机器学习学术速递[8.19]

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217篇

大模型相关(29篇)

【1】MDPO: Overcoming the Training-Inference Divide of Masked Diffusion Language Models
标题：MDPO：克服掩蔽扩散语言模型的训练推理鸿沟
链接：https://arxiv.org/abs/2508.13148

【2】Signal and Noise: A Framework for Reducing Uncertainty in Language Model Evaluation
标题：信号与噪音：减少语言模型评估中不确定性的框架
链接：https://arxiv.org/abs/2508.13144

【3】OptimalThinkingBench: Evaluating Over and Underthinking in LLMs
标题：OptimalThinkingBench：评估法学硕士的过度思考和不足思考
链接：https://arxiv.org/abs/2508.13141

【4】Improving Detection of Watermarked Language Models
标题：改进加水印语言模型的检测
链接：https://arxiv.org/abs/2508.13131

【5】Learning to Steer: Input-dependent Steering for Multimodal LLMs
标题：学习转向：多模式LLM的输入依赖型转向
链接：https://arxiv.org/abs/2508.12815

【6】Bridging Human and LLM Judgments: Understanding and Narrowing the Gap
标题：弥合人类和LLM判断：理解并缩小差距
链接：https://arxiv.org/abs/2508.12792

【7】FedSODA: Federated Fine-tuning of LLMs via Similarity Group Pruning and Orchestrated Distillation Alignment
标题：FedSODA：通过相似性组修剪和启发式蒸馏对齐对LLM进行联邦微调
链接：https://arxiv.org/abs/2508.12727

【8】Energy-Efficient Wireless LLM Inference via Uncertainty and Importance-Aware Speculative Decoding
标题：通过不确定性和重要性感知推测解码的节能无线LLM推断
链接：https://arxiv.org/abs/2508.12590

【9】Deep Learning Model for Amyloidogenicity Prediction using a Pre-trained Protein LLM
标题：使用预训练蛋白质LLM进行淀粉样蛋白生成预测的深度学习模型
链接：https://arxiv.org/abs/2508.12575

【10】Illuminating LLM Coding Agents: Visual Analytics for Deeper Understanding and Enhancement
标题：启发LLM编码代理：视觉分析，以更深入的理解和增强
链接：https://arxiv.org/abs/2508.12555

【11】CorrSteer: Steering Improves Task Performance and Safety in LLMs through Correlation-based Sparse Autoencoder Feature Selection
标题：CorrSteer：Steering通过基于相关性的稀疏自动编码器特征选择提高LLM的任务性能和安全性
链接：https://arxiv.org/abs/2508.12535

【12】Rethinking Safety in LLM Fine-tuning: An Optimization Perspective
标题：重新思考LLM微调中的安全性：优化的角度
链接：https://arxiv.org/abs/2508.12531

【13】Mitigating Hallucinations in Large Language Models via Causal Reasoning
标题：通过因果推理缓解大型语言模型中的幻觉
链接：https://arxiv.org/abs/2508.12495

【14】Cost-Aware Contrastive Routing for LLMs
标题：LLM的成本意识对比路由
链接：https://arxiv.org/abs/2508.12491

【15】Uncovering Emergent Physics Representations Learned In-Context by Large Language Models
标题：揭示大型语言模型在上下文中学习的新兴物理表示
链接：https://arxiv.org/abs/2508.12448

【16】Unlearning at Scale: Implementing the Right to be Forgotten in Large Language Models
标题：大规模放弃学习：在大型语言模型中实现被遗忘的权利
链接：https://arxiv.org/abs/2508.12220

【17】ProtTeX-CC: Activating In-Context Learning in Protein LLM via Two-Stage Instruction Compression
标题：ProtTeX-CC：通过两阶段指令压缩激活蛋白质LLM中的上下文学习
链接：https://arxiv.org/abs/2508.12212

【18】STEM: Efficient Relative Capability Evaluation of LLMs through Structured Transition Samples
标题：STEM：通过结构化过渡样本对LLM进行有效的相对能力评估
链接：https://arxiv.org/abs/2508.12096

【19】J6: Jacobian-Driven Role Attribution for Multi-Objective Prompt Optimization in LLMs
标题：J6：LLM中多目标即时优化的Jacobian驱动角色归属
链接：https://arxiv.org/abs/2508.12086

【20】FutureX: An Advanced Live Benchmark for LLM Agents in Future Prediction
标题：FutureX：LLM代理未来预测的高级实时基准
链接：https://arxiv.org/abs/2508.11987

【21】CORE: Measuring Multi-Agent LLM Interaction Quality under Game-Theoretic Pressures
标题：CORE：在博弈论压力下测量多智能体LLM交互质量
链接：https://arxiv.org/abs/2508.11915

【22】EVTP-IVS: Effective Visual Token Pruning For Unifying Instruction Visual Segmentation In Multi-Modal Large Language Models
标题：EVTP-IVS：有效的视觉标记修剪，以统一多模式大型语言模型中的指令视觉分割
链接：https://arxiv.org/abs/2508.11886

【23】SupraTok: Cross-Boundary Tokenization for Enhanced Language Model Performance
标题：SupraTok：跨境代币化增强语言模型性能
链接：https://arxiv.org/abs/2508.11857

【24】Ontology-Guided Query Expansion for Biomedical Document Retrieval using Large Language Models
标题：使用大型语言模型进行生物医学文档检索的实体引导查询扩展
链接：https://arxiv.org/abs/2508.11784

【25】Enhancing GraphQL Security by Detecting Malicious Queries Using Large Language Models, Sentence Transformers, and Convolutional Neural Networks
标题：通过使用大型语言模型、句子转换器和卷积神经网络检测恶意收件箱来增强GraphQL安全性
链接：https://arxiv.org/abs/2508.11711

【26】Data-Driven Discovery of Interpretable Kalman Filter Variants through Large Language Models and Genetic Programming
标题：通过大型语言模型和遗传编程的数据驱动发现可解释的卡尔曼过滤器变体
链接：https://arxiv.org/abs/2508.11703

【27】Deep Language Geometry: Constructing a Metric Space from LLM Weights
标题：深度语言几何：从LLM权重构建度量空间
链接：https://arxiv.org/abs/2508.11676

