Training Artificial Neural Networks Using TaguchiMethods

Authors:
Chris Macleod;Geva Dror;Grant Maxwell
Affiliations:
The Robert Gordon University, Aberdeen, Scotland;The Robert Gordon University, Aberdeen, Scotland;The Robert Gordon University, Aberdeen, Scotland
Venue:
Artificial Intelligence Review
Year:
1999

Citing 1
Cited 1

Neural computing: theory and practice

Neural computing: theory and practice

Training neural networks using taguchi methods: overcoming interaction problems

ICANN'05 Proceedings of the 15th international conference on Artificial neural networks: formal models and their applications - Volume Part II

Quantified Score

Hi-index	0.00

Visualization

Abstract

This paper shows how the processoptimization methods known as Taguchi methods may beapplied to the training of Artificial NeuralNetworks. A comparison is made between the efficiencyof training using Taguchi methods and the efficiencyof conventional training methods; attention is drawnto the advantages of Taguchi methods. Further, it isshown that Taguchi methods offer potential benefits inevaluating network behaviour such as the ability toexamine interaction of weights and neurons within anetwork.