TY - JOUR
T1 - Global Clipper
T2 - 2024 IJCAI Workshop on Artificial Intelligence Safety, AISafety 2024
AU - Syed, Qutub
AU - Paulitsch, Michael
AU - Pattabiraman, Karthik
AU - Hagn, Korbinian
AU - Oboril, Fabian
AU - Buerkle, Cornelius
AU - Scholl, Kay Ulrich
AU - Hinz, Gereon
AU - Knoll, Alois
N1 - Publisher Copyright:
© 2024 Copyright for this paper by its authors.
PY - 2024
Y1 - 2024
N2 - As transformer-based object detection models progress, their impact in critical sectors like autonomous vehicles and aviation is expected to grow. Soft errors causing bit flips during inference have significantly impacted DNN performance, altering predictions. Traditional range restriction solutions for CNNs fall short for transformers. This study introduces the Global Clipper and Global Hybrid Clipper, effective mitigation strategies specifically designed for transformer-based models. It significantly enhances their resilience to soft errors and reduces faulty inferences to 0%. We also detail extensive testing across over 64 scenarios involving two transformer models (DINO-DETR and Lite-DETR) and two CNN models (YOLOv3 and SSD) using three datasets, totalling approximately 3.3 million inferences, to assess model robustness comprehensively. Moreover, the paper explores unique aspects of attention blocks in transformers and their operational differences from CNNs.
AB - As transformer-based object detection models progress, their impact in critical sectors like autonomous vehicles and aviation is expected to grow. Soft errors causing bit flips during inference have significantly impacted DNN performance, altering predictions. Traditional range restriction solutions for CNNs fall short for transformers. This study introduces the Global Clipper and Global Hybrid Clipper, effective mitigation strategies specifically designed for transformer-based models. It significantly enhances their resilience to soft errors and reduces faulty inferences to 0%. We also detail extensive testing across over 64 scenarios involving two transformer models (DINO-DETR and Lite-DETR) and two CNN models (YOLOv3 and SSD) using three datasets, totalling approximately 3.3 million inferences, to assess model robustness comprehensively. Moreover, the paper explores unique aspects of attention blocks in transformers and their operational differences from CNNs.
UR - http://www.scopus.com/inward/record.url?scp=85213406951&partnerID=8YFLogxK
M3 - Conference article
AN - SCOPUS:85213406951
SN - 1613-0073
VL - 3856
JO - CEUR Workshop Proceedings
JF - CEUR Workshop Proceedings
Y2 - 4 August 2024
ER -