Node Classification on Pubmed (Epsilon Accuracy)
80.5Accuracy (Clean)GPRGNN
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| GPRGNNAdversarial training=true, Attack type=LR-BCD2023.06 | 80.5 | 75.8 | 74.7 | |
| GPRGNNAdversarial training=true, Attack type=PR-BCD2023.06 | 80.4 | 73 | 68.8 | |
| SGFormer2026.07 | 80.3 | — | — | |
| GraphGPS2026.07 | 79.94 | — | — | |
| NodeFormer2026.07 | 79.9 | — | — | |
| NAGphormer2026.07 | 79.73 | — | — | |
| X-LogSMaskConfiguration=ours2026.07 | 79.6 | — | — | |
| Exphormer2026.07 | 79.46 | — | — | |
| GraphSAGE2026.07 | 79.41 | — | — | |
| Polynormer2026.07 | 79.24 | — | — | |
| X-LogSMaskLayers=12026.07 | 79.2 | — | — | |
| GOAT2026.07 | 79.13 | — | — | |
| GAT2026.07 | 79 | — | — | |
| GPRGNNAdversarial training=false, Attack type=LR-BCD2023.06 | 78.8 | 61.1 | 54 | |
| GPRGNNAdversarial training=false, Attack type=PR-BCD2023.06 | 78.8 | 57 | 42.7 | |
| GCN2026.07 | 78.8 | — | — | |
| GCNAdversarial training=false, Attack type=LR-BCD2023.06 | 77.4 | 60.4 | 52.8 | |
| GCNAdversarial training=false, Attack type=PR-BCD2023.06 | 77.4 | 54 | 39.3 |