English

Systematically and efficiently improving $k$-means initialization by pairwise-nearest-neighbor smoothing

Machine Learning 2022-12-12 v4 Machine Learning

Abstract

We present a meta-method for initializing (seeding) the kk-means clustering algorithm called PNN-smoothing. It consists in splitting a given dataset into JJ random subsets, clustering each of them individually, and merging the resulting clusterings with the pairwise-nearest-neighbor (PNN) method. It is a meta-method in the sense that when clustering the individual subsets any seeding algorithm can be used. If the computational complexity of that seeding algorithm is linear in the size of the data NN and the number of clusters kk, PNN-smoothing is also almost linear with an appropriate choice of JJ, and quite competitive in practice. We show empirically, using several existing seeding methods and testing on several synthetic and real datasets, that this procedure results in systematically better costs. In particular, our method of enhancing kk-means++ seeding proves superior in both effectiveness and speed compared to the popular "greedy" kk-means++ variant. Our implementation is publicly available at https://github.com/carlobaldassi/KMeansPNNSmoothing.jl.

Keywords

Cite

@article{arxiv.2202.03949,
  title  = {Systematically and efficiently improving $k$-means initialization by pairwise-nearest-neighbor smoothing},
  author = {Carlo Baldassi},
  journal= {arXiv preprint arXiv:2202.03949},
  year   = {2022}
}

Comments

https://openreview.net/forum?id=FTtFAg3pek 16 pages (+8 appendix), 2 figures, 4 tables (+14 appendix). Transactions on Machine Learning Research, Dec 2022

R2 v1 2026-06-24T09:26:33.992Z