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

Classification With Truncated Distance Kernel.

Xiaolin Huang, Johan A K Suykens, Shuning Wang

    IEEE Transactions on Neural Networks and Learning Systems
    |April 1, 2017
    PubMed
    Summary

    A new truncated distance (TL1) kernel creates classifiers that are nonlinear globally but linear locally. This nonlinear kernel shows promise for classification tasks, offering adaptive performance comparable to established methods.

    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

    SteuerLLM: local specialized large language model for German tax law analysis.

    Scientific reports·2026
    Same author

    Inferring and evaluating network medicine-based disease modules with nextflow.

    Bioinformatics (Oxford, England)·2026
    Same author

    Generating Alzheimer's narratives using large language models.

    BMC medical informatics and decision making·2026
    Same author

    Pushing the boundaries of robotic computed tomography: automated twin-robot CT scan with maximum reachability.

    Scientific reports·2026
    Same author

    Synthesizing vocal tract magnetic resonance imaging sequences with phoneme-aware diffusion models.

    Journal of medical imaging (Bellingham, Wash.)·2026
    Same author

    Illuminating the black box of reservoir computing.

    Scientific reports·2026

    Area of Science:

    • Machine Learning
    • Kernel Methods
    • Classification Algorithms

    Background:

    • Kernel methods are crucial for nonlinear classification.
    • Existing kernels may lack adaptiveness to localized nonlinearities.
    • Need for flexible kernels that handle varying nonlinear patterns.

    Purpose of the Study:

    • To introduce and evaluate a novel truncated distance (TL1) kernel.
    • To develop a classifier with a hybrid global nonlinear and local linear structure.
    • To assess the TL1 kernel's performance and applicability in classification.

    Main Methods:

    • Proposed a truncated distance (TL1) kernel function.
    • Developed a classifier leveraging the TL1 kernel for simultaneous subregion training.
    • Applied classical kernel learning methods with TL1 kernel evaluation replacement.

    Related Experiment Videos

  • Conducted numerical experiments comparing TL1 with radial basis function kernel.

    • Main Results:

    • The TL1 kernel enables classifiers nonlinear globally and linear within subregions.
    • Achieved effective training using all data for subregion structure and local classifiers.
    • Demonstrated good adaptiveness to nonlinearity and suitability for diverse nonlinear problems.
    • TL1 kernel achieved performance similar to or better than tuned RBF kernel in experiments.

    Conclusions:

    • The TL1 kernel is a promising nonlinear kernel for classification.
    • Its hybrid structure offers advantages in adaptiveness and training efficiency.
    • Applicable in standard toolboxes despite not being positive semidefinite.
    • Suitable for complex problems requiring localized nonlinear modeling.