Subjects / Reinforcement learning

Best books to learn Reinforcement learning, in order

RL is unusually theory-dependent: the math of value and policy comes before any code, or the algorithms feel arbitrary. Start with Markov decision processes and dynamic programming, then tabular methods (Q-learning, temporal-difference), then function approximation and deep RL—reaching policy gradients and actor-critic only once you can reason about exploration, credit assignment, and why training is so unstable.

Build your own Reinforcement learning list →Browse all paths