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

Optimal reduced-set vectors for support vector machines with a quadratic kernel.

Thorsten Thies1, Frank Weber

  • 1thorsten.thies@cognitec.com

Neural Computation
|August 5, 2004
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

A year in review in Minerva Anestesiologica 2025. Anesthesia, analgesia, and perioperative medicine.

Minerva anestesiologica·2026
Same author

IL10 is a potent modulator of IL1-mediated inflammation in bovine endometrial explants.

Reproduction, fertility, and development·2026
Same author

Dedicated operating room teams for highly specialized procedures in a tertiary children's hospital: the Rotterdam experience.

Minerva anestesiologica·2026
Same author

Spread and dynamics of Calicophoron daubneyi and Fasciola hepatica in an extensively kept water buffalo herd and suitability of an ELISA for detecting antibodies against F. hepatica.

Parasitology research·2026
Same author

Real-world data show limited efficacy of sequential cabozantinib treatment in RAIR-(P)DTC.

Endocrine-related cancer·2026
Same author

Drug Development.

Alzheimer's & dementia : the journal of the Alzheimer's Association·2025

This study presents a method to reduce computational cost in support vector machines (SVM) by minimizing the number of vectors. The proposed approach offers optimal approximation for quadratic kernels, improving efficiency in machine learning.

Area of Science:

  • Machine Learning
  • Computational Science

Background:

  • Support Vector Machines (SVM) are powerful classification tools.
  • Reducing the computational cost of SVMs is crucial for practical applications.
  • Existing methods for vector reduction in SVMs have limitations.

Purpose of the Study:

  • To develop an explicit solution for representing SVM discriminant functions with minimal vectors.
  • To achieve the best possible approximation for a given number of vectors using a general quadratic kernel.
  • To improve the efficiency and scalability of Support Vector Machines.

Main Methods:

  • The study focuses on a general quadratic kernel: k(x, x') = (C + D x^T x')^2.
  • An explicit solution is derived for minimizing vectors in the SVM discriminant function.

Related Experiment Videos

  • The core idea involves transforming an inhomogeneous kernel into a homogeneous one in a higher-dimensional space, following Burges (1996).

    • Main Results:

    • The proposed solution provides the best approximation for a fixed number of vectors.
    • The method can perfectly recover the discriminant function when a sufficient number of vectors are used.
    • Demonstrates improved computational efficiency for SVMs with quadratic kernels.

    Conclusions:

    • The developed method offers an effective way to reduce the computational cost of SVMs.
    • This approach enhances the practical applicability of SVMs, especially those utilizing quadratic kernels.
    • The technique provides a significant advancement in optimizing Support Vector Machine performance.