English

Online Coresets for Clustering with Bregman Divergences

Data Structures and Algorithms 2020-12-14 v1 Machine Learning

Abstract

We present algorithms that create coresets in an online setting for clustering problems according to a wide subset of Bregman divergences. Notably, our coresets have a small additive error, similar in magnitude to the lightweight coresets Bachem et. al. 2018, and take update time O(d)O(d) for every incoming point where dd is dimension of the point. Our first algorithm gives online coresets of size O~(\mboxpoly(k,d,ϵ,μ))\tilde{O}(\mbox{poly}(k,d,\epsilon,\mu)) for kk-clusterings according to any μ\mu-similar Bregman divergence. We further extend this algorithm to show existence of a non-parametric coresets, where the coreset size is independent of kk, the number of clusters, for the same subclass of Bregman divergences. Our non-parametric coresets are larger by a factor of O(logn)O(\log n) (nn is number of points) and have similar (small) additive guarantee. At the same time our coresets also function as lightweight coresets for non-parametric versions of the Bregman clustering like DP-Means. While these coresets provide additive error guarantees, they are also significantly smaller (scaling with O(logn)O(\log n) as opposed to O(dd)O(d^d) for points in R˜d\~R^d) than the (relative-error) coresets obtained in Bachem et. al. 2015 for DP-Means. While our non-parametric coresets are existential, we give an algorithmic version under certain assumptions.

Keywords

Cite

@article{arxiv.2012.06522,
  title  = {Online Coresets for Clustering with Bregman Divergences},
  author = {Rachit Chhaya and Jayesh Choudhari and Anirban Dasgupta and Supratim Shit},
  journal= {arXiv preprint arXiv:2012.06522},
  year   = {2020}
}

Comments

Work in Progress

R2 v1 2026-06-23T20:54:33.572Z