一种基于遗传算法的特征选择和权重确定方法

摘要

在模式识别中，特征选择是其中非常重要的步骤，特征集的选择直接影响分类器的精度。该文提出了一种基于遗传算法的特征权重确定方法，首先使用传统遗传算法进行特征的初步选择，得到一个粗选的特征集；然后使用实数编码的遗传算法在第一步的基础上进一步精选特征，并确定每个入选特征的权重。通过实验和一些传统特征选择方法进行对比，结果显示，该文提出的算法取得了较好的效果。%Feature selection is a key step in pattern recognition and the selection of feature set affects the accuracy of classifier directly .A feature selection and weights determination method based on genetic algorithm is proposed in this paper . In the first step of this method ,a conventional genetic algorithm (GA) is used to obtain a lower dimensional candidate feature subset .Then a weighted GA is employed to determine the weights of the features .A lot of experiments are carried out . Experimental results demonstrate the effect of the proposed method .