Node Classification on PPI (test)
90.5F1 (micro)GCN-BS
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| GCN-BSArchitecture=GCN2020.06 | 90.5 | — | — | — | |
| ClusterGCNArchitecture=GCN2020.06 | 85.3 | — | — | — | |
| GraphSAINTArchitecture=GCN2020.06 | 78.7 | — | — | — | |
| S-GCNArchitecture=GCN2020.06 | 74.4 | — | — | — | |
| GraphSAGEArchitecture=GCN2020.06 | 68.9 | — | — | — | |
| AS-GCNArchitecture=GCN2020.06 | 59.9 | — | — | — | |
| LADIESArchitecture=GCN2020.06 | 57.4 | — | — | — | |
| FastGCNArchitecture=GCN2020.06 | 50.2 | — | — | — | |
| GCNII*learning_protocol=inductive2021.08 | 0.9958 | — | — | — | |
| GCNIIlearning_protocol=inductive2021.08 | 0.9954 | — | — | — | |
| SGAS (Cri.1 best)Params (M)=23.18, Search Cost (GPU-days)=0.0032019.11 | 0.9946 | — | — | — | |
| SGAS (Cri.2 best)Params (M)=29.73, Search Cost (GPU-days)=0.0032019.11 | 0.9946 | — | — | — | |
| DenseMRGCN-14Params (M)=53.42, Search Cost (GPU-days)=manual2019.11 | 0.9943 | — | — | — | |
| ResMRGCN-28Params (M)=14.76, Search Cost (GPU-days)=manual2019.11 | 0.9941 | — | — | — | |
| Cluster-GCNlayers=5, hidden units=20482019.05 | 0.9936 | — | — | — | |
| Cluster-GCNlearning_protocol=inductive2021.08 | 0.9936 | — | — | — | |
| PDE-GCNMlearning_protocol=inductive2021.08 | 0.9918 | — | — | — | |
| PDE-GCNDlearning_protocol=inductive2021.08 | 0.9907 | — | — | — | |
| SGAS (small)Params (M)=0.40, Search Cost (GPU-days)=0.0032019.11 | 0.9889 | — | — | — | |
| GaAN2019.05 | 0.9871 | — | — | — | |
| GaANlearning_protocol=inductive2021.08 | 0.9871 | — | — | — | |
| MGNNI2022.10 | 0.987 | — | — | — | |
| GeniePath2019.05 | 0.985 | — | — | — | |
| GeniePathlearning_protocol=inductive2021.08 | 0.985 | — | — | — | |
| GraphSAINT2020.04 | 0.981 | — | — | — | |
| EIGNN2022.02 | 0.98 | — | — | — | |
| EIGNN2022.10 | 0.98 | — | — | — | |
| GeniePathParams (M)=1.81, Search Cost (GPU-days)=manual2019.11 | 0.979 | — | — | — | |
| GeniePathDepth=3, Params=1.81M2019.04 | 0.979 | — | — | — | |
| VR-GCN2019.05 | 0.978 | — | — | — | |
| VR-GCNlearning_protocol=inductive2021.08 | 0.978 | — | — | — | |
| JKNetlearning_protocol=inductive2021.08 | 0.976 | — | — | — | |
| IGNN2022.02 | 0.976 | — | — | — | |
| JKNet2022.10 | 0.976 | — | — | — | |
| IGNN2022.10 | 0.976 | — | — | — | |
| GAT2019.05 | 0.973 | — | — | — | |
| GATlearning_protocol=inductive2021.08 | 0.973 | — | — | — | |
| GAT2022.02 | 0.973 | — | — | — | |
| GAT2022.10 | 0.973 | — | — | — | |
| SIGNoperator powers (p, s, t)=(2, 0, 1)2020.04 | 0.97 | — | — | — | |
| GP-BS.MArchitecture=GeniePath2020.06 | 0.965 | — | — | — | |
| Stochastic-GCN2020.04 | 0.963 | — | — | — | |
| GP-BSArchitecture=GeniePath2020.06 | 0.958 | — | — | — | |
| PairETrain ratio=30%, Learning Setting=Unsupervised2022.03 | 0.9483 | — | — | — | |
| NodeETrain ratio=30%, Learning Setting=Unsupervised2022.03 | 0.9148 | — | — | — | |
| DGITrain ratio=30%, Learning Setting=Unsupervised2022.03 | 0.8958 | — | — | — | |
| S-GCN2020.04 | 0.892 | — | — | — | |
| ClusterGCN2020.04 | 0.875 | — | — | — | |
| GAT-BS.MArchitecture=GAT2020.06 | 0.867 | — | — | — | |
