计算机工程与应用2018，Vol.54Issue(4)：128-134,7.DOI:10.3778/j.issn.1002-8331.1608-0557
一种新的L1度量Fisher线性判别分析研究
Study of Fisher linear discriminant analysis based on L1 -norm
摘要
Abstract
Fisher Linear Discriminant Analysis(FLDA)is a classical method of feature extraction with supervised infor-mation,which maximizes the Fisher criterion to find the optimal projection matrix.In the criterion of standard FLDA,the involved metric is based on L2norm metric, which is usually lack of robustness and sensitive to outliers. In order to improve the robustness,this paper proposes a new model and algorithm for FLDA,which is based on L1norm metric. The experimental results show that,FLDA with L1norm outperforms that with L2norm in classification accuracy and robustness in many cases.关键词
Fisher线性判别分析/Fisher准则/L1范数度量/鲁棒性/特征提取Key words
Fisher linear discriminant analysis/Fisher criterion/L1norm metric/robustness/feature extraction分类
信息技术与安全科学引用本文复制引用
余景丽,胡恩良,张涛..一种新的L1度量Fisher线性判别分析研究[J].计算机工程与应用,2018,54(4):128-134,7.基金项目
国家自然科学基金(No.61165012,No.61663049). （No.61165012,No.61663049）
