Three Conflict Methods in Multiple Classifiers that Use Dispersed Knowledge

Abstract

In this paper, a conflict model proposed by Pawlak was applied to the approach in which multiple classifiers make global decisions based on dispersed knowledge. Such an application of Pawlak’s model has not been considered before. In this paper, three new usage approaches of this model have been proposed. The model is used to create coalitions of classifiers on the basis of which common knowledge was generated. The approach of using dispersed knowledge, that is discussed in the article, is different from the classical multimodel approach which is known from the literature. The approach has wider applications and more general assumptions. The proposed methods were tested based on data sets from the UCI repository. The obtained results were compared in terms of the efficiency of the system and the form of created coalitions.