Table 5 The optimization performance (weighted macro-average metric) in binary classification
From: An application of convolutional neural networks with salient features for relation classification
Optimization | Random search | Grid search |
|---|---|---|
Execution time | 1150.446s | 1688.907s |
Accuracy | 90.53% | 87.30% |
Precision | 90.53% | 60.07% |
Recall | 50.00% | 58.04% |
F1-score | 90.53% | 93.06% |
Suggested | learning rate=0.001; l2=1.0E-5; | learning rate=0.1; l2=1.0E-5; |
hyperparameters | beta1=0.001; beta2=0.001; | beta1=0.001; beta2=0.001; |
| Â | epsilon=0.001; activation=ReLU; | epsilon=0.001; activation=ReLU; |