【28】LLM-Based Intelligent Agents for Music Recommendation: A Comparison with Classical Content-Based Filtering
标题：基于LLM的音乐推荐智能代理：与经典基于内容的过滤的比较
链接：https://arxiv.org/abs/2508.11671

【29】Dropping Just a Handful of Preferences Can Change Top Large Language Model Rankings
标题：删除少量偏好就可以改变大型语言模型的顶级排名
链接：https://arxiv.org/abs/2508.11847

Graph相关(图学习|图神经网络|图优化等)(7篇)

【1】Predicting the Performance of Graph Convolutional Networks with Spectral Properties of the Graph Laplacian
标题：用图拉普拉斯算子的谱性质预测图卷积网络的性能
链接：https://arxiv.org/abs/2508.12993

【2】One-Class Intrusion Detection with Dynamic Graphs
标题：使用动态图的一类入侵检测
链接：https://arxiv.org/abs/2508.12885

【3】Defining and Benchmarking a Data-Centric Design Space for Brain Graph Construction
标题：定义和基准以数据为中心的设计空间以进行脑图构建
链接：https://arxiv.org/abs/2508.12533

【4】CRoC: Context Refactoring Contrast for Graph Anomaly Detection with Limited Supervision
标题：CRoC：有限监督下图形异常检测的上下文重构对比
链接：https://arxiv.org/abs/2508.12278

【5】On Balancing Sparsity with Reliable Connectivity in Distributed Network Design with Random K-out Graphs
标题：随机K-out图分布式网络设计中平衡稀疏性与可靠连接性
链接：https://arxiv.org/abs/2508.11863

【6】From Pixels to Graphs: Deep Graph-Level Anomaly Detection on Dermoscopic Images
标题：从像素到图形：皮肤镜图像的深度图形级异常检测
链接：https://arxiv.org/abs/2508.11826

【7】A Graph Neural Network based on a Functional Topology Model: Unveiling the Dynamic Mechanisms of Non-Suicidal Self-Injury in Single-Channel EEG
标题：基于功能布局模型的图神经网络：揭示单通道脑电中非自杀性自伤的动态机制
链接：https://arxiv.org/abs/2508.11684

Transformer(11篇)

【1】Causally-Guided Pairwise Transformer -- Towards Foundational Digital Twins in Process Industry
标题：因果引导的成对Transformer --面向过程工业的基础数字孪生
链接：https://arxiv.org/abs/2508.13111

【2】The Application of Transformer-Based Models for Predicting Consequences of Cyber Attacks
标题：基于转换器的模型在预测网络攻击后果中的应用
链接：https://arxiv.org/abs/2508.13030

【3】Learning In-context $\pmb{n}$-grams with Transformers: Sub-$\pmb{n}$-grams Are Near-stationary Points
链接：https://arxiv.org/abs/2508.12837

【4】Wavy Transformer
标题：波形Transformer
链接：https://arxiv.org/abs/2508.12787

【5】Online Ensemble Transformer for Accurate Cloud Workload Forecasting in Predictive Auto-Scaling
标题：在线注册Transformer，在预测性自动缩放中实现准确的云容量预测
链接：https://arxiv.org/abs/2508.12773

【6】MixCache: Mixture-of-Cache for Video Diffusion Transformer Acceleration
标题：Mixache：用于视频扩散Transformer加速的混合缓存
链接：https://arxiv.org/abs/2508.12691

【7】Bi-Axial Transformers: Addressing the Increasing Complexity of EHR Classification
标题：双向Transformer：解决EHR分类日益复杂的问题
链接：https://arxiv.org/abs/2508.12418

【8】Set-Valued Transformer Network for High-Emission Mobile Source Identification
标题：用于高排放移动源识别的集值Transformer网络
链接：https://arxiv.org/abs/2508.11976

【9】A Comprehensive Review of AI Agents: Transforming Possibilities in Technology and Beyond
标题：人工智能代理的全面回顾：改变技术及其他领域的可能性
链接：https://arxiv.org/abs/2508.11957

【10】Assessing Representation Stability for Transformer Models
标题：评估Transformer模型的表示稳定性
链接：https://arxiv.org/abs/2508.11667

【11】Arabic ASR on the SADA Large-Scale Arabic Speech Corpus with Transformer-Based Models
标题：具有基于转换器的模型的SADA大规模阿拉伯语语音库上的阿拉伯语ASB
链接：https://arxiv.org/abs/2508.12968

GAN|对抗|攻击|生成相关(13篇)

【1】Eyes on the Image: Gaze Supervised Multimodal Learning for Chest X-ray Diagnosis and Report Generation
标题：眼睛看着图像：用于胸部X射线诊断和报告生成的凝视监督多模式学习
链接：https://arxiv.org/abs/2508.13068

【2】Next Visual Granularity Generation
标题：下一个可视化粒度生成
链接：https://arxiv.org/abs/2508.12811

【3】BUILDA: A Thermal Building Data Generation Framework for Transfer Learning
标题：BUILDA：用于迁移学习的热力建筑数据生成框架
链接：https://arxiv.org/abs/2508.12703

【4】ToolACE-MT: Non-Autoregressive Generation for Agentic Multi-Turn Interaction
标题：工具ACE-MT：统计多圈交互的非自回归生成
链接：https://arxiv.org/abs/2508.12685

【5】Robust Federated Learning under Adversarial Attacks via Loss-Based Client Clustering
标题：通过基于损失的客户端集群实现对抗性攻击下的鲁棒联邦学习
链接：https://arxiv.org/abs/2508.12672

【6】Cognitive Structure Generation: From Educational Priors to Policy Optimization
标题：认知结构生成：从教育先验到政策优化
链接：https://arxiv.org/abs/2508.12647

【7】FlowMol3: Flow Matching for 3D De Novo Small-Molecule Generation
标题：FlowMol3：用于3D从头小分子生成的流动匹配
链接：https://arxiv.org/abs/2508.12629

【8】An Improved Algorithm for Adversarial Linear Contextual Bandits via Reduction
标题：一种基于约简的对抗线性上下文盗贼改进算法
链接：https://arxiv.org/abs/2508.11931

