Fine tuning the Naïve Bayesian learning algorithm

Abstract

This work augments the Naive Bayesian learning algorithm with a second training phase in an attempt to improve its classification accuracy. This is achieved by finding more accurate estimations of the needed probability terms. This approach helps in dealing with the problem of the lack of training data. Unlike many previous approaches that deal with this problem, the proposed method is an eager method in the sense that it does most of the work during training and, therefore, it does not increase classification time. It consists of two phases. In the first phase, the algorithm builds a classical Naive Bayesian classifier. The second phase is a fine tuning phase. In this phase each training instance is classified, if it is misclassified, the probability values involved are fine tuned in such a way that increases the chances of correctly classifying this instance in the next round. Our results show significant improvement in the classification accuracy of many benchmark data sets, compared to the classical Naive Bayesian, and two other methods that improve on the Naive Bayesian algorithm.