Node Classification 벤치마크
Node Classification on Wiki-Vote
Accuracy
- 2016-09-09 — GCN_cheby (Kipf and Welling, 2017): Accuracy 49.5
- 2017-10-30 — GAT (Velickovic et al., 2018): Accuracy 59.4
- 2019-06-05 — DEMO-Net(weight): Accuracy 99.8
| Rank | Model | Accuracy | Paper | Code | Year |
|---|---|---|---|---|---|
| 1 | DEMO-Net(weight) | 99.8 | DEMO-Net: Degree-specific Graph Neural Networks for Node and Graph Classification | jwu4sml/DEMO-Net | 2019 |
| 2 | GAT (Velickovic et al., 2018) | 59.4 | Graph Attention Networks | labmlai/annotated_deep_learning_paper_implementations · dmlc/dgl · dmlc/dgl · +90 | 2017 |
| 3 | GCN_cheby (Kipf and Welling, 2017) | 49.5 | Semi-Supervised Classification with Graph Convolutional Networks | dmlc/dgl · dmlc/dgl · tkipf/gcn · +52 | 2016 |
| 4 | Union (Li et al., 2018) | 46.3 | Deeper Insights into Graph Convolutional Networks for Semi-Supervised Learning | liqimai/gcn | 2018 |
| 5 | GCN (Kipf and Welling, 2017) | 32.9 | Semi-Supervised Classification with Graph Convolutional Networks | dmlc/dgl · dmlc/dgl · tkipf/gcn · +52 | 2016 |
| 6 | GraphSAGE (Hamilton et al., [2017a]) | 24.5 | Inductive Representation Learning on Large Graphs | pyg-team/pytorch_geometric · dmlc/dgl · dmlc/dgl · +17 | 2017 |