摘要
针对大多数模型对时间序列预测数据的预测准确率较低,为提升时间序列的预测精度,提出一种基于Biased Drop-weight的偏置剪枝叠式自编码回声状态网络(BD-AE-SGESN)的深度模型。以叠式ESN为多层深度网络框架,提出一种生成式AE算法生成每一层的输入权值,利用BD算法根据输入权重激活值进行剪枝。对比实验结果表明,该模型能够有效提升预测准确率,在3个不同的数据上,相比其它模型有着较小的预测误差和较高的稳定度。
For the low prediction accuracy of most models on time series prediction data,to improve the accuracy of multivariate time series prediction,a depth model based on Biased Drop-weight stacked self-coding echo state network(BD-AE-SGESN)was proposed.The proposed stacked ESN was used as a multilayer depth network framework and a generative self-coding algorithm was proposed to generate the input weights for each layer,and the BD algorithm was used to prune the input weights according to their activation values.The proposed model can effectively improve the prediction accuracy with long memory,and it has smaller prediction error and higher stability than other models on three different data.
作者
刘丽丽
刘玉玺
王河山
LIU Li-li;LIU Yu-xi;WANG He-shan(School of Physics and Information Technology,Shaanxi Normal University,Xi’an 710062,China;College of Electrical Engineering,Zhengzhou University,Zhengzhou 450001,China)
出处
《计算机工程与设计》
北大核心
2024年第1期212-219,共8页
Computer Engineering and Design
基金
国家自然科学基金项目(61603343)
河南省高等学校重点科研基金项目(22A413009)。
关键词
多变量时间序列
回声状态网络
预测模型
剪枝
自编码
深度网络
权重优化
multivariable time series prediction
echo state network
prediction model
pruning
auto-encoder
deep network
weight optimization
作者简介
刘丽丽(1982-),女,陕西西安人,硕士,讲师,研究方向为复杂网络人工智能和神经网络;刘玉玺(1996-),男,河南郑州人,硕士研究生,研究方向为神经网络建模与优化;通讯作者:王河山(1987-),男,河南安阳人,博士,副教授,研究方向为人工智能、递归神经网络建模及优化、时间序列预测。E-mail:h.s.wang@zzu.edu.cn。
