IEICE Technical Committee Submission System
Conference Paper's Information
Online Proceedings
[Sign in]
Tech. Rep. Archives
 Go Top Page Go Previous   [Japanese] / [English] 

Paper Abstract and Keywords
Presentation 2025-02-18 13:15
Reliable Memristor-Based Neural Networks with Fault-Injected Training
Md. Sihabul Islam, Ryota Eguchi, Michiko Inoue (NAIST) DC2024-109
Abstract (in Japanese) (See Japanese page) 
(in English) Energy-efficient and high-performance neural network (NN) accelerators are required to meet the increasing demands for executing neural network (NN) learning and inference on edge devices. Memristor crossbar (MC) devices in neuromorphic computing (NC) provide a promising solution for accelerating NNs. However, immature fabrication technology makes faulty memristors inevitable, and the presence of stuck-at faults (SAFs) in MC devices substantially reduces the inference accuracy of NC systems. This paper proposes a reliability-aware design framework for an NC system utilizing MC devices by combining fault-injected training and IC-specific mapping. We train a neural network where faults are deliberately injected and then map the trained model to MC devices considering fault location for individual devices. The proposed method avoids the requirement for either redundant hardware or retraining for individual MC devices. The experimental results show that the proposed framework achieves high inference accuracy for several different faulty MCs and enhances the reliability of NC system.
Keyword (in Japanese) (See Japanese page) 
(in English) neuromorphic computing / memristor crossbar (MC) / stuck-at-faults / fault-injected training / mapping / / /  
Reference Info. IEICE Tech. Rep., vol. 124, no. 374, DC2024-109, pp. 19-24, Feb. 2025.
Paper # DC2024-109 
Date of Issue 2025-02-11 (DC) 
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)
Download PDF DC2024-109

Conference Information
Committee DC  
Conference Date 2025-02-18 - 2025-02-18 
Place (in Japanese) (See Japanese page) 
Place (in English) Kikai-Shinko-Kaikan Bldg. 
Topics (in Japanese) (See Japanese page) 
Topics (in English)  
Paper Information
Registration To DC 
Conference Code 2025-02-DC 
Language English 
Title (in Japanese) (See Japanese page) 
Sub Title (in Japanese) (See Japanese page) 
Title (in English) Reliable Memristor-Based Neural Networks with Fault-Injected Training 
Sub Title (in English)  
Keyword(1) neuromorphic computing  
Keyword(2) memristor crossbar (MC)  
Keyword(3) stuck-at-faults  
Keyword(4) fault-injected training  
Keyword(5) mapping  
Keyword(6)  
Keyword(7)  
Keyword(8)  
1st Author's Name Md. Sihabul Islam  
1st Author's Affiliation Nara Institute of Science and Technology (NAIST)
2nd Author's Name Ryota Eguchi  
2nd Author's Affiliation Nara Institute of Science and Technology (NAIST)
3rd Author's Name Michiko Inoue  
3rd Author's Affiliation Nara Institute of Science and Technology (NAIST)
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 ()
21st Author's Name  
21st Author's Affiliation ()
22nd Author's Name  
22nd Author's Affiliation ()
23rd Author's Name  
23rd Author's Affiliation ()
24th Author's Name  
24th Author's Affiliation ()
25th Author's Name  
25th Author's Affiliation ()
26th Author's Name / /
26th Author's Affiliation ()
()
27th Author's Name / /
27th Author's Affiliation ()
()
28th Author's Name / /
28th Author's Affiliation ()
()
29th Author's Name / /
29th Author's Affiliation ()
()
30th Author's Name / /
30th Author's Affiliation ()
()
31st Author's Name / /
31st Author's Affiliation ()
()
32nd Author's Name / /
32nd Author's Affiliation ()
()
33rd Author's Name / /
33rd Author's Affiliation ()
()
34th Author's Name / /
34th Author's Affiliation ()
()
35th Author's Name / /
35th Author's Affiliation ()
()
36th Author's Name / /
36th Author's Affiliation ()
()
Speaker Author-1 
Date Time 2025-02-18 13:15:00 
Presentation Time 25 minutes 
Registration for DC 
Paper # DC2024-109 
Volume (vol) vol.124 
Number (no) no.374 
Page pp.19-24 
#Pages
Date of Issue 2025-02-11 (DC) 


[Return to Top Page]

[Return to IEICE Web Page]


The Institute of Electronics, Information and Communication Engineers (IEICE), Japan