Federated Learning Enhances Transcriptomics Privacy with FPLIER

Researchers have developed FPLIER, a federated learning approach that extends the Pathway Level Information Extractor (PLIER) for transcriptomics. This method allows for distributed training of gene-set-aware factorization models across multiple data holders without centralizing sensitive expression data. FPLIER uses secure aggregation to produce training updates equivalent to a centralized approach, while also analyzing privacy risks related to membership inference attacks. The study demonstrates stable convergence and shows that privacy is enhanced by incorporating public data or reducing data dimensionality. AI

RANK_REASON The cluster contains a research paper detailing a new method for transcriptomics analysis. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.LG →

• FPLIER
• K-CLIER
• MultiPLIER

• paper
• safety
• other

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