Algorithm 2: Subroutine of CV

Input: dataset {A, Y} Results: fitness value λn Divide{A,Y}into k disjoint sub-datasets{A1,Y1},{A2,Y2},{A3,Y3},……,{Ak,Yk}; For p = 1:k   Set A as the input variable, Y as the output variable;   Train a new ANN with dataset {A,Y}={A,Y}-{Ak,Yk};   Validate the dataset {Ak,Yk} on the new ANN and the MSE denoted as MSEp; End ${\mathsf{\lambda }}_{\mathrm{n}}=\frac{1}{\mathrm{k}}{\displaystyle {{\displaystyle \sum }}_{\mathrm{p}=1}^{\mathrm{k}}}{\mathrm{MSE}}_{\mathrm{p}};$