arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2307.02693cs.LGstat.ML
收藏

核、数据与物理

Kernels, Data & Physics

École Polytechnique Fédérale de Lausanne (EPFL)(洛桑联邦理工学院);Instituto de Matemática Pura e Aplicada (IMPA)(纯数学与应用数学研究所);Technical University of Munich(慕尼黑工业大学);New York University(纽约大学)

Francesco Cagnetta, Deborah Oliveira, Mahalakshmi Sabanayagam, Nikolaos Tsilivis, Julia Kempe

AI总结:

本文基于Julia Kempe教授的暑期学校讲义,介绍利用神经正切核(NTK)方法理解机器学习难题,并重点展示其在数据蒸馏与对抗鲁棒性中的应用。

摘要:

本文是Julia Kempe教授在Les Houches“机器学习统计物理”暑期学校授课的讲义。该讲义讨论了机器学习问题中所谓的NTK方法,其核心在于通过寻找可处理的核表述来理解通常不可解的问题。讲义主要聚焦于实际应用,例如数据蒸馏和对抗鲁棒性,同时也讨论了归纳偏置的示例。

英文摘要:

Lecture notes from the course given by Professor Julia Kempe at the summer school "Statistical physics of Machine Learning" in Les Houches. The notes discuss the so-called NTK approach to problems in machine learning, which consists of gaining an understanding of generally unsolvable problems by finding a tractable kernel formulation. The notes are mainly focused on practical applications such as data distillation and adversarial robustness, examples of inductive bias are also discussed.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