Is Heterophily A Real Nightmare For Graph Neural Networks To Do Node Classification?

Benchmark Model Rank Results
node-classification-on-citeseerACM-GCNAccuracy: 81.68 ± 0.97
node-classification-on-citeseerACM-Snowball-2Accuracy: 81.58 ± 1.23
node-classification-on-citeseerACMII-Snowball-2Accuracy: 82.07 ± 1.04
node-classification-on-citeseerACMII-Snowball-3Accuracy: 81.56 ± 1.15
node-classification-on-coraACM-GCNAccuracy: 88.62% ± 1.22%
node-classification-on-coraACM-Snowball-2Accuracy: 88.83% ± 1.49%
node-classification-on-coraACMII-GCNAccuracy: 88.95% ± 1.04%
node-classification-on-coraACMII-Snowball-3Accuracy: 89.36% ± 1.26%
node-classification-on-pubmedACM-GCNAccuracy: 90.74 ± 0.5
node-classification-on-pubmedACMII-Snowball-2Accuracy: 90.56 ± 0.39
node-classification-on-pubmedACMII-Snowball-3Accuracy: 91.31 ± 0.6