AI Paper Atlas

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

Characterizing Model Robustness via Natural Input Gradients

Adrian Rodriguez-Munoz, Tongzhou Wang, Antonio Torralba

ECCV 2024 · Adversarial robustness and security › Certified and empirical robustness · approach: Robustness-accuracy trade-off analysis · 3 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
regularize input gradient norm on natural examples only instead of adversarial trainingSwin-B · ImageNet-1k AutoAttack eps=4/255AutoAttack accuracy: 0.00 -> 51.58 (adversarial training 56.12)Improved
replace ReLU with smooth GeLU in ResNet-50 for gradient norm regularization (ablation)ResNet-50 GradNorm · ImageNet-1k AutoAttackAutoAttack accuracy: 6.82 -> 30.00Improved
replace ReLU with GeLU in ResNet-50 for adversarial training (ablation)ResNet-50 adversarial training · ImageNet-1k AutoAttackAutoAttack accuracy: 31.60 -> 32.64Improved
increase GradNorm weight from 1.2 to 1.4 (ablation)Swin-B GradNorm · ImageNet-1k AutoAttack eps=4/255AutoAttack accuracy: 51.58 -> 52.48 AA; clean 77.78 -> 76.20Mixed
double the GradNorm weight on the green channel (ablation)Swin-B GradNorm · ImageNet-1kAutoAttack accuracy: 51.66 -> 51.38Hurt
add edge regularization aligning class gradients with image edgesSwin-B · ImageNet-1k AutoAttack eps=4/255AutoAttack accuracy: 0.00 -> 35.02Improved
regularize saliency maps or loss gradients toward edges instead of class gradients (ablation)Edge regularization · ResNet50+GeLU ImageNetrobustnessHurt
add adversarial-noise inputs to gradient norm regularization (ablation)Swin-B GradNorm · ImageNet-1k AutoAttack eps=4/255AutoAttack accuracy: 51.66 -> 56.49Improved
add Gaussian noise inputs to gradient norm regularization (ablation)Swin-B GradNorm · ImageNet-1k AutoAttack eps=4/255AutoAttack accuracy: 51.66 -> 52.04Improved
gradient norm regularization instead of adversarial training on CIFAR-10PreActResNet-18 · CIFAR-10 eps=8/255AutoAttack accuracy: AA 42.05 -> 39.60; clean 69.81 -> 77.74Mixed

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