Research map / Adversarial robustness and security / Certified and empirical robustness
Rethinking Lipschitz Neural Networks and Certified Robustness: A Boolean Function Perspective
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 |
|---|---|---|---|
| Use SortNet instead of ℓ∞-distance Net by incorporating additional order statistics | Lipschitz classifiers · CIFAR-10 ε=2/255 and ε=8/255 | certified accuracy | Helped |
| Add order statistics beyond the maximum (ρ>0) (ablation) | SortNet · CIFAR-10 | certified accuracy | Helped |
| Increase the SortNet+MLP backbone size | SortNet+MLP · TinyImageNet ε=1/255 | certified accuracy: 17.92; 18.18 | Helped |
| Increase the SortNet+MLP backbone size | SortNet+MLP · ImageNet 64×64 ε=1/255 | certified accuracy: 9.02; 9.54 | Helped |
| Use geometric-series weights and stochastic max estimation instead of explicitly sorting | Specialized SortNet · Training efficiency | training time: 22-fold decrease in training time | Helped |
| Use SortNet instead of relaxation-based certification | Certified classifiers · CIFAR-10 ε=2/255 | certified accuracy: +3∼+6 point for both radii | Mixed |
Closest papers by meaning
- A Recipe for Improved Certifiable Robustness (ICLR 2024 · similarity 0.93)
- Some Fundamental Aspects about Lipschitz Continuity of Neural Networks (ICLR 2024 · similarity 0.92)
- ECLipsE: Efficient Compositional Lipschitz Constant Estimation for Deep Neural Networks (NeurIPS 2024 · similarity 0.91)
- Enhancing Certified Robustness via Block Reflector Orthogonal Layers and Logit Annealing Loss (ICML 2025 · similarity 0.91)
- Certified Robustness Under Bounded Levenshtein Distance (ICLR 2025 · similarity 0.91)
- Efficient Robust Conformal Prediction via Lipschitz-Bounded Networks (ICML 2025 · similarity 0.90)
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.