Research map / Label-efficient and robust learning / Recognition and long-tail
Deep active learning
Selects informative samples for labeling using ensembles, submodular measures and supervision-level optimization.
76 accepted papers at ICML, NeurIPS, ICLR, CVPR and AAAI (2016–2026) · an approach in Recognition and long-tail, keyword cluster “label · classification · classifier”
State History Tried Checklist Preprints Who Export
Where it stands
Not enough benchmark data Fewer than four papers report the same benchmark, so there is no record to follow.
23 papers in 2023–24 and 10 in 2025–26 (at ICML, NeurIPS, ICLR, CVPR and AAAI); its share of the topic shrank, from 1.8% to 0.8%.
How this is decided
A benchmark is tracked when at least four papers of this approach report their own method on the same dataset and metric in at least two different years. Its record is the best number reported up to 2024 compared with the best reported in 2025–26: still improving means at least 1% better, flat means at most 0.2% better with at least two papers from 2025–26 trying. Numbers on a different scale from the rest (a fraction among percentages) and far outliers are left out. Results come from different papers with different setups, so they are signals, not head-to-head comparisons. The share of papers is counted at ICML, NeurIPS, ICLR, CVPR and AAAI.
How it got here
Most cited paper of each year
- 2016 · Search Improves Label for Active Learning (NeurIPS 2016 · 4 citations)
- 2017 · Predicting Medications from Diagnostic Codes with Recurrent Neural Networks (ICLR 2017 · 44 citations)
- 2018 · Active Learning for Convolutional Neural Networks: A Core-Set Approach (ICLR 2018 · 678 citations)
- 2019 · Self-Paced Active Learning: Query the Right Thing at the Right Time (AAAI 2019 · 81 citations)
- 2020 · State-Relabeling Adversarial Active Learning (CVPR 2020 · 130 citations)
- 2021 · Task-Aware Variational Adversarial Active Learning (CVPR 2021 · 140 citations)
- 2022 · Active Learning by Feature Mixing (CVPR 2022 · 120 citations)
- 2023 · Active Finetuning: Exploiting Annotation Budget in the Pretraining-Finetuning Paradigm (CVPR 2023 · 30 citations)
- 2024 · Entropic Open-Set Active Learning (AAAI 2024 · 24 citations)
- 2025 · Rethinking Epistemic and Aleatoric Uncertainty for Active Open-Set Annotation: An Energy-Based Approach (CVPR 2025 · 3 citations)
- 2026 · Learning-To-Measure: In-Context Active Feature Acquisition (ICML 2026 · 0 citations)
Earliest papers here
- Search Improves Label for Active Learning (NeurIPS 2016 · 4 citations)
- Predicting Medications from Diagnostic Codes with Recurrent Neural Networks (ICLR 2017 · 44 citations)
- Querying Partially Labelled Data to Improve a K-nn Classifier (AAAI 2017 · 3 citations)
What has been tried
67 papers here report 650 measured changes (a component added, removed or swapped, compared with a baseline): helped 67%, no clear effect 12%, hurt 16%, mixed 6%. Papers mostly publish changes that helped, so "helped" is over-represented.
