AI Paper Atlas

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

Guided Adversarial Attack for Evaluating and Enhancing Adversarial Defenses

Gaurang Sriramanan, Sravanti Addepalli, Arya Baburaj, Venkatesh Babu R

NeurIPS 2020 · Adversarial robustness and security › Certified and empirical robustness · approach: Adversarial training refinements · 21 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
use GAT instead of FBF single-step defenseRN18 single-step adversarial training · CIFAR-10AutoAttack accuracy (%): 43.12 -> 47.30Improved
add GAMA adversaries to TRADES (TR-GAT)TRADES WRN34 · CIFAR-10AutoAttack accuracy (%): 53.18 -> 53.32Improved
use margin loss instead of cross-entropy loss in PGD attack (ablation)PGD-100 · TRADES-WRN34 CIFAR-10accuracy under attack (%): 55.61 -> 53.94Improved
compute margin on probabilities instead of logits (ablation)PGD margin attack · TRADES-WRN34 CIFAR-10accuracy under attack (%): 54.19 -> 53.94Improved
add decaying L2 relaxation term (graduated optimization) with Bernoulli init (GAMA-PGD) (ablation)PGD margin attack in prob. space · TRADES-WRN34 CIFAR-10 100-stepaccuracy under attack (%): 53.94 -> 53.29Improved
use Frank-Wolfe (GAMA-FW) instead of PGD for 10-step attack (ablation)GAMA · TRADES-WRN34 CIFAR-10 10-stepaccuracy under attack (%): 54.95 -> 54.27Improved
use Adam with true gradients instead of signed SGD in GAMA-PGD (ablation)GAMA-PGD · TRADES-WRN34 CIFAR-10accuracy under attack (%): 53.29 -> 53.66Hurt
remove alternation between CE and GA-CE attacks in GAT (ablation)GAT · CIFAR-10 RN18AutoAttack accuracy (%): 46.52 -> 47.30Improved
use GAMA max-margin loss for training in GAT instead of cross-entropy (ablation)GAT · CIFAR-10 RN18AutoAttack accuracy (%): 47.30 -> 10.98Hurt
use both guided attack and guided defense loss instead of only guided attack (A3) or only defense (A5) (ablation)GAT · CIFAR-10 RN18AutoAttack accuracy (%): A3 32.29, A5 44.21 -> 47.30Improved

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