证强化学习
Kim, Minsu, Yun, Se-Young
Abstract
While reinforcement learning from verifiable rewards (RLVR) typically has relied on a single binary verification signal, symbolic proof assistants in formal reasoning offer rich, fine-grained structured feedback. This gap between structured processes and unstructured rewards highlights the importance of feedback that is both dense and sound. In this work, we demonstrate that the Lean proof assistant itself can serve as a symbolic process oracle, supplying both outcome-level and fine-grained tactic-level verified feedback during training. Proof attempts are parsed into tactic sequences, and Lean's elaboration marks both locally sound steps and the earliest failing step, yielding dense, verifier-grounded credit signals rooted in type theory. We incorporate these structured rewards into a GRPO-style reinforcement learning objective with first-error propagation and first-token credit methods that balances outcome- and process-level advantages. Experiments with STP-Lean and DeepSeek-Prover-V1.5 show that tactic-level supervision outperforms outcome-only baselines in most settings, delivering improvements on benchmarks such as MiniF2F and ProofNet. Beyond empirical gains, our study highlights a broader perspective: symbolic proof assistants are not only verifiers at evaluation time, but can also act as process-level reward oracles during training. This opens a path toward reinforcement learning frameworks that combine the scalability of language models with the reliability of symbolic verification for formal reasoning.
Chinese Translation
虽然基于可验证奖励的强化学习（RLVR）通常依赖于单一的二元验证信号，但在形式推理中的符号证明助手提供了丰富且细致的结构化反馈。这种结构化过程与非结构化奖励之间的差距突显了既密集又可靠的反馈的重要性。在本研究中，我们展示了 Lean 证明助手本身可以作为符号过程的神谕，在训练过程中提供结果级和细粒度战术级的验证反馈。证明尝试被解析为战术序列，Lean 的详细说明标记了局部可靠的步骤和最早的失败步骤，从而产生基于类型理论的密集、验证器基础的信用信号。我们将这些结构化奖励纳入 GRPO 风格的强化学习目标中，采用首次错误传播和首次令牌信用方法，平衡结果级和过程级的优势。与 STP-Lean 和 DeepSeek-Prover-V1.5 的实验表明，在大多数设置中，战术级监督优于仅基于结果的基线，在 MiniF2F 和 ProofNet 等基准上取得了改进。除了经验上的提升，我们的研究还强调了一个更广泛的视角：符号证明助手不仅在评估时充当验证者，还可以在训练过程中作为过程级奖励的神谕。这为结合语言模型的可扩展性与符号验证的可靠性的强化学习框架开辟了一条新路径，以用于形式推理。
View on arXiv Download PDF AI Translation
cs.AI / 44 / 2606.20084

Residual-Space Evolutionary Optimization via Flow-based Generative Models

基于流的生成模型的残差空间进化优化
Cao, Zhuo, Krieger, Lena, Nader, Fernanda, Zhao, Xuan, Scharr, Hanno, Assent, Ira
Abstract
Data editing with generative methods typically requires differentiable objectives and gradient-based search. However, these assumptions break down in flow-based settings, where edits are performed through forward and backward integration and often involve non-differentiable or black-box objectives. We introduce residual-space evolutionary optimization, a model-agnostic framework that addresses this gap by combining flow-based generative editing with evolutionary algorithms. Building on the observation that conditional flow matching (CFM) can disentangle condition-controlled factors from instance-specific residuals, our framework directly operates in residual space and separates two complementary search regimes: self-pollination performs local exploitation through feature-preserving residual refinement, and cross-pollination promotes broader exploration by recombining residuals across heterogeneous samples. As a proof of concept, we validate on MorphoMNIST, a benchmark dataset for counterfactual generation, and on crystal data, demonstrating that this exploration--exploitation decomposition provides a useful mechanism for balancing target alignment, instance preservation, and diversity, and extends beyond images to real-world scientific domains.
Chinese Translation
使用生成方法进行数据编辑通常需要可微分的目标和基于梯度的搜索。然而，这些假设在基于流的设置中失效，因为编辑是通过前向和后向积分进行的，并且通常涉及不可微分或黑箱目标。我们提出了残差空间进化优化，这是一种模型无关的框架，通过将基于流的生成编辑与进化算法相结合来解决这一问题。基于条件流匹配（Conditional Flow Matching, CFM）能够将条件控制因素与特定实例的残差解耦的观察，我们的框架直接在残差空间中操作，并分离出两种互补的搜索机制：自我授粉通过特征保留的残差细化进行局部开发，而交叉授粉则通过在异质样本之间重新组合残差来促进更广泛的探索。作为概念验证，我们在用于反事实生成的基准数据集 MorphMNIST 和晶体数据上进行了验证，证明这种探索-开发分解提供了一种有用的机制，用于平衡目标对齐、实例保留和多样性，并扩展到现实世界的科学领域。
View on arXiv Download PDF AI Translation
cs.AI / 45 / 2606.20087

Multi-Head Attention-Based Feature Extractor Integration with Soft Actor-Critic for Porosity Prediction and Process Parameter Optimization in Additive Manufacturing

基于多头注意力的特征提取器与软演员-评论家集成用于增材制造中的孔隙率预测和过程参数优化
Aqabakee, Kianoush, Stella, Leonardo
Abstract
Additive manufacturing process optimization requires precise parameter control to minimize defects such as porosity. Traditional reinforcement learning (RL) approaches using discrete action spaces suffer from slow convergence and susceptibility to local optima, limiting their effectiveness for high-precision manufacturing tasks. This study addresses these limitations by employing a continuous action space combined with a novel architecture that integrates a multi-head attention mechanism with the Soft Actor-Critic (SAC) algorithm. The attention-based feature extractor enhances the agent's ability to capture subtle variations in low-dimensional input features, enabling more effective exploration-exploitation balance for navigating value spaces with local minima. We validate our approach on porosity prediction and process parameter optimization in laser powder bed fusion, demonstrating faster convergence and higher final reward values compared to standard RL methods including DQN, PPO, TD3, and vanilla SAC. The proposed methodology achieves a convergence value of 322.79 within 14 episodes, outperforming existing approaches while maintaining stability throughout training.
Chinese Translation
增材制造过程的优化需要精确的参数控制，以最小化诸如孔隙率等缺陷。传统的强化学习（RL）方法使用离散动作空间，存在收敛速度慢和易受局部最优影响的问题，限制了其在高精度制造任务中的有效性。本研究通过采用连续动作空间，并结合一种新颖的架构，将多头注意力机制与软演员-评论家（Soft Actor-Critic, SAC）算法集成，来解决这些局限性。基于注意力的特征提取器增强了智能体捕捉低维输入特征中微小变化的能力，从而实现了更有效的探索-利用平衡，以便在具有局部最小值的价值空间中导航。我们在激光粉末床熔融中的孔隙率预测和过程参数优化上验证了我们的方法，结果表明与包括DQN、PPO、TD3和基础SAC在内的标准RL方法相比，收敛速度更快，最终奖励值更高。所提出的方法在14个回合内实现了322.79的收敛值，超越了现有方法，同时在训练过程中保持了稳定性。
View on arXiv Download PDF AI Translation
cs.AI / 46 / 2606.20122

ScaffoldAgent: Utility-Guided Dynamic Outline Optimization for Open-Ended Deep Research

ScaffoldAgent：面向开放式深度研究的效用引导动态大纲优化
Yang, Zhibang, Jiang, Xinke, Xiao, Yuzhen, Zhang, Ruizhe, Fang, Yue, Wan, XinFei, Song, Zhengxing, Liu, Yuxuan, Huang, Yuheng, Chu, Xu, Zhao, Junfeng, Wang, Yasha
Abstract
Open-ended deep research (OEDR) requires systems to acquire knowledge through multi-round retrieval and generate coherent long-form reports. The outline plays a central role as a structural scaffold that coordinates retrieval, evidence organization, and generation. However, existing methods either fix the outline before writing or refine it with local heuristics, leading to scaffold drift under continuous information accumulation and delayed feedback for evaluating outline modifications. We propose ScaffoldAgent, a utility-guided dynamic outline optimization framework for OEDR. ScaffoldAgent models outline evolution as a structured decision process with three operations: Expansion, Contraction, and Revision, enabling controlled updates to the report scaffold. It further introduces a utility-guided feedback mechanism that estimates the downstream value of each outline operation from retrieval gain, structural coherence, and trial-generation quality. The resulting utility signal guides node selection, operation scheduling, and termination during inference. Experiments on DeepResearch Bench and DeepResearch Gym show that ScaffoldAgent consistently improves long-form report generation and factual grounding over existing deep research agents.
Chinese Translation
开放式深度研究（OEDR）要求系统通过多轮检索获取知识，并生成连贯的长篇报告。大纲作为协调检索、证据组织和生成的结构支架，发挥着核心作用。然而，现有方法要么在写作前固定大纲，要么通过局部启发式方法对其进行优化，这导致在持续信息积累和评估大纲修改的反馈延迟下出现支架漂移。我们提出了ScaffoldAgent，一个面向OEDR的效用引导动态大纲优化框架。ScaffoldAgent将大纲演变建模为一个结构化决策过程，包含三个操作：扩展（Expansion）、收缩（Contraction）和修订（Revision），使报告支架的更新可控。它进一步引入了一种效用引导反馈机制，该机制通过检索增益、结构一致性和试生成质量来评估每个大纲操作的下游价值。由此产生的效用信号指导节点选择、操作调度和推理过程中的终止。在DeepResearch Bench和DeepResearch Gym上的实验表明，ScaffoldAgent在长篇报告生成和事实基础方面始终优于现有的深度研究代理。
View on arXiv Download PDF AI Translation
cs.AI / 47 / 2606.20138

Learning to Prompt: Improving Student Engagement with Adaptive LLM-based High-School Tutoring

学习提示：通过自适应基于大型语言模型的高中辅导提高学生参与度
Chang, Po-Chin, Hogan, Nicholas, Plaat, Aske, van der Meer, Michiel T.
Abstract
LLMs can personalize education, although current static-prompt tutoring systems struggle to adapt to diverse academic disciplines. We develop and test a system with subject-aware prompting, based on 14 pedagogical features (e.g., tutor scaffolding, student understanding) extracted from raw transcripts. We first train a prompt routing model in a simulation environment, and then deploy it for online adaptation with actual high-school students. The simulation benchmark shows the router outperforming two static baselines ($0.694$ vs. $0.647$ and $0.64$, $p<0.001$). A/B testing ($N=656$ conversations from 359 students) shows sim-to-real transfer where the model switches from analytical to scaffolding learning strategies. Our adaptive prompt selection mechanism improves instructional efficiency, maintains pedagogical quality and reduces interactions by around 3 turns ($p=0.007$). While a greedy router achieves a comparable exercise conversion rate with the baseline ($19.1\%$ vs. $19.6\%$), a stochastic router that samples strategies leads to a higher conversion rate ($28.1\%$).
Chinese Translation
大型语言模型（LLMs）能够个性化教育，尽管当前的静态提示辅导系统在适应多样化学科方面存在困难。我们开发并测试了一个基于14个教学特征（例如，辅导者支架、学生理解）提取的学科感知提示系统。我们首先在模拟环境中训练一个提示路由模型，然后将其部署用于与实际高中学生的在线适应。模拟基准显示，该路由器的表现优于两个静态基线（$0.694$ 对比 $0.647$ 和 $0.64$，$p<0.001$）。A/B 测试（$N=656$ 次对话，来自359名学生）显示了从模拟到真实的转移，其中模型从分析学习策略切换到支架学习策略。我们的自适应提示选择机制提高了教学效率，保持了教学质量，并减少了约3轮的互动（$p=0.007$）。虽然贪婪路由器与基线的练习转化率相当（$19.1\%$ 对比 $19.6\\%$），但随机路由器通过采样策略导致了更高的转化率（$28.1\\%$）。
View on arXiv Download PDF AI Translation
cs.AI / 48 / 2606.20142

RACL: Reasoning-Agent Control Layers for Continuous Metaheuristic Learning

RACL：用于连续元启发式学习的推理代理控制层
Manzárraga, Antón Asla
Abstract
This paper introduces RACL, a Reasoning-Agent Control Layer for metaheuristics. RACL places a reasoning agent above an existing optimizer. The agent does not replace the optimizer and does not modify business constraints. Instead, it controls the optimizer's internal search behavior by observing operational memory, reasoning over past behavior, formulating bounded hypotheses, testing interventions, evaluating outcomes, applying guardrails, consolidating useful policies and explaining its decisions. The experiment uses vehicle routing as a testbed, but the contribution is not a new routing solver, a particular ALNS configuration or a specific set of routing rules. The contribution is the RACL method: a way for a reasoning agent to discover, validate, consolidate and explain algorithmic control rules for a metaheuristic. In the current experimental setting, RACL improves or ties the Operational Memory Policy in 21 of 21 feasible cases and improves or ties a non-reasoning Stagnation-Triggered Policy in 18 of 21 feasible cases, with an average RACL vs STP cost delta of -0.641%. In the Sevilla-9/10 runtime sample, RACL improves average cost by -8.337% versus Fixed and -1.605% versus STP without showing material computational overhead. During the proof-of-concept, Codex was used as an in-the-loop reasoning agent observing executions, interpreting logs and proposing live bounded interventions. The policy proxy was later used only to make quantitative evaluation reproducible.
Chinese Translation
本文介绍了RACL，一种用于元启发式算法的推理代理控制层。RACL将一个推理代理置于现有优化器之上。该代理并不替代优化器，也不修改业务约束。相反，它通过观察操作记忆、推理过去行为、制定有界假设、测试干预措施、评估结果、应用保护措施、整合有用策略并解释其决策来控制优化器的内部搜索行为。实验使用车辆路径规划作为测试平台，但贡献并不是一个新的路径求解器、特定的自适应大邻域搜索（ALNS）配置或一组特定的路径规则。贡献在于RACL方法：一种让推理代理发现、验证、整合和解释元启发式算法控制规则的方法。在当前的实验设置中，RACL在21个可行案例中改善或持平了操作记忆策略，在21个可行案例中改善或持平了非推理的停滞触发策略，RACL与停滞触发策略（STP）之间的平均成本差异为-0.641%。在Sevilla-9/10运行样本中，RACL相比固定策略平均成本改善了-8.337%，相比STP改善了-1.605%，且没有显示出显著的计算开销。在概念验证过程中，Codex被用作在循环中的推理代理，观察执行过程、解释日志并提出实时的有界干预。政策代理随后仅用于使定量评估可重复。
View on arXiv Download PDF AI Translation
cs.AI / 49 / 2606.20146

BIM-Edit: Benchmarking Large Language Models for IFC-Based Building Information Modeling

BIM-Edit：基于IFC的建筑信息建模大语言模型基准评估
Nithyanantham, Bharathi Kannan, Kujat, Clemens, Sesterhenn, Tobias, Telgmann, Stefan, Plönnigs, Jörn, Lüdtke, Stefan, Bartelt, Christian
Abstract
Large language models (LLMs) are increasingly applied to computer-aided design (CAD) to generate design artifacts from textual instructions. In engineering practice, this requires more than creating new geometry, models must also understand existing scenes, edit them correctly, and preserve semantics and relations. However, many CAD benchmarks focus on creating new models rather than editing existing ones, and mostly evaluate geometric correctness. We introduce BIM-Edit, a benchmark for evaluating LLMs on natural-language editing of Building Information Models (BIM) represented in the Industry Foundation Classes (IFC) format. BIM provides a challenging testbed because building models encode geometry together with semantic and relational structure. BIM-Edit contains 324 editing tasks spanning 11 realistic building models and 36 synthetic scenes. Tasks are expressed using three instruction categories - direct, spatial, and topological - covering both explicit and scene-grounded edits. We evaluate outputs along three dimensions: geometric accuracy, semantic validity, and topological consistency. Across evaluated LLMs, the best-performing model achieves only 49.5% average score across the three metrics, and no model fully solves more than 3.4% of tasks. These results demonstrate a substantial gap between current LLM capabilities and the requirements of structured engineering design workflows.
Chinese Translation
大型语言模型（LLMs）越来越多地应用于计算机辅助设计（CAD），以根据文本指令生成设计成果。在工程实践中，这不仅需要创建新的几何体，模型还必须理解现有场景，正确编辑它们，并保持语义和关系。然而，许多CAD基准测试侧重于创建新模型，而不是编辑现有模型，并且主要评估几何正确性。我们提出了BIM-Edit，这是一个用于评估LLMs在自然语言编辑建筑信息模型（BIM）方面的基准，BIM以工业基础类（IFC）格式表示。BIM提供了一个具有挑战性的测试平台，因为建筑模型同时编码几何体以及语义和关系结构。BIM-Edit包含324个编辑任务，涵盖11个真实建筑模型和36个合成场景。任务通过三种指令类别表达——直接、空间和拓扑——涵盖显式和场景基础的编辑。我们从三个维度评估输出：几何准确性、语义有效性和拓扑一致性。在评估的LLMs中，表现最佳的模型在这三项指标上的平均得分仅为49.5%，且没有模型能够完全解决超过3.4%的任务。这些结果表明，当前LLM的能力与结构化工程设计工作流程的要求之间存在显著差距。
View on arXiv Download PDF AI Translation
cs.AI / 50 / 2606.20156

Modularity-Free Conflict-Averse Training for Generalized PINNs

无模块化的抗冲突训练用于广义物理信息神经网络
Kong, Heejo, Park, Beomchul, Kim, Sung-Jin, Lee, Seong-Whan
Abstract
Physics-informed neural networks (PINNs) have become a powerful framework for solving PDEs by embedding physical laws into differentiable objectives. Despite their advances, training PINNs remains fragile: recent conflict-averse optimization schemes alleviate gradient interference between residual and boundary losses, but we show that their effectiveness deteriorates as model capacity increases. In this paper, we identify a capacity-induced failure mode, where overparameterized networks undergo functional modularity, self-partitioning into task-exclusive modules that suppress cross-objective interaction and hinder convergence toward Pareto-stationary points. To address this issue, we propose a novel framework, Modular-Sparsity Synchronization (ModSync), which integrates structural optimization into conflict-averse training by penalizing task-exclusive connections while preserving interaction-promoting pathways. Extensive experiments across diverse PDE benchmarks demonstrate that ModSync consistently prevents capacity-driven failures, sustains robust cross-objective coupling, and achieves state-of-the-art accuracy. Codes are available at \url{https://github.com/heejokong/ModSync}.
Chinese Translation
物理信息神经网络（PINNs）已成为解决偏微分方程（PDEs）的强大框架，通过将物理定律嵌入可微分目标中。尽管取得了进展，训练PINNs仍然脆弱：最近的抗冲突优化方案缓解了残差损失与边界损失之间的梯度干扰，但我们发现，随着模型容量的增加，其有效性会下降。在本文中，我们识别出一种由容量引起的失效模式，其中过参数化的网络经历功能模块化，自我划分为任务专属模块，从而抑制交叉目标的相互作用并妨碍向帕累托稳态点的收敛。为了解决这个问题，我们提出了一种新颖的框架——模块稀疏同步（Modular-Sparsity Synchronization，ModSync），通过惩罚任务专属连接，同时保留促进相互作用的路径，将结构优化整合到抗冲突训练中。针对多种PDE基准的广泛实验表明，ModSync始终能防止由容量驱动的失效，维持强健的交叉目标耦合，并实现了最先进的准确性。代码可在 {https://github.com/heejokong/ModSync} 获取。
View on arXiv Download PDF AI Translation
cs.AI / 51 / 2606.20162

Implicit Semantic-Aware Communication Based on Hypergraph Reasoning

基于超图推理的隐式语义感知通信
Liao, Yiwei, Tu, Shurui, Xiao, Yong, Li, Yingyu, Shi, Guangming
Abstract
Semantic-aware communication has emerged as a transformative paradigm for next-generation communication systems, shifting the fundamental goal from transmitting bit-level symbols to reliably recovering and understanding the semantic meaning of information. Previous studies have demonstrated that representing the semantic content of source messages as graph-based structures can significantly improve communication efficiency and the accuracy of semantic inference at the receiver. However, existing solutions typically employ graphs that capture only pairwise relationships, thereby neglecting higher-order implicit correlations commonly observed in real-world scenarios, such as group interactions, multi-entity associations, and complex relational contexts. This limitation reduces semantic expressiveness and makes semantic inference susceptible to ambiguity and performance degradation, particularly under noisy or corrupted channel conditions. To address these issues, this paper proposes a novel hypergraph-based implicit semantic reasoning framework, HISR, which leverages hypergraphs to represent complex multi-entity relationships among semantic knowledge entities. In HISR, entities and their associated higher-order relations are mapped into dedicated semantic subspaces tailored to distinct relational contexts. This design not only disentangles diverse semantic interactions to mitigate the over-smoothing effects commonly found in traditional graph embedding methods but also enables robust semantic inference even when partial information loss occurs during transmission. Numerical results show that the proposed HISR achieves up to a 36.6% improvement in implicit semantic interpretation accuracy over the state-of-the-art benchmarks.
Chinese Translation
语义感知通信已成为下一代通信系统的一种变革性范式，其基本目标从传输比特级符号转变为可靠地恢复和理解信息的语义意义。先前的研究表明，将源消息的语义内容表示为基于图的结构可以显著提高通信效率和接收端的语义推理准确性。然而，现有解决方案通常采用仅捕捉成对关系的图，从而忽视了在现实场景中常见的高阶隐式关联，例如群体互动、多实体关联和复杂关系背景。这一局限性降低了语义表达能力，使得语义推理容易受到模糊性和性能下降的影响，尤其是在噪声或信道损坏的条件下。为了解决这些问题，本文提出了一种新颖的基于超图的隐式语义推理框架HISR，该框架利用超图表示语义知识实体之间复杂的多实体关系。在HISR中，实体及其相关的高阶关系被映射到专门针对不同关系背景的语义子空间中。这一设计不仅解开了多样的语义交互，以减轻传统图嵌入方法中常见的过平滑效应，还能够在传输过程中部分信息丢失时实现稳健的语义推理。数值结果表明，所提出的HISR在隐式语义解释准确性上比最先进的基准提高了最高36.6%。
View on arXiv Download PDF AI Translation
cs.AI / 52 / 2606.20205

Apparent Psychological Profiles of Large Language Models are Largely a Measurement Artifact

大型语言模型的表观心理特征在很大程度上是测量伪影
Meyer, Jelena, Garcia, David, Wulff, Dirk U.
Abstract
Psychological instruments designed for humans are increasingly used to assign large language models (LLMs) stable psychological profiles that affect their usability, safety assessment, and use as proxies for human participants in research. Using a formal psychometric framework, we show that these profiles are largely a measurement artifact. Administering a battery of personality and risk-preference instruments spanning self-reports and behavioral tasks to 56 instruction-tuned LLMs alongside large human reference samples, we report four findings. First, differences between models are driven not by the traits an instrument targets but by a directional response bias, a tendency to respond toward one end of the scale, or one labeled option, regardless of item content; a variance decomposition attributes 81-90% of between-model variation to this bias, against 9-16% in humans. Second, the bias declines with model capability but is not eliminated by it. Third, because bias rather than trait drives responding, an instrument's apparent reliability is almost entirely predicted by its response orthogonality, a term we coin for the proportion of items for which trait and bias point in opposite directions. Fourth, the profile a model appears to have shifts with the items used and can be manufactured through item selection. These results demonstrate that the apparent psychological profiles of LLMs are artifacts of the instrument used to measure them, not properties of the models themselves. As instruments borrowed from human psychology are rarely fully orthogonal and may inherently lack validity for LLMs, we call for dedicated assessments centered on response orthogonality.
Chinese Translation
为人类设计的心理测量工具越来越多地用于为大型语言模型（LLMs）分配稳定的心理特征，这些特征影响其可用性、安全评估以及在研究中作为人类参与者的代理使用。我们使用正式的心理测量框架，表明这些特征在很大程度上是测量伪影。通过对56个经过指令调优的LLMs施加一系列涵盖自我报告和行为任务的人格与风险偏好测量工具，并与大规模人类参考样本进行比较，我们报告了四个发现。首先，模型之间的差异并非由测量工具所针对的特征驱动，而是由一种方向性反应偏差驱动，即倾向于向量表的一端或某个标记选项响应，而不考虑项目内容；方差分解显示，81-90%的模型间变异归因于这种偏差，而人类则为9-16%。其次，偏差随着模型能力的提高而下降，但并未被消除。第三，由于偏差而非特征驱动响应，测量工具的表观可靠性几乎完全由其响应正交性预测，我们为此创造了一个术语，指的是特征和偏差指向相反方向的项目比例。第四，模型看似拥有的特征会随着所用项目的变化而变化，并且可以通过选择项目进行制造。这些结果表明，LLMs的表观心理特征是用于测量它们的工具的伪影，而非模型本身的属性。由于借用自人类心理学的测量工具很少完全正交，并且可能固有地缺乏对LLMs的有效性，我们呼吁进行以响应正交性为中心的专门评估。
View on arXiv Download PDF AI Translation
cs.AI / 53 / 2606.20208

Beyond Accuracy: Measuring Logical Compliance of Predictive Models

超越准确性：预测模型的逻辑合规性测量
Delplanque, Guillaume Olivier, Genevès, Pierre, Layaïda, Nabil, Faure, Zephirin
Abstract
Machine learning models are predominantly evaluated through predictive performance metrics such as ranking quality, prediction error, or classification accuracy. While these metrics effectively quantify how closely predictions match the ground truth, they do not assess whether model outputs respect predefined logical or domain-specific constraints. In high-stakes applications, including healthcare, finance, and autonomous systems, logical consistency can be as critical as predictive accuracy, yet no standard metric captures this dimension. We introduce the Rule Violation Score (RVS), a complementary evaluation metric that quantifies the extent to which a predictive model respects a given set of logical rules, independently of predictive accuracy. RVS treats hard rules (strict constraints) and soft rules (statistical regularities) differently, can be evaluated on any dataset and on any predictive model expressed over a relational vocabulary, and can be computed using SQL queries that are automatically generated for Horn rules. Beyond evaluating models, RVS can also evaluate the logical consistency of training datasets and help identify poorly defined rules. We evaluate RVS on three benchmarks covering knowledge graph link prediction and relational regression, including rule-based, embedding-based, and neuro-symbolic predictive models. Our results demonstrate that two models achieving comparable predictive accuracy can exhibit substantially different levels of logical compliance, revealing differences in model behavior that standard metrics fail to capture.
Chinese Translation
机器学习模型主要通过预测性能指标进行评估，如排名质量、预测误差或分类准确率。尽管这些指标有效量化了预测与真实情况的匹配程度，但它们并未评估模型输出是否遵循预定义的逻辑或领域特定的约束。在医疗、金融和自主系统等高风险应用中，逻辑一致性与预测准确性同样关键，但目前没有标准指标能够捕捉这一维度。我们提出了规则违反分数（Rule Violation Score, RVS），这是一种补充评估指标，用于量化预测模型遵循给定逻辑规则的程度，而不依赖于预测准确性。RVS对硬规则（严格约束）和软规则（统计规律）进行不同处理，可以在任何数据集和任何在关系词汇上表达的预测模型上进行评估，并且可以使用为Horn规则自动生成的SQL查询进行计算。除了评估模型，RVS还可以评估训练数据集的逻辑一致性，并帮助识别定义不清的规则。我们在三个基准上评估了RVS，这些基准涵盖知识图谱链接预测和关系回归，包括基于规则、基于嵌入和神经符号预测模型。我们的结果表明，两个具有可比预测准确性的模型可能表现出显著不同的逻辑合规性，揭示了标准指标未能捕捉的模型行为差异。
View on arXiv Download PDF AI Translation
cs.AI / 54 / 2606.20210

