Memetic algorithm

A memetic algorithm is an extension of an evolutionary algorithm, used in computer science and operations research, that speeds up the evolutionary search for an optimum by adding one or more heuristics or local search techniques to improve the solutions the evolutionary algorithm generates. The term was introduced by Pablo Moscato in a technical report in 1989, who saw it as close to a population-based hybrid genetic algorithm coupled with an individual learning procedure that performs local refinements. The name draws on both Darwinian evolution and Dawkins notion of a meme. The idea is to combine the global search of an evolutionary method with local refinement, because evolutionary algorithms perform poorly at fixing the exact position of an optimum while local search is limited by its starting point. In the literature such algorithms are also called Baldwinian evolutionary algorithms, Lamarckian evolutionary algorithms, cultural algorithms or genetic local search.