Stable learning establishes some common ground between causal inference and machine learning

Causal inference has recently attracted substantial attention in the machine learning and artificial intelligence community. It is usually positioned as a distinct strand of research that can broaden the scope of machine learning from predictive modelling to intervention and decision-making. In this Perspective, however, we argue that ideas from causality can also be used to improve the stronghold of machine learning, predictive modelling, if predictive stability, explainability and fairness are important. With the aim of bridging the gap between the tradition of precise modelling in causal inference and black-box approaches from machine learning, stable learning is proposed and developed as a source of common ground. This Perspective clarifies a source of risk for machine learning models and discusses the benefits of bringing causality into learning. We identify the fundamental problems addressed by stable learning, as well as the latest progress from both causal inference and learning perspectives, and we discuss relationships with explainability and fairness problems. Machine learning performs well at predictive modelling based on statistical correlations, but for high-stakes applications, more robust, explainable and fair approaches are required. Cui and Athey discuss the benefits of bringing causal inference into machine learning, presenting a stable learning approach.

Automated summary

Causal inference plays a significant role in improving the stability, explainability, and fairness of predictive modeling in machine learning. Stable learning, as a common ground between causal inference and machine learning, bridges the gap by addressing the source of risk in machine learning models. It offers a more robust approach for high-stakes applications.