Precision and dynamics of two mutationally-augmented g3-pcx evolutionary optimization algorithms

Teo, Jason Tze Wi and Mohd Hanafi Ahmad Hijazi (2008) Precision and dynamics of two mutationally-augmented g3-pcx evolutionary optimization algorithms. In: 4th IASTED International Conference on Advances in Computer Science and Technology (ACST 2008), 2 - 4 April 2008, Langkawi, Malaysia.

Full text not available from this repository.

Abstract

This study investigates the augmentation of the crossoveronly G3-PCX evolutionary optimization algorithm with mutational diversity for solving multimodal global optimization problems. The objective is to empirically test and compare the optimization precision and dynamics of the two mutationally-augmented G3-PCX algorithms against the original G3-PCX algorithm based on a set of well-known test functions. Empirical tests on five benchmark continuous multimodal test functions have shown highly competitive performance in the augmented G3 algorithms which are called G3MD (G3 with Mutational Diversity) and G3SAM (G3 with Self- Adaptive Mutation) respectively where the proposed algorithms outperformed the standard G3 algorithm in terms of solution quality on two of the test problems. Analysis of the optimization dynamics reveals that the improved performance was due to the longer maintenance of genetic diversity through the mutational augmentation. Hence it is shown that the simple addition of Gaussian mutation can greatly enhance the optimization precision of RCGAs for solving large scale problems with highly deceptive fitness landscapes.

Item Type: Conference or Workshop Item (UNSPECIFIED)
Keyword: Evolutionary computation, G3- PCX, Global optimization, Mutational diversity
Subjects: Q Science > QA Mathematics > QA1-939 Mathematics > QA150-272.5 Algebra
Department: SCHOOL > School of Engineering and Information Technology
Depositing User: ADMIN ADMIN
Date Deposited: 25 Mar 2011 13:02
Last Modified: 30 Dec 2014 14:58
URI: https://eprints.ums.edu.my/id/eprint/1608

Actions (login required)

View Item View Item