可能性划分系数和模糊变差相结合的聚类有效性函数
Clustering validity function based on possibilistic partition coefficient combined with fuzzy variation
-
摘要: 基于可能性分布描述因子定义的可能性划分系数有随类数增加而单调递减的趋势,缺乏与数据集几何结构的直接联系。该文考虑到数据集的几何结构信息,对可能性划分系数进行改进,提出了新的聚类有效性标准。实验结果表明,该文提出的方法具有良好的分类性能。Abstract: Possibilistic partition coefficient which based on possibilistic distribution descriptor has decreasing tendency as the classification number increasing and does not directly relate to the geometry structure of data set. To consider the geometry structure information of data set, new clustering validity functions are defined by modify possibilistic partition coefficient. Experimental results show that the new methods have good classification performance.
-
J.C. Bezdek, Pattern Recognition with Fuzzy Objective Function Algorithms, New York, Plenums Press, 1981, 95-107.[2]N.R. Pal, J. C. Bezdek, On cluster validity for the fuzzy C-means model, IEEE Trans. on Fuzzy Systems, 1995, 3(3), 370-379.[3]范九伦,裴继红,谢维信,基于可能性分布的有效性函数,电子学报,1998,26(4),103-105.[4]L.A. Zadeh, Fuzzy sets as a basis for a theory of possibility, Fuzzy Sets and Systems, 1978, 1(1),3-28.[5]L.O. Hall, A. Kandel, Studies in possibilistic recognition, Fuzzy Sets and Systems, 1985, 17(2),167-179.[6]范九伦,模糊聚类新算法与聚类有效性问题研究,[博士论文],西安电子科技大学,1998年. ［7］X.L. Xie, G. A. Beni, Validity measure for fuzzy clustering, IEEE Trans. on Pattern Anal.Machine Intell, 1991, 3(8), 841-846.[7]M.S. Yang, On a class of fuzzy classification maximum likelihood procedures, Fuzzy Sets and Systems, 1993, 57(3), 365-375. -
计量
- 文章访问数: 2600
- HTML全文浏览量: 117
- PDF下载量: 514
- 被引次数: 0
下载:
