Research map / Label-efficient and robust learning / Recognition and long-tail
Active Learning for Open-Set Annotation
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 LfOSA instead of Random sampling | Active learning for open-set annotation · CIFAR10, CIFAR100; mismatch ratios 20%, 30%, and 40% | selection recall: 68.8%, 53.4%, and 35.7% (CIFAR10); 34.3%, 26.7%, and 20.5% (CIFAR100) | Helped |
| Use LfOSA instead of the compared active learning methods | Active learning for open-set annotation · CIFAR10, CIFAR100, and Tiny-Imagenet; mismatch ratios 20%, 30%, and 40% | selection precision | Helped |
| Use LfOSA instead of other active learning methods | CIFAR100 · CIFAR100; mismatch ratios 20%, 30%, and 40% | classification accuracy: about 20%, 15% and 12% | Helped |
| Use LfOSA instead of Uncertainty or BALD sampling | Active learning for open-set annotation · Most tested settings | classification accuracy | Helped |
| Add a low-temperature detector loss instead of using temperature 1 or 2 (ablation) | LfOSA · CIFAR100; 20% mismatch ratio | selection recall: 1.89% | Helped |
| Add a low-temperature detector loss instead of using temperature 1 or 2 (ablation) | LfOSA · CIFAR100; 20% mismatch ratio | classification accuracy: 3.1% | Helped |
| Decouple detection and classification with separate detector and classifier networks (ablation) | LfOSA · CIFAR100; 20% mismatch ratio | classification accuracy | Helped |
| Use a separate classifier rather than removing the classifier (ablation) | LfOSA · CIFAR100; 20% mismatch ratio | classification accuracy | Helped |
| Train the detector with the invalid set rather than omitting it (ablation) | LfOSA · CIFAR100; 20% mismatch ratio | classification accuracy | Helped |
Closest papers by meaning
- Inconsistency-Based Data-Centric Active Open-Set Annotation (AAAI 2024 · similarity 0.87)
- Rethinking Epistemic and Aleatoric Uncertainty for Active Open-Set Annotation: An Energy-Based Approach (CVPR 2025 · similarity 0.87)
- Entropic Open-Set Active Learning (AAAI 2024 · similarity 0.85)
- Instance-wise Supervision-level Optimization in Active Learning (CVPR 2025 · similarity 0.84)
- Towards Cost-Effective Learning: A Synergy of Semi-Supervised and Active Learning (CVPR 2025 · similarity 0.84)
- Joint Out-of-Distribution Filtering and Data Discovery Active Learning (CVPR 2025 · 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.