Research map / Label-efficient and robust learning / Semi-supervised learning
Open-World Semi-Supervised Learning
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 |
|---|---|---|---|
| Use ORCA’s uncertainty-adaptive margin instead of zero margin (ablation) | ORCA · CIFAR-100 | novel accuracy: 32.0 -> 43.0 | Helped |
| Use uncertainty-adaptive margin instead of fixed negative margin (ablation) | ORCA · ImageNet-100 | novel accuracy: 66.2 -> 72.1 | Helped |
| Add the supervised objective (ablation) | ORCA pairwise and regularized objective · CIFAR-100 | overall accuracy: 10.0 -> 48.1 | Helped |
| Add the regularization term (ablation) | ORCA supervised and pairwise objectives · CIFAR-100 | novel accuracy: 25.9 -> 43.0 | Helped |
| Use uncertainty-based margin rather than entropy-based margin estimation (ablation) | ORCA · CIFAR-100 | overall accuracy: 46.5 -> 48.1 | Helped |
| Use uncertainty-based adaptive margin rather than a linear margin schedule (ablation) | ORCA · CIFAR-100 | overall accuracy: 45.9 -> 48.1 | Helped |
| Pretrain on labeled and unlabeled data rather than only labeled data (ablation) | ORCA · CIFAR-100 | overall accuracy: 45.2 -> 48.1 | Helped |
| Use RotationNet rather than SimCLR pretraining (ablation) | ORCA · CIFAR-100 | overall accuracy: 48.1 -> 51.4 | Helped |
| Increase the initialized number of novel classes from 100 to 400 (ablation) | ORCA · CIFAR-100 | overall accuracy: 48.1 -> 43.4 | Hurt |
| Increase uncertainty regularizer lambda from 0.6 to 1.4 (ablation) | ORCA · CIFAR-100 | novel accuracy: 42.6 -> 43.5 | Helped |
| Add maximum-entropy regularization (ablation) | ORCA on unbalanced CIFAR-10 · unbalanced CIFAR-10 | novel accuracy: 61.4 -> 82.9 | Helped |
| Increase the proportion of novel classes in the unlabeled set (ablation) | ORCA · CIFAR-100 | novel accuracy | Hurt |
Closest papers by meaning
- OwMatch: Conditional Self-Labeling with Consistency for Open-World Semi-Supervised Learning (NeurIPS 2024 · similarity 0.89)
- Targeted Representation Alignment for Open-World Semi-Supervised Learning (CVPR 2024 · similarity 0.88)
- CGMatch: A Different Perspective of Semi-supervised Learning (CVPR 2025 · similarity 0.87)
- Revisiting Semi-Supervised Learning in the Era of Foundation Models (NeurIPS 2025 · similarity 0.87)
- Unsupervised Learning for Class Distribution Mismatch (ICML 2025 · similarity 0.87)
- A Unified Framework for Heterogeneous Semi-supervised Learning (CVPR 2025 · 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.