BrainRiem framework uses Riemannian geometry for privacy-preserving brain network diagnosis

Researchers have developed BrainRiem, a novel framework for adapting brain network diagnostic models across different sites without requiring direct access to source data. This method utilizes Riemannian geometry to learn compact brain prototypes that remain valid Symmetric Positive Definite (SPD) matrices, thus avoiding geometric distortions common in Euclidean approaches. By transmitting only these anonymized prototypes, BrainRiem enables training of local models at target sites while adhering to clinical privacy regulations and outperforming existing source-free and graph domain adaptation techniques. AI

RANK_REASON The cluster contains a research paper detailing a new methodology for domain adaptation in brain network analysis. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.LG →

• BrainRiem
• Riemannian geometry

• paper
• other

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →