University of Cambridge > > Machine Learning @ CUED > Bayesian nonparametric methods for non-exchangeable data

Bayesian nonparametric methods for non-exchangeable data

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

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

In this talk I will present a flexible framework to incorporate side-information into Bayesian nonparametric latent variable models. Latent variable models have become increasingly popular in machine learning, examples include mixture models, topic models, and latent feature models. Bayesian nonparametric methods allow the specification of latent variable models that can learn an appropriate dimensionality from the observed data.

The proposed framework has nice analytic properties, admits a simple inference algorithm, and extends previous work in the field. I apply the framework to both a latent feature model and a topic model (applied to the well known corpus of State of Union Addresses) and demonstrate that incorporating side-information improves predictive performance relative to exchangeable versions of the models.

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

Tell a friend about this talk:

This talk is included in these lists:

Note that ex-directory lists are not shown.


© 2006-2023, University of Cambridge. Contact Us | Help and Documentation | Privacy and Publicity