2026-09论文中文摘要
RACE：LLM神经元中功能一致性的可扩展统计估计
RACE: Scalable Statistical Estimation of Functional Consistency in LLM Neurons
arXiv 2608.24758 · 2026-09-02用于联合多接收机无导频Wi-Fi解码的Transformer
A Transformer for Joint Multi-Receiver Pilotless Wi-Fi Decoding
arXiv 2608.24584 · 2026-09-02带路径记忆辅助比特的三缝干涉：亮-暗态表述
Three-slit interference with a which-path memory ancilla: A bright-dark state formulation
arXiv 2608.24243 · 2026-09-02迈向作用量依赖的帕拉蒂尼引力的联络形式化
Towards a Connection Formulation of Action-Dependent Palatini Gravity
arXiv 2608.24140 · 2026-09-02托卡马克磁控制的逆Grad-Shafranov神经网络方法
An Inverse Grad-Shafranov Neural Network Approach to Tokamak Magnetic Control
arXiv 2608.23976 · 2026-09-02让自动体外除颤器（AED）动起来：增强城市流动性的除颤器部署方案
Let AEDs Move: Urban Mobility Enhanced Defibrillator Deployment
arXiv 2608.23810 · 2026-09-02增长移位下的定量对数Chowla相关性
Logarithmic Chowla Correlations Across All Shift Scales
arXiv 2608.23500 · 2026-09-02CyberFactory：利用真实场景实例扩展网络安全能力
CyberFactory: Scaling Cyber Security Capabilities with Instances from the Wild
arXiv 2608.23181 · 2026-09-02自发电流决定活性物质中的毛细上升
Spontaneous currents determine capillary rise in active matter
arXiv 2608.22878 · 2026-09-02伊顿-莫雷托猜想对p-可解群成立
The Eaton--Moretó Conjecture Holds True for \(p\)-Solvable Groups
arXiv 2608.22813 · 2026-09-02对遗忘算法进行压力测试
Stress Testing Unlearning Algorithms
arXiv 2608.22527 · 2026-09-02随机策略proof的设施选址：两个及以上设施
Randomized Strategyproof Facility Location: Two Facilities and Beyond
arXiv 2608.22484 · 2026-09-02超越密集Adam状态：用于内存高效优化器的自适应对数空间量化
Beyond Dense Adam States: Adaptive Log-Space Quantization for Memory-Efficient Optimizers
arXiv 2608.22322 · 2026-09-02带活性末端的化学-机械胶体链的集体动力学
Collective dynamics of chemo-mechanical colloidal chains with active tips
arXiv 2608.22318 · 2026-09-02MRMAD：用于评估大型音频语言模型中声学降级感知的多轮多音频基准
MRMAD: A Multi-Round Multi-Audio Benchmark for Evaluating Acoustic Degradation Perception in Large Audio-Language Models
arXiv 2608.22236 · 2026-09-02PropUQ-MAS：面向大语言模型多智能体系统的传播感知不确定性量化
PropUQ-MAS: Propagation-Aware Uncertainty Quantification for LLM Multi-Agent Systems
arXiv 2608.22130 · 2026-09-02几乎所有图都可由其令牌图重构
Almost Every Graph Is Reconstructible from Its Token Graphs
arXiv 2608.22094 · 2026-09-02HiDiffTIR：面向多轮工具集成推理的分层难度感知策略优化
HiDiffTIR: Hierarchical Difficulty-Aware Policy Optimization for Multi-Turn Tool-Integrated Reasoning
arXiv 2608.21863 · 2026-09-02作为封闭生命支持系统的地球：有限缓冲、动力学阈值与行星可持续性
Earth as a Closed Life-Support System: Finite Buffers, Dynamical Thresholds, and Planetary Sustainability
arXiv 2608.21505 · 2026-09-02基于串并联攻击图与自适应攻击者的安全博弈
Security Games on Series-Parallel Attack Graphs with Adaptive Attackers
arXiv 2608.21259 · 2026-09-02掺杂驱动的莫特相变：基于稳态密度泛函理论的研究
Doping-driven Mott transition from steady-state density functional theory
arXiv 2608.21029 · 2026-09-02薄壳性质通过高斯倾斜蕴含小球偏差
Thin-Shell implies small-ball deviation via Gaussian tilts
arXiv 2608.20816 · 2026-09-02共形微扰论导出的杂弦理论中的阈值修正
Threshold Corrections in Heterotic String Theory from Conformal Perturbation Theory
arXiv 2608.20459 · 2026-09-02DreamHand：复用视频扩散模型实现遮挡鲁棒的第一人称视角3D手部运动恢复
ACE-Ego-Hand: Repurposing Video Diffusion Models for Occlusion-Robust Egocentric 3D Hand Motion Recovery
arXiv 2608.20308 · 2026-09-02电子航海图变化分类
Electronic Navigational Chart Change Classification
arXiv 2608.20218 · 2026-09-02SWE-bench Science：编码智能体能解决科学领域的工程任务吗？
SWE-bench Science: Can Coding Agents Resolve Engineering Tasks in Science?
