Lightweight Online Adaption for Time Series Foundation Model Forecasts
This work addresses the challenge of adapting time series foundation models to current data characteristics efficiently, which is incremental as it builds on existing FMs with a novel adaptation method.
The paper tackles the problem of foundation models (FMs) failing to adapt to new data during deployment due to high computational costs, and proposes ELF, a lightweight mechanism for online adaptation that improves FM forecasts across standard time series datasets.
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.