[1]冯 宇.基于张量偏最小二乘法的高维输出预测模型[J].计算机技术与发展,2019,29(07):114-118.[doi:10. 3969 / j. issn. 1673-629X. 2019. 07. 023]
　FENG Yu.A Multi-dimensional Output Prediction Model Based on Tensor Partial Least Squares[J].,2019,29(07):114-118.[doi:10. 3969 / j. issn. 1673-629X. 2019. 07. 023]
点击复制

基于张量偏最小二乘法的高维输出预测模型()
分享到：

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

卷:
29
期数:
2019年07期
页码:
114-118
栏目:
应用开发研究
出版日期:
2019-07-10

文章信息/Info

Title:
A Multi-dimensional Output Prediction Model Based on Tensor Partial Least Squares
文章编号:
1673-629X(2019)07-0114-05
作者:
冯 宇
长安大学 电子与控制工程学院,陕西 西安 710064
Author(s):
FENG Yu
School of Electronic and Control Engineering,Chang’an University,Xi’an 710064,China
关键词:
预测模型; 高维输出; 张量; 偏最小二乘法; 心脏电生理信息
Keywords:
prediction model; multi-dimensional output; tensor; PLS; cardiac electrophysiological information
分类号:
TP399
DOI:
10. 3969 / j. issn. 1673-629X. 2019. 07. 023
摘要:
针对输出为高维数据的预测问题,构建了一种基于张量偏最小二乘法的预测模型。 该模型的输入和输出均为张量,可在不进行数据降维操作的前提下对多个输出同时预测。 该方法以偏最小二乘法为基础,引入逐块 Tucker 分解和高阶奇异值分解,通过对输入和输出变量的分析,提取同时包含二者最大信息的潜在变量并计算残差,再通过残差计算新的潜在变量,循环直到残差小于给定范围为止。 实验数据来源于心脏传导系统在正常和急性高糖环境下采集的电生理信息,通过最大正振幅、最大负振幅、频率、单次信号持续时间四个维度的输入同时预测急性高糖的浓度和作用时间,并将预测结果与传统的多向偏最小二乘法和多维偏最小二乘法相比较。 实验结果表明,基于张量偏最小二乘法的预测模型预测精度最高。
Abstract:
A prediction model based on tensor partial least squares (TPLS) is built to solve the problem of high dimensional data. The input and output of the model are tensors,and multi-dimensional output can be predicted without dimensional reduction. Based on partial least squares, this method introduces block by block Tucker decomposition and high order singular value decomposition. Through the analysis of input and output variables,the potential variables containing the maximum information of both are extracted and the residual error is calculated. Then,the new potential variable is calculated through the residual error,and the loop runs until the residual error is less than the given range. The experimental data are derived from electrophysiological information of cardiac conduction systems in normal and acute hyperglycemic environments. The input is a four-dimensional tensor which contains maximum positive amplitude,maximum negative amplitude,frequency and single signal duration. The predictive output contains the concentrations and durations of acute hyperglycemia. The prediction results are compared with multi-way PLS and N-way PLS,which shows that the tensor PLS has the highest prediction accuracy.

相似文献/References:

[1]李霞.ID3分类算法在银行客户流失中的应用研究[J].计算机技术与发展,2009,(03):158.
　LI Xia.ID3 Applying to Loss of Bank Clients[J].,2009,(07):158.
[2]邱东 陈爽 仝彩霞 朱里红 王龙山.钢铁企业高炉煤气平衡与综合优化[J].计算机技术与发展,2009,(03):196.
　QIU Dong,CHEN Shuang,TONG Cai-xia,et al.Blast Furnace Gas Balance and Comprehensive Optimization in Iron and Steel Enterprises[J].,2009,(07):196.
[3]杨颖 陈德华.基于小波神经网络的时间序列流数据的研究[J].计算机技术与发展,2006,(06):193.
　YANG Ying,CHEN De-hua.Research for Model of Time Series Streaming Based on Wavelet Neural Network[J].,2006,(07):193.
[4]张瑞敏 黄梦涛 程青涛.智能神经元网络时间序列预测模型的研究[J].计算机技术与发展,2012,(03):74.
　ZHANG Rui-min,HUANG Meng-tao,CHENG Qing-tao.Application of Dynamic Intelligent Neural Network in Time Series Prediction[J].,2012,(07):74.
[5]吕克 徐夫田 舒文迪.基于神经网络的鸟撞预测模型应用研究[J].计算机技术与发展,2012,(05):90.
　LUE Ke,XU Fu-tian,SHU Wen-di.Bird-Strike Prediction Model Application and Research Based on Neural Network[J].,2012,(07):90.
[6]彭勇 陈俞强 严文杰.基于改进BP网络模型的公路流量预测[J].计算机技术与发展,2012,(08):111.
　PENG Yong,CHEN Yu-qiang,YAN Wen-jie.Forecasting Highway Flow Based on Improved BP Neural Network Model[J].,2012,(07):111.
[7]赵伟.一种改进的网络流量预测模型研究[J].计算机技术与发展,2013,(04):20.
　ZHAO Wei.Research on an Improved Prediction Model of Network Traffic[J].,2013,(07):20.
[8]邵鸿玲,吴陈,杨习贝.商业智能在连锁企业进销存管理中的应用研究[J].计算机技术与发展,2013,(07):198.
　SHAO Hong-ling,WU Chen,YANG Xi-bei.Research and Application on Business Intelligence in Chain Enterprise Inventory Management[J].,2013,(07):198.
[9]李炜,潘作舟,杨静.RI地震预测模型的分析及其验证[J].计算机技术与发展,2013,(09):255.
　LI Wei,PAN Zuo-zhou,YANG Jing.Analysis and Verification in RI Earthquake Forecast Model[J].,2013,(07):255.
[10]沈海迪,万振凯. 基于指数平滑法的动态预测机制[J].计算机技术与发展,2017,27(07):6.
　SHEN Hai-di,WAN Zhen-kai. A Dynamic Prediction Mechanism Based on Exponential Smoothing Method[J].,2017,27(07):6.

更新日期/Last Update: 2019-07-10