mlsanityData to 7 Oct 2026

Out-of-distribution Generalization in the Presence of Nuisance-Induced Spurious Correlations

arXiv 2107.00520 ↗First posted 29 Jun 2021cs.LG stat.MLcited by 41

Not on the map. Why?

Abstract (arXiv)

In many prediction problems, spurious correlations are induced by a changing relationship between the label and a nuisance variable that is also correlated with the covariates. For example, in classifying animals in natural images, the background, which is a nuisance, can predict the type of animal. This nuisance-label relationship does not always hold, and the performance of a model trained under one such relationship may be poor on data with a different nuisance-label relationship. To build predictive models that perform well regardless of the nuisance-label relationship, we develop Nuisance-Randomized Distillation (NURD). We introduce the nuisance-randomized distribution, a distribution where the nuisance and the label are independent. Under this distribution, we define the set of representations such that conditioning on any member, the nuisance and the label remain independent. We prove that the representations in this set always perform better than chance, while representations outside of this set may not. NURD finds a representation from this set that is most informative of the label under the nuisance-randomized distribution, and we prove that this representation achieves the highest performance regardless of the nuisance-label relationship. We evaluate NURD on several tasks including chest X-ray classification where, using non-lung patches as the nuisance, NURD produces models that predict pneumonia under strong spurious correlations.

Key facts. Cited by 41 papers in the map (data through 7 Oct 2026). Builds on 39 of its 51 references that are in the map. First posted on arXiv 29 Jun 2021.

Builds on

39 of 51 references.

References · 39

  1. 2014Adam: A Method for Stochastic Optimizationcited by 36,960
  2. 2017On Calibration of Modern Neural Networkscited by 4,860
  3. 2016Simple and Scalable Predictive Uncertainty Estimation using Deep Ensemblescited by 3,888
  4. 2014Deep Learning Face Attributes in the Wildcited by 3,679
  5. 2015Domain-Adversarial Training of Neural Networkscited by 3,023
  6. 2020Shortcut Learning in Deep Neural Networkscited by 1,928
  7. 2019Invariant Risk Minimizationcited by 1,832
  8. 2019CheXpert: A Large Chest Radiograph Dataset with Uncertainty Labels and Expert Comparisoncited by 1,410
  9. 2019Generating Diverse High-Fidelity Images with VQ-VAE-2cited by 1,376
  10. 2019Distributionally Robust Neural Networks for Group Shifts: On the Importance of Regularization for Worst-Case Generalizationcited by 1,232
  11. 2020In Search of Lost Domain Generalizationcited by 1,042
  12. 2017Learning to Generalize: Meta-Learning for Domain Generalizationcited by 762
27 more
  1. 2020Out-of-Distribution Generalization via Risk Extrapolation (REx)cited by 737
  2. 2018Mitigating Unwanted Biases with Adversarial Learningcited by 663
  3. 2019MIMIC-CXR-JPG, a large publicly available database of labeled chest radiographscited by 590
  4. 2018Confounding variables can degrade generalization performance of radiological deep learning modelscited by 374
  5. 2020An Investigation of Why Overparameterization Exacerbates Spurious Correlationscited by 310
  6. 2020Environment Inference for Invariant Learningcited by 306
  7. 2020A Brief Review of Domain Adaptationcited by 287
  8. 2020Fairness without Demographics through Adversarially Reweighted Learningcited by 251
  9. 2018What is the Effect of Importance Weighting in Deep Learning?cited by 250
  10. 2020Domain Generalization using Causal Matchingcited by 224
  11. 2016Statistics of Robust Optimization: A Generalized Empirical Likelihood Approachcited by 182
  12. 2021On Calibration and Out-of-domain Generalizationcited by 134
  13. 2020Invariant Rationalizationcited by 131
  14. 2017Controllable Invariance through Adversarial Feature Learningcited by 127
  15. 2020Model Patching: Closing the Subgroup Performance Gap with Data Augmentationcited by 82
  16. 2021Counterfactual Invariance to Spurious Correlations: Why and How to Pass Stress Testscited by 77
  17. 2021Can Subnetwork Structure be the Key to Out-of-Distribution Generalization?cited by 75
  18. 2020Robustness to Spurious Correlations in Text Classification via Automatically Generated Counterfactualscited by 67
  19. 2020Identifying Spurious Correlations for Robust Text Classificationcited by 59
  20. 2019Can we trust deep learning models diagnosis? The impact of domain shift in chest radiograph classificationcited by 57
  21. 2021Causally motivated Shortcut Removal Using Auxiliary Labelscited by 55
  22. 2018Deep Learning Predicts Hip Fracture using Confounding Patient and Healthcare Variablescited by 50
  23. 2019Adversarial Invariant Feature Learning with Accuracy Constraint for Domain Generalizationcited by 43
  24. 2021X-CAL: Explicit Calibration for Survival Analysiscited by 30
  25. 2021Out-of-distribution Prediction with Invariant Risk Minimization: The Limitation and An Effective Fixcited by 27
  26. 2020Accounting for Unobserved Confounding in Domain Generalizationcited by 14
  27. 2020Learning Invariant Feature Representation to Improve Generalization across Chest X-ray Datasetscited by 6

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2021: 1; 2022: 7; 2023: 15; 2024: 8; 2025: 6; 2026: 4.

