东篱科研大数据发现系统（DRDS）

位置：成果数据库 > 期刊 > 期刊详情页

基于核的四重子空间的LDA特征抽取方法

ISSN号：1673-4807
期刊名称：《江苏科技大学学报：自然科学版》
时间：0
分类：TP391.41[自动化与计算机技术—计算机应用技术;自动化与计算机技术—计算机科学与技术]
作者机构：[1]江苏科技大学计算机科学与工程学院,江苏镇江212003
相关基金：国家自然科学基金资助项目（60632050 60572034 60973094）; 江苏科技大学本科生创新计划课题

关键词：特征抽取, 线性鉴别分析, 子空间学习, 非线性小样本问题, 核主分量, feature extraction, linear discriminant analysis, subspace learning, nonlinear small sample size problem, kernel principal component analysis

中文摘要：

模式分类在面临非线性高维数据下的小样本问题时通常十分困难.文中提出了一种核的四重子空间学习（KFS）方法.首先通过构造基于类内和类间散布矩阵的混合鉴别准则,获得分布在各子空间中降维样本的最优鉴别信息.其次,通过向量点积,核鉴别分析方法（KFD）成为一种有效的抽取非线性鉴别信息的算法,在此基础上,提出了基于核的四重子空间鉴别分析算法,从而有效解决了非线性小样本问题的特征抽取.在ORL和Yale人脸库上的实验结果验证了该方法的有效性.

英文摘要：

Usually,classification of nonlinear high-dimensional data isn′t amenable to standard pattern recognition techniques because of an underlying nonlinear small sample size conditions.To address the problem,a novel kernel fourfold subspace learning（KFS） was developed.First,a hybrid discriminant criterion based on the Fisher theory was proposed by which fourfold subspaces learning derived from the within-class and between-class scatter matrices were constructed,respectively.Second,considering the fact that the kernel Fisher discriminant（KFD） was effective to extract nonlinear discriminative information of the input feature space by using kernel trick,a kernel algorithm was presented subsequently based on the new Fisher discriminant criterion,which had the potential to outperform the traditional subspace learning algorithms,especially in the cases of nonlinear small sample sizes.Experimental results conducted on the ORL and Yale face database demonstrate the effectiveness of the proposed method.