J4 ›› 2015, Vol. 42 ›› Issue (5): 120-124+160.doi: 10.3969/j.issn.1001-2400.2015.05.021

• 研究论文 • 上一篇    下一篇



  1. (北京航空航天大学 经济管理学院,北京  100191)
  • 收稿日期:2014-09-29 出版日期:2015-10-20 发布日期:2015-12-03
  • 通讯作者: 宋晓东
  • 作者简介:蔡艳艳(1976-),女,北京航空航天大学博士研究生,E-mail: caiyanyan@buaa.edu.cn.
  • 基金资助:


New fuzzy SVM model used in imbalanced datasets

CAI Yanyan;SONG Xiaodong   

  1. (School of Economics and Management, Beihang Univ., Beijing  100191, China)
  • Received:2014-09-29 Online:2015-10-20 Published:2015-12-03
  • Contact: SONG Xiaodong



关键词: 支持向量机, 分类, 非平衡数据集, 噪声, 惩罚函数


The paper proposes a new fuzzy SVM, called CI-FSVM(Class Imbalance Fuzzy Support Vector Machine) short for which is based on imbalanced datasets classification. By improving penalty functions, we reduce the sensitivity of the model for imbalanced datasets with “overlap”. In addition, the parameters in SVM models are optimized by the grid-parameter-search algorithm. The results show that the CI-FSVM has a better effect in imbalanced datasets classification compared with other models. It not only has a higher overall accuracy, but also improves are judgment accuracy when dealing with the minority classifications.

Key words: support vector machine, classification, imbalanced datasets, noise samples, penalty function