基于混沌更新策略的蜂群算法在SVM参数优化中的应用

发布时间：2018-03-24 21:20

本文选题：支持向量机　切入点：参数寻优　出处：《计算机工程与科学》2017年01期

【摘要】：针对支持向量机的参数寻优缺乏数学理论指导,传统人工蜂群算法易陷入长期停滞的不足,而混沌搜索算法具有很好的随机性和遍历性,提出了基于混沌更新策略人工蜂群支持向量机参数选择模型(IABC-SVM)。该模型利用混沌搜索对侦察蜂搜索方式进行改进,有效提高蜂群算法搜索效率。以UCI标准数据库中的数据进行数值实验,采用ACO-SVM、PSO-SVM、ABC-SVM作为对比模型,实验表明了IABC在SVM参数优化中的可行性和有效性,具有较高的预测准确率和较好的算法稳定性。
[Abstract]:In view of the lack of mathematical theory guidance for parameter optimization of support vector machines, the traditional artificial bee colony algorithm is easy to fall into long-term stagnation, while the chaotic search algorithm has good randomness and ergodicity. A parameter selection model of artificial bee swarm support vector machine (SVM) based on chaotic updating strategy is proposed in this paper. The model uses chaotic search to improve the search mode of reconnaissance bee. The efficiency of bee colony search is improved effectively. The numerical experiment is carried out on the data of UCI standard database, and ACO-SVMN PSO-SVMU ABC-SVM is used as a comparative model. The experiment shows the feasibility and effectiveness of IABC in the optimization of SVM parameters. It has higher prediction accuracy and better algorithm stability.
【作者单位】：辽宁工程技术大学优化与决策研究所;
【基金】：教育部高校博士学科科研基金联合资助项目(20132121110009) 辽宁省教育厅基金(L2015208)
【分类号】：TP18

【参考文献】