2026-10论文中文摘要
总是10:10：参考图像打破仅提示时钟的偏差
It's Always 10:10: Reference Images Break a Bias That Prompts Only Dent
arXiv 2610.11320 · 2026-10-09极端相对论性物质中的中微子振荡：关于托尔曼VI时空的作用
Neutrino oscillations inside ultrarelativistic matter: on the role of the Tolman VI spacetime
arXiv 2610.11319 · 2026-10-09芬斯勒流匹配：用于单快照轨迹推断的动力学感知测地线插值
Finsler Flow Matching: Dynamics-Aware Geodesic Interpolation for Single-Snapshot Trajectory Inference
arXiv 2610.11318 · 2026-10-09DivMoE：通过跨领域专家组合实现细粒度MoE升级
DivMoE: Fine-Grained MoE Upcycling via Cross-Domain Expert Composition
arXiv 2610.11317 · 2026-10-09MetaEncoder：探索带自然语言接口的多模态System One决策双编码器的极限
MetaEncoder: Exploring the Limit of Bi-Encoders for Multimodal System One Decision Making with Natural Language Interface
arXiv 2610.11316 · 2026-10-09降低海森矩阵：用于多模态扩散Transformer的秩-4 W4A4量化
Deflating the Hessian: Rank-4 W4A4 Quantization for Multimodal Diffusion Transformers
arXiv 2610.11315 · 2026-10-09从检索到重构：为长期对话构建可演化的认知记忆
From Retrieval to Reconstruction: Constructing Evolvable Cognitive Memory for Long-Term Dialogue
arXiv 2610.11314 · 2026-10-09两层隐藏层ReLU神经网络中最小化神经元数量的NP-难解性
NP-Hardness of Minimizing Neurons in Two-Hidden-Layer ReLU Neural Networks
arXiv 2610.11313 · 2026-10-09MedBenchAgent：迈向医学视觉语言模型（VLM）基准构建的系统化自动化
MedBenchAgent: Towards Systematic Automation of Medical VLM Benchmark Construction
arXiv 2610.11312 · 2026-10-09中子多重性与裂变竞争的耦合对$^{48}$Ca+$^{244}$Pu反应蒸发剩余截面的影响
The effect of the coupling between the neutron multiplicity and the fission competition on the evaporation residue cross sections of the $^{48}$Ca$+^{244}$Pu reaction
arXiv 2610.11311 · 2026-10-09FloorSAV：基于二维平面图阐明AV-LLM的空间视听上下文
FloorSAV: Elucidating Spatial Audio-Visual Context with 2D Floormap for AV-LLMs
arXiv 2610.11310 · 2026-10-09从几何到泛化：为何行归一化能胜过Adam与Muon
From Geometry to Generalization: Why Row Normalization Can Beat Adam and Muon
arXiv 2610.11309 · 2026-10-09基于UWB测距与有限通信的同构多机器人系统分布式相对定位
Distributed Relative Localization for Homogeneous Multi-Robot Systems through UWB Ranging and Limited Communications
arXiv 2610.11308 · 2026-10-09gr-ntnisac：面向NTN-ISAC的开源GNU Radio测试平台
gr-ntnisac: An Open GNU Radio Testbed for NTN-ISAC
arXiv 2610.11307 · 2026-10-09基于教师偏好样本的协同引导对抗鲁棒蒸馏
Collaboratively Guided Adversarial Robust Distillation with Teacher-Favorable Examples
arXiv 2610.11306 · 2026-10-09BeliefScope：诊断大语言模型中证据驱动的修正与压力诱导的转变
BeliefScope: Diagnosing Evidence-Driven Revision and Pressure-Induced Shifts in Large Language Models
arXiv 2610.11305 · 2026-10-09有限可解李共形代数的结构理论
Structure theory of finite solvable Lie conformal algebras
arXiv 2610.11304 · 2026-10-09用于放射科报告生成的高效多粒度知识迁移
Efficient Multi-Granularity Knowledge Transfer for Radiology Report Generation
arXiv 2610.11303 · 2026-10-09iCATS：基于交互感知稀疏注意力与时步自适应稀疏性的快速视频生成方法
iCATS: Fast Video Generation via Interaction-Aware Sparse Attention and Timestep-Adaptive Sparsity
arXiv 2610.11302 · 2026-10-09任意维数下定常Navier-Stokes方程的齐次解
Homogeneous solutions of the stationary Navier-Stokes equations in arbitrary dimension
arXiv 2610.11301 · 2026-10-09基于大语言模型的代码生成中过度自信失败的表征
Characterizing Overconfident Failure in LLM-Based Code Generation
arXiv 2610.11300 · 2026-10-09DuplexAgent-RSI：用于全双工语音智能体协作的递归工具改进
DuplexAgent-RSI: Recursive Harness Improvement for Full-Duplex Voice Agent Collaboration
arXiv 2610.11299 · 2026-10-09半代数函数上随机次梯度方法的非极小吸引子
Nonminimizing Attractors of Stochastic Subgradient Methods on Semialgebraic Functions
arXiv 2610.11298 · 2026-10-09高质量六方氮化硼（hBN）中缺陷态的位移场诱导电离
Displacement-Field-Induced Ionization of Defect States in High-Quality hBN
