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

Incremental training of support vector machines.

Alistair Shilton1, M Palaniswami, Daniel Ralph

  • 1Center of Expertise on Networked Decision and Sensor Systems, Department of Electrical and Electronic Engineering, The University of Melbourne, Victoria 3010, Australia. apsh@ee.mu.oz.au

IEEE Transactions on Neural Networks
|March 1, 2005
PubMed
Summary

This study introduces a novel incremental training algorithm for Support Vector Machines (SVMs). This method efficiently updates trained SVMs with new data or parameter changes, outperforming traditional batch retraining.

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

Point cloud dosimetry framework for preclinical microbeam radiation therapy.

Physica medica : PM : an international journal devoted to the applications of physics to medicine and biology : official journal of the Italian Association of Biomedical Physics (AIFB)·2025
Same author

Comparison of Deep Learning Models for fast and accurate dose map prediction in Microbeam Radiation Therapy.

Physica medica : PM : an international journal devoted to the applications of physics to medicine and biology : official journal of the Italian Association of Biomedical Physics (AIFB)·2025
Same author

Dual-Branch Multi-Dimensional Attention Mechanism for Joint Facial Expression Detection and Classification.

Sensors (Basel, Switzerland)·2025
Same author

An Innovative Neighbor Attention Mechanism Based on Coordinates for the Recognition of Facial Expressions.

Sensors (Basel, Switzerland)·2024
Same author

Facial Expression Recognition-You Only Look Once-Neighborhood Coordinate Attention Mamba: Facial Expression Detection and Classification Based on Neighbor and Coordinates Attention Mechanism.

Sensors (Basel, Switzerland)·2024
Same author

Accurate and Fast Deep Learning Dose Prediction for a Preclinical Microbeam Radiation Therapy Study Using Low-Statistics Monte Carlo Simulations.

Cancers·2023

Area of Science:

  • Machine Learning
  • Optimization Algorithms
  • Data Science

Background:

  • Support Vector Machines (SVMs) are powerful classification models.
  • Traditional SVM training can be computationally intensive, especially with large or evolving datasets.
  • Retraining SVMs from scratch for new data or parameter adjustments is inefficient.

Purpose of the Study:

  • To develop an efficient algorithm for incremental training of Support Vector Machines (SVMs).
  • To address challenges posed by sequentially arriving data and rapid changes in constraint parameters.
  • To demonstrate the computational advantages of the proposed incremental training method over batch retraining.

Main Methods:

  • A novel "warm-start" algorithm is proposed for SVM training.

Related Experiment Videos

  • The method leverages the incremental properties of active set approaches for linearly constrained optimization.
  • The algorithm facilitates quick retraining of SVMs with added data or modified parameters.

    • Main Results:

    • The incremental training algorithm significantly reduces retraining time compared to batch methods.
    • The approach is effective for both adding new training vectors and varying constraint parameters.
    • Demonstrated computational superiority in scenarios of sequential data arrival and parameter variation.

    Conclusions:

    • The proposed incremental training algorithm offers a computationally superior alternative to batch retraining for SVMs.
    • This method is particularly beneficial for dynamic machine learning applications with evolving data and parameters.
    • The "warm-start" approach enhances the efficiency of SVM updates in practical, real-world scenarios.