IEEE Trans Cybern. 2022 Sep;52(9):8793-8803. doi: 10.1109/TCYB.2021.3055951. Epub 2022 Aug 18.
Time-series forecasting is a key component in the automation and optimization of intelligent applications. It is not a trivial task, as there are various short-term and/or long-term temporal dependencies. Multiscale modeling has been considered as a promising strategy to solve this problem. However, the existing multiscale models either apply an implicit way to model the temporal dependencies or ignore the interrelationships between multiscale subseries. In this article, we propose a multiscale interactive recurrent network (MiRNN) to jointly capture multiscale patterns. MiRNN employs a deep wavelet decomposition network to decompose the raw time series into multiscale subseries. MiRNN introduces three key strategies (truncation, initialization, and message passing) to model the inherent interrelationships between multiscale subseries, as well as a dual-stage attention mechanism to capture multiscale temporal dependencies. Experiments on four real-world datasets demonstrate that our model achieves promising performance compared with the state-of-the-art methods.
时间序列预测是智能应用自动化和优化的关键组成部分。这不是一项简单的任务，因为存在各种短期和/或长期的时间依赖关系。多尺度建模已被认为是解决这个问题的一种有前途的策略。然而，现有的多尺度模型要么采用隐式方式来建模时间依赖关系，要么忽略多尺度子序列之间的相互关系。在本文中，我们提出了一种多尺度交互递归网络（MiRNN）来联合捕获多尺度模式。MiRNN 采用深度小波分解网络将原始时间序列分解为多尺度子序列。MiRNN 引入了三种关键策略（截断、初始化和消息传递）来建模多尺度子序列之间的固有相互关系，以及双阶段注意力机制来捕获多尺度时间依赖关系。在四个真实数据集上的实验表明，与最先进的方法相比，我们的模型具有有竞争力的性能。
