EUVIMEDEUVIMED entdecken
Crossref

TimeCAP: A Channel-Aware Pre-Training Framework for Multivariate Time Series Forecasting

Chuanru Ren, Yao Lu, Tianjin Huang, Haowen Zheng, Hengde Zhu, Yunyin Li, Hengxiao Li, Lu Liu

Proceedings of the AAAI Conference on Artificial Intelligence · 14.03.2026 · 40 (30) · 25108-25116

DOI: 10.1609/aaai.v40i30.39700 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Proceedings of the AAAI Conference on Artificial Intelligence, online erschienen am 14.03.2026. Verfasst von Chuanru Ren und 7 weiteren Autor:innen.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Amid recent advances for multivariate time series forecasting, self-supervised learning has emerged as a promising paradigm for deriving transferable knowledge from multi-domain data. Despite its effectiveness, existing approaches exhibit two critical limitations: (1) Underestimating the significance of multivariate dependencies in learning generalizable representations and (2) Failing to reconcile the complementary strengths of autoregressive and one-shot generative paradigms. In this work, we propose TimeCAP, a novel channel-aware pre-training framework that internalizes latent causal relationships among variables inherent in multi-domain data, and effectively transfers the acquired knowledge to downstream applications. Technically, we present a flexible channel-grouping learning approach, complemented by an adaptive meta-routing mechanism, enabling TimeCAP to parallel recognize intra-group local patterns while maintaining global coherence. Intra- and inter-group multivariate dependencies are captured through the self- and cross-attention with channel-aware mask, which strictly confine interactions among time-aligned, fine-grained multivariate tokens. To seamlessly unify two advanced generative paradigms, we propose a novel dynamic dual-head decoding and optimization strategy, empowering TimeCAP to leverage critical dependencies in the output series while avoiding cumulative errors over time. In the few-shot evaluation, TimeCAP achieves average MSE and MAE reductions of 11.8% and 6% over leading baselines, while also outperforming state-of-the-art models in full-shot and zero-shot settings by large margins.

Zitieren

APA

Ren, C., Lu, Y., Huang, T., Zheng, H., Zhu, H., Li, Y., Li, H., & Liu, L. (2026). TimeCAP: A Channel-Aware Pre-Training Framework for Multivariate Time Series Forecasting. Proceedings of the AAAI Conference on Artificial Intelligence, 40(30), 25108-25116. https://doi.org/10.1609/aaai.v40i30.39700

Vancouver

Ren C, Lu Y, Huang T, Zheng H, Zhu H, Li Y, et al. TimeCAP: A Channel-Aware Pre-Training Framework for Multivariate Time Series Forecasting. Proceedings of the AAAI Conference on Artificial Intelligence. 2026;40(30):25108-25116. doi:10.1609/aaai.v40i30.39700

BibTeX
@article{Ren2026,
  author = {Ren, Chuanru and Lu, Yao and Huang, Tianjin and Zheng, Haowen and Zhu, Hengde and Li, Yunyin and Li, Hengxiao and Liu, Lu},
  title = {TimeCAP: A Channel-Aware Pre-Training Framework for Multivariate Time Series Forecasting},
  journal = {Proceedings of the AAAI Conference on Artificial Intelligence},
  year = {2026},
  volume = {40},
  number = {30},
  pages = {25108--25116},
  doi = {10.1609/aaai.v40i30.39700}
}
Kostenlos starten →