Research map / Label-efficient and robust learning / Noisy labels
CS-Isolate: Extracting Hard Confident Examples by Content and Style Isolation
Yexiong Lin, Yu Yao, Xiaolong Shi, Mingming Gong, Xu Shen, Dong Xu, Tongliang Liu
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 content/style isolation (CS-Isolate) | DivideMix · CIFAR-10N Worst | test accuracy: 92.48 -> 94.28 | Improved |
| add content/style isolation (CS-Isolate) | DivideMix · CIFAR-10 IDN-0.4 | test accuracy: 94.16 -> 95.54 | Improved |
| add content/style isolation (CS-Isolate) | DivideMix · CIFAR-100 IDN-0.4 | test accuracy: 70.50 -> 73.11 | Improved |
| add content/style isolation (CS-Isolate) | DivideMix · Fashion-MNIST IDN-0.4 | test accuracy: 92.28 -> 94.40 | Improved |
| add content/style isolation (CS-Isolate) | DivideMix · Clothing1M | test accuracy: 74.76 -> 74.92 | Improved |
| add content/style isolation (CS-Isolate) | DivideMix · CIFAR-10 Pair-45% | test accuracy: 70.86 -> 87.54 | Improved |
| add content/style isolation (CS-Isolate) | DivideMix · CIFAR-10 Sym-80% | test accuracy: 92.43 -> 92.57 | No clear effect |
| add content/style isolation (CS-Isolate) | DivideMix · CIFAR-10N Worst | recall of confident examples: 0.966 -> 0.980 | Improved |
| add content/style isolation (CS-Isolate) | DivideMix · CIFAR-10N Worst | precision of confident examples: 0.951 -> 0.956 | No clear effect |
| add content/style isolation (CS-Isolate) | Co-Teaching · CIFAR-10N Worst | test accuracy: 82.04 -> 83.93 | No clear effect |
| add content/style isolation (CS-Isolate) | Me-Momentum · CIFAR-10N Worst | test accuracy: 84.21 -> 86.95 | Improved |
| add content/style isolation (CS-Isolate) | Me-Momentum · CIFAR-10N Worst | recall ratio of confident examples: 92.02 -> 95.76 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- DivideMix — DivideMix: Learning with Noisy Labels as Semi-supervised Learning (ICLR 2020) · better on 17 of 17 shared results
- Me-Momentum — Me-Momentum: Extracting Hard Confident Examples From Noisily Labeled Data (ICCV 2021) · better on 5 of 5 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
- Enhancing Sample Selection Against Label Noise by Cutting Mislabeled Easy Examples (NeurIPS 2025 · similarity 0.85)
- Learning with Complementary Labels Revisited: The Selected-Completely-at-Random Setting Is More Practical (ICML 2024 · similarity 0.84)
- Enhanced Sample Selection with Confidence Tracking: Identifying Correctly Labeled Yet Hard-to-Learn Samples in Noisy Data (AAAI 2025 · similarity 0.84)
- CGMatch: A Different Perspective of Semi-supervised Learning (CVPR 2025 · similarity 0.84)
- Learning with Noisy Labels Using Hyperspherical Margin Weighting (AAAI 2024 · similarity 0.84)
- NoiseGPT: Label Noise Detection and Rectification through Probability Curvature (NeurIPS 2024 · similarity 0.84)
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.
