Optimizing Dynamic Ensemble Selection Procedure by Evolutionary Extreme Learning Machines and a Noise Reduction Filter

Abstract

Ensemble of classifier is an effective way of improving performance of individual classifiers. However, the choice of the ensemble members can become a very difficult task, which, in some cases, can lead to ensembles with no performance improvement. Dynamic ensemble selection systems aim to select a group of classifiers that is most adequate for a specific query pattern. In this paper, we present a strategy that optimizes the dynamic ensemble selection procedure. Initially, a pool of classifiers has been built in an automatic way through an evolutionary algorithm. After, we improved the regions of competence in order to avoid noise and create smoother class boundaries. Finally, we use a dynamic ensemble selection rule. Extreme Learning Machines were used in the classification phase. Performance of the system was compared against other methods.