Model‑free reinforcement learning algorithm that learns action‑value functions to maximize cumulative reward.