Deep learning has very interesting contributions to the general machine learning community, particularly when it comes to deep discriminative and generative models. We are familiar with what a discriminative model is—for example, a Multilayer Perceptron (MLP) is one. In a discriminative model, we are tasked with guessing, predicting, or approximating a desired target, 
, given input data 
. In statistical theory terms, we are modeling the conditional probability density function, 
. On the other hand, by a generative model, this is what most people mean:
A model that can generate data that follows a particular distribution based on an input or stimulus 
.
In deep learning, we can build a neural network that can model this generative process very well. In statistical terms, the neural model approximates the conditional probability density function, 
. While there are several generative models today, in this book, we will talk about three in particular...
