Bayesian modeling with nested random partitions

This talk discusses a Bayesian nonparametric multivariate mixture model where the prior on the latent distribution extends the popular Dirichlet process in allowing random nested partitions, while maintaining a simple predictive rule.