AI Paper Atlas

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

Diversity can be Transferred: Output Diversification for White- and Black-box Attacks

Yusuke Tashiro, Yang Song, Stefano Ermon

NeurIPS 2020 · Adversarial robustness and security › Black-box and transfer attacks · approach: Query-efficient black-box attacks · 52 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
Initialize PGD with 2-step ODI instead of naive uniform restartsPGD · CIFAR-10accuracy: 46.06\pm 0.02\% -> 44.45\pm 0.02\%Improved
Initialize C&W with ODI instead of Gaussian initializationC&W · ImageNetaverage minimum ℓ2 perturbation: 1.58\pm0.00 -> 1.32\pm0.01Improved
Use tuned ODI-PGD rather than MultiTargetedCIFAR-10 robust model · CIFAR-10accuracy: 44.03\% -> 44.00 \pm 0.01\%Improved
Replace SimBA-DCT directions with ODS directions from surrogate modelsSimBA · ImageNet untargetedaverage queries: 909 -> 242Improved
Replace SimBA-DCT directions with ODS directions from surrogate modelsSimBA · ImageNet targetedmedian ℓ2 perturbation: 7.00 -> 3.55Improved
Replace RGF random sampling with ODS using one surrogateRGF · ImageNet ℓ2 untargetedaverage queries: 633 -> 133Improved
Replace RGF random sampling with ODS using one surrogateRGF · ImageNet ℓ∞ targetedsuccess rate: 65.3\% -> 92.0\%Improved
Use ODS-RGF instead of Square AttackBlack-box score-based attacks · ImageNet ℓ2 untargetedaverage queries: 647 -> 144Improved
Use ODS-RGF instead of Square AttackBlack-box score-based attacks · ImageNet ℓ∞ untargetedaverage queries: 60 -> 78Hurt
Replace Boundary Attack Gaussian directions with ODS from four surrogatesBoundary Attack · ImageNet untargeted, 10000 queriesmedian ℓ2 perturbation: 4.30 -> 0.57Improved
Replace Boundary Attack Gaussian directions with ODS from four surrogatesBoundary Attack · ImageNet targeted, 10000 queriesmedian ℓ2 perturbation: 27.05 -> 3.76Improved
Train ODS surrogate models on an OOD dataset rather than the full dataset (ablation)Boundary-ODS · ImageNet untargeted, 10000 queriesmedian ℓ2 perturbation: 0.57 -> 0.98Hurt

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 0 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

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