Requirements | Training | Contributing
This repository is the official implementation of Learning Stable Deep Dynamics Models for Partially Observed or Delayed Dynamical Systems.
All requirements are summarized in the Project.toml file.
This code implements stable Neural DDEs. To reproduce our results simply run the python scripts in src/experiments/:
- scr/experiments/train_ANODE_cos.jl
- scr/experiments/train_NDDE_cos.jl
- src/experiments/train_n_pendulum_LRF_alongtraj.jl
- src/experiments/train_n_pendulum_unstable.jl
- src/experiments/train_inverted_pendulum_LRF_alongtraj.jl
If you would like to contribute to the project please reach out to Andreas Schlaginhaufen. If you found this library useful in your research, please consider citing.
@article{schlaginhaufen2021learning,
title={Learning Stable Deep Dynamics Models for Partially Observed or Delayed Dynamical Systems},
author={Andreas Schlaginhaufen, Philippe Wenk, Andreas Krause and Florian Dörfler},
journal={Advances in Neural Information Processing Systems},
year={2021},
}