arXiv 2610.11297 · 2026-10-09基于MLP-KAN融合的多维信号空间频率感知隐式神经表示
Spatial-Frequency-Aware Implicit Neural Representation of Multidimensional Signals via MLP-KAN Fusion
arXiv 2610.11296 · 2026-10-09正曲率爱因斯坦流形格林函数的黑塞（Hessian）尖锐不等式与单调性
Sharp Hessian inequalities and monotonicity for Green's functions of positively curved Einstein manifolds
arXiv 2610.11295 · 2026-10-09ORDO：面向MIP预求解的操作级回合感知动态排序
ORDO: Operation-level Round-aware Dynamic Ordering for MIP Presolve
arXiv 2610.11294 · 2026-10-09Janus CrMCN4（M = Si/Ge）单层的多功能性：谷电子学物理、压电响应及光催化潜力
Multifunctionality in Janus CrMCN4 (M = Si/Ge) Monolayers: Valleytronic Physics, Piezoelectric Response, and Photocatalytic Potential
arXiv 2610.11293 · 2026-10-09次黎曼测地线轨道流形的结构
On the Structure of Sub-Riemannian Geodesic Orbit Manifolds
arXiv 2610.11292 · 2026-10-09我们何时需要在线策略蒸馏？从离线学生 rollout 中进行蒸馏通常效果更好
When Do We Need On-Policy Distillation? Distilling on Offline Student Rollouts Is Often Better
arXiv 2610.11291 · 2026-10-09ProxyEraseAgent：野外盲水印去除
ProxyEraseAgent: Blind Watermark Removal in the Wild
arXiv 2610.11290 · 2026-10-09基于可能性推理的开放式科学发现
Open-ended Scientific Discovery with Possibilistic Reasoning
arXiv 2610.11289 · 2026-10-09带随机特征的条件扩散模型中的记忆与恶性泛化
Memorization and Malign Generalization in Conditional Diffusion Models with Random Features
arXiv 2610.11288 · 2026-10-09REMORY：学习用于上下文压缩的残差记忆
REMORY: Learning Residual Memory for Context Compaction
arXiv 2610.11287 · 2026-10-09当场景文本劫持场景：揭示、利用与缓解图像生成模型中的渲染文本语义泄漏
When Scene Text Hijacks the Scene: Uncovering, Exploiting, and Mitigating Rendered-Text Semantic Leakage in Image Generation Models
arXiv 2610.11286 · 2026-10-09面向策略感知的私有数据合成的软投票方法
Soft Voting for Policy-Aware Private Data Synthesis
arXiv 2610.11285 · 2026-10-09DynaTE：通过动态令牌执行加速扩散大语言模型
DynaTE: Accelerating Diffusion LLMs via Dynamic Token Execution
arXiv 2610.11284 · 2026-10-09Being-M0.7：一种用于人形机器人的潜在世界-动作模型
Being-M0.7: A Latent World-Action Model for Humanoid Robots
arXiv 2610.11283 · 2026-10-09环半群的仿射Schur--Weyl理论与多项式表示
Affine Schur--Weyl theory for loop Semigroups and polynomial representations
arXiv 2610.11282 · 2026-10-09如何在替代模型上进行后训练：包络采样缓解奖励黑客行为
How to post-train on a surrogate: Envelope sampling mitigates reward hacking
arXiv 2610.11281 · 2026-10-09一种零体积模量的铰接蜂窝结构，保留了其组成材料超过五分之四的剪切模量
A hinged honeycomb with zero bulk modulus retaining more than four-fifths of its constituent's shear modulus
arXiv 2610.11280 · 2026-10-09锚定多重检验：透明使用e-闭包改进FDR程序
Anchored multiple testing: a transparent use of e-closure to improve FDR procedures
arXiv 2610.11279 · 2026-10-09面向准确的说话人清单的在线说话人 diarization（说话人 diarization 指说话人分割与识别）
Toward accurate speaker inventories for online speaker diarization
arXiv 2610.11278 · 2026-10-09H2CE：用于异构两阶段交叉编码器的POI重排序的地理语义交互建模
H2CE: Modeling Geo-Semantic Interactions for POI Reranking with Heterogeneous Two-Stage Cross-Encoders
arXiv 2610.11277 · 2026-10-09投射与内射直和项的互反性
Reciprocity for projective and injective direct summands
arXiv 2610.11276 · 2026-10-09多语言语音模型中的音系干扰
Phonological Interference in Multilingual Speech Models
arXiv 2610.11275 · 2026-10-09全矢量衍射深度神经网络
Full-Vector Diffractive Deep Neural Networks
arXiv 2610.11274 · 2026-10-09具非齐次Dirichlet边界条件的随机多孔介质方程
Stochastic Porous Media Equations With Nonhomogeneous Dirichlet Boundary Conditions
arXiv 2610.11273 · 2026-10-09利用注册声纹库作为证据：面向目标说话人标注的仿射不变分数校准
Using the enrollment gallery as evidence: affine-invariant score calibration for target speaker tagging
arXiv 2610.11272 · 2026-10-09避免长度为4的模式的不同平划分
Distinct flattened partitions avoiding a pattern of length four
arXiv 2610.11271 · 2026-10-09
