Breaking Silos: Adaptive Model Fusion Unlocks Better Time Series Forecasting

TL;DR

TimeFuse is a novel adaptive model fusion framework for time series forecasting that dynamically combines multiple models based on input characteristics, leading to improved accuracy across diverse datasets.

Contribution

The paper introduces TimeFuse, a learnable fusion method that adaptively combines heterogeneous models at the sample level for better forecasting performance.

Findings

TimeFuse outperforms individual models on various benchmarks.

The fusor effectively predicts optimal model weights for diverse time series.

Sample-level adaptive fusion improves generalization to unseen data.

Abstract

Time-series forecasting plays a critical role in many real-world applications. Although increasingly powerful models have been developed and achieved superior results on benchmark datasets, through a fine-grained sample-level inspection, we find that (i) no single model consistently outperforms others across different test samples, but instead (ii) each model excels in specific cases. These findings prompt us to explore how to adaptively leverage the distinct strengths of various forecasting models for different samples. We introduce TimeFuse, a framework for collective time-series forecasting with sample-level adaptive fusion of heterogeneous models. TimeFuse utilizes meta-features to characterize input time series and trains a learnable fusor to predict optimal model fusion weights for any given input. The fusor can leverage samples from diverse datasets for joint training, allowing it to adapt to a wide variety of temporal patterns and thus generalize to new inputs, even from unseen datasets. Extensive experiments demonstrate the effectiveness of TimeFuse in various long-/short-term forecasting tasks, achieving near-universal improvement over the state-of-the-art individual models. Code is available at https://github.com/ZhiningLiu1998/TimeFuse.