Title of article
COMPARING EVOLUTIONARY ALGORITHMS ON TUNING THE PARAMETERS OF FUZZY WAVELET NEURAL NETWORK
Author/Authors
ARAB KHEDRI, P. shahid bahonar university of kerman - Department of Computer Engineering, كرمان, ايران , EFTEKHARI, M. shahid bahonar university of kerman - Department of Computer Engineering, كرمان, ايران , MAAZALLAHI, R. shahid bahonar university of kerman - Department of Computer Engineering, كرمان, ايران
From page
193
To page
198
Abstract
In recent years Fuzzy Wavelet Neural Networks (FWNNs) have been used in manyareas. Function approximation is an important application of FWNNs. One of the main problemsin effective usage of FWNN is tuning of its parameters. In this paper several different evolutionaryalgorithms including Genetic Algorithm (GA), Gravitational Search Algorithm (GSA),Evolutionary Strategy (ES), Fast Evolutionary Strategy (FES) and variants of DifferentialEvolutionary algorithms (DE) are used for adjusting these parameters on five test functions. Theobtained results are compared based on some measures by using multiple non-parametricstatistical tests. The comparison reveals the superiority of some variants of DE in terms ofconvergence behavior and the ability of function approximation.
Keywords
Fuzzy wavelet neural networks , function approximation , evolutionary algorithms , nonparametricstatistical test
Journal title
Iranian Journal of Science and Technology :Transactions of Electrical Engineering
Journal title
Iranian Journal of Science and Technology :Transactions of Electrical Engineering
Record number
2596371
Link To Document