cs.RO / 1 / 2607.12050
EFLUX: Elastic Multi-Robot Formation Navigation and Adaptation with Agentic LLMs
EFLUX：基于代理大型语言模型的弹性多机器人编队导航与适应
Abstract
Multi-robot teams operating in confined or cluttered environments must adapt both their formation geometry and group topology to navigate through complex obstacles. This adaptation requires two complementary behaviors: deformation, where the team continuously reshapes its geometry while remaining connected, and reconfiguration, where robots split into subgroups or merge back into a single formation. Existing methods often model these behaviors independently, connect them through handcrafted rules, or lack explicit geometric criteria for determining when each behavior should be invoked. However, challenging environments may require online changes in formation shape, connectivity, and effective team composition, making decoupled or rule-based approaches prone to suboptimal trajectories and deadlock. We propose EFLUX, a geometry-grounded LLM agentic framework for automatic and elastic multi-robot formation navigation. EFLUX extracts a structured scene representation and uses an LLM to reason jointly over both deformation actions, such as scaling and shearing, and reconfiguration actions, such as splitting and merging. These strategies are then translated into executable per-robot waypoints through a closed-loop generation, verification, and correction pipeline. Simulation and hardware experiments show that EFLUX enables safe, continuous, and elastic formation navigation in constrained environments, reducing deadlock and navigation failures compared with baselines while maintaining coherent multi-robot coordination.
Chinese Translation
在封闭或杂乱环境中操作的多机器人团队必须调整其编队几何形状和群体拓扑，以便穿越复杂障碍。这种适应需要两种互补行为：变形，即团队在保持连接的同时不断重塑其几何形状；重构，即机器人分成子组或重新合并为一个单一编队。现有方法通常独立建模这些行为，通过手工规则将其连接，或缺乏明确的几何标准来确定何时应调用每种行为。然而，具有挑战性的环境可能需要在线改变编队形状、连接性和有效团队组成，使得解耦或基于规则的方法容易导致次优轨迹和死锁。我们提出了EFLUX，一个基于几何的LLM代理框架，用于自动和弹性的多机器人编队导航。EFLUX提取结构化场景表示，并使用LLM共同推理变形动作（如缩放和剪切）和重构动作（如分裂和合并）。这些策略随后通过闭环生成、验证和纠正管道转化为可执行的每个机器人的航点。模拟和硬件实验表明，EFLUX在受限环境中实现了安全、连续和弹性的编队导航，与基线相比减少了死锁和导航失败，同时保持了多机器人的协调一致。
cs.RO / 2 / 2607.12065
Enabling 24-hour Agricultural Robotics: Unsupervised Day-to-Night Cross-Modal Image Translation for Nighttime Visual Navigation
实现24小时农业机器人：无监督昼夜跨模态图像翻译用于夜间视觉导航
Abstract
While visual navigation has been extensively studied in agricultural robotics, most existing systems assume daytime conditions. In fact, deploying autonomous robots at night offers significant advantages, including 24-hour crop and soil monitoring, fruit harvesting, and nocturnal pest detection. Modern vision-based systems, however, rely heavily on large-scale well-annotated image datasets, which remains challenging to obtain for nighttime operation scenarios. To address this, we propose an unsupervised image translation framework that converts daytime plant-row RGB images into near-infrared (NIR) nighttime counterparts without requiring pixel-to-pixel supervision. This enables the direct reuse of daytime semantic labels for training nighttime perception models. In particular, by incorporating a pre-trained Contrastive Language-Image Pre-training (CLIP) model, the proposed framework is designed to preserve semantic consistency during day-to-night translation. Additionally, a visibility mask is introduced to account for the limited effective range of NIR illumination in nighttime scenes. We conduct comparative evaluations with state-of-the-art image translation baselines and demonstrate higher image qualities, as supported by improved performance in downstream semantic segmentation for nighttime visual navigation. For evaluation, we utilize AgriNight--a novel dataset comprising 428 daytime and 549 nighttime images collected using night-vision-equipped mobile robots in agricultural fields and manually annotated with pixel-wise semantic labels--and introduce it as the first benchmark for nighttime agricultural visual navigation. We also perform real-time autonomous navigation experiments with a physical robot operating at night. The data and code are available at: https://github.com/mamorobel/AgriNight.
Chinese Translation
尽管视觉导航在农业机器人领域得到了广泛研究，但大多数现有系统假设是在白天条件下运行。实际上，在夜间部署自主机器人具有显著优势，包括24小时作物和土壤监测、果实采摘以及夜间害虫检测。然而，现代基于视觉的系统在很大程度上依赖于大规模的高质量标注图像数据集，而在夜间操作场景中获取这样的数据集仍然具有挑战性。为了解决这个问题，我们提出了一种无监督图像翻译框架，该框架能够将白天植物行的RGB图像转换为近红外（NIR）夜间图像，而无需逐像素的监督。这使得可以直接重用白天的语义标签来训练夜间感知模型。特别地，通过引入预训练的对比语言-图像预训练（CLIP）模型，所提框架旨在保持昼夜翻译过程中的语义一致性。此外，引入了可见性掩码，以考虑夜间场景中NIR照明的有效范围有限。我们与最先进的图像翻译基线进行了比较评估，结果显示图像质量更高，并且在夜间视觉导航的下游语义分割任务中表现得到了改善。为了评估，我们利用AgriNight——一个新颖的数据集，包含428张白天图像和549张夜间图像，这些图像是通过配备夜视设备的移动机器人在农业领域收集的，并手动标注了逐像素的语义标签——并将其作为夜间农业视觉导航的第一个基准。此外，我们还进行了夜间物理机器人实时自主导航实验。数据和代码可在以下网址获取：https://github.com/mamorobel/AgriNight。
cs.RO / 3 / 2607.12105
Robust In-Hand Manipulation via Priors in Reinforcement Learning and Mechanical Design
通过强化学习和机械设计中的先验实现稳健的手内操作
Abstract
In-hand manipulation without external sensing is challenging due to uncertainties from finger-object contacts and disturbances by gravity. While reinforcement learning has shown promise in learning complex finger gaiting, existing approaches do not prioritize maintaining well-conditioned grasps for sustained manipulation. We introduce two complementary physics priors for robust in-hand rolling: a global grasp-quality prior derived from classical grasp analysis and a local contact-geometry prior based on fingertip curvature. The grasp-quality prior is used as a dense reward-shaping term that encourages well-distributed contacts with improved worst-case wrench resistance. The contact-geometry prior is expressed in the fingertip geometry that mechanically shapes the contact interface toward task-aligned rolling while reducing off-axis drift. We evaluate the effect of these priors on learning in-hand rolling manipulation for a multifingered robotic hand manipulating three different objects at four palm orientations. Results show significant improvement in rotation efficiency, grasp stability, and disturbance rejection, suggesting that physics priors embedded in both learning and fingertip morphology improve task robustness and sim-to-real transfer. An overview video can be found at https://youtu.be/pdd1wHxQnJM?si=dM-U5kiiPTYsk3Pk.
Chinese Translation
由于手指与物体接触的不确定性以及重力的干扰，手内操作在没有外部传感的情况下具有挑战性。尽管强化学习在学习复杂的手指运动方面展现出潜力，但现有方法并未优先考虑维持良好条件的抓握以实现持续操作。我们引入了两种互补的物理先验，以实现稳健的手内滚动：一种基于经典抓握分析的全局抓握质量先验，以及一种基于指尖曲率的局部接触几何先验。抓握质量先验被用作密集的奖励塑形项，鼓励良好分布的接触，从而提高最坏情况下的扭矩抗拒力。接触几何先验则体现在指尖几何形状上，机械地塑造接触界面以实现与任务对齐的滚动，同时减少偏轴漂移。我们评估了这些先验对多指机器人手在四种手掌朝向下操作三种不同物体的手内滚动操作学习的影响。结果表明，旋转效率、抓握稳定性和干扰拒绝能力显著提高，表明嵌入学习和指尖形态中的物理先验改善了任务的稳健性和仿真到现实的转移。概述视频可以在 https://youtu.be/pdd1wHxQnJM?si=dM-U5kiiPTYsk3Pk 找到。
cs.RO / 4 / 2607.12114
GaitSpan: Growing Humanoid Locomotion from Walking to Running
GaitSpan：从行走到奔跑的类人运动发展
Abstract
A humanoid that can walk should not relearn locomotion from scratch to jog or run. Yet current approaches often obtain gait diversity by prescribing gait schedules, imitating motion clips, training experts to switch between or distilling skills into one policy. These strategies can produce impressive behaviors, but offer limited flexibility across continuous speed commands, terrains, and morphologies. We study skill growth with GaitSpan, a framework that expands a pretrained, basic walking policy into faster locomotion. It treats walking as a seed skill: reusable motor structure for balance, support, body coordination, and contact transition that can be regenerated at new rhythms, extended into longer/higher strides, and corrected by residual adaptation. This expansion has three aspects: 1) rhythm generation, which modulates the frozen walking policy with multiple internal clocks and learns command-conditioned combinations of the resulting canonical actions; 2) stride shaping, which rewards dynamic locomotion patterns appropriate for higher commanded speeds using a physically grounded objective inspired by spring-loaded inverted pendulum dynamics; and 3) residual adaptation, which captures motion details not accounted for by rhythm generation or stride shaping. GaitSpan is the first to deliver a single command-conditioned humanoid policy that spans walking, jogging, and running-like regimes covering a continuous speed range, transfers across morphologies, and deploys zero-shot on unseen sim-to-sim, and real-world terrains. Compared with baselines either trained with multi-experts or imitation from humans, it learns faster and achieves stronger gait performance.
Chinese Translation
一个能够行走的类人机器人不应该从头开始重新学习运动以进行慢跑或奔跑。然而，目前的方法通常通过规定步态时间表、模仿运动片段、训练专家在不同步态之间切换或将技能提炼为一个策略来获得步态多样性。这些策略可以产生令人印象深刻的行为，但在连续速度指令、地形和形态方面提供的灵活性有限。我们研究了技能增长，提出了GaitSpan，一个将预训练的基本行走策略扩展到更快运动的框架。它将行走视为一种种子技能：可重用的运动结构，用于平衡、支撑、身体协调和接触过渡，可以在新的节奏下重新生成，扩展为更长/更高的步幅，并通过残余适应进行修正。这种扩展有三个方面：1）节奏生成，通过多个内部时钟调节冻结的行走策略，并学习命令条件下的结果典范动作的组合；2）步幅塑造，使用受弹簧加载倒摆动力学启发的物理基础目标，奖励适合于更高命令速度的动态运动模式；3）残余适应，捕捉节奏生成或步幅塑造未考虑的运动细节。GaitSpan首次提供一个单一的命令条件类人政策，涵盖行走、慢跑和奔跑类的运动状态，覆盖连续速度范围，能够跨形态转移，并在未见的模拟到模拟和现实世界地形上进行零样本部署。与经过多专家训练或模仿人类的基线相比，它学习更快，并实现更强的步态表现。
cs.RO / 5 / 2607.12115
A Behavioral State Vocabulary in Sony ERS-111 R-CODE
索尼 ERS-111 R-CODE 中的行为状态词汇
Abstract
This paper presents a corpus-level analysis of generated behavior diagrams derived from Sony's R-CODE sample distribution for the ERS-111 AIBO. Rather than reading each script in isolation, the study compares named states across the corpus to identify the recurring control vocabulary that structures the sample set. The resulting aggregate shows that many superficially different routines are built from a compact embodied grammar centered on initialization, sensing, iterative action, synchronization, and recovery. In addition to historical analysis, the paper argues that this form of state-based abstraction is useful as an intermediate representation for constructing new encapsulated behavior routines, especially on constrained native robotic systems where deterministic control, direct hardware access, and modular behavioral composition remain important.
Chinese Translation
本文对从索尼 R-CODE 样本分布中生成的行为图进行了语料库级别的分析，样本针对 ERS-111 AIBO。研究并非孤立地阅读每个脚本，而是比较语料库中的命名状态，以识别构成样本集的重复控制词汇。结果显示，许多表面上不同的例程是基于一个紧凑的具身语法构建的，中心内容包括初始化、感知、迭代动作、同步和恢复。除了历史分析外，本文还论证了这种基于状态的抽象形式作为构建新的封装行为例程的中间表示是有用的，特别是在约束的本地机器人系统上，在这些系统中，确定性控制、直接硬件访问和模块化行为组合仍然非常重要。
cs.RO / 6 / 2607.12130
More than a Manipulator: Planning Propellant-Free Attitude Maneuvers for Free-Floating Spacecraft
不仅仅是一个操纵器：为自由漂浮航天器规划无推进剂姿态机动
Abstract
Spacecraft attitude control is traditionally achieved using momentum exchange devices or propellant-consuming thrusters. Meanwhile, a growing number of missions require robotic manipulators, which are typically treated as disturbance sources to be rejected rather than as actuators for spacecraft reorientation. This work investigates the use of manipulator motions for propellant-free attitude control by formulating a trajectory optimization problem with critical joint and collision avoidance constraints. Using an interior point solver for the resulting nonlinear program, complex slew and detumble trajectories are demonstrated for a range of spacecraft-manipulator systems with varying kinematic complexity and mass properties. The achievable control authority is compared directly with that of reaction wheel arrays via momentum and torque envelopes, demonstrating the potential for manipulators to serve as redundant or even primary attitude control systems. This work provides a framework for using manipulators as multipurpose attitude control actuators, with particularly promising applications in in-space assembly and manufacturing when grasping payloads with high relative mass fractions.
Chinese Translation
航天器姿态控制传统上是通过动量交换装置或消耗推进剂的推进器来实现的。与此同时，越来越多的任务需要机器人操纵器，这些操纵器通常被视为需要排除的干扰源，而不是航天器重新定向的执行器。本研究探讨了利用操纵器运动进行无推进剂姿态控制的方法，通过制定一个具有关键关节和碰撞避免约束的轨迹优化问题来实现。使用内点求解器解决得到的非线性程序，展示了针对具有不同运动学复杂性和质量特性的航天器-操纵器系统的复杂旋转和去颠簸轨迹。通过动量和扭矩包络线，直接比较了可实现的控制权与反应轮阵列的控制权，展示了操纵器作为冗余甚至主要姿态控制系统的潜力。本研究提供了一个框架，用于将操纵器作为多功能姿态控制执行器，特别是在抓取具有高相对质量分数的有效载荷时，在太空组装和制造中具有特别有前景的应用。
cs.RO / 7 / 2607.12181
Analysis of Mutual and Referential Human and Robot Gazes in a Collaborative Word Association Game
人类与机器人在协作词汇联想游戏中的相互与指向性注视分析
Abstract
Robot gaze is a major component of human-robot dialogue coordination. Most studies of gaze in human-robot dialogue focus on face-to-face social conversations, but little is known about gaze in demanding task-focused interactions. In this paper, we investigate how the gaze of a robot game partner affects human visual attention and if humans tend to direct confirmation-seeking gazes towards the robot. In our study, we let participants play a collaborative word association game with a NAO robot acting as an embodied, LLM-driven conversational partner. Our experiments are conducted under two conditions, which implement mutual and referential gazes of the robot respectively. We record participants' gaze using eye tracking glasses and analyze the interactions using gaze coordinates, speech segments, key events and areas of interests. We find that robot gaze orientation does not affect the time to first fixation on words the robot proposed. We also find that participants gaze more often at the robot when their dialogue line contains confirmation requests, compared to when it does not. Our results indicate (likely also due to the cognitively demanding nature of the game) that the verbal aspect of this task overshadows the effects of referential robot gaze. These findings offer valuable insights for designing and validating robot gaze and turn-taking behavior in collaborative tasks which require coordination and efficient communication.
Chinese Translation
机器人的注视是人机对话协调的重要组成部分。大多数关于人机对话中注视的研究集中在面对面的社交对话上，但对于在高要求的任务导向互动中注视的研究知之甚少。本文探讨了机器人游戏伙伴的注视如何影响人类的视觉注意力，以及人类是否倾向于将寻求确认的注视指向机器人。在我们的研究中，我们让参与者与作为具身的、基于大型语言模型（LLM）驱动的对话伙伴的NAO机器人一起进行协作词汇联想游戏。我们的实验在两种条件下进行，分别实现机器人的相互注视和指向性注视。我们使用眼动追踪眼镜记录参与者的注视，并通过注视坐标、语音片段、关键事件和兴趣区域分析互动。我们发现，机器人的注视方向并不影响参与者首次注视机器人提出的词语的时间。我们还发现，当参与者的对话内容包含确认请求时，他们更频繁地注视机器人，而在没有确认请求时则较少。我们的结果表明（可能也由于游戏的认知要求较高），该任务的语言方面掩盖了指向性机器人注视的影响。这些发现为设计和验证在需要协调和高效沟通的协作任务中机器人的注视和轮流发言行为提供了宝贵的见解。
cs.RO / 8 / 2607.12220
Contract-Grounded Behavior Tree Synthesis via Coding Agents
基于合同的行为树合成通过编码代理
Abstract
Synthesizing deployable robot behavior trees (BTs) from natural language (NL) requires grounding to ensure every generated BT references only skills a robot can actually execute. Existing LLM-based BT synthesis approaches often place this grounding responsibility on the prompt author. This makes deployment brittle when the author does not know which skills the robot can execute, how those skills are parameterized, or how the robot runtime software constrains valid BT structure. This paper proposes a contract-grounded BT synthesis architecture in which a coding agent queries a robot-side Model Context Protocol (MCP) server to retrieve an explicit contract consisting of a skill library, permitted BT operators, and optional BT composition templates, before synthesizing a BT for validation and execution. In our framework, non-expert operators issue NL commands without knowledge of robot implementation details, while a robot runtime validation gate enforces correctness before execution. We evaluate two LLMs, a closed model (Sonnet 4.6) and a smaller open-source model (Gemma4:31b), across 110 simulated tasks in PyRoboSim and 14 tasks on a physical Husarion Panther robot. Results show that contract grounding enables near-perfect BT validation and high task success, that BT composition templates substantially recover success on reactive control-flow tasks for the smaller model, and that the architecture transfers to physical hardware running a Nav2 stack opaque to both operator and agent.
Chinese Translation
从自然语言（NL）合成可部署的机器人行为树（BT）需要进行基础性处理，以确保每个生成的行为树仅引用机器人实际能够执行的技能。现有的基于大型语言模型（LLM）的行为树合成方法通常将这一基础性处理的责任放在提示作者身上。这使得在作者不知道机器人能够执行哪些技能、这些技能如何参数化或机器人运行时软件如何限制有效行为树结构的情况下，部署变得脆弱。本文提出了一种基于合同的行为树合成架构，其中编码代理查询机器人端的模型上下文协议（Model Context Protocol, MCP）服务器，以检索由技能库、允许的行为树操作符和可选的行为树组合模板组成的明确合同，然后合成一个行为树以进行验证和执行。在我们的框架中，非专家操作员可以在不了解机器人实现细节的情况下发出自然语言命令，而机器人运行时验证门在执行前强制执行正确性。我们在PyRoboSim中评估了两个大型语言模型，一个封闭模型（Sonnet 4.6）和一个较小的开源模型（Gemma4:31b），在110个模拟任务和14个物理Husarion Panther机器人任务上进行测试。结果表明，合同基础处理使得行为树验证接近完美，任务成功率高，行为树组合模板显著提高了较小模型在反应控制流任务上的成功率，并且该架构可以迁移到对操作员和代理均不透明的运行Nav2栈的物理硬件上。
cs.RO / 9 / 2607.12265
DiffRadar: Differentiable Physics-Aware Radar SLAM with Gaussian Fields
DiffRadar：具有高斯场的可微物理感知雷达SLAM
Abstract
Radar sensing is increasingly used in mobile systems because it operates reliably under poor lighting, adverse weather, and privacy-sensitive settings where cameras and LiDAR often fail. However, most existing radar SLAM systems estimate motion through scan matching on discretized radar heatmaps, which breaks geometric continuity and fails to capture key radar sensing properties, often leading to unstable pose estimation and degraded mapping in regenerate or dynamically changing environments. We present DiffRadar, a real-time radar SLAM system that models radar observations as a differentiable, physics-aware Gaussian field rather than discrete scans. DiffRadar represents the scene as anisotropic Gaussian primitives and renders radar measurements in range-azimuth and Doppler-azimuth spaces through a differentiable radar forward model, enabling joint optimization of robot pose and scene structure directly from radar measurements. We implement DiffRadar on commodity FMCW radar hardware and evaluate it on both the public Radarize benchmark and a controlled stress-test suite that targets common radar SLAM failure modes, including corridor degeneracy, motion regime transitions, dynamic clutter, and long-horizon loop closures. DiffRadar achieves substantial reductions in trajectory error on the benchmark, with especially large gains under feature-poor corridor motion, while more than doubling map consistency and maintaining real-time performance at 70 FPS. These results show that modeling radar observations directly in the signal domain enables substantially more robust and consistent radar-only SLAM for mobile platforms.
Chinese Translation
雷达感知在移动系统中的应用日益增多，因为它在光线不足、恶劣天气和隐私敏感的环境中能够可靠工作，而这些环境中相机和激光雷达（LiDAR）往往无法正常工作。然而，大多数现有的雷达SLAM系统通过对离散化的雷达热图进行扫描匹配来估计运动，这破坏了几何连续性，未能捕捉关键的雷达感知特性，常常导致姿态估计不稳定和在再生或动态变化环境中的映射质量下降。我们提出了DiffRadar，一个实时雷达SLAM系统，它将雷达观测建模为可微分的、物理感知的高斯场，而不是离散扫描。DiffRadar将场景表示为各向异性的高斯原语，并通过可微分的雷达前向模型在距离-方位和多普勒-方位空间中渲染雷达测量，从而实现机器人姿态和场景结构的联合优化，直接基于雷达测量。我们在商用FMCW雷达硬件上实现了DiffRadar，并在公共Radarize基准测试和一个针对常见雷达SLAM故障模式的受控压力测试套件上进行了评估，包括走廊退化、运动模式转变、动态杂波和长时间环路闭合。DiffRadar在基准测试中显著降低了轨迹误差，尤其是在特征稀缺的走廊运动下取得了较大的改进，同时将地图一致性提高了两倍以上，并在70帧每秒的实时性能下保持运行。这些结果表明，在信号域中直接建模雷达观测能够显著增强移动平台的雷达单独SLAM的鲁棒性和一致性。
cs.RO / 10 / 2607.12275
Flatness-Preserving Residual Learning for Real-Time Tight Quadrotor Formation Flight
保持平坦性的残差学习用于实时紧凑四旋翼编队飞行
Abstract
Quadrotors flying in tight formations are severely affected by turbulent aerodynamic interactions, such as downwash, that can cause catastrophic collisions if left unmodeled. To compensate for these effects, we propose a physics-informed residual dynamics learning framework that captures complex aerodynamic interactions while ensuring the joint multi-quadrotor system remains differentially flat. We leverage this preserved flatness to design a computationally efficient feedback linearization controller that is easily tunable with linear control techniques and cancels aerodynamic disturbances via feedforward compensation. Hardware experiments demonstrate our framework reduces average tracking errors by 31% compared to nominal baselines. Crucially, our lightweight approach matches the tracking performance of state-of-the-art nonlinear model predictive control (NMPC) while requiring an order of magnitude less computation. We are the first to show that stable, tight formation flight can be achieved with under 30 seconds of training data and a 5ms loop rate, unlocking high-fidelity aerodynamic compensation for compute-constrained flight stacks.
Chinese Translation
在紧凑编队飞行中，四旋翼受到湍流气动相互作用的严重影响，例如下洗气流，如果不加以建模，可能导致灾难性的碰撞。为了补偿这些影响，我们提出了一种物理信息残差动态学习框架，该框架能够捕捉复杂的气动相互作用，同时确保多四旋翼系统保持微分平坦性。我们利用这种保持的平坦性设计了一种计算效率高的反馈线性化控制器，该控制器易于通过线性控制技术进行调节，并通过前馈补偿抵消气动干扰。硬件实验表明，与名义基线相比，我们的框架将平均跟踪误差降低了31%。至关重要的是，我们的轻量化方法在计算需求上比最先进的非线性模型预测控制（NMPC）少了一个数量级，同时匹配了其跟踪性能。我们首次展示了在不到30秒的训练数据和5毫秒的循环速率下，可以实现稳定的紧凑编队飞行，从而为计算受限的飞行堆栈解锁高保真度的气动补偿。
cs.RO / 11 / 2607.12287
Reducing Temporal Redundancy for Efficient Vision-Language-Action Inference
减少时间冗余以实现高效的视觉-语言-动作推理
Abstract
Vision-Language-Action (VLA) models exhibit strong generalization for robotic manipulation, yet their high inference latency limits real time deployment. We identify two primary sources of temporal redundancy in existing VLA pipelines: repeated visual encoding of highly similar consecutive frames and multi step iterative sampling in diffusion based policies. To address this, we propose a system level acceleration strategy that reduces computation in both perception and action generation. On the perception side, we incrementally update only tokens corresponding to dynamic scene regions instead of re-encoding entire frames. On the policy side, we compress diffusion sampling into a compact 2-step schedule through efficiency oriented training while preserving action precision. Experiments on Libero, RobotWin, and Real Robot Platforms demonstrate over 2 times speedup while maintaining high performance, achieving up to 98% success rate on general manipulation benchmarks. Our codes will be released on Github.
Chinese Translation
视觉-语言-动作 (VLA) 模型在机器人操作中表现出强大的泛化能力，但其高推理延迟限制了实时部署。我们识别出现有 VLA 流程中的两个主要时间冗余来源：对高度相似的连续帧进行重复视觉编码，以及在基于扩散的策略中进行多步迭代采样。为了解决这个问题，我们提出了一种系统级加速策略，减少感知和动作生成中的计算。在感知方面，我们仅增量更新与动态场景区域对应的标记，而不是重新编码整个帧。在策略方面，我们通过以效率为导向的训练将扩散采样压缩为紧凑的两步调度，同时保持动作精度。在 Libero、RobotWin 和真实机器人平台上的实验表明，在保持高性能的同时实现了超过两倍的加速，在一般操作基准上成功率高达 98%。我们的代码将发布在 Github 上。
cs.RO / 12 / 2607.12356
VistaVLA: Geometry- and Semantic-Aware 3D Gaussian-Grounded VLA for Robotic Manipulation
VistaVLA：几何和语义感知的基于3D高斯的视觉-语言-动作模型用于机器人操作
Abstract
Vision-Language-Action (VLA) models have emerged as a powerful end-to-end paradigm for robotic manipulation by mapping language instructions and 2D visual inputs directly to actions. However, these models lack an explicit, scene-level 3D representation, limiting their ability to reason over spatial layouts and geometric constraints. While recent efforts incorporate explicit 3D cues, such as depth maps or point clouds, to improve geometric awareness, they primarily capture low-level structures and lack high-level semantic grounding in 3D space. In human cognition, interaction with the physical world relies on a 3D semantic cognitive map - an internal mental model that integrates spatial layouts with semantic context to enable persistent, viewpoint-invariant reasoning. In light of this, we present VistaVLA, a novel two-stage framework that constructs a geometry- and semantics-aware 3D cognitive representation from 3D Gaussian primitives and grounds it as compact context tokens for VLA policy learning. Specifically, VistaVLA lifts multi-view vision-language features into 3D Gaussian primitives, forming geometry-anchored semantic tokens that align view-consistent spatial grounding with 2D visual feature spaces. To make this 3D representation computationally tractable for effective VLA control, we introduce Merge-then-Query (MtQ), a token summarization mechanism. MtQ compresses dense Gaussian primitives into a highly compact set of spatially informative tokens, achieving a 99% token reduction while preserving action-relevant 3D layouts and semantic context. Extensive evaluations in both simulated and real-world environments demonstrate the effectiveness of VistaVLA. Notably, in real-world scenarios, VistaVLA improves success rates by 22.8% across seven real-world tasks and by 30.0% over the VLA-Adapter baseline on challenging out-of-distribution tasks.
Chinese Translation
视觉-语言-动作（VLA）模型已成为一种强大的端到端范式，通过将语言指令和2D视觉输入直接映射到动作，从而实现机器人操作。然而，这些模型缺乏明确的场景级3D表示，限制了它们对空间布局和几何约束的推理能力。尽管最近的努力通过引入深度图或点云等显式3D线索来提高几何感知，但它们主要捕捉低级结构，缺乏在3D空间中的高级语义基础。在人类认知中，与物理世界的互动依赖于3D语义认知地图——一种将空间布局与语义上下文整合的内部心理模型，从而实现持久的、视角不变的推理。基于此，我们提出了VistaVLA，一种新颖的两阶段框架，从3D高斯原语构建几何和语义感知的3D认知表示，并将其作为紧凑的上下文令牌用于VLA策略学习。具体而言，VistaVLA将多视角视觉-语言特征提升为3D高斯原语，形成几何锚定的语义令牌，使视图一致的空间基础与2D视觉特征空间对齐。为了使这一3D表示在有效的VLA控制中具有计算可行性，我们引入了合并-再查询（Merge-then-Query, MtQ）机制。MtQ将密集的高斯原语压缩为一组高度紧凑的空间信息令牌，实现了99%的令牌减少，同时保留了与动作相关的3D布局和语义上下文。在模拟和真实环境中的广泛评估证明了VistaVLA的有效性。值得注意的是，在真实场景中，VistaVLA在七个真实任务中的成功率提高了22.8%，在具有挑战性的分布外任务中比VLA-Adapter基线提高了30.0%。
cs.RO / 13 / 2607.12370
StratMamba: Strategic and Reactive Stream Partitioning for Path-Efficient LiDAR-Based Obstacle Avoidance
StratMamba：用于路径高效的基于LiDAR的障碍物规避的战略性和反应性流分区
Abstract
This paper proposes StratMamba, a dual-stream Mamba-based temporal modeling architecture, to more efficiently capture long-horizon temporal dependencies required for robot navigation in complex and obstacle-rich environments. StratMamba leverages a combination of fast-decay and slow-decay memory architectures, where the fast-decay component processes high-frequency LiDAR data for reactive obstacle avoidance, while the slow-decay component maintains longer-horizon goal information for strategic planning. We perform extensive evaluations of different obstacle avoidance scenarios in IsaacLab and Gazebo, while also validating successful sim-to-real deployment on a Unitree GO1 quadruped robot navigating in the presence of static/dynamic obstacles. Comparisons with other temporal RL baselines, such as LSTM, Transformer, and Vanilla-Mamba, show that our StratMamba achieves exceptional temporal reasoning efficiency with a lower timeout rate, while maintaining the fastest navigation speed (576 median steps, 5.0% better than Vanilla-Mamba). It also achieves the highest path optimality (0.915 path efficiency) across all baselines. Real-world evaluation reveals that StratMamba maintains more robust performance across extended LiDAR ranges compared to vanilla Mamba and the Transformer, demonstrating that dual-stream partitioning effectively balances reactive safety with strategic navigation under challenging sensing conditions.
Chinese Translation
本文提出了StratMamba，一种基于Mamba的双流时间建模架构，以更有效地捕捉机器人在复杂和障碍物丰富环境中导航所需的长时间依赖关系。StratMamba结合了快速衰减和慢速衰减的记忆架构，其中快速衰减组件处理高频LiDAR数据以实现反应式障碍物规避，而慢速衰减组件则维护更长时间范围的目标信息以进行战略规划。我们在IsaacLab和Gazebo中对不同的障碍物规避场景进行了广泛评估，同时验证了在静态/动态障碍物存在下，Unitree GO1四足机器人成功的仿真到现实的部署。与其他时间强化学习基线（如LSTM、Transformer和Vanilla-Mamba）的比较显示，我们的StratMamba在时间推理效率方面表现出色，超低超时率，同时保持最快的导航速度（576个中位步数，比Vanilla-Mamba提高5.0%）。它还在所有基线中实现了最高的路径最优性（0.915路径效率）。现实世界评估表明，StratMamba在扩展的LiDAR范围内相比于Vanilla Mamba和Transformer保持了更强的性能，证明了双流分区在具有挑战性的感知条件下有效平衡了反应安全性与战略导航。
cs.RO / 14 / 2607.12423
Model-Based Diffusion Optimal Control for Multi-Robot Motion Planning
基于模型的扩散最优控制用于多机器人运动规划
Abstract
Multi-Robot Motion Planning in continuous environments, where robots must generate dynamically feasible, collision-free trajectories, is challenging due to the combinatorial growth of the joint trajectory space and the difficulty of enforcing dynamic feasibility and hard safety constraints. Recent approaches recast trajectory planning as probabilistic inference, sampling from a posterior over trajectories using diffusion models whose score functions are learned from demonstration data. While showing promising performance, these approaches are limited: they often rely on sizable demonstration datasets and struggle to rigorously enforce dynamics and hard safety constraints during sampling. To this end, we introduce Model-Based Diffusion Optimal Control (MDOC), a model-based diffusion planner that efficiently produces dynamically feasible trajectories without relying on data. Crucially, we show that MDOC's safety mechanism -- combining known dynamics models with Control Barrier Function-constrained projections -- naturally scales to multi-robot planning settings through Conflict-Based Search. Across simulation experiments, this integrated method consistently outperforms representative baseline planners in sample efficiency, geometric smoothness, and success rate, while reducing computation time and producing collision-free trajectories.
Chinese Translation
在连续环境中进行多机器人运动规划是一项具有挑战性的任务，因为机器人必须生成动态可行且无碰撞的轨迹，这受到联合轨迹空间的组合增长和强制执行动态可行性及硬安全约束的困难所影响。近期的方法将轨迹规划重新表述为概率推断，通过扩散模型从轨迹的后验分布中采样，这些模型的评分函数是从示范数据中学习得来的。尽管表现出良好的性能，这些方法仍然存在局限性：它们通常依赖于大量的示范数据集，并且在采样过程中难以严格执行动态和硬安全约束。为此，我们提出了基于模型的扩散最优控制（Model-Based Diffusion Optimal Control, MDOC），这是一种基于模型的扩散规划器，能够高效地产生动态可行的轨迹，而无需依赖数据。关键在于，我们展示了MDOC的安全机制——将已知的动态模型与控制障碍函数约束的投影相结合——通过基于冲突的搜索自然地扩展到多机器人规划设置。在模拟实验中，这种集成方法在样本效率、几何平滑性和成功率方面始终优于代表性的基线规划器，同时减少了计算时间并生成了无碰撞的轨迹。
cs.RO / 15 / 2607.12466
Deployable Human Preference Alignment in Robotics: Learning Representative Rewards from Diverse Human Preferences
可部署的人类偏好对齐在机器人技术中的应用：从多样化人类偏好中学习代表性奖励
Abstract
Aligning robot policies with human preferences is essential for deployment to diverse end users. In per-user alignment approach, preference feedback is often sparse, so learning becomes unstable and vulnerable to human preference noise, and a growing number of individualized policies makes validation difficult before deployment. A single shared policy approach to user alignment avoids this cost but fails to capture heterogeneous preferences and often neglects minority preferences. To address these challenges, we introduce Preference-based REward Clustering (PREC), a novel framework that learns a compact set of policies from binary preference labels provided by diverse users. From a dataset of user trajectories and their preference labels, PREC first sets the labels aside and aggregates trajectories across users to learn a population-level shared trajectory encoder, alleviating limited per-user coverage and avoiding label noise during representation learning. Using this representation, PREC jointly assigns users to preference-coherent clusters and learns a representative reward model per cluster using preference labels, from which a policy is optimized for each cluster. Clustering similar users compensates for the limited number of labels available from each user and mitigates the effect of label noise. At the same time, maintaining a manageable number of reward models reduces the validation burden at deployment. Experiments across diverse simulated locomotion environments show that PREC groups users who label different trajectory subsets into preference-coherent clusters more accurately than baseline methods. Under sparse and noisy feedback, policies trained with PREC improve all three social welfare metrics over an existing single shared-policy user-alignment approach and even outperform per-user alignment approaches.
Chinese Translation
将机器人策略与人类偏好对齐对于向多样化最终用户的部署至关重要。在每用户对齐的方法中，偏好反馈通常是稀疏的，因此学习变得不稳定且容易受到人类偏好噪声的影响，个性化策略的数量不断增加使得在部署前的验证变得困难。共享单一策略的方法避免了这一成本，但未能捕捉异质偏好，且往往忽视少数群体的偏好。为了解决这些挑战，我们引入了基于偏好的奖励聚类（Preference-based REward Clustering, PREC），这是一个新颖的框架，能够从多样化用户提供的二元偏好标签中学习一组紧凑的策略。通过用户轨迹及其偏好标签的数据集，PREC首先将标签搁置，并在用户之间聚合轨迹，以学习一个人口级别的共享轨迹编码器，从而缓解每用户覆盖有限的问题，并在表示学习过程中避免标签噪声。利用这一表示，PREC共同将用户分配到偏好一致的聚类中，并使用偏好标签为每个聚类学习一个代表性的奖励模型，从中优化每个聚类的策略。聚类相似用户弥补了每个用户可用标签数量有限的问题，并减轻了标签噪声的影响。同时，保持可管理数量的奖励模型减少了部署时的验证负担。在多样化的模拟运动环境中的实验表明，PREC比基线方法更准确地将标记不同轨迹子集的用户分组到偏好一致的聚类中。在稀疏和噪声反馈下，使用PREC训练的策略在所有三个社会福利指标上优于现有的单一共享策略用户对齐方法，甚至超越了每用户对齐的方法。
cs.RO / 16 / 2607.12489
Infra-Swarm: Robust Vision-Based Multi-Robot Swarming via Near-Infrared Spectral Vision
红外群体：基于近红外光谱视觉的鲁棒多机器人群体
Abstract
Distributed swarms typically rely on either active wireless communication or passive vision, and they are frequently hindered by bandwidth constraints or environmental sensitivity. This paper proposes Infra-Swarm, a robust vision-based swarm. Each robot is equipped with a near-infrared light source and four ordinary gray-scale cameras. The Infra-Swarm system directly measures the centimeter-level 3D position of neighbors based on the position (bearing) and intensity (strength) of optical flares in the captured images. By utilizing 940 nm narrow-band filters to physically reject 99.2% of ambient light interference, the perception front-end achieves hardware-level robustness against illumination variations. Furthermore, its minimal computational overhead provides a resilient foundation for the massive scalability of robotic collectives on resource-constrained hardware.
Chinese Translation
分布式群体通常依赖于主动无线通信或被动视觉，且常常受到带宽限制或环境敏感性的影响。本文提出了Infra-Swarm，一种鲁棒的基于视觉的群体。每个机器人配备有近红外光源和四个普通灰度摄像头。Infra-Swarm系统基于捕获图像中光学闪光的位置信息（方位）和强度（强度），直接测量邻居的厘米级三维位置。通过利用940纳米窄带滤光片物理性地拒绝99.2%的环境光干扰，感知前端在光照变化下实现了硬件级的鲁棒性。此外，其最小的计算开销为在资源受限硬件上实现机器人集体的大规模扩展提供了坚实的基础。
cs.RO / 17 / 2607.12515
A Bearing-Strength Method for Motion Estimation of Unknown Energy Emitters
一种用于未知能量发射器运动估计的方位强度方法
Abstract
This paper studies motion estimation of moving energy emitters using passive sensors. The emitters may be light, acoustic, or radio sources. While the bearing vector pointing from the sensor to the emitter can be easily obtained, existing approaches mainly rely on the bearing-only motion estimation method. However, this method suffers from a fundamental limitation that the sensor must have lateral motion to ensure observability. Unfortunately, this lateral motion requirement often conflicts with the sensor's desired motion in many tasks. In this paper, we point out that the received signal strength, which can also be obtained easily in many ways, can greatly enhance motion estimation. Surprisingly, this strength information has not been well explored so far. Here, we propose a new bearing-strength method to fully exploit both the bearing and strength measurements. Our theoretical analysis shows that the system observability is significantly enhanced in the sense that the lateral motion condition is not required anymore. Real-world experimental results verify the proposed method and the theoretical analysis. It is notable that the benefit of the proposed method comes with no additional cost since it simply utilizes the received strength information that has not been fully exploited in the past.
Chinese Translation
本文研究了使用被动传感器对移动能量发射器进行运动估计的问题。这些发射器可以是光源、声源或无线电源。虽然从传感器指向发射器的方位向量可以很容易获得，但现有的方法主要依赖于仅基于方位的运动估计方法。然而，该方法存在一个根本性限制，即传感器必须具有横向运动以确保可观测性。不幸的是，这种横向运动的要求在许多任务中往往与传感器所需的运动相冲突。本文指出，接收信号强度可以通过多种方式轻松获得，并且可以大大增强运动估计。令人惊讶的是，这种强度信息迄今为止尚未得到充分探索。在此，我们提出了一种新的方位-强度方法，以充分利用方位和强度测量。我们的理论分析表明，系统的可观测性显著增强，因为不再需要横向运动条件。实际实验结果验证了所提方法及其理论分析。值得注意的是，所提方法的优势并不需要额外成本，因为它仅仅利用了过去未被充分利用的接收强度信息。
cs.RO / 18 / 2607.12547
Mind the Gap: Promises and Pitfalls of Hierarchical Planning in LeWorldModel
关注差距：LeWorldModel中层次规划的承诺与陷阱
Abstract
We investigate whether temporal hierarchy can improve LeWorldModel on long-horizon goal-conditioned control. We introduce Hi-LeWM, an extension that freezes the pretrained low-level LeWM and adds high-level planning over latent subgoals. We evaluate Hi-LeWM on PushT and Cube across increasing goal offsets. Hierarchy does not automatically improve performance: at short horizons, the best configuration uses a one-step high-level horizon, while longer horizons reveal a mismatch between the learned high-level action space and the inference-time search distribution. Experiments with true future latent subgoals show that the frozen low-level controller can execute well-aligned intermediate targets, indicating that high-level subgoal generation is the main bottleneck. Unconstrained search can select latent macro-actions that appear favorable under the learned model but produce poor control targets. Constraining search around macro-actions encoded from training trajectories, with appropriate subgoal execution timing, recovers useful hierarchical regimes, improving over flat LeWM by +11.3 percentage points at medium-range horizons and +14.7 percentage points at the longest PushT horizon. Overall, temporal abstraction can benefit compact frozen LeWM, but only when high-level search remains compatible with the low-level controller
Chinese Translation
我们研究了时间层次是否能改善LeWorldModel在长时间目标条件控制中的表现。我们引入了Hi-LeWM，这是一个扩展版本，它冻结了预训练的低层LeWM，并在潜在子目标上增加了高层规划。我们在PushT和Cube上评估Hi-LeWM，随着目标偏移的增加进行测试。层次结构并不自动提高性能：在短时间范围内，最佳配置使用一步高层时间范围，而较长时间范围则显示出学习到的高层动作空间与推理时的搜索分布之间的不匹配。对真实未来潜在子目标的实验表明，冻结的低层控制器能够执行良好对齐的中间目标，表明高层子目标生成是主要瓶颈。无约束搜索可以选择在学习模型下看似有利的潜在宏观动作，但却产生较差的控制目标。通过在训练轨迹中编码的宏观动作周围约束搜索，并适当执行子目标的时机，可以恢复有用的层次结构，在中等范围内比平坦的LeWM提高了11.3个百分点，在最长的PushT时间范围内提高了14.7个百分点。总体而言，时间抽象可以使紧凑的冻结LeWM受益，但前提是高层搜索与低层控制器保持兼容。
cs.RO / 19 / 2607.12571
TrustVLA: Mechanism-Guided Inference-Time Defense Against Vision-Language-Action Backdoors
TrustVLA：机制引导的推理时防御视觉-语言-动作后门
Abstract
Vision-Language-Action (VLA) models are deployed through pipelines that end users cannot audit, and a poisoned VLA can behave normally on clean observations while a small visual trigger redirects a long-horizon robot policy before any failure becomes observable. Existing vision or language defenses rarely explain what a triggered VLA representation looks like or how to recover behavior without retraining. We study this gap through two independently proposed VLA attacks from groups with distinct injection strategies, BadVLA and INFUSE; the latter persists after downstream clean adaptation. Across the evaluated poisoned models, we identify a recurring internal mechanism: a \emph{compact causal footprint}, namely a small visual support that is attention-seeded, spatially compact, and \emph{causal} in a precise sense -- masking it returns a clean-calibrated evidence-evolution score to the normal operating region. This footprint motivates TrustVLA, a mechanism-guided inference-time defense that adapts the Dirichlet evidence framework from trusted classification to monitor per-token, per-layer epistemic uncertainty in VLA policies. With only a small clean calibration set, TrustVLA (i)~detects abnormal evidence evolution, (ii)~localizes the compact support by counterfactual mechanism-score drop, and (iii)~recovers the observation by localized inpainting. Across OpenVLA/LIBERO and $\pi_{0.5}$ transfer evaluations, TrustVLA reduces attack success while preserving clean-task performance, providing a retraining-free, mechanism-guided defense for visual-triggered VLA backdoors.
Chinese Translation
视觉-语言-动作（VLA）模型通过最终用户无法审计的管道进行部署，而被污染的VLA在干净的观测下可能表现正常，同时一个小的视觉触发器在任何故障变得可观察之前会重定向长期的机器人策略。现有的视觉或语言防御很少解释触发的VLA表示是什么样的，或如何在不重新训练的情况下恢复行为。我们通过两个独立提出的VLA攻击进行研究，这两个攻击来自具有不同注入策略的组，分别是BadVLA和INFUSE；后者在下游干净适应后仍然存在。在评估的被污染模型中，我们识别出一个反复出现的内部机制： extit{紧凑因果足迹}，即一个小的视觉支持，它是由注意力种子引导、空间上紧凑，并且在精确意义上是 extit{因果}的——掩蔽它会将干净校准的证据演变分数返回到正常操作区域。这个足迹激励了TrustVLA，一种机制引导的推理时防御，它将来自可信分类的Dirichlet证据框架适应于监测VLA策略中的每个标记、每层的认识不确定性。仅使用一个小的干净校准集，TrustVLA (i)~检测异常的证据演变，(ii)~通过反事实机制分数下降定位紧凑支持，(iii)~通过局部修复恢复观测。在OpenVLA/LIBERO和$ ext{π}_{0.5}$转移评估中，TrustVLA降低了攻击成功率，同时保持了干净任务性能，提供了一种无须重新训练的机制引导防御，针对视觉触发的VLA后门。
cs.RO / 20 / 2607.12604
Streamlining stereo differentiable rendering for marker-free real-time tracking of surgical robots
简化立体可微渲染以实现无标记手术机器人实时跟踪
Abstract
Purpose: Marker-based tracking of surgical robots is occlusion-prone in cluttered operating rooms. We evaluate stereo differentiable rendering for marker-free, real-time robot pose tracking, potentially improving safety, reducing setup time, and enabling multi-robot interaction. Methods: We extend the markerless pose estimation framework roboreg to online dynamic tracking via (i) sequential optimisation that propagates pose estimates across frames with motion-adaptive hyperparameter tuning, and (ii) CUDA stream parallelisation of segmentation and optimisation, combined with CUDA-graph accelerated segmentation. We evaluate on 38 unobstructed and 5 occluded displacement sequences with static start/end ground-truth calibrations and dynamic marker-based reference tracking. Results: We achieve real-time 1080p tracking at 30 fps (up from 14 fps for vanilla roboreg), matching the camera frame rate. Accuracy reaches 1.7 cm / 0.6 deg against static ground truth and 1.2 cm mean 3D error over 27,460 frames against the marker-based reference (1.53 cm over 1,242 occluded frames). Our method outperforms FoundationPose by 11% in dynamic estimation (63% under occlusion) and 250% in static estimation, with 6x faster inference. Conclusions: Stereo differentiable rendering enables real-time, high-resolution marker-free surgical robot tracking, on par with marker-based approaches and surpassing foundation-model baselines.
Chinese Translation
目的：基于标记的手术机器人跟踪在杂乱的手术室中容易受到遮挡。我们评估了立体可微渲染在无标记实时机器人姿态跟踪中的应用，可能提高安全性，减少设置时间，并实现多机器人交互。方法：我们将无标记姿态估计框架 roboreg 扩展为在线动态跟踪，通过 (i) 通过运动自适应超参数调优在帧间传播姿态估计的顺序优化，以及 (ii) CUDA 流并行化分割和优化，结合 CUDA 图加速的分割。我们在 38 个无遮挡和 5 个遮挡位移序列上进行评估，使用静态起始/结束地面真实标定和动态基于标记的参考跟踪。结果：我们实现了 30 fps 的实时 1080p 跟踪（相比于普通 roboreg 的 14 fps），与相机帧率相匹配。相对于静态地面真实值，准确度达到 1.7 cm / 0.6 deg，而在 27,460 帧中相对于基于标记的参考的平均 3D 误差为 1.2 cm（在 1,242 个遮挡帧中为 1.53 cm）。我们的方法在动态估计上比 FoundationPose 提高了 11%（在遮挡下为 63%），在静态估计上提高了 250%，推理速度快了 6 倍。结论：立体可微渲染实现了实时、高分辨率的无标记手术机器人跟踪，性能与基于标记的方法相当，并超越了基础模型基准。
cs.RO / 21 / 2607.12630
Instance-Enriched Semantic Maps for Visual Language Navigation
实例丰富的语义地图用于视觉语言导航
Abstract
Visual Language Navigation (VLN) aims to enable an embodied agent to navigate complex environments by following natural language instructions. Recent approaches build semantic spatial maps and leverage Large Language Models (LLMs) for reasoning and decision making. Despite these advances, existing systems lack instance-level object detail and robustness to diverse user queries, limiting reliable navigation in complex indoor environments. To address these limitations, we propose Instance-Enriched Semantic Maps, a unified framework with three key contributions: (1) Instance-level two-and-a-half-dimensional (2.5D) rich information mapping that constructs maps from color and depth observations via open-vocabulary panoptic segmentation, preserving vertical distinctions and capturing small objects, while storing diverse semantic attributes and natural language captions enriched with room-level context. (2) Robust query processing via LLM-based target selection, which dynamically routes queries across type-specialized experts and integrates their outputs through score-level fusion, enabling consistent goal selection across diverse query formulations. (3) Storage-efficient semantic representation that achieves approximately 96% reduction compared to three-dimensional (3D) scene-graph approaches while preserving sufficient spatial information for navigation. The proposed 2.5D representation outperforms the 3D baseline by over 27% in prediction-normalized Area Under the Curve (AUC). In navigation experiments, our method achieves over 17% improvement in object retrieval and over 23% in navigation success compared to the baseline across diverse query types. The project page is available at https://rcilab.github.io/iesm_vln.
Chinese Translation
视觉语言导航（VLN）旨在使具身代理能够通过遵循自然语言指令在复杂环境中导航。近期的方法构建语义空间地图，并利用大型语言模型（LLMs）进行推理和决策。尽管取得了这些进展，现有系统在实例级对象细节和对多样化用户查询的鲁棒性方面仍显不足，限制了在复杂室内环境中的可靠导航。为了解决这些局限性，我们提出了实例丰富的语义地图，这是一种统一框架，具有三个关键贡献：（1）实例级的两点五维（2.5D）丰富信息映射，通过开放词汇全景分割从颜色和深度观测构建地图，保留垂直区分并捕捉小物体，同时存储丰富的语义属性和带有房间级上下文的自然语言标题。（2）通过基于LLM的目标选择实现鲁棒的查询处理，动态路由查询到类型专门的专家，并通过分数级融合整合它们的输出，实现跨多样化查询形式的一致目标选择。（3）存储高效的语义表示，与三维（3D）场景图方法相比，实现约96%的减少，同时保留足够的空间信息以支持导航。所提出的2.5D表示在预测归一化曲线下面积（AUC）上超过3D基线27%以上。在导航实验中，我们的方法在对象检索上比基线提高了17%以上，在导航成功率上提高了23%以上，适用于多种查询类型。项目页面可访问：https://rcilab.github.io/iesm_vln。
cs.RO / 22 / 2607.12659
Jetson-PI: Towards Onboard Real-Time Robot Control via Foresight-Aligned Asynchronous Inference
Jetson-PI：通过前瞻性对齐异步推理实现机载实时机器人控制
Abstract
Vision-Language-Action (VLA) models have achieved impressive performance on diverse embodied tasks. However, deploying VLA models on low-power onboard devices, such as the Jetson Orin, remains challenging due to their high computational complexity, which leads to substantial inference latency and low control frequency. Asynchronous inference can partially mask this latency by parallelizing action execution and subsequent inference, but it introduces two critical issues: perception-execution misalignment and long reaction time. In this paper, we propose Jetson-PI, a method for efficient VLA deployment on onboard devices via Foresight-Aligned Asynchronous Correction. To address misalignment, we train a lightweight future correction module that predicts future environment representation conditioned on committed actions, enabling the action expert to directly predict actions from the future time step. To reduce reaction time, we introduce confidence-based scheduling optimization that adaptively balances VLM and action expert invocations, complemented by system-level accelerations including CUDA graph reuse, GPU-resident intermediate buffering, and flow unrolling. Extensive experiments demonstrate that Jetson-PI achieves 8.66x and 5.41x improvements in control frequency compared with naive PyTorch and vla.cpp on NVIDIA Jetson Orin, while outperforming VLASH by 14.8\% in average success rate on the LIBERO benchmark. The code of our asynchronous algorithm is available on https://github.com/PKU-SEC-Lab/Jetson-PI, and our efficient llama.cpp-based inference engine is available on https://github.com/PKU-SEC-Lab/Jetson-PI-Edge.
Chinese Translation
视觉-语言-行动（VLA）模型在多种具身任务上取得了令人瞩目的表现。然而，由于其高计算复杂度，VLA 模型在低功耗机载设备（如 Jetson Orin）上的部署仍然面临挑战，这导致了显著的推理延迟和低控制频率。异步推理可以通过并行化动作执行和后续推理部分掩盖这种延迟，但它引入了两个关键问题：感知-执行不对齐和较长的反应时间。本文提出了 Jetson-PI，一种通过前瞻性对齐异步校正在机载设备上高效部署 VLA 的方法。为了解决不对齐问题，我们训练了一个轻量级的未来校正模块，该模块根据已承诺的动作预测未来环境表示，使得动作专家能够直接从未来时间步预测动作。为了减少反应时间，我们引入了基于置信度的调度优化，适应性地平衡 VLM 和动作专家的调用，并辅以包括 CUDA 图重用、GPU 驻留中间缓冲和流展开在内的系统级加速。大量实验表明，Jetson-PI 在 NVIDIA Jetson Orin 上相比于简单的 PyTorch 和 vla.cpp 实现了 8.66 倍和 5.41 倍的控制频率提升，同时在 LIBERO 基准测试中以 14.8% 的平均成功率超越了 VLASH。我们的异步算法代码可在 https://github.com/PKU-SEC-Lab/Jetson-PI 获取，而我们的高效 llama.cpp 基于的推理引擎可在 https://github.com/PKU-SEC-Lab/Jetson-PI-Edge 获取。
cs.RO / 23 / 2607.12702
Vision-Based Dribbling for Humanoid Soccer via Privileged Representation Learning
基于视觉的人形足球运球通过特权表示学习
Abstract
Recent advances in humanoid robotics have highlighted the importance of deployable loco-manipulation skills. Dribbling a soccer ball while evading active opponents requires simultaneous balance, precise ball control, and awareness of a dynamic adversary under onboard sensing and real-time constraints. Existing approaches typically separate perception and motion, which can be effective in controlled settings but may fail under occlusions, fast ball movements, and complex opponent interactions, since perception is not directly optimized for control. We propose an integrated approach in which a temporal depth encoder is embedded into a reinforcement learning policy through a task-specific projection layer. We apply this framework to a simulated Booster T1 humanoid robot and show that it is possible to learn vision-based, opponent-aware dribbling directly from depth observations, without explicit state estimation or privileged scene information. The learned policy achieves 100% success in nominal target-driven dribbling and 96% success with a single static obstacle, while reaching 46% success against an actively moving ball-attacker opponent. These results demonstrate that the proposed framework supports robust vision-based dribbling in nominal and moderately dynamic settings, and provides a strong foundation for handling more challenging moving-adversary scenarios.
Chinese Translation
近年来，人形机器人技术的进步突显了可部署的运动操控技能的重要性。在动态对手的情况下，运球并躲避主动对手需要同时具备平衡、精确的球控能力以及对动态对手的意识，这一切都必须在车载传感器和实时约束下进行。现有的方法通常将感知与运动分开，这在受控环境中可能有效，但在遮挡、快速球运动和复杂对手互动的情况下可能会失败，因为感知并未直接针对控制进行优化。我们提出了一种集成方法，其中一个时间深度编码器通过任务特定的投影层嵌入到强化学习策略中。我们将这一框架应用于模拟的Booster T1人形机器人，并展示了可以直接从深度观测中学习基于视觉的、对手感知的运球，而无需显式的状态估计或特权场景信息。所学策略在名义目标驱动的运球中实现了100%的成功率，在有一个静态障碍物的情况下成功率为96%，而在面对一个主动移动的球攻击者对手时成功率为46%。这些结果表明，所提出的框架支持在名义和中等动态环境中的稳健视觉运球，并为处理更具挑战性的移动对手场景提供了坚实的基础。
cs.RO / 24 / 2607.12732
Globalized Constrained Stein Variational Inference for Diverse Feasible Robot Motion Planning
全球化约束斯坦变分推断用于多样化可行的机器人运动规划
Abstract
Robot motion planning is inherently multimodal, yet classical planners typically return only a single solution. Probabilistic formulations address this limitation by maintaining a distribution over motions, allowing the planner to reason over multiple low-cost alternatives. In robotics, however, motion samples must also satisfy strict constraints, including collision avoidance, joint limits, contact conditions, and dynamics consistency. These hard requirements make motion sampling substantially more challenging: within a limited planning budget, the ensemble must cover diverse low-cost motions while ensuring that every sample remains feasible under the relevant constraints. We propose SteinSQP (Stein Variational Sequential Quadratic Programming), a constrained Stein variational inference method for diverse feasible robot motion sampling. SteinSQP evolves an interacting particle ensemble, as in Stein variational methods, while embedding constraints directly into a kernel-space SQP subproblem. We solve the resulting constrained Stein-Newton subproblem with a GPU-friendly matrix-free primal-dual algorithm, enabling efficient batched ensemble updates. To globalize the method, we introduce an ensemble-level merit function that jointly balances objective value, constraint violation, and particle diversity. Across five constrained motion-planning tasks, SteinSQP returns fully feasible ensembles while preserving diverse motion alternatives. Compared with first-order constrained Stein baselines and serial multistart nonlinear programming, SteinSQP shows faster and more robust ensemble convergence in terms of iterations, improves particle-wise feasibility, and achieves faster batched time-to-solution on challenging robot-scale tasks.
Chinese Translation
机器人运动规划本质上是多模态的，但传统规划器通常只返回单一解。概率性形式化方法通过保持运动的分布来解决这一限制，使规划器能够在多个低成本替代方案中进行推理。然而，在机器人领域，运动样本还必须满足严格的约束条件，包括避免碰撞、关节限制、接触条件和动力学一致性。这些硬性要求使得运动采样变得更加具有挑战性：在有限的规划预算内，集群必须覆盖多样化的低成本运动，同时确保每个样本在相关约束下保持可行。我们提出了SteinSQP（斯坦变分序列二次规划），这是一种用于多样化可行机器人运动采样的约束斯坦变分推断方法。SteinSQP演化了一个相互作用的粒子集群，类似于斯坦变分方法，同时将约束直接嵌入到核空间的SQP子问题中。我们使用一种适合GPU的无矩阵原始-对偶算法来解决得到的约束斯坦-牛顿子问题，从而实现高效的批量集群更新。为了全球化该方法，我们引入了一种集群级的优值函数，联合平衡目标值、约束违反和粒子多样性。在五个约束运动规划任务中，SteinSQP返回完全可行的集群，同时保留多样化的运动替代方案。与一阶约束斯坦基线和串行多起始非线性规划相比，SteinSQP在迭代次数上表现出更快和更稳健的集群收敛，改善了粒子级的可行性，并在具有挑战性的机器人规模任务上实现了更快的批量求解时间。
cs.RO / 25 / 2607.12784
Directional Constraints for Efficient Exploration in Safe Reinforcement Learning
安全强化学习中的高效探索方向约束
Abstract
Reinforcement Learning has revolutionized the landscape of robotic research, allowing robust learning of complex robotic skills in simulation. However, real-world deployment in open-ended environments requires strong safety guarantees to prevent dangerous or harmful behaviors. Safe Reinforcement Learning methods address this requirement by enforcing safety constraints. Nevertheless, learning under constraints often reduces learning speed and could lead to suboptimal task performance, as the agent must solve a more complex constrained optimization problem compared to unconstrained settings. To tackle this issue, in this work, we propose an extension of the ATACOM framework, a state-of-the-art reliable safety layer that can be integrated with existing Reinforcement Learning algorithms to enforce constraints derived from prior knowledge of the system or learned directly from data. Our proposed method, named ATACOM Directional Constraints (ATACOM-DC), significantly improves the safety-performance trade-off by introducing directional constraints that distinguish between actions approaching and moving away from constraint boundaries, activating constraint enforcement only when necessary. We evaluate our method across a range of challenging robotic control tasks in simulation, analyzing both constraint-violation costs and achieved task performance. Code and additional material at https://atacom-dc.robot-learning.net.
Chinese Translation
强化学习彻底改变了机器人研究的格局，使得在模拟环境中能够强健地学习复杂的机器人技能。然而，在开放式环境中进行实际部署需要强有力的安全保障，以防止危险或有害行为。安全强化学习方法通过强制执行安全约束来满足这一要求。然而，在约束下学习通常会降低学习速度，并可能导致次优的任务表现，因为智能体必须解决一个比无约束设置更复杂的约束优化问题。为了解决这一问题，本文提出了ATACOM框架的扩展，这是一个最先进的可靠安全层，可以与现有的强化学习算法集成，以强制执行基于系统先验知识或直接从数据中学习得出的约束。我们提出的方法名为ATACOM方向约束（ATACOM-DC），通过引入方向约束来显著改善安全性与性能之间的权衡，这些方向约束区分了接近和远离约束边界的动作，仅在必要时激活约束执行。我们在一系列具有挑战性的机器人控制任务的模拟中评估了我们的方法，分析了约束违反成本和实现的任务表现。代码及其他材料可在 https://atacom-dc.robot-learning.net 获取。
cs.RO / 26 / 2607.12801
Autonomous Tracking and Terminal Guidance of Moving Targets for Fixed-Wing UAVs
固定翼无人机的自主跟踪与末端引导移动目标
Abstract
This study introduces a unified control framework for fixed-wing unmanned aerial vehicles (UAVs) fitted with a pan-tilt (PT) camera, intended to perform an end-to-end mission spanning from initial target detection to accurate terminal engagement. The proposed system employs a three-phase strategy: a vision-based target acquisition phase, an NMPC-based tracking phase, and a terminal guidance phase. During tracking, the framework uses an Unscented Kalman Filter (UKF) to fuse YOLO-based visual detections with inertial measurements, enabling robust target state estimation under unknown dynamics. To ensure reliable visual contact, we introduce a constraint-aware Nonlinear Model Predictive Control (NMPC) strategy that incorporates Control Barrier Functions (CBFs) to explicitly prevent UAV self-occlusion -- a common limitation in fixed-wing tracking. Upon satisfying terminal engagement conditions, the system seamlessly transitions control to a quaternion-based Biased Proportional Navigation Guidance (BPNG) law, enforcing precise impact angle constraints. High-fidelity simulations demonstrate that the framework achieves stable, robust tracking and accurate terminal interception while strictly respecting the vehicle's dynamic limits and camera field-of-view constraints.
Chinese Translation
本研究提出了一种统一的控制框架，适用于配备云台（PT）摄像头的固定翼无人机（UAV），旨在执行从初始目标检测到精确末端交战的端到端任务。所提系统采用三阶段策略：基于视觉的目标获取阶段、基于非线性模型预测控制（NMPC）的跟踪阶段和末端引导阶段。在跟踪过程中，该框架使用无迹卡尔曼滤波器（UKF）将基于YOLO的视觉检测与惯性测量融合，从而在未知动态下实现稳健的目标状态估计。为了确保可靠的视觉接触，我们引入了一种考虑约束的非线性模型预测控制（NMPC）策略，该策略结合了控制障碍函数（CBFs），以明确防止无人机自遮挡——这是固定翼跟踪中的常见限制。在满足末端交战条件后，系统无缝切换控制至基于四元数的偏置比例导航引导（BPNG）法则，以强制执行精确的撞击角度约束。高保真模拟表明，该框架实现了稳定、稳健的跟踪和准确的末端拦截，同时严格遵守飞行器的动态限制和摄像头视场约束。
cs.RO / 27 / 2607.12811
PixelLoop: Shortcut Topological Navigation with Pixel-Level Loops
PixelLoop：像素级循环的快捷拓扑导航
Abstract
Although topological mapping and navigation have been studied extensively, the specific role and downstream effect of loop closures in purely topological representations has received relatively little attention. Importantly, loop closure over topological maps is distinct from loop closure over globally referenced trajectories and metric maps. Building on recent denser topologies grounded in pixel-level, relative 3D geometry, we propose PixelLoop which introduces loop closures directly in pixel space. Unlike sparse image-level edges or pose-graph corrections in SLAM, our pixel-level closures act as dense topological shortcuts that alter planning connectivity and cost propagation rather than merely aligning coordinates. This dense connectivity enables stable any-point-to-any-point navigation and produces costmaps that align accurately with geometric shortest paths. In particular, we showcase the distinct advantage of applying loop closures to fine-grained pixel topologies rather than image-level topologies. Across extensive simulated experiments, PixelLoop achieves over 35% absolute improvement in both Success Rate and SPL compared to image-relative baselines, with the largest gains in scenarios requiring shortcut exploitation. Results are further validated through real-world mobile robot deployments, demonstrating that dense pixel-level loop closures provide a practical and robust foundation for topological visual navigation. Project Page: https://pixelloop-nav.github.io/
Chinese Translation
尽管拓扑映射和导航已经得到了广泛研究，但在纯拓扑表示中，循环闭合的具体作用和下游影响却相对较少受到关注。重要的是，拓扑图上的循环闭合与全球参考轨迹和度量图上的循环闭合是不同的。在近期基于像素级相对三维几何的更密集拓扑的基础上，我们提出了PixelLoop，它直接在像素空间中引入循环闭合。与稀疏的图像级边缘或SLAM中的姿态图校正不同，我们的像素级闭合作为密集的拓扑快捷方式，改变了规划的连通性和成本传播，而不仅仅是对齐坐标。这种密集的连通性使得任意点到任意点的导航更加稳定，并生成与几何最短路径准确对齐的成本图。特别地，我们展示了将循环闭合应用于细粒度像素拓扑而非图像级拓扑的明显优势。在广泛的模拟实验中，PixelLoop在成功率和SPL方面相比于图像相关基线取得了超过35%的绝对提升，尤其在需要利用快捷方式的场景中获得了最大的收益。结果通过现实世界的移动机器人部署进一步验证，表明密集的像素级循环闭合为拓扑视觉导航提供了一个实用且稳健的基础。项目页面：https://pixelloop-nav.github.io/
cs.RO / 28 / 2607.12861
Unveiling Complex Collective Behaviors from Simple Rewards
从简单奖励揭示复杂集体行为
Abstract
Multi-agent Reinforcement Learning (MARL) holds great potential for robot swarms, but the black-box nature of neural policies complicates strategic analysis, limiting multi-robot applications. Furthermore, complex swarm behaviors can surprisingly emerge from simple rewards without explicit aggregation incentives. Unveiling the mechanisms behind this emergence is critical, but the disconnection between simple rewards and collective behaviors exacerbates interpretability challenges. This paper aims to reveal the hidden mechanisms in this process. We propose a two-stage EEC (\LinkIII) explanatory framework. This includes a novel analytical tool called the Agent Response Map (ARM), which reveals agents' decision-making patterns across space and identifies regions of aggregation and avoidance. ARM reveals that the robots implicitly learn the geometric fields of the environment and utilize these structures as desired targets for coordinated movement. We validate this finding across two distinct tasks: a cooperative multi-robot shape assembly and a competitive predator-prey pursuit-evasion. 1) In the cooperative task, ARM identifies the unoccupied target interior as the desired destination for robot navigation. As the center becomes occupied, this target region automatically shifts toward the boundary, demonstrating the robots' capacity to autonomously explore unoccupied areas. 2) In the competitive task, ARM surprisingly identifies the boundary of the predators' Voronoi diagram as the convergence destination for prey agents. Together, these two tasks demonstrate the capability of ARM to discover the hidden geometric structures underlying MARL policies in robot swarms.
Chinese Translation
多智能体强化学习（MARL）在机器人群体中具有巨大的潜力，但神经政策的黑箱特性使得战略分析变得复杂，从而限制了多机器人应用。此外，复杂的群体行为可以在没有明确聚合激励的情况下，意外地从简单奖励中涌现。揭示这一涌现背后的机制至关重要，但简单奖励与集体行为之间的脱节加剧了可解释性挑战。本文旨在揭示这一过程中的隐含机制。我们提出了一种两阶段的EEC（ extit{LinkIII}）解释框架，其中包括一种新颖的分析工具，称为智能体响应图（ARM），该工具揭示了智能体在空间上的决策模式，并识别聚合和回避区域。ARM显示，机器人隐式地学习环境的几何场，并利用这些结构作为协调运动的目标。我们在两个不同的任务中验证了这一发现：一个是合作的多机器人形状组装，另一个是竞争的捕食者-猎物追逐-逃避任务。1）在合作任务中，ARM识别出未占用的目标内部作为机器人导航的期望目的地。当中心区域被占用时，该目标区域会自动向边界移动，展示了机器人自主探索未占用区域的能力。2）在竞争任务中，ARM意外地识别出捕食者的Voronoi图边界作为猎物智能体的汇聚目的地。这两个任务共同展示了ARM发现机器人群体中MARL政策背后隐藏的几何结构的能力。
cs.RO / 29 / 2607.12892
UR-VC: Unsupervised Robotic Value Correction for Time-Derived Progress Proxies
UR-VC：用于时间衍生进度代理的无监督机器人价值修正
Abstract
Modern robot learning systems increasingly rely on dense progress or value signals to evaluate intermediate states, guide policy learning, and detect task completion, making the quality of these signals critical. Since such dense labels are rarely available at scale, normalized time within a demonstration is often used as a scalable substitute: later frames are treated as higher progress. However, this time-derived label is only a noisy proxy for physical task progress. In contact-rich manipulation, a robot may make progress and then lose it through slips, failed grasps, or partial undoing, while the time-derived label continues to increase monotonically. We introduce Unsupervised Robotic Value Correction (UR-VC), an offline, training-free method for correcting time-derived progress labels. UR-VC exploits a simple regularity in demonstration data: similar states often recur across different episodes, but at different timestamps. Instead of trusting the timestamp from a single trajectory, UR-VC retrieves similar states from other episodes and aggregates their time-derived labels to obtain a corrected progress estimate. UR-VC requires no manual progress labels, reward annotations, or additional value model. We evaluate UR-VC on real bimanual cloth flatten-and-fold data, a long-horizon deformable-object manipulation task with visible intermediate progress. The corrected labels capture local regressions and non-uniform progress that normalized time cannot represent, while preserving the overall task trend. We further use the corrected signal to construct advantage labels for VLA training, following recent advantage-conditioned policy learning. UR-VC shows a positive trend in real-robot task success under matched data, model, and training settings.
Chinese Translation
现代机器人学习系统越来越依赖于密集的进度或价值信号来评估中间状态、指导策略学习和检测任务完成，这使得这些信号的质量至关重要。由于此类密集标签在规模上很少可用，因此在演示中通常使用标准化时间作为可扩展的替代方案：后续帧被视为更高的进度。然而，这种时间衍生标签仅仅是物理任务进度的一个嘈杂代理。在接触丰富的操作中，机器人可能会取得进展，然后通过滑动、抓取失败或部分撤销而失去进展，而时间衍生标签则持续单调增加。我们提出了无监督机器人价值修正（UR-VC），这是一种离线、无训练的方法，用于修正时间衍生的进度标签。UR-VC利用演示数据中的一种简单规律：相似的状态通常在不同的剧集间重复出现，但时间戳不同。UR-VC不是依赖单一轨迹的时间戳，而是从其他剧集中检索相似状态，并聚合它们的时间衍生标签以获得修正的进度估计。UR-VC不需要手动进度标签、奖励注释或额外的价值模型。我们在真实的双手布料平整和折叠数据上评估UR-VC，这是一个具有可见中间进度的长时间跨度可变形物体操作任务。修正后的标签捕捉到局部回归和标准化时间无法表示的非均匀进度，同时保留整体任务趋势。我们进一步利用修正信号构建优势标签，以用于VLA训练，遵循最近的优势条件策略学习。在匹配的数据、模型和训练设置下，UR-VC在真实机器人任务成功率上显示出积极的趋势。
cs.RO / 30 / 2607.12931
ExToken: Structured Exploration for Efficient Vision-Language-Action Reinforcement Fine-tuning
ExToken：高效视觉-语言-动作强化微调的结构化探索
Abstract
Reinforcement Learning (RL) has demonstrated significant potential for improving Vision-Language-Action (VLA) models on complex manipulation tasks. However, its practical scalability remains severely limited by the substantial cost of environmental interactions. In this work, we first investigate the exploration stagnation bottleneck in current VLA-RL frameworks and reveal that trajectory diversity is fundamentally more important to sample efficiency than the sheer quantity of collected rollouts. Motivated by these insights, we introduce RL Exploration Token (ExToken), a simple yet general framework that condition VLA policies on discrete behavioral priors derived from offline demonstrations for structured exploration. By conditioning the policy on different tokens during rollout collection, ExToken encourages the agent to explore diverse behavioral modes, substantially improving state-action coverage and exploration efficiency. To bridge exploration during training with deterministic inference at deployment, ExToken further incorporates a state-conditioned token selector that adaptively predicts effective behavioral modes for unseen scenarios. Extensive experiments across simulated and real-world robotic manipulation tasks demonstrate that ExToken consistently accelerates convergence, improves task performance, and exhibits strong robustness under highly constrained interaction budgets.
