机器学习学术速递[9.23]

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274篇

大模型相关(33篇)

【1】Spiffy: Multiplying Diffusion LLM Acceleration via Lossless Speculative Decoding
标题：Spiffy：通过无损推测解码乘以扩散LLM加速
链接：https://arxiv.org/abs/2509.18085

【2】Strategic Dishonesty Can Undermine AI Safety Evaluations of Frontier LLM
标题：战略不诚实可能破坏Frontier LLM的AI安全评估
链接：https://arxiv.org/abs/2509.18058

【3】Adaptive Kernel Design for Bayesian Optimization Is a Piece of CAKE with LLMs
标题：用于Bayesian优化的自适应核设计是LLM的小菜一碟
链接：https://arxiv.org/abs/2509.17998

【4】Variation in Verification: Understanding Verification Dynamics in Large Language Models
标题：验证的变化：了解大型语言模型中的验证动态
链接：https://arxiv.org/abs/2509.17995

【5】Understanding Post-Training Structural Changes in Large Language Models
标题：了解大型语言模型的训练后结构变化
链接：https://arxiv.org/abs/2509.17866

【6】Revealing Multimodal Causality with Large Language Models
标题：用大型语言模型揭示多模式因果关系
链接：https://arxiv.org/abs/2509.17784

【7】ConfClip: Confidence-Weighted and Clipped Reward for Reinforcement Learning in LLMs
标题：ConfTrap：LLM中强化学习的信心加权和缩减奖励
链接：https://arxiv.org/abs/2509.17730

【8】Investigating Bias: A Multilingual Pipeline for Generating, Solving, and Evaluating Math Problems with LLMs
标题：调查偏见：使用LLM生成、解决和评估数学问题的多语言管道
链接：https://arxiv.org/abs/2509.17701

【9】Mechanistic Interpretability with SAEs: Probing Religion, Violence, and Geography in Large Language Models
标题：SAEs的机械解释性：在大型语言模型中探索宗教、暴力和地理
链接：https://arxiv.org/abs/2509.17665

【10】AuditoryBench++: Can Language Models Understand Auditory Knowledge without Hearing?
标题：AuditoryBench++：语言模型可以在没有听力的情况下理解听觉知识吗？
链接：https://arxiv.org/abs/2509.17641

【11】Interpreting Attention Heads for Image-to-Text Information Flow in Large Vision-Language Models
标题：解释大型视觉语言模型中图像到文本信息流的注意力
链接：https://arxiv.org/abs/2509.17588

【12】SilentStriker:Toward Stealthy Bit-Flip Attacks on Large Language Models
标题：SilentStriker：走向对大型语言模型的隐形位翻转攻击
链接：https://arxiv.org/abs/2509.17371

【13】CogAtom: From Cognitive Atoms to Olympiad-level Mathematical Reasoning in Large Language Models
标题：CogAtom：从认知原子到大型语言模型中的奥运级数学推理
链接：https://arxiv.org/abs/2509.17318

【14】Clotho: Measuring Task-Specific Pre-Generation Test Adequacy for LLM Inputs
标题：Clotho：衡量LLM输入的特定任务前一代测试合格性
链接：https://arxiv.org/abs/2509.17314

【15】Probabilistic Token Alignment for Large Language Model Fusion
标题：大型语言模型融合的概率令牌对齐
链接：https://arxiv.org/abs/2509.17276

【16】Can Agents Judge Systematic Reviews Like Humans? Evaluating SLRs with LLM-based Multi-Agent System
标题：代理人可以像人类一样判断系统评价吗？利用基于LLM的多智能体系统评估SLR
链接：https://arxiv.org/abs/2509.17240

【17】SignalLLM: A General-Purpose LLM Agent Framework for Automated Signal Processing
标题：SignalLLM：用于自动信号处理的通用LLM代理框架
链接：https://arxiv.org/abs/2509.17197

【18】LifeAlign: Lifelong Alignment for Large Language Models with Memory-Augmented Focalized Preference Optimization
标题：LifeAlign：通过内存增强集中偏好优化实现大型语言模型的终身一致
链接：https://arxiv.org/abs/2509.17183

【19】The Transfer Neurons Hypothesis: An Underlying Mechanism for Language Latent Space Transitions in Multilingual LLMs
标题：迁移神经元假说：多语言LLM中语言潜空间转换的潜在机制
链接：https://arxiv.org/abs/2509.17030

【20】Advancing Speech Understanding in Speech-Aware Language Models with GRPO
标题：使用GRPO推进语音感知语言模型中的语音理解
链接：https://arxiv.org/abs/2509.16990

【21】PTQTP: Post-Training Quantization to Trit-Planes for Large Language Models
标题：PTQTP：大型语言模型的三平面训练后量化
链接：https://arxiv.org/abs/2509.16989

【22】seqBench: A Tunable Benchmark to Quantify Sequential Reasoning Limits of LLMs
标题：seqBench：量化LLM序列推理限制的可调基准
链接：https://arxiv.org/abs/2509.16866

【23】DISCO: Disentangled Communication Steering for Large Language Models
标题：DISCO：大型语言模型的分离沟通引导
链接：https://arxiv.org/abs/2509.16820

【24】Decoding Uncertainty: The Impact of Decoding Strategies for Uncertainty Estimation in Large Language Models
标题：解码不确定性：解码策略对大型语言模型中不确定性估计的影响
链接：https://arxiv.org/abs/2509.16696

【25】FESTA: Functionally Equivalent Sampling for Trust Assessment of Multimodal LLMs
标题：FESTA：多模式LLM信任评估的功能等效抽样
链接：https://arxiv.org/abs/2509.16648

【26】mmExpert: Integrating Large Language Models for Comprehensive mmWave Data Synthesis and Understanding
标题：mmExpert：集成大型语言模型以实现全面的毫米波数据合成和理解
链接：https://arxiv.org/abs/2509.16521

【27】LLM-Guided Co-Training for Text Classification
标题：LLM引导的文本分类协同训练
链接：https://arxiv.org/abs/2509.16516

【28】GRIL: Knowledge Graph Retrieval-Integrated Learning with Large Language Models
标题：GRIL：知识图谱检索-与大型语言模型的集成学习
链接：https://arxiv.org/abs/2509.16502

【29】Towards Universal Debiasing for Language Models-based Tabular Data Generation
标题：面向基于语言模型的表格数据生成的普遍去偏置
链接：https://arxiv.org/abs/2509.16475

【30】Intrinsic Meets Extrinsic Fairness: Assessing the Downstream Impact of Bias Mitigation in Large Language Models
标题：内在与外在公平：评估大型语言模型中偏差缓解的下游影响
链接：https://arxiv.org/abs/2509.16462

【31】Robust LLM Training Infrastructure at ByteDance
标题：字节跳动强大的法学硕士训练基础设施
链接：https://arxiv.org/abs/2509.16293

【32】Gender and Political Bias in Large Language Models: A Demonstration Platform
标题：大型语言模型中的性别和政治偏见：一个演示平台
链接：https://arxiv.org/abs/2509.16264

【33】How Can Quantum Deep Learning Improve Large Language Models?
标题：量子深度学习如何改进大型语言模型？
链接：https://arxiv.org/abs/2509.16244

Graph相关(图学习|图神经网络|图优化等)(23篇)

