遗传算法优化神经网络的拓扑结构与权值

广东工业大学学报 ›› 2006, Vol. 23 ›› Issue (4): 64-69.

遗传算法优化神经网络的拓扑结构与权值

西北工业大学经济研究中心；北京清河大楼子八陕西西安710072；陕西西安710072；北京清河大楼子八；北京； 100085；北京100085；

出版日期:2006-11-10 发布日期:2006-11-10

Optimization of Topological Structure and Weight Value of Artificial Neural Network Using Genetic Algorithm

(1.Economics Research Center,Northwestern Polytechnical University,Xi’an 710072,China;2.Qing-he Building D8,Beijing 100085,China)

Online:2006-11-10 Published:2006-11-10

摘要/Abstract

摘要： 遗传算法(GA)和人工神经网络(ANN)的相互结合有辅助式和合作式两种方式.本文在此基础上提出了融合、BP_GA和GA_BP三种算法,并采用GA_BP算法同时优化BP神经网络的结构、权值和阈值,研究和实现了一套先进的编码技术和进化策略,克服了传统BP神经网络经验尝试方法的盲目性.实例优化与检验结果表明:遗传算法优化获得的神经网络比由经验尝试法得到的BP网络性能更优异,方法更合理.

关键词: 遗传算法:神经网络；拓扑结构；权值；

Abstract: In crder to make use of the respective advantages of Genetic Algorithm(GA)and BP neural network,integrated algorithm,BP-GA algorithm and GA-BP algorithm are brought for ward.Genetic algorithm is used to optimize the topological structure,the weight values and the biases of BP neural network.Furthermore,a set of advanced coding technique and evolution tactic are studied and realized to overcome the blindness resulted from experiences and tries of traditional BP network.The simulation results of instances show that optimised metwork is of better fitting capability and of more reasonable method than traditional BP network.

Key words: genetic algorithm； artificial neural network； topological structure； weight value；

谷小青；易当祥；刘春和； . 遗传算法优化神经网络的拓扑结构与权值[J]. 广东工业大学学报, 2006, 23(4): 64-69.

GU Xiao-qing~1,YI Dang-xiang~(1,2),LIU Chun-he~2. Optimization of Topological Structure and Weight Value of Artificial Neural Network Using Genetic Algorithm[J]. Journal of Guangdong University of Technology, 2006, 23(4): 64-69.

参考文献

[1] 朱大奇. 人工神经网络研究现状及其展望[J]. 江南大学学报. 2004(01)

[2] 赵正佳,黄洪钟,陈新. 优化设计求解的遗传神经网络新算法研究[J]. 西南交通大学学报. 2000(01)

[3] 赖鑫生. 神经网络结合遗传算法优化应用[J]. 贵州大学学报(自然科学版). 2004(02)

[1] Eric I Chang,,R ichard P L ippm ann.Using genetic algorithm s to improve pattern c lassification perform ance. Pro-ceed ings of the 1990 Conference on advances in neural inform ation processing system s 3 .

[2] Yao X.Areview of evolutionary artific ial neural networks. International Journal of Intelligent System s . 1993

Metrics

Viewed

Full text

Abstract

Cited

Shared

Discussed