Parameter | Value |
Epochs | 2000 |
Population size | 15 |
Selection method | Tournament Κ = 3 (Size of tournament group) |
Recombination method | Random K = 5 (Crossover break size) |
Mutation method | Simple random (Add or subtract a random value) |
Value range, for variable learning rate | [0.3, 0.00001] |
Range of values for the probability of mutation | [200, 80] |
Range of values for the mutation value | [−1, 1] |
Neural network size | A hidden layer with 9 neurons A neuron in the output layer |
Activation function, for hidden layer neurons | Logistic sigmoid [18] |
Network Output Function | Ground-relu [19] (except for the Lorenz dataset [13] , where it has the linear) |