【1】A Knowledge Graph-based Retrieval-Augmented Generation Framework for Algorithm Selection in the Facility Layout Problem
标题：基于知识图的检索增强生成框架用于设施布局问题中的算法选择
链接：https://arxiv.org/abs/2509.18054

【2】Budgeted Adversarial Attack against Graph-Based Anomaly Detection in Sensor Networks
标题：针对传感器网络中基于图的异常检测的潜在对抗攻击
链接：https://arxiv.org/abs/2509.17987

【3】MSGAT-GRU: A Multi-Scale Graph Attention and Recurrent Model for Spatiotemporal Road Accident Prediction
标题：MSGAT-GRU：用于时空道路事故预测的多尺度图注意力和回归模型
链接：https://arxiv.org/abs/2509.17811

【4】A non-smooth regularization framework for learning over multitask graphs
标题：用于多任务图学习的非光滑正规化框架
链接：https://arxiv.org/abs/2509.17728

【5】Fast, Accurate and Interpretable Graph Classification with Topological Kernels
标题：使用拓扑核快速、准确且可解释的图分类
链接：https://arxiv.org/abs/2509.17693

【6】Periodic Graph-Enhanced Multivariate Time Series Anomaly Detector
标题：周期图增强多元时间序列异常检测器
链接：https://arxiv.org/abs/2509.17472

【7】Robust Anomaly Detection Under Normality Distribution Shift in Dynamic Graphs
标题：动态图正态分布漂移下的鲁棒异常检测
链接：https://arxiv.org/abs/2509.17400

【8】Word2VecGD: Neural Graph Drawing with Cosine-Stress Optimization
标题：Word 2 VecGD：利用Cosine-Stress优化的神经图绘制
链接：https://arxiv.org/abs/2509.17333

【9】GraphWeave: Interpretable and Robust Graph Generation via Random Walk Trajectories
标题：GraphWeave：通过随机游走轨迹生成可解释且稳健的图形
链接：https://arxiv.org/abs/2509.17291

【10】Graph Signal Generative Diffusion Models
标题：图信号生成扩散模型
链接：https://arxiv.org/abs/2509.17250

【11】Prospective Multi-Graph Cohesion for Multivariate Time Series Anomaly Detection
标题：用于多元时间序列异常检测的前瞻性多图融合
链接：https://arxiv.org/abs/2509.17235

【12】Unrolled Graph Neural Networks for Constrained Optimization
标题：约束优化的展开图神经网络
链接：https://arxiv.org/abs/2509.17156

【13】Gradient Interference-Aware Graph Coloring for Multitask Learning
标题：用于多任务学习的梯度干扰感知图着色
链接：https://arxiv.org/abs/2509.16959

【14】Adaptive Graph Convolution and Semantic-Guided Attention for Multimodal Risk Detection in Social Networks
标题：自适应图卷积和语义引导注意力用于社交网络中的多模式风险检测
链接：https://arxiv.org/abs/2509.16936

【15】Randomized Space-Time Sampling for Affine Graph Dynamical Systems
标题：仿射图动力系统的随机时空采样
链接：https://arxiv.org/abs/2509.16818

【16】Self-Supervised Learning of Graph Representations for Network Intrusion Detection
标题：用于网络入侵检测的图表示的自监督学习
链接：https://arxiv.org/abs/2509.16625

【17】Bayesian Ego-graph inference for Networked Multi-Agent Reinforcement Learning
标题：网络多智能体强化学习的Bayesian自我图推理
链接：https://arxiv.org/abs/2509.16606

【18】Entropic Causal Inference: Graph Identifiability
标题：熵因果推理：图的可识别性
链接：https://arxiv.org/abs/2509.16463

【19】GRID: Graph-based Reasoning for Intervention and Discovery in Built Environments
标题：GRID：建筑环境中的干预和发现基于图形的推理
链接：https://arxiv.org/abs/2509.16397

【20】Neural Atlas Graphs for Dynamic Scene Decomposition and Editing
标题：用于动态场景分解和编辑的神经阿特拉斯图
链接：https://arxiv.org/abs/2509.16336

【21】GraphMend: Code Transformations for Fixing Graph Breaks in PyTorch 2
标题：GraphMend：修复PyTorch 2中图形中断的代码转换
链接：https://arxiv.org/abs/2509.16248

【22】Self-Supervised Discovery of Neural Circuits in Spatially Patterned Neural Responses with Graph Neural Networks
标题：基于图神经网络的空间模式神经回路自监督发现
链接：https://arxiv.org/abs/2509.17174

【23】TF-DWGNet: A Directed Weighted Graph Neural Network with Tensor Fusion for Multi-Omics Cancer Subtype Classification
标题：TF-DWGNet：一种具有张量融合的有向加权图神经网络，用于多组癌症亚型分类
链接：https://arxiv.org/abs/2509.16301

Transformer(8篇)

【1】Optimizing Inference in Transformer-Based Models: A Multi-Method Benchmark
标题：优化基于转换器的模型中的推理：多方法基准
链接：https://arxiv.org/abs/2509.17894

【2】Conv-like Scale-Fusion Time Series Transformer: A Multi-Scale Representation for Variable-Length Long Time Series
标题：类Conv尺度融合时间序列Transformer：可变长度长时间序列的多尺度表示
链接：https://arxiv.org/abs/2509.17845

【3】MTM: A Multi-Scale Token Mixing Transformer for Irregular Multivariate Time Series Classification
标题：MTM：一种用于不规则多元时间序列分类的多尺度令牌混合Transformer
链接：https://arxiv.org/abs/2509.17809

【4】Transformer-Gather, Fuzzy-Reconsider: A Scalable Hybrid Framework for Entity Resolution
标题：Transformer-Gather、Fuzzy-Reconsider：实体解析的可扩展混合框架
链接：https://arxiv.org/abs/2509.17470

【5】Point-RTD: Replaced Token Denoising for Pretraining Transformer Models on Point Clouds
标题：Point-RTI：取代点云中预训练Transformer模型的令牌去噪
链接：https://arxiv.org/abs/2509.17207

【6】Time Series Forecasting Using a Hybrid Deep Learning Method: A Bi-LSTM Embedding Denoising Auto Encoder Transformer
标题：使用混合深度学习方法进行时间序列预测：Bi-LSTM嵌入去噪自动编码器Transformer
链接：https://arxiv.org/abs/2509.17165

【7】Causality-Induced Positional Encoding for Transformer-Based Representation Learning of Non-Sequential Features
标题：用于基于变换器的非序列特征表示学习的偶然性诱导位置编码
链接：https://arxiv.org/abs/2509.16629

【8】Mental Multi-class Classification on Social Media: Benchmarking Transformer Architectures against LSTM Models
标题：社交媒体上的心理多类别分类：Transformer架构与LSTM模型进行基准测试
链接：https://arxiv.org/abs/2509.16542

GAN|对抗|攻击|生成相关(8篇)

【1】Reinforced Generation of Combinatorial Structures: Applications to Complexity Theory
标题：组合结构的增强生成：复杂性理论的应用
链接：https://arxiv.org/abs/2509.18057

【2】Shilling Recommender Systems by Generating Side-feature-aware Fake User Profiles
标题：通过生成侧面特征感知虚假用户配置文件来支付推荐系统
链接：https://arxiv.org/abs/2509.17918

【3】Is It Certainly a Deepfake? Reliability Analysis in Detection & Generation Ecosystem
标题：难道真的是假的吗？检测与发电生态系统中的可靠性分析
链接：https://arxiv.org/abs/2509.17550

