Learning Conditional Instrumental Variable Representation for Causal Effect Estimation. (arXiv:2306.12453v1 [cs.LG])

Learning Conditional Instrumental Variable Representation for Causal Effect Estimation. (arXiv:2306.12453v1 [cs.LG])
By: <a href="http://arxiv.org/find/cs/1/au:+Cheng_D/0/1/0/all/0/1">Debo Cheng</a>, <a href="http://arxiv.org/find/cs/1/au:+Xu_Z/0/1/0/all/0/1">Ziqi Xu</a>, <a href="http://arxiv.org/find/cs/1/au:+Li_J/0/1/0/all/0/1">Jiuyong Li</a>, <a href="http://arxiv.org/find/cs/1/au:+Liu_L/0/1/0/all/0/1">Lin Liu</a>, <a href="http://arxiv.org/find/cs/1/au:+Le_T/0/1/0/all/0/1">Thuc Duy Le</a>, <a href="http://arxiv.org/find/cs/1/au:+Liu_J/0/1/0/all/0/1">Jixue Liu</a> Posted: June 23, 2023

One of the fundamental challenges in causal inference is to estimate the
causal effect of a treatment on its outcome of interest from observational
data. However, causal effect estimation often suffers from the impacts of
confounding bias caused by unmeasured confounders that affect both the
treatment and the outcome. The instrumental variable (IV) approach is a
powerful way to eliminate the confounding bias from latent confounders.
However, the existing IV-based estimators require a nominated IV, and for a
conditional IV (CIV) the corresponding conditioning set too, for causal effect
estimation. This limits the application of IV-based estimators. In this paper,
by leveraging the advantage of disentangled representation learning, we propose
a novel method, named DVAE.CIV, for learning and disentangling the
representations of CIV and the representations of its conditioning set for
causal effect estimations from data with latent confounders. Extensive
experimental results on both synthetic and real-world datasets demonstrate the
superiority of the proposed DVAE.CIV method against the existing causal effect

Provided by:



Moderator and Editor