New Bayesian Nonparametric Mixture Models

• Activity Number: 497
• Type: Topic Contributed
• Date/Time: Wednesday, August 4, 2010 : 10:30 AM to 12:20 PM
• Sponsor: Section on Bayesian Statistical Science
• Author(s): Stephen Graham Walker*+ and Ramses Mena
• Companies: University of Kent and IIMAS-UNAM
• Keywords: Dirichlet process ; geometric weights ; density estimation

Abstract

The idea of the talk is an alternative discrete random distribution function to the Dirichlet process. Rather than eloborating my making more complex with additional parameters, we take the contrary step of making it simpler, specifically by making the weight process based on a geometric distribution. The support is not diminished by this choice and when applied to the widely used mixture model, the density estimation aspect of the model is as good as with the Dirichlet process. Developments to other aspects of Bayesian nonparametric inference based on the Dirichlet process can also be examind with this new class of random distribution function, such as the construction of dependent random distribution and density functions.