The original paper is in English. Non-English content has been machine-translated and may contain typographical errors or mistranslations. ex. Some numerals are expressed as "XNUMX".
Copyrights notice
The original paper is in English. Non-English content has been machine-translated and may contain typographical errors or mistranslations. Copyrights notice
Dans cet article, un modèle de mémoire associative avec processus d'oubli proposé par Mezard et al. est étudié comme moyen de stocker des modèles peu codés par le SCSNA proposé par Shiino et Fukai. Semblable au cas du stockage de modèles non clairsemés (non biaisés) analysé par Mezard et al., ce modèle de mémoire associative peu codé est également exempt d'une détérioration catastrophique de la mémoire causée par une surcharge des modèles de mémoire. Nous obtenons théoriquement une relation entre la capacité de stockage et le taux d’oubli, et constatons qu’il existe un taux d’oubli optimal conduisant à la capacité de stockage maximale. Nous appelons cela le taux de capacité de stockage optimal. À mesure que la cadence de déclenchement du motif de mémoire diminue, la capacité de stockage optimale augmente et le taux d'oubli optimal diminue. De plus, nous avons montré que le taux de capacité (c'est-à-dire le rapport entre la capacité de stockage pour la règle d'apprentissage par corrélation conventionnelle et la capacité de stockage optimale) est presque constant par rapport à la cadence de déclenchement du motif mémoire.
The copyright of the original papers published on this site belongs to IEICE. Unauthorized use of the original or translated papers is prohibited. See IEICE Provisions on Copyright for details.
Copier
Tomoyuki KIMOTO, Masato OKADA, "Sparsely Encoded Associative Memory Model with Forgetting Process" in IEICE TRANSACTIONS on Information,
vol. E85-D, no. 12, pp. 1938-1945, December 2002, doi: .
Abstract: In this paper, an associative memory model with a forgetting process proposed by Mezard et al. is investigated as a means of storing sparsely encoded patterns by the SCSNA proposed by Shiino and Fukai. Similar to the case of storing non-sparse (non-biased) patterns as analyzed by Mezard et al., this sparsely encoded associative memory model is also free from a catastrophic deterioration of the memory caused by memory pattern overloading. We theoretically obtain a relationship between the storage capacity and the forgetting rate, and find that there is an optimal forgetting rate leading to the maximum storage capacity. We call this the optimal storage capacity rate. As the memory pattern firing rate decreases, the optimal storage capacity increases and the optimal forgetting rate decreases. Furthermore, we shown that the capacity rate (i.e. the ratio of the storage capacity for the conventional correlation learning rule to the optimal storage capacity) is almost constant with respect to the memory pattern firing rate.
URL: https://global.ieice.org/en_transactions/information/10.1587/e85-d_12_1938/_p
Copier
@ARTICLE{e85-d_12_1938,
author={Tomoyuki KIMOTO, Masato OKADA, },
journal={IEICE TRANSACTIONS on Information},
title={Sparsely Encoded Associative Memory Model with Forgetting Process},
year={2002},
volume={E85-D},
number={12},
pages={1938-1945},
abstract={In this paper, an associative memory model with a forgetting process proposed by Mezard et al. is investigated as a means of storing sparsely encoded patterns by the SCSNA proposed by Shiino and Fukai. Similar to the case of storing non-sparse (non-biased) patterns as analyzed by Mezard et al., this sparsely encoded associative memory model is also free from a catastrophic deterioration of the memory caused by memory pattern overloading. We theoretically obtain a relationship between the storage capacity and the forgetting rate, and find that there is an optimal forgetting rate leading to the maximum storage capacity. We call this the optimal storage capacity rate. As the memory pattern firing rate decreases, the optimal storage capacity increases and the optimal forgetting rate decreases. Furthermore, we shown that the capacity rate (i.e. the ratio of the storage capacity for the conventional correlation learning rule to the optimal storage capacity) is almost constant with respect to the memory pattern firing rate.},
keywords={},
doi={},
ISSN={},
month={December},}
Copier
TY - JOUR
TI - Sparsely Encoded Associative Memory Model with Forgetting Process
T2 - IEICE TRANSACTIONS on Information
SP - 1938
EP - 1945
AU - Tomoyuki KIMOTO
AU - Masato OKADA
PY - 2002
DO -
JO - IEICE TRANSACTIONS on Information
SN -
VL - E85-D
IS - 12
JA - IEICE TRANSACTIONS on Information
Y1 - December 2002
AB - In this paper, an associative memory model with a forgetting process proposed by Mezard et al. is investigated as a means of storing sparsely encoded patterns by the SCSNA proposed by Shiino and Fukai. Similar to the case of storing non-sparse (non-biased) patterns as analyzed by Mezard et al., this sparsely encoded associative memory model is also free from a catastrophic deterioration of the memory caused by memory pattern overloading. We theoretically obtain a relationship between the storage capacity and the forgetting rate, and find that there is an optimal forgetting rate leading to the maximum storage capacity. We call this the optimal storage capacity rate. As the memory pattern firing rate decreases, the optimal storage capacity increases and the optimal forgetting rate decreases. Furthermore, we shown that the capacity rate (i.e. the ratio of the storage capacity for the conventional correlation learning rule to the optimal storage capacity) is almost constant with respect to the memory pattern firing rate.
ER -