[New Model] <xxy2024mimir>

[xiaoyunxxy]

Paper Information

• Paper Title: MIMIR: Masked Image Modeling for Mutual Information-based Adversarial Robustness
• Paper URL: https://arxiv.org/abs/2312.04960
• Paper authors: Xiaoyun Xu, Shujian Yu, Zhuoran Liu, Stjepan Picek

Leaderboard Claim(s)

Model 1

• Architecture: Swin-base
• Dataset: ImageNet-1K
• Threat Model: Linf
• eps: 4/255
• Clean accuracy: 76.62
• Robust accuracy: 55.90
• Additional data: false
• Evaluation method: AutoAttack
• Checkpoint and code: checkpoint code log.

Model 2

• Architecture: Swin-large
• Dataset: ImageNet-1K
• Threat Model: Linf
• eps: 4/255
• Clean accuracy: 78.62
• Robust accuracy: 59.68
• Additional data: false
• Evaluation method: AutoAttack
• Checkpoint and code: checkpoint code log

The pre-process of ImageNet-1K for evaluation can be found here, link

Model Zoo:

• I want to add my models to the Model Zoo (check if true)
• I use an architecture that is included among
  those here or in timm. If not, I added the link to the architecture implementation so that it can be added.
• I agree to release my model(s) under MIT license (check if true) OR under a custom license, located here: (put the custom license URL here if a custom license is needed. If no URL is specified, we assume that you are fine with MIT)

[xiaoyunxxy]

Hi,

It has been about one month since we submitted our models. I'd like to kindly ask if you have the chance to review our models. We would appreciate your feedback.

In addition, please use the code provided in the link for Swin architectures. The link will directly go to the code for Swin-base and Swin-large. Because our hyper-parameter may be slightly different from the timm.

Best

[fra31]

Hi,

thanks for the submission, and sorry for the delay. I've added the models with 5ff2500. Let me know if it looks fine.

[xiaoyunxxy]

Hi,

It looks fine. Thank you very much.

[fra31]

Leaderboard updated with RobustBench/robustbench.github.io#16.