Building upon the foundational architectures and training solutions from previous chapters, we now shift focus to applying Graph Neural Networks to more complex graph structures and tasks. This chapter guides you through adapting GNNs for heterogeneous graphs, which contain diverse node and edge types, using approaches such as Relational GCN (RGCN) and Heterogeneous Attention Network (HAN). You will also study techniques for modeling dynamic graphs where the structure or features change over time.

We will examine how GNNs contribute to knowledge graph embeddings and link prediction. Additionally, the chapter covers graph generative models for creating new graph data, self-supervised learning methods for graph representation learning without explicit labels, and advanced graph pooling and readout functions necessary for graph-level prediction tasks. By the end of this chapter, you will be equipped to handle a wider variety of graph-based problems using specialized GNN techniques.