Generalized Dirichlet-process-means for f-separable distortion measures

DP-means clustering was obtained as an extension of $K$-means clustering. While it is implemented with a simple and efficient algorithm, can estimate the number clusters simultaneously. However, specifically designed for average distortion measure. Therefore, vulnerable to outliers in data, cause large maximum clusters. In this work, we extend objective function $f$-separable measures propose unified learning algorithm overcome above problems by selecting $f$. Further, influence estimated cluster center analyzed evaluate robustness against outliers. We demonstrate performance generalized method numerical experiments using real datasets.