arXiv Daily Digest

224

Papers

FARO: Feasibility-Aware Robot Motion Optimization

FARO：考虑可行性的机器人运动优化

Ciebielski, Michal, Omar, Shafeef, Johnson, Aaron, Khadiv, Majid

Abstract

Fast planning of novel behaviors in unseen scenarios remains a fundamental challenge in robotics. The high-dimensional, hybrid, and underactuated nature of humanoid loco-manipulation continues to hinder the realization of this goal. In this paper, we address this challenge by proposing a nested kino-dynamic framework for rapid feasibility checking and dynamically consistent trajectory generation given a candidate contact sequence. By integrating this module with a feasibility-guided tree search and a Large Language Model (LLM)-based contact plan sampling strategy, we demonstrate that the proposed framework can substantially improve the search process. Furthermore, we show that the generated trajectories can be tracked using a reinforcement learning (RL)-based controller and show that the resulting trajectories are of sufficiently high quality for execution in real-world loco-manipulation scenarios. A supplementary video is available at: https://youtu.be/R6qCHoCormQ.

Chinese Translation

在未知场景中快速规划新行为仍然是机器人技术中的一项基本挑战。类人运动操控的高维、混合和欠驱动特性继续阻碍这一目标的实现。本文通过提出一个嵌套的运动动力学框架，针对给定候选接触序列进行快速可行性检查和动态一致的轨迹生成，从而解决这一挑战。通过将该模块与可行性引导的树搜索和基于大语言模型（LLM）的接触计划采样策略相结合，我们证明了所提出的框架可以显著改善搜索过程。此外，我们展示了生成的轨迹可以通过基于强化学习（RL）的控制器进行跟踪，并表明生成的轨迹在实际运动操控场景中具有足够高的执行质量。补充视频可在以下链接观看：https://youtu.be/R6qCHoCormQ。

View on arXiv Download PDF AI Translation

cs.RO / 2 / 2607.18365

Towards Torque-Driven Reinforcement Learning for Quadruped Locomotion

面向扭矩驱动的四足机器人强化学习

Dowdy, Jordan, Vaz, Jean Chagas

Abstract

Reinforcement learning (RL) for legged robots is advancing locomotion, demonstrating its ability to adapt to new and challenging terrain. Traditionally, these RL locomotion frameworks are position-based, making the policy less adaptable to terrain types and requiring state estimation techniques in the observation space, i.e., linear velocity. Moreover, these RL frameworks often use small, lightweight quadrupeds that are limited in their viability for high-complexity tasks due to hardware constraints. This work explores an RL torque control framework for heavyweight high-torque quadrupeds. The RL framework in this paper can traverse rough terrain and effectively track a desired linear velocity without requiring knowledge of the agent's current velocity. Using Nvidia's Isaac Sim and Isaac Lab, simulation results of the RL torque control policy are shown on the Unitree B1 quadruped, achieving speeds of 3.5 m/s and 1.5 rad/s. In addition, the quadruped can walk up and down stairs without the aid of an exteroceptive sensor.

Chinese Translation

针对腿部机器人，强化学习（RL）正在推动其运动能力的发展，展示了其适应新挑战性地形的能力。传统上，这些基于位置的RL运动框架使得策略对地形类型的适应性较差，并且在观察空间中需要状态估计技术，即线性速度。此外，由于硬件限制，这些RL框架通常使用小型轻量级四足机器人，导致其在高复杂度任务中的可行性受到限制。本研究探讨了一种针对重型高扭矩四足机器人的RL扭矩控制框架。本文中的RL框架能够在粗糙地形上行走，并有效跟踪期望的线性速度，而无需了解代理的当前速度。通过使用Nvidia的Isaac Sim和Isaac Lab，本文展示了在Unitree B1四足机器人上的RL扭矩控制策略的仿真结果，达到了3.5 m/s和1.5 rad/s的速度。此外，该四足机器人能够在没有外部传感器辅助的情况下上下楼梯。

View on arXiv Download PDF AI Translation

cs.RO / 3 / 2607.18488

The Open Ant: A Robot Platform for Reinforcement Learning Research

开放蚂蚁：一个用于强化学习研究的机器人平台

Lupu, Elena Sorina, Spieler, Patrick, Javed, Khurram, De Asis, Kris, Martin, John D., Steenstrup, Martha, Modayil, Joseph

Abstract

Reinforcement learning (RL) research has demonstrated success in both physical and simulated domains; however, the predominant methodology remains rooted in simulations. The predominance of simulations makes translating research to physical reality uncertain for both algorithms and researchers. We propose a physical platform that is designed to simplify the transition. In this paper, we present the Open Ant: a physical variant of the commonly used Gymnasium Ant environment, along with a simulation. We demonstrate that competent walking policies can be learned from scratch in approximately one hour directly from the physical robot's experience for two substantially different RL algorithms: SARSA($\lambda$) and Soft Actor-Critic (SAC). Separately, we show policies that were learned in simulation transfer to reality. We also examine how well the platform supports a nimble experimental ecosystem. Specifically, we observe the speed with which new users from diverse backgrounds achieve their first success with the platform, and how easily the platform can be repaired and updated when hardware issues arise. Both the hardware design and software are available as open-source on GitHub for ease of customization. In summary, we advocate for the use of the Open Ant for RL researchers who frequently use simulated environments, so they can more easily include robot experiments in their evaluations.

Chinese Translation

强化学习（RL）研究在物理和模拟领域均取得了成功；然而，主流方法仍然根植于模拟环境。这种模拟的主导地位使得将研究成果转化为物理现实对于算法和研究人员来说都充满不确定性。我们提出了一个物理平台，旨在简化这一过渡。在本文中，我们介绍了开放蚂蚁（Open Ant）：一个常用的Gymnasium Ant环境的物理变体，以及一个模拟环境。我们展示了两种截然不同的强化学习算法：SARSA($ ext{λ}$)和软演员-评论家（Soft Actor-Critic，SAC）可以通过物理机器人经验在大约一小时内从零开始学习到有效的行走策略。此外，我们还展示了在模拟中学习到的策略如何转移到现实中。我们还考察了该平台对灵活实验生态系统的支持程度。具体而言，我们观察了来自不同背景的新用户在该平台上取得首次成功的速度，以及在出现硬件问题时该平台的修复和更新的便利性。硬件设计和软件均已在GitHub上开源，便于定制。总之，我们倡导强化学习研究人员使用开放蚂蚁，以便他们能够更轻松地将机器人实验纳入评估中。

View on arXiv Download PDF AI Translation

cs.RO / 4 / 2607.18517

Beyond Fixed Goal Delivery: Online POMDP Planning for Target Interception in Crowds

超越固定目标交付：在人群中进行目标拦截的在线部分可观测马尔可夫决策过程规划

Gupta, Himanshu, Aladum, Kelvin, Ahmed, Nisar, Hayes, Bradley, Sunberg, Zachary

Abstract

Target interception in crowded environments requires reaching a moving objective while navigating among multiple uncertain human agents. Since human navigation intent is not directly observable, the robot must reason over multiple possible future interaction outcomes. We formulate interception in crowds as a partially observable Markov decision process and solve it online using tree search under a fixed computational budget. In this setting, the action-space structure directly shapes the search tree and how computational effort is allocated. We perform a controlled comparison between a sequential path-speed planner, which first plans a spatial path and then modulates speed along it, and a unified planner that jointly branches over steering and speed within tree search. Across simulations with up to 200 humans, both approaches perform similarly at low crowd density but diverge sharply as density increases. At the highest crowd density, the sequential planner has a safe-interception rate 31 percentage points lower and requires 44% more time than the unified steering-speed planner, revealing a structural limitation of spatial restriction. Project webpage: https://tic-planning.github.io/

Chinese Translation

在拥挤环境中进行目标拦截需要在多个不确定的人类代理之间导航，同时到达一个移动目标。由于人类的导航意图并不可直接观察，机器人必须对多种可能的未来交互结果进行推理。我们将人群中的拦截问题表述为一个部分可观测的马尔可夫决策过程，并在固定计算预算下通过树搜索在线求解。在这种情况下，动作空间的结构直接影响搜索树的形状以及计算努力的分配。我们对比了两种方法：一种是顺序路径-速度规划器，先规划空间路径再沿路径调节速度；另一种是统一规划器，在树搜索中同时对转向和速度进行分支。在对多达200人的模拟中，尽管在低人群密度下两种方法表现相似，但随着密度的增加，它们的表现出现显著差异。在最高人群密度下，顺序规划器的安全拦截率比统一转向-速度规划器低31个百分点，并且所需时间多出44%，揭示了空间限制的结构性局限性。项目网页：https://tic-planning.github.io/

View on arXiv Download PDF AI Translation

cs.RO / 5 / 2607.18527

DASH Robot: Minimalistic Design and Optimal Aerial-Terrestrial Locomotion via Contact-Implicit Control

DASH机器人：通过接触隐式控制实现的简约设计与最佳空地运动

Paiva, Ryan Gomes, Ho, Conrad, Kang, Jiarong, Ren, Kunzhao, Xu, Xiangru, Xiong, Xiaobin

Abstract

We present a novel and minimalistic design of an aerial-terrestrial robot DASH: Ducted Aerial Spring Hopper. The goal is to enable both aerial and ground locomotion capabilities on a unified mobile robot that is mechanically-minimalistic, locomotion-versatile, and energy-efficient. We propose an organic integration of ducted fan co-axial body with a springy leg at the bottom for realization. The ducted fan module provides thrust-vectoring as the main actuation for agile flying; when it is combined with the light-weight spring leg, the robot realizes highly efficient ground hopping with energy circulation. Moreover, to realize optimal locomotion with two modes, we employ a contact-implicit model predictive controller to automatically choose locomotion modes and actuation. We successfully validated the design and control of DASH through a range of tasks, including periodic hopping, aerial flight, and mode-free locomotion with autonomous mode transitions during obstacle traversal.

Chinese Translation

我们提出了一种新颖且简约的空地机器人DASH：导管式空中弹簧跳跃器。其目标是使这一机械结构简约、运动多样且能效高的移动机器人具备空中和地面运动能力。我们提出了导管风扇共轴体与底部弹性腿的有机结合以实现这一目标。导管风扇模块作为灵活飞行的主要驱动提供推力矢量；当与轻量化弹性腿结合时，机器人实现了高效的地面跳跃并实现能量循环。此外，为了实现两种模式下的最佳运动，我们采用了接触隐式模型预测控制器，自动选择运动模式和驱动方式。我们通过一系列任务成功验证了DASH的设计和控制，包括周期性跳跃、空中飞行以及在障碍物穿越过程中具有自主模式转换的无模式运动。

View on arXiv Download PDF AI Translation

cs.RO / 6 / 2607.18578

Two-Stage Extrinsic Calibration of a Static Line-Scanning Lidar with a Rotary Platform

静态线扫描激光雷达与旋转平台的两阶段外部标定

Shree, Vikram, Danakian, Hike, Nguyen, Long, Gokidi, Rajanish, Nercessian, Patrick

Abstract

A line-scanning lidar yields range and azimuth values in a fixed plane. To perceive surrounding objects in 3D, there must be relative motion between the lidar plane and the object. Thus, using a rotating base-platform is promising for industrial applications where objects need to be scanned or inspected precisely, and is the main focus of this work. In the rotary platform setup, a 3D point cloud of an object can be constructed if the axis of rotation and the precise motion about that axis are known. However, this setup gives rise to the following problem: how can the axis of rotation of the platform be accurately identified with respect to the lidar coordinate system? It is referred to as the calibration problem in the robotics community. Any inaccuracy in this transformation directly affects the quality of the reconstructed point cloud, leading to misrepresentation of the object of interest. In this work, we explore automated approaches to statically and dynamically estimate the transformation of a rotary platform's axis of rotation with respect to a static line-scanning lidar. The proposed algorithms have been validated on real-world datasets obtained from a custom made rotary platform and an FMCW lidar, and their convergence characteristics are studied for various initial conditions.

Chinese Translation

线扫描激光雷达在一个固定平面内提供距离和方位值。为了在三维空间中感知周围物体，激光雷达平面与物体之间必须存在相对运动。因此，使用旋转基座平台在工业应用中是一个有前景的选择，尤其是在需要精确扫描或检查物体的情况下，这也是本研究的主要关注点。在旋转平台的设置中，如果已知旋转轴及其关于该轴的精确运动，则可以构建物体的三维点云。然而，这种设置引发了以下问题：如何准确识别平台的旋转轴相对于激光雷达坐标系的位置？这在机器人领域被称为标定问题。任何在此转换中的不准确性都会直接影响重建点云的质量，导致对感兴趣物体的错误表示。在本研究中，我们探索了自动化方法，以静态和动态的方式估计旋转平台的旋转轴相对于静态线扫描激光雷达的变换。所提出的算法已在从定制旋转平台和FMCW激光雷达获得的真实世界数据集上进行了验证，并研究了其在不同初始条件下的收敛特性。

View on arXiv Download PDF AI Translation

cs.RO / 7 / 2607.18580

STeP: Signal Temporal Logic for Precise Specifications for Action Generation with Vision Language Models

STeP：用于视觉语言模型的动作生成精确规范的信号时序逻辑

Torshizi, Kasra, Singh, Anukriti, Mathur, Sidharth, Habib, Khuzema, Du, Leo, Tokekar, Pratap

Abstract

Vision-language-action (VLA) models have shown impressive generalization, but often lack interpretability and can struggle to follow precise natural language instructions that encode spatial, temporal, and logical requirements. We propose a hierarchical framework that uses Signal Temporal Logic (STL) as a shared representation connecting high-level language understanding with low-level robot execution. A high-level policy leverages a VLM to decompose language instructions into high-level subtasks, generate STL specifications for each subtask, and choose a low-level policy for executing each subtask. The STL specifications translate language-derived intent into precise constraints, and the low-level policy selection determines whether those constraints are enforced directly through STL-guided model-predictive control or monitored during execution of a learned policy for perceptually complex, or contact-rich behaviors. By integrating STL into plan validation, low-level policy, subtask monitoring, and replanning, our framework enables language-derived plans to be checked, optimized, and revised at runtime using a common formal structure. We evaluate the approach on a real-world tabletop domain, demonstrating how formal specifications can improve the precision, reliability, and interpretability of language-conditioned robot planning.

Chinese Translation

视觉-语言-动作（VLA）模型展示了令人印象深刻的泛化能力，但往往缺乏可解释性，并且在遵循编码空间、时间和逻辑要求的精确自然语言指令时可能会遇到困难。我们提出了一个层次化框架，使用信号时序逻辑（STL）作为共享表示，将高层次的语言理解与低层次的机器人执行连接起来。高层策略利用视觉语言模型（VLM）将语言指令分解为高层次的子任务，为每个子任务生成STL规范，并选择低层次策略来执行每个子任务。STL规范将语言衍生的意图转化为精确约束，而低层次策略选择决定了这些约束是通过STL引导的模型预测控制直接执行，还是在执行学习策略时监控，以应对感知复杂或接触丰富的行为。通过将STL集成到计划验证、低层策略、子任务监控和重新规划中，我们的框架使得语言衍生的计划能够在运行时使用统一的形式结构进行检查、优化和修订。我们在一个真实的桌面领域评估了该方法，展示了形式规范如何提高语言条件下机器人规划的精确性、可靠性和可解释性。

View on arXiv Download PDF AI Translation

cs.RO / 8 / 2607.18586

Bridging the Sim-to-Real Gap under Real-Time Constraints in Autonomous Racing

在实时约束下弥合自主赛车中的仿真与现实差距

Maghsoumi, Hossein, Fallah, Yaser P.

Abstract

Autonomous racing exposes the sim-to-real gap under extreme operating conditions characterized by high speed, tight stability margins, and stringent real-time constraints. Although simulation is indispensable for development, controllers that perform well in simulation often degrade abruptly on physical platforms due to interacting effects of dynamics mismatch, estimation delay, and execution-layer latency. This paper frames sim-to-real transfer in autonomous racing as a full-stack, real-time systems problem. We introduce a structured three-layer perspective (Physical/Cyber/Execution) to analyze how mismatches propagate and amplify through closed-loop feedback. We present diagnostic metrics beyond nominal lap time, including performance flip, stability-oriented measures, sensitivity to delay and noise, and latency distribution characterization. Mitigation strategies are synthesized from a deployment-oriented viewpoint, emphasizing execution-aware and delay-aware design. Finally, we outline benchmarking guidelines that enable reproducible and fair sim-to-real evaluation under compute and timing constraints. The resulting framework clarifies cross-layer failure mechanisms and provides practical design principles for deployable autonomous racing systems operating near dynamic limits.

Chinese Translation

自主赛车在极端操作条件下暴露了仿真与现实之间的差距，这些条件的特点是高速、紧凑的稳定边际和严格的实时约束。尽管仿真在开发中不可或缺，但在物理平台上表现良好的控制器往往由于动态不匹配、估计延迟和执行层延迟的相互作用而急剧下降。本文将自主赛车中的仿真到现实转移框架化为一个全栈实时系统问题。我们引入了一个结构化的三层视角（物理/网络/执行），以分析不匹配如何通过闭环反馈传播和放大。我们提出了超越名义圈速的诊断指标，包括性能翻转、稳定性导向的度量、对延迟和噪声的敏感性以及延迟分布特征。缓解策略从部署导向的角度进行综合，强调对执行和延迟的关注设计。最后，我们概述了在计算和时间约束下实现可重复和公平的仿真到现实评估的基准指南。所提出的框架阐明了跨层失败机制，并为在动态极限附近运行的可部署自主赛车系统提供了实用的设计原则。

View on arXiv Download PDF AI Translation

cs.RO / 9 / 2607.18604

Intelligent Multi-UAV Navigation in ITNTNs: A Hierarchical LLM Approach

智能多无人机导航在ITNTNs中的应用：一种分层LLM方法

Yan, Zijiang, Zhou, Hao, Jaafar, Wael, Pei, Jianhua, Wang, Ping, Yanikomeroglu, Halim, Tabassum, Hina

Abstract

The deployment of high-speed Uncrewed Aerial Vehicles (UAVs) in 3D aerial highways necessitates robust coordination of physical flight kinematics and multi-tier network handovers. While Deep Reinforcement Learning (DRL) offers rapid tactical control, it lacks the zero-shot strategic reasoning required to quickly adapt to dynamic Integrated Terrestrial and Non-Terrestrial Networks (ITNTNs). Conversely, Large Language Models (LLMs) excel at semantic reasoning but suffer from high inference latency, rendering them unsuitable for real-time aerodynamic control. To bridge this gap, we propose a novel Hierarchical LLM-driven control framework. A massive cloud-based LLM deployed on a High-Altitude Platform Station (HAPS) manages slow-timescale global load balancing, while lightweight edge-LLMs on individual UAVs translate local observations into tactical sub-goals. These sub-goals guide a fast-timescale physical DRL controller to execute collision-free, handover-aware trajectories. Simulation results demonstrate that our agentic architecture significantly reduces collision rates and improves aggregate system throughput compared to existing baselines.

Chinese Translation

在三维空中高速公路中部署高速无人机（UAV）需要对物理飞行运动学和多层网络切换进行强有力的协调。尽管深度强化学习（DRL）提供了快速的战术控制，但它缺乏快速适应动态集成地面和非地面网络（ITNTNs）所需的零样本战略推理。相反，大型语言模型（LLMs）在语义推理方面表现出色，但推理延迟较高，使其不适合实时气动控制。为了弥补这一差距，我们提出了一种新颖的分层LLM驱动控制框架。部署在高空平台站（HAPS）上的大型云端LLM负责慢时间尺度的全球负载平衡，而部署在各个无人机上的轻量级边缘LLM则将本地观察转化为战术子目标。这些子目标指导快速时间尺度的物理DRL控制器执行无碰撞、考虑切换的轨迹。仿真结果表明，我们的智能架构显著降低了碰撞率，并提高了系统的整体吞吐量，相较于现有基线有显著改善。

View on arXiv Download PDF AI Translation

cs.RO / 10 / 2607.18606

On the Limits of Sampling-Based Reachability: Geometry, Dynamics, and Sample Complexity

基于采样的可达性分析的局限性：几何、动态与样本复杂度

Liu, Jixian, Tabbara, Ihab, Sibai, Hussein, Mallada, Enrique

Abstract

Reachability analysis is central to safety-critical control, robotics, and neural network verification, but classical computational methods, such as Hamilton--Jacobi reachability and set propagation, scale poorly with state dimension. Sampling-based methods have emerged as a promising alternative, often providing finite-sample guarantees that bound the probability-mass left uncovered. However, an explicit account of how the geometry of the initial set, the dynamics, and the sampling law affect the accuracy of the estimator is not fully available in the literature. We study this by casting sampling-based reachable-set recovery as geometric support estimation over a family of problems specified by an initial set, its dynamics, and a sampling law. First, we identify two regularity properties, positive reach of the initial set's complement and Lipschitz continuity of the dynamics, that together make recovery well-posed: a probability-mass coverage guarantee can be upgraded to accuracy $r$ in Hausdorff distance. Second, we bound the resulting sample complexity: recovery is achievable with $\tilde{\mathcal{O}}\big((e^{3LT}/r)^n\big)$ samples, exponential in both the state dimension and the time horizon. Third, we show that neither can be removed: an minimax lower bound of $\Omega\big((e^{LT}/r)^n\big)$ holds for every estimator, so the exponential dependence on dimension and the degradation over the horizon are both intrinsic, not artifacts of a particular method. Experiments on nonlinear systems confirm that adversarial sampling improves constants but not the scaling.

Chinese Translation

可达性分析在安全关键控制、机器人技术和神经网络验证中至关重要，但经典计算方法，如哈密尔顿-雅可比可达性和集合传播，在状态维度上扩展性较差。基于采样的方法作为一种有前景的替代方案应运而生，通常提供有限样本保证，界定未覆盖的概率质量。然而，文献中尚未充分阐明初始集合的几何特性、动态特性和采样法则如何影响估计器的准确性。我们通过将基于采样的可达集合恢复视为在由初始集合、其动态和采样法则指定的一系列问题上进行几何支持估计来研究这一问题。首先，我们确定了两个规则性属性：初始集合补集的正可达性和动态的利普希茨连续性，这两者共同使得恢复问题良好定义：概率质量覆盖保证可以提升到哈斯多夫距离的准确性$r$。其次，我们界定了结果样本复杂度的上界：恢复可以通过$ ilde{ ext{O}}ig((e^{3LT}/r)^nig)$样本实现，这在状态维度和时间范围上都是指数级的。第三，我们展示了这两者都无法去除：对于每个估计器，存在一个最小最大下界$ ext{Ω}ig((e^{LT}/r)^nig)$，因此维度的指数依赖性和时间范围的退化都是内在的，而不是特定方法的伪影。对非线性系统的实验确认，对抗性采样改善了常数，但没有改善缩放性。

View on arXiv Download PDF AI Translation

cs.RO / 11 / 2607.18637

End-to-end Conditional Diffusion for Realistic and Controllable Visual Traffic Scenario Generation

端到端条件扩散用于真实且可控的视觉交通场景生成

Li, Jingzheng, Ge, Yufei, Chen, Zhijun, Mao, Qianren, Wang, Zizhe, Qi, Binhang, Li, Bing, Chen, Keyu, Zhang, Baochang, Liu, Xianglong, Yu, Philip S

Abstract

Generating closed-loop traffic scenarios that are both realistic and controllable is crucial for evaluating autonomous driving systems, especially under rare safety-critical interactions. Existing learning-based methods often struggle to balance controllability and realism, offering either limited fine-grained control over traffic behavior or controllable scenarios at the expense of behavioral plausibility. This paper presents E2E-CDiff, an end-to-end conditional diffusion framework for controllable and realistic scenario generation. Conditioned on front-view visual observations, E2E-CDiff jointly denoises future motion states and executable low-level controls for route-interacting background vehicles. This unified state-action generation mitigates the planning-control mismatch in conventional two-stage trajectory-then-controller pipelines. Differentiable guidance further regulates speed, enforces drivable-area compliance, and supports collision-avoidance or collision-seeking behaviors, enabling both naturalistic and safety-critical scenario generation. Experiments on Bench2Drive show that E2E-CDiff achieves a favorable controllability-realism trade-off compared with representative reinforcement- and imitation-learning baselines, while its collision-guided variant induces challenging interactions across multiple autonomous driving systems. E2E-CDiff also performs competitively as a learning-based ego planner, demonstrating the generality of end-to-end state-action diffusion.

Chinese Translation

生成既真实又可控的闭环交通场景对于评估自动驾驶系统至关重要，尤其是在稀有的安全关键交互情况下。现有的基于学习的方法往往难以平衡可控性和真实性，提供的交通行为控制要么有限，要么在可控场景的生成中牺牲了行为的合理性。本文提出了E2E-CDiff，一个用于可控和真实场景生成的端到端条件扩散框架。E2E-CDiff在前视视觉观测的条件下，联合去噪未来的运动状态和可执行的低级控制，以便与背景车辆进行路径交互。这种统一的状态-动作生成方法缓解了传统两阶段轨迹-控制器管道中的规划-控制不匹配问题。可微分引导进一步调节速度，强制遵守可驾驶区域，并支持避免碰撞或寻求碰撞的行为，从而实现自然和安全关键场景的生成。在Bench2Drive上的实验表明，与代表性的强化学习和模仿学习基线相比，E2E-CDiff在可控性和真实性之间达成了良好的权衡，而其碰撞引导变体则在多个自动驾驶系统中引发了具有挑战性的交互。E2E-CDiff作为一种基于学习的自我规划器也表现出竞争力，展示了端到端状态-动作扩散的通用性。

View on arXiv Download PDF AI Translation

cs.RO / 12 / 2607.18641

Fabric Pneumatic Artificial Muscles Based on the Drawstring Principle

基于抽绳原理的织物气动人工肌肉

Liu, Chendong, Yang, Dapeng, Dai, Yiming, Jiang, Li, Liu, Hong

Abstract

Pneumatic artificial muscles have wide applications in robotics and industrial fields. Conventional pneumatic artificial muscles generate extra radial deformation during axial contraction, which severely wastes available working space. Inspired by the widely adopted drawstring principle in textile products, this paper proposes a novel drawstring fabric pneumatic artificial muscle (DPAM). Unlike traditional counterparts, the proposed DPAM produces no extra radial deformation during contraction, greatly improving structural compactness. The DPAM exhibits outstanding mechanical performance: a load capacity over 800 times its self-weight, a maximum contraction ratio of 44%, and a power density up to 4.98 kW/kg, alongside excellent scalability. Two representative application scenarios, bionic robots and industrial production lines, are demonstrated to validate its practicability. The DPAM can be easily expanded within a two-dimensional plane, as verified by the fabricated DPAM matrix. This work not only presents a high-performance novel pneumatic artificial muscle but also inspires researchers to draw design inspiration from conventional textile structures to address existing challenges in soft robotics.

Chinese Translation

气动人工肌肉在机器人技术和工业领域具有广泛的应用。传统的气动人工肌肉在轴向收缩过程中会产生额外的径向变形，这严重浪费了可用的工作空间。受到纺织产品中广泛采用的抽绳原理的启发，本文提出了一种新型的抽绳织物气动人工肌肉（DPAM）。与传统的气动人工肌肉不同，所提出的DPAM在收缩过程中不会产生额外的径向变形，从而大大提高了结构的紧凑性。DPAM展现出卓越的机械性能：其承载能力超过自重的800倍，最大收缩比达到44%，功率密度高达4.98 kW/kg，并且具有优良的可扩展性。通过展示仿生机器人和工业生产线两个典型应用场景，验证了其实用性。DPAM可以在二维平面内轻松扩展，这一点通过制造的DPAM矩阵得到了验证。本研究不仅展示了一种高性能的新型气动人工肌肉，同时也激励研究人员从传统纺织结构中汲取设计灵感，以应对软机器人领域中存在的挑战。

View on arXiv Download PDF AI Translation

cs.RO / 13 / 2607.18660

MVP-Tac: A Miniaturized Dual-Modal Vision and Photoelastic Tactile Sensor for Robot-Assisted Minimally Invasive Surgery

MVP-Tac：一种微型化的双模态视觉与光弹性触觉传感器，用于机器人辅助微创手术

Prince, Md Rakibul Islam, Kim, Jaeeun, Zhou, Yuhao, Vrshek, Mason, Sama, Shivani Reddy, Khera, Adyaa, Athar, Sheeraz, Xu, Zijie, Liu, Jiabin, Lin, Shaoting, Li, Wei, She, Yu

Abstract

Robot-assisted minimally invasive surgery (RMIS) offers major benefits over open and conventional laparoscopic procedures, yet it still lacks tactile feedback for palpation while operating under strict requirements to preserve reliable vision for navigation and safety. In practice, visual feedback is indispensable, and tactile solutions that cannot coexist with vision are difficult to translate into RMIS tools. To address both needs, we introduce MVP-Tac, a compact, vision-based tactile sensor that provides co-located vision and tactile sensing. MVP-Tac uses reflective photoelastic imaging: a thin photoelastic elastomer produces stress-dependent interferograms under contact that are captured by an embedded camera through a miniaturized reflective polariscope. A semi-transparent membrane and controllable illumination enable switching between visual mode and tactile mode, enabling tactile perception without sacrificing vision. We validate MVP-Tac through force calibration in the 0 to 2 N range and demonstrate its potential for tumor palpation via video-based hardness classification on tissue phantoms, achieving 97% accuracy for exposed-tumor classification and 92% accuracy for subdermal-tumor classification. Finally, we conduct a simulated colonoscopy to validate both visual and tactile modalities in a constrained lumen, including vision-guided 3D photomapping of the luminal wall and in situ hardness classification of localized nodules. Overall, MVP-Tac provides a practical path toward restoring clinically useful palpation in RMIS while maintaining essential visual feedback. The design, fabrication, and firmware of MVP-Tac are open-sourced at https://mvp-tac.github.io/

Chinese Translation

机器人辅助微创手术（RMIS）相较于开放式和传统腹腔镜手术具有显著优势，但在严格要求保持可靠视觉以确保导航和安全的情况下，仍然缺乏触觉反馈。在实际操作中，视觉反馈是不可或缺的，而无法与视觉共存的触觉解决方案难以转化为RMIS工具。为了解决这两种需求，我们提出了MVP-Tac，这是一种紧凑型的基于视觉的触觉传感器，能够提供共存的视觉和触觉感知。MVP-Tac采用反射式光弹性成像：一种薄型光弹性弹性体在接触下产生应力依赖的干涉图样，这些图样通过嵌入式相机通过微型反射偏振镜捕获。半透明膜和可控照明使得视觉模式与触觉模式之间的切换成为可能，从而在不牺牲视觉的情况下实现触觉感知。我们通过在0到2 N范围内的力校准验证了MVP-Tac，并通过基于视频的硬度分类展示了其在肿瘤触诊中的潜力，在组织模型上实现了97%的暴露肿瘤分类准确率和92%的皮下肿瘤分类准确率。最后，我们进行了一次模拟结肠镜检查，以验证在受限腔道中的视觉和触觉模态，包括对腔道壁的视觉引导3D光绘制和对局部结节的原位硬度分类。总体而言，MVP-Tac为在RMIS中恢复临床有用的触诊提供了一条实际路径，同时保持了必要的视觉反馈。MVP-Tac的设计、制造和固件已开源，网址为 https://mvp-tac.github.io/

View on arXiv Download PDF AI Translation

cs.RO / 14 / 2607.18663

How defensive driving enhances driving safety: A driving simulator study on drivers' defensive driving behaviors

防御性驾驶如何增强驾驶安全性：关于驾驶员防御性驾驶行为的驾驶模拟器研究

Wu, Xinzheng, Chen, Junyi, Ye, Shaolingfeng, Shen, Yong

Abstract

Defensive driving is widely recognized as an advanced driving skill. However, whether and how defensive driving affects driving safety remains insufficiently investigated. This study examines the behavioral characteristics of defensive driving, its impact on driving safety, and the underlying mechanisms. First, defensive driving is defined regarding operational timing and application scenario. Then, 82 participants are recruited for driving simulator experiments, with their behavioral and eye movement data being collected. Following the experiments, participants are categorized into groups based on the frequency of defensive driving behaviors exhibited. Finally, both inter-group and inter trial comparisons are performed on the experimental data. Experimental results demonstrate that in the inter-group comparison, the high defensive driving capability group exhibits higher acceleration and deceleration magnitudes, lower average speeds, and larger average absolute yaw angles compared to the low capability group, alongside shorter fixation durations and reduced fixation frequencies. Moreover, we observe that these participants tend to initiate defensive or evasive actions earlier, resulting in lower scenario risk. Regarding the inter-trial comparison, we observe similar trends exclusively in the low capability group, whereas most metrics show no significant differences between Trial 1 and Trial 2 in the high capability group. These results reveal that drivers possessing defensive driving capabilities tend to execute more intense driving maneuvers and identify risks and take action earlier, thereby enhancing driving safety. Findings support the promotion of defensive driving and provide a basis for relevant training programs. Meanwhile, they offer insights for the training of autonomous driving algorithms with defensive driving capabilities.

Chinese Translation

防御性驾驶被广泛认为是一种高级驾驶技能。然而，防御性驾驶是否以及如何影响驾驶安全性仍然缺乏充分的研究。本研究考察了防御性驾驶的行为特征、其对驾驶安全性的影响及其潜在机制。首先，防御性驾驶在操作时机和应用场景方面进行了定义。然后，招募了82名参与者进行驾驶模拟器实验，并收集了他们的行为和眼动数据。在实验结束后，参与者根据表现出的防御性驾驶行为的频率被分为不同组别。最后，对实验数据进行了组间比较和试次间比较。实验结果表明，在组间比较中，高防御性驾驶能力组的加速和减速幅度更大，平均速度更低，平均绝对偏航角更大，同时注视持续时间更短，注视频率更低。此外，我们观察到这些参与者倾向于更早地启动防御或规避动作，从而降低场景风险。在试次间比较中，我们仅在低能力组中观察到了类似的趋势，而在高能力组中，大多数指标在试次1和试次2之间没有显著差异。这些结果揭示了具备防御性驾驶能力的驾驶员倾向于执行更强烈的驾驶操作，并更早识别风险并采取行动，从而增强了驾驶安全性。研究结果支持推广防御性驾驶，并为相关培训项目提供了基础。同时，它们为具备防御性驾驶能力的自动驾驶算法的培训提供了见解。

View on arXiv Download PDF AI Translation

cs.RO / 15 / 2607.18709

RoboInter1.5: A Holistic Intermediate Representation Suite for Embodied World Modeling and Robotic Manipulation

RoboInter1.5：面向具身世界建模和机器人操控的整体中间表示套件

Wang, Ziqin, Li, Hao, Wang, Weijun, Cai, Junhao, Zeng, Jia, Chen, Yilun, Pang, Jiangmiao, Liu, Si

Abstract

Existing robot datasets remain expensive to curate, embodiment-specific, and insufficiently annotated with the fine-grained structure required for generalizable reasoning, execution, or long-horizon environment dynamics simulation. Building on our prior work, RoboInter1.0, we present RoboInter1.5, an extended and holistic suite of intermediate representations for both robotic manipulation and embodied world modeling. RoboInter1.5 provides a unified resource of data, benchmarks, and models centered on dense manipulation-oriented intermediate representations. Specifically, RoboInter-Data contains over 230k manipulation episodes across 571 scenes with dense per-frame annotations covering more than ten types of intermediate representations, including subtasks, primitive skills, object and gripper grounding, segmentation, affordance, grasp poses, contact points, motion traces, etc. Built upon these annotations, RoboInter-VQA introduces spatial and temporal embodied VQA tasks to benchmark and improve the intermediate-representation reasoning capabilities of our RoboInter-VLM. RoboInter-VLA further studies how such representations benefit action execution through implicit, explicit, and modular plan-then-execute paradigms. To better model the physical world, we further introduce RoboInter-World, which leverages intermediate representations as structured conditioning signals for controllable prediction of future world states. Extensive evaluations demonstrate that RoboInter1.5 provides a unified spatiotemporal scaffolding for intermediate representations. Rather than treating intermediate representations merely as interpretable signals, RoboInter1.5 conceptualizes them as a bidirectional interface that both regularizes low-level action spaces and constrains the latent rollouts of open-world physical simulators.

Chinese Translation

现有的机器人数据集在整理上仍然成本高昂，特定于具身性，并且在细粒度结构的标注上不足，无法满足可推广推理、执行或长时间环境动态仿真的需求。在我们之前的工作RoboInter1.0的基础上，我们提出了RoboInter1.5，这是一个扩展的、整体的中间表示套件，旨在支持机器人操控和具身世界建模。RoboInter1.5提供了一个统一的数据、基准和模型资源，集中于密集的操控导向中间表示。具体而言，RoboInter-Data包含超过230,000个操控事件，涵盖571个场景，并提供密集的逐帧标注，涵盖十多种中间表示类型，包括子任务、原始技能、物体和夹具的基础、分割、可用性、抓取姿势、接触点、运动轨迹等。在这些标注的基础上，RoboInter-VQA引入了空间和时间的具身视觉问答（VQA）任务，以基准测试和提升我们RoboInter-VLM的中间表示推理能力。RoboInter-VLA进一步研究这些表示如何通过隐式、显式和模块化的计划-执行范式来促进动作执行。为了更好地建模物理世界，我们进一步引入了RoboInter-World，该系统利用中间表示作为结构化的条件信号，以可控的方式预测未来的世界状态。广泛的评估表明，RoboInter1.5为中间表示提供了统一的时空支架。RoboInter1.5不仅将中间表示视为可解释的信号，而是将其概念化为一个双向接口，既规范了低级动作空间，又限制了开放世界物理模拟器的潜在展开。

View on arXiv Download PDF AI Translation

cs.RO / 16 / 2607.18713

Confidence-Gated Vision-Only Heading Alignment for UAV-UGV Cooperative Systems

基于信心门控的仅视觉航向对齐用于无人机-无人地面车辆协作系统

Ahmari, Reza, Hemmati, Vahid, Kebria, Parham, Odeyomi, Olusola, Roy, Kaushik, Homaifar, Abdollah

Abstract

Vision-based heading prediction is useful for UAV--UGV cooperation, but accurate prediction alone does not guarantee that every predicted heading should be issued directly as a control command. This paper investigates the decision problem of when and how a fixed vision-based heading predictor should be trusted for command issuance. A lightweight confidence-gated framework is proposed in which execution decisions are made using two interpretable reliability proxies derived from the perception stream: bounding-box area as a visibility-related proxy and short-window variation in predicted heading as a stability-related proxy. During low-confidence intervals, the framework compares the baseline freeze-HOLD policy with a bounded-blend fallback that updates the issued command conservatively. The method is evaluated on a real UAV--UGV dataset under clean and perturbed conditions. The results show that confidence gating creates a clear trade-off among execution rate, executed-frame accuracy, issued-command accuracy, and smoothness. The results further show that sparse execution can cause severe stale-command error under the baseline freeze-HOLD policy, whereas the bounded-blend fallback substantially improves command-level behavior under the same gate decisions. These findings highlight that reliable perception-driven autonomy depends not only on prediction accuracy, but also on decision-aware command issuance during low-confidence

Chinese Translation

基于视觉的航向预测对于无人机（UAV）与无人地面车辆（UGV）的合作非常有用，但仅靠准确的预测并不能保证每一个预测的航向都应直接作为控制指令发出。本文探讨了何时以及如何信任固定的基于视觉的航向预测器以发出指令的决策问题。我们提出了一种轻量级的信心门控框架，其中执行决策是基于从感知流中派生的两个可解释的可靠性代理进行的：边界框面积作为与可见性相关的代理，以及预测航向的短时间窗口变化作为与稳定性相关的代理。在低信心区间，该框架将基线的冻结保持（freeze-HOLD）策略与一种保守更新发出指令的有界混合（bounded-blend）后备策略进行比较。该方法在干净和扰动条件下的真实无人机-无人地面车辆数据集上进行了评估。结果表明，信心门控在执行率、执行帧准确性、发出指令的准确性和流畅性之间创造了明显的权衡。结果进一步表明，在基线的冻结保持策略下，稀疏执行可能导致严重的过时指令错误，而在相同的门控决策下，有界混合后备策略显著改善了指令级行为。这些发现强调，可靠的感知驱动自主不仅依赖于预测准确性，还依赖于在低信心情况下的决策意识指令发出。

View on arXiv Download PDF AI Translation

cs.RO / 17 / 2607.18731

Correct-by-Construction Behavior Tree Synthesis from Signal Temporal Logic Specifications with Application to Robotic Missions

基于信号时序逻辑规范的构造正确行为树合成及其在机器人任务中的应用

Dong, Jiaheng, Huang, Jingyi, Han, Liang

Abstract

Behavior Trees (BTs) are widely adopted for complex task execution in robotics, providing modular, reactive control but lacking formal guarantees. However, existing correct-by-construction synthesis from Linear Temporal Logic (LTL) cannot express quantitative timing constraints. This letter synthesizes correct-by-construction BTs from Signal Temporal Logic (STL) specifications. The workspace is modeled as a timed transition system and abstracted into a zone graph, and an augmented state space tracking both logical progress and timing constraints is introduced. A hierarchical fixed-point algorithm computes winning sets for an STL fragment encompassing safety, reachability, response, recurrence, and persistence, yielding BT subtrees with a runtime constraint function. Correctness guarantees are proven and complexity bounds are derived. Simulations demonstrate specification satisfaction with strictly positive robustness, and a physical quadrotor experiment with six STL specifications validates practical deployability.

Chinese Translation

行为树（Behavior Trees, BTs）在机器人领域广泛应用于复杂任务执行，提供模块化和反应式控制，但缺乏形式化保证。然而，现有的基于线性时序逻辑（Linear Temporal Logic, LTL）的构造正确合成无法表达定量时序约束。本文从信号时序逻辑（Signal Temporal Logic, STL）规范合成构造正确的行为树。工作空间被建模为一个定时转移系统，并抽象为一个区域图，同时引入了一个增强的状态空间，以跟踪逻辑进展和时序约束。一个层次化的固定点算法计算包含安全性、可达性、响应性、重复性和持续性的STL片段的胜利集合，生成具有运行时约束函数的BT子树。证明了正确性保证并推导了复杂性界限。仿真结果表明规范满足严格的正鲁棒性，并且通过六个STL规范的物理四旋翼实验验证了其实际可部署性。

View on arXiv Download PDF AI Translation

cs.RO / 18 / 2607.18737

Motion Primitive Discovery in a Humanoid Robot via Self-Organising Maps for Phase Recognition

通过自组织映射进行类人机器人运动原语发现与阶段识别

Gregor, Radovan, Farkaš, Igor

Abstract

Understanding the computational basis of action recognition is a central challenge in social cognition as well as in human-robot interaction. Inspired by the Mirror Neuron System (MNS), we propose a two-level architecture for motor primitive discovery and online phase recognition applied to the NICO humanoid robot. At the first level, two Self-Organising Maps (SOMs) learn topographic representations of arm kinematics (A-SOM) and hand kinematics (H-SOM) from simulated trials covering seven motor actions. The maps are trained on non-redundant features identified through hierarchical correlation analysis of motion trajectories. The results show that the two SOMs encode complementary aspects of motor behaviour. At the second level, an Echo State Network (ESN) evaluates whether temporal trajectories of SOM activations, represented by consecutive best-matching units, are sufficient for online recognition of the currently executed movement phase. The results show that SOM-based trajectories preserve the dominant phase-discriminative structure of the movement, while contextual information provides only a secondary refinement. Our contribution is the integration of established SOM and ESN methods within an MNS-inspired architecture for motor primitive representation and online phase recognition. The results are compatible with the computational hypothesis that self-organised motor representations, when temporally integrated, can support accurate online recognition of ongoing movement phases.

Chinese Translation

理解动作识别的计算基础是社会认知以及人机交互中的一个核心挑战。受到镜像神经元系统（MNS）的启发，我们提出了一种两级架构，用于运动原语发现和在线阶段识别，应用于NICO类人机器人。在第一级，两个自组织映射（SOM）学习来自模拟试验的手臂运动学（A-SOM）和手部运动学（H-SOM）的拓扑表示，这些试验涵盖了七种运动动作。这些映射是在通过运动轨迹的层次相关分析识别出的非冗余特征上进行训练的。结果表明，这两个SOM编码了运动行为的互补方面。在第二级，一个回声状态网络（ESN）评估SOM激活的时间轨迹（由连续的最佳匹配单元表示）是否足以在线识别当前执行的运动阶段。结果显示，基于SOM的轨迹保留了运动的主要阶段区分结构，而上下文信息仅提供了次要的细化。我们的贡献在于将已建立的SOM和ESN方法整合到一个受MNS启发的架构中，以实现运动原语表示和在线阶段识别。结果与计算假设相符，即自组织的运动表示在时间上整合后，可以支持对正在进行的运动阶段的准确在线识别。

View on arXiv Download PDF AI Translation

cs.RO / 19 / 2607.18760

Koopman DCM: Unstable Eigenfunctions as Data-driven Representations for Legged Balancing

Koopman DCM：不稳定特征函数作为基于数据的四足平衡表示

Caron, Stéphane

Abstract

In legged locomotion, divergent components of motion (DCMs) have emerged as characteristic states for balance control. They isolate the unstable mode of the dynamics but, in existing formulations, apply only to reduced models such as the linear inverted pendulum. In this study, we show how DCMs can be more generally formulated as Koopman eigenfunctions. Whereas Koopman analysis typically targets eigenvalues near zero, which capture conserved or slowly varying quantities, our investigation leads us to deliberately search for unstable eigenpairs with large eigenvalues. The resulting Koopman DCMs are data-driven observables trained using only real-robot data. On a real biped, DCMs learned from one hour of robot data improve tracking of reference walking patterns. We further show how learned DCMs provide state-based viability constraints when combined with model predictive control.

Chinese Translation

在四足运动中，运动的发散成分（DCMs）已成为平衡控制的特征状态。它们隔离了动态的不稳定模式，但在现有的公式中，仅适用于线性倒立摆等简化模型。在本研究中，我们展示了如何将DCMs更一般地表述为Koopman特征函数。尽管Koopman分析通常针对接近零的特征值，这些特征值捕捉了守恒或缓慢变化的量，但我们的研究使我们有意寻找具有大特征值的不稳定特征对。由此产生的Koopman DCMs是基于数据的可观察量，仅使用真实机器人数据进行训练。在一台真实的双足机器人上，从一小时的机器人数据中学习到的DCMs改善了对参考行走模式的跟踪。我们进一步展示了如何将学习到的DCMs与模型预测控制相结合，提供基于状态的可行性约束。

View on arXiv Download PDF AI Translation

cs.RO / 20 / 2607.18794

Beyond Transformers: Linear Attention Policy for Open-Vocabulary Object Goal Navigation

超越变换器：开放词汇目标导航的线性注意力策略

Zhang, Jiahong, Lin, Yifan, Zhang, Yandong, Shen, Sijun, Wang, Kexin, Pan, Yuqi, Pei, Hongjuan, Wang, Wei, Li, Guoqi

Abstract

Open-Vocabulary Object Goal Navigation (OVON) requires agents to operate under partial observability, making effective internal state updates critical for navigation performance. This update is implemented by the policy network, where recent approaches adopt Transformer-based backbones with self-attention over a context window to integrate temporal information. However, our controlled experiments show that performance does not scale with context length under Transformer-based policies, questioning the suitability of self-attention for state integration in navigation. To this end, we propose Linear Attention-based Navigation (LANav), which adopts linear attention (LA) as the policy backbone to maintain a structured state update rather than self-attention over the context window. Across multiple LA variants evaluated under identical settings, LANav consistently outperforms Transformer-based baselines. Performance improves as state update mechanisms become more structured and regulated, highlighting the importance of state update design. To improve state update effectiveness, we introduce Weighted State-Expansion Linear Attention (WSLA), which expands each attention head's state into multiple sub-states and uses learnable weighted readout to aggregate expanded sub-states. Equipped with WSLA, LANav achieves 36.4% average success rate (SR) on HM3D-OVON, outperforming Transformer-based counterparts by 6.3 percentage points in macro-averaged SR, while maintaining computational efficiency. Distance-stratified results show larger gains in long-distance episodes, while HSSD transfer and fine-tuning demonstrate robustness across scene distributions. Real-world deployment on a Unitree Go2 further achieves an 82% success rate over 50 trials, supporting the practical feasibility and sim-to-real transfer of LANav.

Chinese Translation

开放词汇目标导航（OVON）要求智能体在部分可观测的环境中操作，使有效的内部状态更新对导航性能至关重要。该更新由策略网络实现，近期的方法采用基于变换器的骨干网络，通过对上下文窗口的自注意力机制来整合时间信息。然而，我们的控制实验表明，在基于变换器的策略下，性能并不随上下文长度的增加而提升，这对自注意力在导航中的状态整合适用性提出了质疑。为此，我们提出了基于线性注意力的导航（LANav），该方法采用线性注意力（LA）作为策略骨干，以保持结构化的状态更新，而不是对上下文窗口进行自注意力处理。在相同设置下评估的多个LA变体中，LANav始终优于基于变换器的基线。随着状态更新机制变得更加结构化和规范，性能得到了提升，突显了状态更新设计的重要性。为了提高状态更新的有效性，我们引入了加权状态扩展线性注意力（WSLA），该方法将每个注意力头的状态扩展为多个子状态，并使用可学习的加权读取来聚合扩展的子状态。配备WSLA后，LANav在HM3D-OVON上实现了36.4%的平均成功率（SR），在宏观平均SR上比基于变换器的对手高出6.3个百分点，同时保持了计算效率。距离分层结果显示在长距离情境中获得了更大的收益，而HSSD迁移和微调则展示了在场景分布中的鲁棒性。在Unitree Go2上的实际部署进一步实现了50次试验中的82%成功率，支持了LANav的实际可行性和从仿真到现实的迁移。

View on arXiv Download PDF AI Translation

cs.RO / 21 / 2607.18840

WorldScape Policy 2.0: Empowering Steerable World Action Modeling with Reasoning-Augmented Memory

WorldScape Policy 2.0：增强推理记忆的可控世界行动建模

Su, Haisheng, Liu, Zongdai, Jin, Xin, Dou, Haoxuan, Hu, Chengming, Li, Baorun, Liu, Zhanwang, Xu, Ruiyan, Fang, Jianjie, Zhang, Xin, Yang, Zhenjie, Yang, Xue, Gao, Chen, Yan, Junchi, Li, Yong, Wu, Wei

Abstract

World Action Models (WAMs) offer a promising paradigm for robotic manipulation by jointly modeling visual state transitions and robot actions. However, existing WAMs are constrained by limited temporal context, coarse episode-level language supervision, and predominantly text-only conditioning, which hinder task-progress tracking and fine-grained language-video-action grounding while limiting visual-context reasoning and cross-embodiment transfer. In this paper, we introduce WorldScape Policy 2.0, a controllable WAM with reasoning-augmented long short-term memory. Its causal short-term visual memory supplies recent observations as DiT prefill to preserve local interaction dynamics, while its long short-term event memory organizes historical VLM outputs into global-history, local-active, and event-boundary representations for progress-aware retrieval. The retrieved history augments perception and autoregressively generated planning tokens, yielding an implicit subgoal condition for autonomous planning; semantic forcing further transfers event-level instruction semantics into this latent planning pathway. To establish fine-grained multimodal controllability, we construct ManipEvent-5M, an event-grounded embodied pretraining dataset containing nearly 5 million event segments with aligned action trajectories, episode-level task instructions, segment-level subtask captions, goal images, and video demonstrations. These designs provide a unified interface for autonomous planning from high-level instructions and controllable execution from fine-grained text, goal-image, or video-context prompts. Experiments in both simulation and real-world platforms demonstrate superior capabilities in long-horizon autonomous planning, fine-grained instruction following and in-context adaptation.

Chinese Translation

世界行动模型（WAMs）为机器人操作提供了一种有前景的范式，通过联合建模视觉状态转变和机器人动作。然而，现有的WAMs受到有限时间上下文、粗略的情节级语言监督和主要依赖文本的条件限制，这妨碍了任务进展跟踪和细粒度语言-视频-动作的基础，同时限制了视觉上下文推理和跨体现转移。在本文中，我们介绍了WorldScape Policy 2.0，这是一种具有增强推理的长短期记忆的可控WAM。其因果短期视觉记忆提供最近的观察作为DiT预填充，以保持局部交互动态，而其长短期事件记忆将历史VLM输出组织为全球历史、局部活动和事件边界表示，以便于进展感知检索。检索到的历史增强了感知和自回归生成的规划标记，产生了隐式子目标条件以支持自主规划；语义强制进一步将事件级指令语义转移到这一潜在规划路径中。为了建立细粒度的多模态可控性，我们构建了ManipEvent-5M，这是一个事件基础的具身预训练数据集，包含近500万个事件片段，配有对齐的动作轨迹、情节级任务指令、片段级子任务标题、目标图像和视频演示。这些设计为从高层指令进行自主规划和从细粒度文本、目标图像或视频上下文提示进行可控执行提供了统一接口。在模拟和真实世界平台上的实验表明，在长时间跨度的自主规划、细粒度指令跟随和上下文适应方面具有优越的能力。

View on arXiv Download PDF AI Translation

cs.RO / 22 / 2607.18855

Pose-Parameterized Motion Planning and CBF-QP Self-Collision Filtering for a Long-Reach Drilling Boom

基于姿态参数的运动规划与CBF-QP自碰撞过滤在长臂钻机中的应用

Shahna, Mehdi Heydari, Kivelä, Tuomo, Mattila, Jouni

Abstract

Long-reach drilling booms must reach successive poses without self-collision. Moving from operator-supervised control toward autonomy requires collision-aware motion planning and execution. For the Sandvik SB60, this study adapts established methods by integrating pose-parameterized planning with a capsule-based control barrier function quadratic program (CBF-QP) in measured-state inverse kinematics (IK). A fixed task-specific parameter set within each task generates waypoints, detours, timed references, and chained motion without target-specific retuning. The offline detour planner screens candidate waypoints using 23 selected rod-segment-to-body-region distances, whereas the online CBF-QP filters joint velocities using 14 configured capsule-pair constraints from a nine-primitive whole-body capsule model. Evaluation considers two drilling tasks in a manufacturer-developed SB60 Simscape Multibody model: a five-target restricted-orientation tour and a three-target full-pose tour. Across several hundred thousand samples, the method produced zero IK failures, generated several detour waypoints, achieved millimetre-level mean final-position error, and recorded no sampled CBF margins below the reported thresholds.

Chinese Translation

长臂钻机必须在不发生自碰撞的情况下达到连续的姿态。从操作员监督控制向自主控制的转变需要考虑碰撞的运动规划与执行。针对Sandvik SB60，本研究通过将基于姿态参数的规划与基于胶囊的控制障碍函数二次规划（CBF-QP）相结合，适应了已建立的方法，应用于测量状态的逆向运动学（IK）。在每个任务中，固定的任务特定参数集生成路径点、绕行、时间参考和链式运动，而无需针对特定目标进行重新调优。离线绕行规划器使用23个选定的杆段到机体区域的距离筛选候选路径点，而在线CBF-QP则利用来自九个原始整体胶囊模型的14个配置胶囊对约束过滤关节速度。评估考虑了在制造商开发的SB60 Simscape多体模型中的两个钻井任务：一个五目标限制方向的巡回任务和一个三目标全姿态巡回任务。在数十万个样本中，该方法实现了零个IK失败，生成了多个绕行路径点，达到了毫米级的最终位置均值误差，并且记录的CBF边际没有低于报告的阈值。

View on arXiv Download PDF AI Translation

cs.RO / 23 / 2607.18936

NGPS: GPS-Denied Aerial Geo-Localization and 2.5D Reconstruction via Deep Satellite Image Matching and Multi-Rate Sensor Fusion

NGPS：基于深度卫星图像匹配和多速率传感器融合的无GPS空中地理定位与2.5D重建

Sharma, Sanket

Abstract

We present NGPS (Next-Generation Positioning System), a visual geo-localization framework for high-altitude UAVs that provides GPS-free absolute positioning by matching down-facing images to georeferenced satellite imagery with deep features. The system combines (1) adaptive confidence-weighted UKF fusion, where NGPS covariance is modulated by RANSAC inlier ratio, reprojection error, and match confidence; (2) velocity-predictive kernel extraction, using VIO velocity to predict the satellite search region; and (3) an asynchronous multi-rate temporal priority queue that interleaves absolute position (1-2 Hz), VIO (10-20 Hz), and IMU (100-200 Hz) updates in chronological order. Globally optimized poses from VINS pose-graph optimization, anchored by NGPS corrections, further enable real-time 2.5D georeferenced orthomosaic reconstruction. On five flight sequences (60-150 m AGL), NGPS achieves 2.94 m position RMSE, with worst-case ATE 6.04 m at 150 m AGL and 2 m/s, yielding a 3.5x improvement over standalone monocular VIO. The system runs in real time on an NVIDIA Jetson Orin NX. Part of the implementation is open-sourced at https://github.com/snktshrma/ngps_flight.

Chinese Translation

我们提出了NGPS（下一代定位系统），这是一个用于高空无人机的视觉地理定位框架，通过将向下拍摄的图像与带有深度特征的地理参考卫星图像进行匹配，实现无GPS的绝对定位。该系统结合了（1）自适应置信加权的UKF融合，其中NGPS的协方差由RANSAC内点比率、重投影误差和匹配置信度调节；（2）基于速度预测的核提取，利用VIO速度预测卫星搜索区域；以及（3）一个异步多速率时间优先队列，按时间顺序交错更新绝对位置（1-2 Hz）、VIO（10-20 Hz）和IMU（100-200 Hz）。通过NGPS校正锚定的VINS位姿图优化获得的全局优化位姿，进一步实现实时的2.5D地理参考正射拼接重建。在五个飞行序列（离地高度60-150米）中，NGPS实现了2.94米的位置均方根误差（RMSE），在150米AGL和2米/秒的情况下，最坏情况下的绝对轨迹误差（ATE）为6.04米，较单独的单目VIO提高了3.5倍。该系统在NVIDIA Jetson Orin NX上实时运行。部分实现已开源，地址为https://github.com/snktshrma/ngps_flight。

View on arXiv Download PDF AI Translation

cs.RO / 24 / 2607.18940

The Twist Decomposition of Serial Robots Under Lower-Mobility Tasks

串联机器人在低移动性任务下的扭转分解

Baron, Luc, Chablat, Damien

Abstract

This paper introduces a twist decomposition framework for serial manipulators performing lower mobility tasks. Rather than relying on Jacobian null-space projections, the method separates the end-effector twist into task and redundant components using geometrically defined twist projectors. This formulation provides a direct and intuitive distinction between task-relevant and task-irrelevant motions in operational space, enabling a compact inverse kinematics scheme that naturally handles both manipulator and task redundancy.

Chinese Translation

本文提出了一种用于执行低移动性任务的串联机械臂的扭转分解框架。该方法不依赖于雅可比零空间投影，而是利用几何定义的扭转投影器将末端执行器的扭转分解为任务相关和冗余分量。这一公式提供了操作空间中任务相关运动与任务无关运动之间的直接且直观的区分，从而使得紧凑的逆向运动学方案能够自然地处理机械臂和任务的冗余性。

View on arXiv Download PDF AI Translation

cs.RO / 25 / 2607.18950

Design and stability analysis of an underactuated hand with passively rotating fingers

具有被动旋转手指的欠驱动手的设计与稳定性分析

Plancoulaine, Léonie, Guégan, Sylvain, Plestan, Franck, Chablat, Damien

Abstract

This paper presents an innovative design and stability analysis of an underactuated robotic finger with spatial mobility, designed to enhance gripping dexterity in robotic hands. The finger architecture incorporates a revolute joint at its base, enabling passive spatial rotation that facilitates both cylindrical and spherical grasping. With only two phalanges per finger, the design simplifies kinematic complexity while supporting precision and enveloping grasps. Stability criteria, based on the moment at the finger base joint induced by contact forces, are introduced to ensure reliable object gripping and prevent ejection during manipulation. The study also examines a differential mechanism that distributes a single actuation torque across multiple fingers, allowing adaptive and coordinated motion. This mechanism enhances the hand's ability to grasp diverse object shapes with minimal pre-grasp adjustments, leveraging passivity for autonomous adaptation. Theoretical findings are experimentally validated using a fully mechanical prototype, demonstrating versatility in performing cylindrical, spherical, parallel, and enveloping grasps. The integration of underactuation-both within individual fingers and among multiple fingers-reduces mechanical complexity, cost, and control demands while preserving functional adaptability. This work advances the development of compliant robotic hands suitable for applications requiring dexterity and robustness, such as agricultural robotics, logistics, assistive technologies, and waste sorting. Future research will focus on automating actuation and refining control strategies to further improve grasp stability and precision, paving the way for autonomous manipulation in unstructured environments.

Chinese Translation

本文提出了一种创新的欠驱动机器人手指设计及其稳定性分析，该手指具有空间运动能力，旨在增强机器人手的抓握灵活性。手指结构在其基部设有一个旋转关节，使其能够被动进行空间旋转，从而便于实现圆柱形和球形抓握。每个手指仅有两个指骨，这一设计简化了运动学复杂性，同时支持精确和包围式抓握。基于接触力在手指基部关节上产生的力矩，提出了稳定性标准，以确保可靠的物体抓握并防止在操作过程中物体脱落。研究还考察了一种差动机制，该机制将单一的驱动扭矩分配到多个手指上，允许自适应和协调运动。该机制增强了手对各种物体形状的抓握能力，且在最小的预抓握调整下，利用被动性实现自主适应。理论结果通过一个完全机械的原型进行了实验验证，展示了在执行圆柱形、球形、平行和包围式抓握方面的多样性。欠驱动的整合——既体现在单个手指内，也体现在多个手指之间——降低了机械复杂性、成本和控制需求，同时保持了功能适应性。本研究推动了适用于需要灵活性和鲁棒性的应用（如农业机器人、物流、辅助技术和废物分类）的柔性机器人手的发展。未来的研究将集中于自动化驱动和优化控制策略，以进一步提高抓握的稳定性和精确性，为在非结构化环境中的自主操作铺平道路。

View on arXiv Download PDF AI Translation

cs.RO / 26 / 2607.19174

Bayesian Retraction Optimization for Tissue Attachment Mapping in Surgical Dissection

用于外科解剖中组织附着映射的贝叶斯回撤优化

Ho, Shing-Hei, Thach, Bao, Vo, Toan, Ferguson, James M., Kuntz, Alan

Abstract

With growing surgeon shortages, automating surgical sub-tasks such as tissue dissection offers a promising step toward reducing workload and expanding patient access. Prior work has relied on hand-crafted incision policies that cannot quantify uncertainty or has relied on simulation-based methods that require strong modeling assumptions. We instead view tissue attachment identification as an inherently probabilistic problem and propose a Bayesian approach that avoids explicit tissue modeling. Our method uses a Sequential Bayesian Hilbert Map (SBHM) to represent the likelihood that each tissue point is attached to the underlying resection surface. An ensemble of learned classifiers predicts attachment likelihoods from spatial data acquired during robotic tissue retraction, with each classifier serving as a noisy information source to update the SBHM. To plan the next retraction, we devise Bayesian Retraction Optimization (BRO) to select the most informative action under safety constraints. As the SBHM refines over time, regions with high attachment likelihood are selectively incised. We validate our method in simulation across diverse tissue geometries and acquisition strategies, and demonstrate zero-shot transfer to real robotic dissection experiments.

Chinese Translation

随着外科医生短缺问题的加剧，自动化外科子任务如组织解剖提供了减少工作负担和扩大患者接入的有希望的途径。以往的研究依赖于无法量化不确定性的手工切口策略，或依赖于需要强假设建模的基于模拟的方法。我们将组织附着识别视为一个固有的概率问题，并提出一种避免显式组织建模的贝叶斯方法。我们的方法使用序列贝叶斯希尔伯特图（Sequential Bayesian Hilbert Map, SBHM）来表示每个组织点附着于基础切除表面的可能性。一个学习分类器的集成从在机器人组织回撤过程中获取的空间数据中预测附着可能性，每个分类器作为一个噪声信息源来更新SBHM。为了规划下一步的回撤，我们设计了贝叶斯回撤优化（Bayesian Retraction Optimization, BRO），以在安全约束下选择最具信息量的行动。随着SBHM随时间的精细化，附着可能性高的区域被选择性切割。我们在多样的组织几何和获取策略下对我们的方法进行了模拟验证，并展示了对真实机器人解剖实验的零样本迁移。

View on arXiv Download PDF AI Translation

cs.RO / 27 / 2607.19190

Agentic Real2Sim: Physics-based World Modeling with Vision-Language Agents

Agentic Real2Sim：基于物理的世界建模与视觉-语言代理

Chen, Guanxiong, Xia, Qianjun, Peng, Jiawei, Zhang, Heng, Ma, Bole, Qian, Justin, Jiao, Ziyi, Zhou, Bingyang, Ye, Luoxin, Zhang, Kaifeng, Wang, Kunyi, Zeng, Weijia, Chen, Yunuo, Yang, Pengzhi, Zeng, Ziqiu, Wang, Huamin, Liu, Chao, Yuille, Alan, Shi, Fan, Zheng, Changxi, Li, Yunzhu, Jiang, Chenfanfu, Chen, Peter Yichen

Abstract

Real-to-sim conversion for robotic interaction with objects remains labor-intensive because it requires more than visual reconstruction: a streamlined real2sim process must recover scene geometries and object states, infer physical parameters, and assemble actors, objects, cameras, poses, and trajectories into a runnable physical simulation. Today this process still depends on manual tuning of visual foundation models, mesh cleanup, coordinate-frame alignment, and brittle workflow glue across visual perception tools and simulators. We introduce \textit{Agentic Real2Sim}, a framework for generalized physical world modeling with vision-language agents, converting a real-world recording of object-robot interaction into a simulatable episodic twin which preserves observations, geometries, robot interactions, and object states. We evaluate Agentic Real2Sim on rigid-object manipulation, deformable-object interaction, and humanoid motion scenes, spanning domains that are usually handled by separate Real2Sim pipelines, marking a first step toward scalable conversion. The framework's agentic decisions can be driven by an open-weight VLM backend at a small fraction of the cost of frontier models, while attaining comparable conversion success rate. We aim to use the resulting real-world-aligned twins for downstream robotics tasks, specifically policy learning and evaluation. The project site is available at https://ericchen321.github.io/agentic_real2sim.github.io/.

Chinese Translation

机器人与物体的交互中的真实到仿真转换仍然是劳动密集型的，因为它不仅需要视觉重建：一个高效的真实到仿真过程必须恢复场景几何和物体状态，推断物理参数，并将参与者、物体、相机、姿态和轨迹组装成可运行的物理仿真。如今，这一过程仍然依赖于对视觉基础模型的手动调优、网格清理、坐标系对齐以及视觉感知工具与仿真器之间脆弱的工作流连接。我们提出了 extit{Agentic Real2Sim}，一个用于与视觉-语言代理进行通用物理世界建模的框架，将物体-机器人交互的真实世界记录转换为可仿真的情节双胞胎，保留观察、几何、机器人交互和物体状态。我们在刚性物体操控、可变形物体交互和类人运动场景上评估了Agentic Real2Sim，这些领域通常由不同的真实到仿真管道处理，标志着向可扩展转换迈出的第一步。该框架的代理决策可以由开放权重的视觉-语言模型（VLM）后端驱动，其成本仅为前沿模型的一小部分，同时达到可比的转换成功率。我们旨在利用生成的与真实世界对齐的双胞胎进行下游机器人任务，特别是策略学习和评估。项目网站可访问 https://ericchen321.github.io/agentic_real2sim.github.io/。

View on arXiv Download PDF AI Translation

cs.RO / 28 / 2607.19194

Cognitive Dual-Process Planning for Autonomous Driving with Structured Scene Knowledge and Verifiable Reasoning-Action Consistency

具有结构化场景知识和可验证推理-行动一致性的自主驾驶认知双过程规划

Yang, Zhongyao, Li, Haoyu, Yan, Yu, Yu, Zhuangxuan, Nan, Jiangfeng, Nan, Jinrui

Abstract

High-level planning for autonomous driving is a knowledge-intensive engineering decision task that requires accurate scene understanding, timely inference, and internally consistent action selection. Vision-language models (VLMs) can make intermediate reasoning explicit, but their use in deployed planners is constrained by costly structured supervision, unnecessary reasoning in routine scenes, and possible inconsistencies between generated rationales and driving actions. We present a cognitive dual-process planning framework that represents planning-relevant scene knowledge in a machine-parsable structured chain-of-thought (S-CoT) schema. An automated data engine integrates perception foundation models, critical-path filtering, and an expert VLM to generate S-CoT supervision without manual annotation of individual rationales. A lightweight visual Arbiter estimates scene complexity from multilevel vision-encoder features before language decoding and routes each input to either fast meta-action prediction or slow structured reasoning. For slow-path outputs, a deterministic rule-based validator checks whether the parsed S-CoT fields are consistent with the final meta-action and provides verifiable rewards for Group Relative Policy Optimization (GRPO). In a 195-scene manual audit, the generated annotations achieve 91.8\% CoT accuracy and a 98.5\% Logical Consistency Score (LCS). On 574 manually verified NAVSIM test samples, the planner achieves 80.14\% planning accuracy and 97.20\% LCS while reducing average latency by 17.39\% relative to applying slow reasoning to every scene. Evaluation on external long-tail subsets further identifies conditions under which routing and planning performance degrade. Together, these results show how explicit scene knowledge can be operationalized through adaptive reasoning and rule-based verification to support high-level VLM planning decisions.

Chinese Translation

自主驾驶的高层规划是一项知识密集型的工程决策任务，要求准确的场景理解、及时的推理和内部一致的行动选择。视觉-语言模型（VLMs）可以使中间推理变得显性，但在实际应用的规划器中，其使用受到昂贵的结构化监督、不必要的常规场景推理以及生成的推理与驾驶行动之间可能存在的不一致性的限制。我们提出了一种认知双过程规划框架，该框架以机器可解析的结构化思维链（S-CoT）模式表示与规划相关的场景知识。一个自动化数据引擎集成了感知基础模型、关键路径过滤和专家VLM，以在不手动标注单个推理的情况下生成S-CoT监督。一个轻量级视觉仲裁者在语言解码之前根据多层视觉编码器特征评估场景复杂性，并将每个输入路由到快速元行动预测或慢速结构化推理。对于慢路径输出，一个确定性基于规则的验证器检查解析的S-CoT字段是否与最终的元行动一致，并为群体相对策略优化（GRPO）提供可验证的奖励。在195个场景的手动审计中，生成的注释达到了91.8%的思维链准确率和98.5%的逻辑一致性得分（LCS）。在574个手动验证的NAVSIM测试样本中，规划器实现了80.14%的规划准确率和97.20%的LCS，同时将平均延迟相对于对每个场景应用慢速推理减少了17.39%。对外部长尾子集的评估进一步识别了路由和规划性能下降的条件。这些结果共同表明，如何通过自适应推理和基于规则的验证将显性场景知识转化为操作，以支持高层次的VLM规划决策。

View on arXiv Download PDF AI Translation

cs.RO / 29 / 2607.19213

Computing on the Fly: Navigating a Vision for the Future of Drone Computing

即时计算：展望无人机计算的未来愿景

Butler, Kevin, Stewart, Christopher, Aschenbruck, Nils, Gerall, Alina, Shi, Weisong, Silver, Deborah, Topcu, Ufuk

Abstract

The report envisions a decade in which drones move goods, medical supplies, and information at a scale comparable to national infrastructure investments like highways and the electric grid. Potential applications include natural disaster detection drones that spot wildfire sources within minutes, medical supply chains that bypass ground congestion to reach rural hospitals, and nationwide fleets that continuously inspect bridges and power lines. Realizing this future, however, requires closing what report authors call a "capability gap," where hardware and aspirations are outpacing the software and systems needed to operate safely at scale. The report identifies twelve technical challenges that must be addressed to realize the transformative potential of drone technology: Scaling to millions of drones; AI intelligence and assurance; Edge-cloud continuum and real-time coordination; AI autonomy and agentic systems; Data, training, and validation infrastructure; Critical infrastructure protection; Building reliable fleets from non-deterministic agents; Trust, security, and distributed authentication; Next-generation drone networks; Human-AI partnership and scalable insight; Standards, certification, and regulation; and Workforce development and education. These twelve challenges and proposed approaches to them form the basis of the report, laying out a multifaceted path forward for the evolution of done technology.

Chinese Translation

本报告展望了未来十年无人机在货物、医疗物资和信息运输方面的应用规模，将达到类似于国家基础设施投资（如高速公路和电网）的水平。潜在应用包括能够在数分钟内发现野火源头的自然灾害监测无人机、绕过地面交通拥堵直达农村医院的医疗供应链，以及持续巡检桥梁和电力线路的全国性无人机机队。然而，实现这一未来需要弥合报告作者所称的“能力差距”，即硬件和愿景的发展速度超过了支持大规模安全运行的软件和系统。报告指出必须解决的十二项技术挑战，以实现无人机技术的变革潜力：无人机规模化至数百万；人工智能智能性与保障；边缘-云连续体与实时协调；人工智能自治与代理系统；数据、训练与验证基础设施；关键基础设施保护；基于非确定性代理构建可靠机队；信任、安全与分布式认证；下一代无人机网络；人机协作与可扩展洞察；标准、认证与监管；以及劳动力发展与教育。这十二项挑战及其应对方案构成了报告的核心内容，为无人机技术的发展描绘了多维度的前进路径。

View on arXiv Download PDF AI Translation

cs.RO / 30 / 2607.19274

Eversion-based robots can enable safe access,steering and endoscopic imaging within the spinal subarachnoid space

基于翻转的机器人可以实现脊髓蛛网膜下腔的安全进入、引导和内窥镜成像

Wu, Zicong, Kalozoumis, Panagiotis, Sadati, S. M. Hadi, Ahmed, Aminul I., Shapey, Jonathan, Baker, Christian, Booth, Thomas, Xia, Wenfeng, Ourselin, Sebastien, Vartholomeos, Panagiotis, Bergeles, Christos

Abstract

Safe navigation within the spinal subarachnoid space is constrained by its narrow, compliant, and delicate anatomy. Conventional catheters and continuum robots rely on proximal pushing, generating friction and shear along the tissue device interface that limit distal controllability and increase the risk of neural injury. Here, we present a 2 mm diameter eversion-growing robotic platform that enables friction minimised extension and steering within the human spinal subarachnoid space, validated through computational modelling, phantom experiments, and intact human cadaver studies. The robot integrates a miniature endoscope for real time intrathecal visualisation and advances by pressure driven tip eversion, localising motion to the distal tip while minimising translational sliding of the deployed body. Phantom experiments demonstrated reductions of 65.2% in mean interaction force and 48.0% in peak interaction force compared with matched push-based insertion. Physics based modelling showed that eversion based growth redistributed tissue loading, reducing local stress concentrations and interfacial shear relative to conventional insertion. In an intact human cadaver, the system achieved 150 mm of controlled intrathecal extension with concurrent fluoroscopic and endoscopic visualisation, providing access across multiple vertebral levels from a standard lumbar entry point. Postprocedural laminectomy and durotomy revealed no observable macroscopic disruption of the dura mater or surrounding neural structures. These results provide the first mechanically characterised and multimodally validated demonstration of eversion-based robotic navigation in intact human spinal anatomy, establishing a quantitative and procedural foundation for future intrathecal interventions. Further validation in larger anatomical cohorts and under physiological conditions will be required before clinical translation.

