An enhanced general fuzzy min-max neural network for classification

Abstract

This paper proposes the enhanced general fuzzy min-max neural network (EGFM) classification model to perform supervised classification of data. The aim is to overcome a number of limitations of the general fuzzy min-max neural network (GFMM) and improve its classification performance. New hyperbox expansion, overlap and contraction rules proposed in enhanced fuzzy min-max (EFMM) neural network algorithm are used to improve the GFMM learning algorithm. The proposed EGFM classifier is mainly a merging of GFMM and EFMM classifiers to overcome some unidentified cases of hyperboxes overlap and consequent contraction problems in some regions. Accuracy and efficiency of EGFM classifier are evaluated using different standard datasets taken from UCI machine learning repository and the results are better than those from GFMM classifier.