Research map / Adversarial robustness and security / Black-box and transfer attacks
Nesterov Accelerated Gradient and Scale Invariance for Adversarial Attacks
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 |
|---|---|---|---|
| Add Nesterov look-ahead updates to iterative gradient attacks (ablation) | Inc-v3-crafted attacks · Inc-v3 to Inc-v4 | attack success rate (%): 52.6 -> 76.0 | Helped |
| Add scale-copy gradient optimization (SIM) (ablation) | NI-FGSM on Inc-v3 · Inc-v3 to Inc-v4 | attack success rate (%): 52.6 -> 76.0 | Helped |
| Add scale-copy gradient optimization (SIM) (ablation) | NI-FGSM on Inc-v3 · Inc-v3 to Inc-v3_ens3 | attack success rate (%): 12.9 -> 31.6 | Helped |
| Replace TI-DIM with SI-NI-TI-DIM (ablation) | Inc-v3-crafted attacks · Inc-v3 to Inc-v4 | attack success rate (%): 66.1 -> 85.5 | Helped |
| Replace TI-DIM with SI-NI-TI-DIM (ablation) | attacks on an ensemble of normally trained models · Multi-model attack: Inc-v3_ens3 | attack success rate (%): 85.2 -> 96.0 | Helped |
| Replace TI-DIM with SI-NI-TI-DIM (ablation) | attacks transferred from an ensemble of normally trained models · Advanced defenses (average) | attack success rate (%): 75.6 -> 90.3 | Helped |
| Use Nesterov updates instead of momentum updates (ablation) | iterative attacks crafted on Inc-v3 · Transfer to Inc-v4 and IncRes-v2 | attack success rate (%) | Helped |
| Optimize over scale copies of the input (ablation) | gradient-based attacks · ImageNet transfer attacks | attack success rate (%): 10% ∼ 35% | Helped |
Closest papers by meaning
- Improving Integrated Gradient-based Transferable Adversarial Examples by Refining the Integration Path (AAAI 2025 · similarity 0.87)
- On the Convergence of an Adaptive Momentum Method for Adversarial Attacks (AAAI 2024 · similarity 0.87)
- Enhancing Adversarial Transferability with Adversarial Weight Tuning (AAAI 2025 · similarity 0.86)
- Adversarial Initialization with Universal Adversarial Perturbation: A New Approach to Fast Adversarial Training (AAAI 2024 · similarity 0.86)
- Boosting Adversarial Transferability through Augmentation in Hypothesis Space (CVPR 2025 · similarity 0.85)
- Boosting the Transferability of Adversarial Attack on Vision Transformer with Adaptive Token Tuning (NeurIPS 2024 · similarity 0.85)
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.