Node Classification on Pubmed (test)
91.18AccuracyRGLM-Similarizer
Evaluation results
| Method | Links | ||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| RGLM-SimilarizerTraining set=Cora+Pubmed2026.03 | 91.18 | — | — | — | 91.07 | — | — | — | — | — | — | — | |
| ANS-GT + HDSEstructural encoding=HDSE2023.08 | 90.63 | — | — | — | — | — | — | — | — | — | — | — | |
| Adaptvariance reduction=false2018.09 | 90.6 | — | — | — | — | — | — | — | — | — | — | — | |
| Adaptvariance reduction=true2018.09 | 90.6 | — | — | — | — | — | — | — | — | — | — | — | |
| RDGNN-I2024.06 | 90.37 | — | — | — | — | — | — | — | — | — | — | — | |
| RGLM-DecoderTraining set=Cora+Pubmed2026.03 | 90.37 | — | — | — | 90.32 | — | — | — | — | — | — | — | |
| RGLM-DenoiserTraining set=Cora+Pubmed2026.03 | 90.34 | — | — | — | 90.36 | — | — | — | — | — | — | — | |
| GCNII2024.06 | 90.3 | — | — | — | — | — | — | — | — | — | — | — | |
| GREAD2024.06 | 90.23 | — | — | — | — | — | — | — | — | — | — | — | |
| Fullsampling=full-batch2018.09 | 90.22 | — | — | — | — | — | — | — | — | — | — | — | |
| GCNII2025.11 | 90.15 | — | — | — | — | — | — | — | — | — | — | — | |
| RDGNN-S2024.06 | 90.11 | — | — | — | — | — | — | — | — | — | — | — | |
| Geom-GCN-INumber of Layers=2, Embedding Method=Isomap, Hidden Units=5122020.02 | 90.05 | — | — | — | — | — | — | — | — | — | — | — | |
| Geom-GCNbest model variant=true2021.06 | 90.05 | — | — | — | — | — | — | — | — | — | — | — | |
| GEOM-GCN2022.05 | 90.05 | — | — | — | — | — | — | — | — | — | — | — | |
| Geom-GCN†variant=maximal accuracy2024.06 | 90.05 | — | — | — | — | — | — | — | — | — | — | — | |
| GRAFF†variant=maximal accuracy2024.06 | 90.04 | — | — | — | — | — | — | — | — | — | — | — | |
| Node-Wisesampling=node-wise2018.09 | 90.02 | — | — | — | — | — | — | — | — | — | — | — | |
| Geom-GCN2025.11 | 89.95 | — | — | — | — | — | — | — | — | — | — | — | |
| PDE-GCNMsubscript=M2024.06 | 89.93 | — | — | — | — | — | — | — | — | — | — | — | |
| UDGNNBackbone=SAGE2022.05 | 89.88 | — | — | — | — | — | — | — | — | — | — | — | |
| UDGNNBackbone=GCN2022.05 | 89.85 | — | — | — | — | — | — | — | — | — | — | — | |
| UDGNNBackbone=GAT2022.05 | 89.78 | — | — | — | — | — | — | — | — | — | — | — | |
| DCNNhops=22015.11 | 89.76 | 89.76 | 89.43 | — | — | — | — | — | — | — | — | — | |
| 2-hop DCNN2018.09 | 89.76 | — | — | — | — | — | — | — | — | — | — | — | |
| Coarformer2023.08 | 89.75 | — | — | — | — | — | — | — | — | — | — | — | |
| BundlePolySD2025.11 | 89.75 | — | — | — | — | — | — | — | — | — | — | — | |
| GeneralPolySD2025.11 | 89.73 | — | — | — | — | — | — | — | — | — | — | — | |
| ACMII-GCN++2024.06 | 89.71 | — | — | — | — | — | — | — | — | — | — | — | |
| DiagPolySD2025.11 | 89.7 | — | — | — | — | — | — | — | — | — | — | — | |
| SignGT2023.10 | 89.64 | — | — | — | — | — | — | — | — | — | — | — | |
| UDGNN*Backbone=GAT2022.05 | 89.62 | — | — | — | — | — | — | — | — | — | — | — | |
