Py学习  »  机器学习算法

机器学习学术速递[3.30]

arXiv每日学术速递 • 1 月前 • 187 次点击  

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123篇


大模型相关(9篇)

【1】ALBA: A European Portuguese Benchmark for Evaluating Language and Linguistic Dimensions in Generative LLMs
标题:ALBA:生成式法学硕士中评估语言和语言维度的欧洲葡萄牙语基准
链接:https://arxiv.org/abs/2603.26516

作者:Inês Vieira, Inês Calvo, Iago Paulo, James Furtado, Rafael Ferreira, Diogo Tavares, Diogo Glória-Silva, David Semedo, João Magalhães
备注:PROPOR 2026 - The 17th International Conference on Computational Processing of Portuguese
摘要
摘要


【2】AMALIA Technical Report: A Fully Open Source Large Language Model for European Portuguese
标题:AMALIA技术报告:欧洲葡萄牙语的完全开源大型语言模型
链接:https://arxiv.org/abs/2603.26511

作者:Afonso Simplício, Gonçalo Vinagre, Miguel Moura Ramos, Diogo Tavares, Rafael Ferreira, Giuseppe Attanasio, Duarte M. Alves, Inês Calvo, Inês Vieira, Rui Guerra, James Furtado, Beatriz Canaverde, Iago Paulo, Vasco Ramos, Diogo Glória-Silva, Miguel Faria, Marcos Treviso, Daniel Gomes, Pedro Gomes, David Semedo, André Martins, João Magalhães
备注:PROPOR 2026 - The 17th International Conference on Computational Processing of Portuguese
摘要
摘要


【3】A Formal Framework for Uncertainty Analysis of Text Generation with Large Language Models
标题:大语言模型文本生成不确定性分析的形式化框架
链接:https://arxiv.org/abs/2603.26363

作者:Steffen Herbold, Florian Lemmerich
摘要
摘要


【4】Distilling Conversations: Abstract Compression of Conversational Audio Context for LLM-based ASR
标题:提炼对话:基于LLM的ASB的对话音频上下文的抽象压缩
链接:https://arxiv.org/abs/2603.26246

作者:Shashi Kumar, Esaú Villatoro-Tello, Sergio Burdisso, Kadri Hacioglu, Thibault Bañeras-Roux, Hasindri Watawana, Dairazalia Sanchez-Cortes, Srikanth Madikeri, Petr Motlicek, Andreas Stolcke
备注:11 pages
摘要
摘要


【5】DataFlex: A Unified Framework for Data-Centric Dynamic Training of Large Language Models
标题:DataFlex:大型语言模型以数据为中心的动态训练的统一框架
链接:https://arxiv.org/abs/2603.26164

作者:Hao Liang, Zhengyang Zhao, Meiyi Qiang, Mingrui Chen, Lu Ma, Rongyi Yu, Hengyi Feng, Shixuan Sun, Zimo Meng, Xiaochen Ma, Xuanlin Yang, Qifeng Cai, Ruichuan An, Bohan Zeng, Zhen Hao Wong, Chengyu Shen, Runming He, Zhaoyang Han, Yaowei Zheng, Fangcheng Fu, Conghui He, Bin Cui, Zhiyu Li, Weinan E, Wentao Zhang
摘要
摘要


【6】Are LLM-Enhanced Graph Neural Networks Robust against Poisoning Attacks?
标题:LLM增强型图神经网络对中毒攻击是否稳健?
链接:https://arxiv.org/abs/2603.26105

作者:Yuhang Ma, Jie Wang, Zheng Yan
备注:To appear at 2026 IEEE Symposium on Security and Privacy (SP)
摘要
摘要


【7】Selective Deficits in LLM Mental Self-Modeling in a Behavior-Based Test of Theory of Mind
标题:基于行为的心理理论测试中LLM心理自我建模中的选择性缺陷
链接:https://arxiv.org/abs/2603.26089

作者:Christopher Ackerman
备注:22 pages, 13 figures, 1 table
摘要
摘要


【8】H-Node Attack and Defense in Large Language Models
标题:大型语言模型中的H节点攻击和防御
链接:https://arxiv.org/abs/2603.26045

作者:Eric Yocam, Varghese Vaidyan, Yong Wang
备注:17 pages, 7 figures, 6 tables
摘要
摘要


【9】In-Context Molecular Property Prediction with LLMs: A Blinding Study on Memorization and Knowledge Conflicts
标题:基于LLM的分子性质预测:知识冲突与知识模糊的盲法研究
链接:https://arxiv.org/abs/2603.25857

作者:Matthias Busch, Marius Tacke, Sviatlana V. Lamaka, Mikhail L. Zheludkevich, Christian J. Cyron, Christian Feiler, Roland C. Aydin
摘要
摘要


Graph相关(图学习|图神经网络|图优化等)(6篇)

【1】D-GATNet: Interpretable Temporal Graph Attention Learning for ADHD Identification Using Dynamic Functional Connectivity
标题:D-GATNet:使用动态功能连接性进行ADHD识别的可解释时态图注意力学习
链接:https://arxiv.org/abs/2603.26308

作者:Qurat Ul Ain, Alptekin Temizel, Soyiba Jawed
备注:5 pages , 4 figures
摘要
摘要


【2】Topology-Aware Graph Reinforcement Learning for Energy Storage Systems Optimal Dispatch in Distribution Networks
标题:配电网储能系统优化调度的分布式图强化学习
链接:https://arxiv.org/abs/2603.26264

作者:Shuyi Gao, Stavros Orfanoudakis, Shengren Hou, Peter Palensky, Pedro P. Vergara
备注:15 pages, 10 figures
摘要
摘要


【3】Geometric Evolution Graph Convolutional Networks: Enhancing Graph Representation Learning via Ricci Flow
标题:几何进化图卷积网络:通过Ricci流增强图表示学习
链接:https://arxiv.org/abs/2603.26178

