Linear Regression Games: Convergence Guarantees to Approximate Out-of-Distribution Solutions

Abstract

Recently, invariant risk minimization (IRM) (Arjovsky et al.) was proposed as a promising solution to address out-of-distribution (OOD) generalization. In Ahuja et al., it was shown that solving for the Nash equilibria of a new class of "ensemble-games" is equivalent to solving IRM. In this work, we extend the framework in Ahuja et al. for linear regressions by projecting the ensemble-game on an ${\ell }_{\infty }$ ball. We show that such projections help achieve non-trivial OOD guarantees despite not achieving perfect invariance. For linear models with confounders, we prove that Nash equilibria of these games are closer to the ideal OOD solutions than the standard empirical risk minimization (ERM) and we also provide learning algorithms that provably converge to these Nash Equilibria. Empirical comparisons of the proposed approach with the state-of-the-art show consistent gains in achieving OOD solutions in several settings involving anti-causal variables and confounders.

References in corpus (12)

• Invariant Risk Minimization
• Agnostic Federated Learning
• Out-of-Distribution Generalization via Risk Extrapolation (REx)
• Domain-Adversarial Neural Networks
• On Causal and Anticausal Learning
• Algorithms and Theory for Multiple-Source Adaptation
• Domain Adaptation by Using Causal Inference to Predict Invariant Conditional Distributions
• In Search of Lost Domain Generalization
• Minimax Statistical Learning with Wasserstein Distances
• Unshuffling Data for Improved Generalization
• Invariant Risk Minimization Games
• When is invariance useful in an Out-of-Distribution Generalization problem ?

Cited by in corpus (2)

• Gradient Starvation: A Learning Proclivity in Neural Networks
• A call for better unit testing for invariant risk minimisation