2026-10论文中文摘要
模型植入触发器：多轮大语言模型中的答案侧后门攻击
The Model Plants the Trigger: Answer-Side Backdoor Attacks in Multi-Turn Large Language Models
arXiv 2610.07723 · 2026-10-07干预会破坏你的模型吗？大语言模型干预方法的多维评估套件
Does Steering Break Your Model? A Multi-Dimensional Evaluation Suite for LLM Steering Methods
arXiv 2610.07722 · 2026-10-07体积采样岭回归的精确校准与尖锐风险几何
Exact Calibration and Sharp Risk Geometry for Volume-Sampled Ridge Regression
arXiv 2610.07721 · 2026-10-07RefRoute：通过紧凑残差条件化与空间路由解耦条件化成本与参考图像
RefRoute: Decoupling Conditioning Cost from References via Compact Residual Conditioning and Spatial Routing
arXiv 2610.07720 · 2026-10-07双电荷子霍尔效应
Bichargon Hall Effect
arXiv 2610.07719 · 2026-10-07边删除对图上非交换距离的影响
The effect of edge deletion on noncommutative distances on graphs
arXiv 2610.07718 · 2026-10-07亚高斯数据上测度到测度变换器的稳定性
Stability of Measure-to-Measure Transformers on Sub-Gaussian Data
arXiv 2610.07717 · 2026-10-07仅预填充决策模型中的读出稳定性：零标签预测与推理时计算分配
Readout Stability in Prefill-Only Decision Models:Zero-Label Prediction and Inference-Time Compute Allocation
arXiv 2610.07716 · 2026-10-07当旧事实重现：重读、回退与时间记忆的局限
When Old Facts Return: Re-Reads, Reverts, and the Limits of Temporal Memory
arXiv 2610.07715 · 2026-10-07非对称界面密度分层下多层瑞利-泰勒不稳定性的直接数值模拟
Direct numerical simulations of multi-layer Rayleigh-Taylor instability under asymmetric interfacial density stratification
arXiv 2610.07714 · 2026-10-07数学不变量启用的拓扑神经网络用于分子和材料性质预测
Mathematical Invariant-Enabled Topological Neural Networks for Molecular and Materials Property Prediction
arXiv 2610.07712 · 2026-10-07肾脏病理学中基础细胞核分割模型的综合评估与微调
Comprehensive Evaluation and Fine-Tuning of Foundational Cell Nuclei Segmentation Models in Renal Pathology
arXiv 2610.07711 · 2026-10-07宇宙网在0.4<z<4星系演化中的作用
The Role of the Cosmic Web in Galaxy Evolution Across 0.4<z<4
arXiv 2610.07710 · 2026-10-07采样前的证据：可解释的隐式负候选发现用于推荐
Evidence Before Sampling: Interpretable Implicit Negative Candidate Discovery for Recommendation
arXiv 2610.07708 · 2026-10-07AgentMemGate：解决对话助手记忆中的推测污染问题
AgentMemGate: Addressing Speculation Contamination in Conversational Assistant Memory
arXiv 2610.07707 · 2026-10-07WASD：基于Wasserstein距离的大语言模型知识蒸馏
WASD: Wasserstein-based Knowledge Distillation for Large Language Models
arXiv 2610.07706 · 2026-10-07帧级标签在热成像视频中小型无人机点检测中的能与不能
What Frame-Level Labels Can and Cannot Do for Small-UAV Point Detection in Thermal Video
arXiv 2610.07705 · 2026-10-07基于反事实语义-社会世界模型的独立多智能体强化学习
Independent Multi-Agent Reinforcement Learning with Counterfactual Semantic-Social World Models
arXiv 2610.07704 · 2026-10-07空中液珠：声悬浮中液珠的蒸发动力学
Airborne liquid marble: Evaporation dynamics of liquid marble in acoustic levitation
arXiv 2610.07703 · 2026-10-07基于本地硬件的片段分子轨道量子化学并行多量子处理单元执行演示
Demonstration of Parallel Multi-QPU Execution for Fragment-Based Quantum Chemistry Using On-Premises Hardware
arXiv 2610.07702 · 2026-10-07关于准入阈值的边界：定量策略研究中注入真值验证的证伪优先选择
On the Boundary of Admission Gates: An Injected-Truth Study of Falsification-First Selection in Quantitative Strategy Research
arXiv 2610.07701 · 2026-10-07检测行为操纵下基于击键的越南语LLM辅助写作
Detecting LLM-Assisted Vietnamese Writing via Keystrokes under Behavioral Manipulation
arXiv 2610.07700 · 2026-10-07通过对抗强化学习改进论辩挖掘的合成数据生成
Improving Synthetic Data Generation for Argument Mining via Adversarial Reinforcement Learning
arXiv 2610.07699 · 2026-10-07RBMatch：面向半监督建筑足迹提取的双层级类别再平衡
RBMatch: Dual-Level Class Rebalancing for Semi-Supervised Building Footprint Extraction
