A Multi-input Time Series Prediction Model Based on CNN-BLSTM
Year: 2021
Author: Ting Xiao
Journal of Information and Computing Science, Vol. 16 (2021), Iss. 1 : pp. 41–51
Abstract
Real time series data sets are often composed of multiple variables. For the future trend of data, not only the historical value of the variables but also other implicit influence factors should be considered. In this paper, a deep neural network prediction model based on multivariable input multi-step output named CNN-BLSTM is proposed. CNN-BLSTM is mainly composed of convolutional neural network (CNN) and bi-directional long short memory network (Bi-LSTM). CNN is used to extract spatial features between variables of multivariate raw data, and Bi-LSTM is used to extract and encode features in time direction. The proposed CNN-BLSTM is able to predict the temperature based on a real-life meteorological data. The experimental results show that the prediction accuracy of the proposed CNN-BLSTM model is significantly better than several state-of-the-art baseline methods.
Journal Article Details
Publisher Name: Global Science Press
Language: English
DOI: https://doi.org/2024-JICS-22378
Journal of Information and Computing Science, Vol. 16 (2021), Iss. 1 : pp. 41–51
Published online: 2021-01
AMS Subject Headings:
Copyright: COPYRIGHT: © Global Science Press
Pages: 11
Keywords: Time series bi-directional long-short term memory convolutional neural network prediction multivariable input.
Author Details
Ting Xiao Email