作者投稿和查稿 主编审稿 专家审稿 编委审稿 远程编辑

计算机工程 ›› 2013, Vol. 39 ›› Issue (5): 165-168,173. doi: 10.3969/j.issn.1000-3428.2013.05.036

• 人工智能及识别技术 • 上一篇    下一篇

基于多核支持向量数据描述的单类分类方法

吴定海    a,张培林     a,王怀光     a,傅建平     b   

  1. (军械工程学院 a. 车辆与电气工程系;b. 火炮工程系,石家庄 050003)
  • 收稿日期:2012-02-24 出版日期:2013-05-15 发布日期:2013-05-14
  • 作者简介:吴定海(1981-),男,讲师、博士,主研方向:机械状态监测,信号处理,人工智能;张培林,教授、博士生导师;王怀光,讲师、博士研究生;傅建平,副教授、博士

One-class Classification Method Based on Multi-kernel Support Vector Data Description

WU Ding-hai a, ZHANG Pei-lin a, WANG Huai-guang a, FU Jian-ping b   

  1. (a. Department of Vehicle and Electrical Engineering; b. Department of Ordnance Engineering, Artillery Engineering College, Shijiazhuang 050003, China)
  • Received:2012-02-24 Online:2013-05-15 Published:2013-05-14

摘要: 针对异常检测模型中,单核支持向量数据描述存在映射形式单一以及核函数、核参数选择困难的问题,提出一种多核优化组合的支持向量域描述的单类分类方法。在分析多核映射的核空间基础上,建立多核支持向量数据描述模型,以更灵活地描述训练样本在高维特征空间的边界分布情况。采用目标函数的梯度下降法对该模型的多核组合权重进行分步寻优,并引入异常类测试样本来控制和评价分类器的描述精度和推广能力。仿真实验结果表明,该方法具有更好的学习能力和计算效率。

关键词: 模式识别, 单类分类, 多核学习, 支持向量数据描述, 异常检测

Abstract: Considering the support vector data description model having disadvantages of simple form with only one kernel information and hard to choose the best kernel and its parameters, the multi-kernel one-class classification with a linear combination of multi-kernel is proposed. The multi-kernel support vector data description model which can descript the data distribution boundary in eigenspace more flexibly is built after analysing the space of multi-kernel mapping. The optimal combination kernels’ weight is solved by reduced gradient algorithm. Test dataset which includes abnormal samples is introduced to control and evaluate the description accuracy and expansibility of hyper spherical interface. Experimental results show the method has better learning ability and computing efficiency.

Key words: pattern recognition, one-class classification, multi-kernel learning, support vector data description, anomaly detection

中图分类号: