Learning domain-specific causal discovery from time series

TL;DR

This paper proposes a data-driven, supervised approach to causal discovery from time series data, outperforming traditional human-designed methods across various datasets and suggesting a shift towards learned, domain-specific causal models.

Contribution

It introduces a novel supervised learning framework for domain-specific causal discovery from time series, demonstrating significant improvements over existing methods.

Findings

Supervised models outperform traditional methods on multiple datasets.

Domain-specific causal discovery can be effectively learned from data.

The approach enhances causal inference in neuroscience and medical data.

Abstract

Causal discovery (CD) from time-varying data is important in neuroscience, medicine, and machine learning. Techniques for CD encompass randomized experiments, which are generally unbiased but expensive, and algorithms such as Granger causality, conditional-independence-based, structural-equation-based, and score-based methods that are only accurate under strong assumptions made by human designers. However, as demonstrated in other areas of machine learning, human expertise is often not entirely accurate and tends to be outperformed in domains with abundant data. In this study, we examine whether we can enhance domain-specific causal discovery for time series using a data-driven approach. Our findings indicate that this procedure significantly outperforms human-designed, domain-agnostic causal discovery methods, such as Mutual Information, VAR-LiNGAM, and Granger Causality on the MOS 6502 microprocessor, the NetSim fMRI dataset, and the Dream3 gene dataset. We argue that, when feasible, the causality field should consider a supervised approach in which domain-specific CD procedures are learned from extensive datasets with known causal relationships, rather than being designed by human specialists. Our findings promise a new approach toward improving CD in neural and medical data and for the broader machine learning community.