Decentralized Graph Neural Network for Privacy-Preserving Recommendation

By X. Zheng et al

Published on June 10, 2023

Read the original document by opening this link in a new tab.

Table of Contents

1. Introduction
2. Related Work
3. Method
3.1 Graph Construction
3.2 Local Gradient Calculation
3.3 Global Gradient Passing
4. Conclusion

Summary

The document discusses the development of a novel decentralized Graph Neural Network (GNN) for privacy-preserving recommendations. It addresses the challenges in privacy protection and training efficiency faced by existing methods. The framework consists of three stages: graph construction, local gradient calculation, and global gradient passing. The inter-user graph and inner-item hypergraph are constructed to model user preferences and calculate gradients based on interactions. A secure gradient-sharing mechanism is proposed to protect user privacy. Extensive experiments validate the effectiveness of the proposed framework.