Month: August 2019
A Discussion of ‘Adversarial Examples Are Not Bugs, They Are Features’
On May 6th, Andrew Ilyas and colleagues published a paper outlining two sets of experiments. Firstly, they showed that models trained on adversarial examples ...
A Discussion of ‘Adversarial Examples Are Not Bugs, They Are Features’: Discussion and Author Responses
We want to thank all the commenters for the discussion and for spending time designing experiments analyzing, replicating, and expanding upon our results. These ...
Adversarial Example Researchers Need to Expand What is Meant by ‘Robustness’
The hypothesis in Ilyas et. al. is a special case of a more general principle that is well accepted in the distributional robustness literature — models ...
A Discussion of ‘Adversarial Examples Are Not Bugs, They Are Features’: Robust Feature Leakage
Ilyas et al. report a surprising result: a model trained on adversarial examples is effective on clean data. They suggest this transfer is driven ...
Two Examples of Useful, Non-Robust Features
A Discussion of ‘Adversarial Examples Are Not Bugs, They Are Features’: Two Examples of Useful, Non-Robust Features Ilyas et al. define a feature as ...
Adversarially Robust Neural Style Transfer
A figure in Ilyas, et. al. that struck me as particularly interesting was the following graph showing a correlation between adversarial transferability between architectures ...
Adversarial Examples are Just Bugs, Too
We demonstrate that there exist adversarial examples which are just “bugs”: aberrations in the classifier that are not intrinsic properties of the data distribution. ...
Learning from Incorrectly Labeled Data
Section 3.2 of Ilyas et al. (2019) shows that training a model on only adversarial errors leads to non-trivial generalization on the original test ...
