Roadmap of Reinforcement Learning

A roadmap of study reinforcement learning. Posts are listed from basics to modern works.

Fundamental Concepts in Markov Decision Process and Reinforcement Learning introduces the foundation of Reinforcement Learning. The following are essential to understand modern RL:

• Environmental setup;
• Rewards and discounting;
• State value, Action value, and Advantage;
• Bellman Equation;
• Monte-Carlo & Temporal difference error;

Policy Gradient Theorem is another important topic in Reinforcement Learning. It serves as the theoretical foundation to optimize parameterized policies, bridging modern AI technology and RL. It turns out that RL is empirically proved as the most efficient way to improve AI’s ability.