Conditional Independence Testing via Latent Representation Learning

TL;DR

This paper introduces LCIT, a novel non-parametric method leveraging representation learning to effectively test for conditional independence, especially in complex, high-dimensional data, improving causal discovery tasks.

Contribution

The paper proposes a generative framework that learns latent representations to test for conditional independence, outperforming existing methods in various settings.

Findings

LCIT outperforms state-of-the-art baselines.

Effective in non-linear and high-dimensional data.

Works well on synthetic and real datasets.

Abstract

Detecting conditional independencies plays a key role in several statistical and machine learning tasks, especially in causal discovery algorithms. In this study, we introduce LCIT (Latent representation based Conditional Independence Test)-a novel non-parametric method for conditional independence testing based on representation learning. Our main contribution involves proposing a generative framework in which to test for the independence between X and Y given Z, we first learn to infer the latent representations of target variables X and Y that contain no information about the conditioning variable Z. The latent variables are then investigated for any significant remaining dependencies, which can be performed using the conventional partial correlation test. The empirical evaluations show that LCIT outperforms several state-of-the-art baselines consistently under different evaluation metrics, and is able to adapt really well to both non-linear and high-dimensional settings on a diverse collection of synthetic and real data sets.