Table 4.

Calculated root-mean-squared (RSM) errors for the seven multilayer perceptron (MLP) architectures compared in the study.

	MLP 12 × 3 × 3	MLP 12 × 4 × 3	MLP 12 × 8 × 3	MLP 12 × 2 × 2 × 3	MLP 1 × 4 × 3	MLP 7 × 2 × 2 × 3	MLP 4 × 8 × 7 × 3
Training set	0.063	0.133	0.015	0.036	0.001	0.045	0.170
Verification set	0.085	0.169	0.448	0.087	0.018	0.231	0.286
Test set	0.115	0.3678	0.443	0.481	0.091	0.112	0.352