机器学习学术速递[9.9]

点击阅读原文访问arxivdaily.com，涵盖CS|物理|数学|经济|统计|金融|生物|电气领域，更有搜索、收藏等功能！

cs.LG 方向，今日共计223篇

大模型相关(27篇)

【1】Interleaving Reasoning for Better Text-to-Image Generation
标题：交错推理用于更好的文本到图像生成
链接：https://arxiv.org/abs/2509.06945

【2】Outcome-based Exploration for LLM Reasoning
标题：基于结果的LLM推理探索
链接：https://arxiv.org/abs/2509.06941

【3】Proof-Carrying Numbers (PCN): A Protocol for Trustworthy Numeric Answers from LLMs via Claim Verification
标题：携带证明号（PCE）：通过索赔验证从LLM获得值得信赖的数字答案的协议
链接：https://arxiv.org/abs/2509.06902

【4】Aligning Large Vision-Language Models by Deep Reinforcement Learning and Direct Preference Optimization
标题：通过深度强化学习和直接偏好优化调整大型视觉语言模型
链接：https://arxiv.org/abs/2509.06759

【5】Ban&Pick: Achieving Free Performance Gains and Inference Speedup via Smarter Routing in MoE-LLMs
标题：禁止与选择：通过MoE-LLM中的更智能路由实现免费性能收益和推理加速
链接：https://arxiv.org/abs/2509.06346

【6】A Fragile Number Sense: Probing the Elemental Limits of Numerical Reasoning in LLMs
标题：脆弱的数字感：探索LLM中数字推理的元素限制
链接：https://arxiv.org/abs/2509.06332

【7】Text-Trained LLMs Can Zero-Shot Extrapolate PDE Dynamics
标题：经过文本训练的LLM可以Zero-Shot外推PED动态
链接：https://arxiv.org/abs/2509.06322

【8】RecMind: LLM-Enhanced Graph Neural Networks for Personalized Consumer Recommendations
标题：RecMind：用于个性化消费者推荐的LLM增强图形神经网络
链接：https://arxiv.org/abs/2509.06286

【9】From Implicit Exploration to Structured Reasoning: Leveraging Guideline and Refinement for LLMs
标题：从隐式探索到结构化推理：LLM的利用指南和细化
链接：https://arxiv.org/abs/2509.06284

【10】FineServe: Precision-Aware KV Slab and Two-Level Scheduling for Heterogeneous Precision LLM Serving
标题：FineServe：用于异类精确LLM服务的精确感知GV平板和两级调度
链接：https://arxiv.org/abs/2509.06261

【11】Reasoning Language Model for Personalized Lung Cancer Screening
标题：肺癌个体化筛查的推理语言模型
链接：https://arxiv.org/abs/2509.06169

【12】Benchmarking Gender and Political Bias in Large Language Models
标题：大型语言模型中的性别和政治偏见基准
链接：https://arxiv.org/abs/2509.06164

【13】PolicyEvolve: Evolving Programmatic Policies by LLMs for multi-player games via Population-Based Training
标题：Policy Evolve：LLM通过基于人群的训练为多人游戏制定程序化政策
链接：https://arxiv.org/abs/2509.06053

【14】ALPHA: LLM-Enabled Active Learning for Human-Free Network Anomaly Detection
标题：ALA：LLM支持的主动学习，用于无人网络异常检测
链接：https://arxiv.org/abs/2509.05936

【15】X-SQL: Expert Schema Linking and Understanding of Text-to-SQL with Multi-LLMs
标题：X-SQL：使用多LLM对文本到SQL的专家模式链接和理解
链接：https://arxiv.org/abs/2509.05899

【16】Let's Roleplay: Examining LLM Alignment in Collaborative Dialogues
标题：让我们角色扮演：审视协作对话中的LLM一致性
链接：https://arxiv.org/abs/2509.05882

【17】Finetuning LLMs for Human Behavior Prediction in Social Science Experiments
标题：社会科学实验中人类行为预测的LLM微调
链接：https://arxiv.org/abs/2509.05830

【18】Automating API Documentation with LLMs: A BERTopic Approach
标题：使用LLM自动化API文档：BERTopic方法
链接：https://arxiv.org/abs/2509.05749

【19】Cross-Service Threat Intelligence in LLM Services using Privacy-Preserving Fingerprints
标题：使用隐私保护指纹的LLM服务中的跨服务威胁情报
链接：https://arxiv.org/abs/2509.05608

【20】Neural Breadcrumbs: Membership Inference Attacks on LLMs Through Hidden State and Attention Pattern Analysis
标题：神经面包屑：通过隐藏状态和注意力模式分析对LLM的成员推理攻击
链接：https://arxiv.org/abs/2509.05449

【21】AI-in-the-Loop: Privacy Preserving Real-Time Scam Detection and Conversational Scambaiting by Leveraging LLMs and Federated Learning
标题：AI在环：通过利用LLM和联邦学习保护隐私的实时欺诈检测和会话欺诈
链接：https://arxiv.org/abs/2509.05362

【22】Beyond ROUGE: N-Gram Subspace Features for LLM Hallucination Detection
标题：Beyond ROUGE：用于LLM幻觉检测的N-Gram子空间功能
链接：https://arxiv.org/abs/2509.05360

【23】Privacy-Preserving Offloading for Large Language Models in 6G Vehicular Networks
标题：6G车载网络中大型语言模型的隐私保护卸载
链接：https://arxiv.org/abs/2509.05320

【24】Standard vs. Modular Sampling: Best Practices for Reliable LLM Unlearning
标题：标准抽样与模块抽样：可靠LLM忘记学习的最佳实践
链接：https://arxiv.org/abs/2509.05316

【25】Large Language Model Integration with Reinforcement Learning to Augment Decision-Making in Autonomous Cyber Operations
标题：大型语言模型与强化学习集成以增强自主网络运营中的决策
链接：https://arxiv.org/abs/2509.05311

【26】Are LLM Agents Behaviorally Coherent? Latent Profiles for Social Simulation
标题：LLM代理人行为一致吗？社交模拟的潜在配置文件
链接：https://arxiv.org/abs/2509.03736

【27】Fisher Random Walk: Automatic Debiasing Contextual Preference Inference for Large Language Model Evaluation
标题：Fisher随机漫步：大型语言模型评估的自动去偏上下文偏好推断
链接：https://arxiv.org/abs/2509.05852

