Research map / Adversarial robustness and security / Certified and empirical robustness
A Spectral View of Randomized Smoothing under Common Corruptions: Benchmarking and Improving Certified Robustness
Jiachen Sun, Akshay Mehra, Bhavya Kailkhura, Pin-Yu Chen, Dan Hendrycks, Jihun Hamm, Z. Morley Mao
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (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 |
|---|---|---|---|
| add hierarchical consistency regularization (HCR) in place of JSD to FourierMix (ablation) | FourierMix (ResNet-110) · CIFAR-10-C | mACR: 0.444 -> 0.460 | Improved |
| add HCR in place of JSD to FourierMix (ablation) | FourierMix (ResNet-110) · CIFAR-10-C-bar | mACR: 0.397 -> 0.419 | Improved |
| add HCR in place of JSD to FourierMix (ablation) | FourierMix (ResNet-110) · CIFAR-100-C | mACR: 0.247 -> 0.261 | Improved |
| add HCR in place of JSD to FourierMix (ablation) | FourierMix (ResNet-50) · ImageNet-C | mACR: 0.399 -> 0.397 | No clear effect |
| add JSD consistency loss to Gaussian augmentation (ablation) | Gaussian augmentation (ResNet-110) · CIFAR-10-C | mACR: 0.363 -> 0.439 | Improved |
| add JSD consistency loss to Gaussian augmentation (ablation) | Gaussian augmentation (ResNet-50) · ImageNet-C | mACR: 0.256 -> 0.395 | Improved |
| use FourierMix instead of AugMix augmentation, both with HCR (ablation) | HCR-regularized training (ResNet-110) · CIFAR-10-C | mACR: 0.437 -> 0.460 (5.3% relative gain) | Improved |
| use FourierMix instead of AugMix augmentation, both with HCR (ablation) | HCR-regularized training (ResNet-110) · CIFAR-100-C | mACR: 0.249 -> 0.261 (4.8% relative gain) | Improved |
| use FourierMix+HCR instead of Gaussian augmentation | ResNet-110 · CIFAR-10-C | mACR: 0.363 -> 0.460 (26.7% relative gain) | Improved |
| use FourierMix+HCR instead of Gaussian augmentation | ResNet-110 · CIFAR-100-C | mACR: 0.169 -> 0.261 (54.4% relative gain) | Improved |
| use FourierMix instead of AugMix, both with JSD (ablation) | JSD-regularized training (ResNet-50) · ImageNet-C | mACR: 0.391 -> 0.399 (2.1% relative gain on overall mACR vs AugMix+JSD) | Improved |
| add AutoAugment to Gaussian training (ablation) | Gaussian augmentation (ResNet-110) · CIFAR-10-C-bar | mACR: 0.314 -> 0.304 | Hurt |
Closest papers by meaning
- Average Certified Radius is a Poor Metric for Randomized Smoothing (ICML 2025 · similarity 0.89)
- Effects of Exponential Gaussian Distribution on (Double Sampling) Randomized Smoothing (ICML 2024 · similarity 0.88)
- Certified Adversarial Robustness via Randomized $\alpha$-Smoothing for Regression Models (NeurIPS 2024 · similarity 0.87)
- Certified Robustness for Deep Equilibrium Models via Serialized Random Smoothing (NeurIPS 2024 · similarity 0.87)
- Adaptive Randomized Smoothing: Certified Adversarial Robustness for Multi-Step Defences (NeurIPS 2024 · similarity 0.87)
- Treatment of Statistical Estimation Problems in Randomized Smoothing for Adversarial Robustness (NeurIPS 2024 · 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.
