基于ELM改进层集成架构的时间序列预测
|
摘要
为进一步提高时间序列预测模型的预测精度和时间效率,提出一种基于极限学习机的层集成网络结构。以极限学习机网络作为基学习器,构成两层集成网络,每层网络在构建时利用先分类,再从类中选优的思想同时考虑基学习器的准确性与多样性,其中第一层用以优化参数,第二层实现预测。对比实验结果表明,与基于多层感知器的层集成网络相比,该模型在提高预测准确度的同时将学习用时缩短了1-2个数量级。
To further improve the prediction accuracy and efficiency of the time-series prediction model,an extreme learning machine based layered ensemble network was proposed.Extreme learning machine networks were taken as base predictors and consisted of two ensemble layers.Each layer considered both accuracy and diversity of the individual networks in constructing the ensemble by taking the strategy in which the best was selected after classification.The first ensemble layer was used to find an appropriate lag,while the second one was used to employ the obtained lag for forecasting.The proposed model was tested in comparison experiments.The results reveal clearly that compared with the layered ensemble architecture based on multilayer perceptron,the proposed model not only improves the prediction accuracy,but also improves the time efficiency by 1-2 orders of magnitude.
To further improve the prediction accuracy and efficiency of the time-series prediction model,an extreme learning machine based layered ensemble network was proposed.Extreme learning machine networks were taken as base predictors and consisted of two ensemble layers.Each layer considered both accuracy and diversity of the individual networks in constructing the ensemble by taking the strategy in which the best was selected after classification.The first ensemble layer was used to find an appropriate lag,while the second one was used to employ the obtained lag for forecasting.The proposed model was tested in comparison experiments.The results reveal clearly that compared with the layered ensemble architecture based on multilayer perceptron,the proposed model not only improves the prediction accuracy,but also improves the time efficiency by 1-2 orders of magnitude.
引文
[1]Xu M,Han M.Adaptive elastic echo state network for multivariate time series prediction[J].IEEE Transactions on Cybernetics,2017,46(10):2173-2183.
[2]Taieb SB,Atiya AF.A bias and variance analysis for multistep-ahead time series forecasting[J].IEEE Trans Neural Netw Learn Syst,2016,27(1):62-76.
[3]Chou JS,Nguyen TK.Forward forecast of stock price using sliding-window metaheuristic-optimized machine learning regression[J].IEEE Transactions on Industrial Informatics,2018,14(7):3132-3142.
[4]Yunus K,Thiringer T,Chen P.ARIMA-based frequencydecomposed modeling of wind speed time series[J].IEEETransactions on Power Systems,2016,31(4):2546-2556.
[5]YANG Mao,ZHANG Qiang.Real-time prediction of wind power based on ensemble empirical mode decomposition and correlation vector machine[J].Acta Energiae Solaris Sinica,2016,37(5):1093-1099(in Chinese).[杨茂,张强.基于集合经验模态分解和相关向量机的风电功率实时预测研究[J].太阳能学报,2016,37(5):1093-1099.]
[6]Rafiei M,Niknam T,Khooban MH.Probabilistic forecasting of hourly electricity price by generalization of ELM for usage in improved wavelet neural network[J].IEEE Transactions on Industrial Informatics,2017,13(1):71-79.
[7]Lippi M,Bertini M,Frasconi P.Short-term traffic flow forecasting:An experimental comparison of time-series analysis and supervised learning[J].IEEE Transactions on Intelligent Transportation Systems,2013,14(2):871-882.
[8]YIN Rongwang.Network traffic predicting model based on FCM-LSSVM[J].Computer Engineering and Applications,2016,52(1):105-109(in Chinese).[殷荣网.基于FCM-LSSVM网络流量预测模型[J].计算机工程与应用,2016,52(1):105-109.]
[9]Mohan BR.Resource usage prediction based on ARIMA-ARCHmodel for virtualized server system[J].International Journal of Geomate,2017,12(33):139-146.
[10]Mohamadi S,Amindavar H,Hosseini SM Ali Tayaranian.ARIMA-GARCH modeling for epileptic seizure prediction[C]//IEEE International Conference on Acoustics,Speech and Signal Processing,2017:994-998.
[11]Avnimelech R,Intrator N.Boosting regression estimators[J].Neural Computation,2014,11(2):499-520.
[12]Wang Z,Srinivasan RS.A review of artificial intelligence based building energy prediction with a focus on ensemble prediction models[C]//Winter Simulation Conference.IEEE,2016:3438-3448.
