Presentation | 2022-06-09 A Study on Accelerating Stochastic Weight Difference Propagation with Momentum Term Shahrzad Mahboubi, Hiroshi Ninomiya, |
---|---|
PDF Download Page | PDF download Page Link |
Abstract(in Japanese) | (See Japanese page) |
Abstract(in English) | With the rapid development of the IoT, there has been an increasing need to process the data on microcomputers equipped with neural networks (NN). On the other hand, the NN implemented in the microcomputer is based on the stochastic error backpropagation (SGD) method. In SGD, backpropagation of the derivative of the output layer is required for each training sample to update the weights. This is the reason that hardware and computational costs are increasing. To solve these problems, we proposed the Stochastic Weight Difference Propagation (SWDP), which uses the product of the weights and the difference of that for updating the weights. In this research, the modified algorithm of SWDP is proposed as Momentum in stochastic weighted differential propagation (MoSWDP), which introduces the momentum term to accelerate the SWDP. The proposed method is expected to suppress the increase in the number of iterations, which is one of the disadvantages of SWDP and enables faster training while maintaining training accuracy. |
Keyword(in Japanese) | (See Japanese page) |
Keyword(in English) | Neural network / gradient-based training algorithm / stochastic gradient decsent method / weight diffrence propagation / momentum term |
Paper # | NLP2022-9,CCS2022-9 |
Date of Issue | 2022-06-02 (NLP, CCS) |
Conference Information | |
Committee | CCS / NLP |
---|---|
Conference Date | 2022/6/9(2days) |
Place (in Japanese) | (See Japanese page) |
Place (in English) | |
Topics (in Japanese) | (See Japanese page) |
Topics (in English) | |
Chair | Megumi Akai(Hokkaido Univ.) / Akio Tsuneda(Kumamoto Univ.) |
Vice Chair | Masaki Aida(TMU) / Hidehiro Nakano(Tokyo City Univ.) / Hiroyuki Torikai(Hosei Univ.) |
Secretary | Masaki Aida(TDK) / Hidehiro Nakano(Shibaura Insti. of Tech.) / Hiroyuki Torikai(Sojo Univ.) |
Assistant | Tomoyuki Sasaki(Shonan Instit. of Tech.) / Hiroyasu Ando(Tsukuba Univ.) / Miki Kobayashi(Rissho Univ.) / " Hiroyuki YASUDA(The Univ. of Tokyo) / Yuichi Yokoi(Nagasaki Univ.) / Yoshikazu Yamanaka(Utsunomiya Univ.) |
Paper Information | |
Registration To | Technical Committee on Complex Communication Sciences / Technical Committee on Nonlinear Problems |
---|---|
Language | JPN |
Title (in Japanese) | (See Japanese page) |
Sub Title (in Japanese) | (See Japanese page) |
Title (in English) | A Study on Accelerating Stochastic Weight Difference Propagation with Momentum Term |
Sub Title (in English) | |
Keyword(1) | Neural network |
Keyword(2) | gradient-based training algorithm |
Keyword(3) | stochastic gradient decsent method |
Keyword(4) | weight diffrence propagation |
Keyword(5) | momentum term |
1st Author's Name | Shahrzad Mahboubi |
1st Author's Affiliation | Shonan Institute of Technology(Shonan Inst. of Tech.) |
2nd Author's Name | Hiroshi Ninomiya |
2nd Author's Affiliation | Shonan Institute of Technology(Shonan Inst. of Tech.) |
Date | 2022-06-09 |
Paper # | NLP2022-9,CCS2022-9 |
Volume (vol) | vol.122 |
Number (no) | NLP-65,CCS-66 |
Page | pp.pp.40-45(NLP), pp.40-45(CCS), |
#Pages | 6 |
Date of Issue | 2022-06-02 (NLP, CCS) |