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

Local linear independent component analysis based on clustering.

J Karhunen1, S Mălăroiu, M Ilmoniemi

  • 1Helsinki University of Technology, Neural Networks Research Centre, Espoo, Finland. Juha.Karhunen@hut.fi

International Journal of Neural Systems
|April 20, 2001
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

Physically principled reflection models applied to filtered camera imaging inversions in metal walled fusion machines.

The Review of scientific instruments·2019
Same author

Impact of early angiographic evaluation on the frequency of emergency reoperations after coronary bypass surgery.

Scandinavian journal of surgery : SJS : official organ for the Finnish Surgical Society and the Scandinavian Surgical Society·2010
Same author

An experimental comparison of neural algorithms for independent component analysis and blind separation.

International journal of neural systems·1999
Same author

Irregular expression of hyaluronan and its CD44 receptor is associated with metastatic phenotype in laryngeal squamous cell carcinoma.

Virchows Archiv : an international journal of pathology·1999
Same author

Least-squares methods for blind source separation based on nonlinear PCA.

International journal of neural systems·1999
Same author

On neural blind separation with noise suppression and redundancy reduction.

International journal of neural systems·1997

This study introduces local Independent Component Analysis (ICA) models combined with clustering to better represent nonlinear data. This approach improves upon standard linear ICA without the complexity of nonlinear ICA.

Area of Science:

  • Data analysis
  • Machine learning
  • Signal processing

Background:

  • Standard Independent Component Analysis (ICA) relies on linear models, which are insufficient for complex, nonlinear data distributions.
  • Existing nonlinear ICA methods can be computationally intensive and difficult to implement.

Purpose of the Study:

  • To propose a novel approach using local ICA models combined with data clustering.
  • To achieve a more accurate representation of nonlinear data than linear ICA.
  • To avoid the computational challenges associated with full nonlinear ICA.

Main Methods:

  • Data is first grouped using clustering algorithms (e.g., K-means, Self-Organizing Maps, Neural Gas).
  • Linear ICA models are then applied independently to each identified cluster.

Related Experiment Videos

  • This creates a hybrid model combining global nonlinear structure with local linear components.

    • Main Results:

    • The proposed local ICA method offers improved data representation for nonlinear distributions compared to global linear ICA.
    • Experimental results on artificial data and natural images demonstrate the effectiveness of the approach.
    • The method provides a flexible framework bridging global, dense representations and local, sparse coding.

    Conclusions:

    • Local ICA with clustering offers a practical and effective solution for analyzing nonlinear data.
    • This hybrid approach balances representational accuracy with computational feasibility.
    • The framework generalizes various data representation methods, positioning local ICA between global and sparse coding extremes.