Augmenting Game AI with Deep Reinforcement Learning

通过深度强化学习增强游戏人工智能
Sestini, Alessandro, Bergdahl, Joakim, Baghi, Amir, Barrette-LaPierre, Jean-Philippe, Fuchs, Florian, Gisslén, Linus
Abstract
Immersion in video games depends not only on graphics, audio, and game mechanics, but also on the quality of in-game characters. Producing believable characters, or game AI, remains a significant challenge as behavioral complexity is hard to capture with hand-coded systems. Game AI is a source of immersion and engagement; however, the limitations stemming from the challenges of creating game AI often lead to frustration and the breaking of the illusion of realism within the game. The introduction of machine learning models opens the door to creating more believable, authentic, and relatable characters in games. The promise is that they either learn from interacting with the game, or from player data, to develop true human-like behavior. In this paper, we envision more applications of reinforcement learning for game AI in the future. For this to materialize, current research limitations are prohibitive to broad deployment across game genres. Therefore, we propose a framework for training reinforcement learning models with a set of requirements in mind that are suited towards game AI and game development. We present examples of games with reinforcement learning-augmented game AI and describe the practicalities of deploying player-facing machine learning agents in modern games. Furthermore, we identify bottlenecks and hard problems in these areas, which we believe offer promising research directions to accelerate the adoption of machine learning in game AI for the video game industry.
Chinese Translation
视频游戏的沉浸感不仅依赖于图形、音频和游戏机制，还取决于游戏内角色的质量。制作可信的角色或游戏人工智能仍然是一个重大挑战，因为手动编码系统难以捕捉行为的复杂性。游戏人工智能是沉浸感和参与感的来源；然而，由于创建游戏人工智能所面临的挑战所带来的局限性，往往导致挫败感，并打破游戏中的现实主义幻觉。机器学习模型的引入为在游戏中创建更可信、真实和易于共鸣的角色打开了大门。其前景是，它们要么通过与游戏的互动学习，要么通过玩家数据学习，以发展出真正的人类行为。在本文中，我们展望了未来在游戏人工智能中更多应用强化学习的可能性。为了实现这一目标，目前的研究限制对各类游戏的广泛部署构成了障碍。因此，我们提出了一个框架，用于训练强化学习模型，考虑到适合游戏人工智能和游戏开发的一系列要求。我们展示了增强强化学习的游戏人工智能的游戏示例，并描述了在现代游戏中部署面向玩家的机器学习代理的实际情况。此外，我们识别了这些领域中的瓶颈和难题，我们相信这些问题提供了有前景的研究方向，以加速机器学习在视频游戏行业游戏人工智能中的应用。
View on arXiv Download PDF AI Translation
cs.AI / 55 / 2606.20227

QMFOL: Benchmarking Large Language Model Reasoning via Quantifiable Monadic First-Order Logic Test Case Generation

QMFOL：通过可量化的一元一阶逻辑测试用例生成对大型语言模型推理的基准评估
Zheng, Xinyi, Shi, Ling, Yu, Tianlong, Zhao, Yongxin, Goette, Lorenz, Wang, Kailong
Abstract
Large Language Models (LLMs) have made significant progress in reasoning, particularly in deductive reasoning, which is crucial for high-stakes decision-making. As models improve, evaluation benchmarks should evolve to keep pace. However, existing benchmarks lack fine-grained control over logical complexity and struggle to balance semantic diversity with logical consistency. To address these issues, we propose QMFOL, an automated framework for generating monadic first-order logic reasoning tasks with quantifiable and controllable complexity. It constructs formal logical structures using conjunction and disjunction patterns, enabling precise control over reasoning depth, width, label types, and distractors. These structures are then translated into natural language via LLMs, with logical consistency ensured through round-trip verification using an external prover. Based on our framework, we build QMFOLBench, a benchmark comprising 2880 instances with 960 configurations across diverse logical and semantic dimensions. Evaluations on six large reasoning models (LRMs) and two LLMs show that performance degrades and computational overhead increases with rising logical complexity. Models perform better on True-labeled tasks than on False or Unknown ones, and exhibit sensitivity to semantic variation. Overall, QMFOL offers a scalable and reliable approach for constructing deductive reasoning benchmarks with controllable complexity, enabling more precise evaluation of reasoning capabilities in modern language models.
Chinese Translation
大型语言模型（LLMs）在推理方面取得了显著进展，特别是在演绎推理方面，这对于高风险决策至关重要。随着模型的改进，评估基准也应不断演变以跟上步伐。然而，现有基准缺乏对逻辑复杂性的细粒度控制，并且在语义多样性与逻辑一致性之间难以取得平衡。为了解决这些问题，我们提出了QMFOL，这是一个自动化框架，用于生成具有可量化和可控复杂性的一元一阶逻辑推理任务。该框架使用合取和析取模式构建形式逻辑结构，从而精确控制推理的深度、宽度、标签类型和干扰项。这些结构随后通过LLMs翻译为自然语言，并通过使用外部证明者进行往返验证以确保逻辑一致性。基于我们的框架，我们构建了QMFOLBench，这是一个包含2880个实例和960种配置的基准，涵盖多样的逻辑和语义维度。在六个大型推理模型（LRMs）和两个LLMs上的评估显示，随着逻辑复杂性的增加，性能下降和计算开销增加。模型在真标签任务上的表现优于假标签或未知标签任务，并且对语义变化表现出敏感性。总体而言，QMFOL提供了一种可扩展且可靠的方法，用于构建具有可控复杂性的演绎推理基准，从而更精确地评估现代语言模型的推理能力。
View on arXiv Download PDF AI Translation
cs.AI / 56 / 2606.20231

Thermodynamic Measure of Intelligence

热力学智能测量
Chattopadhyay, Ishanu
Abstract
Can intelligence be measured? We propose that intelligence can be defined as the lawful amplification of rare but valid futures: a system increases the probability of outcomes that would be unlikely under passive dynamics but remain admissible under the constraints of the domain. We start with the premise that an intelligent system must model the world and its own place within it. Because the system is part of the world it models, this leads naturally to recursive self-simulation: the system represents futures in which its own actions are part of the trajectory. Our central results give a necessity statement and a conditional near-sufficiency statement connecting this architecture to a precise thermodynamic measure of lawful amplification of rare-valid futures: high rare-valid lift is impossible unless the internal simulation identifies rare-valid futures with high fidelity; conversely, when rare-valid fidelity is high and the simulation contains an effective policy, the achievable lift approaches the actuation-limited optimum. Thus recursive self-simulation is not merely a plausible feature of intelligence but, under the stated assumptions, is necessary and nearly sufficient for high thermodynamic intelligence. The resulting framework makes intelligence measurable on a universal scale, from passive matter and feedback controllers, large language models, and humans as text generators to Maxwell-demon-like information engines.
Chinese Translation
智能可以被测量吗？我们提出智能可以被定义为对稀有但有效未来的合法放大：一个系统增加了在被动动态下不太可能发生的结果的概率，但在领域的约束下仍然是可接受的。我们从一个前提出发，即智能系统必须对世界及其在其中的位置进行建模。由于该系统是其所建模的世界的一部分，这自然导致了递归自我模拟：该系统表示其自身行为是轨迹一部分的未来。我们的核心结果给出了一个必要性声明和一个条件近充分性声明，将这一架构与稀有有效未来的合法放大的精确热力学测量联系起来：高稀有有效提升是不可能的，除非内部模拟以高保真度识别稀有有效未来；相反，当稀有有效保真度高且模拟包含有效策略时，所能实现的提升接近于激活限制的最优值。因此，递归自我模拟不仅仅是智能的一个合理特征，而是在所述假设下，对于高热力学智能是必要且几乎充分的。由此产生的框架使得智能在一个普遍的尺度上可测量，从被动物质和反馈控制器、大型语言模型，到作为文本生成器的人类，以及类似麦克斯韦妖的信息引擎。
View on arXiv Download PDF AI Translation
cs.AI / 57 / 2606.20236

A Multi-Agent system for Multi-Objective constrained optimization

用于多目标约束优化的多智能体系统
Filippini, Federica
Abstract
Many decision-making problems in computing and networking systems can be naturally formulated as cost-minimization problems under performance constraints. In dynamic environments, reinforcement learning (RL) is often used to solve such problems at runtime by embedding both costs and constraint violations into a single scalar reward through weighted penalty terms, following a Lagrangian-inspired formulation. However, in this context the behavior of the learned policy critically depends on the choice of these weights, which are typically selected manually. This makes it difficult to identify an appropriate trade-off between optimizing the primary objective and effectively avoiding constraint violations, particularly in non-stationary environments where their relative importance may change. This paper presents MAMO (Multi-Agent system for Multi-Objective constrained optimization), an approach to tackle this balancing problem through multi-agent RL. MAMO decouples task execution from objective design by formulating the selection of reward weights as a learning problem, providing a !rst step towards more autonomous and robust RL-based solutions for constrained optimization problems in dynamic environments.
Chinese Translation
许多计算和网络系统中的决策问题可以自然地表述为在性能约束下的成本最小化问题。在动态环境中，强化学习（Reinforcement Learning, RL）通常用于通过加权惩罚项将成本和约束违反嵌入到单一标量奖励中，从而在运行时解决此类问题，遵循拉格朗日启发式的公式。然而，在这种情况下，学习到的策略的行为严重依赖于这些权重的选择，而这些权重通常是手动选择的。这使得在优化主要目标与有效避免约束违反之间识别适当的权衡变得困难，特别是在相对重要性可能变化的非平稳环境中。本文提出了MAMO（用于多目标约束优化的多智能体系统），一种通过多智能体强化学习来解决这一平衡问题的方法。MAMO通过将奖励权重的选择公式化为一个学习问题，从而将任务执行与目标设计解耦，为在动态环境中基于强化学习的约束优化问题提供了更自主和稳健的解决方案的第一步。
View on arXiv Download PDF AI Translation
cs.AI / 58 / 2606.20245

Navigating Unreliable Parametric and Contextual Knowledge: Explicit Knowledge Conflict Resolution for LLM Inference

应对不可靠的参数知识和上下文知识：大语言模型推理中的显式知识冲突解决
Peng, Huang, Tang, Jiuyang, Zeng, Weixin, Xu, Hao, Zhao, Xiang
Abstract
Large language models (LLMs) have achieved strong performance across a wide range of language-based tasks by leveraging both extensive parametric knowledge and in-context learning ability, enabling them to incorporate external information provided in the input prompt. However, the integration of external knowledge can introduce conflicts, not only between the model's internal parametric knowledge and the external information, but also among multiple pieces of external contexts. Existing approaches typically assume that either the model or the provided context is reliable, overlooking the possibility that both sources may contain errors, and avoid conflicts by privileging one source over the other, rather than actively resolving inconsistencies. To address these limitations, we propose a novel framework MACR for LLM knowledge conflict resolution that moves beyond the conventional binary choice paradigm and incorporates an explicit conflict-resolution mechanism based on a multi-agent reasoning approach. Specifically, we first propose an adaptive knowledge assessment and retrieval approach that employs a modified semantic entropy measure to quantify an LLM's confidence in its answer to a given query. Based on this confidence estimation, MACR either externalizes the model's internal knowledge as textual representations or retrieves relevant external knowledge when internal knowledge is insufficient, generating basic contexts for subsequent reasoning. Then we introduce an inductive multi-agent reasoning framework with three specialized agents that, respectively, induce explicit rules, analyze potential conflicts, and resolve inconsistencies across all available contexts. Empirical results demonstrate that MACR significantly outperforms state-of-the-art baselines across benchmarks, while also providing interpretable resolutions of explicit conflicts.
Chinese Translation
大型语言模型（LLMs）通过利用广泛的参数知识和上下文学习能力，在各种基于语言的任务中取得了强劲的表现，使其能够整合输入提示中提供的外部信息。然而，外部知识的整合可能引发冲突，这不仅体现在模型的内部参数知识与外部信息之间，还包括多个外部上下文之间的冲突。现有的方法通常假设模型或提供的上下文是可靠的，忽视了两者都可能包含错误的可能性，并通过优先考虑一个来源而非另一个来源来避免冲突，而不是积极解决不一致性。为了解决这些局限性，我们提出了一种新颖的框架MACR，用于大语言模型知识冲突的解决，该框架超越了传统的二元选择范式，并基于多智能体推理方法引入了显式的冲突解决机制。具体而言，我们首先提出了一种自适应知识评估和检索方法，该方法采用修改的语义熵度量来量化LLM对给定查询答案的信心。基于这种信心估计，MACR要么将模型的内部知识外部化为文本表示，要么在内部知识不足时检索相关的外部知识，从而生成后续推理的基本上下文。然后，我们引入了一个归纳多智能体推理框架，包含三个专门的智能体，分别用于归纳显式规则、分析潜在冲突以及解决所有可用上下文中的不一致性。实证结果表明，MACR在基准测试中显著优于最先进的基线，同时也提供了显式冲突的可解释解决方案。
View on arXiv Download PDF AI Translation
cs.AI / 59 / 2606.20264

Confidence-Aware Automated Assessment of Student-Drawn Scientific Models

基于信心的学生绘制科学模型的自动评估
Fang, Luyang, Zhang, Yingchuan, Park, Jongchan, Wang, Zhaoji, Ma, Ping, Zhai, Xiaoming
Abstract
Student-generated drawings are widely used in science education to assess learners' conceptual understanding in modeling-based tasks aligned with the Next Generation Science Standards (NGSS). However, scoring such drawings requires expert human judgment to interpret complex visual representations, making large-scale assessment costly to implement and sustain in classroom settings. In this work, we study automated scoring of student-generated scientific drawings using a vision-based model. We evaluate a Vision Transformer (ViT) with parameter-efficient adaptation and propose a confidence-aware scoring framework that derives response-level confidence from test-time predictive distributions. This confidence signal enables selective automation by scoring high-confidence responses automatically while deferring uncertain cases for human review. Experiments on six NGSS-aligned middle school assessment items show that the proposed approach improves scoring reliability while supporting a practical trade-off between automated coverage and scoring risk, highlighting the value of confidence-aware methods for trustworthy educational assessment.
Chinese Translation
学生生成的绘图在科学教育中被广泛用于评估学习者在与下一代科学标准（NGSS）对齐的建模任务中的概念理解。然而，对这些绘图的评分需要专家的人工判断来解释复杂的视觉表现，这使得在课堂环境中实施和维持大规模评估的成本高昂。在本研究中，我们使用基于视觉的模型研究学生生成的科学绘图的自动评分。我们评估了一种具有参数高效适应的视觉变换器（Vision Transformer, ViT），并提出了一种基于信心的评分框架，该框架从测试时的预测分布中推导出响应级别的信心。这种信心信号使得选择性自动化成为可能，通过自动评分高信心的响应，同时将不确定的案例推迟给人工审核。对六个与NGSS对齐的中学评估项目的实验表明，所提出的方法提高了评分的可靠性，同时支持自动化覆盖与评分风险之间的实际权衡，突显了基于信心的方法在可信教育评估中的价值。
View on arXiv Download PDF AI Translation
cs.AI / 60 / 2606.20274

Lagrange: An Open-Vocabulary, Energy-Based Sparse Framework for Generalized End-to-End Driving

Lagrange：一种开放词汇、基于能量的稀疏框架用于广义端到端驾驶
Ji, Shihao, Li, HongXi, Song, Zihui, Li, Mingyu
Abstract
Scaling end-to-end autonomous driving to complex, open-world environments requires perceptual models that generalize to anomalous scenarios and planners that produce kinematically valid trajectories. Existing paradigms face a distinct dichotomy between representational efficiency and generalization capacity. Dense models (e.g., occupancy networks), while geometrically robust, incur critical computational bottlenecks and struggle with high-level semantic reasoning. Conversely, sparse, query-based planners are efficient but reliant on closed-set definitions, rendering them vulnerable to out-of-distribution (OOD) events. Although recent Vision-Language-Action (VLA) models offer open-vocabulary reasoning, their autoregressive, discrete token generation fundamentally conflicts with the continuous, high-frequency control requirements of vehicle dynamics. To address this, we propose Lagrange, an open-vocabulary, computationally sparse driving framework based on Masked Latent Fields (MLF). Rather than relying on dense volumetric reconstructions or closed-set query mechanisms, Lagrange exploits Vision-Language Models (VLMs) to encode class-agnostic object proposals into continuous semantic visual tokens. We introduce an intent-driven masked cross-attention module that temporally filters irrelevant entities, decoding the attended tokens into an implicit continuous energy field defined over spatial coordinates. By framing decision-making as a Lagrangian action minimization problem spanning this energy field, we enforce strict compliance with vehicle kinematics while executing collision avoidance. Extensive offline evaluations on both standard (nuScenes) and long-tail (CODA) benchmarks demonstrate that Lagrange establishes a promising framework for robust, interpretable, and kinematically feasible open-world autonomy.
Chinese Translation
将端到端自主驾驶扩展到复杂的开放世界环境需要能够对异常场景进行泛化的感知模型和能够生成运动学有效轨迹的规划器。现有范式在表征效率和泛化能力之间面临明显的二元对立。尽管密集模型（例如，占用网络）在几何上具有鲁棒性，但它们会导致关键的计算瓶颈，并且在高层语义推理方面表现不佳。相反，稀疏的基于查询的规划器虽然高效，但依赖于封闭集定义，使其在面对分布外（OOD）事件时显得脆弱。尽管最近的视觉-语言-动作（VLA）模型提供了开放词汇推理，但它们的自回归离散令牌生成与车辆动力学的连续高频控制需求根本冲突。为了解决这个问题，我们提出了Lagrange，一种基于掩蔽潜在场（Masked Latent Fields, MLF）的开放词汇、计算稀疏的驾驶框架。Lagrange并不依赖于密集的体积重建或封闭集查询机制，而是利用视觉-语言模型（Vision-Language Models, VLMs）将与类别无关的物体提议编码为连续的语义视觉令牌。我们引入了一种基于意图的掩蔽交叉注意模块，能够在时间上过滤无关实体，将关注的令牌解码为定义在空间坐标上的隐式连续能量场。通过将决策制定框架设定为跨越该能量场的拉格朗日作用最小化问题，我们在执行避碰时严格遵循车辆运动学。在标准（nuScenes）和长尾（CODA）基准上的广泛离线评估表明，Lagrange建立了一个有前景的框架，以实现稳健、可解释且运动学上可行的开放世界自主驾驶。
View on arXiv Download PDF AI Translation
cs.AI / 61 / 2606.20323

Leveraging systems' non-linearity to tackle the scarcity of data in the design of Intelligent Fault Diagnosis Systems

利用系统的非线性特性应对智能故障诊断系统设计中的数据稀缺问题
Santamato, Giancarlo, Garavagno, Andrea Mattia, Solazzi, Massimiliano, Frisoli, Antonio
Abstract
Deep Transfer Learning (DTL) allows for the efficient building of Intelligent Fault Diagnosis Systems (IFDS). On the other hand, DTL methods still heavily rely on large amounts of labelled data. Obtaining such an amount of data can be challenging when dealing with machines or structures faults. This document proposes a novel approach to the design of vibration-based IFDS using DTL in condition of strong data scarcity. A periodic multi-excitation level procedure leveraging intrinsic non-linearities of real-world systems is used to produce images that can be conveniently analysed by pre-trained Convolutional Neural Networks (CNNs) to diagnose faults. A new data visualization method and its augmentation technique are proposed in this paper to tackle the typical lack of data encountered during the design of IFDS. Experimental validation on a railway pantograph structure provides effective support for the proposed method.
Chinese Translation
深度迁移学习（Deep Transfer Learning, DTL）使得构建智能故障诊断系统（Intelligent Fault Diagnosis Systems, IFDS）变得高效。然而，DTL 方法仍然在很大程度上依赖大量标记数据。在处理机器或结构故障时，获取如此数量的数据可能会面临挑战。本文提出了一种新颖的方法，利用DTL在数据稀缺的情况下设计基于振动的IFDS。采用一种周期性多激励水平程序，利用真实世界系统的内在非线性特性，生成可由预训练卷积神经网络（Convolutional Neural Networks, CNNs）方便分析的图像，以诊断故障。本文提出了一种新的数据可视化方法及其增强技术，以应对在设计IFDS过程中常见的数据不足问题。在铁路受电弓结构上的实验验证为所提方法提供了有效支持。
View on arXiv Download PDF AI Translation
cs.AI / 62 / 2606.20333

SoftSkill: Behavioral Compression for Contextual Adaptation

SoftSkill：用于上下文适应的行为压缩
Tao, Xijia, Teng, Yihua, Fu, Xinyu, Liu, Ziru, Chen, Kecheng, Zhao, Yuzhi, Zhang, Suiyun, Liu, Rui, Kong, Lingpeng
Abstract
Agent skills are commonly deployed as natural-language Markdown files that encode answer policies, evidence-use habits, and task procedures. These files are readable and portable, but they are consumed indirectly: for each task instance, a frozen language model must translate a long textual artifact into generation-time behavior. This paper asks whether a natural-language skill can instead initialize a compact continuous context object, refined by a trainable soft delta while the base model remains frozen. We propose SoftSkill, a frozen-backbone method that tunes such soft skills with next-token prediction and deploys them as latent behavioral priors at inference time. In our main single-round setting, a length-32 SoftSkill prefix on Qwen3.5-4B improves over no-skill prompting by 8.3 points on SearchQA, 42.1 points on LiveMath, and 1.3 points on DocVQA. Relative to SkillOpt, SoftSkill improves accuracy by 5.2 points on SearchQA and 12.5 points on LiveMath, while replacing hundreds to thousands of Markdown skill tokens with a few virtual tokens. We further study agentic execution as a harder boundary case, where sparse trajectory imitation provides useful signal but does not yet robustly compress long-horizon procedural behavior. More broadly, the results suggest that some task skills are better treated not as additional Markdown to be reinterpreted at inference time, but as compact latent controls over how a frozen model enters the task.
Chinese Translation
代理技能通常以自然语言Markdown文件的形式部署，这些文件编码了回答策略、证据使用习惯和任务程序。这些文件可读且可移植，但它们是间接消耗的：对于每个任务实例，冻结的语言模型必须将一个冗长的文本工件翻译为生成时的行为。本文探讨了是否可以将自然语言技能初始化为一个紧凑的连续上下文对象，并通过可训练的软增量进行精炼，同时基础模型保持冻结。我们提出了SoftSkill，这是一种冻结骨干方法，通过下一个标记预测来调整这种软技能，并在推理时将其作为潜在行为先验进行部署。在我们的主要单轮设置中，Qwen3.5-4B上的长度为32的SoftSkill前缀在SearchQA上比无技能提示提高了8.3分，在LiveMath上提高了42.1分，在DocVQA上提高了1.3分。相较于SkillOpt，SoftSkill在SearchQA上提高了5.2分，在LiveMath上提高了12.5分，同时将数百到数千个Markdown技能标记替换为少量虚拟标记。我们进一步研究代理执行作为一个更难的边界案例，其中稀疏轨迹模仿提供了有用的信号，但尚未稳健地压缩长时间程序行为。更广泛地说，结果表明某些任务技能更适合被视为紧凑的潜在控制，而不是在推理时重新解释的额外Markdown。
View on arXiv Download PDF AI Translation
cs.AI / 63 / 2606.20363

Automating SKILL.md Generation for Computer-Using Agents via Interaction Trajectory Mining

通过交互轨迹挖掘自动生成计算机使用代理的 SKILL.md
Hao, Yuexing, Li, Xiaomin
Abstract
Explicit skill libraries make computer-using agents easier to inspect, but it remains unclear whether such libraries can be mined from interaction data in a way that improves downstream policies. We study this question through a three-stage pipeline that segments GUI trajectories, clusters segments into candidate skills, and trains a skill-aware policy from the resulting annotations. The mined clusters are readable on the source benchmark: five of eight clusters have at least 0.95 purity against InteraSkill Workflows labels. However, readability does not imply transfer. GRPO improves IW skill-step accuracy only from 18.5\% to 20.5\%, leaves BrowseComp+ essentially unchanged, and underperforms trivial frequency priors on key source-domain metrics. We therefore present the method as a diagnostic study: trajectory mining can expose inspectable skill structure, but the current boundary detector, orderless segment representation, and offline reward model are insufficient for reliable cross-domain policy improvement.
Chinese Translation
显式技能库使计算机使用代理更易于检查，但尚不清楚这些库是否可以通过交互数据挖掘，以改善下游策略。我们通过一个三阶段的流程研究这个问题，该流程对 GUI 轨迹进行分段，将段落聚类为候选技能，并从结果注释中训练一个技能感知策略。挖掘的聚类在源基准上是可读的：八个聚类中有五个与 InteraSkill Workflows 标签的纯度至少为 0.95。然而，可读性并不意味着可转移。GRPO 仅将 IW 技能步骤的准确率从 18.5\% 提高到 20.5\%，对 BrowseComp+ 基本没有影响，并且在关键源领域指标上表现不如简单的频率先验。因此，我们将该方法呈现为一项诊断研究：轨迹挖掘可以揭示可检查的技能结构，但当前的边界检测器、无序段表示和离线奖励模型不足以实现可靠的跨领域策略改进。
View on arXiv Download PDF AI Translation
cs.AI / 64 / 2606.20381

Rethinking Shrinkage Bias in LLM FP4 Pretraining: Geometric Origin, Systemic Impact, and UFP4 Recipe

重新思考 LLM FP4 预训练中的收缩偏差：几何起源、系统影响与 UFP4 配方
Zhao, Qian, Chen, Kunlong, Tian, Changxin, Jiang, Zhonghui, Zhang, Haitao, Yu, Chaofan, Jiang, Peijie, Gong, Mingliang, Liu, Jia, Liu, Ziqi, Zhang, Zhiqiang, Zhou, Jun
Abstract
FP4 training promises substantial reductions in memory and computation cost for LLM pretraining, yet current FP4 hardware paths and recipes, including NVIDIA Blackwell/Rubin-class systems and AMD MI350-series GPUs, remain centered on E2M1 data elements. In this study, we identify a fundamental limitation of that choice: non-uniform formats such as E2M1 inherently suffer from Shrinkage Bias, a systematic negative rounding error caused by the geometric asymmetry of their representable bins. We show that this bias accumulates multiplicatively across layers and is amplified by the Random Hadamard Transform (RHT), providing a unified explanation for the training instability observed in existing E2M1-based FP4 recipes. In contrast, uniform grids (E1M2/INT4) bypass this grid-geometry error and better convert the improved bucket utilization from RHT into higher quantization quality. Based on this finding, we propose UFP4, a uniform 4-bit training recipe that applies RHT to all three training GEMMs while restricting stochastic rounding to dY alone. On Dense 1.5B, MoE 7.9B, and MoE 124B long-run pretraining, UFP4 consistently achieves lower BF16-relative loss degradation than strong E2M1-based baselines, supported by scaling-law analysis and ablation studies. Our results suggest that future accelerators should support E1M2/INT4-style uniform 4-bit grids as first-class training primitives alongside E2M1.
Chinese Translation
FP4 训练承诺在 LLM 预训练中显著降低内存和计算成本，但当前的 FP4 硬件路径和配方，包括 NVIDIA Blackwell/Rubin 系列系统和 AMD MI350 系列 GPU，仍然集中于 E2M1 数据元素。在本研究中，我们识别出这一选择的一个基本限制：非均匀格式如 E2M1 本质上受到收缩偏差的影响，这是一种由于其可表示的桶的几何不对称性所导致的系统性负舍入误差。我们展示了这种偏差在各层之间以乘法方式累积，并且通过随机哈达玛变换（Random Hadamard Transform, RHT）被放大，为现有基于 E2M1 的 FP4 配方中观察到的训练不稳定性提供了统一的解释。相比之下，均匀网格（E1M2/INT4）绕过了这种网格几何误差，并更好地将 RHT 提升的桶利用率转化为更高的量化质量。基于这一发现，我们提出了 UFP4，一种均匀的 4 位训练配方，该配方对所有三个训练 GEMM 应用 RHT，同时将随机舍入限制在 dY 上。在 Dense 1.5B、MoE 7.9B 和 MoE 124B 的长期预训练中，UFP4 一直实现了比强大的基于 E2M1 的基线更低的 BF16 相对损失降幅，得到了规模法则分析和消融研究的支持。我们的结果表明，未来的加速器应支持 E1M2/INT4 风格的均匀 4 位网格，作为与 E2M1 并列的一流训练原语。
View on arXiv Download PDF AI Translation
cs.AI / 65 / 2606.20438

Interpretable Sperm Morphology Classification via Attention-Guided Deep Learning

基于注意力引导的深度学习可解释性精子形态分类
Varzaneh, Zahra Asghari, Khoshkangini, Reza, Ebner, Thomas, Johansson, Lars
Abstract
Male infertility is a major cause of couple infertility, often linked to abnormal sperm morphology. While deep learning models offer automated analysis, most lack interpretability, limiting their clinical adoption. This study proposes an attention-guided deep learning framework for sperm morphology classification. We combine a pretrained EfficientNet-B0 with a Convolutional Block Attention Module (CBAM) to focus on key areas of the sperm head, improving both accuracy and interpretability. Evaluated on the SMIDS and HuSHem public datasets, our model achieves accuracies of 90.2% and 93.9% (macro F1 scores of 0.913 and 0.948), outperforming SimpleCNN and standard EfficientNet-B0. Furthermore, we use Grad-CAM++ visualizations to highlight features influencing the model's decisions. The results demonstrate that this accurate and transparent framework is a practical tool for automated sperm analysis in fertility clinics.
Chinese Translation
男性不育是导致夫妇不育的主要原因，通常与异常的精子形态有关。尽管深度学习模型提供了自动化分析，但大多数模型缺乏可解释性，限制了其临床应用。本研究提出了一种基于注意力引导的深度学习框架用于精子形态分类。我们将预训练的 EfficientNet-B0 与卷积块注意力模块（Convolutional Block Attention Module, CBAM）相结合，聚焦于精子头部的关键区域，从而提高了准确性和可解释性。在 SMIDS 和 HuSHem 公共数据集上的评估结果显示，我们的模型达到了 90.2% 和 93.9% 的准确率（宏 F1 分数分别为 0.913 和 0.948），优于 SimpleCNN 和标准 EfficientNet-B0。此外，我们使用 Grad-CAM++ 可视化技术突出影响模型决策的特征。结果表明，该准确且透明的框架是生育诊所中自动化精子分析的实用工具。
View on arXiv Download PDF AI Translation
cs.AI / 66 / 2606.20459

Context-Aware Hierarchical Bayesian Modeling of IVF Laboratory Environmental Conditions

基于上下文的分层贝叶斯 IVF 实验室环境条件建模
Varzaneh, Zahra Asghari, Khoshkangini, Reza, Saldeen, Pia, Johansson, Lars, Ebner, Thomas
Abstract
IVF pregnancy rates are routinely modeled using patient-level variables, while high-resolution laboratory environmental data remain underutilized. We show that this is a missed opportunity. Rather than relying on raw sensor averages, we engineer 55 context-aware temporal features, including rolling thermal stability, simultaneous temperature-humidity adherence, peak stress duration, and post-stress recovery speed, that capture the dynamics of incubator microenvironments. On 61 weeks of data from an Asian IVF clinic, these features reduce cross-validated prediction error to 1.27%, compared to 3-5% for raw averages. We then train a hierarchical Bayesian Beta regression model that shares environmental effects across an Asian and a Northern European clinic via partial pooling, while preserving site-specific baselines. On held-out data from the Northern European clinic, the model achieves R2 = 0.86 and a 64% error reduction for the 35-39 age group over a naive baseline, demonstrating that structured environmental monitoring contains clinically meaningful, transferable signal.
Chinese Translation
IVF（体外受精）妊娠率通常使用患者级变量进行建模，而高分辨率的实验室环境数据却未得到充分利用。我们展示了这是一个被错过的机会。我们不仅依赖原始传感器平均值，而是工程化了 55 个基于上下文的时间特征，包括滚动热稳定性、同时温湿度遵循、峰值应力持续时间和应力后恢复速度，这些特征捕捉了培养箱微环境的动态变化。在来自一家亚洲 IVF 诊所的 61 周数据上，这些特征将交叉验证的预测误差降低至 1.27%，而原始平均值的误差为 3-5%。随后，我们训练了一个分层贝叶斯 Beta 回归模型，通过部分汇聚在亚洲和北欧诊所之间共享环境效应，同时保留特定地点的基线。在来自北欧诊所的保留数据上，该模型实现了 R² = 0.86，并且在 35-39 岁年龄组上相较于简单基线减少了 64% 的误差，证明了结构化的环境监测包含临床上有意义且可转移的信号。
View on arXiv Download PDF AI Translation
cs.AI / 67 / 2606.20508

What Do Safety-Aligned LLMs Learn From Mixed Compliance Demonstrations?

