基于均匀设计的特征选择方法

Feature Selection Based on Uniform Design

针对模式识别系统中有效特征的选择问题,采用支持向量机作为分类器,提出了基于均匀设计的特征选择方法.根据均匀设计表的结构及采用的数据集进行训练、测试,最后检验所选的特征子集.实验结果表明,该方法能够有效地去除数据集的冗余特征,取得比使用特征全集更好的分类性能.

For the feature selection in pattern recognition system, a method based on uniform design, which uses support vector machine as a classifier, is proposed. The experiment is carried out according to the structure of uniform design table and the feature numbers of datasets, and finally the results indicate that the method is efficient in discarding the redundant features and obtaining high classification than that on the entire datasets.