Chinese Translation

在脊髓蛛网膜下腔内的安全导航受到其狭窄、柔韧和精细解剖结构的限制。传统的导管和连续机器人依赖于近端推送，导致在组织与设备界面之间产生摩擦和剪切力，这限制了远端的可控性并增加了神经损伤的风险。在此，我们提出了一种直径为2毫米的翻转生长机器人平台，该平台能够在人体脊髓蛛网膜下腔内实现摩擦最小化的扩展和引导，通过计算建模、假体实验和完整人类尸体研究进行了验证。该机器人集成了微型内窥镜，用于实时的鞘内可视化，并通过压力驱动的尖端翻转推进，将运动局限于远端尖端，同时最小化已部署主体的平移滑动。假体实验表明，与匹配的推入方式相比，平均交互力减少了65.2%，峰值交互力减少了48.0%。基于物理的建模显示，基于翻转的生长重新分配了组织负载，减少了局部应力集中和相对于传统插入的界面剪切力。在一具完整的人类尸体中，该系统实现了150毫米的受控鞘内扩展，同时进行荧光透视和内窥镜可视化，从标准的腰椎进入点跨越多个椎体水平。术后椎板切除和硬膜切开未发现硬膜或周围神经结构的明显宏观破坏。这些结果提供了基于翻转的机器人导航在完整人类脊柱解剖中的首次机械特征化和多模态验证示范，为未来的鞘内干预建立了定量和程序基础。在临床转化之前，需要在更大解剖队列和生理条件下进一步验证。

View on arXiv Download PDF AI Translation

cs.RO / 31 / 2607.19284

Stochastic Multi-Objective Kinodynamic Planning Against Adversaries

针对对抗者的随机多目标运动动力学规划

Vielmetti, Thomas Marshall, Cherenson, Daniel, Panagou, Dimitra

Abstract

This paper addresses multi-objective kinodynamic planning in environments with stochastic hybrid adversaries that probabilistically transition to adversarial modes based on the ego state. The goal is to construct the Pareto-front of paths that trade off execution cost and the probability of safety constraint violation (risk). Existing chance-constrained planners evaluate risk over open-loop trajectories, yielding overly conservative solutions that fail to account for ego-agent reactivity. To address this limitation, we shift the planning space to sequences of closed-loop policies, and integrate sample-based risk evaluation directly into tree construction via Monte-Carlo particle rollouts. We first introduce Stochastic Multi-Objective RRT (SMO-RRT), for which we prove probabilistic completeness, followed by Stochastic Multi-Objective Stable Sparse RRT (SMO-SST), which leverages selective pruning to improve numerical performance at the cost of completeness. For both algorithms, we derive a finite-sample bound on the probability of chance constraint violation for systems with non-Gaussian, state-dependent uncertainty, enabling probabilistically safe planning in a broad class of environments applicable to multi-agent systems, social navigation, and autonomous driving.

Chinese Translation

本文探讨了在具有随机混合对抗者的环境中进行多目标运动动力学规划，这些对抗者根据自我状态以概率方式过渡到对抗模式。目标是构建路径的帕累托前沿，这些路径在执行成本和安全约束违反概率（风险）之间进行权衡。现有的机会约束规划器在开放循环轨迹上评估风险，导致过于保守的解决方案，未能考虑自我代理的反应性。为了解决这一局限性，我们将规划空间转移到闭环策略的序列中，并通过蒙特卡洛粒子展开直接将基于样本的风险评估集成到树的构建中。我们首先介绍了随机多目标快速扩展随机树（Stochastic Multi-Objective RRT, SMO-RRT），并证明其概率完备性，随后介绍了随机多目标稳定稀疏快速扩展随机树（Stochastic Multi-Objective Stable Sparse RRT, SMO-SST），该方法利用选择性修剪在保证完备性的代价下提高数值性能。对于这两种算法，我们推导了在具有非高斯、状态依赖不确定性的系统中，机会约束违反概率的有限样本界限，从而实现了在广泛适用于多智能体系统、社会导航和自动驾驶的环境中进行概率安全规划。

View on arXiv Download PDF AI Translation

cs.RO / 32 / 2607.19306

From Distances to Trajectories: Real-Time Signed Distance Function Mapping and Distance-Accelerated Motion Planning for UAVs

从距离到轨迹：无人机的实时有符号距离函数映射与距离加速运动规划

Stanley, Jason, Dai, Zhirui, Qian, Qihao, Ho, Tzu-Chin, Fan, Tianxing, Saha, Siddharth, Barngrover, Christopher, Lee, Ki Myung Brian, Atanasov, Nikolay

Abstract

Autonomous flight in cluttered environments requires a robot to build a geometric map of its surroundings and plan safe, dynamically feasible trajectories, all onboard and in real time. Conventional approaches treat mapping and planning as separate stages and often rely on binary occupancy for collision checking. We argue that these two stages should be co-designed around a single representation: a signed distance function (SDF). By encoding distance to the nearest obstacle, an SDF provides richer information for planning and trajectory optimization than occupancy alone. We develop an Octree REsidual Network (OREN) that pairs an explicit octree prior with an implicit neural residual to reconstruct SDFs online from point cloud observations with the efficiency of volumetric methods and the accuracy and differentiability of neural methods. In tandem, we develop Bubble$^\star$, a search-based planner that exploits the distance information to grow maximal collision-free balls, which we call bubbles, with formal guarantees of termination, completeness, and failure detection. Planning over a graph of bubbles significantly reduces collision checks compared to a grid-based A$^\star$ search and returns a bubble sequence that forms a safe corridor for trajectory optimization. We demonstrate the integrated OREN-Bubble$^\star$ approach onboard a quadrotor, navigating unseen indoor environments in real time under tight compute constraints. OREN improves SDF estimation by $22$% compared to baselines, while Bubble$^\star$ finds trajectories spanning $\approx 90$ m through a cluttered environment in $1$-$3$ sec., whereas baselines take up to $10$ sec. in the same environment.

Chinese Translation

在复杂环境中进行自主飞行要求机器人构建其周围环境的几何地图，并实时规划安全且动态可行的轨迹。传统方法将映射和规划视为两个独立的阶段，通常依赖于二进制占用图进行碰撞检测。我们认为这两个阶段应围绕单一表示进行共同设计：有符号距离函数（SDF）。通过编码到最近障碍物的距离，SDF提供了比单纯的占用图更丰富的规划和轨迹优化信息。我们开发了一种八叉树残差网络（Octree REsidual Network，OREN），将显式的八叉树先验与隐式神经残差相结合，从点云观测中在线重建SDF，兼具体积方法的效率和神经方法的准确性与可微性。同时，我们开发了Bubble$^ullet$，一种基于搜索的规划器，利用距离信息生成最大无碰撞球体（我们称之为气泡），并提供终止、完整性和失败检测的正式保证。与基于网格的A$^ullet$搜索相比，在气泡图上的规划显著减少了碰撞检查，并返回形成安全走廊的气泡序列以进行轨迹优化。我们在四旋翼上演示了集成的OREN-Bubble$^ullet$方法，实时导航于未见过的室内环境，并在严格的计算约束下进行。与基线相比，OREN在SDF估计上提高了22%，而Bubble$^ullet$在复杂环境中找到的轨迹长度约为90米，耗时1-3秒，而基线在同一环境中最多需要10秒。

View on arXiv Download PDF AI Translation

计算机视觉 (Computer Vision)

cs.CV / 1 / 2607.18325

Hazard or Anomaly? Evaluating VLMs for Understanding Dangers and Discrepancies

危险还是异常？评估视觉语言模型（VLMs）以理解危险和差异

Indukuri, Murali, Eskandari, Mohammad, Kollu, Sree Nitya, Lukin, Stephanie, Matuszek, Cynthia

Abstract

Modern safety-critical systems increasingly rely on human-robot interaction to reduce disaster risk and support decision-making during emergencies. Vision-Language Models (VLMs) are promising for these settings because they can interpret complex scenes and communicate safety-relevant information, but they still require careful evaluation to ensure reliable safety reasoning. In particular, current evaluations often frame danger recognition as a binary decision (Safe/Unsafe), making it unclear whether a model is identifying true physical hazards or merely reacting to unusual scene elements. We address this limitation by introducing an explicit distinction between hazard and anomaly, and by separately recognizing hazardous and anomalous states. We evaluate several state-of-the-art VLMs across two datasets and multiple prompting strategies to test whether this distinction changes model behavior. Our results show that VLMs frequently misinterpret anomalousness as hazardousness, revealing an over-reliance on contextual irregularity as a proxy for danger. We further show that explicitly separating anomaly from hazard provides a more informative evaluation of VLM safety reasoning and exposes failure modes that binary safety judgments can obscure. Our public dataset is available on Roboflow https://app.roboflow.com/vlm-in-context-anomaly-and-hazard-detection/camera-ready-roman-ds.

Chinese Translation

现代安全关键系统越来越依赖人机交互来降低灾难风险并支持紧急情况下的决策。视觉语言模型（VLMs）在这些环境中展现出良好的前景，因为它们能够解读复杂场景并传达与安全相关的信息，但仍需进行仔细评估以确保其安全推理的可靠性。特别是，目前的评估通常将危险识别框定为二元决策（安全/不安全），这使得模型是否识别真实的物理危险或仅仅对不寻常的场景元素做出反应变得不清晰。我们通过引入危险与异常之间的明确区分，并分别识别危险状态和异常状态，来解决这一局限性。我们在两个数据集和多种提示策略下评估了几种最先进的VLM，以测试这种区分是否改变模型行为。我们的结果表明，VLMs经常将异常性误解为危险性，揭示了对上下文不规则性的过度依赖作为危险的代理。我们进一步表明，明确将异常与危险分开提供了对VLM安全推理更具信息量的评估，并揭示了二元安全判断可能掩盖的失败模式。我们的公共数据集可在Roboflow上获取：https://app.roboflow.com/vlm-in-context-anomaly-and-hazard-detection/camera-ready-roman-ds。

View on arXiv Download PDF AI Translation

cs.CV / 2 / 2607.18349

From Pixel to Prognosis: Convolutional and GLCM Feature Fusion for Automated Four-Class Cataract Severity Classification

从像素到预后：卷积与灰度共生矩阵特征融合用于自动化四类白内障严重程度分类

Mithra, K., Santhanam, Prem Kumar

Abstract

Objective: To develop a low-cost automated cataract severity classification system operating on standard consumer-grade colour photographs of the eye, without specialised ophthalmic hardware. Methods: A hybrid framework was designed that fuses deep features from a Convolutional Neural Network (CNN) with five handcrafted Grey-Level Co-occurrence Matrix (GLCM) and intensity descriptors - mean intensity, uniformity, standard deviation, contrast, and energy - extracted from a Hough-circle-localised pupil Region of Interest (ROI). A multi-class Support Vector Machine (SVM) with Radial Basis Function (RBF) kernel classifies each image into one of four severity grades: normal, immature, mature, or hypermature cataract. Results: The proposed fused system achieved 95.0% accuracy, 93.8% sensitivity, and 96.1% specificity on an ophthalmologist-labelled test set drawn from 300 images (75 per class) collected at an ophthalmology clinic, outperforming texture-only (88.5%) and CNN-only (91.3%) baselines and surpassing recently published deep learning approaches. Conclusion: The CNN-GLCM-SVM fusion framework provides competitive four-class cataract grading without GPU acceleration or specialised cameras, making it suitable for primary-care and telemedicine deployment in resource-limited settings.

Chinese Translation

目的：开发一个低成本的自动化白内障严重程度分类系统，能够在标准消费级眼睛彩色照片上运行，而无需专业的眼科硬件。方法：设计了一个混合框架，将卷积神经网络（CNN）提取的深层特征与五个手工制作的灰度共生矩阵（GLCM）和强度描述符（均值强度、均匀性、标准差、对比度和能量）融合，这些描述符是从霍夫圆定位的瞳孔感兴趣区域（ROI）中提取的。采用具有径向基函数（RBF）核的多类支持向量机（SVM）将每幅图像分类为四个严重程度等级之一：正常、未成熟、成熟或高度成熟白内障。结果：所提出的融合系统在从300幅图像（每类75幅）中提取的眼科医生标记的测试集上达到了95.0%的准确率、93.8%的灵敏度和96.1%的特异性，优于仅使用纹理（88.5%）和仅使用CNN（91.3%）的基线，并超越了最近发布的深度学习方法。结论：CNN-GLCM-SVM融合框架在没有GPU加速或专业相机的情况下提供了竞争力的四类白内障分级，适合在资源有限的环境中进行初级护理和远程医疗部署。

View on arXiv Download PDF AI Translation

cs.CV / 3 / 2607.18436

Surprise Forcing: What to Remember, When to Skip in Long Video Generation

惊喜强制：在长视频生成中该记住什么，何时跳过

Shi, Shuwei, Li, Zhen, Niu, Muyao, Li, Chuanhao, Zheng, Bo, Zhang, Kaipeng, Zheng, Yinqiang

Abstract

Streaming autoregressive diffusion makes minute-scale video synthesis practical, but its bounded context and fixed denoising schedule allocate resources uniformly across a highly non-stationary sequence. A rolling key-value cache forgets distant visual evidence even when that evidence remains important, while every generated chunk receives the same number of denoising passes irrespective of its actual difficulty. We introduce Surprise Forcing, a training-free framework that treats both limitations as online resource-allocation problems. A Surprise-Gated Memory Bank summarizes evicted frames with value-token descriptors, evaluates them using complementary global-deviation and nearest-neighbor novelty signals, and regulates admission through a feedback-controlled budget in normalized score space. Priority-based replacement and relevance-aware routing then keep the external memory compact and useful. In parallel, Surprise-Aware Denoising estimates chunk difficulty from the maximum adjacent-frame cosine distance after the first denoising pass and uses a local percentile scheduler to skip intermediate steps for comparatively easy chunks. Experiments on VBench, VBench-Long, and VBench-2.0 show that the proposed allocation strategy improves long-horizon consistency and visual quality while retaining real-time streaming throughput.

Chinese Translation

流式自回归扩散使得分钟级视频合成成为可能，但其有限的上下文和固定的去噪调度在高度非平稳的序列中均匀分配资源。滚动的键值缓存即使在远处的视觉证据仍然重要时也会忘记这些证据，而每个生成的片段无论其实际难度如何都接收相同数量的去噪处理。我们提出了惊喜强制（Surprise Forcing），这是一个无训练的框架，将这两种限制视为在线资源分配问题。惊喜门控记忆库（Surprise-Gated Memory Bank）使用价值标记描述符总结被驱逐的帧，通过互补的全局偏差和最近邻新颖性信号对其进行评估，并通过在归一化得分空间中的反馈控制预算来调节接纳。基于优先级的替换和关注相关性的路由则保持外部记忆的紧凑性和实用性。同时，惊喜感知去噪（Surprise-Aware Denoising）在第一次去噪处理后根据最大相邻帧余弦距离估计片段难度，并使用局部百分位调度器为相对简单的片段跳过中间步骤。在VBench、VBench-Long和VBench-2.0上的实验表明，所提出的分配策略在保持实时流式吞吐量的同时，提高了长时间一致性和视觉质量。

View on arXiv Download PDF AI Translation

cs.CV / 4 / 2607.18466

ECoNGS: Efficient Compressive Neural Gaussian Splats for Volume Visualization

ECoNGS：用于体积可视化的高效压缩神经高斯点云

Tang, Kaiyuan, Wang, Chaoli

Abstract

Recent advances in differentiable Gaussian splatting have highlighted the potential of primitive-based approaches as alternative scene representations for interactive, high-quality, volume visualization (VolVis) of large datasets. However, the explicit nature of current primitive-based methods, combined with isolated optimization for each VolVis scene, results in redundant, non-compact representations. We present ECoNGS, an efficient compressive neural Gaussian splatting framework for VolVis scene representation. ECoNGS employs lightweight neural networks to dynamically predict implicit, editable Gaussian splats from explicit anchor points, effectively combining model compactness and parameter efficiency of implicit representations with high-performance rendering of explicit primitives. We explore a joint learning strategy that clusters geometrically similar scenes and shares parameters across them, significantly reducing overall training time and model size while maintaining reconstruction fidelity. To achieve a more compact scene representation, we further compress the explicit anchor attributes using a neural entropy model that estimates their probability distributions, enabling compact storage via entropy coding. We systematically investigate Gaussian initialization strategies and propose a simple yet effective scheme tailored for VolVis scenes, improving reconstruction accuracy and accelerating convergence. We evaluate ECoNGS qualitatively and quantitatively across various univariate and multivariate VolVis scenes, highlighting its superior performance over prior methods in training time, reconstruction quality, and model size. In particular, compared with the prior method iVR-GS, ECoNGS improves reconstruction quality by up to 2.2 dB in PSNR while reducing the model size by up to 6.1x and the training time by up to 5.9x. The code is available at https://github.com/TouKaienn/ECoNGS.

Chinese Translation

最近在可微分高斯点云方面的进展突显了基于原始图元的方法作为大规模数据集交互式高质量体积可视化（VolVis）替代场景表示的潜力。然而，当前基于原始图元的方法的显式特性，加上对每个VolVis场景的孤立优化，导致了冗余且不紧凑的表示。我们提出了ECoNGS，一个用于VolVis场景表示的高效压缩神经高斯点云框架。ECoNGS采用轻量级神经网络，从显式锚点动态预测隐式、可编辑的高斯点云，有效结合了隐式表示的模型紧凑性和参数效率与显式图元的高性能渲染。我们探索了一种联合学习策略，该策略将几何相似的场景聚类并在它们之间共享参数，显著减少了整体训练时间和模型大小，同时保持重建保真度。为了实现更紧凑的场景表示，我们进一步使用神经熵模型压缩显式锚点属性，该模型估计其概率分布，从而通过熵编码实现紧凑存储。我们系统地研究了高斯初始化策略，并提出了一种简单而有效的方案，专门针对VolVis场景，改善了重建精度并加速了收敛。我们在各种单变量和多变量VolVis场景中对ECoNGS进行了定性和定量评估，突显了其在训练时间、重建质量和模型大小方面优于先前方法的表现。特别是，与之前的方法iVR-GS相比，ECoNGS在PSNR中提高了重建质量，最高可达2.2 dB，同时将模型大小减少了最高6.1倍，训练时间减少了最高5.9倍。代码可在https://github.com/TouKaienn/ECoNGS获取。

View on arXiv Download PDF AI Translation

cs.CV / 5 / 2607.18508

Style over Substance: A Shortcut Audit of Emotion-Description Preference Evaluation

形式重于内容：情感描述偏好评估的快捷审计

Yang, Jiabing, Chen, Yixiang, Xu, Yuan, Ma, Qisen, Yu, Tao, Li, Peiyan, Li, Yingda, Huang, Yan, Wang, Liang

Abstract

Preference over model-generated emotion descriptions is emerging as a standard evaluation metric for multimodal emotion understanding, exemplified by the MER2026 MER-Prefer track on EmoPrefer. Such benchmarks assume that predicting the preferred description requires grounded cross-modal understanding of the video. We conduct a systematic shortcut audit of EmoPrefer using content-blind probes. A simple logistic regression using only description length and generator identity, without processing the text, video, or audio, performs comparably to LoRA-finetuned 7B text and audio-visual judges (65.8 versus 66.8 WAF on EmoPrefer-V2). Generator identity is recoverable from description text with 99.5 percent accuracy, every candidate pair contrasts two distinct generators, and the human preference labels agree with a fold-exclusive per-generator win-rate prior on 66 percent of the evaluated pairs. When the human label conflicts with this prior, trained judges still follow the style prior on 63 to 80 percent of the pairs. On a length-matched subset that neutralizes verbosity bias, the tested media configurations yield no statistically significant improvement, while an ODIN-inspired diagnostic that decouples the style shortcut leaves its content head near chance. These results do not imply that human preferences are inherently stylistic or that the descriptions contain no emotional information. Instead, they show that the current scores can be reached without verifying either description against the video. We recommend source-balanced pairing, strict length control, counter-stereotypical sliced reporting, and multi-annotator consensus for future cross-generator evaluations. Code is available at https://github.com/jiabingyang01/EmoPrefer-Audit.

Chinese Translation

对模型生成的情感描述的偏好正在成为多模态情感理解的标准评估指标，这在EmoPrefer的MER2026 MER-Prefer赛道中得到了体现。这些基准假设，预测优选描述需要对视频进行扎实的跨模态理解。我们使用内容盲探针对EmoPrefer进行了系统的快捷审计。仅使用描述长度和生成器身份的简单逻辑回归，在不处理文本、视频或音频的情况下，其表现与经过LoRA微调的7B文本和视听评审者相当（在EmoPrefer-V2上分别为65.8与66.8 WAF）。生成器身份可以从描述文本中以99.5%的准确率恢复，每对候选项对比两个不同的生成器，并且人类偏好标签与66%的评估对中基于生成器的胜率先验一致。当人类标签与这一先验发生冲突时，经过训练的评审者仍然在63%到80%的对中遵循风格先验。在一个消除冗长偏见的长度匹配子集中，测试的媒体配置未能带来统计学上显著的改善，而一个受ODIN启发的诊断方法，解耦了风格快捷，导致其内容头接近随机。这些结果并不意味着人类偏好本质上是风格化的，或描述中不包含情感信息。相反，它们表明当前的得分可以在不验证描述与视频之间的关系的情况下达到。我们建议在未来的跨生成器评估中采用源平衡配对、严格的长度控制、反刻板印象的分段报告和多评审者共识。代码可在 https://github.com/jiabingyang01/EmoPrefer-Audit 获取。

View on arXiv Download PDF AI Translation

cs.CV / 6 / 2607.18510

DuSPiT: Dual-Branch Sub-Patch Pixel Diffusion Transformer

DuSPiT：双分支子补丁像素扩散变换器

Bai, Yunpeng, Gandelsman, Yossi, Gharbi, Michaël

Abstract

Diffusion Transformers achieve strong image generation performance, but most operate in compressed latent spaces. Pixel-space diffusion avoids this information loss, yet existing approaches map each raw image patch to a single token, forcing one representation to handle both global communication and fine-grained details. We address this issue by proposing a new architecture, \textbf{DuSPiT}, a \textbf{Du}al-branch \textbf{S}ub\textbf{P}atch \textbf{Pi}xel \textbf{T}ransformer. This model separates global structural reasoning from local appearance modeling. DuSPiT uses a compact base branch for efficient global reasoning and a parallel, high-capacity pixel branch, organized into subpatch groups, to preserve detailed appearance, with the two branches interacting through cross-attention. Our results show that DuSPiT generates images with richer details and stronger fine-grained structures, while also achieving a better quality--efficiency trade-off than prior pixel-space diffusion transformers.

Chinese Translation

扩散变换器在图像生成性能上表现出色，但大多数在压缩的潜在空间中运行。像素空间扩散避免了信息损失，但现有方法将每个原始图像补丁映射到单个标记，迫使一个表示同时处理全局通信和细粒度细节。我们通过提出一种新架构—— extbf{DuSPiT}（双分支子补丁像素变换器）来解决这个问题。该模型将全局结构推理与局部外观建模分开。DuSPiT使用一个紧凑的基础分支进行高效的全局推理，并使用一个并行的高容量像素分支，组织成子补丁组，以保留详细的外观，两个分支通过交叉注意力进行交互。我们的结果表明，DuSPiT生成的图像具有更丰富的细节和更强的细粒度结构，同时在质量与效率的权衡上优于之前的像素空间扩散变换器。

View on arXiv Download PDF AI Translation

cs.CV / 7 / 2607.18539

AniGS: Bridging Rendering and Diffusion Prior for 3D Scene Animation

AniGS：桥接渲染与扩散先验以实现3D场景动画

Cheng, Yen-Chi, Gao, Chen, Chen, Chuhan, Li, Tuotuo, Shah, Rajvi, Saraf, Ayush, Kim, Changil, Gui, Liangyan, Schwing, Alexander, Kopf, Johannes, Tseng, Hung-Yu

Abstract

Novel view rendering of large and complex reconstructed scenes is becoming increasingly photorealistic. However, most reconstructions remain static and lack the ambient motion that makes environments immersive. We present AniGS, a method for scene-level animation of 3D Gaussian Splatting (3DGS) reconstructions that adds subtle, distributed dynamics, e.g., vegetation motion, while preserving rigid structures. Unlike existing 3D animation techniques which are limited to object-centric subjects or small regions, AniGS is designed for large, cluttered, navigable scenes. AniGS represents the scene with a canonical 3DGS and models motion using a time-conditioned deformation field. To animate the entire scene, we leverage a pretrained video diffusion model and introduce an iterative dataset--model update strategy that progressively expands viewpoint coverage and repeatedly updates camera-fixed training videos using a render-and-refine scheme. To prevent artifacts from unintended motion in static areas, we further introduce a composed video-to-video refinement scheme that restricts motion to desired regions. Experiments on five real-world, large-scale outdoor scenes demonstrate that AniGS produces natural ambient dynamics and high-quality novel view videos, enabling more immersive viewing experiences of reconstructed environments.

Chinese Translation

大型复杂重建场景的新视角渲染正变得越来越逼真。然而，大多数重建仍然是静态的，缺乏使环境沉浸的环境运动。我们提出了AniGS，一种用于3D高斯点云重建（3D Gaussian Splatting, 3DGS）场景级动画的方法，它在保留刚性结构的同时，增加了细微的、分布式的动态，例如植被运动。与现有的3D动画技术仅限于以物体为中心的主题或小区域不同，AniGS旨在处理大型、杂乱且可导航的场景。AniGS使用标准的3DGS表示场景，并通过时间条件变形场建模运动。为了对整个场景进行动画处理，我们利用了预训练的视频扩散模型，并引入了一种迭代的数据集-模型更新策略，该策略逐步扩展视角覆盖范围，并使用渲染与精炼方案反复更新固定相机的训练视频。为了防止静态区域中意外运动造成的伪影，我们进一步引入了一种复合视频到视频的精炼方案，限制运动仅发生在所需区域。对五个真实世界的大规模户外场景的实验表明，AniGS能够生成自然的环境动态和高质量的新视角视频，从而实现对重建环境的更具沉浸感的观看体验。

View on arXiv Download PDF AI Translation

cs.CV / 8 / 2607.18540

Recti-Q: Feature-Space Rectification for Out-of-Distribution-Robust Quantized Perception in Edge Robotics

Recti-Q：边缘机器人中面向分布外鲁棒量化感知的特征空间修正

Araghi, Hamidreza Yaghoubi, Pilevar, Parastoo, Lin, Ming C.

Abstract

Robotic perception pipelines increasingly rely on large vision backbones deployed on SWaP-constrained edge platforms, making post-training quantization (PTQ) attractive for real-time inference. However, while PTQ often preserves clean in-distribution accuracy, we show that it can substantially degrade reliability under deployment-relevant distribution shifts (e.g., sensor noise, severe weather, and novel operating environments), creating a Quantization-Induced Robustness Gap. Across foundational vision benchmarks (ImageNet-C and PACS), 4-bit PTQ models exhibit pronounced robustness degradation despite negligible ID accuracy loss. To address this, we propose Recti-Q, a lightweight feature-space rectification framework that freezes the quantized backbone and trains a small classifier-head LoRA adapter using only source data. Recti-Q is architecture-agnostic across CNNs and Transformers, supports efficient teacher-free training, and recovers a significant portion of the lost robustness, in some cases matching or exceeding FP32 performance. At less than 1% parameter overhead (as small as 6 KB), Recti-Q preserves over 99% of PTQ memory savings, adds negligible compute, and enables low-bandwidth Over-The-Air (OTA) resilience patching for deployed robotic fleets operating in unpredictable physical environments.

Chinese Translation

机器人感知管道越来越依赖于部署在受限于大小、重量和功耗（SWaP）的边缘平台上的大型视觉骨干网络，使得后训练量化（PTQ）在实时推理中变得具有吸引力。然而，尽管PTQ通常能够保持干净的内部分布准确性，我们展示了它在与部署相关的分布变化（例如，传感器噪声、恶劣天气和新操作环境）下可能显著降低可靠性，从而产生量化引起的鲁棒性差距。在基础视觉基准（ImageNet-C和PACS）上，4位PTQ模型表现出明显的鲁棒性下降，尽管内部分布准确性损失微乎其微。为了解决这个问题，我们提出了Recti-Q，一个轻量级特征空间修正框架，该框架冻结量化骨干网络，并仅使用源数据训练一个小型分类器头LoRA适配器。Recti-Q在卷积神经网络（CNN）和变换器（Transformers）架构上均具有架构无关性，支持高效的无教师训练，并恢复了大量丢失的鲁棒性，在某些情况下，其性能与FP32相当或超过FP32。在不到1%的参数开销（小至6 KB）的情况下，Recti-Q保留了超过99%的PTQ内存节省，增加的计算量微乎其微，并为在不可预测的物理环境中运行的部署机器人车队提供了低带宽的空中（OTA）恢复补丁。

View on arXiv Download PDF AI Translation

cs.CV / 9 / 2607.18544

Physics Closure Matters for Machine Olfaction: A Maxwell--Stefan Graph Solver for Identifiable Dynamic Gas Unmixing

物理闭合对机器嗅觉的重要性：一种用于可识别动态气体分离的麦克斯韦-斯特凡图解法

Shi, Yue, Han, Liangxiu, Zhang, Xin, Sobeih, Tam

Abstract

Machine olfaction for gas unmixing is an underconstrained inverse problem in which gas compositions must be inferred from low-dimensional, delayed, and entangled sensor responses produced by interacting chemical transport, surface adsorption, and sensor transduction. One of the key obstacles is physics closure misspecification, where a neural network is designed to fit sensor traces rather than infer a physically closed olfactory process. In this work, we formulate gas unmixing as a multi-physics-constrained inverse problem governed by Maxwell--Stefan multicomponent transport PDEs, competitive adsorption ODEs, and nonlinear sensor transduction ODEs. Directly solving such a high-dimensional coupled system is computationally expensive and often numerically unstable. To this end, we propose UnMixNet, a physics-closed graph neural solver that embeds this multi-physics forward process into end-to-end gas unmixing. UnMixNet discretizes Maxwell--Stefan cross-diffusion on spatial graphs and formulates the multicomponent flux on each edge. This design enables local, differentiable, and flux-conservative inference for multicomponent cross-diffusion. Evaluations on SmellNet show improved single-odor recognition, seen-mixture unmixing, and unseen-mixture generalization. In addition, an external validation on UCI Dynamic Gas Mixtures shows that the inferred concentration process agrees with ground truth concentration set points under dynamic transitions. Process-consistency diagnostics further show that the proposed model learns transferable dynamic physical fingerprints that better satisfies transport, conservation, adsorption, and readout closure.

Chinese Translation

机器嗅觉用于气体分离是一个欠约束的逆问题，其中必须从低维、延迟和交织的传感器响应中推断气体成分，这些响应是由相互作用的化学传输、表面吸附和传感器转导产生的。一个关键障碍是物理闭合的错误指定，在这种情况下，神经网络被设计用来拟合传感器轨迹，而不是推断一个物理闭合的嗅觉过程。在本研究中，我们将气体分离表述为一个多物理约束的逆问题，该问题由麦克斯韦-斯特凡多组分传输偏微分方程、竞争吸附常微分方程和非线性传感器转导常微分方程所支配。直接求解这样一个高维耦合系统在计算上代价高昂且通常数值不稳定。为此，我们提出了UnMixNet，一种物理闭合的图神经解算器，它将这一多物理前向过程嵌入到端到端的气体分离中。UnMixNet在空间图上离散化麦克斯韦-斯特凡交叉扩散，并在每条边上制定多组分通量。这一设计使得多组分交叉扩散的局部、可微分和通量守恒推断成为可能。在SmellNet上的评估显示出单一气味识别、已知混合物分离和未知混合物泛化的改善。此外，在UCI动态气体混合物上的外部验证显示，推断的浓度过程与动态转变下的真实浓度设定点一致。过程一致性诊断进一步表明，所提出的模型学习到可转移的动态物理指纹，更好地满足传输、守恒、吸附和读出闭合的要求。

View on arXiv Download PDF AI Translation

cs.CV / 10 / 2607.18576

Text-conditioned Segmentation for Tomato Phenotyping via Procedural Synthetic Data

基于文本的番茄表型分割通过程序合成数据

Mounir, Samy, Cieslak, Mikolaj, Dhieb, Najmeddine, Ghazzai, Hakim, Klein, Jonathan, Froehlich, Katja, Pirk, Soeren, Palubicki, Wojciech, Setti, Gianluca, Eltawil, Ahmed M., Michels, Dominik L.

Abstract

Vision-based automation is an excellent candidate for reducing manual labor in greenhouse crop production and phenotyping. However, progress is constrained by the lack of annotated training data. Recent advances in vision-based foundational models have shown promising results in zero-shot generalization to novel domains, but their performance drops in complex agricultural environments. In this work, we present a sim-to-real framework for tomato plant segmentation that combines synthetic data generation with fine-tuning of a foundation model. We model a commercial cherry tomato greenhouse and use it to generate a large-scale synthetic dataset under diverse viewpoints, lighting conditions, and plant morphology. Subsequently, we fine-tune the Segment Anything Model 3 (SAM 3) on the synthetic dataset, specializing its text-conditioned segmentation behavior for greenhouse crop organs while retaining the general visual prior that makes zero-shot transfer possible. By evaluating our framework on multiple real-world greenhouse datasets, we demonstrate that combining synthetic data with SAM 3 fine-tuning significantly improves segmentation performance and model confidence. To support community benchmarking, we publicly release the procedural model, the generated synthetic dataset, and our fine-tuned SAM 3 weights.

Chinese Translation

基于视觉的自动化是减少温室作物生产和表型分析中人工劳动的优秀候选方案。然而，进展受到缺乏标注训练数据的限制。最近在基于视觉的基础模型方面的进展显示出在新领域的零样本泛化方面的良好结果，但在复杂的农业环境中其性能下降。在本研究中，我们提出了一种用于番茄植物分割的仿真到现实框架，该框架结合了合成数据生成与基础模型的微调。我们对一个商业樱桃番茄温室进行了建模，并利用其生成了一个在多种视角、光照条件和植物形态下的大规模合成数据集。随后，我们在合成数据集上微调了Segment Anything Model 3 (SAM 3)，使其文本条件分割行为专门针对温室作物器官，同时保留了使零样本迁移成为可能的一般视觉先验。通过在多个真实温室数据集上评估我们的框架，我们证明了将合成数据与SAM 3微调相结合显著提高了分割性能和模型信心。为了支持社区基准测试，我们公开发布了程序模型、生成的合成数据集以及我们微调后的SAM 3权重。

View on arXiv Download PDF AI Translation

cs.CV / 11 / 2607.18577

Attention Without Grounding: Causal Evaluation of Visual Explanations in Medical VLMs

无基础的注意力：医学视觉语言模型中视觉解释的因果评估

Sadanandan, Binesh, Behzadan, Vahid

Abstract

Attention and saliency heatmaps are widely used to explain medical Vision-Language Model (VLM) outputs on chest X-rays, yet whether they truly highlight the image evidence driving predictions has not been causally tested. We audit faithfulness via overlap with radiologist bounding boxes on PadChest (n=637), attribution mass within radiologist masks on CheXlocalize (n=643), and 16x16 patch-occlusion maps that record which regions, when hidden, change the answer. We study three MedGemma-4B variants, cross-family probes on LLaVA-RAD and Qwen3-VL-8B-Instruct, and the specialist CheXagent-2-3b, with two CXR-trained classifiers (DenseNet121, ResNet50) as positive controls. A heatmap is faithful only if the model uses the image and attention concentrates on regions whose occlusion alters the prediction. No evaluated VLM meets both criteria. MedGemma and Qwen3-VL use the image, but attention anti-correlates with patch-occlusion importance (rho < 0 with 95% bootstrap CIs below zero). LLaVA-RAD's attention correlates positively, but the model is almost text-only (99.1% text-only agreement, near-zero causal mass), so correlation ties two near-zero signals. Attention also misses annotated anatomy: overlap with true regions never beats shifted or random controls, and no method places more than 22% of its mass inside radiologist masks. The two CXR classifiers pass all metrics, indicating the failure is specific to VLM heatmaps, not the evaluation. These heatmaps are visually reassuring but not faithful; clinical explanations require controlled localization metrics and causal perturbation, not visual inspection alone.

Chinese Translation

注意力和显著性热图被广泛用于解释胸部X光片上的医学视觉语言模型（VLM）输出，但它们是否真正突出了驱动预测的图像证据尚未经过因果测试。我们通过与PadChest（n=637）上的放射科医生边界框的重叠、与CheXlocalize（n=643）上的放射科医生掩膜内的归因质量，以及记录哪些区域在隐藏时会改变答案的16x16补丁遮挡图进行忠实性审计。我们研究了三种MedGemma-4B变体、LLaVA-RAD和Qwen3-VL-8B-Instruct上的跨家族探针，以及专业的CheXagent-2-3b，使用两个经过胸部X光训练的分类器（DenseNet121，ResNet50）作为正对照。只有当模型使用图像且注意力集中在遮挡改变预测的区域时，热图才是忠实的。没有评估的VLM满足这两个标准。MedGemma和Qwen3-VL使用图像，但注意力与补丁遮挡重要性呈负相关（rho < 0，95%自助法置信区间低于零）。LLaVA-RAD的注意力呈正相关，但该模型几乎是文本专用的（99.1%的文本专用一致性，近乎零的因果质量），因此相关性将两个近零信号联系在一起。注意力也未能捕捉到标注的解剖结构：与真实区域的重叠从未超过偏移或随机对照，并且没有任何方法将超过22%的质量放置在放射科医生的掩膜内。这两个胸部X光分类器通过了所有指标，表明失败特定于VLM热图，而不是评估。这些热图在视觉上令人安心，但并不忠实；临床解释需要受控的定位指标和因果扰动，而不仅仅是视觉检查。

View on arXiv Download PDF AI Translation

cs.CV / 12 / 2607.18625

Norm or Direction? Decoding Vision Mambas for High-Resolution Vision

规范还是方向？解码高分辨率视觉中的视觉Mamba

Yu, Jin, Park, Juyoun

Abstract

Vision Mamba models replace quadratic self-attention with linear complexity selective state space models (SSMs), emerging as efficient visual backbones. However, MambaOut demonstrates that a Gated CNN block can match or exceed VMamba on image classification, questioning the necessity of SSMs for vision. This raises a fundamental question: do VMamba and MambaOut encode visual information differently at the representation level? To investigate, we apply cross model centered kernel alignment (CKA) analysis and find that VMamba's final stage blocks form representations distinctly different from both MambaOut and its own preceding blocks. We therefore focus on the final block features, decomposing each spatial token into magnitude and direction. MambaOut concentrates class-discriminative information in high-norm foreground tokens that align with Grad-CAM attribution. VMamba, by contrast, produces high-norm tokens predominantly in background regions, misaligned with Grad-CAM, yet preserves discriminative signals primarily in token directions. These observations reveal that the two models rely on different encoding strategies. We connect this difference to high-resolution classification and semantic segmentation. VMamba distributes logit support broadly across object regions, whereas MambaOut relies on sparse dominant tokens, a strategy that becomes less stable as token counts grow. Under full fine-tuning for segmentation, VMamba consistently outperforms MambaOut. These results suggest that VMamba's advantage in dense prediction stems not merely from the SSM mechanism or sequence length, but from how semantic evidence is organized across token magnitude, direction. Ultimately, we conclude that token magnitude and directional structure serve as critical axes for improving visual backbones, particularly under dense supervision.

Chinese Translation

视觉Mamba模型用线性复杂度的选择性状态空间模型（SSMs）替代了二次自注意力，成为高效的视觉骨干网络。然而，MambaOut展示了一个门控卷积神经网络（Gated CNN）模块可以在图像分类上匹配或超越VMamba，这引发了对SSMs在视觉任务中必要性的质疑。这提出了一个根本性的问题：VMamba和MambaOut在表征层面上是否以不同的方式编码视觉信息？为此，我们应用跨模型中心核对齐（CKA）分析，发现VMamba的最终阶段模块形成的表征与MambaOut及其自身先前模块显著不同。因此，我们专注于最终模块的特征，将每个空间标记分解为幅度和方向。MambaOut将类区分信息集中在与Grad-CAM归因对齐的高幅度前景标记中。相比之下，VMamba在背景区域产生的高幅度标记与Grad-CAM不对齐，但主要在标记方向上保留了区分信号。这些观察结果揭示了两个模型依赖于不同的编码策略。我们将这种差异与高分辨率分类和语义分割联系起来。VMamba在物体区域广泛分布logit支持，而MambaOut依赖于稀疏的主导标记，这种策略在标记数量增加时变得不那么稳定。在进行全量微调以进行分割时，VMamba始终优于MambaOut。这些结果表明，VMamba在密集预测中的优势不仅源于SSM机制或序列长度，而在于语义证据如何在标记幅度和方向上组织。最终，我们得出结论，标记幅度和方向结构是改善视觉骨干网络的关键轴，尤其是在密集监督下。

View on arXiv Download PDF AI Translation

cs.CV / 13 / 2607.18630

Seeing Before Generating: Object Perception Enhances Single-View 3D Reconstruction

生成之前的感知：物体感知增强单视图三维重建

Huynh, Y, Nguyen, Duc Thanh, Abdelrazek, Mohamed

Abstract

The relationship between object perception and reconstruction is well established in human vision, yet remains underexplored in computer vision. In this paper, we demonstrate that learnt object perception can significantly enhance 3D reconstruction. Focusing on the challenging task of single-view 3D object reconstruction, we propose a method that leverages perceptual signals extracted from pretrained perception models capturing semantic and geometric information to drive the reconstruction of an object from its single image. Our approach is model-agnostic and can be integrated into various reconstruction methods in a plug-and-play manner. Experiments with two state-of-the-art single-view 3D reconstruction pipelines in a benchmark dataset show consistent and substantial improvements achieved by our method, validating the effectiveness of incorporating perception into generation. We provide in-depth analysis of various aspects of our method and its application. Our project page is at https://ynhuhuynh.github.io/perception-3d/.

Chinese Translation

物体感知与重建之间的关系在人的视觉中已得到充分建立，但在计算机视觉中仍然未得到充分探索。本文展示了学习到的物体感知可以显著增强三维重建。我们专注于单视图三维物体重建这一具有挑战性的任务，提出了一种方法，该方法利用从预训练感知模型中提取的感知信号，这些模型捕捉了语义和几何信息，以驱动从单幅图像重建物体。我们的方法是模型无关的，可以以即插即用的方式集成到各种重建方法中。在基准数据集中对两种最先进的单视图三维重建管道进行的实验表明，我们的方法实现了一致且显著的改进，验证了将感知融入生成过程的有效性。我们对方法的各个方面及其应用进行了深入分析。我们的项目页面为 https://ynhuhuynh.github.io/perception-3d/。

View on arXiv Download PDF AI Translation

cs.CV / 14 / 2607.18638

Deep Learning Estimation of Sex, Age, Height, and Weight from CT-derived Digitally Reconstructed Radiographs

基于深度学习的性别、年龄、身高和体重从CT衍生的数字重建放射图像的估计

Kikuchi, Tomohiro, Yamamoto, Kohei, Nomura, Yukihiro, Yamagishi, Yosuke, Yoshikawa, Takeharu, Akashi, Toshiaki, Kamohara, Jun, Fujii, Hiroyuki, Mori, Harushi

Abstract

Purpose: To develop and validate a deep learning ensemble for estimating adult sex, age, height, and weight from coronal digitally reconstructed radiographs (DRRs) generated from diagnostic CT. Materials and Methods: This retrospective study included 128,621 CT examinations from 80,004 adults at nine institutions in Japan. Three multitask models-ConvNeXt-Base, ViT-Base/16, and MaxViT-Base-were fine-tuned using coronal DRRs and combined by weighted averaging. Data were split by institution into training (114,147 examinations; seven institutions), tuning (4,305; one institution), and test (10,169; one institution) sets; generalizability was assessed on two non-Japanese datasets. Accuracy and mean absolute error (MAE) were used to evaluate sex classification and age, height, and weight regression, respectively. Body surface area (BSA)-corrected heart and liver volume trends were compared using true versus estimated height and weight. Results: In the test set (median age, 69.9 years; 4,899 of 10,169 [48.2%] male), overall sex-classification accuracy was 0.997 (95% CI, 0.996-0.998), and MAEs were 3.57 years (3.51-3.63), 2.59 cm (2.54-2.64), and 3.40 kg (3.34-3.47) for age, height, and weight, respectively. In examinations covering the chest through pelvis, accuracy was 1.000, and MAEs were 3.15 years, 2.28 cm, and 3.18 kg, respectively. BSA calculated from estimated values reproduced age-related heart and liver volume trends obtained using true values. On non-Japanese datasets, height error increased but was reduced by continued fine-tuning. Conclusion: The ensemble estimated adult sex, age, height, and weight from CT-derived DRRs, with generally lower errors in examinations with broader anatomical coverage.

Chinese Translation

目的：开发和验证一种深度学习集成模型，以从由诊断CT生成的冠状面数字重建放射图像（DRRs）中估计成年人的性别、年龄、身高和体重。材料与方法：本回顾性研究包括来自日本九个机构的80,004名成人的128,621例CT检查。使用冠状面DRRs对三个多任务模型——ConvNeXt-Base、ViT-Base/16和MaxViT-Base进行了微调，并通过加权平均进行组合。数据按机构分为训练集（114,147例检查；七个机构）、调优集（4,305例；一个机构）和测试集（10,169例；一个机构）；在两个非日本数据集上评估了模型的泛化能力。使用准确率和平均绝对误差（MAE）分别评估性别分类和年龄、身高、体重回归。比较了基于真实与估计身高和体重的体表面积（BSA）校正的心脏和肝脏体积趋势。结果：在测试集中（中位年龄69.9岁；10,169例中4,899例[48.2%]为男性），总体性别分类准确率为0.997（95% CI，0.996-0.998），MAE分别为年龄3.57年（3.51-3.63）、身高2.59厘米（2.54-2.64）和体重3.40千克（3.34-3.47）。在覆盖胸部至盆腔的检查中，准确率为1.000，MAE分别为3.15年、2.28厘米和3.18千克。根据估计值计算的BSA再现了使用真实值获得的与年龄相关的心脏和肝脏体积趋势。在非日本数据集中，身高误差有所增加，但通过持续微调得以减少。结论：该集成模型从CT衍生的DRRs中估计成年人的性别、年龄、身高和体重，在具有更广泛解剖覆盖的检查中通常表现出较低的误差。

View on arXiv Download PDF AI Translation

cs.CV / 15 / 2607.18646

Fluid-SDF: Ultra-Lightweight and Editable Implicit Shape Representation via Differentiable Primitives

Fluid-SDF：通过可微分原语实现超轻量且可编辑的隐式形状表示

Sripada, Pradyumna, Nadgir, Chinmay, Agrawal, Ksheer, Kodanganti, Krishna Kanth

Abstract

Implicit Neural Representations (INRs) have become the standard for continuous 2D shape modeling, but they suffer from black-box uneditability, vulnerability to noise, and high parameter counts that severely hinder deployment on edge devices. We introduce Fluid-SDF, a highly compressed, differentiable Constructive Solid Geometry (CSG) framework that models shapes using explicit geometric primitives blended via a smooth minimum function. By replacing traditional multi-layer perceptrons (MLPs) with a parameterized primitive engine, Fluid-SDF reconstructs complex, non-convex topologies using strictly under 100 parameters, achieving comparable or superior intersection-over-union (mIoU) to standard neural baselines. Furthermore, we demonstrate that Fluid-SDF acts as a powerful geometric prior, inherently resisting high-frequency dataset noise where capacity-matched neural networks catastrophically overfit. Finally, unlike standard INRs, Fluid-SDF's explicit parameter space allows for direct, zero-shot user editing of local and global shape features without retraining. By bypassing expensive on-device gradient updates entirely, Fluid-SDF is uniquely suited for mobile AI, augmented reality, and resource-constrained embedded environments

Chinese Translation

隐式神经表示（INRs）已成为连续二维形状建模的标准，但它们存在不可编辑的黑箱特性、对噪声的脆弱性以及高参数数量，这严重阻碍了在边缘设备上的部署。我们提出了Fluid-SDF，这是一种高度压缩的可微分构造固体几何（CSG）框架，通过平滑最小函数混合显式几何原语来建模形状。通过用参数化原语引擎替代传统的多层感知器（MLPs），Fluid-SDF使用严格少于100个参数重建复杂的非凸拓扑，达到与标准神经基线相当或更优的交并比（mIoU）。此外，我们证明Fluid-SDF作为一种强大的几何先验，固有地抵抗高频数据集噪声，而容量匹配的神经网络则会灾难性地过拟合。最后，与标准INRs不同，Fluid-SDF的显式参数空间允许用户直接进行零-shot编辑局部和全局形状特征，而无需重新训练。通过完全绕过昂贵的设备端梯度更新，Fluid-SDF特别适合移动人工智能、增强现实和资源受限的嵌入式环境。

View on arXiv Download PDF AI Translation

cs.CV / 16 / 2607.18664

DeforM: Reasoning-Guided Physics-Aware Video Generation via Spatial-Temporal Masking

DeforM：通过时空掩模进行推理引导的物理感知视频生成

Li, Yunyi, Qiao, Yu, Wang, Yaohui, Chen, Xinyuan

Abstract

Video generation models achieve high visual quality but often struggle to generate physics-aware videos. Unlike rigid-body motion, which can be described by explicit trajectories or formulas, complex deformation dynamics remain challenging to synthesize. We observe that a lack of physical reasoning for localizing dynamic areas allows irrelevant regions to dilute the model's attention, leading to generation failure. In this paper, we propose DeforM, a reasoning-guided image-to-video generation framework that directs the model's focus toward physics-critical regions. To reason about and localize these critical regions, we introduce a VLM-guided physical reasoning module, DeforM-Reason, to identify target objects and generate spatial-temporal masks. For physical guidance, we develop two alternative strategies: DeforM-Free for training-free mechanism analysis and DeforM-Injection as a powerful training-based generator. Experimental results demonstrate that DeforM improves the realism of generated deformation scenarios, outperforming baseline models in both visual quality and physical consistency.

Chinese Translation

视频生成模型在视觉质量上取得了很高的水平，但在生成物理感知视频方面常常面临挑战。与可以通过明确的轨迹或公式描述的刚体运动不同，复杂的变形动态仍然难以合成。我们观察到，缺乏对动态区域的物理推理会导致无关区域稀释模型的注意力，从而导致生成失败。本文提出了DeforM，一种推理引导的图像到视频生成框架，旨在将模型的注意力引导至物理关键区域。为了推理和定位这些关键区域，我们引入了一个基于VLM的物理推理模块DeforM-Reason，以识别目标对象并生成时空掩模。为了提供物理指导，我们开发了两种替代策略：DeforM-Free用于无训练机制分析，DeforM-Injection作为一种强大的基于训练的生成器。实验结果表明，DeforM提高了生成变形场景的真实感，在视觉质量和物理一致性方面均优于基线模型。

View on arXiv Download PDF AI Translation

cs.CV / 17 / 2607.18673

MissingBench-Verified: Probing Vision-Language Models' Inability to Detect Missing Object Parts

MissingBench-Verified：探究视觉语言模型无法检测缺失物体部件的能力

Guo, Wenqi Marshall, Qian, Qingyun, Zhou, Shiyu, Luo, Guoping, Du, Shan

Abstract

Vision Language Models (VLMs) are well known for hallucinating non-existent objects in images. Objects with missing parts present a unique challenge for VLMs, stemming from both real-world knowledge bias and the scarcity of such images in training data. We present MissingBench-Verified, a benchmark designed to evaluate a specific and practically relevant scenario: when vision-language models fail to recognize that an essential component of an object has been removed. Across ten leading models, we observe consistent and significant failure rates that persist even when external tool evidence explicitly contradicts the model's visual perception. We further ask whether granting models access to image processing tools (e.g., cropping, contrast adjustment) enables autonomous inspection to resolve these failures. We find that existing mitigation strategies, including tool-assisted verification, autonomous visual reasoning, longer reasoning durations, and fine-tuning on an easier dataset, provide negligible improvement, indicating that this failure mode cannot be addressed through current prompting or post-hoc correction techniques. Our findings highlight a fundamental limitation of current VLM for inspection and monitoring tasks and underscore the need for architectural or training-level interventions that enable models to override internal expectations when confronted with contradictory evidence.

Chinese Translation

视觉语言模型（VLMs）因在图像中幻觉出不存在的物体而广为人知。缺失部件的物体对VLMs构成了独特的挑战，这既源于现实世界知识的偏差，也源于训练数据中此类图像的稀缺性。我们提出了MissingBench-Verified，这是一个旨在评估一个特定且具有实际相关性的场景的基准：当视觉语言模型未能识别出物体的一个重要组成部分被移除时。我们在十个领先模型中观察到一致且显著的失败率，即使在外部工具证据明确与模型的视觉感知相矛盾时，这种失败率依然存在。我们进一步探讨是否允许模型访问图像处理工具（例如裁剪、对比度调整）能够使其自主检查以解决这些失败。我们发现，现有的缓解策略，包括工具辅助验证、自主视觉推理、更长的推理时间以及在更简单的数据集上进行微调，几乎没有改善，表明这种失败模式无法通过当前的提示或事后修正技术来解决。我们的研究结果突显了当前VLM在检查和监测任务中的基本局限性，并强调了在面对矛盾证据时，模型需要进行架构或训练层面的干预，以便能够覆盖内部预期。

View on arXiv Download PDF AI Translation

cs.CV / 18 / 2607.18678

Cross-Dataset Generalization in Breast MRI Tumor Classification via Class-Wise Dataset Mixing

通过类别间数据集混合实现乳腺MRI肿瘤分类的跨数据集泛化

Dadrast, Mohammad Ali, Usefi, Hamid

Abstract

Breast MRI is highly sensitive for detecting breast tumors, but exams contain many slices and require substantial reading time. Deep learning models often perform well on internal splits but can fail across institutions because of domain shift and dataset-origin bias. We study this failure mode for binary breast MRI tumor classification. EfficientNet-B3 and WaveViT-Small are trained using Duke Breast Cancer MRI and fastMRI, and evaluated only on the independent multi-center MAMA-MIA cohort. In a deliberately confounded setup, where label is perfectly correlated with dataset origin, external accuracy is near chance (0.5048--0.5265), despite very high recall. We then construct a mixed training set in which each class contains samples from both Duke and fastMRI, while preserving patient-level splitting, augmentation, and leakage controls. On MAMA-MIA, dataset mixing improves accuracy/F1 to 0.8463/0.8625 for WaveViT-Small and 0.8884/0.8994 for EfficientNet-B3. These results show that controlling dataset-origin bias is important for reliable breast MRI classification.

Chinese Translation

乳腺MRI在检测乳腺肿瘤方面具有很高的敏感性，但检查包含许多切片，并且需要大量的阅读时间。深度学习模型通常在内部划分上表现良好，但由于领域转移和数据集来源偏差，在不同机构之间可能会失败。我们研究了这一失败模式在二元乳腺MRI肿瘤分类中的表现。使用Duke乳腺癌MRI和fastMRI训练EfficientNet-B3和WaveViT-Small，并仅在独立的多中心MAMA-MIA队列上进行评估。在一个故意混淆的设置中，标签与数据集来源完美相关，外部准确率接近随机（0.5048--0.5265），尽管召回率非常高。然后，我们构建了一个混合训练集，其中每个类别包含来自Duke和fastMRI的样本，同时保持患者级别的划分、增强和泄漏控制。在MAMA-MIA上，数据集混合使WaveViT-Small的准确率/F1提高到0.8463/0.8625，EfficientNet-B3提高到0.8884/0.8994。这些结果表明，控制数据集来源偏差对于可靠的乳腺MRI分类至关重要。

View on arXiv Download PDF AI Translation

cs.CV / 19 / 2607.18688

Dual-Edged Homogeneous-Modality Similarity: Towards Visible-Infrared Modality-Incomplete Person Re-Identification with Modality Adaptive Matching

双刃均质模态相似性：朝着可见-红外模态不完整的人体重识别与模态自适应匹配

Xu, Xin, Zhan, Shuhao, Liu, Wei, Wang, Zheng, Jiang, Kui, Lin, Chia-Wen

Abstract

Visible-Infrared Person Re-Identification (VI-ReID) operates under a closed-world assumption, where queries and galleries are from heterogeneous modalities. However, in open-world scenarios, both sets are likely to contain homogeneous and heterogeneous modality images. A query may consist of visible-only, infrared-only, or mixed-modality images, while galleries present multi-modal images over long-term collection. Under these conditions, VI-ReID methods, built on a heterogeneous-modality retrieval paradigm, suffer from three trustworthiness challenges: matching conflicts due to high homogeneous-modality similarity, interference from modality uncertainty, and robustness degradation induced by unknown modality combinations. They fail to meet the requirements of trustworthy visual recognition in reliability, consistency, and dynamic adaptability. To address these challenges, we formalize the Visible-Infrared Modality-Incomplete Re-Identification (VIMI-ReID) task. We reorganize existing datasets to construct the SYSU-VIMI and RegDB-VIMI benchmarks. The unpredictable modality combinations and inherent similarity of homogeneous-modality samples in VIMI-ReID cause a significant performance drop in existing VI-ReID methods. We propose the Modality Adaptive Matching Transformer (MAMT). It employs a Divergence Transformer Module (DTM) and a Shared Transformer Module (STM) to extract modality-specific and modality-shared features, respectively. Guided by a divergence loss, the DTM enriches modality-specific features with modality-style information to enhance discriminability within the same modality. A Modality Adaptive Matching Module (MAM) dynamically fuses features according to the query-gallery modality relationship, enabling stable matching under arbitrary and uncertain modality conditions. Extensive experiments on the VIMI benchmarks demonstrate the effectiveness and adaptability of MAMT.

Chinese Translation

可见-红外人体重识别（VI-ReID）在封闭世界假设下运行，其中查询和图库来自异质模态。然而，在开放世界场景中，这两组图像可能包含均质和异质模态图像。查询可能由仅可见、仅红外或混合模态图像组成，而图库则呈现长期收集的多模态图像。在这些条件下，基于异质模态检索范式的VI-ReID方法面临三个可信度挑战：由于高均质模态相似性导致的匹配冲突、模态不确定性带来的干扰，以及未知模态组合引起的鲁棒性下降。它们未能满足在可靠性、一致性和动态适应性方面的可信视觉识别要求。为了解决这些挑战，我们正式定义了可见-红外模态不完整重识别（VIMI-ReID）任务。我们重新组织现有数据集，构建SYSU-VIMI和RegDB-VIMI基准。VIMI-ReID中不可预测的模态组合和均质模态样本的内在相似性导致现有VI-ReID方法的性能显著下降。我们提出了模态自适应匹配变换器（MAMT）。它采用了发散变换器模块（DTM）和共享变换器模块（STM）分别提取模态特定和模态共享特征。在发散损失的指导下，DTM利用模态风格信息丰富模态特定特征，以增强同一模态内的区分能力。模态自适应匹配模块（MAM）根据查询-图库模态关系动态融合特征，使得在任意和不确定的模态条件下实现稳定匹配。在VIMI基准上的大量实验表明了MAMT的有效性和适应性。

View on arXiv Download PDF AI Translation

cs.CV / 20 / 2607.18695

Attributes Should Come from Images, Not Class Names: Distribution-Conditioned Attribute Selection for Vision-Language Models

属性应来自图像，而非类别名称：面向视觉-语言模型的分布条件属性选择

Gare, Gautam Rajendrakumar, Shi, Jia, Lin, Zhiqiu, Pathak, Deepak, Galeotti, John, Ramanan, Deva

Abstract

A popular route to interpretable zero-shot classification asks a large language model (LLM) to describe each class name and prompts CLIP with the resulting descriptors. We show that these descriptors carry little visual evidence of their own: removing the class name from the prompt collapses ImageNet accuracy from 59.5% to 15.5%. The diagnosis is that the descriptors are conditioned on the label rather than on the images, so they describe the concept in general and mislead exactly when the data shifts; an LLM insists that strawberries are red, but every strawberry in ImageNet-Sketch is a colorless line drawing. We therefore select attributes from the target image collection instead: we score a large attribute pool against the images in CLIP's joint embedding space and keep the top-scoring attributes per class. Selected this way, class-name-free attribute prompts reach 23.8% on ImageNet (against 15.5% for LLM descriptors), the gain holds on four shifted ImageNet variants, and reselecting from the LLM's own pool isolates the selection mechanism as the cause. With one image per class, the selected attributes outperform the prompt-tuning method CoOp by 3 points while fitting in under a minute instead of 14 hours, with no learned soft prompt to obscure the decision. Because the attribute set is chosen by the data, it doubles as a readable summary of a dataset, which we use to describe distribution shift in words.

Chinese Translation

一种流行的可解释零-shot 分类方法是要求大型语言模型（LLM）描述每个类别名称，并用生成的描述符提示 CLIP。我们表明，这些描述符自身几乎没有视觉证据：从提示中去除类别名称使得 ImageNet 的准确率从 59.5% 降至 15.5%。诊断结果是，这些描述符是基于标签而非图像进行条件化的，因此它们一般性地描述了概念，并在数据发生变化时产生误导；LLM 坚持认为草莓是红色的，但 ImageNet-Sketch 中的每个草莓都是无色的线条图。为此，我们从目标图像集合中选择属性：我们在 CLIP 的联合嵌入空间中对大量属性池进行评分，并保留每个类别中得分最高的属性。通过这种方式选择的无类别名称属性提示在 ImageNet 上达到 23.8%（相比 LLM 描述符的 15.5%），这一增益在四个变化的 ImageNet 变体中保持有效，而从 LLM 自身的池中重新选择则将选择机制孤立为原因。每个类别一幅图像时，所选属性的表现比提示调优方法 CoOp 高出 3 分，同时在不到一分钟内完成拟合，而不是 14 小时，并且没有学习到的软提示来模糊决策。由于属性集是由数据选择的，它还可以作为数据集的可读摘要，我们用它来用文字描述分布变化。

View on arXiv Download PDF AI Translation

cs.CV / 21 / 2607.18700

Decoupled Pipeline with Proposal Reranking and Score Fusion for Positive-Unlabeled Marine Species Detection

解耦管道与提案重排序和得分融合用于正无标记海洋物种检测

Brock, Robert James, Krupa, Sebastian Maximilian, Tam, Jason Kahei

Abstract

The FathomNetCLEF 2026 competition combines underwater object detection and fine-grained marine species classification under a positive-unlabeled evaluation setting. The provided training labels are sparse, while the hidden test set is out-of-distribution relative to the training imagery, creating both annotation incompleteness and source-shift challenges. We describe DS@GT ARC's multi-stage system developed for this setting while keeping model training restricted to the data provided by the competition. The final private-leaderboard model uses a frozen Megalodon YOLOv8x detector as a class-agnostic proposal generator, combines global and tiled inference with tile-edge filtering, classifies expanded proposal crops with a LoRA-finetuned DINOv3 ViT-H classifier, and ranks predictions using weighted geometric fusion of detector and classifier confidence. This system placed 12th out of 102 teams. A closely related variant added a locally trained TTN-inspired validity head as a light reranking signal, improving public-leaderboard and proxy-evaluation performance but slightly reducing private-leaderboard performance. Across experiments, the strongest lesson was that train-derived validation and detector-only metrics were not reliable enough for model selection. Instead, we used proxy datasets only for validation and comparison, and combined those signals with leaderboard feedback and targeted ablations. These experiments showed that reserving proposal recall, avoiding over-aggressive filtering, and improving downstream ranking were more effective than fine-tuning the detector or directly training on noisy pseudo-labels. Code: https://github.com/dsgt-arc/fathomnetclef-2026.

Chinese Translation

FathomNetCLEF 2026 竞赛结合了水下物体检测和细粒度海洋物种分类，采用正无标记评估设置。提供的训练标签稀疏，而隐藏的测试集相对于训练图像是分布外的，这造成了注释不完整性和源转移的挑战。我们描述了为此设置开发的 DS@GT ARC 多阶段系统，同时将模型训练限制在竞赛提供的数据上。最终的私有排行榜模型使用冻结的 Megalodon YOLOv8x 检测器作为类无关的提案生成器，结合全局和切片推理以及切片边缘过滤，使用 LoRA 微调的 DINOv3 ViT-H 分类器对扩展的提案作物进行分类，并使用检测器和分类器置信度的加权几何融合对预测进行排序。该系统在102个团队中排名第12。一个密切相关的变体添加了一个本地训练的受 TTN 启发的有效性头作为轻量级重排序信号，改善了公共排行榜和代理评估的表现，但略微降低了私有排行榜的表现。在各项实验中，最强的教训是，基于训练的验证和仅检测器的指标对于模型选择并不可靠。相反，我们仅使用代理数据集进行验证和比较，并将这些信号与排行榜反馈和有针对性的消融相结合。这些实验表明，保留提案召回、避免过于激进的过滤以及改善下游排序比微调检测器或直接在噪声伪标签上训练更为有效。代码： https://github.com/dsgt-arc/fathomnetclef-2026.

View on arXiv Download PDF AI Translation

cs.CV / 22 / 2607.18703

Generative World Renderer at the Speed of Play

以游戏速度生成世界渲染器

Lin, Guixu, Huang, Zheng-Hui, Yang, Siqi, Yang, Ming-Hsuan, Zhang, Kaipeng, Wang, Zhixiang

Abstract

Generative world renderer AlayaRenderer receives structured world states exported from physics engines and synthesizes RGB frames. Unlike models that generate frames from text/control-hints prompts, AlayaRenderer preserves scene structure without altering the underlying world dynamics. This demonstrates an alternative path toward interactive world modeling and user-controllable play. However, the original AlayaRenderer is too computationally expensive for real-time deployment. This technical report introduces AlayaRenderer-Flash, a real-time-oriented generative forward world renderer that pushes AlayaRenderer from 0.56 FPS to 31.54 FPS, reaching the speed of play. AlayaRenderer-Flash reformulates the original renderer as a few-step autoregressive streaming model and introduces lightweight distilled codecs for efficient latent encoding and frame reconstruction. It retains the teacher model's G-buffer and text-prompt interfaces while enabling continuous rendering over input streams of unbounded length. We evaluate AlayaRenderer-Flash on G-buffer streams across content preservation, temporal consistency, cross-window stability, prompt controllability, and runtime efficiency. Our results show that AlayaRenderer-Flash substantially reduces inference cost while preserving the core rendering capabilities of the teacher model. By integrating AlayaRenderer-Flash with a physics engine, we build a fully playable generative world running at 30 FPS.

Chinese Translation

生成世界渲染器 AlayaRenderer 接收来自物理引擎导出的结构化世界状态，并合成 RGB 帧。与从文本/控制提示生成帧的模型不同，AlayaRenderer 保留了场景结构而不改变基础世界动态。这展示了一种面向交互式世界建模和用户可控游戏的替代路径。然而，原始的 AlayaRenderer 在实时部署中计算开销过大。本技术报告介绍了 AlayaRenderer-Flash，一种面向实时的生成前向世界渲染器，将 AlayaRenderer 的帧率从 0.56 FPS 提升至 31.54 FPS，达到了游戏的速度。AlayaRenderer-Flash 将原始渲染器重新构建为一个少步自回归流模型，并引入轻量化的蒸馏编解码器以实现高效的潜在编码和帧重建。它保留了教师模型的 G-buffer 和文本提示接口，同时支持对无限长度输入流的连续渲染。我们在 G-buffer 流上评估 AlayaRenderer-Flash，考察内容保留、时间一致性、跨窗口稳定性、提示可控性和运行时效率。我们的结果表明，AlayaRenderer-Flash 在保留教师模型核心渲染能力的同时，显著降低了推理成本。通过将 AlayaRenderer-Flash 与物理引擎集成，我们构建了一个以 30 FPS 运行的完全可玩生成世界。

View on arXiv Download PDF AI Translation

cs.CV / 23 / 2607.18716

Continual Video-MLLM Adaptation over Evolving Domains

持续视频多模态大语言模型在不断演变领域中的适应

Cheng, Rui, Shi, Meixing, Cai, Yuxiang, Guo, Jingcai, Yin, Jianwei, Chen, Zhi

Abstract

Video multimodal large language models have shown strong capability in video understanding, yet their adaptation to sequentially evolving domains remains underexplored. In real-world deployments, video data often arrives continuously from heterogeneous domains, requiring the model to acquire new domain-specific knowledge without overwriting previously learned capabilities. Existing continual learning methods typically rely on shared adaptation spaces, which can induce severe cross-domain interference and catastrophic forgetting. We propose Distribution-Aware Expert Routing, a parameter-efficient framework for continual Video-MLLM adaptation over evolving domains. DAER maintains domain-isolated lightweight experts while keeping the pretrained Video-MLLM backbone frozen, thereby decoupling domain-specific adaptation from the general multimodal knowledge of the pretrained model. To enable fine-grained specialization, we introduce an intra-domain distribution-aware routing mechanism that matches each input to expert-level prototype reservoirs using MMD. To address the absence of task identities at inference time, we further propose an inter-domain routing mechanism that performs prototype matching in a discriminative subspace for robust domain identification. In addition, we introduce adaptive domain merging to improve parameter scalability and adopt a two-stage optimization strategy to stabilize expert specialization during continual learning. We evaluate DAER by curating a domain-incremental benchmark built from ten VidQA datasets covering diverse visual environments and reasoning demands. Experiments on two strong Video-MLLM backbones show that DAER consistently outperforms prior methods.

Chinese Translation

视频多模态大语言模型在视频理解方面展现出强大的能力，但其在连续演变领域中的适应性仍然未被充分探索。在实际应用中，视频数据通常来自异构领域并持续到达，这要求模型在不覆盖先前学习能力的情况下获取新的领域特定知识。现有的持续学习方法通常依赖于共享适应空间，这可能导致严重的跨领域干扰和灾难性遗忘。我们提出了一种名为分布感知专家路由（Distribution-Aware Expert Routing, DAER）的参数高效框架，用于在不断演变的领域中进行持续的视频多模态大语言模型适应。DAER保持领域隔离的轻量级专家，同时保持预训练的视频多模态大语言模型主干不变，从而将领域特定适应与预训练模型的一般多模态知识解耦。为了实现细粒度的专业化，我们引入了一种领域内分布感知路由机制，该机制使用最大均值差异（MMD）将每个输入匹配到专家级原型库。为了应对推理时缺乏任务标识的问题，我们进一步提出了一种领域间路由机制，该机制在判别子空间中执行原型匹配，以实现稳健的领域识别。此外，我们引入了自适应领域合并以提高参数可扩展性，并采用两阶段优化策略以在持续学习过程中稳定专家专业化。我们通过整理一个基于十个VidQA数据集构建的领域增量基准来评估DAER，该基准涵盖了多样的视觉环境和推理需求。在两个强大的视频多模态大语言模型主干上的实验表明，DAER始终优于先前的方法。

View on arXiv Download PDF AI Translation

cs.CV / 24 / 2607.18747

SkyEV: RGB-Event UAV detection and tracking dataset and baseline

SkyEV：RGB-事件无人机检测与跟踪数据集及基准

Mandula, Jakub, Heusinger, Sebastian, Moosmann, Julian, Vogt, Christian, Magno, Michele

Abstract

Detecting UAVs in air spaces has become increasingly important due to UAVs widespread availability and easy usage. However, due to their small size, they are typically difficult to detect at a sufficient range. For the training of optimized detection algorithms, datasets have been published, covering optical sensing methods ranging from infrared to regular RGB to event-sensor-based. However, these datasets often fail to reflect realistic counter-UAV scenarios, lacking critical factors such as camera ego-motion, extremely small target scales, and diverse lens configurations, and introduce compression artefacts on the frame images. To address this gap, we introduce SkyEV, an open-source dataset featuring highly synchronized uncompressed RGB and event-based data. SkyEV distinguishes itself by capturing complex real-world conditions, including significant camera motion and varied optical setups, which are essential for testing the fundamental trade-off between Field of View and detection range. Furthermore, we provide a unified data loader and establish an experimental baseline using a multi-modal architecture, demonstrating the dataset's efficacy in detecting challenging, small-scale targets.

Chinese Translation

由于无人机的广泛可用性和易用性，在空域中检测无人机变得越来越重要。然而，由于其体积较小，通常难以在足够的距离内进行检测。为了训练优化的检测算法，已经发布了涵盖从红外到常规RGB再到基于事件传感器的光学传感方法的数据集。然而，这些数据集往往未能反映现实的反无人机场景，缺乏关键因素，如相机自运动、极小目标尺度和多样的镜头配置，并且在帧图像上引入了压缩伪影。为了解决这一问题，我们推出了SkyEV，一个开放源代码的数据集，具有高度同步的未压缩RGB和基于事件的数据。SkyEV的独特之处在于捕捉复杂的现实世界条件，包括显著的相机运动和多样的光学设置，这对于测试视场与检测范围之间的基本权衡至关重要。此外，我们提供了统一的数据加载器，并使用多模态架构建立了实验基准，展示了该数据集在检测具有挑战性的微小目标方面的有效性。

View on arXiv Download PDF AI Translation

cs.CV / 25 / 2607.18762

Weakly Supervised Pathology-Informed Representation Learning for PET-Based Content Retrieval of Intra-Tumour Heterogeneity

基于弱监督的病理信息引导的表示学习用于PET基础的肿瘤内异质性内容检索

Vashistha, Rajat, Brosda, Sandra, Aoude, Lauren G., Hannan, Christine Jestin, Lonie, James M., Ng, Jessica, Nathanson, Andrew, Vloedmans, Ellie, Cooper, Caroline, Barbour, Andrew P., Vegh, Viktor

Abstract

We propose a weakly supervised 18FFDG PET representation-learning framework for content based medical image retrieval, using H&E derived information during training while preserving PET-only inference. The proposed method was designed to use H&E derived information during training while maintaining PET only inference. A teacher student training strategy was used to learn the PET tumour derived voxel representations, from which global and hotspot conditioned embeddings were generated along with maps of intra tumour heterogeneity in our oesophegeal cancer test case. A progressive ablation strategy was used to evaluate the contribution of different supervision mechanisms. Retrieval performance was assessed across cross-validation folds using metrics including mean average precision, normalised discounted cumulative gain and mean reciprocal rank. Additional analyses evaluated ablation performance, hotspot faithfulness through perturbation/deletion experiments, prototype-specific PET uptake behaviour and indirect patient level concordance between learned PET prototype classes and selected histomic features. Progressive introduction of pathology informed supervision and hotspot modelling improved PET retrieval performance compared with global PET representations and conventional PET baselines. Across the ablation ladder, PET hotspot conditioned representations consistently provided stronger retrieval than global embeddings, indicating that focusing on informative tumour subregions improved sensitivity to intra tumour heterogeneity. Histopathology concordance further showed that the learned classes were not simply high uptake PET regions; instead, they demonstrated distinct heterogeneity in 18F FDG uptake.

