Analisis Patially Mapped Crossover Algoritma Genetika dalam Mengoptimalkan Pembelajaran Backpropagation Neural Network

Abstract

Backpropagation Neural Network a computational method that is reliable and capable of accurately recognizing patterns but old in the process of training. Behind the success of backpropagation neural network as one of the powerful computational methods, there is a weakness in this method is the length of time needed in training in order to get the best results. That requires an approach to optimize learning so that the method is used partially mapped crossover genetic algorithm that can optimize the learning process in the backpropagation neural network. Where the genetic algorithm used in determining the architecture and optimal weight initialization on the network propagation neural network. Resulting and optimal degree of accuracy in recognizing the pattern of alphanumeric characters.