Research map / Adversarial robustness and security / Certified and empirical robustness
On the Algorithmic Stability of Adversarial Training
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 |
|---|---|---|---|
| inject Gaussian noise into parameters and data during adversarial training (ablation) | adversarial training (linear regression) · simulation d=10, n=1000 | gradient Lipschitz constant (right tail): 323.36 -> 234.43 (99.9%-quantile) | Helped |
| inject Gaussian noise into parameters and data (ablation) | adversarial training (linear regression) · simulation d=10, n=1000 | mean gradient Lipschitz constant: 178.90 -> 167.72 | No clear effect |
| use smaller batch size (ablation) | adversarial training (linear regression) · simulation | gradient Lipschitz constant tail | Hurt |
| introduce error in computing the attack (ablation) | adversarial training · simulations (Appendix B) | generalization gap | Hurt |
| inject noise into adversarial training of deep networks (ablation) | deep neural network · CIFAR10 | generalization gap | Helped |
| use more accurate attack (PGD-k instead of FGM) (ablation) | adversarial training of deep network · CIFAR10 | adversarial training quality | Helped |
Closest papers by meaning
- Uniformly Stable Algorithms for Adversarial Training and Beyond (ICML 2024 · similarity 0.90)
- Algorithmic Stability Based Generalization Bounds for Adversarial Training (ICLR 2025 · similarity 0.90)
- Stability and Generalization of Adversarial Training for Shallow Neural Networks with Smooth Activation (NeurIPS 2024 · similarity 0.89)
- Adversarial Training Should Be Cast as a Non-Zero-Sum Game (ICLR 2024 · similarity 0.87)
- Adversarial Training Can Provably Improve Robustness: Theoretical Analysis of Feature Learning Process Under Structured Data (ICLR 2025 · similarity 0.86)
- Clarifying the Behavior and the Difficulty of Adversarial Training (AAAI 2024 · similarity 0.86)
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.