2026-10论文中文摘要
Dex-One2Many：从单次人类演示中学习灵巧操作
Dex-One2Many: Learning Dexterous Manipulation from a Single Human Demonstration
arXiv 2610.12470 · 2026-10-09Rubric-CEPR：基于奖励验证自蒸馏的自进化图像编辑
Rubric-CEPR: Self-Evolving Image Editing via Reward-Verified Self-Distillation
arXiv 2610.12469 · 2026-10-09DreamTrue：具备反事实后训练的动作忠实机器人世界模型
DreamTrue: Action-Faithful Robot World Model with Counterfactual Post-Training
arXiv 2610.12468 · 2026-10-09CSF：面向运动生成器的上下文安全过滤
CSF: Contextual Safety Filtering for Motion Generators
arXiv 2610.12467 · 2026-10-09关于AI时间范围的估计与有效性——对METR图的统计学分析
On the estimation and validity of AI time horizons---a statistical look at the METR plot
arXiv 2610.12466 · 2026-10-09平衡数据策略：解决机器人控制超大规模强化学习中的探索瓶颈
A Balanced Data Diet: Addressing the Exploration Bottleneck in Mega-Scale RL for Robot Control
arXiv 2610.12465 · 2026-10-0930000小时自我中心视频无法教会的东西
What 30,000 Hours of Ego-centric Video Does Not Teach
arXiv 2610.12464 · 2026-10-09从被动遏制到主动保障：OpenAI、Anthropic与谷歌智能体安全事件的教训
From Reactive Containment to Proactive Assurance: Lessons from OpenAI, Anthropic, and Google Agent Security Incidents
arXiv 2610.12463 · 2026-10-09吸积盘结构可决定活动星系核（AGN）的变异性并解释吸积盘大小问题：无需宽线区
Disk Structure May Determine AGN Variability and Explain the Accretion Disk Size Problem: No Broad Line Region Required
arXiv 2610.12462 · 2026-10-09OuroWorld：将任意3D世界转化为多样、无限循环的3D动态影像
OuroWorld: Bringing Any 3D World Alive as Diverse, Endlessly Looping 3D Cinemagraphs
arXiv 2610.12461 · 2026-10-09无限光线与具有无限多个枢轴的无限簇
Infinite light rays and infinite clusters with infinitely many pivots
arXiv 2610.12460 · 2026-10-09WorldGuide：面向流程任务执行的目标导向视频世界模型
WorldGuide: Goal-Directed Video World Model for Procedural Task Execution
arXiv 2610.12459 · 2026-10-09OmniCapBench：面向细粒度视听字幕生成的深度结构化评估框架
OmniCapBench: A Deep-Structured Evaluation Framework for Fine-Grained Audio-Visual Captioning
arXiv 2610.12458 · 2026-10-09SpatialHarness：面向精细机器人操作的测试时空间支架
SpatialHarness: Test-Time Spatial Scaffolding for Fine Robotic Manipulation
arXiv 2610.12457 · 2026-10-09耦合独立性蕴含零自由性
Coupling Independence Implies Zero-Freeness
arXiv 2610.12456 · 2026-10-09混合电影摄影：生成式视频重拍中的幻觉风险预可视化与管理
Hybrid Cinematography: Previsualizing and Managing Hallucination Risk in Generative Video Reshooting
arXiv 2610.12455 · 2026-10-09PPT纠缠蒸馏研究
On PPT entanglement distillation
arXiv 2610.12454 · 2026-10-09多智能体系统中的心智模型
Mental-Models for Multi-Agent Systems
arXiv 2610.12453 · 2026-10-09BrickBench：评估智能体的乐高套装设计能力
BrickBench: Evaluating Agentic Brick Design
arXiv 2610.12452 · 2026-10-09VersaCamVLA：用于机器人操作的相机可配置VLA策略
VersaCamVLA: Camera-Configurable VLA Policies for Robotic Manipulation
arXiv 2610.12451 · 2026-10-09帕克斯存档数据中河外隐藏脉冲星的发现
Discovery of Hidden Extragalactic Pulsars in the Parkes Archival Data
arXiv 2610.12450 · 2026-10-09Bi-FORK：高维分岔系统的生成建模
Bi-FORK: Generative Modeling of High-Dimensional Bifurcating Systems
arXiv 2610.12449 · 2026-10-09一个块，多个深度：具有深度编程专家的循环视觉Transformer
One Block, Multiple Depths: Recurrent Vision Transformers with Depth-Programmed Experts
arXiv 2610.12448 · 2026-10-09纠缠熵与ZX图的魔力
Entanglement entropy and magic of ZX-diagrams
arXiv 2610.12447 · 2026-10-09单连通流形上极作用的自相交截面
Self-intersecting sections of polar actions on simply connected manifolds
