Reinforcement learning studies how an agent learns to act through interaction with an environment.
It becomes especially relevant when prediction alone is not enough and the goal is sequential decision-making under uncertainty.
Typical themes include policies, value functions, exploration, credit assignment, and planning.