Domain Generalization using Causal Matching

2020/06/12 by Divyat Mahajan, Shruti Tople, Mahajan, Divyat +3 · 27 citations
Computer Science · Mathematics · #Artificial Intelligence (cs.AI) #Domain Adaptation and Few-Shot Learning #FOS: Computer and information sciences #Machine Learning (cs.LG) #Machine Learning (stat.ML) #Multimodal Machine Learning Applications #Topic Modeling #cs.AI #cs.LG #stat.ML

paper · pdf · doi:10.48550/arxiv.2006.07500

Proceedings of the 38th International Conference on Machine Learning (ICML), PMLR 139, 2021. (Long Talk)

openalex publication_date 2020/06/12 · arxiv created 2021/06/29 · arxiv updated 2021/06/30 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28

Abstract

In the domain generalization literature, a common objective is to learn representations independent of the domain after conditioning on the class label. We show that this objective is not sufficient: there exist counter-examples where a model fails to generalize to unseen domains even after satisfying class-conditional domain invariance. We formalize this observation through a structural causal model and show the importance of modeling within-class variations for generalization. Specifically, classes contain objects that characterize specific causal features, and domains can be interpreted as interventions on these objects that change non-causal features. We highlight an alternative condition: inputs across domains should have the same representation if they are derived from the same object. Based on this objective, we propose matching-based algorithms when base objects are observed (e.g., through data augmentation) and approximate the objective when objects are not observed (MatchDG). Our simple matching-based algorithms are competitive to prior work on out-of-domain accuracy for rotated MNIST, Fashion-MNIST, PACS, and Chest-Xray datasets. Our method MatchDG also recovers ground-truth object matches: on MNIST and Fashion-MNIST, top-10 matches from MatchDG have over 50% overlap with ground-truth matches.

Citations

• Domain Generalization via Invariant Feature Representation
• Domain Generalization via Conditional Invariant Representation
• Contrastive Multiview Coding
• On Learning Invariant Representation for Domain Adaptation
• Invariant Risk Minimization
• Domain Generalization via Model-Agnostic Learning of Semantic Features
• Towards Shape Biased Unsupervised Representation Learning for Domain Generalization
• Momentum Contrast for Unsupervised Visual Representation Learning
• Distributionally Robust Neural Networks for Group Shifts: On the Importance of Regularization for Worst-Case Generalization
• A Simple Framework for Contrastive Learning of Visual Representations
• Invariant Risk Minimization Games
• Out-of-Distribution Generalization via Risk Extrapolation (REx)
• Improving out-of-distribution generalization via multi-task self-supervised pretraining
• Sequential Learning for Domain Generalization
• Generalizing to unseen domains via distribution matching
• The Pitfalls of Simplicity Bias in Neural Networks
• In Search of Lost Domain Generalization

Cited by

• Contrast transfer functions help quantify neural network out-of-distribution generalization in HRTEM
• Towards Unbiased Cross-Modal Representation Learning for Food Image-to-Recipe Retrieval
• An Analysis of Causal Effect Estimation using Outcome Invariant Data Augmentation
• Humanoid-inspired Causal Representation Learning for Domain Generalization
• Exploring Scale Shift in Crowd Localization under the Context of Domain Generalization
• MEASURE: Multi-scale Minimal Sufficient Representation Learning for Domain Generalization in Sleep Staging
• Test-Time Adaptation by Causal Trimming
• Single Domain Generalization in Diabetic Retinopathy: A Neuro-Symbolic Learning Approach
• Quantization Meets OOD: Generalizable Quantization-aware Training from a Flatness Perspective
• Class Unbiasing for Generalization in Medical Diagnosis
• Multimodal Causal-Driven Representation Learning for Generalizable Medical Image Segmentation
• Simulate, Refocus and Ensemble: An Attention-Refocusing Scheme for Domain Generalization
• Domain Generalization via Pareto Optimal Gradient Matching
• FEDTAIL: Federated Long-Tailed Domain Generalization with Sharpness-Guided Gradient Matching
• Text-Driven Causal Representation Learning for Source-Free Domain Generalization
• Incorporating Interventional Independence Improves Robustness against Interventional Distribution Shift
• Fair Deepfake Detectors Can Generalize
• Learning Time-Aware Causal Representation for Model Generalization in Evolving Domains
• DISCO: Mitigating Bias in Deep Learning with Conditional Distance Correlation
• Towards Data-Free Domain Generalization
• From Invariant Representations to Invariant Data: Provable Robustness to Spurious Correlations via Noisy Counterfactual Matching
• Bridging Distribution Shift and AI Safety: Conceptual and Methodological Synergies
• LangDAug: Langevin Data Augmentation for Multi-Source Domain Generalization in Medical Image Segmentation
• Iterative Feature Matching: Toward Provable Domain Generalization with Logarithmic Environments
• Energy-Aware Deep Learning on Resource-Constrained Hardware
• Alleviating Privacy Attacks via Causal Learning
• The Risks of Invariant Risk Minimization
• Context-Aware Self-Adaptation for Domain Generalization

Related

• Contrastive ACE: Domain Generalization Through Alignment of Causal Mechanisms
• Domain Generalization -- A Causal Perspective
• Causality Inspired Representation Learning for Domain Generalization
• Meta-causal Learning for Single Domain Generalization
• Domain Generalization via Contrastive Causal Learning