Node Classification on Cora Planetoid (public split)
85.5AccuracyGCNII
Evaluation results
| Method | Links | |
|---|---|---|
| GCNII2026.03 | 85.5 | |
| GCNvariant=enhanced2026.03 | 85.1 | |
| SGFormer2026.03 | 84.82 | |
| GIST2026.03 | 84 | |
| GAT2026.03 | 83 | |
| GATv22026.03 | 82.9 | |
| GATHyperparameter Tuning=per-model, Measurement=mean ± std over 10 seeds2026.03 | 82.8 | |
| APPNPHyperparameter Tuning=per-model, Measurement=mean ± std over 10 seeds2026.03 | 82.8 | |
| Hier(8)Hyperparameter Tuning=per-model, Measurement=mean ± std over 10 seeds, Variant=GHN-Hier (hierarchical with 8 groups)2026.03 | 82.3 | |
| GCNHyperparameter Tuning=per-model, Measurement=mean ± std over 10 seeds2026.03 | 82.2 | |
| LSEHyperparameter Tuning=per-model, Measurement=mean ± std over 10 seeds, Variant=GHN-LSE (flat, softmax retrieval)2026.03 | 82 | |
| LSRHyperparameter Tuning=per-model, Measurement=mean ± std over 10 seeds, Variant=GHN-LSR (flat, Epanechnikov retrieval)2026.03 | 82 | |
| GCNvariant=base2026.03 | 81.6 | |
| NoMemHyperparameter Tuning=per-model, Measurement=mean ± std over 10 seeds, Variant=GHN-NoMem (memory-disabled ablation)2026.03 | 80.3 | |
| GraphSAGEHyperparameter Tuning=per-model, Measurement=mean ± std over 10 seeds2026.03 | 80.2 | |
| GIN2026.03 | 77.6 | |
| GINHyperparameter Tuning=per-model, Measurement=mean ± std over 10 seeds2026.03 | 77.5 | |
| GraphSAGE2026.03 | 71.49 | |
| MLPHyperparameter Tuning=per-model, Measurement=mean ± std over 10 seeds2026.03 | 60 |