mlsanityData to 13 Oct 2026

Diagnosing and Rectifying Fake OOD Invariance: A Restructured Causal Approach

arXiv 2312.09758 ↗First posted 15 Dec 2023cs.LG cs.AI stat.MEcited by 1

Not on the map. Why?

Abstract (arXiv)

Invariant representation learning (IRL) encourages the prediction from invariant causal features to labels de-confounded from the environments, advancing the technical roadmap of out-of-distribution (OOD) generalization. Despite spotlights around, recent theoretical results verified that some causal features recovered by IRLs merely pretend domain-invariantly in the training environments but fail in unseen domains. The fake invariance severely endangers OOD generalization since the trustful objective can not be diagnosed and existing causal surgeries are invalid to rectify. In this paper, we review a IRL family (InvRat) under the Partially and Fully Informative Invariant Feature Structural Causal Models (PIIF SCM /FIIF SCM) respectively, to certify their weaknesses in representing fake invariant features, then, unify their causal diagrams to propose ReStructured SCM (RS-SCM). RS-SCM can ideally rebuild the spurious and the fake invariant features simultaneously. Given this, we further develop an approach based on conditional mutual information with respect to RS-SCM, then rigorously rectify the spurious and fake invariant effects. It can be easily implemented by a small feature selection subnet introduced in the IRL family, which is alternatively optimized to achieve our goal. Experiments verified the superiority of our approach to fight against the fake invariant issue across a variety of OOD generalization benchmarks.

Key facts. Cited by 1 paper in the map (data through 13 Oct 2026). Builds on 33 of its 37 references that are in the map. First posted on arXiv 15 Dec 2023.

Builds on

33 of 37 references.

References · 33

  1. 2015Deep Residual Learning for Image Recognitioncited by 51,950
  2. 2015Domain-Adversarial Training of Neural Networkscited by 3,030
  3. 2019Invariant Risk Minimizationcited by 1,842
  4. 2020In Search of Lost Domain Generalizationcited by 1,048
  5. 2016Deep Variational Information Bottleneckcited by 924
  6. 2021Generalizing to Unseen Domains: A Survey on Domain Generalizationcited by 764
  7. 2020Out-of-Distribution Generalization via Risk Extrapolation (REx)cited by 738
  8. 2018MINE: Mutual Information Neural Estimationcited by 682
  9. 2015Causal inference using invariant prediction: identification and confidence intervalscited by 534
  10. 2021Towards Out-Of-Distribution Generalization: A Surveycited by 432
  11. 2020The Risks of Invariant Risk Minimizationcited by 262
  12. 2020Domain Generalization using Causal Matchingcited by 224
21 more
  1. 2021Invariance Principle Meets Information Bottleneck for Out-of-Distribution Generalizationcited by 220
  2. 2020Invariant Risk Minimization Gamescited by 188
  3. 2020Understanding the Failure Modes of Out-of-Distribution Generalizationcited by 146
  4. 2022Causality Inspired Representation Learning for Domain Generalizationcited by 137
  5. 2020Invariant Rationalizationcited by 131
  6. 2021Does Invariant Risk Minimization Capture Invariance?cited by 116
  7. 2021Open Domain Generalization with Domain-Augmented Meta-Learningcited by 115
  8. 2020Heterogeneous Domain Generalization via Domain Mixupcited by 103
  9. 2021Invariant Information Bottleneck for Domain Generalizationcited by 86
  10. 2022ZIN: When and How to Learn Invariance Without Environment Partition?cited by 58
  11. 2018Stable Prediction across Unknown Environmentscited by 55
  12. 2022Out-of-distribution Generalization with Causal Invariant Transformationscited by 45
  13. 2023Meta-causal Learning for Single Domain Generalizationcited by 43
  14. 2022Invariant and Transportable Representations for Anti-Causal Domain Shiftscited by 37
  15. 2022Invariant Causal Mechanisms through Distribution Matchingcited by 31
  16. 2020Domain adaptation under structural causal modelscited by 28
  17. 2018Domain Randomization for Scene-Specific Car Detection and Pose Estimationcited by 23
  18. 2023Distribution Shift Inversion for Out-of-Distribution Predictioncited by 23
  19. 2022Cross-Domain Ensemble Distillation for Domain Generalizationcited by 21
  20. 2022The Missing Invariance Principle Found -- the Reciprocal Twin of Invariant Risk Minimizationcited by 8
  21. 2023On Heterogeneous Treatment Effects in Heterogeneous Causal Graphscited by 2

Built on by

Cited by year

Citing papers per year: 2025: 1.

1 paper.

  1. 2025Causally Guided Gaussian Perturbations for Out-Of-Distribution Generalization in Medical Imagingcited by 0