Soft Error Reliability Analysis of Vision Transformers

arXiv:2302.10468v3 Announce Type: replace
Abstract: Vision Transformers (ViTs) that leverage self-attention mechanism have shown superior performance on many classical vision tasks compared to convolutional neural networks (CNNs) and gain increasing popularity recently. Existing ViTs works mainly optimize performance and accuracy, but ViTs reliability issues induced by soft errors in large-scale VLSI designs have generally been overlooked. In this work, we mainly study the reliability of ViTs and investigate the vulnerability from different architecture granularities ranging from models, layers, modules, and …

accuracy analysis arxiv attention cnns convolutional neural networks cs.cr error errors large mechanism networks neural networks performance reliability scale transformers