Hybrid Universal Learning Networks

Dazi Li, Takayuki Furuzuki, Junichi Murata, Kotaro Hirasawa

Research output: Contribution to journalArticle

1 Citation (Scopus)

Abstract

A variety of neuron models combine the neural inputs through their summation and sigmoidal functions. Such structure of neural networks leads to shortcomings such as a large number of neurons in hidden layers and huge training data required. We introduce a kind of multiplication neuron which multiplies their inputs instead of summing to overcome the above problems. A hybrid universal learning network constructed by the combination of multiplication units arid summation units is proposed and trained for several well known benchmark problems. Different combinations of the above two are tried. It is clarified that multiplication is an essential computational element in many cases and the combination of the multiplication units with summation units in different layers in the networks improved the performance of the network.

Original languageEnglish
Pages (from-to)552-559
Number of pages8
JournalIEEJ Transactions on Electronics, Information and Systems
Volume123
Issue number3
DOIs
Publication statusPublished - 2003
Externally publishedYes

Fingerprint

Neurons
Neural networks

Keywords

  • hybrid Universal Learning Networks
  • multiplication units
  • summation units

ASJC Scopus subject areas

  • Electrical and Electronic Engineering

Cite this

Hybrid Universal Learning Networks. / Li, Dazi; Furuzuki, Takayuki; Murata, Junichi; Hirasawa, Kotaro.

In: IEEJ Transactions on Electronics, Information and Systems, Vol. 123, No. 3, 2003, p. 552-559.

Research output: Contribution to journalArticle

Li, Dazi ; Furuzuki, Takayuki ; Murata, Junichi ; Hirasawa, Kotaro. / Hybrid Universal Learning Networks. In: IEEJ Transactions on Electronics, Information and Systems. 2003 ; Vol. 123, No. 3. pp. 552-559.
@article{532cb379ab35448f9fe60e26be7736d5,
title = "Hybrid Universal Learning Networks",
abstract = "A variety of neuron models combine the neural inputs through their summation and sigmoidal functions. Such structure of neural networks leads to shortcomings such as a large number of neurons in hidden layers and huge training data required. We introduce a kind of multiplication neuron which multiplies their inputs instead of summing to overcome the above problems. A hybrid universal learning network constructed by the combination of multiplication units arid summation units is proposed and trained for several well known benchmark problems. Different combinations of the above two are tried. It is clarified that multiplication is an essential computational element in many cases and the combination of the multiplication units with summation units in different layers in the networks improved the performance of the network.",
keywords = "hybrid Universal Learning Networks, multiplication units, summation units",
author = "Dazi Li and Takayuki Furuzuki and Junichi Murata and Kotaro Hirasawa",
year = "2003",
doi = "10.1541/ieejeiss.123.552",
language = "English",
volume = "123",
pages = "552--559",
journal = "IEEJ Transactions on Electronics, Information and Systems",
issn = "0385-4221",
publisher = "The Institute of Electrical Engineers of Japan",
number = "3",

}

TY - JOUR

T1 - Hybrid Universal Learning Networks

AU - Li, Dazi

AU - Furuzuki, Takayuki

AU - Murata, Junichi

AU - Hirasawa, Kotaro

PY - 2003

Y1 - 2003

N2 - A variety of neuron models combine the neural inputs through their summation and sigmoidal functions. Such structure of neural networks leads to shortcomings such as a large number of neurons in hidden layers and huge training data required. We introduce a kind of multiplication neuron which multiplies their inputs instead of summing to overcome the above problems. A hybrid universal learning network constructed by the combination of multiplication units arid summation units is proposed and trained for several well known benchmark problems. Different combinations of the above two are tried. It is clarified that multiplication is an essential computational element in many cases and the combination of the multiplication units with summation units in different layers in the networks improved the performance of the network.

AB - A variety of neuron models combine the neural inputs through their summation and sigmoidal functions. Such structure of neural networks leads to shortcomings such as a large number of neurons in hidden layers and huge training data required. We introduce a kind of multiplication neuron which multiplies their inputs instead of summing to overcome the above problems. A hybrid universal learning network constructed by the combination of multiplication units arid summation units is proposed and trained for several well known benchmark problems. Different combinations of the above two are tried. It is clarified that multiplication is an essential computational element in many cases and the combination of the multiplication units with summation units in different layers in the networks improved the performance of the network.

KW - hybrid Universal Learning Networks

KW - multiplication units

KW - summation units

UR - http://www.scopus.com/inward/record.url?scp=10844225644&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=10844225644&partnerID=8YFLogxK

U2 - 10.1541/ieejeiss.123.552

DO - 10.1541/ieejeiss.123.552

M3 - Article

AN - SCOPUS:10844225644

VL - 123

SP - 552

EP - 559

JO - IEEJ Transactions on Electronics, Information and Systems

JF - IEEJ Transactions on Electronics, Information and Systems

SN - 0385-4221

IS - 3

ER -