Diversity Based Improved Bagging Algorithm

Abstract

Bagging is a well known method for designing classifier ensembles. It builds an ensemble of classifier trained on different bootstrap replicates of the training data set. In this paper an improvement to bagging algorithm called DivBagging is presented and studied in depth. The experimental results show that DivBagging is a promising method for ensemble pruning. We believe that it has many advantages over similar methods such as Bagging and Learn because their mechanism is solely based on selecting the most accurate base classifiers.