Fault-Criticality Assessment for AI Accelerators using Graph Convolutional Networks

Owing to the inherent fault tolerance of deep neural networks (DNNs), many structural faults in DNN accelerators tend to be functionally benign. In order to identify functionally critical faults, we analyze the functional impact of stuck-at faults in the processing elements of a 128×128 systolic-arr...

Full description

Saved in:
Bibliographic Details
Published in2021 Design, Automation & Test in Europe Conference & Exhibition (DATE) pp. 1596 - 1599
Main Authors Chaudhuri, Arjun, Talukdar, Jonti, Jung, Jinwook, Nam, Gi-Joon, Chakrabarty, Krishnendu
Format Conference Proceeding
LanguageEnglish
Published EDAA 01.02.2021
Subjects
AI accelerators
Fault diagnosis
Fault tolerance
Fault tolerant systems
Machine learning
Neural networks
Training
Online AccessGet full text

Cover

More Information
Summary:Owing to the inherent fault tolerance of deep neural networks (DNNs), many structural faults in DNN accelerators tend to be functionally benign. In order to identify functionally critical faults, we analyze the functional impact of stuck-at faults in the processing elements of a 128×128 systolic-array accelerator that performs inferencing on the MNIST dataset. We present a 2-tier machine-learning framework that leverages graph convolutional networks (GCNs) for quick assessment of the functional criticality of structural faults. We describe a computationally efficient methodology for data sampling and feature engineering to train the GCN-based framework. The proposed framework achieves up to 90% classification accuracy with negligible misclassification of critical faults.
ISSN:1558-1101
DOI:10.23919/DATE51398.2021.9474128