Node Classification 벤치마크
Node Classification on AMZ Comp
Accuracy
- 2019-10-28 — GCN (Heat Diffusion): Accuracy 86.77
- 2023-08-17 — HH-GCN: Accuracy 90.92
| Rank | Model | Accuracy | Paper | Code | Year |
|---|---|---|---|---|---|
| 1 | HH-GCN | 90.92% | Half-Hop: A graph upsampling approach for slowing down message passing | nerdslab/halfhop | 2023 |
| 2 | GCN | 90.22% | Half-Hop: A graph upsampling approach for slowing down message passing | nerdslab/halfhop | 2023 |
| 3 | GCN (Heat Diffusion) | 86.77% | Diffusion Improves Graph Learning | klicperajo/gdc · jhngjng/naq-pytorch · hazdzz/GDC | 2019 |
| 4 | HH-GraphSAGE | 86.6% | Half-Hop: A graph upsampling approach for slowing down message passing | nerdslab/halfhop | 2023 |
| 5 | SIGN | 85.93 ± 1.21 | SIGN: Scalable Inception Graph Neural Networks | dmlc/dgl · twitter-research/sign · facebookresearch/NARS · +2 | 2020 |
| 6 | GraphSAGE | 84.79% | Half-Hop: A graph upsampling approach for slowing down message passing | nerdslab/halfhop | 2023 |
| 7 | Graph InfoClust (GIC) | 81.5 ± 1.0 | Graph InfoClust: Leveraging cluster-level node information for unsupervised graph representation learning | cmavro/Graph-InfoClust-GIC · cmavro/HeMI | 2020 |