Modeling of genetic algorithms with a finite population
Cross-competition between non-overlapping building blocks can strongly influence the performance of evolutionary algorithms. The choice of the selection scheme can have a strong influence on the performance of a genetic algorithm. This paper describes a number of different genetic algorithms, all involving elitism. Infinite population models are presented for each of these algorithms. A problem involving cross-competition is introduced and we show how we can make use of equivalence-classes to make an efficient tracing of the transmission-function models possible on this type of problems. By adding a small extension to the models it is possible to predict the qualitative behavior of finite population genetic algorithms on this type of problems also. Using this model the reliability of the different genetic algorithms and the influence of population sizing on the reliability is investigated.
|Miscellaneous (acm F.2.m), Optimization (acm G.1.6), Problem Solving, Control Methods, and Search (acm I.2.8)|
|Learning and adaptive systems (msc 68T05), Problem solving (heuristics, search strategies, etc.) (msc 68T20)|
|Software Engineering [SEN]|
van Kemenade, C.H.M. (1997). Modeling of genetic algorithms with a finite population. Software Engineering [SEN]. CWI.