采用双目标优化的核参数选择方法

核函数的参数选取问题是用支持向量机进行非线性学习时的一个重要问题.采用距离标准和夹角标准给出一种基于多目标的核参数确定方法:使核参数满足两类训练样本到各自的中心距离和尽可能小、而到对方类中心距离和尽可能大,并且使核矩阵和目标核矩阵之间的夹角尽可能小.试验结果表明该方法是有效的.

Bi-criterion optimization for kernel parameters selection

Parameter selection of kernel function is an issue of nonlinear study based on Support Vector Machine(SVM).Considering that the parameters of the kernel function affect the result of the nonlinear SVM greatly,two criterions are used for choosing the optimum parameter of a given kernel function.One is the distance criterion that minimizes the sum-square distance between the labeled training sample and its own center,and maximizes the sum-square distance between the training sample and the other labeled-center.The other is the angle criterion that minimizes the angle between the kernel matrix and targets matrix.The experiments show that our methods are efficient.