On Robustness of Kernel Principal Component Analysis using Fast HCS

admin August 28, 2019

Authors:

Lekaa Ali Muhamed,Hayder Yahya mohammed,

DOI NO:

https://doi.org/10.26782/jmcms.2019.08.00024

Keywords:

High-dimensional data,outlier detection,kernel principal component analysis(KPCA),FastHCS (High-dimensional Congruent Subsets),

Abstract

When dealing with multivariate data with higher dimensions, we often use principal component analysis (PCA) to lessen the dimensions, but in the case of nonlinear data it is not possible to deal with classic estimated because of obtaining misleading results and therefore using kernel methods , when data contain outliers the results of the kernel pca (KPCA) for correlation matrix or variance covariance matrix are inaccurate. The aim of this research is to employ Robust KPCA (RKPCA) to solve nonlinear data using kernel function and outlier observation using Robust method termed as FastHCS (High-dimensional Congruent Subsets) that stands for a robust PCA algorithm appropriate for high-dimensional appliances to know the most effect variables on the phenomenon.

Refference:

I. Fan J, Chow TW. Exactly Robust Kernel Principal Component
Analysis. IEEE transactions on neural networks and learning systems.
2019 Apr 29.
II. Hall P,Marshall D, Martin R. Adding and subtracting eigenspaces with
eigenvalue decomposition and singular value decomposition. Image
and Vision Computing. (2002):1009-1016.
III. Huang HH, Yeh YR. An iterative algorithm for robust kernel principal
component analysis. Neurocomputing. 2011 Nov 1;74(18):3921-30.
IV. Lee JM, Yoo C, Choi SW, Vanrolleghem PA, Lee IB. Nonlinear
process monitoring using kernel principal component analysis.
Chemical engineering science. 2004 Jan 1;59(1):223-34.
V. Lu C, Zhang T, Zhang R, Zhang C. Adaptive robust kernel PCA
algorithm. In2003 IEEE International Conference on Acoustics, Speech,
and Signal Processing, 2003. Proceedings.(ICASSP’03). 2003 Apr 6
(Vol. 6, pp. VI-621). IEEE.
VI. Mika S, Schölkopf B, Smola AJ, Müller KR, Scholz M, Rätsch G.
Kernel PCA and de-noising in feature spaces. InAdvances in neural
information processing systems 1999 (pp. 536-542).
VII. Neil HT. Applied multivariate analysis. Board.(NY, USA), Springer,
(2002). Lekaa Ali Muhamed et al. 303
VIII. PITOY C. EMPIRICAL STUDY ON BANDWIDTH
OPTIMIZATION FOR KERNEL PCA IN THE K-MEANS
CLUSTERING OF NON-LINEARLY SEPARABLE DATA. Journal
of Theoretical & Applied Information Technology. 2017 Oct 31;95(20).
IX. Samuel RT, Cao Y. Nonlinear process fault detection and identification
using kernel PCA and kernel density estimation. Systems Science &
Control Engineering. 2016 Jan 1;4(1):165-74.
X. 8Schölkopf B, Smola A, Müller KR. Nonlinear component analysis as
a kernel eigenvalue problem. Neural computation. 1998 Jul
1;10(5):1299-319.
XI. Shams MA. Fault Identification using Kernel Principle Component
Analysis. IFAC Proceedings Volumes. 2011 Jan 1;44(1):4320-5.
XII. Schmitt E, Vakili K. The FastHCS algorithm for robust PCA. Statistics
and Computing. 2016 Nov 1;26(6):1229-42.
XIII. Schmitt E, Vakili K. Finding multivariate outliers with FastPCS.
Computational Statistics & Data Analysis. 2014 Jan 1;69:54-66.
XIV. Wu W, Massart DL, De Jong S. The kernel PCA algorithms for wide
data. Part I: theory and algorithms. Chemometrics and Intelligent
Laboratory Systems. 1997 Apr 1;36(2):165-72.

View Download