Learning the Structure of Dynamic Probabilistic Networks

TL;DR

This paper presents methods for learning the structure of dynamic probabilistic networks from data, extending scoring rules and search algorithms to handle hidden variables, with applications in behavior prediction and biological causal learning.

Contribution

It introduces structure learning techniques for dynamic probabilistic networks, including handling hidden variables and applying to real-world dynamic systems.

Findings

Effective structure learning from data demonstrated in two applications.

Extended scoring rules for dynamic networks.

Successful handling of hidden variables in structure search.

Abstract

Dynamic probabilistic networks are a compact representation of complex stochastic processes. In this paper we examine how to learn the structure of a DPN from data. We extend structure scoring rules for standard probabilistic networks to the dynamic case, and show how to search for structure when some of the variables are hidden. Finally, we examine two applications where such a technology might be useful: predicting and classifying dynamic behaviors, and learning causal orderings in biological processes. We provide empirical results that demonstrate the applicability of our methods in both domains.