Citing
If you make use of the MeDIL package or associated theory, please cite the relevant subset of the following papers:
The UAI paper [MGW20] introduces MeDIL causal models, the strong causal insufficiency assumption, and the reduction of structure learning to minimum edge clique cover.
@InProceedings{Markham_2020_UAI,
title = {Measurement Dependence Inducing Latent Causal Models},
author = {Markham, Alex and Grosse-Wentrup, Moritz},
booktitle = {Proceedings of the 36th Conference on Uncertainty in Artificial Intelligence (UAI)},
pages = {590--599},
year = 2020,
volume = 124,
publisher = {PMLR},
url = {http://proceedings.mlr.press/v124/markham20a.html}
}
The PGM software demonstration [MCGW20] is specifically associated with the MeDIL Python package.
@InProceedings{Markham_2020_PGM,
title = {{MeDIL}: {A} {Python} Package for Causal Modelling},
author = {Markham, Alex and Chivukula, Aditya and Grosse-Wentrup, Moritz},
booktitle = {Proceedings of the 10th International Conference on Probabilistic Graphical Models (PGM)},
pages = {621--624},
year = 2020,
volume = 138,
series = {Proceedings of Machine Learning Research},
publisher = {PMLR},
url = {https://proceedings.mlr.press/v138/markham20a.html}
}
The NCFA preprint [MLAS23] further develops the framework, connects it to factor analysis, introduces identifiability results, and extends MeDIL to deep generative causal mechanisms.
@misc{markham2023neuro,
author = {Alex Markham and Mingyu Liu and Bryon Aragam and Liam Solus},
title = {Neuro-Causal Factor Analysis},
year = 2023,
note = {preprint, arXiv:2305.19802 [stat.ML]},
url = {https://arxiv.org/abs/2305.19802}
}