Policy-Based & Actor-Critic
A 12-module path through policy-based & actor-critic, in study order from foundations to the frontier.
Modules
- Policy Parameterization and Objectives
- The Policy-Gradient Theorem
- REINFORCE: Monte Carlo Policy Gradient
- Baselines and Variance Reduction
- Actor-Critic Methods
- The Advantage Function
- Generalized Advantage Estimation
- Trust-Region Policy Optimization
- Proximal Policy Optimization
- Entropy Regularization
- Continuous Action Spaces and Gaussian Policies
- Deterministic Policy Gradients