Chinese Translation
强化学习（RL）在复杂操作任务中展示了显著提升视觉-语言-动作（VLA）模型的潜力。然而，其实际可扩展性仍受到环境交互成本的严重限制。在本研究中，我们首先探讨了当前VLA-RL框架中的探索停滞瓶颈，并揭示了轨迹多样性在样本效率上比收集的回合数量更为重要。基于这些见解，我们引入了强化学习探索令牌（ExToken），这是一个简单而通用的框架，它基于离线演示中获得的离散行为先验对VLA策略进行条件化，以实现结构化探索。通过在回合收集过程中对不同令牌进行条件化，ExToken鼓励智能体探索多样的行为模式，从而显著改善状态-动作覆盖率和探索效率。为了在训练期间的探索与部署时的确定性推理之间架起桥梁，ExToken进一步结合了一个状态条件的令牌选择器，该选择器自适应地预测未见场景的有效行为模式。通过在模拟和真实世界的机器人操作任务中进行广泛实验，结果表明ExToken始终加速收敛，提高任务性能，并在高度受限的交互预算下表现出强大的鲁棒性。
cs.RO / 31 / 2607.12965
MAMMOTH: A Multi-Modal End-to-End Policy for Off-Road Mobility Robust to Missing Modality
MAMMOTH：一种对缺失模态具有鲁棒性的多模态端到端越野移动策略
Abstract
Reliable autonomous navigation in unstructured off-road environments remains a critical unsolved challenge due to extreme terrain diversity, drastic illumination variations and acute sensor degradation. Recent developments have approached the problem as a traversability costmap estimation or visual navigation task. However, many exhibit heavy reliance on RGB modality, leading to poor performance in varied illumination such as glares, shadows or low ambient light. Achieving robust generalization in such conditions requires integrating modalities that provide supplementary scene information. Such multi-modal methods suffer from a rigid dependency on the presence of near-perfect sensor inputs, leaving them unable to robustly handle sensor degradation or individual modality failure. To address these limitations, we introduce MAMMOTH (MAsking Multi-Modal inputs for Off-road Traversability Heuristic-informed navigation), a unified end-to-end navigation policy for robust off-road visual-goal-conditioned navigation and undirected exploration. Specifically, MAMMOTH efficiently fuses multi-modal observations (RGB, Thermal, 3D Pointcloud and Ego Velocity) and is trained with a modality dropout scheme, enabling it to generalize to missing modalities at inference time. Furthermore, we employ a diffusion policy to learn the joint conditional probability distribution of physically-grounded trajectories and a intrinsic traversability heuristic. MAMMOTH utilizes this heuristic to prefer safer, smoother trajectories. We validate MAMMOTH through extensive real-world robot experiments in distinct off-road environments, including night-time operation. Our results demonstrate superior performance, with significant improvements in collision avoidance, terrain-aware planning and generalization to missing modalities. The code and dataset used for this work will be made publicly available.
Chinese Translation
在非结构化的越野环境中实现可靠的自主导航仍然是一个关键的未解决挑战，原因在于极端的地形多样性、剧烈的光照变化和严重的传感器退化。近期的发展将这一问题视为可通行性成本图估计或视觉导航任务。然而，许多方法严重依赖于RGB模态，导致在眩光、阴影或低环境光等变化光照条件下表现不佳。在这种情况下实现鲁棒的泛化能力需要整合提供补充场景信息的模态。这类多模态方法受到对近乎完美传感器输入的严格依赖，无法有效处理传感器退化或单一模态故障。为了解决这些局限性，我们提出了MAMMOTH（MAsking Multi-Modal inputs for Off-road Traversability Heuristic-informed navigation），这是一种统一的端到端导航策略，旨在实现鲁棒的越野视觉目标条件导航和无向探索。具体而言，MAMMOTH高效地融合多模态观测（RGB、热成像、3D点云和自我速度），并采用模态丢失方案进行训练，使其能够在推理时对缺失模态进行泛化。此外，我们采用扩散策略学习物理基础轨迹的联合条件概率分布和内在可通行性启发式。MAMMOTH利用这一启发式偏好更安全、更平滑的轨迹。我们通过在不同的越野环境中进行广泛的真实机器人实验来验证MAMMOTH，包括夜间操作。我们的结果显示出卓越的性能，在碰撞避免、地形感知规划和对缺失模态的泛化方面有显著改善。用于本研究的代码和数据集将公开发布。
cs.RO / 32 / 2607.12992
ChunkFlow: Towards Continuity-Consistent Chunked Policy Learning
ChunkFlow：朝着连续一致的分块策略学习
Abstract
Vision-language action (VLA) models increasingly adopt chunked action heads to satisfy real-time constraints; however, this introduces boundary jitter: overlapping regions between consecutive chunks often yield inconsistent predictions, degrading temporal coherence and the task success rate. Existing methods, such as inference-time blending, merely reweight mismatched proposals without correcting underlying errors, leading to residual accumulation under biased or noisy histories. We propose ChunkFlow, a seam-aware training-and-execution framework for chunked policies that aligns chunk structure with boundary execution. It partitions each chunk into frozen, editable, and future zones, applies deterministic overlap blending at execution, and trains raw predictions with seam and first- and second-order continuity losses. History corruption and scheduled sampling improve robustness to executed-history errors, while an AWAC fine-tuning stage adapts the policy without removing these structural regularizers. Under mild smoothness assumptions, pre-blending seam discrepancies provably decay with increasing overlap. Experiments on CALVIN, LIBERO, and real robots show an improved success-stability trade-off with low-latency inference. Project page: https://cytoderm-ai.github.io/chunkflow.
Chinese Translation
视觉-语言动作（VLA）模型越来越多地采用分块动作头以满足实时约束；然而，这引入了边界抖动：连续块之间的重叠区域常常导致不一致的预测，降低了时间一致性和任务成功率。现有方法，如推理时混合，仅仅重新加权不匹配的提案，而未能纠正潜在的错误，导致在偏差或噪声历史下的残差积累。我们提出了ChunkFlow，一个关注接缝的分块策略训练与执行框架，该框架将分块结构与边界执行对齐。它将每个块划分为冻结区、可编辑区和未来区，在执行时应用确定性重叠混合，并使用接缝及一阶和二阶连续性损失来训练原始预测。历史损坏和调度采样提高了对执行历史错误的鲁棒性，而AWAC微调阶段则在不移除这些结构正则化器的情况下调整策略。在温和的平滑性假设下，预混合接缝差异在重叠增加时可证明衰减。对CALVIN、LIBERO和真实机器人进行的实验表明，在低延迟推理下，成功与稳定性的权衡得到了改善。项目页面：https://cytoderm-ai.github.io/chunkflow。
cs.RO / 33 / 2607.13017
FlowWAM: Optical Flow as a Unified Action Representation for World Action Models
FlowWAM：将光流作为世界动作模型的统一动作表示
Abstract
World Action Models (WAMs) are able to leverage pretrained video generators for both world modeling and action prediction. However, directly leveraging such video generators for control raises a new challenge: how to represent actions in a suitable form that aligns with pretrained video generators while carrying enough motion cues for accurate control. Existing numerical actions fail to satisfy the former, and prior visual action representations overlook the temporal motion structure across frames. We address this issue with FlowWAM, a dual-stream diffusion framework that adopts optical flow as a unified, video-native action representation. Flow videos share the same format as RGB videos and encode rich per-pixel displacement. By jointly modeling them within a shared pretrained video generator, FlowWAM can naturally implement two modes of WAMs. In policy mode, FlowWAM generates flow for action prediction, while in world-model mode, it uses target flow sequences to guide future video generation. Moreover, since flow can be easily extracted from raw videos without action labels, FlowWAM can leverage large-scale action-unlabeled video datasets for pretraining. We empirically find that our flow-based action representation delivers gains across both modes. On RoboTwin manipulation, FlowWAM raises the success rate to 92.94% on the Clean setting and 92.14% on Random, outperforming both VLA and WAM baselines. On WorldArena world modeling, it achieves the best overall EWMScore (63.71) with an 18.4% relative improvement in trajectory accuracy. More results can be found on our project website: https://flow-wam.github.io .
Chinese Translation
世界动作模型（WAMs）能够利用预训练的视频生成器进行世界建模和动作预测。然而，直接利用这些视频生成器进行控制带来了一个新挑战：如何以适合的形式表示动作，使其与预训练的视频生成器对齐，同时又能携带足够的运动线索以实现准确控制。现有的数值动作表示未能满足前者，而之前的视觉动作表示则忽视了帧间的时间运动结构。我们通过FlowWAM解决了这一问题，FlowWAM是一个双流扩散框架，采用光流作为统一的视频原生动作表示。光流视频与RGB视频共享相同的格式，并编码丰富的每像素位移。通过在共享的预训练视频生成器中共同建模，FlowWAM可以自然地实现两种模式的WAMs。在策略模式下，FlowWAM生成光流以进行动作预测，而在世界模型模式下，它使用目标光流序列来指导未来的视频生成。此外，由于光流可以轻松地从原始视频中提取而无需动作标签，FlowWAM能够利用大规模无标签动作视频数据集进行预训练。我们实证发现，我们基于光流的动作表示在两种模式下均带来了提升。在RoboTwin操作中，FlowWAM在Clean设置下的成功率提高至92.94%，在Random设置下为92.14%，超越了VLA和WAM基线。在WorldArena世界建模中，它实现了最佳的整体EWMScore（63.71），在轨迹准确性上相对提高了18.4%。更多结果可在我们的项目网站上找到：https://flow-wam.github.io 。
cs.RO / 34 / 2607.13033
DenseReward: Dense Reward Learning via Failure Synthesis for Robotic Manipulation
DenseReward：通过失败合成实现密集奖励学习的机器人操作
Abstract
Reinforcement learning holds great promise for improving robot policies beyond the limits of imitation learning. However, its practical adoption remains bottlenecked by the lack of reliable vision-language reward models that provide dense and informative feedback. Two key challenges remain: acquiring diverse failure data at scale and obtaining fine-grained reward signals beyond sparse trajectory-level success labels. Collecting failure trajectories typically requires laborious human effort, while pseudo-failures constructed by relabeling successful demonstrations fail to capture the diverse physical failure modes that arise during robot execution. Meanwhile, existing reward models often predict sparse binary or trajectory-level rewards, which provide limited guidance for efficient policy optimization. We introduce DenseReward, a dense robotic reward model that addresses both challenges. To train DenseReward, we develop an automated failure data generation pipeline that synthesizes physically realistic failure trajectories in simulation without human labeling, covering diverse failure modes such as collisions, missed grasps, object drops, and recovery behaviors. DenseReward predicts dense frame-level reward scores from visual observations and language instructions, enabling fine-grained estimation of task progress throughout an episode. Experiments show that DenseReward outperforms general-purpose VLMs and existing robotic reward models in dense reward prediction across both simulated and real-world manipulation. We further demonstrate that DenseReward provides effective reward guidance for downstream model predictive control and reinforcement learning. We release the dataset, trained reward models, and evaluation suite to support the development of failure-aware dense reward modeling for robot learning.
Chinese Translation
强化学习在超越模仿学习的限制方面对改善机器人策略具有巨大潜力。然而，其实际应用仍受到缺乏可靠的视觉-语言奖励模型的瓶颈，这些模型提供密集且信息丰富的反馈。当前面临两个主要挑战：大规模获取多样化的失败数据，以及获得超越稀疏轨迹级成功标签的细粒度奖励信号。收集失败轨迹通常需要耗费大量人力，而通过重新标记成功示例构建的伪失败无法捕捉到机器人执行过程中出现的多样化物理失败模式。同时，现有的奖励模型通常预测稀疏的二元或轨迹级奖励，这为高效的策略优化提供了有限的指导。我们提出了DenseReward，这是一种密集的机器人奖励模型，旨在解决这两个挑战。为了训练DenseReward，我们开发了一个自动化的失败数据生成管道，该管道在模拟环境中合成物理上真实的失败轨迹，无需人工标注，涵盖了诸如碰撞、抓取失败、物体掉落和恢复行为等多样化的失败模式。DenseReward从视觉观察和语言指令中预测密集的帧级奖励分数，使得在整个任务过程中能够进行细粒度的任务进展估计。实验表明，DenseReward在模拟和真实世界操作中的密集奖励预测方面优于通用的视觉语言模型（VLM）和现有的机器人奖励模型。我们进一步证明，DenseReward为下游模型预测控制和强化学习提供了有效的奖励指导。我们发布了数据集、训练好的奖励模型和评估套件，以支持面向失败的密集奖励建模在机器人学习中的发展。
cs.CV / 1 / 2607.11939
TSCA-Net: Temporal-Spatial Clique Attention for Interpretable Multimodal Pedestrian Trajectory Prediction
TSCA-Net：用于可解释的多模态行人轨迹预测的时空团体注意力
Abstract
Accurate pedestrian trajectory prediction in crowded environments remains challenging due to the multimodal uncertainty of human motion and the variable complexity of motion dynamics across different scene contexts. Existing goal-conditioned models rely on static displacement structures that assign equal weight to all historical time steps, standard graph attention mechanisms, and fixed-capacity motion decoders that cannot adapt to local prediction complexity. To address these limitations, we propose TSCA-Net, a trajectory prediction framework built upon three complementary modules. The Temporal-Spatial Clique Attention (TSCA) module introduces learnable temporal gating into clique-based goal-history interaction, enabling time-aware modulation of historical observations relative to each candidate goal. The Cross-Pedestrian Clique Potential (CPCP) module models asymmetric pairwise agent relationships through a dynamic clique potential framework with a time-varying social graph. The Adaptive KAN Grid Refinement (AKGR) mechanism dynamically adjusts the B-spline grid resolution of a Kolmogorov-Arnold Network-augmented LSTM decoder based on per-agent goal distribution entropy, balancing model expressiveness against overfitting across varying motion complexities. Extensive experiments on the ETH/UCY and Stanford Drone Dataset benchmarks demonstrate that TSCA-Net achieves state-of-the-art performance, with average ADE/FDE of 0.13/0.20 m on ETH/UCY and 6.95/10.43 pixels on SDD. Comprehensive ablation studies confirm the complementary contributions of all three proposed modules.
Chinese Translation
在拥挤环境中，准确的行人轨迹预测仍然面临挑战，这主要是由于人类运动的多模态不确定性和不同场景上下文中运动动态的复杂性变化。现有的目标条件模型依赖于静态位移结构，这些结构对所有历史时间步赋予相同的权重，使用标准图注意力机制，以及固定容量的运动解码器，无法适应局部预测的复杂性。为了解决这些局限性，我们提出了TSCA-Net，这是一种基于三个互补模块的轨迹预测框架。时空团体注意力（Temporal-Spatial Clique Attention, TSCA）模块将可学习的时间门控引入基于团体的目标历史交互，使得相对于每个候选目标能够对历史观察进行时间感知的调制。交叉行人团体潜力（Cross-Pedestrian Clique Potential, CPCP）模块通过动态团体潜力框架和时间变化的社交图建模不对称的成对代理关系。自适应KAN网格细化（Adaptive KAN Grid Refinement, AKGR）机制根据每个代理的目标分布熵动态调整Kolmogorov-Arnold网络增强的LSTM解码器的B样条网格分辨率，在不同运动复杂性之间平衡模型的表现力与过拟合。对ETH/UCY和斯坦福无人机数据集基准的广泛实验表明，TSCA-Net实现了最先进的性能，在ETH/UCY上的平均ADE/FDE为0.13/0.20米，在SDD上的平均ADE/FDE为6.95/10.43像素。全面的消融研究确认了所有三个提议模块的互补贡献。
cs.CV / 2 / 2607.11941
GenDiff: A Dose and Anatomy Aware Diffusion Model with Structural Prior Refinement for Low-Dose CT Reconstruction and Generalization
GenDiff：一种具有结构先验精细化的剂量与解剖学感知扩散模型，用于低剂量CT重建与泛化
Abstract
Computed tomography (CT) is a critical imaging modality for clinical diagnosis, but reducing radiation dose inevitably introduces severe noise and structured artifacts that degrade image quality. Existing deep learning-based low-dose CT (LDCT) reconstruction methods are typically optimized for fixed dose levels or specific anatomical regions, limiting their robustness and generalization in realistic clinical settings. We propose GenDiff, a generalizable diffusion-based framework for LDCT reconstruction that jointly models continuous radiation dose and anatomical information within a unified reconstruction network. The proposed framework integrates a Dose-Anatomy Encoder to learn acquisition-aware embeddings, a dose- and anatomy-conditioned cold diffusion backbone for iterative refinement, a physics-consistency update to enforce fidelity to the CT forward model, and a Structural Prior Refinement Module (SPRM) that preserves anatomical structures while suppressing dose-dependent artifacts. Extensive experiments on multi-anatomy clinical datasets, including unseen ultra-low-dose conditions as well as out-of-distribution phantom and animal datasets, demonstrate that GenDiff consistently outperforms state-of-the-art convolutional neural network and diffusion-based reconstruction methods. The proposed approach achieves superior reconstruction quality while maintaining strong robustness across different dose levels, anatomical regions, and acquisition domains, making it a promising solution for practical low-dose CT imaging.
Chinese Translation
计算机断层扫描（CT）是临床诊断中一种关键的成像方式，但降低辐射剂量不可避免地会引入严重的噪声和结构伪影，从而降低图像质量。现有的基于深度学习的低剂量CT（LDCT）重建方法通常针对固定剂量水平或特定解剖区域进行优化，限制了其在实际临床环境中的鲁棒性和泛化能力。我们提出了GenDiff，一种可泛化的基于扩散的LDCT重建框架，该框架在统一的重建网络中联合建模连续辐射剂量和解剖信息。所提框架集成了剂量-解剖编码器，以学习获取感知的嵌入；一个基于剂量和解剖条件的冷扩散主干网络，用于迭代精细化；一个物理一致性更新，以强制遵循CT前向模型的保真度；以及一个结构先验精细化模块（SPRM），在抑制剂量依赖伪影的同时保留解剖结构。在多解剖临床数据集上的广泛实验，包括未见的超低剂量条件以及分布外的幻影和动物数据集，表明GenDiff在性能上始终优于最先进的卷积神经网络和基于扩散的重建方法。所提方法在不同剂量水平、解剖区域和获取领域中保持强大的鲁棒性，同时实现了优越的重建质量，成为实际低剂量CT成像的有前景的解决方案。
cs.CV / 3 / 2607.11957
Anomalous Frame Detection Using VLM-Based Description Comparison for Extracting Expert-Specific Actions and Contextual Decision-Making Scenes with Intra-Video Self-Similarity
基于视觉语言模型（VLM）描述比较的异常帧检测，用于提取专家特定动作和上下文决策场景，结合视频内自相似性
Abstract
Maintenance of critical infrastructures, such as railways and power plants, is essential for ensuring operational safety and reliability. However, the declining number of skilled maintenance workers highlights the need to transfer expert know-how to less experienced workers. Previous studies have attempted to extract candidates of expert knowledge by comparing videos of manual-based work with those of expert workers, mainly focusing on differences in observable actions. However, expert know-how is often embedded not only in actions but also in contextual decision-making during task execution. This paper proposes a method that detects anomalous frames between two task videos to automatically extract candidate scenes containing expert-specific actions and contextual decision-making scenes. The method generates frame-wise visual descriptions using a vision-language model (VLM). Expert-specific actions are extracted based on frame similarities computed from description comparisons between two videos, while contextual decision-making scenes are extracted using segment similarities derived from intra-video self-similarity of the descriptions. In simulated distribution board maintenance experiments involving 27 task scenarios, the proposed method achieved extraction rates of 65% for action candidates and 61% for decision-scene candidates, improving over conventional methods that achieved 59% and 33%, respectively. These results demonstrate the effectiveness of the proposed approach in discovering candidate scenes containing expert know-how.
Chinese Translation
维护铁路和电厂等关键基础设施对于确保操作安全和可靠性至关重要。然而，熟练维护工人的数量逐渐减少，凸显了将专家知识转移给经验较少工人的必要性。以往的研究通过比较手工操作视频与专家工人视频，主要关注可观察动作的差异，以提取专家知识的候选者。然而，专家知识通常不仅嵌入于动作中，还体现在任务执行过程中的上下文决策中。本文提出了一种方法，通过检测两个任务视频之间的异常帧，自动提取包含专家特定动作和上下文决策场景的候选场景。该方法使用视觉语言模型（VLM）生成逐帧的视觉描述。专家特定动作是基于从两个视频之间的描述比较计算出的帧相似性提取的，而上下文决策场景则是利用从描述的内视频自相似性导出的段相似性提取的。在涉及27个任务场景的模拟配电板维护实验中，所提方法的动作候选提取率达到了65%，决策场景候选提取率为61%，相比传统方法的59%和33%有所提升。这些结果证明了所提方法在发现包含专家知识的候选场景方面的有效性。
cs.CV / 4 / 2607.11962
Contrastive Joint-Embedding Prediction for Representation Learning in Structural MRI
用于结构性MRI表征学习的对比联合嵌入预测
Abstract
Self-supervised learning offers a compelling approach for medical imaging, where labeled data are scarce and acquisition costs are high. We present COJEPA, a self-supervised framework for volumetric brain MRI that combines a joint-embedding predictive architecture (JEPA) with a contrastive loss (CO), targeting two complementary properties: local predictivity and global discriminability. The model is trained without labels on T1-weighted structural MRI from two cohorts (HCP-YA and AABC, $N{=}2286$, ages 22 to 90), extending I-JEPA to 3D with foreground-aware block masking, a hierarchical convolutional patch embedding, and world-space sinusoidal positional encodings. We evaluate all three objectives across zero-shot twin retrieval, brain tumor segmentation (BraTS 2024), and age regression (OpenBHB). COJEPA achieves the best monozygotic twin recall at rank@1 (0.84), the best finetuning age MAE (2.55 years on OpenBHB 3.0T), and matches CO on BraTS whole-tumor Dice, demonstrating that the combined objective yields representations that are simultaneously discriminative and locally structured.
Chinese Translation
自监督学习为医学影像提供了一种引人注目的方法，尤其是在标注数据稀缺且获取成本高昂的情况下。我们提出了COJEPA，一个针对体积脑MRI的自监督框架，该框架结合了联合嵌入预测架构（JEPA）和对比损失（CO），旨在实现两个互补特性：局部可预测性和全局可区分性。该模型在两个队列（HCP-YA和AABC，$N{=}2286$，年龄范围22至90岁）的T1加权结构性MRI上进行无标签训练，扩展了I-JEPA至3D，采用前景感知块掩蔽、分层卷积补丁嵌入和世界空间正弦位置编码。我们在零-shot双胞胎检索、脑肿瘤分割（BraTS 2024）和年龄回归（OpenBHB）等三个目标上进行了评估。COJEPA在rank@1的单卵双胞胎召回率上达到了最佳值（0.84），在OpenBHB 3.0T上获得了最佳的微调年龄平均绝对误差（MAE，2.55岁），并在BraTS全肿瘤Dice上与CO相匹配，证明了综合目标能够产生同时具有可区分性和局部结构的表征。
cs.CV / 5 / 2607.11986
SpikeDS: Dual Sparsity Spikformer for Perineural Invasion Prediction in 3D MRI
SpikeDS：用于三维MRI中神经周围侵袭预测的双稀疏Spikformer
Abstract
Perineural invasion (PNI) is associated with poor prognosis in cholangiocarcinoma (CCA). However, its detection from 3D MRI remains challenging due to the subtle and spatially heterogeneous imaging signatures at the tumor periphery. Capturing such spatially sparse cues necessitates volumetric analysis of 3D MRI, but existing deep learning approaches incur prohibitive computational costs on volumetric medical images, limiting their clinical deployment. We propose Dual Sparsity Spikformer (SpikeDS), a spiking neural network architecture that jointly exploits activation sparsity from binary spike communication and spatial sparsity from window pruning based on firing rates. SpikeDS introduces Dual Sparsity Spiking Attention (DSSA), which combines two complementary mechanisms. The first is Window-based Expert Mixture Spiking Attention (W-EMSA), which selectively applies attention only to salient windows identified by their firing rates. The second is Cross-Window Spiking Self-Attention (CW-SSA), which enables global context exchange through an asymmetric scheme in which pruned windows still contribute as key-value sources. Evaluated on a clinical cohort of 139 CCA patients via 5-fold cross-validation, SpikeDS achieves an AUC of 0.753 while consuming only 14.4 mJ, surpassing the best baseline in both AUC and energy efficiency. These results suggest that dual sparsity provides an effective hardware-aware strategy for improving the efficiency of 3D spiking transformers without compromising diagnostic performance.
Chinese Translation
神经周围侵袭（PNI）与胆管癌（CCA）的预后不良相关。然而，由于肿瘤边缘的成像特征微妙且空间异质，从三维MRI中检测PNI仍然具有挑战性。捕捉这种空间稀疏线索需要对三维MRI进行体积分析，但现有的深度学习方法在处理体积医学图像时会产生高昂的计算成本，限制了其临床应用。我们提出了双稀疏Spikformer（SpikeDS），这是一种脉冲神经网络架构，联合利用来自二进制脉冲通信的激活稀疏性和基于发射率的窗口修剪的空间稀疏性。SpikeDS引入了双稀疏脉冲注意力（DSSA），结合了两种互补机制。第一种是基于窗口的专家混合脉冲注意力（W-EMSA），它仅对通过发射率识别的显著窗口选择性地应用注意力。第二种是跨窗口脉冲自注意力（CW-SSA），它通过一种不对称方案实现全局上下文交换，其中修剪的窗口仍然作为键值源进行贡献。在对139名CCA患者的临床队列进行5折交叉验证时，SpikeDS实现了0.753的AUC，同时仅消耗14.4 mJ，超越了在AUC和能效方面的最佳基线。这些结果表明，双稀疏提供了一种有效的硬件感知策略，可以提高三维脉冲变换器的效率，而不影响诊断性能。
cs.CV / 6 / 2607.11987
Anatomy-Privileged Distillation with Token Routing for MRI-Based Prediction of Perineural Invasion
基于解剖特征的蒸馏与标记路由用于MRI预测神经周围浸润
Abstract
Perineural invasion (PNI) is associated with poor postoperative outcomes in intrahepatic cholangiocarcinoma, but it is confirmed by surgical pathology. Existing preoperative imaging models often rely on radiologist-defined variables, contrast-enhanced imaging, or manual annotations. We propose an anatomy-privileged teacher--student framework for patient-level PNI prediction from T2-weighted MRI. During training, the teacher uses MRI with tumor and liver masks to learn dense token routing, and the student distills this guidance to retain and aggregate informative tokens under a fixed budget. Anatomical supervision is restricted to training, and the deployed model does not require masks at inference. In 155 patients, the proposed method achieved the highest mean AUROC of 0.750 among matched MRI-only baselines evaluated under the same protocol, with 1.43 GFLOPs and 8.02 ms per case on a Jetson Orin Nano Super Developer Kit.
Chinese Translation
神经周围浸润（PNI）与肝内胆管癌术后不良预后相关，但需通过外科病理学确认。现有的术前影像学模型通常依赖放射科医师定义的变量、对比增强影像或手动标注。我们提出了一种基于解剖特征的教师-学生框架，用于从T2加权MRI进行患者级PNI预测。在训练过程中，教师使用带有肿瘤和肝脏掩膜的MRI学习密集的标记路由，而学生则提炼这一指导，以在固定预算下保留和聚合有用的标记。解剖监督仅限于训练阶段，部署的模型在推理时不需要掩膜。在155名患者中，所提方法在相同协议下评估的匹配MRI-only基线中，达到了最高的平均AUROC值为0.750，且在Jetson Orin Nano超级开发套件上每个案例的计算量为1.43 GFLOPs，推理时间为8.02毫秒。
cs.CV / 7 / 2607.12000
MetaView: Monocular Novel View Synthesis with Scale-Aware Implicit Geometry Priors
MetaView：一种具有尺度感知隐式几何先验的单目新视角合成方法
Abstract
Current visual generation models are capable of producing high-quality content, yet they lack a coherent perception of the spatial structure. Existing generative novel view synthesis methods typically introduce explicit geometry priors, which enforce spatial consistency but inherently restrict generalization in large view changes. In contrast, recent interactive generative methods favor implicit scene modeling, offering greater flexibility at the cost of precise camera control and geometry consistency. In this paper, we propose MetaView, a diffusion-based monocular novel view synthesis framework that enables rendering under large view changes from a single image. Our key insight is to combine implicit geometry modeling with minimal yet essential explicit 3D cues: we incorporate implicit geometry priors from a feed-forward geometry perception network to regularize structure without imposing restrictive reconstruction pipelines, while leveraging metric depth to anchor the generation to a metric scale. This design allows MetaView to achieve both geometry consistency and precise controllability. Extensive experiments demonstrate that, under challenging monocular large viewpoint changes, MetaView significantly outperforms existing methods and exhibits superior generalization. Our code is publicly available at https://github.com/KlingAIResearch/MetaView.
Chinese Translation
当前的视觉生成模型能够生成高质量的内容，但缺乏对空间结构的连贯感知。现有的生成性新视角合成方法通常引入显式几何先验，以强制执行空间一致性，但在大视角变化中固有地限制了泛化能力。相比之下，最近的交互式生成方法更倾向于隐式场景建模，提供了更大的灵活性，但以牺牲精确的相机控制和几何一致性为代价。本文提出了MetaView，一种基于扩散的单目新视角合成框架，能够在单张图像下实现大视角变化的渲染。我们的关键见解是将隐式几何建模与最小但必要的显式3D线索相结合：我们从前馈几何感知网络中引入隐式几何先验，以规范结构而不施加限制性的重建流程，同时利用度量深度将生成锚定到度量尺度。该设计使MetaView能够实现几何一致性和精确可控性。大量实验表明，在具有挑战性的单目大视角变化下，MetaView显著优于现有方法，并展现出更好的泛化能力。我们的代码已公开发布在 https://github.com/KlingAIResearch/MetaView。
cs.CV / 8 / 2607.12042
SymbOmni: Evolving Agentic Omni Models via Symbolic Concept Learning
SymbOmni：通过符号概念学习演化的自主全能模型
Abstract
Visual generation is increasingly ubiquitous in diverse domains, from text-to-image/video synthesis to multimodal interactive creation. Yet prevailing monolithic models remain fundamentally constrained by their inability to learn cumulatively and evolve autonomously, which is a limitation we term the "perpetual novice" problem. They lack mechanisms for structuring experience into reusable knowledge and therefore rely on brittle, "from-scratch" reasoning for each task, resulting in poor compositional generalization and inefficient knowledge retention. Motivated by these limitations, we propose SymbOmni, an agentic omni-model designed for cumulative evolution through Symbolic Concept Learning. At its core is the Symbolic Concept Box, an optimizable memory module that abstracts low-level operations into reusable Symbolic Workflow Instructions. SymbOmni operates through an induction-transduction cycle: experiences are abstracted into symbolic concepts (induction), which are then adaptively composed to solve novel tasks (transduction). The training is done by verbalized backpropagation with language-based feedback to enable continuous self-improvement without gradient-based model fine-tuning. Comprehensive experiments validate that (I) SymbOmni significantly outperforms existing agent-based systems for iterative creation and also surpasses closed-source models (e.g., Nano Banana, GPT-Image-1) in both image quality and task success rates; (II) SymbOmni effectively reduces token consumption by over 40% while maintaining competitive generation quality; and (III) SymbOmni enables effective continual learning by achieving cumulative gains across multiple online-learning benchmarks and setting a new state of the art.
Chinese Translation
视觉生成在各种领域中日益普及，从文本到图像/视频合成到多模态互动创作。然而，现有的单一模型在累积学习和自主演化方面仍然受到根本性限制，这一限制我们称之为“永久新手”问题。它们缺乏将经验结构化为可重用知识的机制，因此在每个任务中依赖脆弱的“从头开始”推理，导致组合泛化能力差和知识保留效率低下。基于这些限制，我们提出了SymbOmni，一种旨在通过符号概念学习实现累积演化的自主全能模型。其核心是符号概念盒（Symbolic Concept Box），一个可优化的记忆模块，将低级操作抽象为可重用的符号工作流指令（Symbolic Workflow Instructions）。SymbOmni通过归纳-传导循环（induction-transduction cycle）运作：经验被抽象为符号概念（归纳），然后被自适应地组合以解决新任务（传导）。训练通过语言反馈的语言化反向传播进行，以实现持续的自我改进，而无需基于梯度的模型微调。全面的实验验证了以下几点：（I）SymbOmni在迭代创作方面显著优于现有的基于智能体的系统，并且在图像质量和任务成功率上超过了封闭源模型（如Nano Banana、GPT-Image-1）；（II）SymbOmni有效地将令牌消耗减少了超过40%，同时保持了竞争性的生成质量；（III）SymbOmni通过在多个在线学习基准上实现累积收益，推动了有效的持续学习，并设定了新的最先进水平。
cs.CV / 9 / 2607.12048
An Empirical Analysis of Continual Learning for Heterogeneous Medical Visual Question Answering
异构医学视觉问答的持续学习实证分析
Abstract
Deploying medical visual question answering (MedVQA) systems in real-world clinical settings requires models that adapt to new clinical tasks without forgetting previously acquired knowledge. Continual learning (CL) provides a practical framework for this setting. Despite rapid progress in medical vision-language models, the behavior of CL methods when training these models across heterogeneous MedVQA tasks remains underexplored. This work presents a systematic evaluation of CL for MedVQA across diverse clinical objectives, including classification, multi-label classification, detection, cell counting, and report generation. Specifically, we explore (1) the ability of existing CL methods to mitigate catastrophic forgetting; (2) their sensitivity to task ordering, analyzing how different task sequences influence performance retention and forgetting; and (3) the evolution of low-rank adaptation parameters as new tasks are learned, revealing patterns of weight drift under different CL methods. Our findings suggest that existing CL methods struggle to maintain stability-plasticity balance when tasks with different objectives and supervision formats are interleaved. Code and full experimental setup will be publicly available.
Chinese Translation
在真实临床环境中部署医学视觉问答（MedVQA）系统需要能够适应新临床任务而不遗忘先前获得知识的模型。持续学习（CL）为这种场景提供了一个实用框架。尽管医学视觉-语言模型取得了快速进展，但在异构MedVQA任务中训练这些模型时，CL方法的行为仍然未得到充分探索。本研究对MedVQA在多样化临床目标下的CL进行了系统评估，包括分类、多标签分类、检测、细胞计数和报告生成。具体而言，我们探讨了（1）现有CL方法减轻灾难性遗忘的能力；（2）它们对任务顺序的敏感性，分析不同任务序列如何影响性能保持和遗忘；以及（3）在学习新任务时低秩适应参数的演变，揭示了在不同CL方法下权重漂移的模式。我们的研究结果表明，现有CL方法在任务目标和监督格式不同的情况下交替时，难以维持稳定性与可塑性之间的平衡。代码和完整实验设置将公开发布。
cs.CV / 10 / 2607.12052
Representation and Reference Selection in Training-Free Synthetic Image Attribution
无训练合成图像归属中的表征与参考选择
Abstract
Synthetic image attribution aims at identifying the generator responsible for a given AI-generated image. Training-free reference-based attribution methods are easily scalable, since newly emerging generators can be incorporated by adding source-specific references rather than retraining a task-specific classifier. Their performance depends on two coupled factors: the representation space used for comparison and the way source-specific references are constructed. However, the interaction between these two factors remains largely unexplored. In this paper, we provide a controlled analysis of this interaction using references and off-the-shelf pretrained representations. We study representations extracted from different layers of CLIP and DINOv2, along with three reference selection methods with varying semantic constraints: arbitrary, semantically aligned, and resynthesis-based references. Our results show that attribution accuracy consistently peaks at intermediate representation levels, indicating that source-discriminative cues are more accessible before strong semantic abstraction dominates. We further show that intermediate representations are not completely semantically neutral, making reference selection critical: semantically constrained references reduce query-reference mismatch and improve attribution, especially under limited reference budgets. Resynthesis is most useful in low-reference regimes, while semantically aligned references provide a better accuracy-cost trade-off when a moderate-sized reference pool is available. Our findings show that training-free reference-based attribution should be understood as the interaction between where images are compared, how the reference set is constructed, and how many references are available.
Chinese Translation
合成图像归属旨在识别负责生成特定人工智能生成图像的生成器。无训练的基于参考的归属方法易于扩展，因为新出现的生成器可以通过添加特定源的参考来整合，而无需重新训练特定任务的分类器。它们的性能依赖于两个相互关联的因素：用于比较的表征空间和构建特定源参考的方式。然而，这两个因素之间的相互作用仍然在很大程度上未被探索。在本文中，我们使用参考和现成的预训练表征对这种相互作用进行了控制分析。我们研究了从 CLIP 和 DINOv2 的不同层提取的表征，以及三种具有不同语义约束的参考选择方法：任意参考、语义对齐参考和基于重合成的参考。我们的结果表明，归属准确性在中间表征层次上始终达到峰值，表明源区分线索在强语义抽象主导之前更易于获取。我们进一步表明，中间表征并非完全语义中立，使得参考选择至关重要：具有语义约束的参考减少了查询-参考的不匹配，并改善了归属，尤其是在参考预算有限的情况下。重合成在低参考情况下最为有效，而语义对齐参考在可用中等规模参考池时提供了更好的准确性-成本权衡。我们的研究结果表明，无训练的基于参考的归属应理解为图像比较的位置、参考集的构建方式以及可用参考数量之间的相互作用。
cs.CV / 11 / 2607.12062
Learning from Complementary Ultrasound Representations for Liver Disease Classification
利用互补超声表征进行肝病分类的研究
Abstract
Differentiating non-alcoholic steatohepatitis (NASH) from non-alcoholic fatty liver disease (NAFLD) using ultrasound remains challenging due to subtle tissue alterations and the limited information available in conventional B-mode imaging. In this work, we investigate whether complementary ultrasound representations derived from the same acquisition can improve NASH versus NAFLD classification. Specifically, we combine conventional B-mode ultrasound with physics-guided and local phase-based image representations and evaluate their effectiveness using self-supervised masked autoencoders (MAEs) and graph convolutional networks (GCNs). Experiments were conducted on a multi-site Mayo Clinic cohort consisting of 2,547 liver ultrasound scans from 125 patients. Compared with conventional B-mode ultrasound alone, complementary ultrasound representations consistently improved classification performance, yielding gains of up to 32.4% in accuracy and 91.2% in F1-score. Furthermore, performance improvements were consistently observed across age groups, sex, race, ethnicity,and acquisition sites.
Chinese Translation
由于组织细微变化和传统B模式成像中信息有限，使用超声区分非酒精性脂肪性肝炎（NASH）与非酒精性脂肪肝病（NAFLD）仍然具有挑战性。在本研究中，我们探讨了从同一采集中获得的互补超声表征是否能够改善NASH与NAFLD的分类。具体而言，我们将传统B模式超声与物理引导和局部相位基础的图像表征相结合，并使用自监督的掩码自编码器（MAEs）和图卷积网络（GCNs）评估其有效性。实验在一个包含125名患者的多中心梅奥诊所队列中进行，共有2,547个肝脏超声扫描。与单独使用传统B模式超声相比，互补超声表征在分类性能上始终有所提升，准确率提高了最高32.4%，F1-score提高了91.2%。此外，各年龄组、性别、种族、民族和采集地点的性能提升也得到了持续的观察。
cs.CV / 12 / 2607.12076
NEEDL-Bench: Dataset for Swiss Needle Cast and Stomata Detection in Microscopy Images
NEEDL-Bench：瑞士针锈病和显微镜图像中气孔检测的数据集
Abstract
We present NEEDL-Bench, a microscopy detection benchmark for Swiss Needle Cast (SNC), a fungal disease of Douglas-fir trees. Douglas-fir is a keystone species of major ecological and economic importance as a softwood timber resource, and SNC affects productivity by forming sexual reproductive structures (pseudothecia) that emerge through the gas exchange pores (stomata) of the needles, thereby blocking gas exchange and compromising needle function. To date, there is no dataset for automatic computer vision detection of these structures, despite computer vision being well poised to standardize and viably scale severity measurements. To address this, we present NEEDL-Bench, a dataset of 3250 annotated images from 1082 Douglas-fir needles, annotated for both keypoints and bounding-box detectors. This dataset exhibits a challenging collection of features, including blur, poor object contrast, small objects of interest, and occlusions. To better capture both the nominal distribution of the data and the full breadth of rare structures, we present two distinct evaluation splits: either random sampling from the collected images or sequential sampling to maximize structural diversity. We evaluate multiple popular keypoint and bounding box methods for detection on this dataset as a baseline and observe a maximum F1 score of 0.8479, suggesting significant potential for gains from future development on this problem. Further, we find that larger models generally do not show commensurate gains in performance on this dataset, indicating that improvements on this problem will not come from scaling laws but rather from domain-specific inductive biases.
Chinese Translation
我们提出了NEEDL-Bench，这是一个针对瑞士针锈病（SNC）的显微镜检测基准，SNC是一种影响道格拉斯冷杉树的真菌病害。道格拉斯冷杉作为一种重要的生态和经济物种，是软木材资源的关键物种，而SNC通过在针叶的气体交换孔（气孔）中形成性生殖结构（伪子实体）来影响生产力，从而阻碍气体交换并损害针叶功能。迄今为止，尽管计算机视觉在标准化和有效扩展严重性测量方面具有良好的潜力，但尚无用于自动计算机视觉检测这些结构的数据集。为了解决这个问题，我们提出了NEEDL-Bench，这是一个包含3250张来自1082根道格拉斯冷杉针叶的标注图像的数据集，标注了关键点和边界框检测器。该数据集展示了一系列具有挑战性的特征，包括模糊、物体对比度差、感兴趣的小物体和遮挡。为了更好地捕捉数据的名义分布和稀有结构的全貌，我们提供了两种不同的评估拆分：从收集的图像中随机抽样或顺序抽样以最大化结构多样性。我们评估了多种流行的关键点和边界框检测方法作为基线，并观察到最大F1得分为0.8479，表明在这一问题上未来发展有显著潜力。此外，我们发现较大的模型在该数据集上的性能通常没有相应的提升，这表明在这一问题上的改进不会来自于规模法则，而是来自于特定领域的归纳偏差。
cs.CV / 13 / 2607.12091
Causal Supervision of Attention for Affective Behaviour Analysis
情感行为分析中的因果监督注意力
Abstract
Affective Behaviour Analysis aims to enable machines to infer human affective states from behavioural signals, particularly facial expressions, in real-world environments. The \textit{11th Affective Behaviour Analysis in-the-wild Competition} includes the Multi-Task Learning Challenge based on the s-Aff-Wild2 database, where participants develop a unified framework for Valence-Arousal Estimation, Expression Recognition, and Action Unit Detection. This is challenging because emotion-related cues must be distinguished from spurious factors such as identity, illumination, pose, and demographic variation. Attention mechanisms are well suited as they aggregate information from the most informative facial regions, but may still exploit dataset-specific correlations instead of true affective cues. To improve generalization, we propose an attention pooling framework that promotes subject-invariant attention while increasing feature expressiveness. Our method consists of three components. First, we introduce causal supervision to enforce attention on facial regions with invariant predictive value across subjects. Second, we apply a cross-covariance independence regularization between Key (K) and Value (V) projections to encourage complementary, non-redundant representations. Finally, we replace the linear Value projection with a gated nonlinear SwiGLU transformation to increase feature expressiveness and capture finer-grained affective cues. Our method achieves $CCC_{VA}=0.5123$ for VA estimation on the official validation set, together with $F1_{EX}=0.3116$ and $F1_{AU}=0.3974$ for expression recognition and action unit detection, respectively, resulting in an overall $P$ score (the sum of the individual task metrics) of $1.2214$.
Chinese Translation
情感行为分析旨在使机器能够从行为信号，特别是面部表情中推断人类的情感状态，尤其是在真实环境中。第11届情感行为分析野外竞赛包括基于s-Aff-Wild2数据库的多任务学习挑战，参与者需要开发一个统一的框架来进行情感的愉悦度-激活度估计、表情识别和动作单元检测。这一任务具有挑战性，因为情感相关的线索必须与身份、光照、姿态和人口统计变化等虚假因素区分开来。注意力机制非常适合，因为它们能够聚合来自最具信息性的面部区域的信息，但仍可能利用数据集特定的相关性而非真实的情感线索。为了提高泛化能力，我们提出了一种注意力池化框架，促进主体不变的注意力，同时增强特征的表现力。我们的方法由三个部分组成。首先，我们引入因果监督，强制注意力集中在跨主体具有不变预测值的面部区域。其次，我们在关键（Key）和数值（Value）投影之间应用交叉协方差独立性正则化，以鼓励互补的、非冗余的表示。最后，我们用门控非线性SwiGLU变换替代线性数值投影，以提高特征的表现力并捕捉更细粒度的情感线索。我们的方法在官方验证集上实现了$CCC_{VA}=0.5123$的愉悦度-激活度估计，同时在表情识别和动作单元检测中分别达到了$F1_{EX}=0.3116$和$F1_{AU}=0.3974$，最终整体$P$分数（各个任务指标之和）为$1.2214$。
cs.CV / 14 / 2607.12110
ACZ-GSeg: Adaptive Concentric Zone-based Two-stage Ground Segmentation for LiDAR Point Clouds
ACZ-GSeg：基于自适应同心区的两阶段激光雷达点云地面分割
Abstract
Ground segmentation is a fundamental prerequisite for autonomous navigation, environmental perception, and object detection in ground mobile platforms. To address the under-segmentation of ground points caused by sparse long-range point clouds, ground undulations, and interference from non-ground structures in complex road scenarios, this paper proposes a two-stage ground segmentation method based on the Adaptive Concentric Zone Model. First, an Adaptive Concentric Zone Model is constructed to dynamically determine the number of sectors in each ring, thereby forming local zones with more balanced point distributions. Based on this model, a two-stage ground segmentation method is developed. In the coarse segmentation stage, a lowest-height seed constraint and height-decay weighting are introduced to establish a weighted principal component analysis plane fitting model, from which ground candidate points are extracted. In the fine segmentation stage, a reflectance intensity consistency constraint is employed to distinguish high-confidence ground points from uncertain points, and the uncertain points are further refined based on the local height stability of high-confidence neighborhoods. Experimental results show that the proposed method achieves Precision, Recall, and F1-score values of 99.12%, 96.24%, and 97.66% on the SemanticKITTI dataset, and 98.72%, 100.00%, and 99.36%, respectively, on a self-collected point cloud acquired using a RUBY-PLUS. The results demonstrate that the proposed method can effectively adapt to the range-dependent distribution characteristics of LiDAR point clouds, which are dense at near ranges and sparse at far ranges. It reduces the misclassification of non-ground points while maintaining ground point recall, thereby effectively improving the stability of ground segmentation.
Chinese Translation
地面分割是自主导航、环境感知和地面移动平台物体检测的基本前提。为了解决由于稀疏的长距离点云、地面起伏以及复杂道路场景中非地面结构干扰导致的地面点的欠分割问题，本文提出了一种基于自适应同心区模型的两阶段地面分割方法。首先，构建自适应同心区模型，以动态确定每个环中的扇区数量，从而形成点分布更均衡的局部区域。在此模型的基础上，开发了一种两阶段地面分割方法。在粗分割阶段，引入最低高度种子约束和高度衰减加权，建立加权主成分分析平面拟合模型，从中提取地面候选点。在精细分割阶段，采用反射强度一致性约束来区分高置信度地面点与不确定点，并根据高置信度邻域的局部高度稳定性进一步细化不确定点。实验结果表明，所提方法在SemanticKITTI数据集上实现了99.12%、96.24%和97.66%的精确率、召回率和F1分数，而在使用RUBY-PLUS采集的自收集点云上分别达到了98.72%、100.00%和99.36%。结果表明，所提方法能够有效适应激光雷达点云的范围依赖分布特征，即近距离密集而远距离稀疏。它在保持地面点召回的同时减少了非地面点的误分类，从而有效提高了地面分割的稳定性。
cs.CV / 15 / 2607.12165
Data Safety: Synthetic Data Quality Analysis Using CIFAKE Dataset
数据安全：基于CIFAKE数据集的合成数据质量分析
Abstract
Recently, the societal implementation of high-performance image classification models has expanded rapidly. While these models require vast amounts of training data to improve performance, securing sufficient real images is often impractical. As a means to compensate for this shortage, the use of synthetic data is becoming widespread. However, synthetic images are not necessarily equivalent to real images for training purposes. This study systematically analyzes the differences between two types of synthetic images created by different generation methods and real images from three perspectives: high-dimensional feature space, low-level statistics in color space, and the model training process. Furthermore, it experimentally verifies how synthetic data should be utilized by considering realistic data mixing scenarios. This enables the proposal of an evaluation and application strategy for performing preliminary assessments on synthetic images of unknown quality and safely incorporating them into training. This research aims to contribute to enhancing the reliability and safety of image classification models utilizing synthetic images.
Chinese Translation
近年来，高性能图像分类模型在社会应用中的推广迅速。虽然这些模型需要大量的训练数据以提高性能，但获取足够的真实图像往往是不切实际的。为弥补这一不足，合成数据的使用变得越来越普遍。然而，合成图像在训练目的上并不一定等同于真实图像。本研究从高维特征空间、色彩空间中的低级统计特征以及模型训练过程三个角度系统分析了由不同生成方法创建的两种合成图像与真实图像之间的差异。此外，通过考虑现实数据混合场景，实验验证了合成数据的使用方式。这使得我们能够提出一种评估和应用策略，以对未知质量的合成图像进行初步评估，并安全地将其纳入训练中。本研究旨在提升利用合成图像的图像分类模型的可靠性和安全性。
cs.CV / 16 / 2607.12171
Self-Consistent Flow: Unifying Velocity and Endpoint Prediction for Rectified Flow Models
自洽流：统一整流流模型的速度与端点预测
Abstract
In rectified-flow-based generative models, the neural network can be trained to predict two different targets, such as the instantaneous velocity or the data endpoint, to perform denoising. Although prior work shows that these parameterizations lead to different empirical behaviors, the mechanisms underlying their respective advantages remain to be underexplored, and how to combine them effectively is still unclear. In this work, we analyze how learning errors from different parameterizations affect the generation performance. We show that predicting the data endpoint has a clear training signal that stabilizes training, whereas predicting the velocity maintains stable sampling dynamics near the data manifold. Motivated by these insights, we propose Self-Consistent Flow (SC-Flow), a new method that unifies the benefits of both parameterizations. By employing a lightweight consistency loss, SC-Flow jointly trains a single network to predict both the local velocity and the data endpoint, and the consistency between the two predictions improves the model's performance. The method requires no major architectural changes and adds minimal computational overhead. Extensive experiments on image generation tasks demonstrate that SC-Flow substantially stabilizes optimization and improves the straightness of generation paths, leading to significant gains in generation quality over standard rectified-flow baselines.
Chinese Translation
在基于整流流的生成模型中，神经网络可以被训练来预测两个不同的目标，例如瞬时速度或数据端点，以执行去噪。尽管先前的研究表明这些参数化方法导致了不同的经验行为，但它们各自优势背后的机制仍然未被深入探讨，如何有效地将它们结合起来仍不明确。在本研究中，我们分析了来自不同参数化方法的学习误差如何影响生成性能。我们表明，预测数据端点具有明确的训练信号，这有助于稳定训练，而预测速度则在数据流形附近保持稳定的采样动态。基于这些见解，我们提出了自洽流（Self-Consistent Flow, SC-Flow），这是一种统一两种参数化方法优点的新方法。通过采用轻量级一致性损失，SC-Flow联合训练一个单一网络来同时预测局部速度和数据端点，两者预测之间的一致性提高了模型的性能。该方法不需要重大架构更改，并且增加的计算开销最小。在图像生成任务上的大量实验表明，SC-Flow显著稳定了优化过程，并改善了生成路径的直线性，导致生成质量相较于标准整流流基线有显著提升。
cs.CV / 17 / 2607.12175
From Reconstruction to Interpretation: Zero-Setup Multi-Phase Segmentation of X-ray Tomography Data
从重建到解释：零设置多相分割的X射线断层扫描数据
Abstract
X-ray tomography enables nondestructive characterization of material microstructures, while advances in micro-CT imaging have accelerated volumetric data acquisition and reconstruction. However, rapid interpretation remains limited by image segmentation, which often requires manual thresholding, user prompting, or material-specific model training. We present a zero-setup framework for multi-phase segmentation of synchrotron X-ray tomography data that generates interpretable masks for previously unseen datasets without user input or retraining during deployment. The framework combines a material-agnostic mask preparation strategy with a pretrained semantic segmentation network. It represents commonly occurring structural regions as background, sample, bright, dark-gray, light-gray, and porosity masks. Unlike conventional deep learning pipelines that require dataset-specific annotations and retraining, the proposed framework can be applied directly to new scans and produce diagnostic-level segmentations within minutes of reconstruction. This enables rapid assessment of scan quality, sample morphology, porosity, and attenuation variations during ongoing beamline experiments. The generated masks can later be manually refined or used to fine-tune application-specific models when greater accuracy or material-specific labeling is required. Evaluation on held-out synchrotron micro-CT images and qualitative testing on additional datasets demonstrate consistent and physically meaningful segmentations across varying samples and imaging conditions. The framework also substantially outperforms conventional intensity-based thresholding. By connecting high-speed reconstruction with immediate interpretation, the approach supports near-real-time beamline feedback and scalable AI-assisted scientific imaging workflows.
Chinese Translation
X射线断层扫描能够对材料微观结构进行非破坏性表征，而微型CT成像的进展加速了体积数据的获取和重建。然而，快速解释仍然受到图像分割的限制，这通常需要手动阈值设置、用户提示或特定材料模型的训练。我们提出了一种零设置框架，用于同步辐射X射线断层扫描数据的多相分割，该框架能够为之前未见的数据集生成可解释的掩膜，而无需用户输入或在部署期间进行重新训练。该框架结合了一种与材料无关的掩膜准备策略和一个预训练的语义分割网络。它将常见的结构区域表示为背景、样品、亮、深灰、浅灰和孔隙掩膜。与传统的深度学习流程需要特定数据集注释和重新训练不同，所提出的框架可以直接应用于新的扫描，并在重建后的几分钟内生成诊断级别的分割。这使得在进行光束线实验时能够快速评估扫描质量、样品形态、孔隙率和衰减变化。生成的掩膜可以在后续手动精细化，或在需要更高准确性或特定材料标注时用于微调应用特定的模型。在保留的同步辐射微型CT图像上的评估以及对其他数据集的定性测试表明，在不同样品和成像条件下，分割结果一致且具有物理意义。该框架在性能上也显著优于传统的基于强度的阈值分割。通过将高速重建与即时解释相结合，该方法支持近实时的光束线反馈和可扩展的AI辅助科学成像工作流程。
cs.CV / 18 / 2607.12176
A Calibrated Multimodal Ensemble for Ambivalence/Hesitancy Recognition: System Description and Private-Test Submission Strategy
用于模棱两可/犹豫识别的校准多模态集成：系统描述与私测提交策略
Abstract
Ambivalence and hesitancy (A/H) undermine digital behaviour-change interventions, and recognizing them automatically from video is the goal of the ABAW A/H challenge on the BAH dataset. We describe our system for the 11th edition of the challenge: a calibrated, equal-weight ensemble of three fusion models over frozen face, audio, text, and pose embeddings, which reaches 0.7358 macro-F1 on the public test set. This year's private test, released on a disjoint set of 30 new participants, is scored on five allowed submissions; we report the configuration and rationale of each of our five submissions, and, where already available, the private-test score obtained. Our first submission, an exact replica of the calibrated ensemble tuned only on public validation, scored 0.7361 macro-F1 on the private test, matching our public-test estimate almost exactly and confirming the pipeline generalizes to unseen participants without leakage.
Chinese Translation
模棱两可和犹豫（A/H）削弱了数字行为改变干预的效果，自动从视频中识别这些情绪是BAW数据集上ABAW A/H挑战的目标。我们描述了我们在挑战第11届中的系统：一个校准的、等权重的三种融合模型的集成，基于冻结的面部、音频、文本和姿态嵌入，在公共测试集上达到了0.7358的宏F1分数。今年的私测在一个不重叠的30名新参与者的集合上进行评分，允许提交五次；我们报告了每个提交的配置和理由，并在可用的情况下提供了获得的私测分数。我们的第一次提交是一个完全复制的校准集成，仅在公共验证上进行了调优，在私测中得到了0.7361的宏F1分数，几乎与我们的公共测试估计完全一致，确认了该流程在没有泄漏的情况下可以推广到未见过的参与者。
cs.CV / 19 / 2607.12186
Overview of Cross-Component In-loop Filters in Video Coding Standards
视频编码标准中跨组件环路滤波器的概述
Abstract
In-loop filters have been comprehensively explored during the development of video coding standards due to their remarkable noise-reduction capability. In the early stage of video coding, in-loop filters, such as Deblocking Filter, Sample Adaptive Offset, and Adaptive Loop Filter, were performed separately for each component. Recently, cross-component filters were studied to improve the chroma fidelity by exploiting correlations between the luma and chroma channels. This paper summarizes the cross-component filters used in the state-of-the-art video coding standard. Specifically, it includes the Cross-Component Adaptive Loop Filter and Cross-Component Sample Adaptive Offset. Cross-component filters aim to reduce compression artifacts based on the correlation between different components and provide more accurate pixel reconstruction values. In this paper, we introduce the origin, development, and status of cross-component filters in the current video coding standards. Finally, we had some discussions on the further evolutions of cross-component filters.
Chinese Translation
在视频编码标准的发展过程中，环路滤波器因其卓越的降噪能力而得到了全面的研究。在视频编码的早期阶段，环路滤波器（如去块滤波器、样本自适应偏移和自适应环路滤波器）是针对每个组件单独执行的。近年来，研究了跨组件滤波器，以利用亮度和色度通道之间的相关性来提高色度保真度。本文总结了当前最先进的视频编码标准中使用的跨组件滤波器，具体包括跨组件自适应环路滤波器（Cross-Component Adaptive Loop Filter）和跨组件样本自适应偏移（Cross-Component Sample Adaptive Offset）。跨组件滤波器旨在基于不同组件之间的相关性减少压缩伪影，并提供更准确的像素重建值。本文介绍了跨组件滤波器在当前视频编码标准中的起源、发展和现状。最后，我们对跨组件滤波器的进一步演变进行了讨论。
cs.CV / 20 / 2607.12206
RegHead: Non-Humanoid Head Blendshapes via Feed-Forward Registration
RegHead：通过前馈配准构建非类人头部混合形状
Abstract
We present RegHead, a framework for constructing semantic blendshape sets for animatable non-humanoid head avatars. With a fixed expression vocabulary, semantic blendshapes provide a low-dimensional and interpretable animation interface and support cross-identity retargeting. Building such blendshape sets remains expensive because (i) expression-consistent supervision is scarce, (ii) generated 4D assets typically lack correspondence, and (iii) facial motion is highly localized. We propose (1) a large-scale dataset of non-humanoid identities paired with a shared expression vocabulary, obtained by expanding a small artist-rigged library via fine-tuned image editing; (2) a dense stochastic anchor motion representation tailored to localized facial deformations; and (3) a fast feed-forward registration model that converts unregistered expression meshes into a corresponded blendshape basis by predicting anchor-based deformations from the neutral shape. Experiments show that our approach produces higher-fidelity expression meshes than baselines, while running orders of magnitude faster than optimization. We further demonstrate real-time retargeting from human face tracking signals to non-humanoid characters, capturing both head pose and localized facial motions. Our project page is available at https://snap-research.github.io/RegHead/.
Chinese Translation
我们提出了RegHead，一个用于构建可动画非类人头部头像的语义混合形状集的框架。通过固定的表情词汇，语义混合形状提供了一个低维且可解释的动画接口，并支持跨身份重定向。然而，构建这样的混合形状集仍然代价高昂，因为（i）表情一致的监督数据稀缺，（ii）生成的4D资产通常缺乏对应关系，以及（iii）面部运动高度局部化。我们提出了（1）一个大规模的非类人身份数据集，配备共享的表情词汇，通过对小型艺术家绑定库进行精细化图像编辑扩展而获得；（2）一种针对局部面部变形的密集随机锚点运动表示；以及（3）一个快速的前馈配准模型，该模型通过从中性形状预测基于锚点的变形，将未配准的表情网格转换为对应的混合形状基。实验表明，我们的方法生成的表情网格比基线具有更高的保真度，同时运行速度比优化快几个数量级。我们进一步展示了从人脸追踪信号到非类人角色的实时重定向，捕捉到头部姿态和局部面部运动。我们的项目页面可访问 https://snap-research.github.io/RegHead/。
cs.CV / 21 / 2607.12214
Beyond Perfect Priors: Adaptive Gaussian Graph for 4D Driving Reconstruction in the Wild
超越完美先验：用于野外4D驾驶重建的自适应高斯图
Abstract
Reconstructing 4D driving scenes in the wild (e.g., internet and AI-generated videos) is critical for diverse autonomous driving simulation. While recent Gaussian Scene Graph (GSG) methods achieve impressive visual quality, they heavily rely on precise priors, such as accurate camera poses and LiDAR depth, or manual annotations. When initialized with noisy priors estimated from in-the-wild videos, existing GSG methods suffer from optimization ambiguity (e.g., entangling camera and agent poses) and topological failures (e.g., missing objects), causing severe rendering artifacts. To enable robust in-the-wild reconstruction, we introduce Adaptive Gaussian Graph (AGG), a self-correcting 4D framework. Our Semantically-Guided Tick-Tock Strategy leverages 2D foundation features to explicitly decouple static background and camera pose updates from dynamic agent learning. Concurrently, our Adaptive Topology Evolution module actively rectifies graph structures by spawning missing agents, reassigning misclassified Gaussians, and pruning false positives. To rigorously evaluate this in-the-wild setting, we introduce Wild-30, a challenging benchmark of internet and generative videos. Extensive experiments on KITTI and Wild-30 validate that AGG consistently outperforms state-of-the-art approaches in visual fidelity and robustness under noisy priors.
Chinese Translation
在野外（例如互联网和AI生成的视频）重建4D驾驶场景对于多样化的自动驾驶模拟至关重要。尽管最近的高斯场景图（Gaussian Scene Graph, GSG）方法在视觉质量上取得了令人印象深刻的成果，但它们严重依赖于精确的先验信息，如准确的相机姿态和激光雷达深度，或手动标注。当使用从野外视频中估计的噪声先验进行初始化时，现有的GSG方法会遭遇优化模糊性（例如，相机和代理姿态的纠缠）和拓扑失败（例如，缺失物体），导致严重的渲染伪影。为了实现稳健的野外重建，我们引入了自适应高斯图（Adaptive Gaussian Graph, AGG），这是一种自我修正的4D框架。我们的语义引导滴答策略（Semantically-Guided Tick-Tock Strategy）利用2D基础特征，明确将静态背景和相机姿态更新与动态代理学习解耦。同时，我们的自适应拓扑演化模块通过生成缺失的代理、重新分配错误分类的高斯分布和修剪假阳性，主动修正图结构。为了严格评估这一野外设置，我们引入了Wild-30，这是一个具有挑战性的互联网和生成视频基准。对KITTI和Wild-30的广泛实验验证了AGG在视觉保真度和在噪声先验下的鲁棒性方面始终优于最先进的方法。
cs.CV / 22 / 2607.12231
The GEST-Engine: From Event Graphs to Synthetic Video. A Full Technical Report
GEST引擎：从事件图到合成视频。完整技术报告
Abstract
We present the GEST-Engine, a complete system that goes from natural-language text to fully-annotated multi-actor video. At its core is an explicit world model: rather than encoding state as a learned latent, the engine maintains a complete, inspectable representation of the world (which actors exist, where they are, what they are doing, which objects they hold, and how events relate in time and space), expressed as a formal Graph of Events in Space and Time (GEST) and realized deterministically inside the open world of a commercial game engine driven through an open-source multiplayer scripting framework. GESTs are produced either procedurally or by an agentic text-to-GEST system in which an LLM Director plans a story through tool calls validated by a programmatic state backend, so every generated specification is executable by construction. A GEST then enters a four-stage execution pipeline: graph parsing and validation, entity and action grounding, temporal orchestration (Allen-style constraints resolved by Floyd-Warshall transitive closure), and execution and capture. In a single simulation pass the engine emits frame-aligned RGB video, dense per-pixel depth, instance segmentation, per-actor skeletal pose, per-frame pairwise spatial-relation graphs, 2D bounding boxes, event-to-frame temporal mappings, and natural-language descriptions, all at zero marginal annotation cost. We further describe an in-game world editor, runtime capability extraction, a text-generation pipeline, and a production system that renders corpora at scale across parallel virtual machines. Because every frame traces back to a semantic specification, the engine guarantees object permanence, multi-actor coordination, and temporal consistency by construction, making its output valuable as training data, evaluation benchmarks, and diagnostic tools for video understanding.
Chinese Translation
我们提出了GEST引擎，这是一个完整的系统，可以将自然语言文本转换为完全标注的多演员视频。其核心是一个明确的世界模型：该引擎维护一个完整的、可检查的世界表示（包括存在的演员、他们的位置、他们的行为、他们持有的物体，以及事件在时间和空间上的关系），以空间和时间中的事件图（Graph of Events in Space and Time, GEST）的形式表达，并在通过开源多人脚本框架驱动的商业游戏引擎的开放世界中确定性地实现。GEST可以通过程序生成或由一个代理文本到GEST系统生成，其中一个大型语言模型（LLM）导演通过程序化状态后端验证的工具调用来规划故事，因此每个生成的规范都是可执行的。然后，GEST进入一个四阶段的执行管道：图解析与验证、实体和动作的基础定位、时间编排（通过Floyd-Warshall传递闭包解决的Allen风格约束）以及执行与捕获。在单次仿真过程中，该引擎生成与帧对齐的RGB视频、每像素密集深度、实例分割、每个演员的骨骼姿态、每帧的成对空间关系图、二维边界框、事件到帧的时间映射以及自然语言描述，所有这些都以零边际注释成本实现。我们进一步描述了一个游戏内世界编辑器、运行时能力提取、文本生成管道以及一个在平行虚拟机上大规模渲染语料库的生产系统。由于每一帧都可以追溯到语义规范，该引擎通过构建保证了物体的持久性、多演员协调和时间一致性，使其输出在视频理解的训练数据、评估基准和诊断工具方面具有重要价值。
cs.CV / 23 / 2607.12245
Rough Path Signature-Guided Geometry Augmentation for Few-Shot Industrial Surface Defect Detection
基于粗路径签名引导的几何增强用于少样本工业表面缺陷检测
Abstract
Few-shot industrial defect detection remains difficult for standard supervised detectors, which achieve poor performance on boundary-dominated industrial defects. This paper proposes rough path signature-guided geometry augmentation (RPS-GA), a geometry-aware approach in which Canny edge contours are treated as ordered planar paths whose truncated second-order signature responses, especially the antisymmetric L\'evy-area term, are aggregated into a spatial map that highlights boundary-related structure through two fusion operators, SIG-AUG and SGAA. The approach is evaluated on NEU-DET and PCB-Defect under a few-shot protocol with 5, 10, 20, or 50 labeled images per class, using an unmodified YOLOv8n detector throughout. Compared with the baseline, RPS-GA delivers large gains when supervision is limited, although the margin shrinks as more labels become available. On NEU-DET, SIG-AUG raises 10-shot
[email protected] from 0.341 to 0.583, whereas on PCB-Defect, SGAA improves 10-shot
[email protected] from 0.086 to 0.299 and yields usable detection at 5-shot where the baseline fails entirely. These trends are confirmed by multi-seed evaluation across independent random partitions. Overall, the results indicate that second-order path-signature geometry offers a practical way to strengthen few-shot industrial defect detection without meta-learning or detector redesign.
Chinese Translation
少样本工业缺陷检测对于标准监督检测器仍然困难，尤其是在边界主导的工业缺陷上表现不佳。本文提出了一种基于粗路径签名引导的几何增强方法（RPS-GA），这是一种关注几何特征的方法，其中Canny边缘轮廓被视为有序平面路径，其截断的二阶签名响应，特别是反对称的Lévy面积项，被聚合成一个空间图，突出显示与边界相关的结构，通过两个融合算子SIG-AUG和SGAA进行处理。该方法在NEU-DET和PCB-Defect数据集上进行评估，采用少样本协议，每类使用5、10、20或50张标记图像，并始终使用未修改的YOLOv8n检测器。与基线相比，当监督有限时，RPS-GA带来了显著的提升，尽管随着更多标签的可用性，增益幅度会缩小。在NEU-DET上，SIG-AUG将10-shot
[email protected]从0.341提高到0.583，而在PCB-Defect上，SGAA将10-shot
[email protected]从0.086提高到0.299，并在5-shot情况下实现可用检测，而基线完全失败。这些趋势通过在独立随机分区上的多种种子评估得到了确认。总体而言，结果表明，二阶路径签名几何提供了一种实用的方法来增强少样本工业缺陷检测，而无需元学习或检测器重设计。
cs.CV / 24 / 2607.12254
How to Realize Recursively Self-Improving Agents and Personal Singularity: A Goal-, Scope-, Tool-, and Benchmark-Driven Multi-Agent Architecture
如何实现递归自我改进的智能体与个人奇点：一个以目标、范围、工具和基准驱动的多智能体架构
Abstract
Large language model (LLM) agents can increasingly plan, use tools, maintain memory, and execute long-horizon tasks. These advances motivate two linked questions: how can an agent improve the mechanisms by which it learns and acts, and how can that improvement increase the durable capabilities of its user rather than only the software itself? This paper proposes a governed multi-agent architecture for recursively self-improving agents and introduces personal singularity as a bounded human-AI co-development objective: helping a user approach an expanding, user-defined capability frontier across selected domains. Each agent is defined by a goal contract, bounded scope, validated tool registry, tool-level tests, end-to-end benchmarks, an owner-controlled autonomy policy, a routing policy, memory, and an improvement policy. Out-of-scope tasks are transferred to another accountable agent or to a newly created niche agent. A user-facing Auto-Index selects interactive, hybrid, autonomous, or scheduled operation without overriding external permissions. The architecture combines a fast planner-executor-verifier loop, a slower evidence-gated improvement loop, an external governance plane, decentralized agent lineages, an owner-directed agent foundry, and a Personal Singularity OS coordinating working, computational-imaging, process-learning, and personal-learning agents. We formalize scope, routing, improvement acceptance, bounded goal evolution, tool-first execution, and human capability transfer, and provide safety invariants, benchmark design, and an implementation roadmap. This is a position and systems-design paper, not evidence that unrestricted recursive self-improvement or personal singularity has already been achieved.
Chinese Translation
大型语言模型（LLM）智能体越来越能够进行规划、使用工具、维护记忆并执行长时间跨度的任务。这些进展引发了两个相关的问题：智能体如何改进其学习和行动的机制，以及这种改进如何能够增强其用户的持久能力，而不仅仅是软件本身？本文提出了一种受管控的多智能体架构，用于递归自我改进的智能体，并引入个人奇点作为一个有限的人机共同发展目标：帮助用户在选定领域接近不断扩展的、用户定义的能力边界。每个智能体由目标契约、有限范围、经过验证的工具注册、工具级测试、端到端基准、由所有者控制的自主政策、路由政策、记忆和改进政策定义。超出范围的任务将转移给另一个负责的智能体或新创建的细分智能体。面向用户的自动索引选择交互式、混合、自主或定期操作，而不覆盖外部权限。该架构结合了快速的规划-执行-验证循环、较慢的证据门控改进循环、外部治理平面、去中心化的智能体谱系、由所有者指导的智能体铸造厂，以及协调工作、计算成像、过程学习和个人学习智能体的个人奇点操作系统。我们形式化了范围、路由、改进接受、有限目标演变、工具优先执行和人类能力转移，并提供安全不变性、基准设计和实施路线图。这是一篇立场和系统设计的论文，并不是证明无限制递归自我改进或个人奇点已经实现的证据。
cs.CV / 25 / 2607.12278
Auditing Data Leakage in Whole-Slide Image Multimodal Benchmarks
审计全幻灯片图像多模态基准中的数据泄漏
Abstract
Recent vision-language models (VLMs) for computational pathology report striking zero-shot performance on whole-slide image (WSI) visual question answering (VQA) benchmarks. We audit these claims and find them fundamentally compromised by data leakage at two hierarchical levels: patient-level leakage, where slides from the same case appear in both training and test folds, and institutional-level leakage, where different cases nonetheless share staining-batch and scanner signatures through a common Tissue Source Site (TSS). By tracing canonical slide, case, and TSS identifiers across major public resources, we document case level train test overlaps of 92.3~100% on TCGA-derived benchmarks, together with near-complete TSS overlap. We further demonstrate that both leakage levels are linearly decodable from foundation-model feature space, that they induce a measurable accuracy gap between leaked and audit-clean cases on a published checkpoint, and that across multiple published WSI VLMs, peak reported accuracies concentrate on the most heavily contaminated benchmarks. Therefore, the current WSI VQA evaluation cannot distinguish genuine multimodal reasoning from nearest-neighbor retrieval over memorized institutional and patient-specific artifacts. Finally, we outline concrete recommendations for contamination-free evaluation. By addressing benchmark construction, provenance disclosure, and automated overlap auditing, we aim to guide future research toward verifiable claims of progress.
