Beyond Conjugacy for Chain Event Graph Model Selection

TL;DR

This paper introduces a non-conjugacy-based mixture modeling approach for chain event graph model selection, improving scalability and applicability to real-world data with complex priors.

Contribution

It presents a novel model selection method for chain event graphs that does not depend on conjugate priors, enhancing robustness and scalability.

Findings

Effective on simulated datasets

Outperforms conjugacy-based methods in scalability

Handles complex prior structures

Abstract

Chain event graphs are a family of probabilistic graphical models that generalise Bayesian networks and have been successfully applied to a wide range of domains. Unlike Bayesian networks, these models can encode context-specific conditional independencies as well as asymmetric developments within the evolution of a process. More recently, new model classes belonging to the chain event graph family have been developed for modelling time-to-event data to study the temporal dynamics of a process. However, existing model selection algorithms for chain event graphs and its variants rely on all parameters having conjugate priors. This is unrealistic for many real-world applications. In this paper, we propose a mixture modelling approach to model selection in chain event graphs that does not rely on conjugacy. Moreover, we also show that this methodology is more amenable to being robustly scaled than the existing model selection algorithms used for this family. We demonstrate our techniques on simulated datasets.