In this paper, we postulate some desired behaviors of any evolutionary algorithm (EA) to demonstrate self-adaptive properties. Thereafter, by calculating population mean and variance growth equations, we find bounds on parameter values in a number of EA operators which will qualify them to demonstrate the self-adaptive behavior. Further, we show that if the population growth rates of different EAs are similar, similar performance is expected. This allows us to connect different self-adaptive EAs on an identical platform. This may lead us to find a more unified understanding of the working of different EAs.
CITATION STYLE
Beyer, H. G., & Deb, K. (2000). On the desired behaviors of self-adaptive evolutionary algorithms. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (Vol. 1917, pp. 60–68). Springer Verlag. https://doi.org/10.1007/3-540-45356-3_6
Mendeley helps you to discover research relevant for your work.