Authors: Jingyun Chen, Yading Yuan

Published on: January 11, 2024

Impact Score: 8.15

Arxiv code: Arxiv:2401.0618

Summary

• What is new: Gossip Mutual Learning (GML), a decentralized FL framework using Gossip Protocol for improved local model training.
• Why this is important: Central server vulnerability in traditional FL and poor local performance due to global data property variations.
• What the research proposes: GML allows direct peer-to-peer communication and mutual learning for site-specific model optimization.
• Results: Increased tumor segmentation performance on HECKTOR21 dataset, better generalization, and significantly reduced communication overhead.

Technical Details

Technological frameworks used: Gossip Mutual Learning (GML), Federated Learning (FL), Gossip Protocol

Models used: Tumor segmentation models

Data used: HECKTOR21 dataset from multiple clinical sites

Potential Impact

Healthcare industry, Medical imaging centers, AI-based diagnostics companies

Want to implement this idea in a business?

We have generated a startup concept here: PeerHealthAI.