Euler principal component analysis

Principal Component Analysis (PCA) is perhaps the most prominent learning tool for dimensionality reduction in pattern recognition and computer vision. However, the ℓ 2-norm employed by standard PCA is not robust to outliers. In this paper, we propose a kernel PCA method for fast and robust PCA, whi...

Full description

Bibliographic Details
Main Authors: Liwicki, Stephan, Tzimiropoulos, Georgios, Zafeiriou, Stefanos, Pantic, Maja
Format: Article
Published: Springer 2013
Subjects:
Online Access:https://eprints.nottingham.ac.uk/31427/
_version_ 1848794198906503168
author Liwicki, Stephan
Tzimiropoulos, Georgios
Zafeiriou, Stefanos
Pantic, Maja
author_facet Liwicki, Stephan
Tzimiropoulos, Georgios
Zafeiriou, Stefanos
Pantic, Maja
author_sort Liwicki, Stephan
building Nottingham Research Data Repository
collection Online Access
description Principal Component Analysis (PCA) is perhaps the most prominent learning tool for dimensionality reduction in pattern recognition and computer vision. However, the ℓ 2-norm employed by standard PCA is not robust to outliers. In this paper, we propose a kernel PCA method for fast and robust PCA, which we call Euler-PCA (e-PCA). In particular, our algorithm utilizes a robust dissimilarity measure based on the Euler representation of complex numbers. We show that Euler-PCA retains PCA’s desirable properties while suppressing outliers. Moreover, we formulate Euler-PCA in an incremental learning framework which allows for efficient computation. In our experiments we apply Euler-PCA to three different computer vision applications for which our method performs comparably with other state-of-the-art approaches.
first_indexed 2025-11-14T19:12:23Z
format Article
id nottingham-31427
institution University of Nottingham Malaysia Campus
institution_category Local University
last_indexed 2025-11-14T19:12:23Z
publishDate 2013
publisher Springer
recordtype eprints
repository_type Digital Repository
spelling nottingham-314272020-05-04T20:19:38Z https://eprints.nottingham.ac.uk/31427/ Euler principal component analysis Liwicki, Stephan Tzimiropoulos, Georgios Zafeiriou, Stefanos Pantic, Maja Principal Component Analysis (PCA) is perhaps the most prominent learning tool for dimensionality reduction in pattern recognition and computer vision. However, the ℓ 2-norm employed by standard PCA is not robust to outliers. In this paper, we propose a kernel PCA method for fast and robust PCA, which we call Euler-PCA (e-PCA). In particular, our algorithm utilizes a robust dissimilarity measure based on the Euler representation of complex numbers. We show that Euler-PCA retains PCA’s desirable properties while suppressing outliers. Moreover, we formulate Euler-PCA in an incremental learning framework which allows for efficient computation. In our experiments we apply Euler-PCA to three different computer vision applications for which our method performs comparably with other state-of-the-art approaches. Springer 2013-02 Article PeerReviewed Liwicki, Stephan, Tzimiropoulos, Georgios, Zafeiriou, Stefanos and Pantic, Maja (2013) Euler principal component analysis. International Journal of Computer Vision, 101 (3). pp. 498-518. ISSN 1573-1405 Euler PCA Robust Subspace Online Learning Tracking Background Modeling http://link.springer.com/article/10.1007%2Fs11263-012-0558-z doi:10.1007/s11263-012-0558-z doi:10.1007/s11263-012-0558-z
spellingShingle Euler PCA
Robust Subspace
Online Learning
Tracking
Background Modeling
Liwicki, Stephan
Tzimiropoulos, Georgios
Zafeiriou, Stefanos
Pantic, Maja
Euler principal component analysis
title Euler principal component analysis
title_full Euler principal component analysis
title_fullStr Euler principal component analysis
title_full_unstemmed Euler principal component analysis
title_short Euler principal component analysis
title_sort euler principal component analysis
topic Euler PCA
Robust Subspace
Online Learning
Tracking
Background Modeling
url https://eprints.nottingham.ac.uk/31427/
https://eprints.nottingham.ac.uk/31427/
https://eprints.nottingham.ac.uk/31427/