- Weiqi Chen, Wenwei Wang, Bingqing Peng, Qingsong Wen, Tian Zhou, Liang Sun, "Learning to Rotate: Quaternion Transformer for Complicated Periodical Time Series Forecasting" in Proceedings of the 28th ACM SIGKDD Conference on Knowledge Discovery and Data Mining (KDD 2022), 2022. [paper]
Quaternion Transformer (Quatformer) introduce quternion to model complicated periodical patterns (i.e., muliple periods, variable periods, and phase shift) in time series which also has a linear complexity with decoupling attention. Our empirical studies with six benchmark datasets verify its effectiveness.
- Install Python 3.6, PyTorch 1.9.0.
- Install other dependencies by:
pip install -r requirements.txt- Download data. You can obtain all the six benchmarks from [Autoformer] or [Informer].
- Train the model. We provide the experiment scripts of all benchmarks under the folder
./scripts. You can reproduce the experiment results by:
bash ./scripts/ETT_script/Quatformer.sh
bash ./scripts/ECL_script/Quatformer.sh
bash ./scripts/Exchange_script/Quatformer.sh
bash ./scripts/Traffic_script/Quatformer.sh
bash ./scripts/Weather_script/Quatformer.sh
bash ./scripts/ILI_script/Quatformer.shIf you find this repo useful, please cite our paper.
@inproceedings{chen2022quatformer,
title={Learning to Rotate: Quaternion Transformer for Complicated Periodical Time Series Forecasting},
author={Chen, Weiqi and Wang, Wenwei and Peng, Bingqing and Wen, Qingsong and Zhou, Tian and Sun, Liang},
booktitle={Proceedings of the 28th ACM SIGKDD Conference on Knowledge Discovery and Data Mining},
pages={146--156},
year={2022}
}
If you have any question or want to use the code, please contact jarvus.cwq@alibaba-inc.com.
We appreciate the following github repos a lot for their valuable code base or datasets:
https://github.com/thuml/Autoformer
https://github.com/zhouhaoyi/Informer2020