Special Session: Approximation and Fault Resiliency of DNN AcceleratorsShow others and affiliations
2023 (English)In: Proceedings of the IEEE VLSI Test Symposium, IEEE Computer Society , 2023, Vol. AprilConference paper, Published paper (Refereed)
Abstract [en]
Deep Learning, and in particular, Deep Neural Network (DNN) is nowadays widely used in many scenarios, including safety-critical applications such as autonomous driving. In this context, besides energy efficiency and performance, reliability plays a crucial role since a system failure can jeopardize human life. As with any other device, the reliability of hardware architectures running DNNs has to be evaluated, usually through costly fault injection campaigns. This paper explores approximation and fault resiliency of DNN accelerators. We propose to use approximate (AxC) arithmetic circuits to agilely emulate errors in hardware without performing fault injection on the DNN. To allow fast evaluation of AxC DNN, we developed an efficient GPU-based simulation framework. Further, we propose a fine-grain analysis of fault resiliency by examining fault propagation and masking in networks.
Place, publisher, year, edition, pages
IEEE Computer Society , 2023. Vol. April
Keywords [en]
approximate computing, deep neural networks, fault emulation, reliability, resiliency assessment, Backpropagation, Energy efficiency, Safety engineering, Software testing, Autonomous driving, Efficiency and performance, Fault emulations, Fault injection, Human lives, Performance reliability, Safety critical applications, System failures
National Category
Computer Systems
Identifiers
URN: urn:nbn:se:mdh:diva-63673DOI: 10.1109/VTS56346.2023.10140043ISI: 01011806600022Scopus ID: 2-s2.0-85161889760ISBN: 9798350346305 (print)OAI: oai:DiVA.org:mdh-63673DiVA, id: diva2:1776816
Conference
41st IEEE VLSI Test Symposium, VTS 2023, 24 April 2023 through 26 April 2023, San Diego, USA
2023-06-282023-06-282023-07-26Bibliographically approved