计算机工程与应用 ›› 2019, Vol. 55 ›› Issue (21): 214-219.DOI: 10.3778/j.issn.1002-8331.1903-0381

• 工程与应用 • 上一篇    下一篇

乌鸦搜索算法在SVM参数优化中的应用

王丽婷,张金鑫,张金华   

  1. 1.福州大学 经济与管理学院,福州 350108
    2.湖北大学 商学院,武汉 430062
  • 出版日期:2019-11-01 发布日期:2019-10-30

Application of Crow Search Algorithm in SVM Parameter Optimization

WANG Liting, ZHANG Jinxin, ZHANG Jinhua   

  1. 1.School of Economics and Management, Fuzhou University, Fuzhou 350108, China
    2.School of Business, Hubei University, Wuhan 430062, China
  • Online:2019-11-01 Published:2019-10-30

摘要: 参数的选择对支持向量机(SVM)分类精度和泛化能力有至关重要的影响,而群体智能算法近年来在参数优化方面应用广泛,在此背景下提出CSA-SVM模型。该模型将分类准确率作为目标函数,利用乌鸦搜索算法(CSA)求得SVM的最优参数组合。为了验证CSA-SVM模型的分类性能,将该模型应用于6个标准分类数据集,并分别与遗传算法(GA)和粒子群(PSO)算法优化后的SVM模型进行性能比较。实验结果表明,CSA算法在SVM参数选择中具有更好地寻优能力和更快地寻优速度,CSA-SVM模型具有较高的分类准确率。

关键词: 乌鸦搜索算法, 支持向量机, 参数优化

Abstract: The selection of parameters have a crucial impact on the classification accuracy and generalization capabilities of Support Vector Machine(SVM) and the swarm intelligence algorithm has been widely used in parameter optimization in recent years. In this context, the CSA-SVM model is proposed. The model uses the classification error rate as the objective function, and applies the Crow Search Algorithm(CSA) to obtain the optimal parameter combination of the SVM. In order to verify the classification performance of the CSA-SVM model, the model is applied to six standard classification data sets and compares with the performance of the Genetic Algorithm(GA) and Particle Swarm Optimization(PSO) algorithm respectively. The experimental results show that the CSA algorithm has better searching ability and faster searching speed in the SVM parameter selection, and CSA-SVM model has high classification accuracy.

Key words: crow search algorithm, support vector machine, parameters optimization