安全对齐的大型语言模型从混合合规示范中学到了什么？
Dai, Sihui, Patel, Mann
Abstract
Prior work has shown that in-context demonstrations can jailbreak language models, but it remains unclear how models interpret different types of compliance demonstrations. We study this by mixing benign compliance demonstrations (non-harmful request, helpful response) with harmful compliance demonstrations (harmful request, helpful response) and testing three hypotheses about how demonstration composition drives harmful compliance. Across four models, we find that benign and harmful demonstrations are not interchangeable: benign demonstrations can either reduce or increase harmful compliance depending on the model. We further show that preference optimization is the critical training stage that prevents benign demonstrations from increasing harmful compliance, that demonstration ordering exhibits strong recency bias, and that models differ in how refusal interacts with in-context learning: some adopt demonstrated formatting even when refusing, while others override all in-context signals upon refusal. Taken together, this work moves beyond showing that demonstration-based jailbreaking works to characterizing how it works: what models extract from compliance demonstrations depends on demonstration content, ordering, and training methodology.
Chinese Translation
先前的研究表明，情境示范可以破解语言模型，但模型如何解读不同类型的合规示范仍不清楚。我们通过将良性合规示范（无害请求，积极响应）与有害合规示范（有害请求，积极响应）混合，研究了三种关于示范组合如何驱动有害合规的假设。在四个模型中，我们发现良性和有害示范并不能互换：良性示范可以根据模型的不同而减少或增加有害合规。我们进一步表明，偏好优化是防止良性示范增加有害合规的关键训练阶段，示范顺序表现出强烈的近期偏见，并且模型在拒绝与情境学习的互动方式上存在差异：一些模型在拒绝时仍采用示范格式，而其他模型在拒绝时则覆盖所有情境信号。综上所述，本研究不仅展示了基于示范的破解如何有效，还对其工作机制进行了表征：模型从合规示范中提取的内容取决于示范的内容、顺序和训练方法。
View on arXiv Download PDF AI Translation
cs.AI / 68 / 2606.20517

Multi-LCB: Extending LiveCodeBench to Multiple Programming Languages

Multi-LCB：将 LiveCodeBench 扩展到多种编程语言
Ivanova, Maria, Zadorozhny, Pavel, Levichev, Rodion, Petrov, Ivan, Pavel, Adamenko, Lopatin, Ivan, Kutalev, Alexey, Babaev, Dmitrii
Abstract
LiveCodeBench (LCB) has recently become a widely adopted benchmark for evaluating large language models (LLMs) on code-generation tasks. By curating competitive programming problems, constantly adding fresh problems to the set, and filtering them by release dates, LCB provides contamination-aware evaluation and offers a holistic view of coding capability. However, LCB remains restricted to Python, leaving open the question of whether LLMs can generalize across the diverse programming languages required in real-world software engineering. We introduce Multi-LCB, a benchmark for evaluating LLMs across twelve programming languages, including Python. Multi-LCB transforms Python tasks from the LCB dataset into equivalent tasks in other languages while preserving LCB's contamination controls and evaluation protocol. Because it is fully compatible with the original LCB format, Multi-LCB will automatically track future LCB updates, enabling systematic assessment of cross-language code generation competence and requiring models to sustain performance well beyond Python. We evaluated 24 LLMs for instruction and reasoning on Multi-LCB, uncovering evidence of Python overfitting, language-specific contamination, and substantial disparities in multilingual performance. Our results establish Multi-LCB as a rigorous new benchmark for multi-programming-language code evaluation, directly addressing LCB's primary limitation and exposing critical gaps in current LLM capabilities.
Chinese Translation
LiveCodeBench (LCB) 最近已成为评估大型语言模型 (LLMs) 在代码生成任务中的广泛采用的基准。通过策划具有竞争性的编程问题，不断向集合中添加新问题，并根据发布日期进行筛选，LCB 提供了污染意识的评估，并提供了编码能力的整体视角。然而，LCB 仍然局限于 Python，这引发了一个问题，即 LLMs 是否能够在现实世界软件工程中所需的多种编程语言之间进行泛化。我们介绍了 Multi-LCB，这是一个用于评估 LLMs 在包括 Python 在内的十二种编程语言中的基准。Multi-LCB 将 LCB 数据集中 Python 任务转换为其他语言中的等效任务，同时保留 LCB 的污染控制和评估协议。由于它与原始 LCB 格式完全兼容，Multi-LCB 将自动跟踪未来的 LCB 更新，从而实现对跨语言代码生成能力的系统评估，并要求模型在 Python 之外保持良好的性能。我们在 Multi-LCB 上评估了 24 个 LLMs 的指令和推理，发现了 Python 过拟合、特定语言的污染以及多语言性能之间的显著差异。我们的结果确立了 Multi-LCB 作为多编程语言代码评估的新严格基准，直接解决了 LCB 的主要局限性，并揭示了当前 LLM 能力中的关键缺口。
View on arXiv Download PDF AI Translation
cs.AI / 69 / 2606.20518

FlowEdit: Associative Memory for Lifelong Pronunciation Adaptation in Flow-Matching TTS

FlowEdit：用于流匹配文本到语音系统的终身发音适应的关联记忆
Singh, Harshit, Singh, Ayush Pratap, Mathur, Nityanand
Abstract
Flow-matching text-to-speech systems achieve remarkable zero-shot quality but remain static after deployment: pronunciation errors on out-of-vocabulary proper nouns persist unless the model is retrained. We introduce FlowEdit, a life-long adaptation framework for frozen flow-matching TTS that learns pronunciation corrections as latent conditioning edits rather than weight updates. When corrective feedback is provided, FlowEdit optimizes a token-level perturbation in the text embedding space, then stores the correction in a Modern Hopfield Network serving as content-addressable episodic memory. At inference, corrections are retrieved via soft attention with a similarity gate, enabling fuzzy morphological matching. On our curated benchmark of 312 multilingual proper nouns across 18 language families, FlowEdit reduces target-word Phoneme Error Rate by 92.7% relative to the zero-shot baseline while maintaining identical general-speech quality. Corrections complete in approximately 15 seconds on a single GPU.
Chinese Translation
流匹配文本到语音系统在零样本条件下实现了显著的质量，但在部署后仍然保持静态：对于词汇外专有名词的发音错误持续存在，除非对模型进行重新训练。我们提出了FlowEdit，一个用于冻结流匹配文本到语音系统的终身适应框架，它学习发音修正作为潜在的条件编辑，而不是权重更新。当提供纠正反馈时，FlowEdit在文本嵌入空间中优化一个令牌级的扰动，然后将修正存储在一个现代霍普菲尔德网络中，作为内容可寻址的情节记忆。在推理时，通过带有相似性门的软注意力检索修正，从而实现模糊形态匹配。在我们精心策划的包含18个语言家族的312个多语言专有名词的基准测试中，FlowEdit相对于零样本基线将目标词音素错误率降低了92.7%，同时保持相同的一般语音质量。修正过程在单个GPU上大约完成于15秒。
View on arXiv Download PDF AI Translation
cs.AI / 70 / 2606.20526

DeepSWIP: Quotient-WMC Counterfactuals for Neural Probabilistic Logic Programs

DeepSWIP：神经概率逻辑程序的商-加权模型计数反事实
Habib, Saimun, Belle, Vaishak, He, Fengxiang
Abstract
Neurosymbolic systems such as DeepProbLog combine neural perception with probabilistic logic, but standard inference is associational. Counterfactual reasoning additionally requires a causal semantics for interventions and evidence. We introduce DeepSWIP, a single-world counterfactual semantics for DeepProbLog programs. Using neural materialization, we reduce fixed-context neural predicates to ordinary ProbLog choices, apply Single World Intervention Programs (SWIPs), and compute counterfactuals by weighted model counting (WMC) over a single transformed program. Under finite grounding and unique-supported-model assumptions, DeepSWIP is exact relative to the learned materialized FCM. The standard quotient-WMC form of ProbLog conditionals identifies active neural probabilities and explains intervention cleaning, calibration sensitivity, and rare-evidence instability. Experiments on MPI3D confirm the transformation against a DeepTwin construction against 12,000 queries, as predicted and a 2.14$\times$ inference speedup from avoiding the Twin's endogenous duplication. A SUMO HOV experiment shows that neural calibration degradation biases plug-in estimates, while a correctly scoped randomized-policy AIPW estimator removes most first-order bias for population mean and ATE estimands. Code is at https://github.com/saibib/deep_SWIP.
Chinese Translation
神经符号系统如 DeepProbLog 将神经感知与概率逻辑结合，但标准推理是关联性的。反事实推理还需要对干预和证据的因果语义。我们引入了 DeepSWIP，这是一个针对 DeepProbLog 程序的单世界反事实语义。通过神经物化，我们将固定上下文的神经谓词简化为普通的 ProbLog 选择，应用单世界干预程序（Single World Intervention Programs, SWIPs），并通过对单个变换程序进行加权模型计数（Weighted Model Counting, WMC）来计算反事实。在有限基础和唯一支持模型假设下，DeepSWIP 相对于学习到的物化 FCM 是精确的。ProbLog 条件的标准商-加权模型计数形式识别活跃的神经概率，并解释干预清理、校准敏感性和稀有证据不稳定性。MPI3D 上的实验确认了针对 DeepTwin 构造的变换，与 12,000 个查询的预测一致，并且通过避免 Twin 的内生重复实现了 2.14 倍的推理加速。SUMO HOV 实验表明，神经校准降级会偏倚插件估计，而正确范围的随机政策 AIPW 估计器则消除了大部分针对总体均值和平均处理效应（ATE）估计量的一阶偏倚。代码可在 https://github.com/saibib/deep_SWIP 获取。
View on arXiv Download PDF AI Translation
cs.AI / 71 / 2606.20529

LedgerAgent: Structured State for Policy-Adherent Tool-Calling Agents

LedgerAgent：遵循政策的工具调用代理的结构化状态
Uddin, Md Nayem, Saeidi, Amir, Blanco, Eduardo, Baral, Chitta
Abstract
Policy-adherent tool-calling agents in customer-service domains must maintain task states across turns while calling tools and obeying domain policies. Task states consist of relevant facts, identifiers, constraints, and conditions observed through user interaction and tool calls. In standard agents, task states are not represented separately. Observations, tool returns, and policy instructions are placed in the prompt, leaving agents to reconstruct the relevant states from the prompt each time they decide what to do next. This design makes state management implicit, creating two common failure modes. An agent may retrieve the right facts but later ground its decision in stale, missing, or incorrect information; and a syntactically valid tool call may still violate a domain policy that depends on the current task state. We introduce \textsc{LedgerAgent}, an inference-time method for tool-calling agents that maintains observed task states in a separate ledger and renders the states into the prompt. The ledger is also used to check state-dependent policy constraints before environment-changing tool calls are executed, blocking policy violations. Across four customer-service domains and a mixed panel of open- and closed-weight models, \textsc{LedgerAgent} improves average pass\textasciicircum{}k over a standard prompt-based tool-calling approach, with the largest gains under stricter multi-trial consistency metrics.
Chinese Translation
在客户服务领域，遵循政策的工具调用代理必须在调用工具和遵循领域政策的同时，跨回合维护任务状态。任务状态由通过用户交互和工具调用观察到的相关事实、标识符、约束和条件组成。在标准代理中，任务状态并未单独表示。观察结果、工具返回和政策指令被放置在提示中，代理每次决定下一步行动时都需从提示中重建相关状态。这种设计使得状态管理变得隐式，导致两种常见的失败模式。代理可能检索到正确的事实，但随后基于过时、缺失或不正确的信息做出决策；而一个语法上有效的工具调用仍可能违反依赖于当前任务状态的领域政策。我们提出了 extsc{LedgerAgent}，这是一种用于工具调用代理的推理时方法，它在一个单独的账本中维护观察到的任务状态，并将这些状态呈现到提示中。该账本还用于在执行改变环境的工具调用之前检查依赖于状态的政策约束，从而阻止政策违规。在四个客户服务领域和一个开放与封闭权重模型的混合面板中， extsc{LedgerAgent} 在标准基于提示的工具调用方法上提高了平均通过率（pass extasciicircum{}k），在更严格的多轮一致性指标下获得了最大的提升。
View on arXiv Download PDF AI Translation
cs.AI / 72 / 2606.20532

How Do Instructions Shape Speech? Cross-Attention Attribution for Style-Captioned Text-to-Speech

指令如何塑造语音？风格标注文本到语音的交叉注意力归因
Mathur, Nityanand, Sayed, Hamees, Madha, Wasim, Singh, Apoorv, Khurana, Sameer, Mandloi, Akshat, Kamath, Sudarshan
Abstract
Style-captioned text-to-speech systems use natural language to control voice characteristics, but how individual words influence acoustic output remains unclear. Understanding this is critical for diagnosing failure modes and improving controllability in expressive TTS. We propose cross-attention attribution for speech diffusion models, adapting the DAAM framework to the speech domain for the first time, and apply it to CapSpeech-TTS. Our method extracts per-token heatmaps across 25 layers and 24 ODE steps. We analyze 3,600 (style caption, text transcript) combinations comprising 120 style captions conditioning the generation of 30 text transcripts each, revealing how caption tokens shape waveforms. Results show: (1) style tokens have lower temporal variance than content/function tokens, confirming global conditioning; (2) style attention correlates with F0 and energy; (3) style conditioning peaks in early steps and deep layers; (4) attention entropy reaches its minimum at layer 17, co-occurring with the style importance peak, indicating maximal network selectivity at the most style-critical stage. This is the first study of how natural language influences cross-attention in speech diffusion models
Chinese Translation
风格标注的文本到语音系统使用自然语言来控制语音特征，但单个词汇如何影响声学输出仍不清楚。理解这一点对于诊断失败模式和提高表达性文本到语音（TTS）的可控性至关重要。我们提出了针对语音扩散模型的交叉注意力归因，首次将DAAM框架适配到语音领域，并将其应用于CapSpeech-TTS。我们的方法提取了25层和24个常微分方程（ODE）步骤的每个标记热图。我们分析了3600个（风格标注，文本转录）组合，包括120个风格标注，每个风格标注条件生成30个文本转录，揭示了标注标记如何塑造波形。结果显示：（1）风格标记的时间方差低于内容/功能标记，确认了全局条件；（2）风格注意力与基频（F0）和能量相关；（3）风格条件在早期步骤和深层次达到峰值；（4）注意力熵在第17层达到最低，与风格重要性峰值同时出现，表明在最关键的风格阶段网络选择性达到最大。这是首个研究自然语言如何影响语音扩散模型中交叉注意力的研究。
View on arXiv Download PDF AI Translation
cs.AI / 73 / 2606.20544

Toward Calibrated Mixture-of-Experts Under Distribution Shift

朝着在分布转移下校准的专家混合模型
Wong, Gina, Prinster, Drew, Saria, Suchi, Chellappa, Rama, Liu, Anqi
Abstract
Calibration aligns a model's predictive uncertainty with the frequencies of its empirical outcomes and is important for understanding and trusting reported probabilities. Recent work shows that enforcing calibration at the level of individual predictors can improve ensemble accuracy and calibration, with mixture-of-experts (MoE) models showing strong empirical improvements in particular; however, the conditions under which calibration helps MoE are not well understood. In this work, we study how MoE models behave under distribution shift, focusing on how routing mechanisms interact with expert-level calibration. We show that expert calibration is sufficient to ensure calibration of the overall model under a broad class of distribution shifts in hard-routed models, but is insufficient for calibrating soft-routed models. To address this, we propose an adversarial reweighting that penalizes calibration errors of the routed aggregate under distribution shift, and we demonstrate that it improves the accuracy-calibration tradeoff both on average and on difficult subsets of the data, across model classes, prediction tasks, and distribution shifts.
Chinese Translation
校准使模型的预测不确定性与其经验结果的频率对齐，对于理解和信任报告的概率至关重要。近期的研究表明，在单个预测器层面强制校准可以提高集成的准确性和校准，尤其是专家混合模型（Mixture-of-Experts, MoE）在这方面表现出显著的经验改进；然而，校准对 MoE 的帮助条件尚不明确。在本研究中，我们探讨了 MoE 模型在分布转移下的表现，重点研究路由机制如何与专家级校准相互作用。我们表明，在硬路由模型下，专家校准足以确保整体模型在广泛的分布转移类中的校准，但对于软路由模型则不足以实现校准。为了解决这个问题，我们提出了一种对抗重加权方法，该方法在分布转移下惩罚路由聚合的校准错误，并且我们证明它在模型类别、预测任务和分布转移的各个方面改善了准确性与校准之间的权衡，特别是在困难的数据子集上。
View on arXiv Download PDF AI Translation
计算语言学 (Computation and Language)
56
cs.CL / 1 / 2606.19344

Exposing the Unsaid: Visualizing Hidden LLM Bias through Stochastic Path Aggregation

揭示未言之事：通过随机路径聚合可视化隐藏的语言模型偏见
Pelossi, Matteo, Sevastjanova, Rita, Spinner, Thilo, El-Assady, Mennatallah
Abstract
Large Language Models (LLMs) exhibit representational and syntactic biases that are difficult to evaluate due to the stochastic nature of text generation. Standard auditing methods rely on a single output inspection or static automated metrics. These approaches obscure the underlying probability distributions and fail to capture biases hidden in lower-probability generation branches. This paper introduces TreeTracer, a visual analytics tool designed to evaluate LLM bias through aggregated comparison. Using a systematic perturbation analysis pipeline, the tool replaces ontology-defined terms in each input prompt, aggregates hundreds of stochastic generations into a syntax-aligned hierarchical structure, and then performs classification-aware node merging with an auxiliary language model. The resulting structure is visualized through a custom Sankey diagram. By juxtaposing two ontology-driven trees, the workspace enables direct comparison between semantic contexts and supports systematic bias detection. Because any visualization reflects only a subset of the model's learned behavior, the system further applies contrastive inference to compute and directly display counterfactual token probabilities across contexts, reducing the risk of misinterpreting the presence of bias. We validate the workspace through case studies comparing an unaligned baseline model GPT-2 XL against the constitutionally aligned Apertus models. The visual aggregation successfully exposes hidden representational harms, such as counterfactual pronoun suppression and conversational marginalization of individuals. A preliminary user study confirms that the aggregated comparative interface reduces cognitive load and effectively supports analysts in detecting systemic biases.
Chinese Translation
大型语言模型（LLMs）表现出难以评估的表征和句法偏见，这主要是由于文本生成的随机性。标准审计方法依赖于单一输出检查或静态自动化指标。这些方法掩盖了潜在的概率分布，未能捕捉隐藏在低概率生成分支中的偏见。本文介绍了TreeTracer，一种旨在通过聚合比较评估LLM偏见的可视分析工具。该工具使用系统的扰动分析流程，在每个输入提示中替换本体定义的术语，将数百个随机生成的结果聚合成一个语法对齐的层次结构，然后与辅助语言模型进行分类感知的节点合并。最终结构通过自定义的桑基图进行可视化。通过并列两个本体驱动的树，工作空间实现了语义上下文之间的直接比较，并支持系统的偏见检测。由于任何可视化仅反映模型学习行为的一个子集，该系统进一步应用对比推理计算并直接显示跨上下文的反事实标记概率，从而降低误解偏见存在的风险。我们通过案例研究验证了工作空间，比较了未对齐的基线模型GPT-2 XL与宪法对齐的Apertus模型。可视化聚合成功揭示了隐藏的表征伤害，例如反事实代词抑制和个体的对话边缘化。初步用户研究确认，聚合比较界面减少了认知负担，有效支持分析师检测系统性偏见。
View on arXiv Download PDF AI Translation
cs.CL / 2 / 2606.19345

Ensembles of Large Language Models for Identifying EQ-5D Studies in PubMed Based on Their Abstracts

基于摘要的PubMed中EQ-5D研究识别的大型语言模型集成
Rostam, Zhyar Rzgar K., Péntek, Márta, Czere, János Tibor, Zrubka, Zsombor, Gulácsi, László, Kertész, Gábor
Abstract
The rapid increase in scientific publications leads to the fact that manual study screening in systematic literature reviews (SLRs) is increasingly resource consuming, inefficient, and inconsistent. Classifying studies that clearly report health-related quality-of-life results, such as EQ-5D data, requires a high level of clinical interpretation and poses challenges for human reviewers. This study investigates the use of Google's Gemini and Gemma large language models (LLMs) in automating EQ-5D detection in the PubMed biomedical database based only on published abstracts. A multi-phase framework is proposed that integrates few-shot prompting, weight ensembling aggregation, and a soft stacking meta-classifier. Nine LLMs are evaluated on a dataset of PubMed studies manually labeled by two experts regarding EQ-5D reporting. The weighted ensemble of gemini-2.5-pro, gemma-3-12b, and gemma-3-27b obtained a 0.74 weighted F1-score and 0.74 accuracy, exceeding individually attained results. The ensembling of top-performing models improved the balance between precision and recall compared to individual models, while the soft stacking approach provided greater reliability and interpretability. Feature analysis shows that the probability results from the models are important in guiding the final predictions. The findings suggest that an ensemble-based LLM setup is a reliable and scalable approach for automating screening in biomedical research.
Chinese Translation
科学出版物的快速增加导致系统文献综述（SLRs）中的手动研究筛选变得越来越耗费资源、低效且不一致。对明确报告健康相关生活质量结果的研究进行分类，例如EQ-5D数据，要求较高的临床解读水平，并给人工审阅者带来了挑战。本研究探讨了使用谷歌的Gemini和Gemma大型语言模型（LLMs）在PubMed生物医学数据库中仅基于已发布摘要自动检测EQ-5D的可行性。我们提出了一个多阶段框架，集成了少量示例提示、加权集成聚合和软堆叠元分类器。对九个LLM进行了评估，数据集为由两位专家手动标注的PubMed研究，涉及EQ-5D报告。加权集成的gemini-2.5-pro、gemma-3-12b和gemma-3-27b模型获得了0.74的加权F1分数和0.74的准确率，超过了单独模型的结果。与单个模型相比，表现最佳模型的集成改善了精确度与召回率之间的平衡，而软堆叠方法提供了更大的可靠性和可解释性。特征分析表明，模型的概率结果在指导最终预测中具有重要意义。研究结果表明，基于集成的LLM设置是自动化生物医学研究筛选的可靠且可扩展的方法。
View on arXiv Download PDF AI Translation
cs.CL / 3 / 2606.19346

Disentangling Linguistic Relatedness from Task Alignment in Cross-Lingual Transfer

在跨语言迁移中解开语言相关性与任务对齐的关系
Ahmed, Ahmed Haj, Zhang, Ruochen, Grissom II, Alvin
Abstract
We study cross-lingual transfer by fine-tuning seven large language models (4B--671B parameters) on Arabic and evaluating zero-shot reading comprehension on Semitic languages and non-Semitic controls. Across dense and Mixture-of-Experts architectures, we find no evidence of Semitic-specific transfer: models with weak baselines improve dramatically across all languages, while strong-baseline models show only marginal gains regardless of language family. A chain-of-thought ablation reinforces this finding -- the same models that benefit most from fine-tuning benefit equally from inference-time reasoning, suggesting both mechanisms address task-format alignment rather than cross-lingual knowledge transfer.
Chinese Translation
我们通过对七个大型语言模型（参数规模从4亿到671亿）进行微调，研究跨语言迁移，并在阿拉伯语上评估对闪米特语言和非闪米特对照语言的零-shot 阅读理解。在密集型和专家混合架构中，我们没有发现闪米特特定迁移的证据：基线较弱的模型在所有语言上都有显著提升，而基线较强的模型无论语言家族如何仅显示出微小的增益。一项思维链消融实验进一步强化了这一发现——那些从微调中受益最多的模型在推理时的推理能力上也同样受益，表明这两种机制都关注任务格式对齐而非跨语言知识迁移。
View on arXiv Download PDF AI Translation
cs.CL / 4 / 2606.19347

How LLMs Fail and Generalize in RTL Coding for Hardware Design?

