We've learned a great deal in this chapter, from how to implement MCTS methods to play board games, to creating an advanced network to play an Atari game, and even the technology behind the famous AlphaGo system. Let's recap what we have learned.
Reinforcement learning methods have become the main tools to create AIs for playing games. Whether we are creating systems for real-life board games, or systems for video games, the fundamental concepts of policies, Q-learning, and more that we learned about in Chapter 8, Reinforcement Learning, form the basis for these complex AI systems. When we create AIs for board games, we rely on the building block of the game tree, and use MCTS to simulate various game outcomes from that game tree. For more advanced systems such as AlphaGo and chess-playing AIs, we utilize neural networks to help guide MCTS and make its simulations...
