摘要
本文提出一种SVM参数优化的新方法。应用遗传算法先对SVM参数进行初步的优化,把得到的优化结果邻近的一段区域再作为粒子群算法的搜索区间进行二次优化,以提高支持向量机的泛化能力,缩短SVM参数寻优的时间。仿真实验表示,该方法在样本数据缺失的情况下,同样具有较好的泛化能力。
A new method in optimizing SVM classification's parameter is proposed. Following steps are applied:firstly,applying the GA to optimize the SVM parameters in a rough range,then expanding nearby section of the best result of the GA,finally,using PSO algorithm optimized SVM parameters with new searching section,to improve the generalization ability of SVM and reduce the time consumption of optimizing. Experiment results show that the proposed method has better generalization ability even in condition of losing sample data.
出处
《广西师范大学学报(自然科学版)》
CAS
北大核心
2011年第2期114-118,共5页
Journal of Guangxi Normal University:Natural Science Edition
基金
国家自然科学基金资助项目(61070062)