作者:Jicheng Ma, Yunyan Yang, Juan Zhao, Liang Zhao
摘要
摘要


【4】On the Complexity of Optimal Graph Rewiring for Oversmoothing and Oversquashing in Graph Neural Networks
标题:图神经网络中过度平滑和过度压缩的最佳图重新布线的复杂性
链接:https://arxiv.org/abs/2603.26140

作者:Mostafa Haghir Chehreghani
摘要
摘要


【5】DPD-Cancer: Explainable Graph-based Deep Learning for Small Molecule Anti-Cancer Activity Prediction
标题:DPD-癌症:基于可解释图形的深度学习用于小分子抗癌活性预测
链接:https://arxiv.org/abs/2603.26114

作者:Magnus H. Strømme, Alex G. C. de Sá, David B. Ascher
摘要
摘要


【6】Vision Transformers and Graph Neural Networks for Charged Particle Tracking in the ATLAS Muon Spectrometer
标题:视觉变形器和图形神经网络用于ATLAS μ子光谱仪中的带电粒子跟踪
链接:https://arxiv.org/abs/2603.25793

作者:Jonathan Renusch (on behalf of the ATLAS Collaboration)
摘要
摘要


Transformer(9篇)

【1】A Boltzmann-machine-enhanced Transformer For DNA Sequence Classification
标题:一种用于DNA序列分类的玻尔兹曼机增强Transformer
链接:https://arxiv.org/abs/2603.26465

作者:Zhixuan Cao, Yishu Xu, Xuang WU
备注:19 pages
摘要
摘要


【2】Knowledge Distillation for Efficient Transformer-Based Reinforcement Learning in Hardware-Constrained Energy Management Systems
标题:硬件约束能源管理系统中基于转换器的高效强化学习的知识提炼
链接:https://arxiv.org/abs/2603.26249

作者:Pascal Henrich, Jonas Sievers, Maximilian Beichter, Thomas Blank, Ralf Mikut, Veit Hagenmeyer
摘要
摘要


【3】Finding Distributed Object-Centric Properties in Self-Supervised Transformers
标题:在自我监督的Transformer中寻找分布式以对象为中心的属性
链接:https://arxiv.org/abs/2603.26127

作者:Samyak Rawlekar, Amitabh Swain, Yujun Cai, Yiwei Wang, Ming-Hsuan Yang, Narendra Ahuja
备注:Computer Vision and Pattern Recognition (CVPR) 2026
摘要
摘要


【4】MUST: Modality-Specific Representation-Aware Transformer for Diffusion-Enhanced Survival Prediction with Missing Modality
标题:必须:特定于模式的表示感知Transformer,用于缺失模式的扩散增强生存预测
链接:https://arxiv.org/abs/2603.26071

作者:Kyungwon Kim, Dosik Hwang
备注:Accepted to CVPR 2026. 10 pages, 5 figures, supplementary included
摘要
摘要


【5】Preventing Data Leakage in EEG-Based Survival Prediction: A Two-Stage Embedding and Transformer Framework
标题:防止基于脑电的生存预测中的数据泄露:两阶段嵌入和Transformer框架
链接:https://arxiv.org/abs/2603.25923

作者:Yixin Zhou, Zhixiang Liu, Vladimir I. Zadorozhny, Jonathan Elmer
备注:9 pages, 2 figures. Preliminary version
摘要
摘要


【6】Decoding Defensive Coverage Responsibilities in American Football Using Factorized Attention Based Transformer Models
标题:使用基于因子化注意力的Transformer模型解码美式橄榄球防守覆盖责任
链接:https://arxiv.org/abs/2603.25901

作者:Kevin Song, Evan Diewald, Ornob Siddiquee, Chris Boomhower, Keegan Abdoo, Mike Band, Amy Lee
备注:19 pages, 8 figures, ISACE 2026
摘要
摘要


【7】Do All Vision Transformers Need Registers? A Cross-Architectural Reassessment
标题:所有Vision Transformers都需要寄存器吗?跨架构重新评估
链接:https://arxiv.org/abs/2603.25803

作者:Spiros Baxevanakis, Platon Karageorgis, Ioannis Dravilas, Konrad Szewczyk
备注:Preprint. Submitted to Transactions on Machine Learning Research (TMLR). 26 pages, 17 figures
摘要
摘要


【8】A-SelecT: Automatic Timestep Selection for Diffusion Transformer Representation Learning
标题:A-SelectT:扩散Transformer表示学习的自动时步选择
链接:https://arxiv.org/abs/2603.25758

作者:Changyu Liu, James Chenhao Liang, Wenhao Yang, Yiming Cui, Jinghao Yang, Tianyang Wang, Qifan Wang, Dongfang Liu, Cheng Han
摘要
摘要


【9】On the Expressive Power of Contextual Relations in Transformers
标题:论《Transformer》中语境关系的表现力
链接:https://arxiv.org/abs/2603.25860

作者:Demián Fraiman
摘要
摘要


GAN|对抗|攻击|生成相关(6篇)

【1】Generative Modeling in Protein Design: Neural Representations, Conditional Generation, and Evaluation Standards
标题:蛋白质设计中的生成建模:神经表示、条件生成和评估标准
链接:https://arxiv.org/abs/2603.26378

作者:Senura Hansaja Wanasekara, Minh-Duong Nguyen, Xiaochen Liu, Nguyen H. Tran, Ken-Tye Yong
备注:20 pages, 7 tables, 4 figures
摘要
摘要


【2】Adversarial Bandit Optimization with Globally Bounded Perturbations to Linear Losses
标题:线性损失全局有界扰动的对抗强盗优化
链接:https://arxiv.org/abs/2603.26066

作者:Zhuoyu Cheng, Kohei Hatano, Eiji Takimoto
摘要
摘要


【3】Adversarial-Robust Multivariate Time-Series Anomaly Detection via Joint Information Retention
标题:通过联合信息保留的对抗稳健多元时间序列异常检测
链接:https://arxiv.org/abs/2603.25956

