A Constraint-Based Algorithm for the Structural Learning of Continuous-Time Bayesian Networks

TL;DR

This paper introduces the first constraint-based algorithm for learning the structure of continuous-time Bayesian networks, analyzing its statistical, computational, and empirical performance compared to score-based methods.

Contribution

It presents a novel constraint-based structure learning algorithm for continuous-time Bayesian networks, filling a gap in existing discrete-time-focused research.

Findings

Constraint-based method is more accurate with multi-valued variables.

Score-based method is more accurate with binary variables.

Both methods have comparable computation times.

Abstract

Dynamic Bayesian networks have been well explored in the literature as discrete-time models: however, their continuous-time extensions have seen comparatively little attention. In this paper, we propose the first constraint-based algorithm for learning the structure of continuous-time Bayesian networks. We discuss the different statistical tests and the underlying hypotheses used by our proposal to establish conditional independence. Furthermore, we analyze and discuss the computational complexity of the best and worst cases for the proposed algorithm. Finally, we validate its performance using synthetic data, and we discuss its strengths and limitations comparing it with the score-based structure learning algorithm from Nodelman et al. (2003). We find the latter to be more accurate in learning networks with binary variables, while our constraint-based approach is more accurate with variables assuming more than two values. Numerical experiments confirm that score-based and constraint-based algorithms are comparable in terms of computation time.