TFformer: A time-frequency domain bidirectional sequence-level attention based transformer for interpretable long-term sequence forecasting

Tianlong Zhao, Lexin Fang, Xiang Ma, Xue-mei Li, Caiming Zhang

Abstract: Highlights•A new time–frequency domain bidirectional sequence-level attention is proposed.•A seasonal trend decomposition forecasting model is proposed.•A pair of temporal pattern extraction and reconstruction block is proposed.•A prediction model based on periodic extension is proposed.•A new forecast method with high interpretability and low complexity is proposed.