Hierarchical attention network for multivariate time series long-term forecasting Hierarchical Attention Network for Multivariate Time Series Long-term Forecasting

• Published in: Applied intelligence (Dordrecht, Netherlands) Vol. 53; no. 5; pp. 5060 - 5071
• Main Authors: Bi, Hongjing, Lu, Lilei, Meng, Yizhen
• Format: Journal Article
• Language: English
• Published: New York Springer US 01.03.2023; Springer Nature B.V
• Subjects: Artificial Intelligence; Coders; Computer Science; Error reduction; Forecasting; Machines; Manufacturing; Mechanical Engineering; Multivariate analysis; Processes; Recurrent neural networks; Time series; Long-term forecasting; Multivariate time series; Deep neural network; Hierarchical attention; Multi-modal fusion
• ISSN: 0924-669X; 1573-7497; 1573-7497
• DOI: 10.1007/s10489-022-03825-5

Multivariate time series long-term forecasting has always been the subject of research in various fields such as economics, finance, and traffic. In recent years, attention-based recurrent neural networks (RNNs) have received attention due to their ability of reducing error accumulation. However, the existing attention-based RNNs fail to eliminate the negative influence of irrelevant factors on prediction, and ignore the conflict between exogenous factors and target factor. To tackle these problems, we propose a novel Hierarchical Attention Network (HANet) for multivariate time series long-term forecasting. At first, HANet designs a factor-aware attention network (FAN) and uses it as the first component of the encoder. FAN weakens the negative impact of irrelevant exogenous factors on predictions by assigning small weights to them. Then HANet proposes a multi-modal fusion network (MFN) as the second component of the encoder. MFN employs a specially designed multi-modal fusion gate to adaptively select how much information about the expression of current time come from target and exogenous factors. Experiments on two real-world datasets reveal that HANet not only outperforms state-of-the-art methods, but also provides interpretability for prediction.