[NewModel] Erichson2022NoisyMix

[erichson]

Paper Information

• Paper Title: NoisyMix: Boosting Robustness by Combining Data Augmentations, Stability Training, and Noise Injections:
• Paper URL: https://arxiv.org/pdf/2202.01263.pdf:
• Paper authors: N. Benjamin Erichson, Soon Hoe Lim, Francisco Utrera, Winnie Xu, Ziang Cao, and Michael W. Mahoney:

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: ResNet-50
• Threat Model: Common Corruptions
• eps: N/A
• Clean accuracy: 0.7714
• Robust accuracy: 0.5225493333333333
• Additional data: false
• Evaluation method: ImageNet-C
• Checkpoint and code: https://github.com/erichson/NoisyMix.

Model 2

• Architecture: Wide-ResNet-28x4
• Threat Model: Common Corruptions
• eps: N/A
• Clean accuracy: 0.8116
• Robust accuracy: 0.7206066666666667
• Additional data: false
• Evaluation method: CIFAR-100-C
• Checkpoint and code: https://github.com/erichson/NoisyMix.

Model 2

• Architecture: Wide-ResNet-28x4
• Threat Model: Common Corruptions
• eps: N/A
• Clean accuracy: 0.9673
• Robust accuracy: 0.9277826666666665
• Additional data: false
• Evaluation method: CIFAR-10-C
• Checkpoint and code: https://github.com/erichson/NoisyMix.

Model Zoo:

• [] 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).
• [] 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/70, please let me know if it looks fine.

[max-andr]

seems like there are no concerns with this issue. feel free to reopen if there are some problems!