Research map / Label-efficient and robust learning / Semi-supervised learning
BAL: Balancing Diversity and Novelty for Active Learning
Jingyao Li, Pengguang Chen, Shaozuo Yu, Shu Liu, Jiaya Jia
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 |
|---|---|---|---|
| use rotation-angle pretext instead of SimCLR (ablation) | BAL · CIFAR-10 (10% labeled) | accuracy (%): 50.8 -> 68.7 | Improved |
| use rotation-angle labels instead of image labels (ablation) | BAL · CIFAR-10 (10% labeled) | accuracy (%): 65.0 -> 68.7 | Improved |
| use CDD instead of distance to nearest cluster center (ablation) | BAL · CIFAR-10 (10% labeled) | accuracy (%): 66.6 -> 68.7 | Improved |
| sort CDD ascending instead of descending (ablation) | BAL · CIFAR-10 (10% labeled) | accuracy (%): 57.5 -> 70.5 | Improved |
| use confidence sampler instead of entropy (ablation) | BAL · CIFAR-10 (10% labeled) | accuracy (%): 53.4 -> 68.7 | Improved |
| use confidence sampler instead of random (ablation) | BAL · CIFAR-10 (10% labeled) | accuracy (%): 68.2 -> 68.7 | Improved |
| use adaptive sub-pools instead of uniform sub-pools (ablation) | BAL · Caltech-101 (45% labeled) | accuracy (%): 75.0 -> 76.4 | Improved |
| use CDD instead of pretext task loss as indicator (ablation) | PT4AL · CIFAR-10 | accuracy (%): +1.36% | Improved |
| use BAL instead of PT4AL under large budget (80% labeled) | BAL · Caltech-101 (80% labeled) | accuracy (%): 82.6 -> 92.8 | Improved |
| use BAL instead of PT4AL under large budget (80% labeled) | BAL · CIFAR-10 (80% labeled, from scratch) | accuracy (%): 91.4 -> 85.0 | Hurt |
| use BAL instead of ActiveFT under small budget (4% labeled) | BAL · SVHN (4% labeled) | accuracy (%): 55.3 -> 81.3 | Improved |
| use BAL instead of entropy under small budget (4% labeled) | BAL · CIFAR-10 (4% labeled) | accuracy (%): 25.7 -> 36.7 | Improved |
Closest papers by meaning
- Instance-wise Supervision-level Optimization in Active Learning (CVPR 2025 · similarity 0.87)
- Towards Cost-Effective Learning: A Synergy of Semi-Supervised and Active Learning (CVPR 2025 · similarity 0.87)
- Improved Algorithm for Deep Active Learning under Imbalance via Optimal Separation (ICML 2025 · similarity 0.86)
- Uncertainty Herding: One Active Learning Method for All Label Budgets (ICLR 2025 · similarity 0.86)
- Composite Active Learning: Towards Multi-Domain Active Learning with Theoretical Guarantees (AAAI 2024 · similarity 0.85)
- A Cross-Domain Benchmark for Active Learning (NeurIPS 2024 · similarity 0.85)
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.
