Table 4.
Discrimination of serum samples from H. pylori seropositive vs seronegative individuals based on IR spectra and artificial neural networks.
| No | Topology | Correct classification percentage | Error function | Activation function | ||
|---|---|---|---|---|---|---|
| Hidden neurons | Output neurons | |||||
| Trening subset | Validation subset | |||||
| 1 | MLP 5-5-2 | 98.73 | 96.15 | Entropy | Tanh | Softmax |
| 2 | MLP 5-4-2 | 94.94 | 89.74 | Entropy | Tanh | Softmax |
| 3 | MLP 4-4-2 | 94.94 | 89.74 | Entropy | Tanh | Softmax |
| 4 | MLP 4-5-2 | 96.20 | 89.74 | Entropy | Exponential | Softmax |
| 5 | MLP 5-3-2 | 94.94 | 88.46 | Entropy | Tanh | Softmax |
| 6 | MLP 5-2-2 | 97.47 | 88.46 | Entropy | Tanh | Softmax |
| 7 | MLP 4-3-2 | 93.67 | 88.46 | Entropy | Tanh | Softmax |
| 8 | MLP 3-5-2 | 94.94 | 88.46 | Entropy | Logistic | Softmax |
| 9 | MLP 3-4-2 | 94.94 | 88.46 | SOS | Logistic | Identity |
| 10 | MLP 2-4-2 | 93.67 | 85.90 | SOS | Logistic | Identity |