targeted-causal-reduction 0.0.1a1

Targeted causal reduction learns high-level causes for phenomena occuring in simulations.

Targeted Causal Reduction (TCR)

Overview

TCR is a method for explaining a phenomenon (called target) in high-dimensional simulations (low-level model) by learning a low-dimensional causal model (high-level model) that captures the most important causes of the target. It uses shift interventions in the low-level model and its effects on the target to learn the high-level model. The full mathematical details are explained in the associated paper.

Installation

Clone the repository

git clone [e-mail]:akekic/targeted-causal-reduction.git

and install the package

pip install .

If you want to install the package for development, use

pip install -e .[dev]

this will install the package in editable mode and install the additional dependencies for development.

Hello World!

The package provides an entry point for running the TCR algorithm. To run the algorithm on a synthetic linear low-level causal model, use

tcr

This will run the TCR algorithm and save the results as weights and biases logs in the wandb directory. The full list of arguments can be found by running

tcr --help

or by looking at the argument parser in targeted_causal_reduction/parser.py.

License

This project is licensed under the MIT license. See the LICENSE file for details.

Citation

If you use TCR, please cite the corresponding paper as follows.

Kekić, A., Schölkopf, B., & Besserve, M. (2024). Targeted Reduction of Causal Models. Conference on Uncertainty in Artificial Intelligence (UAI).

Bibtex

@article{
    kekic2024targeted,
    title={Targeted Reduction of Causal Models},
    author={Keki\'c, Armin and Sch\"olkopf, Bernhard and Besserve, Michel},
    journal={Conference on Uncertainty in Artificial Intelligence (UAI)},
    year={2024},
}