(SIGSPATIAL 2023) DiffSTG: Probabilistic Spatio-Temporal Graph Forecasting with Denoising Diffusion Models

This code is a PyTorch implementation of our SIGSPATIAL'23 paper "DiffSTG: Probabilistic Spatio-Temporal Graph Forecasting with Denoising Diffusion Models". [arXiv]

Citing DiffSTG

🌟 If you find this resource helpful, please consider to star this repository and cite our research:

@inproceedings{wen2023diffstg,
  title={{DiffSTG}: Probabilistic spatio-temporal graph forecasting with denoising diffusion models},
  author={Wen, Haomin and Lin, Youfang and Xia, Yutong and Wan, Huaiyu and Wen, Qingsong and Zimmermann, Roger and Liang, Yuxuan},
  booktitle={the 31st ACM International Conference on Advances in Geographic Information Systems},
  year={2023}
}

Model Architecture

Run

1. requirements:

torch
easydict
nni

2. start training

python train.py

Further Reading

1, Diffusion Model for Time Series and SpatioTemporal Data [GitHub Repo]

2, Large Models for Time Series and Spatio-Temporal Data: A Survey and Outlook [arXiv] [GitHub Repo]

• Authors: Ming Jin, Qingsong Wen*, Yuxuan Liang, Chaoli Zhang, Siqiao Xue, Xue Wang, James Zhang, Yi Wang, Haifeng Chen, Xiaoli Li (IEEE Fellow), Shirui Pan*, Vincent S. Tseng (IEEE Fellow), Yu Zheng (IEEE Fellow), Lei Chen (IEEE Fellow), Hui Xiong (IEEE Fellow)