[13]Rahman MM,Islam MM,Murase K,et al.Layered ensemble architecture for time series forecasting[J].IEEE Transactions on Cybernetics,2016,46(1):270-283.
[14]ZHOU Zhihua.Maching learning[M].Beijing:Tsinghua University Press,2016:171-173(in Chinese).[周志华.机器学习[M].北京:清华大学出版社,2016:171-173.]
[15]Tang J,Deng C,Huang GB.Extreme learning machine for multilayer perceptron[J].IEEE Transactions on Neural Networks&Learning Systems,2017,27(4):809-821.
[16]Timeseries forecasting competition for neural networks and computational intelligence[EB/OL].[2017-12-20].http://www.neural-forecasting-competition.com/NN3.
[17]Chen L.A modified Levenberg-Marquardt method with line search for nonlinear equations[J].Computational Optimization&Applications,2016,65(3):753-779.
[18]Gao B,Xu Z.Phaseless recovery using the Gauss-Newton method[J].IEEE Transactions on Signal Processing,2017,65(22):5885-5896.
[2]Taieb SB,Atiya AF.A bias and variance analysis for multistep-ahead time series forecasting[J].IEEE Trans Neural Netw Learn Syst,2016,27(1):62-76.
[3]Chou JS,Nguyen TK.Forward forecast of stock price using sliding-window metaheuristic-optimized machine learning regression[J].IEEE Transactions on Industrial Informatics,2018,14(7):3132-3142.
[4]Yunus K,Thiringer T,Chen P.ARIMA-based frequencydecomposed modeling of wind speed time series[J].IEEETransactions on Power Systems,2016,31(4):2546-2556.
[5]YANG Mao,ZHANG Qiang.Real-time prediction of wind power based on ensemble empirical mode decomposition and correlation vector machine[J].Acta Energiae Solaris Sinica,2016,37(5):1093-1099(in Chinese).[杨茂,张强.基于集合经验模态分解和相关向量机的风电功率实时预测研究[J].太阳能学报,2016,37(5):1093-1099.]
[6]Rafiei M,Niknam T,Khooban MH.Probabilistic forecasting of hourly electricity price by generalization of ELM for usage in improved wavelet neural network[J].IEEE Transactions on Industrial Informatics,2017,13(1):71-79.
[7]Lippi M,Bertini M,Frasconi P.Short-term traffic flow forecasting:An experimental comparison of time-series analysis and supervised learning[J].IEEE Transactions on Intelligent Transportation Systems,2013,14(2):871-882.
[8]YIN Rongwang.Network traffic predicting model based on FCM-LSSVM[J].Computer Engineering and Applications,2016,52(1):105-109(in Chinese).[殷荣网.基于FCM-LSSVM网络流量预测模型[J].计算机工程与应用,2016,52(1):105-109.]
[9]Mohan BR.Resource usage prediction based on ARIMA-ARCHmodel for virtualized server system[J].International Journal of Geomate,2017,12(33):139-146.
[10]Mohamadi S,Amindavar H,Hosseini SM Ali Tayaranian.ARIMA-GARCH modeling for epileptic seizure prediction[C]//IEEE International Conference on Acoustics,Speech and Signal Processing,2017:994-998.
[11]Avnimelech R,Intrator N.Boosting regression estimators[J].Neural Computation,2014,11(2):499-520.
[12]Wang Z,Srinivasan RS.A review of artificial intelligence based building energy prediction with a focus on ensemble prediction models[C]//Winter Simulation Conference.IEEE,2016:3438-3448.
[13]Rahman MM,Islam MM,Murase K,et al.Layered ensemble architecture for time series forecasting[J].IEEE Transactions on Cybernetics,2016,46(1):270-283.
[14]ZHOU Zhihua.Maching learning[M].Beijing:Tsinghua University Press,2016:171-173(in Chinese).[周志华.机器学习[M].北京:清华大学出版社,2016:171-173.]
[15]Tang J,Deng C,Huang GB.Extreme learning machine for multilayer perceptron[J].IEEE Transactions on Neural Networks&Learning Systems,2017,27(4):809-821.
[16]Timeseries forecasting competition for neural networks and computational intelligence[EB/OL].[2017-12-20].http://www.neural-forecasting-competition.com/NN3.
[17]Chen L.A modified Levenberg-Marquardt method with line search for nonlinear equations[J].Computational Optimization&Applications,2016,65(3):753-779.
[18]Gao B,Xu Z.Phaseless recovery using the Gauss-Newton method[J].IEEE Transactions on Signal Processing,2017,65(22):5885-5896.