Closed ghost closed 2 years ago
Dear @AllenPeng-libra,
I did not release the source code since I employed adversarial attacks and pruned models available at other GitHub repositories. However, I have attached (CodePaperICCV.zip) examples employed throughout the experiments, for which you can reproduce the main findings of our paper.
Importantly, on the semantic-preserving attack, you need to download the corrupted images (CIFAR-C) at this link: https://zenodo.org/record/2535967#.YEPqamhKjIU After downloading the data, unzip them and set the "--corruption_path" (line 17 in semantic.py) by the path where the .npy files are.
I hope this helps your research. Best Regards, Artur Jordão.
Sorry, no code was found in the zip file.
Dear @AllenPeng-libra,
Try this file CodePaperICCV.zip
Dear Artur, Hi, I have read your paper "On the Effect of Pruning on Adversarial Robustness". I am very interested in this article and believe it will help a lot with my recent research. But I do not find it here, could you please release the code.