Research map / Label-efficient and robust learning / Domain adaptation and generalization
Conditional distribution alignment
Aligns conditional or disentangled distributions for domain adaptation, including universal and generalized shift.
162 accepted papers at ICML, NeurIPS, ICLR, CVPR and AAAI (2016–2026) · an approach in Domain adaptation and generalization, keyword cluster “domain adaptation · unsupervised domain · uda”
State History Benchmarks Tried Baselines Checklist Preprints Who Export
Where it stands
Few recent reports The tracked benchmarks have at most one report from 2025–26 that could beat the old record, too few to say whether they are still moving.
31 papers in 2023–24 and 35 in 2025–26 (at ICML, NeurIPS, ICLR, CVPR and AAAI); its share of the topic grew, from 9.1% to 11.0%.
Open questions the data points to
- Papers report Source Only on MNIST→SVHN (accuracy (%)) anywhere from 40.2 to 47.8: comparisons against it depend on whose number you take.
- Papers report SymNets on Office-31-DA (accuracy) anywhere from 74.6 to 88.4: comparisons against it depend on whose number you take.
No longer reported by papers here: SVHN→MNIST (accuracy, last in 2020); MNIST→USPS (accuracy, last in 2020); Office-Home Avg. (accuracy, last in 2024).
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 · Learning Transferrable Representations for Unsupervised Domain Adaptation (NeurIPS 2016 · 261 citations)
- 2017 · Asymmetric Tri-training for Unsupervised Domain Adaptation (ICML 2017 · 451 citations)
- 2018 · Learning Semantic Representations for Unsupervised Domain Adaptation (ICML 2018 · 352 citations)
- 2019 · Sliced Wasserstein Discrepancy for Unsupervised Domain Adaptation (CVPR 2019 · 618 citations)
- 2020 · Model Adaptation: Unsupervised Domain Adaptation Without Source Data (CVPR 2020 · 502 citations)
- 2021 · Domain Consensus Clustering for Universal Domain Adaptation (CVPR 2021 · 129 citations)
- 2022 · Active Learning for Domain Adaptation: An Energy-Based Approach (AAAI 2022 · 135 citations)
- 2023 · DARE-GRAM: Unsupervised Domain Adaptation Regression by Aligning Inverse Gram Matrices (CVPR 2023 · 62 citations)
- 2024 · Agile Multi-Source-Free Domain Adaptation (AAAI 2024 · 18 citations)
- 2025 · HiGDA: Hierarchical Graph of Nodes to Learn Local-to-Global Topology for Semi-Supervised Domain Adaptation (AAAI 2025 · 33 citations)
- 2026 · From Blind Transfer to Wise Selection: Prototype-Driven Neighbor-Domain Adaptation for Fake News Detection (AAAI 2026 · 5 citations)
Earliest papers here
- Learning Transferrable Representations for Unsupervised Domain Adaptation (NeurIPS 2016 · 261 citations)
- A New PAC-Bayesian Perspective on Domain Adaptation (ICML 2016 · 30 citations)
- ExTaSem! Extending, Taxonomizing and Semantifying Domain Terminologies (AAAI 2016 · 22 citations)
Benchmark records
The best number each year that beat every earlier one, among papers of this approach reporting their own method. Numbers are extracted by AI from each paper's text; check the paper before citing.
