Identifying the temporal distribution structure in multivariate data for time-series segmentation based on two-sample test

Justyna Witulska

doi:10.1016/j.inffus.2025.103445

Identifying the temporal distribution structure in multivariate data for time-series segmentation based on two-sample test

2025-07-11

0

OA

AI

DOI：10.1016/j.inffus.2025.103445

Original

Original request for help

Save

Abstract

En 中文

• Multisensor monitoring of complex systems by identifying state changes is designed. • MIDAST methodology is proposed for fusion-based multivariate data segmentation. • Two distinct multivariate data models are assessed during computer simulations. • MIDAST outperforms two baseline methods, i.e. e-Divisive and KCPA. • MIDAST enables non-invasive reconstruction of intracranial hypertension events.

Keywords:

Multivariate data segmentation

Two-sample test

Non-Gaussian distributions

Gaussian distribution

Multisensor fusion

Intracranial hypertension detection

Non-invasive measurement

Identifying the temporal distribution structure in multivariate data for time-series segmentation based on two-sample test

Abstract

Journal

Organization