机构地区: 东莞理工学院软件学院
出 处: 《东莞理工学院学报》 2007年第3期38-42,共5页
摘 要: 光滑支持向量机是目前的一个研究热点.牛顿-条件预优共轭梯度法Newton-PCG(Newton- preconditioned congugate gradient)是一种求解优化问题的更有效算法.列出了该算法用于求解光滑支持向量机的基本思想和基本步骤,还比较了原始牛顿法和牛顿-条件预优共轭梯度法的计算效率.结果表明,牛顿-条件预优共轭梯度法的计算效率明显高于原始牛顿法. Smooth support vector machine(SSVM) is an active field in SVM research. This paper presents the solutions to the problems of SSVM with a new optimization technique, the Ncwton-PCG (Newton-preconditioned conjugate gradient) algorithm. By comparison with their computational efficiencies, it is showed that the Newton-PCG algorithm is obviously more efficient than Newton algorithm, which is an important direction for future research on smooth SVMs.
领 域: [生物学]