As we enter the final part of the book, our focus moves to more complex examples in the areas of generative modeling, neuroevolution of augmenting topological networks, reinforcement learning, and instinctual learning. We introduce each of these advanced topics before adding these evolutionary methods.
Chapters 8 and 9 introduce and explore the areas of generative modeling, or generative deep learning. Chapter 8 demonstrates the basic autoencoder and how it can be enhanced into an evolutionary autoencoder. Then, in chapter 9, we introduce the basics of the generative adversarial network. As generative adversarial networks are notorious for being difficult to train, we demonstrate how evolutionary methods can better optimize training.
Neuroevolution for augmenting topologies is showcased in chapters 10 and 11. We first introduce the basics of NEAT in chapter 10, with various examples of how to configure this powerful algorithm to enhance speciation. Then, in chapter 11, we apply NEAT to solve deep reinforcement learning problems found in the OpenAI Gym.