摘要
为提高支持向量机序列最小优化学习算法的学习性能,提出了一种支持向量机改进序列最小优化学习算法,对传统SMO学习方法进行了多方面改进,从优化变量的选择和2个变量的优化方法分别提出具体可行的改进方法.改进后的SMO学习算法提高了学习速度,加快了网络收敛速度.基于改进SMO算法的仿真结果验证了改进SMO算法的有效性和优越性,并通过仿真,与原始算法进行了比较,显示了改进SMO算法的快速性.
In order to improve SMO learning algorithm, an improved learning algorithm of Support Vector Machine is proposed,and many respects of traditional SMO learning algorithm are improved. Practical improvement methods are proposed in details respectively in respects of optimal variable selection and twovaribk optimizing. The improved SMO learning algorithm quickened the learning speed of the algorithm and the convergence speed of the network. The simulation result based on improved SMO algorithm proves the validity and superiority of the improwed SMO algorithm. And the comparisons made between the improved SMO method and original SMO method, which showed the efficiency of the improved SMO algorithm.
出处
《哈尔滨工程大学学报》
EI
CAS
CSCD
北大核心
2007年第2期183-188,共6页
Journal of Harbin Engineering University
