引用本文:周国华, 申燕萍, 殷新春.基于凸壳的在线单类学习机[J].西南大学学报(自然科学版),2018,40(12):163~172
【打印本页】   【HTML】   【下载PDF全文】   查看/发表评论  【EndNote】   【RefMan】   【BibTex】
←前一篇|后一篇→ 过刊浏览    高级检索
本文已被:浏览 33次   下载 76 本文二维码信息
码上扫一扫!
分享到: 微信 更多
基于凸壳的在线单类学习机
周国华, 申燕萍, 殷新春1,2
1. 常州轻工职业技术学院 信息工程系, 江苏 常州 213164;2. 扬州大学 信息工程学院, 江苏 扬州 225127
摘要:
传统的基于支持向量机的单类分类器因计算复杂度高而无法满足大规模数据实时处理的需求,在线学习方法为解决该问题提供了一种有效途径.本文在挖掘样本数据在特征空间分布性状的基础上,提出了一种基于凸壳的在线单类学习机(One-class Online Classifier based on Convex Hull,OOCCH).该方法首先使用凸壳的定义选择能代表特征空间中数据分布的凸壳向量对应的原始样本作为训练样本来缩减训练集的规模;其次在分类器在线更新阶段利用凸壳向量动态地调整分类器的训练样本.理论分析证明了OOCCH的有效性,与现有的在线单类分类器的实验比较,OOCCH在训练时间和分类性能方面有显著优势.
关键词:  在线学习  单类  分类  凸壳
DOI:10.13718/j.cnki.xdzk.2018.12.025
分类号:TP391.4
基金项目:国家自然科学基金资助项目(61472343).
A One-Class Online Classifier Based on Convex Hull
ZHOU Guo-hua, SHEN Yan-ping, YIN Xin-chun1,2
1. Department of Information Engineering, Changzhou Institute of Light Industry Technology, Changzhou Jiangsu 213164, China;2. College of Information Engineering, Yangzhou University, Yangzhou Jiangsu 225127, China
Abstract:
Facing the challenge of large-scale data processing, the traditional SVM(support vector machine) based one-class classifier suffers from its high computational complexity. The online learning technique is an effective way to solve this problem. In this paper, a one-class online classifier based on convex hull (OOCCH) is proposed by considering the distribution characteristics of the data in the feature space. In order to reduce the number of training sets, OOCCH selects the samples corresponding to the convex hull vectors in the feature space as training samples. In the online update stage of the classifier, OOCCH dynamically adjusts the training samples based on the definition of convex hull. Theoretical analysis proves the effectiveness of OOCCH. Compared with the existing online one-class classifiers in experiments, OOCCH has significant advantages in training time and classification performance.
Key words:  online learning  one-class  classification  convex hull
手机扫一扫看