Research map / Label-efficient and robust learning / Noisy labels
How To Prevent the Continuous Damage of Noises To Model Training?
Xiaotian Yu, Yang Jiang, Tianqi Shi, Zunlei Feng, Yuexuan Wang, Mingli Song, Li Sun
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 |
|---|---|---|---|
| Apply gradient switching with gradient-direction pools and semi-supervised learning | CIFAR-10 noisy-label training · CIFAR-10 symmetric noise 40% | accuracy: 92.85 -> 94.20 | Improved |
| Apply gradient switching with gradient-direction pools and semi-supervised learning | CIFAR-100 noisy-label training · CIFAR-100 symmetric noise 80% | accuracy: 45.82 -> 55.04 | Improved |
| Apply gradient switching with gradient-direction pools and semi-supervised learning | Clothing1M training · Clothing1M | accuracy: 74.59 -> 74.88 | Improved |
| Apply gradient switching with gradient-direction pools and semi-supervised learning | WebVision training · WebVision | Top5 accuracy: 91.60 -> 93.09 | Improved |
| Apply gradient switching with gradient-direction pools and semi-supervised learning | ILSVRC12 validation · ILSVRC12 | Top1 accuracy: 75.23 -> 75.18 | Hurt |
| Add predicted labels to direction-pool updates (ablation) | GSS-SB · CIFAR-10 with 40% noisy labels | test accuracy | Improved |
| Add random labels to direction-pool updates (ablation) | GSS-SB · CIFAR-10 with 40% noisy labels | test accuracy | Improved |
| Use two branches instead of a single branch (ablation) | GSS · CIFAR-10 with 40% and 60% noisy labels | accuracy | Improved |
| Add semi-supervised learning to the dual-branch GSS (ablation) | GSS-DB · CIFAR-10 with 40% and 60% noisy labels | accuracy | 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 18 of 19 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
- Handling Label Noise via Instance-Level Difficulty Modeling and Dynamic Optimization (NeurIPS 2025 · similarity 0.87)
- How Does Label Noise Gradient Descent Improve Generalization in the Low SNR Regime? (NeurIPS 2025 · similarity 0.87)
- Noise against noise: stochastic label noise helps combat inherent label noise (ICLR 2021 · similarity 0.86)
- Early Stopping Against Label Noise Without Validation Data (ICLR 2024 · similarity 0.86)
- Understanding and Mitigating the Label Noise in Pre-training on Downstream Tasks (ICLR 2024 · similarity 0.85)
- On the Role of Label Noise in the Feature Learning Process (ICML 2025 · 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.
