Noise distributions in retrieval dynamics of the Hopfield model
- 7 November 1994
- journal article
- Published by IOP Publishing in Journal of Physics A: General Physics
- Vol. 27 (21) , 7061-7068
- https://doi.org/10.1088/0305-4470/27/21/020
Abstract
We examine the dynamical theory, recently proposed by Coolen and Sherrington, which describes the deterministic flow of order parameters in the memory retrieval processes of the Hopfield model. We have performed Monte Carlo simulations to investigate the limit of applicability of their theory. We have found that the qualitative behaviour of retrieval processes follows their prediction fairly well. However, the theory fails to describe the noise distribution quantitatively when the effects of non-retrieved patterns are not negligible.Keywords
This publication has 7 references indexed in Scilit:
- Order-parameter flow in the fully connected Hopfield model near saturationPhysical Review E, 1994
- Dynamics of fully connected attractor neural networks near saturationPhysical Review Letters, 1993
- Retrieval dynamics of associative memory of the Hopfield typeJournal of Physics A: General Physics, 1993
- A probabilistic approach to parallel dynamics for the Little-Hopfield modelJournal of Physics A: General Physics, 1991
- Glauber dynamics of the Little-Hopfield modelZeitschrift für Physik B Condensed Matter, 1988
- Statistical neurodynamics of associative memoryNeural Networks, 1988
- Zero temperature parallel dynamics for infinite range spin glasses and neural networksJournal de Physique, 1987