Lightweight Online Adaption for Time Series Foundation Model Forecasts

Thomas L Lee, William Toner, Rajkarn Singh, Artjom Joosen, Martin Asenov
Proceedings of the 42nd International Conference on Machine Learning, PMLR 267:33736-33764, 2025.

Abstract

Foundation models (FMs) have emerged as a promising approach for time series forecasting. While effective, FMs typically remain fixed during deployment due to the high computational costs of learning them online. Consequently, deployed FMs fail to adapt their forecasts to current data characteristics, despite the availability of online feedback from newly arriving data. This raises the question of whether FM performance can be enhanced by the efficient usage of this feedback. We propose ELF to answer this question. ELF is a lightweight mechanism for the online adaption of FM forecasts in response to online feedback. ELF consists of two parts: a) the ELF-Forecaster which is used to learn the current data distribution; and b) the ELF-Weighter which is used to combine the forecasts of the FM and the ELF-Forecaster. We evaluate the performance of ELF in conjunction with several recent FMs across a suite of standard time series datasets. In all of our experiments we find that using ELF improves performance. This work demonstrates how efficient usage of online feedback can be used to improve FM forecasts.

Cite this Paper

BibTeX
@InProceedings{pmlr-v267-lee25ag, title = {Lightweight Online Adaption for Time Series Foundation Model Forecasts}, author = {Lee, Thomas L and Toner, William and Singh, Rajkarn and Joosen, Artjom and Asenov, Martin}, booktitle = {Proceedings of the 42nd International Conference on Machine Learning}, pages = {33736--33764}, year = {2025}, editor = {Singh, Aarti and Fazel, Maryam and Hsu, Daniel and Lacoste-Julien, Simon and Berkenkamp, Felix and Maharaj, Tegan and Wagstaff, Kiri and Zhu, Jerry}, volume = {267}, series = {Proceedings of Machine Learning Research}, month = {13--19 Jul}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v267/main/assets/lee25ag/lee25ag.pdf}, url = {https://proceedings.mlr.press/v267/lee25ag.html}, abstract = {Foundation models (FMs) have emerged as a promising approach for time series forecasting. While effective, FMs typically remain fixed during deployment due to the high computational costs of learning them online. Consequently, deployed FMs fail to adapt their forecasts to current data characteristics, despite the availability of online feedback from newly arriving data. This raises the question of whether FM performance can be enhanced by the efficient usage of this feedback. We propose ELF to answer this question. ELF is a lightweight mechanism for the online adaption of FM forecasts in response to online feedback. ELF consists of two parts: a) the ELF-Forecaster which is used to learn the current data distribution; and b) the ELF-Weighter which is used to combine the forecasts of the FM and the ELF-Forecaster. We evaluate the performance of ELF in conjunction with several recent FMs across a suite of standard time series datasets. In all of our experiments we find that using ELF improves performance. This work demonstrates how efficient usage of online feedback can be used to improve FM forecasts.} }
Endnote
%0 Conference Paper %T Lightweight Online Adaption for Time Series Foundation Model Forecasts %A Thomas L Lee %A William Toner %A Rajkarn Singh %A Artjom Joosen %A Martin Asenov %B Proceedings of the 42nd International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2025 %E Aarti Singh %E Maryam Fazel %E Daniel Hsu %E Simon Lacoste-Julien %E Felix Berkenkamp %E Tegan Maharaj %E Kiri Wagstaff %E Jerry Zhu %F pmlr-v267-lee25ag %I PMLR %P 33736--33764 %U https://proceedings.mlr.press/v267/lee25ag.html %V 267 %X Foundation models (FMs) have emerged as a promising approach for time series forecasting. While effective, FMs typically remain fixed during deployment due to the high computational costs of learning them online. Consequently, deployed FMs fail to adapt their forecasts to current data characteristics, despite the availability of online feedback from newly arriving data. This raises the question of whether FM performance can be enhanced by the efficient usage of this feedback. We propose ELF to answer this question. ELF is a lightweight mechanism for the online adaption of FM forecasts in response to online feedback. ELF consists of two parts: a) the ELF-Forecaster which is used to learn the current data distribution; and b) the ELF-Weighter which is used to combine the forecasts of the FM and the ELF-Forecaster. We evaluate the performance of ELF in conjunction with several recent FMs across a suite of standard time series datasets. In all of our experiments we find that using ELF improves performance. This work demonstrates how efficient usage of online feedback can be used to improve FM forecasts.
APA
Lee, T.L., Toner, W., Singh, R., Joosen, A. & Asenov, M.. (2025). Lightweight Online Adaption for Time Series Foundation Model Forecasts. Proceedings of the 42nd International Conference on Machine Learning, in Proceedings of Machine Learning Research 267:33736-33764 Available from https://proceedings.mlr.press/v267/lee25ag.html.

Related Material