Face recognition using eigen-faces and extension neural network

In this paper, the extension neural network is used as a classification tool in the application of face recognition. The eigen-faces method is firstly adopted in order to extract the coefficients associated with the most important eigen-faces. Next, the role of ENN comes as a classifier or pattern recognition technique. The performance of ENN is found to be superior over the traditional Multi-Layer Perceptron (MLP) in several aspects. The accuracy of the ENN is higher than MLP with less memory and processing requirements. Moreover, the structure of ENN is simple and fully determined compared with the MLP structure. In addition, the learning speed of ENN is higher than MLP. As a consequence of its simple structure, ENN is easier to be extended to be able to recognize new persons by just adding neurons in the output layer. Furthermore, in this paper, the effect of the learning rate, on both the stability and the speed of learning, is examined. We concluded that there is some optimum value for the learning rate which, in general, depends on the data set.