Computational Technology Resources - CCP

Keywords: CardiffGA, non-disruptive genetic algorithms.

Summary

Genetic Algorithms (GAs) are composed of genetic operators on coded representations of optimisation/search variables, and some fitness biased selection procedure. These are all encased within an iterative framework of "generations" which evolves an initial randomly generated set of solutions towards better and better solutions. Although much work has been carried out on Genetic Algorithms, and many variations now exist, the overall iterative framework of "generations" used by the GA has remained stable. This typical GA framework is intrinsically very disruptive since a large fraction of the entire population (if not all of it) is regenerated in each iterative step (i.e. a "generation"). There is also very little interaction of any sort across the generation gap.

A new framework called CardiffGA (CGA) is now proposed and evaluated. This new framework dispenses with the concept of "generations" and instead iterates with "timesteps". The impetus is to better mimic natural genetics, and create an environment more conducive to refined competition, where only a small fraction of the population is refreshed at each time step. The potential advantage is that such an approach is more suitable for "difficult" searches. Along with the new CGA framework, concepts such as "life span", "age", "natural deletion" and "accidental deletion" are also introduced. Details of the CGA are provided in the paper.

The newly proposed CGA has been tested alongside an ordinary GA for a simple and a difficult problem and the CGA has been found to perform better on a number of counts. In both test problems, but more so with the difficult problem, CardiffGA had a greater success rate in reaching the known optimum solution within a required accuracy than an ordinary GA. The CGA was also typically more efficient with its use of genetic operations and evolved the population with fewer such operations. This was offset to a certain extent by additional computations not required in the ordinary GA, but nonetheless, the overall computational time for CGA was still significantly less.

purchase the full-text of this paper (price £20)

go to the previous paper
go to the next paper
return to the table of contents
return to the book description
purchase this book (price £95 +P&P)

	Computational & Technology Resources an online resource for computational, engineering & technology publications
	not logged in - login
Front Page Browse CCP CSETS CTR IJRT Other Authors Search Purchase Guide FAQ Contact us	Civil-Comp Proceedings ISSN 1759-3433 CCP: 89 PROCEEDINGS OF THE SIXTH INTERNATIONAL CONFERENCE ON ENGINEERING COMPUTATIONAL TECHNOLOGY Edited by: M. Papadrakakis and B.H.V. Topping Paper 77 CardiffGA: A New Genetic Algorithm Framework H. Chen, J.C. Miles and A.S.K. Kwan Cardiff School of Engineering, Cardiff University, United Kingdom doi:10.4203/ccp.89.77 purchase the full-text of this paper Full Bibliographic Reference for this paper H. Chen, J.C. Miles, A.S.K. Kwan, "CardiffGA: A New Genetic Algorithm Framework", in M. Papadrakakis, B.H.V. Topping, (Editors), "Proceedings of the Sixth International Conference on Engineering Computational Technology", Civil-Comp Press, Stirlingshire, UK, Paper 77, 2008. doi:10.4203/ccp.89.77 Keywords: CardiffGA, non-disruptive genetic algorithms. Summary Genetic Algorithms (GAs) are composed of genetic operators on coded representations of optimisation/search variables, and some fitness biased selection procedure. These are all encased within an iterative framework of "generations" which evolves an initial randomly generated set of solutions towards better and better solutions. Although much work has been carried out on Genetic Algorithms, and many variations now exist, the overall iterative framework of "generations" used by the GA has remained stable. This typical GA framework is intrinsically very disruptive since a large fraction of the entire population (if not all of it) is regenerated in each iterative step (i.e. a "generation"). There is also very little interaction of any sort across the generation gap. A new framework called CardiffGA (CGA) is now proposed and evaluated. This new framework dispenses with the concept of "generations" and instead iterates with "timesteps". The impetus is to better mimic natural genetics, and create an environment more conducive to refined competition, where only a small fraction of the population is refreshed at each time step. The potential advantage is that such an approach is more suitable for "difficult" searches. Along with the new CGA framework, concepts such as "life span", "age", "natural deletion" and "accidental deletion" are also introduced. Details of the CGA are provided in the paper. The newly proposed CGA has been tested alongside an ordinary GA for a simple and a difficult problem and the CGA has been found to perform better on a number of counts. In both test problems, but more so with the difficult problem, CardiffGA had a greater success rate in reaching the known optimum solution within a required accuracy than an ordinary GA. The CGA was also typically more efficient with its use of genetic operations and evolved the population with fewer such operations. This was offset to a certain extent by additional computations not required in the ordinary GA, but nonetheless, the overall computational time for CGA was still significantly less. purchase the full-text of this paper (price £20) go to the previous paper go to the next paper return to the table of contents return to the book description purchase this book (price £95 +P&P)
Back to top	©Civil-Comp Limited 2023 - terms & conditions