大型语言模型在硬件设计的RTL编码中如何失败与泛化？
Liu, Guan-Ting, Yang, Chao-Han Huck, Deng, Chenhui, Yu, Zhongzhi, Khailany, Brucek, Wang, Yu-Chiang Frank
Abstract
Translating sequential programming priors into the parallel temporal logic of hardware design remains a crucial bottleneck for large language models(LLM). To investigate this, we introduce a new error taxonomy grounded in problem solvability, inspired by cognitive theory. Our taxonomy categorizes failures into syntactic, semantic, solvable functional, and unsolvable functional types. Evaluations reveal a strict empirical ceiling on the VerilogEval benchmark, as frontier models plateau at a 90.8% initial pass rate. These plateaus are defined by unsolvable functional errors, exposing persistent knowledge gaps immune to test time compute scaling. Furthermore, we expose a striking surface convergence gap: optimization readily eliminates syntax errors but concurrently exacerbates deeper functional failures. Our findings demonstrate that alignment techniques merely teach models to compile. While repeated sampling strategies can patch solvable errors, register-transfer level(RTL) coding capacity remains strictly bounded by pretraining knowledge. Addressing challenges in the current LLM based hardware generation pipeline requires more studies in model reasoning rather than alignment interventions.
Chinese Translation
将顺序编程的先验知识转化为硬件设计的并行时序逻辑仍然是大型语言模型（LLM）面临的一个关键瓶颈。为此，我们引入了一种基于问题可解性的新的错误分类法，灵感来源于认知理论。我们的分类法将失败分为句法、语义、可解功能和不可解功能四种类型。评估结果显示，在VerilogEval基准测试中存在严格的经验上限，前沿模型的初始通过率停滞在90.8%。这些停滞是由不可解功能错误定义的，暴露了对测试时间计算扩展免疫的持续知识差距。此外，我们揭示了一个显著的表面收敛差距：优化能够轻松消除语法错误，但同时加剧了更深层次的功能失败。我们的研究结果表明，调整技术仅仅教会模型编译。虽然重复采样策略可以修补可解错误，但寄存器传输级（RTL）编码能力仍然严格受限于预训练知识。解决当前基于LLM的硬件生成管道中的挑战需要更多关于模型推理的研究，而不是调整干预。
View on arXiv Download PDF AI Translation
cs.CL / 5 / 2606.19348

DeepSeek-V4: Towards Highly Efficient Million-Token Context Intelligence

DeepSeek-V4：迈向高效的百万标记上下文智能
DeepSeek-AI, Xu, Anyi, Lin, Bangcai, Xue, Bing, Wang, Bingxuan, Xu, Bingzheng, Wu, Bochao, Zhang, Bowei, Lin, Chaofan, Dong, Chen, Ling, Chenchen, Lu, Chengda, Zhao, Chenggang, Deng, Chengqi, Hou, Chengyu, Xu, Chenhao, Shao, Chenze, Ruan, Chong, Sun, Conner, Dai, Damai, Guo, Daya, Yang, Dejian, Chen, Deli, Li, Donghao, Ji, Dongjie, Li, Erhang, Wei, Fang, Lin, Fangyun, Yuan, Fangzhou, Xia, Feiyu, Dai, Fucong, Hao, Guangbo, Chen, Guanting, Cao, Guoai, Meng, Guolai, Li, Guowei, Yu, Han, Zhang, Han, Xu, Hanwei, Li, Hao, Liang, Haofen, Zhang, Haoling, Luo, Haoming, Wei, Haoran, Yuan, Haotian, Zhang, Haowei, Luo, Haowen, Chen, Haoyu, Ji, Haozhe, Zhang, Hengqing, Ding, Honghui, Tang, Hongxuan, Cao, Huanqi, Gao, Huazuo, Qu, Hui, Zeng, Hui, Yang, J, Zhu, JQ, Luo, Jia, Song, Jia, Yu, Jia, Huang, Jialiang, Cai, Jialu, Liang, Jian, Zhou, Jiangting, Ye, Jiasheng, Li, Jiashi, Xu, Jiaxin, Hu, Jiewen, Yang, Jieyu, Chen, Jin, Yan, Jin, Chen, Jingchang, Zhou, Jingli, Xiang, Jingting, Yuan, Jingyang, Cheng, Jingyuan, Zhou, Jingzi, Zhu, Jinhua, Yu, Jiping, Sun, Joseph, Ran, Jun, Jiang, Junguang, Qiu, Junjie, Li, Junlong, Zheng, Junmin, Song, Junxiao, Dong, Kai, Gao, Kaige, Guan, Kang, Zhou, Kexing, Huang, Kezhao, Yu, Kuai, Wang, Lean, Zhang, Lecong, Wang, Lei, Xia, Leyi, Zhang, Li, Zhao, Liang, Guo, Lihua, Luo, Lingxiao, Ma, Linwang, Zhu, Linyan, Wang, Litong, Cai, Liyu, Zhang, Liyue, Chen, Longhao, Di, MS, Xu, MY, Mei, Max, Wang, Miaojun, Zhang, Mingchuan, Zhang, Minghua, Tang, Minghui, Li, Mingming, Zhou, Mingxu, Han, Minmin, Wang, Ning, Huang, Panpan, Wang, Panpan, Cong, Peixin, Wang, Peiyi, Zhang, Peng, Wang, Qiancheng, Zhu, Qihao, Li, Qingyang, Chen, Qinyu, Du, Qiushi, Jiang, Qiwei, Tian, Rui, Xu, Ruifan, Lu, Ruijie, Xu, Ruiling, Ge, Ruiqi, Zhang, Ruisong, Pan, Ruizhe, Wang, Runji, Chen, Runqian, Yin, Runqiu, Xu, Runxin, Shen, Ruomeng, Zhang, Ruoyu, Chen, Ruyi, Liu, SH, Lu, Shanghao, Sun, Shangmian, Zhou, Shangyan, Chen, Shanhuang, Cai, Shaofei, Nie, Shaoheng, Wu, Shaoqing, Chen, Shaoyuan, Hu, Shengding, Liu, Shengyu, Hu, Shiqiang, Ma, Shirong, Wang, Shiyu, Yu, Shuiping, Zhou, Shunfeng, Pan, Shuting, Yu, Shuying, Zhou, Songyang, Ni, Tao, Yun, Tao, Jin, Tian, Pei, Tian, Ye, Tian, Lin, Tianle, Ji, Tianran, Cui, Tianyi, Yue, Tianyuan, Yu, Tingting, Wang, Tun, Zhang, W, Xiao, WL, Zeng, Wangding, An, Wei, Zhao, Weilin, Liu, Wen, Liang, Wenfeng, Pang, Wenjie, Luo, Wenjing, Yao, Wenjing, Gao, Wenjun, Yang, Wenkai, Huang, Wenlve, Hou, Wenqing, Zhang, Wentao, Ma, Wenting, Gao, Xi, He, Xiang, Wang, Xiangwen, Wang, Xianzu, Bi, Xiao, Liu, Xiaodong, Wang, Xiaohan, Chen, Xiaokang, Zhang, Xiaokang, Nie, Xiaotao, Sun, Xiaowen, Wang, Xiaoxiang, Cheng, Xin, Liu, Xin, Xie, Xin, Liu, Xingchao, Liu, Xingchen, Yu, Xingkai, Li, Xingyou, Yang, Xinyu, Zhang, Xinyu, Chen, Xu, Wang, Xuanyu, Su, Xuecheng, Chen, Xueyin, Lin, Xuheng, Fu, Xuwei, Yan, YC, Wang, YQ, Ma, YW, Luo, Yanfeng, Zhang, Yang, Xu, Yanhong, Ma, Yanru, Huang, Yanwen, Li, Yao, Li, Yao, Xu, Yao, Zhao, Yao, Sun, Yaofeng, Wang, Yaohui, Qian, Yi, Shao, Yi, Yu, Yi, Zhang, Yichao, Ding, Yifan, Shi, Yifan, Wu, Yijia, Xiong, Yiliang, Ma, Yiling, He, Ying, Tang, Ying, Zhou, Ying, Luo, Yingjia, Zhong, Yinmin, Piao, Yishi, Wang, Yisong, Zhang, Yixiang, Chen, Yixiao, Tan, Yixuan, Wei, Yixuan, Ma, Yiyang, Liu, Yiyuan, Yang, Yonglun, Guo, Yongqiang, Wu, Yongtong, Wu, Yu, Li, YuKun, Cheng, Yuan, Ou, Yuan, Xu, Yuanfan, Li, Yuanhao, Wang, Yuduan, Yang, Yuehan, Xu, Yuer, Wu, Yuhan, Meng, Yuhao, Zou, Yuheng, Zha, Yukun, Xiong, Yunfan, Chen, Yupeng, Lin, Yuping, Cao, Yuqian, Wang, Yuqian, Zhang, Yushun, Yan, Yuting, Lin, Yutong, Gu, Yuxian, Luo, Yuxiang, You, Yuxiang, Liu, Yuxuan, Zhou, Yuxuan, Zhou, Yuyang, Huang, Yuzhen, Wu, ZF, Wang, Zehao, Zhao, Zehua, Ren, Zehui, Zhang, Zekai, Sha, Zhangli, Fu, Zhe, Ju, Zhe, Xu, Zhean, Xie, Zhenda, Zhang, Zhengyan, Gao, Zheren, Hao, Zhewen, Gou, Zhibin, Ma, Zhicheng, Yan, Zhigang, Shao, Zhihong, Huang, Zhixian, Chen, Zhixuan, Wu, Zhiyu, Ren, Zhizhou, Wu, Zhongyu, Li, Zhuoshu, Zhang, Zhuping, Xu, Zian, Wang, Zihao, Qu, Zihua, Gu, Zihui, Zhu, Zijia, Li, Zilin, Zhang, Zipeng, Xie, Ziwei, Gao, Ziyi, Wan, Ziyi, Pan, Zizheng, Yao, Zongqing
Abstract
We present a preview version of DeepSeek-V4 series, including two strong Mixture-of-Experts (MoE) language models -- DeepSeek-V4-Pro with 1.6T parameters (49B activated) and DeepSeek-V4-Flash with 284B parameters (13B activated) -- both supporting a context length of one million tokens. DeepSeek-V4 series incorporate several key upgrades in architecture and optimization: (1) a hybrid attention architecture that combines Compressed Sparse Attention (CSA) and Heavily Compressed Attention (HCA) to improve long-context efficiency; (2) Manifold-Constrained Hyper-Connections (mHC) that enhance conventional residual connections; (3) and the Muon optimizer for faster convergence and greater training stability. We pre-train both models on more than 32T diverse and high-quality tokens, followed by a comprehensive post-training pipeline that unlocks and further enhances their capabilities. DeepSeek-V4-Pro-Max, the maximum reasoning effort mode of DeepSeek-V4-Pro, redefines the state-of-the-art for open models, outperforming its predecessors in core tasks. Meanwhile, DeepSeek-V4 series are highly efficient in long-context scenarios. In the one-million-token context setting, DeepSeek-V4-Pro requires only 27% of single-token inference FLOPs and 10% of KV cache compared with DeepSeek-V3.2. This enables us to routinely support one-million-token contexts, thereby making long-horizon tasks and further test-time scaling more feasible. The model checkpoints are available at https://huggingface.co/collections/deepseek-ai/deepseek-v4.
Chinese Translation
我们展示了DeepSeek-V4系列的预览版本，包括两个强大的混合专家（Mixture-of-Experts, MoE）语言模型——具有1.6万亿参数（激活49亿）的DeepSeek-V4-Pro和具有2840亿参数（激活130亿）的DeepSeek-V4-Flash——这两个模型均支持一百万标记的上下文长度。DeepSeek-V4系列在架构和优化方面进行了几项关键升级：（1）一种混合注意力架构，结合了压缩稀疏注意力（Compressed Sparse Attention, CSA）和重压缩注意力（Heavily Compressed Attention, HCA），以提高长上下文的效率；（2）流形约束超连接（Manifold-Constrained Hyper-Connections, mHC），增强了传统的残差连接；（3）Muon优化器，用于更快的收敛和更大的训练稳定性。我们在超过32万亿多样化和高质量的标记上对这两个模型进行了预训练，随后进行了全面的后训练流程，以解锁并进一步增强它们的能力。DeepSeek-V4-Pro-Max是DeepSeek-V4-Pro的最大推理努力模式，重新定义了开放模型的最新技术，超越了其前身在核心任务中的表现。同时，DeepSeek-V4系列在长上下文场景中表现出高度的效率。在一百万标记的上下文设置中，与DeepSeek-V3.2相比，DeepSeek-V4-Pro仅需27%的单标记推理FLOPs和10%的KV缓存。这使我们能够常规支持一百万标记的上下文，从而使长时间跨度的任务和进一步的测试时间扩展变得更加可行。模型检查点可在https://huggingface.co/collections/deepseek-ai/deepseek-v4获取。
View on arXiv Download PDF AI Translation
cs.CL / 6 / 2606.19349

Where to Place the Query? Unveiling and Mitigating Positional Bias in In-Context Learning for Diffusion LLMs via Decoding Dynamics

查询应放置在哪里？揭示并缓解扩散大语言模型中上下文学习的位置信息偏差
Li, Zhengheng, Li, Panrui, Liu, Xuyang, Xia, Puzhi
Abstract
While In-Context Learning (ICL) is extensively studied in Autoregressive (AR) LLMs, its mechanism within Diffusion Large Language Models (dLLMs) remains largely unexplored. Unlike AR models restricted by unidirectional causal masking, dLLMs intrinsically utilize bidirectional attention, offering extensive spatial flexibility for query placement. Unfortunately, current practices conventionally inherit AR-style trailing-query templates, often overlooking the structural paradigm shift. This paper presents a comprehensive analysis unveiling that query position is actually a first-order variable in dLLMs. Through empirical decoupling, we demonstrate that positional variance impacts generation quality on par with example semantic quality. Internally, this positional sensitivity stems from a spatial ``Recency Effect'' in attention flow and task-dependent shifts in decoding trajectories. To mitigate this instability without ground-truth labels, we reveal that traditional single-step confidence ($C_{decoded}$) fails in dLLMs. Instead, we propose Average Confidence ($\overline{C}$), a novel metric tracking the iterative decoding process. By establishing the foundational spatial ICL baselines, we introduce Auto-ICL, a training-free adaptive routing strategy that dynamically optimizes query placement, robustly approaching oracle performance across heterogeneous reasoning and perception tasks.
Chinese Translation
尽管上下文学习（In-Context Learning, ICL）在自回归（Autoregressive, AR）大语言模型中得到了广泛研究，但其在扩散大语言模型（Diffusion Large Language Models, dLLMs）中的机制仍然 largely 未被探索。与受单向因果掩蔽限制的 AR 模型不同，dLLMs 本质上利用双向注意力，提供了查询放置的广泛空间灵活性。不幸的是，当前的实践通常继承了 AR 风格的尾部查询模板，往往忽视了结构范式的转变。本文提供了全面的分析，揭示查询位置实际上是 dLLMs 中的一个一阶变量。通过经验解耦，我们证明了位置变化对生成质量的影响与示例语义质量相当。在内部，这种位置敏感性源于注意力流中的空间“近期效应”（Recency Effect）和解码轨迹中的任务依赖性变化。为了在没有真实标签的情况下缓解这种不稳定性，我们揭示了传统的单步置信度（$C_{decoded}$）在 dLLMs 中的失效。相反，我们提出了平均置信度（$ar{C}$），这是一种跟踪迭代解码过程的新指标。通过建立基础的空间 ICL 基线，我们引入了 Auto-ICL，这是一种无训练的自适应路由策略，能够动态优化查询放置，在异构推理和感知任务中稳健地接近 oracle 性能。
View on arXiv Download PDF AI Translation
cs.CL / 7 / 2606.19350

Pruning via Causal Attribution Preserves Reasoning Performance in Large Language Models

通过因果归因修剪在大型语言模型中保持推理性能
Sheth, Amogh, Assefa, Biruk, Huang, Yi Wen, Lin, Andrew, Ge, Yuhao
Abstract
Large language models (LLMs) excel at multi-step reasoning but incur substantial inference cost. We introduce Causal Attribution Pruning (CAP), a training-free method that identifies critical attention heads by measuring their causal impact on reasoning tasks and uses these head-level scores to guide fine-grained weight pruning. For each attention head, CAP estimates the expected performance degradation when the head is masked during forward passes on a small calibration set of reasoning problems. These causal scores are then converted into weight-level importance values for the corresponding projection matrices. Unlike magnitude-only or activation-based criteria, CAP's interventional measurement directly captures each head's functional contribution, yielding relative accuracy gains of up to 61% over Wanda on ARC-Challenge at 20% sparsity. We evaluate CAP on GSM8K, StrategyQA, and ARC-Challenge using Llama-3-8B-Instruct and Mistral-7B-Instruct at 10%, 20%, and 50% sparsity. At moderate sparsity (10-20%), CAP improves over Wanda in most model-benchmark configurations. with especially large gains on ARC-Challenge for Llama-3. Our results suggest that attention-head-level causal attribution can better preserve reasoning performance on downstream benchmarks than correlational pruning criteria at equivalent sparsity, while remaining limited by coarse MLP attribution at 50% sparsity.
Chinese Translation
大型语言模型（LLMs）在多步骤推理方面表现出色，但推理成本相当高。我们提出了一种无训练方法——因果归因修剪（Causal Attribution Pruning, CAP），通过测量注意力头对推理任务的因果影响来识别关键的注意力头，并利用这些头级得分指导细粒度的权重修剪。对于每个注意力头，CAP在一小组推理问题的校准集上估计当该头在前向传播中被屏蔽时预期的性能下降。这些因果得分随后被转换为相应投影矩阵的权重级重要性值。与仅基于幅度或激活的标准不同，CAP的干预测量直接捕捉每个头的功能贡献，在20%稀疏度下，相较于Wanda在ARC-Challenge上实现了高达61%的相对准确性提升。我们在GSM8K、StrategyQA和ARC-Challenge上评估了CAP，使用Llama-3-8B-Instruct和Mistral-7B-Instruct，在10%、20%和50%稀疏度下进行测试。在中等稀疏度（10-20%）下，CAP在大多数模型-基准配置中优于Wanda，尤其是在Llama-3的ARC-Challenge上取得了显著的提升。我们的结果表明，注意力头级的因果归因在相同稀疏度下能够更好地保持下游基准的推理性能，相较于相关性修剪标准，但在50%稀疏度时仍受到粗糙的多层感知机归因的限制。
View on arXiv Download PDF AI Translation
cs.CL / 8 / 2606.19351

Detecting Hallucinations for Large Language Model-based Knowledge Graph Reasoning

基于大型语言模型的知识图谱推理中的幻觉检测
Zhu, Xinyan, Liu, Yaoqi, Gao, Yue, Ma, Huadong, Yang, Cheng, Shi, Chuan
Abstract
Knowledge graph (KG) reasoning infers new knowledge from existing facts and is widely applied in question answering, recommendation, and decision support. With the rapid development of large language models (LLMs), LLM-based KG reasoning frameworks have become increasingly popular by leveraging retrieved KG information. However, hallucinations in LLMs remain a critical issue. Even when relevant KG knowledge is incorporated, models may still generate incorrect outputs, leading to misinformation and unreliable decisions. Existing hallucination detection methods either focus on LLM internal states or verify consistency with retrieved contexts, but both overlook the structural information in KGs, resulting in suboptimal performance. To address this gap, we propose LUCID, the first halLUcination deteCtIon method for LLM-based knowleDge graph reasoning frameworks. LUCID jointly leverages LLM attention scores, KG semantics, and structural information. Specifically, it extracts node and edge features from attention scores and semantic similarities, and integrates them with KG structure using a graph neural network. We also construct manually annotated benchmark datasets for evaluation. Experiments on nine datasets show that LUCID achieves state of the art performance compared to 15 baselines.
Chinese Translation
知识图谱（KG）推理从现有事实中推断新知识，广泛应用于问答、推荐和决策支持。随着大型语言模型（LLMs）的快速发展，基于LLM的KG推理框架通过利用检索到的KG信息变得越来越流行。然而，LLM中的幻觉问题仍然是一个关键问题。即使相关的KG知识被纳入，模型仍可能生成不正确的输出，导致错误信息和不可靠的决策。现有的幻觉检测方法要么关注LLM的内部状态，要么验证与检索上下文的一致性，但两者都忽视了KG中的结构信息，导致性能不佳。为了解决这一问题，我们提出了LUCID，这是第一个针对基于LLM的知识图谱推理框架的幻觉检测方法。LUCID联合利用LLM的注意力分数、KG语义和结构信息。具体而言，它从注意力分数和语义相似性中提取节点和边特征，并通过图神经网络将其与KG结构整合。我们还构建了手动标注的基准数据集以供评估。在九个数据集上的实验表明，LUCID相比于15个基线方法达到了最先进的性能。
View on arXiv Download PDF AI Translation
cs.CL / 9 / 2606.19352

Sign-Language Datasets at Scale: A Comprehensive Survey on Resources, Benchmarks, and Annotation Standards

大规模手语数据集：资源、基准和注释标准的综合调查
Ni, Yiming, Cheng, Zhi-Qi, Li, Jiayu, Cheng, Wei
Abstract
Sign languages are expressive visual languages used by Deaf and Hard-of-Hearing (DHH) communities. Despite substantial progress in sign-language recognition, translation, and production, advances remain constrained by fragmented datasets, inconsistent annotations, and limited linguistic coverage. Existing benchmarks often fail to reflect real-world communication needs, and systematic analyses of these limitations remain limited. In this survey, we present a comprehensive index of sign-language datasets, covering 120 resources across 35 sign languages. We analyze key challenges such as modality imbalance, annotation granularity, and signer bias, and outline considerations for future dataset design. We also introduce a 24-field Sign-Language Datasheet and release a public GitHub repository (https://github.com/Ginqwerty/Open-Sign-Language) to support standardized documentation and reproducible evaluation. Overall, our work provides a unified and practical foundation for developing inclusive, robust, and scalable sign-language technologies in real-world applications.
Chinese Translation
手语是聋人和听力受损（DHH）社区使用的富有表现力的视觉语言。尽管在手语识别、翻译和生成方面取得了显著进展，但由于数据集碎片化、注释不一致和语言覆盖有限，进展仍然受到限制。现有基准往往无法反映现实世界的交流需求，对这些局限性的系统分析也相对较少。在本调查中，我们提供了一个全面的手语数据集索引，涵盖了35种手语中的120个资源。我们分析了诸如模态不平衡、注释粒度和签署者偏见等关键挑战，并概述了未来数据集设计的考虑因素。我们还引入了一个包含24个字段的手语数据表，并发布了一个公共GitHub存储库（https://github.com/Ginqwerty/Open-Sign-Language），以支持标准化文档和可重复评估。总体而言，我们的工作为在现实应用中开发包容性、稳健性和可扩展的手语技术提供了统一而实用的基础。
View on arXiv Download PDF AI Translation
cs.CL / 10 / 2606.19353

Quantifying Aleatoric Uncertainty of In-Context Learning for Robust Measure of LLM Prediction Confidence

量化上下文学习中的随机不确定性以增强大语言模型预测置信度的稳健性
Chung, Jinseok, Song, Minkyoung, Jung, Hyunji, Lee, Namhoon
Abstract
In-Context Learning (ICL) allows LLMs to adapt to new tasks from a few demonstrations, but its reliability remains a concern: predictions are highly sensitive to both prompt design and the model's ability to understand the context, obscuring whether failures arise from data properties or model limitations. Uncertainty decomposition-separating aleatoric from epistemic sources-is particularly crucial in this setting, yet existing methods, designed for standard generation tasks, fail to capture the unique dynamics of ICL. To address this, we introduce a concept of self-function vectors, built upon Bayesian views and the mechanistic interpretability of ICL. These vectors leverage internal model representations to model the latent concept learned during in-context prompting, thereby enabling a direct estimation of aleatoric uncertainty within a Bayesian framework and circumventing the reliance on brittle input or decoding manipulations. Given the lack of established benchmarks and suitable evaluation protocols, we also propose the first and rigorous evaluation protocol, in which data is manipulated in controlled ways so as to quantify aleatoric uncertainty precisely and separately from epistemic uncertainty. With this new evaluation framework, initially grounded in synthetic tasks for conceptual development and subsequently extended to real-world datasets, we show that our proposed methodology can measure uncertainty of LLM predictions made under ICL more reliably than existing alternative methods. Moreover, we show it can be used as a practical tool for trustworthy-related applications, such as hallucination detection. Our findings pave a new direction for connecting the quantitative view of uncertainty with the mechanistic understanding of model behavior.
Chinese Translation
上下文学习（In-Context Learning, ICL）使大语言模型（LLMs）能够通过少量示例适应新任务，但其可靠性仍然令人担忧：预测对提示设计和模型理解上下文的能力高度敏感，模糊了失败是由于数据特性还是模型局限性造成的。在这种情况下，不确定性分解——将随机不确定性与认知不确定性分开——尤为重要，然而现有方法主要针对标准生成任务，未能捕捉到ICL的独特动态。为了解决这一问题，我们引入了一种自函数向量的概念，基于贝叶斯视角和ICL的机制可解释性。这些向量利用内部模型表示来建模在上下文提示过程中学习到的潜在概念，从而在贝叶斯框架内直接估计随机不确定性，避免了对脆弱输入或解码操作的依赖。鉴于缺乏既定基准和合适的评估协议，我们还提出了第一个严格的评估协议，在该协议中，数据以受控方式进行操控，以精确量化随机不确定性并与认知不确定性分开。通过这一新的评估框架，最初基于合成任务进行概念开发，随后扩展到真实世界数据集，我们展示了所提方法在ICL下对LLM预测的不确定性测量比现有替代方法更为可靠。此外，我们还表明，该方法可以作为可信相关应用的实用工具，例如幻觉检测。我们的研究成果为将不确定性的定量视角与模型行为的机制理解相连接开辟了新的方向。
View on arXiv Download PDF AI Translation
cs.CL / 11 / 2606.19354

Granularity-Regulated Adaptive Computational Efficiency for Optimal Verification in Test-Time Scaling

基于粒度调节的自适应计算效率在测试时间缩放中的最优验证
Krasniqi, Ardit, Vejsiu, Luan, Dervishi, Elira
Abstract
Test-time scaling (TTS) has emerged as a powerful paradigm for improving the reasoning performance of large language models (LLMs) by investing additional compute at inference time. A central component of TTS is the \emph{verifier}, which selects or scores candidate solutions to guide the search process. While prior work has explored the benefit of verification, a fundamental question remains underexplored: \emph{what is the optimal granularity of verification under a given compute budget?} Coarse-grained outcome reward models (ORMs) and fine-grained process reward models (PRMs) represent two extremes, yet neither alone achieves compute-optimality across all regimes. In this paper, we establish a unified theoretical framework, called \textbf{GRACE} (\underline{G}ranularity-\underline{R}egulated \underline{A}daptive \underline{C}omputational \underline{E}fficiency), that characterizes the optimal verification granularity as an explicit function of problem difficulty, verifier accuracy, and compute budget. We prove that there exists a phase transition: fine-grained verification dominates when either the compute budget is large or the problem is hard, whereas coarse-grained verification is preferred in the low-budget, easy-problem regime. Our theory unifies Best-of-$N$, beam search, and step-level MCTS within a single Pareto-optimality framework, and motivates an adaptive granularity strategy that provably achieves the compute-performance Pareto frontier. Empirical results on MATH-500, GSM8K, and AIME benchmarks corroborate all four theoretical claims, with our adaptive strategy outperforming fixed-granularity baselines by up to 3.1\% accuracy at matched compute.
Chinese Translation
测试时间缩放（TTS）已成为一种强大的范式，通过在推理时投入额外的计算资源来提高大型语言模型（LLMs）的推理性能。TTS的一个核心组成部分是 extit{验证器}，它选择或评分候选解决方案以指导搜索过程。尽管之前的研究探讨了验证的好处，但一个基本问题仍未得到充分研究： extit{在给定的计算预算下，验证的最优粒度是什么？} 粗粒度结果奖励模型（ORMs）和细粒度过程奖励模型（PRMs）代表了两个极端，但单独使用它们在所有情况下都无法实现计算最优性。本文建立了一个统一的理论框架，称为 extbf{GRACE}（ extit{G}ranularity- extit{R}egulated extit{A}daptive extit{C}omputational extit{E}fficiency），该框架将最优验证粒度表征为问题难度、验证器准确性和计算预算的显式函数。我们证明了存在一个相变：当计算预算较大或问题较难时，细粒度验证占优势，而在低预算、简单问题的情况下，粗粒度验证更为优选。我们的理论将Best-of-$N$、束搜索和步级MCTS统一在一个单一的帕累托最优框架内，并激励了一种自适应粒度策略，该策略在理论上证明能够实现计算-性能的帕累托前沿。在MATH-500、GSM8K和AIME基准上的实证结果证实了所有四个理论主张，我们的自适应策略在匹配计算条件下的准确率比固定粒度基线高出最多3.1%。
View on arXiv Download PDF AI Translation
cs.CL / 12 / 2606.19356

Trustworthy Multi-Agent Systems: Mitigating Semantic Drift with the Argent Signaling Protocol

可信赖的多智能体系统：通过 Argent 信号协议减轻语义漂移
Sharma, Anantha
Abstract
When multi-agent LLM systems produce bad answers, not all failures are equal: some answers are grounded in the right material but incomplete, while others are simply ungrounded and should be stopped. Current retry strategies treat both cases identically (try again and hope for the best), leaving human supervisors unable to tell whether a retry was warranted or whether the system should have halted instead. We introduce the Argent Signaling Protocol (ASP), a compact machine-readable header that accompanies every AI-generated response with structured quality signals: certainty (@C), grounding (@G), stochasticity (@S), and an assumption index that classifies the evidentiary basis of each claim. These signals enable a controller to distinguish repairable failures from containment failures and route each case differently. We evaluate ASP in two modes. In standalone mode, a 27-question document-grounded QA benchmark over the Array BioPharma/Ono license agreement compares baseline prompts against ASP-instrumented controller actions across three local GGUF models. On Qwen~(0.8B), ASP improves pass rate from 11.1% to 33.3% and mean term coverage from 36.7% to 65.4%; on Dobby~(8B), ASP produces 4 fail-to-pass recoveries, raising pass rate from 33.3% to 44.4%; on SmolLM3~(3B), ASP alternates between repair and containment per question. Aggregate improvement is meaningful (12/81 to 21/81 passes). In multi-agent mode, an ASP sidecar sits between a retrieval agent and a downstream decision agent; the sidecar blocks 100% of ungrounded upstream outputs from reaching the downstream agent (24/27 blocked, 0 ungrounded propagations).
Chinese Translation
当多智能体大语言模型（LLM）系统产生错误答案时，并非所有失败都是相同的：有些答案基于正确的材料但不完整，而另一些则完全没有依据，应该被停止。目前的重试策略对这两种情况采取相同的处理方式（重试并希望得到更好的结果），这使得人类监督者无法判断重试是否合理，或者系统是否应该停止。我们提出了 Argent 信号协议（ASP），这是一种紧凑的机器可读头部，伴随每个 AI 生成的响应，提供结构化的质量信号：确定性（@C）、基础性（@G）、随机性（@S）以及一个假设指数，用于分类每个主张的证据基础。这些信号使控制器能够区分可修复的失败和需要遏制的失败，并对每种情况采取不同的处理方式。我们在两种模式下评估了 ASP。在独立模式下，我们对 Array BioPharma/Ono 许可协议进行了一项包含 27 个问题的文档基础问答基准测试，比较了基线提示与 ASP 工具化控制器操作在三个本地 GGUF 模型上的表现。在 Qwen~(0.8B) 上，ASP 将通过率从 11.1% 提高到 33.3%，平均术语覆盖率从 36.7% 提高到 65.4%；在 Dobby~(8B) 上，ASP 产生了 4 次失败恢复，将通过率从 33.3% 提高到 44.4%；在 SmolLM3~(3B) 上，ASP 在每个问题之间交替进行修复和遏制。总体改善是显著的（从 12/81 提高到 21/81）。在多智能体模式下，ASP 侧车位于检索代理和下游决策代理之间；侧车阻止了 100% 无依据的上游输出到达下游代理（24/27 被阻止，0 次无依据传播）。
View on arXiv Download PDF AI Translation
cs.CL / 13 / 2606.19468

