Learning K2 and metaheuristics

[kile]

Hi all!

I'm making some experiments on mixing ants colony metaheuristics to create Bayesian Network and I found your website.
I was looking throw the forum and documentation and I dont get exactly what is going with your greedy implementation of K2.

My problem is that I would like to create a ordering of the initial graph using metaheuristics, and after I would like to give this ordering with the dataset to one functions to learn the structure from that. So I could go testing how good is the graph and changing the metaheuristic fitness function.

There're any possibility to do something like that with your library? Any suggestions are welcome.

thank u very much

[shooltz[BayesFusion]]

My problem is that I would like to create a ordering of the initial graph using metaheuristics, and after I would like to give this ordering with the dataset to one functions to learn the structure from that. So I could go testing how good is the graph and changing the metaheuristic fitness function.

You can achieve that by using DSL_greedyThickThinning::tiers:

http://genie.sis.pitt.edu/wiki/Referenc ... ckThinning

This functionality corresponds to tiers in GeNIe's background knowledge editor.

[kile]

Ok thank u very much I'll try to use it