Research map / Label-efficient and robust learning / Semi-supervised learning
Learning to Learn in a Semi-Supervised Fashion
Yun-Chun Chen, Chao-Te Chou, Yu-Chiang Frank Wang
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 semantics-oriented meta-learning stage on the full labeled training set | BoT · Market-1501 (fully supervised, all IDs) | Rank-1 (%): 94.5 -> 94.8 | Improved |
| add semantics-oriented meta-learning stage on the full labeled training set | AlignedReID · Market-1501 (fully supervised, all IDs) | Rank-1 (%): 89.2 -> 91.1 | Improved |
| add semantics-oriented meta-learning stage on the full labeled training set | AlignedReID · DukeMTMC-reID (fully supervised, all IDs) | Rank-1 (%): 79.3 -> 81.7 | Improved |
| add semantics-oriented meta-learning stage on the full labeled training set | BoT · DukeMTMC-reID (fully supervised, all IDs) | Rank-1 (%): 86.4 -> 86.6 | Improved |
| increase labeled category ratio from 25% to 75% (ablation) | semi-supervised retrieval · Car196 retrieval | Recall@1 (%): 54.5 -> 65.9 | Improved |
| increase labeled category ratio from 25% to 75% (ablation) | semi-supervised retrieval · CUB-200 retrieval | Recall@1 (%): 48.4 -> 51.0 | Improved |
| increase labeled ID ratio from 1/12 to 1/3 (ablation) | semi-supervised re-ID · Market-1501 (semi-supervised) | Rank-1 (%): 56.7 -> 80.3 | Improved |
| use meta-learned pseudo-labels instead of clustering-based pseudo labels | MVC · Market-1501 (semi-supervised, 1/3 IDs labeled) | Rank-1 (%): 72.2 -> 80.3 | Improved |
| use meta-learned pseudo-labels instead of clustering-based pseudo labels | MVC · DukeMTMC-reID (semi-supervised, 1/3 IDs labeled) | Rank-1 (%): 52.9 -> 67.5 | Improved |
| apply the meta-learning strategy to limited labeled data only | PyrNet · Market-1501 (fully supervised, 1/3 of IDs) | Rank-1 (%): 74.9 -> 77.0 | Improved |
| apply the meta-learning strategy to limited labeled data only | MGN · Market-1501 (fully supervised, 1/12 of IDs) | Rank-1 (%): 50.6 -> 53.4 | Improved |
| use 25% labeled categories with meta-learned pseudo labels instead of unsupervised Instance | Instance · Car196 retrieval | Recall@1 (%): 41.3 -> 54.5 | Improved |
Closest papers by meaning
- SimPLE: Similar Pseudo Label Exploitation for Semi-Supervised Classification (CVPR 2021 · similarity 0.91)
- InterLUDE: Interactions between Labeled and Unlabeled Data to Enhance Semi-Supervised Learning (ICML 2024 · similarity 0.90)
- Scaling Up Semi-supervised Learning with Unconstrained Unlabelled Data (AAAI 2024 · similarity 0.89)
- Learning with Partial-Label and Unlabeled Data: A Uniform Treatment for Supervision Redundancy and Insufficiency (ICML 2024 · similarity 0.89)
- CGMatch: A Different Perspective of Semi-supervised Learning (CVPR 2025 · similarity 0.88)
- OwMatch: Conditional Self-Labeling with Consistency for Open-World Semi-Supervised Learning (NeurIPS 2024 · similarity 0.87)
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.
