Jove
Visualize
Contact Us

Related Experiment Videos

An information-maximization approach to blind separation and blind deconvolution

A J Bell1, T J Sejnowski

  • 1Howard Hughes Medical Institute, Computational Neurobiology Laboratory, Salk Institute, La Jolla, CA 92037, USA.

Neural Computation
|November 1, 1995
PubMed
Summary

This study introduces a novel self-organizing algorithm that enhances information transfer in nonlinear networks. It successfully separates independent sources and performs blind deconvolution, offering a unified approach to blind signal processing.

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

Presynaptic Vesicles Supply Membrane for Axonal Bouton Enlargement during Long Term Potentiation.

The Journal of neuroscience : the official journal of the Society for Neuroscience·2026
Same author

Presynaptic vesicles supply membrane for axonal bouton enlargement during LTP.

bioRxiv : the preprint server for biology·2025
Same author

Stochastic self-tuning hybrid algorithm for reaction-diffusion systems.

The Journal of chemical physics·2020
Same author

Incidence, causes and correlates of maternal near-miss morbidity: a multi-centre cross-sectional study.

BJOG : an international journal of obstetrics and gynaecology·2018
Same author

Expected background rates of latent TB infection in London inner city schools: lessons from a TB contact investigation exercise in a secondary school.

Epidemiology and infection·2018
Same author

Search for direct pair production of supersymmetric top quarks decaying to all-hadronic final states in pp collisions at [Formula: see text].

The European physical journal. C, Particles and fields·2017
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

Area of Science:

  • Computational Neuroscience
  • Machine Learning
  • Signal Processing

Background:

  • Traditional linear models have limitations in capturing complex input distributions.
  • Understanding information processing in nonlinear neural networks is crucial for advanced AI.
  • Blind signal processing tasks like source separation and deconvolution are challenging.

Purpose of the Study:

  • To develop a new self-organizing learning algorithm for nonlinear networks.
  • To maximize information transfer without prior knowledge of input distributions.
  • To generalize principal component analysis for higher-order statistical components.

Main Methods:

  • Derivation of a novel self-organizing learning algorithm.
  • Application to the source separation problem (cocktail party problem).

Related Experiment Videos

  • Development of a network variant for blind deconvolution.

    • Main Results:

    • The algorithm successfully separates unknown mixtures of up to 10 speakers.
    • Demonstrated blind deconvolution capabilities for speech signals.
    • Identified dependencies of information transfer on time delays.

    Conclusions:

    • Information maximization provides a unifying framework for blind signal processing.
    • Nonlinearities enable the capture of higher-order statistics for redundancy reduction.
    • The algorithm offers a powerful tool for complex signal processing tasks.