Research map / Adversarial robustness and security / Black-box and transfer attacks
Benchmarking Adversarial Robustness on Image Classification
Check 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 |
|---|---|---|---|
| Evaluate defenses across robustness curves instead of a single attack configuration | Image classifiers under adversarial attacks · CIFAR-10 and ImageNet | accuracy | Helped |
| Use PGD-based adversarial training | Image classifiers · CIFAR-10, white-box ℓ∞ attacks | robust accuracy | Helped |
| Use adversarial training for robustness beyond the training threat model | PGD-based adversarially trained models · Other threat models, including ℓ2 | robustness | Helped |
| Add randomization to the defense | Image classifiers · CIFAR-10 and ImageNet, score-based and decision-based black-box attacks | accuracy | Helped |
| Use input transformations such as JPEG compression or bit-depth reduction | Image classifiers · ImageNet, score-based and decision-based black-box attacks | accuracy | Helped |
| Use MIM or DIM instead of BIM for transfer attacks | Image classifiers · CIFAR-10 | transfer attack effectiveness | No clear effect |
| Use MIM or DIM instead of BIM for transfer attacks | Image classifiers · ImageNet | transfer attack effectiveness | Helped |
| Use adversarial training | Image classifiers · Natural accuracy and training cost | natural accuracy | Hurt |
Closest papers by meaning
- RobustBench: a standardized adversarial robustness benchmark (NeurIPS 2021 · similarity 0.93)
- Robust Models are less Over-Confident (NeurIPS 2022 · similarity 0.92)
- Fantastic Robustness Measures: The Secrets of Robust Generalization (NeurIPS 2023 · similarity 0.92)
- Adversarial Training Can Provably Improve Robustness: Theoretical Analysis of Feature Learning Process Under Structured Data (ICLR 2025 · similarity 0.92)
- Towards Million-Scale Adversarial Robustness Evaluation With Stronger Individual Attacks (CVPR 2025 · similarity 0.91)
- Comparing the Robustness of Modern No-Reference Image- and Video-Quality Metrics to Adversarial Attacks (AAAI 2024 · similarity 0.90)
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.