Actor Critic

Explore the foundations of actor-critic methods in reinforcement learning through this comprehensive 35-minute lecture. Delve into key concepts such as the Stochastic Gradient Policy Theorem, REINFORCE Algorithm with baseline, and Temporal Difference updates. Examine performance comparisons and gain insights into advanced techniques like Advantage Actor Critic (A2C), Continuous Actions, and Deterministic Policy Gradient (DPG). Enhance your understanding of reinforcement learning algorithms and their applications in this informative session led by Pascal Poupart.

Intro
Outline
Stochastic Gradient Policy Theorem
REINFORCE Algorithm with a baseline
Performance Comparison
Temporal difference update
Actor Critic Algorithm
Advantage update
Advantage Actor Critic (A2C)
Continuous Actions
Deterministic Policy Gradient (DPG)