Learning Under Adversarial and Interventional Shifts
arXiv:2103.15933
Abstract
Machine learning models are often trained on data from one distribution and deployed on others. So it becomes important to design models that are robust to distribution shifts. Most of the existing work focuses on optimizing for either adversarial shifts or interventional shifts. Adversarial methods lack expressivity in representing plausible shifts as they consider shifts to joint distributions in the data. Interventional methods allow more expressivity but provide robustness to unbounded shifts, resulting in overly conservative models. In this work, we combine the complementary strengths of the two approaches and propose a new formulation, RISe, for designing robust models against a set of distribution shifts that are at the intersection of adversarial and interventional shifts. We employ the distributionally robust optimization framework to optimize the resulting objective in both supervised and reinforcement learning settings. Extensive experimentation with synthetic and real world datasets from healthcare demonstrate the efficacy of the proposed approach.
arXiv comment19 pages including 5 pages appendix, 6 figures, 2 tables. Preliminary version presented at Causal Discovery & Causality-Inspired Machine Learning Workshop 2020
References in corpus (13)
- Invariant Risk Minimization
- Out-of-Distribution Generalization via Risk Extrapolation (REx)
- Measuring Robustness to Natural Distribution Shifts in Image Classification
- Invariant Models for Causal Transfer Learning
- Does Distributionally Robust Supervised Learning Give Robust Classifiers?
- The Risks of Invariant Risk Minimization
- Domain Adaptation by Using Causal Inference to Predict Invariant Conditional Distributions
- Anchor regression: heterogeneous data meets causality
- Counterfactual Off-Policy Evaluation with Gumbel-Max Structural Causal Models
- Understanding and Mitigating the Tradeoff Between Robustness and Accuracy
- Robustness to Spurious Correlations via Human Annotations
- Off-Policy Evaluation and Learning for External Validity under a Covariate Shift
- Optimizing Medical Treatment for Sepsis in Intensive Care: from Reinforcement Learning to Pre-Trial Evaluation