A Note on the Maximum Missed-opportunity Cost Incurred by Minimizing Expected Losses in Genetic Algorithms
暂无分享,去创建一个
The foundational analysis of genetic algorithms in terms of sampling from schemata relies on minimizing expected losses. This criterion does not correspond with the problem of finding the single solution to a problem that affords the best quality. Analysis developed here indicates the maximum cost that might be realized by a search strategy that minimizes expected losses when sampling from non-negative random variables.
[1] John H. Holland,et al. Adaptation in Natural and Artificial Systems: An Introductory Analysis with Applications to Biology, Control, and Artificial Intelligence , 1992 .
[2] Günter Rudolph,et al. Reflections on Bandit Problems and Selection Methods in Uncertain Environments , 1998, ICGA.
[3] Kennetb A. De. Genetic Algorithms Are NOT Function Optimizers , 1992 .
[4] Kenneth A. De Jong,et al. Are Genetic Algorithms Function Optimizers? , 1992, PPSN.