mlsanityData to 7 Oct 2026

Improving Domain Generalization with Domain Relations

arXiv 2302.02609 ↗First posted 6 Feb 2023cs.LGcited by 14

Not on the map. Why?

Abstract (arXiv)

Distribution shift presents a significant challenge in machine learning, where models often underperform during the test stage when faced with a different distribution than the one they were trained on. This paper focuses on domain shifts, which occur when the model is applied to new domains that are different from the ones it was trained on, and propose a new approach called D3G. Unlike previous methods that aim to learn a single model that is domain invariant, D3G leverages domain similarities based on domain metadata to learn domain-specific models. Concretely, D3G learns a set of training-domain-specific functions during the training stage and reweights them based on domain relations during the test stage. These domain relations can be directly obtained and learned from domain metadata. Under mild assumptions, we theoretically prove that using domain relations to reweight training-domain-specific functions achieves stronger out-of-domain generalization compared to the conventional averaging approach. Empirically, we evaluate the effectiveness of D3G using real-world datasets for tasks such as temperature regression, land use classification, and molecule-protein binding affinity prediction. Our results show that D3G consistently outperforms state-of-the-art methods.

Key facts. Cited by 14 papers in the map (data through 7 Oct 2026). Builds on 51 of its 67 references that are in the map. First posted on arXiv 6 Feb 2023.

Builds on

51 of 67 references.

References · 51

  1. 2016Densely Connected Convolutional Networkscited by 6,691
  2. 2018How Powerful are Graph Neural Networks?cited by 4,355
  3. 2016Simple and Scalable Predictive Uncertainty Estimation using Deep Ensemblescited by 3,888
  4. 2017Outrageously Large Neural Networks: The Sparsely-Gated Mixture-of-Experts Layercited by 3,413
  5. 2015Domain-Adversarial Training of Neural Networkscited by 3,023
  6. 2019Invariant Risk Minimizationcited by 1,832
  7. 2020WILDS: A Benchmark of in-the-Wild Distribution Shiftscited by 1,366
  8. 2015Learning Transferable Features with Deep Adaptation Networkscited by 1,323
  9. 2019Distributionally Robust Neural Networks for Group Shifts: On the Importance of Regularization for Worst-Case Generalizationcited by 1,232
  10. 2016Deep CORAL: Correlation Alignment for Deep Domain Adaptationcited by 1,194
  11. 2020In Search of Lost Domain Generalizationcited by 1,042
  12. 2019Strategies for Pre-training Graph Neural Networkscited by 964
38 more
  1. 2020Out-of-Distribution Generalization via Risk Extrapolation (REx)cited by 737
  2. 2016Language Modeling with Gated Convolutional Networkscited by 688
  3. 2014Deep Domain Confusion: Maximizing for Domain Invariancecited by 597
  4. 2022Continual Test-Time Domain Adaptationcited by 520
  5. 2020Improving robustness against common corruptions by covariate shift adaptationcited by 427
  6. 2020Self-Challenging Improves Cross-Domain Generalizationcited by 426
  7. 2021MEMO: Test Time Robustness via Adaptation and Augmentationcited by 376
  8. 2013Learning Factored Representations in a Deep Mixture of Expertscited by 275
  9. 2019Adversarial Domain Adaptation with Domain Mixupcited by 250
  10. 2019Domain Randomization and Pyramid Consistency: Simulation-to-Real Generalization without Accessing Target Domain Datacited by 245
  11. 2020Deep Domain-Adversarial Image Generation for Domain Generalisationcited by 231
  12. 2021Invariance Principle Meets Information Bottleneck for Out-of-Distribution Generalizationcited by 220
  13. 2020Iterative Deep Graph Learning for Graph Neural Networks: Better and Robust Node Embeddingscited by 212
  14. 2022Improving Out-of-Distribution Robustness via Selective Augmentationcited by 202
  15. 2021SelfReg: Self-supervised Contrastive Regularization for Domain Generalizationcited by 189
  16. 2020Improve Unsupervised Domain Adaptation with Mixup Trainingcited by 185
  17. 2020Geometric Dataset Distances via Optimal Transportcited by 157
  18. 2021Open Domain Generalization with Domain-Augmented Meta-Learningcited by 115
  19. 2020Heterogeneous Domain Generalization via Domain Mixupcited by 103
  20. 2021Exploiting Domain-Specific Features to Enhance Domain Generalizationcited by 86
  21. 2022A Whac-A-Mole Dilemma: Shortcuts Come in Multiples Where Mitigating One Amplifies Otherscited by 85
  22. 2021Evading the Simplicity Bias: Training a Diverse Set of Models Discovers Solutions with Superior OOD Generalizationcited by 83
  23. 2021Generalizable Person Re-identification with Relevance-aware Mixture of Expertscited by 77
  24. 2023DELTA: degradation-free fully test-time adaptationcited by 72
  25. 2022Agree to Disagree: Diversity through Disagreement for Better Transferabilitycited by 69
  26. 2021MT3: Meta Test-Time Training for Self-Supervised Test-Time Adaptioncited by 65
  27. 2022C-Mixup: Improving Generalization in Regressioncited by 57
  28. 2022Diversify and Disambiguate: Learning From Underspecified Datacited by 55
  29. 2022DrugOOD: Out-of-Distribution (OOD) Dataset Curator and Benchmark for AI-aided Drug Discovery -- A Focus on Affinity Prediction Problems with Noise Annotationscited by 54
  30. 2021Optimal Representations for Covariate Shiftcited by 51
  31. 2019AdaGraph: Unifying Predictive and Continuous Domain Adaptation through Graphscited by 32
  32. 2022Domain Generalization using Pretrained Models without Fine-tuningcited by 29
  33. 2021Out-of-distribution Prediction with Invariant Risk Minimization: The Limitation and An Effective Fixcited by 27
  34. 2022Graph-Relational Domain Adaptationcited by 19
  35. 2022Improving Out-of-Distribution Generalization by Adversarial Training with Structured Priorscited by 17
  36. 2022Domain Generalization for Activity Recognition via Adaptive Feature Fusioncited by 16
  37. 2022Structured Multi-task Learning for Molecular Property Predictioncited by 12
  38. 2021On Invariance Penalties for Risk Minimizationcited by 10

