AI Paper Atlas

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

Evaluating the Adversarial Robustness of Adaptive Test-time Defenses

Francesco Croce, Sven Gowal, Thomas Brunner, Evan Shelhamer, Matthias Hein, Taylan Cemgil

ICML 2022 · Adversarial robustness and security › Certified and empirical robustness · approach: Attack evaluation and defense critique · 10 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
Apply an adaptive white-box transfer attack rather than the original weaker transfer evaluationSODEF on TRADES · CIFAR-10, ℓ∞ ε=8/255robust accuracy: 60.9% -> 52.2%Hurt
Optimize the full close-loop control defense with APGD and BPDAClose-loop control · CIFAR-10, ℓ∞ ε=2/255robust accuracy: 34.5% -> 5.6%Hurt
Attack the full HD defense adaptively using BPDA and EOTHD · CIFAR-10, ℓ∞ ε=8/255robust accuracy: 63.0% -> 61.0%; 59.2% -> 59.2%; 45.0% -> 45.2%Mixed
Transfer APGD perturbations from the base classifier to Anti-AdversariesAnti-Adversaries · CIFAR-10, ℓ∞ ε=8/255robust accuracy: 63.7%No clear effect
Use APGD with BPDA and gradients averaged over purification iteratesSOAP · CIFAR-10, ℓ∞ ε=8/255robust accuracy: 51.02% -> 3.7%Hurt
Allow gradients through the ATLD+ purification step and attack with APGDATLD+ · CIFAR-10, ℓ∞ ε=8/255robust accuracy: 65.07% -> 12.6%Hurt
Attack AID-purifier directly with APGD and BPDA instead of transferring attacksAID-purifier · CIFAR-10, ℓ∞ ε=8/255robust accuracy: 49.3% -> 43.8%Hurt
Attack Natural Supervision purification with BPDA and EOTNatural Supervision purification · CIFAR-10, ℓ∞ ε=8/255robust accuracy: 59.4% -> 58.4%Hurt
Fix the random seed and remove EOT (ablation)Natural Supervision purification · CIFAR-10, ℓ∞ ε=8/255robust accuracy: 58.4% -> 56.4%Hurt
Keep negative images separate and hidden from the attacker (ablation)Natural Supervision purification · CIFAR-10robust accuracyNo clear effect
Attack ADP directly with APGD and 200 EOT iterationsADP · CIFAR-10, ℓ∞ ε=8/255robust accuracy: 69.71% -> 33.7% ± 0.54Hurt
Attack AID-purifier on SVHN with a larger APGD budget (ablation)AID-purifier · SVHN, ℓ∞ ε=12/255robust accuracy: 29.1% -> 25.0%Hurt

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