A multi-modal perturbation-based selective ensemble algorithm

Abstract

Ensemble techniques have been widely used for improving the classification performance, and recent studies show that ensembling classifiers through multi-modal perturbation can further improve the classification performance. In this paper, we propose a selective ensemble algorithm based on multi-modal perturbation (called SE_MP). In SE_MP, we devise a multi-modal perturbation method based on sampling and reduction to generate diverse base classifiers. The proposed perturbation method can simultaneously disturb the training set and attribute space, which can increase the diversity among base classifiers. In the experimental stage, SE_MP is compared with the existing ensemble algorithm on several UCI data sets, where the KNN (k-nearest neighbors) classification algorithm is used to train base classifiers. Experimental results show that SE_MP can provide competitive solutions for selective ensemble.