作者:Hadi Hojjati, Narges Armanfard
备注:22 pages, 4 figures
摘要
摘要


【4】Speech-Synchronized Whiteboard Generation via VLM-Driven Structured Drawing Representations
标题:通过LMA驱动的结构化绘图表示实现语音同步白板生成
链接:https://arxiv.org/abs/2603.25870

作者:Suraj Prasad, Pinak Mahapatra
摘要
摘要


【5】MAGNET: Autonomous Expert Model Generation via Decentralized Autoresearch and BitNet Training
标题:MAGNET:通过去中心化自动研究和BitNet训练生成自主专家模型
链接:https://arxiv.org/abs/2603.25813

作者:Yongwan Kim, Sungchul Park
备注:20 pages, 4 figures, 8 tables
摘要
摘要


【6】Globalized Adversarial Regret Optimization: Robust Decisions with Uncalibrated Predictions
标题:全球化对抗遗憾优化:具有未校准预测的稳健决策
链接:https://arxiv.org/abs/2603.25948

作者:Jannis Kurtz, Bart P.G. van Parys
摘要
摘要


半/弱/无/有监督|不确定性|主动学习(4篇)

【1】Meta-Learned Adaptive Optimization for Robust Human Mesh Recovery with Uncertainty-Aware Parameter Updates
标题:通过不确定性参数更新进行鲁棒人类网格恢复的元学习自适应优化
链接:https://arxiv.org/abs/2603.26447

作者:Shaurjya Mandal, Nutan Sharma, John Galeotti
摘要
摘要


【2】GLU: Global-Local-Uncertainty Fusion for Scalable Spatiotemporal Reconstruction and Forecasting
标题:GLU:可扩展时空重建和预测的全球-局部-不确定性融合
链接:https://arxiv.org/abs/2603.26023

作者:Linzheng Wang, Jason Chen, Nicolas Tricard, Zituo Chen, Sili Deng
摘要
摘要


【3】SAHMM-VAE: A Source-Wise Adaptive Hidden Markov Prior Variational Autoencoder for Unsupervised Blind Source Separation
标题:SAHMM-VAE:一种用于无监督盲源分离的源自适应隐马尔科夫先验变分自动编码器
链接:https://arxiv.org/abs/2603.25776

作者:Yuan-Hao Wei
摘要
摘要


【4】Uncertainty Quantification for Quantum Computing
标题:量子计算的不确定性量化
链接:https://arxiv.org/abs/2603.25039

作者:Ryan Bennink, Olena Burkovska, Konstantin Pieper, Jorge Ramirez, Elaine Wong
摘要
摘要


迁移|Zero/Few/One-Shot|自适应(6篇)

【1】ARTA: Adaptive Mixed-Resolution Token Allocation for Efficient Dense Feature Extraction
标题:ARTA:用于高效密集特征提取的自适应混合分辨率令牌分配
链接:https://arxiv.org/abs/2603.26258

作者:David Hagerman, Roman Naeem, Erik Brorsson, Fredrik Kahl, Lennart Svensson
摘要
摘要


【2】Dynamic Tokenization via Reinforcement Patching: End-to-end Training and Zero-shot Transfer
标题:通过强化修补的动态代币化:端到端训练和Zero-Shot转移
链接:https://arxiv.org/abs/2603.26097

作者:Yulun Wu, Sravan Kumar Ankireddy, Samuel Sharpe, Nikita Seleznev, Dehao Yuan, Hyeji Kim, Nam H. Nguyen
摘要
摘要


【3】AcTTA: Rethinking Test-Time Adaptation via Dynamic Activation
标题:AcTTA:通过动态激活重新思考测试时适应
链接:https://arxiv.org/abs/2603.26096

作者:Hyeongyu Kim, Geonhui Han, Dosik Hwang
备注:Accepted at CVPR 2026
摘要
摘要


【4】CD-Buffer: Complementary Dual-Buffer Framework for Test-Time Adaptation in Adverse Weather Object Detection
标题:CD-Buffer:用于不利天气物体检测中测试时间自适应的补充双缓冲器框架
链接:https://arxiv.org/abs/2603.26092

作者:Youngjun Song, Hyeongyu Kim, Dosik Hwang
备注:Accepted at CVPR 2026
摘要
摘要


【5】Can Vision Foundation Models Navigate? Zero-Shot Real-World Evaluation and Lessons Learned
标题:Vision Foundation模型可以导航吗?Zero-Shot现实世界评估和吸取的教训
链接:https://arxiv.org/abs/2603.25937

作者:Maeva Guerrier, Karthik Soma, Jana Pavlasek, Giovanni Beltrame
摘要
摘要


【6】A Lightweight, Transferable, and Self-Adaptive Framework for Intelligent DC Arc-Fault Detection in Photovoltaic Systems
标题:一种轻量级、可移植、自适应的太阳能系统智能直流弧故障检测框架
链接:https://arxiv.org/abs/2603.25749

作者:Xiaoke Yang, Long Gao, Haoyu He, Hanyuan Hang, Qi Liu, Shuai Zhao, Qiantu Tuo, Rui Li
备注:10 pages, 13 figures
摘要
摘要


强化学习(1篇)

【1】Empowering Epidemic Response: The Role of Reinforcement Learning in Infectious Disease Control
标题:增强流行病应对能力:强化学习在传染病控制中的作用
链接:https://arxiv.org/abs/2603.25771

作者:Mutong Liu, Yang Liu, Jiming Liu
备注:8 pages, 1 figure, 3 tables
摘要
摘要


符号|符号学习(1篇)

【1】Neuro-Symbolic Process Anomaly Detection
标题:神经符号过程异常检测
链接:https://arxiv.org/abs/2603.26461

作者:Devashish Gaikwad, Wil M. P. van der Aalst, Gyunam Park
摘要
摘要


医学相关(5篇)

【1】Evaluating Interactive 2D Visualization as a Sample Selection Strategy for Biomedical Time-Series Data Annotation
标题:评估交互式2D可视化作为生物医学时间序列数据注释的样本选择策略
链接:https://arxiv.org/abs/2603.26592

