Research map / Learning theory and algorithms / Prediction and decision losses
Learning Algorithms for Active 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 |
|---|---|---|---|
| Learn an end-to-end active selection policy rather than use random label selection | Active MN · Omniglot 10-way 1-shot | accuracy: 67.3%$_{\pm 0.10}$ -> 94.3%$_{\pm 0.24}$ | Helped |
| Use learned selection instead of Min-Max-Cos selection | Active MN · MovieLens-20M cold-start | RMSE: 2.5% improvement after 10 ratings against the best baseline | Helped |
| Add attention temperature to fast prediction (ablation) | Active MN · Omniglot 10-way 1-shot | accuracy: 86.0 -> 94.5 | Helped |
| Reduce matching steps from 3 to 2 or 1 (ablation) | Active MN · Omniglot 10-way 1-shot | accuracy | No clear effect |
| Remove the context-sensitive encoder (ablation) | Active MN · Omniglot 10-way 1-shot | accuracy | No clear effect |
| Remove controller-item similarity features from selection (ablation) | Active MN · Active learning selection | prediction performance | Hurt |
| Remove item-item similarity features from selection (ablation) | Active MN · Active learning selection | prediction performance | Hurt |
| Remove the sharpening term from fast prediction (ablation) | Active MN · Fast prediction | prediction performance | Hurt |
| Use a data-driven personalized policy instead of a fixed diversity heuristic | MovieLens active learner · MovieLens-20M cold-start | RMSE | Helped |
| Use Min-Max-Cos selection instead of a learned policy | Active MN · Omniglot | accuracy | No clear effect |
Closest papers by meaning
- Learning Active Learning from Data (NeurIPS 2017 · similarity 0.89)
- AutoAL: Automated Active Learning with Differentiable Query Strategy Search (ICML 2025 · similarity 0.84)
- Universal Rates for Active Learning (NeurIPS 2024 · similarity 0.84)
- Neural Active Learning Beyond Bandits (ICLR 2024 · similarity 0.84)
- Transformers are almost optimal metalearners for linear classification (NeurIPS 2025 · similarity 0.84)
- A Cross-Domain Benchmark for Active Learning (NeurIPS 2024 · 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.