| GAT-BSArchitecture=GAT2020.06 | 0.841 | — | — | — | |
| SSE2022.02 | 0.836 | — | — | — | |
| SSE2022.10 | 0.836 | — | — | — | |
| GraphSAINT-GATArchitecture=GAT2020.06 | 0.789 | — | — | — | |
| GraphSAGE2022.02 | 0.786 | — | — | — | |
| GraphSAGE2022.10 | 0.786 | — | — | — | |
| SuperGATSDAttention Variant=SD2022.04 | 0.744 | — | — | — | |
| GAT2022.04 | 0.722 | — | — | — | |
| AS-GCN2020.04 | 0.687 | — | — | — | |
| CGAT2022.04 | 0.683 | — | — | — | |
| SuperGATMxAttention Variant=MX2022.04 | 0.672 | — | — | — | |
| ARVGETrain ratio=30%, Learning Setting=Unsupervised2022.03 | 0.6579 | — | — | — | |
| N-SAGE2018.02 | 0.65 | — | — | — | |
| PGETrain ratio=30%, Learning Setting=Semi-supervised2022.03 | 0.6397 | — | — | — | |
| GraphSAGE2020.04 | 0.637 | — | — | — | |
| GCN2022.04 | 0.615 | — | — | — | |
| SAGE-LSTM2018.02 | 0.612 | — | — | — | |
| GraphSAGE-LSTMSupervision protocol=Fully Supervised2017.06 | 0.612 | — | — | — | |
| GraphSAGE2019.05 | 0.612 | — | — | — | |
| GraphSAGE-LSTMLayers=2, Params=0.39M2019.09 | 0.612 | — | — | — | |
| GraphSAGE (LSTM)Params (M)=0.26, Search Cost (GPU-days)=manual2019.11 | 0.612 | — | — | — | |
| GraphSAGE(lstm)Depth=2, Params=0.26M2019.04 | 0.612 | — | — | — | |
| GraphSAGElearning_protocol=inductive2021.08 | 0.612 | — | — | — | |
| SAGE2018.02 | 0.6 | — | — | — | |
| GraphSAGE-poolSupervision protocol=Fully Supervised2017.06 | 0.6 | — | — | — | |
| GraphSAGE-poolLayers=2, Params=0.36M2019.09 | 0.6 | — | — | — | |
| SAGEImplementation=our implementation2018.02 | 0.598 | — | — | — | |
| GraphSAGE-meanSupervision protocol=Fully Supervised2017.06 | 0.598 | — | — | — | |
| GraphSAGE-meanLayers=2, Params=0.11M2019.09 | 0.598 | — | — | — | |
| SAGE-UTrain ratio=30%, Learning Setting=Unsupervised2022.03 | 0.5921 | — | — | — | |
| GCN2022.02 | 0.592 | — | — | — | |
| GCN2022.10 | 0.592 | — | — | — | |
| GraphSAGE2022.04 | 0.59 | — | — | — | |
| AS-GATArchitecture=GAT2020.06 | 0.566 | — | — | — | |
| GAUG-OBackbone=JK-NET2020.06 | 0.531 | — | — | — | |
| TransRTrain ratio=30%, Learning Setting=Unsupervised2022.03 | 0.5228 | — | — | — | |
| DeepWalkTrain ratio=30%, Learning Setting=Unsupervised2022.03 | 0.5156 | — | — | — | |
| GCN2020.04 | 0.515 | — | — | — | |
| FastGCN2020.04 | 0.513 | — | — | — | |
| GraphSAGE-poolSupervision protocol=Unsupervised2017.06 | 0.502 | — | — | — | |
| GraphSAGE-GCNSupervision protocol=Fully Supervised2017.06 | 0.5 | — | — | — | |
| GraphSAGE-GCNLayers=2, Params=0.11M2019.09 | 0.5 | — | — | — | |
| CANTrain ratio=30%, Learning Setting=Unsupervised2022.03 | 0.4991 | — | — | — | |
| GraphSAGE-meanSupervision protocol=Unsupervised2017.06 | 0.486 | — | — | — | |
| GraphSAGE-LSTMSupervision protocol=Unsupervised2017.06 | 0.482 | — | — | — | |
| H2GCNTrain ratio=30%, Learning Setting=Semi-supervised2022.03 | 0.4765 | — | — | — | |
| GAUG-MBackbone=JK-NET2020.06 | 0.474 | — | — | — | |
| N-GCN2018.02 | 0.468 | — | — | — | |
| GAUG-OBackbone=GCN2020.06 | 0.466 | — | — | — | |
| GraphSAGE-GCNSupervision protocol=Unsupervised2017.06 | 0.465 | — | — | — | |
| DROPEDGEBackbone=JK-NET2020.06 | 0.463 | — | — | — |