Research map / Label-efficient and robust learning / Semi-supervised learning
All Labels Are Not Created Equal: Enhancing Semi-Supervision via Label Grouping and Co-Training
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 |
|---|---|---|---|
| Use SemCo (μ=7) instead of FixMatch (μ=7) | CIFAR-10 · CIFAR-10, 250 labels | error rate: 4.55$\pm$0.12 -> 5.12$\pm$0.27 | Hurt |
| Use SemCo instead of FixMatch | CIFAR-10 · CIFAR-10, 4000 labels | error rate: 4.49$\pm$0.05 -> 3.80$\pm$0.08 | Helped |
| Use SemCo instead of FixMatch | CIFAR-100 · CIFAR-100, 2500, 4000, and 10000 labels | error rate: 33.64$\pm$0.07 -> 31.93$\pm$0.01; 31.27$\pm$1.30 -> 28.61$\pm$0.23; 26.13$\pm$0.18 -> 24.45$\pm$0.12 | Helped |
| Use SemCo instead of FixMatch | Mini-ImageNet · Mini-ImageNet, 1000, 4000, and 10000 labels | error rate: 60.97$\pm$0.31 -> 55.35$\pm$0.71; 49.79$\pm$0.11 -> 46.01$\pm$0.93; 44.27$\pm$0.23 -> 41.25$\pm$0.76 | Helped |
| Use SemCo instead of FixMatch | CUB-200 · CUB-200, 1000 and 2000 labels | error rate: 84.35 -> 79.44; 72.15 -> 66.76 | Helped |
| Use SemCo instead of FixMatch | DomainNet Real · DomainNet Real, 6900 and 10350 labels | error rate: 41.34 -> 35.32; 39.04 -> 32.89 | Helped |
| Add label grouping (ablation) | SemCo · Mini-ImageNet, 1000 labels | error rate: 59.60 -> 55.35 | Helped |
| Add label grouping (ablation) | SemCo · CIFAR-100, 2500 labels | error rate: 33.09 -> 31.93 | Helped |
| Add co-training (ablation) | SemCo · Mini-ImageNet, 1000 labels | error rate: 60.39 -> 55.35 | Helped |
| Add co-training (ablation) | SemCo · CIFAR-100, 2500 labels | error rate: 33.19 -> 31.93 | Helped |
| Add label grouping and co-training (ablation) | SemCo · Mini-ImageNet, 1000 labels | error rate: 62.16 -> 55.35 | Helped |
| Add label grouping and co-training (ablation) | SemCo · CIFAR-100, 2500 labels | error rate: 34.25 -> 31.93 | Helped |
Closest papers by meaning
- SimPLE: Similar Pseudo Label Exploitation for Semi-Supervised Classification (CVPR 2021 · similarity 0.92)
- Curriculum Labeling: Revisiting Pseudo-Labeling for Semi-Supervised Learning (AAAI 2021 · similarity 0.92)
- Learning with Partial-Label and Unlabeled Data: A Uniform Treatment for Supervision Redundancy and Insufficiency (ICML 2024 · similarity 0.90)
- InterLUDE: Interactions between Labeled and Unlabeled Data to Enhance Semi-Supervised Learning (ICML 2024 · similarity 0.90)
- Label-Efficient Dataset Pruning via Semi-Supervised Pseudo-Labeling (NeurIPS 2026 · similarity 0.89)
- CGMatch: A Different Perspective of Semi-supervised Learning (CVPR 2025 · 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.