Towards Quantifying Long-Range Interactions in Graph Machine Learning: a Large Graph Dataset and a Measurement

Benchmark Model Rank Results
node-classification-on-log-angelesChebNet#2Average Top-1 Accuracy: 54.1 ± 0.2
node-classification-on-log-angelesGraphSAGE#3Average Top-1 Accuracy: 53.3 ± 0.7
node-classification-on-log-angelesSGFormer#4Average Top-1 Accuracy: 47.2 ± 0.5
node-classification-on-log-angelesGCN#5Average Top-1 Accuracy: 45.9 ± 1.0
node-classification-on-londonChebNet#2Average Top-1 Accuracy: 49.4 ± 0.4
node-classification-on-londonGraphSAGE#3Average Top-1 Accuracy: 48.2 ± 0.8
node-classification-on-londonSGFormer#4Average Top-1 Accuracy: 45.7 ± 0.3
node-classification-on-londonGCN#5Average Top-1 Accuracy: 43.8 ± 0.3
node-classification-on-parisChebNet#2Average Top-1 Accuracy: 49.5 ± 0.4
node-classification-on-parisGraphSAGE#3Average Top-1 Accuracy: 49.1 ± 0.6
node-classification-on-parisGCN#4Average Top-1 Accuracy: 47.3 ± 0.2
node-classification-on-parisSGFormer#5Average Top-1 Accuracy: 45.0 ± 0.2
node-classification-on-shanghaiGraphSAGE#2Average Top-1 Accuracy: 60.4 ± 0.3
node-classification-on-shanghaiChebNet#3Average Top-1 Accuracy: 57.9 ± 1.4
node-classification-on-shanghaiSGFormer#4Average Top-1 Accuracy: 53.5 ± 0.3
node-classification-on-shanghaiGCN#5Average Top-1 Accuracy: 52.4 ± 0.3