Research map / Label-efficient and robust learning / Noisy labels
Centrality and Consistency: Two-Stage Clean Samples Identification for Learning with Instance-Dependent Noisy Labels
Ganlong Zhao, Guanbin Li, Yipeng Qin, Feng Liu, Yizhou Yu
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 |
|---|---|---|---|
| add feature clustering (stage 1) to the CE baseline (ablation) | CE baseline · CIFAR-100 (eta=0.6) | accuracy (%): 25.68 -> 53.60 | Improved |
| add consistency classification (stage 2) to the CE baseline (ablation) | CE baseline · CIFAR-100 (eta=0.6) | accuracy (%): 25.68 -> 51.41 | Improved |
| add consistency classification on top of feature clustering (ablation) | Ours · CIFAR-100 (eta=0.6) | accuracy (%): 53.60 -> 59.40 | Improved |
| add feature clustering on top of consistency classification (ablation) | Ours · CIFAR-100 (eta=0.6) | accuracy (%): 51.41 -> 59.40 | Improved |
| add feature clustering (stage 1) to the CE baseline (ablation) | CE baseline · Clothing1M | accuracy (%): 68.94 -> 73.32 | Improved |
| add consistency classification (stage 2) to the CE baseline (ablation) | CE baseline · Clothing1M | accuracy (%): 68.94 -> 74.26 | Improved |
| add consistency classification on top of feature clustering (ablation) | Ours · Clothing1M | accuracy (%): 73.32 -> 75.40 | Improved |
| add feature clustering on top of consistency classification (ablation) | Ours · Clothing1M | accuracy (%): 74.26 -> 75.40 | Improved |
| replace DivideMix with the proposed method | DivideMix · Inst. CIFAR10 (eta=0.6) | accuracy (%): 85.50 -> 94.95 | Improved |
| replace DivideMix with the proposed method | DivideMix · Inst. CIFAR100 (eta=0.6) | accuracy (%): 46.72 -> 59.40 | Improved |
| replace DivideMix with the proposed method | DivideMix · Inst. CIFAR10 (eta=0.4) | accuracy (%): 95.07 -> 94.97 | Hurt |
| replace DivideMix with the proposed method | DivideMix · Clothing1M | accuracy (%): 74.76 -> 75.40 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- DivideMix — DivideMix: Learning with Noisy Labels as Semi-supervised Learning (ICLR 2020) · better on 8 of 9 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Readiness-Aware Sample Selection for Noisy Labels with Class Imbalance (NeurIPS 2026 · similarity 0.91)
- Uncertainty-Aware Learning against Label Noise on Imbalanced Datasets (AAAI 2022 · similarity 0.90)
- Foster Adaptivity and Balance in Learning with Noisy Labels (ECCV 2024 · similarity 0.90)
- Noise-Robust Learning from Multiple Unsupervised Sources of Inferred Labels (AAAI 2022 · similarity 0.89)
- Sample-wise Label Confidence Incorporation for Learning with Noisy Labels (ICCV 2023 · similarity 0.89)
- Estimating Noisy Class Posterior with Part-level Labels for Noisy Label Learning (CVPR 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.
