Fuzzy clustering with a fuzzy covariance matrix

Fuzzy clustering with a fuzzy covariance matrix
复制标题

DOI：
10.1109/cdc.1978.268028
复制
发表时间：
1978
期刊：
1978 IEEE Conference on Decision and Control including the 17th Symposium on Adaptive Processes
影响因子：
--
通讯作者：
Donald Gustafson;W. Kessel
Donald Gustafson;W. Kessel
中科院分区：
其他
文献类型：
--
作者：
Donald Gustafson;W. Kessel

文献摘要

被引文献

相似文献

已经提出了一类包含模糊均值的模糊ISODATA聚类算法。这类算法被推广到包括模糊协方差。由此得到的算法非常类似于混合密度的最大似然估计。认为使用模糊协方差是一种自然的模糊聚类方法。实验结果表明，利用模糊协方差可以得到更准确的聚类结果。
A class of fuzzy ISODATA clustering algorithms has been developed previously which includes fuzzy means. This class of algorithms is generalized to include fuzzy covariances. The resulting algorithm closely resembles maximum likelihood estimation of mixture densities. It is argued that use of fuzzy covariances is a natural approach to fuzzy clustering. Experimental results are presented which indicate that more accurate clustering may be obtained by using fuzzy covariances.