Improved Boosting Algorithm Using Combined Weak Classifiers

From family of corrective boosting algorithms (i.e. AdaBoost, LogitBoost) to total corrective algorithms (i.e. LPBoost, TotalBoost, SoftBoost, ERLPBoost), we analysis these methods of sample weight updating. Corrective boosting algorithms update the sample weight according to the last hypothesis; comparatively, total corrective algorithms update the weight with the best one of all weak classifiers. However, all these algorithms just use the local information for updating the sample weight ignoring the global information. In light of this context, we show that updating the sample weight using global information of combined weak classifiers maybe accelerate the convergence speed of boosting algorithm. By simply adding the strong classifier to the linear constraints of LPBoost, a new algorithm was proposed. The experimental results show that our algorithm can achieve both better performance and less generalization error compared to some representative boosting algorithm