English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Permanent faults induced due to imperfections in the manufacturing process ofDeep Neural Network (DNN) accelerators are a major concern, as they negativelyimpact the manufacturing yield of the chip fabrication process. Fault-awaretraining is the state-of-the-art approach for mitigating such faults. However,it incurs huge retraining overheads, specifically when used for large DNNstrained on complex datasets. To address this issue, we propose a novelFault-Aware Quantization (FAQ) technique for mitigating the effects of stuck-atpermanent faults in the on-chip weight memory of DNN accelerators at anegligible overhead cost compared to fault-aware retraining while offeringcomparable accuracy results. We propose a lookup table-based algorithm toachieve ultra-low model conversion time. We present extensive evaluation of theproposed approach using five different DNNs, i.e., ResNet-18, VGG11, VGG16,AlexNet and MobileNetV2, and three different datasets, i.e., CIFAR-10,CIFAR-100 and ImageNet. The results demonstrate that FAQ helps in maintainingthe baseline accuracy of the DNNs at low and moderate fault rates withoutinvolving costly fault-aware training. For example, for ResNet-18 trained onthe CIFAR-10 dataset, at 0.04 fault rate FAQ offers (on average) an increase of76.38% in accuracy. Similarly, for VGG11 trained on the CIFAR-10 dataset, at0.04 fault rate FAQ offers (on average) an increase of 70.47% in accuracy. Theresults also show that FAQ incurs negligible overheads, i.e., less than 5% ofthe time required to run 1 epoch of retraining. We additionally demonstrate theefficacy of our technique when used in conjunction with fault-aware retrainingand show that the use of FAQ inside fault-aware retraining enables fastaccuracy recovery.

FAQ: Mitigating the Impact of Faults in the Weight Memory of DNN Accelerators through Fault-Aware Quantization