This repository is the official implementation of Neural Laplace Control for Continuous-time Delayed Systems.

1. Run/Follow steps in install.sh
2. Replicate experimental results by running and configuring run_exp_multi.py.

   python run_exp_multi.py

3. Process the output log file using process_logs.py by updating the LOG_PATH variable to point to the recently generated log file.

   python process_results/process_logs.py

To retrain all models from scratch (much slower), set the following variables to True in run_exp_multi.py before running it:

RETRAIN = True
FORCE_RETRAIN = True

To obtain large files like saved models for this work, please download these from Google Drive here and place them into corresponding directories.

• 💻 Neural Laplace: Neural Laplace: Differentiable Laplace Reconstructions for modelling any time observation with O(1) complexity.

If you use Neural Laplace Control in your research, please cite it as follows:

@inproceedings{holt2023neural,
  title={Neural Laplace Control for Continuous-time Delayed Systems},
  author={Holt, Samuel and H{\"u}y{\"u}k, Alihan and Qian, Zhaozhi and Sun, Hao and van der Schaar, Mihaela},
  booktitle={International Conference on Artificial Intelligence and Statistics},
  pages={1747--1778},
  year={2023},
  organization={PMLR}
}