WaveForM: Graph Enhanced Wavelet Learning for Long Sequence Forecasting of Multivariate Time Series

• DOI: 10.1609/aaai.v37i9.26276
• Publication Date: 2023-06-27T17:43:07Z
• Authors: Fuhao Yang; Xin Li; Min Wang; Hongyu Zang; Wei Pang; Mingzhong Wang
• Keywords: Sequence (biology); Convolution (computer science)

Multivariate time series (MTS) analysis and forecasting are crucial in many real-world applications, such as smart traffic management weather forecasting. However, most existing work either focuses on short sequence or makes predictions predominantly with domain features, which is not effective at removing noises irregular frequencies MTS. Therefore, we propose WaveForM, an end-to-end graph enhanced Wavelet learning framework for long FORecasting of WaveForM first utilizes Discrete Transform (DWT) to represent MTS the wavelet domain, captures both frequency features a sound theoretical basis. To enable further constructor, learns global relationships between variables, graph-enhanced prediction modules, utilize dilated convolution capture correlations predict coefficients different levels. Extensive experiments five datasets show that our model can achieve considerable performance improvement over lengths against competitive baseline each dataset.