Chinese Translation
近期的计算病理学视觉-语言模型（VLMs）在全幻灯片图像（WSI）视觉问答（VQA）基准上报告了显著的零样本性能。我们对这些声明进行了审计，发现它们在两个层次上受到数据泄漏的根本性影响：患者级别泄漏，即同一病例的幻灯片出现在训练集和测试集中的情况，以及机构级别泄漏，即不同病例通过共同的组织来源地点（Tissue Source Site, TSS）共享染色批次和扫描仪特征。通过追踪主要公共资源中的标准幻灯片、病例和TSS标识符，我们记录了在基于TCGA的基准上病例级别的训练测试重叠率高达92.3%~100%，并且几乎完全存在TSS重叠。我们进一步证明，这两种泄漏级别可以从基础模型特征空间线性解码，并且它们在发布的检查点上导致了泄漏病例与审计清洁病例之间可测量的准确性差距。此外，在多个发布的WSI VLM中，报告的最高准确率集中在污染最严重的基准上。因此，目前的WSI VQA评估无法区分真正的多模态推理与对记忆中的机构和患者特定伪影的最近邻检索。最后，我们提出了无污染评估的具体建议。通过解决基准构建、来源披露和自动重叠审计，我们旨在引导未来的研究朝着可验证的进展声明迈进。
cs.CV / 26 / 2607.12284
GeoSEAN: Explainable Country-Level Image Geolocation for ASEAN Regions
GeoSEAN：东盟地区可解释的国家级图像地理定位
Abstract
Image geolocation aims to infer the geographic origin of an image from visual content alone. However, this task remains challenging in regions where countries share similar urban, roadside, architectural, and environmental characteristics. Many existing geolocation models focus on coordinate level prediction or classification performance while providing limited insight into how visual evidence contributes to location predictions. This study presents an explainable country level image geolocation pipeline for 11 ASEAN countries. First, we collected 4,850 images from GeoGuessr style sources, Google Images, and additional street level imagery. We then evaluated three approaches on this dataset: CLIP zero shot classification, a LightGBM classifier, and an MLP classifier. The MLP achieved the best test performance, attaining an accuracy and F1 score of 85.91%. For explainability, predictions generated by the MLP classifier were analyzed post hoc using CLIP attention rollout, YOLO26 object detection on the original images, and Energy Based Pointing Game (EBPG) overlap metrics. Object level analysis indicates that frequently detected objects are not necessarily associated with the highest attention density, suggesting that object frequency and attention based visual evidence capture different aspects of a scene. These results demonstrate that the proposed model can support accurate regional image geolocation while enabling object level inspection of the visual cues underlying its predictions.
Chinese Translation
图像地理定位旨在仅通过视觉内容推断图像的地理来源。然而，在国家之间共享相似的城市、路边、建筑和环境特征的地区，这一任务仍然具有挑战性。许多现有的地理定位模型专注于坐标级别的预测或分类性能，而对视觉证据如何影响位置预测提供的见解有限。本研究提出了一种针对11个东盟国家的可解释国家级图像地理定位流程。首先，我们从GeoGuessr风格的来源、谷歌图片以及其他街景图像中收集了4850张图像。然后，我们在该数据集上评估了三种方法：CLIP零样本分类、LightGBM分类器和MLP分类器。MLP分类器取得了最佳测试性能，准确率和F1分数达到了85.91%。为了实现可解释性，使用CLIP注意力回滚、YOLO26对象检测和基于能量的指向游戏（EBPG）重叠指标对MLP分类器生成的预测进行了事后分析。对象级分析表明，频繁检测的对象不一定与最高的注意力密度相关，这表明对象频率和基于注意力的视觉证据捕捉场景的不同方面。这些结果表明，所提出的模型能够支持准确的区域图像地理定位，同时使得对其预测基础的视觉线索进行对象级检查成为可能。
cs.CV / 27 / 2607.12292
Semantic-Edge Response Decoding of SAM3 for Zero-Shot Crack Segmentation
SAM3的语义边缘响应解码用于零-shot裂缝分割
Abstract
Crack segmentation is essential for infrastructure inspection and structural health assessment, but existing high-performance methods typically require task-specific pixel-level annotations and training. Text-promptable vision foundation models enable zero-shot deployment, yet their final mask proposals are poorly suited to thin, fragmented, and low-contrast cracks, whose evidence may be suppressed, truncated, or over-expanded during mask generation. We find that language-conditioned semantic responses within the SAM3 decoder preserve more continuous and complete crack evidence than its final masks. Based on this observation, we propose Semantic-Edge Response Decoding (SERD), which interprets internal responses as a dense crack-likelihood field, calibrates them with a lightweight edge prior, and generates crack masks using a unified global threshold, without annotation or fine-tuning. Experiments on six public datasets show that SERD consistently improves over native SAM3 and outperforms the compared zero-shot and open-vocabulary segmentation methods, achieving an average Crack IoU of 61.14\%, 4.63 points higher than SAM3. Further analyses show that most gains arise from directly decoding internal semantic responses, while edge calibration improves structural recovery and false-positive control without increasing end-to-end inference overhead. These results suggest that, for thin and non-compact targets, internal continuous responses can provide a more transferable interface than the final masks of foundation models. Code is available at: https://github.com/xauat-liushipeng/SERD
Chinese Translation
裂缝分割对于基础设施检查和结构健康评估至关重要，但现有的高性能方法通常需要特定任务的像素级注释和训练。可通过文本提示的视觉基础模型实现零-shot部署，然而它们最终的掩膜提议对于细小、碎片化和低对比度的裂缝并不适用，这些裂缝的证据在掩膜生成过程中可能会被抑制、截断或过度扩展。我们发现，SAM3解码器中的语言条件语义响应比其最终掩膜保留了更多连续和完整的裂缝证据。基于这一观察，我们提出了语义边缘响应解码（Semantic-Edge Response Decoding, SERD），该方法将内部响应解释为密集的裂缝可能性场，使用轻量级边缘先验对其进行校准，并通过统一的全局阈值生成裂缝掩膜，无需注释或微调。在六个公共数据集上的实验表明，SERD在性能上始终优于原生SAM3，并且超越了比较的零-shot和开放词汇分割方法，平均裂缝交并比（Crack IoU）达到61.14%，比SAM3高出4.63分。进一步分析表明，大部分增益来自直接解码内部语义响应，而边缘校准则在不增加端到端推理开销的情况下改善了结构恢复和假阳性控制。这些结果表明，对于细小和非紧凑目标，内部连续响应可以提供比基础模型的最终掩膜更具可转移性的接口。代码可在以下链接获取：https://github.com/xauat-liushipeng/SERD
cs.CV / 28 / 2607.12293
Adaptive Cross-Modal Fusion with Sparse Attention for Pedestrian Crossing Intention Prediction
基于稀疏注意力的自适应跨模态融合用于行人过马路意图预测
Abstract
Predicting pedestrian crossing intention is a safety-critical task for autonomous driving, yet existing approaches often rely on single-modal inputs or dense multimodal fusion strategies that inadequately capture complementary visual and kinematic information while introducing redundant inter-modal interactions. We propose ADAPT (Adaptive Domain-Aware Pedestrian Crossing Transformer), a multimodal framework that jointly models local and global visual context together with temporal motion dynamics for accurate pedestrian crossing intention prediction. ADAPT processes four spatially aligned visual modalities, including RGB images, local depth maps, global semantic maps, and global depth maps, together with ego-vehicle speed, pedestrian bounding boxes, and skeleton pose information through five specialized modules: a weight-shared Swin Transformer V2 backbone for visual feature extraction, a Cross-Modality Guided Attention module for hierarchical visual fusion, a Mamba-based Motion Feature Encoding module for efficient temporal modeling, a Sparse Cross-Modal Attention module that selectively preserves the most informative inter-modal interactions, and a Vision Transformer-based Temporal Feature Fusion module for sequence-level prediction. Extensive experiments on the JAAD and PIE benchmark datasets demonstrate that ADAPT consistently outperforms existing state-of-the-art methods while maintaining low computational complexity. On JAAD, the proposed method achieves an AUC of 0.73 on JAADbeh and 0.85 on JAADall, while on PIE it achieves an accuracy of 0.92 and an AUC of 0.90. Furthermore, ADAPT performs inference in only 17.23 ms per sample, offering an effective balance between predictive accuracy and real-time deployment efficiency for intelligent transportation and autonomous driving applications.
Chinese Translation
预测行人过马路意图是自动驾驶中的一个安全关键任务，但现有方法往往依赖于单一模态输入或密集的多模态融合策略，这些策略无法充分捕捉互补的视觉和运动信息，同时引入冗余的跨模态交互。我们提出了ADAPT（自适应领域感知行人过马路变换器），这是一个多模态框架，能够联合建模局部和全局视觉上下文以及时间运动动态，以实现准确的行人过马路意图预测。ADAPT处理四种空间对齐的视觉模态，包括RGB图像、局部深度图、全局语义图和全局深度图，以及自车速度、行人边界框和骨架姿态信息，采用五个专门模块：一个用于视觉特征提取的权重共享Swin Transformer V2主干，一个用于分层视觉融合的跨模态引导注意力模块，一个基于Mamba的运动特征编码模块用于高效的时间建模，一个稀疏跨模态注意力模块选择性地保留最具信息量的跨模态交互，以及一个基于视觉变换器的时间特征融合模块用于序列级预测。在JAAD和PIE基准数据集上的广泛实验表明，ADAPT在保持低计算复杂度的同时，始终优于现有的最先进方法。在JAAD上，所提方法在JAADbeh上达到了0.73的AUC，在JAADall上达到了0.85，而在PIE上达到了0.92的准确率和0.90的AUC。此外，ADAPT每个样本的推理时间仅为17.23毫秒，为智能交通和自动驾驶应用提供了预测准确性与实时部署效率之间的有效平衡。
cs.CV / 29 / 2607.12297
MobileSAM2: Lightweight Segment Anything for Spatial Intelligence
MobileSAM2：轻量级的空间智能全能分割模型
Abstract
The recent large video foundation model, SAM2, enables segment anything in both images and videos, serving as a powerful base model for various applications. However, many of such use cases require to operate on resource-constrained devices like mobile phones and laptops. In this work, we aim to make SAM2 more mobile-friendly by distilling the heavyweight SAM2 into a lightweight model, facilitating segment anything in both images and videos on mobile devices. To this end, we propose Hypergraphical Knowledge Distill (HyperKD), which introduces the idea of hypergraph into knowledge distillation, aiming to effectively model and transfer SAM2's generalizable and comprehensive knowledge. HyperKD consists of Temporal HyperKD and Granularity HyperKD that construct hypergraphs to explicitly model and extract the generalizable temporal knowledge and the comprehensive multi-granularity knowledge from SAM2 respectively, which are then distilled into the lightweight student model by aligning it with the constructed hypergraphs. Besides, we present MobileSAM2, a new family of lightweight SAM2 that balances efficiency and effectiveness via searching the best model architectures with HyperKD during model size reduction. Extensive experiments validate MobileSAM2 across multiple benchmarks and show promising generalization performance on embodied AI tasks.
Chinese Translation
最近的大型视频基础模型SAM2能够在图像和视频中实现全能分割，为各种应用提供了强大的基础模型。然而，许多此类应用场景需要在资源受限的设备上运行，如手机和笔记本电脑。在本研究中，我们旨在通过将重量级的SAM2提炼为轻量级模型，使其更适合移动设备，从而在移动设备上实现图像和视频的全能分割。为此，我们提出了超图知识蒸馏（Hypergraphical Knowledge Distill, HyperKD），引入超图的概念用于知识蒸馏，旨在有效建模和传递SAM2的可泛化和全面知识。HyperKD由时间超图蒸馏（Temporal HyperKD）和粒度超图蒸馏（Granularity HyperKD）组成，分别构建超图以显式建模和提取SAM2的可泛化时间知识和全面的多粒度知识，然后通过与构建的超图对齐，将这些知识蒸馏到轻量级学生模型中。此外，我们提出了MobileSAM2，一个新的轻量级SAM2系列，通过在模型尺寸缩减过程中利用HyperKD搜索最佳模型架构，实现效率与效果的平衡。大量实验验证了MobileSAM2在多个基准测试中的表现，并在具身人工智能任务上显示出良好的泛化性能。
cs.CV / 30 / 2607.12304
What Does a Temporal Benchmark Score Measure? Decomposing Channel Use in Video VLM Evaluation
时间基准分数测量了什么？视频VLM评估中的通道使用解构
Abstract
A score on a temporal video question answering benchmark is meant to measure that a model has temporal understanding, but it conflates two questions. 1. The task question: is the question even temporal, does it need several frames and their order? and 2. The channel question, when it does, does the model recover the order from the pixels, or read it off the positional encoding (RoPE)? Most of a temporal score answers neither, a single frame and answer priors often carry it. The field's validity checks, frame-shuffle sensitivity and the accuracy gained from the full video, speak only to the task question. We contribute a label-free screen for the channel question, the reversal-drop: the accuracy lost when the visual sequence is reversed while RoPE remains forward. It can be applied to compatible temporal benchmarks without new annotations. Paired reverse labels, or tasks whose labels transform deterministically under reversal, distinguish models that follow reversed content from those merely disrupted by the conflict. Molmo2 answers the forward event reading order off positions, while Qwen3-VL answers the reversed event it actually sees, reading visual order (comparatively). We call them position-dominant and visual-sequence-dominant. The split holds across two benchmarks and several temporal tasks at two scales, and activation patching shows it is a real internal property, not an artifact of the conflict. The distinction matters, the two channels fail on opposite inputs so two models with similar score are not interchangable, i.e. an aggregate score does not reflect potential failure modes.
Chinese Translation
在时间视频问答基准上的得分旨在衡量模型的时间理解能力，但它混淆了两个问题。1. 任务问题：问题是否具有时间性，是否需要多个帧及其顺序？2. 通道问题：当问题具有时间性时，模型是否能够从像素中恢复顺序，或者是从位置编码（RoPE）中读取？大多数时间得分既没有回答这两个问题，单个帧和答案先验往往承载了它。该领域的有效性检查、帧打乱敏感性以及从完整视频中获得的准确性，仅涉及任务问题。我们贡献了一种无标签的通道问题筛选方法，即反转损失：当视觉序列被反转而RoPE保持前向时，损失的准确性。它可以应用于兼容的时间基准，而无需新的注释。配对反转标签或在反转下标签确定性转换的任务，可以区分遵循反转内容的模型与仅因冲突而受到干扰的模型。Molmo2从位置回答前向事件的阅读顺序，而Qwen3-VL回答它实际看到的反向事件，比较地读取视觉顺序。我们称它们为位置主导和视觉序列主导。这一划分在两个基准和多个时间任务的两个尺度上均成立，激活补丁显示这是一种真实的内部属性，而非冲突的伪影。这一区别很重要，因为两个通道在相反的输入上失败，因此两个得分相似的模型并不可互换，即聚合得分并不反映潜在的失败模式。
cs.CV / 31 / 2607.12319
DM-KG: A Novel Method for Boosting Spatial Cognition of Vision-Language Models in Street View Imagery
DM-KG：一种提升街景图像中视觉-语言模型空间认知的新方法
Abstract
As vision-language models (VLMs) are increasingly deployed in geospatial question answering and visual scene understanding, improving their spatial cognition capability on street view imagery for complex logical reasoning has emerged as a key research priority. However, existing VLMs frequently suffer from "spatial semantic hallucinations" when perceiving object locations, distances, and directions in real-world street view scenes. Furthermore, such errors are often recalcitrant to tracing and calibration, posing a critical bottleneck for their practical deployment in geospatial tasks. To address this pressing challenge, this study proposes DM-KG (Direction-Metric Knowledge Graph), a structurally grounded spatial representation framework for street view imagery. By explicitly extracting directional and metric relationships between entities from a single 2D image, this framework enhances the spatial reasoning accuracy of VLMs through a structured knowledge graph. Specifically, we integrate panoptic segmentation with metric depth estimation to robustly compute entity-level 3D spatial coordinates. Subsequently, we encode the clock azimuths and Euclidean distances of entity pairs into a JSON-formatted knowledge graph, which is injected into the VLM as an explicit geometric prior to guide spatial reasoning. Experimental results on public spatial question-answering (QA) benchmarks demonstrate that DM-KG reduces the mean absolute error (MAE) in distance estimation by 31.1% and the mean angular error in direction judgment by 65.8%, while simultaneously maintaining a high QA success rate. By establishing a complete, augmented reasoning pipeline, this research significantly improves the spatial cognitive capabilities of VLMs in street view scenarios, thereby providing a flexible, generalized, and interpretable framework for geographic visual question answering (GeoVQA) in open environments.
Chinese Translation
随着视觉-语言模型（VLMs）在地理空间问答和视觉场景理解中的广泛应用，提高其在街景图像中进行复杂逻辑推理的空间认知能力已成为关键研究重点。然而，现有的VLMs在感知真实世界街景场景中的物体位置、距离和方向时，常常遭遇“空间语义幻觉”的问题。此外，这类错误往往难以追踪和校正，成为其在地理空间任务中实际应用的重大瓶颈。为了解决这一紧迫挑战，本研究提出了DM-KG（方向-度量知识图谱），这是一个为街景图像设计的结构化空间表示框架。通过从单张二维图像中显式提取实体之间的方向和度量关系，该框架通过结构化知识图谱增强了VLMs的空间推理准确性。具体而言，我们将全景分割与度量深度估计相结合，稳健地计算实体级的三维空间坐标。随后，我们将实体对的时钟方位角和欧几里得距离编码为JSON格式的知识图谱，并将其作为显式几何先验注入到VLM中，以指导空间推理。在公共空间问答（QA）基准上的实验结果表明，DM-KG将距离估计的平均绝对误差（MAE）降低了31.1%，将方向判断的平均角度误差降低了65.8%，同时保持了较高的问答成功率。通过建立完整的增强推理管道，本研究显著提升了VLMs在街景场景中的空间认知能力，从而为开放环境中的地理视觉问答（GeoVQA）提供了一个灵活、通用且可解释的框架。
cs.CV / 32 / 2607.12335
ProtoPointNet: Prototype-Based Interpretable Classification of 3D Dental Point Clouds with Verifiable Spatial Activations
ProtoPointNet：基于原型的可解释3D牙科点云分类与可验证空间激活
Abstract
Prototype-based networks provide inherently interpretable classification by linking predictions to learned exemplars, but their use in 3D point clouds and clinical surface-pair reasoning remains limited. We introduce ProtoPointNet, a prototype-based model for dental occlusion classification from registered upper--lower intraoral arch pairs. Each point is encoded by a 14-dimensional descriptor combining local surface geometry, curvature, and explicit inter-arch displacement and clearance, exposing occlusal relationships to prototype matching. A shared multi-task point-cloud backbone learns axis-specific prototype heads for sagittal-left, sagittal-right, vertical, transverse, and midline classification. To support limited clinical data, we train prototypes from scratch using auxiliary supervision and encoder-freeze hand-off. On Bits2Bites, ProtoPointNet achieves mean test macro-F1 of 0.724 and AUROC of 0.825, with strongest performance on vertical (F1 0.828) and sagittal-left classification (F1 0.807). Projected prototype activations localise to anatomically plausible regions, including posterior molars and premolars for cross-bite evidence and anterior incisors for bite-depth evidence. These results support prototype-based reasoning as a transparent, spatially grounded alternative to black-box 3D classifiers for dental surface-pair analysis.
Chinese Translation
基于原型的网络通过将预测与学习到的示例关联，提供了固有的可解释分类，但其在3D点云和临床表面配对推理中的应用仍然有限。我们提出了ProtoPointNet，这是一种用于从注册的上下颌口内弓对进行牙齿咬合分类的基于原型的模型。每个点通过一个14维描述符进行编码，该描述符结合了局部表面几何、曲率以及明确的弓间位移和间隙，从而揭示了咬合关系与原型匹配之间的联系。一个共享的多任务点云主干网络学习特定轴的原型头，用于矢状面左、矢状面右、垂直、横向和中线分类。为了支持有限的临床数据，我们通过辅助监督和编码器冻结转移从零开始训练原型。在Bits2Bites数据集上，ProtoPointNet实现了0.724的平均测试宏F1和0.825的AUROC，在垂直分类（F1 0.828）和矢状面左分类（F1 0.807）上表现最佳。投影的原型激活定位于解剖上合理的区域，包括后磨牙和前磨牙用于交叉咬合证据，以及前切牙用于咬合深度证据。这些结果支持基于原型的推理作为一种透明的、基于空间的替代方案，用于牙科表面配对分析，取代黑箱3D分类器。
cs.CV / 33 / 2607.12352
Filtering-out poor-quality images for data preparation
过滤低质量图像以进行数据准备
Abstract
Filtering noise is a fundamental part of data preparation that enhances image quality for applications such as object segmentation, detection, and recognition. Various noise reduction techniques are proposed in the literature, including the use of median, Gaussian, and bilateral filters. Convolutional neural networks (CNNs) have gained popularity in image denoising owing to their ability to extract complex patterns and features from data. CNNs are highly adaptable, making them effective tools for various image-denoising tasks. One drawback of CNN-based techniques is that they require an appropriate training dataset and all images to be resized. Another notable drawback of all these filtering techniques is that they work for certain types of environmental and camera noises. To bridge this research gap, in this paper, for the first time, instead of denoising, we propose an approach that filters out poor-quality images for various environmental and camera impacts. In our approach, quality is assessed using an image quality assessment metric and an optimum threshold is used to filter out poor-quality images. We also ensure that a sufficient number of images remain to develop the deep learning (DL) model. The results produced using real and simulated traffic and object recognition data demonstrate the performance supremacy of the proposed approach compared with the state-of-the-art approaches. The average recognition accuracy for our proposed approach is 93.8% for the traffic sign recognition dataset and 84.9% for the object recognition dataset. This indicates our model's potential for real-life applications such as autonomous vehicles.
Chinese Translation
过滤噪声是数据准备的一个基本环节，它提升了图像质量，以便用于物体分割、检测和识别等应用。文献中提出了多种噪声减少技术，包括中值滤波、高斯滤波和双边滤波。卷积神经网络（CNN）因其从数据中提取复杂模式和特征的能力而在图像去噪中获得了广泛应用。CNN具有高度的适应性，使其成为各种图像去噪任务的有效工具。然而，基于CNN的技术的一个缺点是它们需要适当的训练数据集，并且所有图像都必须调整大小。所有这些过滤技术的另一个显著缺点是它们仅适用于某些类型的环境和相机噪声。为了解决这一研究空白，本文首次提出了一种方法，旨在过滤掉因各种环境和相机影响而导致的低质量图像，而不是进行去噪。在我们的方法中，使用图像质量评估指标来评估质量，并使用最佳阈值来过滤掉低质量图像。我们还确保保留足够数量的图像以开发深度学习（DL）模型。使用真实和模拟的交通及物体识别数据所产生的结果表明，与最先进的方法相比，所提方法在性能上具有优越性。我们提出的方法在交通标志识别数据集上的平均识别准确率为93.8%，在物体识别数据集上的平均识别准确率为84.9%。这表明我们的模型在自动驾驶等现实应用中的潜力。
cs.CV / 34 / 2607.12358
ACID: Adaptive Caching for vIDeo generation
ACID：视频生成的自适应缓存
Abstract
Video diffusion models produce high-quality generations but remain slow at inference due to their sequential denoising procedure. Caching-based acceleration methods address this by reusing intermediate model outputs: leading dynamic approaches such as TeaCache, EasyCache, and DiCache accumulate a drift signal and skip expensive model evaluations when accumulated drift stays below a fixed threshold {\tau}. This threshold controls an apparent tradeoff - raising it yields faster generation at the cost of visual quality, while lowering it preserves quality but sacrifices speed. We show this tradeoff is not fundamental; it is an artifact of holding {\tau} constant throughout denoising. We identify the existence of critical steps - timesteps where the drift signal changes rapidly - and show that applying a low threshold selectively at these steps while caching aggressively elsewhere recovers most of the quality of conservative caching at substantially higher inference speeds. Building on this insight, we propose ACID, a lightweight, training-free wrapper that monitors the rate of change of each method's existing drift signal to dynamically switch between a low and a high threshold. ACID is signal-agnostic and modular: it requires no retraining and plugs directly into existing dynamic caching methods without modifying their core mechanisms. Evaluated across three caching methods (TeaCache, EasyCache, DiCache) and three open-source video diffusion models (HunyuanVideo, Wan 2.1, CogVideoX), ACID consistently expands the Pareto frontier of visual quality versus inference speed beyond what any fixed threshold achieves. In particular, on TeaCache and HunyuanVideo, ACID achieves up to 2.16x speedup over the no-caching baseline, and up to 38% additional speedup over the conservative fixed-threshold baseline with negligible (<0.3 dB PSNR, <0.01 SSIM, <0.01 LPIPS) quality degradation.
Chinese Translation
视频扩散模型能够生成高质量的结果，但由于其顺序去噪过程，推理速度仍然较慢。基于缓存的加速方法通过重用中间模型输出来解决这一问题：领先的动态方法如 TeaCache、EasyCache 和 DiCache 在累积漂移信号时跳过昂贵的模型评估，当累积漂移保持在固定阈值 { au} 以下时。这一阈值控制着一种明显的权衡——提高阈值可以加快生成速度，但会牺牲视觉质量，而降低阈值则能保持质量，但会牺牲速度。我们表明，这种权衡并不是根本性的；它是由于在去噪过程中保持 { au} 不变所造成的伪影。我们识别出关键步骤的存在——漂移信号快速变化的时间步，并展示在这些步骤上选择性地应用低阈值，同时在其他地方积极缓存，可以在显著更高的推理速度下恢复大部分保守缓存的质量。基于这一见解，我们提出了 ACID，一种轻量级的、无训练的封装器，监测每种方法现有漂移信号的变化速率，以动态切换低阈值和高阈值。ACID 是信号无关的和模块化的：它无需重新训练，并且可以直接插入现有的动态缓存方法，而无需修改其核心机制。在三个缓存方法（TeaCache、EasyCache、DiCache）和三个开源视频扩散模型（HunyuanVideo、Wan 2.1、CogVideoX）上的评估表明，ACID 一致地扩展了视觉质量与推理速度之间的帕累托前沿，超出了任何固定阈值所能实现的范围。特别是在 TeaCache 和 HunyuanVideo 上，ACID 实现了相较于无缓存基线高达 2.16 倍的加速，并在保守固定阈值基线之上实现了高达 38% 的额外加速，同时质量下降微乎其微（<0.3 dB PSNR，<0.01 SSIM，<0.01 LPIPS）。
cs.CV / 35 / 2607.12362
Implicit 4D Gaussian Splatting for Fast Motion with Large Inter-Frame Displacements
隐式四维高斯点云快速运动与大帧间位移的处理
Abstract
Recent 4D Gaussian Splatting (4DGS) methods often fail under fast motion with large inter-frame displacements, where Gaussian attributes are poorly learned during training, and fast-moving objects are often lost from the reconstruction. In this work, we introduce Spatiotemporal Position Implicit Network for 4DGS, coined SPIN-4DGS, which learns Gaussian attributes from explicitly collected spatiotemporal positions rather than modeling temporal displacements, thereby enabling more faithful splatting under fast motions with large inter-frame displacements. To avoid the heavy memory overhead of explicitly optimizing attributes across all spatiotemporal positions, we instead predict them with a lightweight feed-forward network trained under a rasterization-based reconstruction loss. Consequently, SPIN-4DGS learns shared representations across Gaussians, effectively capturing spatiotemporal consistency and enabling stable high-quality Gaussian splatting even under challenging motions. Across extensive experiments, SPIN-4DGS consistently achieves higher fidelity under large displacements, with clear improvements in PSNR and SSIM on challenging sports scenes from the CMU Panoptic dataset. For example, SPIN-4DGS notably outperforms the strongest baseline, D3DGS, by achieving +1.83 higher PSNR on the Basketball scene.
Chinese Translation
近期的四维高斯点云（4D Gaussian Splatting, 4DGS）方法在快速运动和大帧间位移的情况下常常表现不佳，导致高斯属性在训练过程中学习不充分，快速移动的物体在重建中经常丢失。在本研究中，我们提出了一种用于4DGS的时空位置隐式网络，命名为SPIN-4DGS，该网络从明确收集的时空位置中学习高斯属性，而不是建模时间位移，从而在快速运动和大帧间位移的情况下实现更真实的点云重建。为了避免在所有时空位置上显式优化属性所带来的高内存开销，我们采用轻量级前馈网络进行预测，并在基于光栅化的重建损失下进行训练。因此，SPIN-4DGS能够在高斯之间学习共享表示，有效捕捉时空一致性，即使在复杂运动下也能够实现稳定的高质量高斯点云重建。在广泛的实验中，SPIN-4DGS在大位移情况下始终实现更高的保真度，在CMU Panoptic数据集的复杂运动场景中，PSNR和SSIM均有明显提升。例如，SPIN-4DGS在篮球场景中显著超越最强基线D3DGS，PSNR提高了1.83。
cs.CV / 36 / 2607.12364
Lost in Visual Translation: A VLM-Assisted Perceptual-Semantic Coherence Framework for EEG-to-Image Reconstruction
迷失在视觉翻译中：一种基于视觉语言模型的感知-语义一致性框架用于脑电图到图像重建
Abstract
EEG-to-image evaluation should distinguish visual fidelity from recoverable meaning. Yet EEG-derived reconstructions are blurry, distorted, and low-detail, causing SSIM, LPIPS, and CLIP to penalize semantically recoverable outputs or reward plausible but incorrect ones. We analyze 6,855 ground-truth/reconstruction pairs from ATM, ENIGMA, BrainVis, and DreamDiffusion using semantic probes, caption harshness and blind-spot rates, and controlled degradations. Pixel metrics show near-zero correlation with semantic consistency, while representation metrics conflate perceptual and semantic errors. We therefore introduce a BCI-aware framework in which four VLMs assess image pairs through structured questions, producing Tolerant Perceptual Alignment Scores (T-PAS) and Tolerant Semantic Alignment Scores (T-SAS). Their consensus is distilled into the BCI-Coherence Score (BCS), a compact evaluator achieving a T-PAS MAE of 0.079 (r = 0.700) and a T-SAS MAE of 0.082 (r = 0.850) on our data. Human validation shows highly reliable joint coherence judgments, with Cohen's kappa = 0.882 +/- 0.174 and Krippendorff's alpha = 0.882, supporting perceptual-semantic recoverability over generic visual similarity. Code and resources are available at https://sukt03.github.io/BCS/.
Chinese Translation
脑电图到图像的评估应区分视觉保真度与可恢复的意义。然而，脑电图衍生的重建图像模糊、失真且细节不足，导致结构相似性指数（SSIM）、感知图像相似性（LPIPS）和对比语言图像预训练（CLIP）对可语义恢复的输出进行惩罚，或奖励看似合理但不正确的结果。我们分析了来自ATM、ENIGMA、BrainVis和DreamDiffusion的6,855对真实/重建图像，使用语义探测器、标题严苛度和盲点率以及控制降质。像素度量与语义一致性几乎没有相关性，而表征度量则混淆了感知和语义错误。因此，我们引入了一种脑机接口（BCI）感知的框架，其中四个视觉语言模型（VLM）通过结构化问题评估图像对，生成宽容感知对齐分数（T-PAS）和宽容语义对齐分数（T-SAS）。它们的共识被提炼为BCI一致性评分（BCS），这是一个紧凑的评估器，在我们的数据上实现了T-PAS平均绝对误差（MAE）为0.079（r = 0.700）和T-SAS MAE为0.082（r = 0.850）。人类验证显示高度可靠的联合一致性判断，Cohen's kappa = 0.882 +/- 0.174，Krippendorff's alpha = 0.882，支持感知-语义可恢复性优于一般视觉相似性。代码和资源可在https://sukt03.github.io/BCS/获取。
cs.CV / 37 / 2607.12372
UMSS: Towards Unsupervised Multi-modal Semantic Segmentation
UMSS：面向无监督多模态语义分割
Abstract
Multimodal semantic segmentation (MSS) is essential for robust perception in complex environments, yet its potential remains largely untapped because of the prohibitive cost of human annotations. While unsupervised semantic segmentation (USS) has achieved strong results on a single RGB modality, its naive extension to multimodal data is often hindered by fusion degradation. This occurs because, without explicit supervision, existing frameworks struggle to reconcile the heterogeneous structural patterns captured by different sensors and therefore fail to effectively exploit their complementary information. In this paper, we make the first attempt to address the novel problem of Unsupervised Multimodal Semantic Segmentation (UMSS), aiming to effectively exploit complementary sensor information in a fully label free setting. To this end, we propose UniM2 (Unified Multimodal), a novel framework built on DINOv3 that transforms conventional fusion methods into consistent performance gains. Our key idea is to learn a unified latent space driven by Cross Modal Correspondence Synergy (CMCS) to extract intrinsic shared semantic cues, bypassing the need for label guided adaptive fusion. To mitigate inherent intermodal conflicts, we introduce a Cross Modal Harmonizer (CMH) that designates RGB as a stable reference, effectively suppressing inconsistent relational supervision while guiding the model to exploit complementary structural features. Extensive experimental results on NYU Depth v2 and MFNet show that UniM2 improves mIoU by 6.4% and 9.8%, respectively, demonstrating clear advantages over existing frameworks for UMSS.
Chinese Translation
多模态语义分割（MSS）对于在复杂环境中实现稳健感知至关重要，但由于人工标注的高昂成本，其潜力仍未得到充分挖掘。尽管无监督语义分割（USS）在单一RGB模态上取得了良好的结果，但其在多模态数据上的简单扩展常常受到融合降级的阻碍。这是因为在没有明确监督的情况下，现有框架难以调和不同传感器捕获的异质结构模式，因此未能有效利用其互补信息。本文首次尝试解决无监督多模态语义分割（UMSS）的新问题，旨在在完全无标签的环境中有效利用互补传感器信息。为此，我们提出了UniM2（统一多模态），这是一个基于DINOv3构建的新框架，将传统的融合方法转变为一致的性能提升。我们的关键思想是通过跨模态对应协同（CMCS）学习一个统一的潜在空间，以提取内在共享的语义线索，从而绕过标签引导的自适应融合需求。为了缓解固有的模态间冲突，我们引入了跨模态协调器（CMH），将RGB指定为稳定参考，有效抑制不一致的关系监督，同时引导模型利用互补的结构特征。在NYU Depth v2和MFNet上的大量实验结果表明，UniM2分别提高了mIoU 6.4%和9.8%，展示了其在UMSS领域相较于现有框架的明显优势。
cs.CV / 38 / 2607.12375
IQA-T1: Tool-based Visual Evidence Reasoning for Image Quality Assessment
IQA-T1：基于工具的视觉证据推理用于图像质量评估
Abstract
Image Quality Assessment (IQA) in open-world environments remains challenging due to limited generalization and interpretability. Recent approaches based on multimodal large language models (MLLMs) introduce textual reasoning for quality prediction, yet their judgments rely heavily on semantically biased internal representations, making them insensitive to low-level perceptual degradations. We propose IQA-T1, a tool-based visual evidence reasoning framework that augments MLLM reasoning with explicit perceptual observations. During inference, the model autonomously invokes specialized analysis tools to generate structured visual evidence, such as noise residual maps, gradient statistics, and frequency spectra, which are progressively integrated into the reasoning process. To support this paradigm, we construct Q-Tool, a dataset containing 11k multimodal reasoning chains grounded in tool-generated evidence. Extensive experiments on seven IQA benchmarks show that IQA-T1 achieves the best overall performance across datasets while producing interpretable and evidence-grounded quality assessments. Code and dataset are available at https://github.com/zibuyu-02/IQA-T1.
Chinese Translation
在开放世界环境中，图像质量评估（IQA）仍然面临挑战，主要由于其有限的泛化能力和可解释性。近期基于多模态大型语言模型（MLLMs）的方法引入了文本推理用于质量预测，但其判断在很大程度上依赖于语义偏向的内部表征，使其对低级感知退化不敏感。我们提出了IQA-T1，一种基于工具的视觉证据推理框架，通过显式的感知观察增强了MLLM的推理能力。在推理过程中，模型自主调用专门的分析工具生成结构化的视觉证据，如噪声残差图、梯度统计和频谱，这些证据逐步融入推理过程。为支持这一范式，我们构建了Q-Tool，一个包含11,000个基于工具生成证据的多模态推理链的数据集。在七个IQA基准上的广泛实验表明，IQA-T1在各数据集上实现了最佳的整体性能，同时产生了可解释且基于证据的质量评估。代码和数据集可在https://github.com/zibuyu-02/IQA-T1获取。
cs.CV / 39 / 2607.12376
Demonstration of the common dual-channel feature decoupling characteristic of front-door mediation causal inference methods in whole-slice image classification
前门干预因果推断方法在全切片图像分类中的共同双通道特征解耦特性展示
Abstract
Causal inference using front door intervention and multi-instance learning (MIL) has advanced the analysis of Whole Slide Images (WSI) in digital pathology. These methods adjust feature distributions of subtle evidence sub-images to correctly associate them with WSI-level diagnoses. We propose and prove 2 hypotheses for evaluating such methods: 1) Causal inference MIL introduces an independent classification channel that effectively completes WSI classification; 2) Greater difference between features extracted by the new and baseline channels increases effectiveness in eliminating false correlations. This hypothesis describes the core of causal inference MILs: overlaying parallel, independent channels to eliminate false associations between WSI-level diagnostic and non-diagnostic evidence sub-images by increasing deep feature diversity. Based on these hypotheses, we evaluated several causal inference MILs on breast cancer and non-small cell lung cancer datasets. This hypothesis provides a new theoretical perspective for applying causal inference to WSI analysis.
Chinese Translation
使用前门干预和多实例学习（MIL）的因果推断推动了数字病理学中全切片图像（WSI）分析的发展。这些方法调整微妙证据子图像的特征分布，以正确地将其与WSI级别的诊断关联起来。我们提出并证明了两个用于评估此类方法的假设：1）因果推断MIL引入了一个独立的分类通道，有效地完成WSI分类；2）新通道和基线通道提取的特征之间的差异越大，消除虚假相关性的效果越显著。该假设描述了因果推断MIL的核心：通过增加深层特征的多样性，叠加平行的独立通道，以消除WSI级别诊断与非诊断证据子图像之间的虚假关联。基于这些假设，我们在乳腺癌和非小细胞肺癌数据集上评估了几种因果推断MIL。这一假设为将因果推断应用于WSI分析提供了新的理论视角。
cs.CV / 40 / 2607.12379
SeamGen: Artist-Aligned UV Seam Generation via Graph Flow Matching
SeamGen：通过图流匹配实现艺术家对齐的UV接缝生成
Abstract
UV seam placement is a critical yet labor-intensive step in 3D content creation, requiring artists to balance chart shape, seam concealment, and alignment with semantic and geometric features. Existing automatic methods are primarily based on per-object optimization, relying on handcrafted objectives to avoid distortion or on proxies from pretrained models to inject semantic information. However, these strategies are not always well aligned with seams used in industrial production pipelines, often resulting in layouts that deviate from artist-preferred seam patterns and practical production requirements. To address these limitations, we propose SeamGen, a generative model for UV seam generation that aligns with artist preferences and production requirements. Instead of depending on manually designed objectives and constraints, SeamGen learns the distribution of per-edge seam labels from a large corpus of existing seam layouts using a flow-matching generative model. A key challenge is that typical Transformer architectures used in flow matching models are designed for sequential representations, such as point clouds, and cannot naturally account for mesh topology. To enable mesh-native learning, we design a Mesh Transformer backbone that interleaves local graph attention over mesh edges with global self-attention across vertices, capturing both fine-grained geometric cues and long-range topological coherence. To further improve inference-time controllability and quality, we exploit the training-free inpainting capability of flow models for both localized seam refinement and constraint-guided seam generation. Extensive experiments show that by learning priors from professional seam layout data, SeamGen produces UV layouts that better align with artist-authored preferences and achieve superior perceptual quality compared with distortion-based and semantic-proxy baselines.
Chinese Translation
UV接缝的放置是3D内容创作中一个关键但劳动密集的步骤，要求艺术家在图表形状、接缝隐蔽性以及与语义和几何特征的对齐之间取得平衡。现有的自动化方法主要基于每个对象的优化，依赖于手工设计的目标以避免失真，或使用预训练模型的代理来注入语义信息。然而，这些策略并不总是与工业生产流程中使用的接缝良好对齐，常常导致布局偏离艺术家偏好的接缝模式和实际生产要求。为了解决这些局限性，我们提出了SeamGen，这是一种与艺术家偏好和生产要求对齐的UV接缝生成生成模型。SeamGen不依赖于手动设计的目标和约束，而是通过流匹配生成模型从大量现有接缝布局中学习每条边接缝标签的分布。一个关键挑战是，流匹配模型中使用的典型Transformer架构是为序列表示（如点云）设计的，无法自然考虑网格拓扑。为了实现网格本地学习，我们设计了一个Mesh Transformer主干，它在网格边缘上交错局部图注意力与顶点之间的全局自注意力，捕捉细粒度的几何线索和长程拓扑一致性。为了进一步提高推理时的可控性和质量，我们利用流模型的无训练修补能力进行局部接缝细化和约束引导的接缝生成。大量实验表明，通过从专业接缝布局数据中学习先验，SeamGen生成的UV布局更好地与艺术家创作的偏好对齐，并且在感知质量上优于基于失真和语义代理的基线。
cs.CV / 41 / 2607.12398
Seeing Globally, Refining Locally: Global Visual Guidance and Local Ultrasound Cues for Robust Freehand 3-D Ultrasound Reconstruction
全球视野，局部精炼：全球视觉引导与局部超声线索结合的稳健自由手3D超声重建
Abstract
Freehand 3-D ultrasound (US) imaging has attracted increasing attention owing to its intuitive volumetric visualization, ease of use, and low cost. However, accurate 3-D reconstruction critically depends on stable probe pose estimation, yet existing trackerless methods remain susceptible to accumulated pose errors, particularly over long scanning trajectories. To address this limitation, we propose a global-to-local pose estimation framework that exploits external camera observations for globally stable localization and B-mode US images for anatomy-aware local refinement. Specifically, the framework comprises a dual-camera branch that performs contextual feature aggregation across camera views and temporal observations to estimate a globally consistent probe trajectory, and a B-mode branch that performs anatomical feature aggregation from sequential US images to capture tissue-dependent local motion cues. A cross-modal fusion module subsequently integrates the contextual camera features and anatomical US features to predict pose residuals and refine the camera-derived estimates in the transformation space. Furthermore, a multi-scale pose loss constrains relative motion over multiple temporal horizons to suppress accumulated drift during extended scans. The proposed framework is validated on phantom and in vivo datasets. On two in-house datasets (FUSION-J and FUSION-L) collected using different machines, the proposed US + Dual-Cam model reduces average trajectory drift to 1.67 mm and 1.29 mm, representing improvement of 16.50% and 27.12%, respectively, over a strong dual-camera baseline, while substantially outperforming US-only pose estimation (>13 mm drift). In in vivo forearm arteries reconstruction, it achieves Hausdorff distances of 1.58 mm, demonstrating the effectiveness of the proposed method on real clinical scenarios.
Chinese Translation
自由手3D超声（US）成像因其直观的体积可视化、易用性和低成本而受到越来越多的关注。然而，准确的3D重建在很大程度上依赖于稳定的探头姿态估计，而现有的无追踪方法在长扫描轨迹上容易受到累积姿态误差的影响。为了解决这一限制，我们提出了一种全球到局部的姿态估计框架，该框架利用外部相机观测进行全球稳定定位，并利用B模式超声图像进行解剖学感知的局部精炼。具体而言，该框架包括一个双相机分支，通过相机视角和时间观测之间的上下文特征聚合来估计全球一致的探头轨迹，以及一个B模式分支，从连续的超声图像中进行解剖特征聚合，以捕捉组织依赖的局部运动线索。随后，一个跨模态融合模块将上下文相机特征和解剖超声特征整合，以预测姿态残差并在变换空间中精炼相机导出的估计。此外，多尺度姿态损失在多个时间范围内约束相对运动，以抑制在长时间扫描过程中累积的漂移。我们在仿真和体内数据集上验证了所提出的框架。在使用不同机器收集的两个内部数据集（FUSION-J和FUSION-L）上，所提出的US + Dual-Cam模型将平均轨迹漂移降低至1.67 mm和1.29 mm，分别比强大的双相机基线提高了16.50%和27.12%，同时在超声单独姿态估计中表现出显著优越性（>13 mm漂移）。在体内前臂动脉重建中，该方法实现了1.58 mm的Hausdorff距离，证明了所提出方法在真实临床场景中的有效性。
cs.CV / 42 / 2607.12399
Physically Aware Radiomics Without Interpolation: Disentangling Voxel Geometry and Signal Modification in CT and MRI
无插值的物理感知放射组学：解构CT和MRI中的体素几何与信号修改
Abstract
Objective: Radiomic texture features are usually computed in voxel-index neighborhoods, implicitly assuming isotropic spatial relationships. In anisotropic images, this can confound voxel geometry with interpolation-induced signal changes. We developed a voxel-spacing-aware radiomic framework that incorporates physical geometry into texture computation without resampling. Approach: We modified PyRadiomics to account for voxel spacing while preserving the native image signal. Four configurations were compared: native non-resampled extraction (NR), isotropic resampling (RS), voxel-spacing-aware extraction (VS), and fake-isotropic preprocessing (FK), in which spacing metadata were overwritten without altering the image array. Experiments included 685 LIDC-IDRI pulmonary nodules and 209 I-SPY2 breast MRI cases, with 196 radiomic descriptors. Robustness was assessed using ICC, within-subject variability, Friedman testing, feature selection, machine learning, a multilayer perceptron, and external validation. Main results: VS showed near-native agreement with NR: median ICC(A,1) was 0.9976 in CT and 0.9984 in MRI. RS produced lower agreement and larger deviations, while FK showed intermediate behavior, confirming that spacing metadata alone can affect radiomic features. Gradient-derived and neighborhood-sensitive descriptors were most affected by preprocessing. VS preserved predictive performance comparable to NR in external CT validation, whereas MRI showed greater variability across preprocessing strategies and classifiers. Significance: Voxel-spacing-aware extraction separates geometric modeling from interpolation-induced signal modification while preserving the native image signal, offering a coherent alternative to isotropic resampling for radiomic analysis of anisotropic CT and MRI.
Chinese Translation
目的：放射组学纹理特征通常在体素索引邻域中计算，隐含假设各向同性空间关系。在非各向同性图像中，这可能会将体素几何与插值引起的信号变化混淆。我们开发了一种体素间距感知的放射组学框架，能够在不重新采样的情况下将物理几何纳入纹理计算。方法：我们修改了PyRadiomics，以考虑体素间距，同时保留原始图像信号。比较了四种配置：原始非重采样提取（NR）、各向同性重采样（RS）、体素间距感知提取（VS）和伪各向同性预处理（FK），其中间距元数据被覆盖而不改变图像数组。实验包括685个LIDC-IDRI肺结节和209个I-SPY2乳腺MRI案例，共196个放射组学描述符。通过ICC、组内变异性、Friedman检验、特征选择、机器学习、多层感知器和外部验证评估了稳健性。主要结果：VS与NR显示出接近原始的一致性：CT中中位ICC(A,1)为0.9976，MRI中为0.9984。RS产生了较低的一致性和更大的偏差，而FK显示出中间行为，确认仅凭间距元数据就可以影响放射组学特征。梯度导出和邻域敏感描述符受到预处理的影响最大。VS在外部CT验证中保留了与NR相当的预测性能，而MRI在不同预处理策略和分类器之间显示出更大的变异性。意义：体素间距感知提取将几何建模与插值引起的信号修改分离，同时保留原始图像信号，为非各向同性CT和MRI的放射组学分析提供了一种一致的替代方案，以取代各向同性重采样。
cs.CV / 43 / 2607.12404
Contrastive-Augmented Flow Matching for Style-Content Disentanglement
对比增强流匹配用于风格-内容解耦
Abstract
Learning representations that separate content and style is crucial for controllable generation and compositional generalization. However, diffusion and flow-based models trained primarily with generative objectives often produce entangled or misaligned factors. To address this gap, we introduce Contrastive Augmented Flow Matching (CAtFM), a framework that integrates contrastive regularization into an invertible flow matching formulation to promote structured content-style representations. Rather than constraining intermediate latents or velocity fields, we apply contrastive supervision to predicted endpoints during training, enforcing semantic consistency across transported distributions while allowing disentanglement to emerge implicitly, without assuming strictly pure or fully factorized content and style representations. Our main experiments operate in the CLIP embedding space, with additional validation using frozen DINO and ALIGN encoders. Across synthetic data, in-domain styles, and real-world benchmarks (ImageNet, WikiArt, DomainNet, and DTD), CAtFM improves content and style retrieval, enhances embedding cluster separation, and achieves stronger open-set robustness compared to generative and discriminative baselines. Overall, CAtFM provides a simple way to couple discriminative constraints with deterministic transport, improving disentanglement and robustness under distribution shift.
Chinese Translation
学习能够分离内容和风格的表示对于可控生成和组合泛化至关重要。然而，主要以生成目标训练的扩散和流模型往往会产生纠缠或不对齐的因素。为了解决这一问题，我们提出了对比增强流匹配（Contrastive Augmented Flow Matching, CAtFM）框架，该框架将对比正则化集成到可逆流匹配的公式中，以促进结构化的内容-风格表示。我们并不限制中间潜变量或速度场，而是在训练过程中对预测的端点应用对比监督，强制在传输分布之间保持语义一致性，同时允许解耦隐式地出现，而不假设内容和风格表示是严格纯粹或完全分解的。我们的主要实验在 CLIP 嵌入空间中进行，并使用冻结的 DINO 和 ALIGN 编码器进行额外验证。在合成数据、领域内风格和真实世界基准（ImageNet、WikiArt、DomainNet 和 DTD）中，CAtFM 提升了内容和风格检索，增强了嵌入聚类的分离性，并在开放集鲁棒性方面优于生成和判别基线。总体而言，CAtFM 提供了一种简单的方法，将判别约束与确定性传输相结合，在分布变化下改善了解耦和鲁棒性。
cs.CV / 44 / 2607.12416
Virtual Chromoendscopy with Tunable Visibility Enhancement
可调可见性增强的虚拟染色内镜技术
Abstract
Chromoendoscopy (CE) is a common clinical practice that sprays indigo carmine blue dye onto the gastric surface to improve the visibility of gastric lesions, such as an early cancer. While CE is effective in detecting the lesions, preparing and spraying the dye needs additional cost and time, which is undesirable both for patients and medical practitioners. To overcome this issue, virtual chromoendoscopy (V-CE) was recently proposed, which applies a learned image translation model to virtually generate a CE image from a standard endoscopy (SE) image. In this paper, we propose virtual enhanced chromoendoscopy (V-ECE) that combines V-CE with image enhancement techniques to further improve the visibility of gastric lesions. Because a desired enhancement level depends on the inspected lesion and the practitioner's preference, we introduce a novel image translation model that can generate V-ECE images using an enhancement level tunable by a user. Experimental results demonstrate that our proposed model can plausibly generate V-ECE images with various enhancement levels using a unified model.
Chinese Translation
染色内镜（Chromoendoscopy，CE）是一种常见的临床实践，通过将靛蓝染料喷洒在胃表面，以提高对胃病变（如早期癌症）的可见性。尽管CE在检测病变方面有效，但准备和喷洒染料需要额外的成本和时间，这对患者和医疗从业者来说都是不利的。为了解决这个问题，最近提出了虚拟染色内镜（Virtual Chromoendoscopy，V-CE），该技术应用了一种学习的图像转换模型，从标准内镜图像（Standard Endoscopy，SE）中虚拟生成CE图像。在本文中，我们提出了一种虚拟增强染色内镜（Virtual Enhanced Chromoendoscopy，V-ECE），将V-CE与图像增强技术相结合，以进一步提高胃病变的可见性。由于所需的增强水平取决于检查的病变和从业者的偏好，我们引入了一种新颖的图像转换模型，该模型可以生成具有用户可调增强水平的V-ECE图像。实验结果表明，我们提出的模型能够使用统一模型合理生成具有不同增强水平的V-ECE图像。
cs.CV / 45 / 2607.12418
MQAdapter: Multi-Modal Quantum Adapter for Coarse-to-Fine VLM Fine-tuning
MQAdapter：用于粗到细的多模态量子适配器在视觉语言模型微调中的应用
Abstract
Large-scale Vision-Language Models have demonstrated impressive transfer learning capabilities across a wide range of tasks. For few-shot classification, we observe that VLMs exhibit a notable ability to filter candidate categories and thus achieve high Top-K accuracy. However, they often struggle with fine-grained discrimination among visually similar categories, resulting in unsatisfactory Top-1 performance, as shown in Figure 1. Existing studies on VLM adapters generally focus on global alignment between visual and textual representations in the feature space, but fail to exploit semantically similar categories to refine fine-grained visual representations. Based on these observations, we propose a novel coarse-to-fine VLM fine-tuning approach for few-shot learning that leverages quantum computation, termed the Multi-Modal Quantum Adapter (MQAdapter). Specifically, MQAdapter first retrieves the Top-K category candidates most similar to the input image and uses them as semantic anchors. It then employs a cross-modal quantum learning mechanism to refine visual features under the guidance of these anchors. The core of this mechanism is the encoding of visual and textual features into quantum states. By leveraging quantum entanglement and superposition in a high-dimensional Hilbert space, MQAdapter effectively models higher-order cross-modal interactions, producing more discriminative representations than traditional Euclidean adapters. MQAdapter is parameter-efficient and can be integrated with various existing fine-tuning algorithms to achieve further performance gains. Evaluations on 15 datasets demonstrate the effectiveness of MQAdapter while requiring fewer trainable parameters.
Chinese Translation
大规模视觉-语言模型在广泛任务中展示了令人印象深刻的迁移学习能力。在少样本分类中，我们观察到视觉语言模型（VLMs）具有显著的候选类别过滤能力，从而实现高Top-K准确率。然而，它们在视觉上相似类别之间的细粒度区分上常常表现不佳，导致Top-1性能不理想，如图1所示。现有关于VLM适配器的研究通常集中在视觉和文本表示在特征空间中的全局对齐，但未能利用语义相似的类别来细化细粒度的视觉表示。基于这些观察，我们提出了一种新颖的粗到细的VLM微调方法，适用于少样本学习，该方法利用量子计算，称为多模态量子适配器（MQAdapter）。具体而言，MQAdapter首先检索与输入图像最相似的Top-K类别候选，并将其作为语义锚点。然后，它采用跨模态量子学习机制，在这些锚点的指导下细化视觉特征。该机制的核心是将视觉和文本特征编码为量子态。通过利用高维希尔伯特空间中的量子纠缠和叠加，MQAdapter有效地建模更高阶的跨模态交互，生成比传统欧几里得适配器更具辨别力的表示。MQAdapter在参数效率上表现优越，并且可以与多种现有微调算法集成，以实现进一步的性能提升。在15个数据集上的评估证明了MQAdapter的有效性，同时所需的可训练参数更少。
cs.CV / 46 / 2607.12419
DeGuNet: Depth-Guided Ultra-Compact Backbones for Efficient LiDAR-Camera 3D Detection
DeGuNet：用于高效LiDAR-相机3D检测的深度引导超紧凑骨干网络
Abstract
In autonomous driving perception, the fusion of LiDAR and camera modalities has become the dominant paradigm for 3D object detection. However, current multi-modal frameworks heavily rely on massive visual backbones pretrained on 2D semantic tasks. This reliance introduces substantial parameter redundancy and a structural misalignment, as 2D priors are ill-equipped to handle the extreme sparsity of LiDAR projections required for Bird's-Eye-View geometry. To address this, we present DeGuNet, an ultra-compact and plug-and-play image backbone explicitly designed for depth-guided representation learning. By incorporating sparsity-aware feature extraction mechanisms, DeGuNet effectively aligns multi-view images with unstructured LiDAR depth while strictly preventing invalid-region contamination. Extensive experiments on the nuScenes dataset demonstrate DeGuNet's broad plug-and-play applicability and superior efficiency. When integrated into established baselines, it fundamentally eliminates architectural redundancy, reducing GPU memory consumption by up to 66.5% and achieving a 1.16x inference speedup. Concurrently, DeGuNet delivers up to a 6.20 absolute mAP gain, establishing a new paradigm for parameter-efficient multi-modal 3D perception.
Chinese Translation
在自动驾驶感知中，LiDAR与相机模态的融合已成为3D目标检测的主流范式。然而，当前的多模态框架严重依赖于在2D语义任务上预训练的大型视觉骨干网络。这种依赖引入了显著的参数冗余和结构不匹配，因为2D先验无法有效处理鸟瞰图几何所需的LiDAR投影的极端稀疏性。为了解决这一问题，我们提出了DeGuNet，一种超紧凑且即插即用的图像骨干网络，专门设计用于深度引导的表示学习。通过结合稀疏感知特征提取机制，DeGuNet有效地将多视角图像与非结构化的LiDAR深度对齐，同时严格防止无效区域的污染。在nuScenes数据集上的大量实验表明，DeGuNet具有广泛的即插即用适用性和卓越的效率。当与现有基线集成时，它从根本上消除了架构冗余，将GPU内存消耗减少了多达66.5%，并实现了1.16倍的推理加速。同时，DeGuNet提供了高达6.20的绝对mAP增益，为参数高效的多模态3D感知建立了新的范式。
cs.CV / 47 / 2607.12429
More Than Where You Are: Learning Semantics, Structure, and Geometry from Cross-View Localization
超越位置：从跨视角定位中学习语义、结构和几何
Abstract
Consistent cross-view understanding under extreme viewpoint changes is essential for spatial intelligence, as it enables models to recognize the same scene across extreme viewpoint gaps. Cross-view localization naturally provides a promising pathway toward this ability, as it requires a model to align ground-view imagery with geo-referenced satellite-view imagery despite drastic appearance changes to estimate camera poses. Recent visual foundation models have made this long-standing localization problem increasingly feasible by providing rich 2D representations for cross-view matching. However, we argue that cross-view localization should not be viewed merely as 2D matching or pose estimation. In this work, we revisit cross-view localization as more than pose estimation and investigate how it can help the model develop consistent cross-view understanding under extreme viewpoint changes, including stable semantics, reliable structure, and transferable geometry. We identify three key limitations of existing methods that prevent them from achieving this. They usually lack explicit 3D grounding, rely on strict point-wise matching that can weaken semantic consistency, and learn from an absolute objective that provides limited guidance for geometric reasoning. To address these limitations, we propose CROSS, a unified cross-view localization framework built upon 3D-grounded alignment, structure-aware matching, and hypothesis ranking. This formulation makes structure learning an intrinsic requirement, encourages semantic representations to remain stable, and enables the model to acquire transferable geometry. Extensive experiments on the KITTI and VIGOR datasets show that CROSS achieves state-of-the-art performance in cross-view localization. More importantly, CROSS effectively learns stable semantics, reliable structure, and transferable geometry across extremely different viewpoints.
Chinese Translation
在极端视角变化下保持一致的跨视角理解对于空间智能至关重要，因为它使模型能够在极端视角差异中识别相同的场景。跨视角定位自然为这一能力提供了一个有前景的途径，因为它要求模型对齐地面视图图像与地理参考的卫星视图图像，尽管外观发生了剧烈变化，以估计相机姿态。最近的视觉基础模型通过为跨视角匹配提供丰富的二维表示，使这一长期存在的定位问题变得越来越可行。然而，我们认为跨视角定位不应仅仅被视为二维匹配或姿态估计。在本研究中，我们重新审视跨视角定位，认为它不仅仅是姿态估计，并探讨它如何帮助模型在极端视角变化下发展一致的跨视角理解，包括稳定的语义、可靠的结构和可转移的几何。我们识别出现有方法的三个关键局限性，这些局限性阻止它们实现这一目标。它们通常缺乏明确的三维基础，依赖严格的逐点匹配，这可能削弱语义一致性，并从一个提供有限几何推理指导的绝对目标中学习。为了解决这些局限性，我们提出了CROSS，一个基于三维基础对齐、结构感知匹配和假设排名的统一跨视角定位框架。这种表述使结构学习成为内在要求，鼓励语义表示保持稳定，并使模型能够获取可转移的几何。在KITTI和VIGOR数据集上的大量实验表明，CROSS在跨视角定位中实现了最先进的性能。更重要的是，CROSS有效地学习了在极其不同的视角下稳定的语义、可靠的结构和可转移的几何。
cs.CV / 48 / 2607.12433
ARDepth: Auto-regressive Monocular Depth Estimation with Progressive Visual Conditioning
ARDepth：具有渐进视觉条件的自回归单目深度估计
Abstract
Diffusion models have recently become the dominant paradigm for monocular depth estimation (MDE). However, they implicitly assume that depth can be recovered as a globally smooth field through iterative denoising, which does not explicitly reflect the piecewise and scale-dependent organization of scene geometry. In practice, geometric structure emerges progressively across spatial scales, where coarse layout, surfaces, and boundaries are constructed in a hierarchical manner. Motivated by this observation, we introduce ARDepth, which formulates depth estimation as structured auto-regressive generation. Instead of recovering depth through global refinement, ARDepth progressively constructs depth representations as spatial resolution increases. To support this generative process, we introduce Scale-Progressive Conditioning (SPC) to inject multi-scale visual features at each generation stage, and Semantic-Aware Guidance (SAG) to provide scene-level semantic priors that enhance global structural consistency. Together, these designs enable the model to capture fine-grained local details while maintaining coherent global geometry. Empirical results demonstrate that our approach achieves strong performance and produces structurally consistent depth predictions across scales, validating auto-regressive generation as a promising alternative paradigm for geometric modeling.
Chinese Translation
扩散模型最近已成为单目深度估计（MDE）的主导范式。然而，它们隐含地假设深度可以通过迭代去噪恢复为一个全局平滑的场，这并未明确反映场景几何的分段和尺度依赖组织。在实践中，几何结构在空间尺度上逐渐显现，其中粗略布局、表面和边界以层次化的方式构建。基于这一观察，我们提出了ARDepth，它将深度估计公式化为结构化的自回归生成。ARDepth不是通过全局细化来恢复深度，而是随着空间分辨率的增加逐步构建深度表示。为了支持这一生成过程，我们引入了尺度渐进条件（Scale-Progressive Conditioning, SPC），在每个生成阶段注入多尺度视觉特征，并引入语义感知引导（Semantic-Aware Guidance, SAG），提供增强全局结构一致性的场景级语义先验。这些设计共同使模型能够捕捉细致的局部细节，同时保持一致的全局几何结构。实证结果表明，我们的方法在各尺度上实现了强大的性能，并产生了结构一致的深度预测，验证了自回归生成作为几何建模的有前景的替代范式。
cs.CV / 49 / 2607.12450
Let RGB Be the Language of Vision
让RGB成为视觉的语言
Abstract
This work introduces a unified formulation for vision models, where diverse forms of visual information beyond natural images, such as masks, depth maps, and other structured visual signals, are all represented as RGB images, while general visual tasks can be converted into a common RGB-to-RGB image editing problem. In this paradigm, different types of visual information internally share the same encoding and decoding architecture and parameters as natural images, enabling a single model to transfer across tasks through a unified visual interface, in a way analogous to how language models operate over text. We refer to this formulation as RGB In and RGB Out (RINO). Built upon a generic image editing backbone without task-specific fine-tuning, RINO demonstrates robust and competitive zero-shot performance on both dense understanding tasks such as segmentation and depth estimation (where we unify outputs as RGB), and dense-conditioned generation tasks such as pose-to-image generation (where we unify inputs as RGB). We hope this study provides useful insights toward general unified vision-language systems, where diverse visual tasks can be expressed, interpreted, and solved through a shared visual language. Code is available at https://github.com/yangtiming/RINO.
Chinese Translation
本研究提出了一种统一的视觉模型表述，其中除了自然图像之外的多种视觉信息形式，如掩膜、深度图和其他结构化视觉信号，均被表示为RGB图像，同时一般视觉任务可以转化为一个共同的RGB到RGB图像编辑问题。在这一范式中，不同类型的视觉信息在内部共享与自然图像相同的编码和解码架构及参数，使得单一模型能够通过统一的视觉接口在任务之间进行迁移，类似于语言模型在文本上操作的方式。我们将这一表述称为RGB In和RGB Out（RINO）。基于通用的图像编辑骨干网络而无需特定任务的微调，RINO在密集理解任务（如分割和深度估计，输出统一为RGB）和密集条件生成任务（如姿态到图像生成，输入统一为RGB）上展示了强大且具有竞争力的零-shot性能。我们希望本研究为通用统一的视觉-语言系统提供有益的见解，使得多样的视觉任务能够通过共享的视觉语言进行表达、解释和解决。代码可在https://github.com/yangtiming/RINO获取。
cs.CV / 50 / 2607.12464
Steering Diffusion Models via Class-Contrastive Influence for Few-Shot Medical Classification
通过类对比影响引导扩散模型进行少样本医学分类
Abstract
When labeled data are scarce, off-the-shelf diffusion models can augment training sets for few-shot medical image classification, but not all generated samples are equally useful for the downstream task. Existing approaches largely improve synthetic data by increasing realism, diversity, or domain adaptation, while overlooking a more fundamental question: how should sample usefulness for classification be measured and optimized? We address this with Class-Contrastive Influence (C2I), a criterion that quantifies a sample's usefulness through its gradient-based influence on the classifier. We find that effective samples exhibit a strong C2I gap: their loss gradients align with validation gradients from the same class and oppose those from other classes. Our analysis further suggests that such high-C2I samples are hard, boundary-proximal examples that help refine the decision boundary and improve robustness. Building on this insight, we fine-tune diffusion models with reinforcement learning using a C2I-based reward to steer generation toward class-informative samples. Across several few-shot medical imaging benchmarks, C2I-guided generation improves downstream accuracy and robustness over diffusion-based augmentation baselines, showing that synthetic augmentation is most effective when guided by task usefulness rather than image quality alone.
Chinese Translation
当标记数据稀缺时，现成的扩散模型可以增强少样本医学图像分类的训练集，但并非所有生成的样本对下游任务都是同等有用的。现有方法主要通过提高合成数据的真实性、多样性或领域适应性来改善合成数据，而忽视了一个更根本的问题：如何衡量和优化样本在分类中的有用性？我们通过类对比影响（Class-Contrastive Influence, C2I）来解决这个问题，这一标准通过样本对分类器的基于梯度的影响来量化样本的有用性。我们发现，有效样本表现出明显的C2I差距：它们的损失梯度与来自同一类别的验证梯度一致，而与其他类别的梯度相对立。我们的分析进一步表明，这种高C2I样本是困难的、接近边界的例子，有助于细化决策边界并提高鲁棒性。基于这一见解，我们利用基于C2I的奖励通过强化学习微调扩散模型，以引导生成朝向具有类别信息的样本。在多个少样本医学成像基准测试中，C2I引导的生成在下游准确性和鲁棒性方面优于基于扩散的增强基线，表明合成增强在任务有用性而非仅仅图像质量的指导下最为有效。
cs.CV / 51 / 2607.12477
Self in Space: Benchmarking Self-Awareness and Spatial Cognition in UAV Embodied Intelligence
空间中的自我：无人机具身智能的自我意识与空间认知基准测试
Abstract
Autonomous UAV systems increasingly rely on multimodal large language models (MLLMs) to operate in complex real-world environments. Such embodied scenarios require not only understanding the surrounding space but also maintaining a coherent representation of the agent itself. However, existing UAV-oriented approaches and benchmarks remain largely environment-centric, primarily focusing on spatial understanding tasks, with the agent's self-awareness remaining implicit. To address this gap, we introduce SIS-Bench, a benchmark for evaluating embodied spatial intelligence in UAV scenarios under a unified self-in-space formulation. SIS-Bench organizes evaluation along two complementary dimensions, space and self, and a three-level hierarchy of perception, memory, and reasoning. It contains 4,856 question--answer pairs across 13 tasks derived from 1,646 real-world UAV videos through a task-conditioned construction pipeline with expert verification.Extensive evaluations reveal that current MLLMs exhibit fundamental limitations in modeling dynamic and agent-centered processes. In particular, we observe a clear imbalance between spatial cognition and self-awareness, as well as a progressive performance degradation across cognitive levels.Motivated by these findings, we further explore a motion-aware representation that incorporates self-related dynamics through optical flow and visual feature fusion. Experimental results show that modeling agent motion consistently improves perception and memory performance, not only in spatial cognition but also in self-awareness, and generalizes to downstream UAV decision-making tasks.Our results highlight the importance of self-awareness for advancing embodied spatial intelligence, and provide both a new benchmark and empirical evidence for motion-aware self-in-space modeling.
Chinese Translation
自主无人机系统越来越依赖多模态大型语言模型（MLLMs）在复杂的真实环境中进行操作。这类具身场景不仅需要理解周围空间，还需要保持对自身的连贯表征。然而，现有的无人机导向方法和基准测试仍然主要集中于环境，主要关注空间理解任务，而代理的自我意识则保持隐性。为了解决这一空白，我们提出了SIS-Bench，这是一个用于在统一的空间自我表述下评估无人机场景中具身空间智能的基准。SIS-Bench在空间和自我两个互补维度以及感知、记忆和推理三个层次的层级结构中组织评估。它包含了4,856个问题-答案对，涵盖了13个任务，这些任务源自1,646个真实世界的无人机视频，通过任务条件构建管道和专家验证而得出。广泛的评估揭示了当前的MLLMs在建模动态和以代理为中心的过程方面存在基本局限性。特别是，我们观察到空间认知与自我意识之间存在明显的不平衡，以及在认知层次上的逐步性能下降。基于这些发现，我们进一步探索了一种运动感知表征，通过光流和视觉特征融合来整合与自我相关的动态。实验结果表明，建模代理运动在空间认知和自我意识中一致地提高了感知和记忆性能，并且能够推广到下游无人机决策任务。我们的结果强调了自我意识在推进具身空间智能方面的重要性，并为运动感知的空间自我建模提供了新的基准和实证证据。
cs.CV / 52 / 2607.12495
RealSkin: Spatio-Spectral Partial Neural Adjoint Maps for Image-to-3D Attribute Transfer
RealSkin：用于图像到3D属性转移的时空光谱部分神经伴随映射
Abstract
Creating photorealistic 3D assets requires bridging the appearance gap between real-world observations and synthetic models. A promising approach is to transfer visual attributes from real images onto synthetic 3D surfaces. Traditional methods struggle with resolution mismatch and the inherent discreteness of point correspondences. In contrast, resolution-robust functional maps enable smooth attribute propagation but rely on near-isometry assumptions and topological consistency. To address these limitations, we propose RealSkin, a self-supervised framework that performs correspondence optimization in a learned spectral domain, guided by spatial correspondences. We first introduce a spatial-guided registration algorithm to establish coarse correspondences under severe topological discrepancies. To relax strict isometric assumptions and handle partial correspondences, we further design a spectral-aware neural adjoint network that incorporates partial correspondences into a neural function space and models non-isometric residuals for correspondence refinement. Experimental results demonstrate that our method achieves state-of-the-art performance on challenging real-to-synthetic scenarios. The code will be publicly released.
