董国君, 哈力木拉提·买买提, 余辉. 基于RBF核的SVM核参数优化算法[J]. Journal of Xinjiang University (Natural Science Edition in Chinese and English), 2009, 26(3): 355-358.DOI:
基于RBF核的SVM核参数优化算法
摘要
SVM是一种新型的机器学习方法
其分类性能的优劣主要受核函数及核参数的影响
国内外学者针对SVM核参数的选择已提出许多算法.本文首先分析了RBF核参数对SVM分类性能的影响
然后又对比分析了目前存在的几种基于RBF核的SVM核参数选择方法.通过实验
发现使用遗传算法选择核参数的SVM有比较快的搜索速度.
Abstract
Support vector machine(SVM) is a new kind of machine learning methods
whose classification performance is affected by kernel function and kernel parameters.Nowadays some researchers have proposed many algorithms about choosing kernel parameters.This paper emphasizes on analyzing kernel parameters choice algorithms of SVM with RBF kernel.The paper finds that genetic algorithm which is used in choosing kernel parameters could improve search velocity of SVM.
关键词
Keywords
references
边肇祺,张学工,编著.模式识别[M],第二版。北京:清华大学出版社,2000,284-304.
Vladimir N.Vapnik著,张学工译.统计学习理论的本质[M].北京:清华大学出版社,2000,105-120.
Goldsmith A J,Chua S G.Variable-rate Variable-power Mqam for Fading Channels[J].IEEE Trans.on Communications, 1997,45(10):1218-1230.