Joint Probability of a subset of non-evidence nodes

The engine.
Post Reply
vikramdoshi
Posts: 9
Joined: Mon Mar 04, 2013 10:41 am

Joint Probability of a subset of non-evidence nodes

Post by vikramdoshi »

Hi

I wanted to compute the JP for a subset of the non-evidence nodes. (Similar to MAP, But I dont want only the Maximum, but all)

Since in the GetJPts we may ''get more than one JPT for given node, depending on the triangulation performed during jointree construction''
Can we somehow restrict it to only the nodes of interest to us? (since sometimes it contains more nodes, sometimes less)

Thanks

Regards
Vikram
shooltz[BayesFusion]
Site Admin
Posts: 1417
Joined: Mon Nov 26, 2007 5:51 pm

Re: Joint Probability of a subset of non-evidence nodes

Post by shooltz[BayesFusion] »

vikramdoshi wrote:Can we somehow restrict it to only the nodes of interest to us? (since sometimes it contains more nodes, sometimes less)
When JPT is a superset of your nodeset you can marginalize. Otherwise, you can try setting the explicit targets on the nodes of interest, but this is not guaranteed to give you the required JPT. You can also add an artificial node, which is a child of your nodes. There's no guarantee this will work - the triangulation and the preceeding relevance layer is free to perform any simplifications before clustering runs.
Post Reply