An empirical comparison of two boosting algorithms on real data sets with artificial class noise

Abstract

Boosting algorithms are a means of building a strong ensemble classifier by aggregating a sequence of weak hypotheses. In this paper, multiple TAN classifiers generated by GTAN are combined by a combination method called Boosting-MultiTAN is compared with the Boosting-BAN classifier which is boosting based on BAN combination. We conduct an empirical study to compare the performance of two algorithms, measured in terms of overall test correct rate, on ten real data sets. Finally, experimental results show that the Boosting-BAN has higher classification accuracy on most data sets, but Boosting-MultiTAN has good effect on others. These results argue that boosting algorithms deserve more attention in machine learning and data mining communities. © 2011 Springer-Verlag.