Research map / Continual learning
Domain and prompt-based: research map
138 papers on Domain and prompt-based in Continual learning, from ICML, NeurIPS, ICLR, CVPR, AAAI, ICCV and ECCV, and the journals TPAMI, IJCV and JMLR (2016–2026), grouped into 2 clusters and 4 approaches. The busiest year so far is 2024.
What it is
Here the task stays the same but the data keeps changing, for example new cameras, cities or styles. Many recent methods keep a large pretrained model frozen and learn small "prompts" for each new setting, so the model adapts without overwriting what it knew.
Terms you will see on this page:
- Prompt-based continual learning
- learning a few extra input tokens per task instead of retraining the whole model.
- Domain-incremental learning
- the label set stays fixed while the input distribution shifts over time.
- Test-time adaptation
- updating a model on the fly using only the unlabeled data it sees while deployed.
Start reading with the most cited papers here:
- Learning To Prompt for Continual Learning (CVPR 2022 · 762 citations)
- Continual Test-Time Domain Adaptation (CVPR 2022 · 484 citations)
Within Continual learning, its share shrank from 12.5% in 2023–24 to 8.5% in 2025–26 (44 → 49 papers at ICML, NeurIPS, ICLR, CVPR and AAAI, the venues with data for all four years).
Explore Domain and prompt-based in the interactive map
Working on something in this topic? Describe your idea in scime atlas to see which approach it falls under, the closest papers by meaning and how crowded the spot has become.
What changed recently
Approaches whose share of the topic moved most between 2023–24 and 2025–26 (ICML, NeurIPS, ICLR, CVPR and AAAI):
- Domain-incremental learning with memory rising: 34% → 47% of the topic (15 → 23 papers)
- Lifelong person re-identification rising: 9% → 12% of the topic (4 → 6 papers)
- Prompt-based continual learning falling: 32% → 16% of the topic (14 → 8 papers)
Approaches and key papers
Shifting domains and test-time adaptation · 84 papers
Keywords: domains, time adaptation, test time
Approaches in this cluster:
- Domain-incremental learning with memory (54 papers)
Adapts to sequentially arriving domains using memory, domain-aware representations and randomization. - Continual test-time adaptation (30 papers)
Adapts models online to shifting target domains using adapters, prompts and knowledge fusion.
Most cited and most cited since 2024:
- Continual Test-Time Domain Adaptation (CVPR 2022 · 484 citations)
- Learn From Others and Be Yourself in Heterogeneous Federated Learning (CVPR 2022 · 279 citations)
- Each Test Image Deserves A Specific Prompt: Continual Test-Time Adaptation for 2D Medical Image Segmentation (CVPR 2024 · 44 citations)
- Multi-Domain Incremental Learning for Face Presentation Attack Detection (AAAI 2024 · 30 citations)
Prompts for new tasks and identities · 54 papers
Keywords: prompts, person, prompting
Approaches in this cluster:
- Prompt-based continual learning (32 papers)
Learns small prompt sets over frozen pretrained models without rehearsal buffers. - Lifelong person re-identification (22 papers)
Accumulates and consolidates knowledge across domains for person re-identification with rehearsal or distillation.
Most cited and most cited since 2024:
- Learning To Prompt for Continual Learning (CVPR 2022 · 762 citations)
- CODA-Prompt: COntinual Decomposed Attention-Based Prompting for Rehearsal-Free Continual Learning (CVPR 2023 · 320 citations)
- Unsupervised Continual Anomaly Detection with Contrastively-Learned Prompt (AAAI 2024 · 39 citations)
- Consistent Prompting for Rehearsal-Free Continual Learning (CVPR 2024 · 38 citations)
Related topics in Continual learning
- Online and replay-based (693)
- Class-incremental (450)
- Continual fine-tuning of LLMs (315)
