Suppose the data consist of a set of points , , distributed in a bounded domain , where is a large number. An algorithm is given for finding the sets of dimension , , in a neighborhood of which maximal amount of points lie. The algorithm is different from PCA (principal component analysis)
View on arXiv