Research map / Continual learning / Domain and prompt-based
Domain-incremental learning with memory
Adapts to sequentially arriving domains using memory, domain-aware representations and randomization.
49 accepted papers at ICML, NeurIPS, ICLR, CVPR and AAAI (2018–2026) · an approach in Domain and prompt-based, keyword cluster “domains · time adaptation · test time”
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.
15 papers in 2023–24 and 23 in 2025–26 (at ICML, NeurIPS, ICLR, CVPR and AAAI); its share of the topic grew, from 34.1% to 46.9%.
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 · Life-Long Disentangled Representation Learning with Cross-Domain Latent Homologies (NeurIPS 2018 · 75 citations)
- 2020 · Incremental Multi-Domain Learning with Network Latent Tensor Factorization (AAAI 2020 · 25 citations)
- 2021 · Continual Adaptation of Visual Representations via Domain Randomization and Meta-Learning (CVPR 2021 · 74 citations)
- 2022 · Learn From Others and Be Yourself in Heterogeneous Federated Learning (CVPR 2022 · 279 citations)
- 2023 · Unsupervised Continual Semantic Adaptation Through Neural Rendering (CVPR 2023 · 18 citations)
- 2024 · Multi-Domain Incremental Learning for Face Presentation Attack Detection (AAAI 2024 · 30 citations)
- 2025 · Dual Consolidation for Pre-Trained Model-Based Domain-Incremental Learning (CVPR 2025 · 14 citations)
- 2026 · Continual Learning of Domain-Invariant Representations (ICML 2026 · 0 citations)
Earliest papers here
- Life-Long Disentangled Representation Learning with Cross-Domain Latent Homologies (NeurIPS 2018 · 75 citations)
- Deep Face Detector Adaptation Without Negative Transfer or Catastrophic Forgetting (CVPR 2018 · 12 citations)
- Incremental Multi-Domain Learning with Network Latent Tensor Factorization (AAAI 2020 · 25 citations)
What has been tried
37 papers here report 417 measured changes (a component added, removed or swapped, compared with a baseline): helped 75%, no clear effect 6%, 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 |
|---|---|---|---|---|
| Replace separate expert classifiers with an asymmetric classifier | M, step 4, domain-agnostic · HTER (%) | 24.70 -> 10.62 | Helped | Multi-Domain Incremental Learning for Face Presentation Attack Detection (AAAI 2024 · 30 citations) |
| Add MLP bias to the FT-FC baseline | Pascal VOC series, session 4 · mAP | 44.2 -> 55.3 | Helped | Non-exemplar Domain Incremental Object Detection via Learning Domain Bias (AAAI 2024 · 19 citations) |
| Add weight sharing between sub-networks | CIFAR100/10 CI · ACC_TAG | +2.9% | Helped | MIND: Multi-Task Incremental Network Distillation (AAAI 2024 · 18 citations) |
| add representation consolidation by merging backbone task vectors | CDDB · average accuracy | 66.56 -> 80.36 | Helped | Dual Consolidation for Pre-Trained Model-Based Domain-Incremental Learning (CVPR 2025 · 14 citations) |
| Add pseudo-label blending | ACDC four-target sequence · mIoU Avg. (%) | 32.4 -> 34.0 | Helped | Semantic Segmentation in Multiple Adverse Weather Conditions with Domain Knowledge Retention (AAAI 2024 · 12 citations) |
| Use GMC with K = 2 instead of K = 1 | DomainNet · AT | 64.91 -> 65.25 | Helped | Boosting Domain Incremental Learning: Selecting the Optimal Parameters is All You Need (CVPR 2025 · 10 citations) |
| add global prototypes | ScanNet · MAE | 18.12 -> 14.59 | Helped | ProtoDepth: Unsupervised Continual Depth Completion with Prototypes (CVPR 2025 · 6 citations) |
| Replace [CLS] pooling with HOP | DSC small TIL · mAcc | 55.19 -> 85.63 | Helped | HOP to the Next Tasks and Domains for Continual Learning in NLP (AAAI 2024 · 6 citations) |
| Replace single-level prototypes with dual-level prototypes | DomainNet · average accuracy | 56.22 -> 60.13 | Helped | DualCP: Rehearsal-Free Domain-Incremental Learning via Dual-Level Concept Prototype (AAAI 2025 · 5 citations) |
| Add FFE and CFFS | S-CIFAR-10, Class-IL, buffer 50 · accuracy | 29.42 ±3.53 -> 45.56 ±3.71 | Helped | Continual Learning in the Frequency Domain (NeurIPS 2024 · 5 citations) |
| Increase the Primal-RAIL RHL dimension from 1k to 10k | X-TAIL (10 domains) · Last accuracy | 73.6% to 79.0% | Helped | Advancing Cross-domain Discriminability in Continual Learning of Vision-Language Models (NeurIPS 2024 · 5 citations) |
| add the noise-robust consistency loss | E-MNIST, Office-31, COIL-100, and COIL-20 · clustering performance | Helped | AdaptCMVC: Robust Adaption to Incremental Views in Continual Multi-view Clustering (CVPR 2025 · 4 citations) |
What reviewers will expect
Datasets and baselines named in the 14 papers of this approach from 2025–2026 that have a full-text summary. Leaving out the common ones is a frequent review question.
Datasets
- DomainNet · 4 of 14 papers
- CORe50 · 4 of 14 papers
- CDDB-Hard · 3 of 14 papers
- Office-Home · 2 of 14 papers
Baselines
- EWC · 4 of 14 papers
- LwF · 4 of 14 papers
- L2P · 4 of 14 papers
- DualPrompt · 4 of 14 papers
- DARE · 2 of 14 papers
- DyTox · 2 of 14 papers
- CODA-P · 2 of 14 papers
- S-Prompts · 2 of 14 papers
Recent preprints
Loading recent arXiv preprints…
Who publishes on it
Institutions on the most papers of this approach since 2023 (of 38 papers; from OpenAlex and the papers' affiliations).
- National University of Singapore · 3 papers
- Chinese Academy of Sciences · 2 papers
- Xi’an Jiaotong University · 2 papers
- Intel · 2 papers
- Nanjing University · 2 papers
- University of Chinese Academy of Sciences · 2 papers
- Nanjing University of Science and Technology · 2 papers
Use it
Download BibTeX of 34 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
- Continual test-time adaptation (25 papers)
Elsewhere in the topic
- Prompt-based continual learning (26 papers)
- Lifelong person re-identification (14 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).