[1]丁胜夺,谭 昆,田 琨,等.基于自适应遗传算法的极限学习机改进算法[J].计算机技术与发展,2022,32(S1):26-30.[doi:10. 3969 / j. issn. 1673-629X. 2022. S1. 006]
　DING Sheng-duo,TAN Kun,TIAN Kun,et al.Improved Algorithm of Extreme Learning Machine Based on Adaptive Genetic Algorithm[J].,2022,32(S1):26-30.[doi:10. 3969 / j. issn. 1673-629X. 2022. S1. 006]
点击复制

基于自适应遗传算法的极限学习机改进算法()
分享到：

《计算机技术与发展》[ISSN:1006-6977/CN:61-1281/TN]

卷:
32
期数:
2022年S1期
页码:
26-30
栏目:
大数据分析与挖掘
出版日期:
2022-12-11

文章信息/Info

Title:
Improved Algorithm of Extreme Learning Machine Based on Adaptive Genetic Algorithm
文章编号:
1673-629X(2022)S1-0026-05
作者:
丁胜夺; 谭 昆; 田 琨; 吴顺成
中国石油天然气股份有限公司安全环保技术研究院,北京 102206
Author(s):
DING Sheng-duo; TAN Kun; TIAN Kun; WU Shun-cheng
China National Petroleum Corporation Safety and Environmental Technology Research Institute Co. ,Ltd. ,Beijing 102206,China
关键词:
遗传算法; 极限学习机; 自适应; 分类算法; 回归预测
Keywords:
genetic algorithm; extreme learning machine ( ELM) ; adaption; classification algorithm; regression prediction
分类号:
TP311
DOI:
10. 3969 / j. issn. 1673-629X. 2022. S1. 006
摘要:
针对极限学习机结构参数过多,初始化阶段输入层和隐层间的权值矩阵以及隐层的阈值随机生成导致该算法性能不稳定、易出现过拟合现象的缺陷,在自适应遗传算法的基础上对极限学习机算法进行优化提升。 首先,该算法将传统遗传算法中的交叉变异概率改进为随群体适应度水平和当前遗传代数更新的自适应参数,优化算法对局部和全局最优解的搜索能力;然后,将极限学习机中的待定参数作为染色体带入遗传算法中进行交叉变异操作寻找适应度最佳个体。 通过仿真实验将该改进算法和 ELM、BP 神经网络模型进行比较,实验结果表明该改进算法在分类问题以及回归问题中都得到了更加准确的预测结果,极限学习机的泛化性能得到显著提升。
Abstract:
In view of the defects of the extreme learning machine algorithm,such as too many parameters,the connection weights betweenthe input layer and the hidden layer and the random generation of the threshold of the hidden layer, which lead to the unstableperformance of the algorithm and the over-fitting phenomenon,an improved extreme learning machine algorithm based on the adaptivegenetic algorithm is proposed. Firstly, the crossover mutation probability of traditional genetic algorithm is changed into an adaptiveparameter that changes with population fitness and genetic algebra, and the global and local searching ability of genetic algorithm isenhanced. Then,the parameters to be determined in the limit learning machine are taken as chromosomes into the genetic algorithm forcross mutation operation to find the optimal individual. The improved algorithm was compared with ELM and BP neural network modelsthrough simulation experiments,and the experimental results showed that the improved algorithm obtained more accurate prediction resultsin both classification problems and regression problems,and effectively improved the generalization performance of the extreme learningmachine.

相似文献/References:

[1]冯智明,苏一丹,覃华,等.基于遗传算法的聚类与协同过滤组合推荐算法[J].计算机技术与发展,2014,24(01):35.
　FENG Zhi-ming,SU Yi-dan,QIN Hua,et al.Recommendation Algorithm of Combining Clustering with Collaborative Filtering Based on Genetic Algorithm[J].,2014,24(S1):35.
[2]余晓光 严洪森 殷乾坤.基于Flexsim的车间调度优化[J].计算机技术与发展,2010,(03):44.
　YU Xiao-guang,YAN Hong-sen,YIN Qian-kun.Workshops Scheduling Optimization Based on Flexsim Simulation[J].,2010,(S1):44.
[3]贺计文 宋承祥 刘弘.基于遗传算法的八数码问题的设计及实现[J].计算机技术与发展,2010,(03):105.
　HE Ji-wen,SONG Cheng-xiang,LIU Hong.Design and Implementation of Eight Puzzle Problem Based on Genetic Algorithms[J].,2010,(S1):105.
[4]沈珏萍 庄亚明.基于Agent的二级供应链企业自动谈判研究[J].计算机技术与发展,2010,(03):121.
　SHEN Jue-ping,ZHUANG Ya-ming.A Research for Company Automatic Negotiation in Secondary Supply Chain Based on Agent[J].,2010,(S1):121.
[5]张磊 王晓军.基于遗传算法的业务流程测试[J].计算机技术与发展,2010,(03):155.
　ZHANG Lei,WANG Xiao-jun.Test of Business Process Based on Genetic Algorithm[J].,2010,(S1):155.
[6]曹道友 程家兴.基于改进的选择算子和交叉算子的遗传算法[J].计算机技术与发展,2010,(02):44.
　CAO Dao-you,CHENG Jia-xing.A Genetic Algorithm Based on Modified Selection Operator and Crossover Operator[J].,2010,(S1):44.
[7]范维博 周俊 许正良.应用遗传算法求解第一类装配线平衡问题[J].计算机技术与发展,2010,(02):194.
　FAN Wei-bo,ZHOU Jun,XU Zheng-liang.Appication of Genetic Algorithm to Assembly Line Balancing[J].,2010,(S1):194.
[8]熊伟平 曾碧卿.几种仿生优化算法的比较研究[J].计算机技术与发展,2010,(03):9.
　XIONG Wei-ping,ZENG Bi-qing.Studies on Some Bionic Optimization Algorithms[J].,2010,(S1):9.
[9]余晓光 严洪森.基于禁忌搜索遗传混合算法的装配线平衡[J].计算机技术与发展,2010,(05):5.
　YU Xiao-guang,YAN Hong-sen.Assembly Line Balancing Based on Tabu Search and Genetic Hybrid Algorithm[J].,2010,(S1):5.
[10]黄永聪 张旭[] 吴义纯 吴琦 程家兴.改进的径向基函数网络的研究及应用[J].计算机技术与发展,2010,(05):158.
　HUANG Yong-cong,ZHANG Xu,WU Yi-chun,et al.Research and Application of Improved Genetic Algorithm-Based RBFANN[J].,2010,(S1):158.

更新日期/Last Update: 2022-06-10