Abstract:

A distributed affinity propagation algorithm based on GraphLab was proposed, which was named GStrAP (Graphlab based stream affinity propagation). In GraphLab′s DAG abstraction, the parallel computation was represented as a directed acyclic graph with data flowing along edges between vertices, and the “Gather-Apply-Scatter” paradigm was applied to complete data synchronization and algorithm′s iteration. The experimental results on 3D Clusters, Aggregation, Flame and Pathbased datasets with different scale and the clustering performance were compared with Kmeans, which demonstrated that the proposed GStrAP could achieve high performance on both scalability and accuracy.

Key words: distributed computation, affinity propagation clustering algorithm, GraphLab, clustering ensemble