RRAM Crossbar-Based Fault-Tolerant Binary Neural Networks (BNNs)

Conference Paper (2022)
Author(s)

Anteneh Gebregiorgis (TU Delft - Computer Engineering)

Artemis Zografou (Student TU Delft)

Said Hamdioui (TU Delft - Quantum & Computer Engineering)

Research Group
Computer Engineering
Copyright
© 2022 A.B. Gebregiorgis, Artemis Zografou, S. Hamdioui
DOI related publication
https://doi.org/10.1109/ETS54262.2022.9810414
More Info
expand_more
Publication Year
2022
Language
English
Copyright
© 2022 A.B. Gebregiorgis, Artemis Zografou, S. Hamdioui
Research Group
Computer Engineering
Pages (from-to)
1-2
ISBN (print)
978-1-6654-6707-0
ISBN (electronic)
978-1-6654-6706-3
Reuse Rights

Other than for strictly personal use, it is not permitted to download, forward or distribute the text or part of it, without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license such as Creative Commons.

Abstract

Computation-In Memory (CIM) using RRAM crossbar array is a promising solution to realize energy-efficient neuromorphic hardware, such as Binary Neural Networks (BNNs). However, RRAM faults restrict the applicability of CIM for BNN implementation. To address this issue, we propose a fault tolerance framework to mitigate the impact of RRAM faults on the accuracy of CIM-based BNN hardware. Evaluation results using MNIST, Fashion-MNIST and CIFAR-10 datasets demonstrate that the proposed framework outperforms the related works as it restores more than 99% of the RRAM fault induced accuracy reduction with relatively less overhead.

Files

RRAM_Crossbar_Based_Fault_Tole... (pdf)
(pdf | 1.3 Mb)
- Embargo expired in 01-07-2023
License info not available