Hybrid time Bayesian networks

Capturing heterogeneous dynamic systems in a probabilistic model is a challenging problem. A single time granularity, such as employed by dynamic Bayesian networks, provides insufficient flexibility to capture the dynamics of many real-world processes. The alternative is to assume that time is conti...

Full description

Saved in:
Bibliographic Details
Published in:International journal of approximate reasoning 2017-01, Vol.80, p.460-474
Main Authors: Liu, Manxia, Hommersom, Arjen, van der Heijden, Maarten, Lucas, Peter J.F.
Format: Article
Language:English
Subjects:
Bayesian analysis
Continuous time Bayesian networks
Dynamic Bayesian networks
Dynamic systems
Dynamical systems
Dynamics
Flexibility
Formalism
Mathematical models
Probabilistic methods
Probability theory
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Capturing heterogeneous dynamic systems in a probabilistic model is a challenging problem. A single time granularity, such as employed by dynamic Bayesian networks, provides insufficient flexibility to capture the dynamics of many real-world processes. The alternative is to assume that time is continuous, giving rise to continuous time Bayesian networks. Here the problem is that the level of temporal detail is too precise to match available probabilistic knowledge. In this paper, we present a novel class of models, called hybrid time Bayesian networks, which combine discrete-time and continuous-time Bayesian networks. The new formalism allows us to more naturally model dynamic systems with regular and irregularly changing variables. We also present a mechanism to construct discrete-time versions of hybrid models and an EM-based algorithm to learn the parameters of the resulting BNs. Its usefulness is illustrated by means of a real-world medical problem. •Hybrid time Bayesian networks are defined.•This new class of models allows reasoning with random variables that evolve regularly or irregularly.•A discrete-time characterization of these new models is given.•As an application of hybrid time Bayesian networks a medical example is modeled.
ISSN:0888-613X
1873-4731
DOI:10.1016/j.ijar.2016.02.009