Factorized binary search: Change point detection in the network structure of multivariate high-dimensional time series

Time point High dimensional
DOI: 10.1162/imag_a_00520 Publication Date: 2025-03-13T21:46:31Z
ABSTRACT
Abstract Functional magnetic resonance imaging (fMRI) time series data presents a unique opportunity to understand the behavior of temporal brain connectivity, and models that uncover complex dynamic workings this organ are keen interest in neuroscience. We motivated develop accurate change point detection network estimation techniques for high-dimensional whole-brain fMRI data. To end, we introduce factorized binary search (FaBiSearch), novel method structure multivariate order large-scale characterizations dynamics brain. FaBiSearch employs non-negative matrix factorization, an unsupervised dimension reduction technique, new algorithm identify multiple points. In addition, propose between seek mechanism brain, particularly two sets. The first is resting-state experiment, where subjects scanned over three visits. second task-based read Chapter 9 Harry Potter Sorcerer’s Stone. For set, examine test-retest functional while explore during reading whether points across coincide with key plot twists story. Further, hub nodes their behavior. Finally, make all methods discussed available R package fabisearch on CRAN, as well experiments GitHub
SUPPLEMENTAL MATERIAL
Coming soon ....
REFERENCES (100)
CITATIONS (1)