Discrete structures and prediction in Bayesian Nonparametrics
Igor Pruenster, Bocconi
12-1pm 19th May 2017
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. Several distributional properties are presented and their implications for Bayesian nonparametric inference highlighted. Illustrations in the contexts of mixture modeling, species sampling and curve estimation are provided.
Short Bio
Igor Pruenster received his Ph.D. in Mathematical Statistics from the University of Pavia, Italy, in 2003. Currently he is Professor of Statistics at Bocconi University in Milan, Italy and Director of the Bocconi Institute of Data Science and Analytics (BIDSA). His research interests are mainly related to Bayesian Nonparametrics and include random measures, prediction, asymptotics, mixture models, clustering, species sampling and survival analysis. He is author of 50+ papers in peer-reviewed journals and holds a European Research Council (ERC) grant on Bayesian Nonparametrics. He is also Fellow of the Institute of Mathematical Statistics (IMS), Co-Editor of Bayesian Analysis and Associate Editor of several other journals.