Presentation | 2019-03-06 Efficient Learning for Distillation of DNN by Self Distillation Jumpei Takagi, Motonobu Hattori, |
---|---|
PDF Download Page | PDF download Page Link |
Abstract(in Japanese) | (See Japanese page) |
Abstract(in English) | Knowledge distillation is a method to create a superior student by using knowledge obtained from a trained teacher neural network. Recent studies have shown that much superior students can be obtained by distilling the trained student further as a teacher. Distilling the knowledge through multiple generations, however, takes a long time for learning. In this paper, we propose a self distillation method which can reduce both the number of generations and learning time for knowledge distillation. In self distillation, the most accurate network is obtained during intra-generation learning, and it is used as a teacher of intra-generational distillation. Our experiments for image classification task demonstrate that the proposed self distillation acquires high accuracy with fewer generations and less learning time than the conventional method. |
Keyword(in Japanese) | (See Japanese page) |
Keyword(in English) | Knowledge Distillation / Self Distillation / Deep Learning / Image Classification |
Paper # | NC2018-83 |
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) | Efficient Learning for Distillation of DNN by Self Distillation |
Sub Title (in English) | |
Keyword(1) | Knowledge Distillation |
Keyword(2) | Self Distillation |
Keyword(3) | Deep Learning |
Keyword(4) | Image Classification |
1st Author's Name | Jumpei Takagi |
1st Author's Affiliation | University of Yamanashi(Univ of Yamanashi) |
2nd Author's Name | Motonobu Hattori |
2nd Author's Affiliation | University of Yamanashi(Univ of Yamanashi) |
Date | 2019-03-06 |
Paper # | NC2018-83 |
Volume (vol) | vol.118 |
Number (no) | NC-470 |
Page | pp.pp.209-214(NC), |
#Pages | 6 |
Date of Issue | 2019-02-25 (NC) |