Research map / Label-efficient and robust learning / Self-supervised and contrastive
Using Self-Supervised Learning Can Improve Model Robustness and Uncertainty
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 |
|---|---|---|---|
| add auxiliary rotation prediction to adversarial training | 40-2 Wide ResNet · CIFAR-10 adversarial robustness | 20-step PGD accuracy: 44.8 -> 50.4 | Helped |
| add auxiliary rotation prediction to adversarial training | 40-2 Wide ResNet · CIFAR-10 adversarial robustness | 100-step PGD accuracy: 44.8 -> 50.4 | Helped |
| add auxiliary rotation prediction to TRADES | TRADES · CIFAR-10 adversarial robustness | robust accuracy: 1.22% | Helped |
| add auxiliary rotation prediction | 40-2 Wide ResNet · CIFAR-10 common corruptions | average accuracy across corruptions: 72.3% -> 76.9% | Helped |
| add auxiliary rotation prediction | No Correction classifier · CIFAR-10 label corruption | average error: 27.4 -> 21.8 | Helped |
| add auxiliary rotation prediction | No Correction classifier · CIFAR-100 label corruption | average error: 52.6 -> 47.4 | Helped |
| add auxiliary rotation prediction | GLC (5% Trusted) · CIFAR-10 label corruption | average error: 14.6 -> 10.5 | Helped |
| add auxiliary rotation prediction | GLC (10% Trusted) · CIFAR-10 label corruption | average error: 11.6 -> 9.6 | Helped |
| add auxiliary rotation prediction | GLC · CIFAR-100 label corruption | average error: 48.3 -> 43.2; 39.1 -> 36.8 | Helped |
| add auxiliary rotation prediction | one-class OOD detector · CIFAR-10 one-class OOD detection (mean over classes) | AUROC: 87.3 -> 90.1 | Helped |
| add translation prediction to RotNet (ablation) | ImageNet one-class OOD detector · ImageNet one-class OOD detection | AUROC: 65.3 -> 77.9 | Helped |
| add self-attention to RotNet (ablation) | ImageNet one-class OOD detector · ImageNet one-class OOD detection | AUROC: 65.3 -> 81.6 | Helped |
Closest papers by meaning
- Adversarial Robustness: From Self-Supervised Pre-Training to Fine-Tuning (CVPR 2020 · similarity 0.89)
- 3SAT: A Simple Self-Supervised Adversarial Training Framework (AAAI 2025 · similarity 0.88)
- Annealing Self-Distillation Rectification Improves Adversarial Training (ICLR 2024 · similarity 0.87)
- Self-Supervised Adversarial Training via Diverse Augmented Queries and Self-Supervised Double Perturbation (NeurIPS 2024 · similarity 0.87)
- Ditch the Denoiser: Emergence of Noise Robustness in Self-Supervised Learning from Data Curriculum (NeurIPS 2025 · similarity 0.85)
- Rethinking Robustness of Model Attributions (AAAI 2024 · similarity 0.85)
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.