Are Powerful Graph Neural Nets Necessary? A Dissection on Graph Classification

Benchmark Model Rank Results
graph-classification-on-collabGFN#4Accuracy: 81.50%
graph-classification-on-collabGFN-light#6Accuracy: 81.34%
graph-classification-on-ddGFN#21Accuracy: 78.78%
graph-classification-on-ddGFN-light#25Accuracy: 78.62%
graph-classification-on-enzymesGFN#12Accuracy: 70.17%
graph-classification-on-enzymesGFN-light#14Accuracy: 69.50%
graph-classification-on-imdb-bGFN#32Accuracy: 73.00%
graph-classification-on-imdb-bGFN-light#33Accuracy: 73.00%
graph-classification-on-imdb-mGFN#12Accuracy: 51.80%
graph-classification-on-imdb-mGFN-light#15Accuracy: 51.20%
graph-classification-on-mutagGFN#14Accuracy: 90.84%
graph-classification-on-mutagGFN-light#23Accuracy: 89.89%
graph-classification-on-nci1GFN#22Accuracy: 83.65%
graph-classification-on-nci1GFN-light#31Accuracy: 81.43%
graph-classification-on-proteinsGFN-light#26Accuracy: 77.44%
graph-classification-on-proteinsGFN#38Accuracy: 76.46%
graph-classification-on-re-m5kGFN-light#5Accuracy: 49.75%
graph-classification-on-re-m5kGFN#6Accuracy: 49.43%