基于进化计算方法的支持向量机特征选择

针对支持向量机训练过程中的特征选择问题,提出了基于进化计算方法的支持向量机特征选择方法,该方法采用进化计算方法中2个典型的算法:遗传算法(GA)和微粒群算法(PSO)为支持向量机选择优化的特征子集。通过对UCI标准数据集的实验,比较了遗传算法和微粒群算法的使用效果。仿真实验结果证明了该方法的有效性。

Feature Selection for SVM with Evolutionary Computing Method

Aiming at the problem of feature selection in SVM training process,a new feature selection method for SVM(Support Vector Machine)based on evolutionary computing method is proposed.For the new method,two typical evolutionary computing methods: genetic algorithm(GA) and particle swarm optimal(PSO) algorithm are adopted to select the optimal feature subset for SVM.Practical performance of these two algorithms are compared through the simulation experiments on several UCI benchmark dataset,the simulation results demonstrate the validity of the proposed method.