The well known fuzzy partition clustering algorithms are most based on Euclidean distance function, which can only be used to detect spherical structural clusters. Gustafson-Kessel (GK) clustering algorithm and Gath-Geva (GG) clustering algorithm, were developed to detect non-spherical structural clusters, but both of them based on semi-supervised Mahalanobis distance needed additional prior information. An improved fuzzy C-mean algorithm based on unsupervised Mahalanobis distance, FCM-M, was proposed by our previous work, but it didn't consider the relationships between cluster centers in the objective function. In this paper, we proposed an improved fuzzy C-mean algorithm, FCM-MS, which is not only based on unsupervised Mahalanobis distance, but also considering the relationships between cluster centers, and the relationships between the center of all points and the cluster centers in the objective function, the singular and the initial values problems were also solved. A real data set was applied to prove that the performance of the FCM-MS algorithm gave more accurate clustering results than the FCM and FCM-M methods, and the ratio method which is proposed by us is the better of the two methods for selecting the initial values.
27 March 2017 in DEStech Transactions on Engineering and Technology Research
An improved new fuzzy clustering algorithm is developed to obtain better quality of fuzzy clustering results. The objective function includes the regulating terms about the covariance matrices. The cluster centers and the covariance matrices are directly derived from the Lagrange’s method. The fuzzy c-mean algorithm is different from the GK and GG algorithms. The singular problem and the selecting initial values problem are improved by the Eigenvalue method and the Ratio method. We proposed a proposition which pointed out that the initial memberships of fuzzy c-mean algorithm which was based on Mahalanobis distance Algorithm and the traditional FCM Algorithm can not be all equal. The other important issue is how to approach the global minimum value that can improve the cluster accuracy. The methods to detect the local extreme value were developed by this paper. Focusing attention to these two faults, an improved new algorithm, “Fuzzy C-Means based on Particle Swarm Optimization with Mahalanobis distance (PSO-FCM-M)”, is proposed. We have two aims and goals of our research summary. One is to compare the classification accuracies of fuzzy clustering algorithms based on mahalanobis distances and euclidean distances. The other is to choose the initial membership to promote the classification accuracies.