Simultaneous Adaptation Of Multiple Genetic Algorithm Parameters Using Fuzzy Logic Controllers
Loading...
Date
2010
Authors
Nasser Ghallab, Abdullatif Saleh
Journal Title
Journal ISSN
Volume Title
Publisher
Universiti Sains Malaysia
Abstract
This study aims at designing an online adaptive method to control multiple parameters of the
Genetic Algorithm. The efficiency of Genetic Algorithm requires maintaining an appropriate
balance between exploration and exploitation, which in turn greatly depends on the settings of
several parameters. The parameters are not independent and have complex interactions with each
other during a given run. Ignoring the interaction between the adapted parameters or adapting
one single parameter may have negative impact on the other related parameters, resulting in poor
performance. However, most of the available alternatives cannot solve this problem effectively.
Fuzzy Adaptive Genetic Algorithm techniques have been used recently for parameter control,
but still suffer from some defects. This work contributes towards the design of a robust Fuzzy
Adaptive Genetic Algorithm. It presents a new Fuzzy Adaptive Genetic Algorithm method based
on three fuzzy logic controllers. This method controls multiple parameters simultaneously with
online consideration for their interdependencies during a run. It adapts three main strategic
parameters of the Genetic Algorithm, namely, population size, mutation, and crossover rates.
The validity of the proposed work is illustrated with a series of experiments on a set of standard
test problems often used to evaluate Genetic Algorithm techniques. The empirical results
indicated that the proposed method maintains good adaptive values of parameters with better
performance.
Description
Keywords
Designing an online adaptive method to control , multiple parameters of the Genetic Algorithm