基于改进哈里斯鹰算法的SVM与特征选择同步优化

Support vector machine and feature selection simultaneous optimization based on improved Harris hawk algorithm

标准支持向量机结合封装式特征选择具有冗余特征多、分类准确率低的不足,为此,提出基于改进哈里斯鹰算法的特征选择同步优化策略。为改进特征子集选取能力和支持向量机的分类准确率,利用混沌映射、能量因子非线性调整和小孔成像对立学习对哈里斯鹰算法进行改进,将改进哈里斯鹰算法用于SVM参数调整和特征子集选取同步优化问题。实验结果表明,改进算法能够在降低特征维度的情况下实现更高的分类准确率,实现同步优化效果。

Standard support vector machine combined with encapsulated feature selection has the disadvantages of redundant features and low classification accuracy.Therefore,a synchronous optimization strategy of feature selection based on improved Harris hawk algorithm was proposed.To improve the feature subset selection ability and the classification accuracy of support vector machine SVM,chaos mapping,nonlinear adjustment of energy factor and small hole imaging opposition learning were used to improve Harris hawk algorithm.The improved Harris hawk algorithm was applied to the synchronous optimization of SVM parameter adjustment and feature subset selection.The results show that the improved algorithm can achieve higher classification accuracy and achieve effects of synchronous optimization.