Skip to content

Latest commit

 

History

History
17 lines (12 loc) · 677 Bytes

social_graphs.md

File metadata and controls

17 lines (12 loc) · 677 Bytes

Social Network Classification with Federated TUDataset

Motivation

There has been an skyrocketing interest in supervised ML with graph data, especially using graph neural networks in recent years. TUDataset is the most popular dataset used for for graph classification and regression. Our aim is to provide a federated platform to simulate federated graph classification and regression tasks. Thus, we introduce federated TUDataset benchmark for FedML users. All you need to do is to install FedML and PyG!

Training

Here, we provide scripts and config files under config/ to explore different model and dataset combinations!

sh run_social_networks.sh 4