University of Cambridge > > Machine Learning @ CUED > Exclusive Pólya Urns and their applications

Exclusive Pólya Urns and their applications

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

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

The Dirichlet Process (DP) and its variants have many nice properties which make them popular tools in the machine learning community. For some applications, however, these processes lack certain desirable features, such as support for fast and exact inference, or being easy to interface to an arithmetic coder. Common deployment typically involves approximate inference methods, which either distort the true posterior distribution or come at a high computational cost. I will show that by dropping some of the “nice” mathematical properties of the DP, we can instead construct a novel kind of stochastic process which allows exact Bayesian inference, is fast, easy to implement, and can in many cases be used as a drop-in replacement for Chinese Restaurant Processes and Pitman-Yor processes. I will explain how the new process differs from models in the existing literature, and how it can be applied to interesting tasks, including hierarchical sequence modelling and data compression.

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