Causality-based Fair Machine Learning for Sequential Decision Making

Speaker: Lu Zhang (University of Arkansas)

Dr. Lu Zhang is an assistant professor in the Computer Science and Computer Engineering Department at the University of Arkansas. He received a BEng degree in computer science and engineering from the University of Science and Technology of China in 2008, and a Ph.D. degree in computer science from the Nanyang Technological University in 2013. His research interests lie in the field of data mining, machine learning, and artificial intelligence, particularly in fair machine learning, causal modeling and inference, and robust natural language processing. He has published more than 30 papers in premier conferences and journals in data mining and artificial intelligence including NeurIPS, SIGKDD, AAAI, IJCAI, IEEE Transactions on Knowledge and Data Engineering, IEEE/ACM Transactions on Computational Biology and Bioinformatics, etc. He is a recipient of the NSF CAREER award.

Abstract

Abstract: Fair machine learning aims to build machine learning models such that the predicted decisions made are not subject to discrimination and satisfy fairness requirements. Current fair machine learning literature is mainly focused on static settings where one-shot decisions are made. However, in practical situations, decision-making is more of a sequential nature where each decision may have an impact on subsequent decisions. In this talk, I will discuss the application of Pearl’s Structural Causal Model (SCM) as a general framework and a prime methodology for the study of fair sequential decision-making. I will present results on three situations in sequential decision-making: (1) multiple related decision models at different stages are learned in a partial order; (2) a single decision model is executed repeatedly and could reshape environments through feedback loops; and (3) online recommendation where customers arrive in a sequential and stochastic manner.