Scalable Bayesian Network Structure Learning with Splines

TL;DR

This paper introduces a scalable Bayesian network structure learning method that models both linear and non-linear relationships using splines, improving accuracy on large datasets.

Contribution

It presents a novel approach combining feature selection and spline-based CPD modeling within score-and-search for better structure learning.

Findings

Improves accuracy of Bayesian network structure learning.

Scales effectively to large variable sets.

Handles both linear and non-linear relationships.

Abstract

The graph structure of a Bayesian network (BN) can be learned from data using the well-known score-and-search approach. Previous work has shown that incorporating structured representations of the conditional probability distributions (CPDs) into the score-and-search approach can improve the accuracy of the learned graph. In this paper, we present a novel approach capable of learning the graph of a BN and simultaneously modelling linear and non-linear local probabilistic relationships between variables. We achieve this by a combination of feature selection to reduce the search space for local relationships and extending the score-and-search approach to incorporate modelling the CPDs over variables as Multivariate Adaptive Regression Splines (MARS). MARS are polynomial regression models represented as piecewise spline functions. We show on a set of discrete and continuous benchmark instances that our proposed approach can improve the accuracy of the learned graph while scaling to instances with a large number of variables.