Hierarchical Attention-Based Recurrent Highway Networks for Time Series Prediction

TL;DR

This paper introduces HRHN, a deep learning model that effectively captures complex interactions and temporal dynamics in exogenous data for improved time series prediction, outperforming existing methods especially in sudden change scenarios.

Contribution

The paper presents a novel hierarchical attention-based recurrent highway network that models spatio-temporal features and exogenous data interactions for time series forecasting.

Findings

HRHN outperforms state-of-the-art methods in various datasets.

HRHN effectively captures sudden changes and oscillations.

Hierarchical attention improves feature relevance selection.

Abstract

Time series prediction has been studied in a variety of domains. However, it is still challenging to predict future series given historical observations and past exogenous data. Existing methods either fail to consider the interactions among different components of exogenous variables which may affect the prediction accuracy, or cannot model the correlations between exogenous data and target data. Besides, the inherent temporal dynamics of exogenous data are also related to the target series prediction, and thus should be considered as well. To address these issues, we propose an end-to-end deep learning model, i.e., Hierarchical attention-based Recurrent Highway Network (HRHN), which incorporates spatio-temporal feature extraction of exogenous variables and temporal dynamics modeling of target variables into a single framework. Moreover, by introducing the hierarchical attention mechanism, HRHN can adaptively select the relevant exogenous features in different semantic levels. We carry out comprehensive empirical evaluations with various methods over several datasets, and show that HRHN outperforms the state of the arts in time series prediction, especially in capturing sudden changes and sudden oscillations of time series.