A Parallel Framework for Constraint-Based Bayesian Network Learning

Bayesian networks (BNs) are a widely used graphical model in machine learning. As learning the structure of BNs is NP-hard, high-performance computing methods are necessary for constructing large-scale BNs. In this talk, I will present a parallel framework that we developed to scale BN structure learning algorithms to tens of thousands of variables. I will demonstrate the utility of our framework for parallelizing three different structure learning algorithms: Grow-Shrink (GS), Incremental Association MB (IAMB), and Interleaved IAMB (Inter-IAMB) and present the scalability results obtained using real and simulated data sets.