Research map / Learning theory and algorithms / Prediction and decision losses
Invariance-Aware Randomized Smoothing Certificates
Jan Schuchardt, Stephan Günnemann
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 |
|---|---|---|---|
| Use the tight 2D rotation-invariance certificate instead of black-box smoothing | 2D rotation-invariant models · Adversarial scaling with σ=0.5, ||X||=0.01, p=0.8 | maximum certified ||Δ||₂: 0.4 -> 0.73 | Improved |
| Use rotation-aware gray-box certificates rather than black-box randomized smoothing | Rotation-invariant point-cloud classifiers · MNIST and ModelNet40, randomly perturbed inputs with ||Δ||=σ=0.1 | probabilistic certified accuracy | Improved |
| Use the tight rotation certificate instead of the orbit-based certificate | Rotation-invariant point-cloud classifiers · MNIST and ModelNet40, perturbations with rotational components | probabilistic certified accuracy | No clear effect |
| Enforce translation invariance in addition to rotation invariance | EnsPointNet · MNIST point clouds, σ=0.2, adversarial scaling | maximum certifiable ||Δ||₂: ≈ 0.62 -> 0.8 | Improved |
| Use the tight certificate instead of the orbit-based certificate | Rotation-invariant classifiers · Point-cloud classification in practical smoothing regimes | certified accuracy | No clear effect |
| Increase clean-data norm relative to smoothing standard deviation | Tight 2D rotation certificate · 2D rotation-invariance certificate parameter space | p_min improvement over black-box certificate | Hurt |
Closest papers by meaning
- Certified Robustness for Deep Equilibrium Models via Serialized Random Smoothing (NeurIPS 2024 · similarity 0.85)
- The Lipschitz-Variance-Margin Tradeoff for Enhanced Randomized Smoothing (ICLR 2024 · similarity 0.85)
- Sufficient Invariant Learning for Distribution Shift (CVPR 2025 · similarity 0.84)
- Incremental Randomized Smoothing Certification (ICLR 2024 · similarity 0.84)
- One Sample is Enough to Make Conformal Prediction Robust (NeurIPS 2025 · similarity 0.84)
- Conformal Prediction under Lévy-Prokhorov Distribution Shifts: Robustness to Local and Global Perturbations (NeurIPS 2025 · similarity 0.83)
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.