【9】ComplicitSplat: Downstream Models are Vulnerable to Blackbox Attacks by 3D Gaussian Splat Camouflages
标题：CompicitSplat：下游模型容易受到3D高斯Splat Camemages的黑匣子攻击
链接：https://arxiv.org/abs/2508.11854

【10】FairTabGen: Unifying Counterfactual and Causal Fairness in Synthetic Tabular Data Generation
标题：FairTabGen：统一合成表格数据生成中的反事实公平性和因果公平性
链接：https://arxiv.org/abs/2508.11810

【11】Scalable Geospatial Data Generation Using AlphaEarth Foundations Model
标题：使用AlphaEarth基础模型的可扩展地理空间数据生成
链接：https://arxiv.org/abs/2508.11739

【12】Transfer Learning for Neutrino Scattering: Domain Adaptation with GANs
标题：中微子散射的转移学习：GAN的域适应
链接：https://arxiv.org/abs/2508.12987

【13】Adversarial Robustness in Distributed Quantum Machine Learning
标题：分布式量子机器学习中的对抗鲁棒性
链接：https://arxiv.org/abs/2508.11848

半/弱/无/有监督|不确定性|主动学习(9篇)

【1】SNAP-UQ: Self-supervised Next-Activation Prediction for Single-Pass Uncertainty in TinyML
标题：SNAP-UQ：TinyML中单次不确定性的自我监督下一次激活预测
链接：https://arxiv.org/abs/2508.12907

【2】TCUQ: Single-Pass Uncertainty Quantification from Temporal Consistency with Streaming Conformal Calibration for TinyML
标题：TCUQ：TinyML流共形校准从时间一致性的单次不确定性量化
链接：https://arxiv.org/abs/2508.12905

【3】Multi-Level Knowledge Distillation and Dynamic Self-Supervised Learning for Continual Learning
标题：多层知识提炼和动态自我监督学习以实现持续学习
链接：https://arxiv.org/abs/2508.12692

【4】SSPO: Self-traced Step-wise Preference Optimization for Process Supervision and Reasoning Compression
标题：SSPO：流程监督和推理压缩的自追踪分步偏好优化
链接：https://arxiv.org/abs/2508.12604

【5】DE-VAE: Revealing Uncertainty in Parametric and Inverse Projections with Variational Autoencoders using Differential Entropy
标题：DE-VAE：使用差量的变分自动编码器揭示参数和逆投影中的不确定性
链接：https://arxiv.org/abs/2508.12145

【6】VARAN: Variational Inference for Self-Supervised Speech Models Fine-Tuning on Downstream Tasks
标题：VAR：自我监督语音模型的变分推理对下游任务进行微调
链接：https://arxiv.org/abs/2508.12061

【7】FedUHD: Unsupervised Federated Learning using Hyperdimensional Computing
标题：FedUHD：使用超维计算的无监督联邦学习
链接：https://arxiv.org/abs/2508.12021

【8】Revealing Neurocognitive and Behavioral Patterns by Unsupervised Manifold Learning from Dynamic Brain Data
标题：通过来自动态大脑数据的无监督Manifold学习揭示神经认知和行为模式
链接：https://arxiv.org/abs/2508.11672

【9】Unsupervised Pairwise Learning Optimization Framework for Cross-Corpus EEG-Based Emotion Recognition Based on Prototype Representation
标题：基于原型表示的跨数据库脑电情绪识别的无监督成对学习优化框架
链接：https://arxiv.org/abs/2508.11663

迁移|Zero/Few/One-Shot|自适应(9篇)

【1】Design and Analysis of Robust Adaptive Filtering with the Hyperbolic Tangent Exponential Kernel M-Estimator Function for Active Noise Control
标题：具有双曲切指数核M-估计函数的鲁棒自适应过滤的设计与分析
链接：https://arxiv.org/abs/2508.13018

【2】Fairness-Aware Multi-view Evidential Learning with Adaptive Prior
标题：具有自适应先验的公平意识的多视图证据学习
链接：https://arxiv.org/abs/2508.12997

【3】SL-ACC: A Communication-Efficient Split Learning Framework with Adaptive Channel-wise Compression
标题：SL-ACC：一个具有自适应逐行压缩的通信高效分离学习框架
链接：https://arxiv.org/abs/2508.12984

【4】TTA-DAME: Test-Time Adaptation with Domain Augmentation and Model Ensemble for Dynamic Driving Conditions
标题：TTA-DAME：针对动态驾驶条件的域扩展和模型集合的测试时间自适应
链接：https://arxiv.org/abs/2508.12690

【5】Adaptive Model-Predictive Control of a Soft Continuum Robot Using a Physics-Informed Neural Network Based on Cosserat Rod Theory
标题：基于Cosserat杆理论的物理信息神经网络的软连续体机器人自适应模型预测控制
链接：https://arxiv.org/abs/2508.12681

【6】Local Cluster Cardinality Estimation for Adaptive Mean Shift
标题：自适应均值漂移的局部簇基数估计
链接：https://arxiv.org/abs/2508.12450

【7】Synthetic Data is Sufficient for Zero-Shot Visual Generalization from Offline Data
标题：合成数据足以从离线数据进行Zero-Shot视觉概括
链接：https://arxiv.org/abs/2508.12356

【8】TSLA: A Task-Specific Learning Adaptation for Semantic Segmentation on Autonomous Vehicles Platform
标题：TSLA：自动驾驶汽车平台上用于语义分割的特定任务学习适应
链接：https://arxiv.org/abs/2508.12279

【9】Inductive transfer learning from regression to classification in ECG analysis
标题：心电图分析中从回归到分类的归纳迁移学习
链接：https://arxiv.org/abs/2508.11656

强化学习(6篇)

【1】OPTIC-ER: A Reinforcement Learning Framework for Real-Time Emergency Response and Equitable Resource Allocation in Underserved African Communities
标题：OPTIC-ER：服务不足的非洲社区实时应急响应和公平资源分配的强化学习框架
链接：https://arxiv.org/abs/2508.12943

