An efficient parallel method for canopy clusteing with multi-core platform

This paper introduces an efficient parallel method of canopy clustering algorithm on a multi-core processor. Canopy clustering is a pre-clustering algorithm which is useful when handling large amount of data sets or big dimensions of vectors. Performance of multi-core is outstanding in parallel processing which is suitable for clustering. This paper proposes a solution with hardware-software co-design method rather than hardware or software only. Communication cost in the proposed method is reduced and more predictable. Experiment results show excellent scalability and comparing with using single core solution, it speeds up 4.2 times when using 8 cores.