CISE Seminar: Akshay Krishnamurthy, Principal Researcher, Microsoft Research

March 25, 2022
3:00PM-4:00PM
Hybrid Event
Photonics Center (PHO), 8 St. Mary’s St, Room 210

Representation Learning, Exploration, and Reinforcement Learning

I will discuss new provably efficient algorithms for reinforcement in rich observation environments where the agent operates on complex inputs like images or text. These are challenging settings that require the algorithm to address generalization, exploration, and credit assignment simultaneously. Our algorithms address the generalization challenge by learning a succinct representation of the environment, which they use in an exploration module to acquire new information, thereby addressing exploration and credit assignment.  In the talk, I will cover three algorithms with this flavor that vary in both representation learning and exploration subroutines and come with distinct advantages and disadvantages. All algorithms accommodate nonlinear function approximation and enjoy provable sample and computational efficiency guarantees. I will also present some empirical results that highlight the promise of this broader approach.

Akshay Krishnamurthy is a principal researcher at Microsoft Research, New York City. Previously, he spent two years as an assistant professor in the College of Information and Computer Sciences at the University of Massachusetts, Amherst and a year as a postdoctoral researcher at Microsoft Research, NYC. He completed his PhD in the Computer Science Department at Carnegie Mellon University. His research interests are broadly in machine learning and statistics, and more specifically in interactive learning or learning settings that involve feedback-driven data collection. Recently his work has focused on decision making problems with limited feedback, including contextual bandits and reinforcement learning.

Faculty Host: Venkatesh Saligrama
Student Host: Yuhe Chang