Research Article
GGA-MLP: A Greedy Genetic Algorithm to Optimize Weights and Biases in Multilayer Perceptron
Table 3
Controlling parameters of metaheuristic algorithms.
| Optimization algorithm | Parameter | Value |
| GGA-MLP, GA, ABC, WOA, MMPSO, MVO, GOA | Initial population size | 30 | # iterations | 200 |
| GA, ABC, WOA, MMPSO, MVO, GOA | Initial population generation | Random population initialization (synaptic weights and biases are initialized randomly in the range [−2, 2]) |
| GGA-MLP | Initial population generation | Greedy population initialization |
| GA, GGA-MLP | Probability of crossover | 0.8 | Probability of mutation | 0.05 | Elitism | 30% |
| ABC | Random number (ɸ) | [−1, 1] |
| WOA | Vector | Linearly decreasing from 2 to 0 | Random vector | [0, 1] | Constant (b) | 1 | Random number (p) | [0, 1] | Random number (l) | [−1, 1] |
| MMPSO | Acceleration coefficients | 1.48 | Inertia weights () | 0.729 | Number of swarms | 3 | Swarm size | 10 |
| MVO | Minimum wormhole existence probability | 0.2 | Minimum wormhole existence probability | 1 |
| GOA | | 1 | | 0.00001 | L | 1.5 | F | 0.5 |
|
|