Research map / Label-efficient and robust learning / Semi-supervised learning
Semi-Supervised Learning under Class Distribution Mismatch
Read the paperCheck 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 |
|---|---|---|---|
| UASD unlabelled-data training vs supervised only | WRN-28-2 · CIFAR100 (50% mismatch) | test error rate: 39.79 -> 35.93 | Helped |
| UASD vs supervised only | WRN-28-2 · CIFAR100 + TinyImageNet | test error rate: 48.31 -> 42.83 | Helped |
| pseudo-label SSL under class mismatch | WRN-28-2 · CIFAR100 (50% mismatch) | test error rate: 39.79 -> 43.30 | Hurt |
| VAT under class mismatch | WRN-28-2 · TinyImageNet (50% mismatch) | test error rate: 61.64 -> 63.75 | Hurt |
| Mean-Teacher under class mismatch | WRN-28-2 · CIFAR100 + TinyImageNet | test error rate: 48.31 -> 49.67 | Hurt |
| soft targets instead of one-hot hard targets (ablation) | UASD · CIFAR10 6-class mismatch | test error rate | Helped |
| add OOD filter (ablation) | UASD · CIFAR10 6-class mismatch | test error rate | Helped |
| larger accumulative ensemble (1000+ vs 10+/100+) (ablation) | UASD · CIFAR10 6-class mismatch | test error rate | Helped |
Closest papers by meaning
- On Non-Random Missing Labels in Semi-Supervised Learning (ICLR 2022 · similarity 0.89)
- Open-World Semi-Supervised Learning (ICLR 2022 · similarity 0.89)
- CDMAD: Class-Distribution-Mismatch-Aware Debiasing for Class-Imbalanced Semi-Supervised Learning (CVPR 2024 · similarity 0.88)
- A Realistic Evaluation of Semi-Supervised Learning for Fine-Grained Classification (CVPR 2021 · similarity 0.88)
- DC-SSL: Addressing Mismatched Class Distribution in Semi-Supervised Learning (CVPR 2022 · similarity 0.88)
- OwMatch: Conditional Self-Labeling with Consistency for Open-World Semi-Supervised Learning (NeurIPS 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.