Improving classification with automated selection of a combined classifier

Abstract

In this paper we present how the classification results can be improved using a set of classifiers working together in a combined classifier. This one must be compound of different kind of classifiers in order to get better results. The combined classifier is defined from seven classifiers, each one working alone, and then selecting the best combination of them. The results show that the combined classifier acts like a more efficient classifier. This can be seen from the individual databases used, but it is more significant when different databases are considered.