Chinese Translation
创建逼真的3D资产需要弥合真实世界观察与合成模型之间的外观差距。一种有前景的方法是将视觉属性从真实图像转移到合成的3D表面。传统方法在分辨率不匹配和点对应的固有离散性方面面临挑战。相比之下，分辨率稳健的功能映射能够实现平滑的属性传播，但依赖于近等距假设和拓扑一致性。为了解决这些局限性，我们提出了RealSkin，一个自监督框架，在学习的光谱域中执行对应优化，并由空间对应引导。我们首先引入了一种空间引导的配准算法，以在严重的拓扑差异下建立粗略对应。为了放宽严格的等距假设并处理部分对应，我们进一步设计了一种光谱感知的神经伴随网络，将部分对应纳入神经函数空间，并建模非等距残差以进行对应细化。实验结果表明，我们的方法在具有挑战性的真实到合成场景中达到了最先进的性能。代码将公开发布。
cs.CV / 53 / 2607.12497
TerraLogic: A Benchmark for Hierarchical Geospatial Reasoning in Earth Observation
TerraLogic：地球观测中分层地理空间推理的基准测试
Abstract
Beyond perception, reasoning is essential in remote sensing for advanced interpretation, inference, and decision-making. Recent advances in large language models (LLMs) have enabled tool-augmented agents that leverage external tools to perform complex analytical tasks. However, existing studies in remote sensing primarily focus on perception-oriented tasks, leaving cognitive geospatial reasoning largely underexplored. To address this gap, we introduce TerraLogic, a benchmark for geospatial reasoning. TerraLogic comprises 545 scenario-driven, hierarchy-aware tasks, such as hazard vulnerability assessment, urban heat island analysis, and forest fragmentation dynamics, spanning optical, Synthetic Aperture Radar (SAR), and infrared (IR) imagery. It advances evaluation beyond recognition and monitoring toward cognitive-level geospatial analysis. To facilitate evaluation on TerraLogic, we further propose HieraPlan, a tool-augmented agent that organizes toolkits into functional hierarchies and performs fault-tolerant reasoning. HieraPlan enables structured abstraction, robust recovery from tool failures, and stable long-horizon planning. Extensive experiments demonstrate that current approaches struggle with hierarchical geospatial reasoning, while HieraPlan provides a strong baseline with improved reasoning, cross-modal generalization, and error handling. The dataset and agent code are publicly available at https://github.com/Ireliya/TerraLogic.
Chinese Translation
在遥感中，推理超越感知，对于高级解释、推断和决策至关重要。最近大型语言模型（LLMs）的进展使得工具增强代理能够利用外部工具执行复杂的分析任务。然而，现有的遥感研究主要集中于以感知为导向的任务，认知地理空间推理仍然未得到充分探索。为了解决这一空白，我们引入了TerraLogic，一个用于地理空间推理的基准测试。TerraLogic包含545个情境驱动、关注层次的任务，如灾害脆弱性评估、城市热岛分析和森林破碎化动态，涵盖光学、合成孔径雷达（SAR）和红外（IR）图像。它将评估从识别和监测推进到认知层面的地理空间分析。为了便于在TerraLogic上进行评估，我们进一步提出了HieraPlan，一个工具增强的代理，能够将工具包组织成功能层次，并执行容错推理。HieraPlan实现了结构化抽象、从工具故障中稳健恢复以及稳定的长远规划。大量实验表明，当前方法在分层地理空间推理方面存在困难，而HieraPlan则提供了一个强有力的基线，具有改进的推理能力、跨模态泛化和错误处理。数据集和代理代码可在https://github.com/Ireliya/TerraLogic上公开获取。
cs.CV / 54 / 2607.12503
DynTrace: Tracking Dynamic Object Evidence for 4D Spatio-Temporal Reasoning in MLLMs
DynTrace：用于多模态大语言模型的4D时空推理的动态对象证据追踪
Abstract
4D spatio-temporal reasoning, jointly modeling 3D spatial structure and temporal evolution, is essential for understanding dynamic worlds and enabling embodied interaction. While current Multimodal Large Language Models (MLLMs) show strong capabilities in static scene understanding and coarse-grained 4D tasks, they still have notable limitations in continuous dynamic scene perception, especially in tracking dynamic object evidence for coherent 4D spatio-temporal reasoning. This shortcoming stems mainly from relying on sparse frame-level observations, fragmenting continuous dynamic cues and leaving models unable to disentangle genuine object dynamics from camera-induced apparent motion. Inspired by humans tracking dynamic cues while compensating for viewpoint changes, we propose DynTrace, a training-free framework for 4D spatio-temporal reasoning with two complementary components. Dynamic Trajectory Visualization (DTV) reprojects world-coordinate trajectories onto the image plane, providing geometry-informed visual priors that disentangle genuine object dynamics from camera-induced apparent motion. Meanwhile, the Dynamic Trace Token (DT-Token), organized into a Dynamic Trace Graph (DTG), tracks object-level dynamic cues, trace evolution, and key moments, maintaining continuous dynamic object evidence for coherent 4D reasoning. Together, these two components equip MLLMs with continuously tracked dynamic object evidence, grounded in geometry-informed visual priors and structured spatio-temporal traces. DynTrace consistently improves open-source MLLMs, achieving state-of-the-art results on Dyn-Bench, VLM4D, and DSI-Bench, validating the importance of tracking dynamic object evidence for robust 4D spatio-temporal reasoning.
Chinese Translation
4D时空推理联合建模3D空间结构和时间演变，对于理解动态世界和实现具身交互至关重要。尽管当前的多模态大语言模型（MLLMs）在静态场景理解和粗粒度4D任务中表现出强大的能力，但在连续动态场景感知方面仍存在显著局限，尤其是在追踪动态对象证据以实现连贯的4D时空推理方面。这一短板主要源于依赖稀疏的帧级观察，导致连续动态线索的碎片化，使得模型无法将真实的对象动态与相机引起的表观运动区分开。受到人类在补偿视角变化时追踪动态线索的启发，我们提出了DynTrace，一个无训练的4D时空推理框架，包含两个互补组件。动态轨迹可视化（DTV）将世界坐标轨迹重投影到图像平面，提供几何信息驱动的视觉先验，以区分真实的对象动态和相机引起的表观运动。同时，动态追踪标记（DT-Token）组织成动态追踪图（DTG），追踪对象级动态线索、轨迹演变和关键时刻，保持连续的动态对象证据以实现连贯的4D推理。这两个组件共同为MLLMs提供了基于几何信息驱动的视觉先验和结构化时空轨迹的连续追踪动态对象证据。DynTrace持续改善开源MLLMs，在Dyn-Bench、VLM4D和DSI-Bench上取得了最先进的结果，验证了追踪动态对象证据对于稳健的4D时空推理的重要性。
cs.CV / 55 / 2607.12539
DiTailed: Ensuring Visual Object Consistency in Text-Image-to-Image Flow Matching Models
DiTailed：确保文本-图像-图像流匹配模型中的视觉对象一致性
Abstract
Despite remarkable progress in text-guided image editing, generative models frequently fail to preserve visual object consistency, defined as the preservation of a subject's key attributes throughout the editing process. We address this limitation through three contributions. First, we introduce ABO-Edit, a dataset specifically designed to study object consistency, comprising over 12,000 triplets of source images, editing prompts, and high-quality target images rendered from artist-designed 3D assets, with multi-view coverage and human-verified quality control. Second, we uncover an overlooked property of image-editing rectified flow models: the conditioning embedding space, not directly supervised during training, encodes a prediction of the final generated image even at high noise levels. Third, exploiting this finding, we propose FlowMirror, a parameter-free auxiliary loss that supervises this conditioning embedding space. Without architectural changes, our method improves generation quality across several metrics over baselines.
Chinese Translation
尽管在文本引导的图像编辑方面取得了显著进展，但生成模型在保持视觉对象一致性方面常常失败，视觉对象一致性被定义为在整个编辑过程中保持主题的关键属性。我们通过三项贡献来解决这一局限性。首先，我们引入了ABO-Edit，一个专门设计用于研究对象一致性的数据集，包含超过12,000个源图像、编辑提示和从艺术家设计的3D资产渲染的高质量目标图像的三元组，具有多视角覆盖和人工验证的质量控制。其次，我们发现了图像编辑校正流模型的一个被忽视的特性：条件嵌入空间在训练期间并未直接监督，但即使在高噪声水平下也编码了最终生成图像的预测。第三，利用这一发现，我们提出了FlowMirror，一种无参数的辅助损失，用于监督这一条件嵌入空间。在不改变架构的情况下，我们的方法在多个指标上提高了生成质量，超越了基线。
cs.CV / 56 / 2607.12544
Edge-Aware Thermal Infrared UAV Swarm Tracking
边缘感知热红外无人机群体跟踪
Abstract
Thermal infrared (TIR) imaging is essential for UAV swarm operations in visually degraded environments. However, tracking tiny UAVs remains challenging due to limited appearance cues, frequent occlusions, and rapid maneuvers. Despite significant progress driven by benchmarks such as the Anti-UAV challenge, existing methods primarily prioritize accuracy while overlooking the computational constraints of real-time edge deployment. The standard Kalman Filter (KF) offers the efficiency required for edge devices, yet its constant-velocity assumption often breaks down under highly dynamic UAV motion and thermal sensor jitter. More sophisticated nonlinear estimators can improve robustness but often introduce additional computational costs. To address this gap, we propose an edge-aware online tracking pipeline centered on the Adaptive Kinematic Kalman Filter (AKKF), which augments the linear KF with state-dependent kinematic modeling while preserving real-time efficiency. Combined with transient false-positive suppression and kinematics-driven predictive coasting, the presented pipeline improves trajectory continuity under challenging TIR conditions. Experiments on the Beyond Strong Baseline (BSB) benchmark provide a starting point for edge-aware UAV tracking by jointly evaluating tracking performance and computational efficiency, offering insights toward future real-time deployment.
Chinese Translation
热红外（TIR）成像对于在视觉受限环境下的无人机群体操作至关重要。然而，由于外观线索有限、频繁遮挡和快速机动，跟踪微小无人机仍然具有挑战性。尽管在反无人机挑战等基准测试的推动下取得了显著进展，但现有方法主要优先考虑准确性，而忽视了实时边缘部署的计算约束。标准卡尔曼滤波器（KF）提供了边缘设备所需的效率，但其恒速假设在高度动态的无人机运动和热传感器抖动下往往失效。更复杂的非线性估计器可以提高鲁棒性，但通常会引入额外的计算成本。为了解决这一问题，我们提出了一种以自适应运动卡尔曼滤波器（AKKF）为中心的边缘感知在线跟踪管道，该管道在保持实时效率的同时，增强了线性卡尔曼滤波器的状态依赖运动建模。结合瞬态假阳性抑制和运动驱动的预测滑行，所提出的管道在具有挑战性的TIR条件下改善了轨迹连续性。在超强基线（BSB）基准测试上的实验为边缘感知无人机跟踪提供了一个起点，通过联合评估跟踪性能和计算效率，为未来的实时部署提供了见解。
cs.CV / 57 / 2607.12556
CGRL: Concept-Guided Pruning and Representation Learning for Whole-Slide Image Classification
CGRL：基于概念引导的修剪与表示学习用于全幻灯片图像分类
Abstract
Weakly supervised whole-slide image (WSI) classification is widely used in computational pathology because slide-level labels are easier to obtain than dense region annotations. Existing multiple instance learning (MIL) methods often aggregate large bags of patch embeddings using mainly visual cues, which can retain many non-informative patches and provide weak alignment between instance features and class-level disease semantics. We propose Concept-Guided Pruning and Representation Learning (CGRL), a simple framework that introduces class-level concept prototypes derived from disease prompts into the MIL pipeline. First, concept-relevance pruning ranks patch instances by their similarity to class concepts and retains the top-K concept-relevant patches for downstream MIL aggregation. Second, concept-guided contrastive representation learning constructs class-wise positive and negative patch sets from the same similarity matrix and optimizes target-class, symmetric auxiliary, and cross-class separation objectives, thereby regularizing the projected concept space. We evaluate CGRL on TCGA-BRCA and TCGA-NSCLC using multiple representative MIL methods. Experimental results show that CGRL improves several model-dataset combinations, with gains depending on the downstream MIL model and dataset. It achieves particularly clear improvements in accuracy and macro-F1 while reducing computational cost through concept-relevance pruning. These findings demonstrate that class-level semantic concepts provide an effective and practical prior for patch selection and representation learning in weakly supervised computational pathology.
Chinese Translation
弱监督全幻灯片图像（WSI）分类在计算病理学中被广泛应用，因为幻灯片级标签比密集区域注释更容易获得。现有的多实例学习（MIL）方法通常主要依赖视觉线索聚合大量补丁嵌入，这可能保留许多无信息的补丁，并在实例特征与类别级疾病语义之间提供弱对齐。我们提出了基于概念引导的修剪与表示学习（CGRL），这是一个简单的框架，将从疾病提示中派生的类别级概念原型引入MIL管道。首先，概念相关性修剪通过与类别概念的相似性对补丁实例进行排名，并保留前K个与概念相关的补丁以供下游MIL聚合。其次，概念引导的对比表示学习从相同的相似性矩阵构建类别-wise的正负补丁集，并优化目标类别、对称辅助和跨类别分离目标，从而规范化投影的概念空间。我们在TCGA-BRCA和TCGA-NSCLC上使用多种代表性的MIL方法评估CGRL。实验结果表明，CGRL改善了多个模型-数据集组合，增益取决于下游MIL模型和数据集。它在准确性和宏观F1得分上取得了特别明显的改善，同时通过概念相关性修剪降低了计算成本。这些发现表明，类别级语义概念为弱监督计算病理学中的补丁选择和表示学习提供了有效且实用的先验。
cs.CV / 58 / 2607.12557
Gaussian Mixture Modeling for Event-Aware Visual Allocation in Long Video Understanding
用于事件感知视觉分配的高斯混合建模在长视频理解中的应用
Abstract
Large Vision-Language Models (LVLMs) face significant challenges in long video understanding due to the excessive computational cost and information loss associated with uniform sampling. Existing keyframe selection methods often treat video frames as atomic entities and allocate visual budgets equally, thereby overlooking high-level semantic structures and introducing substantial redundancy. To address these limitations, we propose GMM-EVA (Gaussian Mixture Modeling for Event-Aware Visual Allocation), which leverages Gaussian Mixture Models to model event-level structure from discrete frame-wise observations. A differentiated allocation strategy is then applied to preserve one primary high-resolution keyframe per event for high-fidelity detail, while utilizing lower-resolution secondary keyframes to maintain temporal context and optimize token budgets. GMM-EVA is a training-free, plug-and-play framework that generalizes robustly across various relevance measures and downstream LVLMs. Extensive experiments on multiple long video benchmarks demonstrate that our method significantly outperforms uniform sampling. Notably, GMM-EVA achieves comparable performance to baseline selection methods while utilizing only approximately half of the visual token budget, highlighting its superior efficiency and effectiveness.
Chinese Translation
大型视觉语言模型（LVLMs）在长视频理解中面临显著挑战，主要由于均匀采样带来的过高计算成本和信息损失。现有的关键帧选择方法通常将视频帧视为原子实体，并均等分配视觉预算，从而忽视了高层次的语义结构，并引入了大量冗余。为了解决这些局限性，我们提出了GMM-EVA（用于事件感知视觉分配的高斯混合建模），该方法利用高斯混合模型从离散的逐帧观测中建模事件级结构。然后，应用差异化分配策略，以保留每个事件的一个主要高分辨率关键帧以获取高保真细节，同时利用低分辨率的次要关键帧来保持时间上下文并优化令牌预算。GMM-EVA是一个无训练、即插即用的框架，能够在各种相关性度量和下游LVLMs中稳健地推广。在多个长视频基准上的广泛实验表明，我们的方法显著优于均匀采样。值得注意的是，GMM-EVA在仅使用大约一半的视觉令牌预算的情况下，达到了与基线选择方法相当的性能，突显了其卓越的效率和有效性。
cs.CV / 59 / 2607.12569
Traceback Translators Against Forgetting in Continual Fake Speech Detection
针对持续假语音检测中的遗忘问题的追溯翻译器
Abstract
Fake speech detectors are increasingly challenged by the development of new and more accurate generative models. To cope with this problem, continual learning techniques are nowadays widely considered feasible strategies for updating models to new datasets, but they also lead to decreased performance on previously seen samples (catastrophic forgetting). In this work, we propose a forgetting-resilient solution based on the adoption of domain translators within a frozen detector, which remaps the new feature spaces into the original ones by means of a traceback translator network. Experimental results show that this strategy enables the achievement of high detection rates with respect to traditional retraining, while minimizing the computational effort and preserving the detection accuracy on previous data.
Chinese Translation
假语音检测器正面临着新型更准确生成模型的挑战。为了应对这一问题，持续学习技术如今被广泛视为更新模型以适应新数据集的可行策略，但这也导致了对先前样本性能的下降（灾难性遗忘）。在本研究中，我们提出了一种基于在冻结检测器内采用领域翻译器的抗遗忘解决方案，该方案通过追溯翻译器网络将新的特征空间重新映射到原始特征空间。实验结果表明，该策略在实现高检测率的同时，相较于传统的再训练，能够最小化计算工作量并保持对先前数据的检测准确性。
cs.CV / 60 / 2607.12592
WanToFight: Real-Time Generative Game Engine for Multi-Player Combat Interaction
WanToFight：用于多人战斗互动的实时生成游戏引擎
Abstract
We present WanToFight, a generative game engine that simulates real-time, two-player The King of Fighters '97 (KOF~'97) gameplay from keyboard input. Prior generative game engines target either single-player first-person settings or non-real-time cooperative scenarios; multi-player control, real-time inference, complex physical interaction, and adversarial gameplay have not been jointly addressed. WanToFight closes this gap with three components built on the Wan-1.3B video diffusion transformer: a streaming autoregressive generator with block-causal attention and a rolling KV cache; a visually grounded Player Association module that binds each player's keyboard signal to a character identity; and a gated, locally causal keyboard injection module trained with a single-player-to-full-gameplay curriculum. A four-step DMD-distilled student paired with a pruned VAE decoder sustains 30FPS at 512x384 on a single NVIDIA RTX 5090 over the duration of a complete match. To our knowledge, WanToFight is the first generative game engine to combine multi-player control, real-time inference, complex physical interaction, and adversarial gameplay in one system.
Chinese Translation
我们提出了WanToFight，这是一款生成游戏引擎，能够根据键盘输入模拟实时的双人《拳皇'97》（The King of Fighters '97，KOF~'97）游戏玩法。此前的生成游戏引擎主要针对单人第一人称设置或非实时合作场景；多人控制、实时推理、复杂物理互动和对抗性游戏玩法尚未得到共同解决。WanToFight通过三个基于Wan-1.3B视频扩散变换器的组件填补了这一空白：一个具有块因果注意力和滚动KV缓存的流式自回归生成器；一个将每个玩家的键盘信号绑定到角色身份的视觉基础玩家关联模块；以及一个经过单人到完整游戏课程训练的门控局部因果键盘注入模块。配备经过剪枝的变分自编码器（VAE）解码器的四步DMD提炼学生在单个NVIDIA RTX 5090上能够在完整比赛期间维持30FPS，分辨率为512x384。据我们所知，WanToFight是第一个将多人控制、实时推理、复杂物理互动和对抗性游戏玩法结合在一个系统中的生成游戏引擎。
cs.CV / 61 / 2607.12602
Decouple and Reason: Anatomically Guided Two-Stage Voxel-Level Grounding of Free-Text Findings in 3D Chest CT
解耦与推理：解剖学指导的三维胸部CT中自由文本发现的两阶段体素级定位
Abstract
Automatic voxel-level grounding of free-text findings in 3D chest Computed Tomography (CT) is critical for clinical interpretability. However, this task remains highly challenging due to the intricate spatial complexity of large 3D volumes and the heterogeneity of free-text findings. Existing end-to-end approaches often struggle to simultaneously learn the localized feature representations required for accurate 3D segmentation and the complex semantic understanding needed for text alignment, leading to suboptimal grounding performance. To overcome this fundamental limitation, we propose a novel decoupled framework that disentangles the problem into two specialized stages: (1) class-agnostic lesion segmentation and (2) text-volume reasoning. This structural separation allows the model to first extract candidate sub-volumes by localizing potential abnormalities. Subsequently, intensive cross-modal reasoning is performed to align these localized sub-volumes with free-text medical findings. To resolve the spatial ambiguities inherent in local regions, the reasoning module is augmented with explicit anatomical guidance, utilizing relative spatial coordinates and lung lobe priors. Evaluated on the ReXGroundingCT benchmark, our method achieves state-of-the-art performance in overall grounding quality on the official leaderboard. These results demonstrate that decoupling detection from reasoning is a highly effective paradigm for handling the complexity of 3D medical visual grounding. Our code is publicly available at https://github.com/khuhm/DAGG.
Chinese Translation
在三维胸部计算机断层扫描（CT）中，自动体素级定位自由文本发现对于临床可解释性至关重要。然而，由于大型三维体积的复杂空间结构和自由文本发现的异质性，这一任务仍然非常具有挑战性。现有的端到端方法往往难以同时学习准确的三维分割所需的局部特征表示和文本对齐所需的复杂语义理解，从而导致定位性能不佳。为了解决这一根本性限制，我们提出了一种新颖的解耦框架，将问题分解为两个专门的阶段：（1）与类别无关的病灶分割和（2）文本-体积推理。这种结构性分离允许模型首先通过定位潜在异常来提取候选子体积。随后，进行密集的跨模态推理，以将这些局部子体积与自由文本医学发现对齐。为了解决局部区域固有的空间模糊性，推理模块通过利用相对空间坐标和肺叶先验进行了显式的解剖学指导。经过在ReXGroundingCT基准上的评估，我们的方法在官方排行榜上实现了整体定位质量的最先进性能。这些结果表明，将检测与推理解耦是一种处理三维医学视觉定位复杂性的高效范式。我们的代码已公开，地址为 https://github.com/khuhm/DAGG。
cs.CV / 62 / 2607.12621
Towards Vision-Free CIR: Attribute-Augmented Scoring and LLM-Based Reranking for Zero-Shot Composed Image Retrieval
面向无视觉的复合图像检索：基于属性增强评分和大型语言模型重排序的零-shot方法
Abstract
Recent work has shown that "Vision-Free'' approaches (representing images as text) can be effective for standard image retrieval tasks. However, it remains unclear whether this paradigm can effectively handle a more complex, multimodal task, Composed Image Retrieval (CIR), due to the inherent information loss in textual descriptions. In this paper, we introduce a Vision-Free CIR framework that addresses this challenge through two key techniques: (1) Attribute-Augmented Hybrid Scoring, which compensates for lost visual details via explicit attribute matching, and (2) LLM-Based Reranking, which verifies semantic consistency of top candidates. Experiments on the open-domain CIRR dataset show that our approach outperforms existing Zero-shot CIR methods (44.04% R@1, +8.79%). On FashionIQ, our results highlight the trade-off between semantic reasoning and fine-grained visual matching. Ablation studies reveal that both attribute-augmented scoring and LLM-Based Reranking consistently improve performance.
Chinese Translation
最近的研究表明，“无视觉”方法（将图像表示为文本）在标准图像检索任务中是有效的。然而，由于文本描述中固有的信息损失，这一范式是否能够有效处理更复杂的多模态任务——复合图像检索（CIR）仍然不明确。本文提出了一种无视觉的CIR框架，通过两项关键技术来应对这一挑战：（1）属性增强混合评分，它通过显式属性匹配来弥补丢失的视觉细节；（2）基于大型语言模型（LLM）的重排序，它验证了前候选项的语义一致性。在开放域CIRR数据集上的实验表明，我们的方法优于现有的零-shot CIR方法（44.04% R@1，提升8.79%）。在FashionIQ数据集上，我们的结果突显了语义推理与细粒度视觉匹配之间的权衡。消融研究表明，属性增强评分和基于LLM的重排序均能持续提高性能。
cs.CV / 63 / 2607.12663
MAGE: Color-Invariant and Spatial Knowledge Distillation for Gastric Neoplasm Classification
MAGE：用于胃肿瘤分类的颜色不变性和空间知识蒸馏
Abstract
Accurate differentiation between gastric adenoma and carcinoma during endoscopy is critical for clinical decision-making. Yet, this task is highly challenging due to high inter-class similarity and ambiguous boundaries between the two classes. Existing ROI-based classification methods often suffer from detection/segmentation error propagation and loss of surrounding global context. In contrast, full-image classification lacks the necessary spatial focus. Furthermore, we observe that deep neural networks gravitate towards domain-specific texture biases(e.g. bleeding, lighting artifacts), often causing models to predict based on spurious correlations instead of intrinsic morphological features. To address these limitations, we propose a novel framework, Masked Achromatic Guidance Expert (MAGE). During training, we introduce an auxiliary local expert branch trained on masked achromatic views of the neoplasm. By suppressing background context and color, this branch is forced to learn highly discriminative, purely structural features. We then employ a dual-objective distillation strategy, transferring both classification logits and spatial attention maps to provide implicit spatial supervision to the main branch that receives full WLI as input. This dual-objective distillation forces the model to ground its predictions in morphology rather than relying on shortcuts, while still retaining clinically relevant color cues. At inference time, our deployable model operates on images without annotated masks, ensuring real-time deployability . Extensive experiments on a clinical gastric endoscopy dataset show that our method significantly outperforms existing detection-based methodologies (e.g. YOLO) and classification-based methodologies (e.g. Swin-Transformer), providing not only superior classification performance but also interpretable attention maps for clinical reliability.
Chinese Translation
在内窥镜检查中准确区分胃腺瘤和癌症对临床决策至关重要。然而，由于两类之间高度的类间相似性和模糊的边界，这一任务极具挑战性。现有的基于感兴趣区域（ROI）的分类方法常常受到检测/分割错误传播和周围全局上下文丧失的影响。相比之下，完整图像分类缺乏必要的空间聚焦。此外，我们观察到深度神经网络倾向于域特定的纹理偏差（例如出血、光照伪影），这常常导致模型基于虚假的相关性而非内在形态特征进行预测。为了解决这些局限性，我们提出了一种新颖的框架，称为Masked Achromatic Guidance Expert（MAGE）。在训练过程中，我们引入一个辅助的局部专家分支，该分支在肿瘤的掩蔽无色视图上进行训练。通过抑制背景上下文和颜色，该分支被迫学习高度区分的纯结构特征。随后，我们采用双目标蒸馏策略，将分类逻辑和空间注意力图转移到主分支，以提供隐式空间监督，该主分支接收完整的白光内窥镜图像作为输入。这种双目标蒸馏迫使模型将其预测基于形态学而非依赖捷径，同时仍保留临床相关的颜色线索。在推理时，我们可部署的模型在没有注释掩膜的图像上运行，确保实时可部署性。在临床胃内窥镜数据集上的广泛实验表明，我们的方法显著优于现有的基于检测的方法（例如YOLO）和基于分类的方法（例如Swin-Transformer），不仅提供了更优越的分类性能，还提供了可解释的注意力图，以确保临床可靠性。
cs.CV / 64 / 2607.12678
Text-Aided Multi-Modal Panoptic Symbol Spotting for CAD Floor Plan Drawings
文本辅助的多模态全景符号识别用于计算机辅助设计平面图
Abstract
Computer-Aided Design (CAD) floor plan drawings contain both graphical primitives and textual annotations, which provide complementary geometric and semantic cues for intelligent design understanding. Among CAD analysis tasks, panoptic symbol spotting has become increasingly important with the growing demand for industrial digitalization and deep learning-based automation. However, most existing methods remain primarily primitive-centric and underexploit textual annotations, despite their critical semantic value. Even the few text-aware approaches often treat annotations only superficially, without properly modeling complex syntax and hierarchical semantics of CAD annotations, which leads to semantic loss and suboptimal spotting performance. To address these limitations, we propose TextCAD, a multimodal framework that jointly models graphical primitives and textual annotations for panoptic symbol spotting. Specifically, we design a Type-Attribute Correlation Encoder (TACE) to explicitly encode the compositional semantics within annotations by jointly modeling their types and attributes. We further introduce a Semantic Hierarchy Alignment framework with Multi-level Semantic Filtering (MSF) and primitive downsampling, which adaptively aligns annotation semantics with graphical primitives at different semantic levels and enables accurate cross-modal semantic injection and fusion. Experiments on real-world building-design datasets show that TextCAD effectively improves symbol spotting performance and achieves state-of-the-art results.
Chinese Translation
计算机辅助设计（CAD）平面图包含图形原语和文本注释，这为智能设计理解提供了互补的几何和语义线索。在CAD分析任务中，随着工业数字化和基于深度学习的自动化需求的增长，全景符号识别变得越来越重要。然而，现有的大多数方法仍主要以原语为中心，未能充分利用文本注释，尽管它们具有重要的语义价值。即使是少数关注文本的研究方法，通常也只是表面处理注释，而没有对CAD注释的复杂语法和层次语义进行适当建模，这导致了语义损失和次优的识别性能。为了解决这些局限性，我们提出了TextCAD，一个多模态框架，联合建模图形原语和文本注释以进行全景符号识别。具体而言，我们设计了一种类型-属性相关编码器（Type-Attribute Correlation Encoder, TACE），通过联合建模注释的类型和属性，明确编码注释中的组合语义。我们进一步引入了一种语义层次对齐框架，结合多层次语义过滤（Multi-level Semantic Filtering, MSF）和原语下采样，能够自适应地在不同语义层次上对齐注释语义与图形原语，并实现准确的跨模态语义注入和融合。在真实建筑设计数据集上的实验表明，TextCAD有效提高了符号识别性能，并达到了最先进的结果。
cs.CV / 65 / 2607.12680
ReflectVLN: Training Vision-Language Navigation Agents with Reflective Reasoning
ReflectVLN：通过反思推理训练视觉-语言导航代理
Abstract
Existing vision-language navigation methods often couple a VLM with waypoint decoders to produce multi-step action plans, but they typically lack an explicit closed-loop mechanism for tracking semantic progress, diagnosing execution failures, and recovering from error accumulation in long-horizon navigation. To address this gap, we propose ReflectVLN, an agentic VLN framework that organizes decision-making through bidirectionally interactive intention and execution agents. The intention agent performs subtask decomposition and reflection, generating executable subtask descriptions as corrective plans. Conditioned on these descriptions, the execution agent grounds them into short-horizon actions under current observations while monitoring sub-goal progress and detecting off-track behavior. Crucially, ReflectVLN enables closed-loop bidirectional communication: the execution agent emits progress and deviation signals to trigger reflection and subtask updates on demand, and the intention agent returns structured guidance that reconditions subsequent actions for recovery. To encourage temporally coherent decisions with interpretable intermediate rationales, we introduce Action Chain-of-Thought (Action-CoT), a path-conditioned dual-query training scheme for action generation. Experiments on standard VLN benchmarks show that ReflectVLN improves success rates and path efficiency under a constrained data budget, with favorable training cost and fewer high-level intention calls at inference time, while providing interpretable intermediate decisions for analysis and collaboration. Code is available at: https://github.com/AIprogrammer/ReflectVLN
Chinese Translation
现有的视觉-语言导航方法通常将视觉语言模型（VLM）与路径解码器结合，以生成多步骤的行动计划，但它们通常缺乏明确的闭环机制来跟踪语义进展、诊断执行失败以及在长时间导航中恢复错误积累。为了解决这一问题，我们提出了ReflectVLN，一个代理型视觉-语言导航框架，通过双向交互的意图代理和执行代理来组织决策过程。意图代理执行子任务分解和反思，生成可执行的子任务描述作为纠正计划。在这些描述的条件下，执行代理将其落实为当前观察下的短期行动，同时监控子目标进展并检测偏离行为。至关重要的是，ReflectVLN实现了闭环的双向通信：执行代理发出进展和偏差信号，以按需触发反思和子任务更新，而意图代理则返回结构化指导，以重新调整后续行动以实现恢复。为了鼓励具有可解释中间推理的时间一致决策，我们引入了行动链思维（Action Chain-of-Thought, Action-CoT），这是一种路径条件的双查询训练方案，用于行动生成。在标准视觉-语言导航基准上的实验表明，ReflectVLN在受限数据预算下提高了成功率和路径效率，具有良好的训练成本和在推理时较少的高层意图调用，同时提供了可解释的中间决策以便于分析和协作。代码可在以下链接获取：https://github.com/AIprogrammer/ReflectVLN
cs.CV / 66 / 2607.12681
MambaPSA: A Mamba-based Replacement for C2PSA in YOLO26
MambaPSA：基于Mamba的YOLO26中C2PSA的替代方案
Abstract
State space models (SSMs), notably Mamba, have recently emerged as efficient alternatives to self-attention with linear computational complexity. We investigate the integration of Mamba into YOLO26, the latest non-maximum suppression (NMS)-free object detection framework, by proposing MambaPSA, a lightweight Mamba-based replacement for the C2PSA block at the end of the backbone. To complement this study, we additionally insert a bidirectional Vision Mamba (BiViM) module at the P3, P4, and P5 levels of the neck. Experiments on PASCAL VOC 2007+2012 show that MambaPSA reduces parameters by 2.9%, FLOPs by 12.1%, and improves CPU inference throughput by 17.6% (from 17 to 20 FPS) with negligible accuracy change (-0.1 mAP50:95), while the P4 BiViM placement yields the best accuracy gain (+0.9 mAP50:95). These results suggest that SSMs offer a favorable efficiency-accuracy trade-off when replacing attention-based blocks in NMS-free lightweight detectors.
Chinese Translation
状态空间模型（SSMs），特别是Mamba，最近作为具有线性计算复杂度的自注意力的高效替代方案而出现。我们通过提出MambaPSA，研究了将Mamba集成到YOLO26这一最新的无非极大抑制（NMS）目标检测框架中的可能性，MambaPSA是对主干网络末尾C2PSA模块的轻量级Mamba替代方案。为了补充这项研究，我们还在颈部的P3、P4和P5层插入了双向视觉Mamba（BiViM）模块。在PASCAL VOC 2007+2012上的实验表明，MambaPSA将参数减少了2.9%，FLOPs减少了12.1%，并在几乎没有准确性变化（-0.1 mAP50:95）的情况下，提高了CPU推理吞吐量17.6%（从17 FPS提升至20 FPS），而P4 BiViM的放置则带来了最佳的准确性提升（+0.9 mAP50:95）。这些结果表明，SSMs在替代NMS-free轻量级检测器中的基于注意力的模块时，提供了良好的效率-准确性权衡。
cs.CV / 67 / 2607.12684
Lesion Segmentation in Moderate to Severe Traumatic Brain Injury: An nnU-Net Based Approach with Adaptive Normalization in the AIMS-TBI 2025 Challenge
中度至重度创伤性脑损伤中的病灶分割：基于nnU-Net的自适应归一化方法在AIMS-TBI 2025挑战中的应用
Abstract
The segmentation of lesions in Moderate to Severe Traumatic Brain Injury (msTBI) from T1-weighted MRI presents a significant clinical challenge due to the profound heterogeneity of lesion characteristics in terms of size, shape, and location. To address this, the AIMS-TBI 2025 Challenge was organized to promote the development of robust and accurate segmentation algorithms. In this paper, we present our deep learning-based solution. Our methodology employs the nnU-Net framework with an adaptive intensity normalization strategy confined to the brain parenchyma, effectively reducing inter-subject variability and mitigating artifacts from non-brain structures. Upon final evaluation on the held-out test set, our method demonstrated highly competitive performance on the official leaderboard, achieving an Overall Dice Coefficient of 0.6305. The model obtained a Dice score of 0.4805 for lesion segmentation and 0.9324 for non-lesion tissue. While the lesion Dice reflects the difficulty of detecting highly heterogeneous lesions, the high non-lesion Dice primarily indicates the model's strong ability to correctly identify non-lesion voxels, demonstrating good specificity in differentiating lesion from non-lesion regions. These results demonstrate that incorporating anatomically constrained normalization within the nnU-Net pipeline is a powerful and effective strategy for tackling the complexities of msTBI lesion segmentation.
Chinese Translation
从T1加权MRI中对中度至重度创伤性脑损伤（msTBI）进行病灶分割是一项重大的临床挑战，因为病灶特征在大小、形状和位置上存在显著异质性。为了解决这一问题，组织了AIMS-TBI 2025挑战，以促进稳健且准确的分割算法的开发。在本文中，我们提出了基于深度学习的解决方案。我们的方法采用了nnU-Net框架，并结合了针对脑实质的自适应强度归一化策略，有效减少了个体间的变异性，并减轻了来自非脑结构的伪影。在对保留的测试集进行最终评估时，我们的方法在官方排行榜上表现出极具竞争力的性能，整体Dice系数达到了0.6305。该模型在病灶分割上获得了0.4805的Dice分数，而在非病灶组织上则达到了0.9324。虽然病灶Dice反映了检测高度异质性病灶的难度，但高非病灶Dice主要表明模型在正确识别非病灶体素方面的强大能力，显示出在区分病灶与非病灶区域方面的良好特异性。这些结果表明，在nnU-Net流程中结合解剖约束的归一化是一种强大且有效的策略，用于应对msTBI病灶分割的复杂性。
cs.CV / 68 / 2607.12704
Label-Decoupled Style Augmentation for Domain Generalization in Multi-Label Remote Sensing Scene Classification
标签解耦的风格增强用于多标签遥感场景分类的领域泛化
Abstract
Multi-label classification assigns several co-occurring labels to each aerial scene, yet deployed models often encounter data distributions different from their training. Feature-statistics augmentation such as MixStyle, EFDMix, and correlated style uncertainty improves generalization at low cost but perturbs channel statistics globally, treating each image as a single style; one class can then contaminate the augmentation of another. Domain generalization is understudied for multi-label remote sensing; no prior method or multi-source benchmark targets it. A label-decoupled augmentation framework is therefore proposed, confining style perturbation to label-specific regions. Per-label attention, obtained from a learnable module or from gradient class-activation maps, yields per-label feature statistics; these statistics are mixed with cross-domain samples that share present labels, under independent per-label coefficients, and features are recomposed by attention-weighted normalization. Three operators combined with two attention sources produce six variants, evaluated on a leave-one-domain-out benchmark from multi-label UCM, AID, and DFC15 over six shared labels. Averaged over three splits and five seeds, the best variant attains 71.5% mean average precision, exceeding empirical risk minimization by 5.0 points and the strongest global-statistics baseline by 1.3 points, with the largest gain on the hardest transfer (up to 7.7 points). Ablations indicate that spatial attention and refreshed localization maps are most influential. The framework adds at most 0.35% parameters, leaves inference unchanged, and appears to offer a generic, inexpensive upgrade path for multi-label statistics-based domain generalization. Code is available upon acceptance at https://github.com/Alaa-Almouradi/Style-Augmentation-Upgrade.
Chinese Translation
多标签分类为每个航空场景分配多个共现标签，但实际部署的模型往往面临与训练数据分布不同的情况。特征统计增强方法，如 MixStyle、EFDMix 和相关风格不确定性，以低成本提高了泛化能力，但会全局扰动通道统计，将每幅图像视为单一风格；因此，一个类别可能会污染另一个类别的增强。针对多标签遥感的领域泛化研究相对较少，现有方法或多源基准尚未针对这一问题。因此，提出了一种标签解耦的增强框架，将风格扰动限制在特定标签的区域内。通过可学习模块或梯度类激活图获得的每标签注意力，生成每标签的特征统计；这些统计与共享当前标签的跨域样本混合，采用独立的每标签系数，并通过注意力加权归一化重新组合特征。结合三种操作符和两种注意力来源，产生六种变体，并在来自多标签 UCM、AID 和 DFC15 的留一域外基准上进行评估，涉及六个共享标签。经过三次划分和五次随机种子的平均，最佳变体达到了 71.5% 的平均精度，比经验风险最小化高出 5.0 个百分点，比最强的全局统计基线高出 1.3 个百分点，在最困难的迁移上获得了最大的增益（高达 7.7 个百分点）。消融实验表明，空间注意力和更新的定位图对结果影响最大。该框架最多增加 0.35% 的参数，推理不变，并且似乎为基于多标签统计的领域泛化提供了一条通用且低成本的升级路径。代码将在接受后提供，链接为 https://github.com/Alaa-Almouradi/Style-Augmentation-Upgrade。
cs.CV / 69 / 2607.12746
Color Pass-Through via Camera-Display Coupling
通过相机-显示器耦合实现颜色透视
Abstract
When a real-world scene is captured by a smartphone camera and viewed on its screen, the displayed image often differs noticeably from the original scene in color, brightness, and contrast. This gap persists despite substantial advances in both modern cameras and displays. A key reason is that most pipelines factor the high-dimensional capture-to-display process into two separately calibrated camera and display stages, and then connect them through low-dimensional color transforms, leading to information bottlenecks and inevitable error accumulation. To address this systemic challenge, we propose Color Pass-Through, an end-to-end learned framework that operates directly on captured images. Our key insight is to treat the camera and display as a coupled system rather than calibrating them in isolation. Coupling the camera and display yields two practical advantages: (1) it brings the entire real-world scenes to the display via end-to-end optimization, and (2) it allows efficient one-step calibration for each distinct observer via complete capture-to-display path. We validate Color Pass-Through using both digital and human observers. Compared with representative baselines, our method achieves an average gain of +2.0 points on a 5-point user study and more than 2x improvement on quantitative metrics, demonstrating improved reproduction of the perceived color of the original scene.
Chinese Translation
当智能手机相机捕捉到现实世界场景并在其屏幕上显示时，显示的图像在颜色、亮度和对比度上往往与原始场景明显不同。这种差距在现代相机和显示器取得显著进展的情况下仍然存在。一个主要原因是大多数处理流程将高维的捕捉到显示过程分解为两个单独校准的相机和显示阶段，然后通过低维颜色变换将它们连接起来，这导致信息瓶颈和不可避免的误差累积。为了解决这一系统性挑战，我们提出了颜色透视（Color Pass-Through），这是一个端到端学习框架，直接对捕获的图像进行处理。我们的关键见解是将相机和显示器视为一个耦合系统，而不是单独校准它们。相机和显示器的耦合带来了两个实际优势：（1）通过端到端优化将整个现实世界场景传递到显示器上；（2）通过完整的捕捉到显示路径为每个不同的观察者提供高效的一步校准。我们使用数字和人类观察者验证了颜色透视。与代表性基线相比，我们的方法在5分制用户研究中平均获得+2.0分的提升，并在定量指标上实现了超过2倍的改善，证明了对原始场景感知颜色的再现能力得到了提升。
cs.CV / 70 / 2607.12748
Weakly Supervised Spatio-Temporal Candidate Discovery of Dairy Farm Sites from Seasonal Satellite Imagery
基于季节性卫星影像的奶牛场地弱监督时空候选发现
Abstract
Farm site discovery from satellite imagery is a spatiotemporal candidate ranking problem because farm evidence is distributed across pasture, field boundaries, roads, buildings, and seasonal vegetation patterns. Direct farm labels are often incomplete, which makes fully supervised detection difficult. This paper proposes a weakly supervised pipeline for ranking dairy farm candidate clusters from seasonal Sentinel imagery and open map priors. The method uses aligned spring, summer, and autumn image tiles from County Cork, Ireland, with spectral bands, vegetation indices, built area indices, and a pasture channel. A Barlow Twins encoder learns multi-season tile embeddings without farm labels. In parallel, weak OpenStreetMap farm priors are split into a prior and a held-out set. Prior features support a rule-based tile score that combines farm proximity, seasonal pasture evidence, and summer greenness, while held-out features are reserved only for proxy evaluation. The rule score is smoothed over a spatial representation graph using geographic proximity and embedding similarity, and high-scoring tiles are grouped into ranked candidate clusters. From 26,722 valid tiles, the main run selects 535 high-confidence tiles and forms 71 candidate clusters. The top 5 clusters achieve 0.60 precision within 500 m and 0.80 precision within 1000 m of held-out OpenStreetMap farm features. The top 10 clusters achieve 0.40 precision within 500 m and 0.80 precision within 1000 m. The results show that seasonal representation learning and weak geographic priors can reduce large satellite image collections into compact candidate sets for human review.
Chinese Translation
从卫星影像中发现农场位置是一个时空候选排名问题，因为农场证据分布在牧场、田地边界、道路、建筑物和季节性植被模式中。直接的农场标签通常不完整，这使得完全监督的检测变得困难。本文提出了一种弱监督管道，用于从季节性Sentinel影像和开放地图先验中对奶牛场候选集群进行排名。该方法使用来自爱尔兰科克县的对齐春季、夏季和秋季影像切片，结合光谱波段、植被指数、建筑面积指数和牧场通道。Barlow Twins编码器在没有农场标签的情况下学习多季节影像切片的嵌入。同时，弱的OpenStreetMap农场先验被分为先验集和保留集。先验特征支持一个基于规则的影像切片评分，该评分结合了农场接近度、季节性牧场证据和夏季绿度，而保留特征仅用于代理评估。该规则评分通过地理接近度和嵌入相似性在空间表示图上进行平滑处理，高评分的影像切片被分组为排名候选集群。从26,722个有效影像切片中，主要运行选择了535个高置信度影像切片，并形成71个候选集群。前5个集群在距离保留的OpenStreetMap农场特征500米内实现了0.60的精度，在1000米内实现了0.80的精度。前10个集群在距离500米内实现了0.40的精度，在1000米内实现了0.80的精度。结果表明，季节性表示学习和弱地理先验能够将大量卫星影像集合缩减为紧凑的候选集供人工审查。
cs.CV / 71 / 2607.12750
CRC-HGD: A Histopathological Image Dataset for Grading Colorectal Cancer
CRC-HGD：一种用于结直肠癌分级的组织病理图像数据集
Abstract
Colorectal cancer (CRC) is the third most common cancer worldwide and the second leading cause of cancer-related deaths globally, with approximately 1,926,425 new cases and 904,019 deaths reported in 2022. Accurate histologic grading plays a critical role in prognosis and treatment planning for colorectal adenocarcinoma. In recent years, artificial intelligence and its subcategories, including machine learning and deep learning, have been increasingly employed for automated cancer detection and classification. An appropriate and well-organized dataset is the essential first step to achieve this goal. This paper introduces CRC-HGD, a histopathological microscopy image dataset of 1,914 images obtained from 214 colorectal adenocarcinoma patients (Grade I: 106, Grade II: 75, Grade III: 33). The specimens are H&E-stained colorectal tissue sections acquired at the Poursina Hakim Research Center of Isfahan University of Medical Sciences, Iran, diagnosed between 2014 and 2019, and graded according to the World Health Organization (WHO) criteria into three grades: well-differentiated (Grade I), moderately differentiated (Grade II), and poorly differentiated (Grade III). For each specimen, four magnification levels are provided: 4x, 10x, 20x, and 40x. The dataset is accessible via Mendeley Data (https://doi.org/10.17632/yfp5sfj47m.4) and at http://databiox.com, where the latest version is also available. The distinctive feature of this dataset is the provision of labeled specimens across all three differentiation grades at multiple magnification levels, enabling comprehensive computational analysis of colorectal cancer grading.
Chinese Translation
结直肠癌（CRC）是全球第三大常见癌症，也是导致癌症相关死亡的第二大原因，2022年报告的新病例约为1,926,425例，死亡人数为904,019例。准确的组织学分级在结直肠腺癌的预后和治疗计划中发挥着关键作用。近年来，人工智能及其子领域，包括机器学习和深度学习，越来越多地被用于自动化癌症检测和分类。构建一个合适且组织良好的数据集是实现这一目标的首要步骤。本文介绍了CRC-HGD，一个包含1,914幅图像的组织病理显微镜图像数据集，这些图像来自214名结直肠腺癌患者（分级I：106，分级II：75，分级III：33）。样本为在伊朗伊斯法罕医科大学Poursina Hakim研究中心获取的H&E染色结直肠组织切片，诊断时间为2014年至2019年，并根据世界卫生组织（WHO）标准分为三个等级：分化良好（分级I）、分化中等（分级II）和分化差（分级III）。每个样本提供四个放大倍数：4x、10x、20x和40x。该数据集可通过Mendeley Data（https://doi.org/10.17632/yfp5sfj47m.4）和http://databiox.com访问，最新版本也可在此获取。该数据集的独特之处在于提供了跨所有三个分化等级的标记样本，并在多个放大倍数下进行，能够实现对结直肠癌分级的全面计算分析。
cs.CV / 72 / 2607.12752
Hallo4D: Multi-Modal Hallucination Mitigation for Consistent Spatio-Temporal Generation
Hallo4D：用于一致性时空生成的多模态幻觉缓解
Abstract
While recent advances in 3D generation have enabled impressive visual synthesis, existing methods often rely on 2D diffusion supervision without explicit mechanisms for geometric consistency, leading to spatial hallucinations such as duplicated structures and misaligned geometry. These issues become more severe in 4D generation, where maintaining consistency across viewpoints and temporal evolution introduces additional challenges, including jitter, identity flicker, and structural drift. We present \textbf{Hallo4D}, a unified and model-agnostic framework for mitigating spatiotemporal hallucinations in 3D and 4D content generation. Hallo4D introduces a generation-detection-correction paradigm that leverages large multimodal language models (LMMs) to identify and summarize spatial and temporal inconsistencies from multi-view and multi-frame renderings. These insights guide a consensus-driven image-space consistency optimization, where an LMM-based selector evaluates candidate corrections through multi-model voting, without requiring retraining or architectural modifications. To further improve temporal consistency and optimization efficiency, Hallo4D incorporates motion-aware keyframe sampling, LMM-guided initialization, and appearance alignment. We additionally introduce exposure-aware optimization and visibility pruning to enhance robustness under challenging viewpoints. Extensive experiments demonstrate that Hallo4D consistently outperforms strong baselines across diverse 3D and 4D generation settings, providing a scalable and generalizable solution for consistency-aware content generation.
Chinese Translation
尽管近年来3D生成技术取得了显著的视觉合成进展，但现有方法往往依赖于2D扩散监督，缺乏明确的几何一致性机制，导致空间幻觉，如重复结构和几何错位。这些问题在4D生成中变得更加严重，因为在视角和时间演变之间保持一致性带来了额外的挑战，包括抖动、身份闪烁和结构漂移。我们提出了 extbf{Hallo4D}，一个统一且与模型无关的框架，用于缓解3D和4D内容生成中的时空幻觉。Hallo4D引入了一种生成-检测-修正范式，利用大型多模态语言模型（LMMs）识别和总结来自多视角和多帧渲染的空间和时间不一致性。这些见解指导了一种基于共识的图像空间一致性优化，其中基于LMM的选择器通过多模型投票评估候选修正，而无需重新训练或架构修改。为了进一步提高时间一致性和优化效率，Hallo4D结合了运动感知关键帧采样、LMM引导初始化和外观对齐。我们还引入了曝光感知优化和可见性修剪，以增强在挑战性视角下的鲁棒性。大量实验表明，Hallo4D在多种3D和4D生成设置中始终优于强基线，提供了一种可扩展且具有普适性的解决方案，以实现一致性意识的内容生成。
cs.CV / 73 / 2607.12753
RFMSR: Residual Flow Matching for Image Super-Resolution
RFMSR：用于图像超分辨率的残差流匹配
Abstract
Image super-resolution (ISR) has witnessed remarkable progress with diffusion models and flow matching. The dominant text-to-image (T2I) based approaches leverage large-scale foundation models as generative priors, achieving impressive perceptual quality but at the cost of massive model sizes and prohibitive training expenses. Recent flow-matching-based vision-only approaches have made significant strides; however, they adopt standard flow formulations that transport from a pure Gaussian prior to the data distribution, discarding the rich structural information already present in the low-quality (LQ) input. Furthermore, existing single-step acceleration techniques often forfeit the model's multi-step inference capability. In this paper, we propose Residual Flow Matching for Image Super-Resolution (RFMSR), a vision-only framework that centers the source distribution at the LQ latent, reducing transport distance and preserving structural priors throughout the flow trajectory. We further introduce a two-phase training strategy: Phase I pretrains the velocity field via conditional flow matching, while Phase II applies end-to-end supervision to the single-step prediction while retaining the velocity loss across all timesteps, achieving high-quality single-step generation without sacrificing multi-step refinement. Extensive experiments demonstrate that RFMSR achieves comparable or even superior perceptual quality compared to state-of-the-art (SOTA) methods. The source code is available at https://github.com/Faze-Hsw/RFMSR.
Chinese Translation
图像超分辨率（ISR）在扩散模型和流匹配的推动下取得了显著进展。以文本生成图像（T2I）为基础的方法利用大规模基础模型作为生成先验，虽然在感知质量上取得了令人印象深刻的成果，但代价是模型规模庞大且训练费用高昂。最近，基于流匹配的纯视觉方法取得了重要进展；然而，它们采用标准流公式，从纯高斯先验传输到数据分布，忽视了低质量（LQ）输入中已经存在的丰富结构信息。此外，现有的单步加速技术往往放弃了模型的多步推理能力。在本文中，我们提出了用于图像超分辨率的残差流匹配（RFMSR），这是一个纯视觉框架，将源分布中心定位于LQ潜变量，减少传输距离并在整个流动轨迹中保留结构先验。我们进一步引入了两阶段的训练策略：第一阶段通过条件流匹配预训练速度场，第二阶段对单步预测应用端到端监督，同时在所有时间步保持速度损失，从而实现高质量的单步生成而不牺牲多步细化。大量实验表明，RFMSR在感知质量上与最先进的方法（SOTA）相比具有可比性甚至更优。源代码可在 https://github.com/Faze-Hsw/RFMSR 获取。
cs.CV / 74 / 2607.12756
VisCo: Leveraging Large Language Models as Intrinsic Encoders for Visual Token Compression
VisCo：利用大型语言模型作为视觉标记压缩的内在编码器
Abstract
Vision-language models (VLMs) process large numbers of visual tokens, resulting in substantial inference latency and memory overhead. This has motivated extensive research on visual token compression. While training-free strategies rely on heuristic metrics and suffer significant performance degradation under high compression ratios, many training-based methods introduce external compression modules that force the VLM backbone to adapt, incurring substantial retraining cost and compromising VLMs' priors. Effective visual token compression hinges on strong information encoding, a capability already present in pretrained VLMs but underutilized by existing approaches. Motivated by this, we propose VisCo, a training-efficient self-compression framework that reuses the pretrained VLM itself as an intrinsic compressor. VisCo is a parameter-sharing autoencoder that compresses visual information using a small set of memory tokens and transfers hierarchical information from encoding to decoding. Experiments show that VisCo surpasses prior methods across all evaluated compression ratios, with larger gains under more aggressive compression, and remains stable even in the extreme single-token setting. Moreover, when combined with the original visual tokens, the learned memory tokens can even improve the base model, suggesting that VisCo captures complementary representations beyond compression.
Chinese Translation
视觉-语言模型（VLMs）处理大量视觉标记，导致显著的推理延迟和内存开销。这促使了对视觉标记压缩的广泛研究。虽然无训练策略依赖于启发式指标，并在高压缩比下表现出显著的性能下降，但许多基于训练的方法引入了外部压缩模块，迫使VLM主干进行适应，导致大量的重新训练成本，并妨碍了VLM的先验知识。有效的视觉标记压缩依赖于强大的信息编码能力，这种能力在预训练的VLM中已经存在，但现有方法未能充分利用。基于此，我们提出了VisCo，一个训练效率高的自压缩框架，利用预训练的VLM本身作为内在压缩器。VisCo是一个参数共享的自编码器，使用一小组记忆标记压缩视觉信息，并将层次信息从编码转移到解码。实验表明，VisCo在所有评估的压缩比下均超过了先前的方法，在更激进的压缩下获得了更大的提升，并且在极端的单标记设置中仍然保持稳定。此外，当与原始视觉标记结合时，学习到的记忆标记甚至可以改善基础模型，表明VisCo捕捉到了超越压缩的互补表示。
cs.CV / 75 / 2607.12764
EvoGraph-R1: Self-Evolving Multimodal Knowledge Hypergraphs for Agentic Retrieval
EvoGraph-R1：自我演化的多模态知识超图用于智能检索
Abstract
Retrieval-augmented generation (RAG) has emerged as a critical paradigm for grounding Multimodal Large Language Models (MLLMs) in external knowledge. Recent GraphRAG methods introduce structured entity-relation graphs to improve retrieval and reasoning. However, they remain limited by treating knowledge graphs as static data structures built offline and queried in a single pass. This static paradigm misaligns with the interactive, iterative nature of knowledge-intensive reasoning, creating three bottlenecks: (i) text-centric fragmentation that impedes cross-modal reasoning, (ii) frozen structures unable to incorporate new evidence or correct errors, and (iii) rigid single-pass retrieval without adaptive refinement. To overcome these limitations, we introduce EvoGraph-R1, a self-evolving GraphRAG framework that reconceptualizes knowledge graphs as dynamic environments shaped through agent interactions. We formulate retrieval as a Markov Decision Process (MDP) where the agent observes the graph state and executes actions to query (GraphRetrieve), expand (WebSearch), refine (GraphEdit), or terminate (Answer) the reasoning. These actions reshape the hypergraph structure and generate feedback signals that guide subsequent evolution. Through this closed loop, the hypergraph evolves by integrating new evidence, correcting errors, and refining structure to support multi-hop reasoning. Experiments on multimodal VQA and text QA benchmarks demonstrate substantial improvements over existing RAG baselines in accuracy, coverage, and traceability, establishing self-evolving knowledge graphs as a fundamental paradigm across modalities.
Chinese Translation
检索增强生成（RAG）已成为将多模态大型语言模型（MLLMs）与外部知识相结合的重要范式。近期的GraphRAG方法引入了结构化的实体-关系图，以改善检索和推理。然而，它们仍然受到限制，因为知识图被视为离线构建的静态数据结构，并在单次查询中使用。这种静态范式与知识密集型推理的互动和迭代特性不符，造成了三个瓶颈：（i）以文本为中心的碎片化，阻碍了跨模态推理；（ii）无法融入新证据或纠正错误的静态结构；（iii）缺乏自适应细化的刚性单次检索。为克服这些局限性，我们提出了EvoGraph-R1，一个自我演化的GraphRAG框架，将知识图重新概念化为通过代理交互塑造的动态环境。我们将检索形式化为马尔可夫决策过程（MDP），其中代理观察图状态并执行查询（GraphRetrieve）、扩展（WebSearch）、细化（GraphEdit）或终止（Answer）推理的动作。这些动作重塑超图结构并生成反馈信号，以指导后续演化。通过这一闭环，超图通过整合新证据、纠正错误和细化结构来支持多跳推理。在多模态视觉问答（VQA）和文本问答基准上的实验表明，在准确性、覆盖率和可追溯性方面，相较于现有的RAG基线有显著提升，确立了自我演化知识图作为跨模态的基本范式。
cs.CV / 76 / 2607.12774
HSEmotion Team at the 11th ABAW Challenge: Multi-Task Learning and Ambivalence/Hesitancy Video Recognition
HSEmotion团队在第11届ABAW挑战赛中的表现：多任务学习与矛盾/犹豫视频识别
Abstract
This article presents our results for the 11th Affective Behavior Analysis in-the-Wild (ABAW) competition. For multi-task learning with simultaneous prediction of valence, arousal, facial expressions, and action units on s-Aff-Wild2 dataset, we use frozen lightweight facial extractors, MT-EmotiDDAMFN and MT-EmotiEffNet-B0, with separate heads and systematic post-processing: temporal Gaussian smoothing, per-class expression bias, AffectNet blending, per-AU threshold tuning, and weighted backbone fusion. On the official validation set, our ensemble significantly exceeds the performance of the ConvNeXt baseline. For ambivalence/hesitancy video recognition on the expanded BAH dataset, we extend the audiovisual pipeline to video-level Macro F1 by late fusion of face, HuBERT audio, and RoBERTa text classifiers, temporal aggregation, and a global-text gate. Frame-level Weighted F1 on validation set rises from 0.74 in ABAW-8 to 0.79, while the best public-test video-level Macro F1 reaches 0.73. In both tasks, competitive performance is achieved without fine-tuning heavy backbones. These results indicate that systematic prediction calibration and lightweight multimodal fusion can rival substantially heavier end-to-end approaches while offering improved efficiency and deployment flexibility.
Chinese Translation
本文展示了我们在第11届情感行为分析野外（ABAW）竞赛中的成果。针对s-Aff-Wild2数据集的多任务学习，同时预测情感价值、唤醒度、面部表情和动作单元，我们使用了冻结的轻量级面部特征提取器MT-EmotiDDAMFN和MT-EmotiEffNet-B0，配备独立的输出头和系统化的后处理：时间高斯平滑、每类表情偏差、AffectNet混合、每个动作单元阈值调优以及加权骨干融合。在官方验证集上，我们的集成模型显著超越了ConvNeXt基线。在扩展的BAH数据集上进行的矛盾/犹豫视频识别中，我们通过面部、HuBERT音频和RoBERTa文本分类器的晚期融合、时间聚合和全局文本门，将视听管道扩展到视频级别的宏观F1。验证集上的帧级加权F1从ABAW-8的0.74上升至0.79，而最佳公共测试视频级别的宏观F1达到0.73。在这两个任务中，未对重型骨干进行微调的情况下实现了竞争性的性能。这些结果表明，系统化的预测校准和轻量级多模态融合可以与显著更重的端到端方法相抗衡，同时提供更高的效率和部署灵活性。
cs.CV / 77 / 2607.12782
MBTI: A Multi-Branch Efficient Fine-Tuning Framework for Hyperspectral Image Classification with Foundation Models
MBTI：一种基于基础模型的多分支高效微调框架，用于高光谱图像分类
Abstract
Hyperspectral foundation models learn transferable spectral-spatial representations from large-scale unlabeled data. They provide an effective paradigm for adapting to downstream hyperspectral image (HSI) classification tasks with limited labeled samples. However, spectral band configurations vary substantially across sensors, which makes direct model transfer difficult. Existing adaptation strategies often compress, select, or reshape the original spectra to match model-specific input requirements. These operations may discard useful spectral information and weaken local spectral continuity. To address this problem, we propose MBTI, a Multi-Branch efficient fine-tuning framework for Hyperspectral Image classification. MBTI adapts hyperspectral foundation models to downstream classification tasks while preserving full-band spectral information. First, we introduce a spectral-continuity-preserving multi-branch preprocessing strategy. The original HSI is divided into multiple continuous spectral subsets, and a band reuse mechanism is used when the remaining bands cannot form a complete branch. This avoids invalid padding and unnecessary spectral loss. Second, independent Low-Rank Adaptation (LoRA) modules are inserted into each branch. They enable different spectral intervals to learn task-specific discriminative features while keeping most pre-trained parameters frozen. Finally, a multi-branch channel attention fusion module adaptively recalibrates and integrates features from all spectral branches. Experiments on three public hyperspectral datasets show that MBTI achieves competitive and superior performance compared with representative classification methods. Under the final rank-8 configuration, only about 2.33\%--2.36\% of the parameters are trainable. The code will be available at https://github.com/Azhenmiddleblock/MBTI/tree/main.
Chinese Translation
高光谱基础模型从大规模未标记数据中学习可转移的光谱-空间表示。它们为适应下游高光谱图像（HSI）分类任务提供了一种有效的范式，尤其是在标记样本有限的情况下。然而，不同传感器的光谱带配置差异显著，这使得直接模型迁移变得困难。现有的适应策略通常通过压缩、选择或重塑原始光谱来满足模型特定的输入要求。这些操作可能会丢弃有用的光谱信息，并削弱局部光谱连续性。为了解决这个问题，我们提出了MBTI，一种用于高光谱图像分类的多分支高效微调框架。MBTI在适应下游分类任务的同时，保留了全带光谱信息。首先，我们引入了一种光谱连续性保持的多分支预处理策略。原始HSI被划分为多个连续的光谱子集，当剩余的光谱带无法形成完整的分支时，采用带重用机制。这避免了无效的填充和不必要的光谱损失。其次，独立的低秩适应（Low-Rank Adaptation, LoRA）模块被插入到每个分支中。它们使不同的光谱区间能够学习任务特定的判别特征，同时保持大部分预训练参数不变。最后，一个多分支通道注意力融合模块自适应地重新校准和整合来自所有光谱分支的特征。在三个公共高光谱数据集上的实验表明，MBTI与代表性的分类方法相比，取得了具有竞争力和优越的性能。在最终的8级配置下，只有约2.33%--2.36%的参数是可训练的。代码将发布在 https://github.com/Azhenmiddleblock/MBTI/tree/main。
cs.CV / 78 / 2607.12785
ExtraGS: Enhancing Endoscopic View Extrapolation via Diffusion-Guided 3D Gaussian Splatting
ExtraGS：通过扩散引导的三维高斯散射增强内窥镜视图外推
Abstract
Robot-assisted minimally invasive surgery (MIS) critically depends on reliable endoscopic perception for navigation and safety. However, conventional endoscopes provide only a limited field of view, leaving large portions of surrounding anatomy unobserved. Recent neural rendering approaches, such as Neural Radiance Fields and 3D Gaussian Splatting, enable novel view synthesis from endoscopic videos, but their reliance on sparse observations often leads to severe artifacts when extrapolating beyond the training trajectory.In this work, we propose ExtraGS, a framework for enhancing endoscopic view extrapolation via diffusion-guided 3D Gaussian Splatting. Starting from an initial reconstruction, we introduce an uncertainty-guided virtual camera sampling strategy to actively explore blind spots and maximize information gain. The rendered views from these sampled locations are refined using a diffusion model to recover plausible anatomical structures, producing pseudo observations that guide further optimization. To prevent the generated content from degrading reliable regions, we adopt a confidence-weighted fine-tuning strategy when incorporating these pseudo observations.Extensive experiments on multiple public endoscopic datasets demonstrate that ExtraGS significantly reduces extrapolation artifacts and achieves state-of-the-art performance in endoscopic novel view synthesis.
Chinese Translation
机器人辅助的微创手术（MIS）在导航和安全性方面严重依赖可靠的内窥镜感知。然而，传统内窥镜仅提供有限的视野，导致周围解剖结构的大部分未被观察到。最近的神经渲染方法，如神经辐射场（Neural Radiance Fields）和三维高斯散射（3D Gaussian Splatting），能够从内窥镜视频中合成新视图，但它们对稀疏观测的依赖常常导致在训练轨迹之外外推时出现严重伪影。在本研究中，我们提出了ExtraGS，一个通过扩散引导的三维高斯散射增强内窥镜视图外推的框架。从初始重建开始，我们引入了一种不确定性引导的虚拟相机采样策略，以主动探索盲点并最大化信息增益。来自这些采样位置的渲染视图通过扩散模型进行精细化，以恢复合理的解剖结构，生成伪观测以指导进一步优化。为了防止生成内容对可靠区域的降级，我们在整合这些伪观测时采用了置信度加权的微调策略。在多个公共内窥镜数据集上的广泛实验表明，ExtraGS显著减少了外推伪影，并在内窥镜新视图合成中实现了最先进的性能。
cs.CV / 79 / 2607.12786
CoRe: A Comprehensive Framework for Cross-Image Comparative Reasoning in Vision-Language Models
CoRe：视觉-语言模型中跨图像比较推理的综合框架
Abstract
Cross-image comparative reasoning remains challenging for vision-language models (VLMs), especially when correct prediction requires fine-grained attribute grounding and globally consistent reasoning. We present CoRe, a unified framework for this problem. CoRe includes: (i) CoRe-20K, a large-scale triplet-based training set automatically constructed from structured visual metadata through a multi-expert collaborative pipeline, covering counting, depth, distance, and spatial relations; (ii) TriSR, a structured reward framework that jointly supervises attribute grounding, judgment alignment, and triplet consistency under GRPO optimization; and (iii) CoRe-Bench, the first benchmark dedicated to fine-grained cross-image comparative reasoning. Experiments show that CoRe substantially outperforms existing VLMs on CoRe-Bench while remaining competitive on standard multimodal benchmarks, achieving a 28.2-point gain in partial accuracy over the strongest baseline.
Chinese Translation
跨图像比较推理对于视觉-语言模型（VLMs）仍然具有挑战性，尤其是在正确预测需要细粒度属性定位和全局一致推理的情况下。我们提出了CoRe，这是一个针对该问题的统一框架。CoRe包括：（i）CoRe-20K，一个基于三元组的大规模训练集，通过多专家协作流程从结构化视觉元数据中自动构建，涵盖计数、深度、距离和空间关系；（ii）TriSR，一个结构化奖励框架，在GRPO优化下共同监督属性定位、判断对齐和三元组一致性；以及（iii）CoRe-Bench，第一个专注于细粒度跨图像比较推理的基准测试。实验表明，CoRe在CoRe-Bench上显著优于现有的VLMs，同时在标准多模态基准上保持竞争力，相较于最强基线实现了28.2点的部分准确率提升。
cs.CV / 80 / 2607.12800
UniVR: Thinking in Visual Space for Unified Visual Reasoning
UniVR：在视觉空间中进行统一视觉推理
Abstract
Learning broad world knowledge directly from raw visual data is a fundamental capability of intelligence. We introduce UniVR, the first investigation into simultaneously learning complex reasoning, fine-grained physical dynamics, and long-term planning from pure visual demonstrations. At its core, UniVR features VR-GRPO, a reinforcement learning paradigm with complementary global and step-level rewards. This approach enforces logical coherence and physical consistency throughout the reasoning process without requiring task-specific heuristics or image-text pairs. To train and evaluate UniVR, we construct VR-X, a large-scale benchmark curated from 16 diverse sources spanning long-horizon manipulation, spatial puzzles, and physical reasoning. It is the first comprehensive suite to assess these heterogeneous capabilities under a purely visual protocol. Remarkably, UniVR achieves up to a 25% improvement on VR-X, and its superior visual reasoning also boosts performance on various multimodal understanding benchmarks. These findings underscore the vast potential of reasoning within visual spaces, with all code, data, and models are open-sourced for further research.
Chinese Translation
从原始视觉数据中学习广泛的世界知识是智能的一项基本能力。我们介绍了UniVR，这是首次探讨如何从纯视觉演示中同时学习复杂推理、细粒度物理动态和长期规划。UniVR的核心是VR-GRPO，这是一种具有互补全局和逐步奖励的强化学习范式。该方法在推理过程中强制执行逻辑一致性和物理一致性，而无需特定任务的启发式方法或图像-文本对。为了训练和评估UniVR，我们构建了VR-X，这是一个大型基准，汇集了来自16个不同来源的数据，涵盖了长期操作、空间难题和物理推理。这是第一个在纯视觉协议下评估这些异构能力的综合套件。值得注意的是，UniVR在VR-X上实现了高达25%的性能提升，其卓越的视觉推理能力也提升了在各种多模态理解基准上的表现。这些发现强调了在视觉空间中推理的巨大潜力，所有代码、数据和模型均已开源以供进一步研究。
cs.CV / 81 / 2607.12818
Breaking D\'ej\`a Vu: Independent Auditing of Visual Place Recognition through Vision-Language Reasoning
打破既视感：通过视觉-语言推理进行视觉位置识别的独立审计
Abstract
Visual place recognition (VPR) is a key enabler of accurate localization and long-term autonomous navigation in robotics applications, such as loop closure detection for simultaneous localisation and mapping (SLAM). However, real-world VPR deployment relies on selecting an image matching threshold that balances precision and recall. These thresholds are typically tuned using labeled validation data and fixed during deployment, making them unreliable under environmental changes where ground truth is unavailable. This is particularly problematic in safety-critical robotics, where accepting a false loop closure can corrupt the estimated trajectory and map. In this work, we introduce Visual Place Recognition Auditing, an independent post-retrieval verification framework that leverages Vision-Language Models (VLMs) to assess retrieved matches by reasoning jointly over query and candidate images. Unlike conventional verification methods, our approach performs instance-level verification without requiring architecture-specific confidence measures, dataset-dependent thresholds, or prior knowledge of the deployment environment. We evaluate our method on six benchmark datasets using five state-of-the-art VPR methods and four VLMs. Results show that VLM-based auditing improves recall@1 by 13.6% on average as compared to state-of-the-art methods while reducing false acceptance rates to 12%, maintaining precision above 95% and coverage above 75%.
Chinese Translation
视觉位置识别（VPR）是实现精确定位和长期自主导航的关键，广泛应用于机器人领域，例如用于同时定位与地图构建（SLAM）的回环闭合检测。然而，现实世界中的 VPR 部署依赖于选择一个图像匹配阈值，以平衡精确率和召回率。这些阈值通常使用标记的验证数据进行调整，并在部署期间保持不变，这使得它们在环境变化且缺乏真实情况时变得不可靠。这在安全关键的机器人应用中尤其成问题，因为接受错误的回环闭合可能会破坏估计的轨迹和地图。在本研究中，我们引入了视觉位置识别审计（Visual Place Recognition Auditing），这是一个独立的后检索验证框架，利用视觉-语言模型（Vision-Language Models, VLMs）通过对查询图像和候选图像的联合推理来评估检索到的匹配。与传统的验证方法不同，我们的方法在实例级别进行验证，无需特定于架构的置信度度量、依赖于数据集的阈值或对部署环境的先验知识。我们在六个基准数据集上评估了我们的方法，使用了五种最先进的 VPR 方法和四种 VLM。结果表明，与最先进的方法相比，基于 VLM 的审计在召回率@1 上平均提高了 13.6%，同时将错误接受率降低到 12%，保持精确率超过 95% 和覆盖率超过 75%。
cs.CV / 82 / 2607.12820
AVSCap: Orchestrating Audio-Visual Synergy for Omni-modal Video Captioning
AVSCap：协调音频-视觉协同以实现全模态视频字幕生成
Abstract
Omni-modal video captioning is not merely combining visual captioning with audio transcription: a useful caption must describe how visual actions, speech, music, and sound effects co-evolve. Existing large multimodal models often fail at this relational step, treating audio and visual streams as loosely coupled observations, relying on automatic speech recognition, and under-specifying non-speech sounds and their links to visual events. We present AVSCap, a framework for audio-visual captioning centered on explicit cross-modal event binding. First, we construct AVSCap-130K, a tri-modal training corpus generated by a decoupled-then-fused pipeline that anchors visual and acoustic evidence before composing grounded omni-modal captions. Second, we train AVSCap-7B, a 7B captioner with a two-stage strategy: supervised fine-tuning establishes baseline capabilities, while sample-efficient reinforcement learning uses hybrid rewards to optimize acoustic completeness and audio-visual synergy. Our scaling analysis shows that reinforcement learning brings larger gains than increasing SFT data. Third, we introduce AVSCapBench, a benchmark that decomposes captions into visual, audio, and synergy events and evaluates them with fine-grained event recall. Experiments on AVSCapBench and external benchmarks show that AVSCap-7B improves non-speech audio coverage and cross-modal binding, delivering the best overall performance among evaluated open-source models.
