Statistics seminar 2016: "Are Gibbs-type priors the most natural generalization of the Dirichlet process?"

Igor Pruenster
University of Torino & Collegio Carlo Alberto

Abstract
Discrete random probability measures and the exchangeable random partitions they induce are key tools for addressing a variety of estimation and prediction problems in Bayesian inference. In this talk we focus on the family of Gibbs-type priors, a recent elegant and intuitive generalization of the Dirichlet and the Pitman-Yor process priors.
We highlight their implications for Bayesian nonparametric inference and provide evidence that they indeed represent the most natural generalization of the Dirichlet process.

Contact person
Alessandra Luati