Graph相关(图学习|图神经网络|图优化等)(17篇)

【1】Asynchronous Message Passing for Addressing Oversquashing in Graph Neural Networks
标题：图神经网络中用于解决过度挤压的非同步消息传递
链接：https://arxiv.org/abs/2509.06777

【2】Long-Range Graph Wavelet Networks
标题：长期图子波网络
链接：https://arxiv.org/abs/2509.06743

【3】TrajAware: Graph Cross-Attention and Trajectory-Aware for Generalisable VANETs under Partial Observations
标题：TrajAware：在部分观察下绘制可概括的VANSYS的交叉注意和轨迹感知
链接：https://arxiv.org/abs/2509.06665

【4】A Survey of Generalization of Graph Anomaly Detection: From Transfer Learning to Foundation Models
标题：图异常检测的推广综述：从迁移学习到基础模型
链接：https://arxiv.org/abs/2509.06609

【5】PAC-Bayesian Generalization Bounds for Graph Convolutional Networks on Inductive Node Classification
标题：图卷积网络归纳节点分类的PAC-Bayesian推广界
链接：https://arxiv.org/abs/2509.06600

【6】DyC-STG: Dynamic Causal Spatio-Temporal Graph Network for Real-time Data Credibility Analysis in IoT
标题：DyC-STG：用于物联网实时数据可信性分析的动态因果时空图网络
链接：https://arxiv.org/abs/2509.06483

【7】Graph Neural Networks for Resource Allocation in Interference-limited Multi-Channel Wireless Networks with QoS Constraints
标题：图神经网络在具有服务质量约束的干扰有限多通道无线网络中进行资源分配
链接：https://arxiv.org/abs/2509.06395

【8】A Spatio-Temporal Graph Neural Networks Approach for Predicting Silent Data Corruption inducing Circuit-Level Faults
标题：预测导致电路级故障的无声数据损坏的时空图神经网络方法
链接：https://arxiv.org/abs/2509.06289

【9】MCIGLE: Multimodal Exemplar-Free Class-Incremental Graph Learning
标题：MCIGLE：多模式无示例类增量图学习
链接：https://arxiv.org/abs/2509.06219

【10】Metric Embedding Initialization-Based Differentially Private and Explainable Graph Clustering
标题：基于度量嵌入初始化的差异私有和可解释图集群
链接：https://arxiv.org/abs/2509.06214

【11】Data-Efficient Time-Dependent PDE Surrogates: Graph Neural Simulators vs Neural Operators
标题：数据高效的时间相关的PED替代品：图形神经模拟器与神经操作员
链接：https://arxiv.org/abs/2509.06154

【12】GraMFedDHAR: Graph Based Multimodal Differentially Private Federated HAR
标题：GraMFedDHAR：基于图的多模式差异私有联邦HAR
链接：https://arxiv.org/abs/2509.05671

【13】Distributed Link Sparsification for Scalable Scheduling Using Graph Neural Networks (Journal Version)
标题：使用图神经网络进行可扩展调度的分布式链路稀疏化（期刊版本）
链接：https://arxiv.org/abs/2509.05447

【14】Safeguarding Graph Neural Networks against Topology Inference Attacks
标题：保护图神经网络免受布局推理攻击
链接：https://arxiv.org/abs/2509.05429

【15】RoboBallet: Planning for Multi-Robot Reaching with Graph Neural Networks and Reinforcement Learning
标题：RoboBallet：利用图神经网络和强化学习规划多机器人到达
链接：https://arxiv.org/abs/2509.05397

【16】Learning from one graph: transductive learning guarantees via the geometry of small random worlds
标题：从一张图中学习：通过小型随机世界的几何学保证转换学习
链接：https://arxiv.org/abs/2509.06894

【17】Automated Hierarchical Graph Construction for Multi-source Electronic Health Records
标题：多源电子健康记录自动分层图构建
链接：https://arxiv.org/abs/2509.06576

Transformer(4篇)

【1】H$_{2}$OT: Hierarchical Hourglass Tokenizer for Efficient Video Pose Transformers
标题：H$_{2}$OT：用于高效视频姿势Transformer的分层沙漏代币器
链接：https://arxiv.org/abs/2509.06956

【2】From Noise to Narrative: Tracing the Origins of Hallucinations in Transformers
标题：从噪音到叙事：追踪《Transformer》中幻觉的起源
链接：https://arxiv.org/abs/2509.06938

【3】Lane Change Intention Prediction of two distinct Populations using a Transformer
标题：基于Transformer的两种群换道意图预测
链接：https://arxiv.org/abs/2509.06529

【4】Learning spatially structured open quantum dynamics with regional-attention transformers
标题：使用区域注意力转换器学习空间结构开放量子动力学
链接：https://arxiv.org/abs/2509.06871

GAN|对抗|攻击|生成相关(9篇)

【1】Imitative Membership Inference Attack
标题：模仿会员推断攻击
链接：https://arxiv.org/abs/2509.06796

【2】Group Effect Enhanced Generative Adversarial Imitation Learning for Individual Travel Behavior Modeling under Incentives
标题：激励下个人旅行行为建模的群体效应增强生成对抗模仿学习
链接：https://arxiv.org/abs/2509.06656

【3】On optimal solutions of classical and sliced Wasserstein GANs with non-Gaussian data
标题：基于非高斯数据的经典和切片Wasserstein GAN的最优解
链接：https://arxiv.org/abs/2509.06505

【4】IGAff: Benchmarking Adversarial Iterative and Genetic Affine Algorithms on Deep Neural Networks
标题：IGAff：深度神经网络上的对抗迭代和遗传仿射算法基准
链接：https://arxiv.org/abs/2509.06459

【5】DCMI: A Differential Calibration Membership Inference Attack Against Retrieval-Augmented Generation
标题：DCMI：针对检索增强生成的差异校准成员资格推断攻击
链接：https://arxiv.org/abs/2509.06026

【6】Reasoning Introduces New Poisoning Attacks Yet Makes Them More Complicated
标题：推理引入了新的中毒攻击，但使它们变得更加复杂
链接：https://arxiv.org/abs/2509.05739

