Quickest Causal Change Point Detection by Adaptive Intervention

TL;DR

This paper introduces an adaptive intervention algorithm for quickest change point detection in linear causal models, effectively amplifying change signals and balancing exploration and exploitation, with proven optimality and real-world validation.

Contribution

It presents a novel algorithm that accounts for interventions in causal models, including a new intervention selection method and two adaptive monitoring strategies with theoretical guarantees.

Findings

Algorithm achieves first-order optimality.

Effective amplification of change signals through intervention.

Validated on simulations and real-world case studies.

Abstract

We propose an algorithm for change point monitoring in linear causal models that accounts for interventions. Through a special centralization technique, we can concentrate the changes arising from causal propagation across nodes into a single dimension. Additionally, by selecting appropriate intervention nodes based on Kullback-Leibler divergence, we can amplify the change magnitude. We also present an algorithm for selecting the intervention values, which aids in the identification of the most effective intervention nodes. Two monitoring methods are proposed, each with an adaptive intervention policy to make a balance between exploration and exploitation. We theoretically demonstrate the first-order optimality of the proposed methods and validate their properties using simulation datasets and two real-world case studies.