【4】Training the next generation of physicians for artificial intelligence-assisted clinical neuroradiology: ASNR MICCAI Brain Tumor Segmentation (BraTS) 2025 Lighthouse Challenge education platform
标题：训练下一代医生进行人工智能辅助临床神经放射学：ASNR MICCAE脑肿瘤分割（BraTS）2025 Lighthouse Challenge教育平台
链接：https://arxiv.org/abs/2509.17281

【5】Guided and Unguided Conditional Diffusion Mechanisms for Structured and Semantically-Aware 3D Point Cloud Generation
标题：用于结构化和语义感知3D点云生成的引导和非引导条件扩散机制
链接：https://arxiv.org/abs/2509.17206

【6】Improving User Interface Generation Models from Designer Feedback
标题：根据设计师反馈改进用户界面生成模型
链接：https://arxiv.org/abs/2509.16779

【7】Pain in 3D: Generating Controllable Synthetic Faces for Automated Pain Assessment
标题：3D中的疼痛：生成可控制的合成面部以进行自动疼痛评估
链接：https://arxiv.org/abs/2509.16727

【8】Etude: Piano Cover Generation with a Three-Stage Approach -- Extract, strucTUralize, and DEcode
标题：练习曲：三阶段方法的钢琴翻唱--提取、结构化和解码
链接：https://arxiv.org/abs/2509.16522

半/弱/无/有监督|不确定性|主动学习(10篇)

【1】Unsupervised Learning and Representation of Mandarin Tonal Categories by a Generative CNN
标题：生成式CNN的无监督学习和普通话语气类别表示
链接：https://arxiv.org/abs/2509.17859

【2】Automated Labeling of Intracranial Arteries with Uncertainty Quantification Using Deep Learning
标题：使用深度学习通过不确定性量化自动标记脑动脉
链接：https://arxiv.org/abs/2509.17726

【3】SPICED: A Synaptic Homeostasis-Inspired Framework for Unsupervised Continual EEG Decoding
标题：SPICED：一个受突触Homestasism启发的无监督连续脑电解码框架
链接：https://arxiv.org/abs/2509.17439

【4】Active Learning for Machine Learning Driven Molecular Dynamics
标题：机器学习驱动的分子动力学的主动学习
链接：https://arxiv.org/abs/2509.17208

【5】ScenGAN: Attention-Intensive Generative Model for Uncertainty-Aware Renewable Scenario Forecasting
标题：ScenGAN：用于不确定性感知可再生能源场景预测的注意力密集型生成模型
链接：https://arxiv.org/abs/2509.17119

【6】Uncertainty-Supervised Interpretable and Robust Evidential Segmentation
标题：不确定性监督的可解释和稳健的证据分割
链接：https://arxiv.org/abs/2509.17098

【7】SCAN: Self-Denoising Monte Carlo Annotation for Robust Process Reward Learning
标题：SCAN：用于稳健流程奖励学习的自去噪蒙特卡洛注释
链接：https://arxiv.org/abs/2509.16548

【8】Auto-bidding under Return-on-Spend Constraints with Uncertainty Quantification
标题：具有不确定性量化的支出回报率约束下的自动竞价
链接：https://arxiv.org/abs/2509.16324

【9】Comparison of Deterministic and Probabilistic Machine Learning Algorithms for Precise Dimensional Control and Uncertainty Quantification in Additive Manufacturing
标题：用于增材制造中精确尺寸控制和不确定性量化的确定性和概率机器学习算法的比较
链接：https://arxiv.org/abs/2509.16233

【10】System-Level Uncertainty Quantification with Multiple Machine Learning Models: A Theoretical Framework
标题：多机器学习模型的系统级不确定性量化：理论框架
链接：https://arxiv.org/abs/2509.16663

迁移|Zero/Few/One-Shot|自适应(9篇)

【1】SeqBattNet: A Discrete-State Physics-Informed Neural Network with Aging Adaptation for Battery Modeling
标题：SeqBattNet：一个具有老化适应性的离散状态物理信息神经网络，用于电池建模
链接：https://arxiv.org/abs/2509.17621

【2】Efficient Sliced Wasserstein Distance Computation via Adaptive Bayesian Optimization
标题：通过自适应Bayesian优化进行高效切片Wasserstein距离计算
链接：https://arxiv.org/abs/2509.17405

【3】Adaptive Overclocking: Dynamic Control of Thinking Path Length via Real-Time Reasoning Signals
标题：自适应Overclock：通过实时推理信号动态控制思维路径长度
链接：https://arxiv.org/abs/2509.17000

【4】Dynamic Expert Specialization: Towards Catastrophic Forgetting-Free Multi-Domain MoE Adaptation
标题：动态专家专业化：迈向灾难性的无伪造多领域MoE适应
链接：https://arxiv.org/abs/2509.16882

【5】Domain-Adaptive Pre-Training for Arabic Aspect-Based Sentiment Analysis: A Comparative Study of Domain Adaptation and Fine-Tuning Strategies
标题：基于阿拉伯语情感分析的领域自适应预训练：领域自适应和微调策略的比较研究
链接：https://arxiv.org/abs/2509.16788

【6】Robust, Online, and Adaptive Decentralized Gaussian Processes
标题：稳健、在线和自适应的分散高斯过程
链接：https://arxiv.org/abs/2509.18011

【7】Quantum Adaptive Self-Attention for Financial Rebalancing: An Empirical Study on Automated Market Makers in Decentralized Finance
标题：金融再平衡的量子适应性自我注意力：去中心化金融中自动做市商的实证研究
链接：https://arxiv.org/abs/2509.16955

【8】Overfitting in Adaptive Robust Optimization
标题：自适应鲁棒优化中的过拟
链接：https://arxiv.org/abs/2509.16451

【9】Low-Rank Adaptation of Evolutionary Deep Neural Networks for Efficient Learning of Time-Dependent PDEs
标题：进化深度神经网络的低等级自适应，以有效学习时间相关的PCE
链接：https://arxiv.org/abs/2509.16395

强化学习(5篇)

【1】Strategic Coordination for Evolving Multi-agent Systems: A Hierarchical Reinforcement and Collective Learning Approach
标题：不断发展的多智能体系统的战略协调：分层强化和集体学习方法
链接：https://arxiv.org/abs/2509.18088

【2】Improving After-sales Service: Deep Reinforcement Learning for Dynamic Time Slot Assignment with Commitments and Customer Preferences
标题：改善售后服务：深度强化学习，以实现动态时段分配，包含承诺和客户偏好
链接：https://arxiv.org/abs/2509.17870

【3】HypeMARL: Multi-Agent Reinforcement Learning For High-Dimensional, Parametric, and Distributed Systems
标题：HypeMARL：面向多维、参数和分布式系统的多智能体强化学习
链接：https://arxiv.org/abs/2509.16709

【4】Test-Time Learning and Inference-Time Deliberation for Efficiency-First Offline Reinforcement Learning in Care Coordination and Population Health Management
标题：测试时学习和推理时审议，以实现效率第一-护理协调和人口健康管理中的离线强化学习
链接：https://arxiv.org/abs/2509.16291

【5】Deep Reinforcement Learning in Factor Investment
标题：要素投资中的深度强化学习
链接：https://arxiv.org/abs/2509.16206

元学习(1篇)

