Graphs

8 concepts, 8 reviewed and merged.

Architecture

Graph ConvolutionThe simplest useful graph layer: average each node's neighbours, mix in the node itself, multiply by a weight matrix.verified
Graph TransformerIf message passing is limited by the edges it is given, let every node attend to every other and put the structure…verified

Attention

Graph AttentionA graph convolution treats every neighbour as equally worth listening to.verified

Foundations

Graph ExpressivityThere are pairs of graphs a message-passing network simply cannot tell apart — not because it was trained badly, but…verified
Graph Neural NetworkSome data is not a sequence and not a grid — molecules, road networks, citations, social graphs.verified
Message PassingThe operation nearly every graph network is built from: each node sends a message to its neighbours, each node adds…verified

Regimes

OversmoothingStack enough graph layers and every node ends up looking the same.verified

Systems

Neighborhood SamplingA node's two-hop neighbourhood on a social graph can be most of the graph, so computing one node's representation…verified