作者:Einari Vaaras, Manu Airaksinen, Okko Räsänen
摘要
摘要


【2】EcoFair: Trustworthy and Energy-Aware Routing for Privacy-Preserving Vertically Partitioned Medical Inference
标题:EcoFair:用于保护隐私的垂直分区医疗推理的值得信赖和能源意识的路由
链接:https://arxiv.org/abs/2603.26483

作者:Mostafa Anoosha, Dhavalkumar Thakker, Kuniko Paxton, Koorosh Aslansefat, Bhupesh Kumar Mishra, Baseer Ahmad, Rameez Raja Kureshi
备注:16 pages, 4 figures, 4 tables
摘要
摘要


【3】DuSCN-FusionNet: An Interpretable Dual-Channel Structural Covariance Fusion Framework for ADHD Classification Using Structural MRI
标题:DuSCN-FusionNet:使用结构MRI进行ADHD分类的可解释双通道结构协方差融合框架
链接:https://arxiv.org/abs/2603.26351

作者:Qurat Ul Ain, Alptekin Temizel, Soyiba Jawed
备注:5 pages, 5 figures
摘要
摘要


【4】Improving Risk Stratification in Hypertrophic Cardiomyopathy: A Novel Score Combining Echocardiography, Clinical, and Medication Data
标题:改善肥大性心肌病的风险分层:结合超声心动图、临床和药物数据的新评分
链接:https://arxiv.org/abs/2603.26254

作者:Marion Taconné, Valentina D.A. Corino, Annamaria Del Franco, Sara Giovani, Iacopo Olivotto, Adrien Al Wazzan, Erwan Donal, Pietro Cerveri, Luca Mainardi
摘要
摘要


【5】Doctorina MedBench: End-to-End Evaluation of Agent-Based Medical AI
标题:Doctorina MedBench:基于代理的医疗人工智能的端到端评估
链接:https://arxiv.org/abs/2603.25821

作者:Anna Kozlova, Stanislau Salavei, Pavel Satalkin, Hanna Plotnitskaya, Sergey Parfenyuk
摘要
摘要


自动驾驶|车辆|车道检测等(3篇)

【1】Interpretable long-term traffic modelling on national road networks using theory-informed deep learning
标题 :使用基于理论的深度学习对国家道路网络进行可解释的长期交通建模
链接:https://arxiv.org/abs/2603.26440

作者:Yue Li, Shujuan Chen, Akihiro Shimoda, Ying Jin
摘要
摘要


【2】EngineAD: A Real-World Vehicle Engine Anomaly Detection Dataset
标题:EngineAD:现实世界的车辆发动机异常检测数据集
链接:https://arxiv.org/abs/2603.25955

作者:Hadi Hojjati, Christopher Roth, Rory Woods, Ken Sills, Narges Armanfard
备注:12 pages, 2 figures
摘要
摘要


【3】Collision-Aware Vision-Language Learning for End-to-End Driving with Multimodal Infraction Datasets
标题:使用多模式不合格数据集进行冲突感知视觉语言学习,用于端到端驾驶
链接:https://arxiv.org/abs/2603.25946

作者:Alex Koran, Dimitrios Sinodinos, Hadi Hojjati, Takuya Nanri, Fangge Chen, Narges Armanfard
备注:33 pages, 11 figures
摘要
摘要


联邦学习|隐私保护|加密(1篇)

【1】Optimization Trade-offs in Asynchronous Federated Learning: A Stochastic Networks Approach
标题:同步联邦学习中的优化权衡:随机网络方法
链接:https://arxiv.org/abs/2603.26231

作者:Abdelkrim Alahyane (LAAS-SARA), Céline Comte (CNRS, LAAS-SARA), Matthieu Jonckheere (CNRS, LAAS-SARA)
摘要
摘要


推理|分析|理解|解释(4篇)

【1】PerceptionComp: A Video Benchmark for Complex Perception-Centric Reasoning
标题:PerceptionComp:以复杂感知为中心的推理的视频基准
链接:https://arxiv.org/abs/2603.26653

作者:Shaoxuan Li, Zhixuan Zhao, Hanze Deng, Zirun Ma, Shulin Tian, Zuyan Liu, Yushi Hu, Haoning Wu, Yuhao Dong, Benlin Liu, Ziwei Liu, Ranjay Krishna
备注:Project Page: this https URL
摘要
摘要


【2】A Lyapunov Analysis of Softmax Policy Gradient for Stochastic Bandits
标题:随机盗贼Softmax政策梯度的李雅普诺夫分析
链接:https://arxiv.org/abs/2603.26547

作者:Tor Lattimore
备注:6 pages
摘要
摘要


【3】ExVerus: Verus Proof Repair via Counterexample Reasoning
标题:ExVerus:通过反例推理进行Verus证明修复
链接:https://arxiv.org/abs/2603.25810

作者:Jun Yang, Yuechun Sun, Yi Wu, Rodrigo Caridad, Yongwei Yuan, Jianan Yao, Shan Lu, Kexin Pei
备注:31 pages, 8 figures
摘要
摘要


【4】Generative Score Inference for Multimodal Data
标题:多峰数据的生成性得分推理
链接:https://arxiv.org/abs/2603.26349

作者:Xinyu Tian, Xiaotong Shen
备注:25 pages, 4 figures
摘要
摘要


检测相关(3篇)

【1】Machine Learning Transferability for Malware Detection
标题:恶意软件检测的机器学习可移植性
链接:https://arxiv.org/abs/2603.26632

作者:César Vieira, João Vitorino, Eva Maia, Isabel Praça
备注:12 pages, 1 Figure, 2 tables, World CIST 2026
摘要
摘要


【2】Hardware-Aware Tensor Networks for Real-Time Quantum-Inspired Anomaly Detection at Particle Colliders
标题:硬件感知张量网络用于粒子碰撞机实时量子激励异常检测
链接:https://arxiv.org/abs/2603.26604

作者:Sagar Addepalli, Prajita Bhattarai, Abhilasha Dave, Julia Gonski
备注:28 pages, 9 figures
摘要
摘要