【1】Language Modeling with Learned Meta-Tokens
标题：使用学习元令牌进行语言建模
链接：https://arxiv.org/abs/2509.16278

符号|符号学习(1篇)

【1】Barwise Section Boundary Detection in Symbolic Music Using Convolutional Neural Networks
标题：利用卷积神经网络进行符号音乐中的巴氏段边界检测
链接：https://arxiv.org/abs/2509.16566

分层学习(1篇)

【1】Deep Hierarchical Learning with Nested Subspace Networks
标题：使用嵌套子空间网络的深度分层学习
链接：https://arxiv.org/abs/2509.17874

医学相关(10篇)

【1】ReDepress: A Cognitive Framework for Detecting Depression Relapse from Social Media
标题：ReDepress：从社交媒体上检测抑郁复发的认知框架
链接：https://arxiv.org/abs/2509.17991

【2】Medical priority fusion: achieving dual optimization of sensitivity and interpretability in nipt anomaly detection
标题：医学优先融合：实现乳头异常检测灵敏度和可解释性的双重优化
链接：https://arxiv.org/abs/2509.17924

【3】Causal Representation Learning from Multimodal Clinical Records under Non-Random Modality Missingness
标题：非随机模式缺失下多模式临床记录的因果表示学习
链接：https://arxiv.org/abs/2509.17228

【4】Interpretable Clinical Classification with Kolgomorov-Arnold Networks
标题：使用Kolgomorov-Arnold网络可解释临床分类
链接：https://arxiv.org/abs/2509.16750

【5】Towards a Transparent and Interpretable AI Model for Medical Image Classifications
标题：迈向透明且可解释的医学图像分类人工智能模型
链接：https://arxiv.org/abs/2509.16685

【6】A Novel Metric for Detecting Memorization in Generative Models for Brain MRI Synthesis
标题：一种检测脑MRI合成生成模型中再同步化的新指标
链接：https://arxiv.org/abs/2509.16582

【7】Accurate Thyroid Cancer Classification using a Novel Binary Pattern Driven Local Discrete Cosine Transform Descriptor
标题：使用新型二进制模式驱动的局部离散Cosine变换描述符进行准确的甲状腺癌分类
链接：https://arxiv.org/abs/2509.16382

【8】Estimating Clinical Lab Test Result Trajectories from PPG using Physiological Foundation Model and Patient-Aware State Space Model -- a UNIPHY+ Approach
标题：使用生理基础模型和患者感知状态空间模型从PPV估计临床实验室测试结果轨迹--UNIPHY+方法
链接：https://arxiv.org/abs/2509.16345

【9】Breast Cancer Classification Using Gradient Boosting Algorithms Focusing on Reducing the False Negative and SHAP for Explainability
标题：使用专注于减少假阴性和SHAP的梯度提升算法进行乳腺癌分类
链接：https://arxiv.org/abs/2403.09548

【10】Predicting Chest Radiograph Findings from Electrocardiograms Using Interpretable Machine Learning
标题：使用可解释机器学习根据心电图预测胸部X光检查结果
链接：https://arxiv.org/abs/2509.17674

蒸馏|知识提取(2篇)

【1】Brainprint-Modulated Target Speaker Extraction
标题：脑纹调制目标说话人提取
链接：https://arxiv.org/abs/2509.17883

【2】Knowledge Distillation for Variational Quantum Convolutional Neural Networks on Heterogeneous Data
标题：异类数据上变分量子卷积神经网络的知识提炼
链接：https://arxiv.org/abs/2509.16699

聚类(4篇)

【1】Intra-Cluster Mixup: An Effective Data Augmentation Technique for Complementary-Label Learning
标题：群内混合：一种用于补充标签学习的有效数据增强技术
链接：https://arxiv.org/abs/2509.17971

【2】Cluster Workload Allocation: A Predictive Approach Leveraging Machine Learning Efficiency
标题：集群队列分配：一种利用机器学习效率的预测方法
链接：https://arxiv.org/abs/2509.17695

【3】Machine Learning for Campus Energy Resilience: Clustering and Time-Series Forecasting in Intelligent Load Shedding
标题：校园能源弹性的机器学习：智能甩负荷中的集群和时间序列预测
链接：https://arxiv.org/abs/2509.17097

【4】Kernel K-means clustering of distributional data
标题：分布式数据的核K均值集群
链接：https://arxiv.org/abs/2509.18037

超分辨率|去噪|去模糊|去雾(1篇)

【1】Audio Super-Resolution with Latent Bridge Models
标题：基于潜桥模型的音频超分辨率
链接：https://arxiv.org/abs/2509.17609

自动驾驶|车辆|车道检测等(3篇)

【1】Building Transparency in Deep Learning-Powered Network Traffic Classification: A Traffic-Explainer Framework
标题：在深度学习驱动的网络流量分类中建立透明度：一个逻辑解释器框架
链接：https://arxiv.org/abs/2509.18007

【2】BiLCNet : BiLSTM-Conformer Network for Encrypted Traffic Classification with 5G SA Physical Channel Records
标题：BiLCNet：具有5G SA物理通道记录的加密流量分类的BiLSTM-Conformer网络
链接：https://arxiv.org/abs/2509.17495

【3】Stabilizing Information Flow Entropy: Regularization for Safe and Interpretable Autonomous Driving Perception
标题：稳定信息流熵：安全且可解释的自动驾驶感知的规范化
链接：https://arxiv.org/abs/2509.16277

联邦学习|隐私保护|加密(5篇)

【1】FedEL: Federated Elastic Learning for Heterogeneous Devices
标题：FedEL：针对异类设备的联合弹性学习
链接：https://arxiv.org/abs/2509.16902

【2】Learned Digital Codes for Over-the-Air Federated Learning
标题：用于空中联邦学习的学习数字代码
链接：https://arxiv.org/abs/2509.16577

【3】Federated Learning with Ad-hoc Adapter Insertions: The Case of Soft-Embeddings for Training Classifier-as-Retriever
标题：带有临时适配器插入的联邦学习：软嵌入用于训练分类器作为检索器的案例
链接：https://arxiv.org/abs/2509.16508

【4】orb-QFL: Orbital Quantum Federated Learning
标题：orb-QFL：轨道量子联邦学习
链接：https://arxiv.org/abs/2509.16505

【5】Federated Learning for Financial Forecasting
标题：财务预测联合学习
链接：https://arxiv.org/abs/2509.16393

推理|分析|理解|解释(10篇)

【1】Joint Optimization of Memory Frequency, Computing Frequency, Transmission Power and Task Offloading for Energy-efficient DNN Inference
标题：存储频率、计算频率、传输功率和任务卸载的联合优化以实现节能DNN推理
链接：https://arxiv.org/abs/2509.17970

【2】A Generative Conditional Distribution Equality Testing Framework and Its Minimax Analysis
标题：生成式条件分布均匀性测试框架及其极小极大分析
链接：https://arxiv.org/abs/2509.17729

【3】Comparing Data Assimilation and Likelihood-Based Inference on Latent State Estimation in Agent-Based Models
标题：基于代理的模型中潜在状态估计的数据同化和基于可能性的推断比较
链接：https://arxiv.org/abs/2509.17625

【4】Evaluating the Energy Efficiency of NPU-Accelerated Machine Learning Inference on Embedded Microcontrollers
标题：评估嵌入式微控制器上NPU加速机器学习推理的能源效率
链接：https://arxiv.org/abs/2509.17533

