CASA

The repo is the official implementation for the paper: "CASA: CNN Autoencoder-based Score Attention for Efficient Multivariate Long-term Time-series Forecasting" (IJCAI 2025)

Overall Architecture

CASA regards CNN Autoencoder-based Score Attenton improving channel-wise tokenization and shows Model Agnostic Feature including Computational Efficiency.

<p align="center"> <img src="./figures/CASA.png" alt="" align=center /> </p>

The main result of CASA is as the following:

<p align="center"> <img src="./figures/CASA_main_table.PNG" alt="" align=center /> </p>

Usage

1. Install Pytorch and necessary dependencies.

pip install -r requirements.txt

2. Train and evaluate the model. We provide all the above tasks under the folder ./scripts/. You can reproduce the results as the following examples:

# ECL dataset :  Multivariate forecasting with CASA 
bash ./scripts/long_term_forecast/ECL_script/CASA.sh

Citation

If you find this repo helpful, please cite our paper.

@misc{lee2025casacnnautoencoderbasedscore,
      title={CASA: CNN Autoencoder-based Score Attention for Efficient Multivariate Long-term Time-series Forecasting}, 
      author={Minhyuk Lee and HyeKyung Yoon and MyungJoo Kang},
      year={2025},
      eprint={2505.02011},
      archivePrefix={arXiv},
      primaryClass={cs.LG},
      url={https://arxiv.org/abs/2505.02011}, 
}