Opinion Dynamics with Hopfield Neural Networks
Physics and Society
2008-03-17 v2
Abstract
In Hopfield neural networks with up to 10^8 nodes we store two patterns through Hebb couplings. Then we start with a third random pattern which is supposed to evolve into one of the two stored patterns, simulating the cognitive process of associative memory leading to one of two possible opinions. With probability p each neuron independently, instead of following the Hopfield rule, takes over the corresponding value of another network, thus simulating how different people can convince each other. A consensus is achieved for high p.
Cite
@article{arxiv.0712.4364,
title = {Opinion Dynamics with Hopfield Neural Networks},
author = {Dietrich Stauffer and Przemyslaw A. Grabowicz and Janusz A. Holyst},
journal= {arXiv preprint arXiv:0712.4364},
year = {2008}
}
Comments
Enlarged from 7 to 9 pages with new authors, new simulations, and a theory