摘要
针对现有回归型加权支持向量机直接选择加权系数法存在的不足,提出了一种对加权系数进行寻优的新方法——动态自适应加权算法.通过对权系数进行的自适应迭代调整,以确定其最优值,并进行了实验仿真.仿真结果表明:采用该方法确定的最优加权系数,可以对预测样本数据进行更准确的回归估计.
Against direct selection method of weighting coefficients in the existing weighted support vector machines(WSVM),we present a new weighting coefficient optimization algorithm-adaptive weighted support vector machine(AWSVM).Through adaptive iterations,weighting coefficients are adjusted in order to determine its optimal value.Experimental results show that using the optimal weighting factor determined by this new method,we can predict the sample data for more accurate regression estimates.
出处
《烟台大学学报(自然科学与工程版)》
CAS
北大核心
2009年第4期282-285,共4页
Journal of Yantai University(Natural Science and Engineering Edition)
基金
山东省自然科学基金资助项目(Y2006G22)
关键词
支持向量机
回归
加权系数
自适应加权支持向量回归机
support vector machine(SVM)
regression
weighting factor
adaptive weighted support vector machine(AWSVM)