Topics

1

General Discussion / Free Will

« on: April 27, 2010, 18:47:54  »

Hi - is there a good definition of the 'free will' condition?

I can trap uninstantiated decision nodes in a LIMID within my codes but is this the only condition for the violation of free will?  Is there a list of routines that can fail on violation of 'free will' ... i.e. I get failures from routines such as HAPI.DconnectedNodes?

Thanks

gary

2

General Discussion / Markov Blanket

« on: March 31, 2008, 11:23:43  »

Is there a way to display or list the markov blanket for a particular node in a large network? (Parents, children and childrens parents). For a large network with lots of dependencies this is very difficult to obtain form the diagram (especially if the diagram has been learnt from data so is not displayed in a systematic fashion) or from the net file (where the info is in the potential clauses but spread throughout a large file) ?

Thanks

Gary

3

General Discussion / D connected nodes

« on: July 23, 2007, 16:37:22  »

Is it possible to be more precise about which nodes are included in the collection of D-connected (separated nodes)?

Although a Utility node can not be part of the soft/hard node list it appears that they are included in the list of Connected or Separated nodes produced by the API calls. Are these nodes treated as Chance nodes without evidence? Can a utility be considered as a source node? What is the corresponding situation for decision nodes? Does their status depend on whether the arrow enters or leaves? (i.e time order?).

Is the source node always included in the D-connected list?

I have not been able to get the d-separation analysis tool to work from the GUI for influence diagrams to check these conditions visually?

Thanks

Gary