Home // International Journal On Advances in Networks and Services, volume 8, numbers 3 and 4, 2015 // View article

Analytic Method for Evaluation of the Weights of a Robust Large-Scale Multilayer Neural Network

Authors:
Mikael Fridenfalk

Keywords: analytic; big data; FNN; large-scale; least square method; multilayer; neural network; robust; sigmoid.

Abstract:
The multilayer feedforward neural network is presently one of the most popular computational methods in computer science. However, the current method for the evaluation of its weights is performed by a relatively slow iterative method known as backpropagation. According to previous research on a large-scale neural network with many hidden nodes, attempts to use an analytic method for the evaluation of the weights by the linear least square method showed to accelerate the evaluation process significantly. Nevertheless, the evaluated network showed in preliminary tests to fail in robustness compared to well-trained networks by backpropagation, thus resembling overtrained networks. This paper presents the design and verification of a new method that solves the robustness issues for such a neural network, along with MATLAB code for the verification of key experiments.

Pages: 139 to 148

Copyright: Copyright (c) to authors, 2015. Used with permission.

Publication date: December 30, 2015

Published in: journal

ISSN: 1942-2644