Multivariate Time Series Prediction Based on Temporal Change Information Learning Method
- 4 January 2022
- journal article
- research article
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Neural Networks and Learning Systems
- Vol. 34 (10) , 7034-7048
Abstract
In the multivariate time series prediction tasks, the impact information of all nonpredictive time series on the predictive target series is difficult to be extracted at different time stages. Through the emphasis on optimal-related sequences in the target series, the deep learning model with the attention mechanism achieves a good predictive performance. However, temporal change information in the objective function and optimization algorithm is completely ignored in these models. To this end, a temporal change information learning (CIL) method is proposed in this article. First, mean absolute error (MAE) and mean squared error (MSE) losses are contained in the objective function to evaluate different amplitude errors. Meanwhile, the second-order difference technology is used in the correlation terms of the objective function to adaptively capture the impact of the abrupt and slow change information in each series on the target series. Second, the long short-term memory (LSTM) network with the transformation mechanism is used in the method so that temporal dependence information can be fully extracted (i.e., avoiding the supersaturation region). Third, to effectively obtain the optimal model parameters, the current and historical moment estimation information is adaptively memorized without the introduction of additional hyperparameters, and therefore, the acquisition ability of temporal change information in the error gradient flow is greatly enhanced by the proposed optimization algorithm. Finally, three datasets with different scales are used to verify the advantages of the CIL method in computational overhead and prediction effect.Keywords
Funding Information
- Natural Science Foundation of Hunan Province of China (2021JJ30136)
- Changsha Science and Technology Plan Project (kq2004015)
This publication has 30 references indexed in Scilit:
- A Dual-Stage Attention-Based Recurrent Neural Network for Time Series Prediction2017
- Assessing energy forecasting inaccuracy by simultaneously considering temporal and absolute errorsEnergy Conversion and Management, 2017
- Assessing Beijing's PM 2.5 pollution: severity, weather impact, APEC and winter heatingProceedings of the Royal Society A: Mathematical, Physical and Engineering Sciences, 2015
- A Combined First and Second Order Variational Approach for Image ReconstructionJournal of Mathematical Imaging and Vision, 2013
- Time series prediction using RNN in multi-dimension embedding phase space2002
- The digital TV filter and nonlinear denoisingIEEE Transactions on Image Processing, 2001
- Long Short-Term MemoryNeural Computation, 1997
- Dynamic programming algorithm optimization for spoken word recognitionIEEE Transactions on Acoustics, Speech, and Signal Processing, 1978
- Distribution of Residual Autocorrelations in Autoregressive-Integrated Moving Average Time Series ModelsJournal of the American Statistical Association, 1970
- CONTINUOUS INSPECTION SCHEMESBiometrika, 1954