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

Linear function neurons: structure and training.

S E Hampson, D J Volper

    Biological Cybernetics
    |January 1, 1986
    PubMed
    Summary
    This summary is machine-generated.

    This study develops equivalent representations for thresholded linear equations, revealing similarities between perceptron, Rescorla-Wagner, and Aplysia neural learning. A new conditional probability method accelerates learning, reducing growth rates and eliminating irrelevant feature effects.

    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

    Personality and overweight in 6-12-year-old children.

    Pediatric obesity·2015
    Same author

    Behavioral research on diabetes at the Oregon Research Institute.

    Annals of behavioral medicine : a publication of the Society of Behavioral Medicine·2013
    Same author

    Mothers' negative affectivity during pregnancy and food choices for their infants.

    International journal of obesity (2005)·2009
    Same author

    Statistical detection of chromosomal homology using shared-gene density alone.

    Bioinformatics (Oxford, England)·2004
    Same author

    Comparison of patients' and healthcare professionals' beliefs about and attitudes towards Type 2 diabetes.

    Diabetic medicine : a journal of the British Diabetic Association·2003
    Same author

    Neuropsychological functioning in older people with type 2 diabetes: the effect of controlling for confounding factors.

    Diabetic medicine : a journal of the British Diabetic Association·2002
    Same journal

    Harmonic memory in phasor neural networks.

    Biological cybernetics·2026
    Same journal

    Correction: Decreased spinal inhibition leads to undiversified locomotor patterns.

    Biological cybernetics·2026
    Same journal

    Foundational issues of network models in biology.

    Biological cybernetics·2026
    Same journal

    Dynamical mechanisms for coordinating long-term working memory based on the precision of spike-timing in cortical neurons.

    Biological cybernetics·2026
    Same journal

    Distinct dopaminergic spike-timing-dependent plasticity rules are suited to different functional roles.

    Biological cybernetics·2026
    Same journal

    Fluctuation-response relations for a two-stage population of spiking neurons stimulated by common noise.

    Biological cybernetics·2026
    See all related articles

    Area of Science:

    • Computational Neuroscience
    • Machine Learning
    • Cognitive Science

    Background:

    • Linear equations are fundamental in various scientific domains.
    • Understanding their training characteristics and learning mechanisms is crucial.
    • Existing models like perceptron training offer insights but have limitations.

    Purpose of the Study:

    • To develop and compare different representations for thresholded linear equations.
    • To analyze the training characteristics and learning dynamics of these representations.
    • To explore the connections between mathematical models, animal learning, and neural mechanisms.

    Main Methods:

    • Development of three representationally equivalent forms for thresholded linear equations.
    • Analysis of perceptron training algorithm, including its mathematical properties and bounds.

    Related Experiment Videos

  • Comparison with Rescorla-Wagner learning model and Aplysia gill withdrawal neural mechanism.
  • Introduction of a conditional probability method to accelerate learning.

    • Main Results:

    • Equivalence of representations for binary inputs, with differing training characteristics.
    • Perceptron training lower bound of 2d, upper bound potentially 4d, average 1.4d for d features.
    • Learning time increases linearly with irrelevant/replicated features; (X of N) functions learnable in d3 time.
    • Conditional probability method reduces growth rate to 2d or 1.7d, and eliminates irrelevant feature effects.

    Conclusions:

    • Thresholded linear equations can be represented equivalently, impacting training dynamics.
    • Strong parallels exist between computational, animal, and neural learning mechanisms.
    • Proposed methods significantly accelerate learning and improve robustness to irrelevant data.