Research map / Adversarial robustness and security / Black-box and transfer attacks
Augmented Lagrangian Adversarial Attacks
Jérôme Rony, Eric Granger, Marco Pedersoli, Ismail Ben Ayed
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 changed | Where | Measured | Effect |
|---|---|---|---|
| increase ALMA budget from 100 to 1000 iterations | ALMA l1 · MNIST l1 attack | median l1 distance: 11.45 -> 7.16 | Improved |
| increase ALMA budget from 100 to 1000 iterations | ALMA l1 · CIFAR10 l1 attack | median l1 distance: 4.31 -> 3.69 | Improved |
| increase ALMA budget from 100 to 1000 iterations | ALMA LPIPS · CIFAR10 LPIPS attack | median LPIPS x10^-2: 2.47 -> 0.60 | Improved |
| increase FMN l1 budget from 100 to 1000 iterations | FMN l1 · MNIST l1 attack | ASR (%): 69.87 -> 95.35 | Improved |
| use ALMA l1 instead of FMN l1 (1000 its) on MNIST | ALMA l1 · MNIST l1 attack | ASR (%): 95.35 -> 100 | Improved |
| use ALMA CIEDE2000 instead of PerC-AL (1000 its) | ALMA CIEDE2000 · CIFAR10 CIEDE2000 attack | median CIEDE2000 distance: 2.72 -> 0.78 | Improved |
| use ALMA CIEDE2000 instead of PerC-AL (1000 its) | ALMA CIEDE2000 · ImageNet CIEDE2000 attack | median CIEDE2000 distance: 3.42 -> 0.63 | Improved |
| use ALMA LPIPS instead of LPA with binary search | ALMA LPIPS · CIFAR10 LPIPS attack | median LPIPS x10^-2: 5.39 -> 0.60 | Improved |
| use C&W LPIPS penalty baseline instead of ALMA LPIPS | C&W LPIPS · CIFAR10 LPIPS attack | median LPIPS x10^-2: 0.60 -> 0.47 | Improved |
| use FMN l1 instead of ALMA l1 (1000 its) on ImageNet | FMN l1 · ImageNet l1 attack | median l1 distance: 12.10 -> 7.58 | Improved |
Closest papers by meaning
- Minimally distorted Adversarial Examples with a Fast Adaptive Boundary Attack (ICML 2020 · similarity 0.88)
- Online Adversarial Attacks (ICLR 2022 · similarity 0.87)
- A Frank-Wolfe Framework for Efficient and Effective Adversarial Attacks (AAAI 2020 · similarity 0.87)
- Towards Efficient Training and Evaluation of Robust Models against $l_0$ Bounded Adversarial Perturbations (ICML 2024 · similarity 0.86)
- SuperDeepFool: a new fast and accurate minimal adversarial attack (NeurIPS 2024 · similarity 0.86)
- AttackBench: Evaluating Gradient-based Attacks for Adversarial Examples (AAAI 2025 · similarity 0.86)
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.
