T. Singliar, D. H. Dash: Efficient inference in persistent Dynamic Bayesian Networks

Hi, and thanks for your interest in our paper. We put stuff here that could not fit in the paper but helps elucidate what we are talking about.

First, here's the paper.

Also, you might find it helpful to look at the slides or an early poster that describes this work. Better yet, there is a larger and better version .

Here is the inference code. We don't have a neat manual yet; please contact Tomas <singliar(tadpole)gmail.com> for assistance if you have any trouble with it.

The relative performance advantage of PPT-DBN over Boyen-Koller and similar grows with the size of the system. Here is an experiment that shows what happens when there are about 50 nodes in the system.