Remarks to the direct method of fuzzy clustering based on presentation of initial data by fuzzy powerful tolerance

Abstract

This short paper describes rules for calculation of tolerance threshold in the direct clustering method based on a presentation of initial data by fuzzy powerful tolerances. The concept of the fuzzy representation of an initial set of elements by fuzzy clusters is considered on a basis of the short essential consideration of the problem of fuzzy cluster analysis. General plan of the clustering procedure is presented and rules for tolerance threshold calculation are proposed. Three strategies of clustering are described shortly. Some preliminary conclusions are discussed and ways of perspective investigations are outlined.