University of Cambridge > > Machine Learning @ CUED > Message-passing inference on graphical models

Message-passing inference on graphical models

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

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

Graphical models provide a powerful framework for specifying complex relationships between random variables. However performing inference on these models has remained a major challenge: exact inference is only possible for the simplest of cases and Monte Carlo methods impose large computational requirements.

An alternative approach is using approximate message-passing techniques. The most well known of these is the `loopy’ version of belief propagation, also known as the sum-product algorithm, and includes more recent developments such as generalised belief propagation (GBP) and expectation propagation (EP).

This talk will provide a brief summary of graphical models, and an overview of belief propagation and region-based message-passing methods. The focus of the talk is to present the concept of structured region graphs, which incorporates the GBP and EP algorithms and provides a useful framework for developing more efficient methods applicable to wider variety of problems.

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.


© 2006-2024, University of Cambridge. Contact Us | Help and Documentation | Privacy and Publicity