A Super Classifier with Programming-Based Boosting Using Biometrics for Person Authentication

Abstract

A boosting-based multi-classification system has been designed and established using different biometric features, that is, fingerprints of both right and left hands and handwriting for authenticated person identification. This multi-classifier comprises of three different classifiers. The first classifier is an Optimal Clustering Algorithm (OCA)-based modified Radial Basis Function Network (RBFN) with Back Propagation (BP) learning, second classifier is a Heuristic Based Clustering (HBC) algorithm-based modified RBFN with BP learning and third classifier is a combination of Malsburg learning and BP Network. These three individual classifiers identify fingerprints of both right and left hands and handwriting, respectively, and the super classifier perform the fusion of three conclusions to establish the final decision based on programming-based boosting method for person authentication. The technique of using multiple classifiers in a single system is efficient and effective. Also the accuracy, precision, recall, and F-score of the classifiers are substantially moderate and the training and testing time of all the biometrics are quite low and affordable.