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 learning of feature space and classifier for face recognition.

Seiichi Ozawa1, Soon Lee Toh, Shigeo Abe

  • 1Graduate School of Science and Technology, Kobe University, Kobe 657-8501, Japan. ozawasei@kobe-u.ac.jp

Neural Networks : the Official Journal of the International Neural Network Society
|August 17, 2005
PubMed
Summary

This study introduces an incremental learning model combining Incremental Principal Component Analysis (IPCA) and Resource Allocating Network with Long-Term Memory (RAN-LTM) for enhanced pattern recognition. The approach effectively improves face recognition accuracy by adapting feature spaces and neural classifiers simultaneously.

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

Comparative performance analysis of quantum feature maps for quantum kernel-based machine learning.

Scientific reports·2026
Same author

NEuroMOrphic Neural-Response Decoding System for Adaptive and Personalized Neuro-Prosthetics' Control.

Biomimetics (Basel, Switzerland)·2025
Same author

Decoding Brain Signals in a Neuromorphic Framework for a Personalized Adaptive Control of Human Prosthetics.

Biomimetics (Basel, Switzerland)·2025
Same author

Assessment of first-touch skills in robotic surgical training using hi-Sim and the hinotori surgical robot system among surgeons and novices.

Langenbeck's archives of surgery·2024
Same author

A generalisability theory approach to quantifying changes in psychopathology among ultra-high-risk individuals for psychosis.

Schizophrenia (Heidelberg, Germany)·2024
Same author

Filter and Wrapper Stacking Ensemble (FWSE): a robust approach for reliable biomarker discovery in high-dimensional omics data.

Briefings in bioinformatics·2023

Area of Science:

  • Computer Science
  • Artificial Intelligence
  • Machine Learning

Background:

  • Traditional pattern recognition systems often require retraining on the entire dataset when new data is available.
  • Adapting feature spaces alongside classifiers presents a significant challenge in incremental learning scenarios.

Purpose of the Study:

  • To propose a novel incremental learning framework that simultaneously learns both the feature space and the classifier.
  • To enhance the generalization performance of neural classifiers in dynamic pattern recognition tasks, specifically face recognition.

Main Methods:

  • An extended version of Incremental Principal Component Analysis (IPCA) was combined with a Resource Allocating Network with Long-Term Memory (RAN-LTM).
  • An approximation formula was derived to update memory items in RAN-LTM, enabling adaptation to the evolving feature space from IPCA.

Related Experiment Videos

  • The integrated model was applied to a face recognition system.

    • Main Results:

    • The proposed model demonstrated effective incremental learning of the feature space.
    • The adaptation mechanism allowed the RAN-LTM classifier to adjust to changes in feature space dimensionality and values.
    • Experimental evaluation on a face recognition task showed significant enhancement in generalization performance.

    Conclusions:

    • Simultaneous incremental learning of feature spaces and classifiers is a viable and effective approach for pattern recognition.
    • The developed IPCA and RAN-LTM combination offers a robust solution for adaptive and evolving recognition systems.
    • The method shows promise for real-world applications like face recognition where data distributions can change over time.