Paper Abstract and Keywords |
Presentation |
2022-06-08 15:25
A Compact High-Speed CNN Implementation based on Redundant Computational Analysis and FPGA Acceleration Li Qi, Li Hengyi, Meng Lin (Ritsumeikan Univ.) RECONF2022-21 |
Abstract |
(in Japanese) |
(See Japanese page) |
(in English) |
Convolutional Neural Networks (CNNs) have achieved high performance and are widely used in various applications. However, CNN's are computational-intensive and resource-consuming, causing the development of CNN applications is limited especially in the embedded systems. Therefore, we propose a dynamic CNN pruning method based on redundant computational analysis. The proposal aims to realize model compression within the setting performance degradation through dynamic iterative channel pruning. Experimental results show the proposal reduces about 80% of parameters and FLOPS. Furthermore, the compacted CNN model is implemented on the FPGA and achieved about 40% speedup in inference time. |
Keyword |
(in Japanese) |
(See Japanese page) |
(in English) |
Convolutional neural network / Channel pruning / Redundant calculation analysis / FPGA / / / / |
Reference Info. |
IEICE Tech. Rep., vol. 122, no. 60, RECONF2022-21, pp. 89-94, June 2022. |
Paper # |
RECONF2022-21 |
Date of Issue |
2022-05-31 (RECONF) |
ISSN |
Online edition: ISSN 2432-6380 |
Copyright and reproduction |
All rights are reserved and no part of this publication may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopy, recording, or any information storage and retrieval system, without permission in writing from the publisher. Notwithstanding, instructors are permitted to photocopy isolated articles for noncommercial classroom use without fee. (License No.: 10GA0019/12GB0052/13GB0056/17GB0034/18GB0034) |
Notes on Review |
This article is a technical report without peer review, and its polished version will be published elsewhere. |
Download PDF |
RECONF2022-21 |
Conference Information |
Committee |
RECONF |
Conference Date |
2022-06-07 - 2022-06-08 |
Place (in Japanese) |
(See Japanese page) |
Place (in English) |
CCS, Univ. of Tsukuba |
Topics (in Japanese) |
(See Japanese page) |
Topics (in English) |
Reconfigurable system, etc. |
Paper Information |
Registration To |
RECONF |
Conference Code |
2022-06-RECONF |
Language |
Japanese |
Title (in Japanese) |
(See Japanese page) |
Sub Title (in Japanese) |
(See Japanese page) |
Title (in English) |
A Compact High-Speed CNN Implementation based on Redundant Computational Analysis and FPGA Acceleration |
Sub Title (in English) |
|
Keyword(1) |
Convolutional neural network |
Keyword(2) |
Channel pruning |
Keyword(3) |
Redundant calculation analysis |
Keyword(4) |
FPGA |
Keyword(5) |
|
Keyword(6) |
|
Keyword(7) |
|
Keyword(8) |
|
1st Author's Name |
Li Qi |
1st Author's Affiliation |
Ritsumeikan University (Ritsumeikan Univ.) |
2nd Author's Name |
Li Hengyi |
2nd Author's Affiliation |
Ritsumeikan University (Ritsumeikan Univ.) |
3rd Author's Name |
Meng Lin |
3rd Author's Affiliation |
Ritsumeikan University (Ritsumeikan Univ.) |
4th Author's Name |
|
4th Author's Affiliation |
() |
5th Author's Name |
|
5th Author's Affiliation |
() |
6th Author's Name |
|
6th Author's Affiliation |
() |
7th Author's Name |
|
7th Author's Affiliation |
() |
8th Author's Name |
|
8th Author's Affiliation |
() |
9th Author's Name |
|
9th Author's Affiliation |
() |
10th Author's Name |
|
10th Author's Affiliation |
() |
11th Author's Name |
|
11th Author's Affiliation |
() |
12th Author's Name |
|
12th Author's Affiliation |
() |
13th Author's Name |
|
13th Author's Affiliation |
() |
14th Author's Name |
|
14th Author's Affiliation |
() |
15th Author's Name |
|
15th Author's Affiliation |
() |
16th Author's Name |
|
16th Author's Affiliation |
() |
17th Author's Name |
|
17th Author's Affiliation |
() |
18th Author's Name |
|
18th Author's Affiliation |
() |
19th Author's Name |
|
19th Author's Affiliation |
() |
20th Author's Name |
|
20th Author's Affiliation |
() |
Speaker |
Author-1 |
Date Time |
2022-06-08 15:25:00 |
Presentation Time |
25 minutes |
Registration for |
RECONF |
Paper # |
RECONF2022-21 |
Volume (vol) |
vol.122 |
Number (no) |
no.60 |
Page |
pp.89-94 |
#Pages |
6 |
Date of Issue |
2022-05-31 (RECONF) |
|