2026-09论文中文摘要
华为的τ芯片本应因过热而失效？
Huawei's $τ$ Chip Was Supposed to Melt?
arXiv 2609.04287 · 2026-09-07从匹配模型到招聘智能体：AI招聘系统、评估与治理的系统化叙事综述
From Matching Models to Recruiting Agents: A Systematized Narrative Review of AI Recruitment Systems, Evaluation, and Governance
arXiv 2609.04286 · 2026-09-07论偶维球面及Eilenberg–MacLane空间乘积作为分类空间的有理实现
On The Rational Realization of Even-dimensional Spheres and Products of Eilenberg--MacLane Spaces as Classifying Spaces
arXiv 2609.04285 · 2026-09-07每个亏格为正的闭定向曲面都是Loewner曲面
Every Closed Oriented Surface of Positive Genus Is Loewner
arXiv 2609.04284 · 2026-09-07基于样本引导分布匹配的视频生成联合对齐与蒸馏
Joint Alignment and Distillation for Video Generation via Sample-Guided Distribution Matching
arXiv 2609.04283 · 2026-09-07退一步以进千里：面向视觉生成的感知反思偏好优化
Step Back to Move Forward: Reflection-Aware Preference Optimization for Visual Generation
arXiv 2609.04282 · 2026-09-07当视觉凌驾于认知之上：用于视觉语言模型个性化安全的视觉主导性与基于 deferral 的方法
When Seeing Overrides Knowing: Visual Dominance and Deferral-Based Method for Personalized Safety in VLMs
arXiv 2609.04281 · 2026-09-07关于Σ类中凸函数的Pommerenke问题的一个反例
A Counterexample to a Problem of Pommerenke on Convex Functions in the Class $Σ$
arXiv 2609.04279 · 2026-09-07面向AI驱动的纳米药物发现：用于纳米自组装预测的基准测试与多模态学习框架
Towards AI-Driven Nanomedicine Discovery: A Benchmark and Multimodal Learning Framework for Nano Self-Assembly Prediction
arXiv 2609.04278 · 2026-09-07FailureSpot：面向视觉-语言-动作模型的标签高效时间戳级故障检测
FailureSpot: Label-Efficient Timestamp-Level Failure Detection for Vision-Language-Action Models
arXiv 2609.04277 · 2026-09-07FailSAE：基于稀疏自编码器的视觉语言模型可解释故障预测研究
FailSAE: Towards Interpretable Failure Prediction for Vision-Language Models via Sparse Autoencoders
arXiv 2609.04276 · 2026-09-07二元Laurent多项式环上的一个显式非初等矩阵
An explicit non-elementary matrix over a bivariate Laurent polynomial ring
arXiv 2609.04275 · 2026-09-07多尺度图像表示压缩
Multi-scale Image Representation Compression
arXiv 2609.04274 · 2026-09-07可扩展神经视频表示压缩
Scalable Neural Video Representation Compression
arXiv 2609.04273 · 2026-09-07评估大型语言模型在强迫停运风险预测中的应用：优势及与机器学习的对比
Evaluating Large Language Models for Forced Outage Risk Prediction: Benefits and Comparison to Machine Learning
arXiv 2609.04272 · 2026-09-07量子辅助的参数密集型Wi-Fi人体活动识别的内存高效训练
Quantum-Assisted Memory-Efficient Training for Parameter-Intensive Wi-Fi-Based Human Activity Recognition
arXiv 2609.04271 · 2026-09-07评审者能力决定拒绝目标而非修复技能：来自大语言模型执行-评审-修正流程的证据
Reviewer Capability Governs Rejection Targeting, Not Repair Skill: Evidence from LLM Execute-Review-Revise Pipelines
arXiv 2609.04270 · 2026-09-07企业家族关系解析并非字符串匹配问题：按名称可见性分层的公共基准
Corporate-Family Resolution Is Not a String-Matching Problem: A Public Benchmark Stratified by Name Visibility
arXiv 2609.04269 · 2026-09-07MarkerScout：一种用于从多尺度机制模型预测生物标志物的疾病无关机器学习框架
MarkerScout: A Disease-Agnostic Machine Learning Framework for Biomarker Prediction from Multi-Scale Mechanistic Models
arXiv 2609.04268 · 2026-09-07一种基于风洞实验观测的航空航天代理模型数据融合框架
A Data Fusion Framework for Grounding Aerospace Surrogate Model via Experimental Wind-Tunnel Observations
arXiv 2609.04267 · 2026-09-07存算一体注意力：具有RC可调温度的时域模拟Softmax电路
Compute-in-Memory Attention: A Time-Domain Analog Softmax Circuit with RC-Tunable Temperature
arXiv 2609.04266 · 2026-09-07通过低秩注意力适配实现量化KV缓存的质量恢复
Quality Recovery for Quantized KV Caches via Low-Rank Attention Adaptation
arXiv 2609.04263 · 2026-09-07面向日语音乐搜索查询的低延迟拼写纠错
Low-Latency Spell Correction for Japanese Music Search Queries
arXiv 2609.04262 · 2026-09-07基于LeJEPA的分子图编码器自监督预训练
Self-Supervised Pretraining of Molecular Graph Encoders with LeJEPA
arXiv 2609.04261 · 2026-09-07GhostWord：针对自动语音识别的细粒度后门攻击
GhostWord: A Fine-Grained Backdoor Attack on Automatic Speech Recognition
arXiv 2609.04260 · 2026-09-07面向深度神经网络推理加速器芯片的硬件感知软件训练，用于恢复硬件变异导致的精度下降
