For next week, read the following paper:
Laskey KB, Myers JW (2003)
Population Markov Chain Monte Carlo
Machine Learning 50 (1-2), 175-196
As an exercise, have a look at the talk the authors gave and try to present it at our next meeting.
Familiarize yourself with Kephin Murphy's Bayes Net Toolbox for Matlab.
Take one of the example networks you find in Laskey's and Myer's talk (start with a simple one, like the College network), generate data sets of different size from this network, and try to learn it with (standard, non population) MCMC.
Our next meeting is on Thursday, 13 May at 10.00 a.m.