Presentation | 2002/3/13 H_∞- Learning : Local Optimization Approach Kiyoshi NISIYAMA, Kiyohiko SUZUKI, |
---|---|
PDF Download Page | PDF download Page Link |
Abstract(in Japanese) | (See Japanese page) |
Abstract(in English) | Backpropagation (BP) method is widely known as a learning algorithm of layered neural networks. However, the learning rate is too late, and it is strongly affected by the initial values of weight coefficients and thresholds. In this paper, H_∞- learning of layered neural networks is proposed, and a new learning algorithm, called the l-EHF algorithm, is derived, comparing with the backpropagation (BP) and the extended Kalman filter (EKF) learning algorithms. The robustness of H_∞-learning to variances in the initial weights and thresholds is verified by computer simulations. |
Keyword(in Japanese) | (See Japanese page) |
Keyword(in English) | learning algorithm / H_∞ theory / neural network / robust estimation / back-propagation / Kalman filter |
Paper # | NC2001-224 |
Date of Issue |
Conference Information | |
Committee | NC |
---|---|
Conference Date | 2002/3/13(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) | H_∞- Learning : Local Optimization Approach |
Sub Title (in English) | |
Keyword(1) | learning algorithm |
Keyword(2) | H_∞ theory |
Keyword(3) | neural network |
Keyword(4) | robust estimation |
Keyword(5) | back-propagation |
Keyword(6) | Kalman filter |
1st Author's Name | Kiyoshi NISIYAMA |
1st Author's Affiliation | Department of Computer and Information Science, Iwate University() |
2nd Author's Name | Kiyohiko SUZUKI |
2nd Author's Affiliation | Department of Computer and Information Science, Iwate University |
Date | 2002/3/13 |
Paper # | NC2001-224 |
Volume (vol) | vol.101 |
Number (no) | 737 |
Page | pp.pp.- |
#Pages | 8 |
Date of Issue |