mlsanityData to 7 Oct 2026

Scalable Out-of-distribution Robustness in the Presence of Unobserved Confounders

arXiv 2411.19923 ↗First posted 29 Nov 2024cs.LG stat.MLcited by 5

Not on the map. Why?

Abstract (arXiv)

We consider the task of out-of-distribution (OOD) generalization, where the distribution shift is due to an unobserved confounder (Z) affecting both the covariates (X) and the labels (Y). This confounding introduces heterogeneity in the predictor, i.e., P(Y | X) = EP(Z | X)[P(Y | X,Z)], making traditional covariate and label shift assumptions unsuitable. OOD generalization differs from traditional domain adaptation in that it does not assume access to the covariate distribution (Xte) of the test samples during training. These conditions create a challenging scenario for OOD robustness: (a) Ztr is an unobserved confounder during training, (b) Pte(Z) neq Ptr(Z), (c) Xte is unavailable during training, and (d) the predictive distribution depends on Pte(Z). While prior work has developed complex predictors requiring multiple additional variables for identifiability of the latent distribution, we explore a set of identifiability assumptions that yield a surprisingly simple predictor using only a single additional variable. Our approach demonstrates superior empirical performance on several benchmark tasks.

Key facts. Cited by 5 papers in the map (data through 7 Oct 2026). Builds on 36 of its 57 references that are in the map. First posted on arXiv 29 Nov 2024.

Builds on

36 of 57 references.

References · 36

  1. 2015Domain-Adversarial Training of Neural Networkscited by 3,023
  2. 2019Invariant Risk Minimizationcited by 1,832
  3. 2019Distributionally Robust Neural Networks for Group Shifts: On the Importance of Regularization for Worst-Case Generalizationcited by 1,232
  4. 2016Deep CORAL: Correlation Alignment for Deep Domain Adaptationcited by 1,194
  5. 2020In Search of Lost Domain Generalizationcited by 1,042
  6. 2020Out-of-Distribution Generalization via Risk Extrapolation (REx)cited by 737
  7. 2021Just Train Twice: Improving Group Robustness without Training Group Informationcited by 522
  8. 2021Retiring Adult: New Datasets for Fair Machine Learningcited by 443
  9. 2023Towards Stable Test-Time Adaptation in Dynamic Wild Worldcited by 409
  10. 2022Last Layer Re-Training is Sufficient for Robustness to Spurious Correlationscited by 379
  11. 2018Detecting and Correcting for Label Shift with Black Box Predictorscited by 364
  12. 2020Environment Inference for Invariant Learningcited by 306
24 more
  1. 2020The Risks of Invariant Risk Minimizationcited by 262
  2. 2020Domain Generalization using Causal Matchingcited by 224
  3. 2020Adaptive Risk Minimization: Learning to Adapt to Domain Shiftcited by 183
  4. 2016Identifying Causal Effects With Proxy Variables of an Unmeasured Confoundercited by 138
  5. 2020An Introduction to Proximal Causal Learningcited by 91
  6. 2021Heterogeneous Risk Minimizationcited by 90
  7. 2022Spread Spurious Attribute: Improving Worst-group Accuracy with Spurious Attribute Estimationcited by 88
  8. 2020A causal framework for distribution generalizationcited by 68
  9. 2022ZIN: When and How to Learn Invariance Without Environment Partition?cited by 58
  10. 2022Diagnosing failures of fairness transfer across distribution shift in real-world medical settingscited by 45
  11. 2022Out-of-distribution Generalization with Causal Invariant Transformationscited by 45
  12. 2022Harnessing Out-Of-Distribution Examples via Augmenting Content and Stylecited by 43
  13. 2020Rao-Blackwellizing the Straight-Through Gumbel-Softmax Gradient Estimatorcited by 37
  14. 2022Modeling the Data-Generating Process is Necessary for Out-of-Distribution Generalizationcited by 34
  15. 2023RLSbench: Domain Adaptation Under Relaxed Label Shiftcited by 30
  16. 2023Spuriosity Didn't Kill the Classifier: Using Invariant Predictions to Harness Spurious Featurescited by 21
  17. 2022Causal Balancing for Domain Generalizationcited by 20
  18. 2022Robust Causal Graph Representation Learning against Confounding Effectscited by 12
  19. 2022Adapting to Latent Subgroup Shifts via Concepts and Proxiescited by 12
  20. 2022Beyond Invariance: Test-Time Label-Shift Adaptation for Distributions with "Spurious" Correlationscited by 7
  21. 2024Proxy Methods for Domain Adaptationcited by 6
  22. 2023Kernel Single Proxy Control for Deterministic Confoundingcited by 5
  23. 2014How to Scale Up Kernel Methods to Be As Good As Deep Neural Netscited by 4
  24. 2022Enhancing Distributional Stability among Sub-populationscited by 3

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2025: 3; 2026: 2.

5 papers.

  1. 2025When Shift Happens - Confounding Is to Blamecited by 5
  2. 2026Counterfactual Residual Data Augmentation for Regressioncited by 1
  3. 2025Should Bias be Eliminated? A General Framework to Use Bias for OOD Generalizationcited by 0
  4. 2025Transferring Causal Effects using Proxiescited by 0
  5. 2026Point-Identification of a Robust Predictor Under Latent Shift with Imperfect Proxiescited by 0