Bayesian Network Structural Learning Using Tabu Search

This paper describes the theory and implementation of Bayesian networks structural learning using tabu search algorithm. Bayesian networks providea very general and yet effective graphical language for factoring joint probability distributions. Finding the optimal structure of Bayesian networks from data hasbeen shown to be NP-hard. In this paper, tabu search has been developed to provide more efficient structure. We implemented structural learning in Bayesian networks in the context of data classification. For the purpose of comparison, we considered classification task andapplied general Bayesian networks along with this classifier to some databases. Our experimental results show that the Tabu search can find the good structure with the less time complexity. The simulation results approved that using Tabu search in order to findBayesian networks structure improves the classificationaccuracy.

Keywords:

Graphical Models , Bayesian Networks , Structural learning , Machine learning , Tabu Search