UnetTSF: A Better Performance Linear Complexity Time Series Prediction Model

Li Chu, Xiao Bingjia, Yuan Qiping
1/5/2024
cs.LG

No official code found. Best implementation: lichuustc/UnetTSF (56 stars, last commit 2024-01-09).

Abstract

Recently, Transformer-base models have made significant progress in the field of time series prediction which have achieved good results and become baseline models beyond Dlinear. The paper proposes an U-Net time series prediction model (UnetTSF) with linear complexity, which adopts the U-Net architecture. We are the first to use FPN technology to extract features from time series data, replacing the method of decomposing time series data into trend and seasonal terms, while designing a fusion structure suitable for time series data. After testing on 8 open-source datasets, compared to the best linear model DLiner. Out of 32 testing projects, 31 achieved the best results. The average decrease in mse is 10.1%, while the average decrease in mae is 9.1%. Compared with the complex transformer-base PatchTST, UnetTSF obtained 9 optimal results for mse and 15 optimal results for mae in 32 testing projects.

View on arXivView PDF

Code Implementations(1)

lichuustc/UnetTSFCommunity100%

UnetTSF: A Better Performance Linear Complexity Time Series Prediction Model

569Shell, PythonJan 5, 20242 years ago

Similar Papers

FPN-fusion: Enhanced Linear Complexity Time Series Forecasting Model

90% match2024

U-shaped Transformer: Retain High Frequency Context in Time Series Analysis

85% match2023

Bridging Simplicity and Sophistication using GLinear: A Novel Architecture for Enhanced Time Series Prediction

83% match1 repo62025

A Unified Frequency Domain Decomposition Framework for Interpretable and Robust Time Series Forecasting

82% match2025

Kernel-U-Net: Multivariate Time Series Forecasting using Custom Kernels

82% match2024

A Temporal Linear Network for Time Series Forecasting

82% match1 repo12024

Cite this paper

@article{chu2024unettsf,
  title  = {UnetTSF: A Better Performance Linear Complexity Time Series Prediction Model},
  author = {Li Chu and Xiao Bingjia and Yuan Qiping},
  year   = {2024},
  eprint = {2401.03001},
  archivePrefix = {arXiv},
  url    = {http://arxiv.org/abs/2401.03001v1}
}

Add a badge to your README

Shows whether official code is linked for this paper.

Code of Paper badge
[![Code of Paper](https://codeofpaper.com/badge/2401.03001.svg)](https://codeofpaper.com/paper/2401.03001)

Discussion