Presentation | 2008-11-07 Neural network model with discrete and continuous information representation Jun KITAZONO, Toshiaki OMORI, Masato OKADA, |
---|---|
PDF Download Page | PDF download Page Link |
Abstract(in Japanese) | (See Japanese page) |
Abstract(in English) | An associative memory model and a neural network model with Mexican-hat type interaction are the two most typical attractor networks in the artificial neural network models. The associative memory model has discretely distributed fixed-point attractors, and realizes discrete information representation. On the other hand, the neural network model with Mexican-hat type interaction realizes continuous information representation by a line attractor, seen in working memory in the prefrontal cortex and columnar activity in the visual cortex. In the present study, we propose a neural network model which realizes discrete and continuous information representation. By analysis based on statistical mechanics, we find that the localized retrieval phase exists in the proposed model, where a memory pattern is retrieved in localized subpopulation of the network. In the localized retrieval phase, the discrete and continuous information representation is realized by orthogonality of memory patterns and neutral stability of fixed points along positions of the localized retrieval. The obtained phase diagram suggests that the anti-ferromagnetic interaction and the external field are important to generate the localized retrieval phase. |
Keyword(in Japanese) | (See Japanese page) |
Keyword(in English) | Mexican-hat type interaction / Associative memory model / Localized activity / Statistical mechanics |
Paper # | NC2008-61 |
Date of Issue |
Conference Information | |
Committee | NC |
---|---|
Conference Date | 2008/10/31(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) | Neural network model with discrete and continuous information representation |
Sub Title (in English) | |
Keyword(1) | Mexican-hat type interaction |
Keyword(2) | Associative memory model |
Keyword(3) | Localized activity |
Keyword(4) | Statistical mechanics |
1st Author's Name | Jun KITAZONO |
1st Author's Affiliation | Graduate School of Frontier Science, The University of Tokyo() |
2nd Author's Name | Toshiaki OMORI |
2nd Author's Affiliation | Graduate School of Frontier Science, The University of Tokyo:RIKEN Brain Science Institute |
3rd Author's Name | Masato OKADA |
3rd Author's Affiliation | Graduate School of Frontier Science, The University of Tokyo:RIKEN Brain Science Institute |
Date | 2008-11-07 |
Paper # | NC2008-61 |
Volume (vol) | vol.108 |
Number (no) | 281 |
Page | pp.pp.- |
#Pages | 6 |
Date of Issue |