Although Deep Neural Networks (DNNs) have many successful applications in various domains, it has been acknowledged as a bottle-neck problem of restricting DNNs development that DNNs success lacks rigorous interpretability in a theoretical or logical framework. This tutorial-talk reviews several fundamental concepts of interpretability and summarizes the different viewpoints regarding deep learning interpretability. The challenges of this issue are analyzed and some new views about this issue, i.e., the localized generalization error bounds, data bias, and decision metric analysis of DNN process are shared in this talk.

Xizhao Wang was a Research Fellow in the Hong Kong Polytechnic University until 2001 after receiving his doctorate degree in 1998. He served as the Dean of Computer Sciences College in Hebei University, China from 2000 to 2014. In 2014, he joined Shenzhen University, China as a professor and a director of the Big Data Institute.

Dr. Wang’s current research interests include uncertainty-aware machine learning and big data analysis. He pioneered the research on Machine Learning in uncertain environment. His research highlights can be briefly summarized as, digging the distribution of (big) data by modeling uncertainty, and then using distributed parallel technology to design and implement classification and clustering algorithms for different tasks in cybernetics. He published several monographs and textbooks, and over 280 research papers at conferences and in journals in the above topics. His works have been cited 11,675 times since 2016 according to Google scholar. He was selected as a Highly Cited Researcher by Clarivate in 2019 and 2020, in recognition of exceptional research performance demonstrated by production of multiple highly cited papers that rank in the cross-field top 1%.

Dr. Wang was a recipient of the Outstanding Contribution Award in 2004 and the Best Associate Editor Award in 2006 from the IEEE SMCS Society. He served as a member of the Governing Boards in the IEEE SMC Society (2007-2009, 2012-2014). He is an IEEE Fellow (2012), a CAAI Fellow (2017), and the Editor-in-Chief of Springer Journal Machine Learning and Cybernetics (from 2010 to present)