Learning Fast-Inference Bayesian Networks

Vaidyanathan Peruvemba Ramaswamy, Stefan Szeider

Keywords: Bayesian Network Structure Learning, Exact Probabilistic Reasoning, MaxSAT, Propositional Satisfiability

TL;DR: We propose a new Max-SAT based approach for learning BNs that admit reliably fast exact probabilistic reasoning

Abstract: We propose new methods for learning Bayesian networks (BNs) that reliably support fast inference. We utilize maximum state space size as a more fine-grained measure for the BN's reasoning complexity than the standard treewidth measure, thereby accommodating the possibility that variables range over domains of different sizes. Our methods combine heuristic BN structure learning algorithms with the recently introduced MaxSAT-powered local improvement method (Peruvemba Ramaswamy and Szeider, AAAI'21). Our experiments show that our new learning methods produce BNs that support significantly faster exact probabilistic inference than BNs learned with treewidth bounds.

Code Of Conduct: I certify that all co-authors of this work have read and commit to adhering to the NeurIPS Statement on Ethics, Fairness, Inclusivity, and Code of Conduct.

Code: https://zenodo.org/record/5598257