In this chapter, we learned the important foundations of reinforcement learning, one of the most visible practices in the AI field.
Reinforcement learning is based on the concepts of agents acting in an environment and taking action based on what it sees in its surrounding environment. An agent's actions are guided by either policy optimization methods or dynamic programming methods that help it learn how to interact with its environment. We use dynamic programming methods when we care more about exploration and off-policy learning. On the other hand, we use policy optimization methods when we have dense, continuous problem spaces and we only want to optimize for what we care about.
We'll look at several different real-world applications of reinforcement learning in the upcoming chapter.