PCN: Patch segmentation convolutional networks in time series forecasting tasks

Jinfeng Wang

doi:10.1016/j.knosys.2025.114900

PCN: Patch segmentation convolutional networks in time series forecasting tasks

2025-11-22

0

PRE

AI

DOI：10.1016/j.knosys.2025.114900

Original

Original request for help

Save

Abstract

En 中文

• A patch segmentation convolutional network (PCN) is proposed to more effectively capture temporal dependencies. • Three patch-adaptive convolutional modules are introduced to exploit temporal patterns from multiple perspectives. • A self-supervised learning framework based on PCN is developed to enhance model generalization. • A comprehensive experimental study validates the strengths of PCN.

PCN: Patch segmentation convolutional networks in time series forecasting tasks

Abstract

Journal

Organization