Research map / Label-efficient and robust learning
Domain adaptation and generalization: research map
1,330 accepted papers on Domain adaptation and generalization in Label-efficient and robust learning, from ICML, NeurIPS, ICLR, CVPR and AAAI (2016–2026), grouped into 4 clusters and 17 approaches. The busiest year so far is 2024.
Within Label-efficient and robust learning, its share shrank from 11.9% in 2023–24 to 10.4% in 2025–26 (342 → 319 papers at ICML, NeurIPS, CVPR and AAAI, the venues with data for all four years).
Explore Domain adaptation and generalization 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.
Approaches and key papers
domain adaptation · unsupervised domain · uda · 627 papers
Approaches in this cluster:
- Conditional distribution alignment (162 papers)
Aligns conditional or disentangled distributions for domain adaptation, including universal and generalized shift. - Feature alignment for unsupervised adaptation (105 papers)
Progressively and dynamically aligns source and target features with analysis of limits. - Source-free domain adaptation (92 papers)
Adapts models without source data using distribution estimation and domain-invariant parameters. - Geometric alignment and synthetic-to-real adaptation (93 papers)
Aligns geometry and statistics, simplifies images, and adapts for pose or anytime settings. - Adversarial domain adaptation (95 papers)
Uses domain-adversarial learning with theory, symmetric networks and cluster assumptions. - Statistical alignment and hypothesis testing (80 papers)
Applies covariance, kernel and mutual-information alignment and statistical tests to adaptation.
Most cited and most cited since 2024:
- Adversarial Discriminative Domain Adaptation (CVPR 2017 · 5,087 citations)
- Maximum Classifier Discrepancy for Unsupervised Domain Adaptation (CVPR 2018 · 2,280 citations)
- Prompt-Based Distribution Alignment for Unsupervised Domain Adaptation (AAAI 2024 · 76 citations)
- Source-Free Domain Adaptation with Frozen Multimodal Foundation Model (CVPR 2024 · 50 citations)
cross domain · transfer · domain shot · 327 papers
Approaches in this cluster:
- Cross-domain few-shot with transformers (116 papers)
Adapts token- and adapter-based vision models to cross-domain few-shot learning and detection. - Mixture and distillation for domain shift (78 papers)
Handles domain shift via mixtures of experts, diversification and data mixture optimization. - Statistical transfer learning (79 papers)
Provides statistical methods and bounds for measuring and optimizing transfer between domains. - Cross-domain text and graph transfer (54 papers)
Transfers across domains in named entity recognition and graph recommendation via contrastive learning.
Most cited and most cited since 2024:
- Learning Multi-Domain Convolutional Neural Networks for Visual Tracking (CVPR 2016 · 2,839 citations)
- Learning to Discover Cross-Domain Relations with Generative Adversarial Networks (ICML 2017 · 726 citations)
- Day-Night Cross-domain Vehicle Re-identification (CVPR 2024 · 36 citations)
- Graph Disentangled Contrastive Learning with Personalized Transfer for Cross-Domain Recommendation (AAAI 2024 · 32 citations)
domain generalization · dg · unseen · 219 papers
Approaches in this cluster:
- Optimization-based domain generalization (101 papers)
Studies failure modes and gradient-guided or meta-learning methods for domain generalization. - Foundation-model domain generalization (56 papers)
Uses vision-language models and unsupervised or unknown-domain strategies for generalization. - Augmentation and mixture-of-experts generalization (51 papers)
Improves open-domain generalization via augmentation, meta-learning and sparse experts. - Generalizable person re-identification (11 papers)
Uses style normalization and domain-invariant mappings for person re-identification on unseen domains.
Most cited and most cited since 2024:
- Domain Generalization With Adversarial Feature Learning (CVPR 2018 · 1,271 citations)
- Domain Generalization by Solving Jigsaw Puzzles (CVPR 2019 · 862 citations)
- CFPL-FAS: Class Free Prompt Learning for Generalizable Face Anti-spoofing (CVPR 2024 · 79 citations)
- Gradient Alignment for Cross-Domain Face Anti-Spoofing (CVPR 2024 · 41 citations)
segmentation · semantic · domain adaptive · 157 papers
Approaches in this cluster:
- Adaptation for semantic segmentation (81 papers)
Adapts segmentation across domains with open-set, contextual and coarse-to-fine alignment. - Domain-generalized segmentation (56 papers)
Generalizes segmentation to unseen domains using style augmentation, memory and robust foundation-model tuning. - Cross-modality medical segmentation adaptation (20 papers)
Adapts medical segmentation across imaging modalities using frequency and volumetric alignment.
Most cited and most cited since 2024:
- Learning to Adapt Structured Output Space for Semantic Segmentation (CVPR 2018 · 1,793 citations)
- FDA: Fourier Domain Adaptation for Semantic Segmentation (CVPR 2020 · 1,119 citations)
- MAPSeg: Unified Unsupervised Domain Adaptation for Heterogeneous Medical Image Segmentation Based on 3D Masked Autoencoding and Pseudo-Labeling (CVPR 2024 · 35 citations)
- Style Blind Domain Generalized Semantic Segmentation via Covariance Alignment and Semantic Consistence Contrastive Learning (CVPR 2024 · 33 citations)
Related topics in Label-efficient and robust learning
- Noisy labels (462)
- Zero-/few-shot and meta-learning (784)
- Test-time adaptation (359)
- OOD and anomaly detection (614)
- Recognition and long-tail (4,953)
- Semi-supervised learning (623)
- Self-supervised and contrastive (988)
- Knowledge distillation (544)