LLM-Mixer: Multiscale Mixing in LLMs for Time Series Forecasting

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Kowsher, Md, Sobuj, Md. Shohanur Islam, Prottasha, Nusrat Jahan, Alanis, E. Alejandro, Garibay, Ozlem Ozmen, Yousefi, Niloofar
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866918042421690368
author Kowsher, Md
Sobuj, Md. Shohanur Islam
Prottasha, Nusrat Jahan
Alanis, E. Alejandro
Garibay, Ozlem Ozmen
Yousefi, Niloofar
author_facet Kowsher, Md
Sobuj, Md. Shohanur Islam
Prottasha, Nusrat Jahan
Alanis, E. Alejandro
Garibay, Ozlem Ozmen
Yousefi, Niloofar
contents Time series forecasting remains a challenging task, particularly in the context of complex multiscale temporal patterns. This study presents LLM-Mixer, a framework that improves forecasting accuracy through the combination of multiscale time-series decomposition with pre-trained LLMs (Large Language Models). LLM-Mixer captures both short-term fluctuations and long-term trends by decomposing the data into multiple temporal resolutions and processing them with a frozen LLM, guided by a textual prompt specifically designed for time-series data. Extensive experiments conducted on multivariate and univariate datasets demonstrate that LLM-Mixer achieves competitive performance, outperforming recent state-of-the-art models across various forecasting horizons. This work highlights the potential of combining multiscale analysis and LLMs for effective and scalable time-series forecasting.
format Preprint
id arxiv_https___arxiv_org_abs_2410_11674
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle LLM-Mixer: Multiscale Mixing in LLMs for Time Series Forecasting
Kowsher, Md
Sobuj, Md. Shohanur Islam
Prottasha, Nusrat Jahan
Alanis, E. Alejandro
Garibay, Ozlem Ozmen
Yousefi, Niloofar
Machine Learning
Computation and Language
Time series forecasting remains a challenging task, particularly in the context of complex multiscale temporal patterns. This study presents LLM-Mixer, a framework that improves forecasting accuracy through the combination of multiscale time-series decomposition with pre-trained LLMs (Large Language Models). LLM-Mixer captures both short-term fluctuations and long-term trends by decomposing the data into multiple temporal resolutions and processing them with a frozen LLM, guided by a textual prompt specifically designed for time-series data. Extensive experiments conducted on multivariate and univariate datasets demonstrate that LLM-Mixer achieves competitive performance, outperforming recent state-of-the-art models across various forecasting horizons. This work highlights the potential of combining multiscale analysis and LLMs for effective and scalable time-series forecasting.
title LLM-Mixer: Multiscale Mixing in LLMs for Time Series Forecasting
topic Machine Learning
Computation and Language
url https://arxiv.org/abs/2410.11674