Coresets for clustering in Euclidean spaces: importance sampling is nearly optimal.

Lingxiao Huang, Nisheeth K. Vishnoi

Browse the full STOC paper archive.