Since Faithfulness Fails: The Performance Limits of Neural Causal Discovery

Mateusz Olko, Mateusz Gajewski, Joanna Wojciechowska, Mikołaj Morzy, Piotr Sankowski, Piotr Miłoś
Proceedings of the 42nd International Conference on Machine Learning, PMLR 267:47155-47175, 2025.

Abstract

Neural causal discovery methods have recently improved in terms of scalability and computational efficiency. However, our systematic evaluation highlights significant room for improvement in their accuracy when uncovering causal structures. We identify a fundamental limitation: unavoidable likelihood score estimation errors disallow distinguishing the true structure, even for small graphs and relatively large sample sizes. Furthermore, we identify the faithfulness property as a critical bottleneck: (i) it is likely to be violated across any reasonable dataset size range, and (ii) its violation directly undermines the performance of neural penalized-likelihood discovery methods. These findings lead us to conclude that progress within the current paradigm is fundamentally constrained, necessitating a paradigm shift in this domain.

Cite this Paper

BibTeX
@InProceedings{pmlr-v267-olko25a, title = {Since Faithfulness Fails: The Performance Limits of Neural Causal Discovery}, author = {Olko, Mateusz and Gajewski, Mateusz and Wojciechowska, Joanna and Morzy, Miko{\l}aj and Sankowski, Piotr and Mi{\l}o\'{s}, Piotr}, booktitle = {Proceedings of the 42nd International Conference on Machine Learning}, pages = {47155--47175}, year = {2025}, editor = {Singh, Aarti and Fazel, Maryam and Hsu, Daniel and Lacoste-Julien, Simon and Berkenkamp, Felix and Maharaj, Tegan and Wagstaff, Kiri and Zhu, Jerry}, volume = {267}, series = {Proceedings of Machine Learning Research}, month = {13--19 Jul}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v267/main/assets/olko25a/olko25a.pdf}, url = {https://proceedings.mlr.press/v267/olko25a.html}, abstract = {Neural causal discovery methods have recently improved in terms of scalability and computational efficiency. However, our systematic evaluation highlights significant room for improvement in their accuracy when uncovering causal structures. We identify a fundamental limitation: unavoidable likelihood score estimation errors disallow distinguishing the true structure, even for small graphs and relatively large sample sizes. Furthermore, we identify the faithfulness property as a critical bottleneck: (i) it is likely to be violated across any reasonable dataset size range, and (ii) its violation directly undermines the performance of neural penalized-likelihood discovery methods. These findings lead us to conclude that progress within the current paradigm is fundamentally constrained, necessitating a paradigm shift in this domain.} }
Endnote
%0 Conference Paper %T Since Faithfulness Fails: The Performance Limits of Neural Causal Discovery %A Mateusz Olko %A Mateusz Gajewski %A Joanna Wojciechowska %A Mikołaj Morzy %A Piotr Sankowski %A Piotr Miłoś %B Proceedings of the 42nd International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2025 %E Aarti Singh %E Maryam Fazel %E Daniel Hsu %E Simon Lacoste-Julien %E Felix Berkenkamp %E Tegan Maharaj %E Kiri Wagstaff %E Jerry Zhu %F pmlr-v267-olko25a %I PMLR %P 47155--47175 %U https://proceedings.mlr.press/v267/olko25a.html %V 267 %X Neural causal discovery methods have recently improved in terms of scalability and computational efficiency. However, our systematic evaluation highlights significant room for improvement in their accuracy when uncovering causal structures. We identify a fundamental limitation: unavoidable likelihood score estimation errors disallow distinguishing the true structure, even for small graphs and relatively large sample sizes. Furthermore, we identify the faithfulness property as a critical bottleneck: (i) it is likely to be violated across any reasonable dataset size range, and (ii) its violation directly undermines the performance of neural penalized-likelihood discovery methods. These findings lead us to conclude that progress within the current paradigm is fundamentally constrained, necessitating a paradigm shift in this domain.
APA
Olko, M., Gajewski, M., Wojciechowska, J., Morzy, M., Sankowski, P. & Miłoś, P.. (2025). Since Faithfulness Fails: The Performance Limits of Neural Causal Discovery. Proceedings of the 42nd International Conference on Machine Learning, in Proceedings of Machine Learning Research 267:47155-47175 Available from https://proceedings.mlr.press/v267/olko25a.html.

Related Material