We improved the preprint study by adding confidence-aware forecasting option. Cite both of the paper if you use LMSAutoTSF or LMSAutoTSFV2.

EAAI paper

@article{delibasoglu2026learnable,
  title={Learnable multi-scale decomposition with integrated autocorrelation and confidence-aware time series forecasting},
  author={Delibasoglu, Ibrahim and Chakraborty, Sanjay and Heintz, Fredrik},
  journal={Engineering Applications of Artificial Intelligence},
  volume={181},
  pages={115449},
  year={2026},
  publisher={Elsevier}
}

Arxiv paper

article{delibasoglu2024lms,
  title={LMS-AutoTSF: Learnable Multi-Scale Decomposition and Integrated Autocorrelation for Time Series Forecasting},
  author={Delibasoglu, Ibrahim and Chakraborty, Sanjay and Heintz, Fredrik},
  journal={arXiv preprint arXiv:2412.06866},
  year={2024}
}

Learning Temporal Saliency for Time Series Forecasting with Cross-Scale Attention

CrossScaleNet

@article{delibasoglu2025learning,
  title={Learning Temporal Saliency for Time Series Forecasting with Cross-Scale Attention},
  author={Delibasoglu, Ibrahim and Heintz, Fredrik},
  journal={arXiv preprint arXiv:2509.22839},
  year={2025}
}

Results

Performanced comparison on public benchmark datasets

Acknowledgement

We appreciate the following GitHub repositories:

https://github.com/thuml/Time-Series-Library