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

Concerning the differentiability of the energy function in vector quantization algorithms.

Dominique Lepetz1, Max Némoz-Gaillard, Michaël Aupetit

  • 1EMA-DM, 6 Av. de Clavières, 30319 Alès cedex, France.

Neural Networks : the Official Journal of the International Neural Network Society
|April 10, 2007
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

ISilDR: Isometric Seriation-Based Dimensionality Reduction for Visual Cluster Analysis.

IEEE transactions on visualization and computer graphics·2026
Same author

Efficient and interpretable DNA/RNA representation using Komlós-Hadamard transforms.

BMC bioinformatics·2026
Same author

Distortion-Aware Brushing for Reliable Cluster Analysis in Multidimensional Projections.

IEEE transactions on visualization and computer graphics·2025
Same author

Measuring the Validity of Clustering Validation Datasets.

IEEE transactions on pattern analysis and machine intelligence·2025
Same author

Exploring the Applications of Explainability in Wearable Data Analytics: Systematic Literature Review.

Journal of medical Internet research·2024
Same author

Classes are not Clusters: Improving Label-based Evaluation of Dimensionality Reduction.

IEEE transactions on visualization and computer graphics·2023
Same journal

Exploiting audio-visual modalities in videos: Object detection via multi-stage bilateral coupling network.

Neural networks : the official journal of the International Neural Network Society·2026
Same journal

Reliability-aware modality completion with cross-modal distillation for federated learning with missing modalities.

Neural networks : the official journal of the International Neural Network Society·2026
Same journal

IGFD-Net: Illumination-guided frequency decoupling for polarization image fusion.

Neural networks : the official journal of the International Neural Network Society·2026
Same journal

Multiple-Strategies dung beetle optimizer and its applications in engineering optimization and bankruptcy prediction.

Neural networks : the official journal of the International Neural Network Society·2026
Same journal

Aggregating global-scale pixel-wise forgery cues within a graph.

Neural networks : the official journal of the International Neural Network Society·2026
Same journal

Finite-Time intermittent control for secure synchronization of Neutral-Type stochastic delayed neural networks under aperiodic DoS attacks.

Neural networks : the official journal of the International Neural Network Society·2026
See all related articles

Vector Quantization algorithms, like K-means, rely on energy functions for convergence. This study introduces pseudo-potentials, extending convergence analysis to a broader class of algorithms, including those from Artificial Neural Networks.

Area of Science:

  • * Mathematics
  • * Computer Science
  • * Machine Learning

Background:

  • * The convergence of Vector Quantization (VQ) algorithms is intrinsically linked to the properties of an energy function defined on a topological manifold.
  • * Existing studies on VQ convergence often assume the existence of a potential function, which may not hold for all algorithms.

Purpose of the Study:

  • * To investigate the conditions for the existence of energy functions for a class of VQ algorithms, including K-means and Self-Organizing Maps.
  • * To introduce and define the concept of a pseudo-potential as a generalization of the energy function.
  • * To establish a unified framework for analyzing the convergence of a wide range of VQ algorithms.

Main Methods:

  • * Theoretical analysis of adaptation rules in Vector Quantization algorithms.

Related Experiment Videos

  • * Exploration of topological manifold properties related to energy functions.
  • * Development of the pseudo-potential concept as a uniform limit of potential functions.

    • Main Results:

    • * Demonstrated that the energy function for certain VQ algorithms is not always a potential, but a pseudo-potential.
    • * Identified that many algorithms from the Artificial Neural Networks community belong to this class.
    • * Established a theoretical framework enabling simultaneous convergence analysis for numerous VQ adaptation rules.

    Conclusions:

    • * The concept of pseudo-potential provides a more general condition for VQ algorithm convergence.
    • * The defined framework facilitates unified convergence studies for a broad spectrum of VQ algorithms.
    • * A new theorem offers significant insights into the convergence properties of these algorithms.