SVM Combined with FCM and PSO for Fuzzy Clustering

FCM algorithm is apt to fall into the local optimization, and what fast FCM algorithm can find optimum is greatly depended on the initialization. PSO-based FCM clustering algorithm avoids the local optima, and also is robust to initialization. The fluctuation however has appeared in the new algorithm, and it had been observed that performance of the clustering algorithms deteriorate with more and more overlaps in the data sets. SVM Classifier can handle linear inseparable problems and has the advantages of high accuracy of classification. Motivated by this observation, in this article a new fuzzy clustering technique that SVM Combined with FCM and PSO for Classification Problems has been proposed. Results of numerical experiments on two standard datasets show that the new algorithm is more efficient than the FCM and PSO-based FCM clustering algorithms, it can not only avoids the local optima and is robust to initialization, and also improve accuracy of classification.