Ensemble re-clustering: Refinement of hard clustering by three-way strategy

Abstract

In this paper, we propose a three-way ensemble re-clustering method based on ideas of cluster ensemble and three-way decision. In the proposed method, we use hard clustering methods to produce different clustering results and cluster labels matching to align each clustering results to a given order. The intersection of the clusters with same labels are regarded as the core region and the difference between the union and the intersection of the clusters with same labels are regarded as the fringe region of the specific cluster. Therefore, a three-way result of the cluster is naturally formed. The results on UCI data sets show that such strategy is effective in improving the structure of clustering results and F 1 values.