Learning huge Augmented Naive Bayes Classifier

Abstract

For classification problems, Bayesian networks are often used to infer a class variable when given feature variables. Earlier reports have described that classification accuracies of exact learning augmented naive Bayes (ANB) achieved by maximizing the marginal likelihood (ML) were higher than the Bayesian network of the identification model However, the method cannot learn structures that have more than several dozen variables. To resolve this difficulty, this study proposed exact learning ANB using RAI algolithm. The experimental results show that the proposed method outperforms the other methods.