Research map / Label-efficient and robust learning / Noisy labels
Noisy cross-modal hashing and multi-label
Handles noisy labels in cross-modal retrieval and multi-label classification via robust hashing and loss handling.
49 accepted papers at ICML, NeurIPS, ICLR, CVPR and AAAI (2016–2026) · an approach in Noisy labels, keyword cluster “noisy labels · robust · clean”
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.
11 papers in 2023–24 and 18 in 2025–26 (at ICML, NeurIPS, ICLR, CVPR and AAAI); its share of the topic grew, from 9.1% to 14.8%.
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 · Seeing Through the Human Reporting Bias: Visual Classifiers From Noisy Human-Centric Labels (CVPR 2016 · 215 citations)
- 2017 · Attend in Groups: A Weakly-Supervised Deep Learning Framework for Learning From Web Data (CVPR 2017 · 90 citations)
- 2018 · Synthetic and Natural Noise Both Break Neural Machine Translation (ICLR 2018 · 398 citations)
- 2019 · MetaCleaner: Learning to Hallucinate Clean Representations for Noisy-Labeled Visual Recognition (CVPR 2019 · 114 citations)
- 2020 · Coupled-View Deep Classifier Learning from Multiple Noisy Annotators (AAAI 2020 · 25 citations)
- 2021 · Faster Meta Update Strategy for Noise-Robust Deep Learning (CVPR 2021 · 63 citations)
- 2022 · Learning With Twin Noisy Labels for Visible-Infrared Person Re-Identification (CVPR 2022 · 277 citations)
- 2023 · Noisy Correspondence Learning With Meta Similarity Correction (CVPR 2023 · 47 citations)
- 2024 · Noisy Correspondence Learning with Self-Reinforcing Errors Mitigation (AAAI 2024 · 11 citations)
- 2025 · Robust Self-Paced Hashing for Cross-Modal Retrieval with Noisy Labels (AAAI 2025 · 17 citations)
- 2026 · Neighbor-aware Instance Refining with Noisy Labels for Cross-Modal Retrieval (AAAI 2026 · 1 citation)
Earliest papers here
- Seeing Through the Human Reporting Bias: Visual Classifiers From Noisy Human-Centric Labels (CVPR 2016 · 215 citations)
- Noisy Label Recovery for Shadow Detection in Unfamiliar Domains (CVPR 2016 · 42 citations)
- Distant IE by Bootstrapping Using Lists and Document Structure (AAAI 2016 · 12 citations)
What has been tried
38 papers here report 390 measured changes (a component added, removed or swapped, compared with a baseline): helped 75%, no clear effect 5%, hurt 16%, mixed 4%. Papers mostly publish changes that helped, so "helped" is over-represented.
Ablations in recent papers
| What was changed | Measured on | Result | Effect | Paper |
|---|---|---|---|---|
| add warm-up training | XMedia, noise 0.6, 128 bits (average MAP) · MAP | 84.3 -> 88.6 | Helped | Robust Self-Paced Hashing for Cross-Modal Retrieval with Noisy Labels (AAAI 2025 · 17 citations) |
| Add reconstruction loss Lrec | Scene15, 20% noise · ACC | 18.29 -> 45.65 | Helped | ROLL: Robust Noisy Pseudo-label Learning for Multi-View Clustering with Noisy Correspondence (CVPR 2025 · 14 citations) |
| Add sample filtration | CC152K · image-to-text R@1 | 32.5 -> 37.3 | Helped | Noisy Correspondence Learning with Self-Reinforcing Errors Mitigation (AAAI 2024 · 11 citations) |
| Replace identity weighting with the pre-trained calibrator | MS-COCO, 80% noise · mAP (%) | 59.93 -> 61.48 | Helped | Calibrated Disambiguation for Partial Multi-label Learning (AAAI 2025 · 8 citations) |
| add label filtering and reconstruction rather than training directly on original noisy data | MIR Flickr, NUS-WIDE, and MS COCO, 40% noisy labels, 64 bits · MAP | 0.784 -> 0.846; 0.721 -> 0.755; 0.610 -> 0.631 | Helped | Distribution-Consistency-Guided Multi-modal Hashing (AAAI 2025 · 5 citations) |
| Use noisy constraints rather than unconstrained learning | IGGP, Zendo, program synthesis, Alzheimer, and WN18RR tasks · learning time | 109 ± 2 -> 2 ± 0 | Helped | Learning MDL Logic Programs from Noisy Data (AAAI 2024 · 5 citations) |
| Apply CRL with collaborative refining (LRD and RUS) instead of majority-vote label aggregation | Ablation datasets (Avg.) · AUC | 0.643 -> 0.814 | Helped | Collaborative Refining for Learning from Inaccurate Labels (NeurIPS 2024 · 3 citations) |
| Add the Answer Diffuser (AD) | VQA-RAD 10%-Semantic Noise · overall accuracy | 68.25 -> 71.15 | Helped | DiN: Diffusion Model for Robust Medical VQA with Semantic Noisy Labels (CVPR 2025 · 2 citations) |
| Fine-tune the MAE feature extractor for SimiFeat | Noisy Ostracods test/validation · Hit rate | 27.55 -> 28.18 | Helped | Noisy Ostracods: A Fine-Grained, Imbalanced Real-World Dataset for Benchmarking Robust Machine Learning and Label Correction Methods (NeurIPS 2024 · 1 citation) |
| Add label refinement only | CIFAR-100 · top-1 accuracy | 47.5 -> 48.0 | Helped | GIFT: Unlocking Full Potential of Labels in Distilled Dataset at Near-zero Cost (ICLR 2025 · 0 citations) |
What reviewers will expect
Datasets and baselines named in the 9 papers of this approach from 2025–2026 that have a full-text summary. Leaving out the common ones is a frequent review question.
Datasets
- NUS-WIDE · 3 of 9 papers
- XMediaNet · 2 of 9 papers
- CUB · 2 of 9 papers
- MNIST · 2 of 9 papers
Recent preprints
Loading recent arXiv preprints…
Who publishes on it
Institutions on the most papers of this approach since 2023 (of 29 papers; from OpenAlex and the papers' affiliations).
- Xi’an Jiaotong University · 2 papers
- Sichuan University · 2 papers
- Sichuan National Innovation New Vision UHD Video Technology Co · 2 papers
- Southwest University of Science and Technology · 2 papers
- University of Electronic Science and Technology of China · 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
- Sample selection and co-training (86 papers)
- Noise-aware loss and pairwise methods (66 papers)
- Contrastive learning on noisy data (55 papers)
Elsewhere in the topic
- Instance-dependent noise modeling (106 papers)
- Loss correction and noise robustness theory (77 papers)
- Crowdsourced label aggregation (23 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).