首页 | 本学科首页   官方微博 | 高级检索  
     检索      

基于量子差分进化算法的神经网络优化方法
引用本文:杜文莉,周仁,赵亮,钱锋.基于量子差分进化算法的神经网络优化方法[J].清华大学学报(自然科学版),2012(3):331-335.
作者姓名:杜文莉  周仁  赵亮  钱锋
作者单位:华东理工大学化工过程先进控制和优化技术教育部重点实验室
基金项目:国家“九七三”重点基础研究发展计划(2009CB320603);国家自然科学基金面上项目(20876044);上海科技攻关项目(09DZ1120400,10JC1403400);上海市重点学科建设项目(B504)
摘    要:一般的神经网络的结构是固定的,在实际应用中容易造成冗余连接和高计算成本。该文采用了协同量子差分进化算法(cooperative quantum differential evolution algo-rithm,CQGADE)以同时优化神经网络的结构和参数,即采用量子遗传算法(quantum genetic algorithm,QGA)来优化神经网络的结构和隐层节点数,采用差分算法来优化神经网络的权值。训练后的神经网络的连接开关能有效删除冗余连接,算法的量子概率幅编码和协同机制可以提高神经网络的学习效率、逼近精度和泛化能力。仿真实验结果表明:用训练后的神经网络预测太阳黑子和蒸汽透平流量具有更好的预测精度和鲁棒性。

关 键 词:神经网络  差分进化算法  协同量子差分进化算法(CQGADE)

Cooperative quantum differential evolution algorithm based method for optimizing neural networks
DU Wenli,ZHOU Ren,ZHAO Liang,QIAN Feng.Cooperative quantum differential evolution algorithm based method for optimizing neural networks[J].Journal of Tsinghua University(Science and Technology),2012(3):331-335.
Authors:DU Wenli  ZHOU Ren  ZHAO Liang  QIAN Feng
Institution:(Key Laboratory of Advanced Control and Optimization for Chemical Processes of Ministry of Education,East China University of Science and Technology,Shanghai 200237,China)
Abstract:Neural network structures are fixed,which results in redundant connections and high computing costs.This paper presents a cooperative quantum differential evolution algorithm(CQGADE) that simultaneously optimizes the neural network structure and parameters.The quantum genetic algorithm is used to optimize the neural network structure and the number of hidden nodes,while the differential evolution algorithm is used to optimize the neural network weights.This reduces redundant neural network structures,while the amplitude-based coding method and a cooperation mechanism improve the learning efficiency,approximation accuracy,and generalization.Simulations show that this algorithm has better prediction accuracy and robustness for predicting the number of sunspots and the flow of steam turbine.
Keywords:neural network  differential evolution algorithm  cooperative quantum differential evolution algorithm(CQGADE)
本文献已被 CNKI 等数据库收录!
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司  京ICP备09084417号