TFDNet: Time-Frequency Enhanced Decomposed Network for Long-term Time Series Forecasting

TL;DR

TFDNet is a novel neural network that captures long-term patterns and periodicity in time-frequency domains for improved long-term time series forecasting, outperforming existing methods.

Contribution

The paper introduces TFDNet, a multi-scale time-frequency network with separate trend and seasonal blocks, incorporating diverse kernel strategies for multivariate series.

Findings

Outperforms state-of-the-art methods on eight datasets

Effective in capturing long-term and seasonal patterns

Demonstrates superior efficiency in forecasting tasks

Abstract

Long-term time series forecasting is a vital task and has a wide range of real applications. Recent methods focus on capturing the underlying patterns from one single domain (e.g. the time domain or the frequency domain), and have not taken a holistic view to process long-term time series from the time-frequency domains. In this paper, we propose a Time-Frequency Enhanced Decomposed Network (TFDNet) to capture both the long-term underlying patterns and temporal periodicity from the time-frequency domain. In TFDNet, we devise a multi-scale time-frequency enhanced encoder backbone and develop two separate trend and seasonal time-frequency blocks to capture the distinct patterns within the decomposed trend and seasonal components in multi-resolutions. Diverse kernel learning strategies of the kernel operations in time-frequency blocks have been explored, by investigating and incorporating the potential different channel-wise correlation patterns of multivariate time series. Experimental evaluation of eight datasets from five benchmark domains demonstrated that TFDNet is superior to state-of-the-art approaches in both effectiveness and efficiency.