Research map / Adversarial robustness and security / Certified and empirical robustness
Memorization Weights for Instance Reweighting in Adversarial Training
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 changed | Where | Measured | Effect |
|---|---|---|---|
| Add MeoW instance reweighting to AT | AT with PreActResNet-18 · CIFAR-10 | Best robustness accuracy: 52.81 ± 0.23 -> 55.07 ± 0.27 | Helped |
| Add MeoW instance reweighting to AT | AT with PreActResNet-18 · CIFAR-100 | Best robustness accuracy: 27.09 ± 0.18 -> 30.47 ± 0.25 | Helped |
| Add MeoW instance reweighting to AT | AT with PreActResNet-18 · SVHN | Best robustness accuracy: 54.51 ± 0.13 -> 59.49 ± 0.20 | Helped |
| Add MeoW instance reweighting to AT | AT with WideResNet-34-10 · CIFAR-10 | AutoAttack accuracy: 52.51 ± 0.22 -> 53.92 ± 0.24 | Helped |
| Add MeoW instance reweighting to TRADES | TRADES with WideResNet-34-10 · CIFAR-10 | AutoAttack accuracy: 53.00 ± 0.27 -> 55.60 ± 0.29 | Helped |
| Add MeoW instance reweighting to FAT | FAT with WideResNet-34-10 · CIFAR-10 | AutoAttack accuracy: 52.09 ± 0.23 -> 53.80 ± 0.21 | Helped |
| Add MeoW instance reweighting to AWP | AWP with WideResNet-34-10 · CIFAR-10 | AutoAttack accuracy: 56.03 ± 0.29 -> 56.61 ± 0.22 | Helped |
| Add MeoW instance reweighting to RST | RST with WideResNet-28-10 · CIFAR-10 | AutoAttack accuracy: 59.41 ± 0.20 -> 60.70 ± 0.19 | Helped |
| Add MeoW instance reweighting to AT | AT with WideResNet-34-10 · CIFAR-10 | Natural accuracy: 86.07 ± 0.16 -> 87.03 ± 0.18 | Helped |
| Add MeoW instance reweighting to TRADES | TRADES with WideResNet-34-10 · CIFAR-10 | Natural accuracy: 84.37 ± 0.20 -> 85.07 ± 0.19 | Helped |
| Add MeoW instance reweighting to FAT | FAT with WideResNet-34-10 · CIFAR-10 | Natural accuracy: 87.93 ± 0.23 -> 88.10 ± 0.19 | Helped |
| Add MeoW instance reweighting to AWP | AWP with WideResNet-34-10 · CIFAR-10 | Natural accuracy: 85.22 ± 0.19 -> 85.93 ± 0.19 | Helped |
Closest papers by meaning
- Exploring Memorization in Adversarial Training (ICLR 2022 · similarity 0.89)
- Self-ensemble Adversarial Training for Improved Robustness (ICLR 2022 · similarity 0.87)
- Doubly Robust Instance-Reweighted Adversarial Training (ICLR 2024 · similarity 0.86)
- Soften to Defend: Towards Adversarial Robustness via Self-Guided Label Refinement (CVPR 2024 · similarity 0.84)
- Improving Accuracy-robustness Trade-off via Pixel Reweighted Adversarial Training (ICML 2024 · similarity 0.84)
- The Effectiveness of Random Forgetting for Robust Generalization (ICLR 2024 · similarity 0.84)
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.