Presentation 1999/1/21
A New Associative Memory Algorithm with Multiple Hopfield Neural Networks
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) Generalized Hebbian rule which is sometimes referred to as the outer product method is the most well-known learning algorithm for associative memory with Hopfield neural networks. The algorithm of the generalized Hebbian rule is very simple, but there is a serious problem that if we store many patterns with this rule then a number of spurious memories arise in general. In this paper, we propose a new learning and recalling algorithm which never generates spurious memories. The algorithm is summarized as follows: 1)A given set of memory patterns is devided into some groups, and for each group, we store the memory patterns to a Hopfield neural network by using the generalized Hebbian rule. 2)We impose a given input pattern on every network as the initial state, and find the final state of each network. 3)If some of the networks converged to a spurious memory, these networks and the corresponding memory patterns are removed. 4)We repeat 1) to 3) until only one network is remained. The final state of the remaining Hopfield neural network is regarded as the output of our new algorithm.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) neural networks / Hopfield model / associative memory / spurious memory
Paper # NC98-69
Date of Issue

Conference Information
Committee NC
Conference Date 1999/1/21(1days)
Place (in Japanese) (See Japanese page)
Place (in English)
Topics (in Japanese) (See Japanese page)
Topics (in English)
Vice Chair

Paper Information
Registration To Neurocomputing (NC)
Language JPN
Title (in Japanese) (See Japanese page)
Sub Title (in Japanese) (See Japanese page)
Title (in English) A New Associative Memory Algorithm with Multiple Hopfield Neural Networks
Sub Title (in English)
Keyword(1) neural networks
Keyword(2) Hopfield model
Keyword(3) associative memory
Keyword(4) spurious memory
1st Author's Name Jun TSUZURUGI
1st Author's Affiliation Department of Informatics, Kyushu University()
2nd Author's Name Norikazu TAKAHASHI
2nd Author's Affiliation Department of Informatics, Kyushu University
Date 1999/1/21
Paper # NC98-69
Volume (vol) vol.98
Number (no) 526
Page pp.pp.-
#Pages 7
Date of Issue