【7】Biomedical Literature Q&A System Using Retrieval-Augmented Generation (RAG)
标题：使用检索增强生成（RAG）的生物医学文献问答系统
链接：https://arxiv.org/abs/2509.05505

【8】Ensembling Membership Inference Attacks Against Tabular Generative Models
标题：针对表格生成模型的集合成员推断攻击
链接：https://arxiv.org/abs/2509.05350

【9】Imagining Alternatives: Towards High-Resolution 3D Counterfactual Medical Image Generation via Language Guidance
标题：想象替代方案：通过语言指导实现高分辨率3D反事实医学图像生成
链接：https://arxiv.org/abs/2509.05978

半/弱/无/有监督|不确定性|主动学习(7篇)

【1】Contrastive Self-Supervised Network Intrusion Detection using Augmented Negative Pairs
标题：使用增强负对的对比自监督网络入侵检测
链接：https://arxiv.org/abs/2509.06550

【2】Predicting Fetal Outcomes from Cardiotocography Signals Using a Supervised Variational Autoencoder
标题：使用受监督变分自动编码器从子宫内膜图信号预测胎儿结局
链接：https://arxiv.org/abs/2509.06540

【3】SPINN: An Optimal Self-Supervised Physics-Informed Neural Network Framework
标题：SPINN：一个最佳自我监督的物理知情神经网络框架
链接：https://arxiv.org/abs/2509.05886

【4】Performance of Conformal Prediction in Capturing Aleatoric Uncertainty
标题：保形预测在捕捉感性不确定性中的性能
链接：https://arxiv.org/abs/2509.05826

【5】DQS: A Low-Budget Query Strategy for Enhancing Unsupervised Data-driven Anomaly Detection Approaches
标题：DQS：一种用于增强无监督数据驱动异常检测方法的低预算查询策略
链接：https://arxiv.org/abs/2509.05663

【6】Self-supervised Learning for Hyperspectral Images of Trees
标题：树木高光谱图像的自我监督学习
链接：https://arxiv.org/abs/2509.05630

【7】Uncertainty Quantification in Probabilistic Machine Learning Models: Theory, Methods, and Insights
标题：概率机器学习模型中的不确定性量化：理论、方法和见解
链接：https://arxiv.org/abs/2509.05877

迁移|Zero/Few/One-Shot|自适应(10篇)

【1】Staying in the Sweet Spot: Responsive Reasoning Evolution via Capability-Adaptive Hint Scaffolding
标题：留在最佳点：通过能力自适应提示支架的响应式推理进化
链接：https://arxiv.org/abs/2509.06923

【2】CogGuide: Human-Like Guidance for Zero-Shot Omni-Modal Reasoning
标题：CogGuide：Zero-Shot全模式推理的类人指导
链接：https://arxiv.org/abs/2509.06641

【3】WindFM: An Open-Source Foundation Model for Zero-Shot Wind Power Forecasting
标题：WindFM：零冲击风力预测的开源基础模型
链接：https://arxiv.org/abs/2509.06311

【4】New Insights into Optimal Alignment of Acoustic and Linguistic Representations for Knowledge Transfer in ASR
标题：对ASB中知识转移的声学和语言表示最佳对齐的新见解
链接：https://arxiv.org/abs/2509.05609

【5】ProfilingAgent: Profiling-Guided Agentic Reasoning for Adaptive Model Optimization
标题：ProfilingAgent：用于自适应模型优化的分析引导式推理
链接：https://arxiv.org/abs/2509.05584

【6】Learning Tool-Aware Adaptive Compliant Control for Autonomous Regolith Excavation
标题：自主风化层挖掘的学习工具感知自适应合规控制
链接：https://arxiv.org/abs/2509.05475

【7】Robust and Adaptive Spectral Method for Representation Multi-Task Learning with Contamination
标题：具有污染的表示多任务学习的鲁棒自适应谱方法
链接：https://arxiv.org/abs/2509.06575

【8】Minimax optimal transfer learning for high-dimensional additive regression
标题：多维加性回归的极小最优迁移学习
链接：https://arxiv.org/abs/2509.06308

【9】MOSAIC: Minimax-Optimal Sparsity-Adaptive Inference for Change Points in Dynamic Networks
标题：MOSAIC：动态网络中变化点的极小最优稀疏自适应推理
链接：https://arxiv.org/abs/2509.06303

【10】Predicting Brain Morphogenesis via Physics-Transfer Learning
标题：通过物理迁移学习预测大脑形态发生
链接：https://arxiv.org/abs/2509.05305

强化学习(7篇)

【1】Reinforcement learning meets bioprocess control through behaviour cloning: Real-world deployment in an industrial photobioreactor
标题：强化学习通过行为克隆满足生物过程控制：在工业光生物反应器中的实际部署
链接：https://arxiv.org/abs/2509.06853

【2】Physics-informed Value Learner for Offline Goal-Conditioned Reinforcement Learning
标题：离线目标条件强化学习的了解身体的价值学习者
链接：https://arxiv.org/abs/2509.06782

【3】Toward a Metrology for Artificial Intelligence: Hidden-Rule Environments and Reinforcement Learning
标题：迈向人工智能的计量学：隐藏规则环境和强化学习
链接：https://arxiv.org/abs/2509.06213

【4】Using Reinforcement Learning to Optimize the Global and Local Crossing Number
标题：使用强化学习优化全球和本地交叉数
链接：https://arxiv.org/abs/2509.06108

【5】Learning to Construct Knowledge through Sparse Reference Selection with Reinforcement Learning
标题：通过强化学习通过稀疏引用选择来学习构建知识
链接：https://arxiv.org/abs/2509.05874

【6】Reinforcement Learning with Anticipation: A Hierarchical Approach for Long-Horizon Tasks
标题：带预期的强化学习：长期任务的分层方法
链接：https://arxiv.org/abs/2509.05545

【7】Reward function compression facilitates goal-dependent reinforcement learning
标题：奖励函数压缩促进目标相关强化学习
链接：https://arxiv.org/abs/2509.06810

医学相关(7篇)

【1】Demo: Healthcare Agent Orchestrator (HAO) for Patient Summarization in Molecular Tumor Boards
标题：演示：用于分子肿瘤委员会中患者汇总的医疗保健代理商演示器（HAO）
链接：https://arxiv.org/abs/2509.06602

