学术报告：Understanding Surprising Generalization Phenomena in Deep Learning

报告题目：Understanding Surprising Generalization Phenomena in Deep Learning

报告时间：2023年12月25日14:30

报告地点：计算机学院B404会议室

报告人：胡威

报告人国籍：中国

报告人单位：密歇根大学计算机科学与工程系

报告人简介：Wei Hu is an Assistant Professor in Computer Science and Engineering at the University of Michigan. He obtained his Ph.D. degree from Princeton University and Bachelor's degree from Tsinghua University. His research interest is in the theoretical and scientific foundations of deep learning. He is a recipient of the Google Research Scholar award and the Siebel Scholarship.

报告摘要：Despite the phenomenal empirical successes of deep learning in many application domains, its underlying mathematical mechanisms remain poorly understood. In particular, deep learning has exhibited a number of surprising generalization phenomena that are not captured by classical statistical learning theory. This talk will introduce some of the intriguing generalization phenomena observed in deep learning, including implicit regularization, benign overfitting, and grokking, as well as some recent progress on the theoretical characterizations of these phenomena.

邀请人：王胜