Table 2. Summary of the interaction algorithm parameters.
Parameter | Setting |
Architecture | N-2-1, where N = total number of genes −1 |
Learning algorithm | Backpropagation |
Activation function | Sigmoid |
Epochs | 300 |
Threshold of mean squared error (MSE) | 0.01 |
Window of MSE | 100 |
Momentum | 0.5 |
Learning rate | 0.1 |
Pearson r cutoff | 0.7 |
Cross-validation | Monte Carlo cross-validation |
Random reshuffle | 50 times |
Ratio (%) for training: testing: validation | 60∶20∶20 |
Number of time the whole process repeatfor single variable | 10 |