Learning Structural Causal Models from Ordering: Identifiable Flow Models

TL;DR

This paper introduces flow-based causal models that leverage known causal orderings to recover causal mechanisms, enabling efficient, consistent, and scalable causal inference from observational data.

Contribution

The authors develop a novel flow-based approach that ensures causal consistency, allows simultaneous learning of mechanisms, and reduces computational complexity, outperforming existing methods.

Findings

Outperforms previous state-of-the-art methods in causal inference tasks.

Maintains causal consistency regardless of discretization steps.

Reduces computational time significantly compared to diffusion-based techniques.

Abstract

In this study, we address causal inference when only observational data and a valid causal ordering from the causal graph are available. We introduce a set of flow models that can recover component-wise, invertible transformation of exogenous variables. Our flow-based methods offer flexible model design while maintaining causal consistency regardless of the number of discretization steps. We propose design improvements that enable simultaneous learning of all causal mechanisms and reduce abduction and prediction complexity to linear O(n) relative to the number of layers, independent of the number of causal variables. Empirically, we demonstrate that our method outperforms previous state-of-the-art approaches and delivers consistent performance across a wide range of structural causal models in answering observational, interventional, and counterfactual questions. Additionally, our method achieves a significant reduction in computational time compared to existing diffusion-based techniques, making it practical for large structural causal models.