说明：  针对SVM法线特征筛选算法仅考虑法线对特征筛选的贡献,而忽略了特征分布对特征筛选的贡献的不足,在对SVM法线算法进行分析的基础上,基于特征在正、负例中出现概率的不同提出了加权SVM法线算法,该算法考虑到了法线和特征的分布.通过试验可以看出,在使用较小的特征空间时,与SVM法线算法和信息增益算法相比,加权SVM法线算法具有更好的特征筛选性能.
(Normal feature selection for SVM algorithm only considered normal for the contribution of feature selection, to the neglect of the characteristics of the distribution of feature selection have contributed to the lack of normal SVM algorithm based on the analysis, based on the characteristics of the positive and negative cases emergence of a different probability-weighted normal SVM algorithm, which takes into account the distribution and characteristics of normal. through the test can be seen in the use of smaller feature space, the normal and the SVM algorithm and information gain algorithm, normal weighted SVM algorithm has better performance of feature selection.)