Dynamic Ensemble Selection Using Discriminant Functions and Normalization Between Class Labels – Approach to Binary Classification

Abstract

In the classification task, the ensemble selection methods reduce the available pool of the base classifiers. The dynamic ensemble selection methods allow to find the subset of base classifiers for each test sample separately. In finding the best subset of base classifiers many methods used the so-called competence region determined for the validation data set. In this paper, we propose the dynamic ensemble selection in which the validation data set is not necessary and the competence region for the test sample is not determined. Generally, the described method uses only the decision profiles in the selection process. The experiment results based on ten data sets show that the proposed dynamic ensemble selection is a promising method for the development of multiple classifiers systems.