| Presentation | 1996/5/24 A Sequential Learning Method Using Pseudo Training Data for Multilayer Neural Networks Takayasu Ninagawa, Kenji Nakayama, Kazushi Ikeda, |
|---|---|
| PDF Download Page | PDF download Page Link |
| Abstract(in Japanese) | (See Japanese page) |
| Abstract(in English) | In this paper, the following learning problem is considered. (1)Training data are given step by step. (2)The data used in the previous training are not stored. (3)The neural network should realize the desired functions for all the training data. The problem is how to keep the desired functions for the previous data in training the new data. For this purpose, we propose a training method, which employs "pseudo training data". The pseudo training data are generated randomly in the data space. Some of the pseudo data, with which the neural network outputs the desired value are further selected. The pseudo data locate close to the new training data are removed. Since the neural network under the training is not optimum, some pseudo data may be used to keep undesirable response. Such data are detected and removed during the training process. The proposed training method was applied to two kinds of pattern classification problems. Computer simulation results demonstrate efficiency of this method. |
| Keyword(in Japanese) | (See Japanese page) |
| Keyword(in English) | Multilayer neural network / Supervised learning / Learning / Iterative learning / Pattern classification |
| Paper # | NC96-2 |
| Date of Issue |
| Conference Information | |
| Committee | NC |
|---|---|
| Conference Date | 1996/5/24(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) | A Sequential Learning Method Using Pseudo Training Data for Multilayer Neural Networks |
| Sub Title (in English) | |
| Keyword(1) | Multilayer neural network |
| Keyword(2) | Supervised learning |
| Keyword(3) | Learning |
| Keyword(4) | Iterative learning |
| Keyword(5) | Pattern classification |
| 1st Author's Name | Takayasu Ninagawa |
| 1st Author's Affiliation | Graduate School of Eng., Kanazawa Univ.() |
| 2nd Author's Name | Kenji Nakayama |
| 2nd Author's Affiliation | Faculty of Eng., Kanazawa Univ. |
| 3rd Author's Name | Kazushi Ikeda |
| 3rd Author's Affiliation | Faculty of Eng., Kanazawa Univ. |
| Date | 1996/5/24 |
| Paper # | NC96-2 |
| Volume (vol) | vol.96 |
| Number (no) | 76 |
| Page | pp.pp.- |
| #Pages | 7 |
| Date of Issue | |