【2】Causal Debiasing Medical Multimodal Representation Learning with Missing Modalities
标题：缺失模式的医学多模式表示学习的因果去偏
链接：https://arxiv.org/abs/2509.05615

【3】Advanced Brain Tumor Segmentation Using EMCAD: Efficient Multi-scale Convolutional Attention Decoding
标题：使用EMCAD的高级脑肿瘤分割：高效的多尺度卷积注意力解码
链接：https://arxiv.org/abs/2509.05431

【4】Unmasking COVID-19 Vulnerability in Nigeria: Mapping Risks Beyond Urban Hotspots
标题：揭露尼日利亚的COVID-19脆弱性：绘制城市热点以外的风险
链接：https://arxiv.org/abs/2509.05398

【5】Handling imbalance and few-sample size in ML based Onion disease classification
标题：基于ML的洋葱病分类中处理不平衡和少样本量
链接：https://arxiv.org/abs/2509.05341

【6】Application of discrete Ricci curvature in pruning randomly wired neural networks: A case study with chest x-ray classification of COVID-19
标题：离散Ricci弯曲在修剪随机连线神经网络中的应用：COVID-19胸部X光分类的案例研究
链接：https://arxiv.org/abs/2509.05322

【7】Impact of Labeling Inaccuracy and Image Noise on Tooth Segmentation in Panoramic Radiographs using Federated, Centralized and Local Learning
标题：标签不准确和图像噪音对使用联合、集中和本地学习的全景射线照片中牙齿分割的影响
链接：https://arxiv.org/abs/2509.06553

蒸馏|知识提取(1篇)

【1】QualityFM: a Multimodal Physiological Signal Foundation Model with Self-Distillation for Signal Quality Challenges in Critically Ill Patients
标题：Quality FM：一种具有自蒸馏功能的多模式生理信号基础模型，应对危重患者的信号质量挑战
链接：https://arxiv.org/abs/2509.06516

推荐(2篇)

【1】ARIES: Relation Assessment and Model Recommendation for Deep Time Series Forecasting
标题：ARIES：深度时间序列预测的关系评估和模型推荐
链接：https://arxiv.org/abs/2509.06060

【2】Calibrated Recommendations with Contextual Bandits
标题：针对背景盗贼的校准建议
链接：https://arxiv.org/abs/2509.05460

聚类(1篇)

【1】Causal Clustering for Conditional Average Treatment Effects Estimation and Subgroup Discovery
标题：条件平均治疗效果估计和亚组发现的原因聚集
链接：https://arxiv.org/abs/2509.05775

自动驾驶|车辆|车道检测等(1篇)

【1】Self-Driving Laboratory Optimizes the Lower Critical Solution Temperature of Thermoresponsive Polymers
标题：自动驾驶实验室优化温敏聚合物的下限临界溶液温度
链接：https://arxiv.org/abs/2509.05351

点云|SLAM|雷达|激光|深度RGBD相关(1篇)

【1】LiDAR-BIND-T: Improving SLAM with Temporally Consistent Cross-Modal LiDAR Reconstruction
标题：LiDAR-BIND-T：通过时间一致的跨模式LiDART重建改善SLAM
链接：https://arxiv.org/abs/2509.05728

推理|分析|理解|解释(9篇)

【1】Test-Time Scaling in Reasoning Models Is Not Effective for Knowledge-Intensive Tasks Yet
标题：推理模型中的测试时间缩放对于知识密集型任务还不有效
链接：https://arxiv.org/abs/2509.06861

【2】RT-HCP: Dealing with Inference Delays and Sample Efficiency to Learn Directly on Robotic Platforms
标题：RT-HCP：处理推理延迟和样本效率以直接在机器人平台上学习
链接：https://arxiv.org/abs/2509.06714

【3】Small Vectors, Big Effects: A Mechanistic Study of RL-Induced Reasoning via Steering Vectors
标题：小载体，大效应：通过转向载体的RL诱导推理的机制研究
链接：https://arxiv.org/abs/2509.06608

【4】An Explainable Framework for Particle Swarm Optimization using Landscape Analysis and Machine Learning
标题：使用景观分析和机器学习的粒子群优化的可解释框架
链接：https://arxiv.org/abs/2509.06272

【5】The Measure of Deception: An Analysis of Data Forging in Machine Unlearning
标题：欺骗的衡量：机器取消学习中数据伪造的分析
链接：https://arxiv.org/abs/2509.05865

【6】MambaLite-Micro: Memory-Optimized Mamba Inference on MCUs
标题：MambaLite-Micro：基于MCU的内存优化Mamba推理
链接：https://arxiv.org/abs/2509.05488

【7】Multi-EuP: The Multilingual European Parliament Dataset for Analysis of Bias in Information Retrieval
标题：Multi-EuP：用于信息检索偏差分析的多语言欧洲议会数据集
链接：https://arxiv.org/abs/2311.01870

【8】Robust Analysis for Resilient AI System
标题：弹性人工智能系统的稳健分析
链接：https://arxiv.org/abs/2509.06172

【9】Robust variational neural posterior estimation for simulation-based inference
标题：基于模拟推理的鲁棒变分神经后验估计
链接：https://arxiv.org/abs/2509.05724

检测相关(9篇)

【1】Tackling the Noisy Elephant in the Room: Label Noise-robust Out-of-Distribution Detection via Loss Correction and Low-rank Decomposition
标题：解决房间里的吵闹大象：通过丢失纠正和低等级分解进行标签噪音稳健的分布外检测
链接：https://arxiv.org/abs/2509.06918

【2】Hypergraph-Guided Regex Filter Synthesis for Event-Based Anomaly Detection
标题：用于基于事件的异常检测的超图引导Regex过滤器合成
链接：https://arxiv.org/abs/2509.06911

【3】Detection of trade in products derived from threatened species using machine learning and a smartphone
标题：使用机器学习和智能手机检测受威胁物种产品贸易
链接：https://arxiv.org/abs/2509.06585

【4】CAPMix: Robust Time Series Anomaly Detection Based on Abnormal Assumptions with Dual-Space Mixup
标题：CAPMix：基于双空间混合异常假设的鲁棒时间序列异常检测
链接：https://arxiv.org/abs/2509.06419

【5】Near Real-Time Dust Aerosol Detection with 3D Convolutional Neural Networks on MODIS Data
标题：利用3D卷积神经网络在TLR数据上近实时检测尘埃气溶胶
链接：https://arxiv.org/abs/2509.05887

