英语论文网

deliver adequate performance. Lawrence (1991) reminds us to: “hybridize where possible”, if we want to improve the results achieved by a standard genetic algorithm. This reflects the fact that although good performance can be often achieved by using the most appropriate general-purpose optimization algorithm, much greater gains can often be made by combining it with specific heuristics or operators that incorporate ‘domain knowledge'. In many difficult, well-studied problems, the best results come from such hybrid approaches where very specific heuristics are combined with well proven strategies.

However, despite the message of Davis, it is generally easier to hybridize problem-specific heuristics with a local search based method than it is with a GA, employing recombination. This is because many heuristics use a form of local improvement, but few use any sort of recombination. Nonetheless, it is true that recombination can be a valuable additional operator to use. In these cases, hybrid GAs or memetic algorithms have proved an effective approach. In MAs, there is a local search phase which is often hybridized with some other heuristic technique for the specific problem, and a recombinative phase that can quickly combine parts of promising solutions together, to form better ones.

1.5.1 演算法——1.5.1 Memetic Algorithms

The term memetic algorithms has been used to identify a broad class of hybrid metaheuristics. Memetic Algorithms (MAs) are population-based metaheuristic search methods inspired by both Darwinian principles of natural evolution and Dawkins notion of a meme as a unit of cultural evolution capable of individual learning. In a more diverse context, MA can be defined as a synergy of evolution and individual learning. While genetic algorithms are inspired by the metaphor of genes, memetic algorithms are inspired by the metaphor of memes. A gene is the unit of genetic information that is propagated biologically between generations during the evolution process. A meme is the unit of conceptual information (knowledge, ideas, behaviour, customs, etc.) that is transmitted by imitation from one generation to the next one. Then by incorporating the available knowledge about the problem into an evolutionary algorithm, the working metaphor is that of evolving a population both biologically and culturally.

MA has the potential of exploiting the complimentary advantages of EAs (generality, robustness, global search efficiency), and problem-specific local search (exploiting application-specific problem structure, rapid convergence toward local minima). Such combinations of optimizers are commonly known as hybrid methods. In diverse contexts, hybris EAs are also commonly known as Memetic Algorithms, Baldwinian EAs, Lamarckian EAs, cultural algorithms or genetic local search. Such methods have been demonstrated to converge to high quality solutions more efficiently than their conventional counterparts (Ong, Lim, Zhu & Wong, 2006; Ong, Nair & Lum, 2006). Since we consider evolutionary algorithms that employ individual learning heavily during the entire lifetime of the search, the term Memetic Algorithms is most appropriately used. Besides, the name of Memetic Algorithms is more widely used now since it is believed to be more general and encompasses all the major concepts involved in the others. The pseudo-code of a Memetic Algori本论文由英语论文网提供整理，提供论文代写，英语论文代写，代写论文，代写英语论文，代写留学生论文，代写英文论文，留学生论文代写相关核心关键词搜索。