Research of the Methods of Optimization of Artificial Neural Networks Training in the Tasks of Regression Analysis

Authors:
Publisher website
Google Scholar
Abstract
The paper presents the results of the research of the methods for optimizing the training of artificial neural networks. The mathematical model of the regression analysis of a random analytical dependence is given. The basic method for solving the problems of regression analysis using neural networks is considered. Applied software for the comparative study of the training parameters of standard neural networks in the process of regression analysis is presented. The most promising optimization criteria for existing training algorithms were selected. As the particular optimization model, the group of conjugate gradient methods is considered in more detail. Each of the methods was investigated for the possibility of improvement by the time criterion of optimality.

This publication has 1 reference indexed in Scilit: