Prof. Xiaowei Huang
Biography
Prof. Xiaowei Huang is affiliated with the Department of Computer Science at the University of Liverpool, UK. His research is concerned with the development of automated verification techniques that ensure the correctness and reliability of intelligent systems. He is leading the research direction on the verification and validation of deep neural networks. He has published 50+ papers, most of which appear in top conferences and journals of either Artificial Intelligence, such as the Artificial Intelligence Journal, ACM transactions on Computational Logics, AAAI, IJCAI, AAMAS, etc, Formal Methods, such as CAV, TACAS, and Theoretical Computer Science, or Software Engineering, such as ICSE and ASE. He has given invited talks at several leading conferences, discussing topics related to the safety and security of applying machine learning algorithms to critical applications. He co-chairs the AAAI'19 and IJCAI'19 workshops on Artificial Intelligence Safety. He is the PI of two Dstl projects on Test Metrics for Artificial Intelligence and a co-Investigator of an EPSRC Robotics and Artificial Intelligence Hub on Offshore Robotics for Certification of Assets.
Abstract
The past few years have seen deep learning achieved human-level intelligence in several tasks, such as image processing and video game playing, and been applied to many application areas, including robotics control, healthcare, etc. While the empirical precision has been significantly improved, deep learning suffers from the problem of lacking certification and understanding. This twoday tutorial will introduce some recent progress aiming to open the black-box of the deep learning, from the perspectives of automated verification and information theory. The verification of deep learning, to be presented by Dr Xiaowei Huang, will focus on a few techniques including constraint-solving based methods, search-based methods, and methods based on functional approximation. The understanding of deep learning, to be presented by Dr Xinping Yi, will focus on the information-theoretic measure of the training dynamics through the lens of the information bottleneck method.