Characterizing Narrative Content in Web-scale LLM Pretraining Data

网络规模大语言模型预训练数据中的叙事内容特征
Johnson, Teagan, Ash, Elliott, Piper, Andrew, Antoniak, Maria
Abstract
The narrative composition of web-scale LLM pretraining corpora remains largely unexplored even though narrative is a fundamental mode of human communication. We present the first fine-grained study of narrative features in Dolma, a 3-trillion-token open pretraining corpus. Drawing on narrative theory, we design a framework spanning three core narrative elements (agency, setting, and events) operationalized as 11 interpretable dimensions. After sampling and annotating a diverse set of 400 passages, we finetune and validate NarraBERT, a RoBERTa-based model for fine-grained narrative prediction. We apply NarraBERT to 3M passages, resulting in a new dataset, NarraDolma. We find (i) narrative structure is measurable at scale across extremely heterogeneous data, (ii) we uncover a continuous, multidimensional narrative structure underlying web text, and (iii) narrative qualities are unequally distributed across pretraining sources and topics in ways that current curation practices neither measure nor account for. Our framework, dataset, and analyses provide a foundation for understanding how narrative qualities are distributed in LLM pretraining data and for studying how data composition affects narrative reasoning tasks. We publicly release NarraDolma and NarraBERT.
Chinese Translation
尽管叙事是人类交流的基本方式，网络规模大语言模型（LLM）预训练语料库的叙事构成仍然在很大程度上未被探索。我们首次对Dolma（一个包含3万亿标记的开放预训练语料库）中的叙事特征进行了细致的研究。基于叙事理论，我们设计了一个框架，涵盖三个核心叙事元素（代理、背景和事件），并将其操作化为11个可解释的维度。在对400段多样化文本进行抽样和注释后，我们微调并验证了NarraBERT，一个基于RoBERTa的细粒度叙事预测模型。我们将NarraBERT应用于300万段文本，生成了一个新数据集NarraDolma。我们的研究发现：（i）叙事结构可以在极其异质的数据中进行大规模测量；（ii）我们揭示了潜藏于网络文本中的连续多维叙事结构；（iii）叙事特征在预训练来源和主题中的分布不均，而当前的策展实践既未测量也未考虑这些差异。我们的框架、数据集和分析为理解LLM预训练数据中叙事特征的分布以及研究数据构成如何影响叙事推理任务提供了基础。我们将NarraDolma和NarraBERT公开发布。
View on arXiv Download PDF AI Translation
cs.CL / 14 / 2606.19544

Reliability without Validity: A Systematic, Large-Scale Evaluation of LLM-as-a-Judge Models Across Agreement, Consistency, and Bias

没有有效性的可靠性：对 LLM-as-a-Judge 模型在一致性、连贯性和偏见方面的系统性大规模评估
Norman, Justin D., Rivera, Michael U., Hughes, D. Alex
Abstract
LLM-as-a-Judge has become the dominant evaluation paradigm for language models, but judge validation in practice relies on exact-match agreement, a metric that does not correct for chance and systematically overstates discriminative ability. We present the largest systematic evaluation of LLM-as-a-Judge to date: 21 judges from nine providers across MT-Bench, JudgeBench, and RewardBench, evaluated under three protocols (agreement, consistency, bias audit) over 118 runs and approximately 541,000 individual judgments. Four findings emerge, consistent across the full cohort, including the April 2026 frontier: kappa deflation between exact match and Cohen's kappa is universal (33--41 pp on MT-Bench), judge rankings shift by up to 14 positions across benchmarks, high test--retest reliability (>0.95) coexists with severe position bias (>0.10) in two production-deployed judges (instantiating a consistency--bias paradox), and verbosity bias is small (<0.011) across our cohort under a single pairwise rubric. We distill these into a Minimum Viable Validation Protocol.
Chinese Translation
LLM-as-a-Judge 已成为语言模型的主导评估范式，但实际中的评审验证依赖于精确匹配一致性，这一指标未能纠正偶然性，并系统性地夸大了区分能力。我们展示了迄今为止最大的 LLM-as-a-Judge 系统评估：来自九个提供者的 21 位评审在 MT-Bench、JudgeBench 和 RewardBench 上进行评估，采用三种协议（一致性、连贯性、偏见审计），共进行了 118 次测试，约 541,000 次个体判断。四个发现贯穿整个样本，包括 2026 年 4 月的前沿：精确匹配与 Cohen's kappa 之间的 kappa 衰减是普遍存在的（在 MT-Bench 上为 33-41 个百分点），评审排名在基准之间最多变化 14 个位置，高测试-重测可靠性（>0.95）与两个生产部署的评审中严重的位置偏见（>0.10）共存（体现了一种一致性-偏见悖论），而在我们的样本中，冗长偏见在单一成对标准下较小（<0.011）。我们将这些提炼为最低可行验证协议。
View on arXiv Download PDF AI Translation
cs.CL / 15 / 2606.19552

LaViSA: A Language and Vision Structural Ambiguity Benchmark

LaViSA：语言与视觉结构歧义基准
Sangmyeong, Lee, Inadumi, Shun, Yoshino, Koichiro
Abstract
Structural ambiguity arises when a single sentence admits multiple valid interpretations due to its syntactic structure, posing a fundamental challenge for language understanding. Visual scenes serve as useful cues for resolving such ambiguity, and Vision and Language Models (VLMs) need to be capable of deriving possible semantic interpretations from visual scenes. We introduce Language and Vision Structural Ambiguity (LaViSA), a benchmark designed to evaluate the ability of VLMs to resolve structural ambiguity leveraging visual scenes. LaViSA consists of ambiguous sentences, their disambiguated sentences, and corresponding images of these disambiguated sentences across seven ambiguity categories. Using LaViSA, we conduct a comprehensive evaluation of diverse VLMs, including both proprietary and open-source models with varying parameter scales and reasoning capabilities. Experimental results show that although recent VLMs can leverage visual scenes to resolve structural ambiguity to a some extent, they still struggle with certain ambiguity types and visually subtle semantic distinctions, indicating remaining limitations in resolving structural ambiguity using visual scenes.
Chinese Translation
结构歧义是指由于句子的句法结构而产生的多种有效解释，这对语言理解构成了根本性的挑战。视觉场景作为解决这种歧义的有用线索，视觉与语言模型（Vision and Language Models, VLMs）需要能够从视觉场景中推导出可能的语义解释。我们介绍了语言与视觉结构歧义（Language and Vision Structural Ambiguity, LaViSA），这是一个旨在评估 VLMs 利用视觉场景解决结构歧义能力的基准。LaViSA 包含歧义句子、其消歧句子以及这些消歧句子的对应图像，涵盖七个歧义类别。通过 LaViSA，我们对多种 VLMs 进行了全面评估，包括不同参数规模和推理能力的专有模型和开源模型。实验结果表明，尽管近期的 VLMs 在一定程度上能够利用视觉场景解决结构歧义，但它们在某些歧义类型和视觉上微妙的语义区分方面仍然存在困难，这表明在利用视觉场景解决结构歧义方面仍然存在局限性。
View on arXiv Download PDF AI Translation
cs.CL / 16 / 2606.19591

A BART-based approach with hierarchical strategy for Vietnamese abstractive multi-document summarization

基于BART的越南语抽象多文档摘要的分层策略方法
Xuan, Vu Nguyen Nguyen, Quang, Huy Ngo
Abstract
In this technical report, we focus on solving the challenge of Vietnamese multi-document abstractive summarization, introduced in the International Workshop on Vietnamese Language and Speech Processing (VLSP) 2022. We choose to follow the popular hierarchical approach, i.e. condensing each document followed by aggregation and summarization. We propose a novel yet simple strategy to shorten documents that is driven by the golden summary, thus ensuring high correlation between stages of the hierarchical approach. Our method achieves a ROUGE2-F1 score of 0.2468 on the VLSP's public test set, and can produce fluent and concise summaries. Additionally, we utilize external sources for extra data, which greatly enhances the quantity of data for Vietnamese multi-document summarization. The additional data is made available for the community.
Chinese Translation
在本技术报告中，我们专注于解决在2022年国际越南语言与语音处理研讨会（VLSP）中提出的越南语多文档抽象摘要的挑战。我们选择遵循流行的分层方法，即先对每个文档进行压缩，然后进行聚合和摘要。我们提出了一种新颖而简单的文档缩短策略，该策略由黄金摘要驱动，从而确保分层方法各阶段之间的高度相关性。我们的方法在VLSP的公共测试集上达到了0.2468的ROUGE2-F1分数，并能够生成流畅且简洁的摘要。此外，我们利用外部来源获取额外数据，这大大增强了越南语多文档摘要的数据量。这些额外数据也将提供给社区。
View on arXiv Download PDF AI Translation
cs.CL / 17 / 2606.19625

Where Does Social Reasoning Come From? Capability Provenance in Language Models

社会推理来自何处？语言模型中的能力来源
Matlin, Glenn, Chakraborty, Chandreyi, Eom, Saehee, Okamoto, Mika, Castilla, Rayan, Jaburi, Louis, Deng, Alvin, Min, Taywon, Quirke, Lucia, Biderman, Stella, Riedl, Mark
Abstract
We use training-data attribution as an interpretable tool for capability discovery, mapping which regions of the pretraining corpus support social-reasoning versus STEM-reasoning in OLMo3-7B. Training-data attribution measures how strongly each training document influences a model's predictions on a benchmark, but document-level scores are too noisy to identify which corpus regions support which capabilities, and prior work has emphasized factual knowledge rather than reasoning. We compute gradient-based attribution (TrackStar via Bergson) over a working set drawn from the de-duplicated Dolma3 mix, aggregate influence across WebOrganizer's 24-format x 24-topic taxonomy (576 bins), and contrast benchmark pairs in a 2x2 design that varies domain (social vs. STEM) and capability type (reasoning vs. knowledge): SocialIQA and MMLU Social Sciences against ARC-Challenge and MMLU STEM. Social and STEM reasoning draw on qualitatively distinct corpus regions, and the contrast is sharper at the reasoning level than at the knowledge level. Targeted machine unlearning provides partial causal validation: forgetting high-attribution topic bins (e.g., Literature for SocialIQA) degrades the aligned benchmark more than within-bin random baselines, and we open-source all code, sampling manifests, the bin-level influence matrix, and unlearning checkpoints.
Chinese Translation
我们使用训练数据归因作为一种可解释的工具来发现能力，映射预训练语料库中哪些区域支持社会推理与STEM推理在OLMo3-7B中的表现。训练数据归因测量每个训练文档对模型在基准测试上的预测影响程度，但文档级别的得分过于嘈杂，无法识别哪些语料区域支持哪些能力，之前的研究强调了事实知识而非推理。我们计算了基于梯度的归因（通过Bergson的TrackStar），在从去重的Dolma3混合中抽取的工作集上进行，聚合WebOrganizer的24种格式与24种主题分类法（576个类别）中的影响，并在一个2x2设计中对比基准对，变化域（社会与STEM）和能力类型（推理与知识）：SocialIQA和MMLU社会科学对比ARC-Challenge和MMLU STEM。社会与STEM推理依赖于质上不同的语料区域，且在推理层面的对比比在知识层面更为明显。针对性的机器遗忘提供了部分因果验证：遗忘高归因主题类别（例如，SocialIQA的文学）对对齐基准的影响大于类别内随机基线，我们将所有代码、采样清单、类别级影响矩阵和遗忘检查点开源。
View on arXiv Download PDF AI Translation
cs.CL / 18 / 2606.19637

Before the Labels: How Dataset Construction Shapes Suicidality Detection in Clinical Text

标签之前：数据集构建如何影响临床文本中的自杀倾向检测
Garg, Priyanshi, Rao, Ishita, Ding, Jieqiong, Paullada, Amandalynne
Abstract
Clinical NLP increasingly relies on electronic health record (EHR) data to detect suicidal behaviors, treating clinical documentation as more reliable ground truth than social media. We argue that this framing obscures how EHR-based suicidality datasets encode a particular operationalization of suicidality, shaped by who authors the data, how episodes are bounded, and how ambiguity is resolved. We ground this argument in a case study of the ScAN dataset, built over MIMIC-III clinical notes. We show how governance constraints, ICD-based cohort selection, single-annotator labeling, and hospital-stay-level aggregation produce labels that reflect clinician-documented judgments, treat suicidality as a bounded episode, and assume that intent can be reliably inferred from documentation. A linguistic analysis demonstrates that identical labels subsume heterogeneous clinical framings differing in temporality, negation, and uncertainty. We argue that clinical NLP should examine the assumptions embedded in suicidality datasets before interpreting their labels as ground truth.
Chinese Translation
临床自然语言处理（NLP）越来越依赖电子健康记录（EHR）数据来检测自杀行为，将临床文档视为比社交媒体更可靠的真实依据。我们认为，这种框架掩盖了基于EHR的自杀倾向数据集如何编码自杀倾向的特定操作化，这种操作化受到数据作者、事件界定和模糊性解决方式的影响。我们通过对ScAN数据集的案例研究来支撑这一论点，该数据集建立在MIMIC-III临床笔记之上。我们展示了治理约束、基于ICD的队列选择、单一标注者标注以及住院水平聚合如何产生反映临床医生记录判断的标签，这些标签将自杀倾向视为一个有限的事件，并假设意图可以从文档中可靠推断。语言学分析表明，相同的标签涵盖了在时间性、否定和不确定性上存在差异的异质临床框架。我们认为，临床NLP应该在将自杀倾向数据集的标签解读为真实依据之前，审视其内嵌的假设。
View on arXiv Download PDF AI Translation
cs.CL / 19 / 2606.19638

MiqraBERT: Regression-Based Sentence-BERT Finetuning for Biblical Hebrew Parallel Detection

MiqraBERT：基于回归的句子-BERT微调用于圣经希伯来文平行检测
Smiley, David M.
Abstract
Textual reuse pervades the Hebrew Bible, yet the computational methods used to detect it still rest largely on lexical overlap, and they falter once a parallel involves paraphrase, lexical substitution, or syntactic reworking. This paper introduces MiqraBERT, a Sentence-BERT model finetuned from AlephBERT (a Modern Hebrew encoder) for verse-level semantic similarity in Biblical Hebrew. The training set comprises 1,650 labeled verse and half-verse pairs: 825 true parallels drawn from the Chronicles synoptic material and from foundational studies of poetic parallelism, balanced against 825 randomly sampled negatives. Through cosine-similarity regression, the model learns an embedding space in which parallel verses cluster together and unrelated verses move apart. We evaluate separation with distribution-based metrics, Wasserstein distance and the overlap coefficient, across ten random seeds. MiqraBERT improves distributional separation 2.7-fold over the pre-trained baseline and reduces the ambiguous overlap region from roughly 24% to about 6%. Narrative synoptic parallels reach a recall@10 of 87.1%; poetic parallels remain difficult, below 9%. This genre-dependent asymmetry confines the model's reliable scope to narrative textual reuse. MiqraBERT is publicly available at https://huggingface.co/davidmsmiley/MiqraBERT
Chinese Translation
文本重用贯穿于希伯来圣经，但用于检测文本重用的计算方法仍主要依赖于词汇重叠，一旦平行涉及意译、词汇替换或句法重组，这些方法便显得不足。本文介绍了MiqraBERT，这是一种基于AlephBERT（现代希伯来文编码器）微调的句子-BERT模型，旨在实现圣经希伯来文的经文级语义相似性。训练集包含1,650对标记的经文和半经文对：825对真实平行文本来自《历代志》的对照材料和诗歌平行主义的基础研究，另有825对随机抽样的负例。通过余弦相似度回归，该模型学习了一个嵌入空间，其中平行经文聚集在一起，而无关的经文则分散开来。我们使用基于分布的度量、Wasserstein距离和重叠系数，在十个随机种子上评估分离效果。MiqraBERT在分布分离上比预训练基线提高了2.7倍，并将模糊重叠区域从大约24%减少到约6%。叙事对照平行的召回率@10达到87.1%；而诗歌平行仍然较难，低于9%。这种依赖于体裁的不对称性将模型的可靠范围限制在叙事文本重用上。MiqraBERT可在https://huggingface.co/davidmsmiley/MiqraBERT公开获取。
View on arXiv Download PDF AI Translation
cs.CL / 20 / 2606.19640

Creating Multilingual Mental Health Dialogue Datasets: Limits of Persona-Based Localization via Nationality and Language

创建多语言心理健康对话数据集：基于国籍和语言的角色本地化的局限性
Xu, Yunkai, Abdullah, Saeed
Abstract
AI and large language models (LLMs) have emerged as promising tools to address global mental health challenges. Despite the global nature of these challenges, there remains a critical shortage of high-quality datasets for training and evaluating such systems. To mitigate this gap, researchers increasingly generate synthetic clinical personas to simulate user data and test digital mental health support systems. However, most validated personas rely on English-centric contexts. This paper investigates whether similar persona-based methods can be used to generate multilingual mental health datasets. We modified nationality and language parameters in personas to generate clinical dialogues in Mandarin, Bengali, and Hindi. We then examined how different LLMs perform when evaluating the depression severity of these generated multilingual datasets against the baseline in English. Our findings indicate that just adding nationality and language parameters in personas might not be adequate, as it can introduce clinical inconsistency across languages. LLM judge models often exhibit inaccuracies in assessing depression severity in non-English texts, with performance varying across different models. This exposes the systemic limitations of applying English-centric personas to multilingual contexts. Ultimately, our work highlights the urgent need for culturally responsive data generation to ensure equitable mental health systems globally.
Chinese Translation
人工智能和大型语言模型（LLMs）已成为应对全球心理健康挑战的有前景的工具。尽管这些挑战具有全球性，但用于训练和评估此类系统的高质量数据集仍然严重短缺。为了弥补这一差距，研究人员越来越多地生成合成临床角色，以模拟用户数据并测试数字心理健康支持系统。然而，大多数经过验证的角色依赖于以英语为中心的背景。本文探讨了是否可以使用类似的基于角色的方法来生成多语言心理健康数据集。我们修改了角色中的国籍和语言参数，以生成普通话、孟加拉语和印地语的临床对话。随后，我们评估了不同的LLM在评估这些生成的多语言数据集的抑郁严重性时与英语基线的表现。我们的研究结果表明，仅仅在角色中添加国籍和语言参数可能不足，因为这可能在不同语言之间引入临床不一致性。LLM判断模型在评估非英语文本的抑郁严重性时往往表现出不准确性，不同模型的表现差异明显。这揭示了将以英语为中心的角色应用于多语言背景的系统性局限性。最终，我们的研究强调了迫切需要进行文化响应的数据生成，以确保全球心理健康系统的公平性。
View on arXiv Download PDF AI Translation
cs.CL / 21 / 2606.19647

From 50K to 8.2 Million in 24 Hours: Vozinha's Algorithmic Consecration and the Multilingual Making of World Cup Visibility

从50K到820万：Vozinha的算法祝圣与世界杯可见性的多语言构建
Covas, Vinicius
Abstract
We present a multilingual computational discourse analysis of how language constructed the algorithmic consecration of Vozinha, the 40-year-old Cape Verde goalkeeper, after Spain 0-0 Cape Verde at the 2026 FIFA World Cup. The study contributes a multilingual corpus in Portuguese, Spanish, English, and French; a nine-frame narrative taxonomy with cue-based frame annotation; a reproducible annotation pipeline combining LLM-assisted suggestion with human validation; and an analysis of cross-lingual narrative diffusion across discourse phases. We treat the platform follower count itself, narrated as "50k to 8M", as a linguistic object: a circulating and narratable proof of visibility rather than a mere measurement. The follower-growth timeline is used only as contextual metadata: we reconstruct a conservative phase structure, not a continuous API-native series, and type every datapoint by value class, confidence, and evidence type. The only exact primary scraper anchor is 8,235,652 followers at 2026-06-16 15:47 UTC; all other figures are reported as estimated ranges or thresholds, including an estimated pre-match baseline of 45k-56k. Findings suggest that distinct languages carried distinct frames: Portuguese mobilization, Spanish crisis, English nation-making, and a shared platform-metric spectacle through which peripheral athletic performance became globally visible. As a v0.1 pilot, the paper releases the corpus schema, frame taxonomy, annotation guidelines, hashed visual-evidence log, and typed timeline, while flagging full double annotation and inter-annotator agreement as planned work.
Chinese Translation
我们呈现了一项多语言计算话语分析，探讨语言如何构建Vozinha（这位40岁的佛得角守门员）在2026年国际足联世界杯西班牙与佛得角0-0平局后所获得的算法祝圣。本研究贡献了一个包含葡萄牙语、西班牙语、英语和法语的多语言语料库；一个包含基于线索的框架注释的九框叙事分类法；一个结合大型语言模型（LLM）辅助建议与人工验证的可重复注释流程；以及对跨语言叙事传播在话语阶段的分析。我们将平台关注者数量本身视为一种语言对象：“50k到8M”的叙述，作为可流通和可叙述的可见性证明，而不仅仅是一个简单的度量。关注者增长时间线仅用作上下文元数据：我们重建了一个保守的阶段结构，而不是一个连续的API原生系列，并根据价值类别、置信度和证据类型对每个数据点进行分类。唯一的确切主要抓取锚点是2026年6月16日15:47 UTC的8235652名关注者；所有其他数字均以估计范围或阈值报告，包括估计的赛前基线为45k-56k。研究结果表明，不同语言承载了不同的框架：葡萄牙语的动员、西班牙语的危机、英语的国家构建，以及一个共享的平台指标奇观，通过它边缘运动表现得以全球可见。作为v0.1试点，本文发布了语料库架构、框架分类法、注释指南、哈希视觉证据日志和分类时间线，同时标记了完整的双重注释和注释者间一致性作为计划工作。
View on arXiv Download PDF AI Translation
cs.CL / 22 / 2606.19659

SAGE-OPD: Selective Agent-Guided Intervention for Multi-Turn On-Policy Distillation

SAGE-OPD：针对多回合在线策略蒸馏的选择性代理引导干预
Zhou, Yuhang, Zhang, Lizhu, Wu, Yifan, Wang, Mingyi, Peng, Bo, Liu, Jiayi, Fan, Xiangjun, Zhao, Zhuokai
Abstract
On-policy distillation (OPD) improves student models by training them on trajectories induced by their own policy, making it a promising approach for mitigating exposure bias in agent training. However, most OPD studies focus on single-turn settings, while realistic LLM agents interact with environments over multiple turns. In this regime, early errors can alter future observations and compound across the trajectory, and standard dense token-level OPD becomes brittle, as it may over-penalize semantically valid alternatives, reinforce local degeneracies such as repeated actions, and propagate unreliable teacher supervision on off-distribution histories. We propose SAGE-OPD, a verifier-free selective intervention framework specifically designed for multi-turn OPD. Instead of applying teacher supervision uniformly across all turns, SAGE-OPD first observes environment feedback and uses teacher judgment to decide whether each student response should be skipped or intervened on. To further address compounding errors, SAGE-OPD weights token-level distillation by teacher confidence, reducing the influence of uncertain teacher distributions on corrupted or ambiguous histories. Finally, SAGE-OPD applies loss normalization to preserve the overall loss scale of standard OPD while retaining selective turn-level weighting. Experiments on agent tasks show that SAGE-OPD consistently improves over baselines, achieving up to a 13.3% relative improvement in ALFWorld unseen success rate over standard OPD. Ablation studies further demonstrate that turn-level intervention, teacher confidence weighting, and loss normalization provide complementary benefits. Our results suggest that effective multi-turn OPD should remain on-policy, but teacher supervision should be selectively allocated to turns where intervention is necessary and reliable.
Chinese Translation
在线策略蒸馏（OPD）通过在自身策略诱导的轨迹上训练学生模型来改善其性能，这使其成为减轻代理训练中曝光偏差的有前景的方法。然而，大多数OPD研究集中在单回合设置上，而现实中的大语言模型（LLM）代理与环境的交互通常是多回合的。在这种情况下，早期的错误可能会改变未来的观察结果，并在轨迹中累积，标准的密集令牌级OPD变得脆弱，因为它可能会过度惩罚语义上有效的替代方案，强化局部退化（如重复动作），并在离散历史上传播不可靠的教师监督。我们提出了SAGE-OPD，这是一种无验证的选择性干预框架，专门设计用于多回合OPD。SAGE-OPD并不是在所有回合均匀地应用教师监督，而是首先观察环境反馈，并利用教师的判断决定每个学生响应是否应被跳过或干预。为了进一步解决累积错误，SAGE-OPD通过教师信心对令牌级蒸馏进行加权，从而减少不确定教师分布对损坏或模糊历史的影响。最后，SAGE-OPD应用损失归一化，以保持标准OPD的整体损失规模，同时保留选择性的回合级加权。在代理任务上的实验表明，SAGE-OPD始终优于基线，在ALFWorld未见成功率上相较于标准OPD实现了高达13.3%的相对提升。消融研究进一步表明，回合级干预、教师信心加权和损失归一化提供了互补的好处。我们的结果表明，有效的多回合OPD应保持在线策略，但教师监督应选择性地分配给需要和可靠的干预回合。
View on arXiv Download PDF AI Translation
cs.CL / 23 / 2606.19667

CacheWeaver: Cache-Aware Evidence Ordering for Efficient Grounded RAG Inference

CacheWeaver：基于缓存的证据排序以实现高效的有根据的RAG推理
Tan, Kaizhen, Gu, Rong, Li, Mingyuan
Abstract
Retrieval-Augmented Generation (RAG) improves factual grounding, but it also lengthens prompts and raises prefill cost. Prefix caching in serving engines such as vLLM reduces this cost only when requests share the same token prefix. In grounded generation, however, adjacent queries may retrieve overlapping evidence in different orders, so set overlap does not become reusable prefix overlap. We present CacheWeaver, a lightweight prompt-layer method for cache-aware evidence ordering. The method keeps a prefix tree over recently served evidence sequences and uses a greedy walk to place the most reusable prefix first, while leaving the serving engine and retrieved evidence set unchanged. Across three vLLM configurations, the method lowers median time-to-first-token (TTFT) by about 20-33 percent relative to retrieval-order prefix caching, without hurting answer quality in our QA tests. The greedy policy reaches 97.5 percent of the median TTFT gain from oracle ordering, indicating that most reusable prefix locality can be recovered by a simple scheduling layer between retrieval and inference.
Chinese Translation
检索增强生成（RAG）改善了事实基础，但也延长了提示并提高了预填充成本。在像vLLM这样的服务引擎中，前缀缓存仅在请求共享相同的令牌前缀时才能降低这一成本。然而，在有根据的生成中，相邻查询可能以不同的顺序检索重叠的证据，因此集合重叠并不能转化为可重用的前缀重叠。我们提出了CacheWeaver，一种轻量级的提示层方法，用于基于缓存的证据排序。该方法在最近服务的证据序列上保持一个前缀树，并使用贪婪遍历将最可重用的前缀放在首位，同时保持服务引擎和检索的证据集不变。在三种vLLM配置中，该方法将中位数首次令牌时间（TTFT）降低了约20-33%，而在我们的问答测试中没有影响答案质量。贪婪策略达到了来自oracle排序的中位数TTFT增益的97.5%，表明大多数可重用前缀的局部性可以通过检索与推理之间的简单调度层恢复。
View on arXiv Download PDF AI Translation
cs.CL / 24 / 2606.19668

