The study of classical pattern recognition most closely related to the Kohonen self-organizing algorithms is known as cluster analysis. This class of algorithms is a set of heuristic procedures that suffers from several problems. We present a fuzzy agglomeration Kohonen clustering network which integrates the competitive agglomeration model into the learning rate and updating strategies of the Kohonen network. The objective function of competitive agglomeration composes of two terms: one is similar to the Fuzzy C-means (FCM) objective function; the other is the sum of squares of the cardinalities of clusters which allows us to control the number of clusters. This yields an optimization problem related to competitive agglomeration. Anderson's IRIS data are used to illustrate this method; and results are compared with the standard Kohonen approach and the fuzzy Kohonen clustering network.
第八屆國際模糊系統學會世界年會暨研討會論文集﹝第二冊﹞=Proceedings of the Eighth International Fuzzy Systems Association World Congress ( Vol.II )，頁757-761