| Presentation | 2019-03-05 Novel Backpropagation Algorithm Considering Energy Rintaro Kanada, Masafumi Hagiwara, |
|---|---|
| PDF Download Page |  PDF download Page Link |
| Abstract(in Japanese) | (See Japanese page) |
| Abstract(in English) | In this paper, we propose a novel backpropagation(BP) algorithm considering energy. Neural network (NN) can be classified as the feedfoward type and the interconnecting type. In general, the evaluation function in learning is error in the feedfoward type and energy in the interconnecting type. However, the internal state of the network isn't considered in the error. On the other hand, the internal state of the network is considered in the energy. Therefore, we defined the energy so that the internal state of the feedfoward network could be considered. And, we propose an algorithm that uses the difference between the energy for correct answer output and the actual energy in each layer so that the energy becomes lower. In the evaluation experiments, we compared with existing BP using classification tasks. As a result, it is confirmed that convergence of learning is accelerated and there remains possibility of improvement of accuracy and robustness to noise. |
| Keyword(in Japanese) | (See Japanese page) |
| Keyword(in English) | backpropagation / neural network / energy / information amount |
| Paper # | NC2018-63 |
| Date of Issue | 2019-02-25 (NC) |
| Conference Information | |
| Committee | NC / MBE |
|---|---|
| Conference Date | 2019/3/4(3days) |
| Place (in Japanese) | (See Japanese page) |
| Place (in English) | University of Electro Communications |
| Topics (in Japanese) | (See Japanese page) |
| Topics (in English) | |
| Chair | Yutaka Hirata(Chubu Univ.) / Masaki Kyoso(TCU) |
| Vice Chair | Hayaru Shouno(UEC) / Taishin Nomura(Osaka Univ.) |
| Secretary | Hayaru Shouno(Nagoya Univ.) / Taishin Nomura(NAIST) |
| Assistant | Keiichiro Inagaki(Chubu Univ.) / Takashi Shinozaki(NICT) / Takumi Kobayashi(YNU) / Ｙａｓｕｙｕｋｉ Ｓｕｚｕｋｉ(Osaka Univ.) |
| Paper Information | |
| Registration To | Technical Committee on Neurocomputing / Technical Committee on ME and Bio Cybernetics |
|---|---|
| Language | JPN |
| Title (in Japanese) | (See Japanese page) |
| Sub Title (in Japanese) | (See Japanese page) |
| Title (in English) | Novel Backpropagation Algorithm Considering Energy |
| Sub Title (in English) | |
| Keyword(1) | backpropagation |
| Keyword(2) | neural network |
| Keyword(3) | energy |
| Keyword(4) | information amount |
| 1st Author's Name | Rintaro Kanada |
| 1st Author's Affiliation | Keio University(Keio Univ.) |
| 2nd Author's Name | Masafumi Hagiwara |
| 2nd Author's Affiliation | Keio University(Keio Univ.) |
| Date | 2019-03-05 |
| Paper # | NC2018-63 |
| Volume (vol) | vol.118 |
| Number (no) | NC-470 |
| Page | pp.pp.105-110(NC), |
| #Pages | 6 |
| Date of Issue | 2019-02-25 (NC) |