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

Locally linear discriminant analysis for multimodally distributed classes for face recognition with a single model

Tae-Kyun Kim, J Kittler

    IEEE Transactions on Pattern Analysis and Machine Intelligence
    |March 8, 2005
    PubMed
    Summary

    We introduce Locally Linear Discriminant Analysis (LLDA), a novel nonlinear method for efficient and robust face recognition. LLDA overcomes limitations of linear methods by capturing nonlinear data structures and avoids overfitting issues common in kernel-based approaches.

    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

    Temporal recovery in motor evoked potential during evacuation of cervical spontaneous epidural hematoma: illustrative case.

    Journal of neurosurgery. Case lessons·2026
    Same author

    Superimposed intraoperative neuromonitoring for epidermoid cyst: Two-dimensional operative video.

    Surgical neurology international·2026
    Same author

    Radiation-Induced Meningiomas Have an Aggressive Clinical Course: Genetic Signature Is Limited to <i>NF2</i> Alterations, and Epigenetic Signature Is H3K27me3 Loss.

    Journal of Korean medical science·2025
    Same author

    BP-SGCN: Behavioral Pseudo-Label Informed Sparse Graph Convolution Network for Pedestrian and Heterogeneous Trajectory Prediction.

    IEEE transactions on neural networks and learning systems·2025
    Same author

    Clinical and Genetic Overview of Neurofibromatosis Type 2 (NF2).

    Journal of Korean Neurosurgical Society·2025
    Same author

    Dream360: Diverse and Immersive Outdoor Virtual Scene Creation via Transformer-Based 360° Image Outpainting.

    IEEE transactions on visualization and computer graphics·2024

    Area of Science:

    • Machine Learning
    • Computer Vision
    • Pattern Recognition

    Background:

    • Linear Discriminant Analysis (LDA) is efficient for face recognition but fails to capture nonlinear facial manifolds.
    • Kernel-based methods like GDA and SVM handle nonlinearities but incur high computational costs and overfitting.
    • Pose variations in faces present a significant challenge for traditional linear classification methods.

    Purpose of the Study:

    • To develop a novel, computationally efficient nonlinear discriminant analysis method.
    • To address the limitations of existing linear and nonlinear classification techniques in face recognition.
    • To enable robust face recognition under pose variations using a single model image.

    Main Methods:

    • Locally Linear Discriminant Analysis (LLDA) utilizes locally linear transformations to align nonlinear data structures.

    Related Experiment Videos

  • Input vectors are projected into local feature spaces via linear transformations maximizing between-class covariance and minimizing within-class covariance.
  • A novel gradient-based learning algorithm optimizes local linear bases without local-maxima issues.

    • Main Results:

    • LLDA demonstrates computational efficiency compared to Generalized Discriminant Analysis (GDA).
    • The method exhibits robustness against overfitting due to its linear base structure.
    • Effective face recognition in a low-dimensional subspace under pose variations was achieved using synthetic and real face data.

    Conclusions:

    • LLDA offers a computationally efficient and robust solution for multiclass nonlinear discrimination.
    • The proposed method effectively handles nonlinear data manifolds, particularly in face recognition tasks with pose variations.
    • LLDA provides a promising alternative to existing methods, achieving high classification accuracy with reduced computational complexity.