Fairness is crucial for neural network applications with important societal implications (such as personal credit rating). In this talk, I will present our recent research on neural network discrimination, from the perspective of evaluation, mitigation, and certification. In particular, we propose an approach to formally verify neural networks against fairness. Our method is built upon an approach for learning Markov Chains from a user-provided neural network which is guaranteed to facilitate sound analysis of global properties. We further show that through causality analysis, we can systematically identify the cause of discrimination and repair the neural network effectively and efficiently.

SUN, Jun is a professor at Singapore Management University (SMU). He received Bachelor and PhD degrees in computing science from National University of Singapore (NUS) in 2002 and 2006. In 2007, he received the prestigious LEE KUAN YEW postdoctoral fellowship in School of Computing of NUS. From 2010 to 2019, he was an Assistant/Associate Professor at Singapore University of Technology and Design (SUTD). He was a visiting scholar at MIT from 2011-2012. Since 2019, he joined SMU as an associate professor. Jun’s research interests include software engineering, formal methods, program analysis and cyber-security. He is the co-founder of the PAT model checker.