The Causal Loss: Driving Correlation to Imply Causation

TL;DR

This paper introduces the Causal Loss, a new model-agnostic loss function that enhances the causal interpretability and interventional capabilities of machine learning models by integrating causal regularization.

Contribution

It proposes a novel causal loss function that enables standard models to better reflect causal relationships and perform more reliably under interventions.

Findings

Causal Loss improves models' interventional quality.

Standard models gain causal interpretability with Causal Loss.

Experimental results support theoretical claims.

Abstract

Most algorithms in classical and contemporary machine learning focus on correlation-based dependence between features to drive performance. Although success has been observed in many relevant problems, these algorithms fail when the underlying causality is inconsistent with the assumed relations. We propose a novel model-agnostic loss function called Causal Loss that improves the interventional quality of the prediction using an intervened neural-causal regularizer. In support of our theoretical results, our experimental illustration shows how causal loss bestows a non-causal associative model (like a standard neural net or decision tree) with interventional capabilities.