[New Model] <Pang2022Robustness>

[P2333]

Paper Information

• Paper Title: Robustness and Accuracy Could Be Reconcilable by (Proper) Definition
• Paper URL: https://arxiv.org/pdf/2202.10103.pdf
• Paper authors: Tianyu Pang, Min Lin, Xiao Yang, Jun Zhu, Shuicheng Yan

Leaderboard Claim(s)

Add here the claim for your model(s). Copy and paste the following subsection for the number of models you want to add.

Model 1

• Architecture: WRN-28-10
• Dataset: CIFAR-10
• Threat Model: Linf
• eps: 8/255
• Clean accuracy: 88.61
• Robust accuracy: 61.04
• Additional data: false
• Evaluation method: AutoAttack
• Checkpoint and code: https://github.com/P2333/SCORE#pretrained-checkpoints

Model 2

• Architecture: WRN-70-16
• Dataset: CIFAR-10
• Threat Model: Linf
• eps: 8/255
• Clean accuracy: 89.01
• Robust accuracy: 63.35
• Additional data: false
• Evaluation method: AutoAttack
• Checkpoint and code: https://github.com/P2333/SCORE#pretrained-checkpoints

Model 3

• Architecture: WRN-28-10
• Dataset: CIFAR-100
• Threat Model: Linf
• eps: 8/255
• Clean accuracy: 63.66
• Robust accuracy: 31.08
• Additional data: false
• Evaluation method: AutoAttack
• Checkpoint and code: https://github.com/P2333/SCORE#pretrained-checkpoints

Model 4

• Architecture: WRN-70-16
• Dataset: CIFAR-100
• Threat Model: Linf
• eps: 8/255
• Clean accuracy: 65.56
• Robust accuracy: 33.05
• Additional data: false
• Evaluation method: AutoAttack
• Checkpoint and code: https://github.com/P2333/SCORE#pretrained-checkpoints

Model Zoo:

• [x] I want to add my models to the Model Zoo (check if true)
• [ ] I use an architecture that is not included among those here ( check if true).
• [x] I agree to release my model(s) under MIT license (check if true) OR
• [ ] I want my models to be released under a custom license, located here: (custom license URL here)

[fra31]

Hi,

thanks for the submission! We'll add the models as soon as possible.

[fra31]

Added with https://github.com/RobustBench/robustbench/pull/77, hope it looks good to you.

[P2333]

Hi Francesco,

The update looks good! Thank you!

[P2333]

Hi,

This paper has been accepted to ICML 2022. Could you please update in the venue column? Thank you!

[max-andr]

done (both here and on the website). congrats on the acceptance and i guess see you in Baltimore! :)

[P2333]

Thank you! See you in Baltimore! :)