TensorFI+: A Scalable Fault Injection Framework for Modern Deep Learning Neural Networks

Laskar, Sabuj; Rahman, Md. Hasanur; Li, Guanpeng

doi:10.1109/issrew55968.2022.00074

Cited by 2 publications

(2 citation statements)

References 19 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In our previous study, we proposed TENSORFI+ [48] to work with both sequential and non-sequential DNN models. As we can see, sequential models have only one input and one output for every intermediate layer.…”

Section: Support For Non-sequential Dnn Modelsmentioning

confidence: 99%

See 1 more Smart Citation

Investigating the impact of transient hardware faults on deep learning neural network inference

Rahman,

Laskar,

2024

Software Testing Verif & Rel

Self Cite

View full text Add to dashboard Cite

SummarySafety‐critical applications, such as autonomous vehicles, healthcare, and space applications, have witnessed widespread deployment of deep neural networks (DNNs). Inherent algorithmic inaccuracies have consistently been a prevalent cause of misclassifications, even in modern DNNs. Simultaneously, with an ongoing effort to minimize the footprint of contemporary chip design, there is a continual rise in the likelihood of transient hardware faults in deployed DNN models. Consequently, researchers have wondered the extent to which these faults contribute to DNN misclassifications compared to algorithmic inaccuracies. This article delves into the impact of DNN misclassifications caused by transient hardware faults and intrinsic algorithmic inaccuracies in safety‐critical applications. Initially, we enhance a cutting‐edge fault injector, TensorFI, for TensorFlow applications to facilitate fault injections on modern DNN non‐sequential models in a scalable manner. Subsequently, we analyse the DNN‐inferred outcomes based on our defined safety‐critical metrics. Finally, we conduct extensive fault injection experiments and a comprehensive analysis to achieve the following objectives: (1) investigate the impact of different target class groupings on DNN failures and (2) pinpoint the most vulnerable bit locations within tensors, as well as DNN layers accountable for the majority of safety‐critical misclassifications. Our findings regarding different grouping formations reveal that failures induced by transient hardware faults can have a substantially greater impact (with a probability up to 4 higher) on safety‐critical applications compared to those resulting from algorithmic inaccuracies. Additionally, our investigation demonstrates that higher order bit positions in tensors, as well as initial and final layers of DNNs, necessitate prioritized protection compared to other regions.

show abstract

Section: Support For Non-sequential Dnn Modelsmentioning

confidence: 99%

“…Then we discuss its limitation in supporting non‐sequential DNN models. Finally, we propose our method to design TensorFI+ [48] for injecting transient hardware faults into non‐sequential models.…”

Section: Improvement Of Tensorfi Fault Injection Frameworkmentioning

confidence: 99%