Abstract: One popular algorithm for training support vector machine (SVM) is sequential minimal optimization (SMO), but it still requires a large amount of computation time for solving large size problems. This paper proposes one parallel implementation of SMO for training SVM. Experiments show that the algorithm is effective, and there is great speedup in the parallel SMO when many processors are used.

Key words: support vector machine(SVM), sequential minimal optimization(SMO), parallel algorithm

摘要： 序列最小优化(SMO)是训练支持向量机(SVM)的常见算法，在求解大规模问题时，需要耗费大量的计算时间。该文提出了SMO的一种并行实现方法，验证了该算法的有效性。实验结果表明，当采用多处理器时，并行SMO具有较大的加速比。

关键词: 支持向量机, 序列最小优化, 并行算法

CLC Number: 

• TP18