Presentation 2004/6/18
Stochastic transition of attractors in an associative memory model with correlated noise
Masaki KAWAMURA, Masato OKADA,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) The mechanism of correlated firing has been analyzed in various models. However, its function has not been discussed enough. Aoki and Aoyagi have shown that state of network transits by not thermal noise but correlated noise. We discuss dynamics of a recurrent neural network with correlated noise in order to analyze effect of the correlated noise. In this report, we introduce two types of noise for each neuron : independent noise and common noise. Because of the effects of the common noise, the correlation between neural inputs cannot be ignored, and then behavior of the network has sample dependence. The associative memory model stores hierarchically correlated patterns. We derive a macroscopic dynamical description as a recurrence relation form of a probability density function, when the common noise exists. Furthermore, stochastic transition from a stored pattern to a mixed state by the common noise is analyzed. The results by computer simulations agree with those of theory.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) associative memory model / correlated firing / sample dependent
Paper # NC2004-34
Date of Issue

Conference Information
Committee NC
Conference Date 2004/6/18(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) Stochastic transition of attractors in an associative memory model with correlated noise
Sub Title (in English)
Keyword(1) associative memory model
Keyword(2) correlated firing
Keyword(3) sample dependent
1st Author's Name Masaki KAWAMURA
1st Author's Affiliation Faculty of Science, Yamaguchi University()
2nd Author's Name Masato OKADA
2nd Author's Affiliation Laboratory for Mathematical Neuroscience, RIKEN Brain Science Institute:"Intelligent Cooperation and Control", PRESTO, Japan Science and Technology Agency
Date 2004/6/18
Paper # NC2004-34
Volume (vol) vol.104
Number (no) 140
Page pp.pp.-
#Pages 6
Date of Issue