| Presentation | 2017-05-26 A Parallel Forward-Backward Propagation Learning Rule for Auto-Encoder Yoshihiro Ohama, Takayoshi Yoshimura, |
|---|---|
| PDF Download Page |  PDF download Page Link |
| Abstract(in Japanese) | (See Japanese page) |
| Abstract(in English) | Auto-encoder is known as a hourglass neural network for acquiring essential representations from multi-dimensional data set. Learning identity mapping by the back-propagation rule, the representations will be acquired in general. In order to reduce iterations for updating neural weights, many automated parameter adjustment method for the learning rule and neural activation functions for avoiding vanishing gradients have been proposed. In this work, we introduce that propagating errors both from input and output layer in parallel could decrease the iterations by half. Feasibility of this approach is confirmed by following two simple computer simulations. |
| Keyword(in Japanese) | (See Japanese page) |
| Keyword(in English) | Auto-Encoder / Representation Learning / Identity Mapping / Back-propagation Rule / Forward-propagation Rule |
| Paper # | NC2017-3 |
| Date of Issue | 2017-05-19 (NC) |
| Conference Information | |
| Committee | MBE / NC |
|---|---|
| Conference Date | 2017/5/26(1days) |
| Place (in Japanese) | (See Japanese page) |
| Place (in English) | Toyama Prefectural Univ. |
| Topics (in Japanese) | (See Japanese page) |
| Topics (in English) | |
| Chair | Yutaka Fukuoka(Kogakuin Univ.) / Masafumi Hagiwara(Keio Univ.) |
| Vice Chair | Kazuki Nakajima(Univ. of Toyama) / Yutaka Hirata(Chubu Univ.) |
| Secretary | Kazuki Nakajima(Kogakuin Univ.) / Yutaka Hirata(Toyama Pref. Univ.) |
| Assistant | Ryota Horie(Shibaura Inst. of Tech.) / Kim Juhyon(Univ. of Toyama) / Yoshihisa Shinozawa(Keio Univ.) / Keiichiro Inagaki(Chubu Univ.) |
| Paper Information | |
| Registration To | Technical Committee on ME and Bio Cybernetics / Technical Committee on Neurocomputing |
|---|---|
| Language | JPN |
| Title (in Japanese) | (See Japanese page) |
| Sub Title (in Japanese) | (See Japanese page) |
| Title (in English) | A Parallel Forward-Backward Propagation Learning Rule for Auto-Encoder |
| Sub Title (in English) | |
| Keyword(1) | Auto-Encoder |
| Keyword(2) | Representation Learning |
| Keyword(3) | Identity Mapping |
| Keyword(4) | Back-propagation Rule |
| Keyword(5) | Forward-propagation Rule |
| 1st Author's Name | Yoshihiro Ohama |
| 1st Author's Affiliation | Toyota Central Research and Development Labs., Inc.(Toyota CRDL) |
| 2nd Author's Name | Takayoshi Yoshimura |
| 2nd Author's Affiliation | Toyota Central Research and Development Labs., Inc.(Toyota CRDL) |
| Date | 2017-05-26 |
| Paper # | NC2017-3 |
| Volume (vol) | vol.117 |
| Number (no) | NC-64 |
| Page | pp.pp.13-18(NC), |
| #Pages | 6 |
| Date of Issue | 2017-05-19 (NC) |