Comparative Analysis of Artificial Neural Network Training Algorithms

This study compares training algorithms for artificial neural networks such as genetic, adaptive and hybrid. The “Fisher's Irises” were used as a data for the classification problem and KNIME Analytics Platform was chosen as the experimental environment. Given results comparison analysis of three methods training for neural networks and their parameters are presented. The choice of the optimal architecture was carried out on the basis of the classification algorithm, based on the graph of classification accuracy and error.