摘要
针对支持向量机的参数对分类性能的影响,探讨了基于蚁群算法的支持向量机参数优化方法,建立了支持向量机参数优化模型,给出了基于网格划分策略的连续蚁群算法,并将其用于优化模型求解,通过对支持向量机的惩罚因子和径向基核函数进行优化,使支持向量机的分类性能最优。通过仿真和应用实例,验证了方法的有效性,得到了95%以上的分类正确率。
Parameters of support vector machine is the key factor that impacts its classifying performance. A parameter optimization method for support vector machine using ant colony optimization algorithm is discussed. A parameter optimization model is established. The continuous ant colony optimization method based on gridding partition is given and used to resolve the optimization model. The classifying performance reaches the best state by optimizing the penalty factor and the radial basis function. The validity of the method is tested by simulation and application instances, and more than 95% classified right rate is obtained.
出处
《南京理工大学学报》
EI
CAS
CSCD
北大核心
2009年第4期464-468,共5页
Journal of Nanjing University of Science and Technology
基金
国家自然科学基金(50705097)
关键词
蚁群算法
支持向量机
参数优化
油液分析
故障诊断
ant colony optimization algorithm
support vector machine
parameter optimization
analysis
fault diagnosis
