USING NEURAL NETWORKS PROGRAMMING ON JAVA  FOR SOLVING THE PROBLEM OF SIGNAL RECOGNITION

N. Matveeva

USING NEURAL NETWORKS PROGRAMMING ON JAVA FOR SOLVING THE PROBLEM OF SIGNAL RECOGNITION

Автор(и)

N. Matveeva

Ключові слова:

composite materials, neural networks, multilayer perceptron with back-propagation training, defect, function of activity.

Анотація

The results of the study of signal recognition using neural networks are presented. Scanning of composite materials is performed in the presence of additive noise. A multilayer perceptron with back-propagation error is implemented on Java in the environment NetBeans. Experiments were performed to analyze MSE values and accuracy. The confusion matrix, sensitivity and specificity were obtained and analyzed.