Belief Propagation

User 49 | 3/5/2014, 8:03:18 PM


I was able to run sum-product belief propagation as directed by Scott. Thanks. However, there are a couple of questions: 1. Is it possible to set a tolerance in the testcatbool_joint example? The thing is, the factor graph I am working with does not lead to convergence of bp, and it becomes essential to set a tolerance for it.

  1. I need to implement 'lifted belief propagation', in which the message computation has to be modified slightly. Could you tell me where are the messages being computed, with respect to the testcatbool_joint example?

Thanks a lot!

No Comments