Research map / Adversarial robustness and security / Black-box and transfer attacks
Enhancing Adversarial Transferability by Balancing Exploration and Exploitation with Gradient-Guided Sampling
Zenghao Niu, Weicheng Xie, Siyang Song, Zitong Yu, Feng Liu, Linlin Shen
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 |
|---|---|---|---|
| use gradient-guided sampling instead of momentum-guided sampling (ablation) | RS-based MI-FGSM · Res50 surrogate, 8 other models | avg untargeted ASR: 58.79 -> 79.93 | Improved |
| use gradient-guided sampling instead of random sampling (ablation) | MI-FGSM with inner iterations · Res50 surrogate, 8 other models | avg untargeted ASR: 63.74 -> 79.93 | Improved |
| use momentum-guided sampling instead of random sampling (ablation) | RS · Res50 surrogate, 8 other models | avg untargeted ASR: 63.74 -> 58.79 | Hurt |
| use momentum-guided sampling instead of random sampling (ablation) | RS · Res50 white-box | ASR: 97.3 -> 97.5 | No clear effect |
| add GGS to GRA | GRA · Res50 surrogate, 9 models | avg untargeted ASR: 73.41 -> 78.69 | Improved |
| add GGS to PGN | PGN · Res50 surrogate, 9 models | avg untargeted ASR: 76.53 -> 83.23 | Improved |
| add GGS to PGN | PGN · Res50 surrogate, 9 models | avg targeted ASR: 6.84 -> 12.79 | Improved |
| add GGS to SIM | SIM · Res50 surrogate, 9 models | avg untargeted ASR: 46.80 -> 90.12 | Improved |
| add GGS to DIM | DIM · Res50 surrogate, 9 models | avg untargeted ASR: 51.30 -> 90.13 | Improved |
| add GGS to TIM | TIM · Res50 surrogate, 9 models | avg untargeted ASR: 40.82 -> 84.94 | Improved |
| add GGS to SSM | SSM · Res50 surrogate, 9 models | avg targeted ASR: 8.40 -> 12.54 | Improved |
| add GGS to Admix | Admix · Res50 surrogate, 9 models | avg targeted ASR: 11.28 -> 30.50 | Improved |
Closest papers by meaning
- Seeking Flat Minima Over Diverse Surrogates for Improved Adversarial Transferability: A Theoretical Framework and Algorithmic Instantiation (TPAMI 2026 · similarity 0.91)
- Enhancing Adversarial Transferability with Adversarial Weight Tuning (AAAI 2025 · similarity 0.90)
- Enhancing Adversarial Transferability with Checkpoints of a Single Model's Training (CVPR 2025 · similarity 0.89)
- Towards Transferable Adversarial Attacks with Centralized Perturbation (AAAI 2024 · similarity 0.89)
- Improving Transferable Targeted Adversarial Attacks with Model Self-Enhancement (CVPR 2024 · similarity 0.88)
- Ensemble Diversity Facilitates Adversarial Transferability (CVPR 2024 · 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.