【2】Reinforcement Learning with Rubric Anchors
标题：使用Rubric Interior的强化学习
链接：https://arxiv.org/abs/2508.12790

【3】OS-R1: Agentic Operating System Kernel Tuning with Reinforcement Learning
标题：OS-R1：使用强化学习进行显式操作系统核心调优
链接：https://arxiv.org/abs/2508.12551

【4】Results of the NeurIPS 2023 Neural MMO Competition on Multi-task Reinforcement Learning
标题：NeurIPS 2023年多任务强化学习神经MMO竞赛结果
链接：https://arxiv.org/abs/2508.12524

【5】Cold-RL: Learning Cache Eviction with Offline Reinforcement Learning for NGINX
标题：Cold-RL：通过NGINX的离线强化学习来学习缓存驱逐
链接：https://arxiv.org/abs/2508.12485

【6】Centralized Permutation Equivariant Policy for Cooperative Multi-Agent Reinforcement Learning
标题：协作多智能体强化学习的集中式排列等变策略
链接：https://arxiv.org/abs/2508.11706

元学习(1篇)

【1】Fed-Meta-Align: A Similarity-Aware Aggregation and Personalization Pipeline for Federated TinyML on Heterogeneous Data
标题：Fed-Meta-Align：一种用于在异类数据上联合TinyML的相似性感知聚合和个性化管道
链接：https://arxiv.org/abs/2508.11794

分层学习(1篇)

【1】A Hierarchical Surrogate Model for Efficient Multi-Task Parameter Learning in Closed-Loop Contro
标题：闭环控制中高效多任务参数学习的分层代理模型
链接：https://arxiv.org/abs/2508.12738

医学相关(6篇)

【1】AICRN: Attention-Integrated Convolutional Residual Network for Interpretable Electrocardiogram Analysis
标题：AICRN：用于可解释心电图分析的注意力集成卷积剩余网络
链接：https://arxiv.org/abs/2508.12162

【2】Generative Medical Event Models Improve with Scale
标题：生成性医疗事件模型随着规模的增加而改进
链接：https://arxiv.org/abs/2508.12104

【3】BRIEF: BRain-Inspired network connection search with Extensive temporal feature Fusion enhances disease classification
标题：简介：BRain启发的具有广泛时间特征的网络连接搜索融合增强了疾病分类
链接：https://arxiv.org/abs/2508.11732

【4】Contrastive Regularization over LoRA for Multimodal Biomedical Image Incremental Learning
标题：LoRA上的对比正规化用于多模式生物医学图像增量学习
链接：https://arxiv.org/abs/2508.11673

【5】Collaborative Learning-Enhanced Lightweight Models for Predicting Arterial Blood Pressure Waveform in a Large-scale Perioperative Dataset
标题：用于预测大规模围手术期数据集中动脉血压波动的协作学习增强轻量级模型
链接：https://arxiv.org/abs/2508.11669

【6】Explainable Deep Neural Network for Multimodal ECG Signals: Intermediate vs Late Fusion
标题：多峰心电图信号的可解释深度神经网络：中期与晚期融合
链接：https://arxiv.org/abs/2508.11666

推荐(2篇)

【1】Is This News Still Interesting to You?: Lifetime-aware Interest Matching for News Recommendation
标题：这个消息你还感兴趣吗？：终身感知的新闻推荐兴趣匹配
链接：https://arxiv.org/abs/2508.13064

【2】Leveraging Geometric Insights in Hyperbolic Triplet Loss for Improved Recommendations
标题：利用双曲三重损失中的几何见解改进建议
链接：https://arxiv.org/abs/2508.11978

聚类(1篇)

【1】Constrained Centroid Clustering: A Novel Approach for Compact and Structured Partitioning
标题：约束质心聚类：一种新的紧凑结构划分方法
链接：https://arxiv.org/abs/2508.12758

超分辨率|去噪|去模糊|去雾(1篇)

【1】Denoising diffusion models for inverse design of inflatable structures with programmable deformations
标题：可编程变形充气结构逆设计的降噪扩散模型
链接：https://arxiv.org/abs/2508.13097

自动驾驶|车辆|车道检测等(5篇)

【1】Argos: A Decentralized Federated System for Detection of Traffic Signs in CAVs
标题：Argos：一个用于检测卡韦交通标志的分散联邦系统
链接：https://arxiv.org/abs/2508.12712

【2】A Hybrid Surrogate for Electric Vehicle Parameter Estimation and Power Consumption via Physics-Informed Neural Operators
标题：通过物理信息神经运算符进行电动汽车参数估计和功耗的混合替代方案
链接：https://arxiv.org/abs/2508.12602

【3】Physics-informed deep operator network for traffic state estimation
标题：用于流量状态估计的物理信息深度运营商网络
链接：https://arxiv.org/abs/2508.12593

【4】Scale-Disentangled spatiotemporal Modeling for Long-term Traffic Emission Forecasting
标题：基于尺度分解的交通污染物长期预测时空模型
链接：https://arxiv.org/abs/2508.11923

【5】Lifelong Learner: Discovering Versatile Neural Solvers for Vehicle Routing Problems
标题：终身学习者：发现车辆路径问题的多功能神经求解器
链接：https://arxiv.org/abs/2508.11679

联邦学习|隐私保护|加密(5篇)

【1】Fed-DPRoC:Communication-Efficient Differentially Private and Robust Federated Learning
标题：Fed-DPRoC：通信高效的差分私有和鲁棒联邦学习
链接：https://arxiv.org/abs/2508.12978

【2】FedUNet: A Lightweight Additive U-Net Module for Federated Learning with Heterogeneous Models
标题：FedUNet：一个轻量级添加性U-Net模块，用于具有异类模型的联邦学习
链接：https://arxiv.org/abs/2508.12740

【3】Deploying Models to Non-participating Clients in Federated Learning without Fine-tuning: A Hypernetwork-based Approach
标题：在联邦学习中将模型部署到非参与客户端而无需微调：基于超网络的方法
链接：https://arxiv.org/abs/2508.12673

【4】Fairness Regularization in Federated Learning
标题：联邦学习中的公平正规化
链接：https://arxiv.org/abs/2508.12042

