How the initialization affects the stability of the k-means algorithm - Statistics > Machine LearningReport as inadecuate




How the initialization affects the stability of the k-means algorithm - Statistics > Machine Learning - Download this document for free, or read online. Document in PDF available to download.

Abstract: We investigate the role of the initialization for the stability of thek-means clustering algorithm. As opposed to other papers, we consider theactual k-means algorithm and do not ignore its property of getting stuck inlocal optima. We are interested in the actual clustering, not only in the costsof the solution. We analyze when different initializations lead to the samelocal optimum, and when they lead to different local optima. This enables us toprove that it is reasonable to select the number of clusters based on stabilityscores.



Author: Sebastien Bubeck, Marina Meila, Ulrike von Luxburg

Source: https://arxiv.org/







Related documents