In Chapter 4, Training Neural Networks, we looked at what an artificial neural network (ANN) is and how this kind of model is built. You can say that a deep neural network is an elongated version of an ANN; however, it has got its own set of challenges.

In this chapter, we will learn about the following topics:

• What is a deep neural network?
• How to initialize parameters
• Adversarial networks—generative adversarial networks and Bayesian generative adversarial networks
• Deep Gaussian processes
• Hinton's Capsule network