Naive Bayes Based on Estimation of Distribution Algorithms for Classification

The Nai¿ve Bayesian Classifier (NBC) is a simple effective technique for classifier algorithm. But the independence assumptions of the attribute affect its classification performance, and restrict its application. Here a method based on estimation of distribution algorithms is presented and it further enhances Nai¿ve Bayesian Classifier through using estimation of distribution algorithms. It makes the train set excellent by using estimation of distribution algorithms, and constructs Nai¿ve Bayesian classifiers, then boosts performance by using bagging. Compared with the other attempts to improve the simple Bayes, its results demonstrate excellent performance in experiments.