Code-Switching Reveals Language Anchoring in Multilingual LLMs

代码切换揭示多语言大型语言模型中的语言锚定
Park, Jeonghyun, Yoon, Seunghyun, Jun, Yonghyun, Lee, Hwanhee
Abstract
Multilingual Large Language Models (MLLMs) are increasingly expected to handle Code-Switched (CS) inputs, yet mixing languages frequently degrades performance relative to source- or target-language monolingual counterparts. To understand this degradation, we use grammar-forced CS as a controlled diagnostic setting for locating CS representations relative to their source and target counterparts. We introduce Anchor Bias, a geometric measure that quantifies language anchoring, whether a CS hidden state aligns closer to its source or target language counterpart. Across diverse MLLMs, Anchor Bias reveals a consistent grammar-frame effect: source-framed CS stays source-anchored, whereas target-framed CS shifts target-ward and shows larger Question Answering (QA) degradation. Motivated by this representational pattern, we propose CANVAS (Contextual Anchor-based Neural Vector Alignment Steering), an inference-time intervention that extracts a source-side canvas from the input and softly steers target-language hidden states toward the source anchor during prefill. CANVAS consistently recovers QA F1 across MLLMs and CS conditions, showing that internal anchoring signals provide an actionable target for mitigating CS inference failures.
Chinese Translation
多语言大型语言模型（MLLMs）越来越被期望能够处理代码切换（CS）输入，然而，语言混合通常会导致相较于源语言或目标语言的单语模型性能下降。为了理解这种性能下降，我们使用强制语法的代码切换作为一种受控诊断设置，以定位代码切换表示相对于其源语言和目标语言对应物的位置。我们引入了锚定偏差（Anchor Bias），这是一种几何度量，用于量化语言锚定，即代码切换的隐藏状态是否更接近其源语言或目标语言对应物。在多样化的多语言大型语言模型中，锚定偏差揭示了一种一致的语法框架效应：源框架的代码切换保持源锚定，而目标框架的代码切换则向目标语言偏移，并显示出更大的问答（QA）性能下降。基于这种表示模式，我们提出了CANVAS（基于上下文的锚定神经向量对齐引导），这是一种推理时干预方法，它从输入中提取源侧画布，并在预填充过程中轻柔地引导目标语言的隐藏状态朝向源锚定。CANVAS在多语言大型语言模型和代码切换条件下始终恢复问答F1分数，表明内部锚定信号为减轻代码切换推理失败提供了可行的目标。
View on arXiv Download PDF AI Translation
cs.CL / 25 / 2606.19698

What sentiment analysis can't see: Measuring whether customers were helped, and what went wrong, across 70,000 support conversations

情感分析无法看到的：衡量客户是否得到帮助以及在70,000次支持对话中出现了什么问题
Potteiger, Jason
Abstract
Most companies read their customer support data at scale using sentiment analysis, which measures how customers sound rather than whether they were satisfied with the result. We tested a richer alternative on 70,450 support conversations from a leading online fundraising platform: alongside tone, we used GPT-5.4 to estimate each customer's satisfaction and to flag whether they reported a concrete problem, then validated all three readings against the 1-to-5 ratings customers left on the conversations they rated. The satisfaction estimate tracked those ratings far better than sentiment did, correlating at 0.47 against 0.36 and flagging unhappy customers with far fewer false alarms. The structured read also sees what sentiment cannot: tone and satisfaction disagree in 44% of conversations, a single "Neutral" label hides everything from quietly satisfied customers to ones who quietly gave up, and the largest group of all is "tolerated friction," customers who are satisfied but still reporting a fixable problem, a standing issue that no sentiment-based dashboard can surface. The broader finding is that LLM-based annotation can capture far more than the tonality of a customer's language, offering strong potential for new business metrics grounded instead in the customer's state (whether they were satisfied) and the cause of their problem extracted directly from the raw textual data of interactions and feedback.
Chinese Translation
大多数公司通过情感分析大规模地阅读客户支持数据，该分析测量的是客户的语气，而非他们对结果的满意度。我们在一家领先的在线筹款平台的70,450次支持对话中测试了一种更丰富的替代方案：除了语气外，我们还使用GPT-5.4来估计每位客户的满意度，并标记他们是否报告了具体问题，然后将这三种读数与客户在他们评价的对话中留下的1到5的评分进行验证。满意度估计的跟踪效果远优于情感分析，相关性为0.47，而情感分析的相关性为0.36，并且在标记不满意客户时误报率显著降低。结构化的读取还揭示了情感分析无法看到的内容：在44%的对话中，语气和满意度存在不一致，一个单一的“中立”标签掩盖了从安静满意的客户到默默放弃的客户的所有情况，而最大的群体是“容忍摩擦”的客户，他们虽然满意但仍报告了可解决的问题，这是任何基于情感的仪表盘无法揭示的持续性问题。更广泛的发现是，基于大型语言模型（LLM）的注释能够捕捉到客户语言的语气之外的更多内容，为新的商业指标提供了强大的潜力，这些指标基于客户的状态（他们是否满意）以及直接从互动和反馈的原始文本数据中提取的问题原因。
View on arXiv Download PDF AI Translation
cs.CL / 26 / 2606.19700

TerraMARS: A Domain-Adapted Small-Language-Model Pipeline for Mars Terraforming Literature

TerraMARS：用于火星 terraforming 文献的领域适应小语言模型管道
Singh, Jyotsna, Black, Ash, Larsen, Jeff, Saleska, Scott R.
Abstract
Researchers are interested in learning about Mars so that it may eventually become habitable for humans. To achieve this, there is a need for comprehensive knowledge of the planet's atmosphere, hydrology, surface chemistry, radiation environment, and spatial features through the scientific literature. These contain valuable information and meaningful quantitative constraints that can be used in other models and studies, such as habitability assessment and future terraforming studies. We present TerraMARS, an end-to-end information extraction pipeline that combines a domain-adapted Small Language Model to answer Mars terraforming-related questions and convert unstructured Mars science text into machine-readable structured outputs in JavaScript Object Notation (JSON) format. A corpus of open-access papers is collected and processed using a multistage retrieval and chunking framework. Google Gemma 3 1B was adapted to the domain using Quantized Low-Rank Adaptation (QLoRA) fine-tuning on Mars-specific question-answering and information extraction datasets. The resulting pipeline generates both types of output and provides a foundation for integrating knowledge from scientific literature into downstream applications like digital twins and habitability modeling for Mars. The output from this pipeline looks promising, but further improvements are needed to increase extraction accuracy and factual consistency.
Chinese Translation
研究人员对火星的了解充满兴趣，以期最终使其适合人类居住。为此，需要通过科学文献全面了解该星球的气氛、水文学、表面化学、辐射环境和空间特征。这些文献中包含了宝贵的信息和有意义的定量约束，可用于其他模型和研究，例如适居性评估和未来的 terraforming 研究。我们提出了 TerraMARS，一个端到端的信息提取管道，结合了领域适应的小语言模型，以回答与火星 terraforming 相关的问题，并将非结构化的火星科学文本转换为机器可读的结构化输出，格式为 JavaScript 对象表示法（JSON）。我们收集并处理了一批开放获取的论文，采用多阶段检索和分块框架。Google Gemma 3 1B 通过在火星特定的问答和信息提取数据集上进行量化低秩适应（QLoRA）微调，适应了该领域。最终的管道生成两种类型的输出，并为将科学文献中的知识整合到下游应用（如数字双胞胎和火星适居性建模）提供了基础。该管道的输出看起来很有前景，但仍需进一步改进以提高提取准确性和事实一致性。
View on arXiv Download PDF AI Translation
cs.CL / 27 / 2606.19710

FineREX: Fine-Tuned NER-RE for Human Smuggling Knowledge Graphs

FineREX：针对人类走私知识图谱的微调命名实体识别与关系提取
Feldman, Elijah, Meher, Dipak, Domeniconi, Carlotta
Abstract
Court proceedings contain valuable evidence about human smuggling networks, but this information is often buried within unstructured, jargon-heavy legal documents. While large language models (LLMs) can support knowledge graph construction through automated information extraction, existing approaches rely on general-purpose models that are not tailored to the entity and relationship definitions required in this domain. We introduce FineREX, a streamlined knowledge graph construction pipeline built around a fine-tuned LLM for named entity recognition and relationship extraction (NER-RE). Using a manually annotated dataset of $512$ text chunks, FineREX achieves absolute improvements of 15.50% and 31.46% in entity and relationship F1-score, respectively, compared to a larger general-purpose baseline. These gains translate into higher-quality knowledge graphs, reducing legal noise by nearly half and lowering node duplication on long documents from 17.78% to 11.17%. By eliminating document rewriting and redundant extraction stages, FineREX also reduces end-to-end processing time by 50.0%. Our results demonstrate that domain-specific fine-tuning can substantially outperform larger general-purpose models while improving both the quality and efficiency of knowledge graph construction for illicit network analysis.
Chinese Translation
法庭诉讼程序包含关于人类走私网络的宝贵证据，但这些信息常常埋藏在结构松散、术语繁多的法律文件中。尽管大型语言模型（LLMs）可以通过自动信息提取支持知识图谱的构建，但现有的方法依赖于不针对该领域所需的实体和关系定义的通用模型。我们提出了FineREX，一个围绕微调的LLM构建的简化知识图谱构建管道，用于命名实体识别和关系提取（NER-RE）。使用一个手动标注的$512$个文本块的数据集，FineREX在实体和关系的F1-score上分别实现了15.50%和31.46%的绝对提升，相比于一个更大的通用基线。这些提升转化为更高质量的知识图谱，将法律噪声减少近一半，并将长文档中的节点重复率从17.78%降低到11.17%。通过消除文档重写和冗余提取阶段，FineREX还将端到端处理时间减少了50.0%。我们的结果表明，领域特定的微调可以显著超越更大的通用模型，同时提高非法网络分析的知识图谱构建的质量和效率。
View on arXiv Download PDF AI Translation
cs.CL / 28 / 2606.19727

NRITYAM: Language Models Meet Art and Heritage of Dance

NRITYAM：语言模型与舞蹈艺术与遗产的交汇
Singh, Punit Kumar, Ghosh, Niladri, Joshiınst, Advait, Choudhary, Shailee, Färber, Michael, Yang, Haiqin
Abstract
Language models have become essential tools in shaping modern workflows. However, their global effectiveness hinges on a nuanced understanding of local socio-cultural contexts. To address this gap, we present NRITYAM, a comprehensive benchmark for evaluating the cultural comprehension capabilities of language models in the context of global dance traditions. NRITYAM comprises 9,260 carefully curated question-answer pairs spanning 12 languages, making it the largest dataset dedicated to evaluating cultural knowledge in dance. The dataset has been developed from the ground up through close collaboration with native dance artists and native speakers of the languages, who authored and validated culturally relevant questions specific to their regions. We evaluate a broad set of models, including large language models, small language models, multimodal large language models, and small multimodal language models. As a multilingual and multicultural benchmark, NRITYAM sets a new standard for evaluating the ability of AI systems to understand and reason about traditional performing arts. Detailed dataset samples are available at~\url{https://github.com/niladrighosh03/NRITYAM}.
Chinese Translation
语言模型已成为塑造现代工作流程的重要工具。然而，它们的全球有效性依赖于对地方社会文化背景的细致理解。为了解决这一问题，我们提出了NRITYAM，这是一个全面的基准，用于评估语言模型在全球舞蹈传统背景下的文化理解能力。NRITYAM包含9260对精心策划的问答对，涵盖12种语言，成为评估舞蹈文化知识的最大数据集。该数据集是通过与本土舞蹈艺术家和母语者的紧密合作，从零开始开发的，他们撰写并验证了与其地区相关的文化问题。我们评估了一系列广泛的模型，包括大型语言模型、小型语言模型、多模态大型语言模型和小型多模态语言模型。作为一个多语言和多文化的基准，NRITYAM为评估人工智能系统理解和推理传统表演艺术的能力树立了新的标准。详细的数据集样本可在~\url{https://github.com/niladrighosh03/NRITYAM}获取。
View on arXiv Download PDF AI Translation
cs.CL / 29 / 2606.19744

Beyond Uniform Forgetting: A Study of Sequential Direct Preference Optimization Across Preference Settings

超越均匀遗忘：跨偏好设置的顺序直接偏好优化研究
Bhandari, Pranav, Fay, Nicolas, Datta, Amitava, Naseem, Usman, Nasim, Mehwish
Abstract
Aligning language models with human preferences often requires optimising multiple behavioural objectives. A practical approach is to apply these objectives sequentially using preference optimisation methods such as Direct Preference Optimisation (DPO), but it remains unclear whether later training uniformly degrades preferences learned earlier or whether the effect depends on the relationship between objectives. We study sequential DPO across four preference settings covering distributional conflict, multi-attribute interaction, strong safety signal, and compatible response-quality objectives. Using Llama-3.1-8B-Instruct with LoRA adapters, we evaluate all objectives after every stage with a fixed base-model reference. We find that sequential DPO does not produce a single forgetting pattern; preference change ranges from partial degradation to stability, pair-level redistribution, or positive transfer depending on objective relationship, signal strength, and training order. Pair-level analysis using length-normalised policy margins shows that aggregate metrics can mask heterogeneous changes across preference pairs, whereas quartile decomposition reveals that high-confidence pairs can either degrade or improve depending on the setting. Mechanistic diagnostics show that Stage~2 gradients and adapter updates are near-orthogonal to the previous objective across all settings, providing little evidence that direct gradient opposition is the primary driver. These findings suggest that future sequential alignment pipelines should account for objective compatibility and signal strength, rather than assuming that later objectives affect earlier preferences uniformly.
Chinese Translation
将语言模型与人类偏好对齐通常需要优化多个行为目标。一种实用的方法是使用偏好优化方法（如直接偏好优化（Direct Preference Optimisation, DPO））顺序应用这些目标，但尚不清楚后续训练是否会均匀地降低早期学习的偏好，或者这种影响是否取决于目标之间的关系。我们研究了在四种偏好设置下的顺序DPO，这些设置涵盖了分布冲突、多属性交互、强安全信号和兼容响应质量目标。使用带有LoRA适配器的Llama-3.1-8B-Instruct，我们在每个阶段后使用固定的基础模型参考评估所有目标。我们发现顺序DPO并未产生单一的遗忘模式；偏好变化从部分退化到稳定、成对重新分配或正向转移，具体取决于目标关系、信号强度和训练顺序。使用长度归一化的策略边际进行的成对分析表明，聚合指标可能掩盖偏好对之间的异质变化，而四分位分解则揭示高置信度对在不同设置下可能退化或改善。机制诊断显示，第二阶段的梯度和适配器更新在所有设置中与先前目标几乎正交，几乎没有证据表明直接梯度对立是主要驱动因素。这些发现表明，未来的顺序对齐流程应考虑目标兼容性和信号强度，而不是假设后续目标均匀地影响早期偏好。
View on arXiv Download PDF AI Translation
cs.CL / 30 / 2606.19815

Clusters are All You Need: Pre-Training the Tsetlin Machine with Semantic Clusters from Language Models for Interpretability

聚类是你所需的一切：利用语言模型的语义聚类对 Tsetlin 机器进行预训练以提高可解释性
Gao, Jiechao, Yadav, Rohan Kumar, Li, Yuangang, Pan, Yuandong, Wang, Jie, Liu, Ying, Lepech, Michael
Abstract
Pre-trained language models such as BERT achieve strong text classification performance but lack transparency, limiting their use in high-stakes settings. The Tsetlin Machine (TM) offers fully interpretable, clause-based reasoning but captures little semantic information, and prior attempts to bridge the two rely on static word embeddings that miss contextual meaning. We propose a semantic pre-training framework that transfers knowledge from a pre-trained language model into a TM without using embeddings. Text samples are grouped into semantically coherent clusters with K-means or Top2Vec, and the resulting cluster-sample pairs pre-train a non-negated TM with enhanced Type I feedback. The TM thereby learns interpretable semantic keywords that are fine-tuned on downstream tasks. Across five datasets, our method substantially outperforms vanilla and embedding-based TMs and reaches performance competitive with BERT while remaining interpretable.
Chinese Translation
预训练语言模型如 BERT 在文本分类任务中表现出色，但缺乏透明性，限制了其在高风险环境中的应用。Tsetlin 机器（TM）提供了完全可解释的基于子句的推理，但捕获的语义信息较少，之前尝试将两者结合的方式依赖于静态词嵌入，未能捕捉上下文意义。我们提出了一种语义预训练框架，将预训练语言模型中的知识转移到 TM 中，而无需使用嵌入。文本样本通过 K-means 或 Top2Vec 被分组为语义一致的聚类，生成的聚类-样本对对非否定的 TM 进行预训练，并增强了类型 I 反馈。通过这种方式，TM 学习到可解释的语义关键词，并在下游任务中进行微调。在五个数据集上，我们的方法显著优于普通和基于嵌入的 TM，并在保持可解释性的同时达到了与 BERT 竞争的性能。
View on arXiv Download PDF AI Translation
cs.CL / 31 / 2606.19819

CREDENCE: Claim Reduction for Decomposition & Enhanced Credibility -- Semantic Metrics and Convergence Analysis

CREDENCE：分解与增强可信度的声明减少——语义度量与收敛分析
Tran, Phuong Huu Vu, Mai, Thuan Duc, Le, Bach Xuan
Abstract
Decomposing compound sentences into atomic, verifiable claims is a prerequisite for reliable automated fact-checking. Prior work has relied on token-overlap (Jaccard) metrics that systematically underestimate decomposition quality for paraphrastic claims, and has lacked formal termination analysis for the repair loop. We present Credence, a revised claim decomposition and evaluation framework addressing both shortcomings. Our contributions are: (1) Semantic-F1: we use BGE-large cosine similarity fidelity metric that resolves Jaccard's penalisation and improves downstream fact-checking accuracy; (2) Convergence theorems: we formally characterise four properties of the repair pipeline, establishing that rule-based repair is monotone and finitely terminating under an oracle parser assumption; LLM-based self-repair is provably non-monotone and requires an early-exit guard; (3) Three evaluation benchmarks spanning social-media, encyclopaedic, and news domains for cross-domain generalisation measurement; (4) Multi-model benchmarking across four decomposer models (3.8B-12B) and a closed API model. Experiments on SocialClaimSplit, WikiSplitBench, and ClaimDecompBench show that Semantic-F1 outperforms Jaccard-F1 by +15-32pp. EPR ranges from 0.94 to 1.00 on SocialClaimSplit and WikiSplitBench, while ClaimDecompBench includes lower base EPR cases (down to 0.824) due to harder news-domain constructions, and rule-repair reduces the Atomicity Violation Rate (AVR) by 47-100% relative to the base model without degrading fidelity.
Chinese Translation
将复合句分解为原子、可验证的声明是可靠自动事实核查的前提。以往的研究依赖于令牌重叠（Jaccard）度量，这系统性地低估了对同义声明的分解质量，并且缺乏对修复循环的正式终止分析。我们提出了Credence，一个修订的声明分解与评估框架，解决了这两个缺陷。我们的贡献包括：（1）Semantic-F1：我们使用BGE-large余弦相似度保真度度量，解决了Jaccard的惩罚问题，并提高了下游事实核查的准确性；（2）收敛定理：我们正式表征了修复管道的四个属性，确立了基于规则的修复在假设有神谕解析器的情况下是单调的且有限终止的；基于LLM的自我修复被证明是非单调的，并需要提前退出保护；（3）涵盖社交媒体、百科全书和新闻领域的三个评估基准，用于跨领域泛化测量；（4）在四个分解模型（3.8B-12B）和一个封闭API模型之间进行多模型基准测试。在SocialClaimSplit、WikiSplitBench和ClaimDecompBench上的实验表明，Semantic-F1的表现优于Jaccard-F1，提升幅度为+15-32个百分点。EPR在SocialClaimSplit和WikiSplitBench上范围为0.94到1.00，而ClaimDecompBench由于更复杂的新闻领域构造包含较低的基础EPR案例（最低为0.824），并且规则修复相对于基础模型将原子性违反率（AVR）降低了47-100%，而不降低保真度。
View on arXiv Download PDF AI Translation
cs.CL / 32 / 2606.19831

Leverage Is Not Reach: A Control-Window Law for Single-Neuron Steering in Language Models

杠杆并非触达：语言模型中单神经元引导的控制窗口法则
Liu, Hongliang
Abstract
Aligned language models gate behaviors such as refusal and language routing through sparse feed forward neurons, yet no theory predicts when a single neuron intervention controls a behavior coherently rather than collapsing the output. We develop a budget normalized control window framework for single neuron steering. A dose along one write direction reduces to one control coordinate: the alignment between the residual stream and the write, driven along a universal saturation curve in units of a coherence budget set by the residual norm divided by the write norm. Coherent control exists when a behavior trigger lies below the collapse ceiling. The same coordinate governs benign mode switches and refusal; the ceiling follows from weights and one generic forward pass, while triggers are measured at rollout. On fifteen held out neurons, the predicted ceiling has mean absolute error 0.14, about 0.07 in bulk layers, and the committed open or closed verdict holds on eleven against a ten of fifteen majority baseline. Closed cases expose three failure modes rather than violations: collapse before trigger, too little depth to propagate, or a normalization that caps how far one neuron can push. The law explains why local gradient attribution anti predicts control: true controllers write off the readout axis and carry a near zero first order gradient. A forward only contrastive screen made precise by the window recovers controllers that attribution misses. On refusal, the hardest case, intervention success is typed, not scalar: coherent bypass and strict actionable reach separate, so a neuron can flip refusal in fluent, on task text with no actionable content, and genuine actionable reach appears only for three of six audited Llama pivots and only at later rollout horizons. Single neuron steering is therefore a budgeted, typed audit of controllability rather than a fixed dose anecdote.
Chinese Translation
对齐的语言模型通过稀疏前馈神经元控制诸如拒绝和语言路由等行为，但尚无理论能够预测单个神经元干预何时能够连贯地控制一种行为，而不是导致输出崩溃。我们开发了一种预算归一化的控制窗口框架用于单神经元引导。在一个写入方向上的剂量减少为一个控制坐标：残差流与写入之间的对齐，沿着由残差范数除以写入范数所设定的连贯预算的普适饱和曲线驱动。当行为触发器位于崩溃上限之下时，连贯控制存在。相同的坐标支配良性模式切换和拒绝；上限由权重和一次通用的前向传递决定，而触发器在回滚时被测量。在十五个保留的神经元中，预测的上限的平均绝对误差为0.14，在大规模层中约为0.07，且在十个中的十一项保持开放或关闭的裁决。关闭案例揭示了三种失败模式，而非违反：在触发之前崩溃、传播深度不足，或归一化限制了单个神经元的推动能力。该法则解释了为何局部梯度归因反向预测控制：真正的控制器在读出轴上写入，并保持近零的一阶梯度。通过窗口精确化的前向对比屏恢复了归因遗漏的控制器。在拒绝这一最困难的案例中，干预的成功是类型化的，而非标量的：连贯的旁路和严格的可操作触达是分开的，因此一个神经元可以在流畅的、任务相关的文本中翻转拒绝，而没有可操作内容，真正的可操作触达仅出现在六个审计的Llama枢轴中的三个，并且仅在较晚的回滚视野中出现。因此，单神经元引导是对可控性的预算化、类型化审计，而不是固定剂量的轶事。
View on arXiv Download PDF AI Translation
cs.CL / 33 / 2606.19847

AtomMem: Building Simple and Effective Memory System for LLM Agents via Atomic Facts

AtomMem：通过原子事实构建简单有效的LLM代理记忆系统
Yao, Yanyu, Li, Shangze, Zheng, Zhi, Zheng, Hui, Liu, Qi, Xu, Tong, Chen, Enhong
Abstract
Large language models (LLMs) demonstrate strong reasoning and generation abilities, but their fixed context windows limit long-term information accumulation and reuse across multi-session interactions. Existing memory-augmented systems often construct memory in a coarse and unstable manner, relying on inefficient memory representations or unstable unconstrained updates. To address these challenges, we propose AtomMem, a long-term memory system designed for value-dense storage and stable memory evolution. AtomMem introduces a Fact Executor, which selectively extracts high value atomic facts from long form interactions to serve as highly efficient memory representations. Subsequently, AtomMem organizes these facts into hierarchical event structures and temporal profiles, capturing coherent episodic contexts and tracking dynamically evolving user attributes over time. During retrieval, the system activates an associative memory graph to connect fragmented memories. Experiments on the LoCoMo benchmark confirm that AtomMem achieves state-of-the-art performance across various reasoning tasks, offering a scalable and economically viable solution for deploying intelligent personalized agents.
Chinese Translation
大型语言模型（LLMs）展现出强大的推理和生成能力，但其固定的上下文窗口限制了长期信息的积累和在多会话交互中的重用。现有的增强记忆系统往往以粗糙和不稳定的方式构建记忆，依赖于低效的记忆表示或不稳定的无约束更新。为了解决这些挑战，我们提出了AtomMem，一个旨在实现高价值存储和稳定记忆演化的长期记忆系统。AtomMem引入了一个事实执行器（Fact Executor），该执行器从长篇交互中选择性提取高价值的原子事实，以作为高效的记忆表示。随后，AtomMem将这些事实组织成层次事件结构和时间档案，捕捉连贯的情节上下文，并动态跟踪用户属性的演变。在检索过程中，该系统激活一个关联记忆图，以连接碎片化的记忆。在LoCoMo基准测试中的实验确认，AtomMem在各种推理任务中实现了最先进的性能，为部署智能个性化代理提供了可扩展且经济可行的解决方案。
View on arXiv Download PDF AI Translation
cs.CL / 34 / 2606.19852

Prompt, Plan, Extract: Zero-Shot Agentic LLMs Workflows for Lung Pathology Extraction from Clinical Narratives

提示、规划、提取：用于从临床叙述中提取肺病理信息的零-shot代理大型语言模型工作流程
Pathak, Aman, Peng, Cheng, Lyu, Mengxian, Chen, Ziyi, Solan, Reema, Talankar, Sankalp, Khan, Yasir, Mehta, Hiren, Chen, Aokun, Guo, Yi, Wu, Yonghui
Abstract
Information extraction from pathology reports is essential for cancer staging, tumor registry population. Yet key data remains embedded in narrative reports, making manual extraction labor-intensive and error-prone. Traditional supervised Natural Language Processing pipelines address this through fully supervised Named Entity Recognition and Relation Extraction, but require expensive manual annotation and suffer cascading failures when upstream entities are missed. In this study, we developed a zero-shot, agentic workflow, and evaluated five open-source generative Large Language Models (LLMs) to populate 13 College of American Pathologists synoptic fields from lung resection pathology reports. We compared them against a state-of-the-art supervised GatorTron NER-RE baseline using a novel, registry-aligned evaluation framework. The baseline achieved Micro-F1of 0.960, while the best zero-shot model (GPT-OSS-20B) achieved Micro-F1 of 0.893 (recall: 0.949), accurately extracting complex relations like Pathologic Stage without task-specific training. These results suggest that open-source, zero-shot agentic LLMs are a low-cost solution for extracting lung pathology information.
Chinese Translation
从病理报告中提取信息对于癌症分期和肿瘤登记至关重要。然而，关键数据仍然嵌入在叙述报告中，使得手动提取劳动密集且容易出错。传统的监督自然语言处理管道通过完全监督的命名实体识别（Named Entity Recognition, NER）和关系提取（Relation Extraction, RE）来解决这一问题，但需要昂贵的手动标注，并且在上游实体缺失时会遭遇级联失败。在本研究中，我们开发了一种零-shot代理工作流程，并评估了五种开源生成大型语言模型（LLMs），以从肺切除病理报告中填充13个美国病理学家学院的摘要字段。我们将其与一种最先进的监督GatorTron NER-RE基线进行比较，采用了一种新颖的与登记对齐的评估框架。基线模型的Micro-F1达到了0.960，而最佳的零-shot模型（GPT-OSS-20B）达到了0.893（召回率：0.949），准确提取了如病理分期等复杂关系，而无需特定任务的训练。这些结果表明，开源的零-shot代理LLMs是提取肺病理信息的一种低成本解决方案。
View on arXiv Download PDF AI Translation
cs.CL / 35 / 2606.19857

