/img alt="Imagem da capa" class="recordcover" src="""/>
Trabalho de Conclusão de Curso - Graduação
Análise comparativa entre novos operadores genéticos incluídos no Framework Evolutionary Algorithms
New operators for Genetic Algorithms are being proposed daily by the academic community to improve the performance of this technique. It is necessary to know their performance in order to make good use of these improvements and to know their limitations and strengths. This work is a comparative a...
Autor principal: | BARRETO, Adriano Silva |
---|---|
Outros Autores: | SOUSA, Thales Silva de |
Grau: | Trabalho de Conclusão de Curso - Graduação |
Publicado em: |
2019
|
Assuntos: | |
Acesso em linha: |
http://bdm.ufpa.br/jspui/handle/prefix/1334 |
Resumo: |
---|
New operators for Genetic Algorithms are being proposed daily by the academic community
to improve the performance of this technique. It is necessary to know their performance
in order to make good use of these improvements and to know their limitations
and strengths. This work is a comparative analysis of variants of genetic algorithms that
were created and implemented by the academic community. The purpose of this study
is to perform comparisons between variants of genetic operators to identify the existing
differences in performance offered by them. The genetic operators that were researched
in this work are: the transgenic operator, the operator of parasite diversity and the adaptive
immune operator based on information entropy. These operators were implemented
and evaluated through tests with multimodal functions. An analysis was made among
the genetic algorithms in order to evaluate if the algorithm finds the solution and the
convergence guarantee. Some metrics that were evaluated in the operators were the robustness
to optimize the function with a given error tolerance and a convergence analysis.
It was considered in this work that the solution is found according to various defined
precisions, where the error is less than or equal to 10−3, 10−2 and 10−1. After the tests,
the performance analysis performed among the implemented operators showed that all
the operators obtained good results for the functions with a good convergence and the
operator that obtained the best results was the adaptive immune operator. |