Initialization of the Nguyen-widrow and Kohonen Algorithm on the Backpropagation Method in the Classifying Process of Temperature Data in Medan

Abstract

In this paper, we propose the initialization of the Nguyen-widrow and Kohonen algorithm on the Backpropagation Neural Network in the classification of temperature in Medan. Initialization of Nguyen-widrow and Kohonen weight in Backpropagation could accelerate the training process of temperature data compared to Backpropagation with random weight. The experiment reaches target error 0.007 at 30 epoch. The result of testing show that the initialization of Nguyen-widrow and Kohonen weight in Backpropagation could recognize the test data reaches 96.52% accuracy.