EUVIMEDEUVIMED entdecken
Crossref

OccamVTS: Distilling Vision Models to 1% Parameters for Time Series Forecasting

Sisuo Lyu, Siru Zhong, Weilin Ruan, Qingxiang Liu, Qingsong Wen, Hui Xiong, Yuxuan Liang

Proceedings of the AAAI Conference on Artificial Intelligence · 14.03.2026 · 40 (29) · 24216-24225

DOI: 10.1609/aaai.v40i29.39601 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Proceedings of the AAAI Conference on Artificial Intelligence, online erschienen am 14.03.2026. Verfasst von Sisuo Lyu und 6 weiteren Autor:innen. Die Arbeit wurde bisher 5-mal zitiert (Crossref).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Time series forecasting is fundamental to diverse applications, with recent approaches leverage large vision models (LVMs) to capture temporal patterns through visual representations. We reveal that while vision models enhance forecasting performance, 99% of their parameters are unnecessary for time series tasks. Through cross-modal analysis, we find that time series align with low-level textural features but not high-level semantics, which can impair forecasting accuracy. We propose OccamVTS, a knowledge distillation framework that extracts only the essential 1% of predictive information from LVMs into lightweight networks. Using pre-trained LVMs as privileged teachers, OccamVTS employs pyramid-style feature alignment combined with correlation and feature distillation to transfer beneficial patterns while filtering out semantic noise. Counterintuitively, this aggressive parameter reduction improves accuracy by eliminating overfitting to irrelevant visual features while preserving essential temporal patterns. Extensive experiments across multiple benchmark datasets demonstrate that OccamVTS consistently achieves state-of-the-art performance with only 1% of the original parameters, particularly excelling in few-shot and zero-shot scenarios.

Zitieren

APA

Lyu, S., Zhong, S., Ruan, W., Liu, Q., Wen, Q., Xiong, H., & Liang, Y. (2026). OccamVTS: Distilling Vision Models to 1% Parameters for Time Series Forecasting. Proceedings of the AAAI Conference on Artificial Intelligence, 40(29), 24216-24225. https://doi.org/10.1609/aaai.v40i29.39601

Vancouver

Lyu S, Zhong S, Ruan W, Liu Q, Wen Q, Xiong H, et al. OccamVTS: Distilling Vision Models to 1% Parameters for Time Series Forecasting. Proceedings of the AAAI Conference on Artificial Intelligence. 2026;40(29):24216-24225. doi:10.1609/aaai.v40i29.39601

BibTeX
@article{Lyu2026,
  author = {Lyu, Sisuo and Zhong, Siru and Ruan, Weilin and Liu, Qingxiang and Wen, Qingsong and Xiong, Hui and Liang, Yuxuan},
  title = {OccamVTS: Distilling Vision Models to 1% Parameters for Time Series Forecasting},
  journal = {Proceedings of the AAAI Conference on Artificial Intelligence},
  year = {2026},
  volume = {40},
  number = {29},
  pages = {24216--24225},
  doi = {10.1609/aaai.v40i29.39601}
}
Kostenlos starten →