Learning a maximum-scored Bayesian network structure with bounded treewidth is rather popular in recent years, as stated in the paper A survey on Bayesian network structure learning from data in 2019. However, I failed to find out any recent research on the structural learning of a maximum-scored Bayesian network with bounded number of parents for each variable(node). D.M. Chickering(1996) formalised this problem as $K$-LEARN where $K$ is the upper bounded for number of parents and proved its NP-Completeness in Learning Bayesian Networks is NP-Complete. Since then, to the best of my knowledge, far less attention is paid to $K$-LEARN in the community(in fact, I couldn't find out any related paper), compared with its popular treewidth-bounded version.

So is $K$-LEARN not considered as a good research topic anymore or it has never been the case? Or I just missed the whole line of research for it since 1996? Any help would be appreciated.

Note: The treewidth of a Bayesian network is defined as the treewidth of its moralised graph, by Elidan and Gould in Learning Bounded Treewidth Bayesian Networks(2008)


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.