Baguan-TS: A Sequence-Native In-Context Learning Model for Time Series Forecasting with Covariates
Fuente:
arXiv
Salvato in:
| Autori principali: | , , , , , , , , , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2026
|
| Soggetti: | |
| Accesso online: | |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866914405426397184 |
|---|---|
| author | Yang, Linxiao Jiang, Xue Xu, Gezheng Zhou, Tian Yang, Min Zhu, ZhaoYang Geng, Linyuan Zeng, Zhipeng Chen, Qiming Gu, Xinyue Jin, Rong Sun, Liang |
| author_facet | Yang, Linxiao Jiang, Xue Xu, Gezheng Zhou, Tian Yang, Min Zhu, ZhaoYang Geng, Linyuan Zeng, Zhipeng Chen, Qiming Gu, Xinyue Jin, Rong Sun, Liang |
| contents | Transformers enable in-context learning (ICL) for rapid, gradient-free adaptation in time series forecasting, yet most ICL-style approaches rely on tabularized, hand-crafted features, while end-to-end sequence models lack inference-time adaptation. We bridge this gap with a unified framework, Baguan-TS, which integrates the raw-sequence representation learning with ICL, instantiated by a 3D Transformer that attends jointly over temporal, variable, and context axes. To make this high-capacity model practical, we tackle two key hurdles: (i) calibration and training stability, improved with a feature-agnostic, target-space retrieval-based local calibration; and (ii) output oversmoothing, mitigated via context-overfitting strategy. On public benchmark with covariates, Baguan-TS consistently outperforms established baselines, achieving the highest win rate and significant reductions in both point and probabilistic forecasting metrics. Further evaluations across diverse real-world energy datasets demonstrate its robustness, yielding substantial improvements. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2603_17439 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Baguan-TS: A Sequence-Native In-Context Learning Model for Time Series Forecasting with Covariates Yang, Linxiao Jiang, Xue Xu, Gezheng Zhou, Tian Yang, Min Zhu, ZhaoYang Geng, Linyuan Zeng, Zhipeng Chen, Qiming Gu, Xinyue Jin, Rong Sun, Liang Machine Learning Artificial Intelligence Transformers enable in-context learning (ICL) for rapid, gradient-free adaptation in time series forecasting, yet most ICL-style approaches rely on tabularized, hand-crafted features, while end-to-end sequence models lack inference-time adaptation. We bridge this gap with a unified framework, Baguan-TS, which integrates the raw-sequence representation learning with ICL, instantiated by a 3D Transformer that attends jointly over temporal, variable, and context axes. To make this high-capacity model practical, we tackle two key hurdles: (i) calibration and training stability, improved with a feature-agnostic, target-space retrieval-based local calibration; and (ii) output oversmoothing, mitigated via context-overfitting strategy. On public benchmark with covariates, Baguan-TS consistently outperforms established baselines, achieving the highest win rate and significant reductions in both point and probabilistic forecasting metrics. Further evaluations across diverse real-world energy datasets demonstrate its robustness, yielding substantial improvements. |
| title | Baguan-TS: A Sequence-Native In-Context Learning Model for Time Series Forecasting with Covariates |
| topic | Machine Learning Artificial Intelligence |
| url | https://arxiv.org/abs/2603.17439 |