摘要
针对经典遗传规划算法(CGP)存在容易早熟收敛、运行效率低的缺陷,提出一种将分布式计算与遗传规划算法结合的计算模型。该模型利用个体迁移策略实现对种群的优化,克服易早熟的缺陷。并且采用分布式计算能够有效地节省算法的运行时间。最后通过对语音数据预测误差的比较,验证了改进后算法的有效性。实验表明,基于分布式粗粒度并行计算的遗传规划算法(CGGP)计算性能优于经典遗传规划算法(CGP)。
Based on the deficiencies of classical genetic programming（ CGP）, such as convergencing prematurely and operat- ing efficiency, this paper proposed a computing model which combined distributed computing with genetic programming. The model optimized populations through the individual migration strategy, overcame the deficiency of prematurity, and saved the running time of the algorithm effectively through distributed computing. Finally, it validated the effectiveness of algorithm which improved through comparing the prediction error of the speech data. The experiment demonstrated that the calculated performance of the genetic programming algorithm based on distributed coarse-grained parallel computing （CGGP） is better than that of classical genetic programming（CGP）.
出处
《计算机应用研究》
CSCD
北大核心
2015年第1期48-50,共3页
Application Research of Computers
基金
国家自然科学基金面上资助项目(11172342
11372167)
国家教育部"新世纪优秀人才支持计划"资助项目(NCET-11-0674)
陕西省自然科学基金重点项目(2014JZ021)
关键词
遗传规划
子种群
收敛
分布式计算
genetic programming（GP）
subpopulation
convergence
distributed computing
