Document Detail


Information theoretical approach to the storage capacity of neural networks with binary weights.
MedLine Citation:
PMID:  11970316     Owner:  NLM     Status:  MEDLINE    
Abstract/OtherAbstract:
The storage capacity of the perceptron with binary weights w(i)in[0,1] is derived by introducing the minimum distance d between input patterns. The approach presented in this paper is based on some results in the information theory, and the obtained storage capacity 0.585 is in good agreement with the well-known value 0.59 by the replica method in statistical physics. A strength of the present information theoretical approach is that it provides an easier and more intuitive understanding for the storage capacity than the replica method, which is believed to be more reliable and informative than the Vapnik-Chervonenkis procedure.
Authors:
H Suyari; I Matsuba
Related Documents :
24109726 - Lumped parameter model of cardiovascular-respiratory interaction.
25064016 - Modeling the respiratory chain complexes with biothermokinetic equations - the case of ...
20670346 - Using theories of behaviour change to inform interventions for addictive behaviours.
22559876 - Quantitatively integrating molecular structure and bioactivity profile evidence into dr...
22439796 - A consumption-based ghg inventory for the u.s. state of oregon.
12232466 - Explaining attitudinal ratings of dutch rising contours: morphological structure vs. th...
Publication Detail:
Type:  Journal Article    
Journal Detail:
Title:  Physical review. E, Statistical physics, plasmas, fluids, and related interdisciplinary topics     Volume:  60     ISSN:  1063-651X     ISO Abbreviation:  Phys Rev E Stat Phys Plasmas Fluids Relat Interdiscip Topics     Publication Date:  1999 Oct 
Date Detail:
Created Date:  2002-04-23     Completed Date:  2002-07-16     Revised Date:  2008-11-21    
Medline Journal Info:
Nlm Unique ID:  9887340     Medline TA:  Phys Rev E Stat Phys Plasmas Fluids Relat Interdiscip Topics     Country:  United States    
Other Details:
Languages:  eng     Pagination:  4576-9     Citation Subset:  IM    
Affiliation:
Department of Information and Image Sciences, Faculty of Engineering, Chiba University 1-33, Yayoi-cho, Inage-ku, Chiba-shi, Chiba 263-8522 Japan.
Export Citation:
APA/MLA Format     Download EndNote     Download BibTex
MeSH Terms
Descriptor/Qualifier:
Biophysical Phenomena
Biophysics*
Models, Statistical
Neural Networks (Computer)*

From MEDLINE®/PubMed®, a database of the U.S. National Library of Medicine


Previous Document:  Noise-induced breakdown of coherent collective motion in swarms.
Next Document:  Categorization in the symmetrically dilute Hopfield network.