Ensembling evidential k-nearest neighbor classifiers through multi-modal perturbation

摘要: Ensembling techniques have already been considered for improving the accuracy of k-nearest neighbor classifier. It is shown that using different feature subspaces each member classifier, strong ensembles can be generated. Although it has a more flexible structure which an obvious advantage from diversity point view and observed to provide better classification accuracies compared voting based k-NN ensembling evidential classifier on Dempster-Shafer theory evidence not yet fully studied. In this paper, we firstly investigate performance random subspace method. Taking into account its potential perturbed also in parameter dimension due class dependent parameters, propose through multi-modal perturbation genetic algorithms. Experimental results improved obtained method further surpassed perturbation.