Towards Principled Disentanglement for Domain Generalization

Abstract

A fundamental challenge for machine learning models is generalizing to out-of-distribution (OOD) data, in part due to spurious correlations. To tackle this challenge, we first formalize the OOD generalization problem as constrained optimization, called Disentanglement-constrained Domain Generalization (DDG). We relax this non-trivial constrained optimization problem to a tractable form with finite-dimensional parameterization and empirical approximation. Then a theoretical analysis of the extent to which the above transformations deviates from the original problem is provided. Based on the transformation, we propose a primal-dual algorithm for joint representation disentanglement and domain generalization. In contrast to traditional approaches based on domain adversarial training and domain labels, DDG jointly learns semantic and variation encoders for disentanglement, enabling flexible manipulation and augmentation on training data. DDG aims to learn intrinsic representations of semantic concepts that are invariant to nuisance factors and generalizable across domains. Comprehensive experiments on popular benchmarks show that DDG can achieve competitive OOD performance and uncover interpretable salient structures within data.

arXiv commentCVPR 2022 Oral

References in corpus (19)

• Learning Transferable Features with Deep Adaptation Networks
• Shortcut Learning in Deep Neural Networks
• Unsupervised Data Augmentation for Consistency Training
• Which Training Methods for GANs do actually Converge?
• Disentangling by Factorising
• Domain Generalization via Invariant Feature Representation
• Distributionally Robust Neural Networks for Group Shifts: On the Importance of Regularization for Worst-Case Generalization
• Invariant Risk Minimization
• Fairness Without Demographics in Repeated Loss Minimization
• Invariant Models for Causal Transfer Learning
• Learning Disentangled Semantic Representation for Domain Adaptation
• The Risks of Invariant Risk Minimization
• Generalizing to unseen domains via distribution matching
• Controlling generative models with continuous factors of variations
• When Unseen Domain Generalization is Unnecessary? Rethinking Data Augmentation
• Nonlinear Invariant Risk Minimization: A Causal Approach
• Does Invariant Risk Minimization Capture Invariance?
• Constrained Learning with Non-Convex Losses
• Model-Based Domain Generalization