Summary: | 碩士 === 國立暨南國際大學 === 資訊管理學系 === 99 === Cyber Swarm Algorithm (CSA) combining Particle Swarm Optimization (PSO) and Scatter Search/Path relinking (SS/PR) concepts has been empirically shown to effectively solve continuous optimization problems. However, the extension to discrete optimization problems is still uncharted. This motivates the thesis for proposing a Discrete Cyber Swarm Algorithm (DCSA). In addition to the applications of PSO and SS/PR, DCSA also marries with techniques such as aging, solution guide selection, and uncertainty principle in the machine learning domain. Further, DCSA applies the Apriori algorithm in data mining domain to find the association between attributes. Experimental results manifest that the proposed algorithm outperforms PSO, hybrid PSO, and novel global harmony search.
|