Office-Home · accuracy (higher is better)
9 papers, 2020–2025 · one report since 2024, not better
| Year | Best so far | Set by |
|---|---|---|
| 2020 | 71 | Unsupervised Domain Adaptation via Structured Prediction Based Selective Pseudo-Labeling (AAAI 2020 · 235 citations) |
| 2022 | 76.19 | Partial disentanglement for domain adaptation (ICML 2022) |
| 2024 | 80.1 | Cycle Self-Refinement for Multi-Source Domain Adaptation (AAAI 2024 · 18 citations) |
SVHN→MNIST · accuracy (higher is better)
8 papers, 2017–2020 · not reported since 2020
| Year | Best so far | Set by |
|---|---|---|
| 2017 | 85 | Asymmetric Tri-training for Unsupervised Domain Adaptation (ICML 2017 · 451 citations) |
| 2018 | 99.26 | Self-ensembling for visual domain adaptation (ICLR 2018 · 60 citations) |
| 2019 | 99.3 | Improving Domain-Specific Classification by Collaborative Learning with Adaptation Networks (AAAI 2019 · 12 citations) |
| 2020 | 99.4 | Model Adaptation: Unsupervised Domain Adaptation Without Source Data (CVPR 2020 · 502 citations) |
Office-31 · accuracy (%) (higher is better)
6 papers, 2020–2025 · one report since 2024, not better
| Year | Best so far | Set by |
|---|---|---|
| 2020 | 89.6 | Unsupervised Domain Adaptation via Structured Prediction Based Selective Pseudo-Labeling (AAAI 2020 · 235 citations) |
| 2024 | 92.7 | Cycle Self-Refinement for Multi-Source Domain Adaptation (AAAI 2024 · 18 citations) |
DomainNet · accuracy (higher is better)
5 papers, 2021–2025 · one report since 2024, not better
| Year | Best so far | Set by |
|---|---|---|
| 2021 | 51.3 | Dynamic Transfer for Multi-Source Domain Adaptation (CVPR 2021 · 85 citations) |
| 2024 | 82 | Low Category Uncertainty and High Training Potential Instance Learning for Unsupervised Domain Adaptation (AAAI 2024 · 6 citations) |
MNIST→USPS · accuracy (higher is better)
5 papers, 2018–2020 · not reported since 2020
| Year | Best so far | Set by |
|---|---|---|
| 2018 | 99.54 | Self-ensembling for visual domain adaptation (ICLR 2018 · 60 citations) |
Office-Home Avg. · accuracy (higher is better)
5 papers, 2020–2024 · not reported since 2024
| Year | Best so far | Set by |
|---|---|---|
| 2020 | 72.3 | Selective Transfer With Reinforced Transfer Network for Partial Domain Adaptation (CVPR 2020 · 60 citations) |
| 2022 | 72.7 | Attracting and Dispersing: A Simple Approach for Source-free Domain Adaptation (NeurIPS 2022 · 55 citations) |
| 2024 | 77.1 | Agile Multi-Source-Free Domain Adaptation (AAAI 2024 · 18 citations) |
What has been tried
136 papers here report 1,400 measured changes (a component added, removed or swapped, compared with a baseline): helped 74%, no clear effect 6%, hurt 17%, 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 GAL to HiGDA-T with S+T training | Office-Home 3-shot · accuracy | 5.3% | Helped | HiGDA: Hierarchical Graph of Nodes to Learn Local-to-Global Topology for Semi-Supervised Domain Adaptation (AAAI 2025 · 33 citations) |
| Add intra-domain weights to ATEN | DomainNet Avg. (three-task ablation table) · accuracy | 59.5 -> 60.1 | Helped | Agile Multi-Source-Free Domain Adaptation (AAAI 2024 · 18 citations) |
| Add Lpsd to Lcls | Office-Home · accuracy | 72.4 -> 78.8 | Helped | Cycle Self-Refinement for Multi-Source Domain Adaptation (AAAI 2024 · 18 citations) |
| Add neighborhood invariance learning (NIL) | Office ODA · H-score | 17.1 -> 92.8 | Helped | MLNet: Mutual Learning Network with Neighborhood Invariance for Universal Domain Adaptation (AAAI 2024 · 18 citations) |
| add domain separation | VisDA · accuracy | 84.3 -> 86.4 | Helped | A Separation and Alignment Framework for Black-Box Domain Adaptation (AAAI 2024 · 10 citations) |
| add supervision using reliable known knowledge | Office-31 (Avg) · HOS | 80.3 -> 79.7 | Hurt | Unveiling the Unknown: Unleashing the Power of Unknown to Known in Open-Set Source-Free Domain Adaptation (CVPR 2024 · 8 citations) |