| UDGNN*Backbone=SAGE2022.05 | 89.6 | — | — | — | — | — | — | — | — | — | — | — | |
| LLaGATraining set=Cora+Pubmed2026.03 | 89.6 | — | — | — | 89.46 | — | — | — | — | — | — | — | |
| H2GCN-22022.05 | 89.59 | — | — | — | — | — | — | — | — | — | — | — | |
| ED-HNN2023.08 | 89.56 | — | — | — | — | — | — | — | — | — | — | — | |
| ANS-GT2023.08 | 89.56 | — | — | — | — | — | — | — | — | — | — | — | |
| H2GCN2023.08 | 89.55 | — | — | — | — | — | — | — | — | — | — | — | |
| NSD†variant=maximal accuracy, record_order=12024.06 | 89.49 | — | — | — | — | — | — | — | — | — | — | — | |
| H2GCN2024.06 | 89.49 | — | — | — | — | — | — | — | — | — | — | — | |
| NSD†variant=maximal accuracy, record_order=22024.06 | 89.49 | — | — | — | — | — | — | — | — | — | — | — | |
| NSDrestriction maps=orthogonal2025.11 | 89.49 | — | — | — | — | — | — | — | — | — | — | — | |
| H2GCN2025.11 | 89.49 | — | — | — | — | — | — | — | — | — | — | — | |
| NSDrestriction maps=diagonal2025.11 | 89.42 | — | — | — | — | — | — | — | — | — | — | — | |
| H2GCN-12022.05 | 89.4 | — | — | — | — | — | — | — | — | — | — | — | |
| UDGNN*Backbone=GCN2022.05 | 89.39 | — | — | — | — | — | — | — | — | — | — | — | |
| GPRGNN2023.10 | 89.38 | — | — | — | — | — | — | — | — | — | — | — | |
| GPRGNN2023.08 | 89.38 | — | — | — | — | — | — | — | — | — | — | — | |
| DAGNN2022.05 | 89.37 | — | — | — | — | — | — | — | — | — | — | — | |
| NSDrestriction maps=general2025.11 | 89.33 | — | — | — | — | — | — | — | — | — | — | — | |
| Conn-NSD2025.11 | 89.28 | — | — | — | — | — | — | — | — | — | — | — | |
| DMP†variant=maximal accuracy2024.06 | 89.27 | — | — | — | — | — | — | — | — | — | — | — | |
| GraphSAGETraining set=Cora+Pubmed2026.03 | 89.27 | — | — | — | 89.12 | — | — | — | — | — | — | — | |
| GCNII2022.05 | 89.26 | — | — | — | — | — | — | — | — | — | — | — | |
| NodeFormer2023.10 | 89.24 | — | — | — | — | — | — | — | — | — | — | — | |
| ANSD2025.11 | 89.21 | — | — | — | — | — | — | — | — | — | — | — | |
| UniMPTraining set=Cora+Pubmed2026.03 | 89.2 | — | — | — | 89.14 | — | — | — | — | — | — | — | |
| Specformer2023.10 | 89.19 | — | — | — | — | — | — | — | — | — | — | — | |
| GGCN2024.06 | 89.15 | — | — | — | — | — | — | — | — | — | — | — | |
| GGCN2025.11 | 89.15 | — | — | — | — | — | — | — | — | — | — | — | |
| FAGCN2023.10 | 89.14 | — | — | — | — | — | — | — | — | — | — | — | |
| UniG-Encoder2023.08 | 88.98 | — | — | — | — | — | — | — | — | — | — | — | |
| DeeperGCN2022.05 | 88.8 | — | — | — | — | — | — | — | — | — | — | — | |
| AllDeepSets2023.08 | 88.75 | — | — | — | — | — | — | — | — | — | — | — | |
| AllSet Transformer2023.08 | 88.72 | — | — | — | — | — | — | — | — | — | — | — | |
| GRAPH CASCADESVariant=VCR-TAS2026.06 | 88.7 | — | — | — | — | — | — | — | — | — | — | — | |
| VCR-TASBackbone=VCR, Cascade Strategy=TAS2026.06 | 88.7 | — | — | — | — | — | — | — | — | — | — | — | |
