Title: Formal Verification of Reinforcement Learning
Speaker:  Mark Chevallier
Abstract: 

This talk is in preparation for the one I will give at FMAI 21. I will discuss my work formalising reinforcement learning and the progress I have made in doing so. My methods used to model Markov Decision Processes with rewards, the main results achieved with those methods, and my recent work on vector translation of functions on MDPs will all be featured.