Figure 2. Reinforcement Learning Overview.
An agent (computer algorithm) observes state information and rewards provided by the environment and selects actions to interact with it. This information is processed to maximize rewards through the desired operation.