This talk on Reinforcement Learning in Games was held on Friday October 21, 2016 in MC 4020. The talk was given by Agastya Kalra.
Abstract
Outline
What is supervised machine learning?
What is the reinforcement learning problem?
Examples: Deepmind Atari, Alpha GO, TD Gammon
Basic Q Learning interactive intuition based on human brain
Formalizing Q Learning
Lets set up a problem
Scaling Q Learning interactive intuition
Improvements (if we get this far) Double Q Intuition, Experience Replay, Frozen Model, Prioritized Replay