An iterative MapReduce framework suitable for evolutionary algorithms

An iterative MapReduce framework suitable for evolutionary algorithms

Jin Weijian Wang Branch

MapReduce modular programming greatly reduces the difficulty of realizing the distributed algorithm, but it also limits its application scope. This paper introduces the basic structure of mapreduce and the defects of its iterative algorithm, and proposes an iterative MapReduce computing framework which is suitable for evolutionary algorithm based on the research of MapReduce's computational framework on the basis of the inefficiency of mapreduce evolutionary algorithm. This paper describes the implementation requirements and implementation of the iterative MapReduce computing framework, presents and proves the feasibility of the anomaly mechanism, and validates the proposed framework on the public Hadoop cloud computing platform. The experimental results show that the parallel genetic algorithm based on the iterative MapReduce computing framework is better than the parallel genetic algorithm based on MapReduce.

An iterative MapReduce framework suitable for evolutionary algorithms