【5】MoEs Are Stronger than You Think: Hyper-Parallel Inference Scaling with RoE
标题：MoE比你想象的更强大：使用RoE的超并行推理扩展
链接：https://arxiv.org/abs/2509.17238

【6】Regularizing Extrapolation in Causal Inference
标题：因果推理中的外推规范化
链接：https://arxiv.org/abs/2509.17180

【7】FlagEval Findings Report: A Preliminary Evaluation of Large Reasoning Models on Automatically Verifiable Textual and Visual Questions
标题：FlagEval调查结果报告：自动验证文本和视觉问题上的大型推理模型的初步评估
链接：https://arxiv.org/abs/2509.17177

【8】Spectral Analysis of the Weighted Frobenius Objective
标题：加权弗罗贝尼乌斯目标的谱分析
链接：https://arxiv.org/abs/2509.16783

【9】Discrete Diffusion Models: Novel Analysis and New Sampler Guarantees
标题：离散扩散模型：新颖的分析和新的采样器保证
链接：https://arxiv.org/abs/2509.16756

【10】ProtoVQA: An Adaptable Prototypical Framework for Explainable Fine-Grained Visual Question Answering
标题：ProtoVQA：可解释细粒度视觉问题回答的可适应原型框架
链接：https://arxiv.org/abs/2509.16680

检测相关(8篇)

【1】Toward Affordable and Non-Invasive Detection of Hypoglycemia: A Machine Learning Approach
标题：迈向经济实惠且无创的低血糖检测：机器学习方法
链接：https://arxiv.org/abs/2509.17842

【2】Tailored Transformation Invariance for Industrial Anomaly Detection
标题：用于工业异常检测的定制转换不变性
链接：https://arxiv.org/abs/2509.17670

【3】A Comprehensive Performance Comparison of Traditional and Ensemble Machine Learning Models for Online Fraud Detection
标题：用于在线欺诈检测的传统和集成机器学习模型的全面性能比较
链接：https://arxiv.org/abs/2509.17176

【4】Detecting Urban PM$_{2.5}$ Hotspots with Mobile Sensing and Gaussian Process Regression
标题：利用移动传感和高斯过程回归检测城市PM$_{2.5}$热点
链接：https://arxiv.org/abs/2509.17175

【5】Long-Tailed Out-of-Distribution Detection with Refined Separate Class Learning
标题：基于精细分类学习的长尾分布外检测
链接：https://arxiv.org/abs/2509.17034

【6】DRES: Fake news detection by dynamic representation and ensemble selection
标题：DRES：通过动态表示和整体选择检测假新闻
链接：https://arxiv.org/abs/2509.16893

【7】CommonForms: A Large, Diverse Dataset for Form Field Detection
标题：CommonForm：用于表单字段检测的大型、多样化数据集
链接：https://arxiv.org/abs/2509.16506

【8】On the Detection of Internal Defects in Structured Media
标题：结构化媒体内部缺陷的检测
链接：https://arxiv.org/abs/2509.16216

分类|识别(9篇)

【1】DIVERS-Bench: Evaluating Language Identification Across Domain Shifts and Code-Switching
标题：潜水员长凳：跨领域转移和代码转换评估语言识别
链接：https://arxiv.org/abs/2509.17768

【2】PRNU-Bench: A Novel Benchmark and Model for PRNU-Based Camera Identification
标题：PRNU-Bench：基于PRNU的摄像机识别的新型基准和模型
链接：https://arxiv.org/abs/2509.17581

【3】Path-Weighted Integrated Gradients for Interpretable Dementia Classification
标题：可解释痴呆症分类的路径加权综合要素
链接：https://arxiv.org/abs/2509.17491

【4】MVCL-DAF++: Enhancing Multimodal Intent Recognition via Prototype-Aware Contrastive Alignment and Coarse-to-Fine Dynamic Attention Fusion
标题：MVCL-SYS ++：通过原型感知对比对齐和粗到细的动态注意力融合增强多模式意图识别
链接：https://arxiv.org/abs/2509.17446

【5】Multi-Scenario Highway Lane-Change Intention Prediction: A Physics-Informed AI Framework for Three-Class Classification
标题：多场景高速公路车道变更意图预测：用于三级分类的物理信息人工智能框架
链接：https://arxiv.org/abs/2509.17354

【6】Optimal Transport for Handwritten Text Recognition in a Low-Resource Regime
标题：低资源条件下手写文本识别的最佳传输
链接：https://arxiv.org/abs/2509.16977

【7】Geometric Mixture Classifier (GMC): A Discriminative Per-Class Mixture of Hyperplanes
标题：几何混合分类器（GMC）：超平面的逐类区分混合
链接：https://arxiv.org/abs/2509.16769

【8】Person Identification from Egocentric Human-Object Interactions using 3D Hand Pose
标题：使用3D手势从以自我为中心的人与物互动中识别人
链接：https://arxiv.org/abs/2509.16557

【9】Deep Learning Inductive Biases for fMRI Time Series Classification during Resting-state and Movie-watching
标题：静息状态和观看电影期间fMRI时间序列分类的深度学习诱导偏差
链接：https://arxiv.org/abs/2509.16973

表征(4篇)

【1】Can multimodal representation learning by alignment preserve modality-specific information?
标题：通过对齐进行的多模式表示学习可以保留特定于模式的信息吗？
链接：https://arxiv.org/abs/2509.17943

【2】Persistence Spheres: Bi-continuous Representations of Persistence Diagrams
标题：持久性球：持久性图的双连续表示
链接：https://arxiv.org/abs/2509.16999

【3】$\boldsymbolλ$-Orthogonality Regularization for Compatible Representation Learning
标题：$用于兼容表示学习的旧符号|$-共变性正规化
链接：https://arxiv.org/abs/2509.16664

【4】EMPEROR: Efficient Moment-Preserving Representation of Distributions
标题：皇帝：分布的有效动量保持表示
链接：https://arxiv.org/abs/2509.16379

3D|3D重建等相关(3篇)

【1】Learning and Optimization with 3D Orientations
标题：使用3D方向学习和优化
链接：https://arxiv.org/abs/2509.17274

【2】PMRT: A Training Recipe for Fast, 3D High-Resolution Aerodynamic Prediction
标题：PMRT：快速、3D高分辨率空气动力学预测的训练食谱
链接：https://arxiv.org/abs/2509.17182

【3】LVADNet3D: A Deep Autoencoder for Reconstructing 3D Intraventricular Flow from Sparse Hemodynamic Data
标题：LVADNet 3D：一种用于从稀疏血流动力学数据重建3D心室内流量的深度自动编码器
链接：https://arxiv.org/abs/2509.16860

编码器(1篇)

【1】ViTCAE: ViT-based Class-conditioned Autoencoder
标题：ViTCAE：基于ViT的类别条件自动编码器
链接：https://arxiv.org/abs/2509.16554

优化|敛散性(9篇)

【1】GaussianPSL: A novel framework based on Gaussian Splatting for exploring the Pareto frontier in multi-criteria optimization
标题：GaussianPSL：一个基于高斯飞溅的新型框架，用于探索多准则优化中的帕累托前沿
链接：https://arxiv.org/abs/2509.17889

【2】Global Optimization via Softmin Energy Minimization
标题：通过Softmin能源最小化进行全球优化
链接：https://arxiv.org/abs/2509.17815

