Stochastic resonance and noise effects in threshold systems : from the viewpoint of entropy and applications
| Project/Area Number |
16560052
|
|---|
| Research Category |
Grant-in-Aid for Scientific Research (C)
|
| Allocation Type | Single-year Grants |
|---|
| Section | 一般 |
|---|
| Research Field |
Engineering fundamentals
|
|---|
| Research Institution | Kyoto University |
|---|
Principal Investigator |
MUNAKATA Toyonori Kyoto University, School of Informatics, Professor, 情報学研究科, 教授 (40026357)
|
|---|
| Project Period (FY) |
2004 – 2005
|
| Project Status |
Completed (Fiscal Year 2005)
|
| Budget Amount *help |
¥3,700,000 (Direct Cost: ¥3,700,000)
Fiscal Year 2005: ¥1,000,000 (Direct Cost: ¥1,000,000)
Fiscal Year 2004: ¥2,700,000 (Direct Cost: ¥2,700,000)
|
|---|
| Keywords | Stochastic resonance / Filtering theory / threshold system / self-tuning / adaptive learning of threshold value / ergode-nonergode transition / mutual information / noise effect / stochastic resonance / threshold system |
|---|
| Research Abstract |
Stochastic resonance(SR) in threshold systems was studied from information theoretical standpoints. Especially the mutual information between input and output signals was shown to be a good measure for estimation of performance in SR.
The main results are as follows :
1) Ergode-nonergode transition in a threshold system : When we have time-dependent input signals, we found that the Output signal corresponding to the input can not determined uniquely and the output depends on the history of the Input signals. This was analyzed based on time-dependent probability distribution and the limitation of a mean-field approach was clarified.
2) SR and mutual information : Both for periodic and aperiodic input signals, we showed that the mutual information can be a good measure for information transfer like the signal-to-noise ratio and power norm, extensively used in literatures.
3) Self-tuning and effects of noise : Modelling high performance of auditory systems in animals, we proposed a simple adaptation rule for a threshold value and confirmed that our system nicely processes informations, even if it is put in a quiet circumstance where noise effects are weak.
4) Extension of Macnamara-Wiesenfeld(MW) treatment of SR : We extended the treatment of MW so that our approach is applicable when the input signal is not weak. Self-tuning was also shown to be formulated within this framework.
|
Report
(3 results)
Research Products
(9 results)
