What this book covers

Chapter 1, Introducing Machine Learning and ML-Agents, covers the basics of machine learning and introduces the ML-Agents framework within Unity. This is basically just a setup chapter, but it's essential to anyone new to Unity and/or ML-Agents.

Chapter 2, The Bandit and Reinforcement Learning, introduces many of the basic problems and solutions used to teach reinforcement learning, from the multiarm and contextual bandit problems to a newly-derived connected bandit problem.

Chapter 3, Deep Reinforcement Learning with Python, explores the Python toolset available for your system and explains how to install and set up those tools. Then, we will cover the basics of neural networks and deep learning before coding up a simple reinforcement learning example.

Chapter 4, Going Deeper with Deep Learning, sets up ML-Agents to use the external Python trainers to create some fun but powerful agents that learn to explore and solve problems.

Chapter 5, Playing the Game, explains that ML-Agents is all about creating games and simulation in Unity. So, in this chapter, we will focus on various play strategies for training and interacting with agents in a real game or simulation.

Chapter 6, Terrarium Revisited and a Multi-Agent Ecosystem, revisits a coding game developed previously called Terrarium as a way to build self-learning agents who live in a little ecosystem. We learn how game rules can be applied to building a game or simulation with multiple agents that interact together.