Clustering high-dimensional data