中国科学院机构知识库网格
Chinese Academy of Sciences Institutional Repositories Grid
首页 机构 成果 学者
  
  1. CAS IR Grid
  2. 计算技术研究所
  3. 中国科学院计算技术研究所
  4. 中国科学院计算技术研究所期刊论文
On-Line Fault Protection for ReRAM-Based Neural Networks

文献类型：期刊论文

作者Li, Wen2,3; Wang, Ying2,3; Liu, Cheng2,3; He, Yintao2,3; Liu, Lian2,3; Li, Huawei1,3; Li, Xiaowei2,3
刊名IEEE TRANSACTIONS ON COMPUTERS
出版日期2023-02-01
卷号72期号:2页码:423-437
关键词Training Fault detection Computational modeling Image edge detection Memristors Neural networks Kernel Deep neural network hard fault ReRAM reliability soft fault
ISSN号0018-9340
DOI10.1109/TC.2022.3160345
英文摘要The emerging Resistive RAM (ReRAM) technology significantly boosts the performance and the energy efficiency of the deep learning accelerators (DLAs) via the Computing-in-Memory (CiM) architecture. However, ReRAM-based DLA also suffers a high occurrence rate of memory faults. How to detect and protect against the faults in ReRAM devices poses great challenges to ReRAM-based DLA design. In this work, we propose RRAMedy, an in-situ fault detection and network remedy framework for ReRAM-based DLAs. With the proposed Adversarial Example Testing, which is a lifetime on-device and on-line fault detection technique, it achieves high detection coverage of both hard faults and soft faults at a low run-time cost. In addition, it employs an edge-cloud collaborative model retraining method to tolerate the detected faults by leveraging the inherent fault-adaptive capability of DNNs. Meanwhile, to enable in-situ model remedy when the cloud assistance is absent due to security or overhead issues, we propose to accelerate the fault-masking retraining process on edge devices with parallelized Knowledge Transfer. Our experimental results show that the proposed fault detection technique achieves high fault detection accuracy and delivers real-time testing performance. Meanwhile, the proposed retraining approach greatly alleviates the accuracy degradation problem and achieves excellent performance speedups over the baselines.
资助项目National Key Research and Development Program of China[2020YFB1600201] ; National Natural Science Foundation of China (NSFC)[62090024] ; National Natural Science Foundation of China (NSFC)[61874124] ; National Natural Science Foundation of China (NSFC)[61876173] ; Zhejiang Lab[2021PC0AC01]
WOS研究方向Computer Science ; Engineering
语种英语
WOS记录号WOS:000917782600010
出版者IEEE COMPUTER SOC
源URL[http://119.78.100.204/handle/2XEOYT63/19935]  
专题中国科学院计算技术研究所期刊论文
通讯作者Wang, Ying
作者单位1.Peng Cheng Lab, Shenzhen 518066, Peoples R China
2.Univ Chinese Acad Sci, Sch Comp & Control Engn, Beijing 100049, Peoples R China
3.Chinese Acad Sci, Inst Comp Technol, SKLCA, Beijing 100190, Peoples R China
推荐引用方式
GB/T 7714		 Li, Wen,Wang, Ying,Liu, Cheng,et al. On-Line Fault Protection for ReRAM-Based Neural Networks[J]. IEEE TRANSACTIONS ON COMPUTERS,2023,72(2):423-437.
APA Li, Wen.,Wang, Ying.,Liu, Cheng.,He, Yintao.,Liu, Lian.,...&Li, Xiaowei.(2023).On-Line Fault Protection for ReRAM-Based Neural Networks.IEEE TRANSACTIONS ON COMPUTERS,72(2),423-437.
MLA Li, Wen,et al."On-Line Fault Protection for ReRAM-Based Neural Networks".IEEE TRANSACTIONS ON COMPUTERS 72.2(2023):423-437.

入库方式： OAI收割

来源：计算技术研究所

浏览0
下载0
收藏0
其他版本

除非特别说明，本系统中所有内容都受版权保护，并保留所有权利。