41 papers.

Sort:

By area:

  1. 2023Change is Hard: A Closer Look at Subpopulation Shiftcited by 119
  2. 2023Domain Generalization for Medical Image Analysis: A Reviewcited by 47
  3. 2024On the Challenges and Opportunities in Generative AIcited by 31
  4. 2023Don't blame Dataset Shift! Shortcut Learning due to Gradients and Cross Entropycited by 30
  5. 2022Are All Spurious Features in Natural Language Alike? An Analysis through a Causal Lenscited by 28
  6. 2023Spurious Feature Diversification Improves Out-of-distribution Generalizationcited by 26
  7. 2023Spuriosity Didn't Kill the Classifier: Using Invariant Predictions to Harness Spurious Featurescited by 21
  8. 2022Causal Balancing for Domain Generalizationcited by 20
  9. 2024A Survey on Evaluation of Out-of-Distribution Generalizationcited by 20
  10. 2022Efficient Conditionally Invariant Representation Learningcited by 19
  11. 2022Nuisances via Negativa: Adjusting for Spurious Correlations via Data Augmentationcited by 18
  12. 2023When More is Less: Incorporating Additional Datasets Can Hurt Performance By Introducing Spurious Correlationscited by 18
29 more
  1. 2023Data Augmentations for Improved (Large) Language Model Generalizationcited by 14
  2. 2023When accurate prediction models yield harmful self-fulfilling propheciescited by 12
  3. 2023Beyond Distribution Shift: Spurious Features Through the Lens of Training Dynamicscited by 11
  4. 2022Evaluating Robustness to Dataset Shift via Parametric Robustness Setscited by 10
  5. 2023SFP: Spurious Feature-targeted Pruning for Out-of-Distribution Generalizationcited by 9
  6. 2024Mind the Graph When Balancing Data for Fairness or Robustnesscited by 9
  7. 2022Beyond Invariance: Test-Time Label-Shift Adaptation for Distributions with "Spurious" Correlationscited by 7
  8. 2026To Use or not to Use Muon: How Simplicity Bias in Optimizers Matterscited by 7
  9. 2022Malign Overfitting: Interpolation Can Provably Preclude Invariancecited by 6
  10. 2023Are demographically invariant models and representations in medical imaging fair?cited by 6
  11. 2023Removing Spurious Concepts from Neural Network Representations via Joint Subspace Estimationcited by 5
  12. 2021Learning Invariant Representations with Missing Datacited by 4
  13. 2023Robustness to Spurious Correlations Improves Semantic Out-of-Distribution Detectioncited by 4
  14. 2023A Causal Framework to Unify Common Domain Generalization Approachescited by 3
  15. 2024Explanations that reveal all through the definition of encodingcited by 3
  16. 2023Boosted Control Functions: Distribution generalization and invariance in confounded modelscited by 2
  17. 2025Bridging Distribution Shift and AI Safety: Conceptual and Methodological Synergiescited by 2
  18. 2025Superclass-Guided Representation Disentanglement for Spurious Correlation Mitigationcited by 2
  19. 2024Robust Anomaly Detection for Particle Physics Using Multi-Background Representation Learningcited by 1
  20. 2024Multi-environment Topic Modelscited by 1
  21. 2025Does Weak-to-strong Generalization Happen under Spurious Correlations?cited by 1
  22. 2024Not Eliminate but Aggregate: Post-Hoc Control over Mixture-of-Experts to Address Shortcut Shifts in Natural Language Understandingcited by 0
  23. 2024Considerations for Distribution Shift Robustness of Diagnostic Models in Healthcarecited by 0
  24. 2025Meta Learning not to Learn: Robustly Informing Meta-Learning under Nuisance-Varying Familiescited by 0
  25. 2025DISCO: Mitigating Bias in Deep Learning with Conditional Distance Correlationcited by 0
  26. 2025Invariant Feature Extraction Through Conditional Independence and the Optimal Transport Barycenter Problem: the Gaussian casecited by 0
  27. 2026Shortcut Features as Top Eigenfunctions of NTK: A Linear Neural Network Case and Morecited by 0
  28. 2026Is invariance all you need for algorithmic fairness? Removing demographic information can create new biascited by 0
  29. 2026PhysioTRACE: Provenance-Aware Stress Tests for Physiological Foundation Modelscited by 0