Global Clipper: Enhancing Safety and Reliability of Transformer-based Object Detection Models
This addresses safety and reliability issues for transformer-based object detection in critical sectors, representing an incremental improvement over existing methods for CNNs.
The study tackled the problem of soft errors causing bit flips in transformer-based object detection models, which can alter predictions in critical applications like autonomous vehicles, and introduced the Global Clipper and Global Hybrid Clipper strategies to enhance resilience, reducing faulty inferences to approximately 0%.
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.