2026-10论文中文摘要
揭示残差流动态中的非受控重复
Uncovering Uncontrolled Repetition through Residual Stream Dynamics
arXiv 2609.38802 · 2026-10-01已知类型的不可约辛簇的超奇异Tate猜想：I
Supersingular Tate conjecture for irreducible symplectic varieties of known types: I
arXiv 2609.38801 · 2026-10-01突破散粒噪声极限：用于量子优化的缓存回收方差缩减梯度
Breaking the Shot-Noise Barrier: Cached Recycled Variance-Reduced Gradients for Quantum Optimization
arXiv 2609.38800 · 2026-10-01重叠、独特与冲突：大语言模型能否提取它们能识别的信息？
Overlap, Unique and Conflict: Can LLMs Extract What They Can Recognize?
arXiv 2609.38799 · 2026-10-01GraphCert：通过认证证据准则引导智能体图推理
GraphCert: Bootstrap Agentic Graph Reasoning with Certified Evidence Rubrics
arXiv 2609.38798 · 2026-10-01评估在不断演化的模型知识下的持久校准
Evaluating Persistent Calibration under Evolving Model Knowledge
arXiv 2609.38797 · 2026-10-01径向加速度关系的加速度尺度是否追踪宇宙膨胀速率？
Is the acceleration scale of the radial acceleration relation tracking the cosmic expansion rate?
arXiv 2609.38796 · 2026-10-01屏障还是助推？熟悉度对使用AI语音克隆识别普通话情绪韵律的影响
A barrier or a booster? Familiarity effects on Mandarin emotion prosody recognition using AI-powered voice cloning
arXiv 2609.38794 · 2026-10-01从语言反馈中通过位置选择性自蒸馏训练LLM裁判
Training LLM Judges from Language Feedback via Position-Selective Self-Distillation
arXiv 2609.38792 · 2026-10-01重新审视大地电磁理论：层状感应与地电场
Revisiting magnetotelluric theory: layered induction and geoelectric fields
arXiv 2609.38791 · 2026-10-01同步蒙特卡洛方法及其在平均场自旋玻璃模型中的应用
Synchronous Monte Carlo method and its application to a mean-field spin glass model
arXiv 2609.38790 · 2026-10-01LEARN-TS：基于正态性引导的LLM增强对齐与重构用于多元时间序列异常检测
LEARN-TS: LLM-Enhanced Alignment and Reconstruction with Normality Guidance for Multivariate Time-Series Anomaly Detection
arXiv 2609.38789 · 2026-10-01正面评分，隐藏担忧：AI 中介的组织倾听中员工意见披露
Positive Ratings, Hidden Concerns: Employee Voice Disclosure in AI-Mediated Organizational Listening
arXiv 2609.38788 · 2026-10-01高维Anosov流与部分双曲流中强不稳定叶理的刚性
Rigidity from the strong unstable foliation for higher-dimensional Anosov and partially hyperbolic flows
arXiv 2609.38787 · 2026-10-01相同损失，不同梯度
Same Loss, Different Gradients
arXiv 2609.38786 · 2026-10-01数值精度达到多少才算足够？
How Accurate Is Accurate Enough?
arXiv 2609.38785 · 2026-10-01在概率对比学习中，共享温度是否意味着共享角度尺度？
Does a Shared Temperature Imply a Shared Angular Scale in Probabilistic Contrastive Learning?
arXiv 2609.38784 · 2026-10-01截断死亡下部分识别效用的政策评估与学习
Policy evaluation and learning with partially identified utility under truncation by death
arXiv 2609.38783 · 2026-10-01AI语音代理中的角色与说服性框架：一项针对儿童的$2\ imes2$实地实验
Persona and Persuasive Framing in AI Voice Agents: A $2\times2$ Field Experiment with Children
arXiv 2609.38782 · 2026-10-01ChartDensity-Bench：在视觉密度下对多模态大语言模型进行数值数据重建的基准测试
ChartDensity-Bench: Benchmarking MLLMs for Numerical Data Reconstruction under Visual Density
arXiv 2609.38781 · 2026-10-01RAST：面向低分辨率音频活动识别的分辨率感知特权结构迁移
RAST: Resolution-Aware Privileged Structure Transfer for Low-Resolution Audio Activity Recognition
arXiv 2609.38780 · 2026-10-01Motzkin-Straus优化在熵计算平台上的应用
Motzkin-Straus Optimization on an Entropy-Computing Platform
arXiv 2609.38779 · 2026-10-01动作条件双模拟用于GUI智能体记忆
Action Conditioned Bisimulation For GUI Agent Memory
arXiv 2609.38778 · 2026-10-01蒸馏视觉证据，而非仅蒸馏答案：面向视觉语言模型的跨世界在线策略蒸馏
Distill the Visual Evidence, Not Just the Answer: Cross-World On-Policy Distillation for Vision-Language Models
arXiv 2609.38777 · 2026-10-01浅电路中可实验验证的量子优势