【5】SimQFL: A Quantum Federated Learning Simulator with Real-Time Visualization
标题：SimQFL：具有实时可视化的量子联邦学习模拟器
链接：https://arxiv.org/abs/2508.12477

推理|分析|理解|解释(20篇)

【1】Contrastive Representations for Temporal Reasoning
标题：时间推理的对比表示
链接：https://arxiv.org/abs/2508.13113

【2】Efficient and Verifiable Privacy-Preserving Convolutional Computation for CNN Inference with Untrusted Clouds
标题：具有不可信云的CNN推理的高效且可验证的保留隐私的卷积计算
链接：https://arxiv.org/abs/2508.12832

【3】Deep Semantic Inference over the Air: An Efficient Task-Oriented Communication System
标题：空中深度语义推理：一个高效的面向任务的通信系统
链接：https://arxiv.org/abs/2508.12748

【4】A Multi-Resolution Benchmark Framework for Spatial Reasoning Assessment in Neural Networks
标题：神经网络空间推理评估的多分辨率基准框架
链接：https://arxiv.org/abs/2508.12741

【5】How can we trust opaque systems? Criteria for robust explanations in XAI
标题：我们如何才能信任不透明的系统？XAI中稳健解释的标准
链接：https://arxiv.org/abs/2508.12623

【6】Trust Region Constrained Measure Transport in Path Space for Stochastic Optimal Control and Inference
标题：随机最优控制和推理的路径空间信任域约束度量传输
链接：https://arxiv.org/abs/2508.12511

【7】Root Cause Analysis of Hydrogen Bond Separation in Spatio-Temporal Molecular Dynamics using Causal Models
标题：利用因果模型分析时空分子动力学中的氢键分离的根本原因
链接：https://arxiv.org/abs/2508.12500

【8】Navigating the Exploration-Exploitation Tradeoff in Inference-Time Scaling of Diffusion Models
标题：扩散模型的推理时间标度中的勘探与开发权衡
链接：https://arxiv.org/abs/2508.12361

【9】Convergence Analysis of the Lion Optimizer in Centralized and Distributed Settings
标题：集中式和分布式环境下Lion优化器的收敛分析
链接：https://arxiv.org/abs/2508.12327

【10】Active inference for action-unaware agents
标题：无动作意识主体的主动推理
链接：https://arxiv.org/abs/2508.12027

【11】MOON: Generative MLLM-based Multimodal Representation Learning for E-commerce Product Understanding
标题：MOON：基于MLLM的生成式多模式表示学习，用于电子商务产品理解
链接：https://arxiv.org/abs/2508.11999

【12】Learning Marked Temporal Point Process Explanations based on Counterfactual and Factual Reasoning
标题：基于反事实和事实推理的标记时间点流程解释
链接：https://arxiv.org/abs/2508.11943

【13】Audio Flamingo Sound-CoT Technical Report: Improving Chain-of-Thought Reasoning in Sound Understanding
标题：音频火烈鸟Sound-CoT技术报告：改进声音理解中的思维链推理
链接：https://arxiv.org/abs/2508.11818

【14】Uncalibrated Reasoning: GRPO Induces Overconfidence for Stochastic Outcomes
标题：未经校准的推理：GRPO导致对随机结果的过度自信
链接：https://arxiv.org/abs/2508.11800

【15】Comparative Analysis of Time Series Foundation Models for Demographic Forecasting: Enhancing Predictive Accuracy in US Population Dynamics
标题：人口预测时间序列基础模型的比较分析：提高美国人口动态的预测准确性
链接：https://arxiv.org/abs/2508.11680

【16】Simulation-Based Inference: A Practical Guide
标题：基于模拟的推理：实用指南
链接：https://arxiv.org/abs/2508.12939

【17】The path to a goal: Understanding soccer possessions via path signatures
标题：通往目标的道路：通过路径签名了解足球财产
链接：https://arxiv.org/abs/2508.12930

【18】A Generalized Genetic Random Field Method for the Genetic Association Analysis of Sequencing Data
标题：测序数据遗传关联分析的广义遗传随机场方法
链接：https://arxiv.org/abs/2508.12617

【19】Exploring Multimodal AI Reasoning for Meteorological Forecasting from Skew-T Diagrams
标题：从斜T图探索气象预报的多模式人工智能推理
链接：https://arxiv.org/abs/2508.12198

【20】BaMANI: Bayesian Multi-Algorithm causal Network Inference
标题：BaMANI：Bayesian多算法因果网络推理
链接：https://arxiv.org/abs/2508.11741

检测相关(4篇)

【1】Outlier Detection of Poisson-Distributed Targets Using a Seabed Sensor Network
标题：利用海底传感器网络检测Poisson分布目标的离群点
链接：https://arxiv.org/abs/2508.13099

【2】Randomized PCA Forest for Outlier Detection
标题：用于离群值检测的随机PCA森林
链接：https://arxiv.org/abs/2508.12776

【3】Automated Model Evaluation for Object Detection via Prediction Consistency and Reliablity
标题：通过预测一致性和可靠性进行对象检测的自动模型评估
链接：https://arxiv.org/abs/2508.12082

【4】Track Component Failure Detection Using Data Analytics over existing STDS Track Circuit data
标题：在现有STDS轨道电路数据上使用数据分析来检测轨道部件故障
链接：https://arxiv.org/abs/2508.11693

分类|识别(7篇)

【1】Empirical Evidences for the Effects of Feature Diversity in Open Set Recognition and Continual Learning
标题：特征多样性在开放集识别和连续学习中影响的经验证据
链接：https://arxiv.org/abs/2508.13005

【2】A Sobel-Gradient MLP Baseline for Handwritten Character Recognition
标题：手写字符识别的Sobel梯度MLP基线
链接：https://arxiv.org/abs/2508.11902

【3】An MLP Baseline for Handwriting Recognition Using Planar Curvature and Gradient Orientation
标题：使用平面弯曲和梯度方向的手写识别MLP基线
链接：https://arxiv.org/abs/2508.11803

