当前位置: 首页 > 通知公告 > 正文

学术报告预告:Neural Networks and Deep Learning-Basic Principles and Practical Issues

发布时间:2018-07-14 来源: 作者: 浏览数:

物理与电子科学学院学术报告预告

题目:Neural Networks and Deep Learning-Basic Principles and Practical Issues

主讲人:卢军青

间:2018717日上午10:15

点:6205

报告内容:

Deep learning is to use vast amount of data to train neural networks with multi-hidden layers. This talk will introduce the general structures of the artificial neurons and neural networks, and discuss the basic principles and algorithms used in network training, such as feedforward and backpropagation algorithms, stochastic gradient decent (SGD) technique, and activation and cost functions. We will also discuss some of the important practical issues in network training, such as training data pre-processing, parameters initializing, hyperparameters tuning, cross validation, and overfitting.

主讲人简介:

卢军青:美国东卡大学(East Carolina University)教授,1983年毕业于南开大学物理系,1986年获南开大学物理系硕士研究生,1991年获美国加利福尼亚大学尔湾分校物理系博士研究生。曾在肯特州立大学、加利福尼亚大学 、圣地亚哥市表面光学公司、东卡大学任职,发表SCI论文60多篇,获国内外专利多项,主持美国自然科学基金3项。现为Optics LettersOptical ExpressApplied OpticsIEEE Transactions on Biomedical Engineering 等杂志审稿人。欢迎全校对神经网络和深度学习感兴趣的师生参加。