Efficient online subspace learning with an indefinite kernel for visual tracking and recognition

We propose an exact framework for online learning with a family of indefinite (not positive) kernels. As we study the case of nonpositive kernels, we first show how to extend kernel principal component analysis (KPCA) from a reproducing kernel Hilbert space to Krein space. We then formulate an incre...

Full description

Bibliographic Details
Main Authors: Liwicki, Stephan, Zafeiriou, Stefanos, Tzimiropoulos, Georgios, Pantic, Maja
Format: Article
Published: IEEE 2012
Subjects:
Online Access:https://eprints.nottingham.ac.uk/31426/
_version_ 1848794198652747776
author Liwicki, Stephan
Zafeiriou, Stefanos
Tzimiropoulos, Georgios
Pantic, Maja
author_facet Liwicki, Stephan
Zafeiriou, Stefanos
Tzimiropoulos, Georgios
Pantic, Maja
author_sort Liwicki, Stephan
building Nottingham Research Data Repository
collection Online Access
description We propose an exact framework for online learning with a family of indefinite (not positive) kernels. As we study the case of nonpositive kernels, we first show how to extend kernel principal component analysis (KPCA) from a reproducing kernel Hilbert space to Krein space. We then formulate an incremental KPCA in Krein space that does not require the calculation of preimages and therefore is both efficient and exact. Our approach has been motivated by the application of visual tracking for which we wish to employ a robust gradient-based kernel. We use the proposed nonlinear appearance model learned online via KPCA in Krein space for visual tracking in many popular and difficult tracking scenarios. We also show applications of our kernel framework for the problem of face recognition.
first_indexed 2025-11-14T19:12:23Z
format Article
id nottingham-31426
institution University of Nottingham Malaysia Campus
institution_category Local University
last_indexed 2025-11-14T19:12:23Z
publishDate 2012
publisher IEEE
recordtype eprints
repository_type Digital Repository
spelling nottingham-314262020-05-04T16:34:08Z https://eprints.nottingham.ac.uk/31426/ Efficient online subspace learning with an indefinite kernel for visual tracking and recognition Liwicki, Stephan Zafeiriou, Stefanos Tzimiropoulos, Georgios Pantic, Maja We propose an exact framework for online learning with a family of indefinite (not positive) kernels. As we study the case of nonpositive kernels, we first show how to extend kernel principal component analysis (KPCA) from a reproducing kernel Hilbert space to Krein space. We then formulate an incremental KPCA in Krein space that does not require the calculation of preimages and therefore is both efficient and exact. Our approach has been motivated by the application of visual tracking for which we wish to employ a robust gradient-based kernel. We use the proposed nonlinear appearance model learned online via KPCA in Krein space for visual tracking in many popular and difficult tracking scenarios. We also show applications of our kernel framework for the problem of face recognition. IEEE 2012-09-10 Article PeerReviewed Liwicki, Stephan, Zafeiriou, Stefanos, Tzimiropoulos, Georgios and Pantic, Maja (2012) Efficient online subspace learning with an indefinite kernel for visual tracking and recognition. Neural Networks and Learning Systems, IEEE Transactions on, 23 (10). pp. 1624-1636. ISSN 2162-237X Face Recognition Gradient Methods Learning (Artificial Intelligence) Object Tracking Principal Component Analysis http://ieeexplore.ieee.org/xpl/articleDetails.jsp?arnumber=6269106 doi:10.1109/TNNLS.2012.2208654 doi:10.1109/TNNLS.2012.2208654
spellingShingle Face Recognition
Gradient Methods
Learning (Artificial Intelligence)
Object Tracking
Principal Component Analysis
Liwicki, Stephan
Zafeiriou, Stefanos
Tzimiropoulos, Georgios
Pantic, Maja
Efficient online subspace learning with an indefinite kernel for visual tracking and recognition
title Efficient online subspace learning with an indefinite kernel for visual tracking and recognition
title_full Efficient online subspace learning with an indefinite kernel for visual tracking and recognition
title_fullStr Efficient online subspace learning with an indefinite kernel for visual tracking and recognition
title_full_unstemmed Efficient online subspace learning with an indefinite kernel for visual tracking and recognition
title_short Efficient online subspace learning with an indefinite kernel for visual tracking and recognition
title_sort efficient online subspace learning with an indefinite kernel for visual tracking and recognition
topic Face Recognition
Gradient Methods
Learning (Artificial Intelligence)
Object Tracking
Principal Component Analysis
url https://eprints.nottingham.ac.uk/31426/
https://eprints.nottingham.ac.uk/31426/
https://eprints.nottingham.ac.uk/31426/