Research map / Label-efficient and robust learning / Noisy labels
Learning to Purify Noisy Labels via Meta Soft Label Corrector
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 |
|---|---|---|---|
| Use MSLC instead of U-correction | CIFAR-10 symmetric noise 0.8 classification · CIFAR-10 symmetric noise 0.8 | test accuracy (Best, %): 68.23 -> 69.87 | Helped |
| Use MSLC instead of MW-Net | CIFAR-100 symmetric noise 0.8 classification · CIFAR-100 symmetric noise 0.8 | test accuracy (Best, %): 19.62 -> 24.32 | Helped |
| Use MSLC instead of MW-Net | Clothing1M classification · Clothing1M test | accuracy (%): 73.72 -> 74.02 | Helped |
| Set β to 0 instead of learning β (ablation) | MSLC · CIFAR-10 symmetric noise 0.4 | test accuracy (Best, %): 91.27 -> 89.84 | Hurt |
| Set β to 0.2 instead of learning β (ablation) | MSLC · CIFAR-10 symmetric noise 0.4 | test accuracy (Best, %): 91.27 -> 90.49 | Hurt |
| Set β to 0.4 instead of learning β (ablation) | MSLC · CIFAR-10 symmetric noise 0.4 | test accuracy (Best, %): 91.27 -> 91.04 | Hurt |
| Set β to 0.6 instead of learning β (ablation) | MSLC · CIFAR-10 symmetric noise 0.4 | test accuracy (Best, %): 91.27 -> 90.34 | Hurt |
| Set β to 0.8 instead of learning β (ablation) | MSLC · CIFAR-10 symmetric noise 0.4 | test accuracy (Best, %): 91.27 -> 89.46 | Hurt |
| Set β to 0 instead of learning β (ablation) | MSLC · CIFAR-100 symmetric noise 0.4 | test accuracy (Best, %): 68.84 -> 67.42 | Hurt |
| Set β to 0.2 instead of learning β (ablation) | MSLC · CIFAR-100 symmetric noise 0.4 | test accuracy (Best, %): 68.84 -> 68.52 | Hurt |
| Set β to 0.4 instead of learning β (ablation) | MSLC · CIFAR-100 symmetric noise 0.4 | test accuracy (Best, %): 68.84 -> 68.25 | Hurt |
| Set β to 0.6 instead of learning β (ablation) | MSLC · CIFAR-100 symmetric noise 0.4 | test accuracy (Best, %): 68.84 -> 67.13 | Hurt |
Closest papers by meaning
- Learning From Noisy Labels With Decoupled Meta Label Purifier (CVPR 2023 · similarity 0.93)
- Meta Label Correction for Noisy Label Learning (AAAI 2021 · similarity 0.93)
- Revisiting Interpolation for Noisy Label Correction (AAAI 2025 · similarity 0.88)
- Adaptive Integration of Partial Label Learning and Negative Learning for Enhanced Noisy Label Learning (AAAI 2024 · similarity 0.87)
- Mitigating Label Noise through Data Ambiguation (AAAI 2024 · similarity 0.87)
- $\epsilon$-Softmax: Approximating One-Hot Vectors for Mitigating Label Noise (NeurIPS 2024 · similarity 0.86)
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.