Search research articles

ABOUT JoVE

Overview Leadership Blog JoVE Help Center

AUTHORS

Publishing Process Editorial Board Scope & Policies Peer Review FAQ Submit

LIBRARIANS

Testimonials Subscriptions Access Resources Library Advisory Board FAQ

RESEARCH

JoVE Journal Methods Collections JoVE Encyclopedia of Experiments Archive

EDUCATION

JoVE Core JoVE Business JoVE Science Education JoVE Lab Manual Faculty Resource Center Faculty Site

Terms & Conditions of Use

Search research articles

Related Experiment Videos

Anti-Hebbian learning in a non-linear neural network.

¹Max-Planck-Institut für Plasmaphysik, Garching, Federal Republic of Germany.

Biological Cybernetics

|January 1, 1990

Summary

This study introduces unsupervised learning rules for neural networks, enabling them to classify input patterns. The hierarchical network learns principal components using non-linear response functions and binary codes.

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

Reliability of the Diagnosis of Cerebral Vasospasm Using Catheter Cerebral Angiography: A Systematic Review and Inter- and Intraobserver Study.

AJNR. American journal of neuroradiology·2021

Same author

A NWB-based dataset and processing pipeline of human single-neuron activity during a declarative memory task.

Scientific data·2020

Same author

Safety and efficacy of thrombopoeitin mimetics for refractory immune thrombocytopenia purpura in patients with systemic lupus erythematosus or antiphospholipid syndrome: a case series.

Lupus·2018

Same author

Dermatomyositis panniculitis: a clinicopathological and immunohistochemical study of 18 cases.

Journal of the European Academy of Dermatology and Venereology : JEADV·2018

Same author

Anticancer effects of resveratrol in canine hemangiosarcoma cell lines.

Veterinary and comparative oncology·2017

Same author

A Method to Investigate Seasonal Variation in Hospitalization for Copd in A Midwestern US State.

Value in health : the journal of the International Society for Pharmacoeconomics and Outcomes Research·2016

Area of Science:

Computational Neuroscience
Machine Learning

Background:

The Hebbian rule enables neurons to learn principal components.
Previous models using anti-Hebbian rules extended this to multiple principal components with linear neurons.

Purpose of the Study:

To extend unsupervised learning of principal components to neurons with non-linear response functions.
To develop local, unsupervised learning rules for neural network parameters.

Main Methods:

Application of an anti-Hebbian rule model to neurons with non-linear response functions (threshold and transition width).
Development and illustration of local, unsupervised learning rules for threshold and transition width.
Simulation of a hierarchical network for pattern classification.

Main Results:

The proposed rules enable neurons to learn to respond to principal components even with non-linear activation.
The network successfully sorts input patterns into classes.
Hierarchical structure allows for binary coding of input pattern features.

Conclusions:

Local, unsupervised learning rules can effectively train neural networks with non-linearities for principal component analysis and pattern classification.
The hierarchical network architecture facilitates efficient coding of complex data structures.

Related Experiment Videos