Learning Weighted Representations for Generalization Across Designs

Abstract

Predictive models that generalize well under distributional shift are often desirable and sometimes crucial to building robust and reliable machine learning applications. We focus on distributional shift that arises in causal inference from observational data and in unsupervised domain adaptation. We pose both of these problems as prediction under a shift in design. Popular methods for overcoming distributional shift make unrealistic assumptions such as having a well-specified model or knowing the policy that gave rise to the observed data. Other methods are hindered by their need for a pre-specified metric for comparing observations, or by poor asymptotic properties. We devise a bound on the generalization error under design shift, incorporating both representation learning and sample re-weighting. Based on the bound, we propose an algorithmic framework that does not require any of the above assumptions and which is asymptotically consistent. We empirically study the new framework using two synthetic datasets, and demonstrate its effectiveness compared to previous methods.

References in corpus (7)

• Learning Transferable Features with Deep Adaptation Networks
• Recursive Partitioning for Heterogeneous Causal Effects
• Domain Adaptation: Learning Bounds and Algorithms
• Domain-Adversarial Neural Networks
• Learning Representations for Counterfactual Inference
• On integral probability metrics, ϕ-divergences and binary classification
• Double/Debiased Machine Learning for Treatment and Causal Parameters

Cited by in corpus (17)

• A Review of Generalizability and Transportability
• Deep Generalized Method of Moments for Instrumental Variable Analysis
• Generalization Bounds and Representation Learning for Estimation of Potential Outcomes and Causal Effects
• A Survey on Causal Inference
• Learning Overlapping Representations for the Estimation of Individualized Treatment Effects
• Off-Policy Evaluation and Learning for External Validity under a Covariate Shift
• On Inductive Biases for Heterogeneous Treatment Effect Estimation
• Representation Balancing MDPs for Off-Policy Policy Evaluation
• SurvITE: Learning Heterogeneous Treatment Effects from Time-to-Event Data
• Model-Based Domain Generalization
• Counterfactual Representation Learning with Balancing Weights
• Density Ratio Estimation via Infinitesimal Classification
• Double Robust Representation Learning for Counterfactual Prediction
• Doing Great at Estimating CATE? On the Neglected Assumptions in Benchmark Comparisons of Treatment Effect Estimators
• Combining Incomplete Observational and Randomized Data for Heterogeneous Treatment Effects
• Learning Optimal Distributionally Robust Individualized Treatment Rules
• Quantifying Error in the Presence of Confounders for Causal Inference