AI Paper Atlas

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

Attributional Robustness Training using Input-Gradient Spatial Alignment

Mayank Singh, Nupur Kumari, Puneet Mangla, Abhishek Sinha, Vineeth N Balasubramanian, Balaji Krishnamurthy

ECCV 2020 · Adversarial robustness and security › Certified and empirical robustness · approach: Decision-boundary geometry of robustness · 21 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
use ART instead of PGD-10 adversarial trainingResNet WRN-28-10 · CIFAR-10Top-k intersection: 69.00 -> 92.90Improved
use ART instead of PGD-7 adversarial trainingWRN-40-2 · SVHNTop-k intersection: 39.67 -> 61.37Improved
use ART instead of PGD-7WRN-28-10 · FlowerTop-k intersection: 80.84 -> 79.84Hurt
use ART instead of PGD-7WRN-28-10 · FlowerPGD-40 adversarial accuracy: 69.85 -> 33.08Hurt
use ART instead of PGD-10WRN-28-10 · CIFAR-10PGD-40 adversarial accuracy: 44.07 -> 37.58Hurt
replace cosine distance with l2 distance in L_attr (ablation)ART · CIFAR-10Top-k intersection: 91.25 -> 68.41Hurt
add triplet loss with negative sample to cosine distance (ablation)ART · CIFAR-10Top-k intersection: 91.25 -> 92.90Improved
choose negative as second most likely class instead of argmin logit (ablation)ART · CIFAR-10Kendall correlation: 83.32 -> 91.76Improved
apply ART instead of PGD-10 for ResNet50 localizationResNet50 · CUBGT-Known Loc (Grad): 66.73 -> 82.65Improved
apply ART instead of natural training for localizationVGG-GAP · CUBGT-Known Loc (Grad): 72.54 -> 76.50Improved
ART instead of PGD-10 under targeted attribution attackART · CIFAR-10Kendall tau: 36.29 -> 64.76Improved
raise lambda beyond 0.75 (ablation)ART · CIFAR-10test accuracyHurt

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