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 Video

Updated: Jun 26, 2026

ARL Spectral Fitting as an Application to Augment Spectral Data via Franck-Condon Lineshape Analysis and Color Analysis
07:11

ARL Spectral Fitting as an Application to Augment Spectral Data via Franck-Condon Lineshape Analysis and Color Analysis

Published on: August 19, 2021

Spectral methods in machine learning and new strategies for very large datasets.

Mohamed-Ali Belabbas1, Patrick J Wolfe

  • 1Department of Statistics, School of Engineering and Applied Sciences, Oxford Street, Harvard University, Cambridge, MA 02138, USA.

Proceedings of the National Academy of Sciences of the United States of America
|January 9, 2009
PubMed
Summary

This study introduces two efficient Nyström-based algorithms for approximating positive-semidefinite kernels, crucial for large datasets in statistics and machine learning. These methods offer improved error bounds for scalable spectral analysis.

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

Statistical properties of linear prediction analysis underlying the challenge of formant bandwidth estimation.

The Journal of the Acoustical Society of America·2015
Same author

Network histograms and universality of blockmodel approximation.

Proceedings of the National Academy of Sciences of the United States of America·2014
Same author

Making sense of big data.

Proceedings of the National Academy of Sciences of the United States of America·2013
Same author

Kalman-based autoregressive moving average modeling and inference for formant and antiformant tracking.

The Journal of the Acoustical Society of America·2012
Same author

A CURE for noisy magnetic resonance images: chi-square unbiased risk estimation.

IEEE transactions on image processing : a publication of the IEEE Signal Processing Society·2012
Same author

On landmark selection and sampling in high-dimensional data analysis.

Philosophical transactions. Series A, Mathematical, physical, and engineering sciences·2009

Area of Science:

  • Statistics and Machine Learning
  • Data Science
  • Computational Mathematics

Background:

  • Spectral methods are foundational in statistics and machine learning, underpinning algorithms from Principal Component Analysis (PCA) to manifold learning.
  • A key challenge is computing low-rank approximations of positive-definite kernels, essential for many algorithms.
  • Exact spectral decomposition is computationally prohibitive for very large or high-dimensional datasets due to cubic scaling complexity.

Purpose of the Study:

  • To develop novel, efficient algorithms for approximating positive-semidefinite kernels applicable to massive datasets.
  • To provide improved error bounds compared to existing literature for kernel approximation.
  • To address the computational limitations of exact spectral decomposition in big data scenarios.

More Related Videos

Applying Hyperspectral Reflectance Imaging to Investigate the Palettes and the Techniques of Painters
07:05

Applying Hyperspectral Reflectance Imaging to Investigate the Palettes and the Techniques of Painters

Published on: June 18, 2021

Related Experiment Videos

Last Updated: Jun 26, 2026

ARL Spectral Fitting as an Application to Augment Spectral Data via Franck-Condon Lineshape Analysis and Color Analysis
07:11

ARL Spectral Fitting as an Application to Augment Spectral Data via Franck-Condon Lineshape Analysis and Color Analysis

Published on: August 19, 2021

Applying Hyperspectral Reflectance Imaging to Investigate the Palettes and the Techniques of Painters
07:05

Applying Hyperspectral Reflectance Imaging to Investigate the Palettes and the Techniques of Painters

Published on: June 18, 2021

Main Methods:

  • Introduced two new algorithms based on the Nyström method for efficient kernel approximation.
  • Developed a randomized algorithm using kernel-induced probability distributions on data partitions (sampling-based).
  • Developed a deterministic algorithm for data partition selection based on sorting.

Main Results:

  • Presented two new strategies for approximating positive-semidefinite kernels that are scalable to massive datasets.
  • Achieved error bounds that surpass existing results in the literature.
  • Demonstrated improved performance over existing methods through simulations on various statistical data analysis problems.

Conclusions:

  • The proposed Nyström-based algorithms provide efficient and accurate solutions for kernel approximation in large-scale machine learning.
  • These methods effectively reduce computational complexity, making spectral analysis feasible for big data.
  • The sampling and sorting strategies offer flexible and powerful tools for modern data analysis challenges.