Research map / Adversarial robustness and security / Certified and empirical robustness
SmoothMix: Training Confidence-calibrated Smoothed Classifiers for Certified Robustness
Jongheon Jeong, Sejun Park, Minkyu Kim, Heung-Chang Lee, Doguk Kim, Jinwoo Shin
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 SmoothMix self-mixup loss at η=5.0 instead of η=1.0 (ablation) | SmoothMix · MNIST test, σ=1.00 | ACR: 1.788 -> 1.820 | Improved |
| Use a one-step adversary as the training input (ablation) | SmoothMix · MNIST test, σ=1.00 | ACR: 1.788 -> 1.816 | Improved |
| Use a one-step adversary as the training input (ablation) | SmoothMix · CIFAR-10 test, σ=0.50 | ACR: 0.720 -> 0.737 | Improved |
| Use a one-step adversary as the training input (ablation) | SmoothMix · CIFAR-10 test, σ=0.25 | ACR: 0.553 -> 0.548 | Hurt |
| Increase the relative weight η of the SmoothMix loss (ablation) | SmoothMix · MNIST, σ=1.0 | certified robustness | Improved |
| Keep the product α·T constant while varying step size and number of steps (ablation) | SmoothMix · MNIST, σ=1.0 | ACR | No clear effect |
| Add a hard ℓ2 radius restriction to adversarial-example search (ablation) | SmoothMix · MNIST, σ=1.0 | certified robustness | Hurt |
| Reduce the number of noise samples used to estimate the smoothed prediction to m=1 (ablation) | SmoothMix · MNIST, σ=1.0 | ACR | No clear effect |
| Train with SmoothMix rather than Gaussian augmentation alone | Smoothed classifier · MNIST, σ=1.0 | ACR | Improved |
| Train with SmoothMix rather than Consistency | Smoothed classifier · CIFAR-10, σ=0.50 | ACR: 0.720 -> 0.737 | Improved |
Later papers that compared with this method
5 later accepted papers report SmoothMix as a baseline in their results tables; 4 of them report a better number on most of the results they share with it.
- Average Certified Radius is a Poor Metric for Randomized Smoothing (ICML 2025 · better on 0 of 1 shared result)
- DRF: Improving Certified Robustness via Distributional Robustness Framework (AAAI 2024 · better on 3 of 3 shared results)
- Confidence-Aware Training of Smoothed Classifiers for Certified Robustness (AAAI 2023 · better on 7 of 7 shared results)
- Interpolation for Robust Learning: Data Augmentation on Wasserstein Geodesics (ICML 2023 · better on 1 of 1 shared result)
- Accelerating Certified Robustness Training via Knowledge Transfer (NeurIPS 2022 · better on 15 of 15 shared results)
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Confidence-Aware Training of Smoothed Classifiers for Certified Robustness (AAAI 2023 · similarity 0.93)
- Consistency Regularization for Certified Robustness of Smoothed Classifiers (NeurIPS 2020 · similarity 0.92)
- DRF: Improving Certified Robustness via Distributional Robustness Framework (AAAI 2024 · similarity 0.89)
- The Lipschitz-Variance-Margin Tradeoff for Enhanced Randomized Smoothing (ICLR 2024 · similarity 0.89)
- Mitigating the Curse of Dimensionality for Certified Robustness via Dual Randomized Smoothing (ICLR 2024 · similarity 0.88)
- Incremental Randomized Smoothing Certification (ICLR 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.
