Hello,
I am planning to use GeNIe for learning the structure and parameters of a BN. I noticed that the Bayesian search learning algorithm has recently been added to the software. Is there any particular documentation or publication about the Bayesian search implemented in GeNIe I could refer to?
Thanks in advance.
Bayesian search learning algorithm
Re: Bayesian search learning algorithm
Hi,
There's useful information in the SMILE documentation, http://genie.sis.pitt.edu/wiki/Reference_Manual:_DSL_bs
You'll find explanation of the parameters that can be set there.
Basically the algorithm does a hill climbing search with random restarts.
The BDeu metric is used for the scoring of the candidate network structures.
The EM algorithm is used for learning the parameters.
There are not really any specific references to look at, but I would suggest reading "A Tutorial on Learning With Bayesian Networks" by David Heckerman.
http://research.microsoft.com/apps/pubs ... x?id=69588
Martijn
There's useful information in the SMILE documentation, http://genie.sis.pitt.edu/wiki/Reference_Manual:_DSL_bs
You'll find explanation of the parameters that can be set there.
Basically the algorithm does a hill climbing search with random restarts.
The BDeu metric is used for the scoring of the candidate network structures.
The EM algorithm is used for learning the parameters.
There are not really any specific references to look at, but I would suggest reading "A Tutorial on Learning With Bayesian Networks" by David Heckerman.
http://research.microsoft.com/apps/pubs ... x?id=69588
Martijn