Large Language Models Do Not Always Need Readable Language

大型语言模型并不总是需要可读语言
Zhu, Jiayi, Peng, Haoxuan, Wang, Junxi, Ke, Liang, Zhang, Chen, Zhang, Linfeng
Abstract
Large language models (LLMs) are commonly prompted and interfaced with human-readable natural language, even when the intended reader is another model. This paper investigates whether semantic information can be encoded in compact, non-standard textual forms that sacrifice human readability while remaining recoverable by LLMs. We refer to this class of model-centric textual representations as BabelTele, approached here not as a fixed protocol but as an empirical probe into LLMs' capacity to generate and interpret such representations. Through readability diagnostics, model likelihood measures, human questionnaires, and downstream task evaluations, we find that BabelTele can substantially depart from ordinary natural language while preserving core semantics for instruction-tuned LLMs. As a task-agnostic representational paradigm, BabelTele demonstrates high information density, maintaining 99.5% semantic fidelity even when the text volume is condensed to 27.9% of its original length. We further evaluate its semantic robustness in cross-model transfer, agent memory, and multi-agent communication. Results suggest that BabelTele can reduce context overhead while generally maintaining reliable downstream performance, although its effectiveness depends on the compressor-reader pair and task setting. These findings indicate that human readability, natural-language typicality, and model-side semantic recoverability can be partially decoupled, opening a path toward model-native representations in future exploration of LLM systems.
Chinese Translation
大型语言模型（LLMs）通常使用人类可读的自然语言进行提示和交互，即使预期的读者是另一个模型。本文探讨了语义信息是否可以以紧凑的非标准文本形式编码，这种形式牺牲了人类可读性，但仍能被LLMs恢复。我们将这一类以模型为中心的文本表示称为BabelTele，在此将其视为对LLMs生成和解释此类表示能力的实证探测，而非固定协议。通过可读性诊断、模型似然度测量、人类问卷和下游任务评估，我们发现BabelTele可以在保持核心语义的同时，显著偏离普通自然语言，适用于经过指令调优的LLMs。作为一种与任务无关的表征范式，BabelTele展示了高信息密度，即使文本量缩减至原始长度的27.9%，仍能保持99.5%的语义保真度。我们进一步评估其在跨模型迁移、代理记忆和多代理通信中的语义鲁棒性。结果表明，BabelTele可以减少上下文开销，同时通常保持可靠的下游性能，尽管其有效性依赖于压缩器-读取器对和任务设置。这些发现表明，人类可读性、自然语言典型性和模型端语义恢复能力可以部分解耦，为未来LLM系统中模型本地表示的探索开辟了路径。
View on arXiv Download PDF AI Translation
cs.CL / 36 / 2606.19864

The Almost Intelligent Revolution: Options for Scaling Up Deliberation and Empowering People with AI

几乎智能的革命：扩大审议和赋权的人工智能选项
Sharoff, Serge
Abstract
The increasing prominence of Large Language Models (LLMs) in public discourse presents both opportunities and challenges for democratic deliberation. While red teaming strategies help mitigate specific risks, broader concerns persist regarding linguistic constraints, biases, and the sycophantic tendencies of LLMs. This chapter explores how LLMs can be used to significantly scale up and democratise deliberation, particularly in fostering inclusivity and empowering traditionally marginalised groups. Drawing on concepts from Systemic-Functional Linguistics, the chapter examines how variations across language users (for example, with respect to socio-demographic groups) and across language use (for example, with respect to communicative functions) shape participation in AI-supported deliberation. The chapter presents AI-driven deliberation studies and assesses their potential to scaffold argumentation, enhance access, and reduce the influence of exclusionary linguistic norms and biases which are embedded in prestigious registers. At the same time, the chapter cautions against both overclaiming, which leads to unrealistic expectations, and underclaiming, which risks missed opportunities for AI-assisted engagement. The chapter concludes by identifying future research directions to maximise the democratic potential of AI-assisted participation while embedding ethical safeguards to counteract the reproduction of linguistic inequalities.
Chinese Translation
大型语言模型（LLMs）在公共话语中的日益突出既带来了民主审议的机遇，也带来了挑战。尽管红队策略有助于缓解特定风险，但关于语言限制、偏见以及LLMs的谄媚倾向等更广泛的问题依然存在。本章探讨了如何利用LLMs显著扩大和民主化审议，特别是在促进包容性和赋权传统边缘化群体方面。通过借鉴系统功能语言学的概念，本章考察了语言使用者之间（例如，社会人口群体方面）和语言使用之间（例如，交际功能方面）的差异如何影响AI支持的审议参与。章节展示了基于AI的审议研究，并评估其在支撑论证、增强获取途径以及减少嵌入在权威语域中的排斥性语言规范和偏见影响方面的潜力。同时，本章警告不要过度宣称，以免导致不切实际的期望，也不要低估，以免错失AI辅助参与的机会。最后，本章通过确定未来研究方向，旨在最大化AI辅助参与的民主潜力，同时嵌入伦理保障，以对抗语言不平等的再生产。
View on arXiv Download PDF AI Translation
cs.CL / 37 / 2606.19881

REDACT: A Systematically Controlled Multilingual Benchmark for Personal Information Detection

REDACT：一个系统控制的多语言个人信息检测基准
Vats, Guneesh, Agrawal, Anubha, Singhal, Shikha, Dash, Ajita, Selvaraj, Praison, Jhawar, Vidhan, Chenna, Ranga Prasad, G, Bharadwaj Y M
Abstract
Benchmark infrastructure for personally identifiable information (PII) detection remains limited: existing corpora cover few entity types, use ad hoc generation conditions, and do not show which surface conditions cause detector failures. We present REDACT, a systematically controlled multilingual PII benchmark with 13,427 records, 324,078 entity annotations, 51 entity types, 4,127 surface-form patterns, and 25 languages across 9 scripts. A strength-2 covering-array sampler controls nine generation axes: domain, format, difficulty, length, density, code-switching, language, adjacency, and co-occurrence. Three entity-level metadata fields (disclosure status, disclosure form, and a GDPR-aligned sensitivity tier) enable stratified evaluation beyond aggregate or per-type F1. From the full benchmark, we evaluate five detectors (Presidio, GLiNER, the OpenAI Privacy Filter, GPT-4.1, and Claude Sonnet 4.6) on a locked, language-stratified sample of 1,000 records. Aggregate F1 masks an architecture-dependent failure structure: the rule-based detector performs poorly on the highest-stakes data, including HIGH-sensitivity categories (recall 0.07) and non-verbatim disclosure forms, while the LLM detectors remain more robust, with the HIGH tier as their strongest sensitivity slice. A three-model reference-free LLM-as-judge assessment corroborates that sensitivity-tier assignment is the task's hardest axis. We release the benchmark, schema, prompts, and stratified evaluation harness.
Chinese Translation
个人可识别信息（PII）检测的基准基础设施仍然有限：现有语料库覆盖的实体类型较少，使用的生成条件是临时的，并且未能显示哪些表面条件导致检测器失败。我们提出了REDACT，这是一个系统控制的多语言PII基准，包含13,427条记录、324,078个实体注释、51种实体类型、4,127种表面形式模式以及覆盖9种脚本的25种语言。一个强度为2的覆盖阵列采样器控制九个生成轴：领域、格式、难度、长度、密度、代码切换、语言、邻接和共现。三个实体级元数据字段（披露状态、披露形式和符合GDPR的敏感性等级）使得超越汇总或每种类型F1的分层评估成为可能。在完整基准中，我们在一个锁定的、按语言分层的1,000条记录样本上评估了五个检测器（Presidio、GLiNER、OpenAI隐私过滤器、GPT-4.1和Claude Sonnet 4.6）。汇总F1掩盖了一个依赖于架构的失败结构：基于规则的检测器在最高风险数据上表现不佳，包括高敏感性类别（召回率为0.07）和非逐字披露形式，而LLM检测器则保持更强的鲁棒性，以高敏感性等级作为其最强的敏感性切片。一个三模型无参考的LLM作为评判者的评估证实了敏感性等级分配是该任务最困难的轴。我们发布了基准、模式、提示和分层评估工具。
View on arXiv Download PDF AI Translation
cs.CL / 38 / 2606.19910

Light-weight Pronunciation Assessment via Discrete Speech Token Surprisal

基于离散语音标记惊讶度的轻量级发音评估
Sara, Syeda Faiza Ahmed, Chowdhury, Shammur Absar
Abstract
Training automated pronunciation assessment often relies on labeled learner errors or non-native corpora that are costly to collect. We propose a lightweight framework trained only on native speech resources, operating unsupervised or lightly calibrated with a small set of scored utterances. At inference, learner speech is discretized with an SSL encoder and a K-means codebook. A token language model trained on native sequences computes surprisal where higher surprisal indicates phonotactic deviation. We add a transcript-guided Text2DUnit--DTW module that predicts native token sequences from reference text and aligns them to acoustic tokens to derive error-sensitive features. Surprisal and alignment features are fused via simple regression. On SpeechOcean762, PCC improves from 0.60 to 0.66 with transcript guidance, near supervised baselines. Cross-dataset evaluation on L2-ARCTIC shows consistent gains.
Chinese Translation
自动化发音评估的训练通常依赖于标记的学习者错误或非母语语料，这些语料的收集成本高昂。我们提出了一种仅基于母语语音资源训练的轻量级框架，该框架在无监督或轻微校准的情况下使用一小组评分的发音进行操作。在推理阶段，学习者的语音通过自监督学习（SSL）编码器和K均值码本进行离散化。训练于母语序列的标记语言模型计算惊讶度，其中较高的惊讶度指示音位学偏差。我们增加了一个基于转录的Text2DUnit--DTW模块，该模块从参考文本中预测母语标记序列，并将其与声学标记对齐，以提取对错误敏感的特征。惊讶度和对齐特征通过简单的回归进行融合。在SpeechOcean762上，使用转录指导时，PCC从0.60提高到0.66，接近监督基线。在L2-ARCTIC上的跨数据集评估显示出一致的提升。
View on arXiv Download PDF AI Translation
cs.CL / 39 / 2606.19946

GEMS: Geometric Constraints Enable Multi-Semantic Superposition in LLMs

GEMS：几何约束使得大语言模型中的多语义叠加成为可能
Deng, Yu
Abstract
Activation steering controls model behavior by modifying intermediate hidden states at inference time without retraining. Existing methods handle only single-direction injection; when multiple semantic directions are superposed without constraints, the model collapses. We show that this collapse decomposes into two independently acting sources: distributional deviation, where additive perturbations accumulate in norm across layers and drive activations outside the training distribution, and directional interference, where non-orthogonal semantic vectors mutually dampen when superposed. These two sources define the design constraints that any training-free multi-directional intervention must address. As one instantiation of these principles, we propose GEMS, a training-free method that maps each source to a corresponding geometric constraint: norm-preserving weighted superposition and targeted attention-pathway injection for distributional deviation, and real-time orthogonalization for directional interference. On GSM8K, injecting three concurrent non-mathematical directions preserves accuracy at 98% (baseline 92%), while unconstrained addition collapses to 4%; on Wikitext-2, the same injection incurs only 2.2% PPL increase. Component ablation isolates the causal role of each constraint, and layer-level probes confirm that orthogonalized signals survive the FFN pathway and reach the output distribution with semantic specificity. Qualitative steering effects transfer across architectures from 3B to 31B.
Chinese Translation
激活引导通过在推理时修改中间隐藏状态来控制模型行为，而无需重新训练。现有方法仅处理单向注入；当多个语义方向在没有约束的情况下叠加时，模型会崩溃。我们表明，这种崩溃可以分解为两个独立作用的来源：分布偏差，其中加性扰动在层间累积并驱动激活超出训练分布，以及方向干扰，其中非正交语义向量在叠加时相互减弱。这两个来源定义了任何无训练的多方向干预必须解决的设计约束。作为这些原则的一种实例，我们提出了GEMS，一种无训练的方法，将每个来源映射到相应的几何约束：保持范数的加权叠加和针对分布偏差的目标注意力路径注入，以及实时正交化以应对方向干扰。在GSM8K上，注入三个同时存在的非数学方向使准确率保持在98%（基线为92%），而无约束的叠加崩溃至4%；在Wikitext-2上，相同的注入仅导致2.2%的PPL增加。组件消融实验隔离了每个约束的因果作用，层级探针确认正交化信号在前馈神经网络路径中存活并以语义特异性到达输出分布。定性引导效应在从3B到31B的不同架构间转移。
View on arXiv Download PDF AI Translation
cs.CL / 40 / 2606.20072

Source-Grounded Data Generation for Text-to-JSON Learning

基于源数据的文本到JSON学习的数据生成
Ahn, Sunghee, Son, Guijin, Yu, Youngjae
Abstract
From financial filings to clinical records, legacy industries rely heavily on long, unstructured documents to store high-value information. Reliably extracting this information into structured, machine-readable representations is a key prerequisite to making the contents accessible to automated systems. JSON is a natural target for such structured extraction, yet constructing reliable and scalable text-to-JSON training data remains challenging. To address this gap, we propose STAGE (Spreadsheet-grounded Text-to-JSON Artifact GEneration), a source-grounded data generation pipeline that constructs reports and JSON schema by using LLMs for scalable synthesis while validating ground-truth values against the underlying spreadsheet. Evaluations on STAGE-Eval, our source-grounded benchmark with an 851-example test set, show that STAGE produces stronger training data than existing approaches. This improves Qwen3-4B exact match from 31.37% to 74.27% and value accuracy from 45.46% to 90.69%.
Chinese Translation
从财务报告到临床记录，传统行业在存储高价值信息时严重依赖于冗长的非结构化文档。将这些信息可靠地提取为结构化的、机器可读的表示形式是使内容可被自动化系统访问的关键前提。JSON是这种结构化提取的自然目标，但构建可靠且可扩展的文本到JSON训练数据仍然具有挑战性。为了解决这一问题，我们提出了STAGE（基于电子表格的文本到JSON工件生成），这是一个基于源数据的数据生成管道，通过使用大型语言模型（LLMs）进行可扩展合成，同时将真实值与基础电子表格进行验证，从而构建报告和JSON模式。在我们的源数据基准STAGE-Eval上进行的评估显示，STAGE生成的训练数据优于现有方法。这使得Qwen3-4B的精确匹配率从31.37%提高到74.27%，值准确率从45.46%提高到90.69%。
View on arXiv Download PDF AI Translation
cs.CL / 41 / 2606.20089

IHUBERT: Vector-Based Semantic Deduplication and Domain-Balanced Pretraining for Persian Resources

IHUBERT：基于向量的语义去重与领域平衡预训练用于波斯语资源
Ghafouri, Arash, Firouzmandi, Mahdi, Saberi, Hossein, Ahangar, Mohammad Reza Hasani
Abstract
Persian pretrained language models (PLMs) are still limited by the scarcity of large-scale, high-quality pretraining corpora and by insufficient evaluation beyond standard classification and NER tasks. We present IHUBERT, a monolingual Persian PLM trained from scratch with the RoBERTa-base encoder (125M parameters) on a 45 GB curated subset of the Sepahr-Danesh collection (about 7-8B tokens). To improve corpus quality and reduce redundancy, we employ a multi-stage preprocessing pipeline that includes normalization, exact and near-duplicate removal, anonymization, and vector-database-based semantic deduplication for distribution balancing control across domains and registers. We additionally train a 139k-vocabulary BPE tokenizer on the full pretraining corpus to better capture Persian morphology and orthographic variation. IHUBERT is evaluated on seven Persian NLU benchmarks covering NER, sentiment analysis, topic classification, NLI, extractive question answering, and relation extraction, using task-standard metrics (entity-level F1, Macro-F1, EM/F1). IHUBERT achieves its strongest gains on extractive QA, ranking first on both PQuAD (F1 88.3542) and ParsiNLU-RC (F1 49.0987), and attains the best result on FarsTail (Macro-F1 0.8350). On NER and topic classification, it remains competitive (e.g., 0.8308 F1 on ParsTwiNER; 0.7953 Macro-F1 on DigiMag), while relation extraction remains the main remaining gap (0.6684 Macro-F1 on PERLEX). A controlled tokenizer ablation on the IHUBERT pretraining corpus shows that BPE yields slightly lower subword fragmentation than WordPiece at matched vocabulary size, supporting our tokenization design. Overall, IHUBERT advances Persian language modeling through semantically curated large-scale pretraining and broad evaluation across both classification and comprehension-oriented tasks.
Chinese Translation
波斯语预训练语言模型（PLMs）仍然受到大规模高质量预训练语料稀缺和评估不足的限制，尤其是在标准分类和命名实体识别（NER）任务之外。我们提出了IHUBERT，这是一个从零开始训练的单语波斯语PLM，使用RoBERTa-base编码器（125M参数）在Sepahr-Danesh集合的45GB精心挑选子集上进行训练（约7-8B标记）。为了提高语料质量并减少冗余，我们采用了多阶段预处理管道，包括规范化、精确和近似重复项移除、匿名化，以及基于向量数据库的语义去重，以控制跨领域和注册的分布平衡。此外，我们在完整的预训练语料上训练了一个包含139k词汇的BPE分词器，以更好地捕捉波斯语的形态学和正字法变异。IHUBERT在七个波斯语自然语言理解基准上进行评估，涵盖NER、情感分析、主题分类、自然语言推理（NLI）、抽取式问答和关系抽取，使用任务标准指标（实体级F1、宏F1、EM/F1）。IHUBERT在抽取式问答上取得了最强的进展，在PQuAD（F1 88.3542）和ParsiNLU-RC（F1 49.0987）上均排名第一，并在FarsTail（宏F1 0.8350）上取得最佳结果。在NER和主题分类上，它仍然具有竞争力（例如，在ParsTwiNER上F1 0.8308；在DigiMag上宏F1 0.7953），而关系抽取仍然是主要的差距（在PERLEX上宏F1 0.6684）。对IHUBERT预训练语料的受控分词器消融实验表明，在匹配的词汇大小下，BPE的子词碎片化略低于WordPiece，支持我们的分词设计。总体而言，IHUBERT通过语义精心策划的大规模预训练和广泛的评估，推动了波斯语建模的发展，涵盖了分类和理解导向的任务。
View on arXiv Download PDF AI Translation
cs.CL / 42 / 2606.20093

Self-Preference Is Weak or Absent in Verifiable Instruction-Following Revision: A Four-Model Test Under Genuine Authorship

可验证的遵循指令修订中自我偏好较弱或缺失：在真实作者身份下的四模型测试
Guey, William, Bougault, Pierrick
Abstract
Large language models (LLMs) increasingly review and revise text, including their own. A documented self-preference bias (models favoring their own generations when acting as judges) raises the question of whether models also resist valid corrections to their own writing. We test this in a setting where "valid" is decided not by another model but by a deterministic verifier: instruction-following revision on IFEval. A model writes a draft; the official IFEval checker confirms the draft violates a constraint and that a candidate edit fixes it; the model then accepts or rejects that edit either as the genuine in-context author or as a fresh model that sees the draft neutrally. Across four mid-tier model families and 85 author-versus-fresh comparisons, we find no detectable self-preference: authors reject verified-good fixes to their own drafts at essentially the same rate as fresh models judging the same drafts (gap -5.1 pp, 95% CI [-12.9, +2.7]). A self-skepticism hint from a smaller pilot did not replicate at scale. The one robust observation is qualitative: when authors do reject a verified-good fix, 97% of their stated reasons are flaw-catching rather than preference, that is, about the character of rejections, not an elevated rate. Effects smaller than ~13 pp cannot be excluded at this sample size.
Chinese Translation
大型语言模型（LLMs）越来越多地审查和修订文本，包括它们自己的文本。已有文献记录了自我偏好偏见（模型在担任评审时偏向于自己的生成文本），这引发了一个问题：模型是否也会抵制对其自身写作的有效修正。我们在一个设置中测试这一点，其中“有效”并不是由另一个模型决定，而是由一个确定性的验证器决定：在 IFEval 上进行遵循指令的修订。一个模型撰写草稿；官方 IFEval 检查器确认该草稿违反了某个约束，并且一个候选编辑可以修复该问题；然后，该模型作为真正的上下文作者或作为一个中立地查看草稿的新模型，接受或拒绝该编辑。在四个中等规模模型系列和 85 个作者与新模型的比较中，我们发现没有可检测的自我偏好：作者拒绝对其自身草稿的验证良好修正的比例与新模型评判相同草稿的比例基本相同（差距 -5.1 个百分点，95% 置信区间 [-12.9, +2.7]）。来自一个较小试点的自我怀疑暗示在大规模测试中未能复制。唯一的稳健观察是定性的：当作者拒绝一个验证良好的修正时，他们所陈述的理由中有 97% 是关于缺陷捕捉而非偏好，即关于拒绝的性质，而不是拒绝率的提高。在这个样本大小下，效应小于 ~13 个百分点无法被排除。
View on arXiv Download PDF AI Translation
cs.CL / 43 / 2606.20097

HydraHead: From Head-Level Functional Heterogeneity to Specialized Attention Hybridization

HydraHead：从头级功能异质性到专门化注意力混合
Tan, Zhentao, Chen, Wei, Shen, Jingyi, Liu, Yao, Shen, Xu, Wu, Yue, Ye, Jieping
Abstract
The quadratic complexity of attention poses a critical bottleneck for long-context processing, spurring interest in hybrid attention designs. Most open-source hybrid models adopt a layer-wise strategy. Yet, prior work has noted the inherent difficulty of integrating Linear Attention (LA) with Full Attention (FA), suggesting that the design space of attention hybridization remains underexplored. To probe this space, we conduct interpretability analysis and observe that layers exhibit block-wise functional similarity, while individual heads within the same layer display distinct functional specialization despite sharing input features. This head-level heterogeneity suggests that the head dimension provides a natural and principled granularity for fusing heterogeneous attention signals. Building on this insight, we introduce HydraHead, a novel architecture that hybridizes FA and LA along the head axis. HydraHead features two key innovations: (1) an interpretability-driven selection strategy that identifies retrieval-critical heads and preserves FA only for them, and (2) a scale-normalized fusion module that reconciles the distributional gap between FA and LA head outputs. By leveraging a three-stage transfer pipeline with parameter reuse and distillation, we achieve high-performance hybrid models with minimal training overhead. Under a unified training setup, HydraHead outperforms other hybrid designs in long-context tasks while maintaining strong general reasoning. With interpretability-driven head selection, it matches a 3:1 layer-wise hybrid's long-context performance at a 7:1 LA-to-FA ratio. Crucially, trained on only 15B tokens, HydraHead achieves over 69% improvement over the baseline at 512K context length, approaching Qwen3.5, a leading model of comparable size with a native context length of 256K. This highlights the significant scaling potential of head-level hybridization.
Chinese Translation
注意力的二次复杂性对长上下文处理构成了关键瓶颈，激发了对混合注意力设计的兴趣。大多数开源混合模型采用层级策略。然而，先前的研究指出，将线性注意力（Linear Attention, LA）与全注意力（Full Attention, FA）结合的固有困难，表明注意力混合的设计空间仍未得到充分探索。为了探讨这一空间，我们进行了解释性分析，观察到层之间表现出块级功能相似性，而同一层内的各个头尽管共享输入特征，却展现出明显的功能专业化。这种头级异质性表明，头维度为融合异质注意力信号提供了自然且合理的粒度。基于这一见解，我们提出了HydraHead，这是一种沿头轴混合FA和LA的新型架构。HydraHead具有两个关键创新：（1）一种基于可解释性的选择策略，识别出检索关键头并仅为其保留FA；（2）一种规模归一化的融合模块，调和FA和LA头输出之间的分布差距。通过利用具有参数重用和蒸馏的三阶段转移管道，我们实现了高性能的混合模型，训练开销最小。在统一的训练设置下，HydraHead在长上下文任务中超越了其他混合设计，同时保持强大的通用推理能力。通过基于可解释性的头选择，它在7:1的LA与FA比例下，匹配了3:1层级混合的长上下文性能。重要的是，HydraHead仅在15B标记上训练，在512K上下文长度下实现了超过69%的基线改进，接近Qwen3.5，这是一种具有相似规模且原生上下文长度为256K的领先模型。这突显了头级混合的显著扩展潜力。
View on arXiv Download PDF AI Translation
cs.CL / 44 / 2606.20113

When Does Streaming Tool Use Help? Characterizing Tool-Intent Stabilization in Streaming Retrieval-Augmented Generation

何时工具使用有助于提升效果？流媒体检索增强生成中的工具意图稳定性特征
Galbraith, Elroy
Abstract
Streaming Retrieval-Augmented Generation (Streaming RAG) reduces user-perceived latency by issuing tool queries in parallel with ongoing user input, before the utterance is complete. Reported gains are aggregate, yet the mechanism's benefit is fundamentally query-intrinsic: speculation can only help when the correct tool query becomes determinable before the user stops speaking or typing. We isolate and measure this property -- tool-intent stabilization, the point in the input stream at which a speculative query's retrieval converges to the answer-bearing result. On the CRAG benchmark (1371 validation questions) we (i) measure the distribution of stabilization, (ii) derive a model-agnostic bound H on the portion of tool latency that can be hidden behind the user's remaining input, as a function of tool latency L and input cadence {\delta}, (iii) validate against a working streaming pipeline that realized savings meet or exceed this bound, and (iv) identify which query properties predict early versus late stabilization. The study requires no model training and runs on commodity CPU hardware. We find that at a realistic operating point (L=600ms, {\delta}=3w/s, {\theta}=0.8), 73.9% of queries across the full benchmark admit substantial latency hiding -- a blended figure that mixes sufficiency stabilization on the 21.3% of questions where gold evidence is verbatim-present and BM25-retrievable (95.2% streamable on this favorable slice) with a grounding-free top-1-settling fallback on the remainder. On the favorable slice, {\phi}_suf is bracketed to [0.26, 0.281] by exact and relaxed grounding -- both early. Question type produces a significant but coarse early/late split (Kruskal-Wallis p=0.017, epsilon^2=0.04), directly informing when a learned speculative trigger is worth its cost.
Chinese Translation
流媒体检索增强生成（Streaming RAG）通过在用户输入进行时并行发出工具查询，从而减少用户感知的延迟，甚至在用户话语尚未完成时就开始查询。虽然报告的收益是综合性的，但该机制的好处本质上是查询内在的：只有在用户停止说话或输入之前，正确的工具查询才能被确定，从而推测才有帮助。我们隔离并测量这一特性——工具意图稳定性，即在输入流中，推测查询的检索收敛到答案的结果的点。在CRAG基准测试（1371个验证问题）上，我们（i）测量稳定性的分布，（ii）推导出一个与模型无关的界限H，表示可以隐藏在用户剩余输入后的工具延迟的比例，作为工具延迟L和输入节奏{ heta}的函数，（iii）验证一个实际的流媒体管道，发现节省的延迟符合或超过该界限，以及（iv）识别哪些查询特性预测早期与晚期的稳定性。本研究不需要模型训练，并在普通CPU硬件上运行。我们发现，在一个现实的操作点（L=600ms，{ heta}=3w/s，{ heta}=0.8），在整个基准测试中，73.9%的查询能够显著隐藏延迟——这一综合数据混合了在21.3%问题中金证据逐字存在且可通过BM25检索的充分稳定性（在这一有利样本中，95.2%可流媒体）与其余部分的无基础的前1个结果的后备。对于有利样本，{ heta}_suf被精确和放宽的基础限制在[0.26, 0.281]之间——均为早期。问题类型产生了显著但粗略的早期/晚期分割（Kruskal-Wallis p=0.017，epsilon^2=0.04），直接影响何时学习的推测触发器值得其成本。
View on arXiv Download PDF AI Translation
cs.CL / 45 / 2606.20152

From Texts to Scores: Tracing the Emergence of Essay Quality Representations in Large Language Models

从文本到分数：追踪大型语言模型中论文质量表征的出现
Zuo, Jiaxu, You, Mu, Lan, Kaixin, Fang, Tao, Huo, Yujia, Shen, Henghua, Chao, Lidia S., Wong, Derek F.
Abstract
Recent advances in Large Language Models (LLMs) have substantially transformed Automated Essay Scoring (AES), yet the internal mechanisms underlying LLM-based scoring remain poorly understood. In this work, we systematically analyze the hidden representations of eight LLMs across two English essay datasets (ASAP++, CSEE) and one Portuguese dataset (ENEM). Using linear probing, cross-prompt generalization, dimensionality reduction, and neuron-level analyses, we find consistent evidence that essay quality information is encoded in a linearly accessible form within LLM representations. These representations emerge progressively across layers, remain robust across prompting strategies, and partially transfer across essay prompts despite differences in scoring rubrics. In addition, nonlinear probes provide only marginal and inconsistent improvements over linear probes, suggesting that most essay quality information is already linearly decodable. We further identify individual ``essay scoring neurons'' whose activations strongly correlate with essay scores and whose behavior is sensitive to targeted intervention. Moreover, the layer-wise distribution of these neurons systematically shifts with essay length, with longer essays relying more heavily on deeper layers. Overall, our findings provide evidence that LLMs encode structured representations related to essay quality and offer new insights into the interpretability of LLM-based AES systems.
Chinese Translation
近期大型语言模型（LLMs）的进展显著改变了自动化论文评分（AES），然而基于LLM的评分背后的内部机制仍然不甚明了。在本研究中，我们系统地分析了八个LLM在两个英语论文数据集（ASAP++、CSEE）和一个葡萄牙语数据集（ENEM）上的隐藏表征。通过线性探测、跨提示泛化、降维和神经元级分析，我们发现论文质量信息以线性可访问的形式编码在LLM表征中。这些表征在层级中逐渐出现，在不同的提示策略下保持稳健，并且尽管评分标准存在差异，仍部分转移于不同的论文提示。此外，非线性探测器仅提供了边际和不一致的改进，表明大多数论文质量信息已经可以线性解码。我们进一步识别出个别“论文评分神经元”，其激活与论文分数高度相关，并且其行为对有针对性的干预敏感。此外，这些神经元在层级分布上随着论文长度的变化而系统性地转变，较长的论文更依赖于更深的层级。总体而言，我们的研究结果提供了证据，表明LLMs编码了与论文质量相关的结构化表征，并为基于LLM的AES系统的可解释性提供了新的见解。
View on arXiv Download PDF AI Translation
cs.CL / 46 / 2606.20164

