Possibility for Proactive Anomaly Detection

2025-04-15

AI Summary

Existing time-series anomaly detection methods rely on model residuals, adopting a retrospective paradigm that hinders real-time alerting. This paper proposes a proactive anomaly detection paradigm: it decouples forecasting from discrimination, establishing a synergistic framework comprising a forecasting model—specifically optimized for anomaly detection—and a data-driven detector (e.g., Isolation Forest or VAE). We further design an interpretable, dynamically adaptive threshold calibration mechanism to distinguish between predictable and unpredictable anomalies and model them heterogeneously. Evaluated on four standard benchmarks, our approach achieves significant improvements in F1-score and recall. It demonstrates high sensitivity to early anomalous precursors, enabling timely intervention. The implementation is fully open-sourced.

Technology Category

• Data Mining & Knowledge Management: Anomaly/Outlier Detection
• Machine Learning: Time-Series/Data Streams
• Reasoning under Uncertainty: Sequential Decision Making

Application Category

• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• User Modeling, Personalization and Recommendation: Explainable and interpretable methods for personalization
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

Time-series anomaly detection, which detects errors and failures in a workflow, is one of the most important topics in real-world applications. The purpose of time-series anomaly detection is to reduce potential damages or losses. However, existing anomaly detection models detect anomalies through the error between the model output and the ground truth (observed) value, which makes them impractical. In this work, we present a extit{proactive} approach for time-series anomaly detection based on a time-series forecasting model specialized for anomaly detection and a data-driven anomaly detection model. Our proactive approach establishes an anomaly threshold from training data with a data-driven anomaly detection model, and anomalies are subsequently detected by identifying predicted values that exceed the anomaly threshold. In addition, we extensively evaluated the model using four anomaly detection benchmarks and analyzed both predictable and unpredictable anomalies. We attached the source code as supplementary material.

Problem

• Detects anomalies proactively in time-series data
• Reduces damages by forecasting anomalies beforehand
• Uses data-driven thresholds for accurate anomaly detection

Innovation

• Proactive approach for anomaly detection
• Data-driven anomaly threshold establishment
• Extensive evaluation with benchmarks