Thang D. Chu

I am an M.Sc. student in Computing Science at the University of Alberta, advised by Prof. Csaba Szepesvári. My thesis is Lower Bound on Policy Gradient with Decreasing Stepsizes, with a focus on lower bounds for policy-gradient methods that use decreasing stepsizes.

My research interests are reinforcement learning theory, stochastic optimization, and continual learning. More broadly, I am interested in rigorously understanding when learning algorithms work, when they fail, and how their optimization dynamics shape their statistical performance.