arXiv 2610.07698 · 2026-10-07当表达能力不足时：变分量子电路中的离散路由几何
When Expressivity Is Not Enough: Discrete Routing Geometry in Variational Quantum Circuits
arXiv 2610.07697 · 2026-10-07ESP：一步多模态动作生成的能量分数策略
ESP: Energy-Score Policy for One-Step Multimodal Action Generation
arXiv 2610.07696 · 2026-10-07从零维到连续对偶：一个双范畴解释
From Zero-Dimensional to Continuous Dualities: A Double-Categorical Account
arXiv 2610.07695 · 2026-10-07锚定驱动的多模态多尺度专家选择用于生存预测
Anchor-driven Multi-modal Multi-scale Expert Selection for Survival Prediction
arXiv 2610.07694 · 2026-10-07一维局域模型中任意温度下的自发对称性破缺
Spontaneous symmetry breaking at any temperature in a local one-dimensional model
arXiv 2610.07693 · 2026-10-07ExoBridge：通过人体肢体耦合学习裸手到穿戴式外骨骼的映射
ExoBridge: Learning a Bare Hand to Hand-Worn Exoskeleton Mapping through Human Limb Coupling
arXiv 2610.07692 · 2026-10-07PerSpectron：利用感知器检测微架构攻击的不变足迹
PerSpectron: Detecting Invariant Footprints of Microarchitectural Attacks with Perceptron
arXiv 2610.07691 · 2026-10-07如何修复一个损坏的度量：线性核、紧界与树度量
How to Fix a Broken Metric: A Linear Kernel, Tight Bounds, and Tree Metrics
arXiv 2610.07690 · 2026-10-07解锁CLIP中的细粒度感知：基于结构感知的潜在掩码建模
Unlocking Fine-Grained Perception in CLIP via Structurally-Aware Latent Masked Modeling
arXiv 2610.07689 · 2026-10-07FailBench：评估分布式训练架构中的容错能力
FailBench: Evaluating Fault Tolerance Across Distributed Training Architectures
arXiv 2610.07688 · 2026-10-07第二类曲率算子刚性条件的新结果
New Curvature Operator of the Second Kind Conditions for Rigidity
arXiv 2610.07687 · 2026-10-07BluffJAX：JAX中的对抗性不完美信息博弈
BluffJAX: Adversarial Imperfect Information Games in JAX
arXiv 2610.07686 · 2026-10-07CHiRP：控制流历史重用预测
CHiRP: Control-Flow History Reuse Prediction
arXiv 2610.07685 · 2026-10-07解耦频域感知扩散模型中的双图像参考以实现个性化生成
Disentangling Dual Image References in Frequency Aware Diffusion Models for Personalized Generation
arXiv 2610.07684 · 2026-10-076$\ imes$6量子点阵列中的稀疏量子比特操作
Sparse qubit operation in a 6$\times$6 quantum dot array
arXiv 2610.07683 · 2026-10-07语义分割推理用于远程调制识别
Semantic Split Inference for Remote Modulation Recognition
arXiv 2610.07682 · 2026-10-07EigenDEXplore：基于人类先验的灵巧操作结构化探索
EigenDEXplore: Structured Exploration for Dexterous Manipulation with Human Priors
arXiv 2610.07681 · 2026-10-07利用偶极-声子量子逻辑进行光谱测量
Spectroscopy using dipole-phonon quantum logic
arXiv 2610.07680 · 2026-10-07双曲交叉三角多项式一致范数的采样离散化
Sampling discretization of the uniform norm for hyperbolic-cross trigonometric polynomials
arXiv 2610.07679 · 2026-10-07使用互补经典预测器与图预测器的简单融合实现精确分子性质回归
Accurate Molecular Property Regression Using Simple Fusion of Complementary Classical and Graph Predictors
arXiv 2610.07678 · 2026-10-07自适应模型反演攻击泛化隐私-鲁棒性权衡
Adaptive Model Inversion Attacks Generalize a Privacy-Robustness Tradeoff
arXiv 2610.07677 · 2026-10-07精确解体积与Transformer中的长度泛化
Exact-Solution Volume and Length Generalization in Transformers
arXiv 2610.07676 · 2026-10-07EIO-Agents：AI智能体评估缺失的语义层
EIO-Agents: The Missing Semantic Layer for AI Agent Evaluation
arXiv 2610.07675 · 2026-10-07信念信息混合控制与几乎必然目标集收敛
Belief-Informed Hybrid Control with Almost-Sure Target-Set Convergence
arXiv 2610.07674 · 2026-10-07量子游走与图运算
Quantum walks and graph operations
arXiv 2610.07673 · 2026-10-07从算法边缘化到AI中介的再中心化：文化根基AI能否将客家语言与文化带回主流社会？
From Algorithmic Marginalization to AI-Mediated Re-Centering: Can Culturally Grounded AI Bring Hakka Language and Culture Back into Mainstream Society?
arXiv 2610.07672 · 2026-10-07
