Distributed Clustering on Graphs

This paper provides new algorithms for distributed clustering for two popular center-based objec-tives, k-median and k-means. These algorithms have provable guarantees and improve communicationcomplexity over existing approaches. Following a classic approach in clustering by [13], we reduce theproblem of finding a clustering with low cost to the problem of finding a ‘coreset’ of small size. Weprovide a distributed method for constructing a global coreset which improves over the previous methodsby reducing the communication complexity, and which works over general communication topologies.Experiment results on large scale data sets show that this approach outperforms other coreset-baseddistributed clustering algorithms.