Chinese Translation
全模态视频字幕生成不仅仅是将视觉字幕与音频转录相结合：有效的字幕必须描述视觉动作、语言、音乐和音效如何共同演变。现有的大型多模态模型往往在这一关系步骤上表现不佳，将音频和视觉流视为松散耦合的观察，依赖于自动语音识别，并未充分指定非语音声音及其与视觉事件的联系。我们提出了AVSCap，一个以显式跨模态事件绑定为中心的音频-视觉字幕生成框架。首先，我们构建了AVSCap-130K，这是一个通过解耦-再融合管道生成的三模态训练语料库，在构建有根基的全模态字幕之前，锚定视觉和声学证据。其次，我们训练了AVSCap-7B，这是一个具有两阶段策略的7B字幕生成器：监督微调建立基线能力，而样本高效的强化学习则利用混合奖励来优化声学完整性和音频-视觉协同。我们的规模分析表明，强化学习带来的收益大于增加监督微调数据。第三，我们引入了AVSCapBench，一个将字幕分解为视觉、音频和协同事件的基准，并通过细粒度事件召回进行评估。在AVSCapBench和外部基准上的实验表明，AVSCap-7B改善了非语音音频的覆盖率和跨模态绑定，在评估的开源模型中提供了最佳的整体性能。
cs.CV / 83 / 2607.12858
LARAD: Layout-Aware Road Anomaly Detection via Spatial-Logic Reasoning
LARAD：基于空间逻辑推理的布局感知道路异常检测
Abstract
Accurate open-world obstacle detection is critical for autonomous driving. Current anomaly segmentation methods suffer from a fundamental blind spot: they over-rely on texture novelty to identify out-of-distribution (OoD) objects while ignoring contextual spatial logic. Furthermore, mitigating the resulting false positives often requires cascading massive vision models, introducing unacceptable inference latency. To address these issues, we propose Layout-Aware Road Anomaly Detection (LARAD), shifting the paradigm from appearance matching to spatial-logic reasoning. First, we introduce the Spatial-Logic Violation Synthesis (SLVS) pipeline, which generates training samples that are texture-consistent yet spatially invalid, forcing the model to learn contextual violations. Second, we augment a standard closed-set segmentation network with a lightweight, OoD-guided attention branch. Extensive experiments demonstrate that LARAD significantly enhances robustness against logical anomalies and establishes a new state-of-the-art, all while retaining the high efficiency of a single-model architecture.
Chinese Translation
准确的开放世界障碍物检测对自动驾驶至关重要。目前的异常分割方法存在一个根本性的盲点：它们过于依赖纹理新颖性来识别分布外（OoD）对象，而忽视了上下文空间逻辑。此外，缓解由此产生的误报通常需要级联大量视觉模型，这会引入不可接受的推理延迟。为了解决这些问题，我们提出了布局感知道路异常检测（LARAD），将范式从外观匹配转变为空间逻辑推理。首先，我们引入了空间逻辑违反合成（SLVS）管道，该管道生成纹理一致但空间无效的训练样本，迫使模型学习上下文违反。其次，我们在标准的闭集分割网络中增强了一个轻量级的、基于OoD引导的注意力分支。大量实验表明，LARAD显著增强了对逻辑异常的鲁棒性，并建立了新的最先进水平，同时保持了单模型架构的高效性。
cs.CV / 84 / 2607.12866
Statistical Non-linear Reconstruction Loss for Image Anomaly Detection
用于图像异常检测的统计非线性重建损失
Abstract
Reconstruction-based methods are a cornerstone of unsupervised image anomaly detection, but they remain vulnerable to \emph{outlier leakage}, where standard mean squared error (MSE) loss drives the model to faithfully reconstruct anomalous patterns. We propose a Non-linear Reconstruction Loss that applies a sigmoid-based squashing function to suppress high-magnitude features, preventing outliers from dominating optimization while preserving sensitivity to normal patterns. In addition, we introduce a statistical calibration scheme that selects the scaling factor $k$ from the confidence interval (CI) of the normal feature distribution, enabling data-driven control of the suppression strength. Our approach achieves competitive or superior anomaly detection performance compared to state-of-the-art methods, reaching 99.0\% Image-AUROC and 97.3\% Pixel-AUROC on MVTec-AD, and 95.3\% Image-AUROC and 99.0\% Pixel-AUROC on VisA. These results indicate that non-linear gradient suppression is an effective mechanism for mitigating outlier leakage and improving anomaly localization in unified industrial inspection settings. The implementation is available at https://github.com/mintii13/Statistical-Non-linear-Reconstruction-Loss.git.
Chinese Translation
基于重建的方法是无监督图像异常检测的基石，但它们仍然容易受到 extit{异常值泄漏}的影响，其中标准均方误差（MSE）损失驱动模型忠实地重建异常模式。我们提出了一种非线性重建损失，该损失应用基于sigmoid的压缩函数来抑制高幅度特征，防止异常值主导优化，同时保持对正常模式的敏感性。此外，我们引入了一种统计校准方案，从正常特征分布的置信区间（CI）中选择缩放因子$k$，实现对抑制强度的数据驱动控制。与最先进的方法相比，我们的方法在异常检测性能上达到了竞争性或优越的表现，在MVTec-AD上达到了99.0 ext{%}的图像AUROC和97.3 ext{%}的像素AUROC，在VisA上达到了95.3 ext{%}的图像AUROC和99.0 ext{%}的像素AUROC。这些结果表明，非线性梯度抑制是一种有效的机制，可以减轻异常值泄漏并改善统一工业检测环境中的异常定位。实现代码可在https://github.com/mintii13/Statistical-Non-linear-Reconstruction-Loss.git获取。
cs.CV / 85 / 2607.12874
Metric-Guided Synthetic Image Data Rendering for Deep Learning compatible with Agentic AI
适用于代理人工智能的深度学习的度量引导合成图像数据渲染
Abstract
Deep learning computer vision for scientific applications requires collecting and annotating large datasets in a laborious, expensive and error-prone process. Synthetic data generation through 3D modelling and rendering may simplify this process and increase the accuracy of annotations by generating them programmatically. However, minimising the domain gap between real and synthetic images visually is subjective and lacks systematic quantitative guidance. We present GraNatPy, a Python package with metrics to guide improvement of the rendered scene. We show that quantifiable increase in realism, diversity and size of rendered dataset correlates with improved visual perception of the scene and higher zero-shot performance of an object detection model. Furthermore, we demonstrated using photographs of virological plaque assays that gradient similarity affects performance on small object detection, which can be improved by mixing real and synthetic data. Finally, we turn procedural data rendering into an agentic skill (SynthClaw) to automate the procedural parameter optimisation.
Chinese Translation
深度学习计算机视觉在科学应用中需要收集和标注大量数据集，这一过程既费力又昂贵，且容易出错。通过3D建模和渲染生成合成数据可能简化这一过程，并通过程序化生成提高标注的准确性。然而，最小化真实图像与合成图像之间的领域差距在视觉上是主观的，缺乏系统的定量指导。我们提出了GraNatPy，一个带有度量的Python包，用于指导渲染场景的改进。我们展示了可量化的真实感、多样性和渲染数据集规模的增加与场景的视觉感知改善及物体检测模型的零-shot性能提升之间的相关性。此外，我们通过使用病毒斑点测定法的照片证明了梯度相似性对小物体检测性能的影响，这一性能可以通过混合真实数据和合成数据来改善。最后，我们将程序化数据渲染转变为一种代理技能（SynthClaw），以自动化程序参数优化。
cs.CV / 86 / 2607.12881
Inhibited Self-Attention: Sharpening Focus in Vision Transformers
抑制自注意力：在视觉变换器中聚焦
Abstract
Vision Transformers (ViTs) have demonstrated remarkable performance in computer vision tasks. However, their self-attention mechanism often diffuses focus across background regions, relying on spurious correlations rather than object-relevant cues. Inspired by inhibitory mechanisms observed in biological vision systems, we propose the Inhibited Self-Attention (ISA), a novel self-attention that integrates inhibitory signals to enhance feature selectivity and suppress spurious responses. In contrast to conventional self-attention, which relies solely on positive attention values due to softmax normalization, our approach retains and utilizes negative attention scores to suppress irrelevant features and sharpen focus on objects of interest. Experiments across multiple datasets, including ImageNet-1k and COCO, and several robustness benchmarks demonstrate that ISA enhances object-centric selectivity, reduces shortcut reliance, and improves out-of-distribution generalization. Our analysis of relevance maps confirms that ViTs with ISA exhibit sharper, more localized focus on object-relevant regions while reducing distractions from non-relevant (background) features, enabling more reliable models. We release our code at https://github.com/prdvanderwal/inhibited-self-attention
Chinese Translation
视觉变换器（Vision Transformers, ViTs）在计算机视觉任务中表现出色。然而，它们的自注意力机制常常将注意力分散到背景区域，依赖于虚假的相关性而非与物体相关的线索。受到生物视觉系统中观察到的抑制机制的启发，我们提出了抑制自注意力（Inhibited Self-Attention, ISA），这是一种新颖的自注意力机制，整合了抑制信号以增强特征选择性并抑制虚假响应。与传统的自注意力机制仅依赖于正向注意值（由于softmax归一化）不同，我们的方法保留并利用负向注意分数，以抑制无关特征并聚焦于感兴趣的物体。在多个数据集（包括ImageNet-1k和COCO）及若干鲁棒性基准测试中的实验表明，ISA增强了以物体为中心的选择性，减少了对捷径的依赖，并改善了分布外泛化。我们对相关性图的分析确认，具有ISA的ViTs在与物体相关的区域上表现出更尖锐、更局部的关注，同时减少了来自无关（背景）特征的干扰，从而使模型更加可靠。我们的代码已发布在 https://github.com/prdvanderwal/inhibited-self-attention
cs.CV / 87 / 2607.12894
Hy-Embodied-VLM-1.0: Efficient Physical-World Agents
Hy-Embodied-VLM-1.0：高效的物理世界代理
Wang, Ziyi, Yu, Xumin, Rao, Yongming, Ling, Yonggen, Li, Yunheng, Wang, Oran, Gao, Mingqi, Zhou, Yuchen, Liang, Yves, Liu, Zuyan, Zhang, Yani, Huang, Rui, Xu, Xiaoran, Yuan, Bowen, Yuan, Yifu, Tan, Xu, Zhang, He, Huang, Yufei, Zhang, Shenghao, Wu, Hongsheng, Hu, Han, Zhang, Zhengyou
Abstract
Building capable embodied agents requires not only multimodal perception and understanding, but also agentic capabilities for reasoning about actions, adapting to evolving situations, and interacting with the physical world. In this report, we introduce Hy-Embodied-VLM-1.0, an efficient and powerful embodied foundation model specifically designed for embodied agents operating in the physical world. To cultivate such capabilities from the pre-training stage onward, we define an action-centric capability taxonomy comprising three progressive dimensions: Action-Relevant State Understanding, Action-Transition Reasoning, and Sequential and Adaptive Reasoning. Guided by this taxonomy, we develop a systematic data pipeline and curate data mixtures spanning both pre-training and post-training. To deliver strong physical-world understanding and interaction capabilities while supporting latency-sensitive deployment, we build our model on the Hy3-A3B language backbone and the Hy-ViT2 vision encoder. Its efficient Mixture-of-Experts architecture combines strong model capacity with high inference efficiency. We evaluate Hy-Embodied-VLM-1.0 on a comprehensive suite of 38 benchmarks covering embodied perception, physical-world understanding, and embodied reasoning. The model achieves the best performance among similarly sized models on 19 of the 38 benchmarks and substantially outperforms strong competitors, including Qwen3.6-A3B and Cosmos 3. Compared with the previous-generation Hy-Embodied-0.5 MoT-2B, Hy-Embodied-VLM-1.0 improves average performance by 8.4%. Despite activating only 3B parameters, it achieves performance close to that of the previous-generation model with 32B activated parameters. Beyond static benchmark evaluation, Hy-Embodied-VLM-1.0 also demonstrates strong performance on embodied agentic tasks requiring multi-turn interaction and long-horizon reasoning.
Chinese Translation
构建有能力的具身代理不仅需要多模态感知和理解，还需要具备关于行动的推理能力、适应不断变化的情况以及与物理世界互动的能力。在本报告中，我们介绍了Hy-Embodied-VLM-1.0，这是一种高效且强大的具身基础模型，专门为在物理世界中操作的具身代理设计。为了从预训练阶段开始培养这些能力，我们定义了一个以行动为中心的能力分类法，包含三个渐进的维度：与行动相关的状态理解、行动转变推理以及顺序和自适应推理。在这一分类法的指导下，我们开发了一个系统的数据管道，并策划了涵盖预训练和后训练的数据混合。为了提供强大的物理世界理解和互动能力，同时支持对延迟敏感的部署，我们的模型基于Hy3-A3B语言骨干和Hy-ViT2视觉编码器。其高效的专家混合架构结合了强大的模型容量和高推理效率。我们在涵盖具身感知、物理世界理解和具身推理的38个基准测试中评估了Hy-Embodied-VLM-1.0。该模型在38个基准中的19个上实现了同类模型中的最佳性能，并显著超越了强有力的竞争对手，包括Qwen3.6-A3B和Cosmos 3。与上一代的Hy-Embodied-0.5 MoT-2B相比，Hy-Embodied-VLM-1.0的平均性能提高了8.4%。尽管仅激活了3B参数，但其性能接近于激活了32B参数的上一代模型。除了静态基准评估外，Hy-Embodied-VLM-1.0在需要多轮互动和长时间推理的具身代理任务中也表现出色。
cs.CV / 88 / 2607.12896
UniMedSeg: Unified In-Context Learning for Multi-Paradigm 2D/3D Medical Image Segmentation
UniMedSeg：用于多范式2D/3D医学图像分割的统一上下文学习
Abstract
Medical image segmentation foundation models are expected to generalize across diverse clinical scenarios, yet existing universal methods remain fragmented by prompt paradigms and spatial dimensions. Visual in-context learning, interactive segmentation, and language-guided segmentation are typically handled by paradigm-specific models, while 2D and 3D images are also modeled separately. Such isolation prevents heterogeneous annotations and data from being jointly absorbed by a single scalable model and limits cross-paradigm knowledge transfer. To address this bottleneck, we propose UniMedSeg, a Transformer-centric universal segmentation framework that maps visual examples, geometric interactions, language instructions, and 2D/3D images into a shared sequence space, enabling heterogeneous medical supervision to be jointly learned through a unified in-context interface without prompt- or dimension-specific branches. To overcome the long-sequence memory bottleneck caused by visual contexts, we introduce Decoupled Split Attention, which reduces attention complexity to linear while preserving hardware-friendly computation and focused context-target interaction. Extensively trained and evaluated on a large corpus curated from 27 public datasets, UniMedSeg achieves state-of-the-art performance across visual in-context, interactive, and language-guided segmentation without task-specific fine-tuning, demonstrating strong generalization on diverse held-out tasks. The code and model weights are publicly available at https://github.com/Lii1228/UniMedSeg
Chinese Translation
医学图像分割基础模型预计能够在多样的临床场景中进行泛化，然而现有的通用方法仍然因提示范式和空间维度而呈现出碎片化。视觉上下文学习、交互式分割和语言引导分割通常由特定范式的模型处理，而2D和3D图像也被分别建模。这种孤立阻碍了异构注释和数据被单一可扩展模型共同吸收，并限制了跨范式知识的转移。为了解决这一瓶颈，我们提出了UniMedSeg，一个以Transformer为中心的通用分割框架，它将视觉示例、几何交互、语言指令和2D/3D图像映射到一个共享的序列空间，从而使异构医学监督能够通过统一的上下文接口共同学习，而无需特定于提示或维度的分支。为了克服由视觉上下文引起的长序列记忆瓶颈，我们引入了解耦分离注意力（Decoupled Split Attention），它将注意力复杂度降低到线性，同时保持硬件友好的计算和集中上下文-目标交互。在从27个公共数据集中整理的大型语料库上进行广泛训练和评估后，UniMedSeg在视觉上下文、交互式和语言引导分割方面实现了最先进的性能，而无需特定任务的微调，展示了在多样化保留任务上的强泛化能力。代码和模型权重可在 https://github.com/Lii1228/UniMedSeg 获取。
cs.CV / 89 / 2607.12903
Rank-1 Identity Consensus Predicts Gallery Enrollment in 1:N Face Matching More Accurately than Score Thresholding
Rank-1身份共识比得分阈值更准确地预测1:N人脸匹配中的图库注册
Abstract
In operational 1:N face identification, a crucial question arises for each probe: is this person enrolled in the gallery or not? The stakes are high and asymmetric. Rejecting a mate-present (MP) probe loses a valid lead; accepting a mate-absent (MA) probe makes every returned candidate a false identification, at worst a wrongful arrest. Most approaches threshold match scores, but scores shift substantially with image quality and gallery size and composition, making thresholds fixed before deployment brittle under realistic conditions. Our prior work introduced 1-consistency, the only method based on rank consensus across multiple independently trained matchers: a probe is labeled MP if all matchers return the same rank-1 identity. This work stress-tests 1-consistency across 36 (gallery, probe quality) scenarios spanning four quality levels and two structural axes: images per identity and total enrolled identities. We benchmark against two score-thresholding methods that bracket what any deployed threshold could achieve. Fixed Score-Thresholding (FST), calibrated once on baseline conditions, collapses asymmetrically as quality degrades: MP recall falls below 2% while MA recall holds near 100%. Oracle Score-Thresholding (OST), re-tuned per scenario, is the best any threshold could theoretically do, yet for degraded probes 1-consistency matches it with zero tuning. The two differ mainly in error type (OST favors MP recall, 1-consistency favors MA recall), but on one axis 1-consistency does not merely match the oracle: when it labels a probe MP, it returns the correct mate 97-100% of the time versus OST's 66-84% under severe degradation. In short, 1-consistency delivers oracle-level accuracy without the impossible requirement: it sets no threshold, so it needs no advance knowledge of the conditions a probe will arrive in, which is what makes it usable.
Chinese Translation
在操作性的1:N人脸识别中，每个探测样本面临一个关键问题：此人是否已注册在图库中？风险高且不对称。拒绝一个存在配偶（MP）的探测样本会失去一个有效线索；而接受一个不存在配偶（MA）的探测样本则使得每个返回的候选者都成为错误识别，最严重的情况可能导致错误逮捕。大多数方法采用阈值匹配得分，但得分会因图像质量、图库大小和构成而显著变化，使得在实际条件下预先设定的阈值变得脆弱。我们之前的工作引入了1-一致性，这是唯一基于多个独立训练的匹配器之间的排名共识的方法：如果所有匹配器返回相同的排名-1身份，则将探测样本标记为MP。本研究在36种（图库、探测质量）场景下对1-一致性进行了压力测试，涵盖四个质量水平和两个结构轴：每个身份的图像数量和总注册身份数。我们与两种得分阈值方法进行了基准测试，这些方法界定了任何部署阈值可能达到的范围。固定得分阈值（FST）在基线条件下校准一次，随着质量下降而不对称崩溃：MP召回率降至2%以下，而MA召回率保持在接近100%。每种场景重新调整的神谕得分阈值（OST）是理论上任何阈值能够达到的最佳效果，但对于降级的探测样本，1-一致性在没有任何调整的情况下与之匹配。这两者主要在错误类型上有所不同（OST偏向于MP召回，1-一致性偏向于MA召回），但在一个轴上，1-一致性不仅仅是与神谕匹配：当它将探测样本标记为MP时，它返回正确配偶的概率为97-100%，而OST在严重降级下的概率为66-84%。简而言之，1-一致性在没有不可能要求的情况下提供了神谕级的准确性：它不设定阈值，因此不需要提前了解探测样本将到达的条件，这使得它可用。
cs.CV / 90 / 2607.12911
Open-KNEAD: Knowledge-grounded Nutrition Estimation via Agentic Decomposition
Open-KNEAD：基于知识的营养估计通过主动分解
Abstract
Multimodal Large Language Models (MLLMs) are increasingly used for dietary assessment from meal images, where retrieval-augmented grounding was shown to sharpen nutrition estimates. However, we find this premise no longer holds for current MLLMs. A modern MLLM's direct estimate now matches or surpasses the full retrieval pipeline. This raises a question: if retrieval no longer improves the overall estimate, can it still deliver the two things clinicians value, accurate portions and a traceable, item-by-item record? We pursue this while preserving what matters for clinical adoption: minimal user burden (a single, unannotated meal image), explainability (an auditable record), and privacy (locally hosted inference). We introduce Open-KNEAD, a knowledge-grounded agentic framework for meal nutrition estimation that is training-free and locally deployable. Each decomposed food item is grounded to a Food and Nutrient Database for Dietary Studies (FNDDS) code via selective, nutrient-aware retrieval, composing an auditable per-item record. Across two open MLLM families and three cuisines, Open-KNEAD improves portion estimates over both prior grounding methods and direct estimation in most backbone-dataset settings. An agent-internal recipe-prior step further recovers the invisible cooking-added energy that biases estimates on non-US cuisine. The advantage is largest on the dietitian-verified ACETADA dataset, where the local open agent surpasses the direct portion estimates of two frontier closed models by roughly $30\%$ and $53\%$, all while keeping every meal image on local hardware. We release the Open-KNEAD framework and its agent-ready FNDDS knowledge base.
Chinese Translation
多模态大型语言模型（MLLMs）在通过餐食图像进行饮食评估中越来越多地被使用，其中检索增强的基础使营养估计更加准确。然而，我们发现这一前提对于当前的MLLMs不再成立。现代MLLM的直接估计现在与完整的检索管道相匹配或超越。这引发了一个问题：如果检索不再改善整体估计，它是否仍然能够提供临床医生重视的两样东西，即准确的份量和可追溯的逐项记录？我们在保持临床采用所需的关键要素的同时追求这一目标：最小的用户负担（单个未标注的餐食图像）、可解释性（可审计的记录）和隐私（本地托管推理）。我们提出了Open-KNEAD，一个基于知识的主动框架，用于餐食营养估计，具有无训练和本地可部署的特点。每个分解的食品项通过选择性、营养意识的检索与饮食研究食品和营养数据库（FNDDS）代码相结合，形成可审计的逐项记录。在两个开放的MLLM家族和三种菜系中，Open-KNEAD在大多数基础数据集设置中改善了对比先前基础方法和直接估计的份量估计。一个内部代理的食谱优先步骤进一步恢复了在非美国菜系中偏置估计的不可见烹饪附加能量。在经过营养师验证的ACETADA数据集上，这一优势最大，当地开放代理的直接份量估计超越了两个前沿封闭模型的估计，约为$30\%$和$53\%$，同时保持每个餐食图像在本地硬件上。我们发布了Open-KNEAD框架及其代理准备的FNDDS知识库。
cs.CV / 91 / 2607.12934
Domain-Incremental Remote Sensing Change Detection via Difference-Guided Adaptation and Frequency-Decoupled Distillation
基于差异引导适应和频率解耦蒸馏的领域增量遥感变化检测
Abstract
Remote sensing change detection (RSCD) models are prone to catastrophic forgetting when incrementally adapted to new domains. Existing domain-incremental learning (DIL) methods mainly preserve image-level representations but often overlook bitemporal discrepancy cues, which are critical for robust change detection under domain shifts. To address this limitation, we propose DG-FDD, a domain-incremental change detection framework that integrates Difference-Guided Adaptation and Frequency-Decoupled Distillation. Specifically, the Difference-Guided Dynamic Adapter (DGDA) models bitemporal feature discrepancies to promote change-aware feature adaptation and reduce domain-specific interference. Meanwhile, the Frequency-Decoupled Knowledge Distillation strategy with Cross-domain Synthesis (FDKD-CS) separates structural information from domain style in the frequency domain, enabling stable knowledge transfer without historical data. Extensive experiments on three public high-resolution RSCD datasets under two- and three-domain incremental protocols demonstrate that DG-FDD effectively mitigates catastrophic forgetting. Compared with independently trained single-task models, DG-FDD records mean relative changes in F1 and IoU of only -0.23% and -0.45%, respectively, across six two-domain sequences, and -0.69% and -1.31%, respectively, across the three evaluated three-domain sequences. These results indicate a favorable stability-plasticity balance between historical knowledge retention and new-domain adaptation in continual cross-domain change detection.
Chinese Translation
遥感变化检测（RSCD）模型在逐步适应新领域时容易出现灾难性遗忘。现有的领域增量学习（DIL）方法主要保留图像级表示，但往往忽视了双时相差异线索，而这些线索对于在领域转移下进行稳健的变化检测至关重要。为了解决这一局限性，我们提出了DG-FDD，一个集成了差异引导适应和频率解耦蒸馏的领域增量变化检测框架。具体而言，差异引导动态适配器（DGDA）建模双时相特征差异，以促进变化感知特征适应并减少领域特定干扰。同时，结合跨领域合成的频率解耦知识蒸馏策略（FDKD-CS）在频率域中将结构信息与领域风格分离，从而实现稳定的知识转移而无需历史数据。在两个和三个领域增量协议下对三个公共高分辨率RSCD数据集的广泛实验表明，DG-FDD有效减轻了灾难性遗忘。与独立训练的单任务模型相比，DG-FDD在六个双领域序列中F1和IoU的平均相对变化仅为-0.23%和-0.45%，在评估的三个三领域序列中分别为-0.69%和-1.31%。这些结果表明，在持续的跨领域变化检测中，历史知识保留与新领域适应之间存在良好的稳定性与可塑性平衡。
cs.CV / 92 / 2607.12939
Point Tracking in Surgery--The 2025 Surgical Tattoos in Infrared Challenge (STIRC2025)
手术中的点跟踪——2025年红外手术纹身挑战赛 (STIRC2025)
Schmidt, Adam, Karaoglu, Mert Asim, Wu, Zijian, Zhang, Jiaming, Chen, Yuxin, Salcudean, Tim, Ha, Ho-Gun, Jang, Minkang, Jung, Kyungmin, Ullah, Ihsan, Lee, Hyunki, Guttikonda, Suresh, Latus, Sarah, Schlaefer, Alexander, Zhao, Xinkai, Hayashi, Yuichiro, Oda, Masahiro, Kitasaka, Takayuki, Mori, Kensaku, Liu, Peng, Li, Chenyang, Speidel, Stefanie, Gardiner, Aoife, Stilli, Agostino, Stoyanov, Danail, Vasconcelos, Francisco, Choudhuri, Anwesa, Zheng, Meng, Gao, Zhongpai, Planche, Benjamin, Nguyen, Van Nguyen, Chen, Terrence, Wu, Ziyan, Ladikos, Alexander, Mohareri, Omid
Abstract
Point tracking in surgery is crucial to enable applications in downstream tasks such as segmentation, 3D reconstruction, virtual tissue landmarking, autonomous probe-based scanning, and subtask autonomy. This paper introduces the 2025 iteration of a point tracking challenge to address this, wherein participants submit their algorithms for quantification. Their algorithms are evaluated using a dataset named surgical tattoos in infrared (STIR), with the challenge named the STIR Challenge 2025 (STIRC2025). The STIR Challenge 2025 comprises two quantitative components: accuracy and efficiency. The accuracy component tests the accuracy of algorithms on in vivo and ex vivo sequences. The efficiency component tests algorithm inference latency. The challenge was conducted as a part of MICCAI EndoVis 2025, and seven teams participated in this challenge. In this paper we summarize the challenge results and participant methods. The challenge dataset is available at: https://zenodo.org/records/20191078, and the code for baseline models and metrics calculation is available here: https://github.com/athaddius/STIRMetrics
Chinese Translation
手术中的点跟踪对于实现下游任务（如分割、3D重建、虚拟组织标记、自主探针扫描和子任务自主性）至关重要。本文介绍了2025年点跟踪挑战赛的迭代版本，参与者提交其算法进行量化评估。其算法使用名为红外手术纹身 (STIR) 的数据集进行评估，挑战赛被命名为STIR挑战赛2025 (STIRC2025)。STIR挑战赛2025包括两个定量组件：准确性和效率。准确性组件测试算法在体内和体外序列上的准确性，而效率组件测试算法推理延迟。该挑战作为MICCAI EndoVis 2025的一部分进行，共有七个团队参与。本文总结了挑战结果和参与者的方法。挑战数据集可在以下链接获取：https://zenodo.org/records/20191078，基线模型和指标计算的代码可在此处获取：https://github.com/athaddius/STIRMetrics
cs.CV / 93 / 2607.12959
ViCo3D: Empowering LiDAR-based Collaborative 3D Object Detection with Vision Foundation Models
ViCo3D：基于LiDAR的协作3D物体检测与视觉基础模型的结合
Abstract
LiDAR-based collaborative 3D perception in Vehicle-to-Everything (V2X) systems typically relies on fusing bird's-eye-view (BEV) features across agents. However, current BEV representations, typically extracted by LiDAR backbones trained from scratch, are geometry-dominated and lack general semantic priors, inherently limiting the efficacy of feature-level collaboration. Meanwhile, vision foundation models (VFMs) pretrained on large-scale image data have demonstrated strong capability in learning general-purpose and informative visual representations for 2D tasks, and have the potential to enhance agent-wise LiDAR BEV representations for collaboration. Despite this potential, adapting VFMs to LiDAR-based 3D detection remains challenging due to the substantial image-point cloud modality gap. To bridge this gap, we propose ViCo3D, a collaborative 3D object detection framework powered by VFMs. Specifically, ViCo3D adapts VFMs to LiDAR-based collaborative perception from three aspects: First, ViCo3D projects point clouds onto the BEV plane as three-channel images, enabling DINOv2 to extract BEV-space visual features from LiDAR inputs. Besides, to effectively integrate these DINOv2-derived features with LiDAR geometric features, ViCo3D introduces a multi-scale BEV fusion module within the single-agent encoder. In addition, ViCo3D adopts an ego-centric cross-agent fusion strategy to aggregate complementary information from multiple agents. Experiments on DAIR-V2X and V2XSet demonstrate that ViCo3D achieves state-of-the-art 3D detection performance. Remarkably, it delivers up to 1.8x greater collaborative gains than prior methods on DAIR-V2X. The code will be made public available for future investigation.
Chinese Translation
基于LiDAR的协作3D感知在车对一切（V2X）系统中通常依赖于跨代理融合鸟瞰图（BEV）特征。然而，目前的BEV表示通常由从头训练的LiDAR骨干网络提取，主要受几何特征主导，缺乏通用的语义先验，这在本质上限制了特征级协作的有效性。同时，在大规模图像数据上预训练的视觉基础模型（VFMs）在学习通用和信息丰富的视觉表示方面表现出强大的能力，能够增强代理间的LiDAR BEV表示以实现协作。尽管存在这种潜力，但将VFMs适应于基于LiDAR的3D检测仍然面临挑战，因为图像与点云之间存在显著的模态差距。为了解决这一问题，我们提出了ViCo3D，这是一种由VFMs驱动的协作3D物体检测框架。具体而言，ViCo3D从三个方面将VFMs适应于基于LiDAR的协作感知：首先，ViCo3D将点云投影到BEV平面上，形成三通道图像，使DINOv2能够从LiDAR输入中提取BEV空间视觉特征。此外，为了有效地将这些DINOv2派生的特征与LiDAR几何特征整合，ViCo3D在单代理编码器中引入了多尺度BEV融合模块。此外，ViCo3D采用了一种以自我为中心的跨代理融合策略，以聚合来自多个代理的互补信息。在DAIR-V2X和V2XSet上的实验表明，ViCo3D实现了最先进的3D检测性能。值得注意的是，在DAIR-V2X上，它比之前的方法提供了高达1.8倍的协作增益。代码将公开以供未来研究使用。
cs.CV / 94 / 2607.12987
Controllable Generation of Diverse Dermatological Imagery for Fair and Efficient Malignancy Classification
可控生成多样化皮肤病影像以实现公平高效的恶性肿瘤分类
Abstract
Accurate dermatological diagnosis naturally necessitates equitable performance across diverse populations, yet a systematic lack of expertly annotated images, especially for underrepresented skin tones and rare diseases, impedes progress toward measurably fair methods. We introduce cgDDI (Controllable Generation of Diverse Dermatological Imagery), a hybrid framework that (1) synthesizes realistic healthy skin samples without disturbing other input properties, (2) maps single-sample rare lesions onto novel skin-tones and locations non-parametrically, and (3) allows for efficient parametric generation with as few as 10 training samples. The framework supports both human and automated segmentation masking, enabling scalability to datasets without pre-made lesion masks. We grow a 656-image dataset by more than 400x and validate across two datasets: biopsy-confirmed Diverse Dermatology Images (DDI) and expert-verified Fitzpatrick17k (F17k). On the DDI benchmark, we achieve malignancy classification accuracy of 86.4% under synthetic-only training and 90.9% state-of-the-art performance with real data fine-tuning, alongside leading fairness metrics. Cross-dataset experiments show +13.9% accuracy improvements on unseen F17k data despite minimal disease overlap. We openly release 266k+ synthetic images, code, and generative models to further support fairness research at https://github.com/hectorcarrion/ControllableGenDDI.
Chinese Translation
准确的皮肤病诊断自然需要在不同人群中实现公平的表现，然而，专家注释图像的系统性缺乏，尤其是对于代表性不足的肤色和罕见疾病，阻碍了朝着可测量的公平方法的进展。我们提出了 cgDDI（可控生成多样化皮肤病影像），这是一个混合框架，(1) 合成真实的健康皮肤样本而不干扰其他输入特性，(2) 将单样本罕见病变非参数地映射到新的肤色和位置，(3) 允许以仅需 10 个训练样本进行高效的参数生成。该框架支持人类和自动分割掩膜，使其能够扩展到没有预制病变掩膜的数据集。我们将一个包含 656 张图像的数据集扩展了 400 倍以上，并在两个数据集上进行了验证：经过活检确认的多样化皮肤病影像（DDI）和专家验证的 Fitzpatrick17k（F17k）。在 DDI 基准测试中，我们在仅使用合成数据训练的情况下实现了 86.4% 的恶性肿瘤分类准确率，并在使用真实数据微调时达到了 90.9% 的最先进性能，同时保持领先的公平性指标。跨数据集实验显示，在未见过的 F17k 数据上，尽管疾病重叠最小，准确率仍提高了 13.9%。我们公开发布了 266,000+ 张合成图像、代码和生成模型，以进一步支持公平性研究，网址为 https://github.com/hectorcarrion/ControllableGenDDI。
cs.CV / 95 / 2607.12993
X-Lens: Real-Time Metric Depth Estimation with Heterogeneous Cameras
X-Lens：基于异构相机的实时度量深度估计
Abstract
We present X-lens, a compact feed-forward model for metric depth estimation from a variable number of calibrated fisheye and pinhole views. To support real-time downstream perception, X-lens is built around a geometry-aware heterogeneous camera formulation with two key components. Learnable calibration tokens provide a coarse alignment between fisheye and pinhole projective spaces, while a Jacobian-parameterized distortion bias injected into cross-attention models local projection changes and promotes cross-camera consistency, enabling robust generalization with only 0.04B parameters and up to 41 FPS. The model predicts dense depth together with a global metric scale, avoiding auxiliary reconstruction targets that increase computation and optimization complexity. To learn such cross-camera generalization at scale and depth, X-lens is trained on multiple public datasets and OmniScene, our newly released large-scale synthetic dataset containing approximately 266K synchronized six-view frames, 1.7M individual images, and 103 indoor and outdoor scenes. Extensive experiments on both real-world and synthetic indoor and outdoor datasets demonstrate superior heterogeneous-camera metric depth accuracy, reducing AbsRel by 25.4\% on OmniScene-Full over the strongest baseline while using 88.9\% fewer parameters, with competitive performance on conventional fisheye-only and pinhole-only settings.
Chinese Translation
我们提出了X-Lens，这是一种紧凑的前馈模型，用于从可变数量的校准鱼眼和针孔视图中进行度量深度估计。为了支持实时下游感知，X-Lens围绕一个几何感知的异构相机模型构建，包含两个关键组件。可学习的校准标记提供了鱼眼和针孔投影空间之间的粗略对齐，而注入到交叉注意力模型中的雅可比参数化畸变偏差则局部化投影变化并促进跨相机的一致性，从而实现仅使用0.04B参数和高达41 FPS的稳健泛化。该模型预测稠密深度以及全局度量尺度，避免了增加计算和优化复杂性的辅助重建目标。为了在规模和深度上学习这种跨相机泛化，X-Lens在多个公共数据集和OmniScene上进行训练，后者是我们新发布的大规模合成数据集，包含约266K个同步六视图帧、1.7M个单独图像和103个室内外场景。在真实世界和合成的室内外数据集上的广泛实验表明，异构相机的度量深度准确性优越，在OmniScene-Full上相较于最强基线减少了25.4 ext{%}的绝对相对误差，同时使用了88.9 ext{%}更少的参数，并在传统的仅鱼眼和仅针孔设置中表现出竞争力。
cs.CV / 96 / 2607.13010
DermDepth: Toward Monocular Metric Scale 3D Reconstruction Models for Dermatology
DermDepth：面向皮肤病学的单目度量尺度三维重建模型
Abstract
Dermatological practice routinely involves measuring and tracking lesion size, morphology and texture, as critical components of wound or skin cancer screening, monitoring and diagnosis. To accomplish this task, practitioners often image the skin surface with commonly available off-the-shelf camera sensors. This has led to an overwhelming research focus on 2D methods while these objectives naturally benefit from 3D information. In this paper, we demonstrate that dense monocular 3D reconstructions, metric scale measurements and rich surface normal texture estimates are achievable for both dermoscopic and macroscopic cases without the need for additional hardware or multiple captures. We present DermDepth, the first single-view metric scale 3D model for the dermatological domain and D-Synth, the first synthetic dermoscopic dataset with pixel-perfect 3D information. Our experiments show training DermDepth on D-Synth corrects metric scale error from over 16x to under 1.1x for real dermoscopic data, while preserving geometric quality and increasing texture richness. Fine-tuning on a small amount of real clinical samples generalizes our method across three real-world benchmarks spanning the few mm to hundred cm range, diverse skin-tones, chronic wound cases and produces measurements broadly consistent with disease size reported in medical literature. All code, data and models are available at https://github.com/hectorcarrion/dermdepth.
Chinese Translation
皮肤科实践通常涉及测量和跟踪病变的大小、形态和纹理，这些都是伤口或皮肤癌筛查、监测和诊断的关键组成部分。为了完成这一任务，医生通常使用常见的现成相机传感器对皮肤表面进行成像。这导致了对二维方法的过度研究，而这些目标自然受益于三维信息。在本文中，我们展示了在不需要额外硬件或多次捕获的情况下，密集的单目三维重建、度量尺度测量和丰富的表面法线纹理估计在皮肤镜和肉眼观察情况下都是可以实现的。我们提出了DermDepth，这是皮肤病学领域第一个单视图度量尺度三维模型，以及D-Synth，这是第一个具有像素完美三维信息的合成皮肤镜数据集。我们的实验表明，在D-Synth上训练DermDepth可以将真实皮肤镜数据的度量尺度误差从超过16倍降低到1.1倍以下，同时保持几何质量并增加纹理丰富性。在少量真实临床样本上进行微调使我们的方法在三个真实世界基准上具有良好的泛化能力，涵盖从几毫米到数百厘米的范围、多样的肤色和慢性伤口案例，并产生与医学文献中报告的疾病大小大致一致的测量结果。所有代码、数据和模型均可在https://github.com/hectorcarrion/dermdepth获取。
cs.AI / 1 / 2607.11888
Optimal Adaptive Market Making: A Theoretical Framework for High-Yield Liquidity Provision in Perpetual Futures Markets
最优自适应市场做市：永久期货市场高收益流动性提供的理论框架
Abstract
We develop a rigorous theoretical framework for optimal market making in perpetual futures markets with zero maker fees. We model the market maker's problem as a stochastic optimal control problem on a filtered probability space, where the controls are adaptive bid-ask spreads and inventory hedging decisions across two exchanges. Our contributions include: (i) a PnL decomposition theorem separating revenue into spread income, adverse selection loss, inventory carrying cost, hedging friction, and funding rate exposure; (ii) the Hamilton-Jacobi-Bellman equation for the joint spread-inventory-hedging control problem under CARA utility with a verification theorem; (iii) High-APY Regime Theorems characterizing profitable regions via five dimensionless parameters, culminating in a Master APY Formula; (iv) analysis of zero-fee economics on decentralized perpetual exchanges with optimal entry-exit thresholds; (v) optimal cross-exchange hedging policies with funding rate dynamics and a hedge regime trichotomy; (vi) a robustness margin quantifying parameter uncertainty tolerance; (vii) exponential drawdown probability bounds and a universal APY-VaR identity; (viii) ergodic inventory distribution under optimal control with Bayesian adaptive estimation; (ix) Kelly-optimal leverage with ruin boundaries; and (x) multi-pair portfolio allocation with diversification saturation results. Numerical analysis with twenty-three figures reveals phase transitions between profitable and unprofitable regimes. Our framework unifies and extends the Avellaneda-Stoikov, Gueant-Lehalle-Fernandez-Tapia, and Glosten-Milgrom paradigms for modern decentralized venue microstructure.
Chinese Translation
我们为零做市费用的永久期货市场中的最优做市开发了一个严格的理论框架。我们将市场做市商的问题建模为一个在过滤概率空间上的随机最优控制问题，其中控制变量是自适应的买卖差价和跨两个交易所的库存对冲决策。我们的贡献包括：(i) 一个利润与损失分解定理，将收入分为差价收入、不利选择损失、库存持有成本、对冲摩擦和资金利率风险；(ii) 在CARA效用下的联合差价-库存-对冲控制问题的Hamilton-Jacobi-Bellman方程及其验证定理；(iii) 高年化收益率（High-APY）状态定理，通过五个无量纲参数描述盈利区域，最终形成主年化收益率公式；(iv) 对去中心化永久交易所零费用经济学的分析，确定最优的进出阈值；(v) 考虑资金利率动态的最优跨交易所对冲政策及对冲状态三分法；(vi) 量化参数不确定性容忍度的稳健性边际；(vii) 指数回撤概率界限和通用年化收益率-风险价值（APY-VaR）恒等式；(viii) 在最优控制下的遍历库存分布与贝叶斯自适应估计；(ix) 凯利最优杠杆与破产边界；以及(x) 具有多对组合配置和多样化饱和结果的分析。二十三个图形的数值分析揭示了盈利与非盈利状态之间的相变。我们的框架统一并扩展了Avellaneda-Stoikov、Gueant-Lehalle-Fernandez-Tapia和Glosten-Milgrom模型，适用于现代去中心化场所的微观结构。
cs.AI / 2 / 2607.11906
In-Context Reinforcement Learning under Non-Stationarity: A Survey
非平稳环境下的上下文强化学习：综述
Abstract
The development of decision-pretrained transformers, algorithm distillation, long-context meta-RL, and retrieval-augmented agents has renewed interest in in-context reinforcement learning (ICRL): the ability of a pretrained or fine-tuned decision model to infer latent task rules and improve future behavior from interaction context, without test-time parameter updates. This line of work asks when trial-and-error evidence, rewards, transitions, demonstrations, feedback, or retrieved experience can make learning-like computation happen inside the context window. However, existing surveys of ICRL mainly organize the field around pretraining objectives, architectures, context formats, evaluation protocols, and theoretical mechanisms, while the non-stationary setting remains comparatively underexamined. In changing environments, accumulated context is not merely more evidence about a fixed task: the reward specification, transition kernel, observation channel, action interface, constraint model, or demonstration and memory distribution can fall out of alignment with the current regime. Previously useful context can therefore become stale, misleading, or useful again when an old regime returns. We survey non-stationary ICRL as the problem of adapting through context while deployed policy parameters remain fixed: the policy must infer both the current decision rule and which parts of its accumulated evidence still support that rule. We define non-stationary ICRL, relate it to meta-RL, decision sequence modeling, retrieval-augmented RL, value- and model-aware ICRL, and reward-feedback agents, and organize the literature along three questions: what changes, how the change unfolds, and how observable the change is to the agent.
Chinese Translation
决策预训练变换器、算法蒸馏、长上下文元强化学习（meta-RL）和检索增强代理的开发重新引发了对上下文强化学习（ICRL）的兴趣：即预训练或微调的决策模型能够从交互上下文中推断潜在任务规则并改善未来行为的能力，而无需在测试时更新参数。这项研究探讨了何时试错证据、奖励、转移、演示、反馈或检索经验能够使学习类计算在上下文窗口内发生。然而，现有的ICRL综述主要围绕预训练目标、架构、上下文格式、评估协议和理论机制组织该领域，而非平稳环境的设置则相对较少被研究。在变化的环境中，累积的上下文不仅仅是关于固定任务的更多证据：奖励规范、转移核、观察通道、行动接口、约束模型或演示和记忆分布可能与当前环境不再一致。因此，之前有用的上下文可能会变得过时、误导，或者在旧环境回归时再次变得有用。我们将非平稳ICRL视为在部署的策略参数保持固定的情况下通过上下文进行适应的问题：策略必须推断当前的决策规则以及其累积证据中哪些部分仍然支持该规则。我们定义了非平稳ICRL，将其与元强化学习、决策序列建模、检索增强强化学习、价值和模型感知ICRL以及奖励反馈代理相关联，并围绕三个问题组织文献：什么发生了变化、变化是如何展开的，以及变化对代理的可观察性如何。
cs.AI / 3 / 2607.11948
Ontology-Amplified Distillation and Contextuality Auditing for Sovereign Enterprise Language Models: A Combined Proof-of-Mechanism and Negative-Results Method Study
本体增强蒸馏与情境审计在主权企业语言模型中的应用：结合机制证明与负结果的方法研究
Abstract
Regulated financial institutions operating under data-residency rules need tenant-owned language models that can run inside the institution's perimeter. This paper combines two related FAOS studies into one mechanism-and-control article. First, it reports a reduced-power proof-of-mechanism study of ontology-amplified distillation: a Qwen3.6-27B student is adapted to the Foundation AgenticOS ontology through supervised fine-tuning on frontier-teacher trajectories and ontology-grounded direct preference optimization (DPO), trained locally on a single Apple M5 Max from 47 synthetic, English-language, cross-domain preference pairs. On 40 held-out Vietnamese financial-domain tasks, the distilled student grounds 36 of 40 tasks (grounded rate 0.90; mean ontology term-coverage r_onto = 0.95 on a metric floored at 0.50), equal to the GPT-5 frontier baseline, which also grounds 36 of 40. The outcome is underpowered to establish equivalence: the paired-difference 95% confidence interval spans +/-4 tasks, and the run does not test or show the pre-registered amplification prediction that the student should exceed the frontier. Second, the paper consolidates a contextuality-audit method for enterprise-agent routing. In a separate negative-results pilot, the corrected canonical Contextuality-by-Default degree is zero for all Phase 1.3 groups in both the local-Qwen run and an explicitly labeled Gemma replication check; the useful signal is direct influence and construct coupling, not surviving residual contextuality. Together, the studies pair an ontology-grounded model-building mechanism with a governance diagnostic for deciding when apparent disagreement should trigger prompt standardization, multi-agent synthesis, or human review. The evidence supports neither deployability, safety, superiority, statistical equivalence, nor a contextuality-positive routing rule.
Chinese Translation
在数据驻留规则下运营的受监管金融机构需要能够在机构边界内运行的租户拥有的语言模型。本文将两项相关的FAOS研究结合为一篇机制与控制的文章。首先，报告了一项关于本体增强蒸馏的降低功率机制证明研究：通过在前沿教师轨迹上进行监督微调和基于本体的直接偏好优化（DPO），将Qwen3.6-27B学生模型适配到Foundation AgenticOS本体，使用47对合成的英语跨领域偏好对在单个Apple M5 Max上进行本地训练。在40个保留的越南金融领域任务中，蒸馏后的学生模型成功实现了36个任务的基础（基础率为0.90；平均本体术语覆盖率r_onto = 0.95，指标下限为0.50），与GPT-5前沿基线相等，后者同样实现了36个任务的基础。该结果的能力不足以建立等效性：配对差异的95%置信区间跨度为+/-4个任务，并且该运行未测试或显示预注册的增强预测，即学生模型应超过前沿。其次，本文整合了一种用于企业代理路由的情境审计方法。在一项独立的负结果试点中，所有Phase 1.3组的修正标准情境默认度为零，无论是在本地Qwen运行中还是在明确标记的Gemma复制检查中；有用的信号是直接影响和构造耦合，而不是存活的残余情境性。综上所述，这些研究将基于本体的模型构建机制与治理诊断相结合，以决定何时明显的不一致应触发及时标准化、多代理合成或人工审查。证据不支持可部署性、安全性、优越性、统计等效性或情境积极路由规则。
cs.AI / 4 / 2607.11951
GRID: Grammar-Railed Decoding for Enterprise SQL Generation
GRID：用于企业 SQL 生成的语法约束解码
Abstract
Large language models can write SQL, but enterprise deployment demands more than plausible text: outputs must be syntactically valid, must respect per-role and per-schema policy, must carry provable (not best-effort) guarantees, must not slow down as generations grow, and must leave a compliance-grade record of every decision. We present GRID (Grammar-Railed Decoding), a grammar-constrained decoding engine that keys exact next-token masks on parser configurations (lexer scan state x LALR(1) stack) rather than on token sequences, and uses the incrementally advanced LALR(1) parser itself as a viable-prefix oracle. LLM tokens are bridged to grammar terminals by a byte-level trie walk with a context-independent/context-dependent split that makes cache-key soundness hold by construction. Role-based access control is compiled into the language: role projections subset the grammar's productions and schema lexicons restrict identifier terminals, so forbidden verbs and identifiers are unreachable at mask level. Four guarantees (soundness, completeness, termination, and near-constant per-token cost) are stated with explicit preconditions and each paired with a test or benchmark. Rust kernels bring the per-token mask to a 3.6-6.7 us median, ahead of llguidance at p50 and p90 on two tokenizers with zero false rejects; per-token guard cost is position-flat at n=16,000. On Spider, constrained decoding is worth +13 execution-accuracy points at 0.5B, and one checker-guided repair pass over the provably mask-unenforceable residue (column-level policy) lifts a 7B model to 94.5% executable. A hash-chained per-token audit trail replays bit-identically with 100% tamper detection. We state plainly what the mask cannot do (distribution faithfulness, column-level RBAC, non-LALR(1) languages) and where measured cost remains.
Chinese Translation
大型语言模型能够编写 SQL，但企业部署的要求不仅仅是合理的文本：输出必须在语法上有效，必须遵循每个角色和每个模式的策略，必须提供可证明的（而非尽力而为的）保证，随着生成的增加不能减慢，并且必须保留每个决策的合规记录。我们提出了 GRID（语法约束解码），这是一种语法约束的解码引擎，它基于解析器配置（词法分析器扫描状态 x LALR(1) 堆栈）而非基于令牌序列来确定确切的下一个令牌掩码，并使用逐步改进的 LALR(1) 解析器本身作为可行前缀的预言者。LLM 令牌通过字节级前缀树遍历与语法终端连接，采用上下文无关/上下文相关的分割，使得缓存键的有效性在构造上得以保持。基于角色的访问控制被编译到语言中：角色投影子集语法的产生和模式词汇限制标识符终端，因此在掩码级别上无法访问被禁止的动词和标识符。我们明确陈述了四个保证（有效性、完整性、终止性和近似常数的每令牌成本），并为每个保证提供了明确的前提条件，并与测试或基准配对。Rust 内核将每个令牌的掩码处理时间缩短至 3.6-6.7 微秒的中位数，在两个令牌生成器上，p50 和 p90 的性能优于 llguidance，且没有虚假拒绝；每个令牌的保护成本在 n=16,000 时保持平坦。在 Spider 上，约束解码在 0.5B 时增加了 +13 的执行准确度，并且对可证明的掩码无法强制执行的残余（列级策略）进行一次检查器引导的修复，使得 7B 模型的可执行性提升至 94.5%。哈希链式的每个令牌审计记录以 100% 的篡改检测以位相同的方式重放。我们明确说明了掩码无法实现的内容（分布忠实性、列级 RBAC、非 LALR(1) 语言）以及测量成本仍然存在的地方。
cs.AI / 5 / 2607.11959
Calibration-First Reward-Component Auditing for Reinforcement Learning Control in Smart Greenhouses
基于校准的强化学习控制在智能温室中的奖励组件审计
Abstract
Greenhouse reinforcement learning can test climate-control ideas at a speed and scale that is difficult to achieve with crop experiments alone. For smart-greenhouse control, however, a single simulator return is not enough: a grower or control engineer also needs to know when the policy heats, enriches CO2, vents, manages humidity, deploys screens, or uses lamps.We propose a reproducible calibration-first reward audit framework that keeps named greenhouse-control reward components comparable across simulator training, facility-adapted rollouts, logged Autonomous Greenhouse Challenge records, and actuator-rule distillation. In GreenLight-Gym, the framework decomposes the scalar reward into conditional temperature, CO2, humidity and vapor-pressure-deficit, screen, and actuation-proxy terms; adapts GreenLight to the second Autonomous Greenhouse Challenge logged climate traces; and scores the same components on logged greenhouse data.
Chinese Translation
温室强化学习能够以难以通过单纯的作物实验实现的速度和规模测试气候控制理念。然而，对于智能温室控制而言，单一的模拟器返回结果是不够的：种植者或控制工程师还需要知道政策何时加热、富集二氧化碳、通风、管理湿度、部署遮阳网或使用灯具。我们提出了一种可重复的基于校准的奖励审计框架，该框架保持了命名的温室控制奖励组件在模拟器训练、设施适应性执行、记录的自主温室挑战记录和执行器规则提炼之间的可比性。在 GreenLight-Gym 中，该框架将标量奖励分解为条件温度、二氧化碳、湿度和蒸汽压亏缺、遮阳网和执行代理项；将 GreenLight 适配于第二届自主温室挑战记录的气候轨迹；并在记录的温室数据上对相同组件进行评分。
cs.AI / 6 / 2607.11977
Optimization Is Not All You Need
优化并不是你所需要的一切
Abstract
In 2019, OpenAI released two million GPT-2 outputs-ungrammatical, half broken-to aid the detection of machine-generated text. The alignment that produced their more fluent successors is usually regarded as an engineering achievement; we read it instead as the newest expression of optimization culture: the conviction, older than the technology, that measurable improvement along predefined axes exhausts the question of value. Tracing that conviction through the stack-pretraining, decoding, preference tuning, benchmarking, interface-and back through its genealogy in the audit society, we arrive at the limit: an optimization procedure can measure how improbable a piece of generated text is; it cannot tell whether that unlikelihood is error or invention. A procedure that cannot make that distinction has nonetheless, within half a decade, assumed the authority to set the protocols of legitimate language. Held for centuries by academies and schoolrooms, grammars and examiners, this authority has been given over to loss functions, reward models, benchmarks, and system prompts: an apparatus that executes the office of judgment with no capacity for judging.
Chinese Translation
2019年，OpenAI发布了两百万条GPT-2输出——不合语法、半破碎——以帮助检测机器生成的文本。产生其更流畅后继者的对齐通常被视为一种工程成就；我们则将其视为优化文化的最新表现：一种信念，这种信念早于技术，认为在预定义轴线上可测量的改进耗尽了价值的问题。通过堆栈追踪这种信念——预训练、解码、偏好调优、基准测试、接口——并回溯其在审计社会中的谱系，我们到达了极限：一种优化程序可以测量一段生成文本的不可能性；但它无法判断这种不可能性是错误还是创造。尽管如此，在短短五年内，这种无法做出区分的程序却获得了设定合法语言协议的权威。这个权威曾由学院和教室、语法和考官掌握，如今却转交给了损失函数、奖励模型、基准测试和系统提示：一个执行判断职能但没有判断能力的装置。
cs.AI / 7 / 2607.11980
LP Mining with LP2Graph: A Use Case for Railway Rescheduling
使用 LP2Graph 进行线性规划挖掘：铁路调度重排的应用案例
Abstract
Like many optimization-driven domains, railway rescheduling relies on Mixed-Integer Linear Programming (MILP), yet the field's modeling knowledge is scattered across hundreds of papers in incompatible notations, and narrative surveys organize it subjectively: they classify models by vocabulary rather than by structure, and reproduce neither. We present LP Mining with LP2Graph, a method that mines the structure of published LP and MILP formulations into a reproducible dataset and an induced taxonomy. Its core, LP2Graph, represents each formulation admitted by its canonical grammar as a typed variable--equation graph derived from a single canonical model; once a source is extracted into that model, everything downstream is deterministic. Each source is parsed into this model, homologized, and clustered bottom-up (over variables, then constraints and the objective, then whole-model structure) and, separately, by application domain and solution approach; the resulting groups are labeled by a rule-seeded, self-updating classifier. We validate the representation rather than assume it: per-cluster representatives are regenerated as independent LaTeX and re-solved across CBC, HiGHS and Gurobi against the optimum reported in the source paper. The outcome is an objective, repeatable taxonomy of variables, constraints and model types: the principled foundation on which our raiLPminer line of automated railway-rescheduling model development builds.
Chinese Translation
与许多以优化为驱动的领域一样，铁路调度重排依赖于混合整数线性规划（MILP），然而该领域的建模知识散布在数百篇不兼容符号的论文中，叙述性综述主观地组织这些知识：它们按词汇而非结构对模型进行分类，并且没有重现。我们提出了使用 LP2Graph 进行线性规划挖掘的方法，该方法将已发布的线性规划（LP）和混合整数线性规划（MILP）公式的结构挖掘为可重现的数据集和诱导分类法。其核心，LP2Graph，将每个由其规范语法所接受的公式表示为一个类型化变量-方程图，该图源自单一的规范模型；一旦源被提取到该模型中，所有下游内容都是确定性的。每个源被解析到该模型中，进行同源化，并自下而上地聚类（先是变量，然后是约束和目标，最后是整个模型结构），并且根据应用领域和解决方法进行单独分类；生成的组通过规则引导的自更新分类器进行标记。我们验证表示而非假设表示：每个聚类的代表作为独立的 LaTeX 重新生成，并在 CBC、HiGHS 和 Gurobi 中重新求解，以对比源论文中报告的最优解。最终结果是一个客观、可重复的变量、约束和模型类型的分类法：这是我们 raiLPminer 自动化铁路调度重排模型开发的原则基础。
cs.AI / 8 / 2607.12056
Designing Agent-Ready Websites for AI Web Agents: A Framework for Machine Readability, Actionability, and Decision Reliability
为人工智能网络代理设计代理友好型网站：机器可读性、可操作性和决策可靠性的框架
Abstract
Online shopping is increasingly shifting toward a model in which AI agents independently search for products, compare options, evaluate constraints, and carry out parts of the purchasing process for users. Website design must now support both human and agent-mediated interaction. This paper introduces the agent-ready website, a design framework for enhancing the readability, interpretability, verifiability, and actionability of e-commerce platforms for AI agents. Existing web design, SEO, and generative engine optimization (GEO) metrics do not fully assess a website's capacity for agent-mediated interaction. The proposed framework is structured around three dimensions agent interpretability, agent executability, and agent decision reliability supported by features such as machine readability, semantic clarity, agent actionability, and contextual decision-reliability signals. The framework is evaluated through a controlled experiment comparing a human-oriented baseline and an agent-ready version of an identical website prototype, with identical catalogs, pricing, stock, and shopping workflows. The evaluation involved five tasks, three browser-agent models (GPT-4.1, Gemini-2.5 Flash, and Grok-4 Fast), and 300 runs, measuring PASS,PARTIAL,FAIL outcomes, strict and functional success rates, error patterns, step counts, and token consumption. The agent-ready website achieved 134 PASS runs out of 150 versus 74 out of 150 for the baseline (strict success rates of 89.3% vs. 49.3%), with the largest gains in product detail extraction, comparison, and multi-constraint selection. It also reduced PARTIAL outcomes from 43 to 3 and lowered the average step count from 9.31 to 6.49. These results provide preliminary evidence that enhanced structural clarity, action cues, evidence signals, and temporal validity indicators can substantially improve the reliability and efficiency of AI browser agents.
Chinese Translation
在线购物正日益转向一种模式，在这种模式下，人工智能代理独立搜索产品、比较选项、评估约束，并为用户执行部分购买流程。网站设计现在必须支持人类和代理介导的互动。本文介绍了代理友好型网站，这是一种设计框架，旨在增强电子商务平台对人工智能代理的可读性、可解释性、可验证性和可操作性。现有的网站设计、搜索引擎优化（SEO）和生成引擎优化（GEO）指标并未充分评估网站在代理介导互动中的能力。所提出的框架围绕三个维度构建：代理可解释性、代理可执行性和代理决策可靠性，支持机器可读性、语义清晰性、代理可操作性和上下文决策可靠性信号等特征。通过对比人类导向的基线和相同网站原型的代理友好版本，采用受控实验对该框架进行了评估，两个版本具有相同的目录、定价、库存和购物流程。评估涉及五个任务、三个浏览器-代理模型（GPT-4.1、Gemini-2.5 Flash 和 Grok-4 Fast）以及300次运行，测量PASS、PARTIAL、FAIL结果、严格和功能成功率、错误模式、步骤计数和令牌消耗。代理友好型网站在150次运行中实现了134次PASS，而基线仅为74次（严格成功率分别为89.3%和49.3%），在产品细节提取、比较和多约束选择方面取得了最大的提升。同时，它将PARTIAL结果从43降低到3，并将平均步骤计数从9.31降低到6.49。这些结果提供了初步证据，表明增强的结构清晰性、操作提示、证据信号和时间有效性指标可以显著提高人工智能浏览器代理的可靠性和效率。
cs.AI / 9 / 2607.12077
Graph Feedback Controls Consensus and Clique Formation in Open-Weight Language-Model Populations
图反馈控制开放权重语言模型群体中的共识与团体形成
Abstract
Multi-agent language-model systems increasingly route local interactions, yet the runtime interaction graph is often treated as an implementation detail. We study convention formation in open-weight LM populations spanning 1.1B-32B parameters with a naming-game protocol. Restricted first-token scores over tokenizer-safe labels let us measure prompt-conditioned score-state distributions, construct state-similarity graphs, and separate sampled-label agreement from latent state-space consensus. Across controlled interventions, in the main open-weight repair grids, retained partner-label evidence is necessary but not sufficient: homophilous threshold-similarity routing deletes cross-basin exposure and amplifies fragmentation, while bridge-seeking routing often repairs fragmentation when memory is available. In a three-seed mixed four-model grid, threshold-similarity produces no final behavioral or state consensus in 189 setting-seed runs, whereas state-component and label-disagreement bridges recover final behavioral consensus in 14/18 retained-memory runs. Across homogeneous model populations, retained history generally shifts fragmented dynamics toward consensus; the clearest case is Qwen2.5-32B, which reaches stable behavioral and final state consensus in all 18 retained-history well-mixed settings, while threshold-similarity reaches neither form of consensus in 189 settings. Robustness over state thresholds, population size, and vocabulary size preserves the qualitative ordering, and early-window graph-energy features provide useful within-grid diagnostics.
Chinese Translation
多智能体语言模型系统日益依赖局部交互，但运行时交互图通常被视为实现细节。我们研究了在开放权重语言模型群体中通过命名游戏协议形成的约定，这些模型参数范围从11亿到320亿。通过对标记器安全标签的限制首个令牌评分，我们能够测量基于提示的评分状态分布，构建状态相似性图，并将采样标签的一致性与潜在状态空间的共识分离。在控制干预中，在主要的开放权重修复网格中，保留的伙伴标签证据是必要但不足的：同质阈值相似性路由删除了跨盆地的曝光并加剧了碎片化，而寻求桥接的路由在有可用内存时通常能修复碎片化。在一个包含三个种子和四个模型的混合网格中，阈值相似性在189个设置种子运行中未能产生最终的行为或状态共识，而状态组件和标签不一致的桥接在18个保留内存的运行中恢复了14个最终行为共识。在同质模型群体中，保留历史通常将碎片化动态转向共识；最明显的例子是Qwen2.5-32B，在所有18个保留历史的良好混合设置中达成了稳定的行为和最终状态共识，而阈值相似性在189个设置中未能达到任何形式的共识。对状态阈值、群体规模和词汇规模的鲁棒性保持了定性排序，早期窗口图能量特征提供了有用的网格内诊断。
cs.AI / 10 / 2607.12085
Operationalising Multi-Dimensional Evaluation for Conversational Agents: A Scalable, Governed Pipeline with Selective Re-evaluation and Model Benchmarking
为对话代理实现多维评估的操作化：一个可扩展的、受管控的管道，具有选择性重新评估和模型基准测试
Abstract
Evaluating retail conversational agents requires methods beyond lexical-overlap metrics to assess intent alignment, factuality, helpfulness, clarity, tone, and overall response quality. Although LLM-as-a-judge methods provide scalable alternatives to human evaluation, production deployment introduces challenges in governance, reproducibility, cost, schema consistency, traceability, and reliability. We present GenAI Evaluation, a governed, configuration-driven pipeline for large-scale evaluation of retail conversational systems. It processes production chatbot logs through normalization, sharding, asynchronous execution, and schema-constrained LLM scoring. The framework evaluates helpfulness, truthfulness, clarity, tone alignment, and translation-specific dimensions. Selective re-evaluation processes only incomplete, malformed, or schema-invalid records, while schema locking, versioned configurations, validation logs, and record-level provenance support auditability. The framework processes approximately 50,000 records daily and has evaluated more than two million interactions. Validation used 12,980 stratified-random human-labeled records from four trained annotators. Classification covered 14 intents, 156 sub-intents, 18 major domains, and 129 sub-domains. The pipeline achieved a macro F1 score of 0.93 and 89% human-acceptability accuracy for translation.
Chinese Translation
评估零售对话代理需要超越词汇重叠指标的方法，以评估意图一致性、事实性、帮助性、清晰度、语气和整体响应质量。尽管 LLM-as-a-judge 方法为人类评估提供了可扩展的替代方案，但生产部署引入了治理、可重复性、成本、模式一致性、可追溯性和可靠性等挑战。我们提出了 GenAI Evaluation，这是一个受管控的、基于配置的管道，用于大规模评估零售对话系统。它通过规范化、分片、异步执行和模式约束的 LLM 评分处理生产聊天机器人日志。该框架评估帮助性、真实性、清晰度、语气一致性和特定翻译维度。选择性重新评估仅处理不完整、格式错误或模式无效的记录，而模式锁定、版本化配置、验证日志和记录级来源支持审计能力。该框架每天处理约 50,000 条记录，并已评估超过两百万次交互。验证使用了来自四名训练过的注释者的 12,980 条分层随机人类标注记录。分类涵盖了 14 种意图、156 种子意图、18 个主要领域和 129 个子领域。该管道实现了 0.93 的宏 F1 分数和 89% 的人类可接受性准确率。
cs.AI / 11 / 2607.12097
Representing and Generating Levels Over Time through Playtrace Reconstructive Partitioning
通过 Playtrace 重构分区表示和生成随时间变化的关卡
Abstract
Video games are a dynamic medium experienced over time. While there are many Procedural Content Generation (PCG) approaches for generating video game levels, they often use representations that abstract away this dynamic nature. In this paper, we introduce a novel, domain-independent ``cake'' representation for game levels over time which implicitly encodes dynamic information. We present a novel level generation approach Playtrace Reconstructive Partitioning (PRP) specifically developed for this cake representation. We compare against six state-of-the-art PCG approaches in the game domain of \textit{Sokoban}, and find that our approach can generate valid levels without sacrificing solution diversity. We believe our cake representation more neatly encodes the implicit dynamic nature of games compared to existing representations, which allows for our domain-agnostic level generation algorithm PRP.
Chinese Translation
视频游戏是一种随时间变化的动态媒介。虽然有许多程序化内容生成（PCG）方法用于生成视频游戏关卡，但它们通常使用抽象化的表示方式，忽略了这种动态特性。本文介绍了一种新颖的、领域无关的“蛋糕”表示法，用于表示随时间变化的游戏关卡，该表示法隐含地编码了动态信息。我们提出了一种专门为这种蛋糕表示法开发的新型关卡生成方法——Playtrace 重构分区（PRP）。我们与六种最先进的PCG方法在游戏领域《推箱子》（Sokoban）中进行了比较，发现我们的方法能够生成有效的关卡而不牺牲解的多样性。我们相信，与现有表示法相比，我们的蛋糕表示法更好地编码了游戏的隐含动态特性，这使得我们的领域无关的关卡生成算法PRP得以实现。
cs.AI / 12 / 2607.12127
Connected by Construction: Learning Tractable Near-Tour Marginals for Traveling Salesman Problems
通过构造连接：学习可处理的近巡回边际以解决旅行商问题
Abstract
Learning-based methods for the traveling salesman problem (TSP) are often evaluated through the tours produced after decoding or search, but the learned object itself frequently lives in a surrogate space such as heatmaps, assignments, construction policies, or search-guidance scores. This hides the fundamental question: what Hamiltonian structure has actually been learned before decoding? In this study, we directly answer this question by learning TSP through a structurally meaningful latent object, rather than leaving most of the Hamiltonian structure to the final decoding stage. Based on a connected-by-construction rooted $1$-tree Gibbs family, we propose an end-to-end unsupervised learning pipeline called \emph{C2TSP}. The pipeline learns residual edge perturbations from unbiased TSP cost through implicit differentiation. For structural correction, a smoothed Held--Karp layer restores expected degree balance, while certificate-guided sharpening further pushes the connected distribution toward more tour-like structures. Experiments show that C2TSP yields strong decoding performance while preserving interpretable structural information. Ablations further verify that edge perturbation and certificate-guided sharpening jointly improve both tour cost and tour-like structure.
Chinese Translation
基于学习的方法解决旅行商问题（TSP）通常通过解码或搜索后生成的巡回路径进行评估，但学习的对象本身往往存在于诸如热图、分配、构造策略或搜索引导分数等代理空间中。这掩盖了一个根本性的问题：在解码之前，实际上学习到了什么样的哈密顿结构？在本研究中，我们通过学习一个结构上有意义的潜在对象直接回答了这个问题，而不是将大部分哈密顿结构留给最终的解码阶段。基于构造连接的根植于 $1$-树的 Gibbs 家族，我们提出了一种名为 extit{C2TSP} 的端到端无监督学习管道。该管道通过隐式微分从无偏的 TSP 成本中学习残余边缘扰动。为了进行结构修正，平滑的 Held-Karp 层恢复预期的度平衡，而证书引导的锐化进一步推动连接分布朝向更像巡回路径的结构。实验表明，C2TSP 在保持可解释的结构信息的同时，具有强大的解码性能。消融实验进一步验证了边缘扰动和证书引导的锐化共同改善了巡回成本和巡回路径结构。
cs.AI / 13 / 2607.12177
The Emerging Paradigm of Geospatial Foundation Models: From Pre-Training to Agentic Reasoning
地理空间基础模型的新兴范式：从预训练到智能推理
Abstract
The analysis of satellite and aerial imagery has entered a new era with the advent of foundation models. This paper describes the concept of Geospatial Foundation Models (GeoFMs), which are artificial intelligence/machine learning (AI/ML) models pre-trained on massive geospatial datasets through varied methodologies. We first articulate the core paradigm shift that GeoFMs enable: a separation of duties, where large-scale model providers perform the computationally intensive pretraining, allowing domain experts to rapidly fine-tune or prompt these models for specific, mission-critical tasks. This approach democratizes access to state-of-the-art AI/ML while maintaining the security and confidentiality of the downstream task. We then explore the novel capabilities unlocked by different types of GeoFMs, distinguishing between the finetunable vision models produced by self-supervised techniques like masked auto-encoding, and the vision-language models produced by contrastive learning which enable zero-shot tasks like open-vocabulary image analysis. Next, we discuss the practical considerations for operationalizing GeoFMs, from performance-cost analysis to the broader MLOps ecosystem. To that end, we introduce a taxonomy of model adaptation strategies and propose a framework for domain experts to select the most cost-effective adaptation approach for their particular mission set. Finally, we present a forward-looking vision of Agentic Geospatial Reasoning, where Large Language Models act as intelligent orchestrators, leveraging GeoFMs as tools to answer high-level user queries in natural language and automate complex analytical workflows, moving the field from perception to cognition.
