Part 3: Applications of Dirichlet processes
Seagle Liu, Chao Xiong · 2011
In these notes, we will see examples of how the Dirichlet process can be used as a prior in classical statistical estimation problems. We will see two examples: density estimation (and its cousin, cluster analysis), and nonparametric Bayesian GLMs (which include nonparametric regression and classification as special cases). We will introduce along the way two MCMC techniques that can be used to approximate the posterior. We will also look at the question of how to use the posterior to answer the task at hand.