【6】DCV-ROOD Evaluation Framework: Dual Cross-Validation for Robust Out-of-Distribution Detection
标题：DCV-ROOD评估框架：用于稳健的分布外检测的双重交叉验证
链接：https://arxiv.org/abs/2509.05778

【7】VILOD: A Visual Interactive Labeling Tool for Object Detection
标题：VILOD：用于对象检测的视觉交互标签工具
链接：https://arxiv.org/abs/2509.05317

【8】Robustness and accuracy of mean opinion scores with hard and soft outlier detection
标题：具有硬异常值和软异常值检测的平均意见分数的稳健性和准确性
链接：https://arxiv.org/abs/2509.06554

【9】On detection probabilities of link invariants
标题：关于链接不变量的检测概率
链接：https://arxiv.org/abs/2509.05574

分类|识别(10篇)

【1】AxelSMOTE: An Agent-Based Oversampling Algorithm for Imbalanced Classification
标题：AxelSMOTE：一种基于代理的不平衡分类过采样算法
链接：https://arxiv.org/abs/2509.06875

【2】Improved Classification of Nitrogen Stress Severity in Plants Under Combined Stress Conditions Using Spatio-Temporal Deep Learning Framework
标题：使用时空深度学习框架改进组合胁迫条件下植物氮胁迫严重程度的分类
链接：https://arxiv.org/abs/2509.06625

【3】Signal-Based Malware Classification Using 1D CNNs
标题：基于信号的1D CNN恶意软件分类
链接：https://arxiv.org/abs/2509.06548

【4】MRD-LiNet: A Novel Lightweight Hybrid CNN with Gradient-Guided Unlearning for Improved Drought Stress Identification
标题：MRD-LiNet：一种新的轻量级混合CNN，具有主动引导的学习，用于改进干旱胁迫识别
链接：https://arxiv.org/abs/2509.06367

【5】Exploring approaches to computational representation and classification of user-generated meal logs
标题：探索用户生成的膳食日志的计算表示和分类方法
链接：https://arxiv.org/abs/2509.06330

【6】Enhancing Low-Altitude Airspace Security: MLLM-Enabled UAV Intent Recognition
标题：增强低空空域安全：支持MLLM的无人机意图识别
链接：https://arxiv.org/abs/2509.06312

【7】Ensemble of Precision-Recall Curve (PRC) Classification Trees with Autoencoders
标题：使用自动编码器集成精确召回曲线（PRC）分类树
链接：https://arxiv.org/abs/2509.05766

【8】Systematic Evaluation of Multi-modal Approaches to Complex Player Profile Classification
标题：复杂球员概况分类的多模式方法的系统评估
链接：https://arxiv.org/abs/2509.05624

【9】Quantum spatial best-arm identification via quantum walks
标题：通过量子行走进行量子空间最佳臂识别
链接：https://arxiv.org/abs/2509.05890

【10】Risk-averse Fair Multi-class Classification
标题：规避风险公平多类别分类
链接：https://arxiv.org/abs/2509.05771

表征(2篇)

【1】MSRFormer: Road Network Representation Learning using Multi-scale Feature Fusion of Heterogeneous Spatial Interactions
标题：MSRFormer：基于异构空间交互的多尺度特征融合的道路网络表示学习
链接：https://arxiv.org/abs/2509.05685

【2】PLanTS: Periodicity-aware Latent-state Representation Learning for Multivariate Time Series
标题：PLanTS：多元时间序列的周期感知潜在状态表示学习
链接：https://arxiv.org/abs/2509.05478

3D|3D重建等相关(1篇)

【1】Learning in ImaginationLand: Omnidirectional Policies through 3D Generative Models (OP-Gen)
标题：ImaginationLand学习：通过3D生成模型实现全方位政策（OP-Gen）
链接：https://arxiv.org/abs/2509.06191

编码器(1篇)

【1】mmBERT: A Modern Multilingual Encoder with Annealed Language Learning
标题：mmBERT：具有Annealed语言学习的现代多语言编码器
链接：https://arxiv.org/abs/2509.06888

优化|敛散性(9篇)

【1】COMPACT: Common-token Optimized Model Pruning Across Channels and Tokens
标题：COMACT：跨渠道和代币修剪的公共代币优化模型
链接：https://arxiv.org/abs/2509.06836

【2】Nested Optimal Transport Distances
标题：嵌套最佳运输距离
链接：https://arxiv.org/abs/2509.06702

【3】Learning Optimal Defender Strategies for CAGE-2 using a POMDP Model
标题：使用POMDP模型学习CAGE-2的最佳防御策略
链接：https://arxiv.org/abs/2509.06539

【4】PLRV-O: Advancing Differentially Private Deep Learning via Privacy Loss Random Variable Optimization
标题：PLRV-O：通过隐私损失随机变量优化推进差异隐私深度学习
链接：https://arxiv.org/abs/2509.06264

【5】Smoothed Online Optimization for Target Tracking: Robust and Learning-Augmented Algorithms
标题：目标跟踪的平滑在线优化：鲁棒和学习增强算法
链接：https://arxiv.org/abs/2509.05930

【6】Simple Optimizers for Convex Aligned Multi-Objective Optimization
标题：用于凸对齐多目标优化的简单优化器
链接：https://arxiv.org/abs/2509.05811

【7】OptiProxy-NAS: Optimization Proxy based End-to-End Neural Architecture Search
标题：OptProxy-NAS：基于优化代理的端到端神经架构搜索
链接：https://arxiv.org/abs/2509.05656

【8】STL-based Optimization of Biomolecular Neural Networks for Regression and Control
标题：基于STL的生物分子神经网络回归和控制优化
链接：https://arxiv.org/abs/2509.05481

【9】Data-driven solar forecasting enables near-optimal economic decisions
标题：数据驱动的太阳能预测实现近乎最优的经济决策
链接：https://arxiv.org/abs/2509.06925

预测|估计(9篇)

【1】Video-Based MPAA Rating Prediction: An Attention-Driven Hybrid Architecture Using Contrastive Learning
标题：基于视频的MPA评级预测：使用对比学习的注意力驱动混合架构
链接：https://arxiv.org/abs/2509.06826