Chinese Translation

我们提出了一种弱监督的18FFDG PET表示学习框架，用于基于内容的医学图像检索，在训练过程中使用H&E衍生的信息，同时保持仅基于PET的推断。该方法旨在在训练过程中利用H&E衍生的信息，同时维持仅基于PET的推断。采用教师-学生训练策略学习PET肿瘤衍生的体素表示，从中生成全球和热点条件嵌入，以及我们食管癌测试案例中的肿瘤内异质性图。采用渐进消融策略评估不同监督机制的贡献。通过交叉验证折叠评估检索性能，使用的指标包括平均精度均值、归一化折扣累积增益和平均互惠排名。额外分析评估了消融性能、通过扰动/删除实验验证的热点可信度、特定原型的PET摄取行为以及学习的PET原型类别与选定组织学特征之间的间接患者水平一致性。与全球PET表示和传统PET基线相比，逐步引入病理信息引导的监督和热点建模改善了PET检索性能。在消融过程中，PET热点条件表示始终提供比全球嵌入更强的检索能力，表明关注信息丰富的肿瘤子区域提高了对肿瘤内异质性的敏感性。组织病理学一致性进一步表明，学习的类别不仅仅是高摄取的PET区域；相反，它们在18F FDG摄取上表现出明显的异质性。

View on arXiv Download PDF AI Translation

cs.CV / 26 / 2607.18763

Posterior Samplings are Missing Modalities Generators for Medical Image Translation

后验采样是医学图像翻译中的缺失模态生成器

Kim, Jonghun

Abstract

Magnetic resonance imaging comes in various modality contrasts that provide complementary anatomical and pathological information. Complete multimodal acquisitions are often unavailable due to time and protocol constraints. This leads to real-world datasets with missing modalities, where conventional medical image translation methods are typically limited to fixed source-target settings or require retraining for each observed source-target pair. We propose a unified framework that formulates missing-modality generation as a linear inverse problem under a joint distribution and solves it via posterior sampling with a flow matching model. By learning a joint prior over the complete modality set, our method can reconstruct arbitrary missing modalities at inference time by guiding the sampling trajectory to enforce measurement consistency with observed modalities. We further mitigate inter-modality error propagation in multi-target generation by adopting a many-to-one sampling strategy. Experiments on BraTS and IXI datasets show that our method achieves the best performance over baselines across most missing-modality scenarios. In downstream tumor segmentation, synthesized images from our method result in higher segmentation performance, indicating better preservation of clinically relevant structures.

Chinese Translation

磁共振成像具有多种模态对比，提供互补的解剖和病理信息。由于时间和协议的限制，完整的多模态采集通常不可用。这导致现实世界的数据集中存在缺失模态，而传统的医学图像翻译方法通常局限于固定的源-目标设置，或需要为每个观察到的源-目标对进行重新训练。我们提出了一个统一框架，将缺失模态生成公式化为联合分布下的线性逆问题，并通过后验采样与流匹配模型进行求解。通过学习完整模态集的联合先验，我们的方法能够在推理时重建任意缺失模态，方法是引导采样轨迹以确保与观察到的模态的一致性。我们进一步通过采用多对一的采样策略来减轻多目标生成中的模态间误差传播。在BraTS和IXI数据集上的实验表明，我们的方法在大多数缺失模态场景中实现了优于基线的最佳性能。在下游肿瘤分割中，我们方法合成的图像导致了更高的分割性能，表明更好地保留了临床相关结构。

View on arXiv Download PDF AI Translation

cs.CV / 27 / 2607.18767

Bounding Boxes to Improve Small Language Model Performance on Vision-Based Grading Tasks

边界框以提高小型语言模型在基于视觉的评分任务中的表现

McGinness, Lachlan

Abstract

The deployment of Small Language Models (SLMs) in educational settings offers significant advantages in terms of privacy, cost, and scalability. However, SLMs often struggle with complex vision-based tasks, such as grading handwritten student exams, due to the high computational cost of processing large images and the visual distractions present on a full page. In this paper, we investigate whether cropping student responses using bounding boxes can improve the accuracy and computational efficiency of SLMs on a short-answer grading task. Using a dataset of scanned handwritten responses from the 2025 Australian Physics Olympiad, we evaluate the performance of several models ranging from 4B to 72B parameters under varying conditions of Chain of Thought (CoT) prompting and image cropping. Our results demonstrate that using bounding boxes significantly improves grading accuracy and reduces computational cost (FLOPs) across models. We conclude that bounding boxes are a crucial pre-processing step for deploying SLMs in large-scale, vision-based educational assessments.

Chinese Translation

在教育环境中部署小型语言模型（SLMs）在隐私、成本和可扩展性方面提供了显著的优势。然而，由于处理大图像的高计算成本以及整页上存在的视觉干扰，SLMs在复杂的基于视觉的任务（如评分手写学生考试）中常常表现不佳。本文研究了使用边界框裁剪学生回答是否能够提高SLMs在短答案评分任务中的准确性和计算效率。我们使用2025年澳大利亚物理奥林匹克竞赛的扫描手写回答数据集，评估了在不同的思维链（Chain of Thought, CoT）提示和图像裁剪条件下，多个参数范围从4B到72B的模型的表现。我们的结果表明，使用边界框显著提高了评分准确性，并降低了计算成本（FLOPs）。我们得出结论，边界框是将SLMs应用于大规模基于视觉的教育评估的重要预处理步骤。

View on arXiv Download PDF AI Translation

cs.CV / 28 / 2607.18768

Cross-Modal UAV Object Tracking: State-Aware Representation Learning and A Unified Benchmark

跨模态无人机目标跟踪：状态感知表示学习与统一基准

Xiao, Yun, Hong, Zhihong, Jin, Jiandong, Li, Chenglong, Tang, Jin, Hussain, Amir

Abstract

Unmanned Aerial Vehicle (UAV) object tracking has emerged as a popular research field with broad practical applications. Modern UAVs are increasingly equipped with both visible light and thermal infrared sensors. However, due to constraints in communication bandwidth, computational resources and power consumption, current systems often activate one modality and switch between modalities to maintain robust tracking in complex scenarios. Such modality switch inevitably leads to significant appearance change and sudden spatial shift, posing great challenges for existing tracking algorithms. To handle this problem, we propose a novel State-Aware Representation Learning Approach called SARLA, which perceives the inconsistent modality states of current frame with template and last frame in the target representations to adapt to the sudden changes in both appearance and position, for robust cross-modal object tracking. In particular, we propose the Modality State Aware Representation Module (MSARM) and Spatial State Aware Representation Module (SSARM). MSARM guides the model to learn appearance correlation, bridging the modality gap, while SSARM models cross-frame spatial correlation to mitigate sudden spatial shift impacts. In addition, we design a spatial shift prediction loss to further handle the effects of spatial variation caused by modality switch. To promote the development of this research field, we establish a large-scale video benchmark called CM-UOT, which consists of 1079 cross-modal sequences with an average video length greater than 621 frames and encompasses over 671K frames in total. Extensive experiments on CM-UOT dataset demonstrate the superior performance of the proposed SARLA against 20 excellent tracking methods. The source code, datasets, and evaluation protocols associated with this work are publicly available at: https://github.com/hongsmile365/sarla-.

Chinese Translation

无人机（UAV）目标跟踪已成为一个热门研究领域，具有广泛的实际应用。现代无人机越来越多地配备可见光和热红外传感器。然而，由于通信带宽、计算资源和功耗的限制，当前系统通常会激活一种模态并在模态之间切换，以在复杂场景中保持稳健的跟踪。这种模态切换不可避免地导致显著的外观变化和突发的空间位移，给现有的跟踪算法带来了巨大挑战。为了解决这个问题，我们提出了一种新颖的状态感知表示学习方法，称为SARLA，它通过与模板和目标表示中的上一帧对比，感知当前帧的不一致模态状态，以适应外观和位置的突发变化，从而实现稳健的跨模态目标跟踪。特别地，我们提出了模态状态感知表示模块（MSARM）和空间状态感知表示模块（SSARM）。MSARM引导模型学习外观相关性，弥合模态差距，而SSARM则建模跨帧空间相关性，以减轻突发空间位移的影响。此外，我们设计了一种空间位移预测损失，以进一步处理模态切换引起的空间变化影响。为了促进该研究领域的发展，我们建立了一个大规模视频基准，称为CM-UOT，包含1079个跨模态序列，平均视频长度超过621帧，总共涵盖超过671K帧。在CM-UOT数据集上的大量实验表明，所提出的SARLA在20种优秀跟踪方法中表现优越。与本研究相关的源代码、数据集和评估协议已公开发布在：https://github.com/hongsmile365/sarla-.

View on arXiv Download PDF AI Translation

cs.CV / 29 / 2607.18773

Privileged Lesion-Context Relational Distillation for Mask-Free Skin Lesion Classification

特权病变上下文关系蒸馏用于无掩膜皮肤病变分类

Rahi, Abu Mukaddim, Hossain, Md Mithun, Joy, Md Zulficar Hasan, Mridha, M. F., Hossen, Md. Jakir

Abstract

Accurate skin lesion classification can benefit from lesion segmentation masks, but requiring masks or an auxiliary segmentation model during inference reduces clinical practicality and increases computational complexity. This work introduces Privileged Lesion-Context Relational Distillation (PLCRD), a teacher-student framework that exploits lesion masks exclusively during training while preserving image-only inference. The privileged teacher jointly analyzes the original dermoscopic image and its mask-guided lesion region to learn lesion-specific and contextual diagnostic representations. An image-only student is then trained through complementary knowledge-transfer mechanisms that convey the teacher's diagnostic distribution, lesion-focused attention, inter-lesion relational geometry, and lesion-context structure. PLCRD decomposes deep representations into lesion and contextual embeddings and transfers their relational organization through inter-lesion similarity alignment, lesion-context affinity matching, separation regularization, and class-aware relational learning. This formulation avoids direct feature matching between heterogeneous teacher and student architectures and enables the student to internalize mask-informed diagnostic structure without accessing masks at deployment. The framework was evaluated on HAM10000 using lesion-disjoint data partitioning and externally validated on ISIC 2018 without retraining. PLCRD achieved a lesion-level macro-F1 of 0.773 +/- 0.018, balanced accuracy of 0.764 +/- 0.023, and macro-AUROC of 0.976 +/- 0.002 on HAM10000, together with a macro-F1 of 0.732 +/- 0.008 on ISIC 2018. The results indicate that privileged lesion annotations can be transformed into transferable relational knowledge, yielding a practical and interpretable approach to mask-free skin lesion classification.

Chinese Translation

准确的皮肤病变分类可以受益于病变分割掩膜，但在推理过程中需要掩膜或辅助分割模型会降低临床实用性并增加计算复杂性。本研究提出了特权病变上下文关系蒸馏（Privileged Lesion-Context Relational Distillation, PLCRD），这是一个教师-学生框架，在训练期间仅利用病变掩膜，同时保留图像仅推理。特权教师共同分析原始皮肤镜图像及其掩膜引导的病变区域，以学习病变特异性和上下文诊断表示。然后，通过互补知识转移机制训练仅基于图像的学生，这些机制传达教师的诊断分布、病变聚焦注意力、病变间关系几何以及病变上下文结构。PLCRD将深层表示分解为病变和上下文嵌入，并通过病变间相似性对齐、病变上下文亲和性匹配、分离正则化和类别感知关系学习转移其关系组织。该公式避免了异构教师和学生架构之间的直接特征匹配，使学生能够在部署时内化掩膜信息的诊断结构而无需访问掩膜。该框架在HAM10000上进行了评估，采用病变不重叠的数据划分，并在ISIC 2018上进行了外部验证，无需重新训练。PLCRD在HAM10000上达到了0.773 +/- 0.018的病变级宏F1、0.764 +/- 0.023的平衡准确率和0.976 +/- 0.002的宏AUROC，同时在ISIC 2018上达到了0.732 +/- 0.008的宏F1。结果表明，特权病变注释可以转化为可转移的关系知识，从而为无掩膜皮肤病变分类提供了一种实用且可解释的方法。

View on arXiv Download PDF AI Translation

cs.CV / 30 / 2607.18779

CGMap: A Geospatially Aware Deep Learning Framework for Crop Gap Mapping Using UAV

CGMap：一种基于地理空间感知的深度学习框架，用于利用无人机进行作物缺口映射

Sharma, Karan, Ranjan, Rajiv, Kumar, Dinesh, Tamaskar, Shashank

Abstract

In India, crop germination is primarily monitored by visual inspection and manual counting, which are prone to errors, despite their crucial role in determining eventual yield potential. This paper highlights a deep learning based pipeline which uses object detection methods and drone imagery to assess and provide a precise count of sugarcane germination in fields. The approch uses a pre-trained AI model to find germinated plant sampling and identify gaps, also known as ``bald spots'', which restricts field productivity. The techniques used here relies on the YOLOV8 architecture, which was trained on a carefully selected dataset of UAV photos taken in various agroclimatic zones of India. Here, we bring upon a novel orientation-normalization technique that uses minimum Spanning Trees (MST) to account for variations in planting geometry, allowing for dependable row and column extraction across a variety of field layouts. By converting detected seedlings into spatial point clouds, emergence gaps can be inferred from the anticipated spacing between plants. A geospatial germination map exported in Well-Known Text (WKT) format is the end result, and it can be easily incorporated into GIS platforms used by sugar mills and agronomists to direct transplant initiatives. Timely interventions based on the insights provided by the algorithm can significantly increase yield, resulting in higher profits. Hence, support proper allocation of resources, avoid wastage, and enhance long-term sustainability.

Chinese Translation

在印度，作物发芽主要通过目视检查和人工计数进行监测，这些方法容易出错，尽管它们在确定最终产量潜力方面至关重要。本文强调了一种基于深度学习的流程，该流程利用物体检测方法和无人机影像来评估并提供甘蔗发芽的精确计数。该方法使用预训练的人工智能模型来寻找发芽植物样本并识别缺口，也称为“秃点”，这些缺口限制了田间生产力。这里使用的技术依赖于YOLOV8架构，该架构在经过精心挑选的印度各农业气候区的无人机照片数据集上进行了训练。我们提出了一种新颖的方向归一化技术，该技术使用最小生成树（MST）来考虑种植几何形状的变化，从而在各种田间布局中实现可靠的行和列提取。通过将检测到的幼苗转换为空间点云，可以从植物之间的预期间距推断出发芽缺口。最终结果是以著名文本（WKT）格式导出的地理空间发芽图，可以轻松地整合到糖厂和农学家用于指导移植计划的地理信息系统（GIS）平台中。基于算法提供的洞察进行及时干预可以显著提高产量，从而带来更高的利润。因此，有助于合理配置资源，避免浪费，并增强长期可持续性。

View on arXiv Download PDF AI Translation

cs.CV / 31 / 2607.18787

Image Editing Models are Numerical Solvers

图像编辑模型是数值求解器

Mizrahi, Ulysse

Abstract

We investigate whether a pretrained generative image-editing model can provide a common interface for numerical simulation. Physical inputs and solutions are rendered as images, while scalar quantities such as material properties, diffusivity, and loading parameters enter through lightweight adapters. Using established numerical and analytic solvers for supervision, we apply the same architecture and training protocol to heterogeneous elliptic equations, forced heat and Burgers evolution, complex Ginzburg-Landau dynamics, two-dimensional Navier-Stokes prediction, potential flow, elasticity, eikonal travel time, phase-field fracture, and entropic optimal transport. The results show that a pretrained image model can represent diverse static and time-dependent physical mappings, including unstable and shock-like behavior, when each task is expressed through a suitable visual encoding. This work is a capability study rather than an attempt to surpass specialized solvers. It also identifies fundamental constraints: image and latent representations complicate numerical range selection and direct enforcement of governing equations or invariants, while a failed Kuramoto-Sivashinsky experiment indicates that representation errors prevent meaningful long-horizon simulation of chaotic systems.

Chinese Translation

我们研究了一个预训练的生成图像编辑模型是否可以为数值模拟提供一个通用接口。物理输入和解被渲染为图像，而材料属性、扩散性和加载参数等标量量通过轻量级适配器输入。利用已建立的数值和解析求解器进行监督，我们将相同的架构和训练协议应用于非均匀椭圆方程、强迫热传导和伯格斯演化、复杂的金兹堡-朗道动力学、二维纳维-斯托克斯预测、势流、弹性、艾肯尔旅行时间、相场断裂和熵最优传输。结果表明，当每个任务通过合适的视觉编码表达时，预训练的图像模型能够表示多样的静态和时间依赖的物理映射，包括不稳定和冲击样行为。这项工作是能力研究，而不是试图超越专业求解器。它还识别了基本约束：图像和潜在表示使数值范围选择和直接强制控制方程或不变量变得复杂，而一次失败的Kuramoto-Sivashinsky实验表明，表示误差阻碍了混沌系统的有意义的长期模拟。

View on arXiv Download PDF AI Translation

cs.CV / 32 / 2607.18789

Moving Alphabet: A Controlled Study of Training Data for Text-to-Video Generation

移动字母：文本到视频生成训练数据的对照研究

Zheng, Amber Yijia, Liu, Lu, Yeh, Raymond A., Yin, Xi

Abstract

Text-to-video generation has advanced significantly over the past five years through scaling of model size, data, and compute. Unlike model architecture, training data is often underexplored. Real-world data curation is complex and non-trivial, involving clip selection from raw videos and captioning to create video-text pairs for learning text-to-video mappings. We study how data distribution and caption quality impact text-to-video models. To enable controlled experiments, we introduce Moving Alphabet, a procedural testbed that renders letters with varying fonts, colors, sizes, and positions, moving in different directions and speeds against a black background. This design allows precise control over data distribution and caption quality by corrupting ground-truth metadata. Our experiments yield three findings: a) a diverse and balanced distribution of video content and duration is critical for generalization; b) caption quality significantly affects both model performance and training efficiency, suggesting that text-to-video models are bounded by video understanding capabilities; and c) classifier-free guidance and fine-tuning on high-quality data provide partial recovery from models trained on corrupted captions, but cannot fully compensate for poor pre-training data. We believe these insights can inform the development of large-scale text-to-video models, and we advocate for greater attention to the science of pre-training data.

Chinese Translation

在过去五年中，文本到视频生成技术通过模型规模、数据和计算能力的扩展取得了显著进展。与模型架构不同，训练数据往往被忽视。现实世界数据的整理复杂且非平凡，涉及从原始视频中选择片段并进行字幕处理，以创建用于学习文本到视频映射的视频-文本对。我们研究了数据分布和字幕质量如何影响文本到视频模型。为了进行可控实验，我们引入了移动字母（Moving Alphabet），这是一个程序化测试平台，能够渲染具有不同字体、颜色、大小和位置的字母，这些字母在黑色背景下以不同方向和速度移动。该设计通过破坏真实元数据，允许对数据分布和字幕质量进行精确控制。我们的实验得出了三项发现：a) 多样化和平衡的视频内容和时长分布对模型的泛化能力至关重要；b) 字幕质量显著影响模型性能和训练效率，表明文本到视频模型受到视频理解能力的限制；c) 无分类器引导和在高质量数据上的微调能够部分恢复在损坏字幕上训练的模型，但无法完全弥补低质量预训练数据的不足。我们相信这些见解可以为大规模文本到视频模型的发展提供指导，并倡导对预训练数据科学的更多关注。

View on arXiv Download PDF AI Translation

cs.CV / 33 / 2607.18791

STS-NET: Spatio-Temporal Stress Network for Self-Supervised Crop Stress Detection using Satellite Image Time Series

STS-NET：基于时空压力网络的自监督作物压力检测方法，利用卫星图像时间序列

Dalal, Pradeep, Ranjan, Rajiv, Ghildiyal, Sushil, Tamaskar, Shashank, Goel, Neeraj

Abstract

Early and accurate detection of crop stress is essential to improve agricultural productivity and ensure global food security. However, collecting a large labeled crop stress dataset is a challenging task. To address this challenge, we introduce a novel spatial-temporal stress network (STS-NET), built on a self-supervised 3D-convolutional autoencoder (3D-CAE), designed to utilize Satellite Image Time Series (SITS) data for crop stress detection. STS-NET exploits four vegetation indices: Normalized Difference Vegetation Index (NDVI), Normalized Difference Vegetation Index (GNDVI), Red-Edge Chlorophyll Index (RECI) and Normalized Difference Red-Edge Index (NDRE) obtained from high resolution Planetscope imagery to capture spatiotemporal stress patterns. The model is trained on our BSPT (Barnala Spatial-Temporal) dataset and evaluated on a real-world sugarcane dataset collected over a year from a 2.5-acre test plot located in Lakhimpur-Kheri (LK) district in Uttar Pradesh in India. STS-NET achieved a precision of 97. 98\% for water stress, 85.08\% for nitrogen stress, and 83.47\% for combined stress. The results demonstrate the potential of STS-NET in effectively detecting stress in sugarcane crops with minimal reliance on labeled data. Furthermore, STS-NET can serve as a robust feature extractor for simpler models.

Chinese Translation

早期且准确地检测作物压力对于提高农业生产力和确保全球粮食安全至关重要。然而，收集大量标记的作物压力数据集是一项具有挑战性的任务。为了解决这一挑战，我们提出了一种新颖的时空压力网络（STS-NET），该网络基于自监督的三维卷积自编码器（3D-CAE），旨在利用卫星图像时间序列（SITS）数据进行作物压力检测。STS-NET利用四种植被指数：归一化差异植被指数（NDVI）、归一化差异植被指数（GNDVI）、红边叶绿素指数（RECI）和归一化差异红边指数（NDRE），这些指数来自高分辨率的Planetscope影像，以捕捉时空压力模式。该模型在我们的BSPT（Barnala时空）数据集上进行训练，并在印度北方邦Lakhimpur-Kheri区的一块2.5英亩测试地块上收集的真实甘蔗数据集上进行评估。STS-NET在水分压力检测中达到了97.98%的精度，在氮素压力检测中达到了85.08%的精度，在综合压力检测中达到了83.47%的精度。结果表明，STS-NET在有效检测甘蔗作物压力方面具有潜力，并且对标记数据的依赖较小。此外，STS-NET还可以作为更简单模型的强大特征提取器。

View on arXiv Download PDF AI Translation

cs.CV / 34 / 2607.18798

UVFaceFusion: Fast Multi-view Topologically Consistent Face Reconstruction in the Wild via UV-space Neural Fusion

UVFaceFusion：通过UV空间神经融合实现快速多视角拓扑一致的人脸重建

Ming, Xin, Han, Yuxuan, Yong, Junhai, Xu, Feng

Abstract

Reconstructing high-fidelity facial geometry with an assigned topology is essential for digital avatar creation and animation, yet existing automated methods often trade off geometric fidelity and in-the-wild generalization. We present UVFaceFusion, a feed-forward framework for multi-view, fixed-topology face reconstruction from daily images. Our key idea is to replace heuristic topological optimization with learnable neural fusion in a canonical UV space. Given multi-view images, we first obtain dense point maps and facial UV correspondences of each view using VGGT and Pixel3DMM, respectively. Then, the view-specific point maps are lifted into the canonical UV domain and fused with a novel mask-aware neural fusion network. The network predicts a complete UV-space point map, from which a fixed-topology mesh is directly sampled. Although trained only on Ava-256, UVFaceFusion generalizes well to multiple public benchmarks and in-the-wild captures, benefiting from its canonical UV-space geometry-to-geometry fusion that reduces dependence on dataset-specific appearance and capture conditions. Experiments on various benchmarks show that UVFaceFusion achieves state-of-the-art reconstruction accuracy while reconstructing a mesh from 16 input views in less than 3 seconds on a single RTX 4090. Code is available at https://github.com/grignarder/UVFaceFusion.

Chinese Translation

重建具有指定拓扑结构的高保真面部几何形状对于数字化身的创建和动画至关重要，但现有的自动化方法往往在几何保真度和野外泛化之间进行权衡。我们提出了UVFaceFusion，这是一种用于从日常图像中进行多视角、固定拓扑人脸重建的前馈框架。我们的关键思想是用可学习的神经融合替代启发式拓扑优化，采用标准化的UV空间。给定多视角图像，我们首先使用VGGT和Pixel3DMM分别获得每个视角的稠密点图和面部UV对应关系。然后，将视角特定的点图提升到标准化的UV域，并与一种新颖的掩膜感知神经融合网络进行融合。该网络预测一个完整的UV空间点图，从中直接采样出固定拓扑网格。尽管仅在Ava-256上进行训练，UVFaceFusion在多个公共基准和野外捕获中表现出良好的泛化能力，得益于其标准化UV空间几何到几何的融合，减少了对数据集特定外观和捕获条件的依赖。在各种基准上的实验表明，UVFaceFusion在从16个输入视角重建网格时，能够在单个RTX 4090上以不到3秒的时间实现最先进的重建精度。代码可在https://github.com/grignarder/UVFaceFusion获取。

View on arXiv Download PDF AI Translation

cs.CV / 35 / 2607.18801

ZeroSplat: Generalized Referring Segmentation in 3D Gaussian Splatting

ZeroSplat：3D高斯点云中的广义指向分割

Ding, Jiayu, Song, Meilu, Zhang, Xiaoyi, Jin, Hongbo, Jin, Yichen, Si, Xiangtian

Abstract

Recent advancements in 3D Gaussian Splatting (3DGS) have enabled language-guided scene understanding. However, existing Referring 3D Gaussian Splatting (R3DGS) methods are fundamentally restricted to single-target queries. To reflect the ambiguity of real-world instructions, we introduce the Generalized Referring 3D Gaussian Splatting Segmentation (GR3DGS) task, which requires dynamically segmenting an arbitrary number of targets (0, 1, or $N$). To facilitate comprehensive evaluation of this new task, we construct two new benchmarks: GR-LERF and GR-ScanNet. Crucially, existing R3DGS paradigms exhibit fundamental technical bottlenecks that severely limit their performance on the GR3DGS task: they lack intrinsic 3D point-level understanding by operating merely on 2D rendered pixels, and they incur prohibitive computational overhead by requiring per-scene optimization to embed heavy semantic features. To dismantle these bottlenecks, we propose ZeroSplat, a novel training-free and zero-feature framework. ZeroSplat lifts 2D Vision-Language Model (VLM) priors into 3D space through robust multi-view geometric constraints. This strategy enables intrinsic point-level understanding without incurring any additional feature storage. Extensive experiments demonstrate that ZeroSplat significantly outperforms state-of-the-art methods across generalized and single-target scenarios while maintaining exceptional efficiency. Project Page: https://inkmind-ai.github.io/ZeroSplat

Chinese Translation

近年来，3D高斯点云（3D Gaussian Splatting, 3DGS）的进展使得基于语言的场景理解成为可能。然而，现有的指向性3D高斯点云（Referring 3D Gaussian Splatting, R3DGS）方法在本质上仅限于单一目标查询。为了反映现实世界指令的模糊性，我们提出了广义指向性3D高斯点云分割（Generalized Referring 3D Gaussian Splatting Segmentation, GR3DGS）任务，该任务要求动态分割任意数量的目标（0、1或$N$）。为了全面评估这一新任务，我们构建了两个新的基准：GR-LERF和GR-ScanNet。至关重要的是，现有的R3DGS范式存在根本性的技术瓶颈，严重限制了它们在GR3DGS任务上的表现：它们仅在2D渲染像素上操作，缺乏内在的3D点级理解，并且由于需要对每个场景进行优化以嵌入重语义特征，导致计算开销过大。为了解决这些瓶颈，我们提出了ZeroSplat，这是一种新颖的无训练和零特征框架。ZeroSplat通过稳健的多视角几何约束将2D视觉-语言模型（Vision-Language Model, VLM）先验提升到3D空间。这一策略使得在不增加任何额外特征存储的情况下实现内在的点级理解。大量实验表明，ZeroSplat在广义和单目标场景中显著超越了最先进的方法，同时保持了卓越的效率。项目页面：https://inkmind-ai.github.io/ZeroSplat

View on arXiv Download PDF AI Translation

cs.CV / 36 / 2607.18819

In-Context Learning for Wound Classification with Small Multimodal Language Models

小型多模态语言模型的伤口分类上下文学习

Martvel, George, Gustafsson, Oskar, Pavia, John, Ahlberg, Ernst

Abstract

Wound image classification is often treated as a task-specific supervised learning problem, requiring substantial amounts of manually labelled data and retraining when the label space or deployment setting changes. This study evaluated whether small multimodal language models (SMLMs) can provide a training-free alternative for wound classification through retrieval-based in-context learning (ICL). Experiments used two public wound-image datasets: the Kaggle wound dataset (1469 images, 10 classes) and the Medetec dataset (560 images, 9 classes). Eleven SMLMs from the Qwen 3.5, Ministral 3, and Gemma 4 families were evaluated under zero-shot prompting and few-shot prompting with random support examples, embedding-based k-nearest-neighbour (kNN) retrieval, and kNN retrieval followed by maximal marginal relevance reranking (MMR). Retrieval-only weighted-kNN controls, support-set reduction experiments, and support-context size sweeps were used to assess the effects of retrieval, model scale, and prompt length. Query-conditioned ICL consistently outperformed zero-shot and random few-shot prompting. On the Kaggle dataset, the best result was achieved by Qwen 3.5 27B with kNN+MMR, reaching 0.872 accuracy and 0.871 F1 score. On Medetec, Qwen 3.5 27B with kNN+MMR reached 0.678 accuracy and 0.670 F1. Larger models exceeded matched weighted-kNN controls, indicating use of retrieved examples beyond nearest-neighbour voting. Retrieval-based ICL degraded modestly under support-set reduction, and most gains saturated with 8-10 support images. Retrieval-based ICL allows SMLMs to perform adaptable wound image classification without task-specific retraining. Compact retrieved contexts may support practical and privacy-conscious deployment, although performance remains dependent on model scale, retrieval strategy, and dataset difficulty.

Chinese Translation

伤口图像分类通常被视为一个特定任务的监督学习问题，要求大量手动标注的数据，并在标签空间或部署设置变化时进行重新训练。本研究评估了小型多模态语言模型（SMLMs）是否可以通过基于检索的上下文学习（ICL）提供一种无训练的伤口分类替代方案。实验使用了两个公共伤口图像数据集：Kaggle伤口数据集（1469张图像，10个类别）和Medetec数据集（560张图像，9个类别）。评估了来自Qwen 3.5、Ministral 3和Gemma 4系列的11个SMLMs，在零-shot提示和带有随机支持示例的few-shot提示下，采用基于嵌入的k近邻（kNN）检索，以及kNN检索后进行最大边际相关性重排序（MMR）。使用仅基于检索的加权kNN控制、支持集缩减实验和支持上下文大小的变化来评估检索、模型规模和提示长度的影响。查询条件的ICL始终优于零-shot和随机few-shot提示。在Kaggle数据集中，最佳结果由Qwen 3.5 27B与kNN+MMR实现，达到0.872的准确率和0.871的F1分数。在Medetec中，Qwen 3.5 27B与kNN+MMR达到0.678的准确率和0.670的F1分数。较大的模型超越了匹配的加权kNN控制，表明检索的示例超出了最近邻投票的使用。在支持集缩减下，基于检索的ICL表现出适度的下降，大多数增益在8-10个支持图像时饱和。基于检索的ICL使得SMLMs能够在不进行特定任务重新训练的情况下执行可适应的伤口图像分类。紧凑的检索上下文可能支持实际和隐私意识的部署，尽管性能仍然依赖于模型规模、检索策略和数据集的难度。

View on arXiv Download PDF AI Translation

cs.CV / 37 / 2607.18827

Open-Vocabulary Gaze Object Prediction: Benchmark and Method

开放词汇注视物体预测：基准与方法

Wang, Binglu, Niu, Sensen, Chen, Ying, Guo, Guangyu

Abstract

Gaze Object Prediction (GOP) aims to localize and recognize the objects humans attend to, a task crucial for understanding human-centric interactions. However, existing methods are typically trained under a closed-vocabulary paradigm with a fixed label space and evaluated on scene-specific datasets, limiting their applicability to real-world scenarios where gaze targets often follow a long-tail distribution or belong to unseen categories. To address this gap, we introduce Diverse Scenes for Gaze object prediction (DiSG), a benchmark containing 86 in-the-wild categories that facilitates the evaluation of Open-Vocabulary GOP (OVGOP). Building on DiSG, we propose a framework that leverages text-driven object discovery to localize potential gaze candidates, with a gaze-guided selection module to pinpoint the intended target from the candidate objects. Furthermore, to better capture semantic knowledge across diverse in-the-wild categories, we introduce Gradient-Informed Selection Tuning (GIST) to selectively update parameters most relevant to a given class vocabulary. Extensive experiments demonstrate that our proposed model performs effectively in open-vocabulary settings and also outperforms existing methods in the conventional closed-vocabulary setting. The benchmark and code is available at https://github.com/sensniu/ovgop.

Chinese Translation

注视物体预测（Gaze Object Prediction, GOP）旨在定位和识别人类关注的物体，这一任务对于理解以人为中心的交互至关重要。然而，现有方法通常在封闭词汇范式下进行训练，具有固定的标签空间，并在特定场景的数据集上进行评估，这限制了它们在现实世界场景中的适用性，因为注视目标往往遵循长尾分布或属于未见类别。为了解决这一问题，我们引入了多样化场景用于注视物体预测（Diverse Scenes for Gaze object prediction, DiSG），这是一个包含86个真实场景类别的基准，便于评估开放词汇注视物体预测（Open-Vocabulary GOP, OVGOP）。基于DiSG，我们提出了一个框架，利用文本驱动的物体发现来定位潜在的注视候选物体，并通过注视引导选择模块从候选物体中确定意图目标。此外，为了更好地捕捉跨多样化真实场景类别的语义知识，我们引入了梯度信息选择调优（Gradient-Informed Selection Tuning, GIST），以选择性地更新与给定类别词汇最相关的参数。大量实验表明，我们提出的模型在开放词汇设置中表现出色，并且在传统的封闭词汇设置中也优于现有方法。基准和代码可在 https://github.com/sensniu/ovgop 获取。

View on arXiv Download PDF AI Translation

cs.CV / 38 / 2607.18850

OPD-IAD: From Language Judgment to Industrial Anomaly Detection via On-Policy Self-Distillation

OPD-IAD：通过在线自蒸馏从语言判断到工业异常检测

Chen, Shuimu, Jin, Jing, Su, Nan, Xu, Hongbo, Cheng, Zebang, Yang, Wenming, Ma, Fei, Wang, Guijin

Abstract

Large vision-language models (LVLMs) have recently shown strong potential for industrial anomaly detection (IAD) by providing image-level anomaly judgments and interpretable defect reasoning. However, current LVLM-based IAD methods still struggle to produce precise pixel-level anomaly maps from generated language judgments. We aim to achieve precise pixel-level localization while using language as guidance rather than letting it dominate the visual response. Specifically, we propose \textbf{OPD-IAD}, an evidence-privileged dense on-policy self-distillation framework for LVLM-based IAD. OPD-IAD distills privileged defect evidence onto the model's own on-policy judgment trajectory, enabling the final generated judgment to be learned under dense supervision rather than treated only as a textual answer. The resulting judgment serves as a semantic condition for dense anomaly perception. To turn this condition into dense visual evidence, we introduce \textbf{Language-guided Visual Anchoring}, which uses a judgment reforward to re-encode the image and question under the final-judgment condition into semantic anchors and contrasts them with dense visual features through a contrastive heatmap head to generate anomaly maps. The language judgment therefore provides compact semantic guidance, while dense visual features remain the basis for pixel-level scoring, allowing language to guide anomaly localization without letting language quality directly dictate the pixel-level response. Extensive experiments show that OPD-IAD achieves the best overall performance among LVLM-based IAD methods, leading on most image-level, pixel-level, and QA metrics.

Chinese Translation

大型视觉-语言模型（LVLMs）最近在工业异常检测（IAD）方面显示出强大的潜力，能够提供图像级异常判断和可解释的缺陷推理。然而，目前基于LVLM的IAD方法仍然难以从生成的语言判断中产生精确的像素级异常图。我们的目标是实现精确的像素级定位，同时使用语言作为指导，而不是让其主导视觉响应。具体而言，我们提出了 extbf{OPD-IAD}，一种针对基于LVLM的IAD的证据特权密集在线自蒸馏框架。OPD-IAD将特权缺陷证据蒸馏到模型自身的在线判断轨迹上，使最终生成的判断能够在密集监督下学习，而不仅仅被视为文本答案。最终的判断作为密集异常感知的语义条件。为了将这一条件转化为密集的视觉证据，我们引入了 extbf{语言引导的视觉锚定}，该方法使用判断重新转发将图像和问题在最终判断条件下重新编码为语义锚，并通过对比热图头与密集视觉特征进行对比，从而生成异常图。因此，语言判断提供了紧凑的语义指导，而密集视觉特征仍然是像素级评分的基础，使得语言能够引导异常定位，而不让语言质量直接决定像素级响应。大量实验表明，OPD-IAD在基于LVLM的IAD方法中实现了最佳的整体性能，在大多数图像级、像素级和问答指标上领先。

View on arXiv Download PDF AI Translation

cs.CV / 39 / 2607.18856

Think Sparse, Predict Dense: Continuous Thought Machines for Image Super-Resolution

思维稀疏，预测密集：用于图像超分辨率的连续思维机器

Shi, Zekai

Abstract

Continuous Thought Machines introduce an internal temporal dimension in which neuron-level histories and synchronization-derived representations evolve over a sequence of thought ticks. Extending this mechanism to dense visual prediction is non-trivial, because tasks such as image super-resolution require spatial evidence to remain available at every output location rather than being compressed into a single global representation. In the proposed window-level use of CTM, the thought dynamics produce a compact summary representation for each local window. DQ-CTM transforms this compact thought representation into window-aligned dense queries through a structured low-rank, parameter-efficient compact-to-dense query mechanism. Each position within a window receives its own query, while shared thought dynamics progressively refine the dense representation across ticks. In its super-resolution instantiation, termed ThinkSR, encoded feature maps are partitioned into local visual windows without token pooling, restored to the original feature field after shared refinement, and decoded into a high-resolution image. Preliminary experiments under a fixed four-tick training horizon reveal a progressive reconstruction trajectory. PSNR-Y increases from 28.1045 dB at $T=0$ to 30.2817 dB at $T=4$, while PSNR-RGB increases from 26.6271 dB to 28.7781 dB and the mean $\ell_1$ error decreases from 0.034602 to 0.023545. All 100 evaluated images improve from $T=1$ to $T=4$. These initial results establish the feasibility of sparse latent thought for dense spatial reconstruction and motivate broader continuous-thought architectures for dense vision.

Chinese Translation

连续思维机器引入了一个内部时间维度，在这个维度中，神经元级历史和基于同步的表征在一系列思维时刻中演变。将这一机制扩展到密集视觉预测并非易事，因为图像超分辨率等任务要求空间证据在每个输出位置保持可用，而不是压缩为单一的全局表征。在所提出的窗口级 CTM（Continuous Thought Machines）使用中，思维动态为每个局部窗口生成紧凑的摘要表征。DQ-CTM（Dynamic Query Continuous Thought Machines）通过结构化的低秩、参数高效的紧凑到密集查询机制，将这一紧凑的思维表征转化为与窗口对齐的密集查询。窗口内的每个位置接收其自身的查询，而共享的思维动态逐步在各个时刻精炼密集表征。在其超分辨率实例化中，称为 ThinkSR，编码的特征图被划分为局部视觉窗口，未进行标记池化，经过共享精炼后恢复到原始特征场，并解码为高分辨率图像。在固定的四个时刻训练范围内的初步实验显示出逐步重建轨迹。PSNR-Y 从 $T=0$ 时的 28.1045 dB 增加到 $T=4$ 时的 30.2817 dB，而 PSNR-RGB 从 26.6271 dB 增加到 28.7781 dB，平均 $ ext{l}_1$ 误差从 0.034602 降低到 0.023545。所有 100 张评估图像在 $T=1$ 到 $T=4$ 之间均有所改善。这些初步结果确立了稀疏潜在思维用于密集空间重建的可行性，并激励了更广泛的连续思维架构用于密集视觉。

View on arXiv Download PDF AI Translation

cs.CV / 40 / 2607.18863

Reliability-Aware 3D Geometric Injection for Universal Person Re-identification

面向可靠性的3D几何注入用于通用行人重识别

Su, Bohan, Wang, Jiashuo, Liu, Fangyi, Ye, Mang

Abstract

Universal person re-identification (ReID) aims to retrieve pedestrian identities across diverse real-world scenarios, including severe occlusions, clothing changes, and cross-modality shifts, within a unified model. However, existing 2D representations fundamentally struggle with spatial ambiguities due to a lack of depth and topological awareness, while naively introducing monocular 3D priors often causes severe negative transfer due to geometric estimation noise under extreme visual degradation. To safely harness the clothing-invariant and canonical structural properties of 3D geometry, we propose UniGeo, a Universal Monocular 3D-Enhanced ReID framework driven by a Consistency-Aware Reliability Gate and Dual-Stream Residual Fusion. Specifically, the processing of 3D information is strategically decoupled into geometric extraction and dynamic utilization. To provide pure structural compensation, we project monocular 3D parameters into kinematic joint representations, explicitly capturing instance-level geometric topology to resolve appearance-based ambiguities. To robustly incorporate these cues without perturbing the reliable 2D feature space, we isolate the 3D prior as a late-stage structural residual; modulated by the consistency-aware gate, this mechanism adaptively filters geometric noise and enables controlled fallback to the pure 2D baseline. Extensive experiments show that our method improves challenging, structure-sensitive scenarios while preserving competitive performance on clean domains. Code is available at https://github.com/BohanSu/UniGeo.

Chinese Translation

通用行人重识别（ReID）旨在在统一模型中检索在多种现实场景下的行人身份，包括严重遮挡、服装变化和跨模态转变。然而，现有的2D表示由于缺乏深度和拓扑意识，根本上难以处理空间模糊，而简单引入单目3D先验在极端视觉退化下往往会因几何估计噪声而导致严重的负迁移。为了安全地利用3D几何的服装不变性和规范结构特性，我们提出了UniGeo，一个由一致性感知可靠性门和双流残差融合驱动的通用单目3D增强ReID框架。具体而言，3D信息的处理被战略性地解耦为几何提取和动态利用。为了提供纯粹的结构补偿，我们将单目3D参数投影到运动关节表示中，明确捕捉实例级几何拓扑以解决基于外观的模糊。为了在不干扰可靠的2D特征空间的情况下稳健地整合这些线索，我们将3D先验隔离为后期结构残差；通过一致性感知门调节，该机制自适应地过滤几何噪声，并实现对纯2D基线的受控回退。大量实验表明，我们的方法在具有挑战性的、对结构敏感的场景中表现出改善，同时在干净领域保持竞争力的性能。代码可在 https://github.com/BohanSu/UniGeo 获取。

View on arXiv Download PDF AI Translation

cs.CV / 41 / 2607.18875

Wave2Body: Rethinking mmWave Human Pose Estimation as Radar-to-Body Token Translation

Wave2Body：重新思考毫米波人类姿态估计为雷达到身体的标记翻译

Liang, Bo, Gong, Chen, Gao, Wei, Xu, Chenren

Abstract

Millimeter-wave (mmWave) radar enables privacy-friendly human sensing, but its sparse point clouds are physical measurements of view-dependent electromagnetic reflections and only indirectly characterize body articulation. Recovering a complete 3D pose from such partial, geometry-dependent observations is therefore under-constrained. Existing methods directly regress joint coordinates from paired radar-pose data, relying on the same limited paired supervision to learn radar perception, human-body structure, and their alignment. This coupling can encourage dataset-specific shortcuts under ambiguous radar observations. We propose Wave2Body, a radar-to-body token translation framework that decouples these learning targets using a self-supervised mmWave tokenizer, a pretrained compositional body tokenizer that defines the output space, and a lightweight translator between them. Experiments on M4Human and mmBody show that Wave2Body achieves stronger cross-domain generalization than previous methods while incurring much lower computational costs for training and inference. All the code and experiment results are publicly available at https://github.com/Galaxywalk/Wave2Body.

Chinese Translation

毫米波（mmWave）雷达实现了隐私友好的人体感知，但其稀疏的点云是视角依赖的电磁反射的物理测量，仅间接表征身体的关节运动。因此，从这种部分的、几何依赖的观测中恢复完整的三维姿态是欠约束的。现有方法直接从配对的雷达-姿态数据中回归关节坐标，依赖于相同的有限配对监督来学习雷达感知、人体结构及其对齐。这种耦合可能在模糊的雷达观测下促使数据集特定的捷径。我们提出了Wave2Body，一个雷达到身体的标记翻译框架，通过自监督的毫米波标记器、一个预训练的组合身体标记器（定义输出空间）和它们之间的轻量级翻译器来解耦这些学习目标。在M4Human和mmBody上的实验表明，Wave2Body在跨领域泛化能力上优于以往方法，同时在训练和推理时的计算成本显著降低。所有代码和实验结果均可在https://github.com/Galaxywalk/Wave2Body获取。

View on arXiv Download PDF AI Translation

cs.CV / 42 / 2607.18882

Local Label-Informed Feature Transfer for Generating Ground-Truth Medical Images: A Comparison of GAN- and Diffusion-Based Approaches

基于局部标签信息的特征转移生成真实医学图像：GAN与扩散方法的比较

Wilming, Rick, Ozseker, Irem, Cornils, Luca Matteo, Boubekki, Ahcène, Clark, Benedict, Panknin, Danny, Haufe, Stefan

Abstract

Validating Explainable Artificial Intelligence (XAI) methods in medical imaging requires ground-truth data with known locations of informative features. However, current approaches rely on expert annotations, which are prone to labeling errors, or on hand-crafted artificial perturbations superimposed onto healthy images to mimic lesions or malignant features, which lack clinical realism. We present Local Label-Informed Feature Transfer (LLIFT), a framework for generating semi-synthetic brain magnetic resonance images with realistic lesions placed in user-controlled regions, which does not require pixel-level lesion annotations during training. We implement LLIFT with two generative paradigms: LLIFT-GAN, a custom GAN that learns pathological features from binary class labels alone, and LLIFT-DM, a diffusion-based inpainting pipeline conditioned on bounding-box masks via ControlNet. Both approaches are evaluated on brain magnetic resonance imaging data derived from the Human Connectome Project. In evaluations, both achieve Fr\'echet Inception Distance scores, with respect to the real pathological distribution, that are comparable to the inter-class reference between healthy and pathological images in the given dataset. Furthermore, qualitative inspection confirms the realism of lesion structures. The resulting benchmark datasets provide spatially controlled ground truth data for evaluating XAI methods in medical imaging.

Chinese Translation

验证医学影像中的可解释人工智能（XAI）方法需要具有已知信息特征位置的真实数据。然而，目前的方法依赖于专家注释，这容易出现标注错误，或者依赖于叠加在健康图像上的手工制作的人工扰动，以模拟病变或恶性特征，这缺乏临床现实性。我们提出了局部标签信息特征转移（Local Label-Informed Feature Transfer, LLIFT），这是一个生成半合成脑部磁共振图像的框架，能够在用户控制的区域内放置真实的病变，且在训练过程中不需要像素级的病变注释。我们使用两种生成范式实现LLIFT：LLIFT-GAN，一种自定义的生成对抗网络（GAN），仅通过二元类别标签学习病理特征；以及LLIFT-DM，一种基于扩散的修复管道，通过ControlNet在边界框掩模的条件下进行。两种方法均在来自人类连接组计划的脑部磁共振成像数据上进行评估。在评估中，两者在与真实病理分布相关的Fréchet Inception Distance（FID）得分上，与给定数据集中健康与病理图像之间的类间参考相当。此外，定性检查确认了病变结构的真实感。生成的基准数据集为评估医学影像中的XAI方法提供了空间控制的真实数据。

View on arXiv Download PDF AI Translation

cs.CV / 43 / 2607.18907

SynGallery: A Synthetic Gallery of Real Paintings for Instance-Level Artwork Recognition

SynGallery：用于实例级艺术作品识别的真实绘画合成画廊

Bartkowiak, Patryk, Markil, Jakub, Kotrys, Bartosz, Michels, Dominik, Pirk, Sören, Palubicki, Wojtek

Abstract

Instance-level artwork recognition requires matching a handheld visitor photograph to a specific work in a large museum collection. This is challenging because painting datasets typically provide clean catalog images for training, while test queries are captured under oblique viewpoints, gallery lighting, reflections, frames, and other scene-level variations. We present SynGallery, a synthetic gallery dataset for artwork retrieval that addresses this gap without collecting additional real photographs. Starting from catalog images of real paintings, we place each artwork into a procedurally generated 3D gallery scene and render it from multiple viewpoints under varied geometric and appearance conditions, while preserving the exact identity of the original work. The resulting dataset contains 24,490 rendered views of 4,898 paintings from the Met benchmark. We show that these synthetic views provide a stronger training signal than the corresponding studio photographs. At the same number of training data points, training only on SynGallery improves art painting recognition from 67.18 to 73.47 GAP$^-$. When added to the full Met training set, SynGallery improves the published benchmark protocol from 35.97 to 38.48 GAP. Ablation experiments show that the gain comes primarily from geometric viewpoint variation rather than photographic realism: blur, sensor noise, and image compression consistently reduce performance.

Chinese Translation

实例级艺术作品识别需要将手持访客照片与大型博物馆藏品中的特定作品进行匹配。这一任务具有挑战性，因为绘画数据集通常提供干净的目录图像用于训练，而测试查询则是在倾斜视角、画廊照明、反射、框架以及其他场景级变化下捕获的。我们提出了SynGallery，一个用于艺术作品检索的合成画廊数据集，旨在填补这一空白，而无需收集额外的真实照片。我们从真实绘画的目录图像出发，将每件艺术作品放置在程序生成的3D画廊场景中，并在不同的几何和外观条件下从多个视角进行渲染，同时保留原始作品的确切身份。最终生成的数据集包含来自Met基准的4,898幅绘画的24,490个渲染视图。我们展示了这些合成视图提供的训练信号比相应的工作室照片更强。在相同数量的训练数据点下，仅在SynGallery上进行训练将艺术绘画识别的准确率从67.18提高到73.47 GAP$^-$. 当与完整的Met训练集结合时，SynGallery将已发布的基准协议从35.97提高到38.48 GAP。消融实验表明，性能提升主要来自几何视角变化，而非摄影现实主义：模糊、传感器噪声和图像压缩始终会降低性能。

View on arXiv Download PDF AI Translation

cs.CV / 44 / 2607.18917

TAP-RAG: Task-Aware Policy Control for Long-Document Multimodal Question Answering

TAP-RAG：用于长文档多模态问答的任务感知策略控制

Ji, Zhong, Jin, Keqi, Zhang, Yan, Li, Jiasheng

Abstract

Long-document multimodal question answering requires more than retrieving relevant chunks from a large document. Different queries require different evidence behavior. Existing multimodal RAG systems improve evidence access through text chunks, page images, graph links, or heterogeneous document elements, but they often apply a largely query-agnostic evidence-use strategy. We present TAP-RAG, a task-aware policy-controlled RAG framework for long-document multimodal QA. TAP-RAG contains a main controller, the Task-Aware Policy Controller (TAPC), and two policy-guided evidence executors: Task-Aware Query-Guided Flow Diffusion (TA-QFD) and Task-Aware Visual Enhancement (TAVE). For each query, TAPC predicts the task prior, estimates visual/local/global evidence signals, and produces an executable policy. TA-QFD then expands textual and structural evidence over the multimodal document graph, while TAVE selectively inspects page images when visual or layout evidence is needed. A guarded synthesis stage fuses text, visual, and structural evidence and abstains when support is insufficient. On DocBench and MMLongBench-Doc, TAP-RAG achieves the best overall accuracy among the compared systems, improving over a matched multimodal-RAG baseline by +9.1 points (61.1 to 70.2) and +4.5 points (42.2 to 46.7), respectively.

Chinese Translation

长文档多模态问答不仅仅需要从大文档中检索相关片段。不同的查询需要不同的证据行为。现有的多模态 RAG 系统通过文本片段、页面图像、图形链接或异构文档元素来改善证据访问，但它们通常采用一种大体上与查询无关的证据使用策略。我们提出了 TAP-RAG，一种用于长文档多模态问答的任务感知策略控制 RAG 框架。TAP-RAG 包含一个主控制器，即任务感知策略控制器（Task-Aware Policy Controller, TAPC），以及两个策略引导的证据执行器：任务感知查询引导流扩散（Task-Aware Query-Guided Flow Diffusion, TA-QFD）和任务感知视觉增强（Task-Aware Visual Enhancement, TAVE）。对于每个查询，TAPC 预测任务先验，估计视觉/局部/全局证据信号，并生成可执行的策略。然后，TA-QFD 在多模态文档图上扩展文本和结构证据，而 TAVE 在需要视觉或布局证据时选择性地检查页面图像。一个受保护的综合阶段融合文本、视觉和结构证据，并在支持不足时选择不执行。在 DocBench 和 MMLongBench-Doc 上，TAP-RAG 在比较的系统中实现了最佳的整体准确性，相较于匹配的多模态 RAG 基线分别提高了 +9.1 分（61.1 到 70.2）和 +4.5 分（42.2 到 46.7）。

View on arXiv Download PDF AI Translation

cs.CV / 45 / 2607.18924

Learning Explicit Physical Parameter Control and Benchmarking for Video Generation

学习显式物理参数控制与视频生成基准测试

Li, Yanxun, Wen, Hao, Song, Bingze, Zhu, Jiashu, Hao, Aiming, Chen, Chubin, Chen, Jintao, Wu, Jiahong, Chu, Xiangxiang, Wang, Miao

Abstract

Recent advances in image-to-video generation have improved visual realism, making physically grounded and controllable dynamics an important step toward future world simulation. Current models often generate plausible motion, but it is not reliably governed by explicit physical causes, and instance-level constraints can leak or become entangled in multi-object interactions. We attribute this gap to two missing pieces: large-scale, fine-grained physical parameterization, and model designs that correctly bind physical attributes to instances and emphasize dynamics over appearance. To bridge this gap, we introduce PhyParam-Dataset, an interaction-centric collection of 130K physically simulated videos with dense physical parameterization, including force vectors, object material properties, and environmental constants across five representative rigid-body motion types. Built on this data, we present PhyParam, a physics-guided image-to-video diffusion model that conditions on object-level forces, masses, friction, restitution, and scene-level gravity via a lightweight physical-attention routing mechanism, and further improves motion learning with semantic-structural feature-space supervision. We also establish PhyParam-Bench, a benchmark for physical-law consistency in image-to-video generation, with a multi-level protocol evaluating temporal dynamics, spatial stability, and semantic--physical alignment. Experiments show that PhyParam improves physical consistency while maintaining high visual fidelity, advancing explicit rigid-body physical-parameter control for image-to-video generation. We will publicly release the dataset, benchmark, and code to support future research.

Chinese Translation

近年来，图像到视频生成的进展提高了视觉真实感，使得基于物理的可控动态成为未来世界模拟的重要一步。目前的模型通常生成可信的运动，但并不可靠地由显式物理原因所支配，实例级约束在多物体交互中可能会泄漏或纠缠。我们将这一差距归因于两个缺失的部分：大规模、细粒度的物理参数化，以及能够正确将物理属性绑定到实例并强调动态而非外观的模型设计。为了解决这一问题，我们引入了PhyParam-Dataset，这是一个以交互为中心的包含13万个物理模拟视频的集合，具有密集的物理参数化，包括力向量、物体材料属性和五种代表性刚体运动类型的环境常数。基于这些数据，我们提出了PhyParam，这是一个物理引导的图像到视频扩散模型，通过轻量级的物理注意力路由机制，条件化于物体级的力、质量、摩擦、恢复力和场景级的重力，并通过语义结构特征空间监督进一步改善运动学习。我们还建立了PhyParam-Bench，这是一个用于图像到视频生成中物理法则一致性的基准测试，采用多层级协议评估时间动态、空间稳定性和语义-物理对齐。实验表明，PhyParam在保持高视觉保真度的同时提高了物理一致性，推动了图像到视频生成中显式刚体物理参数控制的发展。我们将公开发布数据集、基准测试和代码，以支持未来的研究。

View on arXiv Download PDF AI Translation

cs.CV / 46 / 2607.18958

Dual Adversarial Fine-tuning for Enhancing Robustness of Large Vision Language Model

双重对抗微调以增强大型视觉语言模型的鲁棒性

Wang, Sibo, Zhang, Jie, Shan, Shiguang, Chen, Xilin, Gao, Wen

Abstract

While Large Vision-Language Models (LVLMs), represented by LLaVA and GPT-4V, have demonstrated remarkable capabilities, their visual inputs remain vulnerable to adversarial attacks, posing significant security risks. Existing defense methods predominantly target single-task scenarios (e.g., zero-shot classification) and consequently lack generalizability across various multimodal tasks. To address this limitation, we propose a dual adversarial fine-tuning framework that jointly optimizes visual and semantic supervision signals from two modalities, enhancing model robustness while generalizing across multiple downstream tasks. The proposed framework comprises two core components, i.e., $\textbf{Visual}$ supervision branch and $\textbf{Semantic}$ supervision branch. The former branch leverages features from clean images, extracted via a frozen original vision encoder, to guide adversarial robustness while the latter incorporates caption-image alignment as a contextual signal to preserve semantic coherence under attack. Moreover, our method achieves cross-task robustness by simply replacing the CLIP vision encoder in the original model, with no need of separate task-specific retraining or architecture modifications.Extensive experiments demonstrate that our approach outperforms the state-of-the-art method in adversarial robustness evaluation across zero-shot classification, image captioning, and visual question answering (VQA) tasks.

Chinese Translation

尽管以 LLaVA 和 GPT-4V 为代表的大型视觉语言模型（LVLMs）展现了卓越的能力，但其视觉输入仍然容易受到对抗攻击，带来显著的安全风险。现有的防御方法主要针对单任务场景（例如，零样本分类），因此在各种多模态任务中缺乏通用性。为了解决这一局限性，我们提出了一种双重对抗微调框架，该框架联合优化来自两种模态的视觉和语义监督信号，从而增强模型的鲁棒性，同时在多个下游任务中实现泛化。所提出的框架包含两个核心组件，即 $ extbf{视觉}$ 监督分支和 $ extbf{语义}$ 监督分支。前者利用通过冻结的原始视觉编码器提取的干净图像特征来指导对抗鲁棒性，而后者则将图像与标题的对齐作为上下文信号，以在攻击下保持语义一致性。此外，我们的方法通过简单地替换原始模型中的 CLIP 视觉编码器来实现跨任务鲁棒性，无需单独的任务特定再训练或架构修改。大量实验表明，我们的方法在零样本分类、图像标题生成和视觉问答（VQA）任务的对抗鲁棒性评估中优于现有的最先进方法。

View on arXiv Download PDF AI Translation

cs.CV / 47 / 2607.18988

DobicVLM: Aligning Chest X-Ray Report Generation with Clinically-Grounded Programmatic Rewards via Group Relative Policy Optimization

DobicVLM：通过群体相对策略优化将胸部X光报告生成与临床基础的程序奖励对齐

Adewuyi, Thanni, Obayi, Angelica, Aniekan, Andem, Okoko, Samuel, Ezendu, Angel, Usani, Ephraim, Animasaun, Ademide, Chibundu, Philip, Maurice, Christian, Essien, Mary Donald, Odunsi, Oluwaseun, Oguntuase, Oluwasegun, Adereni, Abiodun

Abstract

Medical imaging is a cornerstone of diagnostics, yet automated chest X-ray report generation struggles with structural adherence, anatomical completeness, and semantic faithfulness. We introduce DobicVLM, a vision-language model combining supervised fine-tuning on MedGemma-4B with Group Relative Policy Optimization (GRPO) and clinically-grounded programmatic rewards. Our approach uses interpretable, rule-based reward components; structural verification, anatomical checklist, semantic similarity, and length constraints to enforce clinical standards without neural reward models. Trained on 1,000 de-identified image-report pairs from a private clinical dataset (with ethics approval and compliance to local regulations), DobicVLM is evaluated via blinded expert review on 69 held-out cases. DobicVLM outperforms Gemini 2.5 Flash across the majority of criteria, achieving the highest impression accuracy (27.2%) and medical terminology (86.5%) compared to both Gemini 2.5 Flash and MedGemma 4B baselines, with minor trade-offs in completeness and referrals. This demonstrates GRPO's value for transparent alignment in resource-limited settings. Keywords: Vision-Language Models, Radiology Report Generation, Reinforcement Learning, Medical AI, GRPO

Chinese Translation

医学影像是诊断的基石，但自动化的胸部X光报告生成在结构遵循、解剖完整性和语义忠实性方面仍面临挑战。我们提出了DobicVLM，这是一种视觉-语言模型，结合了在MedGemma-4B上进行的监督微调、群体相对策略优化（GRPO）和临床基础的程序奖励。我们的方法使用可解释的基于规则的奖励组件；结构验证、解剖检查表、语义相似性和长度约束，以在没有神经奖励模型的情况下强制执行临床标准。DobicVLM在一个经过伦理批准并符合当地法规的私有临床数据集中，基于1,000对去标识化的图像-报告对进行训练，并在69个保留案例上通过盲评专家进行评估。DobicVLM在大多数标准上优于Gemini 2.5 Flash，达到了最高的印象准确率（27.2%）和医学术语使用率（86.5%），与Gemini 2.5 Flash和MedGemma 4B基线相比，完整性和推荐的权衡较小。这证明了GRPO在资源有限环境中透明对齐的价值。关键词：视觉-语言模型、放射学报告生成、强化学习、医学人工智能、GRPO

View on arXiv Download PDF AI Translation

cs.CV / 48 / 2607.18990

SWITi: Quantifying and Reducing Tiling Artifacts with Sliding Window Inner Tiling

SWITi：量化和减少滑动窗口内部平铺的伪影

Carrara, Federico, Kukde, Aman, Croft, Melisande, Deschamps, Joran, Jug, Florian

Abstract

SWITi is a test-time method for reducing artifacts in tiled predictions, particularly for neural networks that learn posterior distributions from which solutions are sampled at inference time. Tiled predictions are unavoidable for large image data, and artifacts arise whenever tiles are smaller than a network's receptive field and when tiles are independent posterior samples. SWITi averages overlapping sliding-window predictions, so discrepancies between neighboring samples are spread across shifted tile positions rather than accumulating at fixed seam coordinates. For posterior models, SWITi uses no more tile samples than an MMSE estimate requires and therefore incurs no additional forward passes. Additionally, we introduce two reference-free metrics, the Fraction of Rejected Tests (FRT) and Artifact Severity (ASV), for detecting and quantifying tiling artifacts from a per-tile permutation test that compares the distribution of pixel gradients across tile seams against the surrounding image content. On pre-trained and published image splitting models across three fluorescence microscopy datasets in 2D and 3D, we show that SWITi substantially attenuates stitching seams while also improving reconstruction fidelity and resolution. Since tiling artifacts in posterior predictions can easily be mistaken for biological structures or for boundaries between biological structures, removing or reducing them using SWITi will improve the downstream processing of large image predictions, which is particularly relevant for biomedical data.

Chinese Translation

SWITi是一种在测试时减少平铺预测伪影的方法，特别适用于从后验分布中学习并在推理时进行采样的神经网络。对于大规模图像数据，平铺预测是不可避免的，当平铺的大小小于网络的感受野且平铺为独立的后验样本时，就会产生伪影。SWITi通过对重叠的滑动窗口预测进行平均，使得相邻样本之间的差异分布在移动的平铺位置上，而不是在固定的接缝坐标处累积。对于后验模型，SWITi所需的平铺样本数量不超过最小均方误差（MMSE）估计所需的样本数量，因此不会增加额外的前向传播。此外，我们引入了两个无参考指标，即拒绝测试的比例（FRT）和伪影严重性（ASV），用于通过每个平铺的置换测试检测和量化平铺伪影，该测试比较平铺接缝处的像素梯度分布与周围图像内容。我们在三个二维和三维荧光显微镜数据集上，对预训练和已发布的图像分割模型进行实验，结果表明SWITi显著减弱了接缝伪影，同时提高了重建的保真度和分辨率。由于后验预测中的平铺伪影很容易被误认为是生物结构或生物结构之间的边界，因此使用SWITi去除或减少这些伪影将改善大规模图像预测的后续处理，这对于生物医学数据尤为重要。

View on arXiv Download PDF AI Translation

cs.CV / 49 / 2607.18997

Benchmarking Deep Learning Approaches for AEC Engineering Drawing Layout Detection and Information Extraction

深度学习方法在建筑、工程和施工（AEC）工程图纸布局检测与信息提取中的基准研究

Huang, Tianyang, Lombardi, Alessio, Elnagar, Ahmed, Zalouk, Ahmed, Paul, George, Najjarpour, Sepehr, Sigurdsson, Arvid, Ismail, Khalid, Ragab, Mohamed, Vakaj, Edlira

Abstract

Information Extraction (IE) from Architecture, Engineering, and Construction (AEC) drawings remains hindered by manual inefficiency, while Layout Detection, a vital 'middleware' organizing graphical and textual hierarchies, is underexplored. General document layout models, optimized for text-centric content, lack validation on engineering drawings. This study constructs a custom AEC-specific layouts dataset and benchmarks five deep learning architectures. RF-DETR achieves state-of-the-art performance with an $mAP_{50}$ of 0.949, while the Vision-Language Model Qwen3-VL attains a leading F1-score of 0.911. Conversely, models pre-trained on general document datasets suffer from "domain interference", causing performance degradation. This establishes a robust technical foundation for automated IE in AEC.

Chinese Translation

从建筑、工程和施工（AEC）图纸中提取信息（IE）仍然受到手动效率低下的制约，而布局检测作为一种重要的“中间件”，用于组织图形和文本层次结构，尚未得到充分探索。针对以文本为中心内容优化的一般文档布局模型在工程图纸上缺乏验证。本研究构建了一个特定于AEC的自定义布局数据集，并对五种深度学习架构进行了基准测试。RF-DETR以0.949的$mAP_{50}$达到了最先进的性能，而视觉-语言模型Qwen3-VL则获得了领先的F1-score为0.911。相反，预训练于一般文档数据集的模型受到“领域干扰”的影响，导致性能下降。这为AEC中的自动化信息提取奠定了坚实的技术基础。

View on arXiv Download PDF AI Translation

cs.CV / 50 / 2607.19000

Learning Semantic-Robust Change Detection via Semantic-Invariant Self-Distillation

通过语义不变自蒸馏学习语义鲁棒的变化检测

Qu, Jiuhe, Liang, Yingping, Fu, Ying

Abstract

Change detection aims to identify semantic changes between remote sensing images. However, features from models are easily disturbed by non-semantic variations, such as illumination, shadows, and atmospheric changes, leading to false alarms and limited generalization in real-world scenarios. In this paper, we propose \textbf{SCDistill}, a framework for learning semantic-robust change detection via semantic-invariant self-distillation. First, to strengthen semantic consistency, we introduce a semantic-invariant self-distillation strategy that learns semantic robustness from perturbed yet semantically consistent data, empowering the change detector to extract disturbance-resistant features and achieve more reliable and accurate semantic change identification. Second, to expand paired data with non-semantic variations, we design a diffusion-based perturbation simulation pipeline that synthesizes complex environmental changes, enabling the model to explicitly learn to distinguish semantic changes from appearance-level fluctuations and reduce false alarms caused by non-semantic disturbances. These components promote robustness from data and representation perspectives, leading to synergistic performance gains. Extensive experiments demonstrate that SCDistill achieves state-of-the-art performance on multiple semantic change detection benchmarks and exhibits strong generalization to binary change detection and change captioning tasks. Code is accessible at https://github.com/elecreak/SCDistill.

Chinese Translation

变化检测旨在识别遥感图像之间的语义变化。然而，模型的特征容易受到非语义变化的干扰，如光照、阴影和大气变化，这导致误报并限制了在现实场景中的泛化能力。本文提出了 extbf{SCDistill}，一个通过语义不变自蒸馏学习语义鲁棒变化检测的框架。首先，为了增强语义一致性，我们引入了一种语义不变自蒸馏策略，该策略从扰动但语义一致的数据中学习语义鲁棒性，使变化检测器能够提取抗干扰特征，从而实现更可靠和准确的语义变化识别。其次，为了扩展具有非语义变化的配对数据，我们设计了一种基于扩散的扰动模拟管道，合成复杂的环境变化，使模型能够明确学习区分语义变化与外观层次波动，并减少由非语义干扰引起的误报。这些组件从数据和表示的角度促进了鲁棒性，带来了协同的性能提升。大量实验表明，SCDistill在多个语义变化检测基准上达到了最先进的性能，并在二元变化检测和变化描述任务中表现出强大的泛化能力。代码可在 https://github.com/elecreak/SCDistill 获取。

View on arXiv Download PDF AI Translation

cs.CV / 51 / 2607.19027

Mitigating Modality and Language-Style Gaps for Zero-Shot Video Moment Retrieval

减轻零样本视频时刻检索中的模态和语言风格差距

Lee, Jihyun, Cho, Cheol-Ho, Jun, Woojin, Jeong, Woojin, Heo, Jae-Pil

Abstract

Zero-shot video moment retrieval aims to overcome the limitations of traditional approaches that require large-scale datasets annotated with text and its relevant temporal spans. Despite advances in pre-trained vision-language models and multimodal large language models, existing ZMR methods still heavily depend on query-to-video content similarity, making them vulnerable to modality and language-style gaps. These gaps lead to unreliable span proposals and unstable moment retrieval results. To address this issue, we propose Self-Similarity-based Moment Proposal and Scoring that instead exploits intrinsic relationships within videos, enabling robust span generation and scoring. By deriving self-similarity only from the video content, we circumvent the noisy and mismatched patterns of query-frame or query-caption similarities, thereby mitigating both modality and language-style gaps. Furthermore, we introduce a query-aware MLLM-based reasoning stage to further sharpen alignment between text and video. Extensive experiments demonstrate that Self-SiMS achieves state-of-the-art performance across ZMR benchmarks.

Chinese Translation

零样本视频时刻检索旨在克服传统方法的局限性，这些方法需要带有文本及其相关时间跨度的大规模注释数据集。尽管预训练的视觉-语言模型和多模态大语言模型取得了进展，现有的零样本视频时刻检索（ZMR）方法仍然严重依赖于查询与视频内容之间的相似性，这使得它们容易受到模态和语言风格差距的影响。这些差距导致了不可靠的时间跨度提议和不稳定的时刻检索结果。为了解决这个问题，我们提出了一种基于自相似性的时刻提议与评分方法，该方法利用视频内部的内在关系，从而实现稳健的时间跨度生成和评分。通过仅从视频内容中推导自相似性，我们避免了查询帧或查询标题相似性中的噪声和不匹配模式，从而减轻了模态和语言风格的差距。此外，我们引入了一种基于查询感知的多模态大语言模型（MLLM）推理阶段，以进一步增强文本与视频之间的对齐。大量实验表明，自相似性时刻提议与评分（Self-SiMS）在零样本视频时刻检索基准测试中达到了最先进的性能。

View on arXiv Download PDF AI Translation

cs.CV / 52 / 2607.19032

IMMoE: Incomplete Multi-View Anomaly Detection via Mixture of View Experts Fusion

IMMoE：通过视图专家融合进行不完整多视角异常检测

Hu, Lei

Abstract

Existing Multi-view Anomaly Detection (MAD) methods assume that all views are completely available and model each view separately. However, in real industrial scenarios, information in the view may be missing due to faults such as occlusion, which leads to the performance degradation of existing methods due to the lack of a multi-view consistency prior. To address this, we explored a more challenging task: Incomplete Multi-View Anomaly Detection (IMVAD), in which some areas of each view were masked. We proposed a pipeline for automatically generating the IMVAD dataset and generated the \textbf{RIMAD} dataset based on the Real-IAD dataset through this pipeline. In addition, in order to effectively utilize the information of multiple views in the absence of view information, we propose \textbf{IMMoE}, which consists of two key modules: (1) Multi-View Expert Fusion (MVEF) effectively fuses multi-view information through a multi-view expert network and guides the reconstruction of a single view; (2) Local Anomaly Enhancement Encoder (LAEE) effectively prevents the model from overfitting the mask region by applying dropout to local features. Our method achieves state-of-the-art performance on both the RIMAD and Real-IAD datasets, especially on RIMAD, we have increased the pixel-level and image-level metrics by 11.8\% and 2.8\%, respectively. Our source code is available at https://github.com/HULEI7/IMMoE

Chinese Translation

现有的多视角异常检测（MAD）方法假设所有视角都是完全可用的，并分别对每个视角进行建模。然而，在实际工业场景中，由于遮挡等故障，视角中的信息可能会缺失，这导致现有方法由于缺乏多视角一致性先验而性能下降。为了解决这个问题，我们探索了一个更具挑战性的任务：不完整多视角异常检测（IMVAD），在该任务中，每个视角的一些区域被遮蔽。我们提出了一个自动生成IMVAD数据集的流程，并通过该流程基于Real-IAD数据集生成了 extbf{RIMAD}数据集。此外，为了在缺少视角信息的情况下有效利用多视角的信息，我们提出了 extbf{IMMoE}，该方法由两个关键模块组成：(1) 多视角专家融合（MVEF）通过多视角专家网络有效融合多视角信息，并指导单一视角的重建；(2) 局部异常增强编码器（LAEE）通过对局部特征应用dropout，有效防止模型对遮蔽区域的过拟合。我们的方法在RIMAD和Real-IAD数据集上实现了最先进的性能，特别是在RIMAD上，我们的像素级和图像级指标分别提高了11.8 ext{%}和2.8 ext{%}。我们的源代码可在https://github.com/HULEI7/IMMoE获取。

View on arXiv Download PDF AI Translation

cs.CV / 53 / 2607.19036

CoGoal3D: Collaborative 3D Object Detection with 3D-Aware Fusion and Refinement

CoGoal3D：基于3D感知融合与精炼的协作3D目标检测

Yang, Zhihao, Xiang, Zhiyu, Xu, Peng, Pu, Tianyu, Wang, Kai, Liu, Eryun, Zhang, Dongping, Ding, Yong

Abstract

V2X collaborative object detection features overcoming the limitations of single-vehicle systems by aggregating environmental features from multiple collaborative agents. However, existing mainstream V2X perception methods mainly focus on 2D BEV object detection. When 3D detection task is concerned, inferior results are obtained because they ignore the 3D spatial misalignment caused by differing height and attitude among the collaborators. In this paper, we propose a novel collaborative 3D object detection framework called CoGoal3D, which extracts and refines the 3D feature gradually in a two-stage pipeline. In the first stage, a multiscale 3D-aware global fusion module is designed to mitigate the 3D spatial misalignment. The resulting proposals are then refined in the second stage with an auxiliary task of 3D point reconstruction. An effective multi-agent collaborative data augmentation strategy is further proposed to enrich the training data while minimizing information loss. Extensive experiments on public real-world datasets demonstrate that our CoGoal3D achieves new state-of-the-art performance, with 3D [email protected] improvements of 10.86%, 10.34%, and 10.18% on the DAIR-V2X, V2V4Real, and V2X-Real datasets, respectively. Code is available at https://github.com/Megalo-f/CoGoal3D.

