English

A New Notion of Individually Fair Clustering: $\alpha$-Equitable $k$-Center

Machine Learning 2022-02-15 v3 Computers and Society Data Structures and Algorithms

Abstract

Clustering is a fundamental problem in unsupervised machine learning, and fair variants of it have recently received significant attention due to its societal implications. In this work we introduce a novel definition of individual fairness for clustering problems. Specifically, in our model, each point jj has a set of other points Sj\mathcal{S}_j that it perceives as similar to itself, and it feels that it is fairly treated if the quality of service it receives in the solution is α\alpha-close (in a multiplicative sense, for a given α1\alpha \geq 1) to that of the points in Sj\mathcal{S}_j. We begin our study by answering questions regarding the structure of the problem, namely for what values of α\alpha the problem is well-defined, and what the behavior of the \emph{Price of Fairness (PoF)} for it is. For the well-defined region of α\alpha, we provide efficient and easily-implementable approximation algorithms for the kk-center objective, which in certain cases enjoy bounded-PoF guarantees. We finally complement our analysis by an extensive suite of experiments that validates the effectiveness of our theoretical results.

Keywords

Cite

@article{arxiv.2106.05423,
  title  = {A New Notion of Individually Fair Clustering: $\alpha$-Equitable $k$-Center},
  author = {Darshan Chakrabarti and John P. Dickerson and Seyed A. Esmaeili and Aravind Srinivasan and Leonidas Tsepenekas},
  journal= {arXiv preprint arXiv:2106.05423},
  year   = {2022}
}

Comments

To appear at AISTATS 2022