Zero-shot Transfer Learning within a Heterogeneous Graph via Knowledge Transfer Networks

State-of-the-art graph learning methods for HGs known as heterogeneous graph neural networks (HGNNs) are applied to learn deep context-informed node representations.