Dear administrator,
I need to build a hybrid dynamic Bayesian network in a recent study. But I found that neither GeNIe nor SMILE seem to support the hybrid dynamic Bayesian model. For example, I have a simple hybrid dynamic Bayesian network blow：
Node A, B, and C are all continuous variables (equation node). As shown in the figure, node C has an influence on node B in the next time step. However, such a relationship between node A and node B does not seem to be editable in the definition of the equation node.
Does anyone know how to build this type of model? Any help will be greatly appreciated.
Best regards
Jeremy
Hybrid dynamic bayesian networkAn inquiry about developing HDBN model using GeNIe

 Site Admin
 Posts: 1202
 Joined: Mon Nov 26, 2007 5:51 pm
Re: Hybrid dynamic bayesian networkAn inquiry about developing HDBN model using GeNIe
As you noticed, GeNIe does not support hybrid dynamic models yet.
The best approximation would be to use an unrolled structure based on a discrete dynamic model as a starting point. The unrolled model is just a plain Bayesian network, which can be modified by changing types of some nodes from CPT to equation.
The best approximation would be to use an unrolled structure based on a discrete dynamic model as a starting point. The unrolled model is just a plain Bayesian network, which can be modified by changing types of some nodes from CPT to equation.
Re: Hybrid dynamic bayesian networkAn inquiry about developing HDBN model using GeNIe
Dear shooltz
Thank you for your reply and help.
Best regards
Jeremy
Thank you for your reply and help.
Best regards
Jeremy