RETIMECAUSAL: A CONSISTENT EM FRAMEWORK FOR CAUSAL DISCOVERY IN IRREGULAR TIME SERIES

Weihong Li, Anpeng Wu, Kun Kuang, Keting Yin

Keywords: Causal Discovery, Expectation-Maximization (EM), Additive Noise Model (ANM), Irregular Sampling, Time Series

TL;DR: ReTimeCausal is a robust method for causal discovery in multivariate time series with missing and irregular data, employing an EM-style framework grounded in Additive Noise Models to ensure accurate structure recovery across varying conditions.

Abstract: This paper studies causal discovery in irregularly sampled time series—a pivotal challenge in high-stakes domains like finance, healthcare, and climate science, where missing data and inconsistent sampling frequencies distort causal mechanisms. The core challenge arises from the interdependence between missing data imputation and causal structure recovery: an error in either component can cascade into the other, ultimately distorting the inferred causal graph. Existing methods either impute first and then discover, or jointly optimize both via neural representation learning, but lack explicit mechanisms to ensure mutual consistency of imputation and structure learning. We address this challenge with ReTimeCausal, an EM-based framework that alternates between imputation and structure learning, promoting structural consistency throughout the optimization process. Our framework emphasizes theoretical consistency guarantees for structure recovery, extending classical results to settings with irregular sampling and high missingness. Through kernelized sparse regression and structural constraints, ReTimeCausal iteratively refines missing values (E-step) and causal graphs (M-step), resolving cross-frequency dependencies and missing data issues. Extensive experiments on synthetic and real-world datasets demonstrate that ReTimeCausal outperforms existing state-of-the-art methods under challenging irregular sampling and missing data conditions.

Primary Area: causal reasoning

Submission Number: 4401