Table 2.
Summarized information about computed three-layer-perceptron architectures used for prediction of inhibitory data against selected human carboanhydrases (hCAs).
| ANN | hCA I | hCA II | hCA IV | hCA IX | |
|---|---|---|---|---|---|
| Architecture * | 3-MLP | 30-17-1 | 30-13-1 | 30-19-1 | 30-12-1 |
| Activation function | Hidden layer | tanh | tanh | tanh | tanh |
| Output layer | exp | log | log | log | |
| Training set | Performance ** | 0.9643 | 0.9972 | 0.9994 | 0.9927 |
| Error *** | 0.0039 | 0.0003 | 0.0001 | 0.0011 | |
| Validation set | Performance ** | 0.9411 | 0.9531 | 0.9441 | 0.9748 |
| Error *** | 0.0111 | 0.0064 | 0.0097 | 0.0066 | |
* Number of neurons: input layer-hidden layer-output layer; ** Performance-regression coefficient; *** Error—root mean squared error (RMSE).