We propose a new Graph Neural Network that combines re-cent advancements in the field. We give theoretical contributions by prov-ing that the model is strictly more general than the Graph IsomorphismNetwork and the Gated Graph Neural Network, as it can approximate thesame functions and deal with arbitrary edge values. Then, we show howa single node information can flow through the graph unchanged
Theoretically Expressive and Edge-aware GraphLearning
Federico Errica;Davide Bacciu;Alessio Micheli
2020-01-01
Abstract
We propose a new Graph Neural Network that combines re-cent advancements in the field. We give theoretical contributions by prov-ing that the model is strictly more general than the Graph IsomorphismNetwork and the Gated Graph Neural Network, as it can approximate thesame functions and deal with arbitrary edge values. Then, we show howa single node information can flow through the graph unchangedFile in questo prodotto:
Non ci sono file associati a questo prodotto.
I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
