Chapter 10: Evolving AI Using ML-Agents
Machine Learning (ML) has evolved drastically over the last decade. It is no longer solely the domain of scientists or companies making headlines beating world chess champions. Now, we can easily write AI that can master most human components for a wide variety of games. We can even write AI that will learn how to play a game using information only available to a human player (MarI/O comes to mind, an ML algorithm that taught itself to play Super Mario). I remember, many years ago, spending months writing my own evolutionary learning algorithms to do something very similar to what you will accomplish in this very chapter (and hopefully it won't take months to complete!). The best news is, we don't have to write our own ML algorithms (although this is still possible if that's the route you want to take!). Instead, we will use ML-Agents to create an intelligent character (or Agent).
ML-Agents is short for Machine Learning Agents...