中文摘要：

针对支持向量机（SVM）分类器的模型选择问题,提出了一种基于特征空间的类别可分性度量（FCSM）准则,并将该准则用于优化多个高斯函数的线性组合系数.与核矩阵度量（FSM）准则相比,FCSM准则在核函数优化应用中的适用性更广,并且在优化效果上有更好的理论支持.实验结果表明,与交叉验证法、半径间隔误差（RM）界法以及基于FSM准则的优化方法相比,FCSM准则能从更大函数集范围优选出核函数,使SVM分类器获得更好的分类能力.

英文摘要：

To solve the problem of model selection for support vector machine（SVM） classifier,a feature-space-based class separability measure（FCSM） was proposed.With this measure,the combination coefficients of multiple Gaussian functions were optimized.Compared with the kernel matrix evaluation measure（FSM）,the new measure has fewer limitations in the application of kernel optimization,and has better theoretical guarantees.The experimental results show that the proposed algorithm outperforms the cross-validation method,the radius margin bound method and the FSM based method,and moreover,it achieves better performance on SVM classifier with the optimal kernel selected from a wider range of function set.