Online method detects changes faster and explains them clearly
High-Dimensional Online Change Point Detection with Adaptive Thresholding and Interpretability
Machine Learning
Summary
Change point detection is about spotting when something important suddenly changes in a stream of data, like a spike in heart rate or a market crash. Traditional methods struggle when the data has many parts and often can’t explain why they detected a change. The authors present a new approach that simplifies complex data into easy scores, figures out smart thresholds to decide when a change happens, and also explains what caused the change. Their method spots true changes more accurately and gives clear reasons, making it useful for tough real-world settings.
What this means in practice
- •For cybersecurity teams: Detect sudden and meaningful changes in high-dimensional network data streams quickly and reduce false alarms with clear explanations.
- •For financial data analysts: Identify abrupt shifts in complex market data in real time, improving detection reliability and providing insights into the nature of market changes.
Authors
Sven Jacob, Bardh Prenkaj, Weijia Shao, Gjergji Kasneci
Abstract
Change point detection (CPD) identifies abrupt and significant changes in sequential data, with applications in human activity recognition, financial markets, cybersecurity, manufacturing, and autonomous systems. Traditional CPD methods often face computational challenges in high-dimensional settings and typically provide limited explanations for detected changes, which can restrict their practical usability. This paper introduces a CPD framework that improves scalability and interpretability by leveraging the Sliced Wasserstein (SW) distance. Our contributions are fourfold: (1) we transform multivariate sequential data into one-dimensional scores using the SW distance, making the resulting representation compatible with existing CPD methods; (2) we analyze the distributional behavior of random slices of the SW distance and show that, under suitable assumptions, they can be approximated by a Gamma distribution, providing a principled basis for threshold calibration; (3) we propose a self-adapting online CPD algorithm that combines this SW-based score with an adaptive quantile-based threshold; (4) we introduce a model-specific framework for generating contrastive explanations for annotated change points. Empirically, our method reduces false positives by at least $48\%$ on average compared with popular online and offline CPD baselines, while maintaining competitive or superior detection performance. Code is available at https://github.com/jsve96/SWCPD_Code. At the same time, it produces interpretable change-point annotations, making it practical for deployment in high-stakes applications.