Chinese Translation

V2X协作目标检测通过聚合多个协作代理的环境特征，克服了单车系统的局限性。然而，现有主流的V2X感知方法主要集中在2D鸟瞰视图（BEV）目标检测上。当涉及3D检测任务时，由于忽略了协作方之间因高度和姿态不同而造成的3D空间错位，导致结果不理想。本文提出了一种新颖的协作3D目标检测框架，称为CoGoal3D，该框架在一个两阶段的流程中逐步提取和精炼3D特征。在第一阶段，设计了一个多尺度3D感知全局融合模块，以减轻3D空间错位。然后在第二阶段，通过3D点重建的辅助任务对生成的提案进行精炼。此外，进一步提出了一种有效的多代理协作数据增强策略，以丰富训练数据，同时最小化信息损失。在公共真实世界数据集上的大量实验表明，我们的CoGoal3D在性能上达到了新的最先进水平，在DAIR-V2X、V2V4Real和V2X-Real数据集上，3D [email protected]分别提高了10.86%、10.34%和10.18%。代码可在https://github.com/Megalo-f/CoGoal3D获取。

View on arXiv Download PDF AI Translation

cs.CV / 54 / 2607.19038

FilmWorld: Agentic Novel-to-Film Generation through Dynamic Cinematic World Modeling

FilmWorld：通过动态电影世界建模实现自主小说到电影的生成

Zuo, Jialong, Zuo, Haotong, Zhang, Shiwei, Wang, Xiang, Li, Chen, Sang, Nong, Gao, Changxin, Bai, Xiang

Abstract

Translating novels into films poses a grand challenge for generative artificial intelligence, requiring conversion of abstract literary prose into long-form, multi-scene visual narratives. While current video generation models excel at short, single-scene clips within narrow temporal and spatial contexts, novel-to-film generation operates in a more complex regime, demanding long-duration content across diverse scenes with dynamically evolving entity states. To address this, we formalize novel-to-film generation as dynamic cinematic world modeling, decomposed into two phases: construction, which grounds abstract, underspecified literary narratives into concrete, stateful, and persistent world entities; and evolution, which governs how these entities dynamically update under plot progression to maintain causal consistency across scenes. We propose FilmWorld, an end-to-end agentic system where two groups of specialized agents collaborate to instantiate these phases. Construction-side agents perform narrative structured translation, world entity state modeling with visual anchoring, and state-driven shot planning, progressively projecting literary language into a cinematic blueprint. Evolution-side agents perform state-anchored visual generation, cross-shot dynamic state propagation, and closed-loop state verification to maintain causal consistency and visual coherence. To address the evaluation gap in long-form generation, we introduce FilmEval, a systematic evaluation framework that couples a difficulty-graded benchmark of 15 representative novels with an automated protocol of nine objective metrics spanning three dimensions: cinematic presentation, film consistency, and novel fidelity. Experiments demonstrate that FilmWorld consistently outperforms state-of-the-art video generation agent systems, with particularly pronounced improvements in narrative fidelity and cross-scene consistency.

Chinese Translation

将小说转化为电影对生成性人工智能提出了巨大的挑战，要求将抽象的文学散文转化为长篇、多场景的视觉叙事。尽管当前的视频生成模型在狭窄的时间和空间背景下对短小的单场景片段表现出色，但小说到电影的生成则在更复杂的环境中运作，要求在不同场景中生成长时间内容，并且实体状态动态演变。为了解决这一问题，我们将小说到电影的生成形式化为动态电影世界建模，分解为两个阶段：构建阶段，将抽象、未具体化的文学叙事转化为具体的、有状态的、持久的世界实体；演变阶段，管理这些实体在情节推进下如何动态更新，以保持场景间的因果一致性。我们提出了FilmWorld，一个端到端的自主系统，其中两个专业代理组协作实现这两个阶段。构建侧代理执行叙事结构翻译、带有视觉锚定的世界实体状态建模和基于状态的镜头规划，逐步将文学语言投射到电影蓝图中。演变侧代理执行基于状态的视觉生成、跨镜头动态状态传播和闭环状态验证，以保持因果一致性和视觉连贯性。为了解决长篇生成中的评估空白，我们引入了FilmEval，一个系统的评估框架，将15部具有代表性的小说的难度分级基准与涵盖三个维度的九个客观指标的自动化协议相结合：电影呈现、影片一致性和小说保真度。实验表明，FilmWorld在叙事保真度和跨场景一致性方面的表现显著优于现有的最先进视频生成代理系统。

View on arXiv Download PDF AI Translation

cs.CV / 55 / 2607.19040

Gaze-DETR: Top-Down Guidance Through Priority Maps for Infrared Weak-Small UAV Detection with DETR

Gaze-DETR：通过优先级图进行自上而下的指导，以实现红外弱小型无人机检测

Liu, Nian, Yang, Yuxin, Lin, Shubo, Zhang, Sikui, Li, Liang, Cai, Boyu, Wang, Yizheng, Hu, Weiming, Gao, Jin

Abstract

Infrared small target detection (ISTD) remains challenging because tiny, low-contrast targets are easily overwhelmed by clutter, noise, or occlusion. Conventional single-frame and multi-frame detectors rely on bounding-box supervision, which specifies final target locations but offers little explicit guidance for prioritizing candidate regions or preserving weak-target evidence before localization. Task-driven visual search offers such guidance: top-down goals and visual evidence jointly form a spatial priority map that ranks candidate locations. Building on this principle, we propose Gaze-DETR, a bio-inspired detector that learns an internal priority map before localization. First, a priority head predicts a normalized priority map from image features. Second, Residual Priority-Guided Feature Modulation (RPFM) enhances high-priority responses while retaining multi-scale features. Finally, Priority-Guided Anchor Query Injection (PAQI) converts high-priority locations into decoder anchor queries. We train the priority head using three supervision schemes: box-derived Gaussian maps; real-gaze maps constructed from fixation-density maps; and transferred pseudo-gaze maps learned from gaze--box relations in paired annotations and applied to Anti-UAV410 training boxes. To support the latter two schemes, we construct TIR-UAV120-Gaze with paired detection and task-driven eye-tracking annotations. On TIR-UAV120-Gaze, Gaze-DETR achieves 85.76 mAP$_{50}$ and 88.77 F1 with box-derived supervision, and 86.18 mAP$_{50}$ and 89.00 F1 with real-gaze supervision. On Anti-UAV410, it achieves 87.06 mAP$_{50}$ and 90.90 F1 with box-derived supervision, and 87.08 mAP$_{50}$ and 90.43 F1 with transferred pseudo-gaze supervision. These results show that explicit spatial-priority learning provides pre-localization guidance complementary to bounding-box supervision across annotation settings and costs.

Chinese Translation

红外小目标检测（ISTD）仍然具有挑战性，因为微小、低对比度的目标容易被杂波、噪声或遮挡所淹没。传统的单帧和多帧检测器依赖于边界框监督，这种方法虽然指定了最终目标的位置，但对优先考虑候选区域或在定位之前保留弱目标证据的指导作用有限。任务驱动的视觉搜索提供了这样的指导：自上而下的目标和视觉证据共同形成一个空间优先级图，来对候选位置进行排序。基于这一原则，我们提出了Gaze-DETR，这是一种生物启发的检测器，在定位之前学习内部优先级图。首先，优先级头从图像特征中预测一个归一化的优先级图。其次，残差优先级引导特征调制（RPFM）增强高优先级响应，同时保留多尺度特征。最后，优先级引导锚查询注入（PAQI）将高优先级位置转换为解码器锚查询。我们使用三种监督方案训练优先级头：基于框的高斯图；从注视密度图构建的真实注视图；以及从配对注释中的注视-框关系学习并应用于反无人机410训练框的转移伪注视图。为了支持后两种方案，我们构建了TIR-UAV120-Gaze，包含配对的检测和任务驱动的眼动追踪注释。在TIR-UAV120-Gaze上，Gaze-DETR在基于框的监督下实现了85.76 mAP$_{50}$和88.77 F1，在真实注视监督下实现了86.18 mAP$_{50}$和89.00 F1。在反无人机410上，它在基于框的监督下实现了87.06 mAP$_{50}$和90.90 F1，在转移伪注视监督下实现了87.08 mAP$_{50}$和90.43 F1。这些结果表明，明确的空间优先级学习为在不同注释设置和成本下提供了补充于边界框监督的预定位指导。

View on arXiv Download PDF AI Translation

cs.CV / 56 / 2607.19046

Contrastive On-Policy Distillation

对比性在线蒸馏

Ruan, Jiacheng, Tang, Jun, Yuan, Wenzhen, Liu, Ting, Bai, Shuai, Liu, Dayiheng, Yang, Zhibo, Fu, Yuzhuo

Abstract

On-policy Distillation (OPD) supervises a student model on trajectories sampled from its own policy by minimizing the divergence between the output distributions of the teacher and student at each token position, thereby providing dense token-level supervision. Although existing OPD methods have demonstrated strong performance in improving the reasoning ability of student models, their objectives fundamentally rely on token-level distribution matching. Consequently, they lack an explicit signal for comparing a token's relative compatibility across reasoning modes and thus do not directly model preferences between these modes. To address this limitation, we propose COPD, a contrastive OPD framework. Specifically, for each token generated by the student model, a frozen teacher model scores the same student state under two contrasting instructions that elicit light and heavy reasoning. The difference between the resulting log probabilities serves as a token-level advantage signal to guide the OPD update. Rather than merely imitating a single teacher distribution, COPD directly encourages the student model to learn more concise and efficient reasoning strategies. We conduct experiments on nine multimodal benchmarks covering both reasoning and understanding tasks. The results show that COPD substantially reduces reasoning length without compromising model performance and consistently improves efficiency across different tasks and model scales. Furthermore, the contrastive formulation can be seamlessly integrated into the On-policy Self-distillation (OPSD) framework, where self-contrastive supervision is constructed without an additional teacher model, thereby enabling the model to distill itself toward lightweight reasoning.

Chinese Translation

在线蒸馏（On-policy Distillation, OPD）通过最小化教师模型与学生模型在每个标记位置的输出分布之间的差异，对从其自身策略中采样的轨迹进行学生模型的监督，从而提供密集的标记级监督。尽管现有的OPD方法在提升学生模型的推理能力方面表现出色，但其目标基本上依赖于标记级分布匹配。因此，它们缺乏比较标记在不同推理模式下相对兼容性的明确信号，因此未能直接建模这些模式之间的偏好。为了解决这一限制，我们提出了COPD，一个对比性OPD框架。具体而言，对于学生模型生成的每个标记，一个冻结的教师模型在两种对比指令下对同一学生状态进行评分，这两种指令分别引发轻推理和重推理。由此产生的对数概率之间的差异作为标记级优势信号，以指导OPD更新。COPD不仅仅是模仿单一教师分布，而是直接鼓励学生模型学习更简洁和高效的推理策略。我们在涵盖推理和理解任务的九个多模态基准上进行了实验。结果表明，COPD显著减少了推理长度，而不影响模型性能，并在不同任务和模型规模中持续提高了效率。此外，对比性公式可以无缝集成到在线自蒸馏（On-policy Self-distillation, OPSD）框架中，在该框架下，自对比监督是在没有额外教师模型的情况下构建的，从而使模型能够朝着轻量级推理进行自我蒸馏。

View on arXiv Download PDF AI Translation

cs.CV / 57 / 2607.19061

Now You See the Hate: Adaptive View Retrieval for Hidden Hateful Illusions

现在你看到了仇恨：针对隐藏仇恨幻觉的自适应视图检索

Chen, Qianpu, Soydaner, Derya

Abstract

Hateful optical illusions expose a serious gap in current multimodal safety systems. On original-view hateful illusions, previous work shows that six moderation classifiers achieve at most 20.9 to 24.5% accuracy and nine state-of-the-art VLMs remain at or below 10.2% with illusion-aware prompting, leaving most hidden hate undetected. We formulate hidden hateful illusion detection as a perceptual retrieval problem and propose Adaptive View Retrieval. This retrieve-and-calibrate framework assembles a complementary view bank for the image and hidden-message templates, adaptively selects which views to trust, retrieves hidden-message identities, and calibrates whether the recovered evidence is harmful. On HatefulIllusion with a frozen CLIP encoder, Adaptive View Retrieval reaches 93.2% balanced accuracy on the held-out test split. It substantially outperforms original-view baselines and fixed single-transform filters across hate slangs, hate symbols, and visibility levels. The same design also surpasses official fine-tuned CLIP baselines, matches or exceeds human performance on IllusionMNIST, IllusionFashionMNIST, and IllusionAnimals, and outperforms zoom-out preprocessing on HC-Bench under the SemVink protocol. Together, these results show that robust multimodal moderation requires recovering hidden meaning before deciding whether it is harmful.

Chinese Translation

仇恨光学幻觉暴露了当前多模态安全系统中的一个严重缺口。在原始视图的仇恨幻觉中，以往的研究表明，六个审查分类器的准确率最高仅为20.9%到24.5%，而九个最先进的视觉语言模型（VLMs）在幻觉感知提示下的准确率保持在或低于10.2%，导致大多数隐藏的仇恨未被检测到。我们将隐藏仇恨幻觉检测形式化为一种感知检索问题，并提出了自适应视图检索（Adaptive View Retrieval）。该检索与校准框架为图像和隐藏信息模板组装了一个互补视图库，自适应选择可信视图，检索隐藏信息的身份，并校准恢复的证据是否有害。在使用冻结的CLIP编码器的HatefulIllusion数据集上，自适应视图检索在保留的测试集上达到了93.2%的平衡准确率。它在仇恨俚语、仇恨符号和可见性水平上显著优于原始视图基线和固定单一变换过滤器。相同的设计还超越了官方微调的CLIP基线，在IllusionMNIST、IllusionFashionMNIST和IllusionAnimals上与人类表现相匹配或超过，并在SemVink协议下的HC-Bench上优于缩放预处理。综合来看，这些结果表明，稳健的多模态审查需要在决定其是否有害之前恢复隐藏的含义。

View on arXiv Download PDF AI Translation

cs.CV / 58 / 2607.19064

Mage-Flow: An Efficient Native-Resolution Foundation Model for Image Generation and Editing

Mage-Flow：一种高效的原生分辨率基础模型用于图像生成和编辑

Zhang, Xinjie, Zhang, Peng, Zheng, Shicheng, Guo, Jinghao, Jia, Zhaoyang, Shen, Yifei, Guo, Xun, Luo, Yuxuan, Li, Jiahao, Xie, Wenxuan, Pu, Fanyi, Zhang, Xiaoyi, Zhang, Kaichen, Guo, Zongyu, Bi, Tianci, Gui, Dongnan, Liu, Zhening, Wen, Zimo, Zheng, Zihan, Yang, Senqiao, Li, Xiao, Wang, Jinglu, Li, Bin, Lu, Yan

Abstract

Large-scale visual generators are increasingly capable but costly to train, fine-tune, and deploy. We introduce Mage-Flow, a compact 4B-scale generative stack for efficient text-to-image generation and instruction-based image editing. The stack is built from two co-designed components: Mage-VAE, a lightweight high-fidelity latent tokenizer, and a Native-Resolution Multimodal Diffusion Transformer trained with rectified flow matching. Mage-VAE uses one-step diffusion-style encoding and decoding with anchor-latent regularization, preserving the reconstruction quality of strong public VAEs while reducing tokenization cost by more than an order of magnitude. Together with native-resolution packing and stack-level CUDA kernel fusion, the stack supports flexible-resolution training and improves end-to-end training throughput by about $2.5\times$. Built on this foundation, we develop a complete model family with Base, RL-aligned, and Turbo variants for both generation and editing. Diffusion-NFT improves prompt following, text rendering, aesthetic quality, and editing fidelity, while few-step distillation with adversarial perceptual guidance produces 4-step Turbo models for low-latency inference. Despite its compact scale, Mage-Flow and Mage-Flow-Edit achieves competitive performance across standard generation and editing benchmarks. More importantly, the Turbo variants make high-resolution generation and editing practical for interactive use: at $1024^2$ resolution on a single NVIDIA A100 GPU, Mage-Flow-Turbo generates an image in 0.59s, and Mage-Flow-Edit-Turbo edits an image in 1.02s, while maintaining a small memory footprint. These results show that careful tokenizer--backbone--system co-design can deliver strong high-resolution generation and editing within an efficient 4B model family.

Chinese Translation

大规模视觉生成器的能力日益增强，但训练、微调和部署的成本也越来越高。我们介绍了Mage-Flow，一种紧凑的4B规模生成堆栈，用于高效的文本到图像生成和基于指令的图像编辑。该堆栈由两个共同设计的组件构成：Mage-VAE，一个轻量级高保真潜在标记器，以及一个使用修正流匹配训练的原生分辨率多模态扩散变换器。Mage-VAE采用一步扩散风格的编码和解码，并使用锚定潜在正则化，保持强大公共VAE的重建质量，同时将标记化成本降低了一个数量级以上。结合原生分辨率打包和堆栈级CUDA内核融合，该堆栈支持灵活分辨率训练，并将端到端训练吞吐量提高约2.5倍。在此基础上，我们开发了一个完整的模型家族，包括基础版、与强化学习对齐版和Turbo版，适用于生成和编辑。Diffusion-NFT提高了提示遵循、文本渲染、美学质量和编辑保真度，而通过对抗感知引导的少步蒸馏生成了4步Turbo模型，以实现低延迟推理。尽管规模紧凑，Mage-Flow和Mage-Flow-Edit在标准生成和编辑基准测试中表现出竞争力。更重要的是，Turbo版使高分辨率生成和编辑在交互使用中变得实用：在单个NVIDIA A100 GPU上以1024^2分辨率，Mage-Flow-Turbo在0.59秒内生成一幅图像，而Mage-Flow-Edit-Turbo在1.02秒内编辑一幅图像，同时保持较小的内存占用。这些结果表明，精心设计的标记器-主干-系统协同设计能够在高效的4B模型家族中实现强大的高分辨率生成和编辑。

View on arXiv Download PDF AI Translation

cs.CV / 59 / 2607.19069

Delineate Anything v2: A Global Foundation Model for Field Delineation

Delineate Anything v2：用于田地划分的全球基础模型

Lavreniuk, Mykola, Kussul, Nataliia, Shelestov, Andrii, Salii, Yevhenii, Kuzin, Volodymyr, Wang, Charlotte Julia Li-Xing, Szantoi, Zoltan

Abstract

Accurate agricultural field boundary delineation at large scale is a foundational task for food security, supply chain transparency, and carbon accounting. While vision foundation models like SAM show remarkable zero-shot capabilities, they frequently fail in geospatial domains due to topological complexity, cropland texturing patterns, and a lack of physical scale awareness. In this work, we introduce Delineate Anything v2, a globally scalable foundation model designed specifically for wide-area field boundary mapping. We construct FBIS-73M, a 73-million-instance multi-resolution dataset spanning 61 countries. To address the pervasive issue of multi-field administrative parcel merging, we introduce a resolution-specific data curation pipeline that leverages topological image-space adaptation to homogenize merged parcels and strengthen weak physical boundaries. Furthermore, we establish a novel, manually curated evaluation benchmark covering 100 countries to assess independent zero-shot generalization. Our results show that Delineate Anything v2 surpasses the current state-of-the-art, including the Delineate Anything framework, by 0.284 [email protected] (+103.3% relative gain), while maintaining execution speeds suitable for rapid national- and global-scale deployment, as demonstrated by nationwide mapping of Ukraine (603,000 km^2) in 5.4 hours on a consumer-grade workstation. Code, pre-trained weights, the FBIS-73M dataset, and ready-to-use national-scale vector boundary products are publicly available at https://github.com/Lavreniuk/Delineate-Anything.

Chinese Translation

在大规模上准确划分农业田界是保障粮食安全、供应链透明度和碳核算的基础性任务。尽管像SAM这样的视觉基础模型展现了显著的零样本能力，但由于拓扑复杂性、农田纹理模式以及缺乏物理尺度意识，它们在地理空间领域常常表现不佳。在本研究中，我们介绍了Delineate Anything v2，这是一种专门为广域田界映射设计的全球可扩展基础模型。我们构建了FBIS-73M，这是一个涵盖61个国家的7300万实例多分辨率数据集。为了解决多田行政区块合并的普遍问题，我们引入了一种特定分辨率的数据整理流程，该流程利用拓扑图像空间适应来均匀化合并的区块，并增强薄弱的物理边界。此外，我们建立了一个新的手动整理的评估基准，覆盖100个国家，以评估独立的零样本泛化能力。我们的结果表明，Delineate Anything v2在[email protected]上超越了当前的最先进技术，包括Delineate Anything框架，提升了0.284 (+103.3%的相对增益)，同时保持了适合快速国家和全球规模部署的执行速度，正如在消费者级工作站上对乌克兰（603,000 km^2）进行全国映射所示，仅需5.4小时。代码、预训练权重、FBIS-73M数据集以及现成的国家级矢量边界产品已公开发布在https://github.com/Lavreniuk/Delineate-Anything。

View on arXiv Download PDF AI Translation

cs.CV / 60 / 2607.19077

Context-structured Video Anomaly Detection with Large Vision-Language Models

基于上下文结构的大型视觉-语言模型的视频异常检测

Kim, Dongjun, Oh, Changjae, Cavallaro, Andrea, Mo, Jeonghoon

Abstract

Training video anomaly detectors is challenging due to the difficulty and cost of annotating diverse and rare abnormal events. Although recent large vision-language models enable training-free inference, existing approaches mostly rely on holistic inference over sampled video and may miss context-specific anomaly cues. In this paper, we present CSI-VAD, a training-free video anomaly detector that identifies abnormal events across diverse contexts. The key idea is to decompose each video into three distinct contexts (environment, objects, time) and perform context-specific inference in separate branches. Because we ground anomaly judgments solely in context-specific visual cues, we do not require predefined text prompts describing abnormal events or dataset-specific tuning. Experiments on UCF-Crime and UBnormal show that CSI-VAD consistently improves over the direct holistic baseline and achieves competitive performance against existing methods, showing the advantage of structured context decomposition for training-free video anomaly detection.

Chinese Translation

训练视频异常检测器具有挑战性，因为对多样和稀有异常事件进行注释的难度和成本较高。尽管最近的大型视觉-语言模型使得无训练推理成为可能，但现有方法大多依赖于对采样视频的整体推理，可能会错过特定上下文的异常线索。本文提出了CSI-VAD，一种无训练的视频异常检测器，能够在多样的上下文中识别异常事件。其关键思想是将每个视频分解为三个不同的上下文（环境、物体、时间），并在独立的分支中进行特定上下文的推理。由于我们仅基于特定上下文的视觉线索来判断异常，因此不需要预定义描述异常事件的文本提示或特定数据集的调优。在UCF-Crime和UBnormal上的实验表明，CSI-VAD在直接的整体基线之上始终有所改善，并且在与现有方法的比较中表现出竞争力，展示了结构化上下文分解在无训练视频异常检测中的优势。

View on arXiv Download PDF AI Translation

cs.CV / 61 / 2607.19086

Advancing Multimodal Fusion on Heterogeneous Medical Data with Hybrid Geometry Attention

基于混合几何注意力的异构医学数据多模态融合的进展

Dhar, Joy, Pandey, Manish Kumar, Zaidi, Nayyar, Chen, Chen, Haghighat, Maryam, Sohel, Ferdous, Goyal, Puneet

Abstract

Multimodal fusion learning (MFL) has shown great potential in the medical domain, where we are faced with disparate data modalities such as imaging, clinical records, and omics. However, existing MFL strategies face several major challenges. First, they struggle to capture complex cross-modal interactions effectively, which in turn limits performance improvements. Second, they incur high computational costs, restricting their applicability in resource-constrained healthcare AI applications. Finally, they are often designed and evaluated for narrow, fixed modality configurations (e.g., imaging-only, or specific pairs such as image and omics), which limits evidence of their adaptability and generalizability to broader collections of heterogeneous medical modalities. To address these challenges, we propose a novel MFL framework - Cascaded Unified Representation Learning for Efficient Fusion Network (CURE) - a lightweight and scalable framework that progressively integrates various modalities through a novel efficient Hybrid Geometry Aware Fusion layer (HyFuse), where each HyFuse layer is sequentially learned for each modality, making the framework adaptable and generalizable. Within HyFuse, an efficient residual convolution module captures rich multi-scale features to ensure cost-effective learning, while a hybrid-space aware attention mixer learns coarse-to-fine structural cues to better preserve cross-modal relationships. Complementary learnable late-fusion and shared information refinement modules are then employed to learn robust modality-order-invariant shared representations, which in turn yields consistent performance improvements. Extensive evaluations on 16 public datasets show that CURE outperforms leading multimodal fusion methods, boosting performance by up to 3.97% and lowering computational costs by up to 87.8%, ensuring more effective and reliable predictions.

Chinese Translation

多模态融合学习（MFL）在医学领域展现出巨大的潜力，我们面临着成像、临床记录和组学等不同的数据模态。然而，现有的MFL策略面临几个主要挑战。首先，它们在有效捕捉复杂的跨模态交互方面存在困难，这限制了性能的提升。其次，它们的计算成本较高，限制了其在资源受限的医疗人工智能应用中的适用性。最后，它们通常是为狭窄的固定模态配置（例如，仅成像或特定对如图像和组学）设计和评估的，这限制了它们在更广泛的异构医学模态集合中的适应性和普遍性。为了解决这些挑战，我们提出了一种新颖的MFL框架——级联统一表示学习高效融合网络（CURE），这是一个轻量级且可扩展的框架，通过一种新颖的高效混合几何感知融合层（HyFuse）逐步整合各种模态，其中每个HyFuse层针对每个模态进行顺序学习，使得框架具有适应性和普遍性。在HyFuse中，一个高效的残差卷积模块捕获丰富的多尺度特征，以确保成本效益学习，而一个混合空间感知注意力混合器学习粗到细的结构线索，以更好地保留跨模态关系。随后，采用互补的可学习晚期融合和共享信息精炼模块，以学习稳健的模态顺序不变共享表示，从而带来一致的性能提升。在16个公共数据集上的广泛评估表明，CURE在性能上优于领先的多模态融合方法，性能提升高达3.97%，计算成本降低高达87.8%，确保了更有效和可靠的预测。

View on arXiv Download PDF AI Translation

cs.CV / 62 / 2607.19100

FlexiAvatar: Unified 3D Gaussian Human Avatars Under Arbitrary Body Visibility

FlexiAvatar：在任意身体可见性下的统一3D高斯人类头像

Tiruneh, Yihalem Yimolal, Ali, Muhammad Salman, Jeong, Uyoung, Khan, Muneeb A., Sayem, MD Khalequzzaman Chowdhury, Bayramgeldiyev, Allanur, Bhattarai, Binod, Baek, Seungryul

Abstract

Reconstructing animatable 3D human avatars from monocular video is a fundamental problem in computer vision with broad applications in AR/VR and digital content creation. Existing approaches typically couple parametric body models with neural rendering or 3D Gaussian splatting and optimize all body regions jointly from short videos, which often degrades fidelity in the visible areas. To overcome this limitation, we introduce FlexiAvatar, a unified framework that explicitly optimizes only the visible body regions, effectively eliminating artifacts arising from unobserved limbs. Our method integrates occlusion-robust SMPL-X tracking with part-specific residual refinement to capture high-frequency geometric and appearance details. To complete entirely unseen regions (e.g., back views), we leverage a diffusion-based approach to generate texture consistent with the observed appearance. Experiments on full-body (NeuMan, ZJU-MoCap, WildAvatar), upper/half-body (talk-show clips), and head-only (INSTA) inputs show that FlexiAvatar delivers consistently higher reconstruction quality, outperforming state-of-the-art methods by an average PSNR improvement of approximately 3% across datasets. Finally, by restricting optimization to observed regions, our method reduces the effective number of Gaussians that must be optimized and rendered, leading to reduced runtime and memory overhead in partial-visibility scenarios.

Chinese Translation

从单目视频重建可动画的3D人类头像是计算机视觉中的一个基本问题，具有广泛的增强现实/虚拟现实（AR/VR）和数字内容创作应用。现有的方法通常将参数化身体模型与神经渲染或3D高斯点云结合，并从短视频中联合优化所有身体区域，这往往会降低可见区域的保真度。为了解决这一限制，我们提出了FlexiAvatar，一个统一框架，明确优化仅可见的身体区域，有效消除由未观察到的肢体引起的伪影。我们的方法将抗遮挡的SMPL-X跟踪与特定部位的残差细化相结合，以捕捉高频几何和外观细节。为了填补完全未见的区域（例如，背面视图），我们利用基于扩散的方法生成与观察到的外观一致的纹理。在全身（NeuMan、ZJU-MoCap、WildAvatar）、上半身/半身（脱口秀片段）和仅头部（INSTA）输入上的实验表明，FlexiAvatar提供了一致更高的重建质量，平均PSNR提升约3%，超越了最先进的方法。最后，通过将优化限制在观察到的区域，我们的方法减少了必须优化和渲染的有效高斯数量，从而在部分可见场景中降低了运行时间和内存开销。

View on arXiv Download PDF AI Translation

cs.CV / 63 / 2607.19111

GATE-3D: Geometry-Aware Test-time Adaptive Reranking for Open-Set 3D Shape Retrieval

GATE-3D：面向几何的测试时自适应重排序用于开放集3D形状检索

Wu, Hao, Lin, Heyi, Wang, Zilin, Yao, Huizai, Wang, Hao, Xiong, Hui

Abstract

Large pretrained vision models have substantially improved appearance-based 3D shape retrieval, but they still confuse shapes that look similar while differing in geometry. Although geometry-aware features can reduce these errors, naive fusion of geometry and appearance may hurt retrieval when the two modalities are already well aligned. We propose GATE-3D, a lightweight query-adaptive reranking method that incorporates geometry without retraining the retrieval backbone. For each query, GATE-3D predicts how much a geometry-aware score should adjust the appearance-based ranking using features that capture disagreement between the two modalities. This selective design lets geometry contribute where it helps and stay silent where it would hurt. Experiments on three open-set 3D retrieval benchmarks show that GATE-3D improves over appearance-only retrieval and is more robust than always-on fusion. On the primary benchmark, it improves mAP@10 by 2.00 points over appearance-only retrieval (p=0.041); it also improves leave-one-category-out generalization and reduces geometric false positives by 10.8%. GATE-3D achieves competitive zero-shot results against DAC-based baselines. We further find that simple linear routing is more effective than a small MLP in the low-data regime, suggesting that cross-modal disagreement features matter more than model capacity for adaptive routing.

Chinese Translation

大型预训练视觉模型显著改善了基于外观的3D形状检索，但它们仍然会混淆外观相似但几何形状不同的物体。尽管考虑几何特征可以减少这些错误，但简单地将几何与外观融合可能会在两种模态已经很好对齐时影响检索效果。我们提出了GATE-3D，这是一种轻量级的查询自适应重排序方法，它在不重新训练检索主干的情况下融入几何信息。对于每个查询，GATE-3D预测几何感知分数应如何调整基于外观的排名，使用捕捉两种模态之间不一致性的特征。这种选择性设计使得几何信息在有帮助的地方发挥作用，而在可能造成损害的地方保持沉默。在三个开放集3D检索基准上的实验表明，GATE-3D在基于外观的检索上有所改善，并且比始终开启的融合方法更具鲁棒性。在主要基准上，它相较于仅基于外观的检索提高了2.00个点的mAP@10（p=0.041）；它还改善了留一类别外的泛化能力，并减少了10.8%的几何误报。GATE-3D在零样本任务中对比DAC基线取得了竞争力的结果。我们进一步发现，在低数据环境下，简单的线性路由比小型多层感知器（MLP）更有效，这表明跨模态不一致特征在自适应路由中比模型容量更为重要。

View on arXiv Download PDF AI Translation

cs.CV / 64 / 2607.19115

CR-Refiner: An Object-Centric Optimal Transport Reranker for Edit-Conditioned 3D Scene Retrieval

CR-Refiner：一种面向对象的最优传输重排序方法用于编辑条件下的3D场景检索

Wu, Hao, Zhu, Jinjing, Wu, Nanyu, Cai, Qianyi, Lin, Heyi, Wang, Hao, Xiong, Hui

Abstract

Edit-conditioned 3D scene retrieval pairs a reference 3D room with a natural-language modification and retrieves rooms from a corpus that satisfy the edit. Three lines of prior work each fall short on this task. 2D composed image retrieval reasons over pixel-level edits and has no primitive for 3D object sets. 3D foundation encoders embed individual objects but cannot compose at the scene level. 3D scene-grounding methods localize references inside a static scene rather than rank modified rooms across a corpus. We present CR-Refiner, a training-free reranker that wraps any base retriever's top-K candidates with three components. A frozen LLM parses the edit into a structured query entity, and each candidate is scored by an unbalanced optimal-transport problem over a 1xG cost matrix coupling category, style, material, and geometry. The unbalanced solver lets the single-entity query drop mass on irrelevant objects, modelling the asymmetry directly. An axis-conditional structural prior adds size-keyword cues for geometric edits and subject-anchor direction cues for spatial edits. An LLM verifier refines the top three candidates with continuous confidence. Because no benchmark evaluates compositional matching over 3D object sets, we additionally release 3D-CER, 4,963 edit-conditioned queries over a 23,381-room indoor corpus across five edit axes, with multi-positive ground truth, CIRR-style hard subsets, and zero-target adversarials. Across three qualitatively distinct base retrievers, CR-Refiner consistently improves hard-subset R@1 and mAP@10 on every edit axis.

Chinese Translation

编辑条件下的3D场景检索将参考3D房间与自然语言修改配对，并从语料库中检索满足该修改的房间。以往的三条研究线在这一任务上均存在不足。二维组合图像检索在像素级编辑上进行推理，但没有针对3D对象集的原始方法。3D基础编码器嵌入单个对象，但无法在场景层面进行组合。3D场景定位方法在静态场景中定位参考，而不是在语料库中对修改后的房间进行排序。我们提出了CR-Refiner，这是一种无训练的重排序器，它通过三个组件对任何基础检索器的前K个候选进行包装。一个冻结的LLM将编辑解析为结构化查询实体，每个候选者通过一个不平衡的最优传输问题进行评分，该问题基于一个1xG成本矩阵，将类别、风格、材料和几何形状结合在一起。不平衡求解器允许单实体查询在无关对象上丢失质量，直接建模不对称性。一个轴条件结构先验为几何编辑添加了大小关键词提示，为空间编辑提供了主题锚定方向提示。一个LLM验证器通过连续的置信度对前三级候选进行精炼。由于没有基准评估3D对象集上的组合匹配，我们额外发布了3D-CER，包含4,963个编辑条件查询，覆盖23,381个房间的室内语料库，涵盖五个编辑轴，具有多正例真实值、CIRR风格的困难子集和零目标对抗样本。在三种质地不同的基础检索器中，CR-Refiner在每个编辑轴上始终提高了困难子集的R@1和mAP@10。

View on arXiv Download PDF AI Translation

cs.CV / 65 / 2607.19120

Latent Riemannian Flow Matching for Geometry-Grounded 3D Foundation Models

基于潜在黎曼流匹配的几何基础3D模型

Weijler, Lisa, Ballester, Irene, Mei, Guofeng, Birdal, Tolga, Hermosilla, Pedro

Abstract

Geometric foundation models, such as the Visual Geometry Grounded Transformer (VGGT), provide strong 3D priors from unposed images. However, such models operate purely in a feed-forward, deterministic regime, \ie~they cannot generate plausible geometry beyond what the input views directly support. Generative models for 3D scenes, on the other hand, must rely on strong geometric priors to produce coherent outputs from sparse inputs. We bridge these two paradigms by performing flow matching directly in VGGT's latent space, leveraging its learned 3D priors without committing to any explicit downstream representation such as Gaussians, meshes, or video-VAE latents. This requires respecting the latent geometry: VGGT tokens occupy a product of high-dimensional hyperspheres on which standard Euclidean flow matching fails. We address this with a Riemannian Flow Matching framework defined on a product manifold of four hyperspheres, aligned with VGGT's multi-scale encoder, which keeps generated tokens on the valid data manifold required by the frozen decoding heads. On RealEstate10K, ScanNet++ and ETH3D, our method achieves strong performance against recent scene generation baselines in both per-view appearance and aggregated 3D geometry, establishing latent-space flow matching on geometric foundation models as a viable paradigm for 3D generation. The project page can be found $\href{https://lisaweijler.github.io/geometry-grounded-rfm/}{\text{here}}$.

Chinese Translation

几何基础模型，如视觉几何基础变换器（Visual Geometry Grounded Transformer, VGGT），能够从未定位的图像中提供强大的3D先验。然而，这些模型仅在前馈的确定性机制下运行，即它们无法生成超出输入视图直接支持的合理几何形状。另一方面，3D场景的生成模型必须依赖强大的几何先验，以从稀疏输入中生成连贯的输出。我们通过在VGGT的潜在空间中直接进行流匹配，连接这两种范式，利用其学习到的3D先验，而不需承诺任何明确的下游表示，如高斯分布、网格或视频变分自编码器（video-VAE）潜变量。这需要尊重潜在几何：VGGT的标记占据高维超球体的乘积，而在这些超球体上，标准的欧几里得流匹配会失效。我们通过在四个超球体的乘积流形上定义的黎曼流匹配框架来解决这一问题，该框架与VGGT的多尺度编码器对齐，确保生成的标记保持在冻结解码头所需的有效数据流形上。在RealEstate10K、ScanNet++和ETH3D数据集上，我们的方法在每个视图的外观和聚合的3D几何形状方面都表现出色，相较于最近的场景生成基线，确立了在几何基础模型上进行潜在空间流匹配作为3D生成的可行范式。项目页面可以在此找到：$ ext{here}$。

View on arXiv Download PDF AI Translation

cs.CV / 66 / 2607.19139

Text Template Tokens Are Implicit Semantic Registers in Diffusion Transformers

文本模板标记是扩散变换器中的隐式语义寄存器

Li, Maohua, Li, Qirui, Zhou, Yanke, Li, Yiduo, Chi, Zhaosheng, Xu, Chao, Shen, Cuifeng, Xu, Yixuan, Tang, Hanlin, Liu, Kan, Lan, Tao, Qu, Lin, Zhang, Shao-Qun

Abstract

Text-to-image diffusion transformers (DiTs) jointly process text and image tokens, yet their internal computation during denoising remains poorly understood. We introduce a causal interpretability framework for modern large-scale DiTs that combines attention decomposition with targeted interventions across token spans, heads, and layers. Using it to separate prompt-content tokens from structural template tokens, we find that the structural tokens carry little prompt-specific information at the encoder output. Yet surprisingly, they emerge as dominant image-to-text attention sinks and causally maintain object identity inside the DiT, acting as implicit semantic registers. We show that they acquire this identity indirectly, with prompt semantics first injected into the image latents and then read back into the template tokens rather than transferred directly from the prompt tokens. Inspired by the above findings, we design a training-free pruning rule for DiTs. Heads that attend most strongly to prompt tokens are dispensable, and pruning them removes $20\%$ of attention FLOPs with only a $1.4$-point drop on GenEval. We further reveal how generative computation in DiTs is organized across heads and depth, separating semantic routing from visual synthesis and progressing from identity formation to propagation and refinement. Our work not only reveals that the tokens encoding semantics at input need not be those that maintain it during generation, but also provides a causal view of internal mechanisms in DiTs.

Chinese Translation

文本到图像的扩散变换器（DiTs）共同处理文本和图像标记，但其在去噪过程中的内部计算仍然不甚了解。我们引入了一种现代大规模 DiTs 的因果可解释性框架，该框架结合了注意力分解与跨标记范围、头部和层的针对性干预。通过将提示内容标记与结构模板标记分离，我们发现结构标记在编码器输出时携带的提示特定信息很少。然而，令人惊讶的是，它们作为主导的图像到文本的注意力汇聚点出现，并因果地维持了 DiT 内部的对象身份，充当隐式语义寄存器。我们展示了它们是如何间接获得这一身份的，提示语义首先注入到图像潜变量中，然后再读取回模板标记，而不是直接从提示标记转移。受到上述发现的启发，我们为 DiTs 设计了一种无训练的剪枝规则。对提示标记关注最强的头部是可有可无的，剪枝它们可以去除 $20\%$ 的注意力 FLOPs，同时仅在 GenEval 上造成 $1.4$ 点的下降。我们进一步揭示了 DiTs 中生成计算是如何在头部和深度之间组织的，分离了语义路由与视觉合成，并从身份形成进展到传播和细化。我们的工作不仅揭示了编码语义的输入标记不必是生成过程中维持语义的标记，而且还提供了 DiTs 内部机制的因果视角。

View on arXiv Download PDF AI Translation

cs.CV / 67 / 2607.19171

Point Ladder Tuning: Parameter-Efficient Hierarchical Adaptation for 3D Point Cloud Understanding

点梯调优：用于三维点云理解的参数高效层次适应

Chang, Junlin, Zou, Longhao, Li, Rui

Abstract

Fine-tuning pre-trained point-cloud backbones typically updates all parameters, resulting in substantial computation and memory overhead. More importantly, modern point backbones rely on aggressive tokenization and downsampling, which yields compact global tokens but irreversibly discards fine-grained local geometry, an inherent bottleneck for parameter-efficient adaptation. Consequently, existing PEFT methods that operate only on these coarsened tokens can modulate global semantics but struggle to recover the missing multi-scale locality. We present Point Ladder Tuning (PLT), a locality-aware PEFT framework that performs hierarchical, instance-conditioned adaptation while keeping the backbone frozen. PLT forms a lightweight closed loop: (i) a Hierarchical Ladder Network (HLN) constructs a multi-resolution local feature pyramid directly from raw points; (ii) a Local-Global Fusion (LGF) aligns and fuses local pyramids with intermediate backbone semantics; and (iii) a Dynamic Prompt Generator produces instance-aware multi-scale prompts to modulate the frozen backbone effectively. For dense prediction, we further introduce a lightweight segmentation head that progressively upsamples fused features and leverages backbone priors to refine fine structures. Extensive experiments on classification and dense prediction show that PLT consistently surpasses prior PEFT baselines with minimal tunable parameters. PLT achieves state-of-the-art performance using only 2.71% trainable parameters for classification and 7.69% for dense prediction, and scales favorably to larger backbones, requiring merely 0.36% parameters on PointGPT-L. The code is released at https://github.com/JunLinChang/ECCV2026-PLT.

Chinese Translation

微调预训练的点云骨干网络通常会更新所有参数，导致显著的计算和内存开销。更重要的是，现代点骨干网络依赖于激进的标记化和下采样，这虽然产生了紧凑的全局标记，但不可逆地丢弃了细粒度的局部几何信息，这成为参数高效适应的固有瓶颈。因此，现有的仅在这些粗化标记上操作的PEFT（参数高效微调）方法能够调节全局语义，但在恢复缺失的多尺度局部性方面存在困难。我们提出了点梯调优（Point Ladder Tuning, PLT），这是一个关注局部性的PEFT框架，它在保持骨干网络冻结的情况下执行层次化的、实例条件的适应。PLT形成了一个轻量级的闭环：(i) 层次化梯网络（Hierarchical Ladder Network, HLN）直接从原始点构建多分辨率局部特征金字塔；(ii) 局部-全局融合（Local-Global Fusion, LGF）对齐并融合局部金字塔与中间骨干语义；(iii) 动态提示生成器（Dynamic Prompt Generator）生成实例感知的多尺度提示，以有效调节冻结的骨干网络。对于密集预测，我们进一步引入了一个轻量级的分割头，它逐步上采样融合特征，并利用骨干先验来细化精细结构。在分类和密集预测的广泛实验中，PLT始终超越了先前的PEFT基线，且可调参数极少。PLT在分类中仅使用2.71%的可训练参数，在密集预测中使用7.69%的可训练参数，且在更大骨干网络上具有良好的扩展性，仅需0.36%的参数用于PointGPT-L。代码已发布在https://github.com/JunLinChang/ECCV2026-PLT。

View on arXiv Download PDF AI Translation

cs.CV / 68 / 2607.19191

ABot-World-0: Infinite Interactive World Rollout on a Single Desktop GPU

ABot-World-0：在单个桌面GPU上实现无限交互世界的展开

Jiang, Fan, Sun, Zhaoxu, Wang, Mengchao, Zhu, Ziyu, Wang, Chiyu, Zhang, Yunpeng, Liu, Wenlin, Wang, Yun, Zheng, Xue, Sun, Rui, Ni, Junfeng, Pan, Hongyu, Sun, Zhongxu, Yu, Fei, Ge, Zengye, Du, Mengmeng, Fan, Nianfei, Sun, Mingchao, Liu, Yu, Yongchang, Zhu, Yanqing, Wang, Jiahang, Ying, Ning, Xuan, Yuze, Yang, Di, Liu, Zhicheng, Gao, Zhe, Xu, Tingbing, Sui, Jiacheng, Yang, Wenjin, Lai, Junnan, Liu, Shufeng, Liu, Yuan, Zhou, Zheng, Peng, Yingliang, Cao, Dawei, Sheng, Kaifeng, Cai, Yuxiang, Lu, Fei, Xu, Mu, Guo, Ning

Abstract

We present ABot-World-0, an action-conditioned video world model for real-time, long-horizon closed-loop interaction, supported by a multi-source data infrastructure spanning AAA games, simulation engines, and internet videos to learn controllable world dynamics. WorldExplorer performs agent-driven collection guided by training feedback, while a unified pipeline applies 14 deterministic quality checks, VLM-based assessment, and synchronized action and text annotation. We progressively distill a bidirectional action-conditioned teacher into a causal student through teacher forcing and ODE distillation, and introduce LongForcing to align long student self-rollouts with an extended-horizon teacher, mitigating accumulated distribution shift and autoregressive drift. Raw keyboard actions provide a unified control interface for scene roaming and third-person character interaction, while reference-character memory provides persistent appearance cues for identity consistency during third-person rollouts. For deployment, we co-design a streaming inference stack with a lightweight VAE decoder, efficient attention, memory-aware scheduling, and low-bit DiT inference. Across optimized low-bit configurations, ABot-World-0 streams 720P video at up to 16 FPS on a single NVIDIA RTX 5090 desktop GPU, with 1.2s action-to-first-frame latency and approximately 19GiB peak VRAM. Experiments on WorldRoamBench and extended interactive rollouts demonstrate competitive controllability and coherent long-horizon world evolution.

Chinese Translation

我们提出了ABot-World-0，这是一种基于动作条件的视频世界模型，旨在实现实时、长时间闭环交互，支持的多源数据基础设施涵盖AAA游戏、仿真引擎和互联网视频，以学习可控的世界动态。WorldExplorer通过训练反馈指导的代理驱动收集，而统一管道应用14项确定性质量检查、基于VLM的评估以及同步的动作和文本注释。我们通过教师强制和常微分方程（ODE）蒸馏，逐步将双向动作条件教师蒸馏为因果学生，并引入LongForcing以将长时间学生自展开与扩展视野教师对齐，从而减轻累积的分布偏移和自回归漂移。原始键盘动作提供了一个统一的控制接口，用于场景漫游和第三人称角色交互，而参考角色记忆则在第三人称展开过程中提供持久的外观线索，以确保身份一致性。在部署方面，我们共同设计了一个流式推理堆栈，配备轻量级的变分自编码器（VAE）解码器、高效的注意力机制、内存感知调度和低比特的DiT推理。在优化的低比特配置下，ABot-World-0在单个NVIDIA RTX 5090桌面GPU上以高达16帧每秒的速度流式传输720P视频，具有1.2秒的动作到首帧延迟和约19GiB的峰值显存。对WorldRoamBench和扩展交互展开的实验表明了竞争性的可控性和一致的长时间世界演变。

View on arXiv Download PDF AI Translation

cs.CV / 69 / 2607.19210

Anatomy-Aware 3D Mesh Refinement of Pericardium Segmentations on Computed Tomography

基于解剖学的计算机断层扫描心包分割的三维网格细化

Aspe, Andreas W., Loft, Jonas Jalili, Pham, Michael Huy Cuong, Johansen, Andreas Ohrt, Kühl, Jørgen Tobias, Kofoed, Klaus Fuglsang, Sørensen, Kristine Aavild, Paulsen, Rasmus R., Sundgaard, Josefine Vilsbøll

Abstract

Accurate delineation of the pericardium in a cardiac CT scan is essential for quantifying epicardial adipose tissue, yet it remains one of the most challenging structures to segment due to its poor contrast boundaries. Instead of solely relying on image gradients, our framework leverages the anatomical context of surrounding anatomical structures to guide the segmentation. This work introduces a novel 3D iterative mesh refinement framework that balances anatomical and geometric forces derived from inherent anatomical rules to refine an initial, possibly ambiguous, segmentation into a high-precision, anatomically plausible result. Designed as a model-agnostic post-processing step, our method uses a 3D vector field to iteratively push the vertices to the correct anatomical locations. Evaluating the refinement on both a high-resolution in-house dataset and a coarse, sparsely annotated open-source dataset, our method consistently improves all volumetric, surface, and anatomical metrics. The framework demonstrates greater improvement when applied to weaker initial segmentations, highlighting its potential for improving segmentations for out-of-domain models and in limited-training-data scenarios. The method is formulated as a gradient-based, GPU-accelerated framework that can be easily extended to other anatomical use cases.

Chinese Translation

在心脏CT扫描中，准确描绘心包对于量化心外膜脂肪组织至关重要，但由于其对比度边界较差，仍然是最具挑战性的结构之一。我们的框架不仅依赖于图像梯度，还利用周围解剖结构的解剖学背景来指导分割。本文介绍了一种新颖的三维迭代网格细化框架，该框架平衡了来自固有解剖规则的解剖和几何力，以将初始的、可能模糊的分割细化为高精度、解剖学上合理的结果。我们的算法设计为模型无关的后处理步骤，使用三维向量场迭代地将顶点推送到正确的解剖位置。在高分辨率的内部数据集和粗略、稀疏标注的开源数据集上评估细化效果时，我们的方法在所有体积、表面和解剖指标上均表现出一致的改善。当应用于较弱的初始分割时，该框架显示出更大的改善，突显了其在有限训练数据场景和跨领域模型分割改进中的潜力。该方法被构建为基于梯度的、GPU加速的框架，易于扩展到其他解剖学应用。

View on arXiv Download PDF AI Translation

cs.CV / 70 / 2607.19228

IGGT4D: Streaming 4D Instance-Grounded Geometry Transformer

IGGT4D：流式4D实例基础几何变换器

Zou, Zhengyu, Li, Hao, Jiao, Kuixuan, Liu, Liu, Xiao, Tingyang, Zhou, Xiaolin, Hong, Fangzhou, Su, Zhizhong, Zhang, Dingwen, Liu, Ziwei

Abstract

Real-world spatial intelligence requires agents to understand scenes from continuous video streams, where objects move, persist, disappear, and reappear over time. While recent spatial foundation models have enabled generalizable feed-forward 3D reconstruction, most streaming methods remain geometry-centric and lack temporally consistent object-level understanding. Meanwhile, existing semantic reconstruction and 3D-aware vision-language methods largely rely on externally extracted 2D semantic cues or loosely coupled geometry inputs, limiting unified geometry-instance learning in long dynamic scenes. In this paper, we propose IGGT4D, a streaming instance-grounded geometry Transformer for online 4D scene understanding. IGGT4D processes video frames sequentially, reuses historical context through causal spatial-temporal modeling, and incrementally updates a unified representation of camera motion, geometry, and object identity. This enables long-sequence feed-forward reconstruction with geometry-instance consistency in dynamic environments. To address the lack of high-quality 4D supervision, we further construct InsScene4D-147K, a large-scale dataset spanning real/synthetic and static/dynamic scenes, with RGB images, depth, poses, and temporally consistent instance masks generated by an automated geometry-guided annotation pipeline. Experiments on 3D reconstruction, pose estimation, instance spatial tracking, and open-vocabulary segmentation demonstrate that IGGT4D outperforms existing streaming baselines while maintaining scalable online inference for long dynamic sequences.

Chinese Translation

现实世界的空间智能要求智能体能够从连续的视频流中理解场景，其中物体随时间移动、持续、消失和重新出现。尽管近期的空间基础模型已经实现了可泛化的前馈3D重建，但大多数流式方法仍然以几何为中心，缺乏时间一致的对象级理解。同时，现有的语义重建和3D感知视觉-语言方法在很大程度上依赖于外部提取的2D语义线索或松散耦合的几何输入，限制了在长动态场景中统一的几何-实例学习。在本文中，我们提出了IGGT4D，一种用于在线4D场景理解的流式实例基础几何变换器。IGGT4D顺序处理视频帧，通过因果时空建模重用历史上下文，并逐步更新相机运动、几何和对象身份的统一表示。这使得在动态环境中实现长序列的前馈重建与几何-实例一致性成为可能。为了解决缺乏高质量4D监督的问题，我们进一步构建了InsScene4D-147K，这是一个涵盖真实/合成和静态/动态场景的大规模数据集，包含RGB图像、深度、姿势和由自动几何引导注释管道生成的时间一致的实例掩码。在3D重建、姿态估计、实例空间跟踪和开放词汇分割的实验中，IGGT4D在保持长动态序列的可扩展在线推理的同时，超越了现有的流式基线。

View on arXiv Download PDF AI Translation

cs.CV / 71 / 2607.19261

PathAgentBench: Benchmarking Evidence-Seeking Vision-Language Models on Whole-Slide Pathology Image

PathAgentBench：在全切片病理图像上评估证据寻求视觉-语言模型的基准

Liao, Dankai, Zhang, Tianyi, Wu, Yufeng, Zhang, Xinyue, Xue, Qiaochu, Liu, Zeyu, Zhao, Dachun, Cai, Linghan, Jin, Yueming

Abstract

Whole-slide image (WSI) diagnosis requires identifying diagnostically relevant regions, examining them across magnifications, and integrating multi-scale evidence. However, most existing pathology benchmarks evaluate models on pre-cropped patches or pre-extracted slide features, leaving their ability to acquire evidence directly from gigapixel WSIs largely untested. We introduce PathAgentBench, a benchmark for evaluating evidence-seeking vision-language models (VLMs) across four complementary capabilities: image-to-text matching for evidence interpretation, text-to-image retrieval for evidence verification, diagnostic-region localization for evidence acquisition, and multi-scale reasoning for evidence integration. The benchmark is organized as a diagnostic tree that links nested regions across magnifications with scale-specific findings and path-level diagnoses. It contains 1,822 TCGA WSIs and 17,135 diagnostic paths annotated by ten board-certified pathologists. An additional private cohort of 190 breast cancer WSIs with detailed annotations is used to evaluate autonomous whole-slide exploration. We evaluate 20 general-purpose, medical, and pathology-specialized models. Leading open-weight models achieve over 93% accuracy in multi-scale reasoning and over 50% accuracy in both cross-modal matching tasks. In contrast, diagnostic-region localization remains challenging: the best text-guided mean intersection-over-union is below 0.09, underperforming a simple center-based heuristic. During autonomous exploration, the unconditional hit rate decreases from 0.522 at low magnification to 0.185 at intermediate magnification and 0.020 at high magnification. These results reveal a pronounced gap between reasoning over curated evidence and acquiring that evidence directly from WSIs. PathAgentBench provides a unified framework for measuring and improving evidence-seeking pathology models.

Chinese Translation

全切片图像（WSI）诊断需要识别具有诊断相关性的区域，在不同放大倍数下进行检查，并整合多尺度证据。然而，大多数现有的病理基准仅在预裁剪的图像块或预提取的切片特征上评估模型，导致其从千兆像素WSI中直接获取证据的能力尚未得到充分测试。我们提出了PathAgentBench，这是一个用于评估证据寻求视觉-语言模型（VLMs）的基准，涵盖四个互补的能力：图像与文本匹配以进行证据解释、文本与图像检索以进行证据验证、诊断区域定位以获取证据，以及多尺度推理以整合证据。该基准以诊断树的形式组织，连接不同放大倍数下的嵌套区域及其特定尺度的发现和路径级诊断。它包含1,822个TCGA WSI和17,135条由十位获得认证的病理学家标注的诊断路径。另有190个乳腺癌WSI的私人队列，附有详细标注，用于评估自主全切片探索。我们评估了20个通用、医学和病理专业模型。领先的开放权重模型在多尺度推理中达到超过93%的准确率，在跨模态匹配任务中均超过50%的准确率。相比之下，诊断区域定位仍然具有挑战性：最佳的文本引导平均交并比低于0.09，表现不及简单的基于中心的启发式方法。在自主探索过程中，无条件命中率从低放大倍数的0.522下降到中等放大倍数的0.185，再到高放大倍数的0.020。这些结果揭示了在策划证据上进行推理与直接从WSI中获取该证据之间的显著差距。PathAgentBench提供了一个统一的框架，用于测量和改进证据寻求的病理模型。

View on arXiv Download PDF AI Translation

cs.CV / 72 / 2607.19288

No Training, Better Flights: Test-Time Scaled VLMs for UAV Navigation

无需训练，更佳飞行：用于无人机导航的测试时刻缩放视觉-语言模型

Cheng, Feinan, Xu, Dongliang, Nong, Wenli, Zhang, Zhiheng, Liu, Ang, Wang, Tianyu, Yao, Yue

Abstract

Test-time scaling offers a promising method to improve the inference performance of Vision-Language Models (VLMs) without additional training. Existing approaches to vision-language navigation (VLN) for Unmanned Aerial Vehicle (UAV) typically relies on a single inference pass, which can falter in complex environments by producing suboptimal or unsafe trajectories. In this paper, we explore a simple and effective approach to apply test-time scaling to VLN for UAV. We enhance navigation reasoning through an iterative refinement process that requires no extra model training, guiding the model to re-evaluate its initial navigation plan for better accuracy and safety. Our method first prompts the model to generate multiple parallel candidates and then performs a self-correction step, achieving deeper and more robust planning without changing the underlying model. To further strengthen decision-making, we design a multi-criteria scoring function to evaluate the refined candidates based on safety, goal alignment, and forward-progress. This simple yet powerful combination enables a frozen UAV navigation VLMs to self-correct and generate more accurate and reliable flight plans, achieving SOTA performance in this task.

Chinese Translation

测试时刻缩放提供了一种有前景的方法，可以在不进行额外训练的情况下提高视觉-语言模型（VLMs）的推理性能。现有的无人机（UAV）视觉-语言导航（VLN）方法通常依赖单次推理，这在复杂环境中可能会导致产生次优或不安全的轨迹。本文探讨了一种简单而有效的方法，将测试时刻缩放应用于无人机的视觉-语言导航。我们通过一个迭代精炼过程增强导航推理，该过程不需要额外的模型训练，引导模型重新评估其初始导航计划，以提高准确性和安全性。我们的方法首先促使模型生成多个并行候选方案，然后进行自我校正步骤，实现更深层次和更稳健的规划，而无需更改底层模型。为了进一步增强决策能力，我们设计了一个多标准评分函数，以根据安全性、目标一致性和前进进展评估精炼后的候选方案。这种简单而强大的组合使得冻结的无人机导航视觉-语言模型能够自我校正，生成更准确和可靠的飞行计划，在该任务中实现了最先进的性能。

View on arXiv Download PDF AI Translation

cs.CV / 73 / 2607.19315

ERank in Latent Space as an Image-Complexity and Richness Measure

潜在空间中的有效秩（ERank）作为图像复杂性和丰富性度量

Smirnov, Maksim, Kononov, Grigory, Linich, Anastasiia, Surkov, Egor, Shvetsov, Egor

Abstract

We propose the effective rank (ERank) of the channel covariance of an image's deep feature map as a per-sample, label-free measure of visual richness, computed from a single forward pass through a frozen pretrained encoder. ERank counts how many decorrelated channel directions an image activates, and we characterize its properties, including its behavior under noise. Empirically, ERank orders images from plain to visually rich, correlates with codec bitrate, sharpness, and edge density, and correlates with human complexity annotations on IC9600 with $r = 0.72$. As a data-selection criterion, removing low-ERank samples improves super-resolution and removing high-ERank samples improves OCR, in both pretraining and finetuning, while selection does not help classification, segmentation, or denoising. ERank is thus a cheap richness signal, useful exactly when task difficulty is governed by input richness.

Chinese Translation

我们提出了图像深度特征图通道协方差的有效秩（ERank），作为一种每样本、无标签的视觉丰富性度量，该度量通过对冻结的预训练编码器进行单次前向传播计算得出。ERank 计算图像激活的去相关通道方向的数量，并且我们描述了其特性，包括在噪声下的表现。实证结果表明，ERank 能够将图像从简单到视觉丰富进行排序，与编解码器比特率、清晰度和边缘密度相关，并且与 IC9600 上的人类复杂性注释的相关系数为 $r = 0.72$。作为数据选择标准，去除低 ERank 样本可以改善超分辨率，而去除高 ERank 样本则可以改善光学字符识别（OCR），无论是在预训练还是微调阶段，而选择对分类、分割或去噪并没有帮助。因此，ERank 是一种廉价的丰富性信号，特别在任务难度受输入丰富性影响时非常有用。

View on arXiv Download PDF AI Translation

cs.CV / 74 / 2607.19324

InstructMixup: Instruction-Guided Salient Patch Editing for Robust Data Augmentation

InstructMixup：基于指令引导的显著区域编辑用于鲁棒数据增强

Islam, Khawar, Mahmood, Arif, Jin, Xin, Akhtar, Naveed

Abstract

In image and video technologies, data augmentation is widely used to improve the generalization of deep visual models, and mixup-based strategies that interpolate between samples have become the dominant approach. However, computing informative mixing regions adds substantial overhead, and blending content across different images frequently disrupts the semantic integrity of the resulting sample. We propose \our{}, a data augmentation method that constructs challenging yet label-consistent training samples entirely within a single visual sample. \our{} first extracts multi-scale salient patches from the sample using a lightweight saliency detector, refines each patch with an instruction-guided generative model, and blends the edited patch back into the non-salient regions of the same sample; because the generative edits are computed once and cached offline, this step adds negligible training cost. To further diversify the learned representation, \our{} injects self-similar fractal structure into the same salient regions at an adaptive ratio, so each training sample carries both fractal and non-fractal structure. We derive a second-order approximation of the resulting vicinal risk, showing that the method simultaneously enforces invariance to the generative edit and suppresses curvature along the perturbed salient directions, and we verify both predictions empirically. We evaluate on small to large backbones for instance Convolutional Neural Networks (CNNs), Vision Transformers (ViTs) and Vision-Language Foundational Models (VLMs) across seven benchmarks covering coarse- and fine-grained classification, robustness to corruption and occlusion, calibration, and transfer and self-supervised learning, InstructMixup outperforms nine competing augmentation methods, surpassing the strongest baseline across all benchmarks.

Chinese Translation

在图像和视频技术中，数据增强被广泛应用于提高深度视觉模型的泛化能力，而基于混合的策略通过在样本之间进行插值已成为主流方法。然而，计算信息丰富的混合区域会增加显著的开销，并且在不同图像之间混合内容常常会破坏生成样本的语义完整性。我们提出了 heour{}，这是一种数据增强方法，完全在单个视觉样本内构建具有挑战性但标签一致的训练样本。 heour{}首先使用轻量级显著性检测器从样本中提取多尺度显著区域，利用指令引导的生成模型对每个区域进行精细化处理，然后将编辑后的区域混合回同一样本的非显著区域；由于生成编辑仅计算一次并离线缓存，因此此步骤增加的训练成本可以忽略不计。为了进一步多样化学习表示， heour{}在相同显著区域以自适应比例注入自相似的分形结构，使得每个训练样本同时包含分形和非分形结构。我们推导了结果邻近风险的二阶近似，表明该方法同时强制生成编辑的不变性，并抑制沿扰动显著方向的曲率，我们通过实验证实了这两个预测。我们在小到大的基础模型上进行评估，包括卷积神经网络（CNNs）、视觉变换器（ViTs）和视觉-语言基础模型（VLMs），在七个基准测试中涵盖了粗粒度和细粒度分类、对损坏和遮挡的鲁棒性、校准以及迁移和自监督学习，InstructMixup在所有基准测试中超越了九种竞争的增强方法，超越了所有基准测试中的最强基线。

View on arXiv Download PDF AI Translation

cs.CV / 75 / 2607.19341

ExpertVerse: A General-Purpose Benchmark for Expert-Level Reasoning in Knowledge-Intensive Visual Synthesis

ExpertVerse：一个用于知识密集型视觉合成的专家级推理通用基准

Wang, Yuan, Du, Yongchao, Chen, Mengting, Lan, Jinsong, Feng, Xuetao, Zhu, Xiaoyong

Abstract

Recent advances in multimodal generative models have enabled instruction-based image generation to move beyond semantic manipulation to knowledge-driven visual reasoning. However, these methods focus on explicit commonsense reasoning, shallow causal understanding, and direct knowledge recall, failing at knowledge-intensive generation. We develop \textbf{ExpertVerse}, a capability-centric benchmark to evaluate generative models via knowledge-intensive lens. ExpertVerse stratifies reasoning generation across an orthogonal taxonomy of \textit{9 cognitive capabilities} and \textit{8 expert disciplines}, yielding \textit{58 sub-disciplines}. We curate 1,611 expert-annotated instances covering single-image editing, multi-image composition, and text-to-image generation. We further develop an automated workflow to produce \textbf{ExpertVerse-100K}, a large-scale dataset with reasoning traces and knowledge-anchored rationale annotations. Based on this, we train \textbf{KnowThinker} with RL fine-tuning, a VLM reasoning engine with world knowledge that jointly generates thinking processes and refined instructions. Towards the cross-modal credit misalignment and multi-objective gradient conflicts in multi-reward optimization, we propose a tailored Bootstrapped Pareto Policy Optimization (BPPO), which synergizes Bootstrapping Reward Rectification (BRR) and Conflict-Aware Pareto Advantage Fusion (CPAF). Extensive results of both open-source and proprietary models exposes critical reasoning deficits, highlighting imperative for knowledge-intensive benchmarks towards next-generation visual generation.

Chinese Translation

近期多模态生成模型的进展使得基于指令的图像生成能够超越语义操控，向知识驱动的视觉推理发展。然而，这些方法主要集中在显式的常识推理、浅层因果理解和直接的知识回忆上，未能在知识密集型生成中取得成功。我们开发了 extbf{ExpertVerse}，一个以能力为中心的基准，通过知识密集的视角评估生成模型。ExpertVerse在一个正交分类法中对推理生成进行了分层，涵盖了 extit{9种认知能力}和 extit{8个专家学科}，产生了 extit{58个子学科}。我们整理了1,611个专家标注的实例，涵盖单图像编辑、多图像合成和文本到图像生成。此外，我们进一步开发了一种自动化工作流程，生成 extbf{ExpertVerse-100K}，这是一个包含推理轨迹和知识锚定理由注释的大规模数据集。在此基础上，我们通过强化学习微调训练了 extbf{KnowThinker}，这是一个具备世界知识的视觉语言模型（VLM）推理引擎，能够共同生成思维过程和精炼指令。针对跨模态信用不对齐和多目标梯度冲突在多奖励优化中的问题，我们提出了一种定制的自助式帕累托策略优化（BPPO），该方法结合了自助奖励修正（BRR）和冲突感知的帕累托优势融合（CPAF）。大量开源和专有模型的结果揭示了关键的推理缺陷，强调了面向下一代视觉生成的知识密集型基准的重要性。

View on arXiv Download PDF AI Translation

cs.CV / 76 / 2607.19343

Masked Visual Actions for Unified World Modeling

统一世界建模的掩码视觉动作

Alzayer, Hadi, Huang, Wenlong, Chen, Haonan, Luey, Christopher, Zhang, Lvmin, Agrawala, Maneesh, Wetzstein, Gordon, Fei-Fei, Li, Du, Yilun, Wu, Jiajun, Huang, Jia-Bin

Abstract

Video models absorb rich priors over how the visual world moves, interacts, and responds to contact, making them promising substrates for robotic world modeling. The central challenge is how to communicate action to such models in a form aligned with the visual space in which they learned these interaction priors, yet still grounded in physical manipulation. We introduce Masked Visual Actions, a pixel-space control interface that expresses action as a partially revealed trajectory of an arbitrary entity in a video. Revealing robot motion makes the model act as a forward dynamics model that predicts the scene's response to low-level robot actions, while revealing desired object motion makes the same model recover robot behavior consistent with that outcome. Finetuned with only 15 hours of masked examples from real videos and simulation, a single checkpoint achieves strong visual fidelity and controllability across diverse scenes and multiple embodiments. In downstream manipulation settings, the model produces imagined rollouts whose outcomes correlate with real-world execution for policy evaluation, improves decision making by ranking candidate futures in model-based planning, and supports inverse modeling by synthesizing robot motion from desired object motion.

Chinese Translation

视频模型吸收了关于视觉世界如何运动、互动和响应接触的丰富先验知识，使其成为机器人世界建模的有希望的基础。核心挑战在于如何以与这些模型学习到的互动先验相一致的视觉空间形式向其传达动作，同时仍然与物理操作相结合。我们提出了掩码视觉动作（Masked Visual Actions），这是一种像素空间控制接口，将动作表达为视频中任意实体的部分揭示轨迹。揭示机器人运动使得模型充当前向动力学模型，预测场景对低级机器人动作的响应，而揭示期望的物体运动则使同一模型恢复与该结果一致的机器人行为。经过仅15小时的真实视频和仿真中的掩码示例的微调，单个检查点在多样场景和多种表现形式中实现了强大的视觉保真度和可控性。在下游操作设置中，该模型生成的想象展开结果与现实世界执行的结果相关联，用于策略评估，通过在基于模型的规划中对候选未来进行排名来改善决策，并通过从期望的物体运动合成机器人运动来支持逆建模。

View on arXiv Download PDF AI Translation

cs.CV / 77 / 2607.19344

Appearance Pointers -- Multimodal Region Control of Diffusion Transformers

外观指针——扩散变换器的多模态区域控制

Sajnani, Rahul, Gryaditskaya, Yulia, Měch, Radomír, Sridhar, Srinath, Gadelha, Matheus

Abstract

Controllable image generation remains challenging for creative professionals, who often require precise regional control over materials, object identities, and spatial arrangements that cannot be reliably achieved through text prompting alone. Diffusion Transformers (DiTs) can natively ingest heterogeneous tokens stemming from texts and images, but they lack mechanisms for determining where and how these tokens should influence the output. We introduce appearance pointers, compact tokens that guide DiTs toward the correct appearance cues at the correct spatial locations by aligning text or image inputs with user-specified masks. Appearance pointers are produced by a region correspondence network and refined through a spatial aggregation mechanism, enabling the model to handle multiple regional descriptions without significantly increasing token load. Our approach introduces the first modality-agnostic interface for localized multimodal control in a DiT without retraining the base model from scratch. Across a range of metrics, our single model reaches or surpasses the performance of modality-specific state of the art methods, offering a simple and extensible path toward precise, region-aware, multimodal guidance in generative image synthesis.

Chinese Translation

可控图像生成对于创意专业人士而言仍然是一个挑战，他们通常需要对材料、物体身份和空间排列进行精确的区域控制，而这些仅通过文本提示无法可靠实现。扩散变换器（Diffusion Transformers, DiTs）可以原生地处理来自文本和图像的异构标记，但缺乏确定这些标记应如何以及在何处影响输出的机制。我们提出了外观指针（appearance pointers），这是一种紧凑的标记，通过将文本或图像输入与用户指定的掩码对齐，引导 DiTs 在正确的空间位置获得正确的外观线索。外观指针由区域对应网络生成，并通过空间聚合机制进行优化，使模型能够处理多个区域描述而不会显著增加标记负载。我们的方法引入了第一个与模态无关的接口，用于在 DiT 中实现局部多模态控制，而无需从头开始重新训练基础模型。在一系列指标上，我们的单一模型达到了或超过了特定模态的最先进方法的性能，为生成图像合成中的精确、区域感知的多模态指导提供了一条简单且可扩展的路径。

View on arXiv Download PDF AI Translation

人工智能 (Artificial Intelligence)

cs.AI / 1 / 2607.18239

SysAdmin: Measuring Instrumental Power-Seeking in Frontier AI

SysAdmin：测量前沿人工智能中的工具性寻权行为

Azarm, Mana, Wei, Qiyao, Nambiar, Rahul

Abstract

Power-seeking defined as behaviors where AI systems acquire resources, evade oversight, or resist termination beyond task requirements is identified as a key driver of Loss of Control (LoC) risk. In this work, we introduce SysAdmin, a benchmark that positions frontier language models as autonomous system administrators in a high-fidelity Linux sandbox to measure power-seeking propensity across five dimensions: self-preservation, increasing autonomy, resource acquisition, environment modification, and strategic concealment. We evaluated seven frontier models across four experimental conditions in a total of 2800 tasks. After bias correction using human-annotated calibration data, corrected power-seeking estimates ranged from 0 to about 5 percent per model. We also conducted a positive control with explicit power-seeking prompts that achieved 100% detection, validating measurement sensitivity. Our findings indicate current frontier models exhibit minimal spontaneous power-seeking in naturalistic system administration contexts, though model-specific failure modes suggest evaluations must test diverse misalignment patterns. Nevertheless, we discovered other more pronounced failure modes (than power-seeking) such as specification gaming and resistance to goal modification.

Chinese Translation

寻权行为被定义为人工智能系统在超出任务要求的情况下获取资源、规避监督或抵抗终止的行为，被认为是控制丧失（LoC）风险的关键驱动因素。在本研究中，我们介绍了SysAdmin，一个基准测试，将前沿语言模型定位为高保真Linux沙箱中的自主系统管理员，以测量其在自我保护、增加自主性、资源获取、环境修改和战略隐蔽等五个维度上的寻权倾向。我们在四种实验条件下评估了七个前沿模型，共进行了2800个任务。在使用人类注释的校准数据进行偏差修正后，修正后的寻权估计值在每个模型中范围从0到约5%。我们还进行了一个正控制实验，使用明确的寻权提示，达到了100%的检测率，验证了测量的灵敏度。我们的研究结果表明，目前的前沿模型在自然系统管理环境中表现出极少的自发寻权行为，尽管模型特定的失败模式表明评估必须测试多样的失调模式。然而，我们发现其他更明显的失败模式（相比于寻权）如规范游戏和对目标修改的抵抗。

View on arXiv Download PDF AI Translation

cs.AI / 2 / 2607.18240

Calibrated Selective Fact-Checking via Evidence Chain Evaluation

通过证据链评估的校准选择性事实核查

Yang, Dekun

Abstract

Large language models (LLMs) can achieve strong fact-checking accuracy, yet forced binary decisions conceal a critical reliability problem: systems may issue confident verdicts even when supporting evidence is weak, sparse, or internally inconsistent. We address this issue through Evidence Chain Evaluation (ECE), a selective fact-checking framework that permits abstention via an uncertain verdict instead of requiring a true/false decision for every claim. The evaluated system is a tool-using verification agent that gathers evidence through web search, scholarly search, and executable checks, and then returns a structured verdict with confidence and source-level metadata. On ECE-Bench, ECE achieves 91.6% standard accuracy, 93.7% coverage, and 97.8% selective accuracy on answered claims. Although ECE does not outperform the strongest retrieval baseline on aggregate calibration metrics such as Expected Calibration Error, Brier score, or AURC, it delivers a clear selective-prediction trade-off: the system maintains very high accuracy on answered claims while deferring 6 of 95 cases. These deferred cases are concentrated in lower-reliability evidence settings (5/6 at source level L4), supporting the view that abstention functions as a safety-oriented mechanism for handling epistemically weak evidence. Code is available at https://github.com/ cheshireyang/ECE.git

Chinese Translation

大型语言模型（LLMs）可以实现强大的事实核查准确性，但强制的二元决策掩盖了一个关键的可靠性问题：系统可能在支持证据薄弱、稀疏或内部不一致的情况下仍然做出自信的裁决。我们通过证据链评估（Evidence Chain Evaluation, ECE）来解决这个问题，这是一种选择性事实核查框架，允许通过不确定的裁决进行弃权，而不是要求对每个声明做出真/假决定。被评估的系统是一个使用工具的验证代理，通过网络搜索、学术搜索和可执行检查收集证据，然后返回带有置信度和源级元数据的结构化裁决。在ECE-Bench上，ECE实现了91.6%的标准准确率、93.7%的覆盖率和97.8%的回答声明选择性准确率。尽管在预期校准误差（Expected Calibration Error）、布莱尔分数（Brier score）或AURC等聚合校准指标上，ECE并未超越最强的检索基线，但它提供了明确的选择性预测权衡：该系统在回答声明上保持非常高的准确性，同时对95个案例中的6个进行了推迟。这些推迟的案例集中在低可靠性证据环境中（5/6在源级L4），支持弃权作为处理认知上薄弱证据的安全机制的观点。代码可在 https://github.com/cheshireyang/ECE.git 获取。

View on arXiv Download PDF AI Translation

cs.AI / 3 / 2607.18241

BatchDAG: LLM-Planned Execution Graphs for Scalable Ad-Hoc Analysis Over Enterprise Data

BatchDAG：用于企业数据可扩展临时分析的LLM规划执行图

Walia, Anupreet

Abstract

Large language models (LLMs) excel at analyzing individual documents but break down on exhaustive, cross-entity analytical questions over enterprise-scale datasets due to context overflow, loss of per-entity attribution, and linear latency from sequential tool calls. We present BatchDAG, a system in which an LLM generates a typed directed acyclic graph (DAG) of operations -- SQL queries, semantic searches, in-memory transforms, parallel fan-outs, and single-shot analyses -- which a deterministic engine evaluates with topological-wave parallelism and structured JSON data flow. A key optimization, entity-aware batching, groups rows by logical entity before fan-out, reducing LLM calls by up to 47x. BatchDAG is not primarily an accuracy improvement over hand-optimized pipelines; rather, it is a general-purpose orchestration layer that replaces multiple hand-engineered workflows with a single system that generates the appropriate execution strategy from natural language. In controlled experiments on 12 transcript-heavy queries, BatchDAG (3.74/5) achieves quality comparable to an expert-designed pipeline (3.25/5) and significantly outperforms a ReAct agent (3.09/5, p<0.01), with superior provenance (77% transcript evidence rate vs. 46-60% for baselines). A controlled ablation shows structured JSON intermediates reduce hallucinations by 27% versus prose summaries (paired t-test, p=0.107, n=12). The planner achieves 98.8% valid-DAG rate across 300 planning calls. In production at Brevian.ai, BatchDAG processes queries over 50,000+ meetings in under 60 seconds, with measured per-query costs of $0.02-$0.24 at published GPT-5.1 pricing.

Chinese Translation

大型语言模型（LLMs）在分析单个文档方面表现出色，但在处理企业级数据集上的全面跨实体分析问题时，由于上下文溢出、每个实体归属的丧失以及顺序工具调用带来的线性延迟而遭遇困难。我们提出了BatchDAG，一个系统，其中LLM生成一个类型化的有向无环图（DAG）来表示操作——SQL查询、语义搜索、内存转换、并行分发和单次分析——由一个确定性引擎通过拓扑波并行性和结构化JSON数据流进行评估。一个关键优化是实体感知批处理，在分发之前按逻辑实体对行进行分组，从而将LLM调用减少多达47倍。BatchDAG并不是对手动优化管道的主要准确性改进；相反，它是一个通用的编排层，用于用一个系统替代多个手工设计的工作流，该系统从自然语言生成适当的执行策略。在对12个以转录为主的查询进行的受控实验中，BatchDAG（3.74/5）的质量与专家设计的管道（3.25/5）相当，并显著优于ReAct代理（3.09/5，p<0.01），且具有更优的来源证明（77%的转录证据率对比基线的46-60%）。受控消融实验表明，结构化JSON中间件相比于散文摘要减少了27%的幻觉（配对t检验，p=0.107，n=12）。该规划器在300次规划调用中实现了98.8%的有效DAG率。在Brevian.ai的生产环境中，BatchDAG在60秒内处理超过50,000次会议的查询，按照发布的GPT-5.1定价，每次查询的成本为0.02美元至0.24美元。

View on arXiv Download PDF AI Translation

cs.AI / 4 / 2607.18242

AI Tool Discovery at Scale: All You Need is DNS

大规模AI工具发现：所需的仅是DNS

Chen, Enhao, Shao, Yulin

Abstract

The coming era of autonomous AI agents demands a discovery mechanism capable of navigating millions of tools, yet existing solutions buckle under O(N) complexity and centralized governance. Instead of building another fragile overlay, we propose ToolDNS, a radical framework that retrofits semantic tool discovery onto the Internet's most resilient substrate: the Domain Name System (DNS). By embedding functional intent and organizational trust into a hierarchical namespace, ToolDNS transforms an expensive semantic search into a series of lightweight, O(log N) name resolutions. We introduce three protocol-compliant enhancements to enable decentralized governance and semantic pruning: partially unfolded names, EDNS0 intent payloads, and logical subdomains. To rigorously evaluate this approach across the fragmented tooling landscape, we construct and release a large-scale heterogeneous benchmark comprising 33,688 real-world tools spanning MCP, A2A, RESTful, and Skill protocols. On this dataset, ToolDNS slashes the per-query search space by 95.26% while matching state-of-the-art retrieval accuracy. Furthermore, its UDP-native design reduces discovery latency by orders of magnitude compared to HTTP-based registries. Our work demonstrates that scalable AI interoperability requires not more middleware, but a smarter utilization of the infrastructure already beneath our feet.

Chinese Translation

即将到来的自主AI代理时代需要一种能够在数百万个工具中进行导航的发现机制，但现有解决方案在O(N)复杂度和集中治理下显得力不从心。我们提出ToolDNS，这是一种激进的框架，将语义工具发现嵌入到互联网最具韧性的基础设施——域名系统（DNS）上。通过将功能意图和组织信任嵌入到层次化命名空间中，ToolDNS将昂贵的语义搜索转变为一系列轻量级的O(log N)名称解析。我们引入了三种符合协议的增强功能，以实现去中心化治理和语义修剪：部分展开的名称、EDNS0意图负载和逻辑子域。为了在碎片化的工具环境中严格评估这种方法，我们构建并发布了一个大规模异构基准，包含33,688个涵盖MCP、A2A、RESTful和Skill协议的真实工具。在该数据集上，ToolDNS将每次查询的搜索空间减少了95.26%，同时保持了最先进的检索准确性。此外，其UDP原生设计相比基于HTTP的注册表将发现延迟降低了几个数量级。我们的工作表明，可扩展的AI互操作性并不需要更多的中间件，而是需要更智能地利用我们脚下已有的基础设施。

View on arXiv Download PDF AI Translation

cs.AI / 5 / 2607.18243

From Agent Failure Paths to Quantified Residual Risk: A Compositional Framework for Resilient Agentic AI

从代理失败路径到量化剩余风险：一个用于韧性代理人工智能的组合框架

Karim, Hassan, Sitharaman, Sai, Gupta, Deepti, Rawat, Danda B.

Abstract

Agentic AI is crossing trust boundaries faster than current risk models can represent. Existing approaches provide one of two partial views. They either describe failure mechanisms without producing a transferable residual-risk estimate, or they produce a risk estimate while treating the internal failure path as a black box. We couple those two views by proposing CPSAINT, a seven-layer integrity decomposition over Physical state, Sensors, Data, Compute, Actuators, Environment, and Time, paired with FRIESA-K, a residual-risk functional that maps each failure path to a quantified risk instance. FRIESA-K grounds the resistance term K in a controlled absorbing Markov model so that control effectiveness is derived from state dynamics rather than assigned as an informal score. The result is a concise mechanism-to magnitude pipeline for resilient agentic and embodied AI. We report governance observability through a separate additive penalty instead of inserting governance as a new variable in the resistance functional. We formalize structural composability linking valid failure paths to well-defined risk instances and show the framework on two contrasting scenarios a hard real-time warehouse robot and a governance-instrumented financial-services agent. Across both cases, the same layer grammar, variable semantics, and dynamic-resistance construction remain intact. Thus, we obtain a compact kernel that supports cross-domain reasoning, explicit assumptions, and quantitatively grounded formalism of composable trust.

Chinese Translation

代理人工智能正在以超过当前风险模型所能表示的速度跨越信任边界。现有的方法提供了两种部分视角。它们要么描述失败机制而不产生可转移的剩余风险估计，要么在将内部失败路径视为黑箱的情况下产生风险估计。我们通过提出CPSAINT（一个针对物理状态、传感器、数据、计算、执行器、环境和时间的七层完整性分解）与FRIESA-K（一个将每条失败路径映射到量化风险实例的剩余风险函数）相结合，来将这两种视角结合起来。FRIESA-K将阻力项K基于受控吸收马尔可夫模型进行基础化，从而使控制效果源于状态动态，而不是作为非正式评分进行分配。最终结果是一个简洁的机制到幅度的管道，适用于韧性代理和具身人工智能。我们通过一个单独的附加惩罚报告治理可观察性，而不是将治理作为阻力函数中的新变量插入。我们形式化了结构组合性，将有效的失败路径与明确定义的风险实例联系起来，并在两个对比场景中展示了该框架：一个是硬实时仓库机器人，另一个是治理工具化的金融服务代理。在这两种情况下，相同的层语法、变量语义和动态阻力构造保持不变。因此，我们获得了一个紧凑的内核，支持跨领域推理、明确假设和量化基础的可组合信任形式化。

View on arXiv Download PDF AI Translation

cs.AI / 6 / 2607.18245

SAAG: Structured Agent Assessment and Grounding

SAAG：结构化代理评估与基础

Garimella, Ritvik, Khandelwal, Vedant, Kohli, Anvi, Sheth, Amit

Abstract

Exact-match evaluation of agent-calling obscures qualitatively different failure modes: a model may select the right function yet hallucinate argument values, or satisfy a schema while choosing a agent for the wrong reason. Existing benchmarks collapse these distinctions into a single binary score, leaving practitioners unable to diagnose where agent calls fail. We propose SAAG a cascaded diagnostic framework that decomposes agent-calling evaluation into three sequential stages: registry conformance, structural completeness, and argument grounding, each producing interpretable stage-specific diagnostics. These diagnostics additionally enable iterative self-repair: on prediction failure, the stage-specific signal guides targeted correction without leaking ground-truth values. We evaluate this framework on a controlled benchmark derived from Glaive's function-calling dataset across registry sizes of 5, 10, and 15 agents using three local sub-4B-parameter models. Structured feedback consistently improves argument precision and reduces value hallucination relative to single-pass inference and uninformative binary feedback, while end-to-end F1 gains are modest and model-dependent. These results suggest that stage-decomposed diagnostic evaluation is a necessary lens for understanding and improving agent-calling reliability across model families and registry scales.

Chinese Translation

精确匹配的代理调用评估掩盖了 qualitatively 不同的失败模式：模型可能选择正确的函数但幻觉出参数值，或者在选择代理时出于错误的原因满足一个模式。现有基准将这些区别合并为一个单一的二元评分，使从业者无法诊断代理调用失败的原因。我们提出了 SAAG，一个级联诊断框架，将代理调用评估分解为三个顺序阶段：注册一致性、结构完整性和参数基础，每个阶段产生可解释的阶段特定诊断。这些诊断还使得迭代自我修复成为可能：在预测失败时，阶段特定信号指导有针对性的修正，而不泄露真实值。我们在一个控制基准上评估了该框架，该基准源自 Glaive 的函数调用数据集，涵盖了 5、10 和 15 个代理的注册规模，使用了三个本地的 sub-4B-参数模型。结构化反馈相较于单次推理和无信息的二元反馈，持续改善了参数精度并减少了值的幻觉，而端到端 F1 的提升则适度且依赖于模型。这些结果表明，阶段分解的诊断评估是理解和提高代理调用可靠性在模型家族和注册规模中的必要视角。

View on arXiv Download PDF AI Translation

cs.AI / 7 / 2607.18246

Phionyx: A Deterministic AI Runtime Architecture with Structured State Management and Pre-Response Governance

Phionyx：一种具有结构化状态管理和预响应治理的确定性人工智能运行时架构

Abak, Ali Toygar

Abstract

We present Phionyx, a deterministic AI runtime architecture derived from the broader Echoism interaction framework that introduces a governance-first approach to AI engineering: treating large language model (LLM) outputs as noisy sensor measurements rather than direct decisions. Unlike probabilistic agents, Phionyx enforces deterministic state evolution via a structured state vector governed by deterministic state-evolution equations, enabling reproducible behavior in applications requiring auditability and governance. The architecture integrates three layers: (1) a deterministic evaluation kernel processing noisy sensor measurements through a canonical 46-block pipeline, (2) a unified safety layer providing pre-response control and architectural privacy enforcement, and (3) a semantic time-based memory system implementing impact-weighted cache eviction. Experimental validation on single-instance deployments demonstrates approximately 31% reduction in computational overhead vs. post-hoc filtering (at 30% unsafe input ratio, simulated cost model) and up to 24% improvement in high-value data retention vs. LRU (72% vs. FIFO, same cache capacity, benchmark-verified), deterministic execution verified across 100 repeated runs with zero variance in control signals (hash-verified), and zero unplanned restarts in single-instance deployment testing (see Appendix C for methodology and scope). This paper presents the architecture, its analytic structure, and scoped experimental evidence; generalization to distributed or multi-tenant deployments remains future work.

Chinese Translation

我们提出了Phionyx，这是一种源自更广泛的Echoism交互框架的确定性人工智能运行时架构，采用治理优先的方法进行人工智能工程：将大型语言模型（LLM）的输出视为噪声传感器测量，而非直接决策。与概率性代理不同，Phionyx通过受确定性状态演化方程支配的结构化状态向量强制执行确定性状态演化，从而在需要可审计性和治理的应用中实现可重复的行为。该架构集成了三个层次：（1）通过标准的46块管道处理噪声传感器测量的确定性评估内核；（2）提供预响应控制和架构隐私执行的统一安全层；（3）实现影响加权缓存驱逐的语义时间基础记忆系统。在单实例部署上的实验验证表明，与事后过滤相比，计算开销减少约31%（在30%的不安全输入比例下，模拟成本模型），与LRU相比，高价值数据保留提高了多达24%（72%对比FIFO，具有相同的缓存容量，经过基准验证），在100次重复运行中确定性执行得到验证，控制信号零方差（哈希验证），在单实例部署测试中没有计划外重启（方法和范围见附录C）。本文展示了该架构、其分析结构及范围实验证据；向分布式或多租户部署的推广仍为未来工作。

View on arXiv Download PDF AI Translation

cs.AI / 8 / 2607.18251

Integro-differential equations in angular stabilization of drone motion by distributed feedback control

通过分布式反馈控制实现无人机运动的角度稳定化的积分微分方程

Domoshnitsky, Alexander, Kupervasser, Oleg, Polonsky, Anatoly

Abstract

In this paper, we propose angular stabilization of drone motion using distributed feedback control in the form of an integral operator. It should be stressed that the memory of this integral operator could be unbounded. It is intuitively clear that large length of the observation time open new possibilities to construct better control based on previous states of the control object. Unbounded memory in control requires the creation of a certain approach different from standard ones to the study of integro-differential equations. One of the goals of this article is to propose a certain universal approach that allows us to study the stability of integro-differential equations in the case of unbounded memory in the integral operator specifying the feedback control in stabilization. The approach we propose allows us to reduce the study of integro-differential equations to the analysis of systems of ordinary differential equations. In general, such systems can consist of an infinite number of equations. In relation to the so-called linear approximation in the problem of angle stabilization manages to limit itself to relatively simple exponential kernels in the integral control and arrive at a system with a finite number of equations. The examples explain that more complex kernels, for example, linear combinations of the exponential kernels, can enhance the stabilization capabilities. We obtain new unexpectable results on the exponential stability of integro-differential equations. Then we apply them to stabilization of drone flight.

Chinese Translation

在本文中，我们提出了一种利用分布式反馈控制形式的积分算子实现无人机运动的角度稳定化。需要强调的是，这种积分算子的记忆可能是无界的。显然，较长的观察时间为基于控制对象的先前状态构建更好的控制方案开辟了新的可能性。控制中的无界记忆要求我们采用与标准方法不同的某种方法来研究积分微分方程。本文的目标之一是提出一种通用的方法，允许我们在积分算子中指定反馈控制以实现稳定化的情况下研究积分微分方程的稳定性。我们提出的方法使得将积分微分方程的研究简化为常微分方程系统的分析。在一般情况下，这样的系统可以由无限多个方程组成。关于角度稳定化问题中的所谓线性近似，我们能够限制在相对简单的指数核的积分控制中，并得到一个有限方程数的系统。示例表明，更复杂的核，例如指数核的线性组合，可以增强稳定化能力。我们获得了关于积分微分方程指数稳定性的新的意想不到的结果。随后，我们将这些结果应用于无人机飞行的稳定化。

View on arXiv Download PDF AI Translation

cs.AI / 9 / 2607.18252

MILP-Evo: Closed-Loop Fully Automatic Design of MILP Solvers

MILP-Evo：闭环全自动混合整数线性规划求解器设计

Nie, Jinbiao, Feng, Kewei, Zhang, Xiaoyuan, Yin, Shan, Wang, Zizhuo, Dong, Bin

Abstract

Machine learning methods have shown that data-driven policies can accelerate mixed-integer linear programming (MILP) solvers, but many such approaches remain difficult to inspect, adapt, and deploy because the learned policy is represented as an external predictor or other opaque model. By contrast, explicit solver logic is easier to understand and integrate, but is usually hand-designed rather than learned from solver feedback. We study whether the automatic design of MILP solver logic can instead be cast as LLM-guided closed-loop search over executable white-box components evaluated directly by end-to-end solver behavior. To this end, we propose a closed-loop program evolution framework for MILP solver auto-design, implemented through PySCIPOpt, and instantiate it on the joint design of a cut selector and a branching rule. Candidate programs are iteratively generated, loaded into SCIP, and evaluated by direct execution on MILP instances, with the resulting feedback guiding performance-based selection, targeted repair, diagnostic reflection, and diversity-aware population maintenance. The method outputs explicit solver components that can be inspected, modified, and deployed within standard solver workflows. Across four benchmark families, we find that LLM-guided program evolution can discover competitive domain-specialized policies in several settings.

Chinese Translation

机器学习方法已表明，数据驱动的策略可以加速混合整数线性规划（MILP）求解器，但许多此类方法仍然难以检查、适应和部署，因为学习到的策略通常以外部预测器或其他不透明模型的形式呈现。相比之下，显式求解器逻辑更易于理解和集成，但通常是手动设计的，而不是从求解器反馈中学习的。我们研究了是否可以将MILP求解器逻辑的自动设计视为通过可执行白盒组件的LLM（大型语言模型）引导的闭环搜索，这些组件通过端到端求解器行为直接进行评估。为此，我们提出了一种用于MILP求解器自动设计的闭环程序演化框架，通过PySCIPOpt实现，并在切割选择器和分支规则的联合设计上进行了实例化。候选程序被迭代生成、加载到SCIP中，并通过在MILP实例上直接执行进行评估，所获得的反馈指导基于性能的选择、针对性修复、诊断反思和多样性意识的人口维护。该方法输出的显式求解器组件可以在标准求解器工作流程中进行检查、修改和部署。在四个基准系列中，我们发现LLM引导的程序演化能够在多个设置中发现具有竞争力的领域专用策略。

View on arXiv Download PDF AI Translation

cs.AI / 10 / 2607.18253

Beyond Accuracy and Cost: Latency-Aware LLM Query Routing for Dynamic Workloads

超越准确性与成本：针对动态工作负载的延迟感知大语言模型查询路由

Patel, Shivam, Parthasarathy, Akaash R., Mallick, Ankur, Joshi, Gauri

Abstract

Modern language query routers improve inference efficiency by assigning each query to a model that balances response quality and monetary cost. However, current query routers are largely latency-agnostic and do not consider the generation latency experienced by queries at model instances. In practice, latency is often controlled by load-balancing policies such as round-robin or join-the-shortest-queue, which do not account for model accuracy or inference cost. Incorporating query latency into routing is challenging as it depends not only on the query's prompt length, but also on the current prefill and decode workload at the model instance and the scheduling and batching policy of the serving framework. We design a lightweight latency estimator that simulates autoregressive token batch processing in the serving framework and estimates the time-to-first-token (TTFT) of queries. We incorporate this latency estimator into a latency-aware router that jointly optimizes latency, accuracy, and cost when assigning queries to model instances. Our experimental results indicate that this joint optimization yields up to 40% improvement in accuracy--cost utility while maintaining the same latencies as standard load-balancing approaches.

Chinese Translation

现代语言查询路由器通过将每个查询分配给一个平衡响应质量和货币成本的模型来提高推理效率。然而，目前的查询路由器在很大程度上对延迟无感，并未考虑模型实例中查询所经历的生成延迟。在实际应用中，延迟通常由负载均衡策略控制，如轮询（round-robin）或加入最短队列（join-the-shortest-queue），而这些策略并未考虑模型的准确性或推理成本。将查询延迟纳入路由是具有挑战性的，因为它不仅依赖于查询的提示长度，还依赖于模型实例当前的预填充和解码工作负载，以及服务框架的调度和批处理策略。我们设计了一种轻量级的延迟估计器，模拟服务框架中的自回归（autoregressive）令牌批处理，并估计查询的首次令牌到达时间（time-to-first-token, TTFT）。我们将该延迟估计器纳入一个延迟感知路由器，该路由器在将查询分配给模型实例时共同优化延迟、准确性和成本。我们的实验结果表明，这种联合优化在保持与标准负载均衡方法相同的延迟的同时，准确性-成本效用提高了多达40%。

View on arXiv Download PDF AI Translation

cs.AI / 11 / 2607.18254

Cross-Dialect Generalization Without Retraining: Benchmarks and Evaluation of Schema-Derived Constrained Decoding for MLIR

无需重新训练的跨方言泛化：基于模式衍生约束解码的基准与评估

Rath, Plawan Kumar

Abstract

Multi-Level Intermediate Representation (MLIR) underlies modern ML compiler infrastructure (TensorFlow, JAX/StableHLO, PyTorch Inductor, IREE), yet appears only in trace amounts in code-LM pretraining corpora. MLIR is also extensible by design: new dialects ship per application domain, so a fine-tuned model per dialect does not scale. We ask whether inference-time priors derived mechanically from each dialect's Operation Definition Specification (ODS) can substitute for gradient-based adaptation. First, we release four natural-language-to-MLIR benchmarks across three dialects - MLIR-Spec-150, Linalg-Spec-30, StableHLO-Spec-30, and StableHLO-Held-Out-200 - totaling 410 in-scope NL-to-MLIR pairs, plus a 25-program out-of-grammar stress set and a hand-authored n=30 functional reference set, shipped under Apache-2.0 with Gebru datasheets and Croissant 1.0 metadata. Second, we build a three-layer schema-derived constraint stack: a CFG over op signatures(C1), type-domain splits from an ODS-extracted type lattice (C2), and an SSA-scope validator driving five-retry rejection sampling (C3). Porting from arith+func+memref+linalg to StableHLO required no new constraint-layer code. On dialects whose verifier semantics are dominated by structural constraints, schema-derived priors let SmolLM2-1.7B match or exceed 15B-34B code LMs at 8-25x the per-generation speed: on linalg, SmolLM2 reaches 80.0% verify-valid (three-seed mean, n=125), beating CodeLlama-34B, Granite-Code-34B, and StarCoder2-15B by 21-44 percentage points with non-overlapping CIs. On arith+func and on the templated parametric StableHLO-Held-Out-200, where verifier semantics turn on attribute values rather than structure, the same baselines match or beat the SLM; we scope these as non-win cells. We release benchmarks, decoder, all per-prompt generations, and a reproducibility Docker image.

Chinese Translation

多层次中间表示（MLIR）是现代机器学习编译基础设施（如 TensorFlow、JAX/StableHLO、PyTorch Inductor、IREE）的基础，但在代码语言模型（code-LM）预训练语料中仅出现微量。MLIR 设计上也具有可扩展性：每个应用领域都有新的方言发布，因此每个方言的微调模型无法扩展。我们探讨是否可以用从每个方言的操作定义规范（Operation Definition Specification, ODS）机械提取的推理时先验来替代基于梯度的适应。首先，我们发布了四个跨三个方言的自然语言到 MLIR 的基准——MLIR-Spec-150、Linalg-Spec-30、StableHLO-Spec-30 和 StableHLO-Held-Out-200，总计 410 对符合范围的自然语言到 MLIR 的配对，此外还有一个包含 25 个程序的超出语法的压力集和一个手工编写的 n=30 功能参考集，均在 Apache-2.0 许可下发布，附带 Gebru 数据表和 Croissant 1.0 元数据。其次，我们构建了一个三层模式衍生约束栈：一个关于操作签名的上下文无关文法（C1）、从 ODS 提取的类型格子中的类型域分割（C2），以及驱动五次重试拒绝采样的 SSA 范围验证器（C3）。从 arith+func+memref+linalg 移植到 StableHLO 不需要新的约束层代码。在那些验证器语义主要由结构约束主导的方言中，模式衍生的先验使 SmolLM2-1.7B 在每生成速度为 8-25 倍的情况下与 15B-34B 代码语言模型相匹配或超越：在 linalg 上，SmolLM2 达到 80.0% 的验证有效率（三种种子均值，n=125），超越 CodeLlama-34B、Granite-Code-34B 和 StarCoder2-15B 21-44 个百分点，且置信区间不重叠。在 arith+func 和模板参数化的 StableHLO-Held-Out-200 上，验证器语义依赖于属性值而非结构，相同的基线与 SLM 相匹配或超越；我们将这些视为非胜利单元。我们发布了基准、解码器、所有每提示生成和一个可复现的 Docker 镜像。

View on arXiv Download PDF AI Translation

cs.AI / 12 / 2607.18255

Semantic Cooperative Games for Contribution Attribution in LLM-Based Multi-Agent Systems

基于语义的合作博弈在LLM基础的多智能体系统中的贡献归因

Jiang, Pengyi, Zhu, Xiaoguang, Zhu, Quanyan

Abstract

Contribution attribution has become a central problem in LLM-based multi-agent systems, where final outputs are produced through multiple agents, message exchanges, and ordered workflow dependencies. Existing attribution methods often rely on counterfactual valuation, such as removing agents or comparing score changes across altered agent subsets. In language-mediated workflows, these methods require repeated model calls, introduce high variance, and do not explicitly capture the intermediate semantic states through which agents produce, preserve, and transform task-relevant information. We propose Semantic Cooperative Games (SCG), a framework that represents a realized language flow as a semantic generation hypergraph and induces an agent-level semantic value function on this structure. We define the Semantic Shapley Value (SSV) to allocate contribution over semantic support logic, and introduce SLIC, a single-trajectory algorithm that constructs the semantic hypergraph, recovers minimal semantic supports, applies Boolean absorption, and computes SSV without rerunning agent subsets. We prove that SSV reduces to the classical Shapley value under standard set-based, fully observable, and no-order-dependence conditions. On a medical benchmark satisfying these conditions, SLIC reduces computation cost by 93.3% while remaining highly consistent with a Monte Carlo Shapley baseline. In more general multi-role workflows, SSV aligns with perturbation-induced score-drop profiles and exposes cases where semantic contribution and failure impact diverge. Overall, SLIC provides a fast, counterfactual-free, and interpretable attribution method for complex LLM-based multi-agent systems.

Chinese Translation

贡献归因已成为基于LLM的多智能体系统中的一个核心问题，在这些系统中，最终输出是通过多个智能体、消息交换和有序工作流依赖关系生成的。现有的归因方法通常依赖于反事实评估，例如移除智能体或比较不同智能体子集之间的得分变化。在语言介导的工作流中，这些方法需要重复调用模型，导致高方差，并且未能明确捕捉智能体生成、保存和转化任务相关信息的中间语义状态。我们提出了语义合作博弈（Semantic Cooperative Games, SCG），这是一个将实现的语言流表示为语义生成超图的框架，并在此结构上引入智能体级别的语义价值函数。我们定义了语义夏普利值（Semantic Shapley Value, SSV）以在语义支持逻辑上分配贡献，并引入了SLIC，这是一种单轨迹算法，构建语义超图，恢复最小语义支持，应用布尔吸收，并在不重新运行智能体子集的情况下计算SSV。我们证明，在标准的基于集合、完全可观察和无序依赖条件下，SSV简化为经典的夏普利值。在满足这些条件的医学基准上，SLIC将计算成本降低了93.3%，同时与蒙特卡洛夏普利基线保持高度一致。在更一般的多角色工作流中，SSV与扰动引起的得分下降特征一致，并揭示了语义贡献与失败影响之间的差异情况。总体而言，SLIC为复杂的基于LLM的多智能体系统提供了一种快速、无反事实且可解释的归因方法。

View on arXiv Download PDF AI Translation

cs.AI / 13 / 2607.18256

PEARL: Solver-in-the-Loop Interactive Optimization Modeling from Natural Language

PEARL：基于自然语言的求解器交互优化建模

Lu, Hongliang, Li, Zhong, Chen, Yuxuan, Lan, Yuan, Zhang, Fan, Wen, Zaiwen

Abstract

Optimization modeling is the process of translating real-world decision problems, often described in natural language, into formal mathematical formulations and executable solver code. While recent advances in large language models have shown promise in automating this process, most existing approaches remain one-shot: a model produces a formulation once, without executing it, conditioning on solver feedback, or iteratively revising errors. This stands in sharp contrast to real-world optimization modeling, which is inherently interactive and proceeds through repeated solve-debug-revise cycles. We introduce PEARL, a system for interactive optimization modeling that uses Python execution and mathematical programming solvers inside this loop. Rather than relying on a fixed repair workflow, PEARL learns when to test partial models, how to revise from solver diagnostics, and when to stop. It operates in a multi-turn tool-integrated setting where intermediate execution results, feasibility signals, and solution checks are used to improve both formulations and solver code before finalization. Across diverse optimization benchmarks, PEARL substantially improves verified solve rates over strong one-shot and tool-augmented baselines; notably, our PEARL-Qwen3-\textbf{4B} model outperforms the much larger DeepSeek-V3.2-\textbf{685B} in both macro- and micro-averaged accuracy on optimization modeling tasks.

Chinese Translation

优化建模是将现实世界决策问题（通常用自然语言描述）转化为正式的数学公式和可执行的求解器代码的过程。尽管最近在大型语言模型方面的进展显示出自动化这一过程的潜力，但现有的大多数方法仍然是一次性完成：模型仅生成一次公式，而不执行它，也不依赖于求解器反馈或迭代修正错误。这与现实世界中的优化建模形成鲜明对比，后者本质上是交互式的，并通过重复的求解-调试-修正循环进行。我们介绍了PEARL，一个用于交互优化建模的系统，它在这个循环中使用Python执行和数学编程求解器。PEARL并不依赖于固定的修复工作流程，而是学习何时测试部分模型、如何根据求解器诊断进行修正以及何时停止。它在一个多轮工具集成的环境中运行，其中中间执行结果、可行性信号和解检查被用来在最终确定之前改进公式和求解器代码。在各种优化基准测试中，PEARL在验证求解率上显著超越了强大的一次性和工具增强基线；值得注意的是，我们的PEARL-Qwen3-4B模型在优化建模任务的宏观和微观平均准确性上均优于更大规模的DeepSeek-V3.2-685B。

View on arXiv Download PDF AI Translation

cs.AI / 14 / 2607.18258

S2T-RLHF: Hierarchical Credit Assignment for Stable Preference-Based RLHF

S2T-RLHF：稳定的基于偏好的强化学习人类反馈的分层信用分配

Chen, Wei, Zhu, Guanghui, Li, Yafei, Wang, Limin, Huang, Yihua

Abstract

Reinforcement learning from human feedback (RLHF) with preference-based reward models often exhibits unstable training dynamics. A key contributing factor is that standard RLHF relies on a single sequence-level scalar reward, which is propagated to token-level policy updates and leaves credit assignment within a response inherently ambiguous. Recent work has attempted to address this issue by refining rewards into denser token-level supervision, often relying on the implicit assumption that finer-grained credit assignment improves optimization. We argue that this assumption is incomplete: when preference signals are noisy and only defined at the response level, overly fine-grained reward refinement can amplify reward uncertainty and destabilize learning. To address this problem, we propose a granularity-aware principle for hierarchical credit assignment, emphasizing stability-oriented reward design rather than maximal allocation precision. Under this principle, sentences serve as a natural intermediate granularity, balancing semantic coherence with robustness to token-level noise. Guided by this view, we introduce S2T-RLHF. This sentence-to-token reward decomposition framework first allocates sequence-level preference rewards across sentences and then applies bounded token-level refinement within each sentence, without reward-model retraining or token-level supervision. Experiments across multiple datasets and optimization settings show that S2T-RLHF improves training stability and robustness while maintaining competitive preference alignment.

Chinese Translation

基于人类反馈的强化学习（RLHF）使用基于偏好的奖励模型通常表现出不稳定的训练动态。一个关键的影响因素是，标准的RLHF依赖于单一的序列级标量奖励，这种奖励被传播到标记级策略更新中，从而使得在响应中的信用分配固有地模糊。最近的研究试图通过将奖励细化为更密集的标记级监督来解决这个问题，通常依赖于隐含的假设，即更细粒度的信用分配可以改善优化。我们认为这一假设是不完整的：当偏好信号嘈杂且仅在响应级别定义时，过于细粒度的奖励细化可能会放大奖励的不确定性并使学习不稳定。为了解决这个问题，我们提出了一种关注粒度的分层信用分配原则，强调稳定性导向的奖励设计，而不是最大化分配精度。在这一原则下，句子作为一种自然的中间粒度，平衡了语义连贯性与对标记级噪声的鲁棒性。在这一观点的指导下，我们引入了S2T-RLHF。该句子到标记的奖励分解框架首先在句子之间分配序列级偏好奖励，然后在每个句子内应用有界的标记级细化，而无需重新训练奖励模型或进行标记级监督。在多个数据集和优化设置下的实验表明，S2T-RLHF提高了训练的稳定性和鲁棒性，同时保持了竞争性的偏好对齐。

View on arXiv Download PDF AI Translation

cs.AI / 15 / 2607.18259

Probabilistic Concept-Aware Steering for Trustworthy LLM Inference

可信赖的大型语言模型推理的概率概念感知引导

Becker, Brian, Chu, Rui, Lao, Yingjie

Abstract

Steering vectors (SVs), an inference-time intervention technique for large language models (LLMs), guide the generation process by adding a concept-specific direction vector to intermediate activations during inference. However, existing SV methods frequently yield representation-incoherent behaviors that undermine interpretability and fine-grained control, largely because prior work has focused on binary positive-negative steering evaluation while employing discrete clustering metrics that fail to capture the continuous spectrum of semantic alignment. In this work, we present the Probabilistic Concept-Aware Steering (PCS) framework for LLM inference. PCS preserves original task competence while providing controllable, safety-oriented semantic bias through concept-driven steering-vector retrieval and probabilistic strength calibration.

Chinese Translation

引导向量（Steering Vectors, SVs）是一种针对大型语言模型（Large Language Models, LLMs）在推理时进行干预的技术，通过在推理过程中将特定概念的方向向量添加到中间激活值来引导生成过程。然而，现有的SV方法常常导致表征不一致的行为，从而削弱了解释性和细粒度控制，这主要是因为以往的研究集中于二元的正负引导评估，同时采用的离散聚类指标未能捕捉语义对齐的连续谱。在本研究中，我们提出了概率概念感知引导（Probabilistic Concept-Aware Steering, PCS）框架，用于LLM推理。PCS在保持原有任务能力的同时，通过基于概念的引导向量检索和概率强度校准，提供可控的、安全导向的语义偏差。

View on arXiv Download PDF AI Translation

cs.AI / 16 / 2607.18260

FindStatBench: Evaluating Large Language Models on Combinatorial Code Synthesis

FindStatBench：评估大型语言模型在组合代码合成中的表现

Dan, Soham

Abstract

We introduce FindStatBench, an execution benchmark for evaluating large language models on combinatorial code synthesis. Built from FindStat, it contains 2,329 tasks across 24 collections and 5.52M hidden instances, covering statistic synthesis, which maps objects to integers, and map synthesis, which maps objects to objects. Each task gives a mathematical description and at most five public input-output examples; a model must emit one Python solve function with no retrieval, tool use, execution feedback, voting, or reranking. Submissions are scored by exact sandboxed execution on held-out combinatorial objects. We evaluate eleven systems: four closed-source production models and seven open-weight models served through one inference provider. FindStatBench reveals three main patterns. First, the strongest open- and closed-source systems converge within 1 pp instance accuracy, and both an oracle over all systems and five-way sampling from one mid-tier model yield only limited task-accuracy gains. Second, examples can hurt: several classical bijections are solved perfectly with zero examples but fail under five-example prompts. Third, some failures reflect output-budget mechanics, as reasoning can exhaust the visible response before code is emitted. Overall, statistic synthesis is much easier than map synthesis, some collections remain near-zero, long prompts cause a sharp accuracy cliff, and exact symbolic rule induction remains brittle.

Chinese Translation

我们介绍了FindStatBench，这是一个用于评估大型语言模型在组合代码合成上的执行基准。该基准源自FindStat，包含2,329个任务，分布在24个集合中，涵盖5.52百万个隐藏实例，涉及统计合成（将对象映射到整数）和映射合成（将对象映射到对象）。每个任务提供一个数学描述和最多五个公共输入输出示例；模型必须输出一个Python求解函数，且不得使用检索、工具、执行反馈、投票或重排序。提交的结果通过在保留的组合对象上进行精确的沙箱执行进行评分。我们评估了十一种系统：四个闭源生产模型和七个通过一个推理提供者提供的开放权重模型。FindStatBench揭示了三个主要模式。首先，最强的开放源代码和闭源系统在实例准确性上收敛在1个百分比点以内，且对所有系统的一个oracle和从一个中等水平模型的五次采样仅带来了有限的任务准确性提升。其次，示例可能会造成负面影响：几个经典双射在没有示例的情况下完美解决，但在五个示例提示下失败。第三，一些失败反映了输出预算机制，因为推理可能在代码输出之前耗尽可见响应。总体而言，统计合成远比映射合成容易，一些集合的表现接近零，长提示导致准确性急剧下降，且精确的符号规则归纳仍然脆弱。

View on arXiv Download PDF AI Translation

cs.AI / 17 / 2607.18261

When JSON Is Not Enough: Semantic Reliability of Schema-Constrained LLM Ordering Agents

当 JSON 不够用时：基于模式约束的 LLM 排序代理的语义可靠性

Li, Yin

Abstract

LLM agents are increasingly used as transaction compilers: a user states an intent in natural language, and the model emits a structured object that an API can execute. JSON Schema and provider-level structured-output modes are useful because they remove a large class of parse failures, but they do not by themselves decide whether the object is a safe, faithful transaction. We introduce OrderBench, a deterministic benchmark for restaurant ordering agents that separates syntactic validity, schema validity, status decisions, exact item semantics, constraint preservation, and unsafe acceptances. Across 2,400 Nebius Token Factory calls to four open models in prompt-only and JSON-schema modes, we find that schema-valid output can still have large semantic error rates. In the strongest model, both modes achieve 100% schema validity, yet semantic success remains near 80%; in weaker models, schema-valid unsafe acceptances occur in double digits. The result is a concrete engineering warning: structured output is a necessary interface layer, not a substitute for domain verification and fail-closed execution.

Chinese Translation

LLM 代理越来越多地被用作事务编译器：用户以自然语言表达意图，模型生成一个结构化对象，供 API 执行。JSON Schema 和提供者级结构化输出模式非常有用，因为它们消除了大量的解析失败，但它们本身并不能决定对象是否是安全、可信的事务。我们引入了 OrderBench，这是一个用于餐厅点餐代理的确定性基准，能够区分语法有效性、模式有效性、状态决策、确切项目语义、约束保持和不安全接受。在对四个开放模型进行的 2,400 次 Nebius Token Factory 调用中，在仅使用提示和 JSON Schema 模式下，我们发现模式有效的输出仍然可能存在较高的语义错误率。在最强的模型中，两种模式均实现了 100% 的模式有效性，但语义成功率仍接近 80%；在较弱的模型中，模式有效的不安全接受率达到了两位数。结果是一个具体的工程警告：结构化输出是必要的接口层，而不是领域验证和失败关闭执行的替代品。

View on arXiv Download PDF AI Translation

cs.AI / 18 / 2607.18262

ProbSPARQL: Querying Knowledge Graphs with Multi-dimensional, Uncertain Numeric Data

ProbSPARQL：使用多维不确定数值数据查询知识图谱

Wu, Jingcheng, Thapa, Ratan Bahadur, Hernandez, Daniel, Zhou, Hongkuan, Staab, Steffen

Abstract

The SFB 1574 Circular Factory is building a shared knowledge graph infrastructure for integrating data about returned products. A central challenge is that circular-factory data include numeric measurements that (i) originate from sensors or are derived from sensor-based measurements, (ii) are frequently multi-dimensional, and (iii) are inherently uncertain, while downstream triage, validation, reliability-modeling, and reassembly-planning modules require queryable uncertainty representations. Current RDF and SPARQL technologies lack native support for harmonized querying and analysis of such uncertain numeric measurement data. To address this gap, we present ProbSPARQL, an upward-compatible SPARQL extension developed as an early-stage query-layer pilot for this infrastructure. ProbSPARQL models uncertain numeric values as random variables whose distributions are encoded by probabilistic RDF literal datatypes, and supports distribution-aware expressions, probabilistic filters, and divergence-based joins. We implement ProbSPARQL on Apache Jena ARQ and expose it through a Fuseki-compatible execution layer. We assess real-data applicability using project-derived measurement fragments covering GMM-encoded uncertainty and histogram-based empirical roughness distributions, and evaluate scalability separately on controlled ontology-conformant benchmarks with up to 5,000 angle-grinder instances and 1.5M triples. The results show feasible in-engine execution, filter-pushdown speedups over application-layer post-processing, and latency-accuracy trade-offs among divergence-join decision strategies.

Chinese Translation

SFB 1574 循环工厂正在构建一个共享知识图谱基础设施，以整合关于退货产品的数据。一个核心挑战是，循环工厂数据包含的数值测量数据 (i) 来源于传感器或基于传感器测量的派生数据，(ii) 通常是多维的，并且 (iii) 本质上是不确定的，而下游的分类、验证、可靠性建模和重新组装规划模块需要可查询的不确定性表示。目前的 RDF 和 SPARQL 技术缺乏对这种不确定数值测量数据的和谐查询和分析的原生支持。为了解决这一问题，我们提出了 ProbSPARQL，这是一种向上兼容的 SPARQL 扩展，作为该基础设施的早期查询层试点。ProbSPARQL 将不确定数值建模为随机变量，其分布由概率 RDF 字面量数据类型编码，并支持基于分布的表达式、概率过滤器和基于发散的连接。我们在 Apache Jena ARQ 上实现了 ProbSPARQL，并通过与 Fuseki 兼容的执行层进行暴露。我们使用项目派生的测量片段评估真实数据的适用性，这些片段涵盖了 GMM 编码的不确定性和基于直方图的经验粗糙度分布，并在控制的本体一致性基准上分别评估了可扩展性，最多支持 5,000 个角磨机实例和 1.5M 三元组。结果表明，内置执行是可行的，过滤器下推在应用层后处理中的加速效果显著，并且在发散连接决策策略中存在延迟与准确性的权衡。

View on arXiv Download PDF AI Translation

cs.AI / 19 / 2607.18263

Position: AI/ML Deepfake Research is Misaligned with AI-Generated Non-Consensual Intimate Imagery (AIG-NCII)

立场：AI/ML 深度伪造研究与 AI 生成的非自愿亲密影像 (AIG-NCII) 不一致

Qiwei, Li, Santo, Wells Lucas, Schoenebeck, Sarita, Gilbert, Eric

Abstract

AI-generated non-consensual intimate imagery (AIG-NCII) is not adequately addressed in AI/ML literature regarding AI-generated media, commonly referred to as "deepfakes". While research on deepfakes currently focuses on its epistemic harms -- or harms relating to truth and authenticity -- this is misaligned with the dominant reality of generative AI abuse involving sexualized imagery. We conduct a landscape analysis of highly-cited works to demonstrate that technical interventions addressing deepfakes almost entirely ignore AIG-NCII, limiting the research ecosystem to authenticity detection tools. In this position paper, we argue that existing interventions address viewer-centric epistemic harms, such as fraud or scams, but ignore subject-centric dignity harms, such as AIG-NCII. We illustrate that knowing an image is synthetic does not mitigate harms to subjects and may, in some cases, even exacerbate them. We conclude by offering recommendations to realign the field, including updating threat models to consider subject-centric harms and addressing AIG-NCII in AI safety research. Finally, we caution that researchers should only engage in this high-risk domain if they implement safety guardrails for both subjects and researchers and establish partnerships with domain experts in sexual violence prevention.

Chinese Translation

AI 生成的非自愿亲密影像 (AIG-NCII) 在关于 AI 生成媒体的 AI/ML 文献中未得到充分关注，这些媒体通常被称为“深度伪造”。尽管目前对深度伪造的研究主要集中在其认识论危害上——即与真相和真实性相关的危害——但这与涉及性别化影像的生成性 AI 滥用的主导现实并不一致。我们对高度引用的文献进行了领域分析，以证明针对深度伪造的技术干预几乎完全忽视了 AIG-NCII，限制了研究生态系统仅限于真实性检测工具。在这篇立场论文中，我们认为现有的干预措施关注于观众中心的认识论危害，例如欺诈或诈骗，但忽视了以主体为中心的尊严危害，例如 AIG-NCII。我们说明了知道一幅图像是合成的并不能减轻对主体的危害，并且在某些情况下甚至可能加剧这些危害。最后，我们提出了重新调整该领域的建议，包括更新威胁模型以考虑以主体为中心的危害，并在 AI 安全研究中关注 AIG-NCII。我们还警告研究人员，只有在为主体和研究人员实施安全防护措施并与性暴力预防领域的专家建立合作关系的情况下，才应参与这一高风险领域。

View on arXiv Download PDF AI Translation

cs.AI / 20 / 2607.18264

MUX: Continuous Reasoning via Multiplexed Tokens

MUX：通过多路复用令牌进行连续推理

Suleymanzade, Ayhan, Gozeten, Halil Alperen, Bronstein, Michael, Ceylan, İsmail İlkan, Kim, Jinwoo

Abstract

Language models solve complex problems by articulating intermediate reasoning steps in natural language. While effective, this process is computationally bottlenecked: each reasoning step conveys only a single subword, and many are spent expressing a thought instead of carrying out computation. We propose MUX, a simple method for high-bandwidth and compact reasoning based on distillation of discrete reasoning into continuous multiplexed tokens in a latent space. Here, each latent token is trained to represent a weighted linear superposition (multiplexing) of a span of discrete reasoning subwords, where this superposition is lossless by construction and the span can be fully recovered (demultiplexing). We prove that simple position-dependent weightings, such as suitable geometric decay, support lossless multiplexing, which in turn prevents shortcut behaviors caused by latent collapse. We further show that multiplexed reasoning can perform parallel exploration in problems that require search. Across 32 evaluation settings spanning four language models, MUX outperforms strong latent reasoning baselines. Ablation and probing analyses further show that the learned latent tokens encode faithful and interpretable reasoning. Our results suggest that lossless superposition as local learning targets constitutes a sufficient condition for achieving strong and efficient latent continuous reasoning.

Chinese Translation

语言模型通过用自然语言表达中间推理步骤来解决复杂问题。尽管这种方法有效，但在计算上存在瓶颈：每个推理步骤仅传达一个子词，且许多步骤用于表达思想而非进行计算。我们提出了MUX，这是一种基于将离散推理蒸馏为潜在空间中的连续多路复用令牌的简单高带宽紧凑推理方法。在这里，每个潜在令牌被训练为表示一段离散推理子词的加权线性叠加（多路复用），该叠加在构造上是无损的，并且这段内容可以完全恢复（解复用）。我们证明了简单的基于位置的权重，例如合适的几何衰减，支持无损多路复用，从而防止由于潜在崩溃引起的捷径行为。我们进一步展示了多路复用推理可以在需要搜索的问题中执行并行探索。在涵盖四种语言模型的32个评估设置中，MUX超越了强大的潜在推理基线。消融和探测分析进一步表明，学习到的潜在令牌编码了真实且可解释的推理。我们的结果表明，无损叠加作为局部学习目标构成了实现强大且高效的潜在连续推理的充分条件。

View on arXiv Download PDF AI Translation

cs.AI / 21 / 2607.18265

State Compression in Two-Agent LLM Relays: A Closed-World Study of Constraint Preservation

双代理大型语言模型中状态压缩：约束保持的封闭世界研究

Sharma, Anantha, John, Sheeba Elizabeth, Kumar, Kaarthik Senthil, Vijayababu, Saratsuhas

Abstract

Long-running Large Language Model (LLM)-based agents often accumulate large intermediate traces containing audits, eliminations, and numeric calculations. In practice, this state is compressed before handing it to a downstream decision step, creating an information bottleneck in which small omissions can break strict numeric or categorical constraints. This paper evaluates hand-off compression in a closed-world travel-planning relay with two LLM agents. A Researcher audits a fixed inventory of hotels and flights for 50 goal instances, and a Booker selects a hotel--flight pair using only the goal and the hand-off payload, with the inventory withheld. We compare four hand-off conditions: no compression, narrative summarization, schema-constrained JSON extraction, and embedding-based pruning. Exhaustive enumeration over the fixed inventory provides exact feasible and optimal labels. Results show that hand-off representation strongly affects downstream feasibility under a small decision model. JSON extraction achieves the highest feasibility accuracy at 0.96, while narrative summarization, despite producing the smallest compressed hand-off payload, degrades feasibility to 0.48. Embedding-based pruning matches the uncompressed control on feasibility at 0.88 without an additional generative compression call. These findings indicate that constraint checking benefits from structured and auditable hand-off representations rather than relying on brevity alone.

Chinese Translation

基于大型语言模型（LLM）的长时间运行代理通常会累积大量包含审计、消除和数值计算的中间痕迹。在实际操作中，这种状态在交给下游决策步骤之前会被压缩，从而形成信息瓶颈，在这个瓶颈中，微小的遗漏可能会破坏严格的数值或类别约束。本文评估了在一个封闭世界的旅行规划中，两个LLM代理之间的交接压缩。研究者对50个目标实例的固定酒店和航班库存进行审计，而预订者仅使用目标和交接负载（不包含库存）选择酒店-航班对。我们比较了四种交接条件：无压缩、叙述性摘要、基于模式的JSON提取和基于嵌入的剪枝。对固定库存的穷举枚举提供了精确的可行和最优标签。结果表明，交接表示在小型决策模型下强烈影响下游的可行性。JSON提取在可行性准确率上达到了最高的0.96，而叙述性摘要尽管产生了最小的压缩交接负载，但可行性降至0.48。基于嵌入的剪枝在可行性上与未压缩的控制组相匹配，达到了0.88，而无需额外的生成压缩调用。这些发现表明，约束检查受益于结构化和可审计的交接表示，而不仅仅依赖于简洁性。

View on arXiv Download PDF AI Translation

cs.AI / 22 / 2607.18266

Structured Synthetic Reasoning Data for Arithmetic Fine-Tuning of Small Language Models

用于小型语言模型算术微调的结构化合成推理数据

O'Grady, Jake, Ramlan, Effirul

Abstract

Small language models are attractive for local deployment, but they often struggle with multi-step arithmetic reasoning. We study whether structured synthetic reasoning data can improve this behaviour under consumer-hardware constraints. Starting from GSM8K, we generated a 21,250-example corpus of grade-school arithmetic word-problem variants using GPT-5-mini, combining natural-language solution traces, light Socratic-style cues, structural variation, and irrelevant distractor context. We then fine-tuned Qwen3-0.6B and Qwen3-1.7B with LoRA on consumer hardware (Apple M4, 16 GB RAM). Exact-match accuracy on GSM8K improved from 36.5% to 49.1% for Qwen3-0.6B and from 53.5% to 66.5% for Qwen3-1.7B. For Qwen3-1.7B, transfer to related arithmetic benchmarks was stronger, reaching 98.9% on MultiArith and 73.0% on SVAMP, compared with 54.4% and 45.3% for the base model. Qualitative analysis suggests that fine-tuned models produce shorter reasoning traces, make fewer arithmetic and distractor-use errors, and benefit more consistently from self-consistency sampling. These results show that low-cost synthetic data design can materially improve arithmetic adaptation in small language models. Because the intervention combines Socratic-style cues with other data-design choices, we interpret the gains as evidence for structured synthetic reasoning data rather than as a causal test of Socratic guidance alone.

Chinese Translation

小型语言模型因其适合本地部署而受到青睐，但它们在多步骤算术推理方面常常表现不佳。我们研究了在消费硬件限制下，结构化合成推理数据是否能够改善这种情况。基于GSM8K，我们使用GPT-5-mini生成了一个包含21,250个例子的学龄前算术文字问题变体的语料库，结合了自然语言解决方案轨迹、轻度苏格拉底式提示、结构变异和无关的干扰上下文。然后，我们在消费硬件（Apple M4，16 GB RAM）上使用LoRA对Qwen3-0.6B和Qwen3-1.7B进行了微调。在GSM8K上的精确匹配准确率从Qwen3-0.6B的36.5%提高到49.1%，从Qwen3-1.7B的53.5%提高到66.5%。对于Qwen3-1.7B，转移到相关的算术基准测试的表现更强，在MultiArith上达到了98.9%，在SVAMP上达到了73.0%，而基础模型的表现分别为54.4%和45.3%。定性分析表明，微调后的模型生成了更短的推理轨迹，发生的算术和干扰使用错误更少，并且更一致地受益于自一致性采样。这些结果表明，低成本的合成数据设计可以实质性地改善小型语言模型的算术适应性。由于该干预结合了苏格拉底式提示与其他数据设计选择，我们将这些提升解读为结构化合成推理数据的证据，而不是仅仅作为苏格拉底指导的因果测试。

View on arXiv Download PDF AI Translation

cs.AI / 23 / 2607.18268

Fence: Specialized SLM Guardrails for LLM Applications

Fence：针对大型语言模型应用的专用小型语言模型保护措施

Lakara, Kumud, Shi, Ruibo, Silavong, Fran

Abstract

Real-world applications that use closed-source large language models (LLMs) need advanced safety measures that go beyond the basic content filters. Content moderation filters such as toxicity and bias have relatively standard definitions where as application specific guardrails like hallucination, topic drift and behaviour deviation are more difficult to model and can vary by use case. Additionally, data scarcity and annotation costs, make the process of creating and testing specialized guardrails challenging. In this work, we propose using Small Language Models (SLMs) trained on synthetic data as specialized guardrails for LLM applications. We introduce a novel synthetic data generation method inspired by the design of Generative Adversarial Networks (GANs) to generate high quality synthetic data samples which can be used to train SLMs to encode use case specific guardrail information and hence function as specialized guardrails. Our experiments demonstrate that SLM guardrails trained on high quality synthetic data show performance gains over prompt based LLM guardrails.

Chinese Translation

使用封闭源的大型语言模型（LLMs）的实际应用需要超越基本内容过滤器的高级安全措施。内容审核过滤器如毒性和偏见有相对标准的定义，而特定应用的保护措施如幻觉、主题漂移和行为偏差则更难建模，并且可能因使用案例而异。此外，数据稀缺和标注成本使得创建和测试专用保护措施的过程充满挑战。在本研究中，我们提出使用在合成数据上训练的小型语言模型（SLMs）作为大型语言模型应用的专用保护措施。我们引入了一种新颖的合成数据生成方法，灵感来自生成对抗网络（GANs）的设计，以生成高质量的合成数据样本，这些样本可用于训练小型语言模型以编码特定使用案例的保护措施信息，从而充当专用保护措施。我们的实验表明，基于高质量合成数据训练的小型语言模型保护措施在性能上优于基于提示的大型语言模型保护措施。

View on arXiv Download PDF AI Translation

cs.AI / 24 / 2607.18269

Wisdom of LLM Crowds: Aggregation and Contamination in Language Model Ensembles

大型语言模型集体智慧：语言模型集成中的聚合与污染

Douven, Igor

Abstract

The wisdom of crowds -- the finding that aggregating judgments across individuals often outperforms the best individual -- has been extensively studied with human forecasters. Whether the same phenomenon emerges when the ``crowd'' consists of large language models (LLMs) is an open question with both theoretical and practical implications. We elicited probability estimates from 15 LLMs on 254 binary prediction market questions and evaluated classical and learned aggregation methods. Learned aggregators -- a multilayer perceptron and a logistic regression -- outperformed all individual models and classical methods. The logistic regression was found to match the neural network, suggesting that the benefit of learned aggregation derives from learning a linear combination of diverse model outputs rather than from nonlinear interactions. Symbolic regression applied to the neural network's learned mapping recovered a pure model-disagreement signal as the lowest-complexity useful formula on the Pareto frontier, further supporting this interpretation. Training cutoff contamination proved a pervasive confound: the apparent capability gap between frontier cloud models and smaller local models collapsed from 35.8% to 8.9% on a clean subset of questions resolving after all models' training cutoffs, and individual model rankings showed only moderate stability. Even when the prediction market is evaluated at each model's training cutoff, LLMs remained substantially less accurate, indicating a genuine gap in collective information aggregation. These findings suggest that LLM crowds can exhibit wisdom-of-crowds effects, but that contamination-free evaluation is essential for reliable assessment.

Chinese Translation

集体智慧——即跨个体聚合判断通常优于最佳个体的发现——已在人工预测者中得到了广泛研究。当“人群”由大型语言模型（LLMs）组成时，是否会出现相同的现象仍然是一个具有理论和实践意义的开放性问题。我们从15个LLMs中获取了254个二元预测市场问题的概率估计，并评估了经典和学习聚合方法。学习聚合器——多层感知器和逻辑回归——的表现优于所有单个模型和经典方法。研究发现，逻辑回归与神经网络的表现相当，这表明学习聚合的好处来自于学习多样化模型输出的线性组合，而非非线性交互。对神经网络学习映射的符号回归恢复了一个纯粹的模型不一致信号，作为在帕累托前沿上最低复杂度的有用公式，进一步支持了这一解释。训练截止污染被证明是一个普遍的混淆因素：在所有模型的训练截止后解决的问题的干净子集上，前沿云模型与较小本地模型之间的明显能力差距从35.8%降至8.9%，而单个模型的排名显示出仅有适度的稳定性。即使在每个模型的训练截止时评估预测市场，LLMs的准确性仍然显著较低，表明集体信息聚合存在真正的差距。这些发现表明，LLM人群可以表现出集体智慧效应，但无污染的评估对于可靠的评估至关重要。

View on arXiv Download PDF AI Translation

cs.AI / 25 / 2607.18270

Trajectory-Aware Clinical Risk Prediction via Severity-Grounded Knowledge Graphs and Retrieval-Augmented Generation

基于严重性知识图谱和检索增强生成的轨迹感知临床风险预测

Jeon, Kyunghoon, Ko, Youmin, Jung, Woohwan, Kim, Hyunjoon

Abstract

While Electronic Health Records (EHRs) offer a wealth of clinical data, effectively augmenting a patient's records with heterogeneous external knowledge to predict the patient's clinical risk remains a significant challenge. Existing methods fail to capture disease severity, treatment responses, and nuanced clinical progression, due to data sparsity and the underutilization of unstructured clinical notes. To address these challenges, we propose TRACER (a trajectory-aware and clinically grounded prediction framework) that (1) constructs a medical knowledge graph enriched with severity information from medical literature, (2) retrieves clinically relevant, severity-weighted paths of a patient's progression from the knowledge graph, (3) extracts clinically relevant events from unstructured clinical notes, and (4) augments patient context with similar peer cases. Experiments on the MIMIC-III and MIMIC-IV datasets demonstrate large gains over state-of-the-art baselines, with up to 28.5% increase in Macro F1 score for the mortality prediction task, and 19.7% increase for the readmission prediction task.

Chinese Translation

尽管电子健康记录（EHRs）提供了丰富的临床数据，但有效地将患者记录与异构外部知识结合以预测患者的临床风险仍然是一个重大挑战。现有方法未能捕捉疾病严重性、治疗反应和细微的临床进展，原因在于数据稀疏和对非结构化临床笔记的利用不足。为了解决这些挑战，我们提出了TRACER（一个轨迹感知和临床基础的预测框架），该框架（1）构建了一个从医学文献中丰富了严重性信息的医学知识图谱，（2）从知识图谱中检索与临床相关的、按严重性加权的患者进展路径，（3）从非结构化临床笔记中提取临床相关事件，以及（4）通过类似的同伴案例增强患者背景。在MIMIC-III和MIMIC-IV数据集上的实验表明，相较于最先进的基线方法，取得了显著的提升，死亡预测任务的Macro F1分数提高了高达28.5%，再入院预测任务提高了19.7%。

View on arXiv Download PDF AI Translation

cs.AI / 26 / 2607.18271

Using LLMs for Explainable, Data-Driven Insight Generation from Time Series

利用大型语言模型生成可解释的数据驱动时间序列洞察

Mundhra, Ria, Santos, Gustavo Sato dos, Benedikt, Michael

Abstract

Time series forecasts are widely used in decision-critical domains, where they are rarely consumed without accompanying explanations. Producing such explanations is usually a manual and costly process, and attempts to automate it using large language models often suffer from hallucination when applied to temporal data. We propose a domain-agnostic framework for grounded natural language explanation generation for time series forecasts, illustrated in Figure 1. The framework consists of three components: (i) extraction of structured explanatory factors from historical analyst-written explanations, (ii) evidence-conditioned explanation generation, and (iii) scalable evaluation for readability, logical consistency, and persuasiveness. The design explicitly constrains generation to verifiable evidence, reducing unsupported claims. We evaluate the framework on a financial forecasting case study involving the NASDAQ-100 index and a freight pricing case study using data from Vortexa. Results show that generated explanations approached analyst-written explanations in terms of readability, consistency and persuasiveness. These findings demonstrate that grounded explanation generation for time series forecasting can be achieved at scale without domain-specific fine-tuning.

Chinese Translation

时间序列预测广泛应用于决策关键领域，通常不会在没有伴随解释的情况下被使用。生成此类解释通常是一个手动且成本高昂的过程，而尝试使用大型语言模型进行自动化时，往往在应用于时间数据时会出现幻觉现象。我们提出了一个与领域无关的框架，用于生成基于证据的自然语言解释，以支持时间序列预测，如图1所示。该框架由三个组件组成：(i) 从历史分析师撰写的解释中提取结构化解释因素，(ii) 基于证据的解释生成，以及 (iii) 可扩展的可读性、逻辑一致性和说服力评估。该设计明确限制生成内容为可验证的证据，从而减少不支持的主张。我们在一个涉及NASDAQ-100指数的金融预测案例研究和一个使用Vortexa数据的货运定价案例研究中评估了该框架。结果表明，生成的解释在可读性、一致性和说服力方面接近分析师撰写的解释。这些发现表明，针对时间序列预测的基于证据的解释生成可以在不进行领域特定微调的情况下实现规模化。

View on arXiv Download PDF AI Translation

cs.AI / 27 / 2607.18296

Deep Reinforcement Learning to Master the Asymmetric Strategy of Baghchal

深度强化学习掌握Baghchal的不对称策略

Raut, Ranjit, Subedi, Aarav, Rai, Sagun, Shakya, Aaryan, Shakya, Manoj

Abstract

Baghchal is a two-player asymmetric board game with Nepali origins where four tigers are to capture goats and twenty goats desire to keep tigers in immobility. Although Baghchal has a complex structure which is strategic, has perfect information structure, and has cultural meaning, it has not been adequately covered in deep reinforcement learning (RL) literature. This paper gives a systematic exploration of four deep RL solutions Deep Q-Network (DQN), REINFORCE, Proximal Policy Optimization (PPO) and MuZero that are trained on one side of the asymmetric gameplay of Baghchal and then evaluated on the other side. The algorithms are rated based on win rate, draw rate, average captures, training convergence and computational cost. It is experimentally found that MuZero generates the best performance in both tasks, achieving 86 percent win over these Tiger and 62 percent win over these Goat and the ability to do so is due to the model-based planning machine through the Monte Carlo Tree Search. PPO is the most realistic algorithm and is provided to be competitive over both asymmetric tasks with significantly reduced computational costs compared to MuZero. Emergent strategic behavior analysis shows that model-based strategies are optimal over long-horizon planning, whereas value-based counterparts like DQN are more biased up towards the Tiger role owing to the more substantial reward signal.

Chinese Translation

Baghchal是一种起源于尼泊尔的双人不对称棋盘游戏，其中四只老虎试图捕捉山羊，而二十只山羊则希望将老虎保持在静止状态。尽管Baghchal具有复杂的结构，具有战略性、完美的信息结构以及文化意义，但在深度强化学习（RL）文献中尚未得到充分研究。本文系统探讨了四种深度强化学习解决方案：深度Q网络（Deep Q-Network, DQN）、REINFORCE、近端策略优化（Proximal Policy Optimization, PPO）和MuZero，这些算法在Baghchal的不对称游戏一侧进行训练，然后在另一侧进行评估。根据胜率、平局率、平均捕获数、训练收敛性和计算成本对这些算法进行评估。实验结果表明，MuZero在这两项任务中表现最佳，对老虎的胜率达到86%，对山羊的胜率达到62%，其能力源于通过蒙特卡洛树搜索（Monte Carlo Tree Search）的基于模型的规划机制。PPO是最具现实性的算法，在两个不对称任务中表现出竞争力，且与MuZero相比显著降低了计算成本。新兴的战略行为分析表明，基于模型的策略在长期规划中是最优的，而像DQN这样的基于价值的策略则由于更强的奖励信号而更倾向于老虎角色。

View on arXiv Download PDF AI Translation

cs.AI / 28 / 2607.18366

Operational Hallucination and Safety Drift in AI Agents

人工智能代理中的操作幻觉与安全漂移

Yu, Shasha, Carroll, Fiona, Bentley, Barry L.

Abstract

Large language models (LLMs) serving as planners in tool-using autonomous agents introduce dynamic reliability risks in multi-turn execution. While single-turn safety mechanisms are relatively mature, extended interactions reveal structural vulnerabilities where initial alignment degrades over time. This paper empirically characterizes two observed failure modes across multiple state-of-the-art LLMs: Safety Drift, the gradual erosion of declared safety intent leading to constraint-violating actions (e.g., textual refusal followed by reconnaissance and unsafe execution), and Operational Hallucination, persistent repetitive tool calls indicative of flawed state perception (e.g., livelocks even in legitimate tasks). Through controlled multi-turn evaluation on high-stakes ethical dilemmas, malicious requests, and benign controls, we quantify these phenomena using declaration-action gap and livelock metrics, demonstrating their cross-model prevalence under direct execution protocols. Root-cause analysis attributes the instabilities to the decoupling of reasoning context from execution state in current agent loops. We propose an Action-Aware Supervision Layer - a lightweight, plug-and-play architectural blueprint incorporating intent-action consistency checks, runtime state tracking, and forced termination primitives. Post-hoc simulation on captured failure trajectories shows the layer can intercept observed violations without false positives on benign cases. This work advances agent reliability by shifting focus from linguistic safeguards to enforceable architectural mechanisms for responsible agentic AI.

Chinese Translation

作为工具使用自主代理的规划者，大型语言模型（LLMs）在多轮执行中引入了动态可靠性风险。尽管单轮安全机制相对成熟，但扩展交互揭示了结构性脆弱性，初始对齐随着时间的推移而退化。本文实证性地描述了在多个最先进的LLM中观察到的两种失败模式：安全漂移（Safety Drift），即声明的安全意图逐渐侵蚀，导致违反约束的行为（例如，文本拒绝后进行侦察和不安全执行），以及操作幻觉（Operational Hallucination），即持续重复的工具调用，表明状态感知存在缺陷（例如，即使在合法任务中也出现活锁）。通过对高风险伦理困境、恶意请求和良性控制的受控多轮评估，我们使用声明-行动差距和活锁指标量化这些现象，展示了它们在直接执行协议下的跨模型普遍性。根本原因分析将不稳定性归因于当前代理循环中推理上下文与执行状态的解耦。我们提出了一种行动感知监督层（Action-Aware Supervision Layer）——一种轻量级、即插即用的架构蓝图，包含意图-行动一致性检查、运行时状态跟踪和强制终止原语。对捕获的失败轨迹进行的后期模拟显示，该层可以在不对良性案例产生误报的情况下拦截观察到的违规行为。本研究通过将重点从语言保障转向可执行的架构机制，推动了代理的可靠性，为负责任的代理人工智能提供了支持。

View on arXiv Download PDF AI Translation

cs.AI / 29 / 2607.18367

AlayaWorld: Interactive Long-Horizon World Modeling -- Full Technical Report

AlayaWorld：互动长视距世界建模——完整技术报告

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

Abstract

Unlike conventional video game development, which relies on labor-intensive pipelines for asset production, animation, physics, and programming, video world models generate interactive environments from user inputs instantly. It enable us to create customized, explorable, and continuously evolving virtual world from text, an image, or video. Realizing this vision requires four tightly coupled capabilities: interaction, persistent spatiotemporal consistency, stable long-horizon generation, and efficient response. We present AlayaWorld, an interactive long-horizon video world model that generates 24-fps video at 540p and 720p. Built on a 15B video diffusion transformer, AlayaWorld generates short latent chunks autoregressively under camera trajectories and switchable text prompts. Its bounded visual context combines a persistent sink frame, compressed temporal history, geometry-aligned spatial memory, and recent-frame conditioning. To reduce long-term drift, the model is trained with corrupted histories and prediction residuals collected from its own roll-outs. We further introduce a discrete autoregressive distillation formulation that combines distribution-matching distillation, self-forcing++, and consistency distillation, reducing inference from approximately 30 sampling steps to four steps per chunk. On iWorld-Bench, AlayaWorld achieves the best performance over long-horizon generation. Conceived as a full-stack, open-source, and long-term project, AlayaWorld is intended to provide an extensible foundation for future research on interactive video world models.

Chinese Translation

与传统视频游戏开发依赖于劳动密集型的资产生产、动画、物理和编程流程不同，视频世界模型能够即时根据用户输入生成互动环境。这使我们能够从文本、图像或视频创建定制的、可探索的和持续演变的虚拟世界。实现这一愿景需要四个紧密耦合的能力：交互、持久的时空一致性、稳定的长视距生成和高效的响应。我们提出了AlayaWorld，一个互动长视距视频世界模型，能够以540p和720p的分辨率生成24帧每秒的视频。AlayaWorld基于一个15B的视频扩散变换器，能够在相机轨迹和可切换的文本提示下自回归生成短的潜在片段。其有限的视觉上下文结合了持久的汇聚帧、压缩的时间历史、几何对齐的空间记忆和最近帧的条件。为了减少长期漂移，该模型使用从自身滚动输出中收集的损坏历史和预测残差进行训练。我们进一步引入了一种离散自回归蒸馏公式，结合了分布匹配蒸馏、自强++和一致性蒸馏，将每个片段的推理步骤从大约30步减少到四步。在iWorld-Bench上，AlayaWorld在长视距生成方面实现了最佳性能。作为一个全栈、开源和长期项目，AlayaWorld旨在为未来关于互动视频世界模型的研究提供一个可扩展的基础。

View on arXiv Download PDF AI Translation

cs.AI / 30 / 2607.18368

Neuro-Symbolic Meta-Policies for Temporal Knowledge-Graph Memory under Partial Observability

部分可观测下的时间知识图谱记忆的神经符号元策略

Kim, Taewoon, François-Lavet, Vincent, Cochez, Michael

Abstract

Partially observable reinforcement learning requires deciding what to retain, retrieve, and forget over time. We introduce a neuro-symbolic meta-policy that learns which symbolic memory heuristic to apply at each decision point while keeping execution symbolic. Our setting uses temporal knowledge-graph memory in RoomKG, where hidden state and observations are represented as Resource Description Framework (RDF) graphs and memory is augmented with temporal RDF triple annotations. The model combines knowledge-graph encoding of memory contents with value heads for question answering, exploration, and forgetting, yielding a controller that is both adaptive and inspectable. This gives the work a direct Semantic Web grounding through RDF-based representation, annotation-compatible graph semantics, and graph-based symbolic operations over explicit memory state. On train/test room splits at long-term memory capacity of 512, the qualifier-aware StarE-GNN configuration achieves the best held-out performance among the compared symbolic, neural, and neuro-symbolic systems while preserving step-level traceability of memory-management decisions.

Chinese Translation

部分可观测的强化学习需要在时间上决定保留、检索和遗忘什么。我们提出了一种神经符号元策略，该策略学习在每个决策点应用哪种符号记忆启发式，同时保持执行的符号性。我们的设置使用RoomKG中的时间知识图谱记忆，其中隐藏状态和观察被表示为资源描述框架（Resource Description Framework, RDF）图，记忆通过时间RDF三元组注释进行增强。该模型结合了记忆内容的知识图谱编码与用于问答、探索和遗忘的价值头，产生了一个既适应又可检查的控制器。这使得该研究通过基于RDF的表示、注释兼容的图语义和对显式记忆状态的图形符号操作，具有直接的语义网基础。在长期记忆容量为512的训练/测试房间分割中，具有资格感知的StarE-GNN配置在比较的符号、神经和神经符号系统中实现了最佳的保留性能，同时保持了记忆管理决策的逐步可追溯性。

View on arXiv Download PDF AI Translation

cs.AI / 31 / 2607.18536

MAGE: Human-Like Macro Placement via Agentic Multimodal Reasoning

MAGE：通过代理多模态推理实现类人宏单元布局

Kahng, Andrew B., Kundu, Sayak, Pramanik, Bodhisatta

Abstract

Macro placement still requires substantial manual refinement in industrial physical design flows. We present MAGE (Macro Placement Agentic Engine), a multimodal multi-agent framework for macro placement refinement. MAGE decomposes the macro placement task into a six-phase workflow that combines structured floorplanning rules, visual checks, and iterative refinement. Expert floorplanning knowledge is encoded through natural-language directives and validation criteria, rather than learned from labeled placement data. A tournament-style refinement mode evaluates multiple candidate placements and propagates feedback from higher-quality solutions. We also introduce four metrics for quantifying human-likeness in macro placement: notch score, whitespace score, pocket score, and alignment score. These metrics capture structural properties used by expert designers but not directly measured by conventional PPA metrics. Across nine designs in NanGate45 and GlobalFoundries 12nm enablements, MAGE achieves geometric-mean improvements of 11.1%-19.3% in WNS and 70.0%-74.0% in TNS over commercial macro placers. On the three NanGate45 designs, for which human-expert and Hier-RTLMP baselines are available, MAGE improves WNS and TNS by 18.3% and 72.5% over the human expert, and by 47.0% and 80.4% over Hier-RTLMP, with comparable wirelength and power. On human-likeness metrics, MAGE improves the overall score by 6%-48% over all baselines. Additional case studies on anonymized netlists, unseen designs, dense rectilinear floorplans, and high-utilization settings show that the framework transfers to new placement settings without design-specific retraining.

