Modeling Classes

What are generative models

This chapter is an introduction to the important topic of building generative models. These are models that are aimed at understanding the variety of a class, such as cars or trees. A generative model should be able to do the following:

• Generate feature vectors for the instances of the class they represent.
• Characterize the class with all its variations.

We’ll discuss this subject both in a Bayesian and a deep learning context and also within a supervised and unsupervised context.

The following important algorithms are discussed in this chapter:

• $k$-means clustering
• Expectation maximization $\text{(EM)}$
• Naïve Bayes
• Generative adversarial networks $\text{(GANs)}$
• Variational autoencoders $(\text{VAE})$