Experimentally Testable Quantum Advantage in Shallow Circuits
arXiv 2609.38775 · 2026-10-01太阳活动区暗条的磁场诊断
Magnetic field diagnostics of a solar active region filament
arXiv 2609.38774 · 2026-10-012-orbifolds 上的稳态梯度 Ricci Yang-Mills 孤立子
Steady gradient Ricci Yang-Mills solitons on 2-orbifolds
arXiv 2609.38773 · 2026-10-01计数不含固定长度线性环的超图
Counting hypergraphs without linear cycles of fixed length
arXiv 2609.38772 · 2026-10-01瑞利-伍德阈值控制原子阵列中的非线性光子关联
Rayleigh-Wood Threshold Controls Nonlinear Photon Correlations in Atomic Arrays
arXiv 2609.38770 · 2026-10-01遗忘中的学习：随机训练动态中的统计支持选择性保留
Learning Under Forgetting: Statistical Support-Selective Retention in Stochastic Training Dynamics
arXiv 2609.38768 · 2026-10-01PathAnchor：面向科学智能体的路径结构化证据
PathAnchor: Path-Structured Evidence for Scientific Agents
arXiv 2609.38766 · 2026-10-01使用马尔可夫决策过程的多期债券组合优化与交易成本
Multiperiod bond portfolio optimization with transaction costs using a Markov Decision process
arXiv 2609.38765 · 2026-10-01抽象预训练在困惑度之外的持久影响
Lasting Effects of Abstract Pretraining Beyond Perplexity
arXiv 2609.38764 · 2026-10-012000-2025年北极沿海气旋影响交换流的长期变化
Long-term Variations of Cyclone-Influenced Exchange Flows in the Coastal Arctic in 2000-2025
arXiv 2609.38763 · 2026-10-01Adaptive-GEPA：让您的工具适配异构请求
Adaptive-GEPA: Make Your Harness Fit Heterogeneous Requests
arXiv 2609.38762 · 2026-10-01多智能体系统在何处失败？集体机制的证据基础诊断
Where Do Multi-Agent Systems Fail? Evidence-Grounded Diagnosis of Collective Mechanisms
arXiv 2609.38761 · 2026-10-01HALO：面向车辆路径问题的基于局部观测的异构分配
HALO: Heterogeneous Allocation Via Localized Observations for the Vehicle Routing Problem
arXiv 2609.38760 · 2026-10-01BOP2-ENR：一种用于自适应富集设计的贝叶斯最优II期设计，控制家系错误率并限制联合声明功效
BOP2-ENR: a Bayesian optimal phase II design for adaptive enrichment with family-wise error rate control and a bound on joint-claim power
arXiv 2609.38759 · 2026-10-01事件驱动的刷新与循环记忆以减少指代视频对象分割中的过时定位
Event-Driven Refresh and Recurrence Memory to Reduce Stale Grounding in Referring Video Object Segmentation
arXiv 2609.38758 · 2026-10-01自演化算法设计智能体：通过种群策展策略优化逃离上下文演化停滞
Self-Evolving Algorithm-Design Agents: Escaping In-Context Evolutionary Stagnation via Population-Curated Policy Optimization
arXiv 2609.38757 · 2026-10-01单层WSi2P4中多晶型可调的自旋织构与激子结构
Polymorph-Tunable Spin Texture and Excitonic Structures in Monolayer WSi2P4
arXiv 2609.38756 · 2026-10-01基于学习排序与验证的智能体相对相机位姿估计
Agentic Relative Camera Pose Estimation via Learned Ranking and Verification
arXiv 2609.38755 · 2026-10-01重新分类以符合要求
Reclassified to Conform
arXiv 2609.38754 · 2026-10-01证据所在之处：审计AI伴侣的自我描述
Where the Evidence Lives: Auditing AI Companions' Self-Descriptions
arXiv 2609.38753 · 2026-10-01LLM能否帮助发现协议规范中的歧义？
Can LLMs help find Ambiguities in Protocol Specifications?
arXiv 2609.38752 · 2026-10-01广义半 Clifford 猜想在第 4 层成立
The Generalized Semi-Clifford Conjecture Holds at Level 4
arXiv 2609.38751 · 2026-10-01圆二色光谱的视觉理解：相互作用发色团的情形
A Visual Understanding of Circular Dichroism Spectroscopy: The Case of Interacting Chromophores
arXiv 2609.38750 · 2026-10-01具有近似少量ANOVA项的张量的最坏情况完备化
Worst-Case Completion of Tensors with Approximately Few ANOVA Terms
arXiv 2609.38749 · 2026-10-01共识感知的多源融合用于参考引导的伪装目标检测
Consensus-Aware Multi-Source Fusion for Reference-Guided Camouflaged Object Detection
arXiv 2609.38747 · 2026-10-01Matisse：用于主动三维重建的证据空间推理
Matisse: Evidence-Space Reasoning for Active 3D Reconstruction
arXiv 2609.38746 · 2026-10-01
