Research map / Label-efficient and robust learning / Recognition and long-tail
Contrastive learning for imbalanced data
Uses contrastive and anchor-based regularization for imbalanced and multi-label long-tailed data.
22 accepted papers at ICML, NeurIPS, ICLR, CVPR and AAAI (2018–2026) · an approach in Recognition and long-tail, keyword cluster “long tailed · classes · tailed recognition”
State History Tried Checklist Preprints Export
Where it stands
Not enough benchmark data Fewer than four papers report the same benchmark, so there is no record to follow.
7 papers in 2023–24 and 8 in 2025–26 (at ICML, NeurIPS, ICLR, CVPR and AAAI); its share of the topic held steady, from 0.6% to 0.6%.
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
- 2018 · Learning Differences Between Visual Scanning Patterns Can Disambiguate Bipolar and Unipolar Patients (AAAI 2018 · 6 citations)
- 2019 · Visual Place Recognition via Robust ℓ2-Norm Distance Based Holism and Landmark Integration (AAAI 2019 · 11 citations)
- 2021 · Pl@ntNet-300K: a plant image dataset with high label ambiguity and a long-tailed distribution (NeurIPS 2021 · 19 citations)
- 2022 · A Simple Episodic Linear Probe Improves Visual Recognition in the Wild (CVPR 2022 · 22 citations)
- 2023 · A Step Towards Worldwide Biodiversity Assessment: The BIOSCAN-1M Insect Dataset (NeurIPS 2023 · 9 citations)
- 2024 · Decoupled Contrastive Learning for Long-Tailed Recognition (AAAI 2024 · 31 citations)
- 2025 · A Tale of Two Classes: Adapting Supervised Contrastive Learning to Binary Imbalanced Datasets (CVPR 2025 · 12 citations)
- 2026 · Negatives-Dominant Contrastive Learning for Generalization in Imbalanced Domains (ICML 2026 · 0 citations)
Earliest papers here
- Learning Differences Between Visual Scanning Patterns Can Disambiguate Bipolar and Unipolar Patients (AAAI 2018 · 6 citations)
- Visual Place Recognition via Robust ℓ2-Norm Distance Based Holism and Landmark Integration (AAAI 2019 · 11 citations)
- Pl@ntNet-300K: a plant image dataset with high label ambiguity and a long-tailed distribution (NeurIPS 2021 · 19 citations)
What has been tried
17 papers here report 177 measured changes (a component added, removed or swapped, compared with a baseline): helped 68%, no clear effect 8%, hurt 18%, 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 |
|---|---|---|---|---|
| Replace SCL with DSCL | ImageNet-LT · top-1 accuracy (Overall) | 51.2 -> 52.6 | Helped | Decoupled Contrastive Learning for Long-Tailed Recognition (AAAI 2024 · 31 citations) |
| Use fixed random anchors directly without embedding/classifier-aware auxiliary training | Cityscapes · mIoU | 79.8~80.3; 79.2~79.9; 79.8~80.4 | Mixed | Beyond Prototypes: Semantic Anchor Regularization for Better Representation Learning (AAAI 2024 · 10 citations) |
| add reliable multi-labeled instance selection | VG PredCls · F@50/100 | 44.0/47.0 -> 45.7/48.6 | Helped | RA-SGG: Retrieval-Augmented Scene Graph Generation Framework via Multi-Prototype Learning (AAAI 2025 · 7 citations) |
| replace ZoeDepth's indiscriminate distance handling with distance-aware multi-expert grouping | NYUD v2 test · RMSE | 0.277 -> 0.244 | Helped | DME: Unveiling the Bias for Better Generalized Monocular Depth Estimation (AAAI 2024 · 6 citations) |
| Add the quantum recognition method | ALL · F1-score (%) | 77.7 -> 80.8 | Helped | Quantum-inspired Non-homologous Representation Constraint Mechanism for Long-tail Senses of Word Sense Disambiguation (AAAI 2025 · 1 citation) |
| Use CSE on both Attention and FFN instead of Attention only | average · accuracy | 73.43 -> 75.82 | Helped | Once Read is Enough: Domain-specific Pretraining-free Language Models with Cluster-guided Sparse Experts for Long-tail Domain Knowledge (NeurIPS 2024 · 1 citation) |
What reviewers will expect
Datasets and baselines named in the 9 papers of this approach from 2024–2026 that have a full-text summary. Leaving out the common ones is a frequent review question.
Datasets
- CIFAR-100 · 2 of 9 papers
- ImageNet-LT · 2 of 9 papers
Baselines
- DisAlign · 2 of 9 papers
- Reweighting · 2 of 9 papers
- KCL · 2 of 9 papers
- PaCo · 2 of 9 papers
- TSC · 2 of 9 papers
- BCL · 2 of 9 papers
Recent preprints
Loading recent arXiv preprints…
Use it
Download BibTeX of 22 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
- Decoupled and calibrated long-tail training (82 papers)
- Classifier rebalancing and margin losses (58 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)
- Partial and multi-label learning (345 papers)
- Loss design and classifier theory (273 papers)
- Group-robust classification (251 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).