| VCR-TASMethod Category=Tokenized Sparse-Attention GTs, Rewiring=TAS2026.06 | 88.7 | — | — | — | — | — | — | — | — | — | — | — | |
| GloGNN2023.10 | 88.58 | — | — | — | — | — | — | — | — | — | — | — | |
| LE-GCN2023.08 | 88.53 | — | — | — | — | — | — | — | — | — | — | — | |
| WRGAT2021.06 | 88.52 | — | — | — | — | — | — | — | — | — | — | — | |
| WRGAT2024.06 | 88.52 | — | — | — | — | — | — | — | — | — | — | — | |
| H2GCNbest model variant=true2021.06 | 88.5 | — | — | — | — | — | — | — | — | — | — | — | |
| MLP2023.08 | 88.5 | — | — | — | — | — | — | — | — | — | — | — | |
| GraphSage2021.06 | 88.45 | — | — | — | — | — | — | — | — | — | — | — | |
| GraphSAGE2022.05 | 88.45 | — | — | — | — | — | — | — | — | — | — | — | |
| GraphSAGE2025.11 | 88.45 | — | — | — | — | — | — | — | — | — | — | — | |
| NodeFormerTraining set=Cora+Pubmed2026.03 | 88.44 | — | — | — | 88.22 | — | — | — | — | — | — | — | |
| GCN2025.11 | 88.42 | — | — | — | — | — | — | — | — | — | — | — | |
| JKNet-GCN2022.05 | 88.41 | — | — | — | — | — | — | — | — | — | — | — | |
| Gophormer2023.08 | 88.33 | — | — | — | — | — | — | — | — | — | — | — | |
| NAG-TASBackbone=NAG, Cascade Strategy=TAS2026.06 | 88.26 | — | — | — | — | — | — | — | — | — | — | — | |
| NAG-TASMethod Category=Tokenized Sparse-Attention GTs, Rewiring=TAS2026.06 | 88.26 | — | — | — | — | — | — | — | — | — | — | — | |
| UniGCNII2023.08 | 88.25 | — | — | — | — | — | — | — | — | — | — | — | |
| JdSNN2025.11 | 88.19 | — | — | — | — | — | — | — | — | — | — | — | |
| NAG-MASBackbone=NAG, Cascade Strategy=MAS2026.06 | 88.19 | — | — | — | — | — | — | — | — | — | — | — | |
| NAG-MASMethod Category=Tokenized Sparse-Attention GTs, Rewiring=MAS2026.06 | 88.19 | — | — | — | — | — | — | — | — | — | — | — | |
| GCN-TASBackbone=GCN, Cascade Strategy=TAS2026.06 | 88.15 | — | — | — | — | — | — | — | — | — | — | — | |
| GCN-TASMethod Category=GNN-based Architectures, Rewiring=TAS2026.06 | 88.15 | — | — | — | — | — | — | — | — | — | — | — | |
| GCNNumber of Layers=2, Hidden Units=642020.02 | 88.13 | — | — | — | — | — | — | — | — | — | — | — | |
| GCN2024.06 | 88.13 | — | — | — | — | — | — | — | — | — | — | — | |
| Geom-GCN-PNumber of Layers=2, Embedding Method=Poincare, Hidden Units=5122020.02 | 88.09 | — | — | — | — | — | — | — | — | — | — | — | |
| JdSNNweighting=none2025.11 | 88.09 | — | — | — | — | — | — | — | — | — | — | — | |
| GCN-Cheby2021.06 | 88.08 | — | — | — | — | — | — | — | — | — | — | — | |
| FastGCN2018.09 | 88 | — | — | — | — | — | — | — | — | — | — | — | |
| RiSNN2025.11 | 88 | — | — | — | — | — | — | — | — | — | — | — | |
| APPNP2022.05 | 87.94 | — | — | — | — | — | — | — | — | — | — | — | |
| RiSNNtransport=none2025.11 | 87.91 | — | — | — | — | — | — | — | — | — | — | — | |
| GCN-MASBackbone=GCN, Cascade Strategy=MAS2026.06 | 87.87 | — | — | — | — | — | — | — | — | — | — | — | |
| GCN-MASMethod Category=GNN-based Architectures, Rewiring=MAS2026.06 | 87.87 | — | — | — | — | — | — | — | — | — | — | — |