Enabling Time-series Foundation Model for Building Energy Forecasting via Contrastive Curriculum Learning
This work addresses the adaptation of foundation models to building energy forecasting, an incremental improvement for energy management applications.
The paper tackled the poor performance of existing time-series foundation models in building energy forecasting by proposing a contrastive curriculum learning method to optimize training data ordering, resulting in a 14.6% improvement in zero/few-shot performance.
Advances in time-series forecasting are driving a shift from conventional machine learning models to foundation models (FMs) that are trained with generalized knowledge. However, existing FMs still perform poorly in the energy fields, such as building energy forecasting (BEF). This paper studies the adaptation of FM to BEF tasks. We demonstrate the shortcomings of fine-tuning FM straightforwardly from both the perspectives of FM and the data. To overcome these limitations, we propose a new \textit{contrastive curriculum learning}-based training method. Our method optimizes the ordering of training data in the context of TSFM adaptation. Experiments show that our method can improve the zero/few-shot performance by 14.6\% compared to the existing FMs. Our code and new TSFM will be available at <Anonymous Github Repo>.