2026-10论文中文摘要
MatrixReward：基于评分矩阵的开放式生成奖励机制
MatrixReward: Reward from Rubric Matrix for Open-Ended Generation
arXiv 2610.00389 · 2026-10-02T2SPO：面向智能体强化学习的轨迹到步骤策略优化
T2SPO: Trajectory-to-Step Policy Optimization for Agentic Reinforcement Learning
arXiv 2610.00388 · 2026-10-02更快的稳定数值多项式乘法
Faster Stable Numerical Polynomial Multiplication
arXiv 2610.00387 · 2026-10-02FAER：面向语言模型后训练的可审计、效用对齐的轨迹重放
FAER: Auditable Utility-Aligned Trajectory Replay for Language Model Post-Training
arXiv 2610.00385 · 2026-10-02RIQE：一种用于计算机断层扫描的NIQE风格参考模型
RIQE: a NIQE-style reference model for Computed Tomography
arXiv 2610.00384 · 2026-10-02EvoGen-Harness：学习在何处以及如何演化图像生成外部系统
EvoGen-Harness: Learning Where and How to Evolve Image-Generation Harnesses
arXiv 2610.00383 · 2026-10-02关于模型量化与模型反演攻击之间关系的研究
On the Relationship between Model Quantization and Model Inversion Attacks
arXiv 2610.00382 · 2026-10-02OmniMed-Jev：通过系统一校准LVLM置信度以实现可信的医疗多模态决策
OmniMed-Jev: Calibrating LVLM Confidence for Trustworthy Medical Multimodal Decisions via System One
arXiv 2610.00381 · 2026-10-02基于时频图像融合技术预测rTMS抑郁症治疗结果
Fusion techniques of time frequency-based images to predict the outcome of rTMS depression therapy
arXiv 2610.00380 · 2026-10-02Carlitz 模的一族具有无界解析秩的显式扭变
An explicit family of twists of the Carlitz module with unbounded analytic rank
arXiv 2610.00379 · 2026-10-02Higgs 丛的联合模空间的分层
Stratification of Joint Moduli Spaces of Higgs Bundles
arXiv 2610.00378 · 2026-10-02STCFormer：用于站点天气预报的自适应时空建模与动态聚类Transformer
STCFormer: Adaptive Spatio-Temporal Modeling with Dynamic Cluster Transformer for Station-based Weather Forecasting
arXiv 2610.00377 · 2026-10-02Jev在网络流量分类中的初探：准确性、处理时间与成本
A First Glance at Jev for Network Traffic Classification: Accuracy, Processing Time, and Cost
arXiv 2610.00376 · 2026-10-02三角形的最优差异
Optimal discrepancy for triangles
arXiv 2610.00375 · 2026-10-02多模态深度研究中忠实的图表生成：框架-证据协同自适应
Faithful Chart Generation for Multimodal Deep Research: Frame-Evidence Co-Adaptation
arXiv 2610.00374 · 2026-10-02注意力头消融何时支持因果主张？投影层混杂、地板效应与匹配对照
When Do Attention-Head Ablations Support Causal Claims? Projection-Level Confounds, Floor Effects, and Matched Controls
arXiv 2610.00373 · 2026-10-02当脚手架失去信号：LLM智能体恢复的因果评估
When Harnesses Lose the Signal: Causal Evaluation of Recovery in LLM Agents
arXiv 2610.00372 · 2026-10-02拒绝而不禁用：多智能体系统的授权配对评估与控制
Deny Without Disabling: Authorization-Paired Evaluation and Control for Multi-Agent Systems
arXiv 2610.00371 · 2026-10-02M$^2$Weather：联合多站点多变量天气预报的基准
M$^2$Weather: A Benchmark for Joint Multi-Station and Multi-Variable Weather Forecasting
arXiv 2610.00370 · 2026-10-02对模型生成文本的共同偏好："AI-AI 偏见"的生成器-选择器矩阵未显示可检测的自模型溢价
A Shared Taste for Model-Written Text: The Generator-by-Selector Matrices of "AI-AI Bias" Show No Detectable Own-Model Premium
arXiv 2610.00369 · 2026-10-02DeepJEPA：从内部扩展世界模型
DeepJEPA: Scaling World Models from Within
arXiv 2610.00368 · 2026-10-02MoRA：通过路由器偏置学习与专家近似进行MoE剪枝
MoRA: MoE Pruning via Router Bias Learning and Expert Approximation
arXiv 2610.00367 · 2026-10-02智能体应该记住什么？在有限记忆评估中区分保留与检索
What Should an Agent Remember? Disentangling Retention from Retrieval in Bounded-Memory Evaluation
arXiv 2610.00366 · 2026-10-02流形约束的初始噪声优化用于高效生成模型对齐
Manifold-Constrained Initial Noise Optimization for Efficient Generative Model Alignment
arXiv 2610.00365 · 2026-10-02关于代数的广泛可均性
On extensive amenability of algebras
arXiv 2610.00364 · 2026-10-02深度学习在铁路系统异常检测中的应用：结构化综述
