Research map / Adversarial robustness and security / Black-box and transfer attacks
Skip Connections Matter: On the Transferability of Adversarial Examples Generated with ResNets
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 |
|---|---|---|---|
| Apply residual-gradient decay with SGM to PGD | RN152 · ImageNet→IncV3 | attack success rate: 26.56±0.75 -> 57.22±0.51 | Helped |
| Apply residual-gradient decay with SGM to PGD | DN201 · ImageNet→SE154 | attack success rate: 39.78±0.69 -> 72.03±0.53 | Helped |
| Apply residual-gradient decay with SGM to PGD | DN201 · ImageNet→IncV3 | attack success rate: 36.01±0.59 -> 65.48±0.23 | Helped |
| Apply SGM to FGSM | RN152 · ImageNet→VGG19 | attack success rate: 41.96±0.52 -> 47.54±0.14 | Hurt |
| Apply SGM to FGSM | RN152 · ImageNet→IncV3 | attack success rate: 25.66±0.07 -> 29.41±0.24 | Hurt |
| Add SGM to MI+DI | DN201 · ImageNet→VGG19 | attack success rate: 87.16 -> 93.00 | Helped |
| Add SGM to MI+DI | DN201 · ImageNet→SE154 | attack success rate: 79.80 -> 89.86 | Helped |
| Add SGM to TI | RN152 · ImageNet→IncV3 ens3 | attack success rate: 45.36±0.97 -> 52.62±0.40 | Helped |
| Add SGM to TI | DN201 · ImageNet→IncRes ens3 | attack success rate: 34.28±0.88 -> 39.32±0.80 | Helped |
| Decrease residual-gradient decay factor γ (ablation) | DenseNet source models · ImageNet transfer attacks | attack success rate | Mixed |
| Decrease residual-gradient decay factor γ (ablation) | ResNet source models · ImageNet transfer attacks | attack success rate | Helped |
| Use SGM rather than FGSM (ablation) | ResNet and DenseNet source models · ImageNet white-box attacks, ε ≤ 2 | attack success rate | No clear effect |
Closest papers by meaning
- Skip Connections Eliminate Singularities (ICLR 2018 · similarity 0.88)
- Enhancing Adversarial Transferability with Adversarial Weight Tuning (AAAI 2025 · similarity 0.86)
- Enhancing Adversarial Transferability with Checkpoints of a Single Model's Training (CVPR 2025 · similarity 0.85)
- Improving the Adversarial Transferability of Vision Transformers with Virtual Dense Connection (AAAI 2024 · similarity 0.85)
- LRS: Enhancing Adversarial Transferability through Lipschitz Regularized Surrogate (AAAI 2024 · similarity 0.84)
- Improving Transferable Targeted Adversarial Attacks with Model Self-Enhancement (CVPR 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.