基于改进SVM的特征选择

Based on Modified Support Vector Machines Feature Selection

本文在仔细分析特征选择思想的基础上,将特征选择过程嵌入到学习机里面,提出了一种基于改进支持向量机的特征选择算法(Feature selection via Modified Support Vector Machines),该方法通过对特征的权重进行排序来实现特征选择.利用可以将特征选择过程和学习过程有机地统一起来,实验表明,与其它方法比较,该方法能够达到比较好的效果.

Based on the careful analysis of feature selection, we propose FS-MSVM（Feature selection via Modified Support Vector Machines） which realizes feature selection through ranking the weights of the features, and embedding the process of feature selection into learning machines. So that can combine feature selection process and learning process efficiently. It is shown in experiments that the method can perform better than other methods do.