arXiv 2608.19799 · 2026-09-02有界凸区域中受限半拉普拉斯算子的扭转函数的严格凹性
Strict Concavity of the Torsion Function for the Restricted Half-Laplacian in Bounded Convex Domains
arXiv 2608.19586 · 2026-09-02当无关文本起作用时：多模态大语言模型中的仿射间隔偏移
When Irrelevant Text Matters: Affine Margin Shifts in Multimodal Large Language Models
arXiv 2608.19208 · 2026-09-02作为验证方法的限定交叉引用：欧盟人工智能法案的规范环境
Qualified Cross-References as a Verification Method: The Normative Environment of the EU AI Act
arXiv 2608.19194 · 2026-09-02不公平0-1多项式问题与高次三项式
The Unfair 0-1 Polynomial Problem and High-Degree Trinomials
arXiv 2608.19173 · 2026-09-02单模量子宇宙学中的Duru-Kleinert路径积分
Duru-Kleinert Path Integral in Unimodular Quantum Cosmology
arXiv 2608.18983 · 2026-09-02三维Kakeya简化的细胞极大密度分解及修正
Cellular Maximal-Density Factoring: Joint Shadings, Stable Refinements, and a Conditional Kakeya Application
arXiv 2608.18870 · 2026-09-02可验证弃权使AI在供水管网泄漏诊断中具备可问责性
Verifiable abstention makes AI leak diagnosis accountable in urban water distribution networks
arXiv 2608.18836 · 2026-09-02循环称重矩阵的新不存在性结果
Character and Multiplier Obstructions for Circulant Weighing Matrices
arXiv 2608.18468 · 2026-09-020xPass：一种用于通用跨链账户的安全协议
0xPass: A Secure Protocol for Universal Cross-Chain Accounts
arXiv 2608.18359 · 2026-09-02用于大规模推荐解释的LLM评判模型的生命周期
The Lifecycle of LLM-as-a-Judge for Large-Scale Recommendation Explanations
arXiv 2608.18300 · 2026-09-02无黄金标准标签的AI生成数据的去偏推断：通过多个不完美测量进行识别
Debiased Inference for AI-Generated Data without Gold-Standard Labels: Identification via Multiple Imperfect Measurements
arXiv 2608.18294 · 2026-09-02关于规范耦合变化的起源
On the Origins of Varying Gauge Couplings
arXiv 2608.18203 · 2026-09-02大型语言模型（LLMs）能否处理信念与事实取决于表述方式
Whether LLMs Can Navigate Beliefs and Facts Depends on How You Phrase It
arXiv 2608.17809 · 2026-09-02具有周期结果和非周期SG值的无限减法游戏
Infinite Subtraction Games with Periodic Outcomes and Aperiodic SG Values
arXiv 2608.17654 · 2026-09-02$w-c_s^2$ 平面中的暗能量
Dark Energy in the $w-c_s^2$ Plane
arXiv 2608.17208 · 2026-09-02利用JWST NIRISS SOSS对TRAPPIST-1 f进行大气侦察：无凌星光源效应的证据
Atmospheric Reconnaissance of TRAPPIST-1 f with JWST NIRISS SOSS: No Evidence for the Transit Light Source Effect
arXiv 2608.17207 · 2026-09-02CUSTOS：实现捕获-遏制边界处具备取证就绪能力的零信任
CUSTOS: Toward Forensic-Ready Zero Trust at the Capture-Containment Boundary
arXiv 2608.17068 · 2026-09-02HarnessEval-W：将视觉世界评估智能体化
HarnessEval-W: Agentifying the Evaluation of Visual Worlds
arXiv 2608.16859 · 2026-09-02基于推理的汽车电子电气组件鲁棒性验证
Reasoning-supported Robustness Validation of Automotive E/E Components
arXiv 2608.16421 · 2026-09-02深度思维对齐：面向视频推理的轨迹级潜在知识蒸馏
Deep Thought Alignment: Trajectory-Level Latent Distillation for Video Reasoning
arXiv 2608.16316 · 2026-09-02持续学习规则塑造表征漂移
Continual-learning rules shape representational drift
arXiv 2608.16141 · 2026-09-02Auslander条件下自内射维数的对称性
Symmetry of self-injective dimensions under the Auslander condition
arXiv 2608.15974 · 2026-09-02可数范畴Engel群的李方法：4-Engel 5-群的Wilson猜想
Lie methods for countably categorical Engel groups: the Wilson conjecture for $4$-Engel $5$-groups
arXiv 2608.15664 · 2026-09-02考虑强气动力的可重复使用火箭动力着陆的浓缩PIPG序列凸优化
Condensed PIPG Sequential Convex Optimization for Reusable-Rocket Powered Landing with Strong Aerodynamics
arXiv 2608.15582 · 2026-09-02
