Using Graph Convolutional Networks to Address fMRI Small Data Problems Conference: Proceedings of the 28th ACM SIGKDD Conference on Knowledge Discovery and Data Mining; August 14–18, 2022; Washington, DC, USA Proceedings of the 28th ACM SIGKDD Conference on Knowledge Discovery and Data Mining (KDD ’

Thomas Screven, András Necz, Jason Smucny, Ian Davidson · 2025

Read Using Graph Convolutional Networks to Address fMRI Small Data Problems Conference: Proceedings of the 28th ACM SIGKDD Conference on Knowledge Discovery and Data Mining; August 14–18, 2022; Washington, DC, USA Proceedings of the 28th ACM SIGKDD Conference on Knowledge Discovery and Data Mining (KDD ’ by Thomas Screven, András Necz, Jason Smucny, Ian Davidson (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2502-17489.txt.

Description

Thomas Screven and three co-authors use graph convolutional networks to learn from small brain imaging datasets, in work from 2025. Predicting treatment response from functional MRI suffers a data famine, since each patient's connectome is itself a graph of linked regions. Their method works in a spectral form of the connectivity graph, which lets information propagate efficiently across it.

The spectral network beat standard deep learning by about twelve percent on the same small data. The gain is traced to a smoothing effect: the method closes triangle inequality gaps in the data, enforcing a transitivity that raw learning misses. The approach targets prognosis tasks where large cohorts will never exist.

Catalog record

CREATORThomas Screven, András Necz, Jason Smucny, Ian Davidson
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE35 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2502-17489

Browse the full library · About the Tower of Babel