报告题目：Deep Neural Network Architecture Design
Dong Xu is Shumaker Endowed Professor in Department of Electrical Engineering and Computer Science, Director of Information Technology Program, with appointments in the Christopher S. Bond Life Sciences Center and the Informatics Institute at the University of Missouri-Columbia. He obtained his PhD from the University of Illinois, Urbana-Champaign in 1995 and did two years of postdoctoral work at the US National Cancer Institute. He was a Staff Scientist at Oak Ridge National Laboratory until 2003 before joining the University of Missouri, where he served as Department Chair of Computer Science during 2007-2016. His research is in computational biology and bioinformatics, including machine-learning application in bioinformatics, protein structure prediction, post-translational modification prediction, high-throughput biological data analyses, in silico studies of plants, microbes and cancers, biological information systems, and mobile App development for healthcare. He has published more than 300 papers. He was elected to the rank of American Association for the Advancement of Science (AAAS) Fellow in 2015.
To address diverse applications, many families of network structures have been developed in deep learning, such as deep neural networks, convolutional neural network (CNN), recurrent neural networks (RNN), and generative adversarial network (GAN). In addition to these basic network architectures, a number of advanced architectures and combinations of different architectures are also introduced. In this lecture, I will cover the following major types of advanced architectures: (1) residual/dense networks; (2) inception networks; (3) light networks; (4) R-CNN; (5) graph neural networks; and (6) hybrid networks. These advanced architectures often significantly improve the performance of various applications, as demonstrated in many research benchmarks and big data open challenges. I will discuss design principles of these deep learning networks. I will also address building and optimizing networks using auto-ML and evolutionary approaches.