Research map / Label-efficient and robust learning / Noisy labels
Crowdsourced label aggregation
Infers true labels from noisy annotators using worker confusion and ability models.
23 accepted papers at ICML, NeurIPS, ICLR, CVPR and AAAI (2016–2026) · an approach in Noisy labels, keyword cluster “label noise · transition · instance”
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.
10 papers in 2023–24 and 4 in 2025–26 (at ICML, NeurIPS, ICLR, CVPR and AAAI); its share of the topic shrank, from 8.3% to 3.3%.
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 · No Oops, You Won’t Do It Again: Mechanisms for Self-correction in Crowdsourcing (ICML 2016 · 25 citations)
- 2017 · Learning to Aggregate Ordinal Labels by Maximizing Separating Width (ICML 2017 · 6 citations)
- 2018 · Learning From Noisy Singly-labeled Data (ICLR 2018 · 108 citations)
- 2019 · Exploiting Worker Correlation for Label Aggregation in Crowdsourcing (ICML 2019 · 36 citations)
- 2020 · Discrepancy Ratio: Evaluating Model Performance When Even Experts Disagree on the Truth (ICLR 2020 · 5 citations)
- 2021 · Learning from Crowds by Modeling Common Confusions (AAAI 2021 · 37 citations)
- 2022 · CS-Shapley: Class-wise Shapley Values for Data Valuation in Classification (NeurIPS 2022 · 5 citations)
- 2023 · Deep Learning From Crowdsourced Labels: Coupled Cross-Entropy Minimization, Identifiability, and Regularization (ICLR 2023 · 3 citations)
- 2024 · Dual Self-Paced Cross-Modal Hashing (AAAI 2024 · 28 citations)
- 2025 · Label Aggregation for Composite Crowd Tasks by Worker Ability Constraint Satisfaction (AAAI 2025 · 1 citation)
- 2026 · MA$^3$S: Model-Agnostic Active Annotation Strategy for Crowdsourcing (ICML 2026)
Earliest papers here
- No Oops, You Won’t Do It Again: Mechanisms for Self-correction in Crowdsourcing (ICML 2016 · 25 citations)
- Learning with Marginalized Corrupted Features and Labels Together (AAAI 2016 · 9 citations)
- Learning to Aggregate Ordinal Labels by Maximizing Separating Width (ICML 2017 · 6 citations)
What has been tried
20 papers here report 194 measured changes (a component added, removed or swapped, compared with a baseline): helped 71%, no clear effect 7%, hurt 18%, mixed 5%. Papers mostly publish changes that helped, so "helped" is over-represented.
Ablations in recent papers
| What was changed | Measured on | Result | Effect | Paper |
|---|---|---|---|---|
| Replace consensus latent representations with specific-view latent representations | MIRFlickr, I → T, 64 bits · mAP (%) | 78.76 -> 78.24 | Hurt | Dual Self-Paced Cross-Modal Hashing (AAAI 2024 · 28 citations) |
| Increase the average number of annotations per instance | CIFAR-10 with varying annotation sparsity · test accuracy | 17.15% on best and 32.95% on last | Helped | Coupled Confusion Correction: Learning from Crowds with Sparse Annotations (AAAI 2024 · 11 citations) |
| Use Euclidean, KL, L1, or rank worker-ability constraints instead of another constraint variant | Pairwise comparison datasets · pairwise comparison accuracy | No clear effect | Label Aggregation for Composite Crowd Tasks by Worker Ability Constraint Satisfaction (AAAI 2025 · 1 citation) | |
| Add instance weighting only | Leaves · model quality (%) | 59.91 -> 62.23 | Helped | IWBVT: Instance Weighting-based Bias-Variance Trade-off for Crowdsourcing (NeurIPS 2024 · 1 citation) |
| add K-free optimization (KF) | Income · Macro-F1 | 77.95 -> 78.69 | Helped | KFNN: K-Free Nearest Neighbor For Crowdsourcing (NeurIPS 2024 · 1 citation) |
| Replace the unbiased reconstruction loss with BCE loss (CLEAR-B) | Image, Scene, and Mirflickr (transition 0.2/0.5) · F1 | Hurt | Unbiased Multi-Label Learning from Crowdsourced Annotations (ICML 2024) |
What reviewers will expect
Datasets and baselines named in the 7 papers of this approach from 2024–2026 that have a full-text summary. Leaving out the common ones is a frequent review question.
Datasets
- MIRFlickr · 2 of 7 papers
- NUS-WIDE · 2 of 7 papers
Baselines
- MV · 3 of 7 papers
- DoctorNet · 2 of 7 papers
- IWMV · 2 of 7 papers
- LAWMV · 2 of 7 papers
- MNLDP · 2 of 7 papers
Recent preprints
Loading recent arXiv preprints…
Who publishes on it
Institutions on the most papers of this approach since 2023 (of 14 papers; from OpenAlex and the papers' affiliations).
- Chinese Academy of Sciences · 2 papers
- China University of Geosciences · 2 papers
Use it
Download BibTeX of 23 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
- Instance-dependent noise modeling (106 papers)
- Loss correction and noise robustness theory (77 papers)
Elsewhere in the topic
- Sample selection and co-training (86 papers)
- Noise-aware loss and pairwise methods (66 papers)
- Contrastive learning on noisy data (55 papers)
- Noisy cross-modal hashing and multi-label (49 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).