An experimental evaluation of feature selection based classifier ensemble for handwritten numeral recognition

Abstract

The paper is about classifier ensemble approach applied for the recognition of handwritten digits. In this approach we used combination of feature selection and ensemble technique simultaneously. The method based on ensemble of diverse classifier is used to classify the patterns. The following approaches for building an ensemble model are used: Selecting diverse training data from the original source data set, constructing different neural network models, selecting ensemble nets from ensemble candidates and combining ensemble members results. Forward and Backward sequential feature selection is applied with different criterion of distance calculation and an improvement in efficiency is found using the proposed approach. © 2014 IEEE.