Reconfigurable Mapping Algorithm based Stuck-At-Fault Mitigation in Neuromorphic Computing Systems

Reconfigurable Mapping Algorithm based Stuck-At-Fault Mitigation in Neuromorphic Computing Systems
复制标题

神经形态计算系统中基于可重构映射算法的卡故障缓解

DOI:
10.1145/3583781.3590208
复制
发表时间:
2023
期刊:
GLSVLSI '23: Proceedings of the Great Lakes Symposium on VLSI 2023
影响因子:
--
通讯作者:
Wang, Jinhui
Wang, Jinhui
中科院分区:
--
文献类型:
--
作者:
Oli-Uz-Zaman, Md.;Khan, Saleh Ahmad;Oswald, William;Liao, Zhiheng;Wang, Jinhui

文献摘要

参考文献

被引文献

相似文献

不成熟的制造和大量的器件使用所产生的忆阻器的故障(SAF)缺陷使得神经形态计算系统无法商业化。为了解决这个问题,本文提出了一种可重构映射算法(RMA)。基于CIFAR 10数据集对VGG 8模型的分析,实验结果表明,当SA 1:SA 0 = 5:1,1:5和1:1时,RMA在SAFs从0.1%到50%的范围内,能够有效地恢复90%以上的推理准确率(无SAF时的原始准确率)。此外,RMA在高非线性LTP = 4和LTD = -4的情况下将精度提高了50%以上,并且标准电导漂移(85摄氏度下10年)几乎对具有RMA的DNN的推理精度没有影响。
Stuck-At-Fault (SAF) defect of memristor generated from immature fabrication and heavy device utilization makes neuromorphic computing systems commercially unavailable. To mitigate this problem, a Reconfigurable Mapping Algorithm (RMA) is proposed in this paper. Based on the analysis for the VGG8 model with CIFAR10 dataset, the experiment results show that the RMA is efficient in restoring the inference accuracy up to 90% (the original accuracy without SAF) under SAFs from 0.1% to 50%, where Stuck-At-One (SA1): Stuck-At-Zero (SA0) = 5:1, 1:5, and 1:1. Additionally, the RMA improves the accuracy more than 50% in presence of high nonlinearity LTP = 4 and LTD = -4 and the standard conductance drift (10 years at 85 degrees Celsius) nearly has no influence on the inference accuracy of the DNN with the RMA.
使用不准确的 RRAM 器件进行准确推理:联合算法设计解决方案
DOI: 10.1109/jxcdc.2020.2987605
发表时间: 2020
影响因子: 2.4
作者:
Charan, Gouranga;Mohanty, Abinash;Du, Xiaocong;Krishnan, Gokul;Joshi, Rajiv V.;Cao, Yu
通讯作者: Cao, Yu
基于 ReRAM 的深度学习加速器的无成本和无数据集卡住故障缓解
DOI: --
发表时间: 2021
期刊: Design, Automation and Test in Europe
影响因子: --
作者:
Giju Jung;M. Fouda;Sugil Lee;Jongeun Lee;A. Eltawil;F. Kurdahi
通讯作者: F. Kurdahi
改善基于忆阻器的 ANN 在边缘计算中的性能
DOI: 10.1109/tc.2021.3081985
发表时间: 2021
影响因子: 3.7
作者:
Liao, Zhiheng;Fu, Jingyan;Wang, Jinhui
通讯作者: Wang, Jinhui