Introduction to Reinforcement Learning: Chapter 1

3-Second Blueprint (English)

1. Sutton and Barto frame reinforcement learning around an agent sending actions to an environment and receiving states and scalar rewards.
2. Core components separate immediate reward signals from value functions estimating long-term return, stochastic policies, and models.
3. Agents balance exploiting high-value states with random exploration, skipping temporal-difference updates on exploratory moves.