Face Recognition Using MLP and RBF Neural Network with Gabor and Discrete Wavelet Transform Characterization: A Comparative Study

Table 2

The optimal neural network topology.

Parameter	Value	Fixed/variable
Dataset	Feature vectors (DWT or Gabor)	Fixed
DWT coefficients	Haar, db2, db4, bior2.2, Gabor	Variable
Training algorithm	Gradient descent	Fixed
Transfer function	Both layers use log-sigmoid	Fixed
Number of neurons in hidden layer	10, 20, 30, 40, …, 90, 100	Variable
Minimum training MSE	0.001	Fixed
Maximum epochs	3000 to 6000	Variable