【2】CAME-AB: Cross-Modality Attention with Mixture-of-Experts for Antibody Binding Site Prediction
标题：CAME-AB：抗体结合位点预测专家混合的跨模式关注
链接：https://arxiv.org/abs/2509.06465

【3】Beyond the Pre-Service Horizon: Infusing In-Service Behavior for Improved Financial Risk Forecasting
标题：超越服务前视野：注入在职行为以改善财务风险预测
链接：https://arxiv.org/abs/2509.06385

【4】Grasp-MPC: Closed-Loop Visual Grasping via Value-Guided Model Predictive Control
标题：Grasp-MPC：通过价值引导模型预测控制的闭环视觉抓取
链接：https://arxiv.org/abs/2509.06201

【5】Unified Interaction Foundational Model (UIFM) for Predicting Complex User and System Behavior
标题：用于预测复杂用户和系统行为的统一交互基础模型（UIFM）
链接：https://arxiv.org/abs/2509.06025

【6】Select, then Balance: A Plug-and-Play Framework for Exogenous-Aware Spatio-Temporal Forecasting
标题：选择，然后选择平衡：外部感知时空预测的即插即用框架
链接：https://arxiv.org/abs/2509.05779

【7】Real-E: A Foundation Benchmark for Advancing Robust and Generalizable Electricity Forecasting
标题：Real-E：推进稳健和可推广电力预测的基础基准
链接：https://arxiv.org/abs/2509.05768

【8】Neural ARFIMA model for forecasting BRIC exchange rates with long memory under oil shocks and policy uncertainties
标题：石油冲击和政策不确定性下预测金砖四国汇率的长记忆神经ARFIMA模型
链接：https://arxiv.org/abs/2509.06697

【9】Topological Regularization for Force Prediction in Active Particle Suspension with EGNN and Persistent Homology
标题：基于EGNN和持续同调的活性颗粒悬浮液力预测的Topology regulation
链接：https://arxiv.org/abs/2509.06574

其他神经网络|深度学习|模型|建模(37篇)

【1】Deep Reactive Policy: Learning Reactive Manipulator Motion Planning for Dynamic Environments
标题：深度反应策略：学习动态环境的反应式机械手运动规划
链接：https://arxiv.org/abs/2509.06953

【2】Learning words in groups: fusion algebras, tensor ranks and grokking
标题：分组学习单词：融合代数、张量等级和Grokking
链接：https://arxiv.org/abs/2509.06931

【3】Concolic Testing on Individual Fairness of Neural Network Models
标题：神经网络模型个体公平性的协和检验
链接：https://arxiv.org/abs/2509.06864

【4】Curia: A Multi-Modal Foundation Model for Radiology
标题：Curia：放射学多模式基础模型
链接：https://arxiv.org/abs/2509.06830

【5】Dato: A Task-Based Programming Model for Dataflow Accelerators
标题：Dato：数据流加速器的基于任务的编程模型
链接：https://arxiv.org/abs/2509.06794

【6】When Secure Isn't: Assessing the Security of Machine Learning Model Sharing
标题：当不安全时：评估机器学习模型共享的安全性
链接：https://arxiv.org/abs/2509.06703

【7】Probabilistic Modeling of Latent Agentic Substructures in Deep Neural Networks
标题：深度神经网络中潜在统计子结构的概率建模
链接：https://arxiv.org/abs/2509.06701

【8】Barycentric Neural Networks and Length-Weighted Persistent Entropy Loss: A Green Geometric and Topological Framework for Function Approximation
标题：以重心为中心的神经网络和长度加权持续性熵损失：函数逼近的绿色几何和布局框架
链接：https://arxiv.org/abs/2509.06694

【9】Knowledge-Guided Machine Learning for Stabilizing Near-Shortest Path Routing
标题：稳定近最短路径路由的知识引导机器学习
链接：https://arxiv.org/abs/2509.06640

【10】BEAM: Brainwave Empathy Assessment Model for Early Childhood
标题：BEAM：幼儿脑电波同理心评估模型
链接：https://arxiv.org/abs/2509.06620

【11】Information-Theoretic Bounds and Task-Centric Learning Complexity for Real-World Dynamic Nonlinear Systems
标题：现实世界动态非线性系统的信息论界和任务中心学习复杂性
链接：https://arxiv.org/abs/2509.06599

【12】On the Reproducibility of "FairCLIP: Harnessing Fairness in Vision-Language Learning''
链接：https://arxiv.org/abs/2509.06535

【13】NeuroDeX: Unlocking Diverse Support in Decompiling Deep Neural Network Executables
标题：NeuroDeX：解锁反编译深度神经网络可执行文件的多元化支持
链接：https://arxiv.org/abs/2509.06402

【14】A Multi-Modal Deep Learning Framework for Colorectal Pathology Diagnosis: Integrating Histological and Colonoscopy Data in a Pilot Study
标题：用于结直肠病理诊断的多模式深度学习框架：在试点研究中集成组织学和结肠镜检查数据
链接：https://arxiv.org/abs/2509.06351

【15】Modeling shopper interest broadness with entropy-driven dialogue policy in the context of arbitrarily large product catalogs
标题：在任意大的产品目录的背景下，利用信息驱动的对话政策对购物者的兴趣广度进行建模
链接：https://arxiv.org/abs/2509.06185

【16】Tracking daily paths in home contexts with RSSI fingerprinting based on UWB through deep learning models
标题：通过深度学习模型，利用基于UWB的RTI指纹识别跟踪家庭环境中的日常路径
链接：https://arxiv.org/abs/2509.06161

【17】Teaching Precommitted Agents: Model-Free Policy Evaluation and Control in Quasi-Hyperbolic Discounted MDPs
标题：教授预先承诺的代理：准双曲折扣MDP中的无模型政策评估和控制
链接：https://arxiv.org/abs/2509.06094

【18】A Surrogate model for High Temperature Superconducting Magnets to Predict Current Distribution with Neural Network
标题：用神经网络预测电流分布的高温超导磁铁替代模型
链接：https://arxiv.org/abs/2509.06067

【19】A novel biomass fluidized bed gasification model coupled with machine learning and CFD simulation
标题：结合机器学习和计算流体模拟的新型生物质沸腾床气化模型
链接：https://arxiv.org/abs/2509.06056