【3】TinyML for Acoustic Anomaly Detection in IoT Sensor Networks
标题:TinyML用于物联网传感器网络中的声学异常检测
链接:https://arxiv.org/abs/2603.26135

作者:Amar Almaini, Jakob Folz, Ghadeer Ashour
摘要
摘要


分类|识别(3篇)

【1】SPECTRA: An Efficient Spectral-Informed Neural Network for Sensor-Based Activity Recognition
标题:SPECTRA:一种用于基于传感器的活动识别的高效光谱信息神经网络
链接:https://arxiv.org/abs/2603.26482

作者:Deepika Gurung, Lala Shakti Swarup Ray, Mengxi Liu, Bo Zhou, Paul Lukowicz
摘要
摘要


【2】Automatic feature identification in least-squares policy iteration using the Koopman operator framework
标题:使用Koopman操作框架在最小平方策略迭代中自动特征识别
链接:https://arxiv.org/abs/2603.26464

作者:Christian Mugisho Zagabe, Sebastian Petiz
备注:6 pages
摘要
摘要


【3】Identification of Bivariate Causal Directionality Based on Anticipated Asymmetric Geometries
标题:基于预期不对称几何体的二元因果方向性识别
链接:https://arxiv.org/abs/2603.26024

作者:Alex Glushkovsky
备注:12 pages, 8 figure, 3 tables
摘要
摘要


表征(2篇)

【1】A Human-Inspired Decoupled Architecture for Efficient Audio Representation Learning
标题:用于高效音频表示学习的以人为本的去耦合架构
链接:https://arxiv.org/abs/2603.26098

作者:Harunori Kawano, Takeshi Sasaki
摘要
摘要


【2】Beyond identifiability: Learning causal representations with few environments and finite samples
标题:超越可识别性:在少数环境和有限样本的情况下学习因果表示
链接:https://arxiv.org/abs/2603.25796

作者:Inbeom Lee, Tongtong Jin, Bryon Aragam
摘要
摘要


3D|3D重建等相关(1篇)

【1】GLASS: Geometry-aware Local Alignment and Structure Synchronization Network for 2D-3D Registration
标题:GLASS:用于2D-3D配准的几何感知局部对齐和结构同步网络
链接 :https://arxiv.org/abs/2603.26262

作者:Zhixin Cheng, Jiacheng Deng, Xinjun Li, Bohao Liao, Li Liu, Xiaotian Yin, Baoqun Yin, Tianzhu Zhang
备注:Accepted by IEEE Transactions on Circuits and Systems for Video Technology
摘要
摘要


优化|敛散性(1篇)

【1】Curvature-aware Expected Free Energy as an Acquisition Function for Bayesian Optimization
标题:曲率感知的期望自由能作为贝叶斯优化的捕获函数
链接:https://arxiv.org/abs/2603.26339

作者:Ajith Anil Meera, Wouter Kouw
备注:under review
摘要
摘要


预测|估计(7篇)

【1】Benchmarking Tabular Foundation Models for Conditional Density Estimation in Regression
标题:回归中条件密度估计的表格基础模型基准
链接:https://arxiv.org/abs/2603.26611

作者:Rafael Izbicki, Pedro L. C. Rodrigues
摘要
摘要


【2】Characterization and forecasting of national-scale solar power ramp events
标题:全国规模太阳能发电坡道事件的特征和预测
链接:https://arxiv.org/abs/2603.26596

作者:Luca Lanzilao, Angela Meyer
摘要
摘要


【3】KMM-CP: Practical Conformal Prediction under Covariate Shift via Selective Kernel Mean Matching
标题:KMM-CP:通过选择性核均值匹配在协变量漂移下的实用保形预测
链接:https://arxiv.org/abs/2603.26415

作者:Siddhartha Laghuvarapu, Rohan Deb, Jimeng Sun
摘要
摘要


【4】Accurate Precipitation Forecast by Efficiently Learning from Massive Atmospheric Variables and Unbalanced Distribution
标题:有效学习大量大气变量和不平衡分布进行准确降水预报
链接:https://arxiv.org/abs/2603.26108

作者:Shuangliang Li, Siwei Li, Li Li, Weijie Zou, Jie Yang, Maolin Zhang
摘要
摘要


【5】QuitoBench: A High-Quality Open Time Series Forecasting Benchmark
标题:QuitoBench:高质量的开放时间序列预测基准
链接:https://arxiv.org/abs/2603.26017

作者:Siqiao Xue, Zhaoyang Zhu, Wei Zhang, Rongyao Cai, Rui Wang, Yixiang Mu, Fan Zhou, Jianguo Li, Peng Di, Hang Yu
备注:project site: this https URL
摘要
摘要


【6】Pure and Physics-Guided Deep Learning Solutions for Spatio-Temporal Groundwater Level Prediction at Arbitrary Locations
标题:用于任意位置时空地下水位预测的纯粹和物理引导深度学习解决方案
链接:https://arxiv.org/abs/2603.25779

作者:Matteo Salis, Gabriele Sartor, Rosa Meo, Stefano Ferraris, Abdourrahmane M. Atto
摘要
摘要


【7】Conditional Neural Bayes Ratio Estimation for Experimental Design Optimisation
标题:实验设计优化的条件神经Bayes比估计
链接:https://arxiv.org/abs/2603.26489

作者:S. A. K. Leeney, T. Gessey-Jones, W. J. Handley, E. de Lera Acedo, H. T. J. Bevins, J. L. Tutt
备注:11 pages, 5 figures. Submitted to IEEE Transactions on Neural Networks and Learning Systems
摘要
摘要


其他神经网络|深度学习|模型|建模(23篇)

【1】PQuantML: A Tool for End-to-End Hardware-aware Model Compression
标题:PQuantML:端到端硬件感知模型压缩工具
链接:https://arxiv.org/abs/2603.26595

作者:Roope Niemi, Anastasiia Petrovych, Arghya Ranjan Das, Enrico Lupi, Chang Sun, Dimitrios Danopoulos, Marlon Joshua Helbing, Mia Liu, Sebastian Dittmeier, Michael Kagan, Vladimir Loncar, Maurizio Pierini
摘要
摘要


