by jdowner on 12/26/18, 3:06 PM with 0 comments
I have been interested in better understanding the belief propagation algorithm for a while. There are some good resources out there but nothing has really filled in all the details for me. I guess the thing that I would really like to better understand is how this would be implemented in code. Christopher Bishop has a good explanation of how it works in his book 'pattern recognition and machine learning' but I can't quite wrap my head around the notion of passing functions as messages. One thing I am not sure of: is this message passing of functions merely descriptive or is that what you explicitly need to do?
Any help is greatly appreciated!