【20】BranchGRPO: Stable and Efficient GRPO with Structured Branching in Diffusion Models
标题：BranchGRPO：扩散模型中具有结构化分支的稳定有效GRPO
链接：https://arxiv.org/abs/2509.06040

【21】Code2MCP: A Multi-Agent Framework for Automated Transformation of Code Repositories into Model Context Protocol Services
标题：Code2MCP：一个多Agent框架，用于代码仓库到模型上下文协议服务的自动转换
链接：https://arxiv.org/abs/2509.05941

【22】Data-Driven Stochastic Modeling Using Autoregressive Sequence Models: Translating Event Tables to Queueing Dynamics
标题：使用自回归序列模型的数据驱动随机建模：将事件表转化为排队动力学
链接：https://arxiv.org/abs/2509.05839

【23】time2time: Causal Intervention in Hidden States to Simulate Rare Events in Time Series Foundation Models
标题：time 2time：在隐藏状态中进行因果干预以模拟时间序列基础模型中的罕见事件
链接：https://arxiv.org/abs/2509.05801

【24】Offline vs. Online Learning in Model-based RL: Lessons for Data Collection Strategies
标题：基于模型的RL中的离线与在线学习：数据收集策略的课程
链接：https://arxiv.org/abs/2509.05735

【25】Simulation Priors for Data-Efficient Deep Learning
标题：数据高效深度学习的模拟先验
链接：https://arxiv.org/abs/2509.05732

【26】Distributed Deep Learning using Stochastic Gradient Staleness
标题：使用随机梯度停滞的分布式深度学习
链接：https://arxiv.org/abs/2509.05679

【27】DreamPRM-1.5: Unlocking the Potential of Each Instance for Multimodal Process Reward Model Training
标题：DreamPRM-1.5：释放每个实例用于多模式流程奖励模型训练的潜力
链接：https://arxiv.org/abs/2509.05542

【28】Prior Distribution and Model Confidence
标题：先验分布和模型置信度
链接：https://arxiv.org/abs/2509.05485

【29】Long-Horizon Visual Imitation Learning via Plan and Code Reflection
标题：通过计划和代码反思进行长期视觉模仿学习
链接：https://arxiv.org/abs/2509.05368

【30】Spiking Neural Networks for Continuous Control via End-to-End Model-Based Learning
标题：通过基于模型的端到端学习实现连续控制的尖峰神经网络
链接：https://arxiv.org/abs/2509.05356

【31】Sequential Least-Squares Estimators with Fast Randomized Sketching for Linear Statistical Models
标题：线性统计模型的快速随机绘制顺序最小平方估计器
链接：https://arxiv.org/abs/2509.06856

【32】Green Learning for STAR-RIS mmWave Systems with Implicit CSI
标题：具有隐式SI的STAR-RIS毫米波系统的绿色学习
链接：https://arxiv.org/abs/2509.06820

【33】Repeating vs. Non-Repeating FRBs: A Deep Learning Approach To Morphological Characterization
标题：重复与非重复FRB：形态学特征的深度学习方法
链接：https://arxiv.org/abs/2509.06208

【34】Machine learning magnetism from simple global descriptors
标题：来自简单全局描述符的机器学习吸引力
链接：https://arxiv.org/abs/2509.05909

【35】Volatility Modeling via EWMA-Driven Time-Dependent Hurst Parameters
标题：通过EGMA驱动的时间相关Hurst参数进行波动率建模
链接：https://arxiv.org/abs/2509.05820

【36】Hybrid Fourier Neural Operator-Plasma Fluid Model for Fast and Accurate Multiscale Simulations of High Power Microwave Breakdown
标题：混合傅里叶神经操作器-等离子体流体模型用于高功率微波击穿的快速准确多尺度模拟
链接：https://arxiv.org/abs/2509.05799

【37】Causal Multi-fidelity Surrogate Forward and Inverse Models for ICF Implosions
标题：ICF内爆的因果多保真度替代正向和逆模型
链接：https://arxiv.org/abs/2509.05510

其他(42篇)

【1】Directly Aligning the Full Diffusion Trajectory with Fine-Grained Human Preference
标题：将全扩散轨迹与细粒度人类偏好直接对齐
链接：https://arxiv.org/abs/2509.06942

【2】Neutron Reflectometry by Gradient Descent
标题：梯度下降法的中子反射测量
链接：https://arxiv.org/abs/2509.06924

【3】Paper2Agent: Reimagining Research Papers As Interactive and Reliable AI Agents
标题：Paper2Agent：将研究论文重新构想为交互式且可靠的人工智能代理
链接：https://arxiv.org/abs/2509.06917

【4】Not All Samples Are Equal: Quantifying Instance-level Difficulty in Targeted Data Poisoning
标题：并非所有样本都是相等的：量化目标数据中毒中的实例级难度
链接：https://arxiv.org/abs/2509.06896

【5】floq: Training Critics via Flow-Matching for Scaling Compute in Value-Based RL
标题：floq：通过基于价值的RL中的缩放计算的流匹配训练批评者
链接：https://arxiv.org/abs/2509.06863

【6】ToonOut: Fine-tuned Background-Removal for Anime Characters
标题：ToonOut：微调背景删除动漫人物
链接：https://arxiv.org/abs/2509.06839

【7】UMO: Scaling Multi-Identity Consistency for Image Customization via Matching Reward
标题：UMO：通过匹配奖励扩展图像定制的多身份一致性
链接：https://arxiv.org/abs/2509.06818

【8】\texttt{R$^\textbf{2}$AI}: Towards Resistant and Resilient AI in an Evolving World
标题： extttt {R$# extBF{2}$AI}：在不断发展的世界中迈向具有抵抗力和弹性的人工智能
链接：https://arxiv.org/abs/2509.06786

【9】AI for Scientific Discovery is a Social Problem
标题：人工智能用于科学发现是一个社会问题
链接：https://arxiv.org/abs/2509.06580

【10】Approximating Condorcet Ordering for Vector-valued Mathematical Morphology
标题：向量数学形态学的康多塞逼近排序
链接：https://arxiv.org/abs/2509.06577

【11】Tackling Device Data Distribution Real-time Shift via Prototype-based Parameter Editing
标题：基于原型的参数编辑解决设备数据分布实时偏移
链接：https://arxiv.org/abs/2509.06552

