Reinforcement Learning Overview
This section provides a high-level perspective on the reinforcement learning landscape, helping readers build a comprehensive knowledge framework spanning classical methods to cutting-edge applications.
Contents
| Topic | Description |
|---|---|
| RL Landscape | The RL methodology taxonomy: from Bellman equations to RLHF, covering model-free/model-based, on-policy/off-policy, single/multi-agent dimensions |
| RL Milestones | Key breakthroughs in RL history: TD-Gammon, DQN, AlphaGo, ChatGPT, o1, and other landmark achievements |
Recommended Reading Order
- Start with RL Landscape to build a methodological framework
- Then read RL Milestones to understand the historical trajectory
- Dive into specific topic chapters based on your interests
Related Sections
- Classic RL — MDP, Dynamic Programming, Monte Carlo, Temporal Difference
- Deep RL — DQN, PPO, SAC, etc.
- Policy Gradient — Policy gradient methods in detail
- Multi-Agent RL — MARL methods and applications