Can Attention Masks Improve Adversarial Robustness?

Publication
The AAAI-20 Workshop on Engineering Dependable and Secure Machine Learning Systems (EDSMLS)