Table 2 The parameters of the selected ANN training algorithms
Training algorithm | MatLab function | Maximum number of epochs to train | Other parameters |
|---|---|---|---|
LM | Trainlm | 1000 | Mu: 0.001 |
BR | Trainbr | 1000 | Mu: 0.005 |
BFGS | Trainbfg | 1000 | Name of the line search routine to use = srchcha Initial step size = 0.01 |
OSS | Trainoss | 1000 | Name of the line search routine to use = srchcha Initial step size = 0.01 |
SCG | Trainscg | 1000 | Â |
CGB | Traincgb | 1000 | Name of the line search routine to use = srchcha Initial step size = 0.01 |
CGF | Traincgf | 1000 | Name of the line search routine to use = srchcha Initial step size = 0.01 |
CGP | Traincgp | 1000 | Name of the line search routine to use = srchcha Initial step size = 0.01 |
GDX | Traingdx | 1000 | Learning rate = 0.01 Momentum constant = 0.9 |
GDM | Traingdm | 1000 | Learning rate = 0.01 Momentum constant = 0.9 |
GD | Traingd | 1000 | Learning rate = 0.01 |
RP | Trainrp | 1000 | Learning rate = 0.01 |