The repo is the official implementation for the paper: Diffusion-based Decoupled Deterministic and Uncertain Framework for Probabilistic MTS Forecasting
Train and evaluate the model. We provide all the above tasks under the folder ./scripts/. You can reproduce the results as the following examples:
# D3U Multivariate Probabilistic forecasting where SparseVQ functions as the conditioning network and PatchDN serves as the denoising
network.
bash ./scripts/SVQ/exp_study/
Place the checkpoint of the already trained point prediction model into this folder. For example, path: pretrain_checkpoints/SVQ/all/weather/192/checkpoint.pth
The datasets can be obtained from Google Drive or Baidu Cloud.
If you have any questions or want to use the code, feel free to contact:
- Qi Li: li.q@bupt.edu.cn
- Zhenyu Zhang: zhangzhenyucad@bupt.edu.cn
If you find this repo helpful, please cite our paper.
@inproceedings{li2025diffusion,
title={Diffusion-based decoupled deterministic and uncertain framework for probabilistic multivariate time series forecasting},
author={Li, Qi and Zhang, Zhenyu and Yao, Lei and Li, Zhaoxia and Zhong, Tianyi and Zhang, Yong},
booktitle={The Thirteenth International Conference on Learning Representations},
year={2025}
}