VisionTS: Visual Masked Autoencoders Are Free-Lunch Zero-Shot Time Series Forecasters
This provides a novel, cross-modality approach to time series forecasting that could reduce data and computational needs, though it may be incremental in leveraging existing visual models.
The paper tackles time series forecasting by repurposing a visual masked autoencoder pre-trained on ImageNet as a zero-shot forecaster, achieving better performance than existing foundation models without adaptation and state-of-the-art results with fine-tuning.
Foundation models have emerged as a promising approach in time series forecasting (TSF). Existing approaches either repurpose large language models (LLMs) or build large-scale time series datasets to develop TSF foundation models for universal forecasting. However, these methods face challenges due to the severe cross-domain gap or in-domain heterogeneity. This paper explores a new road to building a TSF foundation model from rich, high-quality natural images. Our key insight is that a visual masked autoencoder, pre-trained on the ImageNet dataset, can naturally be a numeric series forecaster. By reformulating TSF as an image reconstruction task, we bridge the gap between image pre-training and TSF downstream tasks. Surprisingly, without further adaptation in the time series domain, the proposed VisionTS could achieve better zero-shot forecast performance than existing TSF foundation models. With fine-tuning for one epoch, VisionTS could further improve the forecasting and achieve state-of-the-art performance in most cases. Extensive experiments reveal intrinsic similarities between images and real-world time series, suggesting that visual models may offer a "free lunch" for TSF and highlight the potential for future cross-modality research. Our code is publicly available at https://github.com/Keytoyze/VisionTS.