Initialization of neural networks by means of decision trees

Irena Ivanova, Miroslav Kubat

Research output: Contribution to journalArticle

30 Citations (Scopus)

Abstract

The performance of neural networks is known to be sensitive to the initial weight setting and architecture (the number of hidden layers and neurons in these layers). This shortcoming can be alleviated if some approximation of the target concept in terms of a logical description is available. The paper reports a successful attempt to initialize neural networks using decision-tree generators. The TBNN (tree-based neural net) system compares very favourably with other learners in terms of classification accuracy for unseen data, and it is also computationally less demanding than the back propagation algorithm applied to a randomly initialized multilayer perceptron. The behavior of the system is first studied for specially designed artificial data. Then, its performance is demonstrated by a real-world application.

Original languageEnglish
Pages (from-to)333-344
Number of pages12
JournalKnowledge-Based Systems
Volume8
Issue number6
DOIs
StatePublished - Jan 1 1995
Externally publishedYes

Fingerprint

Decision trees
Neural networks
Backpropagation algorithms
Multilayer neural networks
Neurons
Decision tree
Neuron
Logic
Generator
Approximation
Back propagation
Neural nets

Keywords

  • decision-tree generators
  • neural networks

ASJC Scopus subject areas

  • Management Information Systems
  • Artificial Intelligence
  • Software
  • Information Systems and Management

Cite this

Initialization of neural networks by means of decision trees. / Ivanova, Irena; Kubat, Miroslav.

In: Knowledge-Based Systems, Vol. 8, No. 6, 01.01.1995, p. 333-344.

Research output: Contribution to journalArticle

@article{c08e99eb86af4db6bce6c82d2d3b8328,
title = "Initialization of neural networks by means of decision trees",
abstract = "The performance of neural networks is known to be sensitive to the initial weight setting and architecture (the number of hidden layers and neurons in these layers). This shortcoming can be alleviated if some approximation of the target concept in terms of a logical description is available. The paper reports a successful attempt to initialize neural networks using decision-tree generators. The TBNN (tree-based neural net) system compares very favourably with other learners in terms of classification accuracy for unseen data, and it is also computationally less demanding than the back propagation algorithm applied to a randomly initialized multilayer perceptron. The behavior of the system is first studied for specially designed artificial data. Then, its performance is demonstrated by a real-world application.",
keywords = "decision-tree generators, neural networks",
author = "Irena Ivanova and Miroslav Kubat",
year = "1995",
month = "1",
day = "1",
doi = "10.1016/0950-7051(96)81917-4",
language = "English",
volume = "8",
pages = "333--344",
journal = "Knowledge-Based Systems",
issn = "0950-7051",
publisher = "Elsevier",
number = "6",

}

TY - JOUR

T1 - Initialization of neural networks by means of decision trees

AU - Ivanova, Irena

AU - Kubat, Miroslav

PY - 1995/1/1

Y1 - 1995/1/1

N2 - The performance of neural networks is known to be sensitive to the initial weight setting and architecture (the number of hidden layers and neurons in these layers). This shortcoming can be alleviated if some approximation of the target concept in terms of a logical description is available. The paper reports a successful attempt to initialize neural networks using decision-tree generators. The TBNN (tree-based neural net) system compares very favourably with other learners in terms of classification accuracy for unseen data, and it is also computationally less demanding than the back propagation algorithm applied to a randomly initialized multilayer perceptron. The behavior of the system is first studied for specially designed artificial data. Then, its performance is demonstrated by a real-world application.

AB - The performance of neural networks is known to be sensitive to the initial weight setting and architecture (the number of hidden layers and neurons in these layers). This shortcoming can be alleviated if some approximation of the target concept in terms of a logical description is available. The paper reports a successful attempt to initialize neural networks using decision-tree generators. The TBNN (tree-based neural net) system compares very favourably with other learners in terms of classification accuracy for unseen data, and it is also computationally less demanding than the back propagation algorithm applied to a randomly initialized multilayer perceptron. The behavior of the system is first studied for specially designed artificial data. Then, its performance is demonstrated by a real-world application.

KW - decision-tree generators

KW - neural networks

UR - http://www.scopus.com/inward/record.url?scp=0029483035&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=0029483035&partnerID=8YFLogxK

U2 - 10.1016/0950-7051(96)81917-4

DO - 10.1016/0950-7051(96)81917-4

M3 - Article

AN - SCOPUS:0029483035

VL - 8

SP - 333

EP - 344

JO - Knowledge-Based Systems

JF - Knowledge-Based Systems

SN - 0950-7051

IS - 6

ER -