ActSTD

The official implementation of: Activity Trajectory Generation via Modeling Spatiotemporal Dynamics (KDD '22).

Installation

Environment

Tested OS: Linux
Python >= 3.7
PyTorch == 1.7.1

Dependencies

Install PyTorch 1.7.1 with the correct CUDA version.
Use the pip install -r requirements. txt command to install all of the Python modules and packages used in this project.

run python src/setup.py build_ext --inplace to create the shared object file in the current directory.

Model Training

cd src

Use the following command to train ActSTD on Mobile dataset with different CNF models:

python app.py --data Mobile --model attncnf --tpp neural --l2_attn --ode_method 'scipy_solver' --ode_solver 'RK45' --cuda_id 0 --tpp_style 'gru' --weekhour

python app.py --data Mobile --model jumpcnf --tpp neural --solve_reverse --ode_method 'scipy_solver' --ode_solver 'RK45' --cuda_id 0 --tpp_style 'gru' --weekhour

Use the following command to train ActSTD on Foursquare dataset with different CNF models:

python app.py --data Foursquare --model attncnf --tpp neural --l2_attn --ode_method 'scipy_solver' --ode_solver 'RK45' --cuda_id 0 --tpp_style 'gru' --weekhour

python app.py --data Foursquare --model jumpcnf --tpp neural --solve_reverse --ode_method 'scipy_solver' --ode_solver 'RK45' --cuda_id 0 --tpp_style 'gru' --weekhour

More Related Works

Learning to Simulate Daily Activities via Modeling Dynamic Human Needs (WWW'23)

Citation

If you find this repository useful in your research, please consider citing the following paper:

@inproceedings{yuan2022activity,
  title={Activity Trajectory Generation via Modeling Spatiotemporal Dynamics},
  author={Yuan, Yuan and Ding, Jingtao and Wang, Huandong and Jin, Depeng and Li, Yong},
  booktitle={Proceedings of the 28th ACM SIGKDD Conference on Knowledge Discovery and Data Mining},
  pages={4752--4762},
  year={2022}
}

Note

The implemention is based on Neural STPP.

tsinghua-fib-lab/Activity-Trajectory-Generation