Data Amount Heterogeneity Improves Performance in Federated Learning for Neuroimaging
Abstract
Federated Learning (FL) allows for secure analysis of distributed data without data sharing. We and others have successfully applied FL to neuroimaging tasks, such as Alzheimer's disease classification from structural magnetic resonance imaging (MRI). FL trains a deep learning model locally at each site, encrypts the local model parameters and sends them to a federation controller that aggregates the local models under homomorphic encryption, the resulting community model is sent back to the sites for further training, and the process repeats for a number of rounds. FL often achieves learning performance (e.g., accuracy) close to centralized training, but no data leaves the sites and model parameters are protected through encryption and other measures. In this paper, we show that federations composed of sites with heterogeneous amounts of data (1 large site plus small sites) perform better than homogeneous (same data per site) federations, particularly in low data regimes, while reaching performance close to centralized training. Having a big site often results in faster convergence and higher final accuracy. Heterogeneous federations better exploit sites with small amounts of data, a common situation in large consortia such as ENIGMA.
Related
- Bundling Alleles Within Haplotype Blocks Improves QTL Detection Under Allelic Heterogeneity
- Interpretable Machine Learning Links Group-Specific Plasma Proteomics to Regional Structural Neuroimaging in Alzheimer's Disease
- SPD Matrix Learning for Neuroimaging Analysis: Perspectives, Methods, and Challenges
- Shedding Light on COVIS: Evidence from Optical Neuroimaging for Dissociable Category Learning Systems
- Capturing Spatial Heterogeneity Using UAV-Derived Multispectral Phenotypes and Random Regression Models Improves Genetic Evaluation in Alfalfa Breeding Trials
BibTeX
@misc{kurian2026data,
title = {{Data Amount Heterogeneity Improves Performance in Federated Learning for Neuroimaging}},
author = {Paul Kurian and Tamoghna Chattopadhyay and Abhijith Shaji and Greg Ver Steeg and Paul M. Thompson and Jose Luis Ambite},
howpublished = {arXiv},
year = {2026},
month = oct,
doi = {10.48550/arxiv.2610.38172},
eprint = {2610.38172},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2610.38172},
note = {CC BY 4.0},
}
RIS
TY - UNPB
TI - Data Amount Heterogeneity Improves Performance in Federated Learning for Neuroimaging
AU - Paul Kurian
AU - Tamoghna Chattopadhyay
AU - Abhijith Shaji
AU - Greg Ver Steeg
AU - Paul M. Thompson
AU - Jose Luis Ambite
PY - 2026
DA - 2026/10/28
PB - arXiv
DO - 10.48550/arxiv.2610.38172
UR - https://arxiv.org/abs/2610.38172
AB - Federated Learning (FL) allows for secure analysis of distributed data without data sharing. We and others have successfully applied FL to neuroimaging tasks, such as Alzheimer's disease classification from structural magnetic resonance imaging (MRI). FL trains a deep learning model locally at each site, encrypts the local model parameters and sends them to a federation controller that aggregates the local models under homomorphic encryption, the resulting community model is sent back to the sites for further training, and the process repeats for a number of rounds. FL often achieves learning performance (e.g., accuracy) close to centralized training, but no data leaves the sites and model parameters are protected through encryption and other measures. In this paper, we show that federations composed of sites with heterogeneous amounts of data (1 large site plus small sites) perform better than homogeneous (same data per site) federations, particularly in low data regimes, while reaching performance close to centralized training. Having a big site often results in faster convergence and higher final accuracy. Heterogeneous federations better exploit sites with small amounts of data, a common situation in large consortia such as ENIGMA.
SN - 2331-8422
ER -