Adversarially Self-Supervised Pre-training Improves Accuracy and Robustness

[PDF] arxiv.org

MIMIR: Masked Image Modeling for Mutual Information-based Adversarial Robustness

X Xu, S Yu, J Wu, S Picek - arXiv preprint arXiv:2312.04960, 2023 - arxiv.org
Vision Transformers (ViTs) achieve superior performance on various tasks compared to
convolutional neural networks (CNNs), but ViTs are also vulnerable to adversarial attacks …
相关文章 所有 2 个版本