2026-08论文中文摘要
LabEvolver：面向安全且务实的湿实验室智能体的无训练经验演化框架
LabEvolver: Training-Free Experience Evolution for Safe and Grounded Wet-Lab Agents
arXiv 2607.27690 · 2026-08-03OneShot：面向大规模检索的结合神经打分的排序内索引方法
OneShot: Index-in-Ranking with Neural Scoring for Large-Scale Retrieval
arXiv 2607.27475 · 2026-08-03AI生成音乐检测的鲁棒性改进
Improved Robustness in AI-Generated Music Detection
arXiv 2607.27454 · 2026-08-03平面图中偶长度路径的最大数量
The maximum number of paths of even length in a planar graph
arXiv 2607.27284 · 2026-08-03基于EEG的疾病诊断再思考：将实例表示学习与受试者级监督解耦
Rethinking EEG-Based Disease Diagnosis: Decoupling Instance Representation Learning from Subject-Level Supervision
arXiv 2607.27274 · 2026-08-03DenseOn与LateOn：面向多语言、长上下文及代码检索的全开放密集型与晚交互模型
DenseOn with the LateOn: Fully Open Dense and Late-Interaction Models for Multilingual, Long-Context, and Code Search
arXiv 2607.27178 · 2026-08-03Stanley–Gasharov猜想的无穷反例族
Two infinite families of counterexamples to the Stanley--Gasharov conjecture
arXiv 2607.27166 · 2026-08-03用于高效眼前节分割的DINOv3特征的步注意力细化
Step-Attention Refinement of DINOv3 Features for Efficient Anterior Eye Segmentation
arXiv 2607.27087 · 2026-08-03InkShield：抵御未授权笔迹模仿的书写风格保护
InkShield: Writing Style Protection Against Unauthorized Handwriting Mimicry
arXiv 2607.26976 · 2026-08-03用于持续指令微调的渐进式多模态对齐
Progressive Multimodal Alignment for Continual Instruction Tuning
arXiv 2607.26947 · 2026-08-03SERPO：面向开放式测试时强化学习的自进化规则策略优化
SERPO: Self-Evolving Rubric Policy Optimization for Open-Ended Test-Time Reinforcement Learning
arXiv 2607.26873 · 2026-08-03CaIRec：面向不完整多模态推荐的校准模态补全方法
CaIRec: Calibrated Modality Imputation for Incomplete Multimodal Recommendation
arXiv 2607.26720 · 2026-08-03指纹驱动自动化：将侦察与POC验证相结合
Fingerprint-Driven Automation: Coupling Reconnaissance with POC Verification
arXiv 2607.26655 · 2026-08-03金字塔与扩展度量测度空间
Pyramids and Extended Metric Measure Spaces
arXiv 2607.26626 · 2026-08-03多通道戴森方程的代数图构造
Algebraic Diagrammatic Construction of the Multichannel Dyson Equation
arXiv 2607.26609 · 2026-08-03用于主张-证据可追溯性的证据账本裁决
Evidence-Ledger Adjudication for Claim-Evidence Traceability
arXiv 2607.26512 · 2026-08-03欧洲XFEL产生高通量高强度硬X射线阿秒脉冲
Generation of high-fluence and high-intensity hard x-ray attosecond pulses at European XFEL
arXiv 2607.26505 · 2026-08-03哪种RAG范式在规模化场景下表现最优？检索增强生成范式的规模化研究
BM25 Wins at Scale: A Scaling Study of Retrieval-Augmented Generation Paradigms
arXiv 2607.26497 · 2026-08-03CMT-RAG：面向多轮多跳检索增强生成（RAG）的互补记忆轨迹
CMT-RAG: Complementary Memory Traces for Multi-turn Multi-hop RAG
arXiv 2607.26470 · 2026-08-03击中SLE$_2$的小布朗环：一个精确的自然含量极限
Small Brownian Loops Hitting SLE$_2$: An Exact Natural-Content Limit
arXiv 2607.26439 · 2026-08-03活性无序介质中阻塞现象的非厄米随机矩阵理论
Non-Hermitian Random Matrix Theory of Jamming in Active Disordered Media
arXiv 2607.26406 · 2026-08-03无法隐藏的秘密：揭示原生大语言模型集成开发环境（LLM-native IDEs，简称LIDEs）的安全与隐私问题
Impossible to hide secret ...: Uncovering Security and Privacy Issues in LLM-native IDEs
arXiv 2607.26390 · 2026-08-03失真低于3的度量投票中的敏感性与差分隐私
Sensitivity and Differential Privacy in Metric Voting with Distortion below Three
arXiv 2607.26388 · 2026-08-03通过微波辅助碰撞对分子阵列进行确定性加载
Deterministic loading of molecular arrays by microwave-assisted collisions
arXiv 2607.25783 · 2026-08-03极性分子碰撞稳定混合物中可调谐的状态依赖相互作用
Tunable state-dependent interactions in collisionally stable mixtures of polar molecules
arXiv 2607.25777 · 2026-08-03超冷极性分子之间的转动-振动范德瓦尔斯相互作用
Ro-vibrational van der Waals interaction between ultracold polar molecules
