Adaptable multiple neural networks using evolutionary computation

Sunghwan Sohn, Cihan H. Dagli

Research output: Contribution to journalConference article

2 Citations (Scopus)

Abstract

The architecture of an artificial neural network has a significant influence on its performance. For a given problem, the proper architecture is found by trial and error. This approach is time consuming and may not always produce the optimal network. In this reason, we can apply the evolutionary computation such as genetic algorithm to implement the automation of network's structure as well as the biological inspiration in neural networks to successfully adapt varying input environment. Moreover, we examine the performance of combining multiple evolving networks that are more likely to model the neurophysiology of the human brain. In the combining module, all individual networks or selected individual networks in the population are used. Also, the dynamic data set is used to provide the networks with diversity and generalization. In this model, each evolving individual network is designed to have a specific feature set and neuron connection links for given data. Then, the results are combined through the combining module to improve the generalization performance of the single network. The Iris and Austrian credit data are used in the experiment.

Original languageEnglish (US)
Pages (from-to)141-149
Number of pages9
JournalProceedings of SPIE - The International Society for Optical Engineering
Volume4739
DOIs
StatePublished - Jan 1 2002
Externally publishedYes
EventApplications and Science Computational Intelligence V - Orlando, FL, United States
Duration: Apr 2 2002Apr 3 2002

Fingerprint

Evolutionary Computation
Evolutionary algorithms
Neurophysiology
Neural Networks
Neural networks
Neurons
Brain
Automation
Genetic algorithms
Experiments
Module
Trial and error
Iris
neurophysiology
modules
Network Structure
inspiration
Artificial Neural Network
Neuron
Likely

Keywords

  • Combining module
  • Evolutionary computation
  • Generalization
  • Genetic algorithm
  • Multiple neural networks
  • Neural networks

ASJC Scopus subject areas

  • Electronic, Optical and Magnetic Materials
  • Condensed Matter Physics
  • Computer Science Applications
  • Applied Mathematics
  • Electrical and Electronic Engineering

Cite this

Adaptable multiple neural networks using evolutionary computation. / Sohn, Sunghwan; Dagli, Cihan H.

In: Proceedings of SPIE - The International Society for Optical Engineering, Vol. 4739, 01.01.2002, p. 141-149.

Research output: Contribution to journalConference article

@article{89b3923f889845f9b0913b07b4591551,
title = "Adaptable multiple neural networks using evolutionary computation",
abstract = "The architecture of an artificial neural network has a significant influence on its performance. For a given problem, the proper architecture is found by trial and error. This approach is time consuming and may not always produce the optimal network. In this reason, we can apply the evolutionary computation such as genetic algorithm to implement the automation of network's structure as well as the biological inspiration in neural networks to successfully adapt varying input environment. Moreover, we examine the performance of combining multiple evolving networks that are more likely to model the neurophysiology of the human brain. In the combining module, all individual networks or selected individual networks in the population are used. Also, the dynamic data set is used to provide the networks with diversity and generalization. In this model, each evolving individual network is designed to have a specific feature set and neuron connection links for given data. Then, the results are combined through the combining module to improve the generalization performance of the single network. The Iris and Austrian credit data are used in the experiment.",
keywords = "Combining module, Evolutionary computation, Generalization, Genetic algorithm, Multiple neural networks, Neural networks",
author = "Sunghwan Sohn and Dagli, {Cihan H.}",
year = "2002",
month = "1",
day = "1",
doi = "10.1117/12.458706",
language = "English (US)",
volume = "4739",
pages = "141--149",
journal = "Proceedings of SPIE - The International Society for Optical Engineering",
issn = "0277-786X",
publisher = "SPIE",

}

TY - JOUR

T1 - Adaptable multiple neural networks using evolutionary computation

AU - Sohn, Sunghwan

AU - Dagli, Cihan H.

PY - 2002/1/1

Y1 - 2002/1/1

N2 - The architecture of an artificial neural network has a significant influence on its performance. For a given problem, the proper architecture is found by trial and error. This approach is time consuming and may not always produce the optimal network. In this reason, we can apply the evolutionary computation such as genetic algorithm to implement the automation of network's structure as well as the biological inspiration in neural networks to successfully adapt varying input environment. Moreover, we examine the performance of combining multiple evolving networks that are more likely to model the neurophysiology of the human brain. In the combining module, all individual networks or selected individual networks in the population are used. Also, the dynamic data set is used to provide the networks with diversity and generalization. In this model, each evolving individual network is designed to have a specific feature set and neuron connection links for given data. Then, the results are combined through the combining module to improve the generalization performance of the single network. The Iris and Austrian credit data are used in the experiment.

AB - The architecture of an artificial neural network has a significant influence on its performance. For a given problem, the proper architecture is found by trial and error. This approach is time consuming and may not always produce the optimal network. In this reason, we can apply the evolutionary computation such as genetic algorithm to implement the automation of network's structure as well as the biological inspiration in neural networks to successfully adapt varying input environment. Moreover, we examine the performance of combining multiple evolving networks that are more likely to model the neurophysiology of the human brain. In the combining module, all individual networks or selected individual networks in the population are used. Also, the dynamic data set is used to provide the networks with diversity and generalization. In this model, each evolving individual network is designed to have a specific feature set and neuron connection links for given data. Then, the results are combined through the combining module to improve the generalization performance of the single network. The Iris and Austrian credit data are used in the experiment.

KW - Combining module

KW - Evolutionary computation

KW - Generalization

KW - Genetic algorithm

KW - Multiple neural networks

KW - Neural networks

UR - http://www.scopus.com/inward/record.url?scp=0036404515&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=0036404515&partnerID=8YFLogxK

U2 - 10.1117/12.458706

DO - 10.1117/12.458706

M3 - Conference article

AN - SCOPUS:0036404515

VL - 4739

SP - 141

EP - 149

JO - Proceedings of SPIE - The International Society for Optical Engineering

JF - Proceedings of SPIE - The International Society for Optical Engineering

SN - 0277-786X

ER -