【2】From Synthetic Data to Real Restorations: Diffusion Model for Patient-specific Dental Crown Completion
标题:从合成数据到真实修复:患者特定齿冠完成的扩散模型
链接:https://arxiv.org/abs/2603.26588

作者:Dávid Pukanec, Tibor Kubík, Michal Španěl
备注:VISAPP 2026 Conference
摘要
摘要


【3】The Climber's Grip -- Personalized Deep Learning Models for Fear and Muscle Activity in Climbing
标题:攀登者的紧握--攀登中恐惧和肌肉活动的个性化深度学习模型
链接:https://arxiv.org/abs/2603.26575

作者:Matthias Boeker, Dana Swarbrick, Ulysse T.A. Côté-Allard, Marc T.P. Adam, Hugo L. Hammer, Pål Halvorsen
摘要
摘要


【4】Sharp Capacity Scaling of Spectral Optimizers in Learning Associative Memory
标题:联想记忆学习中谱优化器的容量缩放
链接:https://arxiv.org/abs/2603.26554

作者:Juno Kim, Eshaan Nichani, Denny Wu, Alberto Bietti, Jason D. Lee
备注:77 pages, 8 figures
摘要
摘要


【5】Foundation Model for Cardiac Time Series via Masked Latent Attention
标题:通过掩蔽隐性注意力的心脏时间序列基础模型
链接:https://arxiv.org/abs/2603.26475

作者:Moritz Vandenhirtz, Samuel Ruipérez-Campillo, Simon Böhi, Sonia Laguna, Irene Cannistraci, Andrea Agostini, Ece Ozkan, Thomas M. Sutter, Julia E. Vogt
备注:First two authors are co-first. Last two authors are co-senior
摘要
摘要


【6】UNIFERENCE: A Discrete Event Simulation Framework for Developing Distributed AI Models
标题:CLARENCE:用于开发分布式人工智能模型的离散事件模拟框架
链接:https://arxiv.org/abs/2603.26469

作者:Doğaç Eldenk, Stephen Xia
摘要
摘要


【7】Maintaining Difficulty: A Margin Scheduler for Triplet Loss in Siamese Networks Training
标题:维持困难:连体网络训练中三重损失的保证金
链接:https://arxiv.org/abs/2603.26389

作者:Roberto Sprengel Minozzo Tomchak, Oge Marques, Lucas Garcia Pedroso, Luiz Eduardo Oliveira, Paulo Lisboa de Almeida
摘要
摘要


【8】Dual-Stage Invariant Continual Learning under Extreme Visual Sparsity
标题:极端视觉稀疏下的双阶段不变连续学习
链接:https://arxiv.org/abs/2603.26190

作者:Rangya Zhang, Jiaping Xiao, Lu Bai, Yuhang Zhang, Mir Feroskhan
摘要
摘要


【9】Can AI Scientist Agents Learn from Lab-in-the-Loop Feedback? Evidence from Iterative Perturbation Discovery
标题:人工智能科学家代理可以从循环实验室反馈中学习吗?迭代微扰发现的证据
链接:https://arxiv.org/abs/2603.26177

作者:Gilles Wainrib, Barbara Bodinier, Haitem Dakhli, Josep Monserrat, Almudena Espin Perez, Sabrina Carpentier, Roberta Codato, John Klein
摘要
摘要


【10】PruneFuse: Efficient Data Selection via Weight Pruning and Network Fusion
标题:剪枝:通过权重剪枝和网络融合的高效数据选择
链接:https://arxiv.org/abs/2603.26138

作者:Humaira Kousar, Hasnain Irshad Bhatti, Jaekyun Moon
备注:Published in TMLR (Featured Certification). arXiv admin note: substantial text overlap with arXiv:2501.01118
摘要
摘要


【11】Online Learning for Dynamic Constellation Topologies
标题:动态星座布局在线学习
链接:https://arxiv.org/abs/2603.25954

作者:João Norberto, Ricardo Ferreira, Cláudia Soares
摘要
摘要


【12】Personalizing Mathematical Game-based Learning for Children: A Preliminary Study
标题:儿童个性化数学游戏学习:初步研究
链接:https://arxiv.org/abs/2603.25925

作者:Jie Gao, Adam K. Dubé
备注:Short research paper accepted at 27th International Conference on AI in Education (AIED 2026)
摘要
摘要


【13】Data-Driven Plasticity Modeling via Acoustic Profiling
标题:通过声学轮廓的数据驱动可塑性建模
链接:https://arxiv.org/abs/2603.25894

作者:Khalid El-Awady
摘要
摘要


【14】DRiffusion: Draft-and-Refine Process Parallelizes Diffusion Models with Ease
标题:DRiffetts:起草和完善流程轻松简化扩散模型
链接:https://arxiv.org/abs/2603.25872

作者:Runsheng Bai, Chengyu Zhang, Yangdong Deng
摘要
摘要


【15】Reconstructing Quantum Dot Charge Stability Diagrams with Diffusion Models
标题:用扩散模型重建量子点电荷稳定性图
链接:https://arxiv.org/abs/2603.26432

作者:Vinicius Hernandes, Joseph Rogers, Rouven Koch, Thomas Spriggs, Brennan Undseth, Anasua Chatterjee, Lieven M. K. Vandersypen, Eliska Greplova
备注:Code available at this https URL. Data available at this https URL
摘要
摘要


【16】Kantorovich--Kernel Neural Operators: Approximation Theory, Asymptotics, and Neural Network Interpretation
标题:Kantorovich-核神经算子:逼近理论、渐近性和神经网络解释
链接:https://arxiv.org/abs/2603.26418

作者:Tian-Xiao He
摘要
摘要


【17】Making Multi-Axis Models Robust to Multiplicative Noise: How, and Why?
标题:使多轴模型对乘性噪音具有鲁棒性:如何以及为什么?
链接:https://arxiv.org/abs/2603.26327

