Advertisement
Published in:
2017 | OriginalPaper | Chapter
Supporting the Page-Hinkley Test with Empirical Mode Decomposition for Change Detection
Authors : Raquel Sebastião, José Maria Fernandes
Published in: Foundations of Intelligent Systems
Publisher: Springer International Publishing
Activate our intelligent search to find suitable subject content or patents.
Select sections of text to find matching patents with Artificial Intelligence. powered by
Select sections of text to find additional relevant content using AI-assisted search. powered by
Abstract
In the dynamic scenarios faced nowadays, when handling non stationary data streams it is of utmost importance to perform change detection tests. In this work, we propose the Intrinsic Page Hinkley Test (iPHT), which enhances the Page Hinkley Test (PHT) eliminating the user-defined parameter (the allowed magnitude of change of the data that are not considered real distribution change of the data stream) by using the second order intrinsic mode function (IMF) which is a data dependent value reflecting the intrinsic data variation. In such way, the PHT change detection method is expected to be more robust and require less tunes. Furthermore, we extend the proposed iPHT to a blockwise approach. Computing the IMF over sliding windows, which is shown to be more responsive to changes and suitable for online settings. The iPHT is evaluated using artificial and real data, outperforming the PHT.