利用粒子群算法优化SVM分类器的超参数  

Utilizing particle swarm optimization to optimize hyper-parameters of SVM classifier

作　　者：王东[1] 吴湘滨[2] 

机构地区：[1]佛山科学技术学院计算机科学与技术系,广东佛山528000 [2]中南大学地学与环境工程学院,长沙410083

出　　处：《计算机应用》2008年第1期134-135,139,共3页journal of Computer Applications

基　　金：国家自然科学基金资助项目(40473029)

摘　　要：利用粒子群算法在求解组合优化问题时具有的全局搜索特性,设计并实现了支持向量机分类器中超参数的优选粒子群算法,扼要地叙述了算法实现中个体编码和适应度函数,通过在国际标准数据集上的实验验证了算法的有效性和高效性,最后列举了一些在上述工作基础上可开展的深入性工作。Particle swarm optimization used for optimization selection for hyper-parameter of support vector machine classifier was designed and implemented utilizing global searching property of particle swarm optimization algorithm while the algorithm was used to solve combinatorial optimization problems. The method of individuals coding and evaluating was described in brief. The experimental statistic results demonstrate that the algorithm is effective and efficacious, In the end, some in-depth works are listed on the base of above-mentioned study.

关 键 词：支持向量机 分类器 参数优化 粒子群优化算法 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]