Causal Effect Estimation using identifiable Variational AutoEncoder with Latent Confounders and Post-Treatment Variables

TL;DR

This paper introduces CPTiVAE, a novel joint VAE approach that accurately estimates causal effects from observational data by modeling latent confounders and post-treatment variables, addressing biases ignored by prior methods.

Contribution

It proposes a new identifiable VAE framework that captures latent confounders and post-treatment variables, with proven identifiability and superior performance in experiments.

Findings

CPTiVAE outperforms existing methods on synthetic datasets.

It effectively models latent confounders and post-treatment variables.

Demonstrates potential in real-world causal inference applications.

Abstract

Estimating causal effects from observational data is challenging, especially in the presence of latent confounders. Much work has been done on addressing this challenge, but most of the existing research ignores the bias introduced by the post-treatment variables. In this paper, we propose a novel method of joint Variational AutoEncoder (VAE) and identifiable Variational AutoEncoder (iVAE) for learning the representations of latent confounders and latent post-treatment variables from their proxy variables, termed CPTiVAE, to achieve unbiased causal effect estimation from observational data. We further prove the identifiability in terms of the representation of latent post-treatment variables. Extensive experiments on synthetic and semi-synthetic datasets demonstrate that the CPTiVAE outperforms the state-of-the-art methods in the presence of latent confounders and post-treatment variables. We further apply CPTiVAE to a real-world dataset to show its potential application.