互联网 qkzz.net
全刊杂志网:首页 > 女性 > 文章正文
刊社推荐

模糊核覆盖分类器及其应用


  摘 要:核覆盖算法是一种性能优秀的分类算法,但在拒识点处理方面存在不足。对核覆盖算法的构造过程进行了分析,修改了算法中覆盖半径的选取原则,对拒识样本引入隶属度函数,将算法推广为模糊核覆盖算法。讨论了孤立覆盖对分类器的影响,对覆盖数进行精简,降低计算量。通过实验验证改进算法的性能,并与其他模糊分类方法进行对比。将模糊核覆盖算法应用于垃圾邮件过滤,实验结果表明过滤器的性能得到了有效提高。

  关键词:覆盖算法; 核函数; 拒识样本; 模糊核覆盖; 垃圾邮件过滤

  中图分类号:TP391文献标志码:A

  文章编号:1001-3695(2010)06-2043-04

  doi:10.3969/j.issn.1001-3695.2010.06.013

  Fuzzy kernel covering classifier and its application

  DUAN Zhena,b, CHENG Jia-xinga,b, ZHANG Linga,b

  (a.Key Laboratory IC & SP, Ministry of Education, b.School of Computer Science & Engineering, Anhui University, Hefei 230039, China)

  Abstract:While kernel covering algorithm (KCA) is a kind of effective classification algorithm, it still falls short in treating rejection points. This paper analyzed the construction process of kernel covering algorithm. By revising the selection criteria of covering radius and introducing the membership function for rejection points, generalized the algorithm as fuzzy kernel covering algorithm (FKCA). Also discussed isolated covering’s impact on classifier’s performance and lowered the computation cost through reducing the number of coverings. Comparing with other classification methods on experiment results show this fuzzy kernel covering algorithm works well. FKCA is applied to spam filtering and the classifier’s performance is improved effectively.

......
很抱歉,暂无全文,若需要阅读全文或喜欢本刊物请联系《计算机应用研究》杂志社购买。
欢迎作者提供全文,请点击编辑
分享:
 

了解更多资讯,请关注“木兰百花园”
分享:
 
精彩图文


关键字
支持中国杂志产业发展,请购买、订阅纸质杂志,欢迎杂志社提供过刊、样刊及电子版。
关于我们 | 网站声明 | 刊社管理 | 网站地图 | 联系方式 | 中图分类法 | RSS 2.0订阅 | IP查询
全刊杂志赏析网 2017