MedRLM: Recursive Multimodal Health Intelligence for Long-Context Clinical Reasoning, Sensor-Guided Screening, Evidence-Grounded Decision Support, and Community-to-Tertiary Referral Optimization

MedRLM：用于长上下文临床推理、传感器引导筛查、基于证据的决策支持和社区到三级转诊优化的递归多模态健康智能
Aueawatthanaphisut, Aueaphum
Abstract
Real-world clinical decision support requires reasoning over heterogeneous and longitudinal patient information rather than answering isolated medical questions. However, current medical large language models and retrieval-augmented generation systems often rely on single-step prompting or retrieval, which can be fragile when clinical evidence is distributed across long electronic health records, medical images, sensor streams, guidelines, and referral constraints. This paper proposes MedRLM, a Recursive Multimodal Health Intelligence framework for long-context clinical reasoning, sensor-guided screening, and community-to-tertiary referral support. Instead of compressing all patient information into one prompt, MedRLM treats the patient case as an external clinical environment that can be recursively inspected, decomposed, retrieved, verified, and synthesized. The framework coordinates specialized agents for clinical text, longitudinal EHR, medical imaging, physiological sensor signals, guideline retrieval, uncertainty auditing, and referral planning. It further introduces a Clinical Evidence Graph Memory to connect patient-specific observations with retrieved evidence, standardized definitions, sensor-derived biomarkers, and referral criteria. A sensor-guided recursive triggering mechanism activates deeper reasoning when abnormal physiological or behavioral patterns are detected, while uncertainty-gated refinement supports clinician review for high-risk or low-confidence cases. We also outline a real-data evaluation design using public and credentialed clinical datasets spanning EHR, radiology, ECG, ICU time series, and referral-proxy outcomes. MedRLM aims to move medical AI from static question answering toward auditable, multimodal, and workflow-aware clinical decision support.
Chinese Translation
现实世界中的临床决策支持需要对异构和纵向的患者信息进行推理，而不是仅仅回答孤立的医学问题。然而，目前的医学大型语言模型和检索增强生成系统通常依赖于单步提示或检索，当临床证据分布在长电子健康记录、医学影像、传感器流、指南和转诊约束中时，这种方法可能会显得脆弱。本文提出了MedRLM，一个递归多模态健康智能框架，旨在支持长上下文的临床推理、传感器引导的筛查以及社区到三级的转诊支持。MedRLM并不将所有患者信息压缩为一个提示，而是将患者案例视为一个可以递归检查、分解、检索、验证和综合的外部临床环境。该框架协调了针对临床文本、纵向电子健康记录、医学影像、生理传感器信号、指南检索、不确定性审计和转诊规划的专业代理。它进一步引入了临床证据图记忆，以将患者特定的观察与检索到的证据、标准化定义、传感器衍生的生物标志物和转诊标准连接起来。当检测到异常的生理或行为模式时，传感器引导的递归触发机制会激活更深层次的推理，而不确定性门控的细化则支持临床医生对高风险或低信心案例的审查。我们还概述了一种使用公共和认证临床数据集的真实数据评估设计，这些数据集涵盖了电子健康记录、放射学、心电图、重症监护时间序列和转诊代理结果。MedRLM旨在将医学人工智能从静态问答转向可审计的、多模态的、并且关注工作流程的临床决策支持。
View on arXiv Download PDF AI Translation
cs.CL / 47 / 2606.20179

ReNikud: Audio-Supervised Hebrew Grapheme-to-Phoneme Conversion

ReNikud：音频监督的希伯来字素到音素转换
Melichov, Maxim, Kolani, Yakov, Alper, Morris
Abstract
Grapheme-to-phoneme (G2P) conversion for Modern Hebrew is needed for applications like text-to-speech (TTS), but is challenging due to the language's abjad writing system, which leaves vowels largely unwritten, creating substantial ambiguity. Standard approaches first predict vowel diacritics (nikud) to produce International Phonetic Alphabet (IPA) transcriptions, but this is limited: vocalization data is scarce and laborious to produce, it does not specify features such as lexical stress, and it reflects formal grammatical rules rather than everyday spoken pronunciation. Direct sequence-to-sequence IPA prediction, meanwhile, struggles on limited data and fails to exploit the character-level alignment characteristic of abjads. Our method, ReNikud, overcomes these limitations with two key insights: (1) Weak audio supervision via a phoneme-based automatic speech recognition (ASR) pseudo-labeling pipeline on thousands of hours of unlabeled Hebrew audio, yielding phonemic transcriptions that reflect natural spoken norms without manual annotation. (2) A pseudo-vocalization architecture that predicts IPA phonemes at each character position, enforcing character-level alignment as an inductive bias. Results on existing Hebrew G2P benchmarks and the new targeted MILIM benchmark for spoken Hebrew show that ReNikud surpasses previous state-of-the-art methods. We will release our code and trained models to support further work on Hebrew TTS and speech technologies.
Chinese Translation
现代希伯来语的字素到音素（G2P）转换在文本转语音（TTS）等应用中是必要的，但由于该语言的辅音字母书写系统，元音大多未被书写，导致了显著的歧义，这使得转换变得具有挑战性。标准方法首先预测元音标记（nikud），以生成国际音标（IPA）转录，但这存在局限性：元音化数据稀缺且生成过程繁琐，无法指定诸如词汇重音等特征，并且反映的是正式语法规则而非日常口语发音。与此同时，直接的序列到序列IPA预测在有限数据上表现不佳，未能利用辅音字母的字符级对齐特性。我们的方法ReNikud通过两个关键见解克服了这些局限性：（1）通过基于音素的自动语音识别（ASR）伪标注管道，对数千小时未标记的希伯来音频进行弱音频监督，从而生成反映自然口语规范的音素转录，无需人工标注。（2）一种伪元音化架构，在每个字符位置预测IPA音素，强制字符级对齐作为归纳偏置。在现有的希伯来G2P基准测试和新的针对口语希伯来语的MILIM基准测试中，ReNikud的表现超越了之前的最先进方法。我们将发布我们的代码和训练模型，以支持进一步的希伯来TTS和语音技术研究。
View on arXiv Download PDF AI Translation
cs.CL / 48 / 2606.20198

Pitch Spelling Jazz Lead Sheets, Solo Transcriptions, Classical Piano and Monophonic Scores

音高拼写爵士乐乐谱、独奏转录、古典钢琴及单声部乐谱
Bouquillard, Augustin, Jacquemard, Florent
Abstract
We present an algorithm for pitch spelling and key estimation. Given an input in MIDI-like format, containing information on note pitches (expressed in semitones relative to the lowest reference note) and bar boundaries, it estimates the appropriate note names, a global Key Signature, and a local scale for each bar. This related information elements are evaluated jointly during two stages of optimisation. During an initial 'modal' stage, a probable scale is proposed for each bar, minimising the number of accidentals to be printed in the printed score with a shortest-path search. Then, during a second stage called 'tonal', these local scales are used to estimate the Key Signature and note names that would result in the best musical notation for the entire piece. We present evaluations conducted on datasets comprising a variety of digital musical scores: jazz lead sheets taken from the Real Book, transcriptions of recordings of jazz soli and bass lines, traditional tunes, as well as classical scores for piano and monophonic instruments. Our procedure was originally designed for use in music transcription, specifically for building digital collections of jazz solos transcribed from audio recordings, for the purposes of music analysis, teaching and the preservation of cultural heritage. This method should also prove useful for other tasks related to the processing of musical notation. Furthermore, to this end, we have defined new distances between various common jazz scales, which may be of some interest to musicological studies.
Chinese Translation
我们提出了一种音高拼写和调性估计的算法。给定一个类似MIDI格式的输入，其中包含音符音高（相对于最低参考音符以半音为单位表示）和小节边界的信息，该算法估计适当的音符名称、全局调号和每个小节的局部音阶。这些相关信息元素在两个优化阶段中共同评估。在初始的“模式”阶段，为每个小节提出一个可能的音阶，通过最短路径搜索最小化打印乐谱中意外音符的数量。然后，在第二个称为“调性”的阶段，这些局部音阶用于估计调号和音符名称，以便为整首作品生成最佳的音乐记谱。我们在包含多种数字音乐乐谱的数据集上进行了评估：来自《Real Book》的爵士乐乐谱、爵士独奏和低音线录音的转录、传统曲调，以及古典钢琴和单声部乐器的乐谱。我们的程序最初是为音乐转录而设计，特别是用于构建从音频录音转录的爵士独奏的数字收藏，以便进行音乐分析、教学和文化遗产的保护。这种方法也应对与音乐记谱处理相关的其他任务有所帮助。此外，为此，我们定义了不同常见爵士音阶之间的新距离，这可能对音乐学研究具有一定的兴趣。
View on arXiv Download PDF AI Translation
cs.CL / 49 / 2606.20212

CzechDocs: A Multiway Parallel Dataset of Formatted Documents for Minority Languages in Czechia

CzechDocs：捷克少数语言格式化文档的多向平行数据集
Jon, Josef, Bojar, Ondřej
Abstract
We present CzechDocs, a multiway parallel dataset of formatted documents (HTML, DOCX, and PDF) covering Czech and minority languages used in Czechia-primarily Ukrainian and English, with smaller portions of Vietnamese, Russian and other languages. The dataset is designed to support the evaluation of machine translation systems that aim to preserve document formatting during translation. We provide a comparison of the most common approaches to format-preserving machine translation on a validation subset of the dataset. This validation split, together with the evaluation toolkit, is publicly released for further research. A held-out test split will be reserved for a future shared task focused on document-level translation with formatting preservation.
Chinese Translation
我们提出了CzechDocs，这是一个涵盖捷克及其少数语言（主要是乌克兰语和英语，以及小部分越南语、俄语和其他语言）的格式化文档（HTML、DOCX和PDF）的多向平行数据集。该数据集旨在支持评估旨在保留文档格式的机器翻译系统。我们对数据集的验证子集上最常见的格式保留机器翻译方法进行了比较。该验证子集及评估工具包已公开发布，以便进一步研究。一个保留的测试子集将用于未来专注于文档级翻译和格式保留的共享任务。
View on arXiv Download PDF AI Translation
cs.CL / 50 / 2606.20225

Actionable Activation Directions for Detecting and Mitigating Emergent Misalignment Across Language Model Families

可操作的激活方向用于检测和缓解语言模型家族中的新兴不一致性
Syed, Abdul Rafay
Abstract
Fine-tuning language models on insecure code induces emergent misalignment with poorly understood internal structure. We investigate whether this misalignment corresponds to a causally actionable activation-space direction shared across architectures. Across four instruction-tuned model families (Qwen2.5-1.5B, Gemma-2-2B, Llama-3.2-1B, Ministral-3-3B) finetuned identically, a difference-in-means direction achieves 99.6% separation of aligned and misaligned activations at each model's final layer. Causal steering by subtracting this direction reduces code spillover by 21-51 points, while a secure-code control confirms content specificity. Cross-architecture transfer via ridge regression maps yields large behavioral suppression (up to 46 points) but fails specificity controls as random and orthogonal directions perform comparably. We identify a two-tier specificity structure: within-model directions are causally specific and actionable; cross-model directions are causally real but non-specific. An asymmetric transfer topology emerges, with Gemma and Qwen acting as geometric donors and Llama as a receiver. These findings define the limits of linear cross-architecture correction and recommend within-model probing for auditing.
Chinese Translation
在不安全代码上微调语言模型会导致与内部结构不明的出现性不一致。我们研究这种不一致是否对应于在不同架构间共享的因果可操作激活空间方向。在四个经过相同微调的指令调优模型家族（Qwen2.5-1.5B、Gemma-2-2B、Llama-3.2-1B、Ministral-3-3B）中，均值差异方向在每个模型的最终层实现了99.6%的对齐和不对齐激活的分离。通过减去该方向进行因果引导，代码溢出减少了21-51个点，而安全代码控制则确认了内容特异性。通过岭回归映射的跨架构转移产生了显著的行为抑制（高达46个点），但在特异性控制方面失败，因为随机和正交方向的表现相当。我们识别出一种两级特异性结构：模型内方向是因果特定且可操作的；模型间方向是因果真实但不特定的。出现了一种不对称的转移拓扑，Gemma和Qwen作为几何捐赠者，Llama作为接收者。这些发现定义了线性跨架构校正的限制，并推荐在模型内进行探测以进行审计。
View on arXiv Download PDF AI Translation
cs.CL / 51 / 2606.20255

The Register Gap: A Meaning Intelligence Framework for Nigerian Public Discourse

注册差距：尼日利亚公共话语的意义智能框架
Achi, Celestine
Abstract
We introduce the Meaning Intelligence Framework (MIF), a nine-dimension annotation and evaluation schema for Nigerian public discourse that separates surface sentiment from true communicative intent. Existing benchmarks for Nigerian languages, including NaijaSenti and AfriSenti, treat sentiment classification as a three-way polarity task (positive, negative, neutral). We argue that the dominant failure mode of AI systems on Nigerian discourse is not translation failure but context failure: the same utterance carries opposite pragmatic force depending on speaker, audience, and situation. The MIF operationalises this insight across nine scored dimensions: register, surface sentiment, true intent, irony, coded subtext, risk tier, annotator confidence, speaker emotion, and recommended communications action. We construct a 30-item calibration dataset spanning Standard English, Nigerian English, Nigerian Pidgin, and code-mixed registers, and evaluate a frontier language model (Gemini 2.5 Flash) under zero-shot and schema-informed prompting conditions. The headline finding is the Register Gap: zero-shot register classification accuracy is 33.3%, rising to 73.3% (+40 points) when the model receives the MIF schema in-context. The composite Meaning Intelligence Score increases by 5.4 points (73.2 to 78.6) under schema-informed prompting, with the largest practical gains in register identification, coded-subtext detection (+10 points), and strategic action recommendation (+10.3 points). We release the framework specification, annotation guidelines, and the 30-item public calibration set to support reproducibility, while retaining a private holdout corpus for contamination-protected evaluation.
Chinese Translation
我们介绍了意义智能框架（Meaning Intelligence Framework, MIF），这是一个针对尼日利亚公共话语的九维注释和评估方案，旨在将表面情感与真实交流意图区分开来。现有的尼日利亚语言基准，包括NaijaSenti和AfriSenti，将情感分类视为三向极性任务（积极、消极、中立）。我们认为，人工智能系统在尼日利亚话语中的主要失败模式不是翻译失败，而是上下文失败：同一句话根据说话者、听众和情境的不同，可能具有相反的语用力量。MIF在九个评分维度上实现了这一洞察：注册、表面情感、真实意图、讽刺、编码潜台词、风险等级、注释者信心、说话者情感和推荐的沟通行动。我们构建了一个包含30个项目的校准数据集，涵盖标准英语、尼日利亚英语、尼日利亚皮钦语和混合代码注册，并在零样本和基于框架的提示条件下评估了前沿语言模型（Gemini 2.5 Flash）。主要发现是注册差距：零样本注册分类准确率为33.3%，当模型在上下文中接收MIF框架时，准确率上升至73.3%（增加40个百分点）。在基于框架的提示下，综合意义智能评分提高了5.4分（从73.2提高到78.6），在注册识别、编码潜台词检测（增加10分）和战略行动推荐（增加10.3分）方面获得了最大的实际收益。我们发布了框架规范、注释指南和30个项目的公共校准集，以支持可重复性，同时保留一个私有的保留语料库以进行防污染评估。
View on arXiv Download PDF AI Translation
cs.CL / 52 / 2606.20287

PsyScore: A Psychometrically-Aware Framework for Trait-Adaptive Essay Scoring and ZPD-Scaffolded Feedback

PsyScore：一个心理测量意识框架，用于特质自适应的作文评分和ZPD支架反馈
Xia, Wei, Wu, Jin, Shi, Haoran, Wang, Xiangyu, Zheng, Chanjin
Abstract
Effective Automated Essay Scoring (AES) are expected to support both reliable assessment and actionable instructional feedback. However, existing approaches often treat scoring and feedback as separate components: neural scoring models provide limited interpretability, while Large Language Model (LLM)-based feedback is typically insensitive to learners proficiency levels. To address this fragmentation, this work proposes PsyScore, a psychometrically-aware framework that integrates diagnostic assessment with instructional scaffolding through a shared latent ability representation. PsyScore comprises three key modules: a Trait-Adaptive Neural IRT Scorer that incorporates the Graded Partial Credit Model (GPCM) into a neural architecture, enabling the precise estimation of student ability while maintaining psychometric interpretability, a ZPD-Scaffolded Feedback Generator, which conditions multi-agent feedback strategies on the diagnosed ability parameter to adapt instructional focus across different proficiency levels, and a Multi-Perspective Feedback Evaluation Strategy that assesses feedback quality via pairwise preference judgements and student revision simulations. Experiments on the ASAP++ dataset demonstrate that PsyScore achieves competitive scoring performance while providing more pedagogically aligned feedback.
Chinese Translation
有效的自动化作文评分（AES）应支持可靠的评估和可操作的教学反馈。然而，现有的方法往往将评分和反馈视为独立的组成部分：神经评分模型提供的可解释性有限，而基于大型语言模型（LLM）的反馈通常对学习者的熟练程度不敏感。为了解决这一碎片化问题，本研究提出了PsyScore，一个心理测量意识框架，通过共享的潜在能力表示将诊断评估与教学支架相结合。PsyScore包括三个关键模块：一个特质自适应神经IRT评分器，它将分级部分信用模型（GPCM）融入神经架构中，使得能够精确估计学生能力，同时保持心理测量的可解释性；一个ZPD支架反馈生成器，它根据诊断的能力参数调整多智能体反馈策略，以适应不同熟练程度的教学重点；以及一个多视角反馈评估策略，通过成对偏好判断和学生修订模拟来评估反馈质量。对ASAP++数据集的实验表明，PsyScore在评分性能上具有竞争力，同时提供了更符合教学要求的反馈。
View on arXiv Download PDF AI Translation
cs.CL / 53 / 2606.20369

CATCH-ME if you RAG: a dataset of Contextually Annotated multi-Turn Counterspeech against Hate and Misinformation Exchanges

如果你能抓住我：一个关于针对仇恨和虚假信息交流的上下文标注多轮反对言论的数据集
Bonaldi, Helena, Martone, Genoveffa, Guerini, Marco
Abstract
Online hate speech and misinformation frequently overlap, yet NLP research has mainly treated them in isolation. While LLMs represent a scalable solution for assisting humans in the generation of counterspeech for both threats, zero-shot models frequently generate repetitive and vague responses, underscoring the need for high-quality examples to steer model generation. However, existing counterspeech datasets against the overlap of hate and misinformation are scarce and limited to single-turn English dialogues, while real-life interactions span across multiple turns and languages. To bridge this gap, we introduce the first large-scale, expert-curated, multilingual dataset of dialogues tackling the intersection of hate and misinformation. To ensure factual grounding, the dialogues are also anchored in verified external knowledge (i.e., fact-checking articles and NGO reports) and include document- and chunk-level span annotations, making it directly applicable for RAG systems. Covering five languages and targeting hate directed at seven marginalized groups, this novel resource enables the training and evaluation of more persuasive, factually grounded counterspeech models.
Chinese Translation
在线仇恨言论和虚假信息经常交织在一起，但自然语言处理（NLP）研究主要将它们孤立对待。尽管大型语言模型（LLMs）代表了一种可扩展的解决方案，能够帮助人类生成针对这两种威胁的反对言论，但零样本模型常常生成重复和模糊的回应，这突显了高质量示例在引导模型生成中的必要性。然而，现有针对仇恨和虚假信息交集的反对言论数据集稀缺，且仅限于单轮英语对话，而现实生活中的互动则跨越多个回合和语言。为了解决这一问题，我们推出了第一个大规模、专家策划的多语言对话数据集，专注于处理仇恨与虚假信息的交集。为了确保事实基础，这些对话还基于经过验证的外部知识（即事实核查文章和非政府组织报告），并包括文档级和片段级的跨度标注，使其能够直接应用于检索增强生成（RAG）系统。该数据集覆盖五种语言，针对七个边缘化群体的仇恨言论，成为训练和评估更具说服力、事实基础的反对言论模型的创新资源。
View on arXiv Download PDF AI Translation
cs.CL / 54 / 2606.20482

Your Mouse and Eyes Secretly Leak Your Preference: LLM Alignment using Implicit Feedback from Users

你的鼠标和眼睛秘密泄露了你的偏好：基于用户隐性反馈的LLM对齐
Chang, Haw-Shiuan, Gomez, Jeffrey, Patwari, Mehul, Sajith, Aryan, Zamani, Hamed
Abstract
To align a Large Language Model (LLM), most existing methods collect explicit human feedback and train a reward model to predict the human preference based on the response text. These existing methods have two key limitations. First, the users rarely provide explicit feedback for LLM responses, which makes the high-quality preference annotation expensive to collect. Second, the methods do not leverage implicit human feedback, which has proven vital to the economic moats of Internet giants. To quantify the value of implicit feedback, we build a new dataset called IFLLM, which collects 1336 multi-turn questions from the 59 Mechanical Turk workers, their mouse trajectories, and eye gazing points to the LLMs' responses from their webcams. IFLLM shows that the users have very diverse types of gazing behavior and mouse trajectories. Our reward model based on the implicit user feedback boosts the accuracy of the text-based reward model from 55% to 64% and nearly triples the relative response quality improvements after applying the DPO to eight LLMs, demonstrating the value of implicit feedback in the wild. Our data collection website, dataset, and codes can be found at https://github.com/themehulpatwari/llm-implicit-feedback/.
Chinese Translation
为了对齐大型语言模型（LLM），大多数现有方法收集显式的人类反馈，并训练奖励模型以根据响应文本预测人类偏好。这些现有方法存在两个主要局限性。首先，用户很少为LLM的响应提供显式反馈，这使得高质量的偏好标注收集成本高昂。其次，这些方法未能利用隐性人类反馈，而隐性反馈在互联网巨头的经济护城河中被证明至关重要。为了量化隐性反馈的价值，我们构建了一个新的数据集，称为IFLLM，该数据集收集了来自59名Mechanical Turk工作者的1336个多轮问题、他们的鼠标轨迹以及通过网络摄像头记录的眼动点。IFLLM显示用户的注视行为和鼠标轨迹类型非常多样。基于隐性用户反馈的奖励模型将基于文本的奖励模型的准确性从55%提升至64%，并在对八个LLM应用DPO后，几乎使相对响应质量的改善增加了三倍，展示了隐性反馈在实际应用中的价值。我们的数据收集网站、数据集和代码可以在https://github.com/themehulpatwari/llm-implicit-feedback/找到。
View on arXiv Download PDF AI Translation
cs.CL / 55 / 2606.20487

Beyond Global Replanning: Hierarchical Recovery for Cross-Device Agent Systems

超越全球重规划：跨设备代理系统的层次恢复
Yao, Shu, Luo, Yuhua, Long, Qian, Fan, Jingru, Yu, Zhuoyuan, Wang, Yuheng, Wu, Lin, Dang, Yufan, Li, Huatao, Qian, Chen
Abstract
Real-world computer-use tasks often span multiple applications and devices, requiring agents to coordinate heterogeneous environments under dynamic runtime failures. Existing multi-device agent systems support task decomposition and cross-device assignment, but recovery remains largely coarse-grained: when execution fails, they typically retry the same strategy, reassign the subtask, or revise the global plan, without systematically modeling the device-local strategy space. This limits their ability to distinguish failures that can be repaired within the current device from those that require cross-device replanning. We propose \textbf{H-RePlan}, a hierarchical replanning framework for multi-device agents with unified API--CLI--GUI execution. H-RePlan equips each device with interchangeable execution strategies and separates device-local strategy recovery from orchestrator-level global replanning through a compact cross-layer failure abstraction. To evaluate this capability, we introduce \textbf{HeraBench}, a fault-injected benchmark that constructs cross-device workflows over Linux and Android devices and injects strategy- and device-level failures. Experiments show that H-RePlan substantially outperforms single-strategy and coarse-grained multi-device baselines, achieving higher completion, instruction adherence, and perfect-pass rates while reducing the token cost required for reliable end-to-end success. These results demonstrate that scope-aware hierarchical recovery is essential for robust multi-device agent execution.
Chinese Translation
现实世界中的计算机使用任务通常跨越多个应用程序和设备，要求代理在动态运行时故障下协调异构环境。现有的多设备代理系统支持任务分解和跨设备分配，但恢复仍然主要是粗粒度的：当执行失败时，它们通常重试相同的策略、重新分配子任务或修订全局计划，而没有系统地建模设备本地的策略空间。这限制了它们区分可以在当前设备内修复的故障与需要跨设备重规划的故障的能力。我们提出了 extbf{H-RePlan}，一个用于多设备代理的层次重规划框架，具有统一的 API--CLI--GUI 执行。H-RePlan 为每个设备配备了可互换的执行策略，并通过紧凑的跨层故障抽象将设备本地策略恢复与协调者级别的全局重规划分开。为了评估这一能力，我们引入了 extbf{HeraBench}，一个故障注入基准，构建了跨 Linux 和 Android 设备的跨设备工作流，并注入策略和设备级故障。实验表明，H-RePlan 显著优于单一策略和粗粒度多设备基线，达到了更高的完成率、指令遵循率和完美通过率，同时减少了实现可靠端到端成功所需的代币成本。这些结果表明，范围感知的层次恢复对于稳健的多设备代理执行至关重要。
View on arXiv Download PDF AI Translation
cs.CL / 56 / 2606.20527

StylisticBias: A Few Human Visual Cues Drive Most Social Biases in MLLMs

StylisticBias：少数人类视觉线索驱动大多数多模态大语言模型中的社会偏见
Kolli, Shaghayegh, Cavelius, Timo, Nikeghbal, Nafiseh, Dalal, Samantha, Diesner, Jana
Abstract
Multimodal large language models (MLLMs) are increasingly deployed in personally and societally consequential settings, yet the visual cues that shape how these models judge people remain poorly understood. Prior work often compares different (groups of) individuals, making it difficult to separate appearance effects from identity differences. We introduce StylisticBias, a controlled benchmark for evaluating attribute-level social bias in MLLMs. We generate 500 photorealistic base faces and create about 50 single-attribute variations per face, producing about 25K images. This design keeps identity fixed and changes one visual attribute at a time. It lets us measure how specific cues shift model judgments. We evaluate six MLLMs across 25 binary social judgment scenarios. We find that age and body type dominate identity-level effects, while fashion style and other visual cues drive the largest attribute-level shifts. We further find that about 15 attributes account for nearly 80\% of the total variation, showing that bias is concentrated in a small set of visual cues. Sensitivity is strongest in judgments that are semantically aligned with appearance, especially socioeconomic and style-related judgments. We release StylisticBias as a benchmark for fine-grained bias evaluation in multimodal models. Code and dataset: https://github.com/timo-cavelius/StylisticBias and https://hf.co/datasets/shaghayegh/stylistic-bias-dataset.
Chinese Translation
多模态大语言模型（MLLMs）在个人和社会影响深远的场景中越来越多地被应用，但塑造这些模型对人们判断的视觉线索仍然不够明确。以往的研究通常比较不同（群体的）个体，使得分离外观效应与身份差异变得困难。我们引入了StylisticBias，这是一个用于评估MLLMs中属性级社会偏见的受控基准。我们生成了500个逼真的基础面孔，并为每个面孔创建约50个单属性变体，产生约25K张图像。该设计保持身份不变，同时一次改变一个视觉属性。这使我们能够测量特定线索如何影响模型判断。我们在25个二元社会判断场景中评估了六个MLLMs。我们发现年龄和身体类型主导了身份层面的效应，而时尚风格和其他视觉线索则驱动了最大的属性层面变化。我们进一步发现，约15个属性占据了近80 ext{%}的总变异，显示出偏见集中在一小组视觉线索中。在与外观语义上对齐的判断中，敏感性最强，尤其是在社会经济和风格相关的判断中。我们发布了StylisticBias作为多模态模型中细粒度偏见评估的基准。代码和数据集： https://github.com/timo-cavelius/StylisticBias 和 https://hf.co/datasets/shaghayegh/stylistic-bias-dataset。
View on arXiv Download PDF AI Translation
arXiv Reader
Automatically generated on 2026-07-16 10:15:58
Powered by OpenAI GPT & arXiv