【12】A machine-learned expression for the excess Gibbs energy
标题：超额吉布斯能量的机器学习表达
链接：https://arxiv.org/abs/2509.06484

【13】Variational Garrote for Statistical Physics-based Sparse and Robust Variable Selection
标题：基于统计物理的稀疏稳健变量选择变分绞喉算法
链接：https://arxiv.org/abs/2509.06383

【14】Breaking SafetyCore: Exploring the Risks of On-Device AI Deployment
标题：打破SafetyCore：探索设备上AI部署的风险
链接：https://arxiv.org/abs/2509.06371

【15】A data-driven discretized CS:GO simulation environment to facilitate strategic multi-agent planning research
标题：数据驱动的离散化CS：GO模拟环境，促进战略多智能体规划研究
链接：https://arxiv.org/abs/2509.06355

【16】Embedding Poisoning: Bypassing Safety Alignment via Embedding Semantic Shift
标题：嵌入中毒：通过嵌入语义转变来实现安全一致
链接：https://arxiv.org/abs/2509.06338

【17】Evaluating the Efficiency of Latent Spaces via the Coupling-Matrix
标题：通过耦合矩阵评估潜在空间的效率
链接：https://arxiv.org/abs/2509.06314

【18】LoaQ: Layer-wise Output Approximation Quantization
标题：LoaQ：逐层输出逼近量化
链接：https://arxiv.org/abs/2509.06297

【19】IPR: Intelligent Prompt Routing with User-Controlled Quality-Cost Trade-offs
标题：IPR：具有用户控制质量成本权衡的智能提示路由
链接：https://arxiv.org/abs/2509.06274

【20】UrbanMIMOMap: A Ray-Traced MIMO CSI Dataset with Precoding-Aware Maps and Benchmarks
标题：UrbanMIMOMap：一个带有预编码感知地图和基准的光线跟踪MIMO CSI数据集
链接：https://arxiv.org/abs/2509.06270

【21】Exploring Urban Factors with Autoencoders: Relationship Between Static and Dynamic Features
标题：用自动编码器探索城市因素：静态和动态特征之间的关系
链接：https://arxiv.org/abs/2509.06167

【22】An Improved Template for Approximate Computing
标题：一种改进的逼近计算模板
链接：https://arxiv.org/abs/2509.06162

【23】If generative AI is the answer, what is the question?
标题：如果生成性人工智能是答案，那么问题是什么？
链接：https://arxiv.org/abs/2509.06120

【24】Khana: A Comprehensive Indian Cuisine Dataset
标题：Khana：全面的印度美食数据集
链接：https://arxiv.org/abs/2509.06006

【25】Benchmarking Robust Aggregation in Decentralized Gradient Marketplaces
标题：在去中心化梯度市场中对稳健聚合进行基准
链接：https://arxiv.org/abs/2509.05833

【26】InterAct: A Large-Scale Dataset of Dynamic, Expressive and Interactive Activities between Two People in Daily Scenarios
标题：InterAct：日常场景中两个人之间动态、表达和互动活动的大规模数据集
链接：https://arxiv.org/abs/2509.05747

【27】Morphological Perceptron with Competitive Layer: Training Using Convex-Concave Procedure
标题：具有竞争层的形态感知器：使用凹凸程序进行训练
链接：https://arxiv.org/abs/2509.05697

【28】Audits Under Resource, Data, and Access Constraints: Scaling Laws For Less Discriminatory Alternatives
标题：资源、数据和访问限制下的审计：为较少歧视性的替代方案调整法律
链接：https://arxiv.org/abs/2509.05627

【29】Self-Aligned Reward: Towards Effective and Efficient Reasoners
标题：自我调整的奖励：走向有效和高效的推理者
链接：https://arxiv.org/abs/2509.05489

【30】FAVAE-Effective Frequency Aware Latent Tokenizer
标题：FAVE有效频率感知潜在代币器
链接：https://arxiv.org/abs/2509.05441

【31】Direct-Scoring NLG Evaluators Can Use Pairwise Comparisons Too
标题：直接评分NLG评估者也可以使用成对比较
链接：https://arxiv.org/abs/2509.05440

【32】Murphys Laws of AI Alignment: Why the Gap Always Wins
标题：墨菲人工智能对齐定律：为什么差距总是获胜
链接：https://arxiv.org/abs/2509.05381

【33】Delta Velocity Rectified Flow for Text-to-Image Editing
标题：用于文本到图像编辑的Delta Speed纠正流
链接：https://arxiv.org/abs/2509.05342

【34】Feed Two Birds with One Scone: Exploiting Function-Space Regularization for Both OOD Robustness and ID Fine-Tuning Performance
标题：一个烤饼喂两只鸟：利用功能空间正规化来实现OOD稳健性和ID微调性能
链接：https://arxiv.org/abs/2509.05328

【35】Nonnegative matrix factorization and the principle of the common cause
标题：非负矩阵分解与共因原理
链接：https://arxiv.org/abs/2509.03652

【36】Integrating Spatial and Semantic Embeddings for Stereo Sound Event Localization in Videos
标题：集成空间和语义嵌入以实现视频中的立体声事件定位
链接：https://arxiv.org/abs/2509.06598

【37】Musculoskeletal simulation of limb movement biomechanics in Drosophila melanogaster
标题：黑腹果蝇肢体运动生物力学的肌肉骨骼模拟
链接：https://arxiv.org/abs/2509.06426

【38】The Efficiency Frontier: Classical Shadows versus Quantum Footage
标题：效率前沿：经典阴影与量子镜头
链接：https://arxiv.org/abs/2509.06218

【39】Additive Distributionally Robust Ranking and Selection
标题：加性分布稳健排名和选择
链接：https://arxiv.org/abs/2509.06147

【40】Spectral Methods in Complex Systems
标题：复杂系统中的谱方法
链接：https://arxiv.org/abs/2509.05793

【41】Vector-based loss functions for turbulent flow field inpainting
标题：湍流场修复的基于载体的损失函数
链接：https://arxiv.org/abs/2509.05787

【42】Cryo-EM as a Stochastic Inverse Problem
标题：Cryo-EM作为随机反问题
链接：https://arxiv.org/abs/2509.05541

机器翻译由腾讯交互翻译提供，仅供参考

点击“阅读原文”获取带摘要的学术速递