SIMULATING HIGHLY DILUTED NEURAL NETWORKS

Authors
ARENZON JJ LEMKE N
Citation
Jj. Arenzon et N. Lemke, SIMULATING HIGHLY DILUTED NEURAL NETWORKS, Journal of physics. A, mathematical and general, 27(15), 1994, pp. 5161-5165
Citations number
8
Categorie Soggetti
Physics
Journal title
Journal of physics. A, mathematical and general → ACNP
ISSN journal
03054470
Volume
27
Issue
15
Year of publication
1994
Pages
5161 - 5165
Database
ISI
SICI code
0305-4470(1994)27:15<5161:SHDNN>2.0.ZU;2-5
Abstract
The asymmetric diluted Hopfield model is studied through numerical sim ulations and reliable results are obtained even when the network sizes are not exponentially large. The mean final overlap, the mean converg ence time and a quantity that gives information on the relative import ance of spurious states are measured as a function of the load capacit y alpha.