计算机工程与应用 ›› 2008, Vol. 44 ›› Issue (35): 138-141.DOI: 10.3778/j.issn.1002-8331.2008.35.042

• 数据库、信号与信息处理 • 上一篇    下一篇

基于超球结构的渐进直推式支持向量机

李丽蓉1,牛惠芳2,薛贞霞3,4   

  1. 1.山西警官高等专科学校 计算机科学与技术系,太原 030021
    2.洛阳师范学院 数学科学学院,河南 洛阳471022
    3.河南科技大学 数学系,河南 洛阳471003
    4.西安电子科技大学 应用数学系,西安 710071
  • 收稿日期:2008-07-01 修回日期:2008-09-16 出版日期:2008-12-11 发布日期:2008-12-11
  • 通讯作者: 李丽蓉

Progressive transductive Support Vector Machines based on hypersphere structure

LI Li-rong1,NIU Hui-fang2,XUE Zhen-xia3,4   

  1. 1.Department of Computer Science and Technology,Shanxi Police Academy,Taiyuan 030021,China
    2.College of Mathematical Science,Luoyang Normal University,Luoyang,Henan 471022,China
    3.Department of Mathematics,Henan Science and Technology University,Luoyang,Henan 471003,China
    4.Department of Applied Mathematics,Xidian University,Xi’an 710071,China
  • Received:2008-07-01 Revised:2008-09-16 Online:2008-12-11 Published:2008-12-11
  • Contact: LI Li-rong

摘要: 针对渐进直推式支持向量机(Progressive Transductive Support Vector Machines,PTSVM)算法训练速度慢,学习性能不稳定的问题,提出一种基于超球结构的渐进直推式支持向量机算法。该算法首先利用支持向量域描述(Support Vector Domain Description,SVDD)得到包含每个类别的有标签样本点的最小包球,选择这个包球边界附近的无标签样本点进行标注,然后对目前所有有标签的样本点进行SVM训练。实验结果表明该算法不仅能保持甚至提高算法的精度,更重要的是能大大提高训练速度。

关键词: 半监督学习, 支持向量机, 直推式学习, 超球结构

Abstract: Progressive Transductive Support Vector Machines(PTSVM) has some drawbacks such as slower training speed,and unstable learning performance,a Progressive Transductive Support Vector Machines learning algorithm based on hypersphere structure is proposed.Labeled data of each class are described by using Support Vector Domain Description(SVDD) and the corresponding smallest enclosing hypersphere is obtained.Then,the method selects new unlabeled samples located near the boundary of the hypersphere.Finally,labeled data available is used to train standard SVM.Experiment results show the method can improve greatly the computing speed.Moreover,it can keep,in fact sometimes improve,the classification accuracy in general.

Key words: semi-supervised learning, Support Vector Machines(SVM), transductive learning, hypersphere structure