2026-09论文中文摘要
神经元引导的微调：解锁大型语言模型的高效对齐机制
Neuron-Guided Fine-Tuning: Unlocking Efficient Alignment Mechanisms for Large Language Models
arXiv 2609.05913 · 2026-09-09结构接近，时间遥远：衡量长时程LLM智能体的安全暴露
Structurally Close, Temporally Distant: Measuring Security Exposure in Long-Horizon LLM Agents
arXiv 2609.05911 · 2026-09-09UniRRM：跨语言与评估范式的统一推理奖励模型
UniRRM: Unified Reasoning Reward Models Across Languages and Evaluation Paradigms
arXiv 2609.05910 · 2026-09-09所有节点都能从知识图谱中同等受益吗？面向推荐的自适应节点感知知识图谱融合
Do All Nodes Benefit Equally from Knowledge Graphs? Adaptive Node-Aware KG Fusion for Recommendation
arXiv 2609.05909 · 2026-09-09具有片上真时延元件的光电储备池计算
Optoelectronic Reservoir Computing with an On-Chip True-Time-Delay Element
arXiv 2609.05907 · 2026-09-09浸没在外部磁场中的量子 Oppenheimer--Snyder 黑洞的薄吸积盘与引力俘获截面
Thin accretion disk and gravitational capture cross sections of a quantum Oppenheimer--Snyder black hole immersed in an external magnetic field
arXiv 2609.05906 · 2026-09-09从叙述到可审计预测：面向智能体预测的结构化脚手架
From Narrative to Auditable Forecasts: A Structured Scaffold for Agentic Forecasting
arXiv 2609.05905 · 2026-09-09边界抵消的分数-哈密顿量采样
Boundary-Cancelled Score-Hamiltonian Sampling
arXiv 2609.05904 · 2026-09-09EvoSafeHarness：为智能体安全演化模型特定与领域特定的防护机制
EvoSafeHarness: Evolving Model- and Domain-Specific Harnesses for Securing Agents
arXiv 2609.05903 · 2026-09-09GenPuzzle：图像生成模型中的视觉推理基准测试
GenPuzzle: Benchmarking Visual Reasoning in Image Generation Models
arXiv 2609.05902 · 2026-09-09从审查到授权：面向LLM智能体的密钥隔离阈值签名
From Review to Authorization: Key-Isolated Threshold Signing for LLM Agents
arXiv 2609.05901 · 2026-09-09AlignDiff：利用模型内在信息进行更好的偏好数据选择
AlignDiff: Exploiting Model-Intrinsic Information for Better Preference Data Selection
arXiv 2609.05899 · 2026-09-09化学衍生多层石墨烯在毫特斯拉磁场下的大能斯特效应
Large Nernst effect in chemically derived multilayer graphene at millitesla magnetic fields
arXiv 2609.05898 · 2026-09-09图的顶点扩张超图的谱矩与特征多项式
Spectral moments and characteristic polynomials of vertex expansion hypergraphs of graphs
arXiv 2609.05897 · 2026-09-09极性不对称结构校准用于链接符号预测
Polarity-Asymmetric Structural Calibration for Link Sign Prediction
arXiv 2609.05896 · 2026-09-09一种具有常数遗憾的在线资源分配一阶学习算法
A First-Order Learning Algorithm for Online Resource Allocation with Constant Regret
arXiv 2609.05895 · 2026-09-09AI指数增长的终结：AI泡沫内外的波动
The End of AI Exponentiation: Fluttering Inside and Outside AI Bubble
arXiv 2609.05894 · 2026-09-09平面中第一Dirichlet阈值处的Neumann--Dirichlet特征值比较
Neumann--Dirichlet eigenvalue comparison at the first Dirichlet threshold in the plane
arXiv 2609.05893 · 2026-09-09A4A：从人类示范中跨实体迁移面向动作的4D可操作性
A4A: Cross-Embodiment Transfer of Action-Oriented 4D Affordances from Human Demonstrations
arXiv 2609.05892 · 2026-09-09弱多项式完备性不蕴含几乎多项式完备性
Weak Polynomial Completeness Does Not Imply Almost Polynomial Completeness
arXiv 2609.05891 · 2026-09-09天基引力波探测器数据中暂现信号的自监督重建
Self-supervised reconstruction of transients in data from space-borne gravitational-wave detectors
arXiv 2609.05890 · 2026-09-09多模态资源耗尽攻击：基于联合像素-提示优化的视觉语言模型
Multimodal Resource-Exhaustion Attacks on Vision-Language Models via Joint Pixel-Prompt Optimization
arXiv 2609.05889 · 2026-09-09UniFusion：通过统一时空深度对齐的稀疏视角4D重建
UniFusion: Sparse-View 4D Reconstruction via Unified Spatio-temporal Depth Alignment
arXiv 2609.05888 · 2026-09-09利用复核改进洛伦兹积分变换方法的精度
Improving the accuracy of the Lorentz Integral Transform method using complex kernels
arXiv 2609.05887 · 2026-09-09基于数字锁相放大的快速高灵敏NV微波场成像用于片上微带线诊断
