Reinforcement learning

Reinforcement learning is a type of machine learning in which an agent learns from the environment. The agent takes actions and, as a result of the actions, the environment returns observations and rewards. From the observation and rewards, the agent learns the policy and takes further actions, thus continuing the sequence of actions, observations, and rewards. In the long run, the agent has to learn the policy such that, when it takes actions based on the policy, it does so in such a way as to maximize the long-term rewards.