Research map / Label-efficient and robust learning / Domain adaptation and generalization
Adversarial domain adaptation
Uses domain-adversarial learning with theory, symmetric networks and cluster assumptions.
95 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 Export
Where it stands
Still moving The best reported result is still improving on 1 of 6 tracked benchmarks.
8 papers in 2023–24 and 6 in 2025–26 (at ICML, NeurIPS, ICLR, CVPR and AAAI); its share of the topic shrank, from 2.3% to 1.9%.
Open questions the data points to
- Office-Home avg (accuracy): best went from 70.5 (up to 2024) to 77.3 in 2025 — A General Representation-Based Approach to Multi-Source Domain Adaptation (ICML 2025 · 0 citations)
No longer reported by papers here: SVHN → MNIST (accuracy, last in 2019); Office-31 (Avg) (accuracy, last in 2020); MNIST → SVHN (accuracy, last in 2020); MNIST → USPS (accuracy, last in 2020); Office-Home (accuracy, last in 2023).
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 · Unsupervised Domain Adaptation with Residual Transfer Networks (NeurIPS 2016)
- 2017 · Adversarial Discriminative Domain Adaptation (CVPR 2017 · 5,087 citations)
- 2018 · Maximum Classifier Discrepancy for Unsupervised Domain Adaptation (CVPR 2018 · 2,280 citations)
- 2019 · Domain-Specific Batch Normalization for Unsupervised Domain Adaptation (CVPR 2019 · 437 citations)
- 2020 · Adversarial Domain Adaptation with Domain Mixup (AAAI 2020 · 396 citations)
- 2021 · Reducing Domain Gap by Reducing Style Bias (CVPR 2021 · 351 citations)
- 2022 · Reusing the Task-Specific Classifier as a Discriminator: Discriminator-Free Adversarial Domain Adaptation (CVPR 2022 · 218 citations)
- 2023 · AutoLabel: CLIP-Based Framework for Open-Set Video Domain Adaptation (CVPR 2023 · 22 citations)
- 2024 · Learning CNN on ViT: A Hybrid Model to Explicitly Class-specific Boundaries for Domain Adaptation (CVPR 2024 · 44 citations)
- 2025 · Open-Set Cross-Network Node Classification via Unknown-Excluded Adversarial Graph Domain Alignment (AAAI 2025 · 5 citations)
- 2026 · $\oslash$ Source Models Leak What They Shouldn't $\nrightarrow$: Unlearning Zero-Shot Transfer in Domain Adaptation Through Adversarial Optimization (CVPR 2026 · 0 citations)
Earliest papers here
- Unsupervised Domain Adaptation with Residual Transfer Networks (NeurIPS 2016)
- The l2,1-Norm Stacked Robust Autoencoders for Domain Adaptation (AAAI 2016)
- Adversarial Discriminative Domain Adaptation (CVPR 2017 · 5,087 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.
