Learning Representations for Counterfactual Inference

Abstract

Observational studies are rising in importance due to the widespread accumulation of data in fields such as healthcare, education, employment and ecology. We consider the task of answering counterfactual questions such as, "Would this patient have lower blood sugar had she received a different medication?". We propose a new algorithmic framework for counterfactual inference which brings together ideas from domain adaptation and representation learning. In addition to a theoretical justification, we perform an empirical comparison with previous approaches to causal inference from observational data. Our deep learning algorithm significantly outperforms the previous state-of-the-art.

Appeared in ICML 2016

References in corpus (3)

• Doubly Robust Policy Evaluation and Learning
• The Variational Fair Autoencoder
• Learning from Logged Implicit Exploration Data

Cited by in corpus (39)

• Deep Reinforcement Learning: An Overview
• Causal Effect Inference with Deep Latent-Variable Models
• A Meta-Transfer Objective for Learning to Disentangle Causal Mechanisms
• Learning Counterfactual Representations for Estimating Individual Dose-Response Curves
• Adapting Text Embeddings for Causal Inference
• Adapting Neural Networks for the Estimation of Treatment Effects
• Learning Weighted Representations for Generalization Across Designs
• Adversarial Machine Learning: Bayesian Perspectives
• DeepMatch: Balancing Deep Covariate Representations for Causal Inference Using Adversarial Training
• Representation Learning for Electronic Health Records
• Treatment effect estimation with disentangled latent factors
• Off-Policy Evaluation and Learning for External Validity under a Covariate Shift
• A Deep Causal Inference Approach to Measuring the Effects of Forming Group Loans in Online Non-profit Microfinance Platform
• MissDeepCausal: Causal Inference from Incomplete Data Using Deep Latent Variable Models
• Inferring Individual Direct Causal Effects Under Heterogeneous Peer Influence
• Learning Optimal Policies from Observational Data
• Causality-aware counterfactual confounding adjustment for feature representations learned by deep models
• Unsupervised Domain Adaptation Using Approximate Label Matching
• Causal Effect Variational Autoencoder with Uniform Treatment
• Fairness Through Causal Awareness: Learning Latent-Variable Models for Biased Data
• Loss Functions for Discrete Contextual Pricing with Observational Data
• Improving pairwise comparison models using Empirical Bayes shrinkage
• Counterfactual-based Incrementality Measurement in a Digital Ad-Buying Platform
• A Minimax Surrogate Loss Approach to Conditional Difference Estimation
• Prophit: Causal inverse classification for multiple continuously valued treatment policies
• Counterfactual Prediction Under Selective Confounding
• DeepRite: Deep Recurrent Inverse TreatmEnt Weighting for Adjusting Time-varying Confounding in Modern Longitudinal Observational Data
• An Interpretable Probabilistic Approach for Demystifying Black-box Predictive Models
• Estimating Individual Advertising Effect in E-Commerce
• Doing Great at Estimating CATE? On the Neglected Assumptions in Benchmark Comparisons of Treatment Effect Estimators
• Predicting Counterfactuals from Large Historical Data and Small Randomized Trials
• Uplift Modeling from Separate Labels
• Neuroevolutionary Feature Representations for Causal Inference
• Representation learning with CGAN for casual inference
• Estimating Causal Effects With Partial Covariates For Clinical Interpretability
• Robust Counterfactual Inferences using Feature Learning and their Applications
• On the Fairness of Randomized Trials for Recommendation with Heterogeneous Demographics and Beyond
• The Bias-Variance Tradeoff of Doubly Robust Estimator with Targeted $L_1$ regularized Neural Networks Predictions
• Causal Embeddings for Recommendation: An Extended Abstract