Improving fuzzy c-means method for unbalanced dataset

Traditional fuzzy c-means method (FCM) is a famous clustering algorithm, but has a poor clustering performance for unbalanced dataset. To tackle this defect, a new FCM is presented by introducing cluster size into the formula of determining the membership values in every iteration. Experimental results on synthetic and UCI datasets showed that the proposed method has a better clustering performance than traditional FCM in terms of dealing with datasets with unbalanced clusters.