Wednesday 26 March 2003 - Expectation Propagation

Please note unusual time and day: Wednesday 11-1pm

Mark and Mike will discuss Expectation Propagation.

The recommended reading is:

Thomas Minka's thesis:
"A family of algorithms for approximate Bayesian inference"

If that is too long, a terse version of the above thesis is: Expectation Propagation for Approximate Bayesian Inference. You might also like to see the rest of Thomas Minka's EP page

Erik Sudderth from MIT has some nice slides:EP slides and EP summary.

Contact Iain if you have problems obtaining the reading.