| Presentation | 1994/5/19 An error-correcting learning algorithm using double hysteresis thresholds for associative memory Katsuaki Nishimura, Kenji Nakayama, |
|---|---|
| PDF Download Page | PDF download Page Link |
| Abstract(in Japanese) | (See Japanese page) |
| Abstract(in English) | An associative memory using fixed and variable hysteresis processes,in learning and recalling process,has been proposed by authors.This model can achieve a large memory capacity and very low noise sensitivity.However,a relation between weight change and the hysteresis threshold has not been well discussed.In this paper, a new learning algorithm is proposed,which is based on an error- correcting method.However,to avoid unstable behavior,double hysteresis thresholds are introduced.Unit states are updated using T.The error is evaluated using T+dT instead of T.This means 'over correction'.Stable and fast convergence can be obtained.Relations between η=dT, T and convergence rate and noise sensitivity are dis cussed,resulting the optimum selection for η.Furthemore,the order of presenting training data is optimized taking correlation into account.In the recalling process,a controlling method for T(n) is proposed to achieve fast recalling from noisy patterns. |
| Keyword(in Japanese) | (See Japanese page) |
| Keyword(in English) | Recurrent neural networks / Hysteresis threshold / Noise performance / Error correcting method / Associative memory |
| Paper # | NC94-12 |
| Date of Issue |
| Conference Information | |
| Committee | NC |
|---|---|
| Conference Date | 1994/5/19(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) | An error-correcting learning algorithm using double hysteresis thresholds for associative memory |
| Sub Title (in English) | |
| Keyword(1) | Recurrent neural networks |
| Keyword(2) | Hysteresis threshold |
| Keyword(3) | Noise performance |
| Keyword(4) | Error correcting method |
| Keyword(5) | Associative memory |
| 1st Author's Name | Katsuaki Nishimura |
| 1st Author's Affiliation | Depertment of Electrical and Computer engineering,Faculty of Technology,Kanazawa University() |
| 2nd Author's Name | Kenji Nakayama |
| 2nd Author's Affiliation | Depertment of Electrical and Computer engineering,Faculty of Technology,Kanazawa University |
| Date | 1994/5/19 |
| Paper # | NC94-12 |
| Volume (vol) | vol.94 |
| Number (no) | 40 |
| Page | pp.pp.- |
| #Pages | 8 |
| Date of Issue | |