[1]苌群康,王杰,彭金柱.极限学习机优化及其拟合性分析[J].郑州大学学报(工学版),2016,37(02):20-24.[doi:10.3969/j.issn.1671-6833.201505001]
　Wang Jiechang,Qun Kang,Peng Jinzhu.Study on Extreme learning machine optimized of the fireworks algorithm[J].Journal of Zhengzhou University (Engineering Science),2016,37(02):20-24.[doi:10.3969/j.issn.1671-6833.201505001]
点击复制

极限学习机优化及其拟合性分析()
分享到：

《郑州大学学报(工学版)》[ISSN:1671-6833/CN:41-1339/T]

卷:
37卷
期数:
2016年02期
页码:
20-24
栏目:
出版日期:
2016-04-18

文章信息/Info

Title:
Study on Extreme learning machine optimized of the fireworks algorithm
作者:
苌群康; 王杰; 彭金柱
郑州大学电气工程学院,河南郑州,450001
Author(s):
Wang Jiechang;  Qun Kang;  Peng Jinzhu
School of Electrical Engineering, Zhengzhou University, Zhengzhou, Henan 450001
关键词:
Keywords:
fireworks algorithm; ELM; test error; node in hidden layer; FWAELMfitting
DOI:
10.3969/j.issn.1671-6833.201505001
文献标志码:
A
摘要:
运用烟花算法(fireworks algorithm,FWA)优化极限学习机(extreme learning machine,ELM).首先烟花算法经过多次的迭代,确定M个最优的烟花,并且以极限学习机测试样本的RMSE作为烟花算法每次迭代的适应度函数,达到优化极限学习机的输入权值矩阵和隐含层偏差的效果.最后根据广义逆求出输出矩阵.通过对一维sinC函数的测试结果表明,烟花算法优化极限学习机能够以较少的隐含层节点数目达到更高的精度,比极限学习机的测试误差降低了29.58％.在以上基础上又做了对高斯正态分布函数的拟合实验,验证了烟花算法优化极限学习机比极限学习机拥有更好的拟合性能.
Abstract:
Use the fireworks algorithm (fireworks algorithm, FWA) to optimize the extreme learning machine (extreme learning machine, ELM). First, the fireworks algorithm undergoes multiple iterations to determine M optimal fireworks, and the RMSE of the extreme learning machine test sample is used as the fireworks algorithm The fitness function of each iteration achieves the effect of optimizing the input weight matrix and hidden layer deviation of the extreme learning machine. Finally, the output matrix is ​​obtained according to the generalized inverse. The test results of the one-dimensional sinC function show that the fireworks algorithm optimizes the limit The learning machine can achieve higher accuracy with fewer nodes in the hidden layer, which is 29.58% lower than the test error of the extreme learning machine. On the basis of the above, the fitting experiment of the Gaussian normal distribution function was done to verify that The fireworks algorithm optimizes the extreme learning machine to have better fitting performance than the extreme learning machine.

相似文献/References:

[1]郝向东,毛晓波,梁静.基于ELM与Mean Shift的抗遮挡目标跟踪算法[J].郑州大学学报(工学版),2016,37(01):1.[doi:10.3969/j.issn.1671-6833.201503043]
　Mao Xiaobo,Hao Xiangdong,Liang Jing.Algorithm of Target Tracking Based on Mean Shift with ELM[J].Journal of Zhengzhou University (Engineering Science),2016,37(02):1.[doi:10.3969/j.issn.1671-6833.201503043]
[2]蔡金收,陈铁军,郭丽.基于投票极限学习机的人脸识别混合算法研究[J].郑州大学学报(工学版),2016,37(02):37.[doi:10.3969/j.issn.1671-6833.201505016]
　Chen Tiejun,Cai Jinshou,Guo Li.Research on Hybrid Face Recognition Algorithm Based on Voting Extreme Learning Machine[J].Journal of Zhengzhou University (Engineering Science),2016,37(02):37.[doi:10.3969/j.issn.1671-6833.201505016]

更新日期/Last Update: