Research map / Adversarial robustness and security / Certified and empirical robustness
On Certifying Non-Uniform Bounds against 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 |
|---|---|---|---|
| Replace uniform perturbation bounds with optimized non-uniform bounds | Certification framework · MNIST, Fashion-MNIST, and SVHN test sets; normal and PGD-trained models | certified bound: 0.0295 -> 0.0349; 0.0692 -> 0.1678; 0.0309 -> 0.0350; 0.0507 -> 0.1404; 0.0319 -> 0.0360; 0.0436 -> 0.1167; 0.0134 -> 0. | Helped |
| Adversarially train with PGD | MNIST classifier, 300-300-300 · MNIST test set | mean cosine similarity of non-uniform bounds: 0.9774 -> 0.9964 | Helped |
| Adversarially train with PGD | MNIST classifier, 300-300-300 · MNIST test set | minimum cosine similarity of non-uniform bounds: 0.5038 -> 0.9104 | Helped |
Closest papers by meaning
- Adversarial Robustness with Non-uniform Perturbations (NeurIPS 2021 · similarity 0.91)
- Probably Approximately Global Robustness Certification (ICML 2025 · similarity 0.87)
- Et Tu Certifications: Robustness Certificates Yield Better Adversarial Examples (ICML 2024 · similarity 0.87)
- Compositional Curvature Bounds for Deep Neural Networks (ICML 2024 · similarity 0.87)
- A Recipe for Improved Certifiable Robustness (ICLR 2024 · similarity 0.87)
- Learning Better Certified Models from Empirically-Robust Teachers (NeurIPS 2026 · 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.