LightGTS: A Lightweight General Time Series Forecasting Model

TL;DR

LightGTS is a lightweight, efficient time series forecasting model that leverages periodic patterns to outperform larger models across multiple benchmarks without heavy computational requirements.

Contribution

The paper introduces Periodical Tokenization and Periodical Parallel Decoding, novel techniques that enable a lightweight model to effectively utilize periodicity in time series forecasting.

Findings

Achieves state-of-the-art performance on 9 benchmarks.

Operates efficiently in resource-constrained scenarios.

Outperforms existing foundation models in both zero-shot and full-shot settings.

Abstract

Existing works on general time series forecasting build foundation models with heavy model parameters through large-scale multi-source pre-training. These models achieve superior generalization ability across various datasets at the cost of significant computational burdens and limitations in resource-constrained scenarios. This paper introduces LightGTS, a lightweight general time series forecasting model designed from the perspective of consistent periodical modeling. To handle diverse scales and intrinsic periods in multi-source pre-training, we introduce Periodical Tokenization, which extracts consistent periodic patterns across different datasets with varying scales. To better utilize the periodicity in the decoding process, we further introduce Periodical Parallel Decoding, which leverages historical tokens to improve forecasting. Based on the two techniques above which fully leverage the inductive bias of periods inherent in time series, LightGTS uses a lightweight model to achieve outstanding performance on general time series forecasting. It achieves state-of-the-art forecasting performance on 9 real-world benchmarks in both zero-shot and full-shot settings with much better efficiency compared with existing time series foundation models.