Table 5.
Performance summary for training algorithms of the ANN model (single layer and ten neurons).
| Algorithm | MSE | R2 |
|---|---|---|
| Levenberg-Marquardt | 1.96E-06 | 1.00000 |
| Bayesian Regularization | 2.40E-06 | 1.00000 |
| BFGS Quasi-Newton | 4.14E-04 | 0.99300 |
| Resilient Backpropagation | 7.25E-04 | 0.98771 |
| Scaled Conjugate Gradient | 7.60E-04 | 0.98787 |
| Conjugate Gradient with Powell/Beale Restarts | 4.31E-05 | 0.99927 |
| Fletcher-Powell Conjugate Gradient | 5.06E-04 | 0.99140 |
| Polak-Ribiére Conjugate Gradient | 6.50E-03 | 0.88285 |
| One Step Secant | 2.80E-03 | 0.96091 |
| Variable Learning Rate Gradient Descent | 8.09E-04 | 0.98661 |
| Gradient Descent with Momentum | 1.68E-02 | 0.65602 |
| Gradient Descent | 1.69E-02 | 0.65406 |