【3】Conditional Policy Generator for Dynamic Constraint Satisfaction and Optimization
标题：动态约束满足和优化的条件策略生成器
链接：https://arxiv.org/abs/2509.17205

【4】GRPOformer: Advancing Hyperparameter Optimization via Group Relative Policy Optimization
标题：GRPOformer：通过群体相对政策优化推进超参数优化
链接：https://arxiv.org/abs/2509.17105

【5】Enhancing Performance and Calibration in Quantile Hyperparameter Optimization
标题：增强分位数超参数优化中的性能和校准
链接：https://arxiv.org/abs/2509.17051

【6】NeuFACO: Neural Focused Ant Colony Optimization for Traveling Salesman Problem
标题：NeuFACO：旅行推销员问题的神经聚焦蚁群优化
链接：https://arxiv.org/abs/2509.16938

【7】Near-Optimal Sample Complexity Bounds for Constrained Average-Reward MDPs
标题：受约束平均回报MDP的近最优样本复杂性界
链接：https://arxiv.org/abs/2509.16586

【8】Guided Sequence-Structure Generative Modeling for Iterative Antibody Optimization
标题：迭代抗体优化的引导序列结构生成建模
链接：https://arxiv.org/abs/2509.16357

【9】ROOT: Rethinking Offline Optimization as Distributional Translation via Probabilistic Bridge
标题：ROOT：通过概率桥将离线优化重新思考为分布式翻译
链接：https://arxiv.org/abs/2509.16300

预测|估计(15篇)

【1】Efficient & Correct Predictive Equivalence for Decision Trees
标题：决策树的高效且正确的预测等效
链接：https://arxiv.org/abs/2509.17774

【2】An AutoML Framework using AutoGluonTS for Forecasting Seasonal Extreme Temperatures
标题：使用AutoGluonTS预测季节性极端气温的AutoML框架
链接：https://arxiv.org/abs/2509.17734

【3】SPRINT: Stochastic Performative Prediction With Variance Reduction
标题：SPRint：具有方差缩减的随机表演预测
链接：https://arxiv.org/abs/2509.17304

【4】On the Simplification of Neural Network Architectures for Predictive Process Monitoring
标题：关于预测过程监控的神经网络架构的简化
链接：https://arxiv.org/abs/2509.17145

【5】Ultra-short-term solar power forecasting by deep learning and data reconstruction
标题：通过深度学习和数据重建进行超短期太阳能发电预测
链接：https://arxiv.org/abs/2509.17095

【6】TSGym: Design Choices for Deep Multivariate Time-Series Forecasting
标题：TSGym：深度多元时间序列预测的设计选择
链接：https://arxiv.org/abs/2509.17063

【7】A Hybrid PCA-PR-Seq2Seq-Adam-LSTM Framework for Time-Series Power Outage Prediction
标题：用于时间序列停电预测的混合PCA-PR-Seq 2Seq-Adam-LSTM框架
链接：https://arxiv.org/abs/2509.16743

【8】FairTune: A Bias-Aware Fine-Tuning Framework Towards Fair Heart Rate Prediction from PPG
标题：FairTune：一个基于PPG的公平心率预测的偏差感知微调框架
链接：https://arxiv.org/abs/2509.16491

【9】Predicting First Year Dropout from Pre Enrolment Motivation Statements Using Text Mining
标题：使用文本挖掘预测入学前动机陈述的第一年辍学率
链接：https://arxiv.org/abs/2509.16224

【10】FastNet: Improving the physical consistency of machine-learning weather prediction models through loss function design
标题：FastNet：通过损失函数设计提高机器学习天气预测模型的物理一致性
链接：https://arxiv.org/abs/2509.17601

【11】Bias-variance Tradeoff in Tensor Estimation
标题：张量估计中的偏方差权衡
链接：https://arxiv.org/abs/2509.17382

【12】AI-based Methods for Simulating, Sampling, and Predicting Protein Ensembles
标题：基于人工智能的蛋白质集成模拟、采样和预测方法
链接：https://arxiv.org/abs/2509.17224

【13】DeepEOSNet: Capturing the dependency on thermodynamic state in property prediction tasks
标题：DeepEOSNet：捕捉属性预测任务中对热力学状态的依赖性
链接：https://arxiv.org/abs/2509.17018

【14】A Study on Stabilizer Rényi Entropy Estimation using Machine Learning
标题：基于机器学习的稳定器Rényi信息量估计研究
链接：https://arxiv.org/abs/2509.16799

【15】Vibrational Fingerprints of Strained Polymers: A Spectroscopic Pathway to Mechanical State Prediction
标题：应变聚合物的振动指纹：机械状态预测的光谱途径
链接：https://arxiv.org/abs/2509.16266

其他神经网络|深度学习|模型|建模(37篇)

【1】Learning functions, operators and dynamical systems with kernels
标题：学习函数、算子与核动力系统
链接：https://arxiv.org/abs/2509.18071

【2】Learning to Rank with Top-$K$ Fairness
标题：学习与顶级排名-$K$公平
链接：https://arxiv.org/abs/2509.18067

【3】Prepare Before You Act: Learning From Humans to Rearrange Initial States
标题：行动前准备：向人类学习以重新安排初始状态
链接：https://arxiv.org/abs/2509.18043

【4】StefaLand: An Efficient Geoscience Foundation Model That Improves Dynamic Land-Surface Predictions
标题：StefaLand：一种有效的地球科学基金会模型，可改善动态陆地表面预测
链接：https://arxiv.org/abs/2509.17942

【5】SingLEM: Single-Channel Large EEG Model
标题：SingLEM：单通道大型脑电模型
链接：https://arxiv.org/abs/2509.17920

【6】Lipschitz-Based Robustness Certification for Recurrent Neural Networks via Convex Relaxation
标题：基于Lipschitz的凸松弛回归神经网络鲁棒性证明
链接：https://arxiv.org/abs/2509.17898

【7】Confidence-gated training for efficient early-exit neural networks
标题：高效提前退出神经网络的置信门控训练
链接：https://arxiv.org/abs/2509.17885

【8】Remote Sensing-Oriented World Model
标题：面向遥感的世界模型
链接：https://arxiv.org/abs/2509.17808

【9】Learning Neural Antiderivatives
标题：学习神经反衍生物
链接：https://arxiv.org/abs/2509.17755

【10】An Unlearning Framework for Continual Learning
标题：持续学习的放弃学习框架
链接：https://arxiv.org/abs/2509.17530

【11】Distributionally Robust Safety Verification of Neural Networks via Worst-Case CVaR
标题：通过最坏情况CVaR进行神经网络的分布式鲁棒安全验证
链接：https://arxiv.org/abs/2509.17413

【12】DiffQ: Unified Parameter Initialization for Variational Quantum Algorithms via Diffusion Models
标题：迪夫Q：通过扩散模型的变分量子算法的统一参数分配器
链接：https://arxiv.org/abs/2509.17324

【13】Physics-Informed Operator Learning for Hemodynamic Modeling
标题：血流动力学建模的知情操作员学习
链接：https://arxiv.org/abs/2509.17293

【14】Dendritic Resonate-and-Fire Neuron for Effective and Efficient Long Sequence Modeling
标题：用于长序列建模的树状共振激发神经元
链接：https://arxiv.org/abs/2509.17186

