Dealing with Uncertainty in Contextual Anomaly Detection

2025-07-06

AI Summary

Contextual anomaly detection (CAD) suffers from inadequate modeling of uncertainty introduced by contextual variables. Method: This paper proposes the NS framework—the first to jointly model aleatoric and epistemic uncertainty in CAD—using heteroscedastic Gaussian process regression to probabilistically calibrate Z-scores, thereby explicitly distinguishing stochastic and model-induced uncertainty sources; it further introduces an uncertainty-aware confidence interval to enable adaptive, interpretable anomaly decisions. Contribution/Results: NS significantly enhances reliability and practicality in high-stakes applications (e.g., cardiac diagnosis). Evaluated on multiple benchmark and real-world medical datasets, NS consistently outperforms state-of-the-art methods, achieving simultaneous improvements in detection accuracy and uncertainty calibration. These results empirically validate the critical role of rigorous uncertainty quantification in advancing both performance and trustworthiness of CAD systems.

Technology Category

• Machine Learning: Calibration & Uncertainty Quantification
• Reasoning under Uncertainty: Uncertainty Representations
• Data Mining & Knowledge Management: Anomaly/Outlier Detection

Application Category

• Graph Algorithms and Modeling for the Web: Algorithms and analysis for heterogeneous, signed, attributed, multi-relational, temporal, higher-order, and annotated Web-related graphs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Web Mining and Content Analysis: Robustness and generalizability of Web mining methods

Abstract

Contextual anomaly detection (CAD) aims to identify anomalies in a target (behavioral) variable conditioned on a set of contextual variables that influence the normalcy of the target variable but are not themselves indicators of anomaly. In many anomaly detection tasks, there exist contextual variables that influence the normalcy of the target variable but are not themselves indicators of anomaly. In this work, we propose a novel framework for CAD, normalcy score (NS), that explicitly models both the aleatoric and epistemic uncertainties. Built on heteroscedastic Gaussian process regression, our method regards the Z-score as a random variable, providing confidence intervals that reflect the reliability of the anomaly assessment. Through experiments on benchmark datasets and a real-world application in cardiology, we demonstrate that NS outperforms state-of-the-art CAD methods in both detection accuracy and interpretability. Moreover, confidence intervals enable an adaptive, uncertainty-driven decision-making process, which may be very important in domains such as healthcare.

Problem

• Modeling uncertainty in contextual anomaly detection tasks
• Improving detection accuracy and interpretability in CAD
• Enabling adaptive decision-making with confidence intervals

Innovation

• Uses heteroscedastic Gaussian process regression
• Models aleatoric and epistemic uncertainties explicitly
• Provides confidence intervals for anomaly assessment