| Presentation | 2017-06-23 Analysis of Sequential Learning Capability of Selective Desensitization Neural Network Tomoki Ichiba, Tomohiro Tanno, Kazumasa Horie, Jun Izawa, Syoichi Someno, Masahiko Morita, |
|---|---|
| PDF Download Page |  PDF download Page Link |
| Abstract(in Japanese) | (See Japanese page) |
| Abstract(in English) | A conventional neural network with high generalization ability is known to entirely forget the previously learned information by sequential learning. Recent applied researches have showed that selective desensitization neural network (SDNN) does not have this problem, but its factors are unclear. The present study clarifies the capability and its factors of SDNN on sequential learning by analyzing the difference in characteristics of SDNN and conventional neural networks through numerical experiments of sequential learning on two-dimensional function approximation tasks. As a result, SDNN was able to not only locally fit to each of the new data but also affect widely to complement between samples when several similar data gathered near. These characteristics of SDNN contribute to its high capability to both preserve the previous model and effectively fit to the new data while keeping high generalization ability. |
| Keyword(in Japanese) | (See Japanese page) |
| Keyword(in English) | Function approximation / Neural network / Selective Desensitization Neural Network / Sequential learning / Global generalization |
| Paper # | NC2017-9 |
| Date of Issue | 2017-06-16 (NC) |
| Conference Information | |
| Committee | NC / IPSJ-BIO / IBISML / IPSJ-MPS |
|---|---|
| Conference Date | 2017/6/23(3days) |
| Place (in Japanese) | (See Japanese page) |
| Place (in English) | Okinawa Institute of Science and Technology |
| Topics (in Japanese) | (See Japanese page) |
| Topics (in English) | Machine Learning Approach to Biodata Mining, and General |
| Chair | Masafumi Hagiwara(Keio Univ.) / / Kenji Fukumizu(ISM) |
| Vice Chair | Yutaka Hirata(Chubu Univ.) / / Masashi Sugiyama(Univ. of Tokyo) |
| Secretary | Yutaka Hirata(Tokyo Inst. of Tech.) / (Nagoya Univ.) / Masashi Sugiyama / (Kyoto Univ.) |
| Assistant | Yoshihisa Shinozawa(Keio Univ.) / Keiichiro Inagaki(Chubu Univ.) / / Ichiro Takeuchi(Nagoya Inst. of Tech.) / Toshihiro Kamishima(AIST) |
| Paper Information | |
| Registration To | Technical Committee on Neurocomputing / Special Interest Group on Bioinformatics and Genomics / Technical Committee on Infomation-Based Induction Sciences and Machine Learning / Special Interest Group on Mathematical Modeling and Problem Solving |
|---|---|
| Language | JPN |
| Title (in Japanese) | (See Japanese page) |
| Sub Title (in Japanese) | (See Japanese page) |
| Title (in English) | Analysis of Sequential Learning Capability of Selective Desensitization Neural Network |
| Sub Title (in English) | |
| Keyword(1) | Function approximation |
| Keyword(2) | Neural network |
| Keyword(3) | Selective Desensitization Neural Network |
| Keyword(4) | Sequential learning |
| Keyword(5) | Global generalization |
| 1st Author's Name | Tomoki Ichiba |
| 1st Author's Affiliation | Tsukuba University(Tsukuba Univ.) |
| 2nd Author's Name | Tomohiro Tanno |
| 2nd Author's Affiliation | Tsukuba University(Tsukuba Univ.) |
| 3rd Author's Name | Kazumasa Horie |
| 3rd Author's Affiliation | Tsukuba University(Tsukuba Univ.) |
| 4th Author's Name | Jun Izawa |
| 4th Author's Affiliation | Tsukuba University(Tsukuba Univ.) |
| 5th Author's Name | Syoichi Someno |
| 5th Author's Affiliation | Tsukuba University(Tsukuba Univ.) |
| 6th Author's Name | Masahiko Morita |
| 6th Author's Affiliation | Tsukuba University(Tsukuba Univ.) |
| Date | 2017-06-23 |
| Paper # | NC2017-9 |
| Volume (vol) | vol.117 |
| Number (no) | NC-109 |
| Page | pp.pp.27-32(NC), |
| #Pages | 6 |
| Date of Issue | 2017-06-16 (NC) |