基于SVM的多变量函数回归分析研究(英文)

Choosing Multiple Parameters for Function Regression Based on SVM

研究了通过自动调整回归因子多项参数方式提高基于SVM函数回归性能的问题。通过使用梯度下降法和遗传算法最小化SVM的归一化误差估计值来实现SVM函数回归性能的提高。大量的试验结果验证了所提出方法的性能。

Performance improvement of function regression based on Support Vector Mmachines （SVM） by automatically tuning multiple parameters of regressors was considered. This was done by minimizing some estimation of the generalization error of SVM using gradient descent and Genetic Algorithm （GA） over the set of parameters. Performance of the proposed method was illustrated by extensive experimental results.