Reinforcement Learning
Reinforcement Learning (RL) is a framework for online planning where an agent learns to act optimally by interacting with an environment whose transition and reward functions are initially unknown. The agent explores by taking actions, observes state transitions and rewards, and gradually learns a policy that maximizes cumulative reward.
At each timestep , each tuple is a sample, where is the state, is the action, and is the reward from the taken action). A sequence of samples until termination forms an episode.