edGNN: a Simple and Powerful GNN for Directed Labeled Graphs

18 Apr 2019Guillaume JaumeAn-phi NguyenMaría Rodríguez MartínezJean-Philippe ThiranMaria Gabrani

The ability of a graph neural network (GNN) to leverage both the graph topology and graph labels is fundamental to building discriminative node and graph embeddings. Building on previous work, we theoretically show that edGNN, our model for directed labeled graphs, is as powerful as the Weisfeiler-Lehman algorithm for graph isomorphism... (read more)

PDF Abstract

Evaluation Results from the Paper


TASK DATASET MODEL METRIC NAME METRIC VALUE GLOBAL RANK COMPARE
Graph Classification MUTAG edGNN(avg) Accuracy(10-fold) 86.9 # 1
Graph Classification PTC edGNN (avg) Accuracy 86.4 # 1