Table 1.

Backpropagation training algorithms. MSE, mean squared error.

Algorithm	Function	Optimal Neuron Number	MSE	R 2
Resilient	trainrp	16	57.48	0.908
Fletcher–Reeves conjugate gradient	traincgf	10	3.84	0.989
Polak–Ribière–Polyak conjugate gradient	traincgp	10	4.78	0.986
Powell–Beale conjugate gradient	traincgb	10	3.88	0.988
Levenberg–Marquardt	trainlm	6	4.50	0.987
Scaled conjugate gradient	trainscg	10	6.79	0.981
BFGS quasi-Newton	trainbfg	18	7.83	0.980
One-step secant	trainoss	8	7.07	0.979