ISE Seminar: Archis Ghate

Details

100 Williamson Hall (in-person) or Zoom.

Archis Ghate
Professor and Department Head
Industrial and Systems Engineering
University of Minnesota

Title:
Inverse Markov Decision Processes

Abstract:
Inverse optimization involves finding parameter values that would render given values of decision variables optimal. This is in contrast with the usual (forward) optimization where the goal is to compute optimal values of decision variables using given values of parameters. This talk will focus on inverse optimization in Markov decision processes (MDPs) for imputing rewards and transition probabilities that make a given policy optimal. We will begin with a review of prior work on imputing rewards and then formulate the problem for imputing transition probabilities. While both problems rely on mathematical programming representations of Bellman’s equations of dynamic programming, we will explain why it is easier to impute rewards than to impute transition probabilities. Specifically, while the former problem is linear, the latter is bi-linear. We will present exact and approximate solution methods for different versions of this bilinear problem. We will show how to extend some of this work to continuous-time MDPs via a technique called uniformization. A further extension to semi-Markov decision processes will be described.

Biography:
Archis is a Professor and Department Head of Industrial and Systems Engineering at the University of Minnesota. Previously, he was a Professor of Industrial Engineering and held the Fluor Endowed Chair at Clemson University. Prior to joining Clemson, he was a Professor of Industrial & Systems Engineering at the University of Washington in Seattle. There he served as the Associate Chair for six years and held a College of Engineering Endowed Professorship for five years. He joined the University of Washington as an Assistant Professor after receiving a PhD in Industrial and Operations Engineering from the University of Michigan in 2006, and an MS in Management Science and Engineering from Stanford in 2003. He completed his undergraduate education at the Indian Institute of Technology, Bombay in 2001. Archis is a recipient of the NSF CAREER award. He has also won an award for excellence in teaching Operations Research and a best paper award from IISE. He has received multiple teaching accolades from the University of Washington, and his students have won the Dantzig dissertation award and the Bonder scholarship in healthcare operations research from INFORMS. Archis has served on the editorial boards of several journals. He was the General Chair of the INFORMS 2019 Annual Meeting, and a Program Co-Chair of the 2021 IISE Annual Conference.