Research map / Label-efficient and robust learning / Noisy labels
Augmentation Strategies for Learning With Noisy Labels
Check 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 |
|---|---|---|---|
| Separate weak augmentation for loss analysis from strong augmentation for optimization | DivideMix · CIFAR-10, 90% symmetric noise | Best accuracy: 76.00 -> 91.88 | Helped |
| Separate weak augmentation for loss analysis from strong augmentation for optimization | DivideMix · CIFAR-100, 90% symmetric noise | Best accuracy: 31.50 -> 41.20 | Helped |
| Use weakly augmented rather than strongly augmented warm-up (ablation) | DivideMix with AugDesc-WS · CIFAR-10, 90% symmetric noise | Best accuracy: 35.3 -> 91.9 | Helped |
| Use weakly augmented rather than strongly augmented warm-up (ablation) | DivideMix with AugDesc-WS · CIFAR-100, 90% symmetric noise | Best accuracy: 17.3 -> 41.2 | Helped |
| Use strong augmentation for loss analysis instead of weak augmentation (ablation) | DivideMix AugDesc · CIFAR-10, 90% symmetric noise | Best accuracy: 91.88 -> 81.77 | Hurt |
| Use weak augmentation for optimization instead of strong augmentation (ablation) | DivideMix AugDesc · CIFAR-10, 90% symmetric noise | Best accuracy: 91.88 -> 36.05 | Hurt |
| Replace AutoAugment with RandAugment (ablation) | DivideMix AugDesc-WS · CIFAR-10, 90% symmetric noise | Best accuracy: 91.9 -> 89.6 | Hurt |
| Replace AutoAugment with RandAugment (ablation) | DivideMix AugDesc-WS · CIFAR-100, 90% symmetric noise | Best accuracy: 41.2 -> 36.8 | Hurt |
| Add AugDesc augmentation | Cross Entropy · CIFAR-10, 20% symmetric noise | Best accuracy: 86.8 -> 89.9 | Helped |
| Add AugDesc augmentation | Co-Teaching+ · CIFAR-100, 20% symmetric noise | Best accuracy: 26.2 -> 25.6 | Hurt |
| Add AugDesc augmentation | M-DYR-H · CIFAR-100, 20% symmetric noise | Best accuracy: 68.2 -> 73.0 | Helped |
| Add AugDesc augmentation | DivideMix · CIFAR-100, 20% symmetric noise | Best accuracy: 77.3 -> 79.5 | Helped |
Closest papers by meaning
- Joint Optimization Framework for Learning With Noisy Labels (CVPR 2018 · similarity 0.90)
- Distilling Effective Supervision From Severe Label Noise (CVPR 2020 · similarity 0.90)
- Understanding and Utilizing Deep Neural Networks Trained with Noisy Labels (ICML 2019 · similarity 0.90)
- Handling Label Noise via Instance-Level Difficulty Modeling and Dynamic Optimization (NeurIPS 2025 · similarity 0.90)
- Learning with Structural Labels for Learning with Noisy Labels (CVPR 2024 · similarity 0.89)
- $\epsilon$-Softmax: Approximating One-Hot Vectors for Mitigating Label Noise (NeurIPS 2024 · similarity 0.89)
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.