PROTEINS
Description
PROTEINS
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 | 75.70 | 2.14 |
| GIN | 75.20 | 3.30 |
| GAT | 76.34 | 1.66 |
| AST | 76.63 | 1.74 |
| EDGNN | 73.91 | 4.39 |
| UniGNN2 | 75.20 | 2.96 |
| CWN | 76.13 | 2.70 |
| CCCN | 73.33 | 2.30 |
| SCCNN | 74.19 | 2.86 |
| SCN | 75.27 | 2.14 |
Key Insights
- AST achieves the best performance with 76.63% accuracy
- Most models perform in the 73-76% range, with GAT and CWN also above 76%
- EDGNN shows the highest variability (±4.39)
- GIN and UniGNN2 have identical mean accuracy (75.20%)
Reproducibility
python -m topobench model=hypergraph/allsettransformer dataset=graph/PROTEINS optimizer.parameters.lr=0.01 model.feature_encoder.out_channels=32 model.backbone.n_layers=2 model.feature_encoder.proj_dropout=0.25 dataset.dataloader_params.batch_size=128 dataset.split_params.data_seed=0,3,5,7,9 trainer.max_epochs=500 trainer.min_epochs=50 trainer.check_val_every_n_epoch=5 callbacks.early_stopping.patience=10 logger.wandb.project=TopoBench --multirun