Finding Time Series Anomalies using Granular-ball Vector Data Description

2025-11-15

AI Summary

Addressing the challenges of modeling normal behavior and achieving robust anomaly detection in dynamic nonlinear time series, this paper proposes the Granular-ball One-Class Network (GBOC). Its core innovation is Granular-ball Vector Data Description (GVDD): a density-guided hierarchical partitioning scheme that automatically constructs high-density, compact granular-ball prototypes—naturally preserving local topological structure while drastically reducing prototype cardinality. GVDD further integrates noise filtering, latent-space alignment optimization, and a distance metric based on nearest granular-ball centers to enable efficient anomaly scoring. Evaluated on multiple standard time-series anomaly detection benchmarks, GBOC significantly outperforms conventional nearest-neighbor and clustering-based methods in detection accuracy, while requiring fewer parameters and lower computational overhead—demonstrating both superior robustness and efficiency.

Technology Category

• Data Mining & Knowledge Management: Anomaly/Outlier Detection
• Machine Learning: Time-Series/Data Streams
• Computer Vision: Other Foundations of Computer Vision

Application Category

• Graph Algorithms and Modeling for the Web: Representation, reconstruction, and subgraph or motif discovery in Web-related graphs
• Web Mining and Content Analysis: Robustness and generalizability of Web mining methods
• Economics, Online Markets and Human Computation: Data quality aspects of human-annotated datasets

Abstract

Modeling normal behavior in dynamic, nonlinear time series data is challenging for effective anomaly detection. Traditional methods, such as nearest neighbor and clustering approaches, often depend on rigid assumptions, such as a predefined number of reliable neighbors or clusters, which frequently break down in complex temporal scenarios. To address these limitations, we introduce the Granular-ball One-Class Network (GBOC), a novel approach based on a data-adaptive representation called Granular-ball Vector Data Description (GVDD). GVDD partitions the latent space into compact, high-density regions represented by granular-balls, which are generated through a density-guided hierarchical splitting process and refined by removing noisy structures. Each granular-ball serves as a prototype for local normal behavior, naturally positioning itself between individual instances and clusters while preserving the local topological structure of the sample set. During training, GBOC improves the compactness of representations by aligning samples with their nearest granular-ball centers. During inference, anomaly scores are computed based on the distance to the nearest granular-ball. By focusing on dense, high-quality regions and significantly reducing the number of prototypes, GBOC delivers both robustness and efficiency in anomaly detection. Extensive experiments validate the effectiveness and superiority of the proposed method, highlighting its ability to handle the challenges of time series anomaly detection.

Problem

• Modeling normal behavior in dynamic nonlinear time series data
• Overcoming rigid assumptions in traditional anomaly detection methods
• Detecting anomalies using compact high-density region prototypes

Innovation

• GVDD partitions latent space into granular-ball regions
• GBOC aligns samples with nearest granular-ball centers
• Anomaly scores computed from distance to nearest granular-ball