Parallel C–fuzzy random forest

Abstract

The C–fuzzy random forest is a novel ensemble classifier which uses C-fuzzy decision trees as unit classifiers. The main problem connected with this classifier is a relatively long learning process time. In this paper the method of reducing the C–fuzzy random forest’s learning time is proposed. Authors proposed and described the method of parallelization of this classifier’s learning process by generating trees which are the parts of the forest in separate threads. The experiments which were designed to check the effectiveness of the proposed method were performed and the results were presented and discussed.