Proposal of new hybrid fuzzy clustering algorithms — Application to breast cancer dataset

Abstract

This paper presents new hybrid fuzzy clustering algorithms. The aims of the proposed modifications are to provide robustness for the initial cluster centers using Subtractive clustering and to reduce the number of iterations using the Fuzzy ckMeans center updating strategy. These modifications are applied in the conventional fuzzy clustering algorithms: Fuzzy c-Means, Gustafson-Kessel and Gath-Geva. The proposed methods are applied to Wisconsin Breast Cancer dataset and results compare the proposed algorithms with their conventional forms considering different validity indices and classification accuracy.