Deep Learning for Anomaly Detection in Railway Systems: A Structured Survey
arXiv 2610.00363 · 2026-10-02Frobenius重数及格拉斯曼二平面簇的F-签名
Frobenius multiplicities and F-signatures of Grassmannians of two-planes
arXiv 2610.00362 · 2026-10-02极化Calabi--Yau流形有限性的多势理论方法
A Pluripotential-Theoretic Approach to Finiteness of Polarized Calabi--Yau Manifolds
arXiv 2610.00361 · 2026-10-02DexPolicy: 轨迹引导灵巧操作的调度探索
DexPolicy: Scheduled Exploration for Trajectory-Guided Dexterous Manipulation
arXiv 2610.00360 · 2026-10-02扩散模型软掩码编辑：图像与视频的像素级可调强度重绘
Diffusion Editing with Soft Mask: Pixel Level Redo of Image and Video with Adjustable Strength
arXiv 2610.00359 · 2026-10-02Mordell判别不等式的五点情形
The Five-Point Case of Mordell's Discriminant Inequality
arXiv 2610.00358 · 2026-10-02隐藏态更新与可观测记录复合在逆因果模型中的研究
Hidden-State Updates and observable-record composition in retrocausal models
arXiv 2610.00357 · 2026-10-02电力系统中强迫振荡源的可辨识性极限
Identifiability Limits of Forced Oscillation Sources in Power Systems
arXiv 2610.00356 · 2026-10-02IndoorBEV：面向室内移动机器人的轻量级实时LiDAR BEV感知系统
IndoorBEV: A Lightweight Real-Time LiDAR BEV Perception System for Indoor Mobile Robots
arXiv 2610.00355 · 2026-10-02证明门控签名：在状态漂移下保持有效的求解器检查交易防护，用于链上AI智能体
Proof-Gated Signing: Solver-Checked Transaction Guards that Hold Under State Drift for Onchain AI Agents
arXiv 2610.00354 · 2026-10-02JusticeAxis：刚性规则适用与无依据自由裁量之间的法律判决基准
JusticeAxis: Benchmarking Legal Judgment between Rigid Rule Application and Ungrounded Discretion
arXiv 2610.00353 · 2026-10-02预测精度可选择所需精度的标度律
Prediction Accuracy Can Select the Scaling Law of Required Precision
arXiv 2610.00352 · 2026-10-02多组分合金中多模态调幅分解的热力学起源
Thermodynamic origins of multimode spinodal decomposition in multicomponent alloys
arXiv 2610.00351 · 2026-10-02Vmem-$\phi$：基于膜电位统计的脉冲神经网络低计算量分布外检测
Vmem-$φ$: Low-Compute Out-of-Distribution Detection in Spiking Neural Networks from Membrane-Potential Statistics
arXiv 2610.00350 · 2026-10-02分布式多智能体委托中的容错预算保持
Fault-Tolerant Budget Conservation in Distributed Multi-Agent Delegation
arXiv 2610.00349 · 2026-10-02移除大海捞针：通过权重正交化在大型语言模型中去除后门
Removing the NEEDLE in the Haystack: Backdoor Removal in LLMs via Weight Orthogonalisation
arXiv 2610.00348 · 2026-10-02自修改AI智能体群体的授权：在替换、分叉和回滚中保持权限
Authorization for Self-Modifying AI Agent Populations: Conserving Authority across Replacement, Forking, and Rollback
arXiv 2610.00347 · 2026-10-02无梯度凸优化与凸-凹鞍点问题的下界
Lower Bounds For Gradient-Free Convex Optimization And Convex-Concave Saddle-Point Problems
arXiv 2610.00345 · 2026-10-02度量扩张定理
A Metric Extension Theorem
arXiv 2610.00344 · 2026-10-02关于线性群的非阿贝尔张量平方的一个注记
A note on the non-abelian tensor squares of linear groups
arXiv 2610.00343 · 2026-10-02自发对称性破缺作为BRST不变形式中Gribov标度的机制
Spontaneous symmetry breaking as a mechanism for the Gribov scale in a BRST-invariant formulation
arXiv 2610.00342 · 2026-10-02UnifiedAttack：评估大型多模态模型在协同有害图文生成中的安全性
UnifiedAttack: Evaluating the Safety of Large Multimodal Models in Synergistic Harmful Image-Text Generation
arXiv 2610.00341 · 2026-10-02短期障碍期权价格展开
Short-term barrier option price expansion
arXiv 2610.00340 · 2026-10-02超越 $\sigma=2p$ 的临界拟线性 Hartree 方程的相变、尖锐渐近与对称性
Phase transition, sharp asymptotics and symmetry for critical quasilinear Hartree equations beyond $σ=2p$
arXiv 2610.00339 · 2026-10-02学生与人工智能互动的三条路径：面向高阶思维的约束优先设计
Three Pathways of Student-AI Interaction: Constraint-First Design for Higher-Order Thinking
arXiv 2610.00338 · 2026-10-02