50 most cited of 51 All 51 (JSON)

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2023: 3; 2024: 3; 2025: 4; 2026: 4.

14 papers.

Sort:

By area:

  1. 2024CrossEarth: Geospatial Vision Foundation Model for Domain Generalizable Remote Sensing Semantic Segmentationcited by 27
  2. 2023Symmetry-Informed Geometric Representation for Molecules, Proteins, and Crystalline Materialscited by 22
  3. 2023GraphMETRO: Mitigating Complex Graph Distribution Shifts via Mixture of Aligned Expertscited by 19
  4. 2023Confidence-Based Model Selection: When to Take Shortcuts for Subpopulation Shiftscited by 5
  5. 2024Hybrid-TTA: Continual Test-time Adaptation via Dynamic Domain Shift Detectioncited by 5
  6. 2025CoDEx: Combining Domain Expertise for Spatial Generalization in Satellite Image Analysiscited by 2
  7. 2025From Invariant Representations to Invariant Data: Provable Robustness to Spurious Correlations via Noisy Counterfactual Matchingcited by 2
  8. 2025Domain Generalization Under Posterior Driftcited by 2
  9. 2024Reliable OOD Virtual Screening with Extrapolatory Pseudo-Label Matchingcited by 1
  10. 2025Latent Domain Modeling Improves Robustness to Geographic Shiftscited by 1
  11. 2026OT on the Map: Quantifying Domain Shifts in Geographic Spacecited by 1
  12. 2026SemAlign: Language Guided Semi-supervised Domain Generalizationcited by 0
2 more
  1. 2026Language-Induced Priors for Domain Adaptationcited by 0
  2. 2026Learning Subset-Shared Invariances for Domain Generalization with Mixture-of-Expertscited by 0