Research map / Label-efficient and robust learning / Recognition and long-tail
Actively Testing Your Model While It Learns: Realizing Label-Efficient Learning in Practice
Dayou Yu, Weishi Shi, Qi Yu
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 |
|---|---|---|---|
| ATL-NF importance-weighted test sampling instead of ARE quizzes | ATL-NF · MNIST | estimation error x1e-3 (round 20): 8.31 -> 1.32 | Improved |
| ATL-NF instead of adapted AT integrate | ATL-NF · CIFAR10 | estimation error x1e-3 (round 20): 38.9 -> 6.20 | Improved |
| ATL-NF instead of adapted ASE integrate | ATL-NF · FashionMNIST | estimation error x1e-3 (round 20): 8.42 -> 3.04 | Improved |
| add active feedback of test samples to training (ATL vs ATL-NF) (ablation) | ATL · MNIST | hold-out test risk (round 20): 0.22 -> 0.19 | Improved |
| add active feedback (ATL vs ATL-NF) (ablation) | ATL · FashionMNIST | hold-out test risk (round 20): 0.56 -> 0.51 | Improved |
| add active feedback (ATL vs ATL-NF) (ablation) | ATL · CIFAR10 | hold-out test risk (round 20): 1.55 -> 1.53 | No clear effect |
| active feedback instead of random feedback (ablation) | ATL · MNIST | hold-out test risk (round 20): 0.21 -> 0.19 | Improved |
| active feedback instead of random feedback (ablation) | ATL · MNIST | estimation error x1e-3 (round 20): 7.62 -> 1.87 | Improved |
| active feedback instead of random feedback (ablation) | ATL · CIFAR10 | estimation error x1e-3 (round 20): 24.8 -> 6.57 | Improved |
| active feedback early in AL (round 4) (ablation) | ATL · MNIST | estimation error x1e-3 (round 4): 26.8 -> 14.6 | Improved |
| larger feedback size (83% vs 20%) (ablation) | ATL · MNIST | hold-out test risk (round 20): 0.24 -> 0.20 | Improved |
| larger feedback size (83% vs 20%) (ablation) | ATL · CIFAR10 | hold-out test risk (round 20): 1.63 -> 1.57 | Mixed |
Closest papers by meaning
- Scaling Up Active Testing to Large Language Models (NeurIPS 2025 · similarity 0.90)
- Prediction-Powered Active Testing (NeurIPS 2026 · similarity 0.89)
- Towards Cost-Effective Learning: A Synergy of Semi-Supervised and Active Learning (CVPR 2025 · similarity 0.87)
- Active Statistical Inference (ICML 2024 · similarity 0.86)
- Instance-wise Supervision-level Optimization in Active Learning (CVPR 2025 · similarity 0.85)
- Active Learning with LLMs for Partially Observed and Cost-Aware Scenarios (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.