arXiv 2610.12446 · 2026-10-09当场被抓：探测工具可有效检测破坏行为并捕捉未明确表达的欺骗
Caught in the Act: Probes Effectively Detect Sabotage and Catch Unverbalized Deception
arXiv 2610.12445 · 2026-10-09预条件空间中的舍入：重新设计4位AdamW优化器状态量化
Rounding in Preconditioner Space: Redesigning 4-bit AdamW Optimizer-State Quantization
arXiv 2610.12444 · 2026-10-09对LUX-ZEPLIN 248 keV excess的标量扩展U(1)_{L_μ-L_τ}解释
A scalar-extended ${\rm U(1)_{L_μ-L_τ}}$ explanation of the LUX-ZEPLIN 248 keV excess
arXiv 2610.12443 · 2026-10-09LEGO：一种用于异中心视角到自我中心视角视频生成的无提升方法
LEGO: A Lifting-Free Approach for Exocentric-to-Egocentric Video Generation
arXiv 2610.12442 · 2026-10-09短程矢量自旋玻璃：帕里西层级与实空间超度量性
Short-Range Vector Spin Glasses: Parisi Hierarchies and Real-Space Ultrametricity
arXiv 2610.12441 · 2026-10-09面向人-机器人灵巧操作的生成式神经重定向
Generative Neural Retargeting for Human-to-Robot Dexterous Manipulation
arXiv 2610.12440 · 2026-10-09八流形上M理论的一个实超势
A real superpotential for M-theory on eight-manifolds
arXiv 2610.12439 · 2026-10-09树的并行边排序
Parallel Edge Ranking of Trees
arXiv 2610.12438 · 2026-10-09基于Stein位移场的密度比估计
Density Ratio Estimation with Stein Displacement Fields
arXiv 2610.12437 · 2026-10-09AI智能体的生态学：协作为起飞创造了种群阈值
Ecology of AI Agents: Collaboration Creates a Population Threshold for Takeoff
arXiv 2610.12436 · 2026-10-09VioLA：从人类数据学习通用人形机器人控制策略
VioLA: Learning Generalist Humanoid Control Policies from Human Data
arXiv 2610.12435 · 2026-10-09齐次多项式向量场的全局渐近稳定性是不可判定的
Global asymptotic stability of homogeneous polynomial vector fields is undecidable
arXiv 2610.12434 · 2026-10-09ℓ₂范数下的Hilton-Milner定理
The Hilton-Milner theorem in the $\ell_2$-norm
arXiv 2610.12433 · 2026-10-09FAITH：面向高维系统的可行性感知安全过滤强化学习
FAITH: Feasibility-Aware Safety-Filtered RL for High-Dimensional Systems
arXiv 2610.12432 · 2026-10-09适用于轨迹扩散的可控制不确定性
Control-Ready Uncertainty for Trajectory Diffusion
arXiv 2610.12431 · 2026-10-09关于具有不可求例外梯度集的完全非线性椭圆方程的估计
Estimates on Fully Nonlinear Elliptic Equations with Unrectifiable Exceptional Gradient Sets
arXiv 2610.12430 · 2026-10-09聆听视界：利用GW250114探测近视界反射率
Listening to the Horizon: Probing Near-Horizon Reflectivity with GW250114
arXiv 2610.12429 · 2026-10-09自适应增长量子分类器中电路深度与训练数据规模的联合优化研究
Toward Joint Optimization of Circuit Depth and Training Data Size in Adaptively Grown Quantum Classifiers
arXiv 2610.12428 · 2026-10-09FastBench：流式视觉语言模型（VLMs）能否感知高动态真实世界流？
FastBench: Can Streaming VLMs Perceive High-Dynamic Real-World Streams?
arXiv 2610.12427 · 2026-10-09随机神经网络的李雅普诺夫谱
Lyapunov spectrum of random neural networks
arXiv 2610.12426 · 2026-10-09关于平均量的弛豫时间的大漂移增强
On Large Drift Enhancement for the Relaxation Time of Averages
arXiv 2610.12425 · 2026-10-09RoboRSI：复杂真实环境中稳定、高效且可复用的机器人自进化系统
RoboRSI: Stable, efficient, and reusable robot self-evolution in complex real-world environments
arXiv 2610.12424 · 2026-10-09Pumpire：用于度量距离估计的统一基准
Pumpire: Unified Benchmark for Metric Distance Estimation
arXiv 2610.12423 · 2026-10-09具有屏蔽禁闭相互作用的夸克物质密度泛函
A Density Functional for Quark Matter with Screened Confinement Interactions
arXiv 2610.12422 · 2026-10-09超越时空先验：一种可泛化的密集对应匹配方法
Beyond Spatio-Temporal Priors: A Generalizable Approach for Dense Correspondence Matching
arXiv 2610.12421 · 2026-10-09