Hardware-conscious Software Training for Deep Neural Network Inference Accelerator Chips to Recover Accuracy Degradation due to Hardware Variabilities
arXiv 2609.04259 · 2026-09-07极小体积等角双曲4-多面体
Minimal-Volume Equiangular Hyperbolic $4$-Polytopes
arXiv 2609.04258 · 2026-09-07关于特施纳束缚数猜想的一个反例
A Counterexample to Teschner's Bondage-Number Conjecture
arXiv 2609.04257 · 2026-09-07探测支持语音的大语言模型（LLM）在心理健康对话中由温暖感介导的伤害
Probing Warmth-Mediated Harm in Speech-Enabled LLMs for Mental-Health Conversations
arXiv 2609.04256 · 2026-09-07SAGE：面向多实体视觉检索的语义属性图
SAGE: Semantic Attribute Graphs for Multi-Entity Visual Retrieval
arXiv 2609.04255 · 2026-09-07扩展不确定原理黑洞的固定ADM重构
Fixed ADM reconstruction of extended uncertainty principle black holes
arXiv 2609.04254 · 2026-09-07AVENUE：音频-视频编辑的理解与评估
AVENUE: Audio-Video EditiNg Understanding and Evaluation
arXiv 2609.04253 · 2026-09-07三电极TiO₂ + rGO电池：经典控制及量子失谐框架内的解释
Three-Electrode TiO2 + rGO Cells: Classical Controls and Interpretation within a Quantum-Discord Framework
arXiv 2609.04252 · 2026-09-07一类二维Lempert流形的Kobayashi等距映射
On the Kobayashi isometries of a class of 2-dimensional Lempert manifolds
arXiv 2609.04251 · 2026-09-07Encore：自适应信号路由的无限视听生成
Encore: Infinite Audio-Video Generation with Adaptive Signal Routing
arXiv 2609.04249 · 2026-09-07面向确定性齿轮数字孪生的物理驱动型FPGA齿面接触计算
Physics-Direct FPGA Tooth-Contact Computation for Deterministic Gear Digital Twins
arXiv 2609.04248 · 2026-09-07CAD：用于减轻全模态大语言模型跨模态幻觉的冲突感知解码
CAD: Conflict-Aware Decoding to Mitigate Cross-Modal Hallucinations in Omnimodal Large Language Models
arXiv 2609.04247 · 2026-09-07借助简化计算机辅助的森多夫猜想的标量优化证明
A Scalar Optimization Proof of Sendov's Conjecture with Reduced Computer Assistance
arXiv 2609.04246 · 2026-09-07基于自适应代码空间对齐与局部大语言模型优化的自回归语音增强的对齐解码
Grounded Decoding for Autoregressive Speech Enhancement via Adaptive Code-Space Grounding and Local LLM Refinement
arXiv 2609.04245 · 2026-09-07MonoMoE：一种用于量化MoE解码的高效融合超大内核
MonoMoE: An Efficient Fused Mega-kernel for Quantized MoE Decoding
arXiv 2609.04244 · 2026-09-07多维度偏好建模中的多维度偏差：评估合成智能体在联合实验中替代人类参与者的能力
Multi-dimensional Bias in Modeling Multi-dimensional Preferences: Evaluating the Ability of Synthetic Agents to Replace Human Participants in Conjoint Experiments
arXiv 2609.04243 · 2026-09-07无训练的以语音为中心的全模态理解：基于冻结的视觉语言模型
Training-Free Speech-Centric Omni Understanding with Frozen VLMs
arXiv 2609.04242 · 2026-09-07VocalCoachBench：评估音频-语言模型在歌唱领域专家反馈上的基准
VocalCoachBench: Benchmarking Audio-Language Models on Expert Feedback for Singing
arXiv 2609.04241 · 2026-09-07用于全双工对话系统的TurnFSM：内化状态机逻辑以实现流式语义语音活动检测与语句级拒绝
TurnFSM for Full-Duplex Dialogue System: Internalizing State-Machine Logic for Streaming Semantic Voice Activity Detection and Utterance-Level Rejection
arXiv 2609.04240 · 2026-09-07字节预算：面向存储受限的大语言模型解码的窗口式存储屋顶线与双预算架构 ablation（消融）实验
Budgeting Bytes: A Windowed Storage Roofline and Dual-Budget Architecture Ablations for Storage-Bound LLM Decoding
arXiv 2609.04238 · 2026-09-07重新思考语音编解码器：从压缩到自回归生成建模
Rethinking Speech Codecs: From Compression to Autoregressive Generative Modeling
arXiv 2609.04237 · 2026-09-07声调-词义冲突下的鲁棒语音情感识别：基准与框架
Robust Speech Emotion Recognition under Tone-Word Conflict: A Benchmark and Framework
arXiv 2609.04236 · 2026-09-07各向异性在$f(Q)$引力中的作用：来自宇宙学观测的见解
The role of anisotropy in $f(Q)$ gravity: insights from cosmological observations
arXiv 2609.04235 · 2026-09-07用相互作用暗能量模型研究巧合问题
Investigating the coincidence problem with Interacting Dark Energy Models
arXiv 2609.04234 · 2026-09-07利用动力系统方法研究Starobinsky-Grisaru-Zanon引力中de Sitter解的稳定性
Stability analysis of de Sitter solution in the Starobinsky-Grisaru-Zanon gravity using the dynamical system method
arXiv 2609.04233 · 2026-09-07