【15】Data-efficient Kernel Methods for Learning Hamiltonian Systems
标题：学习汉密尔顿系统的数据高效核心方法
链接：https://arxiv.org/abs/2509.17154

【16】DocIQ: A Benchmark Dataset and Feature Fusion Network for Document Image Quality Assessment
标题：DocIQ：用于文档图像质量评估的基准数据集和特征融合网络
链接：https://arxiv.org/abs/2509.17012

【17】The Complexity of Finding Local Optima in Contrastive Learning
标题 ：对比学习中寻找局部最优值的复杂性
链接：https://arxiv.org/abs/2509.16898

【18】Robot Learning with Sparsity and Scarcity
标题：机器人学习的稀缺性和稀缺性
链接：https://arxiv.org/abs/2509.16834

【19】On the System Theoretic Offline Learning of Continuous-Time LQR with Exogenous Disturbances
标题：含外生干扰的连续时间LQR的系统论离线学习
链接：https://arxiv.org/abs/2509.16746

【20】Min: Mixture of Noise for Pre-Trained Model-Based Class-Incremental Learning
标题：Min：预训练的基于模型的类增量学习的噪声混合
链接：https://arxiv.org/abs/2509.16738

【21】Safe Guaranteed Dynamics Exploration with Probabilistic Models
标题：使用概率模型进行安全保证的动态探索
链接：https://arxiv.org/abs/2509.16650

【22】ORN-CBF: Learning Observation-conditioned Residual Neural Control Barrier Functions via Hypernetworks
标题：ORN-CBF：通过超网络学习观测条件残差神经控制障碍函数
链接：https://arxiv.org/abs/2509.16614

【23】Checking extracted rules in Neural Networks
标题：检查神经网络中提取的规则
链接：https://arxiv.org/abs/2509.16547

【24】A Closer Look at Model Collapse: From a Generalization-to-Memorization Perspective
标题：仔细观察模型崩溃：从概括到简化的角度
链接：https://arxiv.org/abs/2509.16499

【25】Synergies between Federated Foundation Models and Smart Power Grids
标题：联邦基金会模型和智能电网之间的协同作用
链接：https://arxiv.org/abs/2509.16496

【26】CoUn: Empowering Machine Unlearning via Contrastive Learning
标题 ：CoUn：通过对比学习增强机器反学习能力
链接：https://arxiv.org/abs/2509.16391

【27】Improving Deep Tabular Learning
标题：改进深度表格学习
链接：https://arxiv.org/abs/2509.16354

【28】Architectural change in neural networks using fuzzy vertex pooling
标题：使用模糊点池的神经网络架构改变
链接：https://arxiv.org/abs/2509.16287

【29】Evaluation of Ensemble Learning Techniques for handwritten OCR Improvement
标题：手写OCR改进的集成学习技术评估
链接：https://arxiv.org/abs/2509.16221

【30】Discovering Software Parallelization Points Using Deep Neural Networks
标题：使用深度神经网络发现软件并行化点
链接：https://arxiv.org/abs/2509.16215

【31】Functional effects models: Accounting for preference heterogeneity in panel data with machine learning
标题：功能效应模型：用机器学习解释面板数据中的偏好差异
链接：https://arxiv.org/abs/2509.18047

【32】Deep Learning as the Disciplined Construction of Tame Objects
标题：深度学习作为驯服对象的精确构建
链接：https://arxiv.org/abs/2509.18025

【33】Random functions as data compressors for machine learning of molecular processes
标题：随机函数作为分子过程机器学习的数据压缩器
链接：https://arxiv.org/abs/2509.17937

【34】Robust Mixture Models for Algorithmic Fairness Under Latent Heterogeneity
标题：潜在异方差下数学公平性的鲁棒混合模型
链接：https://arxiv.org/abs/2509.17411

【35】DoubleGen: Debiased Generative Modeling of Counterfactuals
标题：DoubleGen：反事实的无偏生成建模
链接：https://arxiv.org/abs/2509.16842

【36】QASTAnet: A DNN-based Quality Metric for Spatial Audio
标题：QASTAnet：基于DNN的空间音频质量指标
链接：https://arxiv.org/abs/2509.16715

【37】Machine Learning for Quantum Noise Reduction
标题：用于降低量子噪音的机器学习
链接：https://arxiv.org/abs/2509.16242

其他(54篇)

【1】SEQR: Secure and Efficient QR-based LoRA Routing
标题：SEQR：安全有效的基于QR的LoRA路由
链接：https://arxiv.org/abs/2509.18093

【2】Control Disturbance Rejection in Neural ODEs
标题：神经ODE中的控制干扰抑制
链接：https://arxiv.org/abs/2509.18034

【3】Unveiling m-Sharpness Through the Structure of Stochastic Gradient Noise
标题：通过随机梯度噪音的结构揭示m-夏普
链接：https://arxiv.org/abs/2509.18001

【4】The Narcissus Hypothesis:Descending to the Rung of Illusion
标题：水仙假说：下降到幻觉的边缘
链接：https://arxiv.org/abs/2509.17999

【5】Equilibrium flow: From Snapshots to Dynamics
标题：平衡流：从快照到动态
链接：https://arxiv.org/abs/2509.17990

【6】Towards Seeing Bones at Radio Frequency
标题：走向用无线电频率看到骨头
链接：https://arxiv.org/abs/2509.17979

【7】ComposableNav: Instruction-Following Navigation in Dynamic Environments via Composable Diffusion
标题：ComposableNav：通过可组合扩散在动态环境中遵循指令的导航
链接：https://arxiv.org/abs/2509.17941

【8】Elucidating the Design Space of FP4 training
标题：阐明FP 4训练的设计空间
链接：https://arxiv.org/abs/2509.17791

【9】GEM-T: Generative Tabular Data via Fitting Moments
标题：GEM-T：通过匹配矩生成表格数据
链接：https://arxiv.org/abs/2509.17752

【10】Flatness is Necessary, Neural Collapse is Not: Rethinking Generalization via Grokking
标题：平坦是必要的，神经崩溃不是：通过Grokking重新思考概括
链接：https://arxiv.org/abs/2509.17738

【11】MontePrep: Monte-Carlo-Driven Automatic Data Preparation without Target Data Instances
标题：MontePrep：Monte-Carlo驱动的自动数据准备，无需目标数据预处理
链接：https://arxiv.org/abs/2509.17553

【12】Achilles' Heel of Mamba: Essential difficulties of the Mamba architecture demonstrated by synthetic data
标题：曼巴的阿喀琉斯之踵：合成数据证明曼巴架构的基本困难
链接：https://arxiv.org/abs/2509.17514

【13】AI Pangaea: Unifying Intelligence Islands for Adapting Myriad Tasks
标题：人工智能盘古大陆：统一情报岛以适应无数任务
链接：https://arxiv.org/abs/2509.17460

【14】Explainability matters: The effect of liability rules on the healthcare sector
标题：解释性很重要：责任规则对医疗保健行业的影响
链接：https://arxiv.org/abs/2509.17334

【15】Generalizable End-to-End Tool-Use RL with Synthetic CodeGym
标题：带合成代码Gym的可扩展端到端工具使用RL
链接：https://arxiv.org/abs/2509.17325

【16】VQEzy: An Open-Source Dataset for Parameter Initialize in Variational Quantum Eigensolvers
标题：VQEzy：一个用于在变分量子特征解算器中初始化参数的开源数据集
链接：https://arxiv.org/abs/2509.17322

