Research map / Label-efficient and robust learning / Domain adaptation and generalization
Statistical transfer learning
Provides statistical methods and bounds for measuring and optimizing transfer between domains.
79 accepted papers at ICML, NeurIPS, ICLR, CVPR and AAAI (2016–2026) · an approach in Domain adaptation and generalization, keyword cluster “cross domain · transfer · domain shot”
State History Tried 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.
14 papers in 2023–24 and 14 in 2025–26 (at ICML, NeurIPS, ICLR, CVPR and AAAI); its share of the topic held steady, from 4.1% to 4.4%.
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 · A Representation Learning Framework for Multi-Source Transfer Parsing (AAAI 2016 · 104 citations)
- 2017 · Source-Target Similarity Modelings for Multi-Source Transfer Gaussian Process Regression (ICML 2017 · 22 citations)
- 2018 · Partial Transfer Learning With Selective Adversarial Networks (CVPR 2018 · 462 citations)
- 2019 · Characterizing and Avoiding Negative Transfer (CVPR 2019 · 478 citations)
- 2020 · What is being transferred in transfer learning? (NeurIPS 2020 · 188 citations)
- 2021 · In the Light of Feature Distributions: Moment Matching for Neural Style Transfer (CVPR 2021 · 57 citations)
- 2022 · What Makes Transfer Learning Work for Medical Images: Feature Reuse & Other Factors (CVPR 2022 · 106 citations)
- 2023 · Partial Network Cloning (CVPR 2023 · 16 citations)
- 2024 · Decoupling Representation and Knowledge for Few-Shot Intent Classification and Slot Filling (AAAI 2024 · 5 citations)
- 2025 · Drawing Informative Gradients from Sources: A One-stage Transfer Learning Framework for Cross-city Spatiotemporal Forecasting (AAAI 2025 · 10 citations)
- 2026 · Domain Transfer Becomes Identifiable via a Single Alignment (ICML 2026 · 0 citations)
Earliest papers here
- A Representation Learning Framework for Multi-Source Transfer Parsing (AAAI 2016 · 104 citations)
- Transfer Learning for Cross-Language Text Categorization through Active Correspondences Construction (AAAI 2016 · 50 citations)
- Constrained Deep Transfer Feature Learning and Its Applications (CVPR 2016 · 37 citations)
What has been tried
64 papers here report 652 measured changes (a component added, removed or swapped, compared with a baseline): helped 75%, no clear effect 6%, hurt 14%, 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 |
|---|---|---|---|---|
| Remove the Node Prompting Module (NPM) | Didi-Shenzhen · MAE | approximately 20% increase | Hurt | Drawing Informative Gradients from Sources: A One-stage Transfer Learning Framework for Cross-city Spatiotemporal Forecasting (AAAI 2025 · 10 citations) |
| Add I2S to joint modeling | FewJoint 1-shot, Cos similarity · Joint Acc. | 24.29 -> 36.56 | Helped | Decoupling Representation and Knowledge for Few-Shot Intent Classification and Slot Filling (AAAI 2024 · 5 citations) |
| Add multi-source transfer instead of using average single-source performance | Office-31 8-shot · accuracy | 0.8483 -> 0.8644 | Helped | H-ensemble: An Information Theoretic Approach to Reliable Few-Shot Multi-Source-Free Transfer (AAAI 2024 · 5 citations) |
| Add self-distillation loss to vanilla fine-tuning | CIFAR10→STL10 · target accuracy | +34.7; +49.4 | Helped | Your Transferability Barrier is Fragile: Free-Lunch for Transferring the Non-Transferable Learning (CVPR 2024 · 5 citations) |
| Removing FCD (disentanglers) from DA-Net | Wikiann NER, en target (af) · F1 | af decreases by 2.16% without FCD | Helped | DA-Net: A Disentangled and Adaptive Network for Multi-Source Cross-Lingual Transfer Learning (AAAI 2024 · 2 citations) |
| add structured non-linearity regularization | VTAB-1k Mean (All) · accuracy | 69.0 -> 70.9 | Helped | Structured Model Probing: Empowering Efficient Transfer Learning by Structured Regularization (CVPR 2024 · 2 citations) |
| Transfer from source only when source is sufficiently similar (psi_1 below about 0.9) | Simulation, n0=100, n1=200 · RMSPR vs Only Target | Mixed | Wasserstein Transfer Learning (NeurIPS 2025 · 0 citations) | |
| Replace dynamic transfer-quantity updates with static under-fitted transfer quantities | Office-Home 10-shot · accuracy | 78.2 -> 77.2 | Hurt | A High-Dimensional Statistical Method for Optimizing Transfer Quantities in Multi-Source Transfer Learning (NeurIPS 2025 · 0 citations) |
| Add pre-training | Benchmark 1 · AUC | 0.5907 -> 0.6527 | Helped | EvoluNet: Advancing Dynamic Non-IID Transfer Learning on Graphs (ICML 2024 · 0 citations) |
| Add disentanglement loss | OOD detection (CIFAR10 ID, LSUN OOD) · AUROC | 84.94±1.3 -> 86.12±0.6 | Helped | Towards Trustworthy Machine Learning Under Distribution Shifts (AAAI 2025) |
| Add instance filtering | Income · aggregation accuracy | 71.83 -> 74.97 | Helped | TLLC: Transfer Learning-based Label Completion for Crowdsourcing (ICML 2025) |
Recent preprints
Loading recent arXiv preprints…
Who publishes on it
Institutions on the most papers of this approach since 2023 (of 28 papers; from OpenAlex and the papers' affiliations).
- Huazhong University of Science and Technology · 2 papers
- Tsinghua University · 2 papers
- National University of Singapore · 2 papers
- JD Explore Academy · 2 papers
- The University of Sydney · 2 papers
- University of Virginia · 2 papers
- Amazon · 2 papers
Use it
Download BibTeX of 40 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
- Cross-domain few-shot with transformers (116 papers)
- Mixture and distillation for domain shift (78 papers)
- Cross-domain text and graph transfer (54 papers)
Elsewhere in the topic
- Conditional distribution alignment (162 papers)
- Feature alignment for unsupervised adaptation (105 papers)
- Optimization-based domain generalization (101 papers)
- Adversarial domain adaptation (95 papers)
- Geometric alignment and synthetic-to-real adaptation (93 papers)
- Source-free domain adaptation (92 papers)
- Adaptation for semantic segmentation (81 papers)
- Statistical alignment and hypothesis testing (80 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).