Graph Classification

Evaluation script for various methods on common benchmark datasets via 10-fold cross validation, where a training fold is randomly sampled to serve as a validation set. Hyperparameter selection is performed for the number of hidden units and the number of layers with respect to the validation set:

• GCN
• GraphSAGE
• GIN
• Graclus
• Top-K Pooling
• SAG Pooling
• DiffPool
• EdgePool
• GlobalAttention
• Set2Set
• SortPool
• ASAPool

Run (or modify) the whole test suite via

$ python main.py

For more comprehensive time-measurement and memory usage information, you may use

$ python main_performance.py