Roman Empire
Description
Roman Empire
Dataset Overview
Lifting Methods
Structural-based Liftings
- Cellular: Cycle-based lifting
- Simplicial: Clique complex lifting
- Hypergraph: k-hop lifting
Feature Lifting
- Projected sum
Model Performance
| Model | Accuracy (%) | Std Dev (±) |
|---|---|---|
| GCN | 78.16 | 0.32 |
| GIN | 79.56 | 0.20 |
| GAT | 84.02 | 0.51 |
| AST | 79.50 | 0.13 |
| EDGNN | 81.01 | 0.24 |
| UniGNN2 | 77.06 | 0.20 |
| CWN | 81.81 | 0.62 |
| CCCN | 82.14 | 0.00 |
| SCCNN | 89.15 | 0.32 |
| SCN | 88.79 | 0.46 |
Key Insights
- SCCNN achieves the best performance with 89.15% accuracy
- SCN is the second best with 88.79% accuracy
- GAT is the best among graph models with 84.02% accuracy
- Most models perform in the 77-84% range, with SCCNN and SCN outperforming others by a significant margin
Reproducibility
python -m topobench model=simplicial/sccnn_custom dataset=graph/roman_empire optimizer.parameters.lr=0.01 model.feature_encoder.out_channels=64 model.backbone.n_layers=2 model.readout.readout_name=NoReadOut transforms.graph2simplicial_lifting.signed=True model.feature_encoder.proj_dropout=0.5 dataset.dataloader_params.batch_size=1 dataset.split_params.data_seed=0,3,5,7,9 trainer.max_epochs=1000 trainer.min_epochs=50 trainer.check_val_every_n_epoch=1 callbacks.early_stopping.patience=50 logger.wandb.project=TopoBench --multirun