An improvement of a back propagation algorithm by the extended Kalman filter and demand forecasting by layered neural networks

Aritoshi Kimura, Ikuo Arizono, Hiroshi Ohta

Research output: Contribution to journalArticle

5 Citations (Scopus)

Abstract

Recently, Watanabe et al. proposed a back propagation algorithm via the extended Kalman filter, in which the learning rate was time-varying. In their algorithm the weights and biases are treated as independent variables. It is, however, natural that the weights and biases are not always independent, and generally have mutual correlation. In this paper, we improve the back propagation algorithm by considering that there is mutual correlation among the weights and bias directly connected to the unit. Through some numerical examples, our improved learning algorithm is compared with Watanabe et al.'s algorithm in learning ability. Furthermore, we consider demand forecasting as a kind of pattern recognition, and propose a demand forecasting method using layered neural networks with the improved learning algorithm. The effectiveness of this demand forecasting method is also discussed through some simulations.

Original languageEnglish
Pages (from-to)473-482
Number of pages10
JournalInternational Journal of Systems Science
Volume27
Issue number5
Publication statusPublished - 1996
Externally publishedYes

Fingerprint

Demand Forecasting
Backpropagation algorithms
Back-propagation Algorithm
Extended Kalman filters
Kalman Filter
Neural Networks
Neural networks
Learning algorithms
Learning Algorithm
Learning Rate
Pattern Recognition
Pattern recognition
Time-varying
Numerical Examples
Unit
Demand forecasting
Extended Kalman filter
Back propagation
Simulation
Learning algorithm

ASJC Scopus subject areas

  • Computational Theory and Mathematics
  • Management Science and Operations Research
  • Control and Systems Engineering
  • Theoretical Computer Science

Cite this

An improvement of a back propagation algorithm by the extended Kalman filter and demand forecasting by layered neural networks. / Kimura, Aritoshi; Arizono, Ikuo; Ohta, Hiroshi.

In: International Journal of Systems Science, Vol. 27, No. 5, 1996, p. 473-482.

Research output: Contribution to journalArticle

@article{692bc904971941a89288309a46e8c484,
title = "An improvement of a back propagation algorithm by the extended Kalman filter and demand forecasting by layered neural networks",
abstract = "Recently, Watanabe et al. proposed a back propagation algorithm via the extended Kalman filter, in which the learning rate was time-varying. In their algorithm the weights and biases are treated as independent variables. It is, however, natural that the weights and biases are not always independent, and generally have mutual correlation. In this paper, we improve the back propagation algorithm by considering that there is mutual correlation among the weights and bias directly connected to the unit. Through some numerical examples, our improved learning algorithm is compared with Watanabe et al.'s algorithm in learning ability. Furthermore, we consider demand forecasting as a kind of pattern recognition, and propose a demand forecasting method using layered neural networks with the improved learning algorithm. The effectiveness of this demand forecasting method is also discussed through some simulations.",
author = "Aritoshi Kimura and Ikuo Arizono and Hiroshi Ohta",
year = "1996",
language = "English",
volume = "27",
pages = "473--482",
journal = "International Journal of Systems Science",
issn = "0020-7721",
publisher = "Taylor and Francis Ltd.",
number = "5",

}

TY - JOUR

T1 - An improvement of a back propagation algorithm by the extended Kalman filter and demand forecasting by layered neural networks

AU - Kimura, Aritoshi

AU - Arizono, Ikuo

AU - Ohta, Hiroshi

PY - 1996

Y1 - 1996

N2 - Recently, Watanabe et al. proposed a back propagation algorithm via the extended Kalman filter, in which the learning rate was time-varying. In their algorithm the weights and biases are treated as independent variables. It is, however, natural that the weights and biases are not always independent, and generally have mutual correlation. In this paper, we improve the back propagation algorithm by considering that there is mutual correlation among the weights and bias directly connected to the unit. Through some numerical examples, our improved learning algorithm is compared with Watanabe et al.'s algorithm in learning ability. Furthermore, we consider demand forecasting as a kind of pattern recognition, and propose a demand forecasting method using layered neural networks with the improved learning algorithm. The effectiveness of this demand forecasting method is also discussed through some simulations.

AB - Recently, Watanabe et al. proposed a back propagation algorithm via the extended Kalman filter, in which the learning rate was time-varying. In their algorithm the weights and biases are treated as independent variables. It is, however, natural that the weights and biases are not always independent, and generally have mutual correlation. In this paper, we improve the back propagation algorithm by considering that there is mutual correlation among the weights and bias directly connected to the unit. Through some numerical examples, our improved learning algorithm is compared with Watanabe et al.'s algorithm in learning ability. Furthermore, we consider demand forecasting as a kind of pattern recognition, and propose a demand forecasting method using layered neural networks with the improved learning algorithm. The effectiveness of this demand forecasting method is also discussed through some simulations.

UR - http://www.scopus.com/inward/record.url?scp=0030141962&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=0030141962&partnerID=8YFLogxK

M3 - Article

AN - SCOPUS:0030141962

VL - 27

SP - 473

EP - 482

JO - International Journal of Systems Science

JF - International Journal of Systems Science

SN - 0020-7721

IS - 5

ER -