Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Reporting success rate of unbounded attacks is meaningless #10

Open
carlini opened this issue Feb 26, 2019 · 2 comments
Open

Reporting success rate of unbounded attacks is meaningless #10

carlini opened this issue Feb 26, 2019 · 2 comments

Comments

@carlini
Copy link

carlini commented Feb 26, 2019

Three of the attacks presented (EAD, CW2, and BLB) are unbounded attacks: rather than finding the “worst-case” (i.e., highest loss) example within some distortion bound, they seek to find the closest input subject to the constraint that it is misclassified. Unbounded attacks should always reach 100% “success” eventually, if only by actually changing an image from one class into an image from the other class; the correct and meaningful metric to report for unbounded attacks is the distortion required.

@ryderling
Copy link
Owner

Three of the attacks presented (EAD, CW2, and BLB) are unbounded attacks: rather than finding the “worst-case” (i.e., highest loss) example within some distortion bound, they seek to find the closest input subject to the constraint that it is misclassified. Unbounded attacks should always reach 100% “success” eventually, if only by actually changing an image from one class into an image from the other class; the correct and meaningful metric to report for unbounded attacks is the distortion required.

We definitely agree with you that only reporting the success rate of attacks is not as useful. Actually, it is the initial motivation for us to evaluate other performance metrics of attacks such as imperceptibility, robustness, and computation cost in our paper. And also, we reported the L0, L2, and L\inf distortion performance of EAD, CW2, and BLB in Table III.

@carlini
Copy link
Author

carlini commented Mar 16, 2019

So definitely it's good that you do report it somewhere, but nevertheless it's not meaningful to talk the success rate of unbounded attacks. Again, you may want to read https://arxiv.org/abs/1902.06705.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

2 participants