An Approximate Fault-Tolerance Design for a Convolutional Neural Network Accelerator

Today, various domain-specific convolutional neural network (CNN) accelerators are deployed in large-scale systems to satisfy the massive computational demands of current deep CNNs. Although bringing significant performance improvements, the highly integrated CNN accelerators are more susceptible to...

Ausführliche Beschreibung

Bibliographische Detailangaben

Veröffentlicht in:	IT professional 2023-07, Vol.25 (4), p.85-90
Hauptverfasser:	Wei, Wenda, Wang, Chenyang, Zheng, Xinyang, Yue, Hengshan
Format:	Artikel
Sprache:	eng
Schlagworte:	Accelerators Aging Artificial neural networks Convolutional neural networks Error detection Fault tolerance Hardware acceleration Neural networks Philosophical considerations Redundancy Reliability engineering Systolic arrays Termination of employment Terminations
Online-Zugang:	Volltext bestellen
Tags:	Tag hinzufügen Keine Tags, Fügen Sie den ersten Tag hinzu!