Table 7.
Best predictors for differentiation CHD patients infected with H. pylori from H. pylori negative healthy donors based on IR spectra and artificial neural networks.
No | Topology | Correct classifications percentage | Error function | Activation function | |||
---|---|---|---|---|---|---|---|
Hidden neurons | Output neurons | ||||||
Trening subset | Valiadation subset | ||||||
1 | MLP 4-2-2 | 93.98 | 93.58 | SOS | Logistic | Identity | |
2 | MLP 4-3-2 | 93.65 | 92.30 | SOS | Logistic | Identity | |
3 | MLP 4-2-2 | 93.55 | 92.30 | SOS | Logistic | Identity | |
4 | MLP 2-4-2 | 89.87 | 89.74 | Entropy | Tanh | Softmax | |
5 | MLP 2-5-2 | 89.87 | 89.74 | Entropy | Exponential | Softmax | |
6 | MLP 3-3-2 | 91.139 | 87.17 | SOS | Tanh | Tanh | |
7 | MLP 3-5-2 | 91.139 | 85.89 | Entropy | Tanh | Softmax | |
8 | MLP 3-2-2 | 89.873 | 85.89 | Entropy | Tanh | Softmax | |
9 | MLP 3-4-2 | 89.873 | 84.615 | Entropy | Tanh | Softmax |