Q-Learning: A model-free reinforcement learning algorithm that estimates the optimal action-value function using a table or a neural network.