contato   links   


Disciplina: Inteligência Computacional Aplicada (ELE1361)
Projeto: -
Link Original: http://www.inf.ufsc.br/~|3c/artigos/leher02.pdf
Some important concepts belonging to evolutionary game theory are employed to test how they can improve the performance of the operators used in genetic algorithms (GA). The use of rational strategies can provide additional efficiency to the GA regarding the search of satisfactory solutions to difficult problems. In this case, the traditional GA operators, namely, selection, crossover and mutation will no more rely solely on random criteria to carry out the exploration of the chromosomes for better fitness, herein considered as a limited and scarce resource. To accomplish this method, the selected paradigm has been the Hawk-Dove Game, well know as an important model of strategic behavior in ecological studies. The participants of the game are the chromosomes, which exert their respective strategies and strive for improving their individual fitness. To test the method, the traveling salesman problem has been used. A series of simulations has been run and the achieved results are presented especially some comparisons between the usual methods regarding the GA operators. Some evidence has been found that indicates the advantage of the investigated methodology.

Nenhuma palavra chave relacionada.

Nenhum arquivo cadastrado.