PROBABILISTIC NEURAL NETWORKS IN THE PROBLEMS OF TIME SERIES IDENTIFICATION

Authors

  • Roman Kvetniy Vinnytsia National Technical University
  • Vladyslav Kabatciy Vinnytsia National Technical University
  • Olga Chumachenko Vinnytsia National Technical University

Keywords:

probabilistic neural networks (PNN), identification, patterns identification, analysis of time series

Abstract

The given paper considers the possibility of time series identification on the basis of probabilistic neural networks and their modified versions. The influence of kernel function width on adequate restoration of density and classification quality is investigated. Modified versions of probabilistic neural networks and peculiarities of their application are considered. Advantages and disadvantages of probabilistic neural networks are shown.

Author Biographies

Roman Kvetniy, Vinnytsia National Technical University

Head of the Department of Automation and Information – Computing Engineering

Vladyslav Kabatciy, Vinnytsia National Technical University

Assistant Professor, Department of Automation and Information – Computing Engineering

Olga Chumachenko, Vinnytsia National Technical University

Student of the Institute of Post- Graduate studies

Downloads

Abstract views: 138

Published

2015-11-30

How to Cite

[1]
R. Kvetniy, V. Kabatciy, and O. Chumachenko, “PROBABILISTIC NEURAL NETWORKS IN THE PROBLEMS OF TIME SERIES IDENTIFICATION”, Works of VNTU, no. 3, Nov. 2015.

Issue

No. 3 (2010): 2010№3

Section

Information Technologies and Computer Engineering

Metrics

Downloads

Download data is not yet available.