Dataset: MUTAG,
Model: GCN

params={'seed': 41, 'epochs': 1000, 'batch_size': 20, 'init_lr': 0.0007, 'lr_reduce_factor': 0.5, 'lr_schedule_patience': 25, 'min_lr': 1e-06, 'weight_decay': 0.0, 'print_epoch_interval': 5, 'max_time': 30.0}

net_params={'device': device(type='cuda'), 'gpu_id': 0, 'batch_size': 20, 'L': 4, 'hidden_dim': 146, 'out_dim': 146, 'residual': True, 'readout': 'mean', 'in_feat_dropout': 0.0, 'dropout': 0.0, 'batch_norm': True, 'self_loop': False, 'optimizer': 'Adam', 'cluster': True, 'in_dim': 7, 'edge_dim': 7, 'max_num_node': 28, 'n_classes': 2, 'total_param': 209986}

GCNNet(
  (embedding_h): Linear(in_features=7, out_features=146, bias=True)
  (in_feat_dropout): Dropout(p=0.0, inplace=False)
  (layers): ModuleList(
    (0): GCNLayer(in_channels=146, out_channels=146, residual=True)
    (1): GCNLayer(in_channels=146, out_channels=146, residual=True)
    (2): GCNLayer(in_channels=146, out_channels=146, residual=True)
    (3): GCNLayer(in_channels=146, out_channels=146, residual=True)
  )
  (MLP_layer): MLPReadout(
    (FC_layers): ModuleList(
      (0): Linear(in_features=146, out_features=73, bias=True)
      (1): Linear(in_features=73, out_features=36, bias=True)
      (2): Linear(in_features=36, out_features=2, bias=True)
    )
  )
  (cos_sim): CosineSimilarity()
  (sim): CosineSimilarity()
  (mlp): Linear(in_features=292, out_features=146, bias=True)
  (mlp_cluster): Linear(in_features=146, out_features=146, bias=True)
  (pool): SAGPoolBlock(
    (pool): SAGPool(
      (score_layer): GraphConv(in=146, out=1, normalization=both, activation=None)
      (softmax): Softmax(dim=None)
    )
    (avgpool): AvgPooling()
    (maxpool): MaxPooling()
    (mlp): Linear(in_features=292, out_features=146, bias=True)
  )
)

Total Parameters: 209986

 edge_num: 69


    FINAL RESULTS
TEST ACCURACY averaged: 80.3216 with s.d. 8.1642
TRAIN ACCURACY averaged: 89.6141 with s.d. 1.2693


    Average Convergence Time (Epochs): 343.8000 with s.d. 40.2885
Total Time Taken: 0.2221 hrs
Average Time Per Epoch: 0.2240 s


All Splits Test Accuracies: [0.7894736842105263, 0.8421052631578947, 0.6842105263157895, 0.8947368421052632, 0.7368421052631579, 0.9473684210526315, 0.6842105263157895, 0.8421052631578947, 0.7777777777777778, 0.8333333333333334]