S. Paul and P.-Y. Chen, “Vision Transformers Are Robust Learners”, AAAI, vol. 36, no. 2, pp. 2071–2081, Jun. 2022.