[1]刘三阳 杜喆.一种改进的模糊支持向量机算法[J].智能系统学报,2007,2(3):30-33.
LIU San-yang,DU Zhe.An improved fuzzy support vector machine method[J].CAAI Transactions on Intelligent Systems,2007,2(3):30-33.
点击复制
《智能系统学报》[ISSN 1673-4785/CN 23-1538/TP] 卷:
2
期数:
2007年第3期
页码:
30-33
栏目:
学术论文—机器学习
出版日期:
2007-06-25
- Title:
-
An improved fuzzy support vector machine method
- 文章编号:
-
1673-4785(2007)03-0030-04
- 作者:
-
刘三阳 杜喆
-
西安电子科技大学理学院,陕西西安710071
- Author(s):
-
LIU San-yang,DU Zhe
-
School of Science, Xidian University, Xi’an 710071, China
-
- 关键词:
-
模糊支持向量机; 隶属度函数; 分类
- Keywords:
-
fuzzy support vector machine; membership function; classification
- 分类号:
-
TP181
- 文献标志码:
-
A
- 摘要:
-
模糊隶属度函数设计是模糊支持向量机中的关键步骤.Lin & Wang提出的基于类中心距离的模糊隶属度设计方法,不能从样本集中有效区分噪声或野值点,而且可能降低支持向量的隶属度.针对上述不足,提出一种改进的隶属度函数设计方法.通过引入一个半径控制因子,充分利用样本间的信息,更加合理地设计样本的模糊隶属度.与基于类中心的隶属度方法相比,该方法在不增加时间复杂度的情况下,通过数值实验表明了方法的优势,大大提高了模糊支持向量机的分类精度.
- Abstract:
-
A design that improves the classifying ability of an SVM by improving the assignment of fuzzy membership is an important step in solving the fuzzy SVM problem. In this paper, a radius controlling factor is introduced to assign sample membership more accurately. This technique can distinguish noise and outliers and increase accuracy of membership in support vectors, compensating for the disadvantages of assigning fuzzy membership based on the distance between a sample and its cluster center proposed by Lin and Wang. Experimental results verify the effectiveness of this method .
更新日期/Last Update:
2009-05-07