| Summary: | 碩士 === 輔仁大學 === 資訊管理學系碩士班 === 104 === Genetic algorithm (GA) is a well-known and popular meta- heuristic method, and thousands of practice optimal problems are solved by GA in widely fields. For reducing the executing time of GA, it needs to distribute the computing jobs of GA with a distributed computing architecture. Since Hadoop MapReduce is the most popular distributed computing architectures and Apache Spark improve Hadoop’s executing effective by using memory, several frameworks are proposed for discussing the distributed GA in current literatures. However, there is still lack an efficient framework for integrating with GA and Hadoop MapReduce or Apache Spark. For this research gap, this study proposes a novel framework for highly integrating with GA and Hadoop/Spark. The propose framework can more advantage the features of MapReduce through appropriately dispatching the core operators of GA into Mapper and Reducer. An experiment demonstrates that the proposed framework is more efficient and effective than current frameworks, especially when the problem scale is large.
|
|---|
