Presentation | 2018-09-20 Structural Learning of Neural Network based on Outputs of Neurons Koji Kamma, Yuki Isoda, Toshikazu Wada, |
---|---|
PDF Download Page | PDF download Page Link |
Abstract(in Japanese) | (See Japanese page) |
Abstract(in English) | This paper presents a method of compaction on DNNs. Our method includes two steps, 1) Neuro-Coding: feature the neuron behaviors by their outputs, and 2) Neuro-Unification: unify the neurons which behave similarly. While the existing pruning methods just remove the neurons or the weights, our method let another neuron succeed the behavior of the removed one. Therefore, our method can reduce the parameters while the sacrifice of inference accuracy stays small. Moreover, many of the existing methods require people to manually determine the network structure after compaction, but our method can explore an adequate structure automatically. Thus, our method could be valuable as a structural learning method. |
Keyword(in Japanese) | (See Japanese page) |
Keyword(in English) | Deep Neural Network (DNN) / Compaction / Neuro-Coding / Neuro-Unification |
Paper # | PRMU2018-40,IBISML2018-17 |
Date of Issue | 2018-09-13 (PRMU, IBISML) |
Conference Information | |
Committee | PRMU / IBISML / IPSJ-CVIM |
---|---|
Conference Date | 2018/9/20(2days) |
Place (in Japanese) | (See Japanese page) |
Place (in English) | |
Topics (in Japanese) | (See Japanese page) |
Topics (in English) | |
Chair | Shinichi Sato(NII) / Hisashi Kashima(Kyoto Univ.) |
Vice Chair | Yoshihisa Ijiri(Omron) / Toru Tamaki(Hiroshima Univ.) / Masashi Sugiyama(Univ. of Tokyo) / Koji Tsuda(Univ. of Tokyo) |
Secretary | Yoshihisa Ijiri(NEC) / Toru Tamaki(Osaka Univ.) / Masashi Sugiyama(Nagoya Inst. of Tech.) / Koji Tsuda(AIST) |
Assistant | Go Irie(NTT) / Yoshitaka Ushiku(Univ. of Tokyo) / Tomoharu Iwata(NTT) / Shigeyuki Oba(Kyoto Univ.) |
Paper Information | |
Registration To | Technical Committee on Pattern Recognition and Media Understanding / Technical Committee on Infomation-Based Induction Sciences and Machine Learning / Special Interest Group on Computer Vision and Image Media |
---|---|
Language | ENG-JTITLE |
Title (in Japanese) | (See Japanese page) |
Sub Title (in Japanese) | (See Japanese page) |
Title (in English) | Structural Learning of Neural Network based on Outputs of Neurons |
Sub Title (in English) | |
Keyword(1) | Deep Neural Network (DNN) |
Keyword(2) | Compaction |
Keyword(3) | Neuro-Coding |
Keyword(4) | Neuro-Unification |
1st Author's Name | Koji Kamma |
1st Author's Affiliation | Wakayama University(Wakayama Univ.) |
2nd Author's Name | Yuki Isoda |
2nd Author's Affiliation | Wakayama University(Wakayama Univ.) |
3rd Author's Name | Toshikazu Wada |
3rd Author's Affiliation | Wakayama University(Wakayama Univ.) |
Date | 2018-09-20 |
Paper # | PRMU2018-40,IBISML2018-17 |
Volume (vol) | vol.118 |
Number (no) | PRMU-219,IBISML-220 |
Page | pp.pp.31-36(PRMU), pp.31-36(IBISML), |
#Pages | 6 |
Date of Issue | 2018-09-13 (PRMU, IBISML) |