A robust algorithm for automatic extraction of an unknown number of clusters from noisy data

A clustering algorithm that combines the advantages of fuzzy clustering and robust statistical estimators is presented. The algorithm generates two sets of weights (memberships) for each feature vector as a by-product. These two sets of weights are used to partition the data set and to obtain robust estimates of the prototype parameters, respectively. An extension of the algorithm to deal with an unknown number of clusters is also proposed. The extension is based on overspecifying the number of clusters, and merging compatible clusters.