AI Paper Atlas

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

Meta Invariance Defense Towards Generalizable Robustness to Unknown Adversarial Attacks

Lei Zhang, Yuhang Zhou, Yi Yang, Xinbo Gao

TPAMI 2024 · Adversarial robustness and security › Certified and empirical robustness · approach: Attack evaluation and defense critique · 18 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 meta-learning over sampled attack combinations (ablation)MID (LeNet5) · MNIST BIM_Taccuracy: 88.96 -> 90.64Improved
add distillation framework to meta learning (ablation)MID (LeNet5) · MNIST BIM_Taccuracy: 90.64 -> 89.09Hurt
add adversarial consistency to meta+distillation (ablation)MID (LeNet5) · MNIST BIM_Taccuracy: 89.09 -> 97.07Improved
add cyclic consistency to meta+distillation (ablation)MID (LeNet5) · MNIST BIM_Taccuracy: 89.09 -> 97.99Improved
add both consistency constraints (ablation)MID (LeNet5) · MNIST BIM_Taccuracy: 89.09 -> 98.78Improved
add meta-learning (ablation)MID (ResNet18) · CIFAR10 FGSM_Naccuracy: 34.59 -> 37.17Improved
add meta-learning (ablation)MID (ResNet18) · CIFAR10 CW_Taccuracy: 41.81 -> 64.57Improved
add adversarial and cyclic consistency to meta+distillation (ablation)MID (ResNet18) · CIFAR10 FGSM_Naccuracy: 39.09 -> 41.63Improved
add both consistency constraints (ablation)MID (ResNet18) · CIFAR10 benignclean accuracy: 66.81 -> 64.48Hurt
use MID instead of adversarial trainingResNet18 · CIFAR10 black-box PGD_Naccuracy: 13.88 -> 64.84Improved
use MID instead of RSLADResNet18 · CIFAR10 black-box CW_Naccuracy: 68.28 -> 64.34Hurt
use MID instead of ATConvNext-B · Tiny-ImageNet SPAaccuracy: 39.26 -> 33.89Hurt

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