A naukri.com initiative
Arxiv
2d
110
Image Credit: Arxiv
Personalized Subgraph Federated Learning with Differentiable Auxiliary Projections
- Federated learning (FL) on graph-structured data faces challenges with non-IID scenarios where clients have distinct subgraphs from a global graph.
- Introduction of Federated learning with Auxiliary projections (FedAux), a framework in personalized subgraph FL.
- FedAux aligns, compares, and aggregates heterogeneously distributed local models without sharing raw data or node embeddings.
- The approach involves joint training of a local GNN and a learnable auxiliary projection vector, enabling effective client-specific information capture and model personalization.
Read Full Article
6 Likes
For uninterrupted reading, download the app