New structure learning algorithms and evaluation methods for large dynamic Bayesian networks

Abstract : Dynamic Bayesian networks (DBNs) are a class of probabilistic graphical models that has become a standard tool for modeling various stochastic time-varying phenomena. Probabilistic graphical models such as 2-Time slice BN (2TBNs) are the most used and popular models for DBNs. Because of the complexity induced by adding the temporal dimension, DBN structure learning is a very complex task. Existing algorithms are adaptations of score-based BN structure learning algorithms but are often limited when the number of variables is high. Another limitation of DBN structure learning studies, they use their own benchmarks and techniques for evaluation. The problem in the dynamic case is that we don't find previous works that provide details about used networks and indicators of comparison. We focus in this project on DBN structure learning and its methods of evaluation with respectively another family of structure learning algorithms, local search methods, known by its scalability and a novel approach to generate large standard DBNs and metric of evaluation. We illustrate the interest of these methods with experimental results
Document type :
Theses
Liste complète des métadonnées

Cited literature [114 references]  Display  Hide  Download

https://tel.archives-ouvertes.fr/tel-00996061
Contributor : Philippe Leray <>
Submitted on : Monday, May 26, 2014 - 10:42:13 AM
Last modification on : Thursday, April 5, 2018 - 10:36:49 AM
Document(s) archivé(s) le : Tuesday, August 26, 2014 - 10:46:54 AM

Identifiers

  • HAL Id : tel-00996061, version 1

Collections

Citation

Ghada Trabelsi. New structure learning algorithms and evaluation methods for large dynamic Bayesian networks. Machine Learning [cs.LG]. Université de Nantes; Ecole Nationale d'Ingénieurs de Sfax, 2013. English. ⟨tel-00996061⟩

Share

Metrics

Record views

585

Files downloads

1705