ReTimeCausal: EM-Augmented Additive Noise Models for Interpretable Causal Discovery in Irregular Time Series

2025-07-04

AI Summary

Causal discovery in high-stakes domains—such as finance, healthcare, and climate science—is challenged by irregularly sampled time series with high missingness rates, where conventional methods assume regular sampling or rely on opaque aggregation. Method: This paper proposes a physics-guided and statistically grounded interpretable causal discovery framework. It integrates the additive noise model with the expectation-maximization (EM) algorithm, jointly optimizing missing-value imputation, latent-variable estimation, and sparse causal graph learning within EM iterations via kernelized sparse regression and structural constraints. The approach explicitly models multi-scale dynamic dependencies (e.g., hourly events and decadal trends). Contribution/Results: Unlike existing approaches, it imposes no regular-sampling assumption, avoids black-box aggregation, and ensures auditability of causal mechanisms. Experiments on synthetic and real-world datasets demonstrate substantial improvements in causal identification accuracy and robustness under high missingness and strong sampling irregularity.

Technology Category

• Machine Learning: Causal Learning
• Reasoning under Uncertainty: Causality
• Knowledge Representation and Reasoning: Diagnosis and Abductive Reasoning

Application Category

• Graph Algorithms and Modeling for the Web: Algorithms and analysis for incomplete, noisy, or partially observed Web-related graphs
• Web Mining and Content Analysis: Interdisciplinary science discovery with web data mining
• User Modeling, Personalization and Recommendation: Explainable and interpretable methods for personalization

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. Traditional methods (e.g., Granger causality, PCMCI) fail to reconcile multi-scale interactions (e.g., hourly storms vs. decadal climate shifts), while neural approaches (e.g., CUTS+) lack interpretability, stemming from a critical gap: existing frameworks either rigidly assume temporal regularity or aggregate dynamics into opaque representations, neglecting real-world granularity and auditable logic. To bridge this gap, we propose ReTimeCausal, a novel integration of Additive Noise Models (ANM) and Expectation-Maximization (EM) that unifies physics-guided data imputation with sparse causal inference. 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.

Problem

• Causal discovery in irregularly sampled time series
• Handling missing data and inconsistent sampling frequencies
• Resolving cross-frequency dependencies in causal inference

Innovation

• EM-Augmented Additive Noise Models for causal discovery
• Kernelized sparse regression for cross-frequency dependencies
• Physics-guided data imputation with sparse causal inference