International Joint Conference On Theoretical Computer Science – Frontier of Algorithmic Wisdom

August 15-19, 2022, City University of Hong Kong, Hong Kong

 

Invited Speakers

Track D


Unified Theory of Explaining Heuristic Findings in Attribution, Robustness, Generalization, Visual Features in a DNN

Quanshi Zhang, Huiqi Deng

Shanghai Jiaotong University

Abstract:
The interpretability of deep neural networks has received increasing attention in recent years. However, most XAI methods are designed to fit people’s cognition in an experimental manner without solid theoretic foundations. More crucially, many heuristic findings in deep learning cannot be theoretically explained. Thus, the lack of theoretic foundations has hampered the future development of XAI. Therefore, in this talk, the speakers will review several studies of explainable AI theories of their research group, which use game-theoretic interactions to explain common mechanisms shared by different studies. The speakers will provide a unified explanation for the attribution, the adversarial robustness, the generalization power of a DNN, and visual concepts encoded by the DNN.

Bio:
Dr. Quanshi Zhang is an associate professor at Shanghai Jiao Tong University, China. He received the Ph.D. degree from the University of Tokyo in 2014. From 2014 to 2018, he was a post-doctoral researcher at the University of California, Los Angeles. His research interests are mainly machine learning and computer vision. In particular, he has made influential research in explainable AI (XAI) and received the ACM China Rising Star Award. He was the co-chairs of the workshops towards XAI in ICML 2021, AAAI 2019, and CVPR 2019. We is the speaker of the tutorials on XAI at IJCAI 2020 and IJCAI 2021.
Huiqi Deng is a postdoctoral researcher in the team of Prof. Quanshi Zhang at Shanghai JiaoTong University. Her research interests focus on trustworthy machine learning, especially the interpretability of deep neural networks. In 2021, she received her Ph.D. degree from the School of Mathematics, Sun Yat-Sen University. During her Ph.D. period, she has ever visited HongKong Baptist University and Texas A&M University for three years. In the past 3 years, she has published many papers in top-tier conferences and journals such as ICLR, AAAI, KDD, Pattern Recognition. Among them, the ICLR paper was invited to give oral presentations (top 1.5%).