Tournament fuzzy clustering algorithm with automatic cluster number estimation
Yasunori Endo, Shingo Yamaguchi · Electronics and Communications in Japan (Part III Fundamental Electronic Science) · 1998
The theory of tournament clustering algorithms is used to develop a new fuzzy clustering algorithm. In the past, work on fuzzy clustering has focused on the fuzzy C-means (FCM) approach. While this approach is more effective than the “hard clustering” approach, which makes no use of fuzzy theory, it has certain deficiencies: it cannot handle objective or subjective differences between individuals well, and it lacks essential capabilities such as the ability to recognize isolated data items. To resolve these problems, a tournament fuzzy clustering algorithm with automatic cluster number estimation (T-FCA-ACNE) is proposed. The algorithm includes a capability for cluster number estimation and can express subjective and objective differences between individuals. The validity of the new algorithm is demonstrated by tests with real data. © 1998 Scripta Technica. Electron Comm Jpn Pt 3, 81(2): 46–59, 1998