AI Paper Atlas

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

Regularizing Hard Examples Improves Adversarial Robustness

Hyungyu Lee, Saehyung Lee, Ho Bae, Sungroh Yoon

JMLR 2025 · Adversarial robustness and security › Certified and empirical robustness · approach: Adversarial training refinements

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 DPLS (difficulty-based label smoothing)PGD · CIFAR-10 A3 (eps=8)robust accuracy: 51.80 -> 54.03Improved
add DPLSTRADES · CIFAR-10 A3 (eps=8)robust accuracy: 54.08 -> 55.19Improved
add DPLSPGD · SVHN A3 (eps=8)robust accuracy: 35.20 -> 42.01Improved
add DPLSPGD · CIFAR-100 A3 (eps=8)robust accuracy: 27.36 -> 29.33Improved
add uniform label smoothing (LS) with same budgetPGD · CIFAR-10 A3 (eps=8)robust accuracy: 51.80 -> 51.99No clear effect
add hard-example pruning (top 10% difficulty)PGD · CIFAR-10 A3 (eps=8)robust accuracy: 51.80 -> 52.56Improved
prune top 10k hard examples from training (ablation)PGD · CIFAR-10robust accuracy: 51.80 -> 53.36Improved
prune top 10k hard examples from training (ablation)PGD · CIFAR-10standard accuracy: 87.19 -> 84.70Hurt
prune top 10k easy examples from training (ablation)PGD · CIFAR-10robust accuracy: 51.80 -> 49.71Hurt
train only on 10k hard examples instead of 10k normal examples (ablation)PGD · CIFAR-10robust accuracy: 39.56 -> 8.68Hurt
add DPLS and SWA (ablation)PGD · CIFAR-10 A3 (eps=8)robust accuracy: 53.98 -> 55.91Improved
use 0-1 loss for difficulty measure instead of EL2N (ablation)DPLS on TRADES · CIFAR-10 A3robust accuracy: 54.32 -> 55.19Improved

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