GraphSAINT: Graph Sampling Based Inductive Learning Method

Benchmark Model Rank Results
link-property-prediction-on-ogbl-citation2GraphSAINT (GCN aggr)#15Ext. data: NoTest MRR: 0.7985 ± 0.0040Validation MRR: 0.7975 ± 0.0039
node-classification-on-ppiGraphSAINT#3F1: 99.50
node-classification-on-redditGraphSAINT#5Accuracy: 97.0%
node-property-prediction-on-ogbn-magGraphSAINT + metapath2vec#24Test Accuracy: 0.4966 ± 0.0022Ext. data: No
node-property-prediction-on-ogbn-magGraphSAINT (R-GCN aggr)#26Test Accuracy: 0.4751 ± 0.0022Ext. data: No
node-property-prediction-on-ogbn-productsGraphSAINT-inductive#42Test Accuracy: 0.8027 ± 0.0026Ext. data: No
node-property-prediction-on-ogbn-productsGraphSAINT (SAGE aggr)#47Test Accuracy: 0.7908 ± 0.0024Ext. data: No