| Add exemplar replay loss Lrep | Office-31-CI · accuracy | 79.8 -> 88.4 | Helped | Multi-Granularity Class Prototype Topology Distillation for Class-Incremental Source-Free Unsupervised Domain Adaptation (CVPR 2025 · 7 citations) |
| Use target refinement with τ=0.8 instead of τ=1.0, retaining the closest samples | Office-31, 40% label corruption · average accuracy | 83.3 -> 83.6 | Helped | Confusing Pair Correction Based on Category Prototype for Domain Adaptation under Noisy Environments (AAAI 2024 · 7 citations) |
| Add LK to the configuration with no listed components | Office-Home · accuracy | 57.6 -> 73.7 | Helped | Low Category Uncertainty and High Training Potential Instance Learning for Unsupervised Domain Adaptation (AAAI 2024 · 6 citations) |
| add selection training | Office-31 · accuracy | 89.2 -> 87.7 | Helped | Reviewing the Forgotten Classes for Domain Adaptation of Black-Box Predictors (AAAI 2024 · 5 citations) |
| replace weight barcode estimation with JPOT-based mean transport cost | Office-31 (Avg) · OS accuracy (%) | 94.9 -> 96.4 | Hurt | Distinguish Then Exploit: Source-free Open Set Domain Adaptation via Weight Barcode Estimation and Sparse Label Assignment (CVPR 2025 · 4 citations) |
| Add MOD to CE-only training | AVE · accuracy (Image) | 49.90 -> 53.22 | Helped | MODfinity: Unsupervised Domain Adaptation with Multimodal Information Flow Intertwining (CVPR 2025 · 4 citations) |
Baseline numbers that disagree
The same baseline on the same benchmark and metric, reported more than 5% apart by different papers here. Training budgets, data splits and reimplementations differ, so state which number you compare with.
| Baseline | Benchmark | Metric | Reported | Papers |
|---|---|---|---|---|
| Source Only | MNIST→SVHN | accuracy (%) | 40.2 – 47.8 | 2 papers, e.g. AAAI 2019 and AAAI 2020 |
| SymNets | Office-31-DA | accuracy | 74.6 – 88.4 | 2 papers, e.g. AAAI 2020 and CVPR 2020 |
What reviewers will expect
Datasets and baselines named in the 23 papers of this approach from 2025–2026 that have a full-text summary. Leaving out the common ones is a frequent review question.
Datasets
- Office-Home · 6 of 23 papers
- DomainNet · 3 of 23 papers
- Office-31 · 3 of 23 papers
- VisDA2017 · 2 of 23 papers
- MNIST · 2 of 23 papers
- USPS · 2 of 23 papers
- SVHN · 2 of 23 papers
Baselines
- DANN · 5 of 23 papers
- PGL · 3 of 23 papers
- OSBP · 2 of 23 papers
- ANNA · 2 of 23 papers
- DeepJDOT · 2 of 23 papers
- SHOT · 2 of 23 papers
- Source-only · 2 of 23 papers
- GCN · 2 of 23 papers
Recent preprints
Loading recent arXiv preprints…
Who publishes on it
Institutions on the most papers of this approach since 2023 (of 66 papers; from OpenAlex and the papers' affiliations).
- Zhejiang University · 5 papers
- Meta · 4 papers
- University of Electronic Science and Technology of China · 3 papers
- Jilin University · 3 papers
- Georgia Institute of Technology · 3 papers
- NVIDIA · 3 papers
- Key Laboratory of Symbolic Computation and Knowledge Engineering (Jilin University) · 2 papers
- Sun Yat-sen University · 2 papers
- RIKEN · 2 papers
- The University of Tokyo · 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
- Feature alignment for unsupervised adaptation (105 papers)
- Source-free domain adaptation (92 papers)
- Geometric alignment and synthetic-to-real adaptation (93 papers)
- Adversarial domain adaptation (95 papers)
- Statistical alignment and hypothesis testing (80 papers)
Elsewhere in the topic
- Cross-domain few-shot with transformers (116 papers)
- Optimization-based domain generalization (101 papers)
- Adaptation for semantic segmentation (81 papers)
- Statistical transfer learning (79 papers)
- Mixture and distillation for domain shift (78 papers)
- Foundation-model domain generalization (56 papers)
- Domain-generalized segmentation (56 papers)
- Cross-domain text and graph transfer (54 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).