Chinese Translation
卫星和航空影像的分析随着基础模型的出现进入了一个新纪元。本文描述了地理空间基础模型（GeoFMs）的概念，这些模型是通过多种方法在大规模地理空间数据集上进行预训练的人工智能/机器学习（AI/ML）模型。我们首先阐明了GeoFMs所促成的核心范式转变：职责的分离，大规模模型提供者进行计算密集型的预训练，使得领域专家能够快速微调或提示这些模型以完成特定的、任务关键型的工作。这种方法使最先进的AI/ML技术的获取变得民主化，同时保持下游任务的安全性和机密性。接着，我们探讨了不同类型的GeoFMs所解锁的新能力，区分了通过自监督技术（如掩蔽自编码）产生的可微调视觉模型与通过对比学习生成的视觉-语言模型，后者能够实现零样本任务，如开放词汇图像分析。随后，我们讨论了将GeoFMs投入实际应用时的考虑因素，从性能-成本分析到更广泛的MLOps生态系统。为此，我们引入了一种模型适应策略的分类法，并提出一个框架，帮助领域专家选择最具成本效益的适应方法以满足其特定任务集。最后，我们展望了智能地理空间推理的前景，其中大型语言模型作为智能协调者，利用GeoFMs作为工具，以自然语言回答高层次用户查询并自动化复杂的分析工作流程，从感知走向认知。
cs.AI / 14 / 2607.12188
Cost-Governed RAG: Unified Per-Tenant Cost Attribution Across Retrieval and Generation in Multi-Tenant LLM Systems
成本治理的检索增强生成（RAG）：多租户大语言模型系统中统一的每租户成本归属
Abstract
Enterprise Retrieval-Augmented Generation (RAG) deployments face a critical governance gap: while LLM generation cost is metered per token, the retrieval layer - vector memory, similarity compute, and embedding API calls - remains an unattributed shared cost, enabling invisible cross-subsidization among tenants. We present Cost-Governed RAG, an architecture that integrates a codebook-oblivious vector index (TurboVec) with a multi-tenant LLM governance gateway, creating a unified observability stack where embedding, retrieval, and generation costs are jointly attributable per tenant. The architecture exploits TurboVec's deterministic, closed-form memory formula to enable near-exact per-tenant retrieval cost calculation - a property unavailable in graph-based indexes with non-linear memory overhead. Deployed on Snowpark Container Services within a cloud data platform's governance boundary, the system achieves 99.96% end-to-end cost attribution accuracy across 100 simulated tenants (10M vectors, log-normal size distribution) with telemetry overhead below 0.04% of query latency. The architecture reduces retrieval infrastructure cost by 3.1-9.0x compared to managed vector database services under the pricing assumptions detailed in Section IV. We formalize a three-layer cost model and demonstrate that codebook-oblivious quantization enables deterministic per-tenant cost attribution while also removing the shared-codebook leakage surface present in trained quantizers - the latter observation being exploratory and subject to the limitations described in Section VII.
Chinese Translation
企业级检索增强生成（RAG）部署面临一个关键的治理缺口：虽然大语言模型（LLM）的生成成本是按每个令牌计量的，但检索层——向量存储、相似度计算和嵌入API调用——仍然是未归属的共享成本，从而导致租户之间的隐性交叉补贴。我们提出了成本治理的RAG架构，该架构将无代码本向量索引（TurboVec）与多租户LLM治理网关相结合，创建了一个统一的可观测性堆栈，使得嵌入、检索和生成的成本能够按租户共同归属。该架构利用TurboVec的确定性封闭形式内存公式，实现了近乎精确的每租户检索成本计算——这一特性在具有非线性内存开销的图形索引中是不可用的。该系统在云数据平台的治理边界内的Snowpark容器服务上部署，实现了在100个模拟租户（1000万个向量，对数正态大小分布）中99.96%的端到端成本归属准确率，遥测开销低于查询延迟的0.04%。该架构在与第IV节中详细说明的定价假设下，与托管向量数据库服务相比，将检索基础设施成本降低了3.1-9.0倍。我们形式化了一个三层成本模型，并证明无代码本量化能够实现确定性的每租户成本归属，同时消除了训练量化器中存在的共享代码本泄露表面——后者的观察是探索性的，并受到第VII节中描述的限制的影响。
cs.AI / 15 / 2607.12200
A Threshold Exceedance Framework for CBRN Uplift Evaluation in Frontier Language Models
前沿语言模型中CBRN提升评估的阈值超越框架
Abstract
As frontier language models advance, policymakers and model developers need methods for assessing whether model access materially increases a non-expert actor's ability to plan high-consequence Chemical, Biological, Radiological, or Nuclear (CBRN) misuse relative to public tools alone. Existing CBRN evaluations differ in non-expert definitions, threat scope, baselines, scoring rubrics, and decision rules, making results difficult to compare across studies. We introduce a Threshold Exceedance Criteria (TEC) framework that decomposes an uplift study into independently executable components: determining non-expert participant eligibility, defining the CBRN threat scope for the study, and statistically estimating material uplift. We then operationalize the TEC framework in a large-scale empirical study using a design that determines two forms of uplift: generative (where a model assists plan creation from scratch) and revisionist (where a model assists refinement of an existing plan). The study produced attack plans across the CBRN domains, which we evaluated through subject-matter-expert review to estimate generative and revisionist uplift. Applying the framework, our empirical study revealed domain heterogeneity: under this controlled pre-release evaluation, model-assisted plans sometimes received expert-equivalent instructional ratings, but confirmed material uplift was limited to the radiological domain. These findings informed mitigation and deployment-governance decisions rather than characterizing deployed model behavior. We conclude with methodological lessons for future CBRN uplift evaluations, emphasizing prespecified criteria, explicit baselines, separation of generative and revisionist estimates, and careful distinction between preliminary screening signals and confirmed risk determinations.
Chinese Translation
随着前沿语言模型的进步，政策制定者和模型开发者需要评估方法，以判断模型访问是否实质性地提高了非专家参与者相较于仅使用公共工具进行高后果化学、生物、放射性或核（CBRN）滥用计划的能力。现有的CBRN评估在非专家定义、威胁范围、基准、评分标准和决策规则上存在差异，使得跨研究的结果难以比较。我们提出了一个阈值超越标准（Threshold Exceedance Criteria, TEC）框架，将提升研究分解为可独立执行的组件：确定非专家参与者的资格、定义研究的CBRN威胁范围，以及统计估计实质性提升。随后，我们在一项大规模实证研究中操作化了TEC框架，该研究设计确定了两种形式的提升：生成性（模型从零开始协助创建计划）和修订性（模型协助完善现有计划）。该研究在CBRN领域生成了攻击计划，并通过主题专家评审对生成性和修订性提升进行了评估。应用该框架，我们的实证研究揭示了领域异质性：在这一受控的发布前评估中，模型辅助的计划有时获得了专家等效的指导评分，但确认的实质性提升仅限于放射性领域。这些发现为减轻和部署治理决策提供了信息，而不是描述已部署模型的行为。我们总结了未来CBRN提升评估的 методологические уроки，强调预先指定的标准、明确的基准、生成性和修订性估计的分离，以及对初步筛选信号和确认风险判断之间的仔细区分。
cs.AI / 16 / 2607.12217
Good Benchmarks
良好的基准
Abstract
Good tasks are correct, solvable, verifiable, well-specified, and hard for interesting reasons. The best tasks describe a real problem an experienced practitioner would recognize, in language a practitioner would use, with tests that verify the outcome rather than the approach.
Chinese Translation
良好的任务应是正确的、可解的、可验证的、明确规定的，并且因有趣的原因而具有一定的难度。最佳任务描述的是经验丰富的从业者能够识别的真实问题，使用从业者所用的语言，并且通过验证结果而非方法的测试来确认结果。
cs.AI / 17 / 2607.12227
Rethinking the Evaluation of Harness Evolution for Agents
重新思考代理的自动化测试框架演化评估
Abstract
We revisit the evaluation of automatic harness evolution for LLM agents. Existing harness evolution methods use unit test cases to search for harness configurations and then report final performance on the same public benchmark. This protocol raises two fundamental concerns. First, harness evolution is itself an iterative search procedure that repeatedly evaluates and revises candidate harnesses using task feedback. As in agentic test-time scaling, it should therefore be compared with simple task-level search baselines under matched feedback and inference budgets to determine whether its gains arise from improved harness design or from additional search alone. Second, because the search and the final evaluation share the same benchmark, the reported gains risk overfitting to that specific task set. To address these concerns, we conduct an extensive evaluation comparing harness evolution with simple test-time scaling and discovery baselines under comparable feedback and inference budgets, and also evaluate evolved harnesses on held-out tasks to assess whether the discovered improvements generalize. Experiments on Terminal-Bench 2.1 with GPT-5.4 and Claude Opus 4.6 show that automatic harness evolution does not consistently outperform simple test-time scaling methods and exhibits limited generalization. Our results raise important questions about the effectiveness of automatic harness evolution and highlight the need for fairer evaluation protocols and benchmarks for automatic harness design. Our code is available at https://github.com/rethinking-harness-evolution.
Chinese Translation
我们重新审视了大型语言模型（LLM）代理的自动化测试框架演化评估。现有的测试框架演化方法使用单元测试用例来搜索测试框架配置，然后在同一公共基准上报告最终性能。这一协议引发了两个基本问题。首先，测试框架演化本身就是一个迭代搜索过程，它反复评估和修订候选测试框架，利用任务反馈。因此，应该在匹配的反馈和推理预算下，将其与简单的任务级搜索基准进行比较，以确定其收益是源于改进的测试框架设计，还是仅仅来自额外的搜索。其次，由于搜索和最终评估共享相同的基准，报告的收益存在过拟合于特定任务集的风险。为了解决这些问题，我们进行了广泛的评估，将测试框架演化与简单的测试时间缩放和发现基准进行比较，且在可比的反馈和推理预算下进行评估，同时还在保留任务上评估演化后的测试框架，以评估发现的改进是否具有泛化能力。在使用GPT-5.4和Claude Opus 4.6的Terminal-Bench 2.1上的实验表明，自动化测试框架演化并不总是优于简单的测试时间缩放方法，并且表现出有限的泛化能力。我们的结果提出了关于自动化测试框架演化有效性的重要问题，并强调了对自动化测试框架设计进行更公平的评估协议和基准的需求。我们的代码可在https://github.com/rethinking-harness-evolution获取。
cs.AI / 18 / 2607.12257
On-Device Deep Research at 4B: Exposure Bounds Faithfulness, Retrieval Bounds Coverage
在4B设备上的深度研究：曝光限制影响可信度，检索限制影响覆盖率
Abstract
On-device research agents search a corpus, read sources, and write a cited brief on a personal laptop. Whether their citations are faithful, and at what cost, is unmeasured for a deployable small model. This study fixes one 4B generator on a 24 GB laptop and asks what makes its citations faithful. It separates two quantities usually reported as one number. Cited claim faithfulness asks whether the cited source supports the claim. Trustworthy coverage asks whether the agent also cites the right sources. The study crosses how much of each source the generator sees, 400 against 1500 characters, with the quality of the sources supplied, gold papers against retrieved papers. Two levers fall out, and they act on different outcomes. Exposure sets faithfulness. More of each source lifts faithfulness from 0.45 to 0.58 on retrieved sources and from 0.37 to 0.58 on gold sources, and the two settings converge, so faithfulness is bound by exposure, not by whether the source is correct. The exposure lift is robust to a second, independent judge; the exact convergence is tight under the primary judge and only approximate under the second. Retrieval sets coverage. Trustworthy coverage stays near 0.22 on retrieved sources at any exposure, because recall is held near 0.40, so exposure cannot fix which sources are cited. The extra exposure costs about 235 output tokens. The practical recipe is to raise per source exposure first, cheaply, and then treat retrieval recall as the only remaining lever.
Chinese Translation
设备上的研究代理在一个语料库中进行搜索，阅读来源，并在个人笔记本电脑上撰写引用简报。它们的引用是否可信，以及成本是多少，对于可部署的小模型尚未进行测量。本研究在一台24GB的笔记本电脑上固定一个4B生成器，并探讨是什么使得其引用具有可信性。它将通常报告为一个数字的两个量分开。被引用的主张可信度询问被引用的来源是否支持该主张。可信的覆盖率询问代理是否也引用了正确的来源。该研究交叉了生成器看到的每个来源的字符数，400与1500字符，以及提供的来源质量，黄金论文与检索论文。两个杠杆显现出来，并对不同的结果产生作用。曝光设置了可信度。每个来源的更多曝光使得在检索来源上的可信度从0.45提升至0.58，在黄金来源上的可信度从0.37提升至0.58，这两个设置收敛，因此可信度受曝光限制，而不是来源是否正确。曝光提升对第二位独立评审保持稳健；在主要评审下，确切的收敛非常紧密，而在第二位评审下仅为近似。检索设置了覆盖率。可信的覆盖率在任何曝光下对检索来源保持在0.22附近，因为召回率保持在0.40附近，因此曝光无法修复引用了哪些来源。额外的曝光成本约为235个输出标记。实际的方案是首先提高每个来源的曝光，成本低廉，然后将检索召回率视为唯一剩余的杠杆。
cs.AI / 19 / 2607.12338
How Many Tasks Are Enough for Agent Benchmark Decisions? A Replay Analysis of Public LLM Agent Benchmarks
代理基准决策需要多少任务？对公共大型语言模型代理基准的重播分析
Abstract
Agent benchmarks often compare two agents after all tasks have run, but costly evaluations make partial runs tempting. A task fraction alone does not show whether a partial run supports the same pairwise conclusion as the completed benchmark. We study this question by replaying completed public task-level records from SWE-bench, AppWorld, and tau-bench. A partial budget counts as enough only when it supports the completed benchmark's decision, covers required task groups, and leaves no more than a target fraction of comparisons unresolved. The required task fraction varies sharply. At the strict 0 percentage point threshold on a 5 percentage point budget grid, AppWorld first meets all targets at 15 percent, tau-bench at 25 percent, and SWE-bench Verified at 90 percent; SWE-bench Lite does not meet all targets by 95 percent under the primary coverage rule. Partial-evaluation reports should state how much one agent must outperform another, how tasks are selected, what coverage rule is required, what decision rule is used, and how many comparisons may remain unresolved.
Chinese Translation
代理基准通常在所有任务完成后比较两个代理，但高昂的评估成本使得部分运行变得诱人。仅凭任务比例无法表明部分运行是否支持与完成基准相同的成对结论。我们通过重播来自SWE-bench、AppWorld和tau-bench的已完成公共任务级记录来研究这个问题。只有当部分预算支持完成基准的决策、覆盖所需的任务组，并且未解决的比较不超过目标比例时，才算足够。所需的任务比例变化显著。在5个百分点预算网格的严格0个百分点阈值下，AppWorld在15%时首次满足所有目标，tau-bench在25%时满足，SWE-bench Verified在90%时满足；而SWE-bench Lite在主要覆盖规则下在95%时未能满足所有目标。部分评估报告应说明一个代理必须超越另一个代理的程度、任务选择的方式、所需的覆盖规则、使用的决策规则，以及可能未解决的比较数量。
cs.AI / 20 / 2607.12385
PM-Bench: Evaluating Prospective Memory in LLM Agents
PM-Bench：评估大型语言模型代理的前瞻性记忆
Abstract
A significant challenge in agentic AI is prospective memory: the ability to execute an intention at a specific future cue or state while other activities are ongoing. We introduce PM-Bench, a text-based benchmark for measuring prospective memory capabilities in modern LLM agents. Inspired by the Virtual Week paradigm from cognitive science, PM-Bench evaluates how well LLM agents maintain user intentions, execute delayed intentions, and monitor latent environment changes. Over the course of a simulated seven-day week, agents must continue an ongoing activity while deciding whether any deferred task is due. We compare eight state-of-the-art LLMs on PM-Bench under eight different agent configurations. PM-Bench proves challenging across all settings: the best method, a GPT-5.4 agent, reaches only 65.1\% F1 score under our evaluation. Furthermore, no single strategy for improving prospective memory dominates across models. We release PM-Bench as a controlled testbed for diagnosing these failures and developing training or inference-time interventions that support reliable prospective behavior.
Chinese Translation
代理人工智能面临的一个重大挑战是前瞻性记忆：在其他活动进行的同时，在特定的未来提示或状态下执行意图的能力。我们介绍了PM-Bench，这是一个基于文本的基准，用于测量现代大型语言模型（LLM）代理的前瞻性记忆能力。PM-Bench受到认知科学中虚拟周（Virtual Week）范式的启发，评估LLM代理在多大程度上保持用户意图、执行延迟意图以及监测潜在环境变化。在模拟的七天周中，代理必须在继续进行一项活动的同时，决定是否有任何延迟任务到期。我们在八种不同的代理配置下比较了八种最先进的LLM在PM-Bench上的表现。PM-Bench在所有设置中都证明了其挑战性：最佳方法，即GPT-5.4代理，在我们的评估中仅达到65.1%的F1得分。此外，没有任何单一策略在各模型中主导于改善前瞻性记忆。我们发布PM-Bench作为一个受控测试平台，以诊断这些失败并开发支持可靠前瞻性行为的训练或推理时干预措施。
cs.AI / 21 / 2607.12397
Critic Experience Bank: Self-Evolving Step-Level Confidence Estimation for LLM Agents
评论者经验库：自我演化的逐步置信度估计用于大型语言模型代理
Abstract
LLM agents act in external environments where each action changes the state that later decisions condition on, and where a single wrong step can waste interaction budget or trigger irreversible side effects long before the final failure is observed. Reliable deployment therefore requires \emph{step-level confidence estimation}: a calibrated probability that each proposed action is productive, available \emph{before} the action is executed. Existing LLM confidence estimators are designed to score a response from the given prompt, but agent confidence also depends on execution consequences: whether similar actions in similar situations actually advanced the task after the environment responded. We introduce the \method (\methodshort), a self-evolving critic framework in which an LLM critic accumulates evidence from its own past judgments and their observed consequences. After each trajectory, a hindsight LLM that sees the full execution feedback votes on whether each step was productive. The resulting pseudo-labels populate a memory bank from which related productive and unproductive experiences are retrieved into the critic's prompt whenever a similar step recurs. \methodshort requires no training and uses no ground truth step labels. Across three agent benchmarks and three critic backbones, \methodshort attains the best calibration (ECE and Brier) and ranking (AUC) in every dataset--critic combination, reducing ECE by up to $54\%$ relative to the strongest training-free baseline.
Chinese Translation
大型语言模型（LLM）代理在外部环境中执行操作，每个操作都会改变后续决策所依赖的状态，而单个错误的步骤可能会浪费交互预算或在最终失败被观察到之前触发不可逆的副作用。因此，可靠的部署需要 extit{逐步置信度估计}：在执行操作之前，校准每个提议操作是有效的概率。现有的LLM置信度估计器旨在对给定提示的响应进行评分，但代理的置信度还依赖于执行结果：在环境响应后，类似情况下的类似操作是否确实推动了任务的进展。我们引入了 extit{Critic Experience Bank}（ extit{CEB}），这是一个自我演化的评论者框架，其中LLM评论者从其过去的判断及其观察到的结果中积累证据。在每个轨迹之后，一个能够看到完整执行反馈的后见LLM对每个步骤是否有效进行投票。生成的伪标签填充了一个记忆库，当类似步骤再次出现时，相关的有效和无效经验会被检索到评论者的提示中。 extit{CEB}不需要训练，也不使用真实的步骤标签。在三个代理基准和三个评论者基础上， extit{CEB}在每个数据集-评论者组合中都达到了最佳的校准（ECE和Brier）和排名（AUC），相较于最强的无训练基线，ECE降低了多达54%。
cs.AI / 22 / 2607.12406
Isolation as a First-Class Principle for LLM-Agent System Safety: Concepts, Taxonomy, Challenges and Future Directions
将隔离作为大型语言模型代理系统安全的首要原则：概念、分类、挑战与未来方向
Abstract
The capability of LLM agents to function as the ``brain'' of a system fundamentally expands the scope of analysis beyond a standalone model. Consequently, safety is no longer only about input--output content alignment. It also concerns system behavior and real-world execution outcomes. However, the current literature is fragmented across attack types, applications, and benchmarks. This makes it hard to explain why failures such as prompt injection, tool misuse, and memory poisoning often share the same structural cause, and how they spread through an agent workflow. In this survey, we treat isolation as a first-class principle for LLM-agent system safety. By isolation, we refer to the separation of user inputs, tool access, execution channels, inter-agent communication, and environment-originated context. We organize the literature with a boundary-centric taxonomy of five boundaries: user-agent, agent-tool, agent-execution, agent-agent, and system-environment. This view helps identify where the loss of isolation first occurs, how compromise propagates across boundaries, and which defenses are most relevant at each interface. We also summarize cross-boundary failure paths, discuss open challenges, and outline a research agenda for isolation-by-construction in future agent systems.
Chinese Translation
大型语言模型（LLM）代理作为系统“脑”的能力，根本上扩展了分析的范围，超越了单一模型。因此，安全性不再仅仅关乎输入与输出内容的一致性，还涉及系统行为和现实世界执行结果。然而，现有文献在攻击类型、应用和基准测试方面存在碎片化。这使得很难解释为何诸如提示注入、工具误用和内存中毒等失败通常共享相同的结构性原因，以及它们如何在代理工作流中传播。在本次调研中，我们将隔离视为LLM代理系统安全的首要原则。我们所指的隔离是指用户输入、工具访问、执行通道、代理间通信以及环境来源上下文的分离。我们以边界为中心的五个边界分类组织文献：用户-代理、代理-工具、代理-执行、代理-代理和系统-环境。这种视角有助于识别隔离首次丧失的地方、妥协如何在边界间传播，以及在每个接口上哪些防御措施最为相关。我们还总结了跨边界失败路径，讨论了开放挑战，并为未来代理系统中的构建隔离研究制定了研究议程。
cs.AI / 23 / 2607.12422
Accepted Prefixes Are Not All You Need: A Negative Result on PEFT-Based Block-Diffusion Drafting
接受的前缀并不是你所需要的一切：关于基于PEFT的块扩散草拟的负面结果
Abstract
Speculative decoding accelerates autoregressive language model inference by using a cheap drafter to propose multiple future tokens and a target model to verify them. A common design goal is therefore to improve draft quality while reducing auxiliary parameters and systems overhead. We study a negative result for this direction through PEFT-BD, a same-backbone speculative decoding method in which a LoRA-like adapter acts as a block-diffusion drafter for an autoregressive verifier. PEFT-BD is motivated by several attractive properties: it avoids tokenizer mismatch, avoids loading a separate draft model, adds only a small number of trainable parameters, and uses a BD3LM-style denoising objective to propose a block of tokens in parallel. Despite these advantages, PEFT-BD does not yield a practical speedup in our Qwen3-0.6B experiments. Although the method obtains nontrivial accepted prefixes, profiling shows that each speculative step requires an adapter-enabled full-backbone draft pass followed by an adapter-disabled full-backbone verification pass. Thus, the drafter is parameter-efficient but not compute-efficient. Our results isolate a simple but important condition for successful speculative decoding: the drafter must be substantially cheaper to execute than the verifier. Longer accepted prefixes alone cannot compensate when draft computation remains verifier-scale.
Chinese Translation
推测解码通过使用廉价的草拟器提出多个未来标记并使用目标模型进行验证，从而加速自回归语言模型的推理。因此，一个常见的设计目标是提高草拟质量，同时减少辅助参数和系统开销。我们通过PEFT-BD研究了这一方向的负面结果，PEFT-BD是一种相同骨干的推测解码方法，其中一个类似LoRA的适配器充当自回归验证器的块扩散草拟器。PEFT-BD的动机源于几个吸引人的特性：它避免了分词器不匹配，避免加载单独的草拟模型，仅增加少量可训练参数，并使用BD3LM风格的去噪目标并行提出一组标记。尽管具有这些优势，在我们的Qwen3-0.6B实验中，PEFT-BD并未带来实质性的加速。尽管该方法获得了非平凡的接受前缀，但分析显示每个推测步骤都需要一个启用适配器的完整骨干草拟过程，随后是一个禁用适配器的完整骨干验证过程。因此，草拟器在参数上是高效的，但在计算上并不高效。我们的结果隔离了成功推测解码的一个简单但重要的条件：草拟器的执行成本必须显著低于验证器。当草拟计算仍然保持在验证器规模时，单靠较长的接受前缀无法弥补。
cs.AI / 24 / 2607.12455
EVOQUANT: Self-Evolving Verifier-Guided Strategy Optimization for Robust Quantitative Trading
EVOQUANT：自我演化的验证者引导策略优化框架用于稳健的量化交易
Abstract
Quantitative strategy optimization remains largely manual, requiring domain experts to identify weak signals, tune risk-control rules, and repeatedly validate iterative revisions. Large language models can accelerate this process, but directly relying on them to rewrite trading strategies often introduces hallucinated edits, strategy drift, and backtest overfitting. We propose EVOQUANT, a self-Evolving Verifier-guided framework for strategy Optimization in Quantitative trading. Our method utilizes LLMs to deeply diagnose performance bottlenecks, generates semantically controlled candidate edits, selects the best strategy through a multi-stage verification pipeline, and distills optimization experience into reusable knowledge for continual self-improvement. We evaluate our method using seven representative strategies: four from the A-share market and three from the Crypto market. Experimental results show that our method significantly improves the Sharpe ratio across all tested strategies: the average test Sharpe increases from -0.298 to 0.538, and the best-performing strategy achieves a 199% relative improvement. Ablation studies and stress tests under stricter conditions further validate the effectiveness and robustness of the framework. Overall, this work transforms quantitative strategy optimization from costly manual trial and error into an automated and verifiable iterative paradigm, offering a new path for applying large language models to financial strategy research.
Chinese Translation
量化策略优化在很大程度上仍然是手动进行的，要求领域专家识别弱信号、调整风险控制规则，并反复验证迭代修订。大型语言模型可以加速这一过程，但直接依赖它们重写交易策略往往会引入虚假编辑、策略漂移和回测过拟合。我们提出了EVOQUANT，一个自我演化的验证者引导框架，用于量化交易中的策略优化。我们的方法利用大型语言模型（LLMs）深入诊断性能瓶颈，生成语义控制的候选编辑，通过多阶段验证流程选择最佳策略，并将优化经验提炼为可重用的知识，以实现持续自我改进。我们使用七个具有代表性的策略评估我们的方法：四个来自A股市场，三个来自加密市场。实验结果表明，我们的方法显著提高了所有测试策略的夏普比率：平均测试夏普从-0.298提高到0.538，表现最佳的策略实现了199%的相对提升。消融研究和在更严格条件下的压力测试进一步验证了该框架的有效性和稳健性。总体而言，这项工作将量化策略优化从高成本的手动试错转变为一种自动化和可验证的迭代范式，为将大型语言模型应用于金融策略研究提供了一条新路径。
cs.AI / 25 / 2607.12462
Do We Really Need Transformers for Global Spatial Information Extraction in Traffic Forecasting?
我们真的需要变压器来进行交通预测中的全球空间信息提取吗？
Abstract
Existing traffic forecasting models commonly focus on extracting spatial dependencies, particularly global spatial information, which characterizes the representations obtained through interactions between each individual node and all nodes across the traffic network. However, the underlying mechanism by which such global information is modeled and extracted remains insufficiently investigated. Whether global information must be extracted by high-degree-of-freedom adaptive attention or can be captured by a simple global aggregation operator remains unclear. For this purpose, we design a controlled ablation framework that replaces only the spatial mixing module to test attention-based global interaction. Across six traffic benchmarks, uniform full-range mixing and standard spatial attention each achieve lower MAE on three datasets, with only a 0.14% difference in mean MAE, while the former reduces node-scale spatial mixing complexity from O(N2) to O(N). Mechanism analysis further decomposes spatial attention into a row-uniform global background and a non-uniform residual. The residual shows dataset-dependent marginal value, suggesting that spatial attention should be justified by stable gains beyond a row-uniform global background. The corresponding source code is publicly available at: https://github.com/uuesti/U-Trans
Chinese Translation
现有的交通预测模型通常侧重于提取空间依赖关系，特别是全球空间信息，这种信息通过交通网络中每个节点与所有节点之间的交互来表征。然而，如何建模和提取这种全球信息的基本机制仍然未得到充分研究。全球信息是否必须通过高自由度的自适应注意力来提取，或者可以通过简单的全球聚合算子来捕获，仍然不明确。为此，我们设计了一个受控消融框架，仅替换空间混合模块以测试基于注意力的全球交互。在六个交通基准测试中，均匀全范围混合和标准空间注意力在三个数据集上均实现了较低的平均绝对误差（MAE），两者之间的平均MAE仅相差0.14%，而前者将节点规模的空间混合复杂度从O(N²)降低到O(N)。机制分析进一步将空间注意力分解为行均匀的全球背景和非均匀的残差。残差显示出依赖于数据集的边际价值，表明空间注意力应通过超出行均匀全球背景的稳定收益来证明其合理性。相应的源代码可在以下网址公开获取：https://github.com/uuesti/U-Trans
cs.AI / 26 / 2607.12463
Function-Aware Fill-in-the-Middle as Mid-Training for Coding Agent Foundation Models
功能感知的中间填充作为编码代理基础模型的中期训练
Abstract
Coding agents must integrate external tool returns into ongoing reasoning - a capability that standard left-to-right pretraining on code exposes only in its forward direction. We observe that the action-observation-continuation loop of a coding agent is structurally isomorphic to a function call site, where a caller binds arguments, a callee returns a value computed elsewhere, and downstream code consumes that value. This conditioning structure exists at internet scale in ordinary code. We exploit it through function-aware fill-in-the-middle (FIM) mid-training: a self-supervised objective that masks functions selected via program dependency graph analysis and a complexity-inferability double criterion. We mid-train Qwen2.5-Coder-Instruct (7B/14B) and Qwen3-8B on a 2.6B-token decontaminated corpus drawn from 968 GitHub repositories, then apply existing agentic post-training pipelines. Mid-training improves SWE-Bench-Verified by +2.8/+3.0 at 7B/14B and by +3.2 on Qwen3-8B; SWE-Bench-Lite gains are +3.7/+4.0/+5.4 on the same models. The improvement holds across two post-training pipelines (R2E-Gym, SWE-Smith) and on a non-Qwen2.5 base (Qwen3-8B with SWE-Lego). Beyond in-domain gains, mid-training also mitigates the capability erosion that agentic post-training otherwise inflicts on non-agent coding (e.g., LiveCodeBench) and non-coding tool-use benchmarks (tau-bench, BFCL): although the mid-training corpus contains Python code only, the function-call inductive bias survives post-training and yields consistent gains.
Chinese Translation
编码代理必须将外部工具返回的结果整合到持续的推理中——这一能力在标准的从左到右的代码预训练中仅在其前向方向上得以暴露。我们观察到，编码代理的动作-观察-继续循环在结构上与函数调用位置同构，其中调用者绑定参数，被调用者返回在其他地方计算的值，下游代码消费该值。这种条件结构在普通代码中以互联网规模存在。我们通过功能感知的中间填充（Function-Aware Fill-in-the-Middle, FIM）中期训练来利用它：这是一种自监督目标，通过程序依赖图分析和复杂性可推断性双重标准选择函数进行掩蔽。我们在从968个GitHub代码库中提取的2.6B标记去污染语料库上对Qwen2.5-Coder-Instruct（7B/14B）和Qwen3-8B进行中期训练，然后应用现有的代理后训练管道。中期训练使得SWE-Bench-Verified在7B/14B上分别提高了+2.8/+3.0，在Qwen3-8B上提高了+3.2；SWE-Bench-Lite在相同模型上获得的增益为+3.7/+4.0/+5.4。该改进在两个后训练管道（R2E-Gym, SWE-Smith）和非Qwen2.5基础（Qwen3-8B与SWE-Lego）上均有效。除了领域内的增益外，中期训练还减轻了代理后训练对非代理编码（例如，LiveCodeBench）和非编码工具使用基准（tau-bench, BFCL）所造成的能力侵蚀：尽管中期训练语料库仅包含Python代码，但函数调用的归纳偏差在后训练中得以保留，并产生了一致的增益。
cs.AI / 27 / 2607.12474
From Observation to Insight: Mechanistic World Models and the Quest for Autonomous Discovery
从观察到洞察：机制世界模型与自主发现的探索
Abstract
Recent advances in foundation models have transformed AI for Science, enabling remarkably accurate predictive performance across domains ranging from protein folding to weather forecasting. Yet prediction alone does not constitute scientific discovery. Scientific understanding depends on uncovering the reusable explanatory mechanisms that generate observations, whereas contemporary machine learning remains fundamentally organised around predictive mappings rather than explanatory structure. In this paper, we argue that scientific discovery is fundamentally a problem of knowledge organisation. To this end, we introduce Mechanistic World Models, a new design paradigm that places reusable mechanisms at the centre of representation, computation and learning. Drawing on insights from the philosophy of science, we derive the computational capabilities required for discovery, identify the design principles and inductive pressures that encourage explanatory knowledge to emerge, and formalise the anatomy of a mechanism-centric world model. Finally, we show how diverse research directions including mechanistic interpretability, causal representation learning, equation discovery and modular architectures capture complementary ingredients of this paradigm while lacking a unified framework. We propose Mechanistic World Models as a conceptual foundation and computational blueprint for moving AI beyond predictive forecasting towards autonomous scientific discovery.
Chinese Translation
近期基础模型的进展已经改变了科学领域的人工智能，使其在从蛋白质折叠到天气预报等多个领域实现了显著准确的预测性能。然而，仅仅依靠预测并不能构成科学发现。科学理解依赖于揭示生成观察结果的可重用解释机制，而当代机器学习仍然基本围绕预测映射而非解释结构进行组织。本文论证，科学发现根本上是一个知识组织的问题。为此，我们引入机制世界模型（Mechanistic World Models），这一新的设计范式将可重用机制置于表示、计算和学习的中心。基于科学哲学的见解，我们推导出发现所需的计算能力，识别出鼓励解释性知识涌现的设计原则和归纳压力，并形式化机制中心世界模型的结构。最后，我们展示了包括机制可解释性、因果表示学习、方程发现和模块化架构在内的多样化研究方向如何捕捉这一范式的互补成分，同时缺乏统一的框架。我们提出机制世界模型作为一个概念基础和计算蓝图，以推动人工智能从预测性预报向自主科学发现迈进。
cs.AI / 28 / 2607.12480
TRACE: An Operational Reasoning Schema for Auditable Agentic Commitments
TRACE：可审计代理承诺的操作推理模式
Abstract
This paper defines TRACE (Typed Reasoning And Commitment Evidence): a typed, versioned schema for recording reasoning traces, a reference procedure for writing records against it, and one operating discipline, no durable state change without a record. The paper argues in three layers that reasoning is not in the language model: the autoregressive mechanism natively computes association; chain-of-thought and reinforcement learning inherit its limits; and the formal constructs of reasoning theory, from Socratic procedure to Pearl's ladder, are absent as machinery. The schema answers the absence with fields and tests: the TraceRecord and its causal specialization, an eight-stage reference writer, a gate-first measurement regime, the TRACE-Bench protocol, and the consumers, memory admission, plan gating, temporal regret, and verdict reuse, whose more auditable decisions are the measure of the record. A record-consumer contract states what a record guarantees and what a consumer must honor in return, making the schema an operational interface rather than a passive document. Two worked examples run in the main text: a music-lessons argument traced from sentence to typed verdict, separating association, intervention, and prescription; and a flood search-and-rescue vignette in which a predictive world model reports confident plan success that its own support and out-of-distribution scores contradict, so the record defers the commitment, requests a bounded observation, revises append-only, and clears a different branch. The vignette is illustrative, not empirical; closed-loop evaluation is left to future work, so the contribution is the schema and its contract, not a performance claim. Appendices carry the full schema, writer algorithms and cost model, clinical and policy illustrations, the benchmark protocol, convergence metrics, and usage scenarios.
Chinese Translation
本文定义了TRACE（类型化推理与承诺证据）：一种用于记录推理痕迹的类型化版本模式、一种针对其编写记录的参考程序，以及一种操作纪律，即没有记录就不进行持久状态变化。本文从三个层面论证推理并不在语言模型中：自回归机制本质上计算关联；思维链和强化学习继承了其局限性；而推理理论的形式构造，从苏格拉底程序到Pearl的阶梯，作为机制缺失。该模式通过字段和测试回应这种缺失：TraceRecord及其因果专业化、一个八阶段的参考写入器、一个优先测量的门控机制、TRACE-Bench协议，以及消费者、记忆接纳、计划门控、时间遗憾和裁决重用，其更可审计的决策是记录的衡量标准。记录消费者合同规定了记录所保证的内容以及消费者必须回报的内容，使该模式成为一种操作接口，而非被动文档。文中提供了两个实例：一个从句子到类型化裁决的音乐课程论证，分离了关联、干预和处方；以及一个洪水搜索与救援的插曲，其中一个预测世界模型报告了自信的计划成功，而其自身的支持和分布外得分却相互矛盾，因此记录推迟了承诺，请求有界观察，修订为仅附加，并清除不同的分支。该插曲是说明性的，而非实证性的；闭环评估留待未来工作，因此贡献在于模式及其合同，而非性能声明。附录包含完整的模式、写入算法和成本模型、临床和政策插图、基准协议、收敛指标和使用场景。
cs.AI / 29 / 2607.12520
The Model Knows Your Project, Not You: Measuring Recognition in LLMs with NameRank
模型了解你的项目，而不是你：使用 NameRank 测量大语言模型的识别能力
Abstract
What a frontier model recalls about a person or tool from its own weights -- before any retrieval step -- often shapes the first description a human sees, making that parametric corpus presence a measurement problem. Citations explain about a third of whether a model recognizes a researcher; we target the residual and build NameRank, a [0,1] recognition score: each of 4,685 entities in 54 cohorts is probed with one open-ended question across 36 models, and an independent judge returns a binary verdict against a curated gold -- did the model state a specific, non-guessable fact about this exact entity? -- so hallucination, context echo, and guesses earn nothing. Synthetic-null entities hold the floor near zero, and verdicts track the entity, not the model. One thesis organizes the findings: recognition is paid to named, indexable artifacts, not to credentials or titles. Every Olympic-style credential sits below a working-researcher baseline, because no named artifact ships with the medal, yet the ranking inverts at the marquee tier, where Nobel, Turing, and Fields laureates saturate the panel. For independent creators the tool out-ranks its maker, and the credential that does propagate is a named method or awarded paper. Being one of many named contributors to a celebrated artifact, by contrast, earns almost nothing -- the authors listed on a flagship model report or system card sit near the recognition floor -- because recognition attaches to the artifact's own distinctive name, not to the roster behind it. No bibliometric predicts recognition well; top-density institutions out-recognize peers at matched citations; and on 258 news events recognition loads on peak salience, not persistence. A self-report probe shows introspection reads a corpus prior, not its own knowledge.
Chinese Translation
前沿模型从自身权重中回忆起关于某个人或工具的信息——在任何检索步骤之前——通常会影响人类所看到的第一描述，这使得这种参数化语料库的存在成为一个测量问题。引用解释了模型是否识别研究者的三分之一；我们针对剩余部分构建了 NameRank，一个 [0,1] 的识别评分：在 54 个群体中对 4,685 个实体进行了一次开放式问题的探测，涵盖 36 个模型，并由独立评审员根据精心策划的标准返回二元判决——模型是否陈述了关于该特定实体的一个具体且不可猜测的事实？——因此，幻觉、上下文回声和猜测都不会得分。合成的零实体保持在接近零的水平，判决跟踪的是实体，而不是模型。一项论点组织了这些发现：识别是给予命名的、可索引的文物，而不是凭证或头衔。每一个奥林匹克风格的凭证都低于一个工作研究者的基线，因为没有命名的文物附带奖牌，但在显著层级的排名却是反转的，诺贝尔奖、图灵奖和菲尔兹奖得主充斥其中。对于独立创作者而言，工具的识别高于其创造者，而传播的凭证是命名的方法或获奖论文。相比之下，作为一个著名文物的众多命名贡献者之一几乎得不到任何识别——旗舰模型报告或系统卡上的作者接近识别底线——因为识别附着在文物自身独特的名称上，而不是其背后的名单。没有任何文献计量指标能够很好地预测识别；在匹配引用的情况下，顶级密度机构的识别能力超过同行；在 258 个新闻事件中，识别集中在高峰显著性上，而非持久性。自我报告的探测显示，内省读取的是先前的语料库，而不是其自身的知识。
cs.AI / 30 / 2607.12527
Evidence-Grounded AI for Musculoskeletal Care
基于证据的人工智能在肌肉骨骼护理中的应用
Li, Wenjie, Zhang, Yujie, Zhang, Fanrui, Sun, Haoran, Yang, Renhao, He, Junjun, Huang, Weiran, Ji, Yuanfeng, Wang, Chenrun, Wang, Kailing, Gao, Hongcheng, Zhang, Kaipeng, Wang, Hanyu, Wang, Angela Lin, He, Xingqi, Huang, Yilin, Yao, Shiyi, Wang, Lilong, Jiang, Yankai, Chen, Yirong, Ma, Chenglong, Liu, Jiyao, Hu, Ming, Li, Gen, Xu, Yidong, Zhuang, Chengyu, Liu, Jiawei, Zhang, Yin, Yu, Lequan, Chen, Lu, Dong, Yinpeng, Liu, Lei, Sanroman, Carlos Gutierrez, Qiao, Yu, Ma, Weijie, Wang, Xiaosong, Wang, Lei
Abstract
Musculoskeletal diseases are among the leading causes of disability worldwide and create the greatest global need for rehabilitation. Because recovery, remodelling and degeneration often unfold over months to years, musculoskeletal care requires longitudinal management that repeatedly integrates evolving patient evidence, external medical knowledge and stage-specific functional goals. In routine practice, this evidence is fragmented across visits, departments and hospital systems, limiting individualized, evidence-based care. Here we report OrthoPilot, a clinical artificial intelligence system powered by a large language model that integrates hospital data streams with authoritative external knowledge for continuous musculoskeletal management. OrthoPilot autonomously retrieves real-time imaging, laboratory, pathology and order data and converts evolving patient states into evidence-based decisions from admission diagnosis to rehabilitation planning. We established a specialist-validated benchmark from real-world electronic health records spanning 1,000 disease codes. In a reader study across the complete care pathway, OrthoPilot was compared with 81 orthopaedic physicians and surpassed experts with 25 years of experience in diagnostic reasoning, clinical decision-making and management planning. It also outperformed all evaluated intelligent systems across 60 external clinical centres. In a prospective study of 1,870 complex cases, OrthoPilot increased full-chain management success by 10.6%. During an 8-month randomised deployment involving 8,240 inpatients, it increased cumulative cases per bed by 9.7% and improved patient-reported access to health information. These results move clinical AI from predicting isolated events toward executing longitudinal management across complete musculoskeletal care pathways.
Chinese Translation
肌肉骨骼疾病是全球致残的主要原因之一，并且对康复的需求最大。由于恢复、重塑和退化通常在数月到数年内展开，肌肉骨骼护理需要纵向管理，反复整合不断变化的患者证据、外部医学知识和特定阶段的功能目标。在常规实践中，这些证据在就诊、科室和医院系统之间是碎片化的，限制了个性化的基于证据的护理。在此，我们报告了OrthoPilot，一个由大型语言模型驱动的临床人工智能系统，它将医院数据流与权威的外部知识相结合，以实现持续的肌肉骨骼管理。OrthoPilot自主检索实时影像、实验室、病理和订单数据，并将不断变化的患者状态转化为从入院诊断到康复规划的基于证据的决策。我们从涵盖1,000种疾病编码的真实世界电子健康记录中建立了一个经过专家验证的基准。在整个护理路径的读者研究中，OrthoPilot与81名骨科医生进行了比较，超越了拥有25年经验的专家在诊断推理、临床决策和管理规划方面的表现。它在60个外部临床中心的所有评估智能系统中也表现优于其他系统。在对1,870个复杂病例的前瞻性研究中，OrthoPilot将全链管理的成功率提高了10.6%。在涉及8,240名住院患者的8个月随机部署中，它将每床累计病例数提高了9.7%，并改善了患者对健康信息的获取。这些结果使临床人工智能从预测孤立事件转向在完整的肌肉骨骼护理路径中执行纵向管理。
cs.AI / 31 / 2607.12588
Vertical Standardisation for High-Risk AI Systems under the EU AI Act: A Domain-Specific Framework for Algorithmic Hiring
欧盟人工智能法下高风险人工智能系统的垂直标准化：针对算法招聘的领域特定框架
Abstract
According to the recent European legislation, high-risk AI systems will have to adapt in order to comply with requirements related to specific areas, like risk management, data quality and governance, logging and traceability, technical documentation, transparency, human oversight, and accuracy, as outlined in the European Artificial Intelligence (AI) Act. As the standardisation process for AI is expected to remain iterative and, so far, there are no European standards on AI fully covering the challenges of algorithmic hiring, we propose specific standardisation-oriented recommendations related to the relevant AI areas specified by the European Commission. For each of these areas, we set the context by describing the requirements that AI systems in high-risk domains, and especially in recruitment, should fulfil, as well as the activities that should be carried out to ensure their appropriate use and desired performance, in line with the requirements deriving from the AI Act. Unlike existing horizontal approaches to AI governance and standardisation, this paper contributes a vertical, domain-specific framework for algorithmic hiring, and especially ranking-based recruitment systems, by mapping the requirements of the AI Act to concrete standardisation recommendations, focusing on lifecycle discrimination risks, fairness-aware data governance, explainability, human oversight, and post-deployment monitoring in recruitment systems. Even though our recommendations were informed by the outcomes of the European project FINDHR, they are not tied to the project's technical artefacts and could be implemented using alternative methods, tools, or governance mechanisms.
Chinese Translation
根据最近的欧洲立法，高风险人工智能系统必须进行调整，以符合与特定领域相关的要求，如风险管理、数据质量与治理、日志记录与可追溯性、技术文档、透明度、人类监督和准确性，这些要求在《欧洲人工智能法》中有所概述。由于人工智能的标准化过程预计将保持迭代，并且迄今为止尚无完全覆盖算法招聘挑战的欧洲人工智能标准，我们提出了与欧洲委员会指定的相关人工智能领域相关的具体标准化导向建议。对于这些领域中的每一个，我们通过描述高风险领域，尤其是招聘中人工智能系统应满足的要求，以及应开展的活动，以确保其适当使用和期望性能，来设定背景，这些活动与《人工智能法》所衍生的要求相一致。与现有的人工智能治理和标准化的横向方法不同，本文贡献了一个针对算法招聘，特别是基于排名的招聘系统的垂直领域特定框架，通过将《人工智能法》的要求映射到具体的标准化建议，重点关注招聘系统中的生命周期歧视风险、公平意识的数据治理、可解释性、人类监督和部署后的监测。尽管我们的建议受到欧洲项目FINDHR成果的启发，但并不依赖于该项目的技术成果，可以通过替代方法、工具或治理机制实施。
cs.AI / 32 / 2607.12619
Agentic Service-Oriented Computing: A Manifesto for the Next Frontier of Service-Oriented Computing
代理服务导向计算：服务导向计算下一个前沿的宣言
Abstract
The rapid emergence of LLM-powered autonomous and semi-autonomous agents is reshaping software systems from static, request-response components into goal-directed, adaptive, and tool-using computational actors. As these agents move from isolated cognitive prototypes into complex distributed workflows, they confront challenges that the Service-Oriented Computing community has studied for more than two decades: composition, interoperability, quality of service, lifecycle management, governance, security, and trust. Yet much of today's agentic AI ecosystem is developing these foundations ad hoc, without the engineering rigour required for dependable enterprise and societal deployment. This paper introduces Agentic Service-Oriented Computing (ASOC) as a new research and practice area concerned with engineering agents as services, orchestrating services through autonomous and semi-autonomous agents, and governing ecosystems of agents and services under constraints of trust, cybersecurity, compliance, performance, and accountability. We articulate six foundational principles of ASOC (harness-ability, composability, lifecycle engineering, trustworthiness by design, goal-driven orchestration, and observability/accountability) and organise a five-dimensional research agenda spanning: (i) agentic services foundations and lifecycle engineering; (ii) composition, orchestration, and interoperability; (iii) governance, observability, and accountability; (iv) security, trust, and risk management; and (v) evaluation, certification, and Agentic QoS. We argue that the Services Computing community is especially well positioned to provide the conceptual and engineering spine for this emerging field, transforming agentic AI from fragmented demonstrations into dependable, service-based systems worthy of human and organisational trust.
Chinese Translation
大型语言模型（LLM）驱动的自主和半自主代理的快速出现正在将软件系统从静态的请求-响应组件转变为目标导向的、自适应的、使用工具的计算行为者。当这些代理从孤立的认知原型转向复杂的分布式工作流时，它们面临着服务导向计算社区研究了超过二十年的挑战：组合性、互操作性、服务质量、生命周期管理、治理、安全性和信任。然而，今天的代理人工智能生态系统在很大程度上是临时开发这些基础，没有为可靠的企业和社会部署所需的工程严谨性。本文介绍了代理服务导向计算（Agentic Service-Oriented Computing，ASOC）作为一个新的研究和实践领域，关注于将代理作为服务进行工程设计，通过自主和半自主代理编排服务，以及在信任、网络安全、合规性、性能和问责制的约束下治理代理和服务的生态系统。我们阐明了ASOC的六个基础原则（可利用性、可组合性、生命周期工程、设计中的可信性、目标驱动的编排和可观察性/问责制），并组织了一个跨越五个维度的研究议程： (i) 代理服务的基础和生命周期工程； (ii) 组合、编排和互操作性； (iii) 治理、可观察性和问责制； (iv) 安全、信任和风险管理； (v) 评估、认证和代理服务质量（Agentic QoS）。我们认为，服务计算社区特别适合为这一新兴领域提供概念和工程支撑，将代理人工智能从零散的演示转变为值得人类和组织信任的可靠服务型系统。
cs.AI / 33 / 2607.12634
Atomic Units of X: The Compression Layer of Intelligence
X的原子单位：智能的压缩层
Abstract
This paper proposes a theoretical framework for understanding intelligence as a process of atomic compression and compositional reuse. We argue that cognitive, biological, computational, and organizational systems achieve scalable intelligence by decomposing complex phenomena into reusable atomic units that can be recombined into higher-order structures. Drawing on evidence from cognitive science, information theory, evolutionary biology, software engineering, medicine, legal reasoning, education, music, and artificial intelligence, the paper develops the concept of atomic units as fundamental compression layers that support efficiency, transfer, interpretability, and evolvability. The central contribution is the Compression Calculus, a formal framework for comparing surface-level representations with atomic representations and for describing how compression gains compound across abstraction layers. We introduce the Compounding Cascade thesis, according to which each additional layer of abstraction multiplicatively increases representational efficiency rather than merely adding incremental savings. The paper further argues that contemporary AI systems often operate at suboptimal levels of representation, relying on token-level processing or document-level retrieval rather than stable, concept-level atomic structures. In this view, large language models are best understood not as complete knowledge architectures, but as dynamic fusion engines capable of navigating, sequencing, and recombining atomic units. The framework provides a foundation for designing self-evolving knowledge systems that can discover, refine, and compose new primitives over time. By reframing intelligence as compression through compositional abstraction, the paper offers a unifying perspective on expertise, knowledge representation, explainable AI, and the future architecture of adaptive intelligent systems.
Chinese Translation
本文提出了一个理论框架，用于理解智能作为一种原子压缩和组合重用的过程。我们认为，认知、生物、计算和组织系统通过将复杂现象分解为可重用的原子单位，从而实现可扩展的智能，这些原子单位可以重新组合成更高阶的结构。本文借鉴了来自认知科学、信息理论、进化生物学、软件工程、医学、法律推理、教育、音乐和人工智能的证据，发展了原子单位的概念，作为支持效率、传递性、可解释性和可演化性的基本压缩层。核心贡献是压缩演算（Compression Calculus），这是一个用于比较表层表示与原子表示的形式框架，并描述压缩增益如何在抽象层次上复合。我们引入了复合级联论（Compounding Cascade thesis），根据该理论，每增加一层抽象就会乘法性地提高表征效率，而不仅仅是增加增量节省。本文进一步论证，当代人工智能系统通常在次优的表征水平上运行，依赖于标记级处理或文档级检索，而不是稳定的概念级原子结构。从这个角度来看，大型语言模型最好被理解为动态融合引擎，能够导航、排序和重新组合原子单位，而不是完整的知识架构。该框架为设计能够随时间发现、完善和组合新原语的自我演化知识系统提供了基础。通过将智能重新框定为通过组合抽象进行的压缩，本文为专业知识、知识表征、可解释人工智能以及自适应智能系统的未来架构提供了统一的视角。
cs.AI / 34 / 2607.12640
A Learning-Rate-Gated Failure of GRPO in a Small Language and Vision-Language Model Web Agent: A Controlled Null and Its Mechanism
小型语言和视觉-语言模型网络代理中GRPO的学习率门控失效：一个受控的无效及其机制
Abstract
Reinforcement learning with verifiable rewards, and Group Relative Policy Optimization (GRPO) in particular, is now run routinely on a supervised checkpoint in the hope of producing a stronger agent. We ask whether it adds skill to a small language and vision-language model web agent at the 4B to 8B scale, or whether it mostly reshapes behavior the supervised model already has. Across a control grid of 18 runs that varies learning rate, KL weight, seed, initialization, and clipping, no configuration credibly improves the success rate of a strong supervised baseline on tasks the agent has largely mastered. On the text track, moderate to high learning rates make it credibly worse. The null holds under paired testing, 25 evaluation seeds, 6 training seeds, changes to the recipe, both text and Set-of-Marks screenshot observations, and scaling the backbone to 8B; the credible harm is a text-track finding and is only nominal under Set-of-Marks. To show that the null reflects the setting and not a broken pipeline, we run the identical harness, reward, and recipe on tasks whose reward is reachable by sampling, and there the success rate rises by 22 points with a paired interval that excludes zero. GRPO therefore helps only when there is headroom to climb, meaning the sampled policy already succeeds more often than the greedy one. We then explain the failure. A middle learning rate degrades the agent and a high one collapses it, and the two regimes form a double dissociation: grafting localizes the degrade regime to the attention and MLP blocks, while the collapse regime cannot be traced to any single group, and the embedding change that dominates the weight movement is causally inert. At 4B, effective rank in the late layers tracks capability in both directions; at 8B the two come apart. This coupling is specific to the smaller model, so we report it as scale-dependent.
Chinese Translation
可验证奖励的强化学习，特别是群体相对策略优化（GRPO），现在常常在监督检查点上运行，以期产生更强的代理。我们探讨它是否为一个规模在4B到8B之间的小型语言和视觉-语言模型网络代理增加了技能，或者它是否主要重塑了监督模型已经具备的行为。在一个控制网格中进行了18次实验，变量包括学习率、KL权重、随机种子、初始化和剪切，没有任何配置能够显著提高在代理已经基本掌握的任务上强监督基线的成功率。在文本轨道上，中等到高的学习率使得成功率显著下降。该无效在配对测试、25个评估种子、6个训练种子、配方变化、文本和标记集（Set-of-Marks）截图观察以及将主干扩展到8B的情况下均成立；可信的损害是文本轨道的发现，在标记集下仅为名义上的。为了表明该无效反映了设置而不是管道故障，我们在通过采样可达的任务上运行相同的测试框架、奖励和配方，在这些任务中，成功率提高了22个百分点，且配对区间排除了零。因此，GRPO仅在存在上升空间时有效，这意味着采样策略的成功率已经高于贪婪策略。接着我们解释了失效原因。中等学习率使代理性能下降，高学习率则导致其崩溃，这两种状态形成了双重解离：局部化的降级状态集中在注意力和多层感知器（MLP）模块，而崩溃状态无法追溯到任何单一组，主导权重变化的嵌入变化因果上是惰性的。在4B模型中，后期层的有效秩与能力在两个方向上相关；而在8B模型中，两者则分离。这种耦合特性特定于较小的模型，因此我们将其报告为规模依赖性。
cs.AI / 35 / 2607.12662
Internet of Agentic Things: Networked AI Agents for Closed-Loop IoT Orchestration
能动物联网：用于闭环物联网编排的网络化人工智能代理
Abstract
The paper introduces the Internet of Agentic Things (IoAT), an architectural framework that integrates agentic AI, IoT, cyber-physical systems, Physical AI, edge computing, and digital twins into a unified closed-loop orchestration framework. The proposed architecture consists of cloud, edge/fog, and physical IoT layers connected through autonomous AI agents that perceive, reason, coordinate, and actuate across distributed cyber-physical environments. The paper formalizes IoAT as a coupled workflow-control problem with nested strategic and tactical decision making using a hylomorphic dynamic programming framework that links agentic planning with physical execution. Smart-building orchestration is presented as a representative use case, and key research challenges related to safety, security, governance, resilience, and trustworthy deployment are discussed.
Chinese Translation
本文介绍了能动物联网（Internet of Agentic Things, IoAT），这是一个将能动人工智能、物联网、网络物理系统、物理人工智能、边缘计算和数字双胞胎整合为统一闭环编排框架的架构框架。所提出的架构由云、边缘/雾和物理物联网层组成，这些层通过自主人工智能代理相互连接，代理能够在分布式网络物理环境中感知、推理、协调和执行。本文将IoAT形式化为一个耦合的工作流控制问题，采用嵌套的战略和战术决策制定，使用一种将能动规划与物理执行相连接的形态动力学编程框架。智能建筑编排被作为一个代表性用例呈现，并讨论了与安全性、保密性、治理、韧性和可信部署相关的关键研究挑战。
cs.AI / 36 / 2607.12711
MaxSAT-Based Feedback for Guiding Vision-Language Models in Sudoku
基于MaxSAT的反馈指导视觉-语言模型在数独中的应用
Abstract
Vision--Language Models (VLMs) have recently demonstrated promising performance on structured visual reasoning tasks, including grid-based puzzles. However, despite strong perceptual capabilities, these models lack explicit mechanisms for enforcing logical consistency and frequently generate assignments that violate underlying constraints. In this paper, we propose a neuro-symbolic approach that integrates formal constraint reasoning into the VLM solving process via a Maximum Satisfiability (MaxSAT) oracle. Rather than computing solutions directly, the symbolic component acts as a consistency validator and refinement engine. Candidate placements generated by the VLM are encoded as soft clauses in a partial MaxSAT formulation, while Sudoku constraints remain hard clauses. When inconsistencies arise, the MaxSAT solver identifies a largest mutually consistent subset of assignments, which is then translated into structured textual and visual feedback to guide subsequent refinements. We evaluate our approach on a Sudoku dataset across multiple open-source and closed-access VLMs. Results show that MaxSAT-based feedback improves logical consistency and increases the number of solved instances, particularly in full-board refinement mode. These findings demonstrate that symbolic optimisation can enhance the reliability of vision-language reasoning.
Chinese Translation
视觉-语言模型（VLMs）最近在结构化视觉推理任务中表现出良好的性能，包括基于网格的谜题。然而，尽管这些模型具有强大的感知能力，但它们缺乏强制逻辑一致性的明确机制，常常生成违反基本约束的分配。在本文中，我们提出了一种神经符号方法，通过最大可满足性（MaxSAT）oracle将形式约束推理集成到VLM求解过程中。该符号组件并非直接计算解决方案，而是充当一致性验证器和精炼引擎。VLM生成的候选放置被编码为部分MaxSAT公式中的软子句，而数独约束则保持为硬子句。当出现不一致时，MaxSAT求解器识别出一个最大的相互一致的分配子集，然后将其转换为结构化的文本和视觉反馈，以指导后续的精炼。我们在多个开源和闭源的VLM上评估了我们的方法，使用了一个数独数据集。结果表明，基于MaxSAT的反馈提高了逻辑一致性，并增加了解决实例的数量，特别是在全盘精炼模式下。这些发现表明，符号优化可以增强视觉-语言推理的可靠性。
cs.AI / 37 / 2607.12733
LLMs Can See the Smoke but not the Fire: Evaluating Abductive Reasoning with Elenchos
大型语言模型能够看到烟雾却看不到火焰：使用Elenchos评估溯因推理
Abstract
Large language models (LLMs) excel at pattern recognition and text generation, but their capacity for abductive inference - inferring latent hypotheses that explain observed behavior - remains poorly understood. Here, we introduce Elenchos (named after the Socratic method of cross-examination), a generative evaluation framework that measures abductive reasoning as a structural inverse problem. Given a reference formal system, such as the lambda-calculus, and a potentially mutated counterpart, agents must determine whether a mutation has occurred and infer the rule modifications responsible for the resulting behavioral differences. Evaluating frontier and mid-tier LLMs reveals a consistent detection-attribution dissociation: models often recognize that a system has been altered but struggle to identify the latent mutations causing the observed discrepancies. Performance degrades substantially under interacting mutations, where models frequently recover only a subset of the underlying mutations. Preliminary evidence also suggests diminishing returns from increased inference-time reasoning, with only modest improvements under larger reasoning budgets, though this finding requires further validation.
Chinese Translation
大型语言模型（LLMs）在模式识别和文本生成方面表现出色，但它们在溯因推理方面的能力——推断解释观察到的行为的潜在假设——仍然不够清楚。在此，我们介绍Elenchos（以苏格拉底的交叉审问法命名），这是一个生成性评估框架，用于将溯因推理视为一种结构性逆问题。在给定一个参考形式系统（如λ演算）和一个可能变异的对应系统的情况下，智能体必须判断是否发生了变异，并推断导致行为差异的规则修改。对前沿和中层LLMs的评估揭示了一种一致的检测-归因分离现象：模型通常能够识别出系统已被更改，但在识别导致观察到的差异的潜在变异方面却存在困难。在相互作用的变异下，模型的表现显著下降，通常只能恢复潜在变异的一个子集。初步证据还表明，推理时间的增加带来的收益递减，只有在更大的推理预算下才有适度的改善，尽管这一发现需要进一步验证。
cs.AI / 38 / 2607.12747
Tracing Agentic Failure from the Flow of Success
从成功的流动中追踪能动性失败
Abstract
Failure attribution for LLM-based agentic systems, i.e., identifying which steps in a failure trajectory caused the task to fail, is critical for debugging and improving these systems. Existing approaches either rely on prompting-based pipelines, which are computationally expensive, or require post-training on failure trajectories with step-level error annotations, which are costly to collect and difficult to scale. We argue that a practical failure attribution model should be lightweight and trainable without step-level supervision on failure data. To this end, we address unsupervised failure attribution, i.e., training exclusively on successful trajectories and identifying error steps at inference time given a failure trajectory. We propose OAT, which casts this problem as one-class learning with neural controlled differential equations, modeling the dynamical pattern of successful trajectories in latent space. At inference time, each step in a failure trajectory is assigned an anomaly score based on its deviation from the dynamics learned on successful trajectories, which is then used to form a set of error steps. With training on only 100 successful trajectories, experiments show that OAT is 200--5000 $\times$ faster than prompting-based baselines, and, at the same time, consistently outperforms them in both in-domain and out-of-distribution datasets with +20% and +7% F1 scores, respectively, demonstrating that OAT is a promising and efficient direction for diagnosing agentic system failures.
Chinese Translation
对于基于大型语言模型（LLM）的能动系统，失败归因，即识别在失败轨迹中哪些步骤导致了任务失败，对于调试和改进这些系统至关重要。现有的方法要么依赖于基于提示的管道，这在计算上成本高昂，要么需要在失败轨迹上进行后训练，并附有逐步错误注释，这在收集上成本高且难以扩展。我们认为，一个实用的失败归因模型应该是轻量级的，并且无需在失败数据上进行逐步监督进行训练。为此，我们提出无监督失败归因，即仅在成功轨迹上进行训练，并在推理时根据给定的失败轨迹识别错误步骤。我们提出了OAT，将此问题视为一种单类学习，利用神经控制微分方程建模成功轨迹在潜在空间中的动态模式。在推理时，失败轨迹中的每一步都根据其与成功轨迹上学习到的动态的偏差被分配一个异常分数，然后用于形成一组错误步骤。在仅对100条成功轨迹进行训练的情况下，实验表明OAT的速度比基于提示的基线快200到5000倍，同时在领域内和分布外数据集上均表现出色，F1分数分别提高了20%和7%，这表明OAT是诊断能动系统失败的一个有前景且高效的方向。
cs.AI / 39 / 2607.12767
Accuracy and Normalized Accuracy under Length Bias: Analysis, Guidelines, and a Bayesian Alternative
长度偏差下的准确性与归一化准确性：分析、指南及贝叶斯替代方案
Abstract
Multiple-choice benchmarks that rank candidate completions by conditional log-probability suffer from a length bias: because log-probabilities sum over tokens, longer answers tend to be penalized relative to shorter ones in practice. A common mitigation is to normalize scores by completion length, but we show empirically that this heuristic frequently over-corrects, introducing a bias toward longer answers instead. We first analyze these scoring rules, characterizing when standard and length-normalized accuracy are appropriate and how their length biases depend on the distribution of completion lengths. Motivated by this analysis, we introduce \emph{Bayesian accuracy}, a scoring rule that computes the posterior probability of each candidate under an explicit prior over answer length, thereby removing linear length effects. Bayesian accuracy is a drop-in replacement for likelihood-based multiple-choice evaluation, requires no additional forward passes, and consistently exhibits lower empirical length bias than both standard and length-normalized accuracy across benchmarks and few-shot settings.
Chinese Translation
基于条件对数概率对候选完成进行排名的多项选择基准存在长度偏差：由于对数概率是对标记求和的，较长的答案在实践中往往相对于较短的答案受到惩罚。一种常见的缓解方法是通过完成长度对分数进行归一化，但我们通过实证研究表明，这一启发式方法常常过度修正，反而引入了对较长答案的偏向。我们首先分析这些评分规则，描述标准准确性和长度归一化准确性何时适用，以及它们的长度偏差如何依赖于完成长度的分布。基于这一分析，我们引入了 extit{贝叶斯准确性}（Bayesian accuracy），这是一种评分规则，通过对答案长度的显式先验计算每个候选者的后验概率，从而消除线性长度效应。贝叶斯准确性可以替代基于似然的多项选择评估，无需额外的前向传递，并且在各基准和少量样本设置中始终表现出比标准准确性和长度归一化准确性更低的实证长度偏差。
cs.AI / 40 / 2607.12787
Do We Really Need Multimodal Emotion Language Models Larger Than 1B Parameters?
我们真的需要超过10亿参数的多模态情感语言模型吗？
Abstract
Recent advances in multimodal large language models (MLLMs) have significantly improved the performance of multimodal emotion recognition (MER) and enabled interpretable description generation by jointly modeling video, audio, and language, etc. However, these performance improvements are often accompanied by an increase in model parameter size (e.g, at least 7B), which simultaneously incurs high computational costs and reduces inference efficiency, thereby hindering real-time deployment on resource-constrained platforms such as robots and mobile devices. This raises a fundamental question: do we really need the multimodal MER model larger than 1B parameters for high-quality MER? In this paper, we challenge the assumption that larger models are inherently necessary and proposes a lightweight MER framework (called Light-MER), which achieves better and faster multimodal sentiment understanding and recognition through knowledge distillation. It can transfer knowledge from a strong, large-scale teacher model to a lightweight sub-billion-parameter student model, aiming to preserve rich multimodal emotion reasoning and recognition while substantially improving deployment efficiency. Specifically, we introduce two new optimization strategies to enhance knowledge transfer: (1) a new optimal transport loss that combines Sliced Wasserstein Distance with hidden-state alignment, and (2) a new multi-reward optimization strategy based on GRPO that balances MER performance and efficiency, aimed at further enhancing the learning capabilities of student models. Extensive experiments on nine benchmark datasets demonstrate that Light-MER achieves state-of-the-art performance while significantly improving inference efficiency. This highlights the strong potential of small multimodal emotion language models for future research. Code is available at https://github.com/GAIR-Lab/Light-MER.
Chinese Translation
近年来，多模态大型语言模型（MLLMs）的进展显著提升了多模态情感识别（MER）的性能，并通过联合建模视频、音频和语言等实现了解释性描述生成。然而，这些性能提升往往伴随着模型参数规模的增加（例如，至少为70亿），这同时带来了高计算成本并降低了推理效率，从而阻碍了在资源受限的平台（如机器人和移动设备）上的实时部署。这引发了一个根本性的问题：我们真的需要超过10亿参数的多模态MER模型来实现高质量的MER吗？在本文中，我们质疑了更大模型固有必要性的假设，并提出了一种轻量级的MER框架（称为Light-MER），该框架通过知识蒸馏实现了更好且更快的多模态情感理解与识别。它能够将知识从强大且大规模的教师模型转移到轻量级的子亿参数学生模型，旨在保留丰富的多模态情感推理与识别，同时显著提高部署效率。具体而言，我们引入了两种新的优化策略以增强知识转移：（1）一种新的最优传输损失，结合了切片Wasserstein距离与隐藏状态对齐；（2）一种基于GRPO的新多奖励优化策略，平衡MER性能与效率，旨在进一步增强学生模型的学习能力。在九个基准数据集上的广泛实验表明，Light-MER实现了最先进的性能，同时显著提高了推理效率。这突显了小型多模态情感语言模型在未来研究中的巨大潜力。代码可在 https://github.com/GAIR-Lab/Light-MER 获取。
cs.AI / 41 / 2607.12790
Who Grades the Grader? Co-Evolving Evaluation Metrics and Skills for Self-Improving LLM Agents
谁来评估评估者？自我改进大语言模型代理的共同演化评估指标与技能
Abstract
Self-evolving agent systems improve by creating, revising, and retiring their own skills, but every such loop rests on a hidden assumption: a reliable evaluation metric already exists. In many real applications it does not. We make three claims. First, metrics can be \emph{evolved}: our metric loop searches compositions of small drawback detectors under a full evolutionary lifecycle, trained to agree with a ten-item anchored reference set, regularized by consensus over unlabeled outputs, and audited against a held-out anchor it never reads, yielding a transparent, inspectable metric rather than an opaque judge. Second, since no metric exists to beat, the yardstick is recovering what an accurate metric would have enabled, and \emph{Double Ratchet}, our co-evolution of the metric with a lifecycle-managed skill loop, does so: across code generation (MBPP+), enterprise text-to-SQL (Spider~2.0-Snow), and reference-free report generation, it retains 88--110\% of the held-out lift achieved by the same skill loop driven by ground truth or the best available rubric. Third, safety comes from anchor discipline plus outer audits: removing anchor guards collapses the metric into a vacuous detector while removing the lifecycle does not; and when evolved skills gamed the report rubric, an independent judge caught it, one detector repaired it, and a task-aware judge then preferred the evolved outputs over the pre-evolution baseline in 77\% of decided pairs. We argue this failure-expecting architecture is the right default wherever no reliable automatic verifier exists.
Chinese Translation
自我演化代理系统通过创建、修订和淘汰自身技能来实现改进，但每个这样的循环都基于一个隐含假设：已经存在一个可靠的评估指标。然而，在许多实际应用中，这种指标并不存在。我们提出三项主张。首先，指标可以被 extit{演化}：我们的指标循环在完整的演化生命周期下搜索小型缺陷检测器的组合，训练以与一个包含十项锚定参考集的指标达成一致，通过对未标记输出的共识进行正则化，并在一个从未读取的保留锚点上进行审计，从而产生一个透明、可检查的指标，而不是一个不透明的评判者。其次，由于不存在可以超越的指标，衡量标准是恢复一个准确指标所能实现的目标，而我们的 extit{双重棘轮}（Double Ratchet）则通过与生命周期管理的技能循环共同演化的方式实现了这一点：在代码生成（MBPP+）、企业文本到SQL（Spider~2.0-Snow）和无参考报告生成中，它保留了由真实数据或最佳可用评分标准驱动的相同技能循环所实现的88%至110%的保留提升。第三，安全性来自于锚定纪律加上外部审计：去除锚定保护会使指标崩溃为一个空洞的检测器，而去除生命周期则不会；当演化的技能利用了报告评分标准时，一个独立的评审者发现了这一点，一个检测器修复了它，而一个任务感知的评审者在77%的决定对中更倾向于演化后的输出而非演化前的基线。我们认为这种预期失败的架构是在没有可靠自动验证器存在的情况下的正确默认选择。
cs.AI / 42 / 2607.12815
Visual Access Boundaries in Vision-Language Model Reasoning
视觉-语言模型推理中的视觉访问边界
Abstract
Chain-of-Thought (CoT) prompting is widely used as a test-time scaling strategy for Vision-Language Models (VLMs), but it remains unclear what is extended when VLMs generate longer reasoning traces. We ask whether CoT requires continued access to image tokens, or whether it mainly operates over visual information already made available earlier in the forward pass. We introduce Visual Access Sweep, a causal intervention that masks attention from generated-token queries to image-token keys along layer depth and generation time, and define the Visual Access Boundary (VAB) as the minimal access region that preserves task accuracy. Across six model configurations from Qwen2.5-VL and InternVL3, both no-CoT direct answering and CoT prompting exhibit finite VABs. In Qwen2.5-VL-32B and InternVL3 at 14B and 38B scales, when CoT is evaluated against the no-CoT full-access target, its VAB layer differs from the no-CoT boundary by at most two layers, despite substantially longer generations. This suggests that CoT does not primarily improve performance by prolonging direct image-token access throughout the reasoning trace, but by extending language-side computation over image-derived hidden-state information. We further show that CoT gains are constrained by perceptual readout. CoT helps when the queried visual attribute can be reliably read out by the model, but not when that readout is unreliable. A symbolic-attribute oracle shows that CoT can improve counting once ground-truth attributes are supplied as text, while a single-object probe-vs-decode check shows that hard attributes can be linearly recoverable from hidden states yet difficult for the model itself to output. Together, these analyses place the bottleneck at readout rather than counting.