【4】On the Importance of Behavioral Nuances: Amplifying Non-Obvious Motor Noise Under True Empirical Considerations May Lead to Briefer Assays and Faster Classification Processes
标题：关于行为细微差别的重要性：在真正的经验考虑下放大不明显的运动噪音可能会导致更简短的分析和更快的分类过程
链接：https://arxiv.org/abs/2508.12742

【5】Towards Generalizable Human Activity Recognition: A Survey
标题：迈向可推广的人类活动识别：一项调查
链接：https://arxiv.org/abs/2508.12213

【6】Towards Generalizable Learning Models for EEG-Based Identification of Pain Perception
标题：基于脑电的疼痛感知识别的可推广学习模型
链接：https://arxiv.org/abs/2508.11691

【7】Energy-Efficient Real-Time 4-Stage Sleep Classification at 10-Second Resolution: A Comprehensive Study
标题：10秒分辨率的节能实时4阶段睡眠分类：一项全面研究
链接：https://arxiv.org/abs/2508.11664

表征(3篇)

【1】HRS: Hybrid Representation Framework with Scheduling Awareness for Time Series Forecasting in Crowdsourced Cloud-Edge Platforms
标题：HRA：具有众包云边缘平台中时间序列预测调度感知的混合表示框架
链接：https://arxiv.org/abs/2508.12839

【2】Bongard-RWR+: Real-World Representations of Fine-Grained Concepts in Bongard Problems
标题：Bongard-RWR+：Bongard问题中细粒度概念的真实世界表示
链接：https://arxiv.org/abs/2508.12026

【3】Unfolded Laplacian Spectral Embedding: A Theoretically Grounded Approach to Dynamic Network Representation
标题：展开拉普拉斯谱嵌入：动态网络表示的理论基础方法
链接：https://arxiv.org/abs/2508.12674

编码器(1篇)

【1】What Matters for Bioacoustic Encoding
标题：生物声学编码的重要性
链接：https://arxiv.org/abs/2508.11845

优化|敛散性(4篇)

【1】DIT: Dimension Reduction View on Optimal NFT Rarity Meters
标题：DID：最佳NFT稀有度计的降维观点
链接：https://arxiv.org/abs/2508.12671

【2】DynamixSFT: Dynamic Mixture Optimization of Instruction Tuning Collections
标题：DynamixSFT：指令调优集合的动态混合优化
链接：https://arxiv.org/abs/2508.12116

【3】Optimal Condition for Initialization Variance in Deep Neural Networks: An SGD Dynamics Perspective
标题：深度神经网络中RST方差的最佳条件：新元动力学角度
链接：https://arxiv.org/abs/2508.12834

【4】An Introduction to Sliced Optimal Transport
标题：分层最佳运输简介
链接：https://arxiv.org/abs/2508.12519

预测|估计(9篇)

【1】Hierarchical Evaluation Function (HEF): A Multi-Metric Approach for Optimizing Demand Forecasting Models
标题：分层评估函数（HEF）：优化需求预测模型的多指标方法
链接：https://arxiv.org/abs/2508.13057

【2】Short-Term Forecasting of Energy Production and Consumption Using Extreme Learning Machine: A Comprehensive MIMO based ELM Approach
标题：基于极限学习机的能源生产和消费短期预测--一种综合MIMO的ELM方法
链接：https://arxiv.org/abs/2508.12764

【3】Deep Learning-Based Financial Time Series Forecasting via Sliding Window and Variational Mode Decomposition
标题：基于深度学习的滑动窗口和变分模式分解金融时间序列预测
链接：https://arxiv.org/abs/2508.12565

【4】Machine Learning-Based Manufacturing Cost Prediction from 2D Engineering Drawings via Geometric Features
标题：通过几何特征从2D工程图中基于机器学习的制造成本预测
链接：https://arxiv.org/abs/2508.12440

【5】STM3: Mixture of Multiscale Mamba for Long-Term Spatio-Temporal Time-Series Prediction
标题：STM 3：用于长期时空时间序列预测的多尺度曼巴混合
链接：https://arxiv.org/abs/2508.12247

【6】CC-Time: Cross-Model and Cross-Modality Time Series Forecasting
标题：CC-Time：跨模型和跨模式时间序列预测
链接：https://arxiv.org/abs/2508.12235

【7】Extending Straight-Through Estimation for Robust Neural Networks on Analog CIM Hardware
标题：在模拟TIM硬件上扩展鲁棒神经网络的直通估计
链接：https://arxiv.org/abs/2508.11940

【8】BConformeR: A Conformer Based on Mutual Sampling for Unified Prediction of Continuous and Discontinuous Antibody Binding Sites
标题：BConformmeR：基于互采样的一致性模型，用于统一预测连续和不连续抗体结合位点
链接：https://arxiv.org/abs/2508.12029

【9】Age-Normalized HRV Features for Non-Invasive Glucose Prediction: A Pilot Sleep-Aware Machine Learning Study
标题：用于无创血糖预测的时间标准化心率变异特征：一项试点睡眠感知机器学习研究
链接：https://arxiv.org/abs/2508.11682

其他神经网络|深度学习|模型|建模(26篇)

【1】Training Machine Learning Models on Human Spatio-temporal Mobility Data: An Experimental Study [Experiment Paper]
标题：基于人类时空移动性数据训练机器学习模型：实验研究[实验论文]
链接：https://arxiv.org/abs/2508.13135

【2】Monte Carlo Functional Regularisation for Continual Learning
标题：用于连续学习的Monte Carlo函数正则化
链接：https://arxiv.org/abs/2508.13006

【3】Toward Storage-Aware Learning with Compressed Data An Empirical Exploratory Study on JPEG
标题：利用压缩数据实现感知学习JPEG的实证探索性研究
链接：https://arxiv.org/abs/2508.12833

【4】A Self-Ensemble Inspired Approach for Effective Training of Binary-Weight Spiking Neural Networks
标题：一种自我激励的有效训练二进制尖峰神经网络的方法
链接：https://arxiv.org/abs/2508.12609

【5】Constructing Invariant and Equivariant Operations by Symmetric Tensor Network
标题：用对称张量网络构造不变与等变运算
链接：https://arxiv.org/abs/2508.12596