作者:Bailey Andrew, David R. Westhead, Luisa Cutillo
备注:9 pages (26 with supplemental), 4 figures (+2 in supplemental), preprint
摘要
摘要


【18】STN-GPR: A Singularity Tensor Network Framework for Efficient Option Pricing
标题:STN-GPT:有效期权定价的奇异张量网络框架
链接:https://arxiv.org/abs/2603.26318

作者:Dominic Gribben, Carolina Allende, Alba Villarino, Aser Cortines, Mazen Ali, Román Orús, Pascal Oswald, Noureddine Lehdili
备注:15 pages, 2 figures
摘要
摘要


【19】Semi-structured multi-state delinquency model for mortgage default
标题:抵押贷款违约的半结构化多州拖欠模型
链接:https://arxiv.org/abs/2603.26309

作者:Victor Medina-Olivares, Wangzhen Xia, Stefan Lessmann, Nadja Klein
摘要
摘要


【20】On associative neural networks for sparse patterns with huge capacities
标题:基于大容量稀疏模式的联想神经网络
链接:https://arxiv.org/abs/2603.26217

作者:Matthias Löwe, Franck Vermet
备注:22 pages
摘要
摘要


【21】Asymptotic Optimism for Tensor Regression Models with Applications to Neural Network Compression
标题:张量回归模型的渐进优化及其在神经网络压缩中的应用
链接:https://arxiv.org/abs/2603.26048

作者:Haoming Shi, Eric C. Chi, Hengrui Luo
备注:62 pages, 11 figures
摘要
摘要


【22】Spectral Coherence Index: A Model-Free Metric for Protein Structural Ensemble Quality Assessment
标题:光谱一致性指数:蛋白质结构集合质量评估的无模型指标
链接:https://arxiv.org/abs/2603.25880

作者:Yuda Bi, Huaiwen Zhang, Jingnan Sun, Vince D Calhoun
摘要
摘要


【23】KANEL: Kolmogorov-Arnold Network Ensemble Learning Enables Early Hit Enrichment in High-Throughput Virtual Screening
标题:KANEL:Kolmogorov-Arnold网络包围学习使高吞吐量虚拟筛选中的早期命中富集成为可能
链接:https://arxiv.org/abs/2603.25755

作者:Pavel Koptev, Nikita Krainov, Konstantin Malkov, Alexander Tropsha
备注:8 Pages
摘要
摘要


其他(28篇)

【1】Tunable Soft Equivariance with Guarantees
标题:带保证的可调软等效性
链接:https://arxiv.org/abs/2603.26657

作者:Md Ashiqur Rahman, Lim Jun Hao, Jeremiah Jiang, Teck-Yian Lim, Raymond A. Yeh
摘要
摘要


【2】An LP-based Sampling Policy for Multi-Armed Bandits with Side-Observations and Stochastic Availability
标题:基于LP的具有侧向观察和随机可用性的多臂盗贼抽样策略
链接:https://arxiv.org/abs/2603.26647

作者:Ashutosh Soni, Peizhong Ju, Atilla Eryilmaz, Ness B. Shroff
摘要
摘要


【3】Automatic Laplace Collapsed Sampling: Scalable Marginalisation of Latent Parameters via Automatic Differentiation
标题:自动拉普拉斯折叠抽样:通过自动分化实现潜在参数的可扩展边缘化
链接:https://arxiv.org/abs/2603.26644

作者:Toby Lovick, David Yallup, Will Handley
备注:28 Pages, 7 Figures. Comments welcome
摘要
摘要


【4】Context-specific Credibility-aware Multimodal Fusion with Conditional Probabilistic Circuits
标题:具有条件概率电路的特定上下文可信度感知多模式融合
链接:https://arxiv.org/abs/2603.26629

作者:Pranuthi Tenali, Sahil Sidheekh, Saurabh Mathur, Erik Blasch, Kristian Kersting, Sriraam Natarajan
摘要
摘要


【5】Sustainability Is Not Linear: Quantifying Performance, Energy, and Privacy Trade-offs in On-Device Intelligence
标题:可持续发展不是线性的:量化设备上智能中的性能、能源和隐私权衡
链接:https://arxiv.org/abs/2603.26603

作者:Eziyo Ehsani, Luca Giamattei, Ivano Malavolta, Roberto Pietrantuono
备注:Under review at Empirical Software Engineering (EMSE)
摘要
摘要


【6】Machine Unlearning under Retain-Forget Entanglement
标题:保留-忘记纠缠下的机器取消学习
链接:https://arxiv.org/abs/2603.26569

作者:Jingpu Cheng, Ping Liu, Qianxiao Li, Chi Zhang
备注:ICLR 2026 camera-ready
摘要
摘要


【7】The internal law of a material can be discovered from its boundary
标题:物质的内在规律可以从其边界发现
链接:https://arxiv.org/abs/2603.26517

作者:Francesco Regazzoni
摘要
摘要


【8】Shapley meets Rawls: an integrated framework for measuring and explaining unfairness
标题:沙普利遇见罗尔斯:衡量和解释不公平的集成框架
链接:https://arxiv.org/abs/2603.26476

作者:Fadoua Amri-Jouidel, Emmanuel Kemel, Stéphane Mussard
摘要
摘要


【9】Fair Data Pre-Processing with Imperfect Attribute Space
标题:属性空间不完美的公平数据预处理
链接:https://arxiv.org/abs/2603.26456

作者:Ying Zheng, Yangfan Jiang, Kian-Lee Tan
备注:Accepted at SIGMOD 2026
摘要
摘要


【10】Label-Free Cross-Task LoRA Merging with Null-Space Compression
标题:无标签跨任务LoRA与空空间压缩合并
链接:https://arxiv.org/abs/2603.26317

作者:Wonyoung Lee, Wooseong Jeong, Kuk-Jin Yoon
备注:Accepted at CVPR 2026
摘要
摘要


【11】SALMUBench: A Benchmark for Sensitive Association-Level Multimodal Unlearning
标题:SALMUBench:敏感关联级多模式遗忘的基准
链接:https://arxiv.org/abs/2603.26316

