Relationship between Recognition Accuracy and Numerical Precision in Convolutional Neural Network Models
Summary :
Quantization is an important technique for implementing convolutional neural networks on edge devices. Quantization often requires relearning, but relearning sometimes cannot be always be applied because of issues such as cost or privacy. In such cases, it is important to know the numerical precision required to maintain accuracy. We accurately simulate calculations on hardware and accurately measure the relationship between accuracy and numerical precision.
- Publication
- IEICE TRANSACTIONS on Information Vol.E103-D No.12 pp.2528-2529
- Publication Date
- 2020/12/01
- Publicized
- 2020/08/13
- Online ISSN
- 1745-1361
- DOI
- 10.1587/transinf.2020PAL0002
- Type of Manuscript
- Special Section LETTER (Special Section on Parallel, Distributed, and Reconfigurable Computing, and Networking)
- Category
- Computer System
Authors
Yasuhiro NAKAHARA
Kumamoto University
Masato KIYAMA
Kumamoto University
Motoki AMAGASAKI
Kumamoto University
Masahiro IIDA
Kumamoto University
Keyword
Latest Issue
Copyrights notice of machine-translated contents
The copyright of the original papers published on this site belongs to IEICE. Unauthorized use of the original or translated papers is prohibited. See IEICE Provisions on Copyright for details.
Cite this
Copy
Yasuhiro NAKAHARA, Masato KIYAMA, Motoki AMAGASAKI, Masahiro IIDA, "Relationship between Recognition Accuracy and Numerical Precision in Convolutional Neural Network Models" in IEICE TRANSACTIONS on Information,
vol. E103-D, no. 12, pp. 2528-2529, December 2020, doi: 10.1587/transinf.2020PAL0002.
Abstract: Quantization is an important technique for implementing convolutional neural networks on edge devices. Quantization often requires relearning, but relearning sometimes cannot be always be applied because of issues such as cost or privacy. In such cases, it is important to know the numerical precision required to maintain accuracy. We accurately simulate calculations on hardware and accurately measure the relationship between accuracy and numerical precision.
URL: https://global.ieice.org/en_transactions/information/10.1587/transinf.2020PAL0002/_p
Copy
@ARTICLE{e103-d_12_2528,
author={Yasuhiro NAKAHARA, Masato KIYAMA, Motoki AMAGASAKI, Masahiro IIDA, },
journal={IEICE TRANSACTIONS on Information},
title={Relationship between Recognition Accuracy and Numerical Precision in Convolutional Neural Network Models},
year={2020},
volume={E103-D},
number={12},
pages={2528-2529},
abstract={Quantization is an important technique for implementing convolutional neural networks on edge devices. Quantization often requires relearning, but relearning sometimes cannot be always be applied because of issues such as cost or privacy. In such cases, it is important to know the numerical precision required to maintain accuracy. We accurately simulate calculations on hardware and accurately measure the relationship between accuracy and numerical precision.},
keywords={},
doi={10.1587/transinf.2020PAL0002},
ISSN={1745-1361},
month={December},}
Copy
TY - JOUR
TI - Relationship between Recognition Accuracy and Numerical Precision in Convolutional Neural Network Models
T2 - IEICE TRANSACTIONS on Information
SP - 2528
EP - 2529
AU - Yasuhiro NAKAHARA
AU - Masato KIYAMA
AU - Motoki AMAGASAKI
AU - Masahiro IIDA
PY - 2020
DO - 10.1587/transinf.2020PAL0002
JO - IEICE TRANSACTIONS on Information
SN - 1745-1361
VL - E103-D
IS - 12
JA - IEICE TRANSACTIONS on Information
Y1 - December 2020
AB - Quantization is an important technique for implementing convolutional neural networks on edge devices. Quantization often requires relearning, but relearning sometimes cannot be always be applied because of issues such as cost or privacy. In such cases, it is important to know the numerical precision required to maintain accuracy. We accurately simulate calculations on hardware and accurately measure the relationship between accuracy and numerical precision.
ER -
English