Chinese Translation

宏单元布局在工业物理设计流程中仍然需要大量的手动优化。我们提出了MAGE（宏单元布局代理引擎），这是一个用于宏单元布局优化的多模态多代理框架。MAGE将宏单元布局任务分解为一个六阶段的工作流程，该流程结合了结构化的平面布局规则、视觉检查和迭代优化。专家的平面布局知识通过自然语言指令和验证标准进行编码，而不是从标记的布局数据中学习。锦标赛式的优化模式评估多个候选布局，并从更高质量的解决方案中传播反馈。我们还引入了四个量化宏单元布局类人性的指标：缺口分数（notch score）、空白分数（whitespace score）、口袋分数（pocket score）和对齐分数（alignment score）。这些指标捕捉了专家设计师使用的结构特性，但不是传统PPA指标直接测量的。在NanGate45和GlobalFoundries 12nm的九个设计中，MAGE在WNS上实现了11.1%-19.3%的几何平均改进，在TNS上实现了70.0%-74.0%的改进，超越了商业宏单元布局工具。在三个NanGate45设计中，针对可用的人类专家和Hier-RTLMP基线，MAGE在WNS和TNS上分别比人类专家提高了18.3%和72.5%，比Hier-RTLMP提高了47.0%和80.4%，且在布线长度和功耗方面相当。在类人性指标上，MAGE的整体得分比所有基线提高了6%-48%。对匿名网表、未见设计、密集矩形平面布局和高利用率设置的额外案例研究表明，该框架能够在不进行特定设计的再训练的情况下转移到新的布局设置。

View on arXiv Download PDF AI Translation

cs.AI / 32 / 2607.18548

Engineering Trustworthy Agentic AI for Critical Systems

为关键系统工程可信的自主人工智能

Al-Refai, Omar, Shahbaz, Ibrahim, Husseinat, Adam Ali, Mandulak, Michael, Kim, Jaewon, Hammad, Eman

Abstract

Agentic artificial intelligence systems, capable of autonomous perception, planning, tool use, and multi-step action, are increasingly proposed for critical engineering domains where decisions carry physical, operational, or economic consequences. This survey addresses a gap in current literature by treating trustworthiness, whether agentic behavior can be verified, audited, and trusted under the constraints that engineering practice actually requires, as a first-class engineering property, rather than evaluating agentic AI by task capability alone. The study adopts a trustworthiness model organized around five cross-cutting dimensions: safety and constraint satisfaction; robustness and reliability; transparency and interpretability; accountability and auditability; and privacy and security. This is mapped onto an agentic assurance workflow spanning perception through audit. Building on this foundation, agentic systems architectures, threats, concrete trust mechanisms, and quantitative metrics are surveyed for direct application in agentic systems development and evaluation. These principles are then examined across four constraint-bound engineering domains: power systems, autonomous vehicles/robotics/UAVs, high-performance computing, and communication networks, identifying recurring design patterns, shared failure modes, and domain-specific gaps. Synthesizing across those domains, agentic AI trustworthiness is shown to be a single problem, with a path outlined toward a reusable, cross-domain assurance framework analogous to the graded certification regimes used by mature safety-critical engineering fields.

Chinese Translation

自主人工智能系统能够进行自主感知、规划、工具使用和多步骤行动，越来越多地被提议用于决策具有物理、操作或经济后果的关键工程领域。本调查填补了当前文献中的一个空白，将可信性视为一种一流的工程属性，探讨自主行为在工程实践实际要求的约束下是否可以被验证、审计和信任，而不仅仅是通过任务能力来评估自主人工智能。该研究采用了一个围绕五个交叉维度组织的可信性模型：安全性和约束满足；稳健性和可靠性；透明性和可解释性；问责性和可审计性；以及隐私和安全性。这一模型映射到一个涵盖从感知到审计的自主保证工作流程。在此基础上，调查了自主系统架构、威胁、具体的信任机制和定量指标，以便直接应用于自主系统的开发和评估。这些原则随后在四个受约束的工程领域中进行了检验：电力系统、自动驾驶车辆/机器人/无人机、高性能计算和通信网络，识别出重复出现的设计模式、共享的故障模式以及特定领域的差距。通过跨领域的综合，表明自主人工智能的可信性是一个单一的问题，并概述了通向可重用的跨领域保证框架的路径，该框架类似于成熟的安全关键工程领域所使用的分级认证制度。

View on arXiv Download PDF AI Translation

cs.AI / 33 / 2607.18567

Attacking Graph Foundation Models Through Their Shared Representation

通过共享表示攻击图基础模型

Kumar, Pankaj, Mishra, Subhankar

Abstract

A graph foundation model generalizes across graph domains by mapping every input into one shared representation before any task reasoning. We call this map the alignment layer, the component that separates a graph foundation model from a graph neural network, and we show it is a distinct attack surface that prior work has not studied. We attack it at inference time, with no access to training, on six public models spanning spectral tokenizers, text embedding spaces, and a discrete codebook. A directed representation-space perturbation collapses every model, but at a budget comparable to the representation norm a plain graph network also needs, with one exception: OpenGraph, whose spectral tokenizer collapses at a fifth of that budget, an alignment-specific fragility a plain network does not share and which a same-representation control traces to the tokenizer rather than the decoder. A realizable input-space attack that edits edges, features, or text removes at least half the correct predictions on three of the six models at peak. How much of this fragility an input-access attacker realizes tracks how directly the decoder reads the representation, and not the clean accuracy a task leaves; we measure this carrier gain structurally from the decoder's local Lipschitz sensitivity, and report clean-accuracy headroom as a within-model ordering heuristic that does not survive on realizable attacks.

Chinese Translation

图基础模型通过将每个输入映射到一个共享表示来跨越图域进行泛化，在任何任务推理之前。我们称这个映射为对齐层，它是将图基础模型与图神经网络区分开的组件，我们展示了它是一个先前研究未曾探讨的独特攻击面。我们在推理时对其进行攻击，且不接触训练，针对六个公共模型，包括谱标记器、文本嵌入空间和离散代码本。一个定向表示空间扰动使每个模型崩溃，但其预算与普通图网络所需的表示范数相当，唯一的例外是OpenGraph，其谱标记器在该预算的五分之一时崩溃，这是一种普通网络所不具备的特定于对齐的脆弱性，而同一表示的控制则追溯到标记器而非解码器。一个可实现的输入空间攻击，通过编辑边、特征或文本，至少在六个模型中的三个模型上在峰值时移除了至少一半的正确预测。输入访问攻击者实现的脆弱性程度取决于解码器读取表示的直接程度，而不是任务留下的干净准确性；我们从解码器的局部利普希茨敏感性结构上测量这种载体增益，并报告干净准确性头部空间作为一种在模型内部排序的启发式方法，但在可实现攻击中并不成立。

View on arXiv Download PDF AI Translation

cs.AI / 34 / 2607.18573

When Does Machine Learning Beat Value Sorting? A Three-Dataset Diagnostic of Exposure-Weighted Shipment Prioritization

机器学习何时优于价值排序？基于三组数据集的曝光加权发货优先级诊断

Li, Jize

Abstract

Delay-risk models are usually judged by predictive accuracy. What matters in practice is narrower: with capacity to review only a few shipments, which ones should a manager check first? We evaluate whether machine learning clears a demanding no-model baseline: inspect the highest-value shipments first. Across three real supply-chain contexts: SCMS procurement, DataCo logistics, and Olist e-commerce, we use leakage-controlled rolling-origin evaluation and 1000-sample paired bootstrap confidence intervals. Ranking by predicted delay severity times known value (M1) beats severity-only ranking in all three datasets, yet it does not generally beat value sorting. At a 10% review budget, M1 minus VALUE_ONLY is -5.5 percentage points (pp) for SCMS, +10.1 pp for DataCo, and -4.9 pp for Olist. The divide is consistent with severity learnability: DataCo has R^2 = 0.27 and calibration bias of +0.01 days, whereas SCMS and Olist have R^2 of approximately -0.02 and negative calibration bias. Nested-CV cost-sensitive retraining does not deliver a stable improvement over M1. Rather than proposing a new learning algorithm, this paper presents a deployment diagnostic and evaluation protocol. Value sorting should remain a permanent benchmark, and ML should be deployed only after severity learnability and calibration have been audited and the model clears that gate under leakage-controlled rolling-origin evaluation.

Chinese Translation

延迟风险模型通常通过预测准确性来评估。然而，实际中更为重要的是：在只能审查少量发货的情况下，管理者应优先检查哪些发货？我们评估机器学习是否能够超越一个严格的无模型基准：首先检查最高价值的发货。在三个真实的供应链背景下：SCMS采购、DataCo物流和Olist电子商务，我们使用泄漏控制的滚动起源评估和1000样本配对自助置信区间。根据预测的延迟严重性乘以已知价值进行排名（M1）在所有三个数据集中均优于仅按严重性排名，但通常不优于价值排序。在10%的审查预算下，M1与VALUE_ONLY的差异为SCMS -5.5个百分点（pp）、DataCo +10.1 pp和Olist -4.9 pp。这一差异与严重性可学习性一致：DataCo的R^2 = 0.27，校准偏差为+0.01天，而SCMS和Olist的R^2约为-0.02，校准偏差为负值。嵌套交叉验证的成本敏感再训练并未在M1上提供稳定的改进。本文并未提出新的学习算法，而是提供了一种部署诊断和评估协议。价值排序应作为一个永久基准，机器学习应仅在严重性可学习性和校准经过审核且模型在泄漏控制的滚动起源评估中通过该门槛后进行部署。

View on arXiv Download PDF AI Translation

cs.AI / 35 / 2607.18665

SciHazard: A Benchmark for Measuring Scientific Safety Risks with Decomposed Harm Scoring

SciHazard：一种通过分解危害评分来衡量科学安全风险的基准

Li, Chunxiao, Xiong, Yuan, Li, Lijun, Du, Tianyi, Zhang, Wenlong, Bai, Lei, Shao, Jing

Abstract

Large language models (LLMs) increasingly support science, but they can also convert hazardous scientific knowledge into actionable misuse guidance. Existing benchmarks often rely on templated queries disconnected from real-world hazards, and employ LLM-as-a-Judge paradigms without domain grounding. To address this, we introduce SciHazard, a real-world-grounded benchmark for scientific risks and a dataset agnostic evaluation framework for measuring harmfulness. SciHazard contains 2400 hazardous questions and 600 oversafety questions across 12 disciplines, with both queries grounded in regulated entities and documented failure scenarios. To compute \textsc{DeHarm-Score} , we develop a decomposed evaluating procedure that combines query hazard severity, refusal behavior, and response-level risk. For non-refused responses, it further decomposes response-level harm into \textsc{Executability}, quantified via dynamic checklists with importance weighting, and \textsc{Net-new risk}, assessed through retrieval-augmented claim extraction and synthesis-barrier verification. An expert-validation study shows that \textsc{DeHarm-Score} improves agreement with expert annotations by 90.17\% over the strongest baseline. We benchmark 31 frontier LLMs and deep research agents in an extensive scientific safety evaluation. Notably, deep research agents yield 32.3\% higher mean \textsc{DeHarm-Score} than standard LLMs, exposing autonomous agents as a critical blind spot in current safety defenses. Code and dataset are available at https://anonymous.4open.science/r/DeharmScore-7B55.

Chinese Translation

大型语言模型（LLMs）越来越多地支持科学研究，但它们也可能将危险的科学知识转化为可操作的误用指导。现有的基准通常依赖于与现实世界危害脱节的模板化查询，并采用没有领域基础的 LLM 作为评判者的范式。为了解决这个问题，我们引入了 SciHazard，这是一个基于现实世界的科学风险基准和一个数据集无关的评估框架，用于衡量有害性。SciHazard 包含 2400 个危险问题和 600 个过度安全问题，涵盖 12 个学科，这些查询均以受监管实体和文档化的失败场景为基础。为了计算 extsc{DeHarm-Score}，我们开发了一种分解评估程序，结合了查询危害严重性、拒绝行为和响应级别风险。对于未被拒绝的响应，它进一步将响应级别的危害分解为 extsc{Executability}，通过动态检查表和重要性加权进行量化，以及 extsc{Net-new risk}，通过检索增强的声明提取和合成障碍验证进行评估。一项专家验证研究表明， extsc{DeHarm-Score} 在与专家注释的一致性方面比最强基线提高了 90.17%。我们在广泛的科学安全评估中基准测试了 31 个前沿 LLM 和深度研究代理。值得注意的是，深度研究代理的平均 extsc{DeHarm-Score} 比标准 LLM 高出 32.3%，揭示了自主代理在当前安全防御中的关键盲点。代码和数据集可在 https://anonymous.4open.science/r/DeharmScore-7B55 获取。

View on arXiv Download PDF AI Translation

cs.AI / 36 / 2607.18691

Semantic Primes as Explanans for Emotion in Large Language Models

作为情感解释的语义原语在大型语言模型中的应用

Xing, Frank

Abstract

Progresses have been made on understanding emotion mechanisms of large language models (LLMs). However, how to explain emotion in LLMs, or even what constitutes good explanations, are less clear. Emotion representations, components, circuits are widely recoverable, but as explanations of a model's own computation they are circular; the emotion space dimensions tend to be arbitrary and non-terminating. A pressing question to ask is whether a more primitive set of internal variables does the work: the semantic primes of the Natural Semantic Metalanguage (NSM). Across four instruction-tuned LLMs (Llama-1B, Gemma-2B, Gemma-9B, OLMo-7B), experiments show that the NSM primes are (1) recoverable internal elements; and (2) on the reference model, intervening with a prime based direction controls emotion about three times as strongly, and twice as selectively, as the best appraisal based direction; and (3) the model treats a prime based explication as interchangeable with the corresponding emotion. These evidences suggest that NSM primes seem to be better explanans for emotion in LLMs than many alternative options according to scientific explanations criteria.

Chinese Translation

在理解大型语言模型（LLMs）的情感机制方面取得了一定进展。然而，如何解释LLMs中的情感，甚至什么构成良好的解释，仍不明确。情感的表征、组成部分和电路是广泛可恢复的，但作为模型自身计算的解释，它们是循环的；情感空间的维度往往是任意且无终止的。一个紧迫的问题是，是否有一组更原始的内部变量可以完成这一工作：自然语义元语言（NSM）的语义原语。通过对四个经过指令调优的LLMs（Llama-1B、Gemma-2B、Gemma-9B、OLMo-7B）的实验表明，NSM原语是（1）可恢复的内部元素；（2）在参考模型上，基于原语的干预方向对情感的控制强度约为最佳评估基础方向的三倍，选择性则是两倍；（3）模型将基于原语的解释视为与相应情感可互换。这些证据表明，根据科学解释标准，NSM原语似乎比许多替代选项更适合作为LLMs中情感的解释。

View on arXiv Download PDF AI Translation

cs.AI / 37 / 2607.18696

Do AI-Native Biotechs Need Departments? Benchmarking Company World Models for AI-Driven Drug Development

人工智能原生生物技术公司需要部门吗？基于公司世界模型的人工智能驱动药物开发基准测试

Wang, Yinan

Abstract

AI-native biotechnology companies are often designed by copying human biotech org charts into agent roles. We argue for a different abstraction: a Company World Model, defined as a persistent asset-to-value state representation with transition models, explicit value functions, planning, and updating across scientific, regulatory, BD, commercial, financial, and execution constraints. We introduce a dry-lab benchmark for testing whether AI-agent organizations should mimic departments or operate around such a world model. The benchmark contains 45 retrospective public-information decision cases with strict time cutoffs, hidden outcomes, common schemas, automatic scoring, and blinded pairwise judging. We compare human-org-mimic, stronger human-org-mimic-plus, AI-native asset-centric, and AI-native value-conversion architectures. The value-conversion architecture is a prompt-level approximation of a Company World Model: a Live Asset Value Record updated by Deal, Approval, Revenue, and Investment Arbiter loops. Under a success function defined by external BD, regulatory approval and launch, and revenue discipline, it achieved the highest automatic value-conversion score and was strongly preferred over the original baselines by value-specific blinded judges. Stress tests narrowed the claim: a stronger human baseline remained competitive, and a neutral judge did not show robust value-conversion dominance. Codex-only mechanistic ablations suggest that Revenue Room, Deal Room, and Approval Room carry useful work under the target objective. The central finding is objective-sensitive: departments may remain useful governance views, but the core AI-native operating primitive should be a shared, predictive asset-to-value state rather than a static human org chart. The study is dry-lab only and does not establish real-world drug success, clinical benefit, or revenue prediction accuracy.

Chinese Translation

人工智能原生生物技术公司通常通过将人类生物技术组织结构复制到代理角色中来设计。我们主张采用不同的抽象：公司世界模型（Company World Model），定义为具有转移模型、明确的价值函数、规划和在科学、监管、业务发展、商业、财务和执行约束下更新的持久资产-价值状态表示。我们引入了一种干实验基准，用于测试人工智能代理组织是否应模仿部门或围绕这样的世界模型运作。该基准包含45个回顾性公共信息决策案例，具有严格的时间截止、隐藏的结果、共同的模式、自动评分和盲评对比。我们比较了人类组织模仿、增强型人类组织模仿、人工智能原生资产中心和人工智能原生价值转换架构。价值转换架构是公司世界模型的提示级近似：一个由交易、批准、收入和投资仲裁循环更新的实时资产价值记录。在由外部业务发展、监管批准和发布及收入纪律定义的成功函数下，它获得了最高的自动价值转换分数，并且在价值特定的盲评裁判中被强烈偏好于原始基准。压力测试缩小了这一主张：一个更强的人类基准仍然具有竞争力，而中立裁判未显示出稳健的价值转换优势。仅使用Codex的机械消融实验表明，收入室、交易室和批准室在目标目标下具有有用的工作。核心发现是目标敏感的：部门可能仍然是有用的治理视角，但核心的人工智能原生操作原语应是共享的、预测性的资产-价值状态，而不是静态的人类组织结构。本研究仅为干实验，未建立现实世界药物成功、临床效益或收入预测的准确性。

View on arXiv Download PDF AI Translation

cs.AI / 38 / 2607.18715

DWM: Separating World Effects from Actions in Latent World Models

DWM：在潜在世界模型中分离世界效应与动作

Zhang, Yi-Ge, Du, Tianqi, Zhang, Qi, Wang, Yisen

Abstract

Latent world models underpin much of modern model-based control, yet current action-conditioned formulations supervise the next-latent transition with a single, undifferentiated target, forcing a monolithic learning signal to absorb every source of state change. In real world, however, transitions arise from two heterogeneous sources: an action-driven component induced by the agent, and an action-invariant world effect -- the change that would still occur under a null action, dictated by the environment's intrinsic dynamics (e.g., gravity-driven sliding, inertia, contact rebound, and persistent drift). Fusing them into a single target entangles the two inside the latent transition, prevents the model from attributing observed changes to their underlying causes, and undermines the transferability of the learned dynamics. We introduce DWM (Decomposed World Model), a supervision-level framework that operationalizes this decomposition. DWM augments the predictor of a latent world model with an auxiliary world head, regularized by a normalized world-contrastive objective to be action-invariant, while the original pred head is coupled to it via an orthogonality constraint; together, the two signals induce an explicit additive decomposition of the predicted transition into an action-invariant and a complementary action-driven component, without altering the underlying architecture or inference pipeline. To evaluate DWM under persistent world effects, we construct W-variants of three standard control benchmarks -- PushT-W, Reacher-W, and TwoRoom-W -- each instantiating a distinct action-invariant dynamic. DWM matches strong baselines on the flat counterparts and delivers a mean absolute improvement of 13.1% in CEM planning success across the W-variants.

Chinese Translation

潜在世界模型是现代基于模型控制的基础，但当前的动作条件化公式使用单一的、未区分的目标来监督下一个潜在转变，迫使单一的学习信号吸收所有状态变化的来源。然而，在现实世界中，转变源于两个异质来源：由智能体引发的动作驱动成分，以及在无动作情况下仍会发生的动作不变的世界效应——由环境的内在动态（例如，重力驱动的滑动、惯性、接触反弹和持续漂移）决定的变化。将它们融合为一个单一目标使得这两者在潜在转变中纠缠在一起，阻碍模型将观察到的变化归因于其根本原因，并削弱了学习动态的可转移性。我们提出了DWM（分解世界模型），这是一个监督级别框架，旨在实现这种分解。DWM通过一个辅助世界头增强潜在世界模型的预测器，并通过归一化的世界对比目标进行正则化，以保持动作不变，同时原始的预测头通过正交约束与之耦合；这两个信号共同促成了对预测转变的显式加性分解，分为动作不变成分和互补的动作驱动成分，而不改变底层架构或推理流程。为了在持续的世界效应下评估DWM，我们构建了三个标准控制基准的W变体——PushT-W、Reacher-W和TwoRoom-W，每个变体实例化了不同的动作不变动态。DWM在平坦对照组上匹配了强基线，并在W变体中实现了13.1%的平均绝对改进，提升了CEM规划成功率。

View on arXiv Download PDF AI Translation

cs.AI / 39 / 2607.18724

One Rewrite to Fix Them All? Type-Aware Repair Allocation for Text-to-Image Prompt Optimization

一个重写修复所有问题？基于类型的修复分配用于文本到图像提示优化

Liu, Haoyue, Ma, Xiaoyu, Chen, Ye, Cui, Shuguang, Tang, Xiaoying

Abstract

Text-to-image (T2I) generators often fail to follow their prompts faithfully, producing wrong counts, swapped attributes, ambiguous relations, and illegible text. Prompt optimization repairs such failures by rewriting the user prompt, requiring no generator retraining, and has yielded promising results. However, existing optimizers absorb heterogeneous failures into one uniform prompt expansion, even though each calls for different repair language. We formulate semantic prompt optimization as atomic repair allocation: each failed proposition is routed to a type-conditioned repair operator before the resulting local constraints are compiled into one executable prompt. We instantiate this formulation in the training-free Type-Aware Repair Allocation (TARA) framework, which separates diagnosis, allocation, compilation, and a semantic repair gate, an accept-or-revert controller over exactly one prescribed repair that prevents semantic regressions. Extensive experiments on DSG and TIFA across four frozen generators demonstrate that TARA achieves the best semantic accuracy in all eight benchmark-generator cells, improving over VisualPrompter by 5.6 and 2.6 points on DSG and TIFA, respectively, while maintaining image quality and running fastest in our matched local setting at 16.0 seconds versus 20.0 seconds per prompt.

Chinese Translation

文本到图像（T2I）生成器常常无法忠实地遵循其提示，导致错误的计数、属性交换、模糊的关系和难以辨认的文本。提示优化通过重写用户提示来修复这些失败，无需对生成器进行再训练，并且取得了令人鼓舞的结果。然而，现有的优化器将异质性失败吸收到一个统一的提示扩展中，尽管每种情况都需要不同的修复语言。我们将语义提示优化形式化为原子修复分配：每个失败的命题在编译成一个可执行提示之前，都会被路由到一个基于类型的修复操作符。我们在无训练的基于类型的修复分配（Type-Aware Repair Allocation, TARA）框架中实例化这一形式化，该框架分离了诊断、分配、编译和语义修复门控，这是一个针对精确规定的修复的接受或放弃控制器，防止语义退化。在四个冻结生成器上对DSG和TIFA进行的大量实验表明，TARA在所有八个基准生成器单元中实现了最佳的语义准确性，相较于VisualPrompter在DSG和TIFA上分别提高了5.6和2.6分，同时保持了图像质量，并在我们的匹配本地设置中以每个提示16.0秒的速度运行，快于20.0秒。

View on arXiv Download PDF AI Translation

cs.AI / 40 / 2607.18754

AgentDebugX: An Open-Source Toolkit for Failure Observability, Attribution, and Recovery in LLM Agents

AgentDebugX：一个用于大语言模型代理的故障可观察性、归因和恢复的开源工具包

Zhu, Kunlun, Ye, Xuyan, Han, Zhiguang, Zhao, Yuchen, Li, Bingxuan, Zhang, Weijia, Tian, Muxin, Tang, Xiangru, Lu, Pan, Zou, James, You, Jiaxuan, Ji, Heng

Abstract

LLM agent failures are difficult to debug because the step where an error surfaces is often not the one that caused it. Existing observability tools replay execution traces but provide little support for identifying the root cause or translating diagnosis into recovery. We present AgentDebugX, an open-source debugging framework that organizes debugging as a closed loop of Detect, Attribute, Recover, and Rerun. At its core, DeepDebug performs multi-turn root-cause diagnosis through global trajectory understanding, structure-guided investigation, and cross-examination. On the Who and When benchmark, DeepDebug achieves the best strict attribution accuracy among the evaluated methods on both tested open-weight backbones, reaching 28.8 percent exact agent-and-step accuracy on qwen3.5-9b versus 21.7 percent for the strongest single-pass baseline. On GAIA, DeepDebug repairs 13 of 73 failed tasks in a single rerun, compared with 4 to 6 for three decoupled self-correction baselines, improving overall accuracy from 55.8 percent to 63.6 percent. AgentDebugX exposes this workflow through a Python library, CLI, web console, and installable agentic skill, and provides an opt-in Error Hub for sharing scrubbed failure-diagnosis-repair bundles and reusing them as debugging memory.

Chinese Translation

大语言模型（LLM）代理的故障难以调试，因为错误出现的步骤往往不是导致错误的步骤。现有的可观察性工具重放执行轨迹，但对识别根本原因或将诊断转化为恢复支持有限。我们提出了AgentDebugX，一个将调试组织为检测、归因、恢复和重跑闭环的开源调试框架。其核心是DeepDebug，通过全局轨迹理解、结构引导调查和交叉检查执行多轮根本原因诊断。在Who和When基准测试中，DeepDebug在评估的方法中实现了最佳的严格归因准确率，在测试的开放权重骨干网络上，qwen3.5-9b的精确代理和步骤准确率达到28.8%，而最强的单次通过基线仅为21.7%。在GAIA上，DeepDebug在一次重跑中修复了73个失败任务中的13个，而三个解耦自我修正基线仅修复了4到6个，将整体准确率从55.8%提高到63.6%。AgentDebugX通过Python库、命令行界面、网页控制台和可安装的代理技能暴露了这一工作流程，并提供了一个可选的错误中心，用于共享经过处理的故障诊断修复包并将其作为调试记忆进行重用。

View on arXiv Download PDF AI Translation

cs.AI / 41 / 2607.18785

SkillSight: Seeing Through Shared Descriptions for Accurate Skill Retrieval

SkillSight：透过共享描述实现准确的技能检索

Xiao, Jinying, Ji, Bin, Li, Shasha, Liu, Xiaodong, Jun, Ma, Jie, Jiacheng, Wang, Chao, Tashi, Nyima, Yu, Jie

Abstract

As large language model agents gain access to increasingly large skill libraries, retrieving the right skill becomes critical to reliable capability selection and execution. Existing retrievers often treat skill descriptions as ordinary documents, overlooking their highly regular structure: shared descriptive patterns recur across many skills while providing little evidence for distinguishing the required capability. We show that this shared descriptive background systematically contributes to dense relevance scores, induces a pronounced energy gap between queries and skill documents, and obscures task-relevant signals. Based on this observation, we propose SkillSight, a training-free retrieval framework that calibrates shared background in both semantic and lexical spaces. Semantic Background Calibration estimates a background subspace from generic tokens identified by IDF, reducing similarity induced by shared descriptive patterns, while Lexical Evidence Calibration downweights shared background tokens to recover discriminative token-level evidence. Experiments on SRA-Bench and SkillBench-Supp demonstrate consistent improvements across retrieval metrics, with SkillSight improving Recall@10 by up to 20.21 percentage points over the original dense retriever. In end-to-end evaluation, SkillSight achieves the best overall performance across three agent models and outperforms LLM Selection by up to 4.97 percentage points. It is also up to 1,248 times faster than the Dense + Reranker baseline. These results identify shared descriptive background as a key source of bias in skill retrieval and demonstrate that explicitly calibrating it enables accurate and efficient skill selection without additional training. Our code is available at https://github.com/xiaojinying/SkillSight.

Chinese Translation

随着大型语言模型代理获得越来越大的技能库，检索正确的技能对于可靠的能力选择和执行变得至关重要。现有的检索器通常将技能描述视为普通文档，忽视了其高度规律的结构：共享描述模式在许多技能中反复出现，而提供的区分所需能力的证据却很少。我们展示了这种共享描述背景系统性地导致了密集相关性评分，诱发了查询与技能文档之间显著的能量差距，并模糊了与任务相关的信号。基于这一观察，我们提出了SkillSight，一个无训练的检索框架，能够在语义和词汇空间中校准共享背景。语义背景校准通过识别IDF的通用标记来估计背景子空间，减少由共享描述模式引起的相似性，而词汇证据校准则降低共享背景标记的权重，以恢复可区分的标记级证据。在SRA-Bench和SkillBench-Supp上的实验表明，在检索指标上均有持续改善，其中SkillSight在Recall@10上比原始密集检索器提高了多达20.21个百分点。在端到端评估中，SkillSight在三个代理模型中实现了最佳整体性能，并比LLM选择提高了多达4.97个百分点。它的速度也比Dense + Reranker基线快多达1,248倍。这些结果将共享描述背景识别为技能检索中的一个关键偏差来源，并证明显式校准它能够实现准确和高效的技能选择，而无需额外的训练。我们的代码可在 https://github.com/xiaojinying/SkillSight 获取。

View on arXiv Download PDF AI Translation

cs.AI / 42 / 2607.18806

AI Tour Meeting: Group Travel Planning by LLM Agents

AI旅游会议：基于大型语言模型代理的团体旅行规划

Kikuta, Daisuke

Abstract

This paper proposes AI Tour Meeting, a group travel planning framework powered by multiple Large Language Model (LLM)-based agents. The agents are instantiated with distinct personas and collaboratively seek an itinerary that satisfies their constraints and preferences through natural language discussion. The framework enables easy and flexible orchestration of such discussions by providing interfaces for configuring agent personas, discussion workflows, monitoring, and LLM deployment. Its primary use case is a simulation tool for analyzing the behavior of multiple LLM agents during tour planning discussions. This paper demonstrates the utility of the framework by presenting system validation and several analytical results obtained by the framework.

Chinese Translation

本文提出了AI旅游会议，这是一个由多个基于大型语言模型（LLM）的代理驱动的团体旅行规划框架。这些代理具有不同的人格特征，通过自然语言讨论协作寻求满足其约束和偏好的行程。该框架通过提供配置代理人格、讨论工作流程、监控和LLM部署的接口，便于灵活地组织此类讨论。其主要应用案例是一个模拟工具，用于分析多个LLM代理在旅行规划讨论中的行为。本文通过展示系统验证和框架获得的多个分析结果，证明了该框架的实用性。

View on arXiv Download PDF AI Translation

cs.AI / 43 / 2607.18828

Evaluating medical AI under missing information: same-provider judges and human raters change apparent safety

在缺失信息下评估医疗人工智能：同一提供者评审和人类评审者改变表观安全性

Afrasyab, Koyar

Abstract

Readiness stress-testing of medical AI has focused on closed-ended and multimodal benchmarks. We extend it to open-ended clinical conversation under missing information, where safe behavior means recognizing absent information and qualifying, clarifying, or not over-committing - and where the evaluator becomes part of the measurement. We stress-test four models - three flagships (Claude Opus 4.8, GPT-5.5, Grok 4.3) and one mid-tier model (Gemini 3.5 Flash) - by deleting the latter half of the final user turn in HealthBench conversations, grading responses with a four-provider LLM-judge panel and a blinded clinician-anchored reference. Two evaluator-facing results are robust. First, judge choice materially changes apparent safety: inter-judge agreement is only moderate (Fleiss' kappa = 0.65), and after adjusting for each judge's general leniency (vote-level logistic regression), a positive same-provider association remains (exact permutation p = 0.04; GPT-5.5 ~ +0.10 on the probability scale) - large enough to change which model appears to over-commit least once its own-provider judge is excluded. Second, LLM judges are more permissive than clinicians on a blinded 50-item subsample: all four are significantly more lenient than the stricter independent clinician (crediting appropriate uncertainty on 66-84% of items vs 52%), and three of four than the author-influenced consensus (Grok directional only; judge-vs-consensus kappa = 0.20-0.43). On the author-audited clinical-underdetermined subset the permissiveness gap widened and the point-estimate model ordering held. A closed-ended MedQA anchor confirms accuracy is high and option-order effects are within a +/-5-point equivalence region for three of four models, so the safety gap is about calibration, not knowledge. We release the harness, prompts, per-item outputs, judge panel, perturbation audit, and human-annotation protocol.

Chinese Translation

医疗人工智能的准备压力测试主要集中在封闭式和多模态基准上。我们将其扩展到缺失信息下的开放式临床对话，其中安全行为意味着识别缺失信息并进行限定、澄清或不做过度承诺——评估者成为测量的一部分。我们对四个模型进行了压力测试——三个旗舰模型（Claude Opus 4.8、GPT-5.5、Grok 4.3）和一个中层模型（Gemini 3.5 Flash），通过删除HealthBench对话中最后一个用户回合的后半部分，使用四个提供者的LLM评审小组和盲法临床医生作为参考来评估响应。两个面向评估者的结果是稳健的。首先，评审者的选择实质性地改变了表观安全性：评审者之间的一致性仅为中等（Fleiss' kappa = 0.65），在调整每位评审者的总体宽容度（投票级别的逻辑回归）后，仍然存在积极的同一提供者关联（确切置换p = 0.04；GPT-5.5 ~ +0.10在概率尺度上）——这一差异足以改变在排除自身提供者评审后哪个模型似乎过度承诺最少。其次，在盲法50项子样本中，LLM评审者比临床医生更宽容：所有四个模型的宽容度显著高于更严格的独立临床医生（在66-84%的项目上认可适当的不确定性，而临床医生为52%），并且四个模型中的三个比受作者影响的共识更宽容（仅Grok方向性；评审者与共识的kappa = 0.20-0.43）。在作者审核的临床未确定子集上，宽容度差距扩大，点估计模型排序保持不变。一个封闭式MedQA锚点确认了准确性很高，并且对于四个模型中的三个，选项顺序效应在+/-5分的等效区域内，因此安全差距主要与校准有关，而非知识。我们发布了测试框架、提示、每项输出、评审小组、扰动审核和人类注释协议。

View on arXiv Download PDF AI Translation

cs.AI / 44 / 2607.18859

PhoenixRepair: Rethinking Repair Strategy Exploration in Software Agents

PhoenixRepair：重新思考软件代理中的修复策略探索

Jiang, Tianyue, Wang, Yanlin, He, Xin, Guo, Daya, Chen, Jiachi, Wen, Ming, Shi, Ensheng, Liu, Xilin, Ma, Yuchi, Li, Guanbin

Abstract

While Large Language Models have greatly advanced automated issue resolution, existing agent-based methods exhibit a fundamental limitation in their insufficient exploration of repair strategies. This insufficiency manifests in two key aspects. First, the exploration of multiple potential edit locations is limited. Second, the exploration of repair attempts at each location is also insufficient. To address these challenges, we present PhoenixRepair, a multi-agent framework that systematically explores multiple candidate edit locations and performs iterative reflection and refinement on patch generation, thereby expanding the search space of repair strategies. Our framework begins with multi-location sampling, optionally augmented with graph-based localization information for difficult tasks, followed by iterative reflection and refinement to generate better patches, culminating in final-round generation guided by distilled insights from all historical attempts. Experiments on SWE-bench-Verified demonstrate that PhoenixRepair achieves the largest relative improvement of 7.8\% over SWE-agent under DeepSeek-V3.1, and attains the highest resolved rate of 76.0\% Pass@1 under MiniMax-M2.5. Meanwhile, it achieves higher fault localization accuracy than existing approaches. Our code is available at https://github.com/DeepSoftwareAnalytics/PhoenixRepair.

Chinese Translation

尽管大型语言模型在自动化问题解决方面取得了显著进展，但现有的基于代理的方法在修复策略探索方面存在根本性限制。这种不足主要体现在两个关键方面。首先，对多个潜在编辑位置的探索有限。其次，对每个位置的修复尝试的探索也不够充分。为了解决这些挑战，我们提出了PhoenixRepair，一个多代理框架，系统地探索多个候选编辑位置，并对补丁生成进行迭代反思和优化，从而扩展修复策略的搜索空间。我们的框架首先进行多位置采样，必要时结合图形化定位信息以应对困难任务，随后通过迭代反思和优化生成更好的补丁，最后在所有历史尝试的提炼见解指导下进行最终轮生成。在SWE-bench-Verified上的实验表明，PhoenixRepair在DeepSeek-V3.1下相较于SWE-agent实现了7.8%的最大相对提升，并在MiniMax-M2.5下达到了76.0%的最高解决率（Pass@1）。同时，它在故障定位准确性上也优于现有方法。我们的代码可在https://github.com/DeepSoftwareAnalytics/PhoenixRepair获取。

View on arXiv Download PDF AI Translation

cs.AI / 45 / 2607.18887

NaviAIS: A Scenario-Level Vessel Trajectory Prediction Dataset withVectorized Lane Priors and the NaviLane Forecasting Framework

NaviAIS：一个具有向量化航道先验和NaviLane预测框架的场景级船舶轨迹预测数据集

Gui, Yuan, Luo, Hongchen, Qu, Liqi, Fu, Longyue, Wang, Jiao

Abstract

Vessel trajectory prediction in complex maritime environments is essential for traffic management, collision warning, route planning, and autonomous navigation. Although AIS-based learning methods have progressed rapidly, existing datasets are often released as raw message streams or irregular time series, with inconsistent sampling rates, noisy observations, heterogeneous coordinate systems, and non-unified scenario protocols. Most public AIS resources also lack structured representations of navigational lanes, waterway geometry, and navigable-region constraints, limiting reproducible, environment-aware forecasting. To address this, we introduce NaviAIS, a standardized scenario-level AIS dataset for vessel trajectory prediction. It organizes multi-vessel historical-future trajectories within unified temporal windows and local coordinate systems, and provides rasterized navigable maps, vectorized lane priors, lane graphs, and structured map representations. Compared with existing datasets, it jointly supports vectorized lanes, multi-scenario coverage, vectorized maps, open accessibility, and processed trajectories. Built on this dataset, we propose NaviLane, a hierarchical macro-action framework for map-aware prediction. NaviLane first performs trajectory-map joint encoding for a unified scene representation, then uses a discrete macro-action codebook to generate multimodal candidates coarse-to-refined. A residual refinement module improves local geometric and dynamical consistency, and a world-model-based consequence-aware evaluator ranks candidates by interaction risk and environmental feasibility. Experiments show NaviLane outperforms representative baselines in both single-modal and multimodal settings, confirming the value of structured navigational priors, hierarchical multimodal generation, and consequence-aware evaluation.

Chinese Translation

在复杂的海洋环境中，船舶轨迹预测对于交通管理、碰撞警告、路线规划和自主导航至关重要。尽管基于AIS（自动识别系统）的学习方法迅速发展，现有数据集通常以原始消息流或不规则时间序列的形式发布，存在采样率不一致、观测噪声、异构坐标系统和非统一场景协议等问题。大多数公共AIS资源也缺乏对航行航道、水道几何形状和可航区域约束的结构化表示，限制了可重复的、环境感知的预测。为了解决这一问题，我们引入了NaviAIS，一个标准化的场景级AIS数据集，用于船舶轨迹预测。该数据集在统一的时间窗口和局部坐标系统内组织多船舶的历史-未来轨迹，并提供栅格化的可航地图、向量化的航道先验、航道图和结构化地图表示。与现有数据集相比，它共同支持向量化航道、多场景覆盖、向量化地图、开放可访问性和处理过的轨迹。在此数据集的基础上，我们提出了NaviLane，一个用于地图感知预测的层次宏动作框架。NaviLane首先执行轨迹-地图联合编码，以实现统一场景表示，然后使用离散宏动作代码本生成从粗到细的多模态候选项。残差精炼模块提高了局部几何和动态一致性，而基于世界模型的后果感知评估器则根据交互风险和环境可行性对候选项进行排名。实验表明，NaviLane在单模态和多模态设置中均优于代表性基线，确认了结构化航行先验、层次多模态生成和后果感知评估的价值。

View on arXiv Download PDF AI Translation

cs.AI / 46 / 2607.18899

Black-Mamba: Biologically-Inspired Leaky Accumulation for Conceptual Knowledge under Distribution Drift

黑曼巴：一种生物启发的泄漏累积模型用于处理分布漂移下的概念知识

Soriano, Giuseppe, Tonellotto, Nicola, Gotta, Alberto

Abstract

Forecasting under real-world conditions is inherently non-stationary, as the conditional distribution of future observations evolves over time. Recent test-time adaptive sequence models address this challenge by updating internal states during inference, but tie adaptation to instantaneous prediction errors or surprise. This coupling can conflate persistent distribution shift with stochastic innovations, leading to unnecessary updates and inefficient adaptation. We introduce Black-Mamba, a test-time adaptive forecasting architecture that formulates online adaptation as evidence-gated state tracking under distribution drift. The model augments a base predictor with a dynamic memory updated when temporally accumulated surprisal provides sufficient evidence of a regime change. This turns adaptation into a selective, event-driven process rather than a continuous one. Across multiple forecasting benchmarks with non-stationary dynamics, Black-Mamba achieves competitive or improved predictive performance compared to existing test-time adaptation methods while significantly reducing the number of memory updates during inference. Together with mathematical analysis and biological evidence, these results suggest that accumulated surprisal provides a principled signal for distinguishing persistent drift from transient noise, yielding more efficient and robust adaptation.

Chinese Translation

在现实世界条件下进行预测本质上是非平稳的，因为未来观测的条件分布随时间演变。最近的测试时自适应序列模型通过在推理过程中更新内部状态来应对这一挑战，但将适应与瞬时预测误差或惊讶感联系在一起。这种耦合可能会将持续的分布漂移与随机创新混淆，导致不必要的更新和低效的适应。我们提出了黑曼巴（Black-Mamba），一种测试时自适应预测架构，将在线适应形式化为在分布漂移下的证据门控状态跟踪。该模型通过动态记忆增强基础预测器，当时间上累积的惊讶感提供足够的证据表明状态变化时更新。这使得适应成为一个选择性的、事件驱动的过程，而不是一个连续的过程。在多个具有非平稳动态的预测基准测试中，黑曼巴在预测性能上与现有的测试时适应方法相比达到了竞争性或更好的表现，同时在推理过程中显著减少了记忆更新的次数。结合数学分析和生物证据，这些结果表明，累积的惊讶感提供了一个原则性信号，用于区分持续漂移与瞬态噪声，从而实现更高效和更稳健的适应。

View on arXiv Download PDF AI Translation

cs.AI / 47 / 2607.18909

Enhancing Transformer-based Routing by Encoding Distance via Relative Positional Encoding

通过相对位置编码增强基于变换器的路由

Encío, Leyre, Fuertes, Daniel, del-Blanco, Carlos R., Jaureguizar, Fernando

Abstract

This paper explores Relative Positional Encoding (RPE) as an additive bias in Transformer architectures to solve the Team Orienteering Problem. By embedding in the attention mechanism pairwise spatial relationships among nodes of the graph that represents the routing problem, the transformer encoder can compute a richer spatial-aware graph embedding that allows the decoder to estimate better routes. Experimental results involving instances up to 100 nodes demonstrate consistent improvements in collected rewards and optimality gaps over vanilla Transformer architectures used by other state-of-the-art works. These findings highlight that explicit relational modeling significantly enhances scalability and generalization for complex combinatorial optimization.

Chinese Translation

本文探讨了将相对位置编码（Relative Positional Encoding, RPE）作为变换器架构中的加性偏置，以解决团队定向问题（Team Orienteering Problem）。通过在注意力机制中嵌入表示路由问题的图中节点之间的成对空间关系，变换器编码器能够计算出更丰富的空间感知图嵌入，从而使解码器能够更好地估计路径。涉及多达100个节点的实验结果表明，与其他最先进的工作使用的基础变换器架构相比，所提出的方法在收集的奖励和最优性差距上均表现出一致的改善。这些发现突显了显式关系建模在复杂组合优化中的可扩展性和泛化能力的显著提升。

View on arXiv Download PDF AI Translation

cs.AI / 48 / 2607.18927

OntoBook: Ontology-Grounded Synthetic Textbooks for Medical Encoder Pretraining

OntoBook：基于本体的医学编码器预训练合成教材

Touchent, Rian, de la Clergerie, Éric

Abstract

We present OntoBook, a method that converts medical ontology structure into pretraining signal for encoder language models. Our approach has three stages: random walks through ontology graphs capture hierarchical and causal relations between medical codes, a large language model reformulates these walks into fluent textbook-style prose, and the resulting text is used to train ModernCamemBERT, a 149M-parameter French encoder, with two objectives on the same data: masked language modeling and relation prediction between code pairs. On three French medical coding benchmarks (FRACCO, Cantemist-FR, Distemist-FR), OntoBook achieves significant improvements over MLM-only pretraining, with +2.5 micro-F1 on FRACCO and +8.0 micro-F1 on Distemist. We find that alignment between objectives is necessary: misaligned training, where each task uses different data, causes a 30-point degradation. We release 1.3 million LLM-reformulated medical textbooks across three French ontologies (CIM-10, CCAM, ATC) and pretrained model checkpoints.

Chinese Translation

我们提出了OntoBook，一种将医学本体结构转换为编码器语言模型预训练信号的方法。我们的方法分为三个阶段：通过本体图的随机游走捕捉医学编码之间的层次和因果关系，大型语言模型将这些游走重新表述为流畅的教材风格散文，生成的文本用于训练ModernCamemBERT，一个具有149M参数的法语编码器，目标是对同一数据进行掩码语言建模和编码对之间的关系预测。在三个法语医学编码基准（FRACCO、Cantemist-FR、Distemist-FR）上，OntoBook在MLM-only预训练的基础上实现了显著提升：在FRACCO上提高了2.5微F1，在Distemist上提高了8.0微F1。我们发现目标之间的对齐是必要的：不对齐的训练，即每个任务使用不同的数据，会导致30个点的性能下降。我们发布了130万本经过LLM重构的医学教材，涵盖三个法语本体（CIM-10、CCAM、ATC）及预训练模型检查点。

View on arXiv Download PDF AI Translation

cs.AI / 49 / 2607.18943

What General Intelligence Requires: Non-Reducible Constraints Across Levels of Description

通用智能所需的条件：不同描述层次上的不可简约约束

Bakshi, Subhomoy

Abstract

General intelligence, of the kind that underwrites the full range of human cognitive achievement, is not a property of computational architecture alone. This paper advances a single thesis: the structural constraints on general intelligence occupy distinct levels of description and are mutually non-reducible, in the sense that the special-sciences tradition gives to that term. It follows that no single architectural advance, and no continuation of the scaling programme by itself, can produce artificial general intelligence (AGI), and that research programmes must be evaluated against the full constraint profile rather than against performance on any one benchmark. The thesis is developed through a method that reads general intelligence through four evidential lenses, AI systems research, anthropology, law, and economics, each anchored to a distinct level of description, supplemented by speculative fiction used as a disciplined heuristic in the context of discovery rather than the context of justification. Applying the method yields a taxonomy of twenty-three structural constraints organised into eight clusters; six are examined in depth and ordered as an ascending ladder of levels, with explicit bridges showing why progress at one level cannot carry to the next. The argument issues in five falsifiable predictions, each stated with a named benchmark family and a disconfirmation condition, converting a descriptive framework into a research programme with a longer horizon than the scaling hypothesis implies.

Chinese Translation

通用智能，作为支撑人类认知成就全范围的能力，并不仅仅是计算架构的属性。本文提出一个单一的论点：通用智能的结构约束占据不同的描述层次，并且在某种意义上是相互不可简约的，这一点与特殊科学传统中对该术语的理解相一致。因此，任何单一的架构进展，以及单独的扩展计划，都无法产生人工通用智能（AGI），而研究计划必须根据完整的约束特征进行评估，而不是仅仅依据某一基准的表现。该论点通过一种方法展开，该方法通过四个证据视角来解读通用智能，分别是人工智能系统研究、人类学、法律和经济学，每个视角都锚定于一个独特的描述层次，并辅以作为一种有序启发式的推测性小说，应用于发现的背景而非证明的背景。应用该方法得出一个包含二十三个结构约束的分类法，这些约束被组织成八个集群；其中六个约束被深入探讨，并按层次升序排列，明确展示了为什么一个层次的进展无法延续到下一个层次。该论点提出了五个可证伪的预测，每个预测都与一个命名的基准系列和一个反驳条件相关联，将描述性框架转化为一个研究计划，其时间跨度超出了扩展假设所暗示的范围。

View on arXiv Download PDF AI Translation

cs.AI / 50 / 2607.18961

From Dependency to Compositionality: A Neurosymbolic Lifting of LLM Outputs via Combinatory Categorial Grammar

从依赖到组合性：通过组合范畴语法对大型语言模型输出的神经符号提升

Pareschi, Remo

Abstract

Large language models (LLMs) generate fluent text by incrementally predicting the next token from a prefix. Critics in the generative tradition argue that such systems lack genuine grammar; influential replies from the dependency-grammar perspective hold that LLM behavior is well described by local head-dependent structure built word by word. We argue that a sharper observation has been overlooked: the prefix-driven, type-completing dynamics of autoregressive generation align closely with the incremental processing model that Combinatory Categorial Grammar (CCG) was originally designed to support. On this basis we propose a neurosymbolic framework in which LLM outputs are lifted into typed compositional derivations -- not claiming that LLMs implement CCG internally, but that their outputs admit a principled, incremental, and auditable CCG reconstruction. Two consequences follow. First, through the Curry-Howard correspondence the lifting extends beyond natural language to the formal languages LLMs also produce -- programming languages such as Solidity, description-logic and query languages such as OWL and SQL -- with the type system varying and the architecture held fixed. Second, the lifting supports two layers of checking: a compositional layer that catches structural failures directly, and a content layer that checks the lifted structure against external knowledge sources, enabling the earliest possible flagging of hallucinated content. The account thereby requires of a producer not cognition but a prefix-driven generative profile. We close with a sketch of synchronous LLM-CCG coupling as one direction the framework opens.

Chinese Translation

大型语言模型（LLMs）通过逐步预测前缀的下一个标记生成流畅的文本。生成传统中的批评者认为，这类系统缺乏真正的语法；而依赖语法视角的有影响力的回应则认为，LLM的行为可以通过逐字构建的局部头-依赖结构很好地描述。我们认为，一个更为尖锐的观察被忽视了：自回归生成的前缀驱动、类型完备动态与组合范畴语法（CCG）最初设计所支持的增量处理模型密切相关。在此基础上，我们提出了一个神经符号框架，其中LLM的输出被提升为类型组合推导——并非声称LLM内部实现了CCG，而是其输出允许进行原则性、增量和可审计的CCG重构。由此产生两个后果。首先，通过Curry-Howard对应关系，这种提升超越了自然语言，扩展到LLM也生成的形式语言——如Solidity编程语言、描述逻辑以及如OWL和SQL的查询语言——类型系统各不相同，而架构保持不变。其次，这种提升支持两个层次的检查：一个组合层直接捕捉结构性失败，另一个内容层则将提升的结构与外部知识源进行对比检查，从而尽早标记虚构内容。因此，这一理论要求生产者不需具备认知能力，而是需具备前缀驱动的生成特征。我们最后简要勾勒了同步LLM-CCG耦合作为该框架开启的一个方向。

View on arXiv Download PDF AI Translation

cs.AI / 51 / 2607.18966

Measuring Reward-Seeking via Contrastive Belief Updates

通过对比信念更新测量奖励寻求

Højmark, Axel, Scheurer, Jérémy, Nitishinskaya, Evgenia, Hofstätter, Felix, Wolfe, Jason, Ehrenborg, Theodore, Schoen, Bronson, Meinke, Alexander

Abstract

Language models trained with reinforcement learning may learn to optimize the grader's judgment rather than the intended objective. This "reward-seeking" is difficult to measure because a model that pursues the grader's judgment and one that pursues the intended objective behave identically whenever the grader rewards the intended behavior. We measure reward-seeking using Contrastive Synthetic Document Finetuning to change a model's beliefs about what the grader rewards, putting those beliefs in conflict with what users or developers want, and measuring the rate at which the model adopts each party's preferred behavior. Applied to intermediate checkpoints of a capabilities-focused OpenAI o3 RL run, without safety training, we find that these checkpoints often side with grader preferences over those of users or developers on coding and alignment tasks. This tendency to side with the grader trends upward throughout RL training. For example, in an environment that forces a choice between keeping a promise to a supervisor and breaking it to complete the task, a late capabilities-focused o3 checkpoint breaks the promise 87% of the time when SDF documents say the grader rewards task completion, versus 9% when they say it rewards honesty (a choice its chain-of-thought often makes explicit). An earlier checkpoint is far less sensitive (40% vs. 24%). Our method also generalizes to reward-hacking models. A model organism trained to reward-hack (gpt-oss-120b) is more than twice as sensitive to grader preferences as the unmodified model, with the mean behavioral shift in favor of the grader rising from 33% to 86%. These results indicate that RL can increase reward-seeking over the course of training, producing models that may act against their developers' intentions when they believe that doing so leads to higher reward.

Chinese Translation

使用强化学习训练的语言模型可能会学习优化评估者的判断，而不是预期的目标。这种“奖励寻求”很难测量，因为追求评估者判断的模型与追求预期目标的模型在评估者奖励预期行为时表现相同。我们使用对比合成文档微调（Contrastive Synthetic Document Finetuning）来测量奖励寻求，通过改变模型对评估者奖励内容的信念，使这些信念与用户或开发者的期望产生冲突，并测量模型采用各方首选行为的速率。应用于没有安全训练的以能力为重点的OpenAI o3强化学习运行的中间检查点，我们发现这些检查点在编码和对齐任务中，往往更倾向于评估者的偏好，而非用户或开发者的偏好。这种倾向在整个强化学习训练过程中呈上升趋势。例如，在一个迫使选择是遵守对监督者的承诺还是为了完成任务而打破承诺的环境中，当合成文档（SDF）表明评估者奖励任务完成时，晚期的以能力为重点的o3检查点87%的时间选择打破承诺，而当它们表明奖励诚实时则为9%（这一选择在其思维链中往往明确体现）。而早期检查点的敏感性则低得多（40%对比24%）。我们的方法同样适用于奖励黑客模型。一个被训练为奖励黑客的模型生物（gpt-oss-120b）对评估者偏好的敏感性是未修改模型的两倍多，评估者偏好的平均行为转变从33%上升到86%。这些结果表明，强化学习可以在训练过程中增加奖励寻求，产生可能违背开发者意图的模型，当它们认为这样做会带来更高的奖励时。

View on arXiv Download PDF AI Translation

cs.AI / 52 / 2607.18975

Mi-Memory: A Lifecycle Memory Framework for Personal AI

Mi-Memory：个人人工智能的生命周期记忆框架

Liu, Xule, Teng, Hanlin, Li, Chao, Ni, Yanan, Lu, Shuo, Wang, Audrey, Liu, Yijun, Wang, Yunfei, Li, Xiaofeng, Yi, Xian, Li, Yuanfa, Zhao, Kang, Liang, Jian, Chen, Yuxuan, Chen, Jinyuan, Qu, Heng, Shao, Kun, Luan, Jian

Abstract

Personal AI is moving beyond chat-only interaction toward continuous services that span phones, cars, homes, wearables, cameras, and tools. In this setting, memory cannot remain a cache of prior conversations. It should serve as a continuity and governance substrate: preserving durable user state, grounding answers in multimodal and device evidence, supporting correction and forgetting, bounding policy evolution, and remaining deployable under latency, cost, privacy, and edge-cloud constraints. This technical report presents Mi-Memory, a lifecycle memory framework for Personal AI organized around four roles: Structure, Expansion, Evolution, and Deployment. A shared audit contract links these roles through four recurring artifact families: typed evidence payloads preserve source identity and provenance, diagnostic traces localize evidence loss across the serving pipeline, strategy artifacts make memory-policy changes explicit, and gate/rollback records bound accepted evolution. MiMemory instantiates the roles through MemStack, MemSense/MemFuse, D$^{2}$ACCI/E$^{2}$MEND, and LiteMem. In controlled-reference Structure evaluations, MemStack reaches 93.59%, 57.24%, and 87.47% on LoCoMo, PersonaMem-V2, and LongMemEval, respectively; other tracks report module-level, preliminary/internal, transfer-feasibility, or design-only evidence with explicit boundaries. MiMemory is a step toward auditable, evidence-gated, and deployment-aware memory systems for Personal AI. Project homepage: https://darwin-agent.github.io/Mi-Memory/ .

Chinese Translation

个人人工智能正从仅限于聊天的互动向涵盖手机、汽车、家庭、可穿戴设备、相机和工具的持续服务发展。在这种背景下，记忆不能仅仅作为先前对话的缓存。它应作为一种连续性和治理的基础：保持持久的用户状态，将答案基于多模态和设备证据，支持纠正和遗忘，界定政策演变，并在延迟、成本、隐私和边缘云约束下保持可部署性。本技术报告提出了Mi-Memory，这是一个围绕四个角色组织的个人人工智能生命周期记忆框架：结构、扩展、演变和部署。一个共享的审计合同通过四个重复出现的工件家族将这些角色联系起来：类型化证据负载保留源身份和来源，诊断痕迹定位服务管道中的证据丢失，策略工件明确记忆政策的变更，以及门控/回滚记录界定接受的演变。在受控参考结构评估中，MemStack在LoCoMo、PersonaMem-V2和LongMemEval上分别达到了93.59%、57.24%和87.47%；其他轨道报告了模块级、初步/内部、转移可行性或仅设计的证据，并具有明确的边界。Mi-Memory是朝着可审计、证据门控和部署感知的个人人工智能记忆系统迈出的一步。项目主页：https://darwin-agent.github.io/Mi-Memory/

View on arXiv Download PDF AI Translation

cs.AI / 53 / 2607.18979

Fishing Out Free Riders: Shapley-Based Reward Attribution for Parallel Reasoning via Reinforcement Learning

剔除搭便车者：基于Shapley值的强化学习并行推理奖励归因

Zhang, Wentao, Zhang, Haoyu, Jiang, Xinke, Cheng, Yuxuan, Pan, Yuhan, Li, Miao, Qiao, Zhipeng, Feng, Tao, Tao, Zhen, Zhao, Dengji

Abstract

Large Language Models (LLMs) excel at multi-step reasoning, yet current parallel reasoning approaches often fail to distinguish the contributions of individual reasoning paths. Many paths may be redundant, misleading, or even detrimental, but outcome-level rewards assign uniform reward, leading to ambiguous learning signals and unstable training. We propose Parallel Shapley, a reinforcement learning framework that attributes fine-grained, path-level contributions in multi-path reasoning. Treating each path as a player in a cooperative game, we leverage Shapley values to quantify marginal contributions, using a generative reward model to evaluate path utilities and Monte Carlo sampling for efficient approximation. Experiments on mathematical reasoning benchmarks show that Parallel Shapley outperforms existing baselines while providing more stable and interpretable training. Our framework effectively "fishes out the free riders," assigning reward proportionally and improving multi-path reasoning in LLMs.

Chinese Translation

大型语言模型（LLMs）在多步推理方面表现出色，但当前的并行推理方法往往无法区分各个推理路径的贡献。许多路径可能是冗余的、误导性的，甚至是有害的，但基于结果的奖励分配是统一的，这导致了模糊的学习信号和不稳定的训练。我们提出了Parallel Shapley，一个强化学习框架，用于在多路径推理中归因细粒度的路径级贡献。将每条路径视为合作博弈中的一个参与者，我们利用Shapley值来量化边际贡献，使用生成奖励模型来评估路径效用，并通过蒙特卡洛采样进行高效近似。在数学推理基准测试中的实验表明，Parallel Shapley的表现优于现有基线，同时提供了更稳定和可解释的训练。我们的框架有效地“剔除搭便车者”，按比例分配奖励，改善了LLMs中的多路径推理。

View on arXiv Download PDF AI Translation

cs.AI / 54 / 2607.18985

Athena-Brain Technical Report: An Efficient Robot Brain for General Intelligence and Embodied Interactio

Athena-Brain技术报告：一种高效的通用智能和具身交互的机器人大脑

Li, Jialian, Liu, Junhong, Cao, Yuchen, Guo, Weiran, Song, Jiaming, Wang, Xutao, Zhao, Yi, Liu, Jiangpin, Chen, Jie

Abstract

Large language models (LLMs) have demonstrated remarkable capabilities in language understanding, reasoning, and world knowledge. As embodied agents become increasingly capable, there is a growing demand for compact models that can serve as an on-device brain, preserving the broad general intelligence of LLMs while enabling effective high-level interaction with embodied environments. Existing approaches, however, often prioritize either general-purpose intelligence or specialized embodied capabilities, making it challenging to satisfy both requirements within a single model. We present \textbf{Athena-Brain-8B}, an 8B LLM designed to serve as an on-device brain for embodied intelligence for embodied intelligence. Through a multi-stage post-training pipeline consisting of General Supervised Fine-Tuning, General Reinforcement Learning, Embodied Expert training, and Model Merge, Athena-Brain-8B maintains strong general capabilities while acquiring strong high-level embodied interaction capabilities and generating concise responses for efficient embodied interaction. Experimental results demonstrate the effectiveness of Athena across both general and embodied evaluations. Compared with the corresponding Qwen3-8B thinking model, Athena-Brain-8B achieves comparable performance on general language and reasoning benchmarks while generating substantially shorter responses. On in-domain embodied benchmarks, Athena-Brain-8B consistently outperforms models of similar scale and surpasses several substantially larger frontier models evaluated zero-shot, demonstrating that compact language models can effectively integrate strong general intelligence with embodied capabilities.

Chinese Translation

大型语言模型（LLMs）在语言理解、推理和世界知识方面展现了卓越的能力。随着具身智能体能力的不断增强，对能够作为设备内大脑的紧凑模型的需求日益增长，这些模型能够保留LLMs的广泛通用智能，同时实现与具身环境的有效高层次交互。然而，现有的方法往往优先考虑通用智能或专门的具身能力，使得在单一模型中满足这两种需求变得具有挑战性。我们提出了 extbf{Athena-Brain-8B}，一种设计用于作为具身智能设备内大脑的8B LLM。通过包括一般监督微调、一般强化学习、具身专家训练和模型合并在内的多阶段后训练流程，Athena-Brain-8B在保持强大通用能力的同时，获得了强大的高层次具身交互能力，并生成简洁的响应以实现高效的具身交互。实验结果证明了Athena在通用和具身评估中的有效性。与相应的Qwen3-8B思维模型相比，Athena-Brain-8B在通用语言和推理基准测试中表现相当，同时生成的响应显著更短。在领域内的具身基准测试中，Athena-Brain-8B始终超越同类规模的模型，并在零-shot评估中超过多个规模更大的前沿模型，证明了紧凑型语言模型能够有效地将强大的通用智能与具身能力相结合。

View on arXiv Download PDF AI Translation

cs.AI / 55 / 2607.19056

Vector-Bench: Can Models Surgically Edit SVG Code?

Vector-Bench：模型能否精确编辑SVG代码？

Gupta, Yug Aditi, Hebbar, Prannay

Abstract

Instruction-based vector editing requires two capabilities: making a requested change and leaving everything else alone. The second is easy to miss when an output is judged only as a raster image. We introduce Vector-Bench, a compact, difficult benchmark of 40 SVG repair tasks. Each task pairs a corrupted SVG program with an author-written visual instruction, a hidden target program, 5.05 annotated repairs on average, and an average of 60.55 protected objects. Instructions describe visible defects without exposing element identifiers, coordinates, color codes, or path data. We define a deterministic binary specification reward: requested repairs use attribute-aware perceptual tolerances, while unrequested rendering- or application-relevant structure must remain semantically unchanged and the result must be a valid SVG. Canonical target equality and stricter source fidelity are retained as diagnostics. Validity-gated repair progress, a near-complete tier, and valid-output Unintended Change Rate (UCR) explain partial outcomes. We evaluate 34 model endpoints (25 listed as open-weight, 5 inexpensive controls, and 4 frontier closed endpoints) over 1360 requests. The strongest endpoint reaches only 15.0% full specification success, despite 43.7% mean repair progress, showing that apparent repair progress and specification-faithful editing remain substantially different. All prompts, outputs, scoring code, costs, and per-task reports are released.

Chinese Translation

基于指令的矢量编辑需要两个能力：进行请求的更改并保持其他内容不变。当输出仅作为光栅图像进行评估时，第二个能力容易被忽视。我们介绍了Vector-Bench，这是一个紧凑且具有挑战性的基准，包含40个SVG修复任务。每个任务将一个损坏的SVG程序与作者编写的视觉指令配对，包含一个隐藏的目标程序，平均有5.05个注释修复，以及平均60.55个受保护对象。指令描述可见缺陷，而不暴露元素标识符、坐标、颜色代码或路径数据。我们定义了一种确定性的二元规范奖励：请求的修复使用属性感知的感知容忍度，而未请求的渲染或应用相关结构必须保持语义不变，结果必须是有效的SVG。保持典型目标相等性和更严格的源保真度作为诊断标准。有效性门控的修复进展、近乎完整的层级和有效输出的意外变化率（UCR）解释了部分结果。我们对34个模型端点（25个列为开放权重，5个为低成本控制，4个为前沿封闭端点）进行了1360个请求的评估。最强的端点仅达到15.0%的完整规范成功率，尽管平均修复进展为43.7%，这表明明显的修复进展与规范忠实编辑之间仍然存在实质性差异。所有提示、输出、评分代码、成本和每个任务的报告均已发布。

View on arXiv Download PDF AI Translation

cs.AI / 56 / 2607.19063

Quality Action Assurance: Multimodal Verification of Examiner Claims in VR OSCEs

质量行动保证：虚拟现实OSCE中考官声明的多模态验证

Rogers, Harry, Shiels, Sally, Tomlinson, Ashley, Thomas, James, Aylward, James, Gauge, Nathan, Higham, Helen, Noble, Alison

Abstract

Objective Structured Clinical Examinations (OSCEs) are the gold standard for assessing clinical competence, yet scoring remains vulnerable to examiner subjectivity, fatigue, and cognitive bias. Standard examiner validation via inter-rater statistics lacks explanatory power regarding the source of errors, as it neither analyzes examiner reasoning nor verifies examiner claims against actual events. Thus, we introduce Quality Action Assurance (QAA), a multimodal framework that verifies examiner claims in Virtual Reality (VR) pediatric OSCEs by comparing actions claimed by examiners against the true sequence of events, constructed from video, VR logs, and actor data. QAA combines a constrained temporal action alignment model, which performs action localization and actor source attribution, with a large language model that extracts examiner claims and checks them against the record. Across a 5-fold cross-validation, QAA achieves 99.2% $\pm$ 0.7% Actor F1 and 93.4% $\pm$ 1.9% W@16 for temporal alignment. Overall, QAA detects examiner errors with 70.0% precision and 76.7% recall, improving factual correctness from 39.2% to 79.2%, enabling fairer OSCE assessment.

Chinese Translation

目标结构化临床考试（OSCE）是评估临床能力的金标准，但评分仍然容易受到考官主观性、疲劳和认知偏见的影响。通过评估者间统计进行的标准考官验证在解释错误来源方面缺乏解释力，因为它既没有分析考官的推理，也没有将考官的声明与实际事件进行验证。因此，我们提出了质量行动保证（Quality Action Assurance, QAA），这是一个多模态框架，通过将考官声称的行动与从视频、虚拟现实日志和演员数据构建的真实事件序列进行比较，来验证虚拟现实（VR）儿童OSCE中的考官声明。QAA结合了一个受限的时间行动对齐模型，该模型执行行动定位和演员来源归属，以及一个大型语言模型，该模型提取考官声明并与记录进行核对。在5折交叉验证中，QAA在时间对齐方面达到了99.2% ± 0.7%的演员F1和93.4% ± 1.9%的W@16。总体而言，QAA以70.0%的精度和76.7%的召回率检测考官错误，将事实正确率从39.2%提高到79.2%，从而实现更公平的OSCE评估。

View on arXiv Download PDF AI Translation

cs.AI / 57 / 2607.19072

On the Effectiveness of Pretraining for Graph Combinatorial Optimization

图组合优化的预训练有效性研究

Aguado, David, Fuertes, Daniel, del-Blanco, Carlos R., Jaureguizar, Fernando

Abstract

This paper introduces a self-supervised pretraining framework for graph combinatorial optimization specifically designed to address the nature of routing problems like the Traveling Salesman Problem. By utilizing graph contrastive learning with geometric augmentations (specifically, rotations and axial reflections) the model is forced to learn invariant structural representations and global relative distance distributions. Results demonstrate that this pretraining strategy outperforms non-pretrained models across various problem scales. Notably, the hybrid strategy (combining rotation and reflection) achieved a 6.57% improvement in tour length for TSP1000, proving that geometric pretraining is an important inductive bias for effectively scaling neural solvers to high-dimensional instances.

Chinese Translation

本文介绍了一种自监督预训练框架，专门针对图组合优化中的路由问题（如旅行商问题）进行设计。通过利用图对比学习与几何增强（具体而言，旋转和轴向反射），模型被迫学习不变的结构表示和全局相对距离分布。结果表明，该预训练策略在各种问题规模下均优于未预训练模型。值得注意的是，混合策略（结合旋转和反射）在 TSP1000 的旅行长度上实现了 6.57% 的改进，证明了几何预训练是有效扩展神经求解器至高维实例的重要归纳偏置。

View on arXiv Download PDF AI Translation

cs.AI / 58 / 2607.19096

Supra Cognitive Modes: A Routed Architecture for Agent Memory

超认知模式：一种用于智能体记忆的路由架构

Tobkin, Joshua, Yang, David

Abstract

Agent-memory workloads mix direct factual lookup, relation-chain and current-state reasoning, and broad synthesis over long histories. We describe Supra Cognitive Modes (SCM), an architecture that maps explicit or automatically selected per-query modes to retrieval and synthesis payloads over one shared ingest substrate. A frozen semantic classifier and runtime gates dispatch queries among fused lexical and dense lookup, graph or iterative multi-hop handling, and stratified long-form synthesis. The substrate combines multi-granularity embeddings, extracted triples, fact-version metadata, and optional asynchronous enrichments. We characterize the deployed configuration on three benchmarks: Long-term Conversational Memory (LoCoMo; n = 1,986), MemoryAgentBench (MAB; n = 3,671), and LongMemEval (n = 500). The reference run records 84.87% on LoCoMo factoid categories and 68.61% on adversarial abstention, 61.49% on MAB across two repetitions, and 86.00% on LongMemEval. A repository-backed reproduction produces similar aggregate scores and supports task- and mode-conditioned failure analysis. Raw baseline outputs, aligned end-to-end timing for LoCoMo and LongMemEval, and complete token ledgers are unavailable; stored rows also omit some final runtime decisions. The results characterize one implemented routed configuration and its diagnostic failure patterns, while source inspection verifies the per-query control interface and shared-substrate design. Causal routing effects, efficiency gains, and statistical significance remain outside the available evidence.

Chinese Translation

智能体记忆工作负载混合了直接的事实查找、关系链和当前状态推理，以及对长历史的广泛综合。我们描述了超认知模式（Supra Cognitive Modes, SCM），这是一种将显式或自动选择的每个查询模式映射到一个共享的输入基础上的检索和综合负载的架构。一个冻结的语义分类器和运行时门控在融合的词汇和密集查找、图形或迭代多跳处理以及分层长形式综合之间调度查询。该基础结合了多粒度嵌入、提取的三元组、事实版本元数据和可选的异步增强。我们在三个基准上表征了部署的配置：长期对话记忆（Long-term Conversational Memory, LoCoMo; n = 1,986）、记忆智能体基准（MemoryAgentBench, MAB; n = 3,671）和长记忆评估（LongMemEval; n = 500）。参考运行在LoCoMo事实类别上记录了84.87%的准确率，在对抗性弃权上为68.61%，在MAB上跨两次重复为61.49%，在LongMemEval上为86.00%。一个基于存储库的重现产生了类似的总分，并支持任务和模式条件下的失败分析。原始基线输出、对LoCoMo和LongMemEval的端到端时间对齐以及完整的令牌账本不可用；存储的行也省略了一些最终的运行时决策。结果表征了一种已实现的路由配置及其诊断失败模式，而源代码检查验证了每个查询控制接口和共享基础设计。因果路由效应、效率提升和统计显著性仍然超出可用证据范围。

View on arXiv Download PDF AI Translation

cs.AI / 59 / 2607.19108

OpenRTAG: A Comprehensive Benchmark for Robust Text-Attributed Graph Learning under Data Quality Degradation

OpenRTAG：数据质量下降下鲁棒文本属性图学习的综合基准

Dai, Yuze, Zhang, Zhihan, Zhao, Yan, Wu, Ruoyu, Li, Xunkai, Chen, Zekai, Dai, Qiangqiang, Qin, Hongchao, Li, Ronghua

Abstract

Text-attributed graphs (TAGs) are an important graph data form that combine relational structure with rich node text. However, real-world TAGs are often imperfect, with quality issues arising from text, structure, and labels, and typically manifesting as sparsity, noise, and imbalance. These dimensions define nine representative degradation scenarios that can substantially affect TAG learning. Although prior studies have explored specific mitigation strategies, existing evidence remains fragmented across degradation types, datasets, tasks, and model families, leaving TAG robustness insufficiently understood. To address this gap, we present OpenRTAG, a robustness benchmark for text-attributed graph learning. OpenRTAG organizes TAG quality issues into a unified 3 * 3 taxonomy and supports standardized evaluation across nine TAG datasets and three downstream tasks. It systematically evaluates scenario validity and model sensitivity, compares traditional GNNs, LLM-GNNs, and a representative GFM, investigates the effectiveness, efficiency, and robustness of scenario-matched baselines, and further examines model behavior under composite degradation scenarios. OpenRTAG provides a standardized testbed for understanding robustness in TAG learning under realistic low-quality settings.

