Research Article
Modeling and Optimization of Electrodeposition Process for Copper Nanoparticle Synthesis Using ANN and Nature-Inspired Algorithms
Table 5
Performance of NN with different topologies and transfer functions.
| | Performance measures/training algorithm | Trainlm with linear output neurons (fitnet) | Trainbr with linear output neurons (fitnet) | Trainscg with linear output neurons (fitnet) | Trainlm with softmax output neurons (patternnet) | Trainbr with linear output neurons (patternnet) | Trainscg with linear output neurons (patternnet) |
| | Mean square error (MSE) | 3.73952e–1 | 9.08727e–1 | 5.24238e–1 | 2.5244e–30 | 6.25453e–19 | 1.21640e–1 | | Root mean square error (RMSE) | 0.611516 | 0.953272 | 0.724043 | 1.49E–15 | 7.91E–10 | 0.348769 | | Mean absolute error (MAE) | 0.11798 | 0.12765 | 0.13756 | 0.0092e–3 | 0.00256e–5 | 0.14874 | | R-square | 0.92086 | 0.91046 | 0.91038 | 0.99994 | 0.98984 | 0.87564 | | Variance accounted for (VAF) | 0.9215 | 0.9120 | 0.9132 | 0.9815 | 0.9752 | 0.8256 |
|
|
