On the fuzzy joint points method for the problem of fuzzy clustering

The present article considers the fuzzy joint points (FJP) method for the problem of fuzzy clustering. The FJP method is based on the creation of joint sets starting with use of the fuzzy neighbor relation between elements. The principal distinctive property here is the existence of a built-in mechanism of defining an optimal number of classes. However, the degree of adequacy of the number of classes depends on the structure of the data. In the present article the FJP method is investigated and a sufficient condition for correct recognition of the optimal number of classes is proved. © Allerton Press, Inc., 2006.