Research map / Adversarial robustness and security / Certified and empirical robustness
Understanding and Increasing Efficiency of Frank-Wolfe Adversarial Training
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 |
|---|---|---|---|
| Use distortion-adaptive FW steps instead of 10-step PGD training | Adversarial training · CIFAR-10 ε=8/255 | AutoAttack accuracy: 45.38 -> 45.09 | No clear effect |
| Use distortion-adaptive FW steps instead of 10-step PGD training | Adversarial training · CIFAR-10 ε=8/255 | training time: 49.8 -> 16.9 min | Helped |
| Use distortion-adaptive FW steps instead of the strongest listed single-step baseline | Adversarial training · CIFAR-10 ε=16/255 | AutoAttack accuracy: 15.09 -> 21.06 | Helped |
| Use distortion-adaptive FW steps instead of 10-step PGD training | Adversarial training · CIFAR-10 ε=16/255 | AutoAttack accuracy: 22.62 -> 21.06 | Hurt |
| Use distortion-adaptive FW steps instead of 10-step PGD training | Adversarial training · CIFAR-10 ε=16/255 | training time: 49.8 -> 22.4 min | Helped |
| Use distortion-adaptive FW steps instead of the strongest listed single-step baseline | Adversarial training · CIFAR-100 ε=8/255 | AutoAttack accuracy: 19.27 -> 21.64 | Helped |
| Use distortion-adaptive FW steps instead of 10-step PGD training | Adversarial training · CIFAR-100 ε=8/255 | training time: 49.8 -> 15.1 min | Helped |
| Use distortion-adaptive FW steps instead of the strongest listed single-step baseline | Adversarial training · CIFAR-100 ε=16/255 | AutoAttack accuracy: 6.34 -> 10.47 | Helped |
| Use distortion-adaptive FW steps instead of 10-step PGD training | Adversarial training · CIFAR-100 ε=16/255 | AutoAttack accuracy: 11.30 -> 10.47 | Hurt |
| Use distortion-adaptive FW steps instead of 10-step PGD training | Adversarial training · CIFAR-100 ε=16/255 | training time: 49.8 -> 24.0 min | Helped |
| Increase the minimum distortion ratio r (ablation) | FW-Adapt · CIFAR-10 and CIFAR-100 | adversarial accuracy | Mixed |
Closest papers by meaning
- Revisiting and Advancing Fast Adversarial Training Through The Lens of Bi-Level Optimization (ICML 2022 · similarity 0.90)
- Taxonomy Driven Fast Adversarial Training (AAAI 2024 · similarity 0.90)
- Clarifying the Behavior and the Difficulty of Adversarial Training (AAAI 2024 · similarity 0.89)
- Adversarial Training Can Provably Improve Robustness: Theoretical Analysis of Feature Learning Process Under Structured Data (ICLR 2025 · similarity 0.89)
- Adversarial Initialization with Universal Adversarial Perturbation: A New Approach to Fast Adversarial Training (AAAI 2024 · similarity 0.88)
- Kernel Learning with Adversarial Features: Numerical Efficiency and Adaptive Regularization (NeurIPS 2025 · similarity 0.87)
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.