AI Paper Atlas

Research map / Adversarial robustness and security / Certified and empirical robustness

On the Tradeoff Between Robustness and Fairness

Xinsong Ma, Zekai Wang, Weiwei Liu

NeurIPS 2022 · Adversarial robustness and security › Certified and empirical robustness · approach: Robustness-accuracy trade-off analysis · 1 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
increase AT perturbation radius eps_train from 4/255 to 16/255Madry AT, ResNet18 · CIFAR-10, FGSM eps_test=16/255average robust accuracy: 29.04 -> 42.94Improved
increase AT perturbation radius eps_train from 4/255 to 16/255Madry AT, ResNet18 · CIFAR-10, FGSM eps_test=16/255variance of class-wise robust accuracy: 1.90 -> 3.52Hurt
increase AT perturbation radius eps_train from 4/255 to 16/255Madry AT, ResNet50 · CIFAR-10, FGSM eps_test=16/255variance of class-wise robust accuracy: 1.24 -> 3.89Hurt
increase AT perturbation radius eps_train from 4/255 to 16/255Madry AT, WRN-32-10 · CIFAR-10, PGD-20 eps_test=16/255average robust accuracy: 5.15 -> 17.05Improved
increase eps_train from 4/255 to 16/255 on the cat classMadry AT, ResNet50 · CIFAR-10, FGSM eps_test=16/255class robust accuracy (cat): 12.0 -> 7.8Hurt
increase eps_train from 4/255 to 16/255 on the ship classMadry AT, ResNet50 · CIFAR-10, FGSM eps_test=16/255class robust accuracy (ship): 34.2 -> 62.4Improved
increase eps_train from 4/255 to 16/255 on the deer classMadry AT, ResNet50 · CIFAR-10, FGSM eps_test=16/255class robust accuracy (deer): 17.5 -> 13.7Hurt
add variance-of-class-wise-adversarial-risk regularizer (FAT)Madry AT, ResNet18, eps_train=16/255 · CIFAR-10, FGSM eps_test=16/255variance of class-wise robust accuracy: 3.52 -> 3.32Improved
add FAT regularizerMadry AT, ResNet18, eps_train=16/255 · CIFAR-10, PGD-20 eps_test=16/255variance of class-wise robust accuracy: 1.28 -> 1.06Improved
add FAT regularizerMadry AT, ResNet18, eps_train=16/255 · CIFAR-10, C&W eps_test=16/255variance of class-wise robust accuracy: 1.38 -> 1.22Improved
add FAT regularizerMadry AT, ResNet18, eps_train=12/255 · CIFAR-10, PGD-20 eps_test=20/255variance of class-wise robust accuracy: 0.35 -> 0.35No clear effect
add FAT regularizerMadry AT, ResNet18, eps_train=16/255 · CIFAR-10, FGSM eps_test=16/255average robust accuracy: 42.97 -> 43.40No clear effect

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy