Compilation as a Defense: Enhancing DL Model Attack Robustness via Tensor Optimization

Adversarial Machine Learning (AML) investigates attacks on Deep Learning (DL) models and their underlying hardware/software assets.