【6】Widening the Network Mitigates the Impact of Data Heterogeneity on FedAvg
标题：拓宽网络可减轻数据异构对FedAvg的影响
链接：https://arxiv.org/abs/2508.12576

【7】The Yokai Learning Environment: Tracking Beliefs Over Space and Time
标题：横井学习环境：在空间和时间上追踪信念
链接：https://arxiv.org/abs/2508.12480

【8】CarelessWhisper: Turning Whisper into a Causal Streaming Model
标题：CarelessWhisper：将Whisper转变为因果流媒体模型
链接：https://arxiv.org/abs/2508.12301

【9】DHG-Bench: A Comprehensive Benchmark on Deep Hypergraph Learning
标题：DHG-Bench：深度超图学习的综合基准
链接：https://arxiv.org/abs/2508.12244

【10】Distribution Matching via Generalized Consistency Models
标题：通过广义一致性模型进行分布匹配
链接：https://arxiv.org/abs/2508.12222

【11】Time-Scale Coupling Between States and Parameters in Recurrent Neural Networks
标题：回归神经网络状态与参数之间的时间尺度耦合
链接：https://arxiv.org/abs/2508.12121

【12】Content Accuracy and Quality Aware Resource Allocation Based on LP-Guided DRL for ISAC-Driven AIGC Networks
标题：ISAC驱动的AIGC网络基于LP引导的DRL的内容准确性和质量感知资源分配
链接：https://arxiv.org/abs/2508.12079

【13】Universal Learning of Nonlinear Dynamics
标题：非线性动力学的通用学习
链接：https://arxiv.org/abs/2508.11990

【14】Efficient Modular Learning through Naive LoRA Summation: Leveraging Orthogonality in High-Dimensional Models
标题：通过朴素LoRA总和进行高效模块化学习：利用多维模型中的随机性
链接：https://arxiv.org/abs/2508.11985

【15】HPD: Hybrid Projection Decomposition for Robust State Space Models on Analog CIM Hardware
标题：DPD：模拟TIM硬件上稳健状态空间模型的混合投影分解
链接：https://arxiv.org/abs/2508.11935

【16】Reduced-order modeling of Hamiltonian dynamics based on symplectic neural networks
标题：基于辛神经网络的Hamilton动力学降维建模
链接：https://arxiv.org/abs/2508.11911

【17】PCA- and SVM-Grad-CAM for Convolutional Neural Networks: Closed-form Jacobian Expression
标题：卷积神经网络的PCA-和SVM-Grad-CAM：封闭式雅可比表达
链接：https://arxiv.org/abs/2508.11880

【18】Combinations of Fast Activation and Trigonometric Functions in Kolmogorov-Arnold Networks
标题：Kolmogorov-Arnold网络中快速激活函数与三角函数的组合
链接：https://arxiv.org/abs/2508.11876

【19】Limitation Learning: Catching Adverse Dialog with GAIL
标题：局限性学习：与GAIL进行负面对话
链接：https://arxiv.org/abs/2508.11767

【20】Causal Structure Learning in Hawkes Processes with Complex Latent Confounder Networks
标题：复杂潜在混杂网络Hawkes过程中的因果结构学习
链接：https://arxiv.org/abs/2508.11727

【21】From Heuristics to Data: Quantifying Site Planning Layout Indicators with Deep Learning and Multi-Modal Data
标题：从启发式到数据：利用深度学习和多模式数据量化场地规划布局指标
链接：https://arxiv.org/abs/2508.11723

【22】Toward Practical Equilibrium Propagation: Brain-inspired Recurrent Neural Network with Feedback Regulation and Residual Connections
标题：迈向实际平衡传播：具有反馈调节和剩余连接的脑启发回归神经网络
链接：https://arxiv.org/abs/2508.11659

【23】Enhancing Corrosion Resistance of Aluminum Alloys Through AI and ML Modeling
标题：通过AI和ML建模提高铝合金的耐腐蚀性
链接：https://arxiv.org/abs/2508.11685

【24】Robust Sparse Bayesian Learning Based on Minimum Error Entropy for Noisy High-Dimensional Brain Activity Decoding
标题：基于最小误差量的鲁棒稀疏Bayesian学习用于有噪的多维脑活动解码
链接：https://arxiv.org/abs/2508.11657

【25】HetSyn: Versatile Timescale Integration in Spiking Neural Networks via Heterogeneous Synapses
标题：HetSyn：通过异类突触在尖峰神经网络中进行多功能时间尺度集成
链接：https://arxiv.org/abs/2508.11644

【26】Vibe2Spike: Batteryless Wireless Tags for Vibration Sensing with Event Cameras and Spiking Networks
标题：Vibe 2 Spike：用于振动传感的无电池无线标签，使用活动摄像机和Spiking网络
链接：https://arxiv.org/abs/2508.11640

其他(37篇)

【1】Has GPT-5 Achieved Spatial Intelligence? An Empirical Study
标题：GPT-5实现了空间智能吗？实证研究
链接：https://arxiv.org/abs/2508.13142

【2】A Perfectly Truthful Calibration Measure
标题：完全真实的校准措施
链接：https://arxiv.org/abs/2508.13100

【3】Seeing the Many: Exploring Parameter Distributions Conditioned on Features in Surrogates
标题：多见：探索以代理人特征为条件的参数分布
链接：https://arxiv.org/abs/2508.13088

【4】Beyond Internal Data: Bounding and Estimating Fairness from Incomplete Data
标题：超越内部数据：从不完整数据中界定和估计公平性
链接：https://arxiv.org/abs/2508.13040

【5】Kourkoutas-Beta: A Sunspike-Driven Adam Optimizer with Desert Flair
标题：Kourkoutas-Beta：一个具有沙漠天赋的Sunspike驱动的Adam优化器
链接：https://arxiv.org/abs/2508.12996

【6】Fully Automated Segmentation of Fiber Bundles in Anatomic Tracing Data
标题：解剖追踪数据中纤维束的全自动分割
链接：https://arxiv.org/abs/2508.12942

【7】SparseMap: A Sparse Tensor Accelerator Framework Based on Evolution Strategy
标题：SparseMap：基于进化策略的稀疏张量加速器框架
链接：https://arxiv.org/abs/2508.12906

