AI Paper Atlas

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

Catastrophic Overfitting: A Potential Blessing in Disguise

MN Zhao, Lihe Zhang, Yuqiu Kong, Baocai Yin

ECCV 2024 · Adversarial robustness and security › Certified and empirical robustness · approach: Adversarial training refinements · 2 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
add feature-activation-difference suppression loss at node M_B (alpha3 200) (ablation)FGSM-MEP fast adversarial training · CIFAR-10 ResNet18, train 12/255clean accuracy / AA: stable at 74.08 clean, 19.72 AAImproved
suppress feature differences at node A, C, D or E instead of B (ablation)FGSM-MEP · CIFAR-10 ResNet18, train 12/255CO avoidance: CO persists at nodes A, C, DNo clear effect
use random uniform noise at inference on CO-affected modelCO-affected ResNet18 · CIFAR-10 train 16/255AutoAttack accuracy (%): PGD-50 58.48, AA 43.53 with clean 88.45Improved
add input noise at inference to a clean-trained model (ablation)ResNet18 clean-trained · CIFAR-10PGD-50 accuracy (%): 0.00 -> 0.00No clear effect
add input noise at inference to a clean-trained model (ablation)ResNet18 clean-trained · CIFAR-10clean accuracy (%): 94.40 -> 80.63Hurt
add input noise at inference to CO modelCO-affected ResNet18 · CIFAR-10 clean accuracyclean accuracy (%): 89.81 -> 89.64No clear effect
use uniform noise instead of Gaussian noise at inference (ablation)CO-affected model · CIFAR-10 ResNet18AutoAttack accuracy (%): 41.64 -> 43.53Improved
increase uniform noise range from 10/255 to 16/255 (ablation)CO-affected model · CIFAR-10 ResNet18AutoAttack accuracy (%): 23.27 -> 43.53Improved
train with noise-augmented data instead of relying on CO (ablation)noise-augmented models · CIFAR-10 ResNet18adversarial robustnessHurt
induce CO with L_co on top 10 percent channels (ablation)FGSM-MEP train 8/255 · CIFAR-10 ResNet18CO occurrenceImproved
use proposed feature-difference regularizer instead of ConvergeSmooth at 12/255stable FAT · CIFAR-10 ResNet18AutoAttack accuracy (%): 19.11 -> 22.08Improved
compare high-variance channel masking against feature-difference masking (ablation)CO-affected model · CIFAR-10 ResNet18clean and adversarial accuracyImproved

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