In this chapter, we're going to be using deep neural networks in a slightly different way. Rather than predicting the membership of a class, estimating a value, or even generating a sequence, we're going to be building an intelligent agent. While the terms machine learning and artificial intelligence are often used interchangeably, in this chapter we will talk about an artificial intelligence as an intelligent agent that can perceive it's environment, and take steps to accomplish some goal in that environment.

Imagine an agent that can play a strategy game such as Chess or Go. A very naive approach to building a neural network to solve such a game might be to use a network architecture where we one hot encode every possible board/piece combination and then predict every possible next move. As massive and complex as that network would be, it probably wouldn't do a very good job. To play Chess well, you have to consider not only your next move, but the moves that follow. Our intelligent agent is going to need to consider the optimal next move given future moves, in a non-deterministic world.

This is an exciting field. It's in this domain of intelligent agents that researchers are making progress towards artificial general intelligence or strong AI, which is the lofty goal of creating intelligent agents that can perform any intellectual task that a human can. This notion of strong AI is typically contrasted with weak AI, which is the ability to solve some single task or application.

This chapter is going to be a challenge for both the author (me) and the readers (you) because reinforcement learning deserves it's own book and needs to summarize work done on math, psychology, and computer science. As such, please forgive the quick reference treatment and know that I'm attempting to give you exactly enough and not a drop more in the coming sections.

Reinforcement learning, Markov Decision Processes, and Q-learning are the building blocks to an intelligent agent, and we will talk about those next.

We will discuss the following topics in this chapter:

• Reinforcement learning overview
• Keras reinforcement learning framework
• Building a reinforcement learning agent in Keras