Research map / Adversarial robustness and security / Black-box and transfer attacks
Blurred-Dilated Method for Adversarial Attacks
Yang Deng, Weibin Wu, Jianping Zhang, Zibin Zheng
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 |
|---|---|---|---|
| replace RN50 source with BD RN50 | source model for MI-FGSM transfer attacks · ImageNet, eps=16/255 | average attack success rate: 72.8% -> 95.9% | Improved |
| replace VGG16 source with BD VGG16 | source model for MI-FGSM transfer attacks · ImageNet, eps=16/255 | average attack success rate: 72.8% -> 86.4% | Improved |
| replace DN121 source with BD DN121 | source model for MI-FGSM transfer attacks · ImageNet, eps=16/255 | average attack success rate: 76.2% -> 88.3% | Improved |
| replace MobV2 source with BD MobV2 | source model for MI-FGSM transfer attacks · ImageNet, eps=16/255 | average attack success rate: 61.2% -> 82.3% | Improved |
| BD vs LinBP, the strongest baseline, with RN50 source | source model for MI-FGSM transfer attacks · ImageNet, eps=16/255 | average black-box success rate: 84.2% -> 95.3% | Improved |
| BD vs LinBP at small budget | source model for MI-FGSM transfer attacks · ImageNet, eps=4/255 | average black-box success rate: 35.8% -> 56.8% | Improved |
| add BD to ILA | source model for MI-FGSM transfer attacks · ImageNet, eps=16/255 | average black-box success rate: 78.4% -> 96.3% | Improved |
| add BD to LinBP | source model for MI-FGSM transfer attacks · ImageNet, eps=16/255 | average black-box success rate: 84.2% -> 97.3% | Improved |
| add BD to ILA+LinBP | source model for MI-FGSM transfer attacks · ImageNet, eps=16/255 | average black-box success rate: 88.7% -> 97.9% | Improved |
| remove dilated convolutions (BlurPool only) (ablation) | source model for MI-FGSM transfer attacks · ImageNet, eps=16/255 | average black-box success rate: 91.0% -> 85.4% | Hurt |
| remove BlurPool (dilated conv only) (ablation) | source model for MI-FGSM transfer attacks · ImageNet, eps=16/255 | average black-box success rate: 91.0% -> 69.9% | Hurt |
| BD ResNet-20 source on low-resolution data | source model for MI-FGSM transfer attacks · CIFAR-10, eps=16/255 | average attack success rate: 95.2% -> 97.2% | Improved |
Closest papers by meaning
- Transferable Adversarial Perturbations (ECCV 2018 · similarity 0.89)
- Yet Another Intermediate-Level Attack (ECCV 2020 · similarity 0.88)
- Improving the Transferability of Adversarial Samples With Adversarial Transformations (CVPR 2021 · similarity 0.88)
- Boosting Black-Box Attack With Partially Transferred Conditional Adversarial Distribution (CVPR 2022 · similarity 0.88)
- Evading Defenses to Transferable Adversarial Examples by Translation-Invariant Attacks (CVPR 2019 · similarity 0.88)
- Beyond ImageNet Attack: Towards Crafting Adversarial Examples for Black-box Domains (ICLR 2022 · similarity 0.88)
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.
