ML4CPS

ML4CPS is a Python package for learning and analysis of hybrid dynamical systems, with the focus on Cyber-Physical Systems (CPS). The code was developed for several research publications (bibtex).

• Website (ml4cps.ai4cps.com)
• Contact (contact@ai4cps.com)

Installation

Install the core package without PyTorch or MLflow:

pip install ml4cps

For DEBTA models and tensor-based tools, install the optional PyTorch dependency:

pip install "ml4cps[torch]"

For experiment tracking and MLflow model integration, install:

pip install "ml4cps[mlflow]"

Combine both extras with pip install "ml4cps[torch,mlflow]". Core CPS simulation, visualization, and DEBTA with log_mlflow=False work without MLflow. Tracking calls require the mlflow extra. The dev extra includes both PyTorch and MLflow for development and tests.

Data

In the folder "data" there are several datasets that can be easily loaded using the ml4cps.examples module.

Bugs

If you find any bugs, please contact us at bugs@ai4cps.com.

License

See LICENSE.
If you use this code in your research, please cite our work.