Host: Japan SOciety for Fuzzy Theory and intelligent informatics
Co-host: The Korea Fuzzy Logic and Intelligent Systems Society, IEEE Computational Intelligence Society, The International Fuzzy Systems Association, 21th Century COE Program "Creation of Agent-Based Social Systems Sciences"
To realize a reliable classification of data, a new clustering method by the K-Means algorithm using a split and merge procedure is proposed. After ordinary clustering by the K-Means algorithm, a method determining whether or not each decision region should be split is introduced. After splitting each target decision region into subregions by using the K-Means algorithm again, the other subregions except one region are merged into appropriate adjacent decision regions. The goodness of this procedure is demonstrated under some classification experiments.