arXiv 2607.25774 · 2026-08-03三维临界Hartree型布雷齐斯-尼伦伯格问题的爆破渐近性
Blow-up asymptotics for a critical Hartree-type Brézis--Nirenberg problem in dimension three
arXiv 2607.25759 · 2026-08-03最小超空间广义不确定性原理变形的晚期宇宙学测试
Late-Time Cosmological Tests of a Minisuperspace Generalized-Uncertainty-Principle Deformation
arXiv 2607.25745 · 2026-08-03超越背景偏差：用于数据集蒸馏的显著性驱动原型对齐
Dataset Distillation Based on Saliency-Driven Prototype Alignment
arXiv 2607.25318 · 2026-08-03FORGE：用于长视频理解的相关几何中的帧正交性
FORGE: Frame Orthogonality in Relevance Geometry for Long-Form Video Understanding
arXiv 2607.25266 · 2026-08-03共形级联：多层大语言模型推理的无分布精度保证
Conformal Cascade: Distribution-Free Accuracy Guarantees for Multi-Tier LLM Inference
arXiv 2607.25018 · 2026-08-03真空中引力波的衰减：量子化引力的低能效应
Gravitational waves decay in vacuum
arXiv 2607.24930 · 2026-08-03自由积的迹根与余中心
Trace radicals and cocenters of free products
arXiv 2607.24697 · 2026-08-03用于审计大语言模型社会模拟器的基于推理的行为模型
Reason-Mediated Behavioral Models for Auditing LLM Social Simulators
arXiv 2607.24649 · 2026-08-03使用不精确二阶预言机的最优凸优化
Optimal Convex Optimization with Inexact Second-Order Oracles
arXiv 2607.24520 · 2026-08-03LEX-EC：黑盒设置下用于零样本大语言模型人格分类的词汇证据通道审计框架
LEX-EC: A Lexical Evidence-Channel Audit Framework for Zero-Shot LLM Personality Classification in Black-Box Settings
arXiv 2607.24435 · 2026-08-03一个用于时变太阳中微子通量的高精度数值框架，对全球地下实验室进行全地球物质振荡校正
A High-Precision Numerical Framework for Time-Varying Solar Neutrino Flux with Full Earth Matter Oscillation Corrections for Global Underground Laboratories
arXiv 2607.24421 · 2026-08-03超越总体风险：用于大语言模型工具调用的角色分层共形风险控制
Beyond Aggregate Risk: Role-Stratified Conformal Risk Control for LLM Tool Calls
arXiv 2607.24343 · 2026-08-03WorldDiT：用于世界和动作建模的统一扩散架构
WorldDiT: A Unified Diffusion Architecture for World and Action Modeling
arXiv 2607.23909 · 2026-08-03通过电池文献的多模态挖掘利用X射线吸收光谱数据
Harnessing X-ray Absorption Spectroscopy Data through Multimodal Mining of Battery Literature
arXiv 2607.23886 · 2026-08-03OmniVAE：用于联合生成的具有跨模态对齐的音频-视频变分自编码器
OmniVAE: An Audio-Video VAE with Cross-Modal Alignment for Joint Generation
arXiv 2607.23855 · 2026-08-03从可验证奖励强化学习到自验证奖励强化学习：任务转换为开放式语言模型自我改进带来自验证奖励
From RLVR to RLSVR: Task Transformation Induces Self-Verifiable Rewards for Open-Ended LLM Self-Improvement
arXiv 2607.23802 · 2026-08-03基于图神经网络的稀疏车载自组织网络中交通激波的多智能体控制
GNN-based Multi-Agent Control of Traffic Shockwaves in Sparse Vehicular Ad-hoc Networks
arXiv 2607.23792 · 2026-08-03SMARM+：分析与增强实时物联网环境中用于远程认证的混洗测量
SMARM+: Analyzing and Enhancing Shuffled Measurements for Remote Attestation in Real-Time IoT Settings
arXiv 2607.23698 · 2026-08-03有限向量空间、仿射空间和衰减空间中 $\ell_2$-范数下的 Erdős--Ko--Rado 定理
Erdős--Ko--Rado theorems in $\ell_2$-norm for three finite spaces
arXiv 2607.23663 · 2026-08-03用于多屏幕拍摄的抗勾结图像无关水印
Screen-Conditioned Watermarking Against Multi-Screen Collusion Attacks
arXiv 2607.23553 · 2026-08-03基于验证感知架构的大语言模型辅助情报、监视与侦察集群任务级运行时保障
Mission-Level Runtime Assurance for LLM-Assisted ISR Swarms over a Verification-Aware Fabric
arXiv 2607.23532 · 2026-08-03错误且更自信：语言模型参加研究生经济学考试的实地实验
Wrong and More Confident: A Field Experiment on Large Language Models Taking a Graduate Economics Exam
arXiv 2607.23424 · 2026-08-03AlloBench：测量大语言模型智能体中的在线工具分配能力
AllocBench: Measuring Online Tool Allocation Capability in LLM Agents
arXiv 2607.23332 · 2026-08-03基于有噪学生策略性自蒸馏的自增强视觉语言模型
Self-Boosting Vision-Language Models with Noisy Student On-Policy Self-Distillation
arXiv 2607.23125 · 2026-08-03
