Studying the robustness of ensembles of classifiers used for cancer diagnosis using microarrays datasets

Ensembles of classifiers were shown to provide better accuracy than single classifiers. However, the classification robustness is an important performance measure for classifiers and ensembles, besides accuracy, that should be considered. Increasing the robustness of classification systems results in reducing the probability of over-fitting. The robustness, as defined in this study, has not been studied in the literature. In this paper, a framework is used to prove that ensembles of classifiers are more robust than single classifiers. The framework selects different ensembles of classifiers and compares their robustness to the robustness of their members. The experiments performed on six different microarray datasets showed that ensembles of classifiers are more robust than their members.