We present a novel soft clustering approach based on Support Vector Clustering.
Data points outside the clusters found form a fuzzy boundary region.
Clusters with any shape as well as outliers can be identified.
Membership degrees were calculated in a natural way.