GPUSCAN: GPU-Based Parallel Structural Clustering Algorithm for Networks
S. Kockara,Recep Avci,Thomas Ryan Stovall
DOI: https://doi.org/10.1109/TPDS.2014.2374607
IF: 5.3
2015-12-01
IEEE Transactions on Parallel and Distributed Systems
Abstract:This paper presents a massively parallel implementation of a prominent network clustering algorithm, the structural clustering algorithm for networks (SCAN), on a graphical processing unit (GPU). SCAN is a fast and efficient clustering technique for finding hidden communities and isolating hubs/outliers within a network. However, for very large networks, it still takes considerable amount of time. With the introduction of massively parallel Compute Unified Device Architecture (CUDA) by Nvidia, applications properly employing GPUs are demonstrating high speed up. In current study, GPUSCAN, a CUDA based parallel implementation of SCAN, is presented. SCAN's computation steps have been carefully redesigned to run very efficiently on the GPU by transforming SCAN into a series of highly regular and independent concurrent operations. All intermediate data structures are created in the GPU to efficiently benefit from GPU's memory hierarchy. How these structures reformed and represented in the GPU memory hierarchy are illustrated. Now, through GPUSCAN, a large network or a batch of disjoint networks can be offloaded to the GPU for very fast and equivalent structural clustering. The performance of the GPU accelerated structural clustering has been shown to be much faster than the sequential CPU implementation. Both GPUSCAN and SCAN are tested on different size artificial and real-world networks. Results indicate that network becomes larger GPUSCAN significantly over performs SCAN. In tested datasets, speed-up of over 500-fold is achieved. For instance, calculating structural similarity and clustering of 5.5 million edges of the California road network in GPUSCAN is 513-fold faster than the serial version of SCAN.
Computer Science,Engineering