| Summary: | 碩士 === 輔仁大學 === 資訊管理學系 === 95 === In recent years the Evolutionary Computation continually flourish, and successfully applied in many domains, The Genetic Algorithms for dealing with linear and nonlinear problems is a good effect. However, the genetic algorithm has some shortcomings exist. For example, when genetic algorithm deals with some specific problem, it can only search for the optimal solution in the local region and the inability to find the global optimal solution, cause adverse evolution of the phenomenon. Our research use the concept of genomic to enhance exploration, and use the concept of chance discovery, supplemented by genetic engineering with simple genetic algorithms, Try to improve simple genetic algorithm, to enhance its efficiency and effectiveness.
Hence, our research's experimental subjects are the member of TSEC Taiwan 50 index and TSEC Taiwan 100 index, discuss the different parameter sets and settings caused by the phenomenon of evolution. And then, we try to buy and holders of the portfolio, observation remuneration trends and then compare the simple GA, the evolutionary genomics and the adaptive evolutionary genomic. Experimental results show that the trend on the actual investment, the alternative space is the member of TSEC Taiwan 50 index, its need to be brought and held until five months for pay. Besides, the member of TSEC Taiwan 100 index need to be brought for held until three months for pay. From the perspective of algorithm, evolutionary genomic has excellent ability to explore, and it can find the great genomics (the gene fragments) to be preserved, and then escape from the region as the opportunity to solutions. When genetic engineering interacts with genetic algorithms, they can use this opportunity to escape form the local region to find the optimal solution in the direction of the global region. The adaptive evolutionary genomic use the enhance exploration of evolutionary genomic and combine the exploitation by genetic algorithm, the effectiveness is better than simple genetic algorithm.
|
|---|
