A Memory-Network Based Solution for Multivariate Time-Series Forecasting
This work addresses forecasting problems in domains like finance and traffic, but it appears incremental as it builds on existing memory network ideas for time-series data.
The authors tackled the challenge of capturing long-term patterns and dependencies in multivariate time-series forecasting by proposing MTNet, a memory-network based model that achieved competitive results on benchmark datasets, though specific numerical gains were not detailed.
Multivariate time series forecasting is extensively studied throughout the years with ubiquitous applications in areas such as finance, traffic, environment, etc. Still, concerns have been raised on traditional methods for incapable of modeling complex patterns or dependencies lying in real word data. To address such concerns, various deep learning models, mainly Recurrent Neural Network (RNN) based methods, are proposed. Nevertheless, capturing extremely long-term patterns while effectively incorporating information from other variables remains a challenge for time-series forecasting. Furthermore, lack-of-explainability remains one serious drawback for deep neural network models. Inspired by Memory Network proposed for solving the question-answering task, we propose a deep learning based model named Memory Time-series network (MTNet) for time series forecasting. MTNet consists of a large memory component, three separate encoders, and an autoregressive component to train jointly. Additionally, the attention mechanism designed enable MTNet to be highly interpretable. We can easily tell which part of the historic data is referenced the most.