Reinforcement Learning
Learning from consequences rather than labels: an agent acts, the world answers with a reward, and the agent slowly learns what each move is worth.
In this chapter
■□□ Foundations ■■□ Practitioner ■■■ Research- § 1 Markov Decision Processes In preparation Level: Foundations
- § 2 Q-Learning In preparation Level: Practitioner
- § 3 Policy Gradients In preparation Level: Practitioner
- § 4 Learning from Human Feedback In preparation Level: Research
Topics: MDPs · Q-learning · Policy gradients · RLHF · Formats: Plates · Simulations · Exercises