Universal Learning Networks with Branch Control

Kotaro Hirasawa, Takayuki Furuzuki, Qingyu Xiong, Junichi Murata

Research output: Contribution to journalArticle

Abstract

In this paper, Universal Learning Network with Branch Control (ULN with BC) is proposed, which consists of basic networks and branch control networks. The branch control network can be used to determine which branches of the basic network should be used. This determination depends on the inputs or the network flow of the basic network. Therefore, by using the ULN with BC, locally functions distributed networks can be realized depending on the values of the inputs of the network or the information of the network flow. The proposed network is applied to some function approximation problems. The simulation results show that the ULN with BC exhibits better performance than the conventional networks with comparable complexity.

Original languageEnglish
Pages (from-to)125-131
Number of pages7
JournalResearch Reports on Information Science and Electrical Engineering of Kyushu University
Volume4
Issue number2
Publication statusPublished - 1999 Sep
Externally publishedYes

ASJC Scopus subject areas

  • Hardware and Architecture
  • Engineering (miscellaneous)
  • Electrical and Electronic Engineering

Cite this

Universal Learning Networks with Branch Control. / Hirasawa, Kotaro; Furuzuki, Takayuki; Xiong, Qingyu; Murata, Junichi.

In: Research Reports on Information Science and Electrical Engineering of Kyushu University, Vol. 4, No. 2, 09.1999, p. 125-131.

Research output: Contribution to journalArticle

@article{1bd0cda2ddd9412b94f662be94edf7f2,
title = "Universal Learning Networks with Branch Control",
abstract = "In this paper, Universal Learning Network with Branch Control (ULN with BC) is proposed, which consists of basic networks and branch control networks. The branch control network can be used to determine which branches of the basic network should be used. This determination depends on the inputs or the network flow of the basic network. Therefore, by using the ULN with BC, locally functions distributed networks can be realized depending on the values of the inputs of the network or the information of the network flow. The proposed network is applied to some function approximation problems. The simulation results show that the ULN with BC exhibits better performance than the conventional networks with comparable complexity.",
author = "Kotaro Hirasawa and Takayuki Furuzuki and Qingyu Xiong and Junichi Murata",
year = "1999",
month = "9",
language = "English",
volume = "4",
pages = "125--131",
journal = "Research Reports on Information Science and Electrical Engineering of Kyushu University",
issn = "1342-3819",
publisher = "Kyushu University, Faculty of Science",
number = "2",

}

TY - JOUR

T1 - Universal Learning Networks with Branch Control

AU - Hirasawa, Kotaro

AU - Furuzuki, Takayuki

AU - Xiong, Qingyu

AU - Murata, Junichi

PY - 1999/9

Y1 - 1999/9

N2 - In this paper, Universal Learning Network with Branch Control (ULN with BC) is proposed, which consists of basic networks and branch control networks. The branch control network can be used to determine which branches of the basic network should be used. This determination depends on the inputs or the network flow of the basic network. Therefore, by using the ULN with BC, locally functions distributed networks can be realized depending on the values of the inputs of the network or the information of the network flow. The proposed network is applied to some function approximation problems. The simulation results show that the ULN with BC exhibits better performance than the conventional networks with comparable complexity.

AB - In this paper, Universal Learning Network with Branch Control (ULN with BC) is proposed, which consists of basic networks and branch control networks. The branch control network can be used to determine which branches of the basic network should be used. This determination depends on the inputs or the network flow of the basic network. Therefore, by using the ULN with BC, locally functions distributed networks can be realized depending on the values of the inputs of the network or the information of the network flow. The proposed network is applied to some function approximation problems. The simulation results show that the ULN with BC exhibits better performance than the conventional networks with comparable complexity.

UR - http://www.scopus.com/inward/record.url?scp=0033355706&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=0033355706&partnerID=8YFLogxK

M3 - Article

AN - SCOPUS:0033355706

VL - 4

SP - 125

EP - 131

JO - Research Reports on Information Science and Electrical Engineering of Kyushu University

JF - Research Reports on Information Science and Electrical Engineering of Kyushu University

SN - 1342-3819

IS - 2

ER -