New framework enhances federated learning privacy and robustness

Researchers have developed a new framework called DP-BR-FedAvg to enhance the security and privacy of federated learning in sensitive sectors like banking and healthcare. This framework combines differential privacy using a Gaussian mechanism with a Byzantine-robust aggregation rule, specifically a coordinate-wise trimmed-mean. Evaluations demonstrated that DP-BR-FedAvg significantly improves model performance and robustness against adversarial attacks compared to standard FedAvg, while also bounding privacy loss. AI

IMPACT Enhances security and privacy for sensitive data applications like banking and healthcare, potentially enabling wider adoption of federated learning.

RANK_REASON Academic paper detailing a new framework for federated learning. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.LG →

• alphaXiv
• arXiv
• CatalyzeX
• coordinate-wise trimmed-mean
• CORE Recommender
• DagsHub
• DP-BR-FedAvg
• FedAvg
• Gaussian mechanism
• Gotit.pub
• Hugging Face
• Influence Flower
• ScienceCast

• paper
• safety
• infra

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