Soft-Voting Clustering Ensemble

• Affiliation of Author(s):西南交通大学
• Journal:11th International Conference on Multiple Classifier Systems
• Key Words:Clustering ensemble, Majority voting, Soft-Voting Clustering Ensemble.
• Abstract:Clustering ensemble is a framework for combining multiple based clustering results of a set of objects without accessing the original feature of the objects. The majority voting method is widely used in clustering ensemble because of its simplicity, robustness and stability. In general, the existing voting methods only accept hard clustering results as input. In this paper we propose a new algorithm, Soft-Voting Clustering Ensemble (SVCE), which has better flexibility and generalization. The theory of SVCE is illustrated and the algorithm of SVCE is stated in detail firstly. Then 15 UCI datasets are used for the experiment and the results show that the proposed method has a better performance than state of the art ensemble methods in most cases, such as Majority Voting, Weighted Majority Voting, CSPA, MCLA, HGPA.
• Co-author:Hongjun Wang, Chen Dahai
• First Author:Haishen Wang
• Indexed by:Academic papers
• Correspondence Author:Yan Yang
• Document Code:20140517252241
• Discipline:Engineering
• First-Level Discipline:Computer Science and Technology
• Volume:LNCS 7872
• Issue:MCS 2013
• Page Number:pp. 307–318, 2013.
• Translation or Not:no