Improved Fuzzy K-means Clustering Based on Imbalanced Measure of Cluster Sizes

Fuzzy k-means (FKM) algorithm is an extension of the K-means algorithm, which improves the clustering accuracy of the K-means algorithm for overlapping data sets. However, it has a poor clustering performance for imbalanced datasets. In order to cope with this issue, a measuring method with imbalanced cluster size is introduced. An improved fuzzy k-means algorithm based on imbalanced measure of cluster size is further proposed, by which the imbalanced datasets can be directly processed at the algorithm level. Experimental results on synthetic and UCI datasets showed that the proposed method has better clustering performance than traditional FKM algorithm in case of that there are imbalanced clusters.