基于GA和Bootstrap的最小二乘支持向量机参数优选

Method for Selecting Parameters of Least Squares Support Vector Machines Based on GA and Bootstrap

支持向量机是一种新型的学习方法,该方法以结构风险最小化原则取代传统机器学习中的经验风险最小化原则,在小样本的机器学习中显示出了优异的性能。传统的支持向量机是解凸二次规划问题,而最小二乘支持向量机是解等式线性方程,显得尤为方便。针对最小二乘支持向量机的特点,通过Bootstrap建立适当的性能指标,用遗传算法(GA)优化最小二乘支持向量机的有关参数,并在非线性经济系统中应用。用最小二乘支持向量机对非线性经济系统进行预测的结果与神经网络预测的结果比较证明,该模型的预测精确度是令人满意的,文中提出的方法是可行的。

Support Vector Machines（SVM）is based on Structural Risk Minimization principle. SVM has shown powerful ability in learning with limited samples. In Least Squares SVM （LS-SVM）, a least squares cost function is proposed so as to obtain a linear set of equations in dual space. In order to select hyper-parameters of LS-SVM, the performance index could be built by Bootstrap, then the performance index could be optimized by genetic algorithm （GA）, at last hyper-parameters selection could be solved. The model was then used to forecast nonlinear macroeconomic system. It is shown the LS-SVM based on selecting parameters by GA and Bootstrap is simple and effective.