Wyniki wyszukiwania - BazTech

Ograniczanie wyników

2 Foundations of Computing and Decision Sciences

Znaleziono wyników: 2

Liczba wyników na stronie

Wyniki wyszukiwania

Wyszukiwano:
w słowach kluczowych: multiple objective genetic algorithms

Sortuj według:

Ogranicz wyniki do:

An empirical investigation on elitism in multiobjective genetic algorithms

Parks G., Jianping L., Balazs M., Miller I.

Foundations of Computing and Decision Sciences

2001

Vol. 26, No. 1

51-74

This paper describes an investigation of the efficacy of various elitist selection strategies in a multiobjective Genetic Algorithm implementation, with parents being selected both from the current population and from the archive record of nondominated solutions encountered during search. The algorithm is tested both on a mathematical test function and a difficult real-world problem. It is concluded that, because the archive of nondominated solutions is inherently diverse, it is possible to improve the search performance of the algorithm through the use of strongly elitist selection strategies exploiting the archive (archive elitism) and the use of high selection pressures without the population suffering premature convergence. Archive elitism serves to maintain diversity in the population.

Multi-species Pareto frontiers in preliminary submarine design

Thomas M. W.

Foundations of Computing and Decision Sciences

2000

Vol. 25, No. 4

273-289

A preliminary design code for submarines is used as the evaluator for a Pareto genetic algorithm. The algorithm seeks the code's Pareto (nondominated) frontier in terms of hull volume and propulsive efficiency, in order to investigate the feasibility of so-called "full stern" submarines. Three different selection schemes--two ranking methods and a tournament method--are installed in the algorithm and compared in terms of their ability to locate and define the frontier. A new concept in evolutionary computation--non-interbreeding competitive species--is introduced, allowing simultaneous optimization of four incompatible propulsor configurations. This produces a feasibility frontier with optimal propulsor configuration as a function of stern fullness.