6533b82bfe1ef96bd128dfaa

RESEARCH PRODUCT

Robust Synchronization-Based Graph Clustering

subject

description

Complex graph data now arises in various fields like social networks, protein-protein interaction networks, ecosystems, etc. To reveal the underlying patterns in graphs, an important task is to partition them into several meaningful clusters. The question is: how can we find the natural partitions of a complex graph which truly reflect the intrinsic patterns? In this paper, we propose RSGC, a novel approach to graph clustering. The key philosophy of RSGC is to consider graph clustering as a dynamic process towards synchronization. For each vertex, it is viewed as an oscillator and interacts with other vertices according to the graph connection information. During the process towards synchronization, vertices with similar connectivity patterns tend to naturally synchronize together to form a cluster. Inherited from the powerful concept of synchronization, RSGC shows several desirable properties: (a) it provides a novel perspective for graph clustering based on proposed interaction model; (b) RSGC allows discovering natural clusters in graph without any data distribution assumption; (c) RSGC is also robust against noise vertices. We systematically evaluate RSGC algorithm on synthetic and real data to demonstrate its superiority.