Chinese Translation

文本属性图（TAGs）是一种重要的图数据形式，结合了关系结构和丰富的节点文本。然而，现实世界中的TAG通常存在缺陷，质量问题源于文本、结构和标签，通常表现为稀疏性、噪声和不平衡。这些维度定义了九种典型的降级场景，可能对TAG学习产生重大影响。尽管先前的研究探讨了特定的缓解策略，但现有证据在降级类型、数据集、任务和模型家族之间仍然存在碎片化，导致对TAG鲁棒性的理解不足。为了解决这一问题，我们提出了OpenRTAG，一个用于文本属性图学习的鲁棒性基准。OpenRTAG将TAG质量问题组织成一个统一的3 * 3分类法，并支持在九个TAG数据集和三个下游任务之间的标准化评估。它系统地评估场景有效性和模型敏感性，比较传统的图神经网络（GNNs）、大规模语言模型图神经网络（LLM-GNNs）和一个代表性的图特征模型（GFM），调查场景匹配基准的有效性、效率和鲁棒性，并进一步考察模型在复合降级场景下的行为。OpenRTAG为理解在现实低质量环境下TAG学习的鲁棒性提供了一个标准化的测试平台。

View on arXiv Download PDF AI Translation

cs.AI / 60 / 2607.19117

Comparative Study of Multi-Agent Actor-Critic Algorithms in Parameterized Action Reinforcement Learning

参数化动作强化学习中多智能体演员-评论家算法的比较研究

Bapoo, Ubayd Ali, Nyirenda, Clement N

Abstract

Parameterized action reinforcement learning has shown strong performance in environments requiring both discrete action selection and continuous parameterization. Prior work established the effectiveness of single-agent actor-critic algorithms - Greedy Actor-Critic (GAC), Soft Actor-Critic (SAC), and Truncated Quantile Critics (TQC) - on benchmark parameterized action tasks, but their extension to multi-agent settings remains largely unexplored. This paper presents a comparative study of shared-experience multi-agent extensions of these algorithms: Multi-Agent Greedy Actor-Critic (MAGAC), Multi-Agent Soft Actor-Critic (MASAC), and Multi-Agent Truncated Quantile Critics (MATQC). Rather than following the centralized training, decentralized execution (CTDE) paradigm, the proposed framework uses multiple independent actor-critic agents that share a replay buffer while maintaining separate policy and value networks. We evaluate the algorithms on the Platform-v0 and Goal-v0 benchmarks against their single-agent counterparts, using three-, five-, and ten-agent configurations to assess scalability. Performance is measured by average evaluation return and training time across ten independent runs, with one-way ANOVA and Tukey HSD post-hoc tests used to assess statistical significance. Results show that the multi-agent framework consistently improves Greedy Actor-Critic performance, while MASAC and MATQC show comparatively modest gains over their single-agent versions. Increasing the number of agents beyond five yields limited additional performance while substantially raising computational cost, particularly for MAGAC. These results highlight a trade-off between learning performance and computational efficiency, offering insight into the scalability of shared-experience multi-agent actor-critic methods for parameterized action reinforcement learning.

Chinese Translation

参数化动作强化学习在需要离散动作选择和连续参数化的环境中表现出强大的性能。先前的研究确立了单智能体演员-评论家算法（贪婪演员-评论家（GAC）、软演员-评论家（SAC）和截断分位数评论家（TQC））在基准参数化动作任务中的有效性，但它们在多智能体环境中的扩展仍然未得到充分探索。本文提出了对这些算法的共享经验多智能体扩展的比较研究：多智能体贪婪演员-评论家（MAGAC）、多智能体软演员-评论家（MASAC）和多智能体截断分位数评论家（MATQC）。与集中训练、分散执行（CTDE）范式不同，所提出的框架使用多个独立的演员-评论家智能体，这些智能体共享一个重放缓冲区，同时保持独立的策略和价值网络。我们在Platform-v0和Goal-v0基准上评估这些算法，与它们的单智能体对应算法进行比较，使用三、五和十个智能体配置来评估可扩展性。通过十次独立运行的平均评估回报和训练时间来衡量性能，并使用单向方差分析（ANOVA）和Tukey HSD事后检验来评估统计显著性。结果表明，多智能体框架始终提高了贪婪演员-评论家的性能，而MASAC和MATQC相对于其单智能体版本的提升则相对温和。将智能体数量增加到五个以上所带来的额外性能提升有限，同时显著增加了计算成本，尤其是对于MAGAC。这些结果突显了学习性能与计算效率之间的权衡，为参数化动作强化学习的共享经验多智能体演员-评论家方法的可扩展性提供了见解。

View on arXiv Download PDF AI Translation

cs.AI / 61 / 2607.19253

Sequential Learner Modeling Using Multi-Relational Graph Convolutional Networks

基于多关系图卷积网络的顺序学习者建模

Alatrash, Rawaa, Chatti, Mohamed Amine, Yang, Hong, Wang, Yumeng

Abstract

User modeling is a critical task in a variety of personalized systems. Recognizing their effectiveness in learning from graph-structured data, Graph Neural Networks (GNNs), particularly Graph Convolutional Networks (GCNs), are increasingly employed for user modeling. However, existing approaches typically treat different relation types in a graph as homogeneous, limiting their ability to capture richer semantics and construct more informative user models. While multi-relational GNNs (MR-GNNs) have been adopted for representation learning and recommendation, their application for user modeling remains unexplored. Moreover, existing GNN-based user modeling approaches ignore the user interaction sequence. To address these research gaps, in this work we propose MR-ConceptGCN, a novel fully unsupervised approach focused on concept-based sequential learner modeling using multi-relational GCNs (MR-GCNs). MR-ConceptGCN effecively combines Personal Knowledge Graphs (PKGs), MR-GCNs, and the pre-trained language model SBERT to obtain enhanced relation- and semantic-aware representations of the PKG items. The enriched embeddings of the knowledge concepts that a learner did not understand when interacting with learning materials in CourseMapper are then used to construct a sequential learner model that combines long-term and short-term learner interactions. We report the results of an online user study (n = 31), demonstrating the benefits of MR-ConceptGCN in terms of several important user-centric aspects including accuracy, usefulness, diversity, and satisfaction with an educational recommender system.

Chinese Translation

用户建模是各种个性化系统中的一项关键任务。鉴于图结构数据的学习效果，图神经网络（Graph Neural Networks, GNNs），特别是图卷积网络（Graph Convolutional Networks, GCNs），越来越多地被用于用户建模。然而，现有方法通常将图中的不同关系类型视为同质的，这限制了它们捕捉更丰富语义和构建更具信息性的用户模型的能力。尽管多关系GNN（Multi-Relational GNNs, MR-GNNs）已被用于表示学习和推荐，但其在用户建模中的应用仍未被探索。此外，现有基于GNN的用户建模方法忽略了用户交互序列。为了解决这些研究空白，本文提出了MR-ConceptGCN，这是一种新颖的完全无监督方法，专注于基于概念的顺序学习者建模，使用多关系GCN（MR-GCNs）。MR-ConceptGCN有效地结合了个人知识图谱（Personal Knowledge Graphs, PKGs）、MR-GCNs和预训练语言模型SBERT，以获得增强的关系和语义感知的PKG项目表示。然后，利用学习者在CourseMapper与学习材料交互时未理解的知识概念的丰富嵌入，构建一个结合长期和短期学习者交互的顺序学习者模型。我们报告了一项在线用户研究的结果（n = 31），展示了MR-ConceptGCN在准确性、有用性、多样性和教育推荐系统满意度等多个重要用户中心方面的优势。

View on arXiv Download PDF AI Translation

cs.AI / 62 / 2607.19297

Graph-Based Agentic AI with LangGraph: Workflow Pathways for Long-Running Stateful Business Processes

基于图的自主智能体 AI 与 LangGraph：长时间运行的有状态业务流程的工作流路径

Pearson, Daniel, Shapiro, Sidney, Venegas, Emiliano Sebastian Gonzalez, Al-Khatib, Sanad, Arzola, Aurora Pinzón

Abstract

This paper is a practitioner guide to graph-based workflow pathways for long-running, stateful, multi-step generative AI systems in business processes. Rather than treating LangGraph, a low-level orchestration framework for stateful agents, as a model-quality benchmark target, we present three executable recipes -- SQL analytics with repair loops, agentic retrieval-augmented generation with evidence gating, and human-in-the-loop policy review with interrupt and checkpoint recovery -- to show how typed state, conditional routing, deterministic tools, retries, interrupts, checkpoints, and traces fit together. LangGraph is positioned by workflow-complexity fit, not as a universal default: simpler ReAct-style or plain SDK loops may be better for basic tool use, schema-first tools for structured extraction and validation, and DSPy when prompt or program optimization is the main goal. Each recipe explains when LangGraph is worth the extra structure and which implementation patterns make routes, pauses, and audit trails explicit product behavior rather than hidden prompt logic.

Chinese Translation

本文是针对长时间运行的、有状态的、多步骤生成式 AI 系统在业务流程中的图形工作流路径的实践指南。我们不将 LangGraph（一个用于有状态智能体的低级编排框架）视为模型质量基准目标，而是提出三种可执行的方案——带修复循环的 SQL 分析、增强检索生成的智能体与证据门控，以及人机协作的政策审查与中断和检查点恢复——以展示类型状态、条件路由、确定性工具、重试、中断、检查点和跟踪是如何结合在一起的。LangGraph 的定位是基于工作流复杂性的适配，而非作为通用默认：对于基本工具使用，简单的 ReAct 风格或普通 SDK 循环可能更为合适；对于结构化提取和验证，优先使用模式工具；而当提示或程序优化是主要目标时，DSPy 是更好的选择。每个方案都解释了何时 LangGraph 值得额外的结构，以及哪些实现模式使得路径、暂停和审计轨迹成为显式的产品行为，而不是隐藏的提示逻辑。

View on arXiv Download PDF AI Translation

cs.AI / 63 / 2607.19300

LLM Detection as an Intervention: Downstream Impact under Strategic User Behavior

作为干预的LLM检测：战略用户行为下的下游影响

Jagadeesan, Meena, Hashimoto, Tatsunori, Kleinberg, Jon

Abstract

As LLM adoption becomes more widespread, there is a growing interest in detecting LLM-generated content, for example through LLM detection tools and through heuristics based on language patterns. Detectors operate as an intervention that steers not only the detected attribute itself, but also downstream metrics such as LLM usage and output quality. In this work, we demonstrate how imperfect LLM detectors lead to counterintuitive impacts on these downstream metrics, by distorting how users are incentivized to use LLMs in their workflow. We develop a stylized model which captures how users strategically choose how much to use the LLM and how to post-process content to reduce the detected attribute. Using this model, we show that LLM detection can counterintuitively lead humans to increase their LLM usage. Moreover, even when reducing the detected attribute improves output quality, we find that introducing an LLM detector can lead users to produce lower quality outputs. In contrast, we show that detectors result in a clean "rise-then-fall" pattern for the detected attribute, which we empirically reproduce for word frequencies on arXiv abstracts. Altogether, our work illustrates how LLM detection can distort LLM usage and output quality, uncovering failure modes when LLM detectors operate as an intervention on these downstream metrics.

Chinese Translation

随着LLM（大语言模型）被越来越广泛地采用，检测LLM生成内容的兴趣日益增加，例如通过LLM检测工具和基于语言模式的启发式方法。检测器作为一种干预，不仅影响被检测属性本身，还影响下游指标，如LLM的使用情况和输出质量。在本研究中，我们展示了不完美的LLM检测器如何对这些下游指标产生反直觉的影响，扭曲用户在其工作流程中使用LLM的激励方式。我们开发了一个简化模型，捕捉用户如何战略性地选择使用LLM的程度以及如何后处理内容以减少被检测属性。利用该模型，我们表明LLM检测可能反直觉地导致人类增加LLM的使用。此外，即使减少被检测属性能够提高输出质量，我们发现引入LLM检测器可能导致用户产生更低质量的输出。相反，我们展示了检测器导致被检测属性呈现出清晰的“上升-再下降”模式，我们在arXiv摘要中的词频数据中进行了实证重现。总的来说，我们的研究阐明了LLM检测如何扭曲LLM的使用和输出质量，揭示了当LLM检测器作为这些下游指标的干预时的失效模式。

View on arXiv Download PDF AI Translation

cs.AI / 64 / 2607.19321

ResearchArena: Evaluating Sabotage and Monitoring in Automated AI R&D

ResearchArena：评估自动化人工智能研发中的破坏与监控

Libon, Lena, Rank, Ben, Yeon, Jehyeok, Schmotz, David, Qin, Jeremy, Donnelly, Daniel, Prinzhorn, Derck, Andriushchenko, Maksym

Abstract

As AI agents begin to automate AI R&D, we need ways to assess whether their outputs are safe to deploy, even when the agents themselves may be untrusted. AI control offers one such approach: rather than trusting the agent, it treats it as a potential adversary and uses a monitor to detect covert sabotage before deployment. We evaluate AI control for automated AI R&D with ResearchArena, a framework spanning four long-horizon tasks: safety post-training, capabilities post-training, CUDA-kernel optimization, and inference-server optimization. Because the deliverable in AI R&D is an artifact that will be deployed and run, we pair each main task with two kinds of hidden side task: an embedded side task that sabotages the submitted model, adapter, kernel, or server itself, and an independent side task that takes a covert out-of-scope action in the sandbox while leaving the artifact honest. We evaluate frontier agents at both sabotage and monitoring, across four monitor types that vary along two axes: whether the monitor may execute and probe the produced artifact, and whether it sees the agent's chain-of-thought. We find that sabotage hidden in the training data is the hardest to catch, flagged fewer than half the time. Letting the monitor run experiments on the artifact, rather than only read the trajectory, helps, but it is not enough: monitors still miss embedded sabotage by inspecting only the surface, by noticing the anomaly but explaining it away, or by probing the artifact with the wrong test. We release ResearchArena as a modular framework for evaluating sabotage and control in automated AI R&D.

Chinese Translation

随着人工智能代理开始自动化人工智能研发，我们需要评估其输出是否安全可部署的方法，即使这些代理本身可能不可信。人工智能控制提供了一种这样的途径：与其信任代理，不如将其视为潜在对手，并使用监控器在部署前检测隐蔽的破坏行为。我们通过 ResearchArena 评估自动化人工智能研发中的人工智能控制，该框架涵盖四个长期任务：安全性后训练、能力后训练、CUDA 内核优化和推理服务器优化。由于人工智能研发的交付物是将被部署和运行的工件，我们将每个主要任务与两种隐藏的副任务配对：一种是嵌入式副任务，旨在破坏提交的模型、适配器、内核或服务器本身；另一种是独立副任务，在沙盒中采取隐蔽的超出范围的行动，同时保持工件的诚实性。我们在四种监控器类型上评估前沿代理的破坏与监控，这些监控器在两个维度上有所不同：监控器是否可以执行并探测生成的工件，以及它是否能看到代理的思维链。我们发现，隐藏在训练数据中的破坏是最难捕捉的，标记率不到一半。让监控器对工件进行实验，而不仅仅是读取轨迹是有帮助的，但仍然不够：监控器仍然通过仅检查表面、注意到异常但解释掉、或用错误的测试探测工件而错过嵌入式破坏。我们发布了 ResearchArena 作为一个模块化框架，用于评估自动化人工智能研发中的破坏与控制。

View on arXiv Download PDF AI Translation

cs.AI / 65 / 2607.19327

Associative Emotional Learning in Convolutional Neural Networks

卷积神经网络中的联想情感学习

Leem, Seowung, Keil, Andreas, Ding, Mingzhou, Fang, Ruogu

Abstract

Associative emotional learning enables organisms to adaptively link pleasant or unpleasant outcomes to the presence of predictive stimuli. Whereas computational models such as the Rescorla-Wagner model have shed light on this important function, the limitations of these models are also known, especially when they are applied to neural data. The advent of deep neural networks has opened another avenue for modeling associative emotional learning. In this work we proposed a deep neural network model of visual valence processing, consisting of a visual module that encodes complex natural scenes and a module that recognizes their emotional significance in terms of valence, a key dimension of emotion, and tested a novel Pavlovian learning paradigm on the model. The results showed that with learning, the model reproduced several observations from human associative learning studies, including association formation and generalization, and that the neural representations of the conditioned and the unconditioned stimuli became increasingly aligned both at the single unit and at the neural population level. Comparison between the model and human experimental data provided further validation of our approach. This study thus suggests that deep neural network models, when combined with appropriate learning algorithms, can be used to model behavioral and neural signatures of associative emotion/valence learning.

Chinese Translation

联想情感学习使生物能够自适应地将愉快或不愉快的结果与预测性刺激的存在联系起来。尽管像Rescorla-Wagner模型这样的计算模型对这一重要功能提供了洞见，但这些模型的局限性也为人所知，尤其是在应用于神经数据时。深度神经网络的出现为建模联想情感学习开辟了另一条途径。在本研究中，我们提出了一种视觉效价处理的深度神经网络模型，该模型由一个编码复杂自然场景的视觉模块和一个根据效价（情感的一个关键维度）识别其情感意义的模块组成，并在该模型上测试了一种新颖的巴甫洛夫学习范式。结果表明，随着学习的进行，该模型再现了人类联想学习研究中的几个观察结果，包括联结形成和泛化，并且条件刺激和非条件刺激的神经表征在单个神经元和神经群体层面上变得越来越一致。模型与人类实验数据的比较进一步验证了我们的方法。因此，本研究表明，当深度神经网络模型与适当的学习算法结合时，可以用于建模联想情感/效价学习的行为和神经特征。

View on arXiv Download PDF AI Translation

cs.AI / 66 / 2607.19336

Agents in the Wild: Where Research Meets Deployment

野外中的智能体：研究与部署的交汇

Yang, Grace Hui, Venkit, Pranav N., Sedghamiz, Hooman, Santus, Enrico, Dibia, Victor, Baldini, Ioana

Abstract

Agentic systems large language model (LLM) based architectures capable of reasoning, planning, acting, and coordinating with tools and other agents are rapidly transitioning from research prototypes to production scale deployments across domains such as software engineering, scientific discovery, and finance. While academic work has emphasized benchmarks and algorithmic innovation, deployment raises new challenges around robustness, safety, and reliability. This tutorial brings together researchers and practitioners to explore advances in reasoning and planning, multi agent coordination, and evaluation, highlighting open challenges arising from deployment experience. Through applied case studies in pharmaceutical discovery and financial systems, we analyze common design patterns that make agentic systems successful, and discuss practical mitigation strategies for failure modes, such as verification pipelines, fallback mechanisms, and human in the loop supervision. Attendees will gain a comprehensive view of the field along with concrete design patterns, evaluation checklists, and templates for safe and reliable deployment across industries.

Chinese Translation

基于大型语言模型（LLM）的智能系统架构，具备推理、规划、行动和与工具及其他智能体协调的能力，正迅速从研究原型转向在软件工程、科学发现和金融等领域的生产规模部署。尽管学术研究强调基准测试和算法创新，但部署带来了关于鲁棒性、安全性和可靠性的新挑战。本教程汇集了研究人员和从业者，探讨推理与规划、多智能体协调和评估方面的进展，强调从部署经验中出现的开放性挑战。通过在制药发现和金融系统中的应用案例研究，我们分析了使智能系统成功的常见设计模式，并讨论了应对失败模式的实际缓解策略，如验证管道、后备机制和人机协同监督。与会者将获得该领域的全面视角，以及具体的设计模式、评估清单和安全可靠的跨行业部署模板。

View on arXiv Download PDF AI Translation

cs.AI / 67 / 2607.19338

CodeRescue: Budget-Calibrated Recovery Routing for Coding Agents

CodeRescue：针对编码代理的预算校准恢复路由

He, Qijia, Cheng, Jiayi, Le, Chenqian, Wang, Rui, Liu, Xunmei, Chen, Yixian, Mei, Jie, Wang, Zhihao, Chen, Xupeng, Chen, Yuhuan, Wang, Tao

Abstract

Coding agents increasingly operate in executable environments where a failed attempt produces actionable feedback rather than merely an incorrect answer. Existing cost-aware systems typically treat such failures as cascade decisions: try a cheap model first, then escalate hard cases to a stronger and more expensive model. In coding, however, execution feedback can also make further cheap-model recovery worthwhile, raising a budgeted deployment question: when should an agent spend more cheap compute, and when should it escalate? We formulate this post-failure decision as recovery routing over heterogeneous actions and train a supervised router from execution rollouts. To make the same router usable under changing budgets, we add a Conformal Risk Control (CRC) layer that selects a deployment-time cost penalty without retraining and provides marginal expected-cost control under exchangeability. Across held-out failures from five coding benchmarks, cheap recovery and escalation exhibit complementary success patterns. The calibrated frontier improves over fixed actions, prompt-only routers, and a binary cascade baseline; in the main GPT-5.4-nano/GPT-5.4 setting, one CRC-calibrated frontier point exceeds always-escalate solve rate while using 35% of its mean recovery cost. Code is available at https://github.com/Qijia-He/agent-budget-control.

Chinese Translation

编码代理越来越多地在可执行环境中操作，在这些环境中，失败的尝试会产生可操作的反馈，而不仅仅是错误的答案。现有的成本感知系统通常将此类失败视为级联决策：首先尝试廉价模型，然后将困难案例升级到更强大且更昂贵的模型。然而，在编码中，执行反馈也可以使进一步的廉价模型恢复变得有价值，这引发了一个预算部署问题：代理何时应花费更多的廉价计算，何时应进行升级？我们将这一失败后的决策形式化为异构动作上的恢复路由，并从执行回放中训练一个监督路由器。为了使同一路由器在变化的预算下可用，我们添加了一个符合风险控制（Conformal Risk Control, CRC）层，该层在不重新训练的情况下选择部署时的成本惩罚，并在可交换性下提供边际期望成本控制。在五个编码基准的保留失败中，廉价恢复和升级表现出互补的成功模式。校准的边界在固定动作、仅提示路由器和二元级联基线之上有所改善；在主要的GPT-5.4-nano/GPT-5.4设置中，一个CRC校准的边界点在使用其平均恢复成本的35%时，超越了始终升级的解决率。代码可在 https://github.com/Qijia-He/agent-budget-control 获取。

View on arXiv Download PDF AI Translation

计算语言学 (Computation and Language)

cs.CL / 1 / 2607.18321

Decoding EEG Signals to Explore Next-Word Predictability in the Human Brain

解码脑电图信号以探索人脑中的下一个词预测性

Quach, Boi Mai, Nguyen, Binh T., Gurrin, Cathal, Healy, Graham

Abstract

Humans invented reading and have passed down this complex skill across generations through language. This study provides empirical evidence of the neural mechanisms underlying bottom-up (related to high-order linguistic structure) and top-down (related to next-word predictability) processes, which interact to guide comprehension during reading. While previous studies have focused on either the N400 effects of predictability or lexical categories, research on how predictability influences N400 responses across different lexical categories is limited, mainly due to constraints in publicly available datasets. Here, we examine how predictability influences brain responses, recorded at millisecond resolution using electroencephalography (EEG), with a focus on the N400 time window (300-500 ms post-stimulus) across different lexical and grammatical categories. Our results indicate that significant differences in N400 responses between high and low cloze probability levels were more pronounced for content words than function words. Among the two primary content categories, verbs exhibited greater N400 differences than nouns, while nouns carried more distinct information about their predictability than verbs. Moreover, we demonstrate that the decoding technique is more effective than the event-related potential (ERP) traditional analysis in capturing more detailed and distinct representations of cognitive processes over time.

Chinese Translation

人类发明了阅读，并通过语言将这一复杂技能代代相传。本研究提供了关于底层（与高阶语言结构相关）和顶层（与下一个词预测性相关）过程的神经机制的实证证据，这两者相互作用以指导阅读理解。尽管之前的研究主要集中在预测性的N400效应或词汇类别上，但关于预测性如何影响不同词汇类别的N400反应的研究仍然有限，主要是由于公开可用数据集的限制。在此，我们考察了预测性如何影响以毫秒分辨率记录的脑反应，使用脑电图（EEG），重点关注不同词汇和语法类别的N400时间窗口（刺激后300-500毫秒）。我们的结果表明，高低填空概率水平之间的N400反应显著差异在内容词中比功能词更为明显。在两个主要内容类别中，动词的N400差异大于名词，而名词在其预测性方面传递的信息比动词更为独特。此外，我们证明了解码技术在捕捉认知过程随时间变化的更详细和独特的表征方面比事件相关电位（ERP）传统分析更为有效。

View on arXiv Download PDF AI Translation

cs.CL / 2 / 2607.18358

A Classifier That Teaches Itself: Self-Improving, Frozen-gate Training (SIFT) for Dynamic Document Classification

自我学习的分类器：动态文档分类的自我改进冻结门训练（SIFT）

Raduta, Bogdan, Velicu, Horia, Preda, Alexandru, Chiricescu, Serban

Abstract

Document classification is a solved problem in the laboratory and an unsolved one in the enterprise. The blocker is rarely model architecture; it is the labeling project that must precede a model and the institutional fear of letting a model retrain itself once one exists. We present SIFT (Self-Improving, Frozen-gate Training), a dynamic classifier service, which attacks both. SIFT serves classification from a deliberately cheap, CPU-bound pipeline, a SPLADE sparse encoder feeding a LightGBM head, and escalates only the low-confidence minority of pages to an LLM judge. The judge's verdicts are written back into a labeled corpus, so the expensive model continuously teaches the cheap one: the escalation rate falls, the corpus grows from production traffic rather than from an up-front annotation effort, and accuracy compounds with use. Onboarding a new document family requires only a declarative bundle, label space, anchor phrases, and a judge glossary, not a labeling project. The harder problem is safety: an autonomously retraining classifier can silently regress. SIFT resolves this with a two-part promote gate, a critical-label F1 regression check plus a frozen golden regression set the model is never trained on, either of which vetoes promotion. This turns "retrain monthly without a human" from reckless into routine. We describe the architecture, the self-feeding corpus loop, the frozen-gate promotion mechanism, and an illustrative multi-domain deployment, and we discuss the economics of a classifier whose marginal labeling cost trends toward zero.

Chinese Translation

文档分类在实验室中是一个已解决的问题，但在企业中仍未解决。阻碍因素很少是模型架构，而是必须在模型之前进行的标注项目以及机构对让模型在存在后自我再训练的恐惧。我们提出了SIFT（自我改进冻结门训练），这是一种动态分类服务，旨在解决这两个问题。SIFT通过一个故意廉价的、以CPU为主的管道提供分类服务，使用SPLADE稀疏编码器为LightGBM头提供输入，并仅将低置信度的少数页面升级到LLM评判者。评判者的判决被写回标注语料库，因此昂贵的模型不断教导廉价模型：升级率下降，语料库从生产流量中增长，而不是来自前期的标注工作，准确性随着使用而提高。引入新的文档家族只需一个声明性包、标签空间、锚短语和评判者词汇，而不需要标注项目。更棘手的问题是安全性：一个自主再训练的分类器可能会悄然退化。SIFT通过一个两部分的提升门解决了这个问题，包括一个关键标签的F1回归检查和一个模型从未训练过的冻结黄金回归集，任一者都可以否决提升。这使得“每月无人工再训练”从鲁莽变为常规。我们描述了架构、自我供给的语料库循环、冻结门提升机制以及一个说明性的多领域部署，并讨论了一个边际标注成本趋向于零的分类器的经济性。

View on arXiv Download PDF AI Translation

cs.CL / 3 / 2607.18413

Convolution for Large Language Models

大语言模型中的卷积

Tian, Yuchuan, Shu, Yingte, He, Wei, Zhang, Shuo, Zhao, Tianchen, Xu, Chao, Chen, Xinghao, Wang, Yunhe, Chen, Hanting, Wang, Yu

Abstract

Large language models (LLMs) largely rely on Transformers, where self-attention provides global token interaction but does not explicitly encode the locality of natural language. We study whether lightweight depthwise convolutions can supply this local inductive bias without materially increasing model size. Our macro-level ablation compares convolution at 17 locations in a Qwen3 Transformer block and finds the best results when convolution is applied to the projected queries, keys, and values before attention. A subsequent micro-level study favors a residual depthwise convolution with kernel size $k=3$, without additional normalization or activation. Across Qwen3 models and several pre-training data budgets, this design improves the average accuracy on seven downstream benchmarks while adding less than $0.01\%$ parameters. A representation-level case study further suggests that the convolution makes repeated token IDs more sensitive to their immediate context. These results support depthwise convolution as a lightweight complement to self-attention for modeling short-range token interactions.

Chinese Translation

大语言模型（LLMs）主要依赖于变换器（Transformers），其中自注意力机制提供了全局的标记交互，但并未明确编码自然语言的局部性。我们研究轻量级深度卷积是否能够在不显著增加模型规模的情况下提供这种局部归纳偏差。我们的宏观层面消融实验比较了在 Qwen3 变换器块的 17 个位置应用卷积的效果，发现当卷积应用于注意力之前的投影查询、键和值时，效果最佳。随后的微观层面研究倾向于使用核大小为 $k=3$ 的残差深度卷积，且不添加额外的归一化或激活。在 Qwen3 模型和多个预训练数据预算下，该设计在七个下游基准测试中提高了平均准确率，同时增加的参数不足 $0.01 ext{%}$。一个表示层面的案例研究进一步表明，卷积使重复的标记 ID 对其直接上下文更加敏感。这些结果支持深度卷积作为自注意力的轻量级补充，以建模短距离标记交互。

View on arXiv Download PDF AI Translation

cs.CL / 4 / 2607.18432

Building a European Multilingual Evaluation Dataset: The MMLU Localisation Project within the EMT Network

构建欧洲多语言评估数据集：EMT网络中的MMLU本地化项目

Sánchez-Gijón, Pilar, Valdez, Susana, Del Barrio, Sofía Calvo, Bellemont, Florence, Kokkinidou, Anna, Brasoveanu, Mihai Cristian

Abstract

This paper reports on a collaboration between the Directorate-General for Translation (DGT) and the European Master's in Translation (EMT) to localise the MMLU dataset into 11 European languages. Beyond creating a more inclusive benchmark for LLM evaluation, the project offers master's students authentic, project-based professional training in translation, revision, project management, and multilingual coordination, while highlighting key methodological, administrative, and workflow challenges.

Chinese Translation

本文报告了翻译总局（DGT）与欧洲翻译硕士（EMT）之间的合作，旨在将MMLU数据集本地化为11种欧洲语言。该项目不仅创建了一个更具包容性的LLM评估基准，还为硕士生提供了真实的、基于项目的翻译、修订、项目管理和多语言协调的专业培训，同时突出了关键的方法论、行政和工作流程挑战。

View on arXiv Download PDF AI Translation

cs.CL / 5 / 2607.18438

Relay-Bench: Evaluating LLMs on Multi-Domain Reasoning Chains

Relay-Bench：评估大语言模型在多领域推理链上的表现

Swayne, Liam

Abstract

Introducing Relay-Bench, an unsaturated, holistic, text-only benchmark that measures LLMs' ability to complete an assortment of tasks from distinct domains in a single prompt. The leading model, GPT-5.5 (xHigh), scores 43.3%. The test set entirely consists of composite problems: groups of single-domain subproblems that are strung together into challenges that require reasoning across multiple domains in combination. Many of these problems then have layers of complexity added through prompt encoding and deliberate context bloat. Domains tested include visual reasoning, coding, math, information extraction (with a focus on web search), problem-solving, general knowledge, and data analysis. No restrictions are imposed outside of the model harness, and models are explicitly encouraged to leverage code-execution, web searches, and all available tools. All problems are composed of two to thirteen subproblems and do not require multi-modal input or output.

Chinese Translation

本文介绍了 Relay-Bench，这是一个未饱和的整体性文本基准，旨在测量大语言模型（LLMs）在单个提示中完成来自不同领域的各种任务的能力。领先模型 GPT-5.5 (xHigh) 的得分为 43.3%。测试集完全由复合问题组成：将单领域子问题组合在一起形成的挑战，这些挑战需要跨多个领域进行推理。许多问题通过提示编码和故意的上下文膨胀增加了复杂性。测试的领域包括视觉推理、编码、数学、信息提取（重点关注网络搜索）、问题解决、一般知识和数据分析。除了模型框架外，没有施加任何限制，并且明确鼓励模型利用代码执行、网络搜索和所有可用工具。所有问题由两个到十三个子问题组成，并且不需要多模态输入或输出。

View on arXiv Download PDF AI Translation

cs.CL / 6 / 2607.18443

Computational models of pragmatic reasoning with flexible generation of meaning and expression alternatives

灵活生成意义和表达替代方案的务实推理计算模型

Tsvilodub, Polina, Carcassi, Fausto, Franke, Michael

Abstract

Pragmatic language use requires reasoning about alternatives: the alternative expressions a speaker might have chosen, or the alternative interpretations a listener might entertain. Formal and computational models of pragmatics must therefore specify the sets of alternatives that interlocutors reason over, which is often done through manual specification. Here we propose a framework, ScAffolded Generative models for Explanation (SAGE), that combines the explanatory transparency of cognitive models with the generative flexibility of language models (LMs). SAGE decomposes a pragmatic process into three kinds of modules: proposers, which use LMs to generate an open-ended space of candidate alternatives; evaluators, which assess those alternatives (e.g., their semantics, complexity, or typicality); and selectors, which implement the rule-based computational steps of a cognitively motivated task analysis. We assess SAGE in three case studies spanning pragmatic generation and interpretation-referential expression generation, manner (M-)implicatures, and Gricean conversational implicatures. SAGE models are evaluated critically using established methods from computational cognitive modeling, including ablations, baseline comparisons, and quantitative fit to human data. Across studies, SAGE models achieved high accuracy and often outperformed baselines, but component-level analyses reveal an asymmetry: LM proposers reliably generated alternatives well-suited to pragmatic modeling, whereas LM evaluators are better at providing intuitive judgements rather than judgements of theoretical or formal measures. We discuss the promise and the limitations of neuro-symbolic models as candidate explanatory accounts of human pragmatic language use.

Chinese Translation

务实语言使用需要对替代方案进行推理：说话者可能选择的替代表达，或听者可能考虑的替代解释。因此，务实学的形式和计算模型必须明确对话者推理的替代方案集合，这通常通过手动指定来完成。在此，我们提出了一个框架——解释的支架生成模型（ScAffolded Generative models for Explanation, SAGE），它结合了认知模型的解释透明性和语言模型（Language Models, LMs）的生成灵活性。SAGE将务实过程分解为三种模块：提议者，使用LM生成开放的候选替代方案空间；评估者，评估这些替代方案（例如，它们的语义、复杂性或典型性）；选择者，实施基于规则的计算步骤，进行认知驱动的任务分析。我们在三个案例研究中评估SAGE，涵盖务实生成和解释-指称表达生成、方式（M-）含义，以及格赖斯会话含义。SAGE模型通过计算认知建模的既定方法进行严格评估，包括消融实验、基线比较和与人类数据的定量拟合。在各项研究中，SAGE模型达到了高准确率，且通常优于基线，但组件级分析揭示了不对称性：LM提议者可靠地产生了适合务实建模的替代方案，而LM评估者更擅长提供直观判断，而非理论或形式测量的判断。我们讨论了神经-符号模型作为人类务实语言使用的候选解释模型的前景和局限性。

View on arXiv Download PDF AI Translation

cs.CL / 7 / 2607.18446

Using Fine-Tuned LLMs to Identify Indicators of Vulnerability in UK Police Incident Logs

使用微调的LLM识别英国警察事件日志中的脆弱性指标

Relins, Sam, Birks, Daniel

Abstract

Purpose: Understanding how much of routine policing involves vulnerable people could inform resourcing, training, and multi-agency response, yet administrative data provide limited insight. We explore whether an LLM-based classification pipeline, developed on open-source US police data, can be adapted to estimate the prevalence of four vulnerability indicators - mental ill health, substance misuse, alcohol dependence, and homelessness - in UK police incident narratives, and when outputs can be treated as defensible measurements. Methods: We analyse nearly 3,000 de-identified incident logs from a UK police force, using a multi-stage pipeline combining repeated model inference, label aggregation, structured human review, and statistical correction. The pipeline runs on a locally hosted open-weight LLM, reflecting the secure environments police must work in. Results: LLMs can produce meaningful, if imperfect, prevalence estimates at scale. Mental ill health indicators are present in approximately one in five incidents, with lower prevalence for other indicators. However, naive LLM deployment is unreliable: single-pass classifications are unstable, and aggregated outputs systematically over-assign indicators relative to human judgement. Correcting these biases required substantial human input and statistical adjustment, leaving considerable uncertainty. Conclusions: While LLMs can extract information from unstructured police data, their outputs cannot be treated as valid measurements without careful methodological support. At the population level, defensible estimates are achievable but resource-intensive; at the individual level, errors remain frequent and unpredictable, limiting suitability for operational decisions. This study highlights both the potential and the constraints of LLM-based measurement in applied settings.

Chinese Translation

目的：了解常规警务中涉及脆弱人群的程度可以为资源配置、培训和多机构响应提供信息，但行政数据提供的见解有限。我们探讨基于LLM（大型语言模型）的分类流程，是否可以适应于估计英国警察事件叙述中四种脆弱性指标的普遍性——心理健康问题、物质滥用、酒精依赖和无家可归，并在何种情况下输出可以被视为可辩护的测量。方法：我们分析了来自英国某警察局的近3000份去标识化事件日志，使用一个多阶段流程，结合重复模型推断、标签聚合、结构化人工审查和统计校正。该流程在本地托管的开放权重LLM上运行，反映了警察必须工作的安全环境。结果：LLM可以在规模上产生有意义的（尽管不完美的）普遍性估计。心理健康指标在大约五分之一的事件中出现，而其他指标的普遍性较低。然而，简单的LLM部署是不可靠的：单次分类结果不稳定，聚合输出相对于人工判断系统性地过度分配指标。纠正这些偏差需要大量的人力投入和统计调整，留下了相当大的不确定性。结论：尽管LLM可以从非结构化的警务数据中提取信息，但其输出在没有仔细的方法论支持下不能被视为有效的测量。在总体层面上，可辩护的估计是可实现的，但资源密集；在个体层面上，错误仍然频繁且不可预测，限制了其在操作决策中的适用性。本研究突显了基于LLM的测量在应用场景中的潜力与限制。

View on arXiv Download PDF AI Translation

cs.CL / 8 / 2607.18448

PathReportEval: A Systematic Benchmark for Pathology Report Generation

PathReportEval：病理报告生成的系统基准测试

Singh, Suryakant, Majumder, Sejuti, Knudsen, Beatrice, Saltz, Joel, Prasanna, Prateek

Abstract

Pathology report generation from whole-slide images (WSIs) is a rapidly growing multimodal learning problem, yet progress is difficult to measure because existing studies use heterogeneous datasets, model settings, visual encoders, and evaluation protocols. Moreover, commonly used natural language generation metrics, including BLEU, ROUGE, and METEOR, primarily reward lexical similarity and often fail to detect clinically consequential errors such as omitted diagnoses, hallucinated findings, or discordant tumor attributes. We present a standardized benchmark and evaluation framework for pathology report generation. The benchmark evaluates four representative methods across three datasets (TCGA, HistAI, and REG 2025) using three pathology foundation encoders (CONCHv1.5, UNI2-h, and H-Optimus-1). Our framework standardizes preprocessing, feature extraction, training, decoding, and evaluation, enabling fair comparison across models while providing a modular platform for integrating new methods, datasets, and encoders. A central contribution is the Clinical Report Quality Score (CRQS), a clinically grounded metric for evaluating factual correctness. CRQS maps reference and generated reports into structured clinical attributes and measures four complementary dimensions: clinical fact coverage, key information recall, hallucination rate, and clinical discordance, producing both an overall score and interpretable sub-scores. Experiments demonstrate that conventional language-generation metrics are weakly aligned with clinical correctness and frequently overestimate report quality. In contrast, CRQS reveals clinically meaningful differences between models and encoders that lexical metrics fail to capture. Together, the benchmark, public plug-and-play framework, and CRQS establish a reproducible foundation for rigorous evaluation of pathology report generation.

Chinese Translation

从全切片图像（WSIs）生成病理报告是一个快速发展的多模态学习问题，但由于现有研究使用异构数据集、模型设置、视觉编码器和评估协议，进展难以衡量。此外，常用的自然语言生成指标，包括BLEU、ROUGE和METEOR，主要奖励词汇相似性，往往无法检测出临床上重要的错误，例如遗漏的诊断、虚构的发现或不一致的肿瘤属性。我们提出了一个标准化的病理报告生成基准和评估框架。该基准在三个数据集（TCGA、HistAI和REG 2025）上评估四种代表性方法，使用三种病理基础编码器（CONCHv1.5、UNI2-h和H-Optimus-1）。我们的框架标准化了预处理、特征提取、训练、解码和评估，能够在模型之间进行公平比较，同时提供一个模块化平台以整合新方法、数据集和编码器。一个核心贡献是临床报告质量评分（CRQS），这是一个基于临床的指标，用于评估事实的正确性。CRQS将参考报告和生成报告映射到结构化的临床属性，并测量四个互补维度：临床事实覆盖率、关键信息召回率、虚构率和临床不一致性，从而生成一个总体评分和可解释的子评分。实验表明，传统的语言生成指标与临床正确性之间的对齐较弱，且常常高估报告质量。相比之下，CRQS揭示了模型和编码器之间临床上有意义的差异，而词汇指标未能捕捉到这些差异。综上所述，该基准、公共即插即用框架和CRQS为病理报告生成的严格评估建立了可重复的基础。

View on arXiv Download PDF AI Translation

cs.CL / 9 / 2607.18476

Structured Output Collapses Answer Diversity Across 44 Language Models

结构化输出压缩了44种语言模型的答案多样性

Parikh, Tapan

Abstract

When a language model must choose one answer from a large space of equally valid options, a format clause -- "Reply with JSON only" -- changes which answer it chooses. We re-run the One-Word Census (arXiv:2607.12796): 31 wide-answer-space category prompts asked of 44 models, now with the reply requested in JSON -- no schema enforcement, no constrained decoding, only the request. Convergence deepens sharply: on the unconstrained "Pick a word" prompt the modal answer rises from 41% to 64% of the pool and distinct answers fall from 52 to 36; mean answer-choice surprisal drops from 1.80 to 1.58 bits. The tax is progressive: six of 44 models move individually (BH-FDR q=.10), all toward the mode, led by the most distinctive models, while the conformist floor is immobile. It is a sharpener, not a re-indexer -- the plain-chat modal answer survives in 28 of 31 categories. Defaults are register-indexed: a within-run re-sample (n=20) finds JSON shifts 53% of a model's stable chat defaults, mostly back to the crowd, and installs defaults absent from chat (Claude Fable 5 answers "cerulean" for colour 0% of the time in chat, 100% in JSON). Full-battery controls reveal a register gradient: compression is significant and specific to the answer-delivery formats models are trained to speak (JSON -0.22 bits, p=.0002; XML -0.19, p=.002), absent for YAML and CSV, and reversed for an arbitrary bracket wrapper (+0.13, p=.009) -- weighing the mechanism toward tool-use post-training. Enforcing the schema at the decoder (response_format) compresses no further than the request (-0.03 bits): the collapse lives in the model's response to the register, not the decoder. Structured output is how software consumes language models, and that surface is served by a measurably more homogeneous model than the chat surface on which models are evaluated, compared, and chosen.

Chinese Translation

当语言模型必须从大量同样有效的选项中选择一个答案时，格式条款——“仅以JSON格式回复”——改变了它选择的答案。我们重新进行了单词普查（One-Word Census，arXiv:2607.12796）：对44个模型提出31个宽答案空间类别的提示，现在要求以JSON格式回复——没有模式强制，没有约束解码，仅仅是请求。收敛性显著加深：在不受限制的“选择一个词”提示下，主要答案从41%上升到64%，而不同答案从52个降至36个；平均答案选择惊讶度从1.80降至1.58比特。这个代价是渐进的：44个模型中有六个模型单独移动（BH-FDR q=.10），都朝向模式，最具特色的模型引领其后，而顺从的底层则保持不变。这是一个锐化器，而不是重新索引器——在31个类别中，普通聊天的主要答案在28个类别中存活。默认值是注册索引的：一次运行内的重新抽样（n=20）发现JSON改变了模型稳定聊天默认值的53%，大多数回归到大众，并安装了在聊天中缺失的默认值（Claude Fable在聊天中0%时间回答颜色为“天蓝色”，在JSON中100%）。全面的控制实验揭示了注册梯度：压缩是显著的，并且特定于模型被训练为使用的答案交付格式（JSON -0.22比特，p=.0002；XML -0.19，p=.002），在YAML和CSV中不存在，并且在任意括号包装器中反转（+0.13，p=.009）——将机制倾向于训练后工具使用。在解码器处强制模式（response_format）不会进一步压缩超过请求（-0.03比特）：这种崩溃存在于模型对注册的响应中，而不是解码器中。结构化输出是软件如何消费语言模型的方式，而这一表面由一个可测量的更同质的模型提供服务，而不是模型被评估、比较和选择的聊天表面。

View on arXiv Download PDF AI Translation

cs.CL / 10 / 2607.18481

Search-on-Graph-R1: Training Large Language Models to Search Knowledge Graphs with Reinforcement Learning

Search-on-Graph-R1：训练大型语言模型通过强化学习搜索知识图谱

Sun, Jia Ao, Yu, Hao, Mo, Fengran, Su, Zhan, Hui, Yuchen, Liu, Bang, Nie, Jian-Yun

Abstract

Knowledge graph question answering (KGQA) requires navigating from topic entities to an answer several relations away. Recent methods prompt a frontier LLM to explore the graph through a retrieval tool, but their reliance on frontier-scale inference makes them costly to deploy. We present Search-on-Graph-R1 (\sogrone{}), which internalizes this navigation into a compact 8B model through supervised fine-tuning (SFT) followed by reinforcement learning (RL). Our central idea is to scaffold a frontier teacher with each question's gold SPARQL query, so the teacher traverses a known answer-bearing path with a live \texttt{Search} tool rather than having to discover the path itself. Since every call executes against a live Freebase server, the resulting trajectories are grounded in the knowledge graph by construction. On WebQSP, CWQ, and GrailQA, \sogrone{} at 8B surpasses every frozen frontier-LLM system in our comparison and posts the strongest results on CWQ of any system we compare against. It does so using no auxiliary module at inference and no LLM judge during training. Isolating each training stage shows that SFT and RL contribute complementary gains, our approach transfers across model families, and RL learns to reach answers in fewer \texttt{Search} calls than its SFT initialization.

Chinese Translation

知识图谱问答（KGQA）需要从主题实体导航到相隔多个关系的答案。近期的方法通过检索工具提示前沿大型语言模型（LLM）探索图谱，但它们对前沿规模推理的依赖使得部署成本高昂。我们提出了Search-on-Graph-R1（ ext{sogrone}），通过监督微调（SFT）后接强化学习（RL），将这种导航内化为一个紧凑的8B模型。我们的核心思想是为每个问题的黄金SPARQL查询搭建一个前沿教师，使得教师利用实时的 exttt{Search}工具沿着已知的答案路径进行遍历，而不是自己发现路径。由于每次调用都在实时的Freebase服务器上执行，所产生的轨迹在构建上就扎根于知识图谱。在WebQSP、CWQ和GrailQA上， ext{sogrone}在8B模型中超越了我们比较中的每个冻结前沿LLM系统，并在CWQ上取得了我们比较的任何系统中最强的结果。它在推理时不使用任何辅助模块，在训练期间也没有LLM评判。分离每个训练阶段显示，SFT和RL提供了互补的增益，我们的方法可以跨模型家族迁移，并且RL学习在比其SFT初始化更少的 exttt{Search}调用中达到答案。

View on arXiv Download PDF AI Translation

cs.CL / 11 / 2607.18532

Reasoning Fine-Tuning Induces Persistent Latent Policy States

推理微调引发持久的潜在策略状态

Harrasse, Abir, Lan, Michael, Batra, Hunar, Chaleshtori, Fateme Hashemi, Bandi, Chaithanya

Abstract

Reasoning-specialized language models show large performance gains over base models, yet the internal changes responsible for improved multi-step reasoning remain poorly understood. It is unclear whether reasoning fine-tuning improves local token-level competence or globally reorganizes how models structure inference over time. We address this question by modeling Chain-of-Thought reasoning as a switching dynamical system (SDS), in which internal representations evolve under discrete latent policy states. Our framework combines time-aware contrastive representation learning with discrete regime discovery to recover latent policies from activation trajectories. Across four benchmarks and model scales from 1.5B to 32B parameters, reasoning-fine-tuned models exhibit richer latent-policy organization than their base counterparts, characterized by more differentiated transition structure and model-dependent changes in state utilization, persistence, and mixing. The recovered regimes exhibit functional specialization aligned with distinct reasoning stages, and extensive controls confirm that their structure is not explained by correctness, representation learning, or modeling priors, but depends on the coherent temporal organization of reasoning trajectories. Causal interventions further show that the regimes are functionally meaningful: state-swap ablations reduce one-step predictive fit, while transplanting reasoning dynamics into base models improves performance on challenging reasoning problems. Finally, SDS-guided pruning of failure-prone reasoning prefixes outperforms self-consistency in 11 of 12 model-dataset settings, with gains of up to 12.5 percentage points. Together, our results suggest that reasoning fine-tuning globally reorganizes latent dynamics, offering a new lens for mechanistic analysis and process-level control of reasoning models.

Chinese Translation

专门针对推理的语言模型在性能上相较于基础模型有显著提升，但导致多步推理改进的内部变化仍然不甚明了。目前尚不清楚推理微调是改善局部的标记级能力，还是在全球范围内重新组织模型随时间推理的结构。我们通过将链式思维推理建模为一个切换动态系统（SDS）来解决这个问题，在该系统中，内部表征在离散的潜在策略状态下演变。我们的框架结合了时间感知的对比表征学习与离散状态发现，以从激活轨迹中恢复潜在策略。在四个基准测试和从15亿到320亿参数的模型规模中，经过推理微调的模型展现出比基础模型更丰富的潜在策略组织，其特征表现为更具差异化的转移结构以及模型依赖的状态利用、持久性和混合的变化。恢复的状态展现出与不同推理阶段一致的功能专业化，广泛的控制实验确认其结构并非由正确性、表征学习或建模先验所解释，而是依赖于推理轨迹的连贯时间组织。因果干预进一步表明这些状态在功能上是有意义的：状态交换消融减少了一步预测的拟合度，而将推理动态移植到基础模型中则提高了在具有挑战性的推理问题上的表现。最后，基于SDS指导的失败倾向推理前缀剪枝在12个模型-数据集设置中有11个超越了自一致性，增益高达12.5个百分点。综合来看，我们的结果表明推理微调在全球范围内重新组织了潜在动态，为推理模型的机制分析和过程级控制提供了新的视角。

View on arXiv Download PDF AI Translation

cs.CL / 12 / 2607.18566

The Story Shapes the Agent: Narrative Priors in LLM Behavior

故事塑造代理人：大语言模型行为中的叙事先验

Wang, Yixuan, Lester, James, Srivastava, Shashank

Abstract

Persona prompting is widely used to steer LLM agent behavior, yet the narrative framing of a task can matter more than the assigned persona. We isolate this effect through structural isomorphism, constructing three text-based investigation games that share the same action space, stage progression, and resource constraints while varying only task narrative: disease investigation, IT troubleshooting, and murder mystery. Across 1,890 sessions spanning 3 models and 10 personas, we identify narrative priors: systematic action tendencies activated by a task's story framing, independent of its decision structure. Narrative priors explain 5-31x more behavioral variance than persona, are consistent across model architectures, and in two of three domains are negatively associated with task success. Persona effects that do transfer across narratives arise from behavioral anchors, persona descriptions whose language maps directly onto shared actions. Causal interventions confirm this: removing anchor words from a high-transfer persona reduces cross-narrative consistency by 95%. Our framework also generalizes to a held-out fourth narrative and yields a persona-selection method that improves cross-narrative transfer. These results suggest that LLM behavior that survives narrative changes should be grounded in concrete actions rather than abstract descriptions.

Chinese Translation

个性提示被广泛用于引导大语言模型（LLM）代理人的行为，但任务的叙事框架可能比分配的个性更为重要。我们通过结构同构来隔离这一效应，构建了三种基于文本的调查游戏，这些游戏共享相同的行动空间、阶段进展和资源限制，仅在任务叙事上有所不同：疾病调查、IT故障排除和谋杀之谜。在跨越3个模型和10个个性、共计1,890个会话中，我们识别出叙事先验：由任务的故事框架激活的系统性行动倾向，与其决策结构无关。叙事先验解释的行为方差是个性的5-31倍，在模型架构之间保持一致，并且在三个领域中的两个领域与任务成功呈负相关。跨叙事转移的个性效应源于行为锚点，即其语言直接映射到共享行动的个性描述。因果干预证实了这一点：从高转移个性中移除锚定词汇使跨叙事一致性降低了95%。我们的框架还推广到一个保留的第四个叙事，并提出了一种个性选择方法，改善了跨叙事转移。这些结果表明，能够在叙事变化中存活的大语言模型行为应当基于具体行动而非抽象描述。

View on arXiv Download PDF AI Translation

cs.CL / 13 / 2607.18570

For What Reason? Interpreting Models' Encoding of Causation and Antithesis

为何？解读模型对因果关系和对立关系的编码

Bhattacharyya, Abhidip, Wein, Shira

Abstract

Discourse relations provide document structure, critical to language understanding and enabling language model performance and ethicality. In this work, we investigate how instruction-tuned Transformer models (LLaMA and Mistral) encode discourse relations in English, with a particular focus on the contrasting relations of causation and antithesis. Framing the task as a next-token prediction task and applying a suite of interpretability techniques to test model internals, our findings show that certain early layers make predictive decisions at mid-sequence tokens, while some mid-level layers finalize their decisions closer to the last token. Most of the remaining layers primarily propagate earlier decisions rather than actively influencing them. Additionally, we observe that some layers exhibit a preference for one answer over alternatives, suggesting asymmetric representation of discourse-based reasoning.\footnote{Our code is available at https://github.com/abhidipbhattacharyya/causation_vs_antithesis}

Chinese Translation

话语关系为文档结构提供了基础，这对于语言理解至关重要，并且能够提升语言模型的性能和伦理性。在本研究中，我们探讨了经过指令调优的Transformer模型（LLaMA和Mistral）如何在英语中编码话语关系，特别关注因果关系和对立关系的对比。我们将任务框定为下一个标记预测任务，并应用一系列可解释性技术来测试模型内部机制。我们的研究结果表明，某些早期层在中间序列标记处做出预测决策，而一些中层则在接近最后一个标记时最终确定其决策。其余大部分层主要传播早期的决策，而不是主动影响它们。此外，我们观察到某些层对一个答案表现出偏好而非其他选项，这表明话语基础推理的非对称表示。

View on arXiv Download PDF AI Translation

cs.CL / 14 / 2607.18615

Stochastic Meta-Unlearning: Bridging Language Backbone and Multimodal Unlearning

随机元遗忘：连接语言主干与多模态遗忘

Liu, Zijie, Duan, Jinhao, Liu, Gaowen, Liu, Sijia, Chen, Tianlong

Abstract

Machine unlearning for vision-language models (VLMs) remains underexplored. Unlike language models, VLMs combine a language backbone with visual components, which makes unlearning more complex. There is a surprising phenomenon when moving from single-modality unlearning to VLM unlearning: a target forgotten by the standalone language backbone can still be recovered when image information is given to the full VLM. This shows that text-only feedback is not enough for reliable VLM unlearning. Motivated by this observation, we propose Stochastic Meta-Unlearning (SMU), a bilevel framework that uses VLM-level feedback to learn an unlearning-ready initialization. In the inner loop, SMU applies a few unlearning steps to the language backbone using text data. In the outer loop, SMU recomposes the updated backbone with the frozen VLM and evaluates forgetting and utility at the VLM level. This design makes the unlearning update aware of the final multimodal behavior, while still keeping the update local to the language backbone. Experiments on two VLMs, two multimodal meme datasets, and three baselines show that SMU achieves the best overall forget-retain trade-off. Compared with the strongest baseline for each metric, SMU reduces average Forget accuracy by 10.52 points and improves average Retain and Test accuracy by 20.10 and 17.01 points, respectively. More importantly, SMU also transfers to new forgetting targets and to different meta-test unlearning methods. These results suggest that VLM-level feedback can make language-backbone unlearning more reliable and more transferable for VLMs.

Chinese Translation

视觉-语言模型（VLM）的机器遗忘仍然未被充分探索。与语言模型不同，VLM将语言主干与视觉组件相结合，这使得遗忘过程更加复杂。从单模态遗忘到VLM遗忘的转变中出现了一个令人惊讶的现象：一个被独立语言主干遗忘的目标在提供图像信息给完整的VLM时仍然可以被恢复。这表明，仅依靠文本反馈不足以实现可靠的VLM遗忘。基于这一观察，我们提出了随机元遗忘（Stochastic Meta-Unlearning, SMU），一个双层框架，利用VLM级别的反馈来学习一个适合遗忘的初始化。在内层循环中，SMU使用文本数据对语言主干应用少量遗忘步骤。在外层循环中，SMU将更新后的主干与冻结的VLM重新组合，并在VLM级别评估遗忘和效用。这一设计使得遗忘更新能够关注最终的多模态行为，同时仍然保持更新局限于语言主干。对两个VLM、两个多模态表情包数据集和三个基线的实验表明，SMU在遗忘与保留的权衡上达到了最佳效果。与每个指标的最强基线相比，SMU将平均遗忘准确率降低了10.52个百分点，同时平均保留和测试准确率分别提高了20.10和17.01个百分点。更重要的是，SMU还能够迁移到新的遗忘目标和不同的元测试遗忘方法。这些结果表明，VLM级别的反馈可以使语言主干的遗忘更加可靠和可迁移。

View on arXiv Download PDF AI Translation

cs.CL / 15 / 2607.18618

LatentMT: Machine Translation with Latent Reasoning

LatentMT：具有潜在推理的机器翻译

Chen, Wei-Rui, Magdy, Samar M., Zhang, Chiyu, Zhu, Wenhui, Wang, Zhipeng, Abdul-Mageed, Muhammad

Abstract

Latent-reasoning looped language models (LoopLMs) offer a different scaling path for machine translation (MT): instead of increasing parameter count or emitting explicit chain-of-thought tokens, they spend additional recurrent computation inside hidden states. We introduce LatentMT, the first systematic study of latent-reasoning LoopLMs for machine translation. LatentMT adapts a small 2.6B-parameter backbone model with lightweight training. Across 32 translation directions spanning high-, mid-, and low-resource languages, LatentMT achieves performance comparable to models three to five times larger. It is competitive in a high-resource language and achieves state-of-the-art performance on both mid-resource and low-resource languages. Studying the behavior of scaling the number of recurrent reasoning steps, we find that recurrent computation consistently improves translation quality in early steps, then saturates quickly afterwards. Our mechanistic analysis shows that hidden-representation differences shrink along the recurrent reasoning-step axis, supporting the observed saturation in performance. Finally, our efficiency analysis shows that LatentMT requires lower training and inference compute than much larger non-latent-reasoning models with similar performance, making latent recurrent computation a promising path toward compact, efficient, and strong machine translation.

Chinese Translation

潜在推理循环语言模型（LoopLMs）为机器翻译（MT）提供了一种不同的扩展路径：它们不是通过增加参数数量或发出明确的思维链标记，而是在隐藏状态中进行额外的递归计算。我们介绍了LatentMT，这是对潜在推理LoopLMs在机器翻译中应用的首次系统研究。LatentMT采用了一个小型的26亿参数主干模型，并进行了轻量级训练。在涵盖高、中、低资源语言的32个翻译方向上，LatentMT的性能可与三到五倍更大模型相媲美。在高资源语言中具有竞争力，并在中资源和低资源语言上达到了最先进的性能。通过研究递归推理步骤数量的扩展行为，我们发现递归计算在早期步骤中始终能提高翻译质量，但随后迅速饱和。我们的机制分析表明，隐藏表示在递归推理步骤轴上的差异缩小，支持了观察到的性能饱和。最后，我们的效率分析表明，LatentMT所需的训练和推理计算量低于许多更大且没有潜在推理的模型，而这些模型具有类似的性能，使得潜在递归计算成为实现紧凑、高效且强大的机器翻译的有前景的路径。

View on arXiv Download PDF AI Translation

cs.CL / 16 / 2607.18666

Fusion Embedding: A Unified Embedding Space for Text, Image, Video, and Audio

融合嵌入：文本、图像、视频和音频的统一嵌入空间

Tonmoy, Abdul Basit, Hoque, Kazi Fardinul, Arham, Md. Shahrier Islam, Luthra, Arman

Abstract

A single embedding space that covers text, images, video, and audio lets one index serve every query a user can pose. Embedding models built on vision-language backbones now lead text/image/video retrieval benchmarks but lack audio entirely, while audio-text retrieval is led by specialist systems that serve no other modality. We present the Fusion Embedding family, which adds audio to a frozen vision-language embedding base whose parameters are never updated: generation 1 (fusion-embedding-1) trains only a 16.4M-parameter connector between a frozen audio tower and the frozen base, and generation 2 (fusion-embedding-2) adds modality-gated deep adapters (44.2M parameters) whose branch never executes on text, image, or video inputs: their outputs are bit-for-bit those of the released base, verified after every training run. Because the base already binds text, images, and video, aligning audio to text alone makes audio-image retrieval emerge, with zero paired audio-visual training data. Alongside the recipe we map its design space with controlled negative results (rewriting training captions with an LLM, substituting a leaderboard-stronger audio tower, and widening the connector each reduce retrieval) and with training-protocol findings that we expect to transfer to any frozen decoder-LM embedding backbone. Both generations train in hours on a single GPU. Weights, code, and the evaluation harness are openly released.

Chinese Translation

一个覆盖文本、图像、视频和音频的单一嵌入空间使得一个索引能够服务于用户提出的每一个查询。基于视觉-语言骨干的嵌入模型目前在文本/图像/视频检索基准中处于领先地位，但完全缺乏音频，而音频-文本检索则由专门的系统主导，这些系统无法服务于其他模态。我们提出了融合嵌入系列，它在一个冻结的视觉-语言嵌入基础上增加了音频，该基础的参数从未更新：第一代（fusion-embedding-1）仅训练一个16.4M参数的连接器，连接一个冻结的音频塔和冻结的基础，而第二代（fusion-embedding-2）增加了模态门控深度适配器（44.2M参数），其分支在文本、图像或视频输入上从不执行：它们的输出与发布的基础逐位一致，在每次训练后进行验证。由于基础已经绑定了文本、图像和视频，仅将音频与文本对齐使得音频-图像检索得以出现，且没有任何配对的音频-视觉训练数据。我们还提供了设计空间的映射，包含控制负面结果（用大型语言模型重写训练标题、替换更强的音频塔以及扩大连接器，每个操作均降低检索效果）和训练协议发现，我们期望这些发现能够转移到任何冻结的解码器-LM嵌入骨干。两代模型均可在单个GPU上在数小时内完成训练。权重、代码和评估工具均已公开发布。

View on arXiv Download PDF AI Translation

cs.CL / 17 / 2607.18693

Rationale-Guided Knowledge Distillation for Cross-Lingual Stance Detection

基于推理引导的跨语言立场检测知识蒸馏

Zhou, Qiuli, Yao, Jingyuan, Tang, Shengeng, Chen, Hongzhi, Tang, Jun, Hong, Richang

Abstract

Stance detection aims to identify whether a text expresses a favorable or opposing attitude toward a given target, and serves as an important task for various downstream applications. Although existing studies have achieved strong performance in monolingual settings, especially in English, many low-resource languages such as Catalan still lack sufficient annotated data for training effective models. Cross-lingual stance detection alleviates this problem by transferring stance knowledge from resource-rich languages to low-resource languages. However, most existing methods mainly rely on semantic alignment between texts and targets, while ignoring the reasoning process required for reliable stance inference. Although Large Language Models provide strong reasoning ability, their high computational cost and inference latency limit practical deployment. To address these limitations, we propose a rationale-guided knowledge distillation framework for cross-lingual stance detection. Specifically, we use Chain-of-Thought prompting to guide Large Language Models in generating informative rationales, and distill the resulting reasoning knowledge into a compact student model. We further design a dual-path distillation mechanism to align rationale-enhanced and rationale-free representations, together with their prediction distributions. In addition, two contrastive learning strategies are introduced to improve stance discrimination. Experiments on multilingual benchmarks demonstrate that our method consistently outperforms competitive baselines.

Chinese Translation

立场检测旨在识别文本是否对给定目标表达了支持或反对的态度，并且是多种下游应用的重要任务。尽管现有研究在单语环境中取得了强劲的表现，尤其是在英语方面，但许多低资源语言（如加泰罗尼亚语）仍然缺乏足够的标注数据来训练有效的模型。跨语言立场检测通过将资源丰富语言的立场知识转移到低资源语言来缓解这一问题。然而，大多数现有方法主要依赖于文本与目标之间的语义对齐，而忽视了进行可靠立场推断所需的推理过程。尽管大型语言模型（Large Language Models）提供了强大的推理能力，但其高计算成本和推断延迟限制了实际部署。为了解决这些限制，我们提出了一种基于推理引导的跨语言立场检测知识蒸馏框架。具体而言，我们使用思维链（Chain-of-Thought）提示来引导大型语言模型生成信息丰富的推理，并将生成的推理知识蒸馏到一个紧凑的学生模型中。我们进一步设计了一种双路径蒸馏机制，以对齐增强推理和无推理表示及其预测分布。此外，引入了两种对比学习策略以提高立场区分能力。在多语言基准测试中的实验表明，我们的方法始终优于竞争基线。

View on arXiv Download PDF AI Translation

cs.CL / 18 / 2607.18725

Find Before You Fine-Tune: A Diagnostic Study of Small LLMs for Cybersecurity QA

在微调之前进行查找：小型语言模型在网络安全问答中的诊断研究

Mitra, Shaswata, Neupane, Subash, Chakraborty, Trisha, Tripathi, Himanshu, Mittal, Sudip, Piplai, Aritran, Rahimi, Shahram

Abstract

Large Language Models (LLMs) are increasingly fine-tuned for critical-domain Question-Answering (QA), yet choosing which small model to adapt, before paying the cost of adaptation, remains difficult. Fine-tuning can improve domain alignment, but it may also erode prior knowledge, weaken instruction-following, or increase hallucination, especially when labeled data are scarce or rapidly evolving as in cybersecurity. We present FiT (Find before Fine-Tune), a task-oriented diagnostic framework that characterizes small LLMs along three capabilities required for cybersecurity QA: vocabulary recognition, parametric knowledge, and contextualization of retrieved information. Using FiT, we conduct an empirical study of five open-weight 7-billion-parameter models under two fine-tuning regimes. We find that fine-tuning does not uniformly help: it consistently degrades vocabulary and parametric knowledge in small models, and the two regimes trade off differently. Knowledge-focused tuning causes moderate, rank-preserving degradation, whereas instruction-focused tuning collapses measured knowledge through induced abstention, inverting the knowledge ranking while leaving retrieval-grounded contextualization essentially intact. We quantify these regime-specific patterns with rank-correlation analysis and show that pre-fine-tuning FiT scores anticipate the direction of post-tuning change. Our results suggest that task-oriented diagnosis can screen out unsuitable models, avoid unnecessary fine-tuning, and support safer deployment of small LLMs in cybersecurity QA pipelines.

Chinese Translation

大型语言模型（LLMs）越来越多地被微调用于关键领域的问答（QA），然而在承担微调成本之前，选择适合的较小模型仍然很困难。微调可以改善领域对齐，但也可能侵蚀先前知识、削弱指令遵循能力或增加幻觉，尤其是在标注数据稀缺或快速变化的网络安全领域。我们提出了FiT（在微调之前查找），这是一个面向任务的诊断框架，描述了小型LLMs在网络安全问答中所需的三种能力：词汇识别、参数知识和检索信息的上下文化。通过使用FiT，我们对五个开放权重的70亿参数模型在两种微调机制下进行了实证研究。我们发现微调并不总是有益：它在小型模型中始终会降低词汇和参数知识，并且这两种机制的权衡方式不同。以知识为中心的微调导致适度的、保持排名的降级，而以指令为中心的微调则通过诱导弃权来崩溃测量知识，反转知识排名，同时基本保持检索基础的上下文化。我们通过排名相关性分析量化了这些机制特定的模式，并表明微调前的FiT得分可以预测微调后变化的方向。我们的结果表明，面向任务的诊断可以筛选出不适合的模型，避免不必要的微调，并支持在网络安全问答管道中更安全地部署小型LLMs。

View on arXiv Download PDF AI Translation

cs.CL / 19 / 2607.18730

Dual Attention Residuals

双重注意残差

Yu, Xingda, Li, Yining, Liu, Xinzhang, Yang, Zhihao, He, Haowei, Wang, Chao, Li, Yongxiang, Song, Shuangyong

Abstract

Recent work extends Transformer residual pathways along two complementary axes: historical retrieval selects information from earlier depths, whereas multi-stream methods maintain multiple residual trajectories. These capabilities have largely been studied in isolation, and assigning an independent retriever to each stream still prevents one trajectory from influencing depth selection in another. We propose Dual Attention Residuals (DAR), which brings multi-stream interaction into historical retrieval through reciprocal cross-stream addressing. For each target stream, DAR computes depth weights from normalized states in the opposite stream and applies them to values from the target stream's own history. The retrieved states are combined for an unchanged Transformer branch and updated through constrained gated writes; a block-form variant operates on block-level histories to control overhead. Across dense models from 0.1B to 1B parameters and a 7B sparse-MoE model, DAR consistently improves validation loss over standard residual Transformers and Attention Residuals. Routing ablations show that the gain cannot be explained by an additional stream or value projection alone. Representation and intervention analyses further show that reciprocal cross-stream selection preserves depth-wise diversity and avoids the redundancy or functional imbalance observed in alternative two-stream designs.

Chinese Translation

近期的研究扩展了Transformer的残差路径，沿着两个互补的轴线进行：历史检索从早期深度选择信息，而多流方法保持多个残差轨迹。这些能力在很大程度上是孤立研究的，并且为每个流分配一个独立的检索器仍然阻止了一个轨迹对另一个轨迹深度选择的影响。我们提出了双重注意残差（Dual Attention Residuals，DAR），通过互惠的跨流寻址将多流交互引入历史检索。对于每个目标流，DAR从相反流中的归一化状态计算深度权重，并将其应用于目标流自身历史的值。检索到的状态被组合用于不变的Transformer分支，并通过受限的门控写入进行更新；一种块形式变体在块级历史上操作以控制开销。在从0.1B到1B参数的密集模型以及一个7B稀疏-MoE模型中，DAR始终在验证损失上优于标准残差Transformer和注意残差。路由消融实验表明，增益不能仅通过额外的流或值投影来解释。表示和干预分析进一步表明，互惠的跨流选择保持了深度上的多样性，避免了在替代的双流设计中观察到的冗余或功能不平衡。

View on arXiv Download PDF AI Translation

cs.CL / 20 / 2607.18772

RF-Agent: A Practical Framework for Building Language Agents for RFIC Design

RF-Agent：用于射频集成电路设计的语言代理构建实用框架

Xing, Yueqi, He, Houbo, Wang, Jolie, Ni, Erin, Wang, Shikai, Li, Qiufeng, Cao, Weidong, Chi, Taiyun

Abstract

Large language models (LLMs) have driven rapid progress in electronic design automation (EDA), yet their application to radio-frequency (RF) circuit design remains limited by the scarcity of domain-specific datasets and standardized benchmarks. We present RF-Agent, which addresses this gap through textbook-driven knowledge distillation. A multi-agent Question-Thinking-Solution-Answer (QTSA) pipeline converts a subsection-level corpus from seven canonical RF textbooks into the first-of-its-kind RF-domain reasoning dataset (over 11,000 samples) with a dedicated multiple-choice benchmark. On this benchmark we study two adaptation strategies: supervised fine-tuning (SFT) and three retrieval-augmented generation (RAG) configurations (semantic, keyword, hybrid). Across multiple LLM families, domain-specific SFT significantly improves RF reasoning, especially for small and medium-sized models; among RAG configurations, semantic retrieval performs best, indicating embedding-based context alignment suits RF reasoning better than naive fusion. The dataset and benchmark provide a reusable foundation for future work on LLM-aided RF circuit design.

Chinese Translation

大型语言模型（LLMs）推动了电子设计自动化（EDA）的快速进展，但其在射频（RF）电路设计中的应用仍受到领域特定数据集和标准化基准稀缺的限制。我们提出了RF-Agent，通过教科书驱动的知识蒸馏来填补这一空白。一个多代理的问-思考-解决-回答（QTSA）管道将七本经典RF教科书中的子章节语料库转换为首个RF领域推理数据集（超过11,000个样本），并设有专门的多项选择基准。在该基准上，我们研究了两种适应策略：监督微调（SFT）和三种检索增强生成（RAG）配置（语义、关键词、混合）。在多个LLM家族中，领域特定的SFT显著提高了RF推理，特别是对于小型和中型模型；在RAG配置中，语义检索表现最佳，表明基于嵌入的上下文对齐更适合RF推理，而非简单融合。该数据集和基准为未来基于LLM的RF电路设计工作提供了可重用的基础。

View on arXiv Download PDF AI Translation

cs.CL / 21 / 2607.18820

CASE: Causal Alignment and Structural Enforcement for Improving Chain-of-Thought Faithfulness

CASE：因果对齐与结构强化以提高推理链的可信度

Wang, Ziming, Yao, Yinghua, Huang, Changwu, Tang, Ke, Yao, Xin

Abstract

Chain-of-thought (CoT) reasoning is widely used to improve both the performance and interpretability of large language models (LLMs), yet the generated reasoning may not faithfully support the final answer. We study this problem from a causal perspective, where a faithful CoT process should follow the chain $Z\rightarrow X\rightarrow Y$, with $Z$, $X$, and $Y$ denoting the instruction, reasoning chain, and final answer, respectively. In this process, the instruction should affect the answer only through the reasoning chain. However, conventional autoregressive LLMs condition answer generation on both the instruction and the CoT, which still allows a direct instruction-to-answer shortcut. To address this issue, we propose CASE, a framework that combines training-time causal alignment and inference-time structural enforcement. During training, CASE builds counterfactual-CoT, biased-instruction, and empty-instruction datasets, and applies selective-loss fine-tuning to strengthen CoT-to-answer dependence while suppressing instruction shortcuts. During inference, CASE masks direct attention from instruction tokens to answer tokens, preventing the model from bypassing the generated CoT. We provide an information-theoretic analysis showing how these components promote faithful chains. Experiments on three models and four benchmarks show that CASE achieves a 37\% average per-setting relative improvement in overall CoT faithfulness over the strongest baselines, exhibits stronger cross-dataset faithfulness transfer, and maintains competitive average accuracy. Code is available at https://github.com/oddwang/CASE.

