Optimization of the Backpropagation Method with Nguyen-widrow in Face Image Classification

Abstract

In this resarch, the Backpropagation method can be optimized using the Nguyen-Widrow algorithm which is related to the initialization of weights and bias. With the optimization of the Backpropagation method using the Nguyen-Widrow algorithm, facial images can be recognized more quickly and accurately. In the trial process carried out on hidden layer 6 neurons with a target error of 0.01, by default the Backpropagation method produces an accuracy rate of 96%, whereas if the Backpropagation method is optimized it will produce a higher accuracy rate of 98%. Similarly, if it is done on hidden layers 7, 8, 9 and 10 neurons. From the results of this research, these algorithms and methods can be used to improve information technology and play a role in facilitating all aspects of human activities. With the speed and high level of accuracy in taking pictures of faces, it can be used as a policy or provision that is applied to traffic cameras, cameras at train stations to airports and surveillance cameras placed in other strategic locations and can help in the social life and human security.