A problem that reinforcement learning can solve is a type of situation where there is no complete knowledge about the target assumption and the situation is likely to change based on the action toward the target. A subject to pursue the correct goal is called an agent. The agent tries to achieve the goal under the given environment. Imagine a situation where a person is drifting in an uninhabited island. Although the person tries to escape from the island and get back home, they do not have the complete knowledge to do so. It is necessary for them to seek the solution alongside exploring the environment, such as collecting food and building a temporary house on the island. The following diagram is an example of the problem reinforcement learning attempts to solve. The person is an agent and the environment is the island in this case:
What should an agent...
