Clustering Ensembles Based on Multi-classifier Fusion
Yu Huang, Dorothy Monekosso, Hui Wang · 2010
Clustering ensembles can combine multiple partitions generated by different clustering methods into a final superior clustering result. Compared to single clustering algorithm, it can provide better solutions in terms of robustness, novelty and stability. In this paper, we proposed a new method named CEMF, i.e., Clustering Ensembles Based on Multi-classifier Fusion. We combine the clustering ensembles method and multi-classifier method to deal with the clustering consensus problem. CEMF generates multiple partitions and create subspaces which can be used to constructs the local optimum classifiers. CEMF makes use of the advantage of multi-classifiers to assist clustering ensembles in different subspaces of data set. Experiments carried out on some public data sets show that CEMF is comparable or better than classical clustering algorithms and traditional clustering ensembles methods. It's an effective and feasible method.