Online Kernel Principal Component Analysis: A Reduced-Order Model

被引:76
作者
Honeine, Paul [1 ]
机构
[1] Univ Technol Troyes, Inst Charles Delaunay, Lab Modelisat & Surete Syst, CNRS,UMR 6279, F-10010 Troyes, France
关键词
Principal component analysis; online algorithm; machine learning; reproducing kernel; Oja's rule; recursive algorithm; APPROXIMATION;
D O I
10.1109/TPAMI.2011.270
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
Kernel principal component analysis (kernel-PCA) is an elegant nonlinear extension of one of the most used data analysis and dimensionality reduction techniques, the principal component analysis. In this paper, we propose an online algorithm for kernel-PCA. To this end, we examine a kernel-based version of Oja's rule, initially put forward to extract a linear principal axe. As with most kernel-based machines, the model order equals the number of available observations. To provide an online scheme, we propose to control the model order. We discuss theoretical results, such as an upper bound on the error of approximating the principal functions with the reduced-order model. We derive a recursive algorithm to discover the first principal axis, and extend it to multiple axes. Experimental results demonstrate the effectiveness of the proposed approach, both on synthetic data set and on images of handwritten digits, with comparison to classical kernel-PCA and iterative kernel-PCA.
引用
收藏
页码:1814 / 1826
页数:13
相关论文
共 58 条
[21]   Merging and splitting eigenspace models [J].
Hall, P ;
Marshall, D ;
Martin, R .
IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE, 2000, 22 (09) :1042-1049
[22]  
Honeine P., 2009, P 19 IEEE WORKSH MAC
[23]   On-line nonlinear sparse approximation of functions [J].
Honeine, Paul ;
Richard, Cedric ;
Bermudez, Jose Carlos M. .
2007 IEEE INTERNATIONAL SYMPOSIUM ON INFORMATION THEORY PROCEEDINGS, VOLS 1-7, 2007, :956-+
[24]   Preimage Problem in Kernel-Based Machine Learning [J].
Honeine, Paul ;
Richard, Cedric .
IEEE SIGNAL PROCESSING MAGAZINE, 2011, 28 (02) :77-88
[25]  
Jolliffe I., 2002, PRINCIPAL COMPONENT, DOI [10.1007/978-1-4757-1904-8_7, 10.1016/0169-7439(87)80084-9]
[26]  
Kallas M., 2011, P 19 EUR C SIGN PROC
[27]  
Keerthi SS, 2006, J MACH LEARN RES, V7, P1493
[28]  
Kim KI, 2005, IEEE T PATTERN ANAL, V27, P1351, DOI 10.1109/TPAMI.2005.181
[29]  
KIM KI, 2003, 109 MAX PLANCK I BIO
[30]   SOME RESULTS ON TCHEBYCHEFFIAN SPLINE FUNCTIONS [J].
KIMELDORF, G ;
WAHBA, G .
JOURNAL OF MATHEMATICAL ANALYSIS AND APPLICATIONS, 1971, 33 (01) :82-+