| Article ID: | iaor19991954 |
| Country: | United Kingdom |
| Volume: | 25 |
| Issue: | 11 |
| Start Page Number: | 895 |
| End Page Number: | 911 |
| Publication Date: | Nov 1998 |
| Journal: | Computers and Operations Research |
| Authors: | Novkovic Sonja, verko Davor |
| Keywords: | artificial intelligence |
Using Goldberg's minimal deceptive problem, we devise a simple method for setting the initial parameters of a genetic algorithm based on the role of nonsense codons in natural genetics. ‘Genetic waste’ provides random parameters throughout the simulations, enabling the abandonment of trial and error search for the optimal combination of initial parameters and thereby adding credibility to the genetic algorithm as a tool of adaptive search.