Improvement and optimization of a fuzzy C-means clustering algorithm

In this paper, an improved FCM clustering algorithm is proposed. Unlike a traditional FCM clustering algorithm whose convergence is sensitive to its initial parameters, the proposed algorithm based on fuzzy decision theory can automatically and adaptively select these parameters with optimal values. The simulation results indicate that the modified algorithm not only overcomes the ill phenomena of the FCM algorithms available now, but also is robust to the selection of the weighting constants.