作者:Cai Selvas-Sala, Lei Kang, Lluis Gomez
备注:Accepted to CVPR 2026. Project page: this http URL
摘要
摘要


【12】Preference-Aligned LoRA Merging: Preserving Subspace Coverage and Addressing Directional Anisotropy
标题:偏好一致的LoRA合并:保留子空间覆盖并解决方向各向异性
链接:https://arxiv.org/abs/2603.26299

作者:Wooseong Jeong, Wonyoung Lee, Kuk-Jin Yoon
备注:Accepted at CVPR 2026
摘要
摘要


【13】Contrastive Conformal Sets
标题:对比保形集
链接:https://arxiv.org/abs/2603.26261

作者:Yahya Alkhatib, Wee Peng Tay
摘要
摘要


【14】PEANUT: Perturbations by Eigenvalue Alignment for Attacking GNNs Under Topology-Driven Message Passing
标题:PEANUT:特征值对齐对攻击GNN的扰动在布局驱动的消息传递下
链接:https://arxiv.org/abs/2603.26136

作者:Bhavya Kohli, Biplab Sikdar
备注:8 content pages, 12 total pages including references
摘要
摘要


【15】Constitutive parameterized deep energy method for solid mechanics problems with random material parameters
标题:随机材料参数固体力学问题的本构参数化深能方法
链接:https://arxiv.org/abs/2603.26030

作者:Zhangyong Liang, Huanhuan Gao
摘要
摘要


【16】Second-Order, First-Class: A Composable Stack for Curvature-Aware Training
标题:二阶、一流:用于曲线感知训练的可组合堆栈
链接:https://arxiv.org/abs/2603.25976

作者:Mikalai Korbit, Mario Zanon
备注:22 pages, 3 figures. Code available at this https URL
摘要
摘要


【17】Do Neurons Dream of Primitive Operators? Wake-Sleep Compression Rediscovers Schank's Event Semantics
标题:神经元梦想着原始操作符吗?醒睡压缩重新发现了沙克的事件语义
链接:https://arxiv.org/abs/2603.25975

作者:Peter Balogh
摘要
摘要


【18】On the Objective and Feature Weights of Minkowski Weighted k-Means
标题:Minkowski加权k-均值的客观权重和特征权重
链接:https://arxiv.org/abs/2603.25958

作者:Renato Cordeiro de Amorim, Vladimir Makarenkov
摘要
摘要


【19】Parameter-Free Dynamic Regret for Unconstrained Linear Bandits
标题:无约束线性盗贼的无参数动态遗憾
链接:https://arxiv.org/abs/2603.25916

作者:Alberto Rumi, Andrew Jacobsen, Nicolò Cesa-Bianchi, Fabio Vitale
备注:10 pages. v1: AISTATS 2026
摘要
摘要


【20】Why Safety Probes Catch Liars But Miss Fanatics
标题:为什么安全探测器能抓到说谎者,却错过了狂热分子
链接:https://arxiv.org/abs/2603.25861

作者:Kristiyan Haralambiev
备注:18 pages, 4 figures, 14 tables
摘要
摘要


【21】Incorporating contextual information into KGWAS for interpretable GWAS discovery
标题:将上下文信息融入KGWAS以实现可解释的GWAS发现
链接:https://arxiv.org/abs/2603.25855

作者:Cheng Jiang, Brady Ryan, Megan Crow, Kipper Fletez-Brant, Kashish Doshi, Sandra Melo Carlos, Kexin Huang, Burkhard Hoeckendorf, Heming Yao, David Richmond
摘要
摘要


【22】A Compression Perspective on Simplicity Bias
标题:简单性偏见的压缩视角
链接:https://arxiv.org/abs/2603.25839

作者:Tom Marty, Eric Elmoznino, Leo Gagnon, Tejas Kasetty, Mizu Nishikawa-Toomey, Sarthak Mittal, Guillaume Lajoie, Dhanya Sridhar
摘要
摘要


【23】A Neural Score-Based Particle Method for the Vlasov-Maxwell-Landau System
标题:Vlasov-Maxwell-Landau系统的基于神经分数的粒子方法
链接:https://arxiv.org/abs/2603.25832

作者:Vasily Ilin, Jingwei Hu
备注:presented at ICLR AI&PDE workshop
摘要
摘要


【24】A Judge Agent Closes the Reliability Gap in AI-Generated Scientific Simulation
标题:判断代理缩小了人工智能生成的科学模拟中的可靠性差距
链接:https://arxiv.org/abs/2603.25780

作者:Chengshuai Yang
备注:36 pages, 5 figures, 22 tables, includes Supplementary Information
摘要
摘要


【25】Identifying Connectivity Distributions from Neural Dynamics Using Flows
标题:使用流从神经动力学中识别连通性分布
链接:https://arxiv.org/abs/2603.26506

作者:Timothy Doyeon Kim, Ulises Pereira-Obilinovic, Yiliu Wang, Eric Shea-Brown, Uygar Sümbül
摘要
摘要


【26】A Power-Weighted Noncentral Complex Gaussian Distribution
标题:功率加权非中心复高斯分布
链接:https://arxiv.org/abs/2603.26344

作者:Toru Nakashika
摘要
摘要


【27】Privacy-Accuracy Trade-offs in High-Dimensional LASSO under Perturbation Mechanisms
标题:微扰机制下多维LASO的隐私度与准确度权衡
链接:https://arxiv.org/abs/2603.26227

作者:Ayaka Sakata, Haruka Tanzawa
备注:53 pages, 11 figures
摘要
摘要


【28】A Priori Sampling of Transition States with Guided Diffusion
标题:具有引导扩散的过渡状态的初步抽样
链接:https://arxiv.org/abs/2603.25980

作者:Hyukjun Lim, Soojung Yang, Lucas Pinède, Miguel Steiner, Yuanqi Du, Rafael Gómez-Bombarelli
摘要
摘要


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

Python社区是高质量的Python/Django开发社区
本文地址:http://www.python88.com/topic/194554