DMS Colloquium: Professor Samy Tindel

Speaker: Professor Samy Tindel (Purdue University)

Title: Some rough paths techniques in reinforcement learning

Abstract: In this talk, I will start by reviewing some classical results relating machine learning problems with control theory. I will mainly discuss basic notions of supervised learning, as well as reinforcement learning. I will then show how noisy environments lead to very natural equations involving rough paths. This will include a couple of motivating examples. In a second part of the talk, I will explain the techniques used to solve noisy reinforcement learning problems with a minimal amount of technicality. In particular, I will focus on rough HJB type equations and their respective viscosity solutions. If time permits, I will also give an overview of our current research program in this direction.  

This talk is based on ongoing joint work with Prakash Chakraborty from Penn State and Harsha Honnappa from Purdue's Industrial Engineering department.

Short bio: Professor Tindel obtained his Ph.D. in 1996 in France and began his academic career in France before moving to the USA to become a faculty member at Purdue University in 2015. He has been very successful in obtaining grants, securing his first NSF standard grant in 2016, which he renewed in 2020. In 2022, he obtained a joint multidisciplinary grant for the project presented in this talk.

Faculty host: Le Chen