evolutionary programming
evolutionary programming
[‚ev·ə¦lü·shə‚ner·ē ′prō‚gram·iŋ]evolutionary programming
An initially random population of individuals (trialsolutions) is created. Mutations are then applied to eachindividual to create new individuals. Mutations vary in theseverity of their effect on the behaviour of the individual.The new individuals are then compared in a "tournament" toselect which should survive to form the new population.
EP is similar to a genetic algorithm, but models only thebehavioural linkage between parents and their offspring, ratherthan seeking to emulate specific genetic operators from naturesuch as the encoding of behaviour in a genome andrecombination by genetic crossover.
EP is also similar to an evolution strategy (ES) althoughthe two approaches developed independently. In EP, selectionis by comparison with a randomly chosen set of otherindividuals whereas ES typically uses deterministicselection in which the worst individuals are purged from thepopulation.