FAT-RABBIT: Fault-Aware Training towards Robustness AgainstBit-flip Based Attacks in Deep Neural Networks

• Published in: Proceedings - International Test Conference pp. 106 - 110
• Main Authors: Pourmehrani, Hossein, Bahrami, Javad, Nooralinejad, Parsa, Pirsiavash, Hamed, Karimi, Naghmeh
• Format: Conference Proceeding
• Language: English
• Published: IEEE 03.11.2024
• Subjects: Accuracy; Computational modeling; Deep learning; fault aware training; fault injection attacks; Force; Hardware; machine learning accelerator; Optimization; Perturbation methods; Robustness; Sensitivity; Training

Machine learning and in particular deep learning is used in a broad range of crucial applications. Implementing such models in custom hardware can be highly beneficial thanks to their low power and computation latency compared to GPUs. However, an error in their output can lead to disastrous outcome...