Towards Trustworthy Machine Learning via Distribution Matching

Abstract: Distribution matching (DM) has emerged as a cornerstone of trustworthy machine learning, finding application in fairness, robustness, causality, and explainability. Distribution matching is the task of learning a representation such that two or more distributions match (i.e., they become equal); this can also be seen as an enforcing an independence constraint on the representation. The distributions are usually defined by user-specified domain labels. Intuitively, DM is the natural complement of classification: while classification labels define what is important, domain labels define what is not important. Despite the broad applicability, research on DM remains fragmented, scattered across specific applications and methods. By unifying research along multiple dimensions, I aim to provide a cohesive perspective, present ongoing work, and outline future directions. I will anchor the unifying perspective around several concrete applications of DM. Just as classification was the cornerstone of current ML, the unification and advancement of DM could become a critical tool for enabling the next generation of trustworthy ML.

Speaker Bio: Prof. David I. Inouye is an assistant professor in the Elmore Family School of Electrical and Computer Engineering at Purdue University. His lab focuses on trustworthy machine learning (ML), which aims to make ML systems more robust, causal and explainable. Currently, he is interested in advancing distribution matching fundamentals, algorithms, and applications such as causality, domain generalization, and distribution shift explanations. He is also interested in highly robust distributed learning algorithms on a network of devices, called Internet Learning. His research is funded by ARL, ONR, and NSF. Previously, he was a postdoc at Carnegie Mellon University working with Prof. Pradeep Ravikumar. He completed his Computer Science PhD at The University of Texas at Austin in 2017 advised by Prof. Inderjit Dhillon and Prof. Pradeep Ravikumar. He was awarded the NSF Graduate Research Fellowship (NSF GRFP).