报告题目：A Random Matrix Approach to Neural Networks: From Linear to Nonlinear, and from Shallow to Deep

报告人：廖振宇 副研究员 华中科技大学

报告时间：2025年12月12日 (星期五) 16:30-17:15

报告地点：三楼多功能厅2

校内邀请人：邹婷婷 [email protected]

报告摘要：

Deep neural networks have become the cornerstone of modern machine learning, yet their multi-layer structure, nonlinearities, and intricate optimization processes pose considerable theoretical challenges.

In this talk, I will review recent advances in random matrix analysis that shed new light on these complex ML models. Starting with the foundational case of linear regression, I will demonstrate how the proposed analysis extends naturally to shallow nonlinear and ultimately deep nonlinear network models. I will also discuss practical implications (e.g., compressing and/or designing "equivalent" NN models) that arise from these theoretical insights. The talk is based on a recent review paper //arxiv.org/abs/2506.13139 joint with Michael W. Mahoney.

报告人简介：

廖振宇，于法国巴黎萨克雷大学获数学与计算机博士学位，后在美国加州大学伯克利分校统计系和ICSI从事博士后研究工作，2021年起至今在华中科技大学电信脱衣舞 工作，任副研究员。主要研究方向是机器学习理论与应用、高维统计和随机矩阵理论，成果发表于ICML、NeurIPS、ICLR、COLT、IEEE汇刊和AAP等机器学习和数据处理的会议与期刊，合著专著Random Matrix Methods for Machine Learning。任ICML、NeurIPS、ICLR、AISTATS和IJCNN等会议的领域主席和Statistics and Computing期刊编委。