A simple and effective method for removal of hidden units and weights

Research output: Contribution to journalArticle

40 Citations (Scopus)

Abstract

The objective of this paper is to present a simple and effective method for removal of both hidden units and weights. In this paper, we propose two methods, the 'Consuming energy' method and the 'Weights power' method, and compare them with the conventional method. According to our computer simulations using the mirror symmetry problem, the Weights power method has shown the best performance with respect to size reduction (removal of units and weights), generalization performance, and the amount of computation required. For example, the number of hidden units reduced to about 40% of the initial state, and the number of weights reduced to less than a fourth of the initial state. In addition, generalization performance was improved more than 10%.

Original languageEnglish
Pages (from-to)207-218
Number of pages12
JournalNeurocomputing
Volume6
Issue number2
DOIs
Publication statusPublished - 1994

Fingerprint

Weights and Measures
Mirrors
Computer simulation
Computer Simulation

ASJC Scopus subject areas

  • Artificial Intelligence
  • Cellular and Molecular Neuroscience

Cite this

A simple and effective method for removal of hidden units and weights. / Hagiwara, Masafumi.

In: Neurocomputing, Vol. 6, No. 2, 1994, p. 207-218.

Research output: Contribution to journalArticle

@article{f5ef2fa39bf74f77a67c988f30aab93b,
title = "A simple and effective method for removal of hidden units and weights",
abstract = "The objective of this paper is to present a simple and effective method for removal of both hidden units and weights. In this paper, we propose two methods, the 'Consuming energy' method and the 'Weights power' method, and compare them with the conventional method. According to our computer simulations using the mirror symmetry problem, the Weights power method has shown the best performance with respect to size reduction (removal of units and weights), generalization performance, and the amount of computation required. For example, the number of hidden units reduced to about 40{\%} of the initial state, and the number of weights reduced to less than a fourth of the initial state. In addition, generalization performance was improved more than 10{\%}.",
author = "Masafumi Hagiwara",
year = "1994",
doi = "10.1016/0925-2312(94)90055-8",
language = "English",
volume = "6",
pages = "207--218",
journal = "Neurocomputing",
issn = "0925-2312",
publisher = "Elsevier",
number = "2",

}

TY - JOUR

T1 - A simple and effective method for removal of hidden units and weights

AU - Hagiwara, Masafumi

PY - 1994

Y1 - 1994

N2 - The objective of this paper is to present a simple and effective method for removal of both hidden units and weights. In this paper, we propose two methods, the 'Consuming energy' method and the 'Weights power' method, and compare them with the conventional method. According to our computer simulations using the mirror symmetry problem, the Weights power method has shown the best performance with respect to size reduction (removal of units and weights), generalization performance, and the amount of computation required. For example, the number of hidden units reduced to about 40% of the initial state, and the number of weights reduced to less than a fourth of the initial state. In addition, generalization performance was improved more than 10%.

AB - The objective of this paper is to present a simple and effective method for removal of both hidden units and weights. In this paper, we propose two methods, the 'Consuming energy' method and the 'Weights power' method, and compare them with the conventional method. According to our computer simulations using the mirror symmetry problem, the Weights power method has shown the best performance with respect to size reduction (removal of units and weights), generalization performance, and the amount of computation required. For example, the number of hidden units reduced to about 40% of the initial state, and the number of weights reduced to less than a fourth of the initial state. In addition, generalization performance was improved more than 10%.

UR - http://www.scopus.com/inward/record.url?scp=0028413960&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=0028413960&partnerID=8YFLogxK

U2 - 10.1016/0925-2312(94)90055-8

DO - 10.1016/0925-2312(94)90055-8

M3 - Article

AN - SCOPUS:0028413960

VL - 6

SP - 207

EP - 218

JO - Neurocomputing

JF - Neurocomputing

SN - 0925-2312

IS - 2

ER -