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

Information criteria for support vector machines.

Kei Kobayashi1, Fumiyasu Komaki

  • 1Institute of Statistical Mathematics, Tokyo 106-8569, Japan. kkoba@stat.t.u-tokyo.ac.jp

IEEE Transactions on Neural Networks
|May 26, 2006
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

Multicenter comparison of hybrid and Norwood procedures in patients with Fontan circulation.

The Journal of thoracic and cardiovascular surgery·2026
Same author

Impact of Primary Graft Dysfunction on Neurodevelopmental Outcomes in Pediatric Heart Transplant Recipients.

medRxiv : the preprint server for health sciences·2026
Same author

Social motivation modulates early visual processing for a few others' emotions.

International journal of psychophysiology : official journal of the International Organization of Psychophysiology·2026
Same author

Targeted Regulation of Protein Expression in <i>Vibrio parahaemolyticus</i>.

Biology·2026
Same author

Missing the Act on the Mission of Maintaining Quality Care.

JAMA surgery·2026
Same author

Enzyme-Directed Assembly of Antiparallel Cellulose II Nanocrystals: Unraveling the Mechanism Beyond Spontaneous Crystallization.

JACS Au·2026
Same journal

Universal perceptron and DNA-like learning algorithm for binary neural networks: LSBF and PBF implementations.

IEEE transactions on neural networks·2013
Same journal

Guest editorial: special section on white box nonlinear prediction models.

IEEE transactions on neural networks·2011
Same journal

Data-based fault-tolerant control of high-speed trains with traction/braking notch nonlinearities and actuator failures.

IEEE transactions on neural networks·2011
Same journal

Guest editorial: special section on data-based control, modeling, and optimization.

IEEE transactions on neural networks·2011
Same journal

Neural network-based multiple robot simultaneous localization and mapping.

IEEE transactions on neural networks·2011
Same journal

Data-driven model-free adaptive control for a class of MIMO nonlinear discrete-time systems.

IEEE transactions on neural networks·2011
See all related articles

A new Kernel Regularization Information Criterion (KRIC) efficiently tunes regularization parameters for kernel logistic regression (KLR) and support vector machines (SVMs). KRIC offers comparable performance to cross-validation but with significantly reduced computational cost.

Area of Science:

  • Machine Learning
  • Statistical Learning Theory

Background:

  • Kernel logistic regression (KLR) and support vector machines (SVMs) are powerful machine learning models.
  • Tuning regularization parameters is crucial for optimal model performance.
  • Existing methods for parameter tuning can be computationally expensive.

Purpose of the Study:

  • To introduce a novel criterion, the Kernel Regularization Information Criterion (KRIC), for tuning regularization parameters.
  • To enhance the computational efficiency of parameter tuning in KLR and SVMs.

Main Methods:

  • The KRIC is developed based on the principles of the Regularization Information Criterion (RIC).
  • An eigenvalue equation is derived for KRIC calculation.
  • The Nyström approximation is employed to reduce computational complexity.

Related Experiment Videos

Main Results:

  • The KRIC provides a computationally efficient method for tuning regularization parameters.
  • Parameter tuning using KRIC yields test error rates comparable to cross-validation and evidence evaluation.
  • The computational cost of KRIC is significantly lower than existing criteria.

Conclusions:

  • KRIC offers an effective and computationally efficient alternative for regularization parameter tuning in KLR and SVMs.
  • The Nyström approximation significantly reduces the computational burden of KRIC.