English

Efficient Online Learning for Dynamic k-Clustering

Machine Learning 2021-06-09 v1

Abstract

We study dynamic clustering problems from the perspective of online learning. We consider an online learning problem, called \textit{Dynamic kk-Clustering}, in which kk centers are maintained in a metric space over time (centers may change positions) such as a dynamically changing set of rr clients is served in the best possible way. The connection cost at round tt is given by the \textit{pp-norm} of the vector consisting of the distance of each client to its closest center at round tt, for some p1p\geq 1 or p=p = \infty. We present a \textit{Θ(min(k,r))\Theta\left( \min(k,r) \right)-regret} polynomial-time online learning algorithm and show that, under some well-established computational complexity conjectures, \textit{constant-regret} cannot be achieved in polynomial-time. In addition to the efficient solution of Dynamic kk-Clustering, our work contributes to the long line of research on combinatorial online learning.

Keywords

Cite

@article{arxiv.2106.04336,
  title  = {Efficient Online Learning for Dynamic k-Clustering},
  author = {Dimitris Fotakis and Georgios Piliouras and Stratis Skoulakis},
  journal= {arXiv preprint arXiv:2106.04336},
  year   = {2021}
}
R2 v1 2026-06-24T02:57:30.769Z