This is an old revision of the document!
the_advantages_of_careful_seeding._proceedings_of_the_eighteenth_annual_acm-siam_symposium_on_discrete_algorithms._society_for_industrial_and_applied_mathematics_philadelphia_pa_usa._pp._1027_1035._abstract_the_k-means_method_is_a_widely_used_clustering_technique_that_seeks_to_minimize_the_average_squared_distance_between_points_in_the_same_cluster._although_it_offers_no_accuracy_guarantees_its_simplicity_and_speed_are_very_appealing_in_practice._by_augmenting_k-means_with_a_simple_randomized_seeding_technique_we_obtain_an_algorithm_that_is_o_log_k_-competitive_with_the_optimal_clustering._experiments_show_our_augmentation_improves_both_the_speed_and_the_accuracy_of_k-means_often_quite_dramatically}}
