message passing does in GNNs: each node aggregates features from its neighbors

GAT (Graph Attention Network) adds: learned attention weights between neighbors

GAT adds learned attention weights between neighbors, enhancing node feature aggregation

the over-smoothing problem is in GNNs: deep GNNs make all node features converge

Over-smoothing in GNNs: Deeper layers cause node features to converge too much, losing unique node identities

GraphSAGE does: samples and aggregates a fixed-size neighborhood

GraphSAGE samples and aggregates a fixed-size neighborhood

GCN (Graph Convolutional Network) does: spectral convolution approximated by neighbor averaging

GCN approximates spectral convolution via neighbor averaging

Graph neural network

Graph pooling reduces graphs to single vectors for graph-level prediction

the lottery ticket hypothesis says: sparse subnetworks can match full network performance

Lottery ticket hypothesis posits sparse subnetworks can match full network performance