Title:Distributed Graph Clustering by Load Balancing

Abstract:Graph clustering is a fundamental problem with a number of applications in algorithm design, machine learning, data mining, and analysis of social networks. Over the past decades, researchers have proposed a number of algorithmic design methods for graph clustering. However, most of these methods are based on complicated spectral techniques or convex optimization, and cannot be applied directly for clustering most real-world networks, whose information is often collected on different sites. Designing a simple clustering algorithm that works in the distributed setting is of important interest, and has wide applications for processing big datasets.
In this paper we present a simple and distributed algorithm for graph clustering: for a wide class of graphs with n nodes that can be characterized by a well-defined cluster-structure, our algorithm finishes in a poly-logarithmic number of rounds, and recovers a partition of the graph with at most o(n) misclassified nodes. The main component of our algorithm is an application of the random matching model of load balancing, which is a fundamental protocol in distributed computing and has been extensively studied in the past 20 years. Hence, our result highlights an intrinsic and interesting connection between graph clustering and load balancing.