【8】CAMAR: Continuous Actions Multi-Agent Routing
标题：CAVAR：连续动作多代理路由
链接：https://arxiv.org/abs/2508.12845

【9】SIS-Challenge: Event-based Spatio-temporal Instance Segmentation Challenge at the CVPR 2025 Event-based Vision Workshop
标题：SIS挑战：CVPR 2025基于事件的视觉研讨会上的基于事件的时空实例分割挑战
链接：https://arxiv.org/abs/2508.12813

【10】Maximum Score Routing For Mixture-of-Experts
标题：混合专家的最大得分路由
链接：https://arxiv.org/abs/2508.12801

【11】A Shift in Perspective on Causality in Domain Generalization
标题：领域泛化中因果关系研究视角的转变
链接：https://arxiv.org/abs/2508.12798

【12】Unlearning Comparator: A Visual Analytics System for Comparative Evaluation of Machine Unlearning Methods
标题：取消学习比较器：用于比较评估机器取消学习方法的视觉分析系统
链接：https://arxiv.org/abs/2508.12730

【13】Score-informed Neural Operator for Enhancing Ordering-based Causal Discovery
标题：用于增强基于顺序的因果发现的分数知情神经运算符
链接：https://arxiv.org/abs/2508.12650

【14】Synthesizing Accurate and Realistic T1-weighted Contrast-Enhanced MR Images using Posterior-Mean Rectified Flow
标题：使用后验平均校正流合成精确和真实的T1加权对比增强MR图像
链接：https://arxiv.org/abs/2508.12640

【15】FLARE: Fast Low-rank Attention Routing Engine
标题：DART：快速低级别注意力路由引擎
链接：https://arxiv.org/abs/2508.12594

【16】Data-driven particle dynamics: Structure-preserving coarse-graining for emergent behavior in non-equilibrium systems
标题：数据驱动的粒子动力学：非平衡系统中涌现行为的结构保持粗粒化
链接：https://arxiv.org/abs/2508.12569

【17】Data-driven Trust Bootstrapping for Mobile Edge Computing-based Industrial IoT Services
标题：基于移动边缘计算的工业物联网服务的数据驱动信任引导
链接：https://arxiv.org/abs/2508.12560

【18】Toward Architecture-Agnostic Local Control of Posterior Collapse in VAEs
标题：走向建筑-VAE后部塌陷的不可知局部控制
链接：https://arxiv.org/abs/2508.12530

【19】Inverse-LLaVA: Eliminating Alignment Pre-training Through Text-to-Vision Mapping
标题：Inverse-LLaVA：通过文本到视觉映射消除对齐预训练
链接：https://arxiv.org/abs/2508.12466

【20】L-SR1: Learned Symmetric-Rank-One Preconditioning
标题：L-SR 1：习得的对称一级预处理
链接：https://arxiv.org/abs/2508.12270

【21】Interpreting Time Series Forecasts with LIME and SHAP: A Case Study on the Air Passengers Dataset
标题：用LIME和SHAP解释时间序列预测：航空乘客数据集的案例研究
链接：https://arxiv.org/abs/2508.12253

【22】Communication-Efficient Distributed Asynchronous ADMM
标题：通信高效的分布式同步ADMM
链接：https://arxiv.org/abs/2508.12233

【23】Belief-Conditioned One-Step Diffusion: Real-Time Trajectory Planning with Just-Enough Sensing
标题：信念条件一步扩散：具有恰到好处感知的实时轨迹规划
链接：https://arxiv.org/abs/2508.12166

【24】RealTalk: Realistic Emotion-Aware Lifelike Talking-Head Synthesis
标题：RealTalk：现实的描述感知逼真的说话头合成
链接：https://arxiv.org/abs/2508.12163

【25】Optimizing Neural Architectures for Hindi Speech Separation and Enhancement in Noisy Environments
标题：优化神经架构以实现高噪环境中印地语语音分离和增强
链接：https://arxiv.org/abs/2508.12009

【26】M3OOD: Automatic Selection of Multimodal OOD Detectors
标题：M3 OOD：自动选择多模式OOD检测器
链接：https://arxiv.org/abs/2508.11936

【27】Optimizing Token Choice for Code Watermarking: A RL Approach
标题：代码水印中标记选择的优化：一种强化学习方法
链接：https://arxiv.org/abs/2508.11925

【28】ENA: Efficient N-dimensional Attention
标题：ENA：高效的N维注意力
链接：https://arxiv.org/abs/2508.11921

【29】Singing Syllabi with Virtual Avatars: Enhancing Student Engagement Through AI-Generated Music and Digital Embodiment
标题：用虚拟化身演唱教学大纲：通过人工智能生成的音乐和数字体现增强学生参与度
链接：https://arxiv.org/abs/2508.11872

【30】Ovis2.5 Technical Report
标题：Ovis 2.5技术报告
链接：https://arxiv.org/abs/2508.11737

【31】Sparse Attention across Multiple-context KV Cache
标题：跨多上下文KV缓存的稀疏注意力
链接：https://arxiv.org/abs/2508.11661

【32】Shapley Values: Paired-Sampling Approximations
标题：沙普利值：配对抽样逼近
链接：https://arxiv.org/abs/2508.12947

【33】Towards SISO Bistatic Sensing for ISAC
标题：面向ISAC的SISO双站感知
链接：https://arxiv.org/abs/2508.12614

【34】Quantum Flow Matching
标题：量子流匹配
链接：https://arxiv.org/abs/2508.12413

【35】ATLAS: AI-Native Receiver Test-and-Measurement by Leveraging AI-Guided Search
标题：ATLAS：利用人工智能引导搜索进行人工智能原生接收器测试和测量
链接：https://arxiv.org/abs/2508.12204

【36】Robust Data Fusion via Subsampling
标题：通过二次抽样进行稳健的数据融合
链接：https://arxiv.org/abs/2508.12048

【37】Tightening the mixed integer linear formulation for the piecewise linear approximation in general dimensions
标题：一般维数下分段线性逼近的混合整数线性形式的紧化
链接：https://arxiv.org/abs/2508.09395

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