Bayesian Nonparametrics: Latent Feature and Prediction Models, and Efficient Inference

Add to your list(s) Download to your calendar using vCal

• Piyush Rai (University of Utah)
• Tuesday 11 October 2011, 11:30-12:30
• Engineering Department, CBL Room 438.

If you have a question about this talk, please contact Zoubin Ghahramani.

Nonparametric Bayesian approaches offer a flexible modeling paradigm for data without limiting the model-complexity a priori. The flexibility comes from the fact that the model-complexity can grow adaptively with data. The Indian Buffet Process (IBP) is an example of a nonparametric Bayesian model in which a set of observations are assumed to be generated from a small set of latent features, and the number of latent features need not be known a priori. In this talk, I will describe some of my recent work on the IBP based models; in particular, (1) A variant of the IBP which removes the independent latent features assumption, and allows the latent features to be related via a hierarchy, (2) A nonparametric Bayesian multitask learning model which uses a combination of the Dirichlet Process mixture model and the IBP as the prior distribution on the weight vectors of multiple tasks, and (3) An efficient, search-based inference method for finding an approximate MAP estimate of the latent feature assignment matrix in the IBP based models.

This talk is part of the Machine Learning @ CUED series.

This talk is included in these lists:

• All Talks (aka the CURE list)
• Biology
• CBL important
• Cambridge University Engineering Department Talks
• Centre for Smart Infrastructure & Construction
• Engineering Department, CBL Room 438
• Featured lists
• Inference Group Summary
• Joint Machine Learning Seminars
• Life Science
• Life Sciences
• Machine Learning @ CUED
• Machine Learning Summary
• Neuroscience
• Neuroscience Seminars
• Neuroscience Seminars
• Neuroscience Seminars
• Required lists for MLG
• School of Technology
• Stem Cells & Regenerative Medicine
• engineering

Note that ex-directory lists are not shown.