Research map / Adversarial robustness and security / Certified and empirical robustness
Evading Adversarial Example Detection Defenses with Orthogonal Projected Gradient Descent
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 |
|---|---|---|---|
| Use Orthogonal PGD instead of the original evaluation attack | Honeypot defense · CIFAR-10, ε=8/255 | attack success rate: 2% -> 100% | Helped |
| Use Orthogonal PGD instead of FGSM | Honeypot defense · CIFAR-10 | attack success rate: FGSM accuracy .06; ROC AUC .994 -> attack success rate above 99.9% | Helped |
| Use Orthogonal PGD instead of the original evaluation attack | Dense Layer Analysis · ε=0.01 | attack success rate: 13% -> 83% | Helped |
| Use Orthogonal PGD under the ε=8/255 constraint | Dense Layer Analysis · ε=8/255 | attack success rate: 100% | Helped |
| Apply Orthogonal PGD under ε=.03 at a 5% false positive rate | Dense Layer Analysis · CIFAR-10, ε=.03, 5% FPR | accuracy: 0 | Hurt |
| Use Orthogonal PGD instead of PGD | Sensitivity Inconsistency Detector · ε=0.01 | attack success rate: 9% -> 91% | Helped |
| Apply Orthogonal PGD under ε=.03 at a 5% false positive rate | Sensitivity Inconsistency Detector · CIFAR-10, ε=.03, 5% FPR | accuracy: 0 | Hurt |
| Use Orthogonal PGD with a differentiable SPAM approximation | SPAM steganalysis detector · ImageNet, 5% FPR | attack success rate: less than 3% -> 98.8% | Helped |
| Use Orthogonal PGD with a differentiable SPAM approximation | SPAM steganalysis detector · ImageNet | ROC AUC: less than 0.5 | Helped |
Closest papers by meaning
- Beating Attackers At Their Own Games: Adversarial Example Detection Using Adversarial Gradient Directions (AAAI 2021 · similarity 0.91)
- Are Adversarial Examples Created Equal? A Learnable Weighted Minimax Risk for Robustness under Non-uniform Attacks (AAAI 2021 · similarity 0.90)
- One Man's Trash Is Another Man's Treasure: Resisting Adversarial Examples by Adversarial Examples (CVPR 2020 · similarity 0.89)
- GAT: Generative Adversarial Training for Adversarial Example Detection and Robust Classification (ICLR 2020 · similarity 0.89)
- Adversarial Example Games (NeurIPS 2020 · similarity 0.89)
- One Stone, Two Birds: Enhancing Adversarial Defense Through the Lens of Distributional Discrepancy (ICML 2025 · similarity 0.87)
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.