基于矩阵计算的并行谱聚类方法.pdf

文档介绍：

1,
1,
1

1
214125

1
2
MPI
-
TP301
A
1000-7180 (2008) xx-xxxx-x
Parallel Spectral Clustering Based on putation
ZHANG Lu-fei1
HAO Zi-yu 1
CHEN Zuo-ning 1
(1 State Key Laboratory of Mathematical Engineering and puting, Wuxi 214125,
China)
Abstract
munity structure of large-scale graph is an important problem. However, some recently proposed
clustering algorithms such as spectral clustering can hardly handle large-scale applications due to plexity issue. It
requires the eigen position of the graph Laplacian matrix, which is memory-consumptive and difficult to explore its
parallelism. To e this problem, we propose the following methods: 1) use state-of-art putation tools; 2)
develop some basic operators such as matrix-vector multiplication with MPI programming model. Finally, we implemented a
parallel spectral clustering algorithm. The experiment result validates the scalability and veracity of the algorithm.
Key words