Graph Neural Networks

Graph neural networks generalize deep learning to graph-structured data using message-passing and aggregation. Architectures (GCN, GAT, MPNN) model node and edge features for tasks like node classification, link prediction, and graph-level regression.