Jove
Visualize
Contact Us
JoVE
x logofacebook logolinkedin logoyoutube logo
ABOUT JoVE
OverviewLeadershipBlogJoVE Help Center
AUTHORS
Publishing ProcessEditorial BoardScope & PoliciesPeer ReviewFAQSubmit
LIBRARIANS
TestimonialsSubscriptionsAccessResourcesLibrary Advisory BoardFAQ
RESEARCH
JoVE JournalMethods CollectionsJoVE Encyclopedia of ExperimentsArchive
EDUCATION
JoVE CoreJoVE BusinessJoVE Science EducationJoVE Lab ManualFaculty Resource CenterFaculty Site
Terms & Conditions of Use
Privacy Policy
Policies

Related Experiment Videos

A robust subspace algorithm for principal component analysis.

Andreas Weingessel1, Kurt Hornik

  • 1Institut für Statistik und Wahrscheinlichkeitstheorie, Technische Universität Wien, Vienna, Austria. Andreas.Weingessel@ci.tuwien.ac.at

International Journal of Neural Systems
|December 4, 2003
PubMed
Summary
This summary is machine-generated.

Related Concept Videos

You might also read

Related Articles

Articles linked to this work by shared authors, journal, and citation graph.

Sort by
Same author

Two Monotonicity Results for Beta Distribution Functions.

Entropy (Basel, Switzerland)·2024
Same author

Motivation, values, and work design as drivers of participation in the R open source project for statistical computing.

Proceedings of the National Academy of Sciences of the United States of America·2015
Same author

The grand old party - a party of values?

SpringerPlus·2014
Same author

On maximum likelihood estimation of the concentration parameter of von Mises-Fisher distributions.

Computational statistics·2014
Same author

Amos-type bounds for modified Bessel function ratios.

Journal of mathematical analysis and applications·2014
Same author

On standard conjugate families for natural exponential families with bounded natural parameter space.

Journal of multivariate analysis·2014

We developed a noise-robust principal component analysis (PCA) algorithm, extending Oja's method. This algorithm allows adjustable noise sensitivity and identifies stable equilibria by minimizing a derived loss function for noisy data.

Area of Science:

  • Machine Learning
  • Statistical Analysis
  • Signal Processing

Background:

  • Principal Component Analysis (PCA) is a fundamental dimensionality reduction technique.
  • Standard PCA is sensitive to noisy data, limiting its application in real-world scenarios.
  • Oja's subspace algorithm provides a foundation for online PCA but requires adaptation for noise robustness.

Purpose of the Study:

  • To introduce a novel noise-robust PCA algorithm.
  • To enable tunable noise sensitivity in PCA.
  • To analyze the stability properties of the proposed algorithm in noisy environments.

Main Methods:

  • Extension of the Oja subspace algorithm.
  • Derivation of a novel loss function.
  • Local stability analysis of the algorithm's equilibria.

Related Experiment Videos

Main Results:

  • The proposed algorithm demonstrates robustness to noise.
  • Noise sensitivity can be adjusted by tuning parameters.
  • Locally stable equilibria correspond to the minima of the derived loss function.

Conclusions:

  • The developed algorithm offers a robust alternative to standard PCA for noisy datasets.
  • The theoretical analysis confirms the algorithm's stability properties.
  • This work provides a principled approach to handling noise in PCA.