Research map / Adversarial robustness and security / Certified and empirical robustness
Formal Guarantees on the Robustness of a Classifier against Adversarial Manipulation
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 |
|---|---|---|---|
| Apply Cross-Lipschitz regularization instead of standard kernel regularization | Gaussian-kernel classifier · MNIST test | test error: 1.48% -> 1.44% | Helped |
| Apply Cross-Lipschitz regularization instead of standard kernel regularization | Gaussian-kernel classifier · MNIST test | average adversarial perturbation L2: 1.91 -> 3.12 | Helped |
| Apply Cross-Lipschitz regularization instead of standard kernel regularization | Gaussian-kernel classifier · MNIST test | average robustness guarantee L2: 0.058 -> 0.045 | Hurt |
| Use a local rather than global cross-Lipschitz bound | One-hidden-layer neural networks · MNIST and CIFAR10 test sets | robustness guarantee: The local guarantees are up to eight times better | Helped |
| Apply Cross-Lipschitz regularization | One-hidden-layer neural network · MNIST | robustness | Helped |
| Apply data augmentation | One-hidden-layer neural network · MNIST | robustness | No clear effect |
| Use adversarial training | One-hidden-layer neural network · MNIST | robustness guarantees and adversarial resistance | Helped |
| Apply Cross-Lipschitz regularization | One-hidden-layer neural network · CIFAR10 augmented | test error and adversarial perturbation upper bound | Helped |
| Apply Cross-Lipschitz regularization | One-hidden-layer neural network · CIFAR10 augmented | robustness guarantee | No clear effect |
| Apply data augmentation | One-hidden-layer neural network · CIFAR10 | test error | Helped |
| Apply data augmentation | One-hidden-layer neural network · CIFAR10 | robustness properties | No clear effect |
| Use adversarial training | One-hidden-layer neural network · CIFAR10 | robustness bounds | Helped |
Closest papers by meaning
- Consistency Regularization for Certified Robustness of Smoothed Classifiers (NeurIPS 2020 · similarity 0.91)
- Probably Approximately Global Robustness Certification (ICML 2025 · similarity 0.91)
- Certified Robustness via Dynamic Margin Maximization and Improved Lipschitz Regularization (NeurIPS 2023 · similarity 0.90)
- Regularized Training and Tight Certification for Randomized Smoothed Classifier with Provable Robustness (AAAI 2020 · similarity 0.90)
- Tight Certificates of Adversarial Robustness for Randomly Smoothed Classifiers (NeurIPS 2019 · similarity 0.89)
- Adversarial vulnerability for any classifier (NeurIPS 2018 · similarity 0.89)
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.