Deep multi-classifier learning for medical data sets

Abstract

This paper presents a comparison among the different classifier such as Sequential Minimal Optimization (SMO), decision tree (J48), random forests (RFs), Naïve Bayes (NB) and Instance Based for K-Nearest neighbor (IBK) on medical data sets such as Breast Cancer Wisconsin and Hepatitis. Classification accuracy was used in this research based on 10-fold cross validation method. Then, a combination at classification level between these classifiers using deep learning approach was applied to get the highest accuracy and see which the most suitable Deep Multi-classifier Learning (DMCL) approach for the data sets. These medical data sets were taken from the UCI Repository. The results showed that the combination SMO+RF+IBK+NB achieved the highest accuracy for Breast Cancer Wisconsin data set with percentage 96.63%. While for Hepatitis data set, the combination IBk+NB+J48+SMO achieved the highest percentage with 92.50 %. It showed that the proposed method are able to produce the highest prediction accuracy than single and combination of classifier that using majority voting for all these medical data sets.