Fair k-Center Clustering in MapReduce and Streaming Settings

Center-based clustering techniques are fundamental to many realworld applications such as data summarization and social network analysis. In this work, we study the problem of fairness aware 𝑘center clustering over large datasets. We are given an input dataset comprising a set of 𝑛 points, where each point belongs to a specific demographic group characterized by a protected attribute, such as race or gender. The goal is to identify 𝑘 clusters such that all clusters have considerable representation from all groups and the maximum radius of these clusters is minimized. The majority of the prior techniques do not scale beyond 100𝐾 points for 𝑘 = 50. To address the scalability challenges, we propose an efficient 2-round algorithm for the MapReduce setting that is guaranteed to be a 9-approximation to the optimal solution. Additionally, we develop a 2-pass streaming algorithm that is efficient and has a low memory footprint. These theoretical results are complemented with an empirical evaluation on million-scale datasets, demonstrating that our techniques are effective to identify highquality fair clusters and efficient as compared to the state-of-the-art. CCS CONCEPTS • Theory of computation → Unsupervised learning and clustering.