Meta Learning for Causal Direction

TL;DR

This paper introduces a meta learning approach with a novel generative model to identify causal directions in small observational datasets, addressing the challenge of limited data in causal inference.

Contribution

It presents a new meta learning-based generative model and an end-to-end algorithm for causal direction detection using limited data and similar training datasets.

Findings

High accuracy in causal direction detection across dataset sizes

Effective on both synthetic and real-world data

Maintains performance with limited observational data

Abstract

The inaccessibility of controlled randomized trials due to inherent constraints in many fields of science has been a fundamental issue in causal inference. In this paper, we focus on distinguishing the cause from effect in the bivariate setting under limited observational data. Based on recent developments in meta learning as well as in causal inference, we introduce a novel generative model that allows distinguishing cause and effect in the small data setting. Using a learnt task variable that contains distributional information of each dataset, we propose an end-to-end algorithm that makes use of similar training datasets at test time. We demonstrate our method on various synthetic as well as real-world data and show that it is able to maintain high accuracy in detecting directions across varying dataset sizes.