Rapid and high-sensitive NV-based microwave field imaging via digital lock-in amplification for on-chip microstrip diagnostics
arXiv 2609.05886 · 2026-09-09一个学习率不够：用于LoRA微调的自适应各向异性学习率
One Rate Is Not Enough: Adaptive Anisotropic Learning Rates for LoRA Fine-Tuning
arXiv 2609.05885 · 2026-09-09CALM：面向去中心化联邦学习的类别一致性与标签门控分歧调制
CALM: Class-wise Agreement and Label-gated Disagreement Modulation for Decentralized Federated Learning
arXiv 2609.05884 · 2026-09-09分位数回归的固定平滑均匀推断
Fixed-smoothing Uniform Inference for Quantile Regression
arXiv 2609.05883 · 2026-09-09如果大语言模型自食其言：多轮交互中的因果历史效应
What if LLMs Ate Their Words: Causal History Effects in Multi-Turn Interaction
arXiv 2609.05882 · 2026-09-09到达即损坏：公共模型注册表中静默缺陷的LLM工件及其捕获方法
Broken on Arrival: Silently Defective LLM Artifacts in Public Model Registries and How to Catch Them
arXiv 2609.05881 · 2026-09-09基于证据图约束的管道仪表图可靠推理：利用恢复的证据图约束视觉语言模型
Grounded and Faithful P&ID Reasoning: Constraining Vision-Language Models with Recovered Evidence Graphs
arXiv 2609.05880 · 2026-09-09正确的测试并不足够：规范驱动的测试生成中预言机转换的度量与训练
Correct Tests Are Not Enough: Measuring and Training Oracle Conversion in Specification-Based Test Generation
arXiv 2609.05879 · 2026-09-09预算依赖的基于覆盖与基于响应的训练集选择在机器学习原子间势中的交叉
A budget-dependent crossover between coverage- and response-based training-set selection for machine-learned interatomic potentials
arXiv 2609.05877 · 2026-09-09FACT：一种具有编译工具使用轨迹的取证代理，用于AI生成图像检测
FACT: A Forensic Agent with Compiled Tool-Use Trajectories for AI-Generated Image Detection
arXiv 2609.05876 · 2026-09-09纯向列量子临界超导体中的扩展奇异金属区
Extended strange metal regime in a pure nematic quantum critical superconductor
arXiv 2609.05875 · 2026-09-09Ulam 排名聚合在四个排名情况下难以近似
Ulam Rank Aggregation Is Hard to Approximate for Four Rankings
arXiv 2609.05874 · 2026-09-09非齐次扩散与自然边界条件下非线性Fokker-Planck方程的长时渐近行为
Long-time asymptotics of nonlinear Fokker-Planck equations with inhomogeneous diffusion and natural boundary conditions
arXiv 2609.05873 · 2026-09-09规范固定无关性的代数证明
An Algebraic Proof of Gauge-Fixing Independence
arXiv 2609.05872 · 2026-09-09声音去向何方？追踪音频条件大语言模型中的声学信息损失
Where Does the Sound Go? Tracing Acoustic Information Loss in Audio-Conditioned LLMs
arXiv 2609.05871 · 2026-09-09Reed-Solomon 码在容量下的算法列表译码与最优接近间隙
Algorithmic List Decoding at Capacity and Optimal Proximity Gaps for Reed-Solomon Codes
arXiv 2609.05870 · 2026-09-09精确有限积分几何：方向核与球面设计
Exact Finite Integral Geometry: Directional Kernels and Spherical Designs
arXiv 2609.05869 · 2026-09-09吸积白矮星上的磁约束山脉
Magnetically Confined Mountains on Accreting White Dwarfs
arXiv 2609.05868 · 2026-09-09触觉搜索：增强三维空间中的目标定位
Tactile Search: Enhancing Targeting in 3D Space
arXiv 2609.05867 · 2026-09-09实旋转下的退禁闭相变：矩阵模型研究
Deconfining Phase Transition under Real Rotation: A Matrix Model Study
arXiv 2609.05865 · 2026-09-09用于域泛化分割的层级提示注入器
Hierarchical Prompt Injector for Domain Generalization Segmentation
arXiv 2609.05864 · 2026-09-09最优镇定前因子的端点渐近性
Endpoint Asymptotics of Optimal Stabilization Prefactors
arXiv 2609.05863 · 2026-09-09预算约束下任务感知的动态网络获取
Budgeted Task-Aware Acquisition of Dynamic Networks
arXiv 2609.05862 · 2026-09-09SLA安全的AI原生NG-RAN能量控制：基于稳定性感知的约束PPO
SLA-Safe Energy Control for AI-Native NG-RAN Using Stability-Aware Constrained PPO
arXiv 2609.05861 · 2026-09-09选择性后验边际正则化用于前向校正分类
Selective Posterior Margin Regularization for Forward-Corrected Classification
arXiv 2609.05859 · 2026-09-09自适应多阶段采样变换用于求解模型未知的有限时域马尔可夫决策过程
Transformation of Adaptive Multistage Sampling for Solving Finite-Horizon Markov Decision Processes with Unknown Model
arXiv 2609.05856 · 2026-09-09
