Root Cause Analysis of Outliers in Unknown Cyclic Graphs

2025-10-08

AI Summary

This paper addresses root cause localization of outliers in cyclic causal graphs, focusing on outlier propagation mechanisms under linear structural equation models (SEMs). We propose a structure-agnostic root cause identification method that reconstructs latent anomaly sources directly from observational data, by modeling perturbation propagation paths within system dynamics and integrating causal inference with graph-theoretic analysis. Our key contribution is extending root cause localization to cyclic causal graphs with unknown topology; under a strong perturbation assumption, we prove that the algorithm outputs a short candidate set containing—w.h.p.—both the true root cause and its immediate parents along the causal cycle. Experiments demonstrate the framework’s feasibility and accuracy on systems with unknown cyclic structures, significantly enhancing interpretability and practicality of anomaly diagnosis in complex feedback systems.

Technology Category

• Reasoning under Uncertainty: Causality
• Machine Learning: Causal Learning
• Data Mining & Knowledge Management: Anomaly/Outlier Detection

Application Category

• Graph Algorithms and Modeling for the Web: Algorithms and analysis for incomplete, noisy, or partially observed Web-related graphs
• Responsible Web: Human-perceived consequences of algorithmic deployment on the web
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

We study the propagation of outliers in cyclic causal graphs with linear structural equations, tracing them back to one or several "root cause" nodes. We show that it is possible to identify a short list of potential root causes provided that the perturbation is sufficiently strong and propagates according to the same structural equations as in the normal mode. This shortlist consists of the true root causes together with those of its parents lying on a cycle with the root cause. Notably, our method does not require prior knowledge of the causal graph.

Problem

• Identifying root causes of outliers in unknown cyclic graphs
• Analyzing outlier propagation through linear structural equations
• Detecting root causes without prior causal graph knowledge

Innovation

• Identifies root causes in unknown cyclic graphs
• Uses linear structural equations for outlier propagation
• Requires no prior knowledge of causal graph