SVHN → MNIST · accuracy (higher is better)
10 papers, 2017–2019 · not reported since 2019
| Year | Best so far | Set by |
|---|---|---|
| 2017 | 87.2 | Few-Shot Adversarial Domain Adaptation (NeurIPS 2017 · 205 citations) |
| 2018 | 99.4 | A DIRT-T Approach to Unsupervised Domain Adaptation (ICLR 2018 · 431 citations) |
Office-31 (Avg) · accuracy (higher is better)
8 papers, 2016–2020 · not reported since 2020
| Year | Best so far | Set by |
|---|---|---|
| 2016 | 73.7 | Unsupervised Domain Adaptation with Residual Transfer Networks (NeurIPS 2016) |
| 2019 | 89.5 | Transferable Normalization: Towards Improving Transferability of Deep Neural Networks (NeurIPS 2019 · 132 citations) |
Office-Home avg · accuracy (higher is better)
8 papers, 2019–2025 · still improving
| Year | Best so far | Set by |
|---|---|---|
| 2019 | 67.6 | Domain-Symmetric Networks for Adversarial Domain Adaptation (CVPR 2019 · 341 citations) |
| 2020 | 70.5 | Domain Conditioned Adaptation Network (AAAI 2020 · 107 citations) |
| 2025 | 77.3 | A General Representation-Based Approach to Multi-Source Domain Adaptation (ICML 2025 · 0 citations) |
MNIST → SVHN · accuracy (higher is better)
7 papers, 2017–2020 · not reported since 2020
| Year | Best so far | Set by |
|---|---|---|
| 2017 | 47 | Few-Shot Adversarial Domain Adaptation (NeurIPS 2017 · 205 citations) |
| 2018 | 76.5 | A DIRT-T Approach to Unsupervised Domain Adaptation (ICLR 2018 · 431 citations) |
| 2019 | 87.48 | Gotta Adapt 'Em All: Joint Pixel and Feature-Level Domain Adaptation for Recognition in the Wild (CVPR 2019 · 48 citations) |
MNIST → USPS · accuracy (higher is better)
7 papers, 2017–2020 · not reported since 2020
| Year | Best so far | Set by |
|---|---|---|
| 2017 | 94.4 | Few-Shot Adversarial Domain Adaptation (NeurIPS 2017 · 205 citations) |
| 2018 | 97.6 | From Source to Target and Back: Symmetric Bi-Directional Adaptive GAN (CVPR 2018 · 252 citations) |
| 2019 | 99.3 | Unsupervised Domain Adaptation Using Feature-Whitening and Consensus Loss (CVPR 2019 · 172 citations) |
Office-Home · accuracy (higher is better)
6 papers, 2019–2023 · not reported since 2023
| Year | Best so far | Set by |
|---|---|---|
| 2019 | 68.1 | Bridging Theory and Algorithm for Domain Adaptation (ICML 2019 · 326 citations) |
| 2020 | 70.9 | Spherical Space Domain Adaptation With Robust Pseudo-Label Loss (CVPR 2020 · 143 citations) |
| 2022 | 71.8 | Reusing the Task-Specific Classifier as a Discriminator: Discriminator-Free Adversarial Domain Adaptation (CVPR 2022 · 218 citations) |
What has been tried
88 papers here report 900 measured changes (a component added, removed or swapped, compared with a baseline): helped 76%, no clear effect 5%, hurt 16%, 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 CNN-to-ViT pseudo-label teaching to ViT-to-CNN teaching | DomainNet SSDA mean, 3-shot · accuracy | 79.4 -> 85.5 | Helped | Learning CNN on ViT: A Hybrid Model to Explicitly Class-specific Boundaries for Domain Adaptation (CVPR 2024 · 44 citations) |
| add a separation stage before adaptation | C→D · OS | 64.3 -> 69.2 | Helped | Open-Set Cross-Network Node Classification via Unknown-Excluded Adversarial Graph Domain Alignment (AAAI 2025 · 5 citations) |
| add VAE-related losses | Office-Home Avg · accuracy | 75.9 -> 77.3 | Helped | A General Representation-Based Approach to Multi-Source Domain Adaptation (ICML 2025 · 0 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 |
|---|---|---|---|---|
| DAN | Office-31 Avg | accuracy | 71.7 – 78.5 | 2 papers, e.g. NeurIPS 2016 and AAAI 2019 |
| DANN | Office-31 A→W | accuracy | 79.3 – 85.97 | 2 papers, e.g. CVPR 2018 and CVPR 2019 |
| DANN | VisDA | accuracy | 57.4 – 61.88 | 2 papers, e.g. CVPR 2018 and NeurIPS 2020 |
What reviewers will expect
Datasets and baselines named in the 5 papers of this approach from 2024–2026 that have a full-text summary. Leaving out the common ones is a frequent review question.
Datasets
- Office-Home · 2 of 5 papers
Baselines
- DANN · 2 of 5 papers
- MCD · 2 of 5 papers
Recent preprints
Loading recent arXiv preprints…
Use it
Download BibTeX of 27 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
- Conditional distribution alignment (162 papers)
- Feature alignment for unsupervised adaptation (105 papers)
- Source-free domain adaptation (92 papers)
- Geometric alignment and synthetic-to-real adaptation (93 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).