Angle 2DPCA: A New Formulation for 2DPCA
Angle 2DPCA: A New Formulation for 2DPCA
复制标题
DOI：
10.1109/tcyb.2017.2712740
复制
发表时间：
2018-05-01
影响因子：
11.8
通讯作者：
Nie, Feiping
中科院分区：
文献类型：
--
作者：
Gao, Quanxue;Ma, Lan;Nie, Feiping
二维主成分分析（2DPCA）采用平方 F 范数作为距离度量，已广泛应用于数据表示和分类的降维中。然而，众所周知，平方 F 范数对异常值非常敏感。为了解决这个问题，我们提出了一种新的 2DPCA 公式，即 Angle-2DPCA。它采用F-范数作为距离度量，并考虑重建误差与目标函数方差之间的关系。我们提出了一种快速迭代算法来求解 Angle-2DPCA。 Extended Yale B、AR 和 PIE 人脸图像数据库上的实验结果说明了我们提出的方法的有效性。
2-D principal component analysis (2DPCA), which employs squared F-norm as the distance metric, has been widely used in dimensionality reduction for data representation and classification. It, however, is commonly known that squared F-norm is very sensitivity to outliers. To handle this problem, we present a novel formulation for 2DPCA, namely Angle-2DPCA. It employs F-norm as the distance metric and takes into consideration the relationship between reconstruction error and variance in the objective function. We present a fast iterative algorithm to solve the solution of Angle-2DPCA. Experimental results on the Extended Yale B, AR, and PIE face image databases illustrate the effectiveness of our proposed approach.