Retrieval Property of Associative Memory Based on Inverse Function Delayed Neural Networks

Hongge LI; Yoshihiro HAYAKAWA; Koji NAKAJIMA

doi:10.1093/ietfec/e88-a.8.2192

Retrieval Property of Associative Memory Based on Inverse Function Delayed Neural Networks

Hongge LI, Yoshihiro HAYAKAWA, Koji NAKAJIMA

Full Text Views

0

Share
Cite this

Summary :

Self-connection can enlarge the memory capacity of an associative memory based on the neural network. However, the basin size of the embedded memory state shrinks. The problem of basin size is related to undesirable stable states which are spurious. If we can destabilize these spurious states, we expect to improve the basin size. The inverse function delayed (ID) model, which includes the Bonhoeffer-van der Pol (BVP) model, has negative resistance in its dynamics. The negative resistance of the ID model can destabilize the equilibrium states on certain regions of the conventional neural network. Therefore, the associative memory based on the ID model, which has self-connection in order to enlarge the memory capacity, has the possibility to improve the basin size of the network. In this paper, we examine the fundamental characteristics of an associative memory based on the ID model by numerical simulation and show the improvement of performance compared with the conventional neural network.

Publication: IEICE TRANSACTIONS on Fundamentals Vol.E88-A No.8 pp.2192-2199

Publication Date: 2005/08/01

Publicized

Online ISSN

DOI: 10.1093/ietfec/e88-a.8.2192

Type of Manuscript: PAPER

Category: Nonlinear Problems

Cite this

Copy

Hongge LI, Yoshihiro HAYAKAWA, Koji NAKAJIMA, "Retrieval Property of Associative Memory Based on Inverse Function Delayed Neural Networks" in IEICE TRANSACTIONS on Fundamentals, vol. E88-A, no. 8, pp. 2192-2199, August 2005, doi: 10.1093/ietfec/e88-a.8.2192.
Abstract: Self-connection can enlarge the memory capacity of an associative memory based on the neural network. However, the basin size of the embedded memory state shrinks. The problem of basin size is related to undesirable stable states which are spurious. If we can destabilize these spurious states, we expect to improve the basin size. The inverse function delayed (ID) model, which includes the Bonhoeffer-van der Pol (BVP) model, has negative resistance in its dynamics. The negative resistance of the ID model can destabilize the equilibrium states on certain regions of the conventional neural network. Therefore, the associative memory based on the ID model, which has self-connection in order to enlarge the memory capacity, has the possibility to improve the basin size of the network. In this paper, we examine the fundamental characteristics of an associative memory based on the ID model by numerical simulation and show the improvement of performance compared with the conventional neural network.
URL: https://globals.ieice.org/en_transactions/fundamentals/10.1093/ietfec/e88-a.8.2192/_p

Copy

@ARTICLE{e88-a_8_2192,
author={Hongge LI, Yoshihiro HAYAKAWA, Koji NAKAJIMA, },
journal={IEICE TRANSACTIONS on Fundamentals},
title={Retrieval Property of Associative Memory Based on Inverse Function Delayed Neural Networks},
year={2005},
volume={E88-A},
number={8},
pages={2192-2199},
abstract={Self-connection can enlarge the memory capacity of an associative memory based on the neural network. However, the basin size of the embedded memory state shrinks. The problem of basin size is related to undesirable stable states which are spurious. If we can destabilize these spurious states, we expect to improve the basin size. The inverse function delayed (ID) model, which includes the Bonhoeffer-van der Pol (BVP) model, has negative resistance in its dynamics. The negative resistance of the ID model can destabilize the equilibrium states on certain regions of the conventional neural network. Therefore, the associative memory based on the ID model, which has self-connection in order to enlarge the memory capacity, has the possibility to improve the basin size of the network. In this paper, we examine the fundamental characteristics of an associative memory based on the ID model by numerical simulation and show the improvement of performance compared with the conventional neural network.},
keywords={},
doi={10.1093/ietfec/e88-a.8.2192},
ISSN={},
month={August},}

Copy

TY - JOUR
TI - Retrieval Property of Associative Memory Based on Inverse Function Delayed Neural Networks
T2 - IEICE TRANSACTIONS on Fundamentals
SP - 2192
EP - 2199
AU - Hongge LI
AU - Yoshihiro HAYAKAWA
AU - Koji NAKAJIMA
PY - 2005
DO - 10.1093/ietfec/e88-a.8.2192
JO - IEICE TRANSACTIONS on Fundamentals
SN -
VL - E88-A
IS - 8
JA - IEICE TRANSACTIONS on Fundamentals
Y1 - August 2005
AB - Self-connection can enlarge the memory capacity of an associative memory based on the neural network. However, the basin size of the embedded memory state shrinks. The problem of basin size is related to undesirable stable states which are spurious. If we can destabilize these spurious states, we expect to improve the basin size. The inverse function delayed (ID) model, which includes the Bonhoeffer-van der Pol (BVP) model, has negative resistance in its dynamics. The negative resistance of the ID model can destabilize the equilibrium states on certain regions of the conventional neural network. Therefore, the associative memory based on the ID model, which has self-connection in order to enlarge the memory capacity, has the possibility to improve the basin size of the network. In this paper, we examine the fundamental characteristics of an associative memory based on the ID model by numerical simulation and show the improvement of performance compared with the conventional neural network.
ER -