Along with generative networks, reinforcement learning algorithms have provided the most visible advances in Artificial Intelligence (AI) today. For many years, computer scientists have worked toward creating algorithms and machines that can perceive and react to their environment like a human would. Reinforcement learning is a manifestation of that, giving us the wildly popular AlphaGo and self-driving cars. In this chapter, we'll cover the foundations of reinforcement learning that will allow us to create advanced artificial agents later in this book.
Reinforcement learning plays off the human notion of learning from experience. Like generative models, it learns based on evaluative feedback. Unlike instructive feedback in supervised learning where the network learns by us telling it how to do something, evaluative feedback helps algorithms learn...
