Posted by Laura on November 17, 2000 at 00:38:13:
I was trying to integrate Ben and my code this afternoon, and I was getting different values for Asia that Ch. 7 of Nepolotion says. I believe this is because when I use the carditality serach algorithm I get a different alpha, which in turn results in a different ordering of the cliques. From what I understand about the message passing part of the algorithm I belive that the problem will go away. I'm not 100% sure of that, but if not, I have an idea on how to perform the cardinality search to select the correct alpha, in the example they seem to pick the nodes in a certian pattern, based on parents and some other things, even though it says that it is random in the algorithm. But I will know for certian when we can test the entire itinery.