ABSTRACT

The Dirichlet process is a infinite-dimensional generalization of the Dirichlet distribution that can be used to set a prior on unknown distributions. Furthermore these unknown densities can be used to extend finite component mixture models to infinite component mixture models.