Applying Classification Problems via a Data Mining Approach Based on a Cerebellar Model Articulation Controller

Abstract

Although applied to classification, neural network (NN) classifiers have certain limitations, including slow training time, complex interpretation and difficult implementation in terms of optimal network topology. To overcome these disadvantages, this study presents an efficient and simple classifier based on the cerebellar model articulation controller NN (CMAC NN), which has the advantages of very fast learning, reasonable generalization ability and robust noise resistance. The performance of the proposed CMAC NN classifier is measured using PROBEN1 benchmark data sets taken from the UCI Machine Learning Repository for diabetes and glass, each of which include, respectively, three permutations of the available patterns. Numerical results show that the proposed CMAC NN classifier was efficient for tested data sets. Therefore, the CMAC NN classifier can be considered as a data mining tool to classification.