Off-Policy Policy Optimization

Explore off-policy policy optimization in reinforcement learning with Dale Schuurmans from Google Brain and the University of Alberta in this 53-minute lecture. Delve into key concepts including the RL problem, batch policy optimization, and optimization objectives. Compare supervised and reinforcement learning approaches, and examine missing data inference in the context of sequential decision making. Gain insights into the emerging challenges in deep learning as applied to reinforcement learning algorithms and policy optimization techniques.

Intro
The RL problem
Batch policy optimization
Optimization objectives
Supervised vs reinforcement learning
Missing data inference
Sequential decision making
Sequential RL