Poster (After Call)

FZJ-2024-07621

http://join2-wiki.gsi.de/foswiki/pub/Main/Artwork/join2_logo100x88.png FASTGPR: DIVIDE-AND-CONQUER TECHNIQUE IN NEUROIMAGING DATA SHORTENS TRAINING TIME AND IMPROVES ACCURACY

Raimondo, F. (Corresponding author)FZJ* ; Antonopoulos, G. (Corresponding author)FZJ* ; Eickhoff, S.FZJ* ; Patil, K.FZJ*

2024

This record in other databases:

Please use a persistent id in citations: doi:10.34734/FZJ-2024-07621

Abstract: Gaussian process regression (GPR) has shown great potential for studying healthy aging and disease via brain-age prediction (BAP) using structural MRI[1].A big drawback of GPR is the training complexity which is an O(N^3) operation (N=number of data points).The need for expansive datasets and the high dimensionality of MRI data, renders the training of GPR impractical with conventional computing resources.We investigated whether a divide-and-conquer approach can be used together with the GPR model

---

Contributing Institute(s):

1. Gehirn & Verhalten (INM-7)

Research Program(s):

1. 5253 - Neuroimaging (POF4-525) (POF4-525)

Appears in the scientific report 2024

---

Database coverage:

---