Presentation 2012-01-26
Statistic and dynamics of the Hopfield model with unlearning term
Haruka OTANI, Midori YOSHIDA, Tatsuya UEZU,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) In the Hopfield model, there is a low temperature region where the Hopfield attractor representing a memory state and mixed states representing mixture of several memory states are stable. In this paper, modifying the Hopfield model with a few memory patterns by subtracting a mixed state as an unlearning term, we studied whether it is possible to make the mixed state unstable or not. We drew a phase diagram in the parameter space of the unlearning coefficient ri and the 'temperature' T which is a measure of probabilistic nature of the system, and found that the mixed state can be deleted by unlearning. We studied dynamical behavior of the system, and proved that the static and dynamic stabilities coincide. Furthermore, we studied the Hopfield model with several unlearning terms, and confirmed that the system is structurally sable. Theoretical results and numerical results by Monte Carlo simulations agree quite well.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) Hopfield model / unlearning / Hopfield sttractor / mixed state
Paper # NC2011-100
Date of Issue

Conference Information
Committee NC
Conference Date 2012/1/19(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) Statistic and dynamics of the Hopfield model with unlearning term
Sub Title (in English)
Keyword(1) Hopfield model
Keyword(2) unlearning
Keyword(3) Hopfield sttractor
Keyword(4) mixed state
1st Author's Name Haruka OTANI
1st Author's Affiliation Graduate School of Humanities and Sciences, Nara Women's University()
2nd Author's Name Midori YOSHIDA
2nd Author's Affiliation Graduate School of Humanities and Sciences, Nara Women's University
3rd Author's Name Tatsuya UEZU
3rd Author's Affiliation Graduate School of Humanities and Sciences, Nara Women's University
Date 2012-01-26
Paper # NC2011-100
Volume (vol) vol.111
Number (no) 419
Page pp.pp.-
#Pages 6
Date of Issue