WFTNet: Exploiting Global and Local Periodicity in Long-term Time Series Forecasting

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Liu, Peiyuan, Wu, Beiliang, Li, Naiqi, Dai, Tao, Lei, Fengmao, Bao, Jigang, Jiang, Yong, Xia, Shu-Tao
Natura: Preprint
Pubblicazione: 2023
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866910286971142144
author Liu, Peiyuan
Wu, Beiliang
Li, Naiqi
Dai, Tao
Lei, Fengmao
Bao, Jigang
Jiang, Yong
Xia, Shu-Tao
author_facet Liu, Peiyuan
Wu, Beiliang
Li, Naiqi
Dai, Tao
Lei, Fengmao
Bao, Jigang
Jiang, Yong
Xia, Shu-Tao
contents Recent CNN and Transformer-based models tried to utilize frequency and periodicity information for long-term time series forecasting. However, most existing work is based on Fourier transform, which cannot capture fine-grained and local frequency structure. In this paper, we propose a Wavelet-Fourier Transform Network (WFTNet) for long-term time series forecasting. WFTNet utilizes both Fourier and wavelet transforms to extract comprehensive temporal-frequency information from the signal, where Fourier transform captures the global periodic patterns and wavelet transform captures the local ones. Furthermore, we introduce a Periodicity-Weighted Coefficient (PWC) to adaptively balance the importance of global and local frequency patterns. Extensive experiments on various time series datasets show that WFTNet consistently outperforms other state-of-the-art baseline. Code is available at https://github.com/Hank0626/WFTNet.
format Preprint
id arxiv_https___arxiv_org_abs_2309_11319
institution arXiv
publishDate 2023
record_format arxiv
spellingShingle WFTNet: Exploiting Global and Local Periodicity in Long-term Time Series Forecasting
Liu, Peiyuan
Wu, Beiliang
Li, Naiqi
Dai, Tao
Lei, Fengmao
Bao, Jigang
Jiang, Yong
Xia, Shu-Tao
Machine Learning
Recent CNN and Transformer-based models tried to utilize frequency and periodicity information for long-term time series forecasting. However, most existing work is based on Fourier transform, which cannot capture fine-grained and local frequency structure. In this paper, we propose a Wavelet-Fourier Transform Network (WFTNet) for long-term time series forecasting. WFTNet utilizes both Fourier and wavelet transforms to extract comprehensive temporal-frequency information from the signal, where Fourier transform captures the global periodic patterns and wavelet transform captures the local ones. Furthermore, we introduce a Periodicity-Weighted Coefficient (PWC) to adaptively balance the importance of global and local frequency patterns. Extensive experiments on various time series datasets show that WFTNet consistently outperforms other state-of-the-art baseline. Code is available at https://github.com/Hank0626/WFTNet.
title WFTNet: Exploiting Global and Local Periodicity in Long-term Time Series Forecasting
topic Machine Learning
url https://arxiv.org/abs/2309.11319