【17】TraceHiding: Scalable Machine Unlearning for Mobility Data
标题：TraceHiding：移动数据的可扩展机器去学习
链接：https://arxiv.org/abs/2509.17241

【18】Virtual Consistency for Audio Editing
标题：音频编辑的虚拟一致性
链接：https://arxiv.org/abs/2509.17219

【19】Flow-Induced Diagonal Gaussian Processes
标题：流动诱导对角高斯过程
链接：https://arxiv.org/abs/2509.17153

【20】Delay compensation of multi-input distinct delay nonlinear systems via neural operators
标题：多输入非线性时滞系统的神经网络时滞补偿
链接：https://arxiv.org/abs/2509.17131

【21】On the Limits of Tabular Hardness Metrics for Deep RL: A Study with the Pharos Benchmark
标题：深RL的平板硬度限值：Pharos基准研究
链接：https://arxiv.org/abs/2509.17092

【22】Localizing Malicious Outputs from CodeLLM
标题：从CodeLLM本地化恶意输出
链接：https://arxiv.org/abs/2509.17070

【23】Equip Pre-ranking with Target Attention by Residual Quantization
标题：通过剩余量化为预排名配备目标注意力
链接：https://arxiv.org/abs/2509.16931

【24】Auditability and the Landscape of Distance to Multicalibration
标题：可审计性和与多元校准的距离格局
链接：https://arxiv.org/abs/2509.16930

【25】Cross-Attention with Confidence Weighting for Multi-Channel Audio Alignment
标题：多通道音频对齐的交叉注意和置信加权
链接：https://arxiv.org/abs/2509.16926

【26】Towards Interpretable and Efficient Attention: Compressing All by Contracting a Few
标题：迈向可解释和高效的注意力：通过减少一些来压缩所有内容
链接：https://arxiv.org/abs/2509.16875

【27】PhysHDR: When Lighting Meets Materials and Scene Geometry in HDR Reconstruction
标题：PhysHDR：当光线在HDR重建中遇到材料和场景几何时
链接：https://arxiv.org/abs/2509.16869

【28】ShadowServe: Interference-Free KV Cache Fetching for Distributed Prefix Caching
标题：ShadowServe：无干扰的KV缓存提取，用于分布式前置缓存
链接：https://arxiv.org/abs/2509.16857

【29】SOLAR: Switchable Output Layer for Accuracy and Robustness in Once-for-All Training
标题：SORAL：可切换输出层，在一次性训练中实现准确性和稳健性
链接：https://arxiv.org/abs/2509.16833

【30】KANO: Kolmogorov-Arnold Neural Operator
标题：KANO：Kolmogorov-Arnold神经运算符
链接：https://arxiv.org/abs/2509.16825

【31】Sublinear Time Quantum Sensitivity Sampling
标题：次线性时间量子灵敏度采样
链接：https://arxiv.org/abs/2509.16801

【32】Angular Dispersion Accelerates $k$-Nearest Neighbors Machine Translation
标题：角分散加速$k$-最近邻机器翻译
链接：https://arxiv.org/abs/2509.16729

【33】Segment-to-Act: Label-Noise-Robust Action-Prompted Video Segmentation Towards Embodied Intelligence
标题：从分段到动作：标签噪音稳健的嵌入式视频分段，迈向嵌入式智能
链接：https://arxiv.org/abs/2509.16677

【34】On the de-duplication of the Lakh MIDI dataset
标题：关于Lakh Buttons数据集的重复数据消除
链接：https://arxiv.org/abs/2509.16662

【35】Causal Fuzzing for Verifying Machine Unlearning
标题：机器遗忘的因果模糊
链接：https://arxiv.org/abs/2509.16525

【36】Revisiting Broken Windows Theory
标题：重新审视破窗理论
链接：https://arxiv.org/abs/2509.16490

【37】Local Mechanisms of Compositional Generalization in Conditional Diffusion
标题：条件扩散中成分概括的局部机制
链接：https://arxiv.org/abs/2509.16447

【38】End-to-end RL Improves Dexterous Grasping Policies
标题：端到端RL改进灵巧抓取政策
链接：https://arxiv.org/abs/2509.16434

【39】Hierarchical Retrieval: The Geometry and a Pretrain-Finetune Recipe
标题：分层检索：几何学和预微调食谱
链接：https://arxiv.org/abs/2509.16411

【40】Dynamic Objects Relocalization in Changing Environments with Flow Matching
标题：使用流匹配在变化环境中动态对象重新定位
链接：https://arxiv.org/abs/2509.16398

【41】Highly Imbalanced Regression with Tabular Data in SEP and Other Applications
标题：SDP和其他应用中使用表格数据的高度不平衡回归
链接：https://arxiv.org/abs/2509.16339

【42】SubDyve: Subgraph-Driven Dynamic Propagation for Virtual Screening Enhancement Controlling False Positive
标题：SubDyve：子图驱动的动态传播，用于虚拟屏幕增强控制假阳性
链接：https://arxiv.org/abs/2509.16273

【43】Core-elements Subsampling for Alternating Least Squares
标题：交替最小二乘的核元素二次抽样
链接：https://arxiv.org/abs/2509.18024

【44】Fréchet Geodesic Boosting
标题：弗雷谢特大地测量增强
链接：https://arxiv.org/abs/2509.18013

【45】RAVEN: RAnking and Validation of ExoplaNets
标题：RAVEN：ExoplaNets的运行和验证
链接：https://arxiv.org/abs/2509.17645

【46】Whitening Spherical Gaussian Mixtures in the Large-Dimensional Regime
标题：大维度区域中的球形高斯混合物白化
链接：https://arxiv.org/abs/2509.17636

【47】Bilateral Distribution Compression: Reducing Both Data Size and Dimensionality
标题：双边分布压缩：减少数据大小和特殊性
链接：https://arxiv.org/abs/2509.17543

【48】Risk Comparisons in Linear Regression: Implicit Regularization Dominates Explicit Regularization
标题：线性回归中的风险比较：隐性正规化主导显正规化
链接：https://arxiv.org/abs/2509.17251

【49】$\texttt{DiffSyn}$: A Generative Diffusion Approach to Materials Synthesis Planning
标题：$ extttt {DiffSyn}$：材料合成规划的生成扩散方法
链接：https://arxiv.org/abs/2509.17094

【50】Differential Privacy for Euclidean Jordan Algebra with Applications to Private Symmetric Cone Programming
标题：欧几里得乔丹代数的差异保密性及其在私有对称锥规划中的应用
链接：https://arxiv.org/abs/2509.16915

【51】Increase Alpha: Performance and Risk of an AI-Driven Trading Framework
标题：增加Alpha：人工智能驱动交易框架的性能和风险
链接：https://arxiv.org/abs/2509.16707

【52】Conditional Multidimensional Scaling with Incomplete Conditioning Data
标题：不完整条件数据的条件多维缩放
链接：https://arxiv.org/abs/2509.16627

【53】Similarity-Guided Diffusion for Long-Gap Music Inpainting
标题：相似性引导的长间隙音乐修复扩散
链接：https://arxiv.org/abs/2509.16342

【54】Motional representation; the ability to predict odor characters using molecular vibrations
标题：运动表示;使用分子振动预测气味特征的能力
链接：https://arxiv.org/abs/2509.16245

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