Model Selection for Gaussian Mixture Models

Speaker:Peng Heng, Associate Professor, Mathematics Department of Hong Kong Baptist University

Title: Model Selection for Gaussian Mixture Models

Date&Time:April 28th, 2015 3:30pm-4:30pm

Location:Lecture Hall, Floor 4th, Mathematics and Physics Building

Abstract:An important issue in finite mixture modeling, the selection of the number of mixing components, is concerned. A new penalized likelihood method is proposed for finite multivariate Gaussian mixture models, and is shown to be statistically consistent in determining the number of components. A modified EM algorithm is developed to select the number of components, and simultaneously to estimate the mixing probabilities and the unknown parameters of Gaussian distributions. Simulations and a real data analysis are presented to illustrate the performance of the proposed method.

About the speaker:

Peng Heng, associate professor of Hong Kong Baptist University, his main research directions are: non parametric and semi parametric regression, high dimensional data analysis and biological statistics. Professor Peng obtained a PhD degree in Statistics at the Chinese University of Hong Kong in 2003. From 2003 to 2006 he was in post doctoral researches of statistics in Princeton University. From 2006 to the present, he works in Mathematics Department of Hong Kong Baptist University. So far Professor Peng has published more than 30 high level academic dissertations on worldwide reputed statistics journals like “The Annals of Statistics”, “Biometrika”, “Journal of the American Statistical Association”, etc.