Exact Structure Learning of Bayesian Networks by Optimal Path Extension

TL;DR

This paper presents a novel exact structure learning method for Bayesian networks that significantly improves computational efficiency and memory usage by leveraging relationships between partial structures and remaining variables.

Contribution

The paper introduces an innovative approach that constrains network extensions, leading to up to three times faster runtime and much lower memory consumption compared to existing algorithms.

Findings

Up to three times faster runtime.

Orders of magnitude reduction in memory usage.

Effective in handling large datasets.

Abstract

Bayesian networks are probabilistic graphical models often used in big data analytics. The problem of exact structure learning is to find a network structure that is optimal under certain scoring criteria. The problem is known to be NP-hard and the existing methods are both computationally and memory intensive. In this paper, we introduce a new approach for exact structure learning. Our strategy is to leverage relationship between a partial network structure and the remaining variables to constraint the number of ways in which the partial network can be optimally extended. Via experimental results, we show that the method provides up to three times improvement in runtime, and orders of magnitude reduction in memory consumption over the current best algorithms.