Issues in designing automated minimal resource allocation neural networks

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

Artificial Neural Networks (ANNs) have a long record of generally promising results in hydrology. The earlier applications were mainly based on the back-propagation feed-forward method, which often used a lengthy trial-and-error method to determine the final network parameters. An attempt to overcome this shortcoming of the traditional applications is the Minimal Resource Allocation Network (MRAN). MRAN is online adaptive method which automatically configures the number of hidden nodes based on the input-output patterns presented to the network. Although MRAN demonstrated superior accuracy and more compact network, when compared with the traditional back-propagation method, some additional questions need to be addressed. While the number of hidden nodes is estimated automatically, other user-defined parameters are selected arbitrarily, and adjusted through simulations. This research addresses determining the user-defined parameters prior to the model run. The research also compares MRAN results from two applications, and discusses a pathway towards designing a fully automated MRAN.

Original languageEnglish (US)
Title of host publicationProceedings of the International Joint Conference on Neural Networks, IJCNN 2005
Pages2671-2673
Number of pages3
Volume5
DOIs
StatePublished - 2005
EventInternational Joint Conference on Neural Networks, IJCNN 2005 - Montreal, QC, Canada
Duration: Jul 31 2005Aug 4 2005

Other

OtherInternational Joint Conference on Neural Networks, IJCNN 2005
CountryCanada
CityMontreal, QC
Period7/31/058/4/05

ASJC Scopus subject areas

  • Software

Fingerprint Dive into the research topics of 'Issues in designing automated minimal resource allocation neural networks'. Together they form a unique fingerprint.

Cite this