Initialization Heuristics for Greedy Bayesian Network Structure Learning

A popular and effective approach for learning Bayesian network structures is to perform a greedy search on the space of variable orderings followed by an exhaustive search over the restricted space of compatible parent sets. Usually, the greedy search is initialized with a randomly sampled order. In this article we develop heuristics for producing informed initial solutions to order-based search motivated by the Feedback Arc Set Problem on data sets without missing values.

Conference paper