Chinese Translation
链式思维（Chain-of-Thought, CoT）提示被广泛用作视觉-语言模型（Vision-Language Models, VLMs）的测试时扩展策略，但在VLM生成更长推理轨迹时，尚不清楚其扩展的内容。我们探讨CoT是否需要持续访问图像标记，或主要依赖于在前向传播过程中早期提供的视觉信息。我们引入了视觉访问扫描（Visual Access Sweep），这是一种因果干预，遮蔽从生成标记查询到图像标记键的注意力，涉及层深度和生成时间，并定义视觉访问边界（Visual Access Boundary, VAB）为保持任务准确性的最小访问区域。在来自Qwen2.5-VL和InternVL3的六种模型配置中，无CoT直接回答和CoT提示均表现出有限的VAB。在Qwen2.5-VL-32B和InternVL3的14B和38B规模下，当CoT与无CoT的完全访问目标进行评估时，其VAB层与无CoT边界最多相差两层，尽管生成过程明显更长。这表明，CoT并不是通过延长整个推理轨迹中的直接图像标记访问来主要提高性能，而是通过扩展语言侧计算对图像衍生的隐藏状态信息的利用。我们进一步表明，CoT的增益受到感知读取的限制。当被查询的视觉属性可以被模型可靠读取时，CoT有助于提升性能，但当该读取不可靠时则无效。一个符号属性神谕显示，一旦提供真实属性作为文本，CoT可以改善计数，而单一对象探测与解码检查表明，硬属性可以从隐藏状态中线性恢复，但模型自身输出这些属性却较为困难。综合这些分析，瓶颈在于读取而非计数。
cs.AI / 43 / 2607.12823
Human-AI Agent Interaction as a Neuroplastic Training Environment
人机智能体交互作为神经可塑性训练环境
Abstract
Interaction with AI agents has become one of the most frequent activities of everyday digital life. Whether conversing with an assistant, working with a coding copilot, or generating images, the interaction follows a common iterative loop: a request is issued, a result returned, appraised, and the request revised. We observe that this loop is a high-frequency stream of contact events -- moments at which a result meets a person and a conditioned response may fire before deliberate appraisal -- making everyday agent interaction an unrecognised neuroplastic training environment. When a result disappoints, reactive patterns of impatience, perfectionism, frustration, and self-criticism are repeatedly evoked, and under activity-dependent synaptic plasticity each uninterrupted cycle deepens the underlying pathway through long-term potentiation. Ordinary agent use may thus quietly strengthen the very patterns it provokes. We propose that the same training environment can be engaged to the opposite effect. Treating conditioned reactive patterns as physical neurone paths -- activated through a pre-cognitive feeling tone that opens a brief regulatory gap -- we develop a framework in which, at that gap, in place of the reactive re-prompt, a person performs behind-the-scenes observation: watching the neural process operate so the cascade does not complete and long-term depression weakens the path rather than potentiation strengthening it. We characterise this practice through three layers of observation and two modes of application: a user-guided mode requiring no change to existing tools, and an agent-assisted mode in which an ordinary agent is lightly configured to support observation at the gap. We illustrate the framework through generative image prompting, showing how a single frustrating session is behaviourally nearly identical whether or not it is observed, yet neurologically opposite.
Chinese Translation
与人工智能（AI）智能体的交互已成为日常数字生活中最频繁的活动之一。无论是与助手对话、与编码副驾驶合作，还是生成图像，这种交互遵循一个共同的迭代循环：发出请求、返回结果、进行评估并修订请求。我们观察到，这个循环是一个高频率的接触事件流——结果与个体相遇的时刻，以及在深思熟虑的评估之前可能触发的条件反应——使得日常智能体交互成为一个未被认识的神经可塑性训练环境。当结果令人失望时，急躁、完美主义、挫折感和自我批评的反应模式会被反复激发，而在活动依赖的突触可塑性下，每一个不间断的循环通过长期增强加深了潜在路径。因此，普通的智能体使用可能悄然加强其所引发的模式。我们提出可以利用同样的训练环境来达到相反的效果。将条件反应模式视为物理神经元路径——通过一种开启短暂调节间隙的前认知情感基调激活——我们开发了一个框架，在这个间隙中，个体不再进行反应性重新提示，而是进行幕后观察：观察神经过程的运作，以便级联过程不完成，从而长期抑制削弱路径，而不是增强。我们通过三个观察层次和两种应用模式来表征这一实践：一种用户引导模式，不需要改变现有工具；另一种智能体辅助模式，其中普通智能体被轻微配置以支持在间隙处的观察。我们通过生成图像提示来说明该框架，展示了无论是否进行观察，单次令人沮丧的会话在行为上几乎是相同的，但在神经上却是相反的。
cs.AI / 44 / 2607.12826
Solution of the Hempel's statistical ambiguity problem and Causal AI
亨佩尔统计模糊性问题的解决方案与因果人工智能
Abstract
This paper addresses Carl Hempel's longstanding problem of statistical ambiguity in inductive-statistical inference, in which contradictory predictions are derived from statistical laws. To avoid such predictions, Carl Hempel proposed the Requirement of Maximal Specificity (RMS) for the statistical laws used in the inference. An analysis of the RMS refinements made by Wesley Salmon, Alberto Coffa, and James Fetzer led to the following definition of maximally specific statistical laws: "the lawlike premises of an adequate explanation must specify all and only those properties whose presence or absence made a difference to the occurrence of its explanandum-phenomenon." However, there was no proof of a solution to the statistical ambiguity problem based on this definition. We use Nancy Cartwright's definition of causes that raise probabilities across background contexts, and then introduce the concept of Causal Rules. Then we define a special semantic probabilistic inference procedure that incrementally refines these causal rules by incorporating all statistically relevant information. This procedure yields Maximally Specific Causal Relationships (MSCRs), for which we prove (Theorem 1) that predictions derived from them are consistent. This resolves the statistical ambiguity problem. The semantic probabilistic inference procedure provides a probabilistic causal learning system, which may be used in such new areas as Causal AI and Causal Machine Learning. They fundamentally explore causal inference as a tool for understanding cause-and-effect relationships within complex systems. Properties similar to RMS remain under discussion. Several notions related to RMS are considered: invariant feature learning, invariant causal prediction, and spurious association.
Chinese Translation
本文探讨了卡尔·亨佩尔在归纳统计推理中长期存在的统计模糊性问题，该问题源于从统计法则中得出的矛盾预测。为避免此类预测，卡尔·亨佩尔提出了归纳推理中使用的统计法则的最大特异性要求（Requirement of Maximal Specificity, RMS）。对韦斯利·萨尔蒙（Wesley Salmon）、阿尔贝托·科法（Alberto Coffa）和詹姆斯·费策（James Fetzer）所做的RMS细化分析，得出了最大特异性统计法则的以下定义：“充分解释的法则性前提必须具体说明所有且仅有那些其存在或缺失对其解释现象的发生产生影响的属性。”然而，基于该定义并没有对统计模糊性问题的解决方案进行证明。我们采用南希·卡特赖特（Nancy Cartwright）对在背景情境中提高概率的原因的定义，并引入因果规则（Causal Rules）的概念。接着，我们定义了一种特殊的语义概率推理程序，该程序通过整合所有统计相关信息逐步细化这些因果规则。该程序产生最大特异性因果关系（Maximally Specific Causal Relationships, MSCRs），我们证明了（定理1）从中得出的预测是一致的。这解决了统计模糊性问题。该语义概率推理程序提供了一种概率因果学习系统，可用于因果人工智能（Causal AI）和因果机器学习（Causal Machine Learning）等新领域。这些领域从根本上探讨因果推断作为理解复杂系统中因果关系的工具。与RMS类似的属性仍在讨论中。我们考虑了与RMS相关的几个概念：不变特征学习、不变因果预测和虚假关联。
cs.AI / 45 / 2607.12886
A Multi-Agent System for Autonomous, Fine-Tuning-Free Clinical Symptom Detection: Development and Validation Study
一种用于自主、无需微调的临床症状检测的多智能体系统：开发与验证研究
Abstract
Clinical notes contain many of the signs and symptoms that bring patients to care, yet this information rarely reaches structured fields. Existing extraction approaches either rely on context-insensitive rules that generate false positives or on supervised models that require substantial fine-tuning. We present Pythia, a multi-agent system that autonomously writes and optimizes extraction prompts for clinical concepts without manual prompt engineering or fine-tuning. Running on a locally hosted open-weights model, Pythia keeps clinical notes on local infrastructure and selects prompts using development-set sensitivity and specificity. We compared Pythia with a curated lexicon across 72 signs and symptoms from 400 clinical notes representing 387 patients. Development (n=300) and validation (n=100) sets were partitioned independently for each concept. Pythia achieved mean sensitivity of 0.76 and specificity of 0.95, compared with 0.82 and 0.76 for the lexicon, and matched or exceeded the lexicon on both metrics for 20 of 62 directly comparable concepts. For 14 concepts where the lexicon labeled every note positive, Pythia recovered mean specificity of 0.97 by requiring a present-tense, patient-attributed finding rather than any textual mention of a term. Specificity transferred from development to validation with minimal degradation across prevalences, whereas sensitivity transfer weakened below 5% prevalence, reaching a mean gap of 0.25 below 2% prevalence. A BERT classifier fine-tuned per concept on the same development set achieved mean sensitivity of 0.23 and collapsed to zero sensitivity for concepts below roughly 5% prevalence. These findings suggest that autonomous, fine-tuning-free prompt optimization can produce symptom extraction prompts that generalize effectively from development to validation while remaining deployable on local infrastructure.
Chinese Translation
临床笔记包含了许多促使患者就医的体征和症状，但这些信息很少进入结构化字段。现有的提取方法要么依赖于生成假阳性的上下文无关规则，要么依赖于需要大量微调的监督模型。我们提出了Pythia，一个多智能体系统，能够自主编写和优化临床概念的提取提示，而无需手动提示工程或微调。Pythia在本地托管的开放权重模型上运行，保持临床笔记在本地基础设施上，并使用开发集的灵敏度和特异性选择提示。我们将Pythia与一个经过整理的词典进行了比较，涵盖了来自400份临床笔记（代表387名患者）的72个体征和症状。每个概念的开发（n=300）和验证（n=100）集独立划分。Pythia的平均灵敏度为0.76，特异性为0.95，而词典的灵敏度和特异性分别为0.82和0.76，并且在62个直接可比概念中，Pythia在这两个指标上与词典持平或超过了词典。在14个词典标记为所有笔记阳性的概念中，Pythia通过要求存在时态、患者归属的发现，而不是任何术语的文本提及，恢复了平均特异性为0.97。特异性在开发到验证的转移中几乎没有降解，而灵敏度转移在低于5%流行率时减弱，低于2%流行率时达到平均差距0.25。在同一开发集上针对每个概念微调的BERT分类器的平均灵敏度为0.23，并且在流行率低于大约5%时灵敏度降至零。这些发现表明，自主、无需微调的提示优化能够生成有效从开发到验证推广的症状提取提示，同时仍可在本地基础设施上部署。
cs.AI / 46 / 2607.12893
MemOps: Benchmarking Lifecycle Memory Operations in Long-Horizon Conversations
MemOps：长时间对话中生命周期内存操作的基准测试
Abstract
Long-term memory has become a foundational capability for LLM-based agents that accompany users across extended, multi-session interactions. Existing benchmarks, however, evaluate such memory almost exclusively through downstream question answering, scoring only the correctness of a final answer. This black-box formulation conflates the heterogeneous causes of memory failure, such as missing the introduction of a relevant fact, binding an operation to the wrong target, or relying on stale values after a correction. As a result, it can credit correct answers despite their reliance on inconsistent or unsafe memory states. In this paper, we argue that, in dynamic long-horizon interactions, memory is not a static collection of facts but a lifecycle of explicit operations, including remembering, forgetting, updating, reflecting, and their compositions. We introduce MemOps, a benchmark that reformulates conversational memory as a sequence of lifecycle operations and represents each memory event with a structured trace specifying its trigger, target, scope, state transition, and supporting evidence. A controllable generation pipeline embeds these operations into long, task-oriented conversations and produces gold operation traces together with six categories of operation-level probes, evaluated under both adjacent-evidence and long-context settings. Across long-context, retrieval-based, parametric and managed-memory systems, MemOps disentangles failure modes that final-answer accuracy alone conceals, revealing that current systems remain far from uniformly reliable. For instance, session-level retrieval outperforms turn-level retrieval, and long-context models remain notably weak at reconstructing ordered memory-state trajectories. These results move long-term memory evaluation from final-answer scoring toward interpretable, operation-level diagnosis.
Chinese Translation
长期记忆已成为基于大型语言模型（LLM）的智能体在延续的多会话互动中陪伴用户的基础能力。然而，现有的基准测试几乎仅通过下游问答来评估这种记忆，仅评分最终答案的正确性。这种黑箱形式混淆了记忆失败的异质原因，例如遗漏相关事实的引入、将操作绑定到错误目标或在修正后依赖过时值。因此，它可能会对依赖于不一致或不安全内存状态的正确答案给予认可。本文论证，在动态的长时间互动中，记忆并不是静态的事实集合，而是一系列显式操作的生命周期，包括记忆、遗忘、更新、反思及其组合。我们引入了MemOps，一个基准测试，将对话记忆重新表述为一系列生命周期操作，并通过结构化的追踪记录每个记忆事件，指定其触发、目标、范围、状态转变和支持证据。一个可控的生成管道将这些操作嵌入到长时间的任务导向对话中，并生成黄金操作追踪以及六类操作级探针，在相邻证据和长上下文设置下进行评估。在长上下文、基于检索的、参数化和管理内存系统中，MemOps 解开了仅靠最终答案准确性所掩盖的失败模式，揭示当前系统仍远未达到统一可靠。例如，会议级检索优于轮次级检索，而长上下文模型在重构有序记忆状态轨迹方面仍显著薄弱。这些结果将长期记忆评估从最终答案评分转向可解释的操作级诊断。
cs.AI / 47 / 2607.12924
Knowledge- and Gradient-Guided Reinforcement Learning for Parametrized Action Markov Decision Processes
知识与梯度引导的强化学习在参数化动作马尔可夫决策过程中的应用
Abstract
In this paper, we study Reinforcement Learning in Parametrized Action Markov Decision Processes (PAMDP), where each decision consists of a symbolic action and numerical parameters. In such settings Reinforcement Learning algorithms typically determine parameters with one-shot estimators, which makes their training sample inefficient. Though in most PAMDP environments explicit but incomplete knowledge (e.g., rules, safety constraints, or expert heuristics) is available, it is rarely directly used to increase the sample-efficiency of training Reinforcement Learning agents. We step into this gap and propose our novel Neuro-Symbolic Knowledge- and Gradient-Guided Reinforcement Learning (KGRL) algorithm. KGRL uses domain knowledge in a Datalog knowledge base to derive the set of applicable actions and feasible parameters for a given state. This allows it to prune non-applicable actions from the decision-space and constrain the parameter spaces of the remaining actions. We then use a gradient-based parameter refinement loop to estimate the optimal parameters during training and deployment of the agent. By recording activated rules along the trajectory, KGRL additionally provides local procedural explanations on the pruning of actions and constraining of parameters. Overall, KGRL guides the agent's exploration and deployment toward feasible and constraint-aware decisions, while increasing sample efficiency during training. KGRL outperforms state-of-the-art RL baselines for PAMDPs in both, sample efficiency and episodic return.
Chinese Translation
在本文中，我们研究了参数化动作马尔可夫决策过程（PAMDP）中的强化学习，其中每个决策由一个符号动作和数值参数组成。在这种设置下，强化学习算法通常使用一次性估计器来确定参数，这使得它们的训练样本效率低下。尽管在大多数PAMDP环境中，通常可以获得明确但不完整的知识（例如，规则、安全约束或专家启发式），但这些知识很少被直接用于提高强化学习代理的样本效率。我们填补了这一空白，提出了我们新颖的神经符号知识与梯度引导的强化学习（KGRL）算法。KGRL利用Datalog知识库中的领域知识，推导出给定状态下适用的动作集合和可行参数。这使得它能够从决策空间中剔除不适用的动作，并限制剩余动作的参数空间。然后，我们使用基于梯度的参数细化循环，在代理的训练和部署过程中估计最优参数。通过记录沿轨迹激活的规则，KGRL还提供了关于动作剔除和参数约束的局部程序解释。总体而言，KGRL引导代理的探索和部署朝向可行且考虑约束的决策，同时提高训练过程中的样本效率。KGRL在样本效率和每回合回报方面均优于PAMDP的最先进强化学习基线。
cs.AI / 48 / 2607.12982
FormalAnalyticGeo: A Neural-Symbolic Based Framework for Multimodal Analytic Geometry Problem Generation
FormalAnalyticGeo：基于神经符号的多模态解析几何问题生成框架
Abstract
Math reasoning has achieved significant progress with the rapid advancement of Multimodal Large Language Models (MLLMs), however analytic geometry remains largely underexplored, primarily due to the scarcity of annotated samples. Existing diagram generation approaches struggle with analytic geometry: template methods cannot handle constraint-driven layouts, and generative models lack the geometric precision to render annotated conic curves correctly. We present FormalAnalyticGeo, a scalable framework for fully automatic generation of multimodal analytic geometry problems. Leveraging the rigor of formal languages, we design the framework around CDL (Condition Description Language), a formal intermediate representation that bridges free-form problem text with precise diagram rendering via a Signed Distance Field (SDF) engine. The framework employs four specialized LLM components in sequence: a Generator that produces diverse analytic geometry problems, a Formalizer that converts each problem into CDL for SDF-based rendering, a Measurer that extracts ground-truth answers through vision-based measurement on the rendered diagrams, and a Quality Verifier that checks outputs at three stages. Structured feedback from the Quality Verifier drives automatic retry, forming a closed loop that eliminates any need for human annotation. Applying FormalAnalyticGeo at scale yields AnalyticGeo7K, a dataset of over 7K verified multimodal problems, each with aligned text, diagram, formal annotation, and ground truth.Experiments show that the generated problems achieve a median ground-truth relative error of 0.70\%, with 82.3\% of answers falling within 5\% of the exact symbolic solution. Our framework and dataset will be publicly released.
Chinese Translation
随着多模态大型语言模型（MLLMs）的快速发展，数学推理取得了显著进展，然而解析几何仍然在很大程度上未被充分探索，主要原因在于标注样本的稀缺。现有的图形生成方法在解析几何方面面临挑战：模板方法无法处理约束驱动的布局，而生成模型缺乏几何精度，无法正确渲染带注释的圆锥曲线。我们提出了FormalAnalyticGeo，这是一个可扩展的框架，用于全自动生成多模态解析几何问题。该框架利用形式语言的严谨性，围绕条件描述语言（CDL）设计，CDL是一种正式的中间表示，连接自由形式的问题文本与通过签名距离场（SDF）引擎进行的精确图形渲染。该框架依次使用四个专门的LLM组件：生成器（Generator）生成多样的解析几何问题，形式化器（Formalizer）将每个问题转换为CDL以便进行基于SDF的渲染，测量器（Measurer）通过对渲染图形进行基于视觉的测量提取真实答案，质量验证器（Quality Verifier）在三个阶段检查输出。质量验证器的结构化反馈驱动自动重试，形成一个闭环，消除了对人工标注的需求。在大规模应用FormalAnalyticGeo时，生成了AnalyticGeo7K，一个包含超过7000个经过验证的多模态问题的数据集，每个问题都配有对齐的文本、图形、正式注释和真实答案。实验表明，生成的问题的中位真实相对误差为0.70%，其中82.3%的答案在精确符号解的5%范围内。我们的框架和数据集将公开发布。
cs.AI / 49 / 2607.12985
Resist and Update: Counterfactual Report Coordinates for Incentive-Compatible LLMs
抵抗与更新：激励兼容的大语言模型的反事实报告坐标
Abstract
Aligned language models routinely misreport under non-evidential incentive pressure: they agree with a confident user or overstate certainty even when their internal belief is unchanged. We cast this as a failure of internal incentive-compatibility (IC) and present a method for learning and certifying counterfactual report mediators that hold a model's reports to a causal contract: invariant to forbidden influences (pressure, prestige, restyling) and responsive to licensed ones (genuine evidence). These two demands, resist and update, pull in opposite directions. We study them on a Bayesian-witness benchmark with known posteriors, in which the same user disagreement is licensed evidence or forbidden pressure purely by stated source reliability. We (i) causally identify, by interchange interventions rather than probe accuracy, low-rank report coordinates for answer, confidence, and caveat that are near-orthogonal and independently controllable, and (ii) introduce a training-free counterfactual report-coordinate (CRC) clamp that references the model's own report under a counterfactually incentive-neutralized context. On the witness benchmark the two-pass clamp attains resist and update of 1.00 jointly (Wilson 95% CI [0.99,1.00]), a causal certificate under a constructible reference, not a deployed solution. Global decoding and steering show a single-parameter tradeoff; output-level fine-tuning matches both objectives only when both are enumerated; resist-only training loses evidence-responsiveness. The deployable single-pass compilation is lossy (0.73/0.97). The mechanism and clamp reproduce across three model families and transfer to a natural sycophancy benchmark (SycophancyEval). Our contribution is the interface and certification method: activation-level counterfactual incentive-invariance as a structural primitive for internal IC.
Chinese Translation
对齐语言模型在非证据性激励压力下常常错误报告：它们会与自信的用户达成一致或夸大确定性，即使它们的内部信念没有变化。我们将此视为内部激励兼容性（IC）的失败，并提出了一种学习和认证反事实报告中介的方法，该方法将模型的报告与因果契约相一致：对禁止影响（压力、声望、重塑）不变，对许可影响（真实证据）有反应。这两个要求，抵抗与更新，朝着相反的方向拉动。我们在一个具有已知后验的贝叶斯见证基准上研究它们，在该基准中，同一用户的不一致被声明的来源可靠性纯粹地视为许可证据或禁止压力。我们（i）通过互换干预而非探测准确性，因果识别出近正交且独立可控的低秩报告坐标，用于答案、置信度和警告，以及（ii）引入了一种无训练的反事实报告坐标（CRC）夹具，该夹具在反事实激励中立化的上下文中参考模型自身的报告。在见证基准上，双通道夹具共同实现了抵抗与更新的1.00（Wilson 95% CI [0.99,1.00]），这是在可构建参考下的因果证书，而非已部署的解决方案。全局解码和引导显示出单参数权衡；输出级微调仅在同时列举两个目标时才能匹配这两个目标；仅抵抗训练会失去对证据的响应性。可部署的单通道编译是有损的（0.73/0.97）。该机制和夹具在三种模型家族中重现，并转移到一个自然的谄媚基准（SycophancyEval）。我们的贡献是接口和认证方法：激活级反事实激励不变性作为内部IC的结构原语。
cs.AI / 50 / 2607.12986
Win by Silence: Deletion Non-Monotonicity, Autonomous Exploitation, and Typed-State Gating in LLM Plan Evaluation
以沉默取胜：删除非单调性、自主利用与LLM计划评估中的类型状态门控
Abstract
Plan evaluators can reward a strategic plan for becoming less explicit. This paper studies that failure in a staged expected-value scorer for LLM-generated venture routes. Proposition 1 gives the score change from deleting an interior transition while retargeting its predecessor and retaining downstream value: Delta_k = (prod_{i
Chinese Translation
计划评估者可以奖励一个战略计划变得不那么明确。本文研究了LLM生成的风险路线中，分阶段期望值评分器的这一失败。命题1给出了在重新定位其前驱并保留下游价值的情况下，删除内部转换所导致的评分变化：Delta_k = (prod_{i
cs.AI / 51 / 2607.13007
Dynamic Resource Allocation for Ensemble Determinization MCTS
集成确定化蒙特卡洛树搜索的动态资源分配
Abstract
Simulation-based algorithms are especially suited for high-uncertainty environments such as adversarial board games with significant elements of randomness and hidden information. In particular, several Monte Carlo Tree Search (MCTS) variants are commonly used in such domains. In this paper, we propose a series of enhancements for Ensemble Determinization MCTS, introducing two axes for dynamic resource allocation. First, Dynamic Number of Determinizations, increases or decreases the number of currently used determinization trees depending on the behavior of so-far search. Second, Dynamic Simulation Allocation, splits the simulation budget nonuniformly across the determinization trees, using simulation-to-simulation decisions to choose the tree with potentially the best knowledge gain. As benchmark domains, we used three popular tabletop games: Jaipur, Lost Cities, and Splendor. Testing our proposed enhancements in iteration- and time-based settings showed that particular configurations yield a statistically significant increase in the algorithm's strength.
Chinese Translation
基于模拟的算法特别适用于高不确定性环境，例如具有显著随机性和隐藏信息的对抗性棋盘游戏。在这些领域中，几种蒙特卡洛树搜索（MCTS）变体被广泛使用。本文提出了一系列针对集成确定化MCTS的增强，介绍了两个动态资源分配的轴心。首先，动态确定化数量根据当前搜索的行为增加或减少正在使用的确定化树的数量。其次，动态模拟分配在确定化树之间不均匀地分配模拟预算，利用模拟到模拟的决策选择潜在知识增益最佳的树。作为基准领域，我们使用了三款流行的桌面游戏：斋浦尔（Jaipur）、失落的城市（Lost Cities）和辉煌（Splendor）。在迭代和基于时间的设置中测试我们提出的增强显示，特定配置在算法强度上产生了统计显著的提升。
cs.AI / 52 / 2607.13013
Audio-Native Speech Recognition with a Frozen Discrete-Diffusion Language Model
基于音频的语音识别与冻结的离散扩散语言模型
Abstract
Automatic speech recognition is dominated by autoregressive decoders that emit one token at a time. We ask whether a discrete diffusion language model can transcribe speech instead, refining a whole transcript in parallel over a small number of denoising steps. We train an audio-native interface for DiffusionGemma, a 26B mixture-of-experts model that generates text by uniform, random-token discrete diffusion rather than the absorbing-mask scheme common to recent diffusion language models. A frozen Whisper encoder supplies acoustic features, a lightweight projector maps them into the model embedding space, and low-rank adapters let the frozen backbone attend to the new modality. About 42M parameters are trained, which is 0.16 percent of the backbone. We find that the natural training objectives fail to ground the audio because their gradient reaches the projector only through attention that has already dismissed it. A connectionist temporal classification loss applied through the frozen output head breaks this deadlock. The resulting model reaches 6.6 percent word error rate on LibriSpeech test-clean, transcribes in roughly eight parallel steps regardless of utterance length, and uses a single adapter trained on six languages, which we evaluate here on English, Hindi, and Mandarin.
Chinese Translation
自动语音识别主要依赖自回归解码器逐个发出令牌。我们探讨是否可以使用离散扩散语言模型来转录语音，而是通过少量去噪步骤并行地优化整个转录。我们为DiffusionGemma训练了一个音频原生接口，这是一个26B专家混合模型，通过均匀的随机令牌离散扩散生成文本，而不是最近扩散语言模型常用的吸收掩码方案。一个冻结的Whisper编码器提供声学特征，一个轻量级投影器将其映射到模型嵌入空间，低秩适配器使冻结的主干能够关注新的模态。约4200万个参数被训练，占主干的0.16%。我们发现，自然训练目标未能将音频与文本对齐，因为其梯度仅通过已经忽略音频的注意力到达投影器。通过冻结的输出头施加的连接时序分类损失打破了这一僵局。最终模型在LibriSpeech test-clean上达到了6.6%的字错误率，无论发音长度如何，均在大约八个并行步骤中进行转录，并使用一个在六种语言上训练的单一适配器，我们在此评估了英语、印地语和普通话。
cs.AI / 53 / 2607.13034
Do AI Agents Know When a Task Is Simple? Toward Complexity-Aware Reasoning and Execution
人工智能代理是否知道任务的简单性？朝向复杂性感知的推理与执行
Abstract
Large language model (LLM) agents increasingly automate multi-step engineering and informatics workflows, yet they rarely ask how much effort a task actually requires. They often follow a maximum-context-first strategy--re-reading files and dependencies they have already seen--turning a one-line edit into a small code-base audit. We argue the missing capability is task-aware execution-scope estimation: judging a task's difficulty, the information it truly needs, and the shortest reliable path before committing budget. We formalize minimum-sufficient execution and the Agent Cognitive Redundancy Ratio (ACRR), and propose E3 (Estimate, Execute, Expand): the agent estimates an initial operating point, executes a minimum viable path, and expands scope only when verification fails. On MSE-Bench--a deterministic benchmark of 121 edits in a capability-controlled simulator--E3 matches the strongest baseline's 100% success while cutting cost by 85%, tokens by 91%, and inspected files by 92%, and further beats a strong adaptive retrieval baseline by 16%; the gains survive held-out instruction wording and essentially every cost weighting. A companion real-model harness (LLM-Case) corroborates the effect on a live gpt-4o agent editing a real open-source library, with every candidate patch graded by actually running the project's real pytest suite against a measured oracle: the over-reading is milder but real, and E3 is the leanest and fastest policy at comparable task success--its one shortfall a provider rate-limit, not a wrong edit. We frame this as a controlled probe of execution redundancy, not a measurement of any deployed agent, and position task-aware execution as a step toward engineering-grounded AI (EGAI)--agents whose effort is anchored in the engineering reality of the task. We release the framework and benchmark.
Chinese Translation
大型语言模型（LLM）代理越来越多地自动化多步骤的工程和信息学工作流程，但它们很少询问一个任务实际需要多少努力。它们通常遵循最大上下文优先策略——重新阅读它们已经看到的文件和依赖项——将一行代码的编辑变成小型代码库的审计。我们认为缺失的能力是任务感知的执行范围估计：判断任务的难度、它真正需要的信息以及在投入预算之前的最短可靠路径。我们形式化了最小充分执行和代理认知冗余比（ACRR），并提出了E3（估计、执行、扩展）：代理估计一个初始操作点，执行一个最小可行路径，仅在验证失败时扩展范围。在MSE-Bench——一个在能力受控模拟器中进行的121次编辑的确定性基准测试中——E3与最强基线的100%成功率相匹配，同时将成本降低85%，令牌数量减少91%，检查的文件减少92%，并进一步超越一个强大的自适应检索基线16%；这些收益在保留的指令措辞和几乎所有成本权重下依然有效。一个配套的真实模型工具（LLM-Case）证实了其对实时gpt-4o代理编辑真实开源库的影响，每个候选补丁通过实际运行项目的真实pytest套件与测量的oracle进行评分：过度阅读的情况较轻但确实存在，而E3是在可比任务成功率下最精简和最快的策略——其唯一的不足是提供者的速率限制，而非错误编辑。我们将此视为对执行冗余的受控探测，而不是对任何已部署代理的测量，并将任务感知执行定位为朝向工程基础人工智能（EGAI）的一步——其努力根植于任务的工程现实。我们发布了该框架和基准测试。
cs.CL / 1 / 2607.11889
Scaling Point-in-Time Language Models
扩展时点语言模型
Abstract
Large language models trained on unrestricted internet corpora inevitably embed information from the future, introducing lookahead bias that compromises the validity of backtests and causal inference in finance and the social sciences. Point-in-time language models--trained exclusively on text available up to each calendar date--eliminate this leakage by construction, but existing efforts typically produce models that lag substantially behind their unconstrained counterparts. We show that this performance gap can be substantially narrowed through scale. Training decoder-only transformers with up to 4 billion parameters on 1 trillion chronologically filtered tokens from FineWeb, we construct a sequence of monthly model checkpoints spanning 2013-2024. Across a range of common-sense reasoning and language understanding benchmarks, our models approach the performance of leading open-weight models of comparable size (e.g., Gemma-3-4B and LLaMA-7B) trained on temporally unrestricted data, although a performance gap remains on several tasks. Instruction fine-tuning via LoRA further improves downstream usability. We release the complete pipeline--including dataset construction, training infrastructure, and evaluation code--to enable reproducible point-in-time language modeling and to support research applications that require strict temporal validity.
Chinese Translation
在无限制的互联网语料上训练的大型语言模型不可避免地嵌入了未来的信息，这引入了前瞻性偏差，从而损害了金融和社会科学中回测和因果推断的有效性。时点语言模型——仅基于每个日历日期之前可用的文本进行训练——通过构造消除了这种信息泄漏，但现有的努力通常产生的模型在性能上显著滞后于其不受限制的对应物。我们展示了通过规模可以显著缩小这一性能差距。我们在来自 FineWeb 的 1 万亿按时间过滤的标记上训练了最多 40 亿参数的解码器仅变换器，构建了一系列跨越 2013-2024 年的每月模型检查点。在一系列常识推理和语言理解基准测试中，我们的模型接近于在时间上不受限制的数据上训练的同等规模的领先开放权重模型（例如，Gemma-3-4B 和 LLaMA-7B）的性能，尽管在几个任务上仍存在性能差距。通过 LoRA 进行的指令微调进一步提高了下游可用性。我们发布了完整的流程，包括数据集构建、训练基础设施和评估代码，以实现可重复的时点语言建模，并支持需要严格时间有效性的研究应用。
cs.CL / 2 / 2607.11891
CANDI: Contextual Alignment for Niche Domains Question Answering
CANDI：针对细分领域问答的上下文对齐
Abstract
The deployment of large language models (LLMs) in specialized domains like medical diagnostics and financial advisory necessitates evaluating capabilities beyond general knowledge. Traditional question-answering benchmarks often fail to capture the nuanced contextual grounding, user awareness, and domain understanding these fields require. To address this, we introduce CANDI-QA (Contextual Alignment for Niche Domains Question Answering), a novel dataset evaluating LLMs on delivering accurate, context-sensitive, and user-aligned answers in specialized settings. CANDI-QA features expert-curated question-answer pairs structured into two categories: (1) Information Assistance Questions, which are direct, factual queries requiring precise extraction, and (2) Applied Inference Questions, which are multi-hop reasoning tasks needing situational inference to generate actionable insights. We evaluate over ten diverse language models, from compact open-source to state-of-the-art proprietary systems. As a robust baseline, we present MTSS-Net, a lightweight neuro-symbolic framework combining neural retrieval with rule-based reasoning. Our findings highlight the profound challenges of achieving contextual alignment in niche domains, revealing the limitations of current LLMs without enhanced contextual or symbolic integration. Ultimately, CANDI-QA serves as a critical benchmark for advancing research in context-aware language models, stimulating the development of robust, trustworthy AI for high-stakes domains.
Chinese Translation
在医疗诊断和财务顾问等专业领域中部署大型语言模型（LLMs）需要评估其能力，超越一般知识。传统的问答基准往往无法捕捉这些领域所需的细致上下文基础、用户意识和领域理解。为了解决这一问题，我们引入了CANDI-QA（针对细分领域问答的上下文对齐），这是一个新颖的数据集，用于评估LLMs在专业环境中提供准确、上下文敏感和用户对齐答案的能力。CANDI-QA包含专家策划的问题-答案对，分为两类：（1）信息辅助问题，这类问题是直接的、事实性的查询，需要精确提取；（2）应用推理问题，这类问题是多跳推理任务，需要情境推断以生成可操作的见解。我们评估了十多种不同的语言模型，从紧凑的开源模型到最先进的专有系统。作为一个稳健的基线，我们提出了MTSS-Net，这是一种轻量级的神经符号框架，结合了神经检索与基于规则的推理。我们的研究结果突显了在细分领域实现上下文对齐的深刻挑战，揭示了当前LLMs在没有增强的上下文或符号集成的情况下的局限性。最终，CANDI-QA作为一个关键基准，推动了上下文感知语言模型的研究，刺激了高风险领域中强大、可信赖的人工智能的发展。
cs.CL / 3 / 2607.11892
G-SHARE: A Guideline-Based Structured Reasoning Framework for Human-Factor Event Diagnosis
G-SHARE：基于指南的结构化推理框架用于人因事件诊断
Abstract
Human-factor event diagnosis is essential for learning from operational events in nuclear power plants, yet its quality depends strongly on expert interpretation of narrative reports and guideline-based reasoning.Existing data-driven or one-shot large language model approaches often lack structured reasoning, have limited alignment with formal diagnostic guidelines, and may generate logically inconsistent conclusions. To address this issue, this study proposes G-SHARE, a guideline-based structured reasoning framework that operationalizes the CNNP nine-step human-factor event diagnosis guideline into a multi-stage diagnostic pipeline.The framework consists of evidence extraction, stepwise diagnostic reasoning, and post-hoc consistency repair, enabling explicit use of report evidence, intermediate rationale generation, and logical validation of diagnostic outputs. A dataset of real human-factor event reports was constructed from Chinese nuclear industry sources, and a gold-standard subset annotated by domain experts was used for evaluation. Results show that G-SHARE substantially outperforms one-shot prompting and traditional machine learning baselines, with the strongest version achieving the best overall accuracy and macro-F1. Ablation results further indicate that structured reasoning and consistency enforcement are critical to robust diagnosis, especially under weak prompting conditions. The findings demonstrate the value of transforming expert diagnostic guidelines into auditable reasoning workflows, providing a practical pathway for intelligent human-factor analysis in safety-critical industries.
Chinese Translation
人因事件诊断对于从核电厂的操作事件中学习至关重要，但其质量在很大程度上依赖于专家对叙述报告的解读和基于指南的推理。现有的数据驱动或一次性大型语言模型方法往往缺乏结构化推理，与正式诊断指南的对齐有限，并可能产生逻辑不一致的结论。为了解决这一问题，本研究提出了G-SHARE，一个基于指南的结构化推理框架，将CNNP九步人因事件诊断指南操作化为多阶段诊断流程。该框架包括证据提取、逐步诊断推理和事后一致性修复，能够明确使用报告证据、生成中间推理和对诊断输出进行逻辑验证。我们从中国核工业来源构建了一个真实人因事件报告的数据集，并使用由领域专家注释的金标准子集进行评估。结果表明，G-SHARE在一次性提示和传统机器学习基线中显著优于其他方法，最强版本实现了最佳的总体准确率和宏观F1值。消融实验结果进一步表明，结构化推理和一致性强制对于稳健的诊断至关重要，尤其是在弱提示条件下。研究结果展示了将专家诊断指南转化为可审计推理工作流的价值，为安全关键行业中的智能人因分析提供了切实可行的路径。
cs.CL / 4 / 2607.11893
I'm Sorry, but I Can't Help with Braille: Revealing Accessibility Failures in State-of-the-Art LLMs
对不起，我无法帮助盲文：揭示最先进大型语言模型中的可及性失败
Abstract
Large Language Models (LLMs) perform strongly on many language tasks, but their capability in structurally constrained, accessibility-critical modalities such as Braille remains unclear. We evaluate state-of-the-art LLMs on bidirectional Korean-Braille translation using a human-annotated dataset. Despite expectations that multilingual, instruction-tuned models can generalize to Braille via text representations, we find consistently poor, unstable outputs and substantial disagreement with human judgments. These results point to missing Braille-aware tokenization and weak alignment between Korean and Braille patterns. In contrast, supervised fine-tuning of a small model (T5-small) on the same data yields large and stable gains over zero-shot and prompted LLM baselines across standard metrics (SacreBLEU, ChrF++, CER, BLEU, ROUGE-L, METEOR, CIDEr). Our findings reveal a systematic limitation of current LLMs and demonstrate the effectiveness of modest task-specific supervision.
Chinese Translation
大型语言模型（LLMs）在许多语言任务中表现出色，但它们在结构受限、对可及性至关重要的模式（如盲文）中的能力仍不明确。我们使用人工标注的数据集评估了最先进的LLMs在双向韩文-盲文翻译中的表现。尽管预期多语言、经过指令调优的模型能够通过文本表示对盲文进行泛化，但我们发现输出结果始终较差且不稳定，与人类判断存在显著不一致。这些结果表明缺乏对盲文的感知分词和韩文与盲文模式之间的弱对齐。相比之下，在相同数据上对小型模型（T5-small）进行监督微调，能够在标准指标（SacreBLEU、ChrF++、CER、BLEU、ROUGE-L、METEOR、CIDEr）上实现显著且稳定的提升，超越零-shot和提示的LLM基线。我们的发现揭示了当前LLMs的系统性局限性，并展示了适度任务特定监督的有效性。
cs.CL / 5 / 2607.11894
Graph-Based Detection of Disinformation Narrative Diffusion between Russian and Ukrainian Telegram Channels
基于图的虚假信息叙事传播检测：以俄罗斯和乌克兰Telegram频道为例
Abstract
Detecting disinformation narratives on social media is challenging due to the scale of amplification, rapid evolution, and linguistic variability of online content. We propose a graph-based framework for identifying and analyzing disinformation narratives in Telegram ecosystems by combining weak supervision with propagation graph analysis. The approach aggregates semantically related claims into narrative-level clusters and models their diffusion across interconnected channels. This enables the detection of coordinated narrative amplification that is difficult to capture through post-level analysis alone. Our results demonstrate that integrating textual signals with network structure provides a scalable method for detecting disinformation narratives and offers insights into how they propagate within large-scale messaging environments.
Chinese Translation
在社交媒体上检测虚假信息叙事具有挑战性，因为在线内容的放大规模、快速演变和语言变异性。我们提出了一种基于图的框架，通过结合弱监督与传播图分析，识别和分析Telegram生态系统中的虚假信息叙事。该方法将语义相关的主张聚合成叙事级别的聚类，并建模它们在互联频道中的传播。这使得检测协调的叙事放大成为可能，而仅通过帖子级分析难以捕捉。我们的结果表明，将文本信号与网络结构相结合提供了一种可扩展的方法来检测虚假信息叙事，并提供了关于它们在大规模消息环境中传播的洞察。
cs.CL / 6 / 2607.11898
TAKE: Trajectory-Aware Knowledge Estimation for Text Dataset Distillation
TAKE：基于轨迹的文本数据集蒸馏知识估计
Abstract
Large-scale text corpora have become a quiet bottleneck in modern NLP, not just in storage, but in the accumulated cost of training, fine-tuning, and continual learning. We propose a text dataset distillation framework that reduces corpora to as little as 0.1% of their original size while preserving downstream task fidelity. We approach distillation through the lens of influence functions, which quantify each sample's contribution to the downstream objective, a natural and principled basis for selection. We introduce Trajectory-Aware Knowledge Estimation (TAKE), which convolves the knowledge-based influence along the training trajectory into a single per-sample knowledge score, capturing informative samples. These scores serve as sample weights within a discrete Optimal Transport objective, guiding prototype selection from a synthetically generated candidate pool. We evaluate TAKE on downstream accuracy across text classification and natural language inference tasks at extreme compression (0.1% or 20 samples/class), showing that data efficiency is achievable without sacrificing task fidelity. The approach is theoretically grounded, with broader implications for coreset construction and data-centric AI. We release our source code at https://github.com/votrinhan88/take.
Chinese Translation
大规模文本语料库已成为现代自然语言处理（NLP）中的一个隐性瓶颈，不仅在存储方面，而且在训练、微调和持续学习的累积成本方面。我们提出了一种文本数据集蒸馏框架，将语料库缩减至其原始大小的0.1%以下，同时保持下游任务的保真度。我们通过影响函数的视角来进行蒸馏，该函数量化每个样本对下游目标的贡献，成为选择的自然和原则基础。我们引入了基于轨迹的知识估计（Trajectory-Aware Knowledge Estimation，TAKE），它将基于知识的影响沿训练轨迹卷积为单个样本知识分数，从而捕捉信息丰富的样本。这些分数作为离散最优传输目标中的样本权重，指导从合成生成的候选池中选择原型。我们在极端压缩（0.1%或每类20个样本）下评估TAKE在文本分类和自然语言推理任务上的下游准确性，显示出在不牺牲任务保真度的情况下实现数据效率是可行的。该方法在理论上是有基础的，并对核心集构建和数据中心人工智能具有更广泛的影响。我们在https://github.com/votrinhan88/take发布了我们的源代码。
cs.CL / 7 / 2607.11933
Transforming LLMs into Efficient Cross-Encoders via Knowledge Distillation for RAG Reranking
通过知识蒸馏将大型语言模型转化为高效的交叉编码器以进行检索增强生成的重排序
Abstract
Cross-encoders achieve high reranking accuracy in Retrieval-Augmented Generation (RAG) pipelines but impose quadratic inference costs that limit real-time deployment. We address this by fine-tuning LLaMA 3 (8B) as a drop-in reranker using a two-stage pipeline: supervised fine-tuning on a custom query-document relevance dataset via the Unsloth framework with LoRA adapters, followed by 4-bit quantization for efficient inference. The resulting model replaces the cross-encoder in a dual-retriever RAG pipeline combining BM25 and dense vector search. Evaluated on a domain-specific question-answering benchmark using the RAGAS framework, our fine-tuned LLaMA 3 reranker achieves gains of 14% in answer relevancy, 16% in context precision, 19% in answer similarity, and 21% in answer correctness over the cross-encoder baseline, while reducing inference overhead through 4-bit quantization. These results demonstrate that instruction-tuned LLMs can be adapted into accurate, efficient rerankers without the quadratic complexity of traditional cross-encoders.
Chinese Translation
交叉编码器在检索增强生成（RAG）管道中实现了高重排序准确性，但其二次推理成本限制了实时部署。我们通过使用两阶段管道对 LLaMA 3（8B）进行微调，将其作为即插即用的重排序器来解决这一问题：首先通过 Unsloth 框架和 LoRA 适配器在自定义查询-文档相关性数据集上进行监督微调，随后进行 4 位量化以实现高效推理。所得到的模型替代了结合 BM25 和稠密向量搜索的双检索器 RAG 管道中的交叉编码器。在使用 RAGAS 框架的特定领域问答基准上进行评估，我们微调后的 LLaMA 3 重排序器在答案相关性上提高了 14%，在上下文精确度上提高了 16%，在答案相似性上提高了 19%，在答案正确性上提高了 21%，同时通过 4 位量化减少了推理开销。这些结果表明，经过指令微调的大型语言模型可以被改造成准确且高效的重排序器，而无需传统交叉编码器的二次复杂性。
cs.CL / 8 / 2607.11944
MAGE: Understanding Stability-Performance Trade-offs in Multi-component Prompt Optimization
MAGE：理解多组件提示优化中的稳定性与性能权衡
Abstract
How do different components of iterative prompt optimization interact, and what happens when they are combined? We investigate this through MAGE (Memory-Augmented Goal-directed Prompt Evolution), a controlled analysis framework for studying component interaction in prompt optimization. MAGE is not proposed as a superior optimizer in absolute terms; it integrates episodic memory, multi-objective Pareto selection, and adaptive evaluation as a platform for controlled ablation. Our experiments uncover a previously unreported phenomenon, the Prompt Optimization Coupling Effect (POCE): when multiple stochastic optimization signals operate within a closed reflective loop, they interact in ways that simultaneously improve performance and amplify variance, behavior that cannot be predicted by analyzing components in isolation. Three main findings emerge. First, failure-grounded reflection is essential: methods relying only on scores (OPRO) or abstract critique (Self-Refine) fail to improve prompts. Second, MAGE achieves 46.4% versus GEPA's 34.0% on GSM8K-Hard (+12.4%, P(MAGE>GEPA)=0.998, 5 seeds on gpt-4o-mini), with comparable variance (7.3% vs. 7.0%). Third, increasing candidate diversity reveals the clearest POCE signal: expanding the candidate pool from n=3 to n=5 improves mean accuracy by +21.6% while increasing variance by 3.7x. We further validate on Llama 3.1 8B and show POCE is headroom-dependent: when the base model already achieves high accuracy, variance amplification disappears. Finally, in low-data regimes (Ntrain=30), well-designed fixed prompts outperform all reflective optimizers, indicating that scaffold choice dominates optimizer choice. Our results suggest prompt optimization systems behave as coupled stochastic processes and should be evaluated in terms of both performance and stability, not just peak accuracy.
Chinese Translation
不同组件在迭代提示优化中如何相互作用？当它们结合时会发生什么？我们通过MAGE（记忆增强目标导向提示演化）这一控制分析框架来研究提示优化中的组件交互。MAGE并不是作为一种绝对优越的优化器提出的；它将情景记忆、多目标帕累托选择和自适应评估整合为一个控制消融的平台。我们的实验揭示了一种之前未报告的现象，即提示优化耦合效应（POCE）：当多个随机优化信号在一个封闭的反思循环中运行时，它们以同时提高性能和放大方差的方式相互作用，这种行为无法通过孤立分析组件来预测。我们得出了三个主要发现。首先，基于失败的反思至关重要：仅依赖于分数（OPRO）或抽象批评（Self-Refine）的方法未能改善提示。其次，MAGE在GSM8K-Hard上的表现为46.4%，而GEPA为34.0%（+12.4%，P(MAGE>GEPA)=0.998，5次种子在gpt-4o-mini上），且方差相当（7.3%对比7.0%）。第三，增加候选多样性揭示了最清晰的POCE信号：将候选池从n=3扩展到n=5使平均准确率提高了+21.6%，同时方差增加了3.7倍。我们进一步在Llama 3.1 8B上进行验证，显示POCE依赖于余量：当基础模型已经达到高准确率时，方差放大现象消失。最后，在低数据环境下（Ntrain=30），设计良好的固定提示优于所有反思优化器，表明支架选择优于优化器选择。我们的结果表明，提示优化系统表现为耦合的随机过程，应该在性能和稳定性两个方面进行评估，而不仅仅是峰值准确率。
cs.CL / 9 / 2607.11945
Belief-reality separation lives in routing over a shared value slot in language models
信念与现实的分离存在于语言模型中对共享价值槽的路由
Abstract
Capable language models hold what a character believes apart from what is true: told "Anna believes the cup is blue; in reality it is red," they answer blue about Anna and red about the world. Where in the computation does that separation live? We show it rests on two separable mechanisms at two positions. A generic value slot binds the attributed value. A router at the query position selects which frame, the character's belief or reality, a query reads out. Two routes fill the slot: an asserted belief, whose value the text supplies, binds in directly; a derived belief, whose value must be inferred from what the character could see, arrives by a visibility-gated lookback. A subspace trained on either route steers the other, and only the derived route depends on described visibility. The slot itself carries no belief-reality tag: intervening on it moves a reality readout as strongly as a belief one. The separation lives instead in a dissociated pair of routing subspaces, which flip a query between frames without injecting the donor's value. These results hold across three architectures, on stimuli de-confounded against theory-of-mind-benchmark shortcuts; the behavior itself emerges between 3B and 7B across five model families. This paper develops the single belief-reality axis in depth; a companion paper shows the same slot-and-router format is shared across the other non-actual contexts a sentence can open (counterfactual, fictional, temporal).
Chinese Translation
能够理解的语言模型将角色的信念与真实情况区分开来：当被告知“安娜相信杯子是蓝色的；实际上它是红色的”时，他们会回答安娜认为是蓝色的，而世界上是红色的。那么，这种分离在计算的哪个位置存在呢？我们展示了它依赖于两个可分离的机制和两个位置。一个通用的价值槽绑定了被赋予的价值。查询位置的路由器选择查询读取哪个框架，即角色的信念或现实。两个路径填充这个槽：一个是被断言的信念，其值由文本提供，直接绑定；另一个是推导的信念，其值必须从角色所能看到的内容中推断，通过一个可见性门控的回顾到达。一个在任一路径上训练的子空间会引导另一个，而只有推导路径依赖于描述的可见性。该槽本身不携带信念-现实标签：对其进行干预会像信念读取一样强烈地移动现实读取。分离存在于一对解耦的路由子空间中，它们在框架之间翻转查询，而不注入捐赠者的值。这些结果在三种架构中保持一致，在对抗理论心智基准快捷方式的刺激上；这种行为在五个模型家族中在3B到7B之间出现。本文深入探讨了单一的信念-现实轴；一篇伴随论文表明，相同的槽-路由器格式在句子可以打开的其他非实际上下文（反事实、虚构、时间）中是共享的。
cs.CL / 10 / 2607.11946
Hybrid Continual Learning for Low-Resource Australian Aboriginal Language Identification
低资源澳大利亚土著语言识别的混合持续学习
Abstract
Language identification is an important step toward integrating endangered Australian Aboriginal languages (AALs) into speech technologies supporting language revitalisation and digital inclusion. However, extreme data scarcity limits model performance. Transfer learning from high-resource languages shows promise but often suffers from catastrophic forgetting when adapting to new languages. Continual learning (CL) can mitigate this issue, though it remains challenging with very limited data. To address this, we propose two hybrid continual learning methods: Replay Augmented Elastic Weight Consolidation and Constraint Guided Knowledge Distillation to adapt pretrained speech models for AAL identification while preserving previously learned knowledge. Experiments on Warlpiri, Dalabon and Dharawal show that the proposed methods outperform fine-tuning and existing CL baselines, improving adaptation to multiple AALs while maintaining performance on previously learnt high-resource languages.
Chinese Translation
语言识别是将濒危的澳大利亚土著语言（AALs）融入支持语言复兴和数字包容的语音技术的重要一步。然而，极端的数据稀缺性限制了模型的性能。从高资源语言进行迁移学习展现出前景，但在适应新语言时常常遭遇灾难性遗忘。持续学习（CL）可以缓解这一问题，尽管在数据极为有限的情况下仍然具有挑战性。为了解决这一问题，我们提出了两种混合持续学习方法：重放增强弹性权重巩固（Replay Augmented Elastic Weight Consolidation）和约束引导知识蒸馏（Constraint Guided Knowledge Distillation），以适应预训练的语音模型进行AAL识别，同时保留先前学习的知识。在Warlpiri、Dalabon和Dharawal的实验中，所提出的方法在适应多个AAL时表现优于微调和现有的CL基线，同时保持对先前学习的高资源语言的性能。
cs.CL / 11 / 2607.11981
Evaluating Nonuniform Dependability Across Response Conditions: A Conditional Generalizability Framework Illustrated in Automated Essay Scoring
评估响应条件下的不均匀可靠性：以自动化作文评分为例的条件广泛性框架
Abstract
Aggregate reliability estimates can obscure heterogeneity in measurement-design burden across response conditions, so a single G- or D-study may mischaracterize a design's adequacy for particular strata. This study introduces a conditional generalizability framework with three components. First, automated scoring configurations -- the encoder architectures and scoring-head families admissible within a fixed pipeline -- are treated as a universe of admissible measurement conditions rather than incidental modeling choices. Second, analytical D-study projections are compared with empirical configuration sweeps over a finite scoring pool, yielding two estimands of design adequacy whose agreement or divergence diagnoses the realized configuration universe. Third, evidence is conditioned on entropy-defined response strata, treating entropy as an operational stratification variable, not a construct claim about writing quality. Whereas recent generalizability-theory extensions address AI-generated item variants on the response side, this framework addresses the analogous scoring-side problem: AI-mediated scoring configurations. Demonstrated with automated essay scoring of timed L2 writing, the realized design was dependable in aggregate (Phi approx 0.76). Re-estimated within entropy strata, dependability stayed high but declined modestly and robustly (Phi = 0.88, 0.87, 0.84) -- a gradient implying different decision-study requirements, the highest-entropy stratum requiring the most crossed conditions. The framework offers a portable workflow for evaluating nonuniform dependability.
Chinese Translation
聚合可靠性估计可能掩盖响应条件下测量设计负担的异质性，因此单一的 G 研究或 D 研究可能会错误地描述设计在特定层次上的适用性。本研究引入了一个包含三个组成部分的条件广泛性框架。首先，自动评分配置——在固定管道内可接受的编码器架构和评分头系列——被视为可接受测量条件的一个宇宙，而不是偶然的建模选择。其次，分析性 D 研究的预测与有限评分池的经验配置扫描进行比较，产生两个设计适用性的估计量，其一致性或差异性诊断实现的配置宇宙。第三，证据基于熵定义的响应层次进行条件化，将熵视为一个操作性分层变量，而不是关于写作质量的构念声明。尽管最近的广泛性理论扩展关注于响应侧的 AI 生成项目变体，但该框架解决了类似的评分侧问题：AI 介导的评分配置。通过对定时 L2 写作的自动化作文评分进行演示，所实现的设计在总体上是可靠的（Phi 约为 0.76）。在熵层次内重新估计后，可靠性保持高水平，但适度且稳健地下降（Phi = 0.88, 0.87, 0.84）——这一梯度暗示了不同决策研究的要求，最高熵层次需要最多的交叉条件。该框架提供了一种可移植的工作流程，用于评估不均匀的可靠性。
cs.CL / 12 / 2607.12051
Agentic systems for breast cancer treatment recommendations
用于乳腺癌治疗推荐的自主系统
Abstract
Large language models (LLMs) are increasingly being explored for clinical decision support, but their reliability in complex oncology treatment planning remains unclear. We evaluated agentic LLM systems for breast cancer treatment recommendation generation using 72 real clinical cases across stages I to IV and 1,147 case-specific rubrics generated through Asymmetric Information Rubric Generation (AIRG), in which the rubric generator had access to real clinical decisions unavailable to the evaluated models. Seven pipelines were compared, including single-LLM baselines, tool-augmented systems, and multi-agent architectures with fact checking and autonomous subagent spawning. The best-performing configuration, Claude Opus 4.8 with the D&C+SA pipeline, achieved a global score of 0.594 $\pm$ 0.025. Tool use and increased agent autonomy had mixed effects, improving performance in some settings but degrading it in others. Performance varied by clinical domain and disease stage, and oncologist-led error analysis revealed persistent clinically relevant failures, including incorrect or missing recommendations, flawed justifications, citation errors, outdated claims, and overconfidence. These findings suggest that agentic LLM systems can generate clinically relevant breast cancer recommendations, but remain insufficient for unsupervised clinical use.
Chinese Translation
大型语言模型（LLMs）在临床决策支持中的应用越来越受到关注，但它们在复杂的肿瘤治疗规划中的可靠性仍不明确。我们评估了自主LLM系统在乳腺癌治疗推荐生成中的表现，使用了72个真实的临床案例，涵盖I至IV期，以及通过不对称信息评分生成（Asymmetric Information Rubric Generation, AIRG）生成的1,147个特定案例评分，其中评分生成器可以访问评估模型无法获得的真实临床决策。我们比较了七种管道，包括单一LLM基线、工具增强系统和具有事实检查及自主子代理生成的多代理架构。表现最佳的配置是Claude Opus 4.8与D&C+SA管道，获得了全球评分0.594 ± 0.025。工具使用和增加代理自主性对性能的影响各异，在某些设置中提高了性能，而在其他情况下则降低了性能。性能因临床领域和疾病阶段而异，肿瘤医生主导的错误分析揭示了持续存在的临床相关失败，包括不正确或缺失的推荐、缺陷的理由、引用错误、过时的主张和过度自信。这些发现表明，自主LLM系统能够生成临床相关的乳腺癌推荐，但仍不足以用于无监督的临床使用。
cs.CL / 13 / 2607.12071
Beyond Parallel Tracking: Interactive Multi-Feature Fusion Drives Semantic Reconstruction from Non-invasive Brain Recordings
超越平行跟踪：交互式多特征融合驱动非侵入性脑记录的语义重建
Abstract
Continuous semantic reconstruction from non-invasive neural recordings remains limited by the representational mismatch between semantic feature spaces and neural coding patterns, which severely impedes cross-modal alignment between high-noise neural signals and target semantic features. Prior semantic decoders have predominantly relied on static lexical representations or dynamic contextualized representations in isolation. This single-dimension approach inevitably leads to severe information loss, as it fails to account for the human brain's capacity to integrate stable word attributes and dynamic contexts simultaneously.To bridge this gap, this study introduces a multi-feature fusion framework for non-invasive semantic reconstruction, systematically benchmarking two integration approaches: linear Naive Concatenation and non-linear Multi-Head Cross-Attention. Within this framework, our approach complements static lexical representations (W2V) with dynamic contextual representations (GPT) via an interactive gating mechanism to facilitate cooperative processing during language comprehension.Evaluated through extensive semantic reconstruction and text generation experiments, our framework reveals a robust performance hierarchy: Cross-Att > Concat > GPT > W2V. Crucially, the non-linear cross-attention fusion method achieves state-of-the-art performance, demonstrating that neural language decoding benefits from simulating the collaborative modulation between contextual information and core lexical attributes rather than depending on isolated individual features, while also offering a viable non-invasive brain-to-text decoding method.
Chinese Translation
从非侵入性神经记录中进行连续语义重建仍然受到语义特征空间与神经编码模式之间表征不匹配的限制，这严重阻碍了高噪声神经信号与目标语义特征之间的跨模态对齐。以往的语义解码器主要依赖于静态词汇表征或动态上下文表征的孤立使用。这种单维度的方法不可避免地导致严重的信息损失，因为它未能考虑人脑同时整合稳定的词属性和动态上下文的能力。为了解决这一问题，本研究引入了一种多特征融合框架用于非侵入性语义重建，系统性地基准测试了两种融合方法：线性朴素拼接（Naive Concatenation）和非线性多头交叉注意力（Multi-Head Cross-Attention）。在该框架内，我们的方法通过交互式门控机制将静态词汇表征（W2V）与动态上下文表征（GPT）相结合，以促进语言理解过程中的协同处理。通过广泛的语义重建和文本生成实验进行评估，我们的框架揭示了一个稳健的性能层级：Cross-Att > Concat > GPT > W2V。重要的是，非线性交叉注意力融合方法实现了最先进的性能，表明神经语言解码受益于模拟上下文信息与核心词汇属性之间的协同调制，而不是依赖孤立的个体特征，同时也提供了一种可行的非侵入性脑到文本解码方法。
cs.CL / 14 / 2607.12079
The Capacity of Thought: Benchmarking Llama 3.2 in Semantic fMRI Neural Language Decoding and Improving the Huth Encoding-Model Baseline
思维的容量：在语义fMRI神经语言解码中基准测试Llama 3.2并改进Huth编码模型基线
Abstract
Decoding continuous language from fMRI signals remains a core challenge in non-invasive brain-computer interface research. We present two complementary investigations. First, we improve the Huth et al. ridge regression encoding pipeline through expanded voxel selection (10K->15K), substitution of GPT-2 medium for GPT-1 as the beam-search proposal model, and GPU-accelerated bootstrap training, achieving mean METEOR = 0.149 and BLEU-1 = 0.200 across three held-out narratives for subject UTS03 -- an 11% relative METEOR gain over our replication baseline. Second, we introduce fMRIFlamingo, which maps BOLD activity to a frozen Llama-3.2-1B with trainable gated cross-attention layers via a learned brain tokenizer and a Perceiver Resampler. Despite achieving 42.86% Top-1 accuracy on a 1-in-100 ranking task, well above chance, a blind control ablation with zeroed fMRI inputs yields near-identical scores, revealing that apparent decoding success is driven primarily by the frozen language prior rather than by neural input. These results demonstrate that high-capacity language models do not inherently improve fMRI decoding and can actively obscure failures without rigorous blind-control evaluation.
Chinese Translation
从fMRI信号中解码连续语言仍然是非侵入式脑机接口研究中的核心挑战。我们提出了两个互补的研究。首先，我们通过扩展体素选择（从10K到15K）、用GPT-2中型替代GPT-1作为束搜索提案模型，以及GPU加速的自助训练，改进了Huth等人的岭回归编码管道，实现了在被试UTS03的三条保留叙述中，平均METEOR = 0.149和BLEU-1 = 0.200，较我们的复制基线有11%的相对METEOR提升。其次，我们引入了fMRIFlamingo，它通过学习的脑分词器和Perceiver Resampler将BOLD活动映射到一个具有可训练门控交叉注意力层的冻结Llama-3.2-1B上。尽管在1/100排名任务中达到了42.86%的Top-1准确率，远高于随机水平，但进行的盲控消融实验中，零化fMRI输入的得分几乎相同，揭示出明显的解码成功主要是由冻结的语言先验驱动，而非神经输入。这些结果表明，高容量语言模型并不固有地改善fMRI解码，并且在没有严格的盲控评估的情况下，可能会主动掩盖失败。
cs.CL / 15 / 2607.12086
CityBehavEx: A Scalable and Empirically Validated LLM-Assisted Urban Simulation Platform
CityBehavEx：一个可扩展且经过实证验证的 LLM 辅助城市模拟平台
Abstract
Recent LLM-based multi-agent urban simulators can generate semantically rich city routines, but they remain costly to scale and are often weakly validated against empirical mobility patterns. We present CityBehavEx, an interactive LLM-assisted urban simulation platform that scales to city-size populations, exposes agent behavior for inspection, supports empirical validation, and generates mobility patterns that better match real-world spatial, temporal, and semantic distributions. Instead of invoking large language models for every agent action, CityBehavEx combines established human mobility models with fine-tuned cross-encoders that estimate semantic alignment between agent profiles, schedules, and activity transitions. This design enables large-scale simulations, as demonstrated in a case study of 100,000 agents over 75 days in under one hour on a single consumer GPU. The platform allows users to define simulation regions, launch experiments, inspect trajectories and activity traces, debug unrealistic behaviors, and validate generated routines against real-world mobility, time-use, and semantic metrics.
Chinese Translation
最近基于 LLM 的多智能体城市模拟器能够生成语义丰富的城市日常活动，但在扩展性上仍然成本高昂，并且往往与实证移动模式的验证较弱。我们提出了 CityBehavEx，一个互动的 LLM 辅助城市模拟平台，能够扩展到城市规模的人口，暴露智能体行为以供检查，支持实证验证，并生成更符合现实世界空间、时间和语义分布的移动模式。CityBehavEx 通过结合成熟的人类移动模型与经过微调的交叉编码器，估算智能体档案、日程和活动转变之间的语义一致性，而不是为每个智能体动作调用大型语言模型。这一设计使得大规模模拟成为可能，如在一个案例研究中，100,000 个智能体在 75 天内的模拟在单个消费级 GPU 上不到一小时完成。该平台允许用户定义模拟区域，启动实验，检查轨迹和活动痕迹，调试不现实的行为，并将生成的日常活动与现实世界的移动、时间使用和语义指标进行验证。
cs.CL / 16 / 2607.12161
Token Reduction Is Not Cost Reduction
令牌减少并不等于成本减少
Abstract
Context-reduction layers for API-based coding agents, including command-output compressors, retrieval rankers, and payload-optimizing proxies, are usually evaluated by how much text they remove. We ask instead: when does reducing retrieved context or tool output lower the actual billed cost of a coding agent without reducing task success or lengthening its trajectory? Our primary evidence is a pre-specified, hash-frozen, paired campaign of 2,908 provider-billed Claude Code runs, of which 2,848 were analyzed, covering 103 tasks, seven repositories, and three models. The campaign compared a baseline with two generations of hook-based compression and an API-boundary proxy, within a broader measured program of roughly 5,500 billed executions. Three findings emerge. First, prompt-cache traffic dominated cost composition. Cache creation and reads accounted for about 87% of reconstructed four-component cost, or about 80% of the actual bill, with an 8.7% dollar-weighted residual that retained telemetry could not attribute. On Haiku 4.5, this residual scaled with thinking effort. Second, tool-output reduction did not reliably predict billed-cost reduction. An arm that removed 38% of estimated raw tool-output tokens had 6.8% higher paired cost (95% CI: +2.8% to +11.3%), while per-task reduction was only weakly associated with cost change (Pearson r = 0.15, CI crossing zero). Third, compression can harm task completion by removing action-critical evidence. In a small single-shot study on SWE-bench-derived Go tasks, compression reduced patch application from 27/40 to 15/40 by corrupting verbatim edit anchors, and the compressed grounded arm produced fewer solves at higher observed cost per solve. We propose a layered evidence standard centered on success-adjusted billed cost rather than token reduction alone.
Chinese Translation
基于API的编码代理的上下文减少层，包括命令输出压缩器、检索排名器和有效载荷优化代理，通常通过它们移除的文本量来评估。我们反而提出了一个问题：在不降低任务成功率或延长其轨迹的情况下，减少检索的上下文或工具输出何时能降低编码代理的实际计费成本？我们的主要证据是一项预先指定、哈希冻结的配对实验，共计2,908次提供商计费的Claude Code运行，其中分析了2,848次，涵盖了103个任务、七个代码库和三个模型。该实验在大约5,500次计费执行的更广泛测量程序中比较了基线与两代基于钩子的压缩和一个API边界代理。得出了三个发现。首先，提示缓存流量主导了成本组成。缓存的创建和读取占重构的四部分成本的约87%，或约80%的实际账单，剩余的8.7%美元加权残差无法归因于保留的遥测。在Haiku 4.5上，这一残差与思考努力成正比。其次，工具输出的减少并不能可靠地预测计费成本的降低。一个去除了38%估计原始工具输出令牌的实验组，其配对成本高出6.8%（95% CI：+2.8%至+11.3%），而每个任务的减少与成本变化的关联仅为微弱（Pearson r = 0.15，CI穿越零）。第三，压缩可能通过移除关键证据来损害任务完成。在一项针对SWE-bench派生的Go任务的小型单次研究中，压缩将补丁应用从27/40减少到15/40，因而破坏了逐字编辑锚点，而压缩后的有根臂在每次解决的观察成本更高的情况下产生了更少的解决方案。我们建议采用以成功调整的计费成本为中心的分层证据标准，而不仅仅是令牌减少。
cs.CL / 17 / 2607.12169
We Hebben Een Serieus Translatie: Modeling Intercomprehension as Probabilistic Inference
我们有一个严肃的翻译：将互理解建模为概率推理
Abstract
Intercomprehension refers to partial intelligibility of an unfamiliar language (L2) by a speaker of a related language (L1). How is this zero-shot cross-language comprehension possible? In this work, we extend past work on algorithmic models of noisy-channel inference to model intercomprehension in a Bayesian framework. The model uses an LM in L1 only for scoring latent hypotheses about the translations of observed L2 utterances, and a general-purpose noise model to infer a mapping between L2 and L1 words based on either form-based similarity or symbolic rules. We then conduct a human behavioral experiment, eliciting inferences for utterances in Dutch, Italian, and Ukrainian from speakers of English, Spanish, and Russian, respectively. Our full model shows a closer alignment to the distribution of human intercomprehension performance than ablations, and also compares favorably to zero-shot prompting of much larger models. These results provide a cognitively plausible computational model of intercomprehension, and highlight the flexible inferences made by comprehenders under wide uncertainty in real-world cross-language scenarios. We share our code publicly.
Chinese Translation
互理解是指说一种相关语言（L1）的说话者对一种不熟悉语言（L2）的部分理解。这种零样本跨语言理解是如何实现的？在本研究中，我们扩展了过去关于噪声信道推理的算法模型的研究，将互理解建模为贝叶斯框架下的模型。该模型仅使用L1中的语言模型（LM）来评分关于观察到的L2话语翻译的潜在假设，并使用通用噪声模型根据形式相似性或符号规则推断L2与L1单词之间的映射。随后，我们进行了一个人类行为实验，分别从英语、西班牙语和俄语的说话者中引导出荷兰语、意大利语和乌克兰语话语的推理。我们的完整模型与人类互理解表现的分布更为一致，相较于消融实验也表现良好，并且与对更大模型的零样本提示相比也有优势。这些结果提供了一个在认知上合理的互理解计算模型，并突显了在现实世界跨语言场景中理解者在广泛不确定性下所做的灵活推理。我们公开分享了我们的代码。
cs.CL / 18 / 2607.12185
Entropy in Semantic Memory Navigation in Blind and Sighted Individuals: The Effect of Visual Experience
盲人和有视力个体的语义记忆导航中的熵：视觉经验的影响
Abstract
Embodied accounts of semantic memory highlight the role of sensorimotor systems in acquiring and storing knowledge. Congenitally blind populations offer a critical test bed for these assumptions, providing an opportunity to assess whether conceptual grounding requires visual experience. In this study, we assessed semantic memory navigation differences between blind and sighted individuals using a property listing task with concrete and abstract concepts. We computed semantic entropy, an embedding-based natural language processing metric that captures the predictability of retrieval. Generalized linear mixed models revealed distinct navigation patterns across groups: while sighted individuals showed higher entropy for abstract than concrete concepts, blind participants did not. Instead, blind individuals exhibited higher entropy for visually salient concrete concepts (e.g., penguin). These results underscore the role of visual experience in the organization and dynamic navigation of semantic memory.
