Presentation 2009-01-19
Statistical mechanics of the Hopfield model with replacing units
Yasunao KOMATSU, Toru AONISHI, Koji KURATA,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) There is a critical memory capacity in the Hopfield model. If the number of embed patterns surpasses the crirical number, the network cannot retrieve all patterns due to the overloading. On the other hands, It has been reported that the network, in which a number of units are replaced by newborn ones when the networks learn, can keep up memory patterns studied recently without the overloading. We analyze the network with replacing units by utilizing statistical mechanics. We show that the critical number of replacing units is R_c=3.2 avoiding the overloading and independent of the network size. In addition, we demonstrate that there is the optimal number of replacing units to maximize the memory capacity. Finally, we compare a model with forgetting process with one with replacing process.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) associative memory / rebirth neuron / overloading / statistical mechanics
Paper # NC2008-90
Date of Issue

Conference Information
Committee NC
Conference Date 2009/1/12(1days)
Place (in Japanese) (See Japanese page)
Place (in English)
Topics (in Japanese) (See Japanese page)
Topics (in English)
Chair
Vice Chair
Secretary
Assistant

Paper Information
Registration To Neurocomputing (NC)
Language JPN
Title (in Japanese) (See Japanese page)
Sub Title (in Japanese) (See Japanese page)
Title (in English) Statistical mechanics of the Hopfield model with replacing units
Sub Title (in English)
Keyword(1) associative memory
Keyword(2) rebirth neuron
Keyword(3) overloading
Keyword(4) statistical mechanics
1st Author's Name Yasunao KOMATSU
1st Author's Affiliation Interdisciplinary Graduate School of Science and Engineering, Tokyo Institute of Technology()
2nd Author's Name Toru AONISHI
2nd Author's Affiliation Interdisciplinary Graduate School of Science and Engineering, Tokyo Institute of Technology
3rd Author's Name Koji KURATA
3rd Author's Affiliation Faculty of Engineering, University of the Ryukyus
Date 2009-01-19
Paper # NC2008-90
Volume (vol) vol.108
Number (no) 383
Page pp.pp.-
#Pages 6
Date of Issue