Adversarial balancing-based representation learning for causal effect inference with observational data (2019-04-30T00:00:00.000000Z)

TL;DR

A neural network framework called Adversarial Balancing-based representation learning for Causal Effect Inference (ABCEI), based on recent advances in representation learning, is proposed that is robust against treatment selection bias, and matches/outperforms the state-of-the-art approaches.

Abstract

Learning causal effects from observational data greatly benefits a variety of domains such as health care, education, and sociology. For instance, one could estimate the impact of a new drug on specific individuals to assist clinical planning and improve the survival rate. In this paper, we focus on studying the problem of estimating the Conditional Average Treatment Effect (CATE) from observational data. The challenges for this problem are two-fold: on the one hand, we have to derive a causal estimator to estimate the causal quantity from observational data, in the presence of confounding bias; on the other hand, we have to deal with the identification of the CATE when the distributions of covariates over the treatment group units and the control units are imbalanced. To overcome these challenges, we propose a neural network framework called Adversarial Balancing-based representation learning for Causal Effect Inference (ABCEI), based on recent advances in representation learning. To ensure the identification of the CATE, ABCEI uses adversarial learning to balance the distributions of covariates in the treatment and the control group in the latent representation space, without any assumptions on the form of the treatment selection/assignment function. In addition, during the representation learning and balancing process, highly predictive information from the original covariate space might be lost. ABCEI can tackle this information loss problem by preserving useful information for predicting causal effects under the regularization of a mutual information estimator. The experimental results show that ABCEI is robust against treatment selection bias, and matches/outperforms the state-of-the-art approaches. Our experiments show promising results on several datasets, encompassing several health care (and other) domains.

References79 items

Identifiability of Cause and Effect using Regularized Regression

Identification of Causal Effects in the Presence of Selection Bias

Causal analysis of the impact of homecare services on patient discharge disposition

Causality

Identifying Exceptional Responders in Randomized Trials: An Optimization Approach

Robust Estimation of Causal Effects via High-Dimensional Covariate Balancing Propensity Score.

Adversarial Balancing for Causal Inference

Learning deep representations by mutual information estimation and maximization

Stable Prediction across Unknown Environments

Estimating the causal effects of chronic disease combinations on 30-day hospital readmissions based on observational Medicaid data

DeepMatch: Balancing Deep Covariate Representations for Causal Inference Using Adversarial Training

GANITE: Estimation of Individualized Treatment Effects using Generative Adversarial Nets

Mutual Information Neural Estimation

Automated versus Do-It-Yourself Methods for Causal Inference: Lessons Learned from a Data Analysis Competition

Estimating Individual Treatment Effect from Educational Studies with Residual Counterfactual Networks

Causal Effect Inference with Deep Latent-Variable Models

Balanced Policy Evaluation and Learning

Improved Training of Wasserstein GANs

Estimating Individual Treatment Effect in Observational Data Using Random Forest Methods

NIPS 2016 Tutorial: Generative Adversarial Networks

Nonparametric survival analysis using Bayesian Additive Regression Trees (BART)

Estimating individual treatment effect: generalization bounds and algorithms

f-GAN: Training Generative Neural Samplers using Variational Divergence Minimization

Learning Representations for Counterfactual Inference

MIMIC-III, a freely accessible critical care database

Fast and Accurate Deep Network Learning by Exponential Linear Units (ELUs)

Doubly Robust Off-policy Value Evaluation for Reinforcement Learning

Estimation and Inference of Heterogeneous Treatment Effects using Random Forests

Heterogeneous Treatment Effects in Digital Experimentation

Adam: A Method for Stochastic Optimization

Distinguishing Cause from Effect Using Observational Data: Methods and Benchmarks

Estimating Conditional Average Treatment Effects

Covariate balancing propensity score

Auto-Encoding Variational Bayes

An optimization approach for making causal inferences

Causal inference in public health.

Balance Optimization Subset Selection (BOSS): An Alternative Approach for Causal Inference with Observational Data

Equitability, mutual information, and the maximal information coefficient

Detecting Latent Heterogeneity

MatchIt: Nonparametric Preprocessing for Parametric Causal Inference

Controlling Selection Bias in Causal Inference

Bayesian Nonparametric Modeling for Causal Inference

Inference on Counterfactual Distributions

Bounds on Direct Effects in the Presence of Confounded Intermediate Variables

Extracting and composing robust features with denoising autoencoders

Covariate Shift Adaptation by Importance Weighted Cross Validation

Vitamin D supplementation and total mortality: a meta-analysis of randomized controlled trials.

Genetic Matching for Estimating Causal Effects: A General Multivariate Matching Method for Achieving Balance in Observational Studies

Domain Adaptation for Statistical Classifiers

Nonparametric Tests for Treatment Effect Heterogeneity

Stratification and weighting via the propensity score in estimation of causal treatment effects: a comparative study

The Costs of Low Birth Weight

A general identification condition for causal effects

A tutorial on Principal Components Analysis

Propensity Score-Matching Methods for Nonexperimental Causal Studies

Using Propensity Scores to Help Design Observational Studies: Application to the Tobacco Litigation

Does Matching Overcome Lalonde's Critique of Nonexperimental Estimators?

A comparison of observational studies and randomized, controlled trials.

Evaluating the Econometric Evaluations of Training Programs with Experimental Data

The central role of the propensity score in observational studies for causal effects

Asymptotic evaluation of certain Markov process expectations for large time

MIMIC-III clinical database demo (version 1.4). PhysioNet

Representation Learning for Treatment Effect Estimation from Observational Data

GENERATIVE ADVERSARIAL NETS

Identification of predicted individual treatment effects in randomized clinical trials

Matching on Balanced Nonlinear Representations for Treatment Effects Estimation

The American economic review pp

Mutual Information Based Matching for Causal Inference with Observational Data

Batch learning from logged bandit feedback through counterfactual risk minimization

Counterfactual risk minimization: learning from logged bandit feedback

Visualizing Data using t-SNE

A Mathematical Theory of Communication

Matching estimators of causal eﬀects: Prospects and pitfalls in theory and practice

Causal inference using potential outcomes: design, modeling, decisions

Random Forests

This Paper Is Included in the Proceedings of the 12th Usenix Symposium on Operating Systems Design and Implementation (osdi '16). Tensorflow: a System for Large-scale Machine Learning Tensorflow: a System for Large-scale Machine Learning

Journal of the American Statistical Association Using Mixed Integer Programming for Matching in an Observational Study of Kidney Failure after Surgery Using Mixed Integer Programming for Matching in an Observational Study of Kidney Failure after Surgery

Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations