基于多核系统的混合神经网络及应用  

Application of Hybrid Neural Network based on Multi-Core System

作　　者：杨蕾 林红[2] YANG Lei,LIN Hong（1 College of Science;China University of Petroleum,Qingdao Shandong 266555;China） 2 College of Pipeline and Civil Engineering;China University of Petroleum,Qingdao Shandong 266555;China）

机构地区：[1]中国石油大学（华东）理学院,山东青岛266555 [2]中国石油大学（华东）储运与建筑工程学院,山东青岛266555

出　　处：《智能计算机与应用》2011年第4期4-6,共3页Intelligent Computer and Applications

基　　金：中央高校基本科研业务费专项资金资助（09CX04064A）

摘　　要：借助混沌免疫遗传优化算法对于BP神经网络进行训练,建立基于混沌免疫遗传算法的混合神经网络模型。针对混沌免疫遗传神经网络计算工作量大,训练速度慢的缺点,利用Matlab的Parallel Computing Toolbox对于所建立的混沌免疫遗传神经网络模型进行并行化算法设计实现,并对渤海海区年极值冰厚数据进行预测,对比分析了串行和并行算法的计算效率和加速比,表明基于多核系统的并行化设计算法可以提高加速比和计算效率。Through chaos immune genetic optimization algorithm,BP neural network is optimized and neural network models based on chaos immune genetic algorithm is established to optimize the network weights.For the shortcoming of heavy computing workload in the hybrid neural network,parallel algorithm is designed and programmed by Parallel Computing Toolbox in Matlab.The ice thickness in Bohai sea is predicted by the model as an example.The computing efficiency and speedup ratio are compared between serial and parallel algorithm.The results show that the parallel algorithm based on multi-core system can increase the speedup ratio and raise computing efficiency.

关 键 词：混沌免疫遗传算法 神经网络 多核系统 计算效率 海冰厚度 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论;自动化与计算机技术—计算机科学与技术]