Channel, Trend and Periodic-Wise Representation Learning for Multivariate Long-term Time Series Forecasting

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Song, Zhangyao, Jiang, Nanqing, He, Miaohong, Zhao, Xiaoyu, Guo, Tao
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866915738820804608
author Song, Zhangyao
Jiang, Nanqing
He, Miaohong
Zhao, Xiaoyu
Guo, Tao
author_facet Song, Zhangyao
Jiang, Nanqing
He, Miaohong
Zhao, Xiaoyu
Guo, Tao
contents Downsampling-based methods for time series forecasting have attracted increasing attention due to their superiority in capturing sequence trends. However, this approaches mainly capture dependencies within subsequences but neglect inter-subsequence and inter-channel interactions, which limits forecasting accuracy. To address these limitations, we propose CTPNet, a novel framework that explicitly learns representations from three perspectives: i) inter-channel dependencies, captured by a temporal query-based multi-head attention mechanism; ii) intra-subsequence dependencies, modeled via a Transformer to characterize trend variations; and iii) inter-subsequence dependencies, extracted by reusing the encoder with residual connections to capture global periodic patterns. By jointly integrating these levels, proposed method provides a more holistic representation of temporal dynamics. Extensive experiments demonstrate the superiority of the proposed method.
format Preprint
id arxiv_https___arxiv_org_abs_2509_23583
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Channel, Trend and Periodic-Wise Representation Learning for Multivariate Long-term Time Series Forecasting
Song, Zhangyao
Jiang, Nanqing
He, Miaohong
Zhao, Xiaoyu
Guo, Tao
Computational Engineering, Finance, and Science
Downsampling-based methods for time series forecasting have attracted increasing attention due to their superiority in capturing sequence trends. However, this approaches mainly capture dependencies within subsequences but neglect inter-subsequence and inter-channel interactions, which limits forecasting accuracy. To address these limitations, we propose CTPNet, a novel framework that explicitly learns representations from three perspectives: i) inter-channel dependencies, captured by a temporal query-based multi-head attention mechanism; ii) intra-subsequence dependencies, modeled via a Transformer to characterize trend variations; and iii) inter-subsequence dependencies, extracted by reusing the encoder with residual connections to capture global periodic patterns. By jointly integrating these levels, proposed method provides a more holistic representation of temporal dynamics. Extensive experiments demonstrate the superiority of the proposed method.
title Channel, Trend and Periodic-Wise Representation Learning for Multivariate Long-term Time Series Forecasting
topic Computational Engineering, Finance, and Science
url https://arxiv.org/abs/2509.23583