Research map / Label-efficient and robust learning / Semi-supervised learning
Pseudo-labeling for specialized tasks
Applies pseudo-label teachers to mirror detection, referring expressions, motion and rotation regression.
48 accepted papers at ICML, NeurIPS, ICLR, CVPR and AAAI (2016–2026) · an approach in Semi-supervised learning, keyword cluster “pseudo labels · segmentation · pseudo labeling”
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.
15 papers in 2023–24 and 18 in 2025–26 (at ICML, NeurIPS, ICLR, CVPR and AAAI); its share of the topic grew, from 9.4% to 11.4%.
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 · Neural Universal Discrete Denoiser (NeurIPS 2016 · 0 citations)
- 2018 · ARC: Adversarial Robust Cuts for Semi-Supervised and Multi-Label Classification (AAAI 2018 · 12 citations)
- 2019 · Distantly Supervised Entity Relation Extraction with Adapted Manual Annotations (AAAI 2019)
- 2020 · Generating Accurate Pseudo-Labels in Semi-Supervised Learning and Avoiding Overconfident Predictions via Hermite Polynomial Activations (CVPR 2020 · 30 citations)
- 2021 · Memory Oriented Transfer Learning for Semi-Supervised Image Deraining (CVPR 2021 · 118 citations)
- 2022 · Self-Taught Metric Learning Without Labels (CVPR 2022 · 22 citations)
- 2023 · Contrastive Semi-Supervised Learning for Underwater Image Restoration via Reliable Bank (CVPR 2023 · 269 citations)
- 2024 · Unleashing Unlabeled Data: A Paradigm for Cross-View Geo-Localization (CVPR 2024 · 44 citations)
- 2025 · Point-to-Region Loss for Semi-Supervised Point-Based Crowd Counting (CVPR 2025 · 8 citations)
- 2026 · Perceive, Act and Correct: Confidence Is Not Enough for Hyperspectral Classification (AAAI 2026 · 2 citations)
Earliest papers here
- Neural Universal Discrete Denoiser (NeurIPS 2016 · 0 citations)
- ARC: Adversarial Robust Cuts for Semi-Supervised and Multi-Label Classification (AAAI 2018 · 12 citations)
- Selective Verification Strategy for Learning From Crowds (AAAI 2018 · 3 citations)
What has been tried
39 papers here report 469 measured changes (a component added, removed or swapped, compared with a baseline): helped 77%, no clear effect 6%, hurt 14%, mixed 3%. Papers mostly publish changes that helped, so "helped" is over-represented.
Ablations in recent papers
| What was changed | Measured on | Result | Effect | Paper |
|---|---|---|---|---|
| Add geometric projection (BEV) to the cold-start stage | CVACT · R@1 | 0.0 -> 44.81 | Helped | Unleashing Unlabeled Data: A Paradigm for Cross-View Geo-Localization (CVPR 2024 · 44 citations) |
| add Colabator to CORUN | RTTS · BRISQUE | 16.541 -> 11.956 | Helped | Real-world Image Dehazing with Coherence-based Pseudo Labeling and Cooperative Unfolding Network (NeurIPS 2024 · 17 citations) |
| add temporal sampling to Mean-Teacher | nuScenes, 1% labeled, fast · mean error | 3.0868 to 2.9037 | Helped | Semi-supervised Class-Agnostic Motion Prediction with Pseudo Label Regeneration and BEVMix (AAAI 2024 · 10 citations) |
| Add target-domain unlabeled data adaptation | A→B, A→Q, B→A, and B→Q · MAE | Helped | Point-to-Region Loss for Semi-Supervised Point-Based Crowd Counting (CVPR 2025 · 8 citations) | |
| add geometric scoring to the labeled-data baseline | PMD · IoU | 0.679 -> 0.693 | Helped | Detect Any Mirrors: Boosting Learning Reliability on Large-Scale Unlabeled Data with an Iterative Data Engine (CVPR 2025 · 5 citations) |
| add multi-view prototype-based clustering without pseudo-label assignment | VG150 PredCls · M@100 | 42.3 -> 45.0 | Helped | Semi-Supervised Clustering Framework for Fine-grained Scene Graph Generation (AAAI 2025 · 4 citations) |
| Add Progressive Sample Assignment | Total-Text, 2% labeled · E2E H-mean (None) | 68.6 -> 72.4 | Helped | SemiETS: Integrating Spatial and Content Consistencies for Semi-Supervised End-to-end Text Spotting (CVPR 2025 · 4 citations) |
| Add pseudo-label filtering during self-training | Middlebury Eval · EPE | 6.15 -> 2.85 | Helped | Consistency-aware Self-Training for Iterative-based Stereo Matching (CVPR 2025 · 2 citations) |
| add confidence distribution learning | NL27k · MSE | 0.015 -> 0.009 | Helped | Uncertain Knowledge Graph Completion via Semi-Supervised Confidence Distribution Learning (NeurIPS 2025 · 1 citation) |
| Add large-scale unlabeled data to labeled training | MPIIFaceGaze zero-shot · angular error | 6.17 -> 4.97 | Helped | OmniGaze: Reward-inspired Generalizable Gaze Estimation in the Wild (NeurIPS 2025 · 1 citation) |
| Add Semantic Query Adversarial Fusion | nuScenes to Waymo (target) · mIoU | 15.2 -> 16.7 | Helped | Enhancing Generalizability via Utilization of Unlabeled Data for Occupancy Perception (AAAI 2025 · 0 citations) |
| add EPNL | Matterport3D · RMSE | 0.5132 -> 0.5062 | Helped | PanDA: Towards Panoramic Depth Anything with Unlabeled Panoramas and Mobius Spatial Augmentation (CVPR 2025 · 0 citations) |
What reviewers will expect
Datasets and baselines named in the 11 papers of this approach from 2025–2026 that have a full-text summary. Leaving out the common ones is a frequent review question.
Baselines
- MT · 2 of 11 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).
- Nanyang Technological University · 2 papers
- Huazhong University of Science and Technology · 2 papers
- Zhejiang University · 2 papers
- Beihang University · 2 papers
- The Chinese University of Hong Kong · 2 papers
- City University of Hong Kong · 2 papers
- Tsinghua University · 2 papers
Use it
Download BibTeX of 38 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
- Confidence-based pseudo-label selection (134 papers)
- Consistency training for semi-supervised segmentation (49 papers)
- Semi-supervised medical segmentation (50 papers)
Elsewhere in the topic
- Holistic semi-supervised pipelines (125 papers)
- Self-training theory (86 papers)
- Safe and realistic semi-supervised learning (70 papers)
- GAN-based semi-supervised learning (61 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).