AI Paper Atlas

Research map / Adversarial robustness and security / Black-box and transfer attacks

Adversarial Defense via Learning to Generate Diverse Attacks

Yunseok Jang, Tianchen Zhao, Seunghoon Hong, Honglak Lee

ICCV 2019 · Adversarial robustness and security › Black-box and transfer attacks · approach: Robustness benchmarks and defenses · 114 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 recursive generation of perturbations (ablation)L2L-DA base · MNIST white-boxclassification accuracy: +7.85%pImproved
add diversity loss (single-step generation) (ablation)L2L-DA base · MNIST white-boxMin accuracy: +4.33%pImproved
add diversity loss on top of recursive generation (ablation)L2L-DA with recursion · MNIST white-boxMin accuracy: 93.97 -> 95.15 (+1.18%p)Improved
add noise input delta to the generator (ablation)L2L-DA full · MNIST white-boxMin accuracy: 92.46 -> 95.15 (+2.69%p)Improved
add class label y with class-conditional batch norm (ablation)L2L-DA full · MNIST white-boxMin accuracy: 93.44 -> 95.15 (+1.71%p)Improved
add gradient normalization to generator input (ablation)L2L-DA full · CIFAR-10 white-boxPGD10 accuracy: about 1% without itImproved
use gradient normalization, class label and noise inputs (L2L-DA base) instead of L2L* (ablation)generator-based adversarial training · MNIST white-boxMin accuracy: 0.02 -> 86.12Improved
adversarial training with L2L-DA instead of best baselineLeNet-5 · MNIST white-boxMin accuracy: +5.68%pImproved
adversarial training with L2L-DA instead of best baselineResNet-20 · CIFAR-10 white-boxMin accuracy: +1.60%pImproved
train with L2L-DA instead of PGD10LeNet-5 · MNIST black-boxMin accuracy: 94.13 -> 95.08Improved
adversarial training on PGD10 onlyclassifier · CIFAR-10 white-boxMin accuracy vs unseen attacks: 37.37 vs 0.32 MinHurt

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