Table 2.
The most efficient 8 networks selected from 2000 networks based on classification accuracy and test loss.
| Network Topology * | Dropout | Epoch | Classification Accuracy (%) | Test loss (CCE) |
|---|---|---|---|---|
| 60-242-7 | 0.4 | 814 | 100 | 0.03804 |
| 60-178-7 | 0.1 | 738 | 100 | 0.05484 |
| 60-240-7 | 1 | 746 | 100 | 0.06449 |
| 60-106-7 | 1 | 1417 | 100 | 0.06616 |
| 60-199-7 | 0.4 | 852 | 100 | 0.07257 |
| 60-226-7 | 0.1 | 499 | 100 | 0.07568 |
| 60-201-7 | 0.7 | 1388 | 100 | 0.07896 |
| 60-238-7 | 0.4 | 712 | 100 | 0.089558 |
* number of neurons in input layer, hidden layer, and output layer, respectively.