| A new non-monotone fitness scaling for genetic algorithm |
| |
| 作者姓名： | LI Minqiang KOU Jisong |
| |
| 作者单位： | Institute of Systems Engineering, Tianjin University, Tianjin 300072, China,Institute of Systems Engineering, Tianjin University, Tianjin 300072, China |
| |
| 基金项目： | Project supported by the National Natural Science Foundation of China (Grant No. 69974026). |
| |
| 摘 要： | The properties of selection operators in the genetic algorithm (GA) are studied in detail. It is indicated that the selection of operations is significant for both improving the general fitness of a population and leading to the schema deceptiveness. The stochastic searching characteristics of GA are compared with those of heuristic methods. The influence of selection operators on the GA' s exploration and exploitation is discussed, and the performance of selection operators is evaluated with the premature convergence of the GA taken as an example based on One-Max function. In order to overcome the schema deceptiveness of the GA, a new type of fitness scaling, non monotone scaling, is advanced to enhance the evolutionary ability of a population. The effectiveness of the new scaling method is tested by a trap function and a needle-in-haystack (NiH) function.
|
| 关 键 词： | genetic algorithms selection operator non-monotone fitness scaling |
A new non-monotone fitness scaling for genetic algorithm |
| |
| LI Minqiang,KOU Jisong.A new non-monotone fitness scaling for genetic algorithm[J].Progress in Natural Science,2001,11(8):622-630. |
| |
| Authors: | Li Minqiang KOU Jisong |
| |
| Institution: | Institute of Systems Engineering, Tianjin University, |
| |
| Abstract: | The properties of selection operators in the genetic algorithm (GA) are studied in detail. It is indicated that the selection of operations is significant for both improving the general fitness of a population and leading to the schema deceptiveness. The stochastic searching characteristics of GA are compared with those of heuristic methods. The influence of selection operators on the GA' s exploration and exploitation is discussed, and the performance of selection operators is evaluated with the premature convergence of the GA taken as an example based on One-Max function. In order to overcome the schema deceptiveness of the GA, a new type of fitness scaling, non monotone scaling, is advanced to enhance the evolutionary ability of a population. The effectiveness of the new scaling method is tested by a trap function and a needle-in-haystack (NiH) function. |
| |
| Keywords: | genetic algorithms selection operator non-monotone fitness scaling |
| 本文献已被 万方数据 等数据库收录！ |
| 点击此处可从《自然科学进展(英文版)》浏览原始摘要信息 |
| 点击此处可从《自然科学进展(英文版)》下载免费的PDF全文 |
