HUSCAP logo Hokkaido Univ. logo

Hokkaido University Collection of Scholarly and Academic Papers >
Graduate School of Information Science and Technology / Faculty of Information Science and Technology >
Peer-reviewed Journal Articles, etc >

A study on a low power optimization algorithm for an edge-AI device

Files in This Item:

The file(s) associated with this item can be obtained from the following URL:https://doi.org/10.1587/nolta.10.373


Title: A study on a low power optimization algorithm for an edge-AI device
Authors: Kaneko, Tatsuya1 Browse this author
Orimo, Kentaro2 Browse this author
Hida, Itaru3 Browse this author
Takamaeda, Shinya4 Browse this author →KAKEN DB
Ikebe, Masayuki5 Browse this author →KAKEN DB
Motomura, Masato6 Browse this author
Asai, Tetsuya7 Browse this author →KAKEN DB
Authors(alt): Yamazaki, Shinya4
Keywords: machine learning
edge AI
training algorithm
backpropagation
quantization
low power
Issue Date: Oct-2019
Publisher: 電子情報通信学会(The Institute of Electronics, Information and Communication Engineers / IEICE)
Journal Title: Nonlinear theory and its applications, IEICE
Volume: 10
Issue: 4
Start Page: 373
End Page: 389
Publisher DOI: 10.1587/nolta.10.373
Abstract: Although research on the inference phase of edge artificial intelligence (AI) has made considerable improvement, the required training phase remains an unsolved problem. Neural network (NN) processing has two phases: inference and training. In the training phase, a NN incurs high calculation cost. The number of bits (bitwidth) in the training phase is several orders of magnitude larger than that in the inference phase. Training algorithms, optimized to software, are not appropriate for training hardware-oriented NNs. Therefore, we propose a new training algorithm for edge AI: backpropagation (BP) using a ternarized gradient. This ternarized backpropagation (TBP) provides a balance between calculation cost and performance. Empirical results demonstrate that in a two-class classification task, TBP works well in practice and compares favorably with 16-bit BP (Fixed-BP).
Rights: Copyright ©2019 The Institute of Electronics, Information and Communication Engineers
https://search.ieice.org/
Type: article
URI: http://hdl.handle.net/2115/76016
Appears in Collections:情報科学院・情報科学研究院 (Graduate School of Information Science and Technology / Faculty of Information Science and Technology) > 雑誌発表論文等 (Peer-reviewed Journal Articles, etc)

Export metadata:

OAI-PMH ( junii2 , jpcoar )

MathJax is now OFF:


 

Feedback - Hokkaido University