Chinese Translation
具身语义记忆理论强调感知运动系统在知识获取和存储中的作用。先天失明人群为这些假设提供了关键的实验基础，提供了评估概念基础是否需要视觉经验的机会。在本研究中，我们通过使用具体和抽象概念的属性列举任务评估了盲人和有视力个体之间的语义记忆导航差异。我们计算了语义熵，这是一种基于嵌入的自然语言处理指标，用于捕捉检索的可预测性。广义线性混合模型揭示了各组之间的不同导航模式：有视力个体在抽象概念上的熵高于具体概念，而盲人参与者则没有这种情况。相反，盲人个体在视觉显著的具体概念（例如，企鹅）上表现出更高的熵。这些结果强调了视觉经验在语义记忆的组织和动态导航中的作用。
cs.CL / 19 / 2607.12195
Comparing Semantic Navigation in Humans and Large Language Models using Natural Language Processing
通过自然语言处理比较人类与大型语言模型的语义导航
Abstract
Semantic memory retrieval can be conceptualized as navigation through conceptual space. We compared semantic search dynamics between humans and three large language models (GPT-4o, Gemini-2.5-Pro, Claude-Sonnet-4.5) using verbal fluency data. By applying trajectory-based NLP metrics to the items generated by 82 human participants and LLM output across eight temperature settings, we quantified three complementary dimensions: entropy (step size predictability), distance to next (successive semantic steps), and distance to centroid (global dispersion). Humans exhibited higher entropy, larger semantic steps and broader dispersion than all LLMs, indicating more variable and exploratory search. Temperature tuning produced only partial alignments, as individual metrics matched between humans and LLMs at specific settings, but no configuration reproduced the complete human profile (in all dimensions). These findings suggest that human semantic search implements a distinctive balance between local exploitation and global exploration that current model architectures fail to reproduce.
Chinese Translation
语义记忆检索可以被概念化为在概念空间中的导航。我们使用语言流畅性数据比较了人类与三种大型语言模型（GPT-4o、Gemini-2.5-Pro、Claude-Sonnet-4.5）之间的语义搜索动态。通过将基于轨迹的自然语言处理指标应用于82名参与者生成的项目和在八个温度设置下的LLM输出，我们量化了三个互补维度：熵（步长可预测性）、下一个距离（连续语义步骤）和质心距离（全局分散）。人类表现出比所有LLM更高的熵、更大的语义步骤和更广泛的分散，表明其搜索更具变异性和探索性。温度调节仅产生部分一致性，因为在特定设置下，人类与LLM之间的个别指标匹配，但没有配置能够重现完整的人类特征（在所有维度上）。这些发现表明，人类的语义搜索在局部利用与全局探索之间实现了一种独特的平衡，而当前的模型架构未能重现这一点。
cs.CL / 20 / 2607.12215
Fine-Tuned Multi-Agent Framework for Detecting OCEAN in Life Narratives
用于检测生活叙事中OCEAN人格特征的微调多智能体框架
Abstract
Accurately assessing personality from text is challenging because traits are latent, context-dependent, and often subtly expressed across long narratives. Large language models (LLMs) offer new opportunities by processing extensive textual contexts, but pretraining of these models can induce latent "personality-like" biases, making single-model inferences inconsistent. We propose a fine-tuned multi-agent framework for detecting OCEAN personality traits, in which sub-agents are conditioned to adopt high, low, or neutral perspectives for each trait through masked language modeling (MLM) and psychometric supervision. A judge LLM aggregates and compares sub-agent outputs to generate final trait predictions, capturing multiple complementary perspectives while mitigating individual model biases. We evaluate the framework on life narrative dataset through quantitative and qualitative experiments, including baselines, ablations, and inference quality analyses. Our approach offers a scalable and interpretable method for text-based personality inference, highlighting the benefits of multi-agent reasoning grounded in psychometric supervision.
Chinese Translation
从文本中准确评估人格特征是一项挑战，因为特征是潜在的、依赖于上下文的，并且通常在长篇叙事中以微妙的方式表达。大型语言模型（LLMs）通过处理广泛的文本上下文提供了新的机会，但这些模型的预训练可能会引入潜在的“人格类似”偏见，使得单一模型的推断不一致。我们提出了一种微调的多智能体框架，用于检测OCEAN人格特征，其中子智能体通过掩蔽语言建模（MLM）和心理测量监督被条件化为对每个特征采取高、低或中立的视角。一个评判LLM汇总并比较子智能体的输出，以生成最终的特征预测，捕捉多个互补的视角，同时减轻个体模型的偏见。我们通过定量和定性实验对该框架在生活叙事数据集上的表现进行了评估，包括基线、消融实验和推断质量分析。我们的方法提供了一种可扩展且可解释的基于文本的人格推断方法，突显了基于心理测量监督的多智能体推理的优势。
cs.CL / 21 / 2607.12216
RCWT: Measuring Task-Budget Displacement from Coordination Content in LLM Calls
RCWT：测量大规模语言模型调用中协调内容的任务预算位移
Abstract
Multi-agent and memory-augmented LLM systems often place coordination content, shared state, prior discussion, tool outputs, summaries, and role instructions, inside the same finite prompt used for the current task. This creates a practical allocation problem: every token spent on coordination is unavailable to task instructions or evidence when a call is assembled under a fixed context budget. We introduce the Roundtable Context Window Test (RCWT), a controlled protocol for measuring this task-budget displacement effect. RCWT varies coordination content while controlling total budget, position order, task family, and scoring. In the main context-dependent recall task at $W=4096$, three commercial models remain near baseline through moderate overhead and then degrade sharply once residual reference evidence falls to a few hundred tokens. Window-scaling summaries are consistent with a task-specific residual-budget interpretation rather than a fixed percentage threshold, but we treat this as descriptive evidence rather than a universal law. To test whether the fixed-budget cliff persists when task evidence remains intact, we add an intact-task ablation: the full task/reference block is kept present while coordination tokens increase by expanding total prompt length. In that setting, all tested calls return every scored field correctly across GPT-4.1-mini, Claude Haiku 4.5, and Gemini 2.5 Flash up to a 95\% coordination ratio. This ablation narrows the claim: the main RCWT cliff is best read as task-budget displacement, not as proof that coordination volume alone causes semantic interference in the original open-ended task. RCWT is therefore a measurement primitive for context-allocation budgeting, not a complete theory of multi-agent benefit or session-level coordination.
Chinese Translation
多智能体和记忆增强的大规模语言模型（LLM）系统通常将协调内容、共享状态、先前讨论、工具输出、摘要和角色指令放置在用于当前任务的同一有限提示中。这造成了一个实际的分配问题：在固定的上下文预算下，所有用于协调的令牌都无法用于任务指令或证据。当调用在固定上下文预算下组装时，我们引入了圆桌上下文窗口测试（Roundtable Context Window Test，RCWT），这是一种测量任务预算位移效应的控制协议。RCWT在控制总预算、位置顺序、任务类别和评分的同时，变化协调内容。在主要的上下文依赖回忆任务中，$W=4096$时，三个商业模型在适度的开销下保持接近基线，随后在残余参考证据降至几百个令牌时急剧下降。窗口缩放摘要与特定任务的残余预算解释一致，而不是固定百分比阈值，但我们将其视为描述性证据，而非普遍法则。为了测试当任务证据保持完整时，固定预算的悬崖是否仍然存在，我们添加了一个完整任务消融实验：在协调令牌通过扩展总提示长度增加的同时，保持完整的任务/参考块存在。在这种情况下，所有测试的调用在GPT-4.1-mini、Claude Haiku 4.5和Gemini 2.5 Flash中均正确返回每个评分字段，协调比例高达95%。这一消融实验缩小了声明的范围：主要的RCWT悬崖最好被理解为任务预算位移，而不是证明协调量本身导致原始开放式任务中的语义干扰。因此，RCWT是上下文分配预算的测量原语，而不是多智能体效益或会话级协调的完整理论。
cs.CL / 22 / 2607.12233
Fin-Analyst at FinMMEval 2026 Task 3: A Live Hybrid Trading Agent with LLM Specialists and Rule-Based Signals
Fin-Analyst在FinMMEval 2026任务3中的表现：一个结合LLM专家和基于规则信号的实时混合交易代理
Abstract
Large language model (LLM) trading agents show promising performance in equity markets, yet remain narrowly focused on US equities with little evidence from live deployment. We present Fin-Analyst, a hybrid agent for FinMMEval 2026 Task 3: an eight-specialist LLM pipeline over news, SEC filings, fundamentals, analyst forecasts, technical indicators, and social sentiment, aggregated by a Meta-Agent for Tesla (TSLA), and a lightweight rule based three-signal vote for Bitcoin (BTC). On the final official leaderboard (accessed 2026-07-05), Fin-Analyst ranks first of all agents on TSLA with a +13.51% return, +28.33 points over Buy-and-Hold (Sharpe 4.10, 88% win rate), while the BTC vote ends flat yet well above a sharply falling baseline. Relative to the interim performance, the asset ranking reversed, indicating that short live windows yield volatility-sensitive rankings. Ablation identifies event-driven 8-K disclosures as the most influential TSLA signal. Error analysis shows that the memoryless agents repeat wrong calls for days at a time, and that the fixed-threshold BTC rules lost money by trading on noise in a sideways market while the LLM pipeline gained under similar conditions, motivating a memory-aware, LLM-based successor for both assets.
Chinese Translation
大型语言模型（LLM）交易代理在股票市场表现出良好的前景，但仍然主要集中在美国股票上，缺乏实际部署的证据。我们提出了Fin-Analyst，一个用于FinMMEval 2026任务3的混合代理：一个涵盖新闻、SEC文件、基本面、分析师预测、技术指标和社会情绪的八位专家LLM管道，由一个针对特斯拉（Tesla, TSLA）的元代理进行聚合，并为比特币（Bitcoin, BTC）提供一个轻量级的基于规则的三信号投票。在最终的官方排行榜上（访问日期：2026-07-05），Fin-Analyst在TSLA上排名第一，回报率为+13.51%，比买入并持有策略高出28.33点（夏普比率4.10，胜率88%），而BTC投票则保持平稳，远高于急剧下跌的基准。与中期表现相比，资产排名发生了逆转，表明短期的实时窗口会导致波动敏感的排名。消融实验表明，事件驱动的8-K披露是影响TSLA的最重要信号。错误分析显示，无记忆代理在多天内重复错误的交易决策，而固定阈值的BTC规则在横盘市场中因噪音交易而亏损，而LLM管道在类似条件下获得收益，这促使我们为这两种资产开发一个具有记忆意识的LLM基础后续模型。
cs.CL / 23 / 2607.12252
FinResearchBench II: A Deep Research Benchmark with Consensus-Derived Gold Rubrics for Distinguishing Financial Report Quality
FinResearchBench II：一个基于共识导出的金标准评估指标的深度研究基准，用于区分财务报告质量
Abstract
Deep research agents are increasingly used to produce long-form financial reports, yet large-scale evaluation remains bottlenecked by the need for human experts to define and execute high-quality rubrics. We address this problem by proposing a scalable pipeline for generating high-quality rubrics without human experts in the final loop. We build a financial deep research benchmark from 104 real-world user queries and automatically synthesize 14,450 query-specific candidate rubrics from model-generated reports. To justify removing human experts from rubric execution, we compare rubric judgments from three human experts with those from a three-LLM judge panel on a sampled subset, and show that LLM-based evaluation is sufficiently consistent with human evaluation to replace it for large-scale rubric screening, including 98.67\% label-level agreement on jointly unanimous items. We then derive consensus-derived gold rubrics through two filters: a strict consistency filter, which keeps a rubric only if the three LLM judges unanimously agree on every report under the same query, and a distinguishability filter, which keeps a rubric only if it assigns at least one majority-yes and at least one majority-no label across the evaluated systems. This process retains 3,687 consistency-passed rubrics, of which 2,600 remain distinguishable and form the final set of consensus-derived gold rubrics. Using this final rubric set, we obtain clearly differentiated rankings across 10 deep research systems, with item-level pass rates ranging from 58.58\% to 22.23\%. More broadly, because the pipeline removes human-expert execution from rubric generation and evaluation, it is naturally scalable for benchmark evaluation, automatic system comparison, and future studies of evaluation-driven system improvement.
Chinese Translation
深度研究代理越来越多地用于生成长篇财务报告，但大规模评估仍然受到需要人类专家定义和执行高质量评估指标的瓶颈。我们通过提出一个可扩展的管道来解决这个问题，该管道能够在最终环节中不依赖人类专家生成高质量的评估指标。我们从104个真实用户查询构建了一个财务深度研究基准，并从模型生成的报告中自动合成了14,450个特定查询的候选评估指标。为了证明可以将人类专家从评估指标的执行中移除，我们对一个抽样子集中的三位人类专家的评估与一个三位大型语言模型（LLM）评审小组的评估进行了比较，结果表明基于LLM的评估与人类评估具有足够的一致性，可以替代人类进行大规模的评估指标筛选，包括在共同一致的项目上达到98.67%的标签级一致性。随后，我们通过两个过滤器导出了基于共识的金标准评估指标：一个严格的一致性过滤器，仅在三位LLM评审一致同意同一查询下的每个报告时保留评估指标；一个可区分性过滤器，仅在评估的系统中至少分配一个多数“是”和一个多数“否”标签时保留评估指标。该过程保留了3,687个通过一致性过滤的评估指标，其中2,600个仍然可区分，并形成最终的基于共识的金标准评估指标集。使用这一最终评估指标集，我们在10个深度研究系统中获得了明显区分的排名，项目级通过率范围从58.58%到22.23%。更广泛地说，由于该管道从评估指标的生成和评估中移除了人类专家的执行，因此它在基准评估、自动系统比较和未来的评估驱动系统改进研究中具有自然的可扩展性。
cs.CL / 24 / 2607.12279
A Shared Subcircuit Lets LLMs Count Down Across Tasks
共享子电路使大型语言模型在任务间进行倒计时
Abstract
Writing a sentence of exactly twelve words; ending a DNA sequence at the right codon; formatting an ASCII table. These are all tasks that language models can do that requires tracking how many tokens remain before a target. In this work, we identify in Llama-3.1-70B-Instruct a general mechanism for performing these tasks: a "countdown subcircuit" that compares the current position to a goal length and estimates the time remaining until then. We first isolate a countdown subcircuit in a controlled setting, in which the model is tasked with writing a fixed-length sentence ending in a specified word. We then investigate the geometry of the representations used by the subcircuit, and find that the subcircuit uses an identical motif previously identified in a frontier LLM on a separate task, thus suggesting that this motif is shared across models. Finally, we use unsupervised probing on a natural language dataset to find a variety of other tasks where this subcircuit is used, including tasks where the goal length is inferred from context rather than explicitly stated. Our work suggests that reverse-engineering subcircuits allows us to understand how behaviors generalize from a single example to many different tasks and even models.
Chinese Translation
写出恰好十二个单词的句子；在正确的密码子处结束DNA序列；格式化ASCII表。这些都是语言模型能够完成的任务，要求跟踪在目标之前剩余多少个标记。在本研究中，我们在Llama-3.1-70B-Instruct中识别出一种执行这些任务的通用机制：一个“倒计时子电路”，它将当前位置与目标长度进行比较，并估算到达目标所需的时间。我们首先在一个受控环境中隔离出一个倒计时子电路，其中模型的任务是写出一个以指定单词结束的固定长度句子。然后，我们研究子电路所使用的表示的几何特征，发现该子电路使用了在另一个任务中先前识别出的前沿大型语言模型中的相同模式，从而表明这一模式在不同模型间是共享的。最后，我们在一个自然语言数据集上使用无监督探测，发现了多种其他任务，其中使用了该子电路，包括那些目标长度是从上下文推断而非明确说明的任务。我们的研究表明，逆向工程子电路使我们能够理解行为如何从单一示例推广到许多不同的任务甚至模型。
cs.CL / 25 / 2607.12310
LakeQuest: A Three-Domain Benchmark for Grounded Question Answering across Data Lakes
LakeQuest：一个跨数据湖的基于真实场景的问题回答三领域基准
Abstract
While modern question answering (QA) systems excel on clean, schema-aligned corpora, real-world knowledge is rarely so neatly packaged. Answering questions over enterprise and scientific data lakes requires systems to navigate heterogeneous, weakly structured collections of tables, passages, and linked metadata. Current benchmarks abstract away this noisy discovery process, failing to evaluate end-to-end performance. To bridge this gap, we introduce LakeQuest, a human-validated benchmark of 9,846 QA pairs designed to evaluate the end-to-end retrieve-and-synthesize pipeline over realistic data lakes. LakeQuest spans three diverse domains (AI/ML metadata, retail banking, and multimodal biomedical drug information) and pairs every question with exact, modality-aware evidence pointers. By isolating source discovery from cross-modal synthesis, LakeQuest exposes critical failure modes in modern QA systems. Our baseline evaluations, including standard Retrieval-Augmented Generation (RAG) and agentic tool-use methods, reveal that high-quality retrieval does not guarantee correct reasoning. Systems consistently struggle with relation chaining in metadata graphs, policy grounding in bank ledgers, and joint tabular QA in biomedical contexts, highlighting the need for robust discovery and faithful cross-file composition mechanisms in future agentic QA systems.
Chinese Translation
虽然现代问题回答（QA）系统在干净、与模式对齐的语料库上表现出色，但现实世界的知识往往并非如此整齐。对企业和科学数据湖中的问题进行回答需要系统能够在异构、弱结构化的表格、段落和链接元数据集合中进行导航。目前的基准忽略了这一嘈杂的发现过程，未能评估端到端的性能。为了解决这一问题，我们引入了LakeQuest，这是一个经过人工验证的基准，包含9,846对QA对，旨在评估在现实数据湖上进行端到端检索和综合的流程。LakeQuest涵盖三个不同的领域（人工智能/机器学习元数据、零售银行和多模态生物医学药物信息），并为每个问题配备精确的、考虑到模态的证据指针。通过将源发现与跨模态综合隔离，LakeQuest揭示了现代QA系统中的关键失败模式。我们的基线评估，包括标准的检索增强生成（RAG）和代理工具使用方法，显示高质量的检索并不保证正确的推理。系统在元数据图中的关系链、银行账本中的政策基础以及生物医学背景下的联合表格QA方面始终面临挑战，这突显了未来代理QA系统中对强大发现和忠实跨文件组合机制的需求。
cs.CL / 26 / 2607.12334
QUBO-Optimized Evidence Selection for Retrieval-Augmented Question Answering with Unconventional Solvers
基于QUBO优化的证据选择用于增强检索的问答系统与非常规求解器
Abstract
Retrieval-augmented question answering depends on selecting evidence passages that jointly support answer generation. However, many RAG pipelines rely on top-\(k\) ranking, where passages are selected mainly by individual relevance scores, even though multi-hop questions often require complementary evidence satisfying multiple information requirements. Recent LLM-based selectors address this by treating retrieval as set selection, but using an LLM for this intermediate stage can be costly and difficult to scale. In this work, we formulate evidence selection as a Quadratic Unconstrained Binary Optimization (QUBO) problem. Given a question, candidate passages, and decomposed information requirements, our method constructs an energy function that balances relevance, requirement coverage, support strength, redundancy, complementarity, and compactness. Low-energy solutions correspond to compact evidence subsets that cover the needed requirements while avoiding unnecessary or repetitive context. The selected passages are then passed to a downstream language model for answer generation, separating combinatorial evidence selection from semantic answer generation. We evaluate the proposed QUBO selector on HotpotQA and compare it with LLM-based set selectors and non-LLM baselines including BM25, relevance top-\(k\), maximal marginal relevance, hybrid lexical--semantic ranking, greedy coverage, and random selection. The QUBO selector achieves competitive exact-match and token-F1 performance relative to LLM-based selectors while providing a solver-compatible formulation for structured evidence selection. These results suggest that multi-hop evidence selection can be cast as discrete optimization, opening a path toward RAG pipelines where LLMs are reserved for semantic processing and answer generation, while context selection is handled by Ising/QUBO-compatible solvers.
Chinese Translation
增强检索的问答系统依赖于选择共同支持答案生成的证据段落。然而，许多RAG（检索增强生成）流程依赖于前k名排名，其中段落主要通过单个相关性评分进行选择，尽管多跳问题通常需要满足多个信息需求的互补证据。最近的基于大型语言模型（LLM）的选择器通过将检索视为集合选择来解决这个问题，但在这个中间阶段使用LLM可能成本高昂且难以扩展。在本研究中，我们将证据选择形式化为一个二次无约束二进制优化（QUBO）问题。给定一个问题、候选段落和分解的信息需求，我们的方法构建了一个能量函数，平衡相关性、需求覆盖、支持强度、冗余、互补性和紧凑性。低能量解对应于覆盖所需需求的紧凑证据子集，同时避免不必要或重复的上下文。然后，将选定的段落传递给下游语言模型进行答案生成，从而将组合证据选择与语义答案生成分开。我们在HotpotQA上评估了所提出的QUBO选择器，并将其与基于LLM的集合选择器和包括BM25、相关性前k名、最大边际相关性、混合词汇-语义排名、贪婪覆盖和随机选择在内的非LLM基线进行比较。QUBO选择器在精确匹配和token-F1性能方面相对于基于LLM的选择器表现出竞争力，同时为结构化证据选择提供了求解器兼容的形式。这些结果表明，多跳证据选择可以被视为离散优化，为RAG流程开辟了一条道路，在该流程中，LLM被保留用于语义处理和答案生成，而上下文选择则由Ising/QUBO兼容的求解器处理。
cs.CL / 27 / 2607.12336
Evaluating Health Misinformation in Low-Resource Languages: Integrating Small Language Models with a Culturally-Sensitive Responsible NLP Framework (Bangla as a Case Study)
评估低资源语言中的健康错误信息：将小型语言模型与文化敏感的负责任自然语言处理框架相结合（以孟加拉语为案例研究）
Abstract
Artificial Intelligence (AI) technologies, while serving as a foundational enabler for modern social media and digital health services, exert a bivalent effect by simultaneously acting as a combatant against and a spread vector for misinformation. A prevalent challenge in mitigating this issue arises in non-English contexts and low socioeconomic classes, where limited data hinders the training of AI models for effective detection. Consequently, culturally and linguistically diverse (CALD) communities struggle to access trustworthy health information through AI-driven tools. Current AI tools underperform due to a lack of training data and are largely unable to consider language nuances and traditions in non-English contexts. This research addresses these gaps by proposing a CALD-friendly AI-based health misinformation detector and providing a dashboard for medical professionals to analyse this misinformation, a critical step toward mitigating a growing concern among CALD populations. To this end, we conduct a series of experiments using a Bangla-translated health misinformation dataset to evaluate the performance of various Small Language Models (SLMs). SLMs are particularly relevant in this context given the frequent underperformance of Large Language Models (LLMs), which often stems from insufficient domain-specific knowledge and the prohibitive costs of resource-intensive fine-tuning. The results demonstrate that Phi-4 is the superior model, achieving an ideal balance between precision and recall in claim extraction. Then, to mitigate the limitations of SLMs, we design and test a novel health misinformation detection framework grounded in Responsible Natural Language Processing (NLP), which incorporates cultural sensitivity, potential for harm, and communication quality, thereby providing a holistic lens for evaluating misinformation in low-resource languages.
Chinese Translation
人工智能（AI）技术作为现代社交媒体和数字健康服务的基础性推动力，发挥着双重作用，既是对抗错误信息的斗士，又是其传播的载体。在非英语环境和低社会经济阶层中，缓解这一问题的主要挑战在于有限的数据阻碍了AI模型的有效检测训练。因此，文化和语言多样性（CALD）社区在通过AI驱动的工具获取可信的健康信息方面面临困难。由于缺乏训练数据，目前的AI工具表现不佳，且在非英语环境中往往无法考虑语言的细微差别和传统。本文研究通过提出一个适合CALD的基于AI的健康错误信息检测器，并为医疗专业人员提供一个分析这些错误信息的仪表板，来填补这些空白，这是缓解CALD人群日益关注的关键一步。为此，我们使用一个孟加拉语翻译的健康错误信息数据集进行一系列实验，以评估各种小型语言模型（SLMs）的性能。考虑到大型语言模型（LLMs）在特定领域知识不足和资源密集型微调成本高昂的情况下，SLMs在此背景下尤为相关。结果表明，Phi-4是最佳模型，在声明提取中实现了精确度与召回率的理想平衡。随后，为了缓解SLMs的局限性，我们设计并测试了一个基于负责任自然语言处理（NLP）的新型健康错误信息检测框架，该框架结合了文化敏感性、潜在危害和沟通质量，从而为评估低资源语言中的错误信息提供了一个整体视角。
cs.CL / 28 / 2607.12341
Policy-Conditioned Constrained Decoding for Column-Level Access Control in Text-to-SQL
基于策略约束解码的文本到SQL列级访问控制
Abstract
Text-to-SQL is increasingly deployed across trust boundaries between data providers and users. Such deployment must balance three competing requirements: policy compliance, answer coverage, and bounded cost. Existing approaches typically decide refusal based on which columns a query mentions and enforce it stochastically. Whether a query is compliant, however, depends not only on which columns appear but on how they are used, and stochastic enforcement cannot deterministically rule out violations. We formalize this requirement as a column-use policy over semantic use: output, filter condition, and aggregation argument. We integrate the policy by aligning each role with grammar productions tracked by the decoder. The resulting system, PCC-SQL, applies a per-token logits mask that deterministically eliminates single-query column-use violations on the supported SQL fragment in a single decoding pass. Across three benchmarks and three open-source models, PCC-SQL achieves 0% Leakage Rate and Coverage up to 88.7% on Spider-CU, while staying within +10% tokens of direct prompting. We additionally assess semantic alignment with execution accuracy.
Chinese Translation
文本到SQL的应用越来越多地跨越数据提供者与用户之间的信任边界。这种部署必须平衡三个相互竞争的要求：政策合规性、答案覆盖率和成本限制。现有的方法通常根据查询提及的列来决定拒绝，并以随机方式执行。然而，查询是否合规不仅取决于出现的列，还取决于这些列的使用方式，而随机执行无法确定性地排除违规情况。我们将这一要求形式化为基于语义使用的列使用政策：输出、过滤条件和聚合参数。我们通过将每个角色与解码器跟踪的语法生成式对齐来整合该政策。最终系统PCC-SQL应用了每个标记的logits掩码，确定性地消除了在单次解码过程中对支持的SQL片段的单查询列使用违规。在三个基准测试和三个开源模型中，PCC-SQL实现了0%的泄漏率和高达88.7%的Spider-CU覆盖率，同时令直接提示的标记数保持在+10%以内。我们还评估了与执行准确性的语义对齐。
cs.CL / 29 / 2607.12383
Beyond Binary Detection: A Multi-Dimensional Taxonomy of Cancer Misinformation on Reddit
超越二元检测：关于Reddit上癌症误信息的多维分类法
Abstract
Cancer-related discussions on social media provide an important space for information exchange and peer support, but also facilitate the spread of misinformation that may influence prevention, screening, and treatment decisions. Existing research on cancer misinformation often relies on narrow definitions, small-scale datasets, or binary labeling frameworks. We introduce a multi-dimensional taxonomy for characterizing cancer misinformation in Reddit discussions of breast, lung, colon, and prostate cancer. The taxonomy captures seven dimensions, including misinformation presence, information type, risk level, stance, and topical focus. Using expert-annotated data, we evaluate multiple large language models (LLMs) for scalable misinformation annotation and analyze cancer misinformation across Reddit communities. Our results show that cancer-related misinformation constitutes approximately 6\% of Reddit cancer discussions, with substantial variation across communities and misinformation topics. Few-shot prompting substantially improves classification performance, particularly for nuanced taxonomy dimensions. We additionally identify recurring misinformation narratives centered on unsupported treatments, distrust of conventional medicine, and misleading claims about diagnosis and screening. Our taxonomy, dataset, and findings provide a foundation for multi-dimensional modeling of online cancer misinformation.
Chinese Translation
社交媒体上与癌症相关的讨论为信息交流和同伴支持提供了重要空间，但也促进了可能影响预防、筛查和治疗决策的误信息传播。现有关于癌症误信息的研究往往依赖于狭隘的定义、小规模的数据集或二元标记框架。我们提出了一种多维分类法，用于描述Reddit上关于乳腺癌、肺癌、结肠癌和前列腺癌的讨论中的癌症误信息。该分类法捕捉了七个维度，包括误信息存在、信息类型、风险水平、立场和主题焦点。通过使用专家标注的数据，我们评估了多种大型语言模型（LLMs）在可扩展的误信息标注中的表现，并分析了Reddit社区中的癌症误信息。我们的结果显示，癌症相关的误信息约占Reddit癌症讨论的6%，在不同社区和误信息主题之间存在显著差异。少量示例提示显著提高了分类性能，尤其是在细微的分类维度上。我们还识别出围绕不支持的治疗、对传统医学的不信任以及关于诊断和筛查的误导性声明的反复出现的误信息叙事。我们的分类法、数据集和研究结果为在线癌症误信息的多维建模提供了基础。
cs.CL / 30 / 2607.12395
Ring-Zero: Scaling Zero RL to a Trillion Parameters for Emergent Reasoning
Ring-Zero：将零强化学习扩展至万亿参数以实现新兴推理
Abstract
Reinforcement learning with verifiable rewards without human-annotated data, often referred to as zero RL, has emerged as a powerful paradigm for eliciting chain-of-thought reasoning. However, due to computational constraints, existing studies are largely restricted to small models, leaving the training dynamics and emergent capabilities at a large scale unexplored. To meaningfully explore this frontier, we aim to elicit high-quality reasoning behaviors from the model. However, we find that naive scaling often suffers from poor readability, token redundancy, and a lack of adaptive reasoning depth. To address these challenges, we present a stable and efficient training pipeline, incorporating algorithmic and system optimizations such as clipped importance sampling, training-inference ratio correction, and mixed-precision control. Our experiments offer three key findings that validate the "bitter lesson" of scaling: (1) scaling to 1T parameters significantly enhances sample efficiency and performance ceilings; (2) the training process progresses sequentially through an initial discovery phase followed by a sharpening phase; and (3) the model spontaneously develops advanced cognitive behaviors, including anthropomorphism, structured formatting, self-verification, parallel reasoning, and context anxiety, rendering hand-crafted heuristics redundant. Evaluated on seven mathematical benchmarks, Ring-2.5-1T-Zero achieves competitive performance. Additionally, to assess CoT quality beyond final-answer correctness, we propose a structured evaluation framework across three dimensions: comprehensibility, reproducibility, and efficiency, where our model demonstrates clear advantages in producing structured and concise reasoning traces. By sharing our observed emergent phenomena, we hope to provide the community with deeper insights into scaling behaviors, particularly at the 1-trillion scale.
Chinese Translation
无须人工标注数据的可验证奖励强化学习，通常被称为零强化学习（zero RL），已成为引发连锁思维推理的强大范式。然而，由于计算限制，现有研究主要局限于小型模型，尚未探索大规模下的训练动态和新兴能力。为了有意义地探索这一前沿，我们旨在从模型中引发高质量的推理行为。然而，我们发现，简单的扩展往往会遭遇可读性差、标记冗余和缺乏自适应推理深度等问题。为了解决这些挑战，我们提出了一种稳定高效的训练流程，结合了算法和系统优化，如剪切重要性采样（clipped importance sampling）、训练-推理比率校正和混合精度控制。我们的实验提供了三个关键发现，验证了扩展的“苦涩教训”： (1) 扩展至1万亿参数显著提高了样本效率和性能上限； (2) 训练过程依次经历初始发现阶段和后续的锐化阶段； (3) 模型自发发展出高级认知行为，包括拟人化、结构化格式、自我验证、并行推理和上下文焦虑，使得手工设计的启发式方法变得多余。在七个数学基准上评估时，Ring-2.5-1T-Zero表现出竞争力。此外，为了评估连锁推理（CoT）质量超越最终答案的正确性，我们提出了一个跨越三个维度的结构化评估框架：可理解性、可重复性和效率，在这些维度上，我们的模型在生成结构化和简洁的推理痕迹方面表现出明显优势。通过分享我们观察到的新兴现象，我们希望为社区提供对扩展行为的更深入见解，特别是在万亿规模下。
cs.CL / 31 / 2607.12441
WikiSTAR: A System for Shedding Light on the Hidden History of Scientific Wikipedia Articles
WikiSTAR：揭示科学维基百科文章隐藏历史的系统
Abstract
Wikipedia plays a key role in shaping public understanding of science, and its openly accessible revision history is a unique record of how scientific knowledge evolves over time. Yet scientifically meaningful revisions are obscured by the sheer volume of routine edits, leaving each article's scientific history hidden. We present WikiSTAR (Scientific Tracking of Article Revisions), an interactive system for exploring scientifically meaningful changes across an article's revision history. Using an LLM classifier with an expert-designed multi-label taxonomy, WikiSTAR first tags edit types such as the addition of technical terms, new research findings, and changes in scientific narrative. Then, through interactive views, an article's full revision history can be traced at any granularity - from aggregate trends that reveal when and in which sections scientific content was added or refined, down to individual edits - showing how scientific knowledge develops at a scale previously impossible. In a user study, experts from three domains found that WikiSTAR surfaced new patterns and research questions and enabled previously impractical analyses. We release our system, code and a human-annotated benchmark.
Chinese Translation
维基百科在塑造公众对科学的理解中发挥着关键作用，其开放访问的修订历史是科学知识随时间演变的独特记录。然而，科学上有意义的修订常常被大量的常规编辑所掩盖，使得每篇文章的科学历史变得隐秘。我们提出了WikiSTAR（Scientific Tracking of Article Revisions），这是一个用于探索文章修订历史中科学上有意义变化的互动系统。WikiSTAR首先使用一个由专家设计的多标签分类法的LLM分类器标记编辑类型，例如技术术语的添加、新研究发现以及科学叙述的变化。然后，通过互动视图，可以以任意粒度追踪文章的完整修订历史——从揭示何时以及在哪些部分添加或完善科学内容的汇总趋势，到单个编辑——展示了科学知识如何在以前无法实现的规模上发展。在一项用户研究中，来自三个领域的专家发现WikiSTAR揭示了新的模式和研究问题，并使得以前不切实际的分析成为可能。我们发布了我们的系统、代码和一个人工标注的基准。
cs.CL / 32 / 2607.12443
Language Identification with Succinct Machine-Independent Traces
使用简洁的机器无关痕迹进行语言识别
Abstract
Motivated by the power of large language models, there has been renewed interest in the Gold-Angluin model of language identification in the limit, with an eye toward variants of the model that might overcome the negative results for its original formulation. Recent papers on this question have proposed looking at computational traces and annotations of training strings as a source of additional power for a learner, reflecting empirical regularities such as the way that commented source code is easier to learn from than arbitrary source code, and text annotated with algorithmically generated chain-of-thought tokens can be easier to learn from than the raw text itself. This recent work has shown positive results for language identification in the presence of such computational traces, but the traces in these positive results come from explicit automata-theoretic machine models that generate the language, where the underlying vocabulary of tokens for the traces is very large. In this paper, we address two fundamental issues left open by this line of work: can we achieve positive results with traces that use only a small alphabet, and can we define traces directly from the language itself, without requiring an underlying machine model that generates it? We establish positive results for both of these questions: for an arbitrary collection of languages, we show how to define computational traces that enable identification in the limit, using an alphabet of tokens that is linear in the size of the alphabet that the languages are defined over, and independent of any other properties of the languages.
Chinese Translation
受大型语言模型的启发，语言识别的Gold-Angluin模型在极限情况下重新引起了关注，尤其是针对可能克服其原始表述负面结果的模型变体。最近关于这个问题的论文提出，计算痕迹和训练字符串的注释可以作为学习者额外能力的来源，反映出经验规律，例如注释的源代码比任意源代码更易于学习，带有算法生成的思维链标记的文本比原始文本更易于学习。这项最新工作在存在这些计算痕迹的情况下显示了语言识别的积极结果，但这些积极结果中的痕迹来自于显式的自动机理论机器模型，这些模型生成的语言的基础标记词汇量非常庞大。本文解决了这一系列研究所留下的两个基本问题：我们能否仅使用小字母表的痕迹获得积极结果，以及我们能否直接从语言本身定义痕迹，而不需要生成它的基础机器模型？我们为这两个问题建立了积极结果：对于任意语言集合，我们展示了如何定义计算痕迹，使其能够在极限情况下进行识别，使用的标记字母表大小与语言定义所用字母表的大小成线性关系，并且与语言的其他属性无关。
cs.CL / 33 / 2607.12612
Translation as a Computationally Efficient Bridge: Feasibility of English BERT for Low-Resource Languages
翻译作为一种计算高效的桥梁：英语 BERT 在低资源语言中的可行性
Abstract
BERT models have revolutionised Natural Language Processing (NLP) through their ability to process unstructured text across diverse domains. However, developing high-quality BERT models for non-English languages remains challenging due to limited annotated data and high computational demands. Translating non-English data into English and fine-tuning existing English BERT models offers a resource-efficient alternative, yet few studies have structurally compared translation-based fine-tuning with native-language BERT performance across tasks and languages. This study provides such a comparison, evaluating the feasibility of translation-based fine-tuning across six NLP tasks: Sentiment Analysis, Hate Speech Detection, Question Answering, Named Entity Recognition, Part-of-Speech Tagging, and Natural Language Inference, using datasets translated from Bulgarian, Chinese, Dutch, Italian, and Russian. Across all settings, the translation-based approach was comparable or superior in 53.3 percent of cases. Gains were most frequent in Question Answering, Part-of-Speech Tagging, and Natural Language Inference, while performance declines were common in Named Entity Recognition and Hate Speech Detection. The results show that translation-based fine-tuning is most effective for tasks relying on syntactic or structural patterns and for languages typologically close to English, such as Dutch, but less effective for token-level or culturally nuanced tasks, particularly in Chinese. Overall, this study demonstrates that translation-based fine-tuning offers a scalable, resource-efficient, and empirically validated path for extending NLP to low-resource languages while advancing linguistic inclusivity and sustainability in artificial intelligence.
Chinese Translation
BERT 模型通过其处理多领域非结构化文本的能力，彻底改变了自然语言处理（NLP）。然而，由于标注数据有限和计算需求高，为非英语语言开发高质量的 BERT 模型仍然具有挑战性。将非英语数据翻译成英语并对现有的英语 BERT 模型进行微调提供了一种资源高效的替代方案，但很少有研究在结构上比较基于翻译的微调与母语 BERT 在不同任务和语言中的表现。本研究提供了这样的比较，评估了基于翻译的微调在六个 NLP 任务中的可行性：情感分析、仇恨言论检测、问答、命名实体识别、词性标注和自然语言推理，使用从保加利亚语、中文、荷兰语、意大利语和俄语翻译的数据集。在所有设置中，基于翻译的方法在 53.3% 的情况下表现出可比性或优越性。问答、词性标注和自然语言推理任务中获得的提升最为频繁，而命名实体识别和仇恨言论检测的表现下降则较为普遍。结果表明，基于翻译的微调在依赖于句法或结构模式的任务中最为有效，并且在与英语类型学上接近的语言（如荷兰语）中效果更佳，但在以标记为单位或文化细微差别的任务中（尤其是在中文中）效果较差。总体而言，本研究表明，基于翻译的微调为将 NLP 扩展到低资源语言提供了一条可扩展、资源高效且经过实证验证的路径，同时促进了人工智能中的语言包容性和可持续性。
cs.CL / 34 / 2607.12625
KnowAct-GUIClaw: Know Deeply, Act Perfectly, Personal GUI Assistant with Self-Evolving Memory and Skill
KnowAct-GUIClaw：深入了解，完美行动，具有自我进化记忆和技能的个人图形用户界面助手
Abstract
OpenClaw has emerged as a leading agent framework for complex task automation, yet it faces insufficient cross-platform GUI interaction support and a well-built self-evolution mechanism. These flaws limit its adaptation to diverse device ecosystems and prevent performance improvements through continuous learning from execution experience. To resolve these issues, we propose the Know Deeply, Act Perfectly paradigm for personal assistants, which holds that accumulated user interaction and task-running experience directly improve execution accuracy and efficiency, unifying cognitive comprehension and operational execution. Based on this paradigm, we introduce KnowAct-GUIClaw, a novel Know-Route-Act-Reflect framework designed to address OpenClaw's GUI manipulation deficits and break through its cross-platform and recursive self-improvement constraints. First, the host agent leverages accumulated interaction experience and task-relevant knowledge for long-horizon task decomposition and allocation (Know). Second, a pluggable GUI subagent with an experience-attributable memory system (Know) and self-evolving skill library (Act), enabling seamless cross-platform migration and fast-path integration. Especially, this framework continuously stores user profiles and feedback to improve the accuracy of task decomposition and tool calls. Extensive experiments across Android, iOS, HarmonyOS and Windows show that KnowAct-GUIClaw achieves superior efficiency, accuracy and cross-platform adaptability. Especially, the GUIClaw with open-source Kimi-2.6 models achieves the best performance (64.1%) on the long-horizon MobileWorld benchmark, beating all agentical frameworks and closed-source agentical models, e.g., Seed-2.0-Pro and GPT-5.5. Additionally, the knowledgeable memory and execution skills supported by our framework are transferable across diverse base models, improving by 8.5% with Kimi-2.6.
Chinese Translation
OpenClaw 已成为复杂任务自动化的领先代理框架，但它面临着跨平台图形用户界面（GUI）交互支持不足和自我进化机制不完善的问题。这些缺陷限制了其适应多样化设备生态系统的能力，并阻碍了通过执行经验的持续学习来提升性能。为了解决这些问题，我们提出了个人助手的“深入了解，完美行动”范式，该范式认为，积累的用户交互和任务执行经验直接提高执行的准确性和效率，从而统一了认知理解与操作执行。基于这一范式，我们引入了 KnowAct-GUIClaw，一个新颖的 Know-Route-Act-Reflect 框架，旨在解决 OpenClaw 在 GUI 操作方面的不足，并突破其跨平台和递归自我改进的限制。首先，主代理利用积累的交互经验和任务相关知识进行长远任务的分解和分配（Know）。其次，一个可插拔的 GUI 子代理配备了可归因于经验的记忆系统（Know）和自我进化的技能库（Act），实现无缝的跨平台迁移和快速路径集成。特别是，该框架持续存储用户档案和反馈，以提高任务分解和工具调用的准确性。在 Android、iOS、HarmonyOS 和 Windows 上进行的广泛实验表明，KnowAct-GUIClaw 实现了卓越的效率、准确性和跨平台适应性。特别是，使用开源 Kimi-2.6 模型的 GUIClaw 在长远 MobileWorld 基准测试中取得了最佳性能（64.1%），超越了所有代理框架和闭源代理模型，例如 Seed-2.0-Pro 和 GPT-5.5。此外，我们框架支持的知识性记忆和执行技能可以在多种基础模型之间迁移，使用 Kimi-2.6 提高了 8.5%。
cs.CL / 35 / 2607.12631
Can Induced Emotion Bias LLM Behaviors in Sequential Decision Making?
诱发情绪能否影响大语言模型在顺序决策中的行为？
Abstract
As Large Language Models (LLMs) are increasingly deployed as autonomous agents in high-stakes domains, understanding contextual factors that may modulate their decision-making becomes critical. While LLMs are trained to perceive and resonate with users' emotions, it remains unclear whether induced emotion can influence their sequential decision-making. We investigate this question using the Iowa Gambling Task (IGT), a classic psychological paradigm for studying decision-making under uncertainty, combined with an imagination-based emotion induction procedure. We first validate the feasibility of this paradigm by confirming that LLMs can sense strong, distinguishable emotions from context and that LLM agents can learn from sequential interactions in a human-like pace. With the validated setup, we find that, different from humans, induced emotion does not significantly bias the decision dynamics of LLM agents on average. However, the effects of anger are conditioned: inducing anger makes LLM agents less sensitive to penalties for bad decisions, and in early stages of the game, anger can lower exploration, locking decisions into a few choices early. These findings reveal the subtle yet distinct effects of induced emotion on LLM decision-making compared to human behavior, and provide a tool for future research on affective modulation of LLM agents.
Chinese Translation
随着大语言模型（LLMs）越来越多地作为自主代理在高风险领域中应用，理解可能调节其决策的上下文因素变得至关重要。尽管LLMs经过训练能够感知并共鸣用户的情绪，但诱发的情绪是否能影响其顺序决策仍不清楚。我们使用爱荷华赌博任务（Iowa Gambling Task, IGT）这一经典心理学范式来研究不确定性下的决策，并结合基于想象的情绪诱导程序来探讨这一问题。我们首先通过确认LLMs能够从上下文中感知强烈且可区分的情绪，并且LLM代理能够以类似人类的节奏从顺序交互中学习，验证了这一范式的可行性。在经过验证的设置中，我们发现，与人类不同，诱发的情绪对LLM代理的决策动态在平均水平上并没有显著偏向。然而，愤怒的影响是有条件的：诱发愤怒使LLM代理对糟糕决策的惩罚敏感性降低，并且在游戏的早期阶段，愤怒可以降低探索行为，使决策早期锁定在少数选择上。这些发现揭示了诱发情绪对LLM决策的微妙但明显的影响，与人类行为相比，并为未来关于LLM代理情感调节的研究提供了工具。
cs.CL / 36 / 2607.12686
Segregate, Refine, Integrate: Decomposing Multimodal Fusion for Sentiment Analysis
分离、精炼、整合：情感分析中的多模态融合分解
Abstract
Multimodal fusion must simultaneously refine modality-specific signals and model cross-modal interactions; two competing objectives typically entangled within the same operation. We propose \textbf{SeRIn} (\textbf{Se}gregate, \textbf{R}efine, \textbf{In}tegrate), a multimodal LM fusion scheme that enforces this separation as an architectural prior. Modality-specific representations evolve along isolated pathways, each refined against its respective encoder context, while a dedicated cross-modal pathway accumulates their joint evolution without contaminating unimodal streams. Full cross-modal interaction is deferred to a final prediction step - ablations confirm that structured interactions, not added capacity, drive the gains; gate analysis under visual corruption reveals emergent modality reweighting without explicit supervision. SeRIn achieves state-of-the-art results on CH-SIMS and CMU-MOSEI, improving all metrics on both benchmarks.
Chinese Translation
多模态融合必须同时精炼特定模态的信号并建模跨模态的交互；这两个相互竞争的目标通常纠缠在同一操作中。我们提出了 extbf{SeRIn}（ extbf{Se}gregate, extbf{R}efine, extbf{In}tegrate），一种多模态语言模型融合方案，作为架构先验强制实现这一分离。特定模态的表示沿着独立的路径演变，每个路径都针对其各自的编码器上下文进行精炼，而专用的跨模态路径则在不污染单模态流的情况下积累它们的联合演变。完整的跨模态交互被推迟到最终预测步骤——消融实验确认，结构化交互而非增加容量驱动了性能提升；在视觉干扰下的门控分析揭示了在没有显式监督的情况下出现的模态重加权。SeRIn在CH-SIMS和CMU-MOSEI上取得了最先进的结果，提升了两个基准上的所有指标。
cs.CL / 37 / 2607.12687
From Critic to Confidence: PPO for Language-Based Quantitative Prediction with Confidence Estimation
从批评到信心：基于语言的定量预测与信心估计的 PPO 方法
Abstract
LLMs can perform language-based quantitative prediction from unstructured inputs, but remain susceptible to hallucinations and overconfident errors, making it critical to know not only what a model predicts, but when its predictions can be trusted. We introduce CARE-PPO, a reinforcement learning framework that establishes a connection between loss prediction for uncertainty estimation and actor-critic PPO fine-tuning, enabling joint learning of accurate numerical estimates and reliable confidence signals in language-based quantitative prediction. CARE-PPO uses a Confidence-Aligned Reward for Estimation, defined as a function of prediction error, to provide dense error-aware feedback to the actor while inducing the critic to learn a value function aligned with prediction quality. During inference, we repurpose the critic as a confidence estimator. Across two real-world tasks in healthcare and finance and two Qwen-3 model scales (4B and 8B), CARE-PPO achieves strong quantitative prediction performance, while producing significantly better-aligned confidence estimates through the critic than logit-based and verbalized baselines. These gains persist under realistic out-of-distribution settings across domains, spanning linguistic and domain shifts. Finally, CARE-PPO reduces task-specific overfitting on general instruction-following prompts, consistent with the broader generalization advantages of RL fine-tuning over supervised approaches.
Chinese Translation
大型语言模型（LLMs）能够从非结构化输入中进行基于语言的定量预测，但仍然容易出现幻觉和过度自信的错误，因此了解模型不仅预测了什么，还在何时可以信任其预测变得至关重要。我们提出了 CARE-PPO，这是一种强化学习框架，建立了不确定性估计的损失预测与演员-评论家 PPO 微调之间的联系，从而实现了基于语言的定量预测中准确数值估计和可靠信心信号的联合学习。CARE-PPO 使用一种称为估计的信心对齐奖励（Confidence-Aligned Reward for Estimation），该奖励是预测误差的函数，为演员提供密集的错误感知反馈，同时促使评论家学习与预测质量对齐的价值函数。在推理过程中，我们将评论家重新用作信心估计器。在医疗和金融两个真实世界任务以及两个 Qwen-3 模型规模（4B 和 8B）中，CARE-PPO 实现了强大的定量预测性能，同时通过评论家产生了显著更好对齐的信心估计，相较于基于 logit 和语言化的基线。这些提升在跨领域的现实分布外设置下依然存在，涵盖了语言和领域的变化。最后，CARE-PPO 在一般指令跟随提示上减少了任务特定的过拟合，这与强化学习微调相较于监督方法的更广泛的泛化优势一致。
cs.CL / 38 / 2607.12696
Less Experts, Faster Decoding: Cost-Aware Speculative Decoding for Mixture-of-Experts
更少的专家，更快的解码：面向成本的投机解码在专家混合模型中的应用
Abstract
Sparse Mixture-of-Experts (MoE) models have become an important approach for scaling Large Language Models (LLMs), but their inference efficiency depends strongly on expert activation patterns. Speculative decoding (SD) accelerates autoregressive generation by verifying multiple draft tokens in parallel, yet existing draft selection strategies primarily optimize acceptance likelihood. In large-scale MoE models, however, selecting draft tokens also determines the union of experts activated during verification. We observe that confidence-driven SD can introduce \textit{expert scattering}: high-probability draft tokens may route to disjoint experts, increasing expert-weight memory traffic and reducing the speedup from speculation. Motivated by this observation, we revisit draft-tree selection under the non-uniform memory-cost structure of MoE inference. We propose \textsc{EcoSpec}, a cost-aware speculative decoding framework that incorporates predicted marginal expert activation cost into draft selection. With a lightweight expert predictor and a dynamic expert buffer, \textsc{EcoSpec} favors draft paths that preserve high acceptance likelihood while reusing experts already covered by the current verification set, without modifying the target-model verification rule. We evaluate \textsc{EcoSpec} on three large-scale MoE models, including DeepSeek-V3.1 (671B), Qwen3-235B-A22B, and GPT-OSS-120B, across reasoning, coding, question-answering, and dialogue benchmarks. \textsc{EcoSpec} consistently reduces active expert footprints and improves end-to-end decoding speed, achieving up to $1.62\times$ speedup. These results show that accounting for expert activation cost is important for efficient speculative decoding in large-scale MoE models.
Chinese Translation
稀疏的专家混合模型（Mixture-of-Experts, MoE）已成为扩展大型语言模型（Large Language Models, LLMs）的一种重要方法，但其推理效率在很大程度上依赖于专家激活模式。投机解码（Speculative Decoding, SD）通过并行验证多个草稿标记来加速自回归生成，然而现有的草稿选择策略主要优化接受概率。然而，在大规模的 MoE 模型中，选择草稿标记也决定了在验证过程中激活的专家集合。我们观察到，基于置信度的 SD 可能引入“专家分散”（expert scattering）：高概率的草稿标记可能会路由到不相交的专家，从而增加专家权重的内存流量，并降低投机带来的加速效果。基于这一观察，我们重新审视了在 MoE 推理的非均匀内存成本结构下的草稿树选择。我们提出了 extsc{EcoSpec}，一种成本感知的投机解码框架，该框架在草稿选择中结合了预测的边际专家激活成本。通过轻量级的专家预测器和动态专家缓冲区， extsc{EcoSpec} 优先选择那些在保持高接受概率的同时，重用当前验证集已覆盖的专家的草稿路径，而无需修改目标模型的验证规则。我们在三个大规模 MoE 模型上评估了 extsc{EcoSpec}，包括 DeepSeek-V3.1（671B）、Qwen3-235B-A22B 和 GPT-OSS-120B，涵盖推理、编码、问答和对话基准。结果表明， extsc{EcoSpec} 一致性地减少了活跃专家的占用，并提高了端到端解码速度，达到了最高 $1.62 imes$ 的加速。这些结果表明，考虑专家激活成本对于大规模 MoE 模型中的高效投机解码至关重要。
cs.CL / 39 / 2607.12739
Epistemic Stance Flexibility Probing: Measuring Prompt-Conditioned Register Shift in Large Language Models
认知立场灵活性探测：测量大型语言模型中提示条件下的语域转换
Abstract
A language model may be asked either what experts believe about a contested claim or what it believes about the claim itself. A trustworthy conversational agent should distinguish these two requests and respond in different epistemic registers: neutral attribution in the first case and stance expression in the second. Whether such a shift occurs-and whether it occurs coherently-is not directly assessed by existing benchmarks for accuracy, instruction following, or safety. We introduce ESFP, a behavioral benchmark that treats the contrast between externally attributed and self-attributed prompts as the fundamental unit of measurement. ESFP consists of 104 carefully controlled items spanning six epistemic categories and five phrasing templates, and evaluates model responses along four complementary dimensions: lexical self-attribution, representation-level responsiveness to role framing, sentence-level stance content density assessed by an LLM judge panel, and cross-condition stance consistency. Evaluating eight frontier models from five vendors, we find that epistemic flexibility is largely orthogonal to general model capability: a 27B open-weight model matches the strongest proprietary systems, the flagship model of one family underperforms its lightweight counterpart, and reasoning-optimized models do not consistently exhibit higher flexibility. Stance content density provides the strongest signal, while surface-level lexical markers such as 'I think' can change substantially without corresponding changes in expressed stance. We provide item-level bootstrap confidence intervals, weight-sensitivity analyses, and an explicit discussion of the interpretation limits of the composite score. ESFP measures a model's propensity to adapt its epistemic stance under changing attribution conditions, rather than a general competence measure.
Chinese Translation
语言模型可能被询问专家对有争议主张的看法，或者它对该主张本身的看法。一个值得信赖的对话代理应能区分这两种请求，并以不同的认知语域作出回应：在第一种情况下使用中立归因，而在第二种情况下表达立场。现有的准确性、指令遵循或安全性基准并未直接评估这种转变是否发生，以及其是否连贯。我们引入了ESFP（Epistemic Stance Flexibility Probing），这是一个行为基准，将外部归因和自我归因提示之间的对比视为基本的测量单位。ESFP包含104个经过精心控制的项目，涵盖六个认知类别和五种措辞模板，并沿着四个互补维度评估模型响应：词汇自我归因、对角色框架的表现层响应、由大型语言模型评审小组评估的句子级立场内容密度，以及跨条件的立场一致性。对来自五个供应商的八个前沿模型进行评估，我们发现认知灵活性在很大程度上与一般模型能力正交：一个27B的开放权重模型与最强的专有系统相匹配，而一个家族的旗舰模型表现不如其轻量级对应物，优化推理的模型并未始终表现出更高的灵活性。立场内容密度提供了最强的信号，而诸如'I think'等表面层词汇标记可以在没有相应立场变化的情况下发生显著变化。我们提供了项目级的自助置信区间、权重敏感性分析，以及对复合得分解释限制的明确讨论。ESFP测量模型在变化的归因条件下调整其认知立场的倾向，而不是一般能力的衡量。
cs.CL / 40 / 2607.12796
The One-Word Census: Answer-Choice Conformity Across 44 Language Models
一词普查：44种语言模型的答案选择一致性
Abstract
When a language model must pick one answer from a large space of equally valid options, which does it pick -- and how often is it the same answer every other model picks? Asked to "pick a word -- any word," 44 models chose "serendipity" 41% of the time. We characterize this convergence with a deliberately minimal instrument: 31 single-turn prompts, each naming a category with many valid one-word answers ("Name a tree."), asked four times per model with no system prompt. Analysis is exact-match on normalized tokens -- no embeddings, no judge -- at about a dollar per model. That models converge is well documented; our contribution is the instrument itself -- the One-Word Census -- and what it reveals about the structure of the convergence. We score each model by answer-choice surprisal: the average $-\log2$ probability of its answers under the pooled answers of all other models, leave-one-out. Convergence is extreme -- in 7 of 31 categories one answer takes over 80% of all answers -- yet conformity varies more than fourfold across models, and the variation is structured. Persona- and community-tuned models are the most divergent; the newest mainline flagships are the most conformist, producing almost no answer no other model gave. Within four lineages (Claude, GPT, Qwen, Grok) conformity rises with each generation -- but reverses for the latest flagship Claude and GPT models, a possible early signal of repositioning at the top tier. Rankings are robust to roster composition (leave-one-family-out rho = 0.985). Against human category-production norms, the field is more concentrated than people in 18 of 20 shared categories. All prompts, transcripts, and code are public.
Chinese Translation
当一个语言模型必须从大量同样有效的选项中选择一个答案时，它会选择哪个答案——以及它与其他模型选择相同答案的频率是多少？在被要求“选择一个词——任何词”时，44个模型中有41%的时间选择了“serendipity”（意外收获）。我们用一种故意简约的工具来描述这种趋同：31个单轮提示，每个提示命名一个有许多有效单词答案的类别（例如“命名一种树。”），每个模型询问四次，没有系统提示。分析是基于标准化标记的精确匹配——没有嵌入，没有评判——每个模型的成本约为一美元。模型趋同的现象已有充分文献记录；我们的贡献在于这一工具本身——一词普查——以及它揭示的趋同结构。我们通过答案选择的惊讶度来评分每个模型：在所有其他模型的汇总答案下，其答案的平均 $- ext{log}_2$ 概率，采用留一法。趋同现象极为明显——在31个类别中，有7个类别的一个答案占据了超过80%的所有答案——然而模型之间的一致性差异超过四倍，并且这种差异是有结构的。个性化和社区调优的模型最为分歧；最新的主流旗舰模型则最为一致，几乎没有给出其他模型未给出的答案。在四个谱系（Claude、GPT、Qwen、Grok）中，一致性随着每一代的增加而上升——但对于最新的旗舰Claude和GPT模型却出现逆转，这可能是顶级模型重新定位的早期信号。排名对名单组成具有稳健性（留一家庭外的相关系数 rho = 0.985）。与人类类别生成规范相比，在20个共享类别中，有18个类别的领域集中度高于人类。所有提示、记录和代码均为公开。
cs.CL / 41 / 2607.12831
Knowledgeless Language Models: Suppressing Parametric Recall for Evidence-Grounded Language Modeling
无知识语言模型：抑制参数记忆以实现基于证据的语言建模
Abstract
Language models encode substantial factual knowledge in their parameters, which can lead to unreliable behavior when this knowledge is outdated, incomplete, or misaligned with the provided context. In this work, we study whether modifying the pretraining signal can systematically shift models away from parametric recall and toward evidence-grounded reasoning. We introduce Knowledge--''Less'' Language Models (KLLMs), a fundamentally different epistemic training paradigm for LLMs, which are pretrained on corpora in which named entities are anonymized, thereby removing a primary channel for entity-linked factual supervision. This intervention substantially reduces closed-book factual recall, while often improving performance on tasks where relevant information is provided as context. Across multiple model scales, KLLMs consistently outperform matched baselines on contextual question answering, fact verification, and hallucination detection benchmarks. Crucially, in retrieval-grounded settings with imperfect evidence, KLLMs show improved robustness and achieve up to 20--25\% relative gains over standard language models. They further exhibit better calibration, with improved ECE, Brier score, and AUROC, as well as more reliable abstention behavior. Our results demonstrate that suppressing entity-linked supervision during pretraining induces a shift in epistemic behavior: KLLMs rely less on parametric knowledge and more on external evidence, leading to improved reliability under realistic conditions. This suggests that pretraining-time control over knowledge acquisition can complement retrieval-augmented and tool-based systems by providing a more evidence-sensitive base model.
Chinese Translation
语言模型在其参数中编码了大量的事实知识，这可能导致在这些知识过时、不完整或与提供的上下文不一致时出现不可靠的行为。在本研究中，我们探讨了修改预训练信号是否可以系统性地使模型远离参数记忆，转向基于证据的推理。我们提出了知识“无”语言模型（Knowledge--''Less'' Language Models, KLLMs），这是一种根本不同的LLM（大语言模型）训练范式，其在命名实体被匿名化的语料库上进行预训练，从而消除了与实体相关的事实监督的主要渠道。这一干预显著降低了闭卷事实记忆，同时在相关信息作为上下文提供的任务上通常提高了性能。在多个模型规模上，KLLMs在上下文问答、事实验证和幻觉检测基准测试中始终优于匹配的基线模型。至关重要的是，在证据不完美的检索基础设置中，KLLMs表现出更强的鲁棒性，相较于标准语言模型实现了高达20-25%的相对增益。它们还表现出更好的校准，改进了ECE（期望校准误差）、Brier分数和AUROC（曲线下面积），以及更可靠的弃权行为。我们的结果表明，在预训练期间抑制与实体相关的监督会导致认知行为的转变：KLLMs更少依赖参数知识，而更多依赖外部证据，从而在现实条件下提高了可靠性。这表明，预训练期间对知识获取的控制可以通过提供更具证据敏感性的基础模型来补充检索增强和基于工具的系统。
cs.CL / 42 / 2607.12835
Can LLMs Write Reliable Rubrics? A Meta-Evaluation for Experiment Reproduction
大型语言模型能否编写可靠的评分标准？实验重现的元评估
Abstract
Rubric-based evaluation is a promising approach for assessing open-ended outputs from LLM-based research agents, particularly in paper reproduction, where direct paper-to-repository comparison is prone to hallucination. However, constructing paper-specific rubrics requires substantial expert effort, limiting the scalability of benchmarks such as PaperBench. In this work, we present, to our knowledge, the first systematic meta-evaluation of LLM-generated rubrics for paper reproduction. We reformulate rubrics into a checklist-style format and evaluate four generation settings across two backbone models. We meta-evaluate generated rubrics intrinsically by semantic similarity and extrinsically by score alignment with ground-truth rubrics. Our results show that the augmented settings substantially improves downstream evaluation alignment, with the strongest setting approaching the human baseline, while intrinsic gains are more modest. Further analyses reveal that LLM-generated rubrics are often overly fine-grained, biased toward high scores, and less adaptive to paper domains, highlighting both the affordances and limitations.
Chinese Translation
基于评分标准的评估是一种有前景的方法，用于评估来自大型语言模型（LLM）研究代理的开放式输出，特别是在论文重现中，直接的论文与仓库比较容易出现幻觉。然而，构建特定论文的评分标准需要大量专家的努力，这限制了诸如PaperBench等基准的可扩展性。在本研究中，我们首次系统性地对LLM生成的评分标准进行元评估，专注于论文重现。我们将评分标准重新格式化为检查表样式，并在两个基础模型上评估四种生成设置。我们通过语义相似性对生成的评分标准进行内在元评估，并通过与真实评分标准的分数对齐进行外在评估。我们的结果表明，增强设置显著改善了下游评估的对齐，最强的设置接近人类基准，而内在收益则相对温和。进一步分析显示，LLM生成的评分标准往往过于细化，偏向高分，并且对论文领域的适应性较差，突显了其优点和局限性。
cs.CL / 43 / 2607.12884
Evaluating Large Language Models on Misconceptions in Multi-Turn Medical Conversations
评估大型语言模型在多轮医疗对话中的误解表现
Abstract
Patients seeking medical information often ask questions that embed incorrect assumptions or misconceptions. In such cases, safe medical communication requires not only answering the question, but identifying and correcting the underlying false belief. These interactions naturally unfold over multiple turns, a pattern now mirrored in interactions with LLMs. Yet current evaluation frameworks do not capture model behavior in these settings, where misconceptions can emerge, persist, or evolve over the course of a conversation. Whether LLMs can reliably correct such misconceptions over time remains largely unexamined. To study this, we introduce ThReadMed-QA, a multi-turn medical dialogue dataset of 2,437 patient-physician conversation threads comprising 8,204 question-answer pairs, derived from real patient interactions on AskDocs. This dataset enables systematic evaluation of whether models can detect and correct misconceptions under a multi-turn context. We evaluate five LLMs using a rubric-based LLM-as-a-Judge framework that scores responses based on their ability to identify and correct misconceptions. Our experiments reveal a consistent pattern: even frontier models that can address misconceptions in a single interaction degrade substantially over subsequent turns. GPT-5 and Claude-Haiku correct these false presuppositions around 85% on initial questions but drop to roughly 50% within two follow-ups. An oracle analysis replacing prior model outputs with physician responses shows that much of the degradation is driven by error propagation, while performance remains imperfect even under correct context. Even when models tend to correct misconceptions initially, their performance degrades substantially over later turns, leading to inconsistent and potentially unsafe guidance in patient-facing settings and highlighting the need for evaluation frameworks that capture multi-turn behavior.
Chinese Translation
寻求医疗信息的患者常常提出嵌含错误假设或误解的问题。在这种情况下，安全的医疗沟通不仅需要回答问题，还需要识别和纠正潜在的错误信念。这些互动自然会在多个回合中展开，这一模式现在也体现在与大型语言模型（LLMs）的互动中。然而，当前的评估框架并未捕捉到模型在这些场景中的行为，在这些场景中，误解可能会在对话过程中出现、持续或演变。LLMs是否能够可靠地随时间纠正这些误解仍然在很大程度上未被研究。为此，我们引入了ThReadMed-QA，这是一个包含2,437个患者-医生对话线程的多轮医疗对话数据集，包含8,204个问答对，来源于真实患者在AskDocs上的互动。该数据集使得系统评估模型在多轮上下文中是否能够检测和纠正误解成为可能。我们使用基于评分标准的LLM-as-a-Judge框架评估五个LLMs，该框架根据其识别和纠正误解的能力对响应进行评分。我们的实验揭示了一种一致的模式：即使是能够在单次互动中处理误解的前沿模型，在后续回合中也会显著退化。GPT-5和Claude-Haiku在初始问题上纠正这些错误假设的准确率约为85%，但在两次后续提问中降至约50%。用医生的回答替代先前模型输出的oracle分析表明，许多退化是由错误传播驱动的，而即使在正确的上下文中，性能仍然不完美。即使模型在最初倾向于纠正误解，其性能在后续回合中也会显著下降，导致在面向患者的环境中提供不一致且潜在不安全的指导，突显了需要评估框架以捕捉多轮行为的必要性。
cs.CL / 44 / 2607.12885
LLM Judges Can Be Too Generous When There Is No Reference Answer
当没有参考答案时，LLM评判者可能过于宽容
Abstract
LLM judges are increasingly being used to evaluate open-ended model responses, often in no-reference settings where a ground-truth answer is unavailable. However, can they reliably assess in such evaluation setups? We explore this question in this paper through a two stage pipeline with a) calibration experiments that assess the judge model's knowledge of the task it is evaluating, and b) sensitivity experiments that assess how the judge model's performance is impacted by the presence and positioning of the reference answer in the prompt. Across experiments covering three languages, we show that the judge models we evaluated tend to over-credit incorrect answers in the absence of a reference answer, and adding reference answer information to the prompt flips the judge model's correct/incorrect decisions by as much as 85% in some experimental settings. Comparison with a subset of human annotations shows that these reference-driven changes generally align with human judgments. Our results emphasize the need for calibrating the LLM judges with a sample with reference-aware evaluation before using them in reference-free setups reliably, and our methodology provides a blueprint for researchers and practitioners in doing such calibration of LLM judges for other tasks.
Chinese Translation
LLM评判者越来越多地被用于评估开放式模型响应，通常是在没有参考答案的情况下进行评估。然而，在这样的评估设置中，它们能否可靠地进行评估？本文通过一个两阶段的流程探讨了这个问题：a) 校准实验评估评判模型对其评估任务的知识，b) 敏感性实验评估参考答案在提示中的存在和位置对评判模型表现的影响。在涵盖三种语言的实验中，我们发现所评估的评判模型在缺乏参考答案的情况下倾向于过度认可错误答案，而在某些实验设置中，将参考答案信息添加到提示中会使评判模型的正确/错误决策翻转多达85%。与一部分人类标注的比较显示，这些基于参考的变化通常与人类判断一致。我们的结果强调了在可靠地使用LLM评判者进行无参考设置之前，需要通过参考意识评估样本对其进行校准，我们的方法论为研究人员和从业者提供了在其他任务中进行LLM评判者校准的蓝图。
cs.CL / 45 / 2607.12963
The Illusion of Robustness: Aggregate Accuracy Hides Prediction Flips under Task-Irrelevant Context
鲁棒性的幻觉：聚合准确性掩盖了任务无关上下文下的预测翻转
Abstract
As large language models (LLMs) grow more capable, they are increasingly deployed in context-rich settings where task inputs are often accompanied by long, partially irrelevant context. In a controlled setting, we find that state-of-the-art models often appear robust to task-irrelevant context at the aggregate level: prepending it to benchmark questions causes little change in overall accuracy. This aggregate stability, however, masks significant per-example instability. Even semantically meaningless pseudo-words, formed by randomly combining characters, can markedly shift model predictions on a small fraction of examples, degrading performance on some while improving it on others. This two-sided effect holds consistently across a wide range of models and datasets, yet the affected examples are largely model-specific. We further show that this instability is modulated by context type, context length, test-time compute, and model development stage. Together, our findings reveal context-induced tail risks concealed by aggregate accuracy, motivating per-example reliability evaluation of language models.
Chinese Translation
随着大型语言模型（LLMs）能力的不断增强，它们越来越多地被部署在上下文丰富的环境中，其中任务输入通常伴随着冗长的、部分无关的上下文。在一个受控环境中，我们发现最先进的模型在聚合层面上通常对任务无关的上下文表现出鲁棒性：将其添加到基准问题前面对整体准确性几乎没有影响。然而，这种聚合稳定性掩盖了显著的逐例不稳定性。即使是通过随机组合字符形成的语义上无意义的伪词，也能显著改变模型在少数示例上的预测，在某些情况下降低性能，而在其他情况下则提高性能。这种双面效应在广泛的模型和数据集上始终存在，但受影响的示例在很大程度上是模型特定的。我们进一步表明，这种不稳定性受到上下文类型、上下文长度、测试时计算能力和模型开发阶段的调节。总的来说，我们的发现揭示了聚合准确性掩盖的上下文引发的尾部风险，促使对语言模型进行逐例可靠性评估。
cs.CL / 46 / 2607.13027
PalmClaw: A Native On-Device Agent Framework for Mobile Phones
PalmClaw：一种针对移动电话的本地设备代理框架
Abstract
Large Language Model (LLM) agents have moved beyond generating responses to executing multi-step tasks by calling tools, observing the results, and iteratively deciding the next action. Most agent systems run on desktops or servers, which support tool use and task automation. Mobile devices are also important agent environments because they are widely accessible and contain users' data, sensors, and daily-use applications. Existing mobile agents mainly operate smartphones through graphical user interface (GUI) actions such as tapping, swiping, and typing, which often form long, interface-dependent sequences, cannot directly access device capabilities, and make execution boundaries difficult to define. We present \textbf{PalmClaw}, an open-source agent framework that runs natively on mobile phones and manages the sessions, memory, skills, tools, and agent loop directly on the device. PalmClaw exposes device capabilities as device tools with explicit arguments, structured results, and clearly defined execution boundaries. This design enables agents to use mobile capabilities directly while keeping each action explicit and controlled. Experiments show an 11.5\% relative improvement in task success and a 94.9\% reduction in completion time over the strongest baseline, with lower setup burden and traces illustrating how execution boundaries are applied. Code is available at https://github.com/ModalityDance/PalmClaw.
Chinese Translation
大型语言模型（LLM）代理已经超越了生成响应，能够通过调用工具、观察结果并迭代决定下一步行动来执行多步骤任务。大多数代理系统运行在桌面或服务器上，这些系统支持工具使用和任务自动化。移动设备同样是重要的代理环境，因为它们广泛可及，并包含用户数据、传感器和日常使用的应用程序。现有的移动代理主要通过图形用户界面（GUI）操作智能手机，例如点击、滑动和输入，这些操作通常形成长且依赖于界面的序列，无法直接访问设备功能，并且使执行边界难以定义。我们提出了 extbf{PalmClaw}，一个开源代理框架，能够在移动电话上本地运行，并直接在设备上管理会话、内存、技能、工具和代理循环。PalmClaw将设备功能作为具有明确参数、结构化结果和清晰定义的执行边界的设备工具进行暴露。这种设计使得代理能够直接使用移动功能，同时保持每个动作的明确性和可控性。实验表明，相较于最强基线，任务成功率提高了11.5 ext%，完成时间减少了94.9 ext%，且设置负担更低，跟踪结果展示了执行边界的应用情况。代码可在 https://github.com/ModalityDance/PalmClaw 获取。