2014/10/25 by Apoorv Agarwal, Anna Choromanska, Agarwal, Apoorv +4 · 1 citation
Computer Science · Mathematics · #Data Management and Algorithms #FOS: Computer and information sciences #Machine Learning (cs.LG) #Point processes and geometric inequalities #Random Matrices and Applications #Stochastic processes and statistical mechanics #cs.LG
paper · pdf · doi:10.48550/arxiv.1410.6975
openalex publication_date 2014/10/25 · arxiv created 2014/10/26 · arxiv updated 2014/10/28 · openalex created_date 2025/10/24 · openalex updated_date 2026/07/28
In this paper, we compare three initialization schemes for the KMEANS clustering algorithm: 1) random initialization (KMEANSRAND), 2) KMEANS++, and 3) KMEANSD++. Both KMEANSRAND and KMEANS++ have a major that the value of k needs to be set by the user of the algorithms. (Kang 2013) recently proposed a novel use of determinantal point processes for sampling the initial centroids for the KMEANS algorithm (we call it KMEANSD++). They, however, do not provide any evaluation establishing that KMEANSD++ is better than other algorithms. In this paper, we show that the performance of KMEANSD++ is comparable to KMEANS++ (both of which are better than KMEANSRAND) with KMEANSD++ having an additional that it can automatically approximate the value of k.