Learning and forgetting on asymmetric, diluted neural networks
Abstract
It is possible to construct diluted asymmetric models of neural networks for which the dynamics can be calculated exactly. The authors test several learning schemes, in particular, models for which the values of the synapses remain bounded and depend on the history. Our analytical results on the relative efficiencies of the various learning schemes are qualitatively similar to the corresponding ones obtained numerically on fully connected symmetric networks.
- Authors:
- Publication Date:
- Research Org.:
- CEN-Saclay, Gif-sur-Yvette (France)
- OSTI Identifier:
- 5371563
- Resource Type:
- Journal Article
- Journal Name:
- J. Stat. Phys.; (United States)
- Additional Journal Information:
- Journal Volume: 49:5/6
- Country of Publication:
- United States
- Language:
- English
- Subject:
- 71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSICS; 99 GENERAL AND MISCELLANEOUS//MATHEMATICS, COMPUTING, AND INFORMATION SCIENCE; ISING MODEL; ARTIFICIAL INTELLIGENCE; STATISTICAL MECHANICS; ASYMMETRY; COMPUTER ARCHITECTURE; COMPUTERIZED SIMULATION; MEMORY DEVICES; PHASE DIAGRAMS; RANDOMNESS; SPIN; ANGULAR MOMENTUM; CRYSTAL MODELS; DIAGRAMS; MATHEMATICAL MODELS; MECHANICS; PARTICLE PROPERTIES; SIMULATION; 657002* - Theoretical & Mathematical Physics- Classical & Quantum Mechanics; 990210 - Supercomputers- (1987-1989)
Citation Formats
Derrida, B, and Nadal, J P. Learning and forgetting on asymmetric, diluted neural networks. United States: N. p., 1987.
Web. doi:10.1007/BF01017556.
Derrida, B, & Nadal, J P. Learning and forgetting on asymmetric, diluted neural networks. United States. https://doi.org/10.1007/BF01017556
Derrida, B, and Nadal, J P. 1987.
"Learning and forgetting on asymmetric, diluted neural networks". United States. https://doi.org/10.1007/BF01017556.
@article{osti_5371563,
title = {Learning and forgetting on asymmetric, diluted neural networks},
author = {Derrida, B and Nadal, J P},
abstractNote = {It is possible to construct diluted asymmetric models of neural networks for which the dynamics can be calculated exactly. The authors test several learning schemes, in particular, models for which the values of the synapses remain bounded and depend on the history. Our analytical results on the relative efficiencies of the various learning schemes are qualitatively similar to the corresponding ones obtained numerically on fully connected symmetric networks.},
doi = {10.1007/BF01017556},
url = {https://www.osti.gov/biblio/5371563},
journal = {J. Stat. Phys.; (United States)},
number = ,
volume = 49:5/6,
place = {United States},
year = {Tue Dec 01 00:00:00 EST 1987},
month = {Tue Dec 01 00:00:00 EST 1987}
}
Other availability
Save to My Library
You must Sign In or Create an Account in order to save documents to your library.