Conditional anomaly detection with soft harmonic functions

2026-04-23

AI Summary

This study addresses the challenge of identifying reliable anomaly labels in conditional anomaly detection, where annotated anomalies may include distributional boundaries or isolated outliers rather than true anomalies. To tackle this issue, the authors propose a novel non-parametric approach based on soft harmonic functions that explicitly models label confidence and incorporates a regularization mechanism to distinguish genuine anomalies from ambiguous cases. This work is the first to apply soft harmonic functions to conditional anomaly detection, synergistically combining non-parametric estimation with regularized optimization to significantly enhance anomaly label identification performance. Extensive experiments demonstrate that the proposed method consistently outperforms existing baselines across multiple synthetic datasets, UCI benchmarks, and real-world electronic health records, successfully uncovering anomalous patient management decisions.

Technology Category

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

Application Category

• Economics, Online Markets and Human Computation: Data quality aspects of human-annotated datasets
• 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

In this paper, we consider the problem of conditional anomaly detection that aims to identify data instances with an unusual response or a class label. We develop a new non-parametric approach for conditional anomaly detection based on the soft harmonic solution, with which we estimate the confidence of the label to detect anomalous mislabeling. We further regularize the solution to avoid the detection of isolated examples and examples on the boundary of the distribution support. We demonstrate the efficacy of the proposed method on several synthetic and UCI ML datasets in detecting unusual labels when compared to several baseline approaches. We also evaluate the performance of our method on a real-world electronic health record dataset where we seek to identify unusual patient-management decisions.

Problem

• conditional anomaly detection
• anomalous mislabeling
• unusual labels
• electronic health record

Innovation

• conditional anomaly detection
• soft harmonic functions
• non-parametric approach
• label confidence
• regularization