OPTIMASI PARAMETER ARTIFICIAL NEURAL NETWORK DENGAN MENGGUNAKAN ALGORITMA GENETIKA UNTUK MEMPREDIKSI NILAI TUKAR RUPIAH

Khairul Fahmi
Stefanus Santosa
Ahmad Zainul Fanani

Jurnal Teknologi Informasi Cyberku
Vol 11 No 2 (2015): Jurnal Teknologi Informasi CyberKU Vol.11 no 2

To predict foreign exchange rate is not easy, accurate prediction is necessary for investor to reduce higrisk about exchange rate volatility. In predicting foreign exchange rate is used Artificial Neural NetworkBackpropagation as a model that applied. There are several parameters to implement Artificial NeuralNetwork that must be determined as training cyclel, learning rate, and momentum, the problem is the lackof standard guidelines in determining the parameters that will be used, therefore in this method used theexperimental method. So that we need a method that can resolve the problem, then that the parametersobtained become more optimal. Solutions that can be applied is to apply the genetic algorithm (GA) onArtificial Neural Networks, in order to optimize the value of training cycle, learning rate and momentumparameters. The results are the application of optimization techniques with Genetic Algorithm canfacilitate the search for optimal parameter values and reduce error (RMSE) or increase the value of theaccuracy of the Artificial Neural Network algorithm, thus the model obtained can be used by investor topredict foreign exchange rate.