耦合人工鱼群算法在支持向量机参数优化中的应用  

Application of Coupled Artificial Fish Swarm Algorithm in Support Vector Machine Parameters Optimization

作　　者：高雷阜[1] 赵世杰[1] 高鼎[1] 

机构地区：[1]辽宁工程技术大学优化与决策研究所,辽宁阜新123000

出　　处：《小型微型计算机系统》2015年第1期172-176,共5页Journal of Chinese Computer Systems

基　　金：教育部高等学校博士学科点专项科研基金联合项目(20132121110009)资助;辽宁省教育厅基金项目(L2012105)资助

摘　　要：针对支持向量机的参数寻优到目前为止仍无一套系统的、完整的理论,同时如采用网格搜索法寻找最优参数又较为费时,单独使用人工鱼群算法寻优时需要反复调用预测数据集的情况下,提出了耦合人工鱼群算法用于支持向量机参数的优化选择.该模型利用训练好的神经网络来拟合人工鱼的食物浓度值,避免了反复调用预测数据集的麻烦;该模型既具有神经网络算法的较强非线性拟合能力,又具有鱼群算法的并行搜索能力和全局寻优性能等优点.数值试验的结果表明:耦合人工鱼群算法在支持向量机的参数寻优中具有较好的预测效果和较高的预测准确率.For the problems that the parameter optimization of support vector machines has＇t still a systematic and complete theory so far, the grid search method used to find the optimal parameter is relatively time consuming, and that the artificial fish swarm algorithm optimization is separately used to find the optimal parameter requires repeated calls the prediction data set, the coupled artificial fish swarm algorithm is proposed and is used to the parameter optimization of the support vector machine. The model uses the trained neu- ral network to fit the artificial fish food concentration value in order to avoiding the trouble of repeatedly calling the prediction data set. Meanwhile the model has a strong non-linear fitting ability of the neural network algorithm,and has the fish swarm algorithm＇s ad- vantages of parallel search ability and global searching performance. The results of the numerical experiments show that the coupled ar- tificial fish swarm algorithm in the optimization parameters of the support vector machine has better forecast effect and higher predic- tion accuracy.

关 键 词：支持向量机 人工鱼群算法 参数优化 神经网络算法 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]