In the chapter 8, Reinforcement Learning Theory, we introduced Reinforcement Learning (RL), a way to make a computer interact with an environment. In this chapter, we'll build upon that knowledge and we'll explore some more advanced RL algorithms and tasks. But don't worry, we won't create the Terminator just yet. We're aiming a little lower, so we'll just see how to teach a machine to play games such as Atari Breakout and Go.
This chapter will cover the following:
- Introduction to genetic algorithms playing games
- Deep Q-learning (DQN)
- Policy gradients
- Actor-critic methods
- Monte Carlo tree search