Dear UAIers,
I wonder if someone could shed light on the following
mathematical question:
Given:
- a Bayesian network with N nodes (possibly with some parameters
to indicate how dense the network is, the average number of values
for each variable, etc);
- a number M of observed nodes (possibly with some parameters to
indicate distribution of observations, etc);
- a query consisting of K nodes.
Wanted:
- the average number of nodes that are d-separated from the query
nodes by the observed nodes.
Is this a difficult mathematical problem or is this
something simple? Maybe this problem has been solved
already; I would be grateful for any pointers.
Thanks,
Fabio Cozman