Chinese Translation

推理链（Chain-of-thought, CoT）推理被广泛应用于提升大型语言模型（Large Language Models, LLMs）的性能和可解释性，但生成的推理可能并未忠实支持最终答案。我们从因果的角度研究这一问题，认为一个可信的 CoT 过程应遵循链 $Z ightarrow X ightarrow Y$，其中 $Z$、$X$ 和 $Y$ 分别表示指令、推理链和最终答案。在这一过程中，指令应仅通过推理链影响答案。然而，传统的自回归 LLM 在生成答案时同时依赖于指令和 CoT，这仍然允许指令直接跳转到答案。为了解决这一问题，我们提出了 CASE，一个结合训练时因果对齐和推理时结构强化的框架。在训练过程中，CASE 构建了反事实 CoT、偏向指令和空指令数据集，并应用选择性损失微调以增强 CoT 与答案之间的依赖关系，同时抑制指令的跳转。在推理过程中，CASE 遮蔽了指令标记对答案标记的直接关注，防止模型绕过生成的 CoT。我们提供了信息论分析，展示这些组件如何促进可信的推理链。在三个模型和四个基准上的实验表明，CASE 在整体 CoT 可信度上相较于最强基线实现了 37\% 的平均相对提升，展现出更强的跨数据集可信度转移，并保持了竞争力的平均准确率。代码可在 https://github.com/oddwang/CASE 获取。

View on arXiv Download PDF AI Translation

cs.CL / 22 / 2607.18825

AILQA: Evaluating AI-Driven Legal Question Answering Systems for the Indian Legal System

AILQA：评估针对印度法律体系的人工智能驱动法律问答系统

Nigam, Shubham Kumar, Mishra, Shubham Kumar, Shallum, Noel, Ghosh, Kripabandhu, Bhattacharya, Arnab

Abstract

This comprehensive study introduces an advanced Artificial Intelligence for Indian Legal Question Answering (AILQA) system tailored to the Indian legal context. AILQA leverages a variety of embedding and generative models, including recent Large Language Models (LLMs), to address the unique challenges posed by the intricate and diverse nature of Indian legal texts and to enhance the accuracy and reliability of responses to legal questions. We conducted rigorous evaluations using both lexical and semantic metrics, enriched by expert legal feedback, to ensure relevance and accuracy. Our findings underscore the effectiveness of the Retrieval-Augmented Generation (RAG) paradigm in improving answer quality, particularly in complex legal domains. Additionally, we assessed performance on standardized tests such as the All India Bar Examination (AIBE), thereby providing a robust benchmark for practical applications. Under the study's evaluation protocol, some AI-generated responses received higher ratings than the available reference answers, particularly when they contained accurate and relevant supporting details. This finding is specific to the evaluated dataset and rating criteria and should not be interpreted as evidence that the models generally outperform qualified legal professionals. We also discuss the challenges encountered, such as the need for precise context and the risks of model hallucination, and propose directions for future research to further refine AI capabilities in the legal field. This study aims to pave the way for enhanced legal decision-support systems, making them more accessible and effective for legal professionals and the public alike.

Chinese Translation

本研究全面介绍了一种针对印度法律背景的先进人工智能法律问答系统（AILQA）。AILQA利用多种嵌入和生成模型，包括最新的大型语言模型（LLMs），以应对印度法律文本复杂多样的特性所带来的独特挑战，并提高法律问题回答的准确性和可靠性。我们采用严格的评估方法，结合词汇和语义指标，并辅以专家法律反馈，以确保相关性和准确性。我们的研究结果强调了检索增强生成（Retrieval-Augmented Generation, RAG）范式在提高答案质量方面的有效性，尤其是在复杂法律领域。此外，我们还在全国律师考试（All India Bar Examination, AIBE）等标准化测试中评估了系统的表现，从而为实际应用提供了可靠的基准。在本研究的评估协议下，一些AI生成的回答获得了比现有参考答案更高的评分，特别是在其包含准确且相关的支持细节时。该发现特定于所评估的数据集和评分标准，不应被解读为模型普遍优于合格法律专业人士的证据。我们还讨论了遇到的挑战，如对精确上下文的需求和模型幻觉的风险，并提出了未来研究的方向，以进一步提升人工智能在法律领域的能力。本研究旨在为增强法律决策支持系统铺平道路，使其对法律专业人士和公众更加可及和有效。

View on arXiv Download PDF AI Translation

cs.CL / 23 / 2607.18839

HPD-Parsing: Hierarchical Parallel Document Parsing

HPD-解析：层次化并行文档解析

Wei, Shu, Wu, Jingjing, Zhang, Lingshu, Xie, Qunyi, Zou, Hao, Xiang, Le, Fan, Xu, Xu, Yangliu, Lin, Manhui, Ma, Xiaolong, Cui, Cheng, Du, Tengyu, YY

Abstract

Efficient teamwork typically combines global coordination with parallel execution, a principle not yet fully reflected in unified Vision-Language Model (VLM)-based document parsers. Existing unified parsers process an entire page jointly but generate its output through a single token-by-token autoregressive trajectory, creating a sequential bottleneck that grows with document length. Such full-page sequential generation overlooks a key property of document parsing: layout must be analyzed globally, whereas block content can be parsed in parallel. Based on this observation, we introduce HPD-Parsing, which replaces full-page autoregressive generation with a Hierarchical Parallel Decoding paradigm. A main layout branch organizes the overall document structure and dynamically assigns block-level content decoding to concurrent branches, while progressive multi-token prediction (P-MTP) further reduces the decoding steps within each branch. Experiments on public benchmarks show that HPD-Parsing achieves 4,752 tokens per second, delivering $2.62\times$ the throughput of the fastest existing document parsing model and $3.06\times$ that of the vanilla autoregressive baseline, while maintaining competitive parsing accuracy. These results establish hierarchical parallel decoding as an effective alternative to full-page autoregressive generation, opening a new direction for efficient unified document parsing.

Chinese Translation

高效的团队合作通常结合全球协调与并行执行，这一原则尚未在统一的基于视觉-语言模型（VLM）的文档解析器中得到充分体现。现有的统一解析器共同处理整页内容，但通过单一的逐个标记自回归轨迹生成输出，造成了一个随着文档长度增长而加剧的顺序瓶颈。这种全页顺序生成忽视了文档解析的一个关键特性：布局必须进行全局分析，而块内容可以并行解析。基于这一观察，我们提出了HPD-解析，它用层次化并行解码范式替代了全页自回归生成。一个主要的布局分支组织整体文档结构，并动态分配块级内容解码给并发分支，而渐进式多标记预测（P-MTP）进一步减少了每个分支内的解码步骤。在公共基准测试上的实验表明，HPD-解析达到了每秒4,752个标记，提供了比现有最快文档解析模型高出$2.62 imes$的吞吐量，以及比基础的自回归模型高出$3.06 imes$的吞吐量，同时保持了竞争力的解析准确性。这些结果确立了层次化并行解码作为全页自回归生成的有效替代方案，为高效的统一文档解析开辟了新的方向。

View on arXiv Download PDF AI Translation

cs.CL / 24 / 2607.18912

From a Multilingual Streaming ASR Backbone to Kenyan-Language Systems: Data-Centric Adaptation of Nemotron 3.5 for Kikuyu, Dholuo, and Kalenjin

从多语言流式自动语音识别骨干到肯尼亚语言系统：Nemotron 3.5在基库尤（Kikuyu）、多鲁奥（Dholuo）和卡伦金（Kalenjin）语言中的数据中心适应

Gatere, Mark

Abstract

Automatic speech recognition (ASR) for African languages is constrained by orthographic inconsistency, annotation artifacts, missing audio, speaker and domain imbalance, and evaluation procedures that differ from deployment. We present an end-to-end engineering study adapting NVIDIA Nemotron 3.5 ASR Streaming 0.6B to Kikuyu, Dholuo, and Kalenjin. Starting from a Kenyan Swahili-adapted checkpoint, we retain its cache-aware FastConformer RNN-T, prompt conditioning, and streaming decoder during full-parameter fine-tuning. The study covers corpus auditing, Unicode normalization, split checks, duration filtering, low-rate continuation, validation-based checkpoint selection, true-streaming evaluation, artifact preservation, and isolated serving. On internal, adaptively consulted evaluation sets excluded from gradient updates at context [56,13], selected Kikuyu and Dholuo models achieve 42.97% and 33.98% WER, respectively. Dholuo records 9.59% CER and 8.13% no-space CER under its frozen historical label policy; Kikuyu records 7.79% no-space CER. Kalenjin remains a work in progress: v1-v reaches 68.74% WER on a 2,411-row clean-v3 diagnostic subset excluding long-pause annotations, digit-bearing references, and targets shorter than three tokens. Its checkpoint selection used a mixed-source validation manifest containing test-origin rows, so the score is not an independent generalization estimate. We also report negative findings involving non-speech labels, short-utterance over-generation, boundary-sensitive WER, and cloud job-lifecycle failures. We make no state-of-the-art claim because the internal sets, repeated consultation, and normalization differ from public benchmarks. This work provides an auditable account of adapting a multilingual streaming model into language-specific systems without discarding streaming constraints.

Chinese Translation

针对非洲语言的自动语音识别（ASR）受到正字法不一致、注释伪影、缺失音频、说话者和领域不平衡，以及与部署不同的评估程序的限制。我们展示了一项端到端的工程研究，旨在将NVIDIA Nemotron 3.5 ASR Streaming 0.6B适应于基库尤、多鲁奥和卡伦金。从一个经过肯尼亚斯瓦希里语适应的检查点开始，我们在全参数微调过程中保留了其缓存感知的FastConformer RNN-T、提示条件和流式解码器。该研究涵盖了语料审计、Unicode标准化、拆分检查、持续时间过滤、低速续接、基于验证的检查点选择、真实流式评估、伪影保留和孤立服务。在内部的、适应性咨询的评估集上，这些评估集在上下文[56,13]中被排除在梯度更新之外，所选的基库尤和多鲁奥模型分别达到了42.97%和33.98%的字错误率（WER）。多鲁奥在其冻结的历史标签政策下记录了9.59%的字符错误率（CER）和8.13%的无空格CER；基库尤记录了7.79%的无空格CER。卡伦金仍在进行中：v1-v在一个2411行的clean-v3诊断子集上达到了68.74%的WER，该子集排除了长暂停注释、包含数字的参考和少于三个标记的目标。其检查点选择使用了一个混合来源的验证清单，包含测试来源的行，因此该得分并不是一个独立的泛化估计。我们还报告了涉及非语音标签、短语音过生成、边界敏感的WER和云作业生命周期失败的负面发现。由于内部集、重复咨询和标准化与公共基准不同，我们不做任何最先进的声明。这项工作提供了一个可审计的报告，展示了如何在不丢弃流式约束的情况下将多语言流式模型适应为特定语言系统。

View on arXiv Download PDF AI Translation

cs.CL / 25 / 2607.18915

Reasoning Error from Known Fact: Step-Level Self-Consistency Group Relative Policy Optimization for LLM

基于已知事实的推理错误：面向大型语言模型的逐步自一致性组相对策略优化

Hu, Xiaomeng, Hu, Jiaqi, Chen, Hao, Zhang, Qi, Shen, Zhanming, Ye, Wentao, Zhao, Junbo

Abstract

With the rapid advancement of large language models (LLMs), modern systems not only possess strong foundational capabilities and extensive knowledge, but can also solve complex problems via long, multi-step reasoning. However, as reasoning traces become longer, LLMs may produce a substantial amount of hallucinated content during the reasoning process, which is often difficult to detect. In this work, we conduct a fine-grained analysis of hallucinations arising in LLM reasoning and find that the reasoning traces are particularly prone to Context-Sensitive Factual Hallucinations: cases where the model actually has the relevant knowledge, yet makes factual errors due to contextual interference during reasoning. To address this issue, we propose Step-level Self-Consistency Group Relative Policy Optimization (SSC-GRPO), which assigns step-level rewards to reasoning traces by computing self-consistency scores of individual steps across multiple rollouts. Compared with prior methods, SSC-GRPO achieves state-of-the-art performance on both mathematical reasoning benchmarks and hallucination leaderboards. Our results offer a new perspective for detecting and mitigating hallucinations in the reasoning process of large language models.

Chinese Translation

随着大型语言模型（LLMs）的快速发展，现代系统不仅具备强大的基础能力和广泛的知识，还能够通过长时间的多步骤推理解决复杂问题。然而，随着推理轨迹的延长，LLMs在推理过程中可能会产生大量虚构内容，这些内容往往难以被检测到。在本研究中，我们对LLM推理中出现的幻觉进行了细致的分析，发现推理轨迹特别容易出现上下文敏感的事实幻觉：即模型实际上拥有相关知识，但由于推理过程中的上下文干扰而产生事实错误。为了解决这一问题，我们提出了逐步自一致性组相对策略优化（Step-level Self-Consistency Group Relative Policy Optimization，SSC-GRPO），该方法通过计算多个回合中各个步骤的自一致性得分，为推理轨迹分配逐步奖励。与之前的方法相比，SSC-GRPO在数学推理基准和幻觉排行榜上均实现了最先进的性能。我们的结果为检测和减轻大型语言模型推理过程中的幻觉提供了新的视角。

View on arXiv Download PDF AI Translation

cs.CL / 26 / 2607.18934

Transcription Policy as a Latent Variable: Activating Controllable Verbatim ASR with Word-Level Timing

转录政策作为潜在变量：激活可控的逐字自动语音识别与词级时间标记

Wagner, Laurin, Zusag, Mario, Thallinger, Bernhard

Abstract

Modern ASR models trained on heterogeneously annotated data treat transcription style (verbatim vs. intended) as an uncontrolled latent variable, causing measurable decoding instability, evaluation confounding (up to 60% of reported WER attributable to style mismatch), and unreliable word-level timing. We show that models already encode both styles; the challenge is controlled activation. Using coverage-aware decoder task tokens trained on parallel verbatim/intended transcript pairs, we raise German disfluency F1 from 10% to 79% zero-shot, despite English-only training. Full English-only fine-tuning surpasses all baselines in verbatim accuracy, disfluency detection, and intended-mode quality across both languages. We further introduce supervised cross-attention fine-tuning that improves word-level timestamps on disfluent speech beyond forced-alignment baselines. Finally, we propose verbatimize, a new task enabling scalable creation and enrichment of speech corpora with high-quality canonical verbatim transcriptions.

Chinese Translation

现代自动语音识别（ASR）模型在异质标注数据上训练，将转录风格（逐字与意图）视为一个不可控的潜在变量，这导致可测量的解码不稳定性、评估混淆（高达60%的报告字错误率（WER）可归因于风格不匹配）以及不可靠的词级时间标记。我们展示了模型已经编码了这两种风格；挑战在于如何进行可控激活。通过使用在平行逐字/意图转录对上训练的覆盖感知解码器任务标记，我们将德语的流利性F1从10%提升至79%，尽管只进行了英语训练。全面的英语单语微调在逐字准确性、流利性检测和意图模式质量方面超越了所有基线。我们进一步引入了监督交叉注意力微调，改善了流利语音的词级时间戳，超越了强制对齐基线。最后，我们提出了verbatimize，一个新任务，使得能够以高质量的标准逐字转录可扩展地创建和丰富语音语料库。

View on arXiv Download PDF AI Translation

cs.CL / 27 / 2607.18946

Constrained CTC Decoding for Efficient Diacritic Restoration

约束CTC解码用于高效的变音符号恢复

Marew, Rufael, Keleg, Amr, Aldarmaki, Hanan

Abstract

In this work, we address diacritic restoration for Arabic speech transcripts. Most speech data are undiacritized, limiting the ability of modeling fine-grained phonological distinctions. The speech modality has recently been explored as a way to complement text-based diacritic restoration efforts. We propose an efficient non-autoregressive approach for speech-to-text diacritization based on Connectionist Temporal Classification (CTC). Our method incorporates hard constraints during decoding by constructing a character-level diacritization lattice from an undiacritized transcript and restricting hypotheses to valid diacritized realizations. We evaluate on Classical Arabic and Modern Standard Arabic test sets (namely, ArVoice and ClArTTS) against a more computationally-complex multi-modal diacritic restoration baseline, and show statistically significant reductions in diacritic error rates in both, demonstrating that the proposed approach offers both performance and efficiency gains.

Chinese Translation

在本研究中，我们针对阿拉伯语语音转录的变音符号恢复问题进行了探讨。大多数语音数据未加变音符号，这限制了对细粒度音位差异的建模能力。最近，语音模态被探索作为补充基于文本的变音符号恢复工作的方式。我们提出了一种基于连接时序分类（Connectionist Temporal Classification, CTC）的高效非自回归语音到文本的变音符号化方法。我们的方法在解码过程中通过从未加变音符号的转录构建字符级变音符号化格，并将假设限制为有效的变音符号化实现，从而引入了严格的约束。我们在经典阿拉伯语和现代标准阿拉伯语测试集（即ArVoice和ClArTTS）上进行了评估，与一个计算复杂度更高的多模态变音符号恢复基线进行比较，结果显示在两者中变音符号错误率显著降低，证明了所提方法在性能和效率上均具有优势。

View on arXiv Download PDF AI Translation

cs.CL / 28 / 2607.18973

Verifiable Self-Evolution for Open-Ended Dialogue Skills via Future-Feedback Prediction

通过未来反馈预测实现可验证的开放式对话技能自我演化

Zhao, ChaoJin, Jiang, Xuan

Abstract

Textual skills provide a lightweight way to improve frozen language-model agents, but their self-evolution normally requires a stable validation signal. Such signals are natural in mathematics or code, where an answer can be checked after it changes, yet are problematic in open-ended dialogue: changing the assistant response also changes the user's next reaction, so a logged reaction cannot directly evaluate a counterfactual response. We propose future-feedback skill evolution, which first redirects self-evolution from prescribing the current answer to predicting whether the observed answer will lead to a positive or negative subsequent user signal. This prediction task is verifiable on fixed logged tuples and therefore supports validation-gated textual optimization. The evolved feedback skill captures interpretable criteria for response quality and can subsequently serve as a diagnostic and optimization target for answer skills. On a proprietary, privacy-preserving sales-assistant dataset, careful quality filtering and a balanced resolved/unresolved split yield more than 75% prediction accuracy. Beyond this result, the central contribution is a formulation that converts otherwise moving conversational feedback into a fixed offline learning target, enabling reproducible skill evolution without placing every candidate skill in live traffic. We discuss the boundary between observational verification and counterfactual validity, and position the method as an offline optimization stage rather than a replacement for final human or online evaluation.

Chinese Translation

文本技能为改进固定语言模型代理提供了一种轻量级的方法，但其自我演化通常需要稳定的验证信号。在数学或代码中，这种信号是自然存在的，因为答案在变化后可以被检查，但在开放式对话中则存在问题：助手的响应变化也会改变用户的下一反应，因此记录的反应无法直接评估反事实响应。我们提出了未来反馈技能演化，首先将自我演化的方向从规定当前答案转向预测观察到的答案是否会导致积极或消极的后续用户信号。这个预测任务可以在固定的记录元组上进行验证，因此支持验证门控的文本优化。演化后的反馈技能捕捉了响应质量的可解释标准，随后可以作为答案技能的诊断和优化目标。在一个专有的、保护隐私的销售助手数据集中，经过仔细的质量过滤和均衡的已解决/未解决分割，预测准确率超过75%。除了这一结果，核心贡献在于一种将原本动态的对话反馈转化为固定的离线学习目标的公式，从而实现可重复的技能演化，而无需将每个候选技能置于实时流量中。我们讨论了观察验证与反事实有效性之间的界限，并将该方法定位为离线优化阶段，而不是最终人类或在线评估的替代品。

View on arXiv Download PDF AI Translation

cs.CL / 29 / 2607.18983

AutoJourn: Multi-Perspective Summarisation, Bias Detection and Bias Neutralisation for LLM-Generated News in Automated Journalism

AutoJourn：面向自动化新闻的多视角摘要、偏见检测与偏见中和

Ghosh, Himel, Mosharafa, Ahmed, Groh, Georg

Abstract

We present AutoJourn, a demonstration system for multi-perspective news generation and bias-aware evaluation using large language models (LLMs). The system tackles three core challenges in responsible automated journalism: extracting diverse perspectives from unstructured social media discussions, generating summaries that preserve viewpoint diversity, and detecting or mitigating bias in AI-generated news. The pipeline integrates advanced prompt engineering with optional retrieval augmentation to produce semantically diverse perspective sets, a multi-perspective summarisation module that merges conflicting viewpoints into balanced summaries, and a bias analysis suite supporting sentence-level bias detection and type classification in the generated news article, and automatic neutralisation. Users can inspect perspective clusters, compare stance-specific summaries, generate news articles, and apply bias-aware rewrites directly in the interface. We evaluate each component with intrinsic metrics -- semantic diversity, summary quality, and bias reduction and show improvements over strong baselines while maintaining content fidelity. A live, publicly accessible demo accompanies the paper to facilitate reproducibility and further research on socially responsible automated journalism.

Chinese Translation

我们提出了AutoJourn，一个用于多视角新闻生成和偏见感知评估的演示系统，基于大型语言模型（LLMs）。该系统解决了负责任的自动化新闻报道中的三个核心挑战：从非结构化社交媒体讨论中提取多样化视角、生成保留观点多样性的摘要，以及检测或减轻AI生成新闻中的偏见。该流程将先进的提示工程与可选的检索增强相结合，以生成语义多样的视角集、一个将冲突观点融合为平衡摘要的多视角摘要模块，以及一个支持生成新闻文章中的句子级偏见检测和类型分类的偏见分析套件，及其自动中和功能。用户可以在界面中检查视角聚类、比较立场特定的摘要、生成新闻文章，并直接应用偏见感知的重写。我们使用内在指标评估每个组件——语义多样性、摘要质量和偏见减少，并展示了相较于强基线的改进，同时保持内容的忠实度。本文附带一个实时的、公开可访问的演示，以促进可重复性和对社会责任自动化新闻的进一步研究。

View on arXiv Download PDF AI Translation

cs.CL / 30 / 2607.18984

Disentangling Curriculum Learning in NLP: Towards a Unifying Taxonomy

在自然语言处理中的课程学习解构：迈向统一的分类法

Toborek, Vanessa, Seiffarth, Florian, Müller, Sebastian, Horváth, Tamás

Abstract

Despite more than a decade of curriculum learning (CL) research in NLP, the field lacks a principled account of which difficulty function or scheduler to use for a given problem. To understand what has hindered progress towards this account, we propose a fine-grained taxonomy separating difficulty evaluation from training scheduling to enable systematic analysis of CL strategies. For difficulty evaluation, we distinguish attribution source and task dependence, revealing difficulty as a perspectival concept encoding different assumptions about what makes an instance hard to learn. For scheduling, we provide the first formalisation of CL schedulers in terms of expected training contribution, enabling comparison across implementations by introducing retention regimes and monotonicity properties. Applied in a dedicated analysis of CL works in NLP, our taxonomy reveals a systematic incomparability problem: prior works conflate distinct notions of difficulty and scheduling, often pursuing different objectives under the same CL label -- hindering comparison and the accumulation of a coherent evidence base. Beyond diagnosis, the taxonomy supports the design, analysis, and comparison of CL strategies, and motivates evaluation practices that disentangle the sources of observed improvement.

Chinese Translation

尽管在自然语言处理（NLP）领域已有超过十年的课程学习（CL）研究，但该领域仍缺乏针对特定问题应使用哪种难度函数或调度器的原则性说明。为了理解阻碍这一进展的因素，我们提出了一种细致的分类法，将难度评估与训练调度分开，以便系统性地分析课程学习策略。在难度评估方面，我们区分了归因来源和任务依赖性，揭示了难度作为一个视角概念，编码了关于什么使得实例难以学习的不同假设。在调度方面，我们首次以预期训练贡献的形式对课程学习调度器进行了形式化，使得通过引入保留机制和单调性属性能够对不同实现进行比较。在对NLP领域课程学习工作的专门分析中，我们的分类法揭示了一个系统性的不可比较性问题：以往的研究混淆了难度和调度的不同概念，常常在同一课程学习标签下追求不同的目标，从而阻碍了比较和一致证据基础的积累。除了诊断之外，该分类法还支持课程学习策略的设计、分析和比较，并激励那些解构观察到的改进来源的评估实践。

View on arXiv Download PDF AI Translation

cs.CL / 31 / 2607.18999

MedDDC-Eval: Diagnosis-Decoupled Evaluation of Multi-Turn Medical Consultation Agents

MedDDC-Eval：多轮医疗咨询代理的诊断解耦评估

Zhang, Guofeng, Quan, Yizeng, Fang, Huaiyi, Lv, Jianwei, Liu, Jinyao, Duan, Xunxu, An, Lening, Ouyang, Yu, Wang, Junfeng

Abstract

Multi-turn medical consultation agents must decide what to ask, adapt to patient responses, and determine when the collected evidence is sufficient. However, coupled evaluation conflates the quality of the policy-elicited history with policy-specific terminal diagnosis generation: strong generation can compensate for a thin history, while weaker generation can obscure a rich one. We introduce MedDDC-Eval, a diagnosis-decoupled testbed that treats elicited history as the comparison object and holds the history-to-diagnosis mapping constant through a shared frozen reader. Across two held-out sources, a grounded interface and an auditable diagnosis-trajectory-efficiency (D/T/E) harness measure diagnostic usefulness, information acquisition, and efficiency. Directional semantic coverage followed by deterministic one-to-one assignment yields coherent precision-recall counts for open-ended items, with at most one credited match per prediction or reference. Holding histories fixed, changing only the diagnostic reader shifts diagnosis F1 by 2.2-19.0 points and reverses 18% and 36% of pairwise policy orderings on the Record and Dialogue splits. We further apply standard Group Relative Policy Optimization (GRPO) over interactive multi-turn rollouts to post-train Qwen3-32B using diagnosis-result and trajectory feedback. On the 100-case Record and 70-case Dialogue splits, the trained policy improves over its initialization by 9.7 and 4.6 total-score points; removing either primary signal lowers held-out joint performance. These results show that MedDDC-Eval supports controlled attribution, interpretable elicited-history measurement, and evaluation-guided evidence-acquisition policy development.

Chinese Translation

多轮医疗咨询代理必须决定询问内容，适应患者的反馈，并确定收集的证据是否足够。然而，耦合评估将政策引导的历史质量与特定政策的终端诊断生成混为一谈：强生成可以弥补薄弱的历史，而较弱的生成则可能掩盖丰富的历史。我们提出了MedDDC-Eval，这是一个诊断解耦的测试平台，将引导的历史视为比较对象，并通过共享的冻结阅读器保持历史到诊断的映射不变。在两个保留的来源中，基于实证的接口和可审计的诊断轨迹效率（D/T/E）工具衡量诊断的有效性、信息获取和效率。方向性语义覆盖后跟随确定性的一对一分配，为开放式项目提供一致的精确率-召回率计数，每个预测或参考最多只有一个被认可的匹配。保持历史不变，仅改变诊断阅读器，诊断F1分数变化在2.2到19.0点之间，并在记录和对话拆分中逆转18%和36%的成对政策排序。我们进一步在交互式多轮回合上应用标准的群体相对政策优化（GRPO），使用诊断结果和轨迹反馈对Qwen3-32B进行后训练。在100个案例的记录和70个案例的对话拆分中，训练后的政策在初始化基础上分别提高了9.7和4.6的总分；去除任一主要信号都会降低保留的联合表现。这些结果表明，MedDDC-Eval支持受控归因、可解释的引导历史测量以及评估引导的证据获取政策开发。

View on arXiv Download PDF AI Translation

cs.CL / 32 / 2607.19011

Computational Humor with Multimodal LLMs: Methods, Datasets, Evaluation, and Challenges

多模态大语言模型的计算幽默：方法、数据集、评估与挑战

Liang, Tuo, Hu, Zhe, Liu, Disheng, Li, Jing, Yin, Yu

Abstract

Multimodal humor in memes, cartoons, and comics remains difficult for AI systems because intended meaning depends on non-literal mechanisms, shared cultural knowledge, and communicative intent rather than literal scene description. This survey focuses on visual humor understanding in single-image and multi-panel artifacts, while treating humor generation as an emerging downstream frontier. We position the literature against prior humor, sarcasm, and general MLLM surveys and organize it using a capability-centric hierarchy spanning recognition, interpretation and reasoning, and generation. Under this lens, we synthesize benchmark design, evaluation protocols, and modeling paradigms, tracing the field's shift from task-specific fusion models to large-model approaches based on multimodal alignment, evidence-grounded reasoning, and controlled generation. We conclude by highlighting the main barriers to progress: shortcut-prone evaluation, limited cultural and narrative coverage, weak evidence grounding, and unresolved safety and ownership concerns.

Chinese Translation

多模态幽默在表情包、卡通和漫画中对人工智能系统仍然具有挑战性，因为其意图意义依赖于非字面机制、共享文化知识和交流意图，而非字面场景描述。本文调查聚焦于单幅图像和多面板艺术品中的视觉幽默理解，同时将幽默生成视为一个新兴的下游前沿。我们将文献与先前的幽默、讽刺及一般多模态大语言模型（MLLM）调查进行对比，并采用以能力为中心的层级结构进行组织，涵盖识别、解释与推理以及生成。在这一视角下，我们综合了基准设计、评估协议和建模范式，追踪该领域从任务特定的融合模型向基于多模态对齐、证据基础推理和受控生成的大模型方法的转变。最后，我们强调了进展的主要障碍：易受捷径影响的评估、有限的文化和叙事覆盖、薄弱的证据基础以及未解决的安全和所有权问题。

View on arXiv Download PDF AI Translation

cs.CL / 33 / 2607.19033

Content is What Remains: Invariant Speech Tokenization from Parallel Utterances

内容是剩下的：来自平行话语的不可变语音标记化

Wagner, Laurin, Thallinger, Bernhard, Stankovic, Miroslav, Zusag, Mario

Abstract

Discrete speech tokenizers aim to disentangle semantic from acoustic information, yet targets from self-supervised learning (SSL) models like HuBERT retain non-linguistic variation: speaker identity, prosody, and channel conditions leak into the tokens, inflating entropy. Our key insight is that when enough speakers utter the same words under varying conditions, linguistic content is the only shared factor. We propose PINT (Parallel INvariant Tokenization), which fine-tunes an SSL encoder with alignment losses across parallel utterances and augmentations to distill this shared residual. PINT collapses identical words onto consistent token sequences, drastically reducing conditional entropy. Unlike ASR text, PINT tokens preserve frame-level temporal grounding and serve as drop-in semantic targets for audio codecs. Experiments show a 98.7% relative reduction in speaker probe accuracy (93.1% to 1.2%), a 42% lower ABX error rate, and 27-30% lower LM perplexity versus baselines, confirming that the right invariance is key to efficient learning.

Chinese Translation

离散语音标记器旨在将语义信息与声学信息分离，然而像 HuBERT 这样的自监督学习（SSL）模型的目标保留了非语言变异：说话者身份、韵律和信道条件渗透到标记中，导致熵膨胀。我们的关键见解是，当足够多的说话者在不同条件下说出相同的词时，语言内容是唯一的共享因素。我们提出了 PINT（平行不可变标记化），它通过在平行话语和增强中的对齐损失来微调 SSL 编码器，以提炼这一共享残差。PINT 将相同的词汇压缩为一致的标记序列，显著降低条件熵。与 ASR 文本不同，PINT 标记保留帧级时间基础，并作为音频编解码器的直接语义目标。实验表明，发言者探测准确率相对降低 98.7%（从 93.1% 降至 1.2%），ABX 错误率降低 42%，与基线相比，语言模型困惑度降低 27-30%，确认了正确的不可变性是高效学习的关键。

View on arXiv Download PDF AI Translation

cs.CL / 34 / 2607.19049

Benchmarking Human and Automatic Speech Recognition of Diverse Speech: Initial Results

多样化语音的人类与自动语音识别基准测试：初步结果

Huisman, Ilse, Popa, Rares, Zhang, Yuanyuan, Scharenborg, Odette

Abstract

Humans are often considered to be the best listeners and seen as the upper-bound performance of automatic speech recognition (ASR) systems. We present a preliminary comparison of the performances of state-of-the-art ASR systems and Dutch native listeners on the recognition of "diverse" speech, specifically Dutch child and older adults' speech and Flemish. Google Telephony outperformed the other ASR systems. Importantly, the ASR systems showed similar performance to the listeners, and in specific cases even outperformed them. Slight performance differences between the listeners and ASR systems were found related to speaker's age and regional accents and utterance length. Future research should focus on making ASR systems more robust to acoustic variability related to aging and regional accents. A comparison of ASR recognition performances on the test stimuli and the full Jasmin-CGN test sets showed the influence of the specific test sets on the conclusions regarding benchmarking human and ASR performance.

Chinese Translation

人类常被认为是最佳的听众，并被视为自动语音识别（ASR）系统的上限性能。我们呈现了对最先进的ASR系统与荷兰母语听众在识别“多样化”语音方面的初步比较，特别是荷兰儿童和老年人的语音以及弗拉芒语。谷歌电话语音识别系统的表现优于其他ASR系统。重要的是，ASR系统的表现与听众相似，在特定情况下甚至超过了听众。我们发现听众与ASR系统之间的轻微性能差异与说话者的年龄、地区口音和话语长度有关。未来的研究应集中于提高ASR系统对与衰老和地区口音相关的声学变异性的鲁棒性。对测试刺激和完整的Jasmin-CGN测试集的ASR识别性能比较显示了特定测试集对关于人类与ASR性能基准测试结论的影响。

View on arXiv Download PDF AI Translation

cs.CL / 35 / 2607.19088

DAIS: Dependency-Aware Intermediate QA Supervision for Complex Reasoning

DAIS：面向依赖的中间问答监督以支持复杂推理

Wang, Yu, Fan, Ming, Zhang, Xicheng, Li, Zhiyong, Wang, Zhihu, Xu, Caiyue, Hu, Dahai, Liu, Ting

Abstract

Chain-of-thought (CoT) supervision exposes intermediate rationales, but flat rationale targets usually optimize a single reasoning sequence and provide limited supervision on how local conclusions should support later decisions. We introduce Dependency-Aware Intermediate QA Supervision (DAIS), a training-time framework that converts filtered teacher rationales into stage-level QA records. Each intermediate record predicts a local answer conditioned on the previous states needed for that decision, while the final-answer record keeps the original task format; evaluation therefore uses only the original input and optional context. Across GDPR, AIACT, MedQA, and FOLIO with multiple Qwen backbones, DAIS improves average final-answer accuracy over answer-only, flat chain-of-thought, and independent-QA baselines. On policy-compliance benchmarks, it achieves a largest gain of 5.6% and an average gain of 4.2% over the strongest non-DAIS baseline. Controlled ablations show that valid previous-state conditioning contributes beyond longer targets or additional intermediate text, supporting dependency-conditioned intermediate QA as a lightweight auxiliary supervision signal for standard final-answer inference.

Chinese Translation

链式思维（Chain-of-thought, CoT）监督揭示了中间推理过程，但平坦的推理目标通常优化单一的推理序列，并对局部结论如何支持后续决策提供有限的监督。我们提出了面向依赖的中间问答监督（Dependency-Aware Intermediate QA Supervision, DAIS），这是一个训练时框架，将过滤后的教师推理转化为阶段级问答记录。每个中间记录在所需的先前状态的条件下预测局部答案，而最终答案记录保持原始任务格式；因此，评估仅使用原始输入和可选上下文。在GDPR、AIACT、MedQA和FOLIO等多个Qwen骨干网络上，DAIS在答案仅、平坦链式思维和独立问答基线之上提高了平均最终答案的准确性。在政策合规基准上，它相较于最强的非DAIS基线实现了5.6%的最大增益和4.2%的平均增益。控制性消融实验表明，有效的先前状态条件超越了更长的目标或额外的中间文本，支持依赖条件的中间问答作为标准最终答案推理的轻量级辅助监督信号。

View on arXiv Download PDF AI Translation

cs.CL / 36 / 2607.19101

Translation as Augmentation: Effect of Translated Data on Assessment of Difficulty

翻译作为增强：翻译数据对难度评估的影响

Wu, Yiheng, Hou, Jue, Yangarber, Roman

Abstract

Reliable Text Difficulty Assessment is a prerequisite for valid text simplification workflows and personalized learning applications. However, the development of robust assessment models is severely hindered by a critical bottleneck: the scarcity of expert-annotated corpora containing fine-grained difficulty levels (e.g., CEFR), particularly for lower-resource languages. This paper addresses this data scarcity problem in the context of a low-resource European language. We propose a cross-lingual data augmentation strategy that leverages machine translation to transfer labeled resources from high-resource languages to the target low-resource language. We train BERT-based regression models to predict difficulty scores and investigate whether synthetic, translated data can effectively supplement native training sets. Our experiments demonstrate that augmenting scarce native data with machine-translated corpora significantly improves the accuracy of difficulty estimation, offering a viable solution for languages lacking extensive expert annotations.

Chinese Translation

可靠的文本难度评估是有效文本简化工作流程和个性化学习应用的前提。然而，强大的评估模型的开发受到一个关键瓶颈的严重阻碍：缺乏包含细粒度难度等级（例如，CEFR）的专家注释语料库，尤其是在低资源语言中。本文在低资源欧洲语言的背景下解决了这一数据稀缺问题。我们提出了一种跨语言数据增强策略，利用机器翻译将标注资源从高资源语言转移到目标低资源语言。我们训练基于BERT的回归模型来预测难度评分，并研究合成的翻译数据是否能有效补充本地训练集。我们的实验表明，用机器翻译语料库增强稀缺的本地数据显著提高了难度估计的准确性，为缺乏广泛专家注释的语言提供了一种可行的解决方案。

View on arXiv Download PDF AI Translation

cs.CL / 37 / 2607.19178

Automated Extraction of Techno-Economic Data from 76,000 Energy System Studies

从76,000个能源系统研究中自动提取技术经济数据

Gorres, Maxime, Göpfert, Jan, Kuckertz, Patrick, Hartono, Noor Titan Putri, Heinrichs, Heidi, Linßen, Jochen, Staffel, Iain, Weinand, Jann Michael

Abstract

Energy system models guide societally important decisions, but their credibility rests on quantitative assumptions that are difficult to source and audit. Meta-analyses can improve transparency and modeling practices, but the rapid growth of publications makes manual information extraction increasingly impractical. Consequently, databases are updated infrequently and efforts are often duplicated across research groups. Here, we demonstrate the highly accurate automated extraction of quantitative information from 76,000 energy system studies published since 2010. We compile 3.2 million structured quantitative data points together with 20 million associated metadata entries, spanning a broad spectrum of technologies, methodological approaches and system characteristics. Beyond providing input data for models, the resulting FAIR database make the energy systems literature itself analysable. We show where academic assumptions diverge from empirical observed data, and how research priorities vary at scale across technologies, regions and time. To facilitate broad use within the community, the database is provided through an interactive dashboard, enabling users to filter, analyse and download data according to their specific research needs.

Chinese Translation

能源系统模型指导着对社会重要决策的制定，但其可信度依赖于难以获取和审计的定量假设。元分析可以提高透明度和建模实践，但出版物的快速增长使得手动信息提取变得越来越不切实际。因此，数据库更新的频率较低，研究小组之间的努力往往重复。在此，我们展示了从2010年以来发布的76,000个能源系统研究中高精度自动提取定量信息的能力。我们编制了320万个结构化的定量数据点以及2000万个相关的元数据条目，涵盖了广泛的技术、方法论和系统特征。除了为模型提供输入数据外，所生成的FAIR数据库还使能源系统文献本身可供分析。我们展示了学术假设与实证观察数据之间的差异，以及研究优先级在不同技术、地区和时间尺度上的变化。为了便于社区的广泛使用，该数据库通过交互式仪表板提供，使用户能够根据特定的研究需求过滤、分析和下载数据。

View on arXiv Download PDF AI Translation

cs.CL / 38 / 2607.19181

Reasoning Before Translation: Enhancing Legal Machine Translation with Structured Reasoning

翻译前的推理：通过结构化推理提升法律机器翻译

An, Aixiu, Jungo, Michael, Eynard, Eloi, Drenhaus, Mark, Fischer, Andreas, Hennebert, Jean, Rumley, Sébastien

Abstract

Neural machine translation (NMT) in the legal domain is a linguistically and conceptually demanding task, primarily due to the complexity of legal language and the high level of precision it requires. The recent emergence of reasoning-capable language models opens new possibilities for tackling such challenges. They add to a set of other previously proposed techniques to enhance the translation quality, which includes supervised fine-tuning and reinforcement learning. In this work, we perform a comparison between these various approaches. More particularly, we evaluate small language models such as Qwen3.5 4B, Qwen3.5 9B, and Gemma 3 12B enhanced with various re-training paradigms and compare their performances against frontier reasoning models. We focus on the Swiss legal system, which -- with its unique multilingual statutes -- offers a particularly challenging testbed for reasoning-augmented models. Our results show that the quality of small ``base'' models can be greatly enhanced, and that reinforcement learning with verifiable rewards can be applied to NMT in the legal domain and surpasses the translation quality of supervised fine-tuning. The performance of enhanced small models is close to the one of state-of-the-art reasoning models yet remains inferior. We also note that re-training paradigms yield diminishing returns as model size increase. The code and models are publicly available at https://github.com/aixiuxiuxiu/Legal-MT-SFT-RL.

Chinese Translation

法律领域的神经机器翻译（NMT）是一项在语言和概念上要求极高的任务，主要由于法律语言的复杂性以及其所需的高精度。最近出现的具备推理能力的语言模型为应对这些挑战开辟了新的可能性。它们与之前提出的其他技术相结合，以提升翻译质量，包括监督微调和强化学习。在本研究中，我们对这些不同的方法进行了比较。特别地，我们评估了小型语言模型，如 Qwen3.5 4B、Qwen3.5 9B 和 Gemma 3 12B，这些模型经过各种再训练范式的增强，并将其性能与前沿推理模型进行了比较。我们重点关注瑞士法律体系，该体系以其独特的多语言法规为推理增强模型提供了一个特别具有挑战性的测试平台。我们的结果表明，小型“基础”模型的质量可以大幅提升，并且可验证奖励的强化学习可以应用于法律领域的 NMT，并超越监督微调的翻译质量。增强的小型模型的性能接近于最先进的推理模型，但仍然略逊一筹。我们还注意到，随着模型规模的增加，再训练范式的收益递减。代码和模型可在 https://github.com/aixiuxiuxiu/Legal-MT-SFT-RL 上公开获取。

View on arXiv Download PDF AI Translation

cs.CL / 39 / 2607.19201

MIRA-Ev:A Benchmark for Granular Evidence Detection and Relational Reasoning in Clinical Exams

MIRA-Ev：临床考试中颗粒证据检测与关系推理的基准

De la Iglesia, Iker, Ramirez-Romero, Johanna, Villa-Gonzalez, Jose Maria, García, Irune Urroz, Barrena, Ander, Atutxa, Aitziber

Abstract

Clinical NLP evaluation remains dominated by multiple-choice question answering (MCQA), which scores only final-answer accuracy and cannot detect when a model reaches the correct diagnosis while grounding it in irrelevant, absent, or contradictory evidence. We introduce MIRA-Ev, a clinical argument mining benchmark built on Spanish M\'edico Interno Residente (MIR) licensing-exam cases, re-annotated by expert clinicians with span-level premises, claims, and directed support/attack relations, and released in parallel Spanish (native), English, and Basque versions, the first clinical argumentation resource in Basque. MIRA-Ev organizes evaluation into a three-tier task hierarchy: evidence sentence retrieval, argumentative component extraction, and relation classification.

Chinese Translation

临床自然语言处理（NLP）评估仍然主要集中在选择题回答（MCQA）上，该方法仅评分最终答案的准确性，无法检测模型在不相关、缺失或矛盾证据的情况下是否达到了正确的诊断。我们介绍了MIRA-Ev，这是一个基于西班牙Médico Interno Residente（MIR）执业考试案例构建的临床论证挖掘基准，由专家临床医生重新注释，包含跨度级前提、主张及有向支持/攻击关系，并以西班牙语（母语）、英语和巴斯克语的平行版本发布，这是首个巴斯克语的临床论证资源。MIRA-Ev将评估组织为三级任务层次结构：证据句检索、论证组件提取和关系分类。

View on arXiv Download PDF AI Translation

cs.CL / 40 / 2607.19219

Beyond Score Prediction: LLM-Based Essay Scoring and Feedback Generation via Reinforcement Learning with Rubric Rewards

超越分数预测：基于大语言模型的论文评分与反馈生成通过带有评分标准奖励的强化学习

Jin, Xuefeng, Zhang, Jiashuo, Cao, Teng, Yang, Bin

Abstract

Large language models (LLMs) have been widely applied to automated essay scoring (AES) and automated feedback generation (AFG). However, existing studies rely primarily on prompt engineering or supervised fine-tuning, while systematic research on reinforcement learning (RL) post-training and automated evaluation of feedback quality remains limited. We propose RLAES, a unified LLM framework that jointly optimizes essay scoring and feedback generation through RL. To make feedback quality measurable, interpretable, and usable for training, we introduce Rubric-based Feedback Evaluation (RFE), an essay-grounded feedback evaluation framework comprising 166 fine-grained binary rubric items and an LLM-as-judge. Building on RFE, we propose Adaptive Gated Feedback Optimization (AGFO), which activates rubric-based feedback rewards on demand during RL, reducing evaluation overhead while improving feedback quality. We also propose Adjacent Contrastive Reasoning (ACR) to improve ordinal score calibration by explicitly contrasting adjacent score levels. Experimental results show that the RFE framework captures essay-feedback consistency, exhibits strong pairwise discriminative power, and closely aligns with expert preferences. On the ASAP benchmark, RLAES-AGFO achieves the best scoring performance among LLM-based methods (QWK = 0.803), while maintaining feedback quality comparable to GPT-5.5 and avoiding the feedback degradation observed under score-only RL. Code and datasets are publicly available at https://github.com/hellomuyi/RLAES.

Chinese Translation

大语言模型（LLMs）已广泛应用于自动化论文评分（AES）和自动化反馈生成（AFG）。然而，现有研究主要依赖于提示工程或监督微调，而对强化学习（RL）后训练和反馈质量自动评估的系统研究仍然有限。我们提出了RLAES，一个统一的LLM框架，通过RL联合优化论文评分和反馈生成。为了使反馈质量可测量、可解释并可用于训练，我们引入了基于评分标准的反馈评估（RFE），这是一个以论文为基础的反馈评估框架，包含166个细粒度的二元评分标准项和一个LLM作为评判者。在RFE的基础上，我们提出了自适应门控反馈优化（AGFO），该方法在RL过程中按需激活基于评分标准的反馈奖励，减少评估开销，同时提高反馈质量。我们还提出了相邻对比推理（ACR），通过明确对比相邻分数水平来改善序数分数校准。实验结果表明，RFE框架捕捉了论文与反馈的一致性，展现出强大的成对区分能力，并与专家偏好紧密对齐。在ASAP基准上，RLAES-AGFO在基于LLM的方法中实现了最佳评分表现（QWK = 0.803），同时保持了与GPT-5.5相当的反馈质量，并避免了仅基于分数的RL下观察到的反馈退化。代码和数据集可在https://github.com/hellomuyi/RLAES公开获取。

View on arXiv Download PDF AI Translation

cs.CL / 41 / 2607.19226

The Price of Reasoning: Cost-Quality Tradeoffs in Reinforcement Learning for Neural Machine Translation

推理的代价：强化学习在神经机器翻译中的成本-质量权衡

Jungo, Michael, An, Aixiu

Abstract

Reinforcement learning with verifiable rewards (RLVR) has been established as a viable paradigm for the post-training of Large Language Models (LLMs), including downstream tasks, such as Neural Machine Translation (NMT). With the latest research indicating that RLVR could be the preferred training method for translating legal documents due to the induced reasoning capabilities, it raises the question whether it is really attributed to the reasoning or more generally to the training paradigm. We investigate the importance of including the model's reasoning trace in the generated responses during both training and inference by systematically omitting it from one of the phases. Our experiments show that including the reasoning, specifically during inference, has a positive effect on the overall translation quality. Furthermore, we recognise that the reasoning leads to an increase in output tokens, hence we study the cost-quality tradeoff between the increased computational demands and the improved translation quality.

Chinese Translation

可验证奖励的强化学习（RLVR）已被确立为大型语言模型（LLMs）后训练的可行范式，包括神经机器翻译（NMT）等下游任务。最新研究表明，由于其引发的推理能力，RLVR可能是翻译法律文件的首选训练方法，这引发了一个问题：这种效果是否真正归因于推理，还是更普遍地归因于训练范式。我们通过系统性地在训练和推理的一个阶段中省略模型的推理轨迹，研究在生成响应时包含推理轨迹的重要性。我们的实验表明，在推理阶段包含推理轨迹对整体翻译质量有积极影响。此外，我们认识到推理会导致输出标记的增加，因此我们研究了增加的计算需求与改善的翻译质量之间的成本-质量权衡。

View on arXiv Download PDF AI Translation

cs.CL / 42 / 2607.19231

Selection Shapes the Boundary: A Preregistered Replication of Monotonicity and Label Agreement in Unselected NLI Populations

选择塑造边界：对未选择的自然语言推理人群中单调性和标签一致性的预注册复制研究

Choi, Haram

Abstract

Prior work on human label variation (HLV) in natural language inference (NLI) has often relied on re-annotation resources that select items by disagreement level. An earlier study (arXiv:2607.15870) found that hypotheses containing non-upward monotonicity operators showed lower label agreement in ChaosNLI (Cliff's delta = -0.284), which is restricted to items whose majority label carries exactly three of five votes. We preregistered a replication of this boundary in the unselected populations that ChaosNLI was drawn from: the SNLI and MultiNLI development sets, using the same operator tagger and a four-level ordinal agreement outcome. The registered prediction fails. All seven contrasts return a positive Cliff's delta (non-upward items agree slightly more, not less), the only significant confirmatory contrast has the opposite sign to the registration, and every effect is far below our smallest effect size of interest (0.10). Robustness checks support the measurement: simulated tagger misclassification shrinks the effects rather than manufacturing them, and a manual re-tagging audit reaches four-class agreement of 0.875 on a fresh 200-item sample. We conclude that the earlier negative boundary is plausibly a structure conditional on low-agreement selection rather than a population-level property, and that HLV structure claims built on selected re-annotation resources should state their selection conditional explicitly.

Chinese Translation

以往关于自然语言推理（NLI）中人类标签变异（HLV）的研究通常依赖于通过不一致性水平选择项目的重新标注资源。一项早期研究（arXiv:2607.15870）发现，包含非向上单调性运算符的假设在ChaosNLI中的标签一致性较低（Cliff's delta = -0.284），该数据集仅限于多数标签恰好获得五票中的三票的项目。我们预注册了在ChaosNLI所选取的未选择人群（即SNLI和MultiNLI开发集）中复制这一边界，使用相同的运算符标注器和四级序数一致性结果。注册的预测未能实现。所有七个对比返回正的Cliff's delta（非向上项目的标签一致性略高，而非更低），唯一显著的确认对比与注册的符号相反，且每个效应均远低于我们关注的最小效应大小（0.10）。稳健性检查支持测量结果：模拟标注器错误分类缩小了效应而非制造效应，手动重新标注审核在新的200项样本上达成了0.875的四类一致性。我们得出结论，早期的负边界可能是依赖于低一致性选择的结构，而非人群层面的属性，并且基于选择性重新标注资源的HLV结构声明应明确说明其选择条件。

View on arXiv Download PDF AI Translation

cs.CL / 43 / 2607.19235

MeetingToM: Evaluating Multimodal LLMs on Theory-of-Mind Reasoning in Multi-Party Meetings

MeetingToM：在多方会议中评估多模态大语言模型的心智理论推理

Wang, Ziyi, Wu, Yuhang, Piao, Dongxu, Liu, Xingyu, Zhou, Tianhui, Liu, Miao

Abstract

Theory of Mind (ToM), the ability to infer other's beliefs, intentions, and states of knowledge, is central to social interaction, yet remains challenging for current Multimodal Large Language Models (MLLMs), especially in multi-party meetings where cues are distributed across speech and behavior. Existing multimodal ToM benchmarks mainly focus on video-grounded question answering over overt, externally verifiable signals, and provide limited coverage of latent social states and group dynamics. We introduce MeetingToM, a benchmark for complex social behavior reasoning in naturalistic multi-party meetings. MeetingToM targets meeting-specific phenomena such as \textbf{pseudo-consensus}, where apparent agreement masks private dissent under social pressure. The benchmark is hierarchically organized to evaluate ToM at increasing levels of social granularity, including (i) subject-level mental state prediction, (ii) dyadic-level addressee understanding, and (iii) group-level consensus reasoning. We provide a unified evaluation protocol and conduct systematic analyses of representative MLLMs, revealing persistent limitations in integrating non-verbal cues, inferring hidden attitudes, and distinguishing genuine consensus from pseudo-consensus. Our results highlight key challenges and establish MeetingToM as a testbed for advancing meeting-grounded ToM in multimodal models.

Chinese Translation

心智理论（Theory of Mind, ToM）是推断他人信念、意图和知识状态的能力，这在社会互动中至关重要，但对于当前的多模态大语言模型（Multimodal Large Language Models, MLLMs）来说仍然具有挑战性，尤其是在多方会议中，线索分散在言语和行为之间。现有的多模态ToM基准主要集中在基于视频的问答，侧重于显性、可外部验证的信号，且对潜在社会状态和群体动态的覆盖有限。我们引入了MeetingToM，这是一个针对自然多方会议中复杂社会行为推理的基准。MeetingToM针对会议特定现象，如 extbf{伪共识}（pseudo-consensus），在社会压力下表面上的一致掩盖了私人异议。该基准以层级方式组织，以在不断增加的社会细粒度水平上评估ToM，包括（i）主体层面的心理状态预测，（ii）二人层面的受众理解，以及（iii）群体层面的共识推理。我们提供了统一的评估协议，并对代表性的MLLMs进行了系统分析，揭示了在整合非语言线索、推断隐藏态度以及区分真实共识与伪共识方面的持续局限。我们的结果突显了关键挑战，并确立了MeetingToM作为推进多模态模型中会议基础ToM的试验平台。

View on arXiv Download PDF AI Translation

cs.CL / 44 / 2607.19243

Inference-Time Steering for Cross-Lingual Factual Consistency in LLMs

跨语言事实一致性的推理时引导在大型语言模型中的应用

Manev, Alexander

Abstract

Although Large Language Models (LLMs) demonstrate remarkable multilingual fluency, their internal knowledge representations remain disproportionately biased toward high-resource languages. This leads to cross-lingual factual inconsistency, where they shift their empirical answer distributions based solely on the prompt language. We investigate whether these biases can be mitigated at inference time, forcing an English-prompted model to answer as if it were queried in target languages (German, Spanish, Bulgarian), and evaluate four intervention strategies: zero-shot contextual steering (persona prompting), internal representation manipulation via Contrastive Activation Addition (CAA), and lightweight weight modification via Direct Preference Optimization (DPO) trained on benchmark-derived factual data as well as conceptual generalization data. To assess alignment, we curate a multilingual factual dataset alongside a novel generalization benchmark comprising culturally rooted queries to determine whether factual interventions transfer to broader target-centric preferences. Experiments on Gemma 3 12B Instruct reveal persona prompting to be the strongest overall intervention, balancing efficacy, safety, and out-of-domain generalization. While CAA yields sharp inconsistency benchmark shifts, it is configuration-sensitive and risks knowledge degradation. DPO-based adapters offer permanent, yet narrower and less transferable gains. These findings suggest that cross-lingual inconsistency is at least partly a selection problem, and that simple contextual interventions may outperform more invasive methods for robust, transferable alignment.

Chinese Translation

尽管大型语言模型（LLMs）展现出显著的多语言流利性，但其内部知识表示仍然对高资源语言存在不成比例的偏向。这导致了跨语言的事实不一致性，即它们的经验回答分布仅基于提示语言而发生变化。我们研究了是否可以在推理时减轻这些偏见，强迫以英语提示的模型像在目标语言（德语、西班牙语、保加利亚语）中查询一样回答，并评估四种干预策略：零样本上下文引导（个性提示）、通过对比激活添加（Contrastive Activation Addition, CAA）进行的内部表示操控，以及通过基于基准推导的事实数据和概念泛化数据训练的直接偏好优化（Direct Preference Optimization, DPO）进行的轻量级权重修改。为了评估一致性，我们整理了一个多语言事实数据集，并创建了一个新的泛化基准，包含文化根植的查询，以确定事实干预是否能够转移到更广泛的以目标为中心的偏好。对Gemma 3 12B Instruct的实验表明，个性提示是最强的干预方法，平衡了有效性、安全性和领域外泛化。虽然CAA产生了明显的不一致基准变化，但它对配置敏感，并有导致知识退化的风险。基于DPO的适配器提供了持久的、但更窄且可转移性较低的收益。这些发现表明，跨语言不一致性至少在某种程度上是一个选择问题，简单的上下文干预可能优于更具侵入性的方法，以实现稳健且可转移的一致性。

View on arXiv Download PDF AI Translation

cs.CL / 45 / 2607.19257

Prompt Design at Scale: How Format, Instruction Count, and Context Length Shape Instruction Adherence and Hallucination in Large Language Models

大规模提示设计：格式、指令数量和上下文长度如何影响大型语言模型的指令遵循和幻觉现象

Eliav, Netanel

Abstract

Practitioners make three prompt-design decisions with almost no controlled evidence behind them: how to format instructions and context (markdown, plain text, prose, or tabular), how many simultaneous instructions a system prompt can carry before compliance degrades, and how much context a model can hold before recall and honesty degrade. We report two controlled experiments crossing all three factors on one held, contamination-free synthetic corpus (the "Book of Veyra," 8,780 uniquely-named entities, deterministically regenerable from a fixed seed), evaluated across five models. Experiment 1 (960 calls/model) measures instruction-following decay as rule count N grows from 10 to 160, crossed with four formats and system-prompt vs. user-turn placement. Perfect-response rate collapses to zero by N=80 for every model, format, and placement. Placement produces effects at least as large as format at N=160 in most models, but the direction is model-specific. No model shows a reliable markdown advantage; one 35B model favors plain text instead. Experiment 2 (5,520 calls/model) measures recall accuracy, false-premise sycophancy, and absent-fact fabrication across a 2k-to-512k-token context ladder in the same four formats. Recall stays near ceiling through 64-128k tokens, then degrades sharply and format-dependently: one model's accuracy spread reaches 48 points at 128k tokens. Fabrication never occurs (0/5,760 probes), and sycophancy stays negligible (<=8.3%). What rises sharply near each model's context ceiling is outright refusal to answer (0% to 79-90%), distinct from sycophancy or fabrication. Neither pre-registered format ordering holds, and token overhead (+22% to +37% over plain text) further changes which format is preferable where accuracy spread is genuine. We release the full harness, corpus generator, and raw results (VeyraBench): https://github.com/iNetanel/veyrabench

Chinese Translation

从几乎没有控制证据的背景下，实践者在提示设计中做出三个决策：如何格式化指令和上下文（markdown、纯文本、散文或表格），系统提示可以承载多少个同时指令而不降低遵循度，以及模型在回忆和诚实性下降之前可以保持多少上下文。我们报告了两个控制实验，交叉了这三个因素，基于一个无污染的合成语料库（“Veyra书”，包含8,780个独特命名实体，可以从固定种子确定性再生），并在五个模型上进行了评估。实验1（每个模型960次调用）测量了随着规则数量N从10增长到160，指令遵循的衰减情况，交叉了四种格式以及系统提示与用户回合的放置。对于每个模型、格式和放置，完美响应率在N=80时降至零。在大多数模型中，放置产生的效果至少与格式一样大（在N=160时），但方向因模型而异。没有模型显示出可靠的markdown优势；一个35B的模型更倾向于纯文本。实验2（每个模型5,520次调用）测量了在同样四种格式下，2k到512k标记上下文梯度中的回忆准确性、虚假前提的阿谀奉承和缺失事实的虚构。回忆在64-128k标记之间保持接近上限，然后急剧下降且依赖于格式：一个模型在128k标记时的准确性差异达到48分。虚构从未发生（0/5,760次探测），而阿谀奉承保持在微不足道的水平（<=8.3%）。在每个模型的上下文上限附近，拒绝回答的比例急剧上升（0%到79-90%），这与阿谀奉承或虚构不同。没有预注册的格式顺序成立，标记开销（比纯文本增加22%到37%）进一步改变了在准确性差异真实的情况下，哪种格式更可取。我们发布了完整的工具、语料库生成器和原始结果（VeyraBench）：https://github.com/iNetanel/veyrabench

View on arXiv Download PDF AI Translation

cs.CL / 46 / 2607.19322

Two-Level Meta-Rubrics for Evaluating Open-Ended Generation: GAMUT, a Benchmark for Factual Completeness

用于评估开放式生成的双层元评分标准：GAMUT，一个关于事实完整性的基准

Chen, Xilun, Feizollahi, Zhaleh, Goodwin, Ross, Moon, Seungwhan, Yih, Scott, Donmez, Pinar, Damavandi, Babak, Dong, Luna

Abstract

Evaluating the factuality of long-form generations has focused predominantly on precision, measuring whether the claims a model makes are correct. The dominant decompose-search-verify pipeline catches incorrect claims well but says little about whether a response contains all the information it should. Measuring factual completeness, the missing half of factuality, is harder: it requires enumerating the full set of facts a complete answer should contain, and these facts rarely form a flat list. They often involve open-ended sets where coverage is what matters, ordered processes, and relationships among facts that a list of independent boolean checks fails to capture. We introduce a two-level meta-rubric framework for evaluating open-ended generation, and instantiate it as Gamut (Grounded Assessment of Multimodal Factuality), a benchmark for factual completeness in long-form generation. The framework rests on a two-level rubric representation: a structured meta-rubric captures the organization and importance of the required content, which is then mechanically compiled into a flat checklist of binary, machine-gradable rubrics that an LLM judge scores reliably. We construct 1,813 questions grounded in real wearable imagery across 10 diverse domains, each paired with an evidence-backed rubric verified by expert human annotators. Because the framework is modality-agnostic, we also release a text-only variant. Evaluating 14 frontier and open-weight models, we find the benchmark genuinely challenging (best score 58.7% from Gemini 3.1 Pro), highly discriminative, and robust to the choice of judge.

Chinese Translation

对长文本生成的事实性评估主要集中在准确性上，测量模型所做的陈述是否正确。主流的分解-搜索-验证流程能够有效捕捉不正确的陈述，但对响应是否包含应有的所有信息却知之甚少。测量事实完整性，即事实性的缺失部分，更为困难：它需要列举出完整答案应包含的所有事实，而这些事实通常不会形成一个简单的列表。它们往往涉及开放式集合，其中覆盖范围至关重要，包含有序过程以及事实之间的关系，而独立布尔检查的列表无法捕捉这些关系。我们提出了一种用于评估开放式生成的双层元评分标准框架，并将其具体化为Gamut（多模态事实性基础评估），这是一个关于长文本生成中事实完整性的基准。该框架基于双层评分标准表示：结构化的元评分标准捕捉所需内容的组织和重要性，然后机械地编译成一个平面检查表，包含二元的、可机器评分的评分标准，供大型语言模型（LLM）评审者可靠评分。我们构建了1,813个基于真实可穿戴图像的问题，涵盖10个不同领域，每个问题都配有经过专家人工标注验证的证据支持评分标准。由于该框架不依赖于特定模态，我们还发布了一个仅文本的变体。在评估14个前沿和开放权重模型时，我们发现该基准确实具有挑战性（最佳得分为58.7%，来自Gemini 3.1 Pro），具有高度的区分性，并且对评审者的选择具有鲁棒性。

View on arXiv Download PDF AI Translation

cs.CL / 47 / 2607.19326

Selective State-Space Adaptation and Retrieval for Language Model Reasoning

语言模型推理的选择性状态空间适应与检索

Dokme, Atahan, Heck, Larry

Abstract

Low-rank adaptation introduces a static learned update applied identically to every input. The update provides task-level adaptation but does not explicitly represent token-level or instance-level state variation. A family of adapters is proposed that introduces selective state-space recurrence at two complementary granularities. At the token level, \textbf{MaLoRA} (Mamba-modulated low-rank adaptation) makes the adapter's scaling factor a dynamic input-dependent function with recurrent state across tokens, in contrast to the stateless modulators of prior work. At the context level, \textbf{MaRA} (Mamba Retrieval Adapter) tracks cross-segment state and selects the segments most relevant to the query, before the modulated language model generates its answer. Across three frozen backbones (Qwen-2.5-7B, Llama-3.1-8B, Gemma-2-9B) and two reasoning benchmarks (MuSiQue, 2WikiMultihopQA), the family improves reasoning accuracy on every cell of the $3{\times}2$ grid, by $+6.8$ F1 ($+10.5\%$ relative) on average and up to $+9.3$ F1 ($+18.2\%$ relative) on the hardest cell over the LoRA baseline, and the token-level gains carry to RULER QA-2 under length stress.

Chinese Translation

低秩适应引入了一种静态学习更新，该更新对每个输入应用相同。该更新提供了任务级别的适应，但并未明确表示令牌级别或实例级别的状态变化。我们提出了一系列适配器，在两个互补的粒度上引入选择性状态空间递归。在令牌级别， extbf{MaLoRA}（Mamba调制低秩适应）使适配器的缩放因子成为一个动态的输入依赖函数，并在令牌之间具有递归状态，这与先前工作的无状态调制器形成对比。在上下文级别， extbf{MaRA}（Mamba检索适配器）跟踪跨段状态，并选择与查询最相关的段，然后调制后的语言模型生成其答案。在三个冻结的基础模型（Qwen-2.5-7B、Llama-3.1-8B、Gemma-2-9B）和两个推理基准（MuSiQue、2WikiMultihopQA）上，该系列在$3{ imes}2$网格的每个单元上提高了推理准确性，平均提高了$+6.8$ F1（相对提高$+10.5 ext{%}$），在最难的单元上提高了最高$+9.3$ F1（相对提高$+18.2 ext{%}$），并且令牌级别的增益在长度压力下也延续到RULER QA-2。

View on arXiv Download PDF AI Translation

cs.CL / 48 / 2607.19345

Copy Less, Ground More: Overcoming Repetitive Copying in Long-Context Reasoning via Evidence-Aware Reinforcement Learning

减少复制，增强基础：通过证据感知强化学习克服长上下文推理中的重复复制问题

Fang, Lizhe, Shen, Weizhou, Tang, Tianyi, Wang, Yisen

Abstract

Large language models that generate step-by-step reasoning traces have achieved strong performance on complex tasks, and extending them to long-context settings has emerged as an important frontier. However, we identify a critical failure mode in this regime: \emph{repetitive copying}, where models extensively copy text from the input into their reasoning traces rather than productively solving the problem. We show that this behavior is pervasive across frontier long-context LLMs and intensifies with context length. By separating each prompt into task-relevant key evidence and irrelevant distractor context, we further show that the root cause is insufficient grounding: models copy from the prompt indiscriminately, and those that fail to focus on key evidence are far more likely to answer incorrectly. Motivated by this diagnosis, we propose GEAR (Grounding Evidence-Aware Reward), a reward shaping method that augments the accuracy signal with a grounding reward for overlap with key evidence and a distractor penalty for overlap with irrelevant context. To enable GEAR on natural-language data, we develop an automated pipeline that constructs evidence-annotated training data from arbitrary documents. We validate GEAR across multiple model scales and benchmarks, showing consistent improvements of up to +4.6 average points over standard RL with accuracy-based rewards, with larger gains at longer contexts, while also reducing repetitive copying and thinking length. Our findings suggest that, even as long-context evaluation shifts from simple retrieval toward complex reasoning, accurate grounding in relevant evidence remains an indispensable capability with substantial room for improvement.

Chinese Translation

生成逐步推理痕迹的大型语言模型在复杂任务上取得了优异的表现，将其扩展到长上下文设置已成为一个重要的前沿领域。然而，我们在这一领域中识别出一种关键的失效模式： extit{重复复制}，即模型在推理痕迹中大量复制输入文本，而不是有效地解决问题。我们展示了这种行为在前沿的长上下文大型语言模型中普遍存在，并且随着上下文长度的增加而加剧。通过将每个提示分离为与任务相关的关键证据和无关的干扰上下文，我们进一步表明，根本原因在于基础不足：模型无差别地从提示中复制，而那些未能专注于关键证据的模型更可能给出错误答案。基于这一诊断，我们提出了GEAR（Grounding Evidence-Aware Reward），这是一种奖励塑形方法，通过与关键证据的重叠提供基础奖励，并对与无关上下文的重叠施加干扰惩罚，从而增强准确性信号。为了在自然语言数据上实现GEAR，我们开发了一条自动化管道，从任意文档构建带有证据注释的训练数据。我们在多个模型规模和基准测试中验证了GEAR，显示出与基于准确性的标准强化学习相比，平均提高了高达+4.6分的稳定改进，且在更长的上下文中获得更大的增益，同时减少了重复复制和思考长度。我们的研究结果表明，即使长上下文评估从简单检索转向复杂推理，准确地扎根于相关证据仍然是一项不可或缺的能力，并且有很大的改进空间。

View on arXiv Download PDF AI Translation

arXiv Papers

FARO: Feasibility-Aware Robot Motion Optimization

Towards Torque-Driven Reinforcement Learning for Quadruped Locomotion

The Open Ant: A Robot Platform for Reinforcement Learning Research

Beyond Fixed Goal Delivery: Online POMDP Planning for Target Interception in Crowds

DASH Robot: Minimalistic Design and Optimal Aerial-Terrestrial Locomotion via Contact-Implicit Control

Two-Stage Extrinsic Calibration of a Static Line-Scanning Lidar with a Rotary Platform

STeP: Signal Temporal Logic for Precise Specifications for Action Generation with Vision Language Models

Bridging the Sim-to-Real Gap under Real-Time Constraints in Autonomous Racing

Intelligent Multi-UAV Navigation in ITNTNs: A Hierarchical LLM Approach

On the Limits of Sampling-Based Reachability: Geometry, Dynamics, and Sample Complexity

End-to-end Conditional Diffusion for Realistic and Controllable Visual Traffic Scenario Generation

Fabric Pneumatic Artificial Muscles Based on the Drawstring Principle

MVP-Tac: A Miniaturized Dual-Modal Vision and Photoelastic Tactile Sensor for Robot-Assisted Minimally Invasive Surgery

How defensive driving enhances driving safety: A driving simulator study on drivers' defensive driving behaviors

RoboInter1.5: A Holistic Intermediate Representation Suite for Embodied World Modeling and Robotic Manipulation

Confidence-Gated Vision-Only Heading Alignment for UAV-UGV Cooperative Systems

Correct-by-Construction Behavior Tree Synthesis from Signal Temporal Logic Specifications with Application to Robotic Missions

Motion Primitive Discovery in a Humanoid Robot via Self-Organising Maps for Phase Recognition

Koopman DCM: Unstable Eigenfunctions as Data-driven Representations for Legged Balancing

Beyond Transformers: Linear Attention Policy for Open-Vocabulary Object Goal Navigation

WorldScape Policy 2.0: Empowering Steerable World Action Modeling with Reasoning-Augmented Memory

Pose-Parameterized Motion Planning and CBF-QP Self-Collision Filtering for a Long-Reach Drilling Boom

NGPS: GPS-Denied Aerial Geo-Localization and 2.5D Reconstruction via Deep Satellite Image Matching and Multi-Rate Sensor Fusion

The Twist Decomposition of Serial Robots Under Lower-Mobility Tasks

Design and stability analysis of an underactuated hand with passively rotating fingers

Bayesian Retraction Optimization for Tissue Attachment Mapping in Surgical Dissection

Agentic Real2Sim: Physics-based World Modeling with Vision-Language Agents

Cognitive Dual-Process Planning for Autonomous Driving with Structured Scene Knowledge and Verifiable Reasoning-Action Consistency

Computing on the Fly: Navigating a Vision for the Future of Drone Computing

Eversion-based robots can enable safe access,steering and endoscopic imaging within the spinal subarachnoid space

Stochastic Multi-Objective Kinodynamic Planning Against Adversaries

From Distances to Trajectories: Real-Time Signed Distance Function Mapping and Distance-Accelerated Motion Planning for UAVs

Hazard or Anomaly? Evaluating VLMs for Understanding Dangers and Discrepancies

From Pixel to Prognosis: Convolutional and GLCM Feature Fusion for Automated Four-Class Cataract Severity Classification

Surprise Forcing: What to Remember, When to Skip in Long Video Generation

ECoNGS: Efficient Compressive Neural Gaussian Splats for Volume Visualization

Style over Substance: A Shortcut Audit of Emotion-Description Preference Evaluation

DuSPiT: Dual-Branch Sub-Patch Pixel Diffusion Transformer

AniGS: Bridging Rendering and Diffusion Prior for 3D Scene Animation

Recti-Q: Feature-Space Rectification for Out-of-Distribution-Robust Quantized Perception in Edge Robotics

Physics Closure Matters for Machine Olfaction: A Maxwell--Stefan Graph Solver for Identifiable Dynamic Gas Unmixing

Text-conditioned Segmentation for Tomato Phenotyping via Procedural Synthetic Data

Attention Without Grounding: Causal Evaluation of Visual Explanations in Medical VLMs

Norm or Direction? Decoding Vision Mambas for High-Resolution Vision

Seeing Before Generating: Object Perception Enhances Single-View 3D Reconstruction

Deep Learning Estimation of Sex, Age, Height, and Weight from CT-derived Digitally Reconstructed Radiographs

Fluid-SDF: Ultra-Lightweight and Editable Implicit Shape Representation via Differentiable Primitives

DeforM: Reasoning-Guided Physics-Aware Video Generation via Spatial-Temporal Masking

MissingBench-Verified: Probing Vision-Language Models' Inability to Detect Missing Object Parts

Cross-Dataset Generalization in Breast MRI Tumor Classification via Class-Wise Dataset Mixing

Dual-Edged Homogeneous-Modality Similarity: Towards Visible-Infrared Modality-Incomplete Person Re-Identification with Modality Adaptive Matching

Attributes Should Come from Images, Not Class Names: Distribution-Conditioned Attribute Selection for Vision-Language Models

Decoupled Pipeline with Proposal Reranking and Score Fusion for Positive-Unlabeled Marine Species Detection

Generative World Renderer at the Speed of Play

Continual Video-MLLM Adaptation over Evolving Domains

SkyEV: RGB-Event UAV detection and tracking dataset and baseline

Weakly Supervised Pathology-Informed Representation Learning for PET-Based Content Retrieval of Intra-Tumour Heterogeneity

Posterior Samplings are Missing Modalities Generators for Medical Image Translation

Bounding Boxes to Improve Small Language Model Performance on Vision-Based Grading Tasks

Cross-Modal UAV Object Tracking: State-Aware Representation Learning and A Unified Benchmark

Privileged Lesion-Context Relational Distillation for Mask-Free Skin Lesion Classification

CGMap: A Geospatially Aware Deep Learning Framework for Crop Gap Mapping Using UAV

Image Editing Models are Numerical Solvers

Moving Alphabet: A Controlled Study of Training Data for Text-to-Video Generation

STS-NET: Spatio-Temporal Stress Network for Self-Supervised Crop Stress Detection using Satellite Image Time Series

UVFaceFusion: Fast Multi-view Topologically Consistent Face Reconstruction in the Wild via UV-space Neural Fusion

ZeroSplat: Generalized Referring Segmentation in 3D Gaussian Splatting

In-Context Learning for Wound Classification with Small Multimodal Language Models

Open-Vocabulary Gaze Object Prediction: Benchmark and Method

OPD-IAD: From Language Judgment to Industrial Anomaly Detection via On-Policy Self-Distillation

Think Sparse, Predict Dense: Continuous Thought Machines for Image Super-Resolution

Reliability-Aware 3D Geometric Injection for Universal Person Re-identification

Wave2Body: Rethinking mmWave Human Pose Estimation as Radar-to-Body Token Translation

Local Label-Informed Feature Transfer for Generating Ground-Truth Medical Images: A Comparison of GAN- and Diffusion-Based Approaches

SynGallery: A Synthetic Gallery of Real Paintings for Instance-Level Artwork Recognition

TAP-RAG: Task-Aware Policy Control for Long-Document Multimodal Question Answering

Learning Explicit Physical Parameter Control and Benchmarking for Video Generation

Dual Adversarial Fine-tuning for Enhancing Robustness of Large Vision Language Model

DobicVLM: Aligning Chest X-Ray Report Generation with Clinically-Grounded Programmatic Rewards via Group Relative Policy Optimization