特邀上海交通大学许志钦教授作线上学术报告

发布者:朱亚宾发布时间:2020-05-22浏览次数:10

报告题目:Understanding Deep learning from Fourier perspective: The F-Principle

报告时间:2020524日(周11:2512:10

人:许志钦教授

报告地点:Zoom云会议(ID937 6354 7091, 密码:nanxinda60

报告摘要:We demonstrate a very universal Frequency Principle (F-Principle) --- DNNs often fit target functions from low to high frequencies --- on high-dimensional benchmark datasets such as MNIST/CIFAR10 and deep neural networks such as VGG16. We then utilize the F-Principle to understand the strengths and limitations of deep learning, which leads to a better use of DNNs. Our work of the F-Principle makes a step towards a quantitative understanding of the learning and generalization of DNNs.

欢迎广大师生踊跃参加!

数学与统计学院

2020522

附:专家简介

许志钦,上海交通大学副教授。分别于2012年和2016年在上海交通大学获得物理学士学位(致远学院)和数学博士学位,2016年至2019年在纽约大学阿布扎分校做博士后和在纽约大学柯朗研究所做访问学者。主要研究方向为深度学习理论和计算神经科学。