Research map / Adversarial robustness and security / Certified and empirical robustness
Learning perturbation sets for robust machine learning
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 |
|---|---|---|---|
| Train adversarially using the learned CVAE perturbation set instead of training directly on CIFAR10-C corruptions | CIFAR10 classifier · CIFAR10-C | robust accuracy (ε=3.9): 17.8 -> 66.1 | Helped |
| Use CVAE perturbation-set samples for augmentation rather than training directly on CIFAR10-C corruptions | CIFAR10 classifier · CIFAR10-C | accuracy (perturbed): 87.7 -> 90.5 | Helped |
| Use CVAE perturbation-set data augmentation rather than direct CIFAR10-C data augmentation | CIFAR10 classifier · CIFAR10-C OOD | accuracy: 85.0 -> 89.6 | Helped |
| Use CVAE adversarial training rather than CVAE data augmentation | CIFAR10 classifier · CIFAR10-C | robust accuracy (ε=10.2): 43.4 -> 55.6 | Helped |
| Use CVAE adversarial training rather than MI data augmentation | Image segmentation model · MI | robust accuracy (ε=17): 27.1 -> 35.4 | Helped |
| Use CVAE perturbation-set augmentation rather than MI data augmentation | Image segmentation model · MI | accuracy (perturbed): 45.2 -> 41.5 | Hurt |
| Use CVAE adversarial training rather than MI data augmentation | Image segmentation model · MI | accuracy (perturbed): 45.2 -> 41.7 | Hurt |
Closest papers by meaning
- Contrasting Adversarial Perturbations: The Space of Harmless Perturbations (AAAI 2025 · similarity 0.91)
- Lower bounds on the robustness to adversarial perturbations (NeurIPS 2017 · similarity 0.90)
- Attribute-Guided Adversarial Training for Robustness to Natural Perturbations (AAAI 2021 · similarity 0.90)
- Adversarially Robust Learning with Uncertain Perturbation Sets (NeurIPS 2023 · similarity 0.90)
- Adversarial Training Can Provably Improve Robustness: Theoretical Analysis of Feature Learning Process Under Structured Data (ICLR 2025 · similarity 0.88)
- Wide Two-Layer Networks can Learn from Adversarial Perturbations (NeurIPS 2024 · similarity 0.88)
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.