Ablations in recent papers
| What was changed | Measured on | Result | Effect | Paper |
|---|---|---|---|---|
| Combine closed-set entropy and distance-based entropy in the query strategy | CIFAR-100, 20% mismatch ratio · classification accuracy | Helped | Entropic Open-Set Active Learning (AAAI 2024 · 24 citations) | |
| add evidence fusion rather than Max-prediction | All · top-1 success rate | 62.9 -> 64.4 | Helped | Evidential Active Recognition: Intelligent and Prudent Open-World Embodied Perception (CVPR 2024 · 12 citations) |
| Add the Distributed Calibration Module | CIFAR10 (0.1%, 0.2%) · Top-1 accuracy | 47.1 -> 59.5; 64.5 -> 70.6 | Helped | ActiveDC: Distribution Calibration for Active Finetuning (CVPR 2024 · 6 citations) |
| Use energy-based epistemic uncertainty with energy-based aleatoric uncertainty, rather than either alone | CIFAR-100, 40% mismatch ratio · test performance | Helped | Rethinking Epistemic and Aleatoric Uncertainty for Active Open-Set Annotation: An Energy-Based Approach (CVPR 2025 · 3 citations) | |
| replace IDC denoising with DB denoising | CIFAR100 2% · Top-1 accuracy | 43.48 -> 41.06 (−2.42) | Hurt | Boundary Matters: A Bi-Level Active Finetuning Method (NeurIPS 2024 · 3 citations) |
| Remove instance-wise uncertainty from VCR calculation | CIFAR100 · accuracy | Hurt | Instance-wise Supervision-level Optimization in Active Learning (CVPR 2025 · 2 citations) | |
| use CLIP instead of ImageNet-pretrained ResNet features | CIFAR10 and CIFAR100 · accuracy | 97.48 ± 0.26 -> 98.15 ± 0.14; 60.75 ± 0.58 -> 63.68 ± 0.67 | Helped | Inconsistency-Based Data-Centric Active Open-Set Annotation (AAAI 2024 · 2 citations) |
| cluster raw data instead of classifier embeddings | Diverging Sine · rank | Hurt | A Cross-Domain Benchmark for Active Learning (NeurIPS 2024 · 2 citations) | |
| Use temperature scaling for the uncertainty function | CIFAR-10, all budget regimes · ΔAcc | Helped | Uncertainty Herding: One Active Learning Method for All Label Budgets (ICLR 2025 · 0 citations) | |
| Use GradCAM saliency maps instead of Deep Taylor | CheXpert, 10% training data · AUC | 72.24 -> 71.63 | Hurt | Combining Graph Transformers Based Multi-Label Active Learning and Informative Data Augmentation for Chest Xray Classification (AAAI 2024 · 0 citations) |
| Use CAMS instead of the variance-based query strategy | CIFAR10 · query cost | 71% | Helped | Contextual Active Model Selection (NeurIPS 2024 · 0 citations) |
| Replace information gain with entropy and use Joint-Entropy-UCB | Synthetic data · accuracy | Helped | SEL-BALD: Deep Bayesian Active Learning with Selective Labels (NeurIPS 2024 · 0 citations) |
What reviewers will expect
Datasets and baselines named in the 6 papers of this approach from 2025–2026 that have a full-text summary. Leaving out the common ones is a frequent review question.
Datasets
- CIFAR-100 · 4 of 6 papers
- CIFAR-10 · 3 of 6 papers
- Tiny-ImageNet · 3 of 6 papers
Baselines
- Random · 3 of 6 papers
- Coreset · 3 of 6 papers
- BADGE · 3 of 6 papers
- Margin · 2 of 6 papers
- Entropy · 2 of 6 papers
Recent preprints
Loading recent arXiv preprints…
Who publishes on it
Institutions on the most papers of this approach since 2023 (of 33 papers; from OpenAlex and the papers' affiliations).
- Nanjing University of Aeronautics and Astronautics · 2 papers
- Mohamed bin Zayed University of Artificial Intelligence · 2 papers
- University of Texas at Dallas · 2 papers
- Shanghai Jiao Tong University · 2 papers
- University of Chicago · 2 papers
Use it
Download BibTeX of 39 key papersOpen the topic in the interactive map
Get alerts. Describe your idea in scime atlas and save it (free scime account): you get a weekly email when new preprints land close to it.
Related approaches
Same keyword cluster
- Partial and multi-label learning (345 papers)
- Loss design and classifier theory (273 papers)
- Group-robust classification (251 papers)
- Learning from unlabeled and positive-unlabeled data (240 papers)
- Confidence calibration objectives (129 papers)
Elsewhere in the topic
- Multilingual language representation pretraining (507 papers)
- Contrastive representation theory (495 papers)
- Dense features and foundation priors (482 papers)
- Correspondence and causal representation learning (447 papers)
- Data augmentation and generalization bias (409 papers)
- Recommendation representation learning (207 papers)
- Margin-based face recognition losses (124 papers)
- Supervised person re-identification (108 papers)
Benchmark numbers, ablations and baseline numbers are extracted by AI from each paper's full text and can be wrong; the paper is the authority. Data: results index of 2026-11-02, citation counts from OpenAlex (2026-10-09).