Evolutionary computing methods have come to be of considerable interest in various practical applications that require some cost function to be optimized. However, the practitioner has to choose from a plethora of operators described in the literature on evolutionary computing. The relations between these operators are unclear, and there are no clear guidelines to assist in the task of deciding which operator is best for a new problem. This paper surveys various operators, and provides formal definitions for the more frequently used ones. A framework of relative generality is given to compare various operators. An experimental strategy to select suitable operators is formulated. An algorithm is also outlined, showing how the probabilities of applying various operators may be adapted in a generational evolutionary algorithm. Holland's Schema Theorem is extended, addressing the regeneration of instances of schema when evolutionary operators are applied.
|Original language||English (US)|
|Number of pages||20|
|Journal||International Journal of Smart Engineering System Design|
|State||Published - Jan 1 1998|
ASJC Scopus subject areas