Authors: Qi Li, Zhuotao Liu, Qi Li, Ke Xu

Published on: September 03, 2023

Impact Score: 8.22

Arxiv code: Arxiv:2309.01098

Summary

• What is new: A novel federated learning architecture, martFL, designed for a secure utility-driven data marketplace.
• Why this is important: Challenges in existing Federated Learning architectures such as private model evaluation, malicious participant handling, and fair billing mechanisms.
• What the research proposes: martFL introduces a quality-aware model aggregation protocol and a verifiable data transaction protocol.
• Results: Improvement of model accuracy by up to 25% and reduction in data acquisition cost by up to 64%.

Technical Details

Technological frameworks used: martFL Federated Learning Architecture

Models used: nan

Data used: nan

Potential Impact

Data marketplaces, companies dealing with private-domain data, sectors prioritizing data privacy like healthcare, finance, and personalized advertising.

Want to implement this idea in a business?

We have generated a startup concept here: DataHarmonics.