Identification of time-series signals using a dynamic neural network with GA-based training

Kunihiko Nakazono, Kouhei Ohnishi, Hiroshi Kinjo

Research output: Contribution to journalArticle

Abstract

We propose a dynamic neural network (DNN) that realizes a dynamic property and has a network structure with the properties of inertia, viscosity, and stiffness without time-delayed input elements, and a training algorithm based on a genetic algorithm (GA). In a previous study, we proposed a modified training algorithm for the DNN based on the error back-propagation method. However, in the previous method it was necessary to determine the values of the DNN property parameters by trial and error. In the newly proposed DNN, the GA is designed to train not only the connecting weights but also the property parameters of the DNN. Simulation results show that the DNN trained by the GA obtains good performance for time-series patterns generated from an unknown system, and provides a higher performance than the conventional neural network.

Original languageEnglish
Pages (from-to)102-105
Number of pages4
JournalArtificial Life and Robotics
Volume10
Issue number2
DOIs
Publication statusPublished - 2006 Nov

Fingerprint

Time series
Genetic algorithms
Neural networks
Viscosity
Backpropagation
Weights and Measures
Stiffness

Keywords

  • Dynamic neural network
  • Genetic algorithm
  • Identification
  • Property parameter

ASJC Scopus subject areas

  • Artificial Intelligence

Cite this

Identification of time-series signals using a dynamic neural network with GA-based training. / Nakazono, Kunihiko; Ohnishi, Kouhei; Kinjo, Hiroshi.

In: Artificial Life and Robotics, Vol. 10, No. 2, 11.2006, p. 102-105.

Research output: Contribution to journalArticle

Nakazono, Kunihiko ; Ohnishi, Kouhei ; Kinjo, Hiroshi. / Identification of time-series signals using a dynamic neural network with GA-based training. In: Artificial Life and Robotics. 2006 ; Vol. 10, No. 2. pp. 102-105.
@article{2c73e8fe8188461f8c2b30f6bf54494e,
title = "Identification of time-series signals using a dynamic neural network with GA-based training",
abstract = "We propose a dynamic neural network (DNN) that realizes a dynamic property and has a network structure with the properties of inertia, viscosity, and stiffness without time-delayed input elements, and a training algorithm based on a genetic algorithm (GA). In a previous study, we proposed a modified training algorithm for the DNN based on the error back-propagation method. However, in the previous method it was necessary to determine the values of the DNN property parameters by trial and error. In the newly proposed DNN, the GA is designed to train not only the connecting weights but also the property parameters of the DNN. Simulation results show that the DNN trained by the GA obtains good performance for time-series patterns generated from an unknown system, and provides a higher performance than the conventional neural network.",
keywords = "Dynamic neural network, Genetic algorithm, Identification, Property parameter",
author = "Kunihiko Nakazono and Kouhei Ohnishi and Hiroshi Kinjo",
year = "2006",
month = "11",
doi = "10.1007/s10015-005-0356-6",
language = "English",
volume = "10",
pages = "102--105",
journal = "Artificial Life and Robotics",
issn = "1433-5298",
publisher = "Springer Japan",
number = "2",

}

TY - JOUR

T1 - Identification of time-series signals using a dynamic neural network with GA-based training

AU - Nakazono, Kunihiko

AU - Ohnishi, Kouhei

AU - Kinjo, Hiroshi

PY - 2006/11

Y1 - 2006/11

N2 - We propose a dynamic neural network (DNN) that realizes a dynamic property and has a network structure with the properties of inertia, viscosity, and stiffness without time-delayed input elements, and a training algorithm based on a genetic algorithm (GA). In a previous study, we proposed a modified training algorithm for the DNN based on the error back-propagation method. However, in the previous method it was necessary to determine the values of the DNN property parameters by trial and error. In the newly proposed DNN, the GA is designed to train not only the connecting weights but also the property parameters of the DNN. Simulation results show that the DNN trained by the GA obtains good performance for time-series patterns generated from an unknown system, and provides a higher performance than the conventional neural network.

AB - We propose a dynamic neural network (DNN) that realizes a dynamic property and has a network structure with the properties of inertia, viscosity, and stiffness without time-delayed input elements, and a training algorithm based on a genetic algorithm (GA). In a previous study, we proposed a modified training algorithm for the DNN based on the error back-propagation method. However, in the previous method it was necessary to determine the values of the DNN property parameters by trial and error. In the newly proposed DNN, the GA is designed to train not only the connecting weights but also the property parameters of the DNN. Simulation results show that the DNN trained by the GA obtains good performance for time-series patterns generated from an unknown system, and provides a higher performance than the conventional neural network.

KW - Dynamic neural network

KW - Genetic algorithm

KW - Identification

KW - Property parameter

UR - http://www.scopus.com/inward/record.url?scp=33750698274&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=33750698274&partnerID=8YFLogxK

U2 - 10.1007/s10015-005-0356-6

DO - 10.1007/s10015-005-0356-6

M3 - Article

VL - 10

SP - 102

EP - 105

JO - Artificial Life and Robotics

JF - Artificial Life and Robotics

SN - 1433-5298

IS - 2

ER -