Research map / Adversarial robustness and security / Certified and empirical robustness
Certified Robustness for Top-k Predictions against Adversarial Perturbations via Randomized Smoothing
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 |
|---|---|---|---|
| Increase k from top-1 to top-2 (ablation) | Randomized smoothing · CIFAR10, ℓ2 radius 0.5 (127/255) | certified top-k accuracy: 45.2% -> 58.8% | Helped |
| Increase k from top-2 to top-3 (ablation) | Randomized smoothing · CIFAR10, ℓ2 radius 0.5 (127/255) | certified top-k accuracy: 58.8% -> 67.2% | Helped |
| Increase k from top-1 to top-3 (ablation) | Randomized smoothing · ImageNet, ℓ2 radius 0.5 | certified top-k accuracy: 46.6% -> 57.8% | Helped |
| Increase k from top-3 to top-5 (ablation) | Randomized smoothing · ImageNet, ℓ2 radius 0.5 | certified top-k accuracy: 57.8% -> 62.8% | Helped |
| Increase the confidence level (ablation) | Randomized smoothing certification · CIFAR10 and ImageNet, certified top-3 accuracy | certified top-3 accuracy | No clear effect |
| Use a smaller Gaussian noise level σ (ablation) | Randomized smoothing · CIFAR10 and ImageNet, certified top-3 accuracy | certified top-3 accuracy | Mixed |
| Replace BinoCP with SimuEM (ablation) | Randomized smoothing certification · CIFAR10 and ImageNet, k=3 | certified top-3 accuracy | Mixed |
Closest papers by meaning
- Almost Tight L0-norm Certified Robustness of Top-k Predictions against Adversarial Perturbations (ICLR 2022 · similarity 0.95)
- Certified Adversarial Robustness via Randomized $\alpha$-Smoothing for Regression Models (NeurIPS 2024 · similarity 0.90)
- Provably Cost-Sensitive Adversarial Defense via Randomized Smoothing (ICML 2025 · similarity 0.90)
- Probably Approximately Global Robustness Certification (ICML 2025 · similarity 0.89)
- Treatment of Statistical Estimation Problems in Randomized Smoothing for Adversarial Robustness (NeurIPS 2024 · similarity 0.89)
- Mitigating the Curse of Dimensionality for Certified Robustness via Dual Randomized Smoothing (ICLR 2024 · 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.