From Guess2Graph: When and How Can Unreliable Experts Safely Boost Causal Discovery in Finite Samples?

Sujai Hiremath, Dominik Janzing, Philipp Michael Faller, Patrick Blöbaum, Elke Kirschbaum, Shiva Kasiviswanathan, Kyra Gan
Proceedings of The 29th International Conference on Artificial Intelligence and Statistics, PMLR 300:1459-1467, 2026.

Abstract

Causal discovery algorithms often perform poorly with limited samples. While integrating expert knowledge (including from LLMs) as constraints promises to improve performance, guarantees for existing methods require perfect predictions or uncertainty estimates, making them unreliable for practical use. We propose the Guess2Graph (G2G) framework, which uses expert guesses to guide the sequence of statistical tests rather than replacing them. This maintains statistical consistency while enabling performance improvements. We develop two instantiations of G2G: PC-Guess, which augments the PC algorithm, and gPC-Guess, a learning-augmented variant designed to better leverage high-quality expert input. Theoretically, both preserve correctness regardless of expert error, with gPC-Guess provably outperforming its non-augmented counterpart in finite samples when experts are "better than random". Empirically, both show monotonic improvement with expert accuracy, with gPC-Guess achieving significantly stronger gains.

Cite this Paper

BibTeX
@InProceedings{pmlr-v300-hiremath26a, title = { From Guess2Graph: When and How Can Unreliable Experts Safely Boost Causal Discovery in Finite Samples? }, author = {Hiremath, Sujai and Janzing, Dominik and Faller, Philipp Michael and Bl{\"o}baum, Patrick and Kirschbaum, Elke and Kasiviswanathan, Shiva and Gan, Kyra}, booktitle = {Proceedings of The 29th International Conference on Artificial Intelligence and Statistics}, pages = {1459--1467}, year = {2026}, editor = {Khan, Emtiyaz and Li, Yingzhen and Solin, Arno and Ramdas, Aaditya}, volume = {300}, series = {Proceedings of Machine Learning Research}, month = {02--05 May}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v300/main/assets/hiremath26a/hiremath26a.pdf}, url = {https://proceedings.mlr.press/v300/hiremath26a.html}, abstract = { Causal discovery algorithms often perform poorly with limited samples. While integrating expert knowledge (including from LLMs) as constraints promises to improve performance, guarantees for existing methods require perfect predictions or uncertainty estimates, making them unreliable for practical use. We propose the Guess2Graph (G2G) framework, which uses expert guesses to guide the sequence of statistical tests rather than replacing them. This maintains statistical consistency while enabling performance improvements. We develop two instantiations of G2G: PC-Guess, which augments the PC algorithm, and gPC-Guess, a learning-augmented variant designed to better leverage high-quality expert input. Theoretically, both preserve correctness regardless of expert error, with gPC-Guess provably outperforming its non-augmented counterpart in finite samples when experts are "better than random". Empirically, both show monotonic improvement with expert accuracy, with gPC-Guess achieving significantly stronger gains. } }
Endnote
%0 Conference Paper %T From Guess2Graph: When and How Can Unreliable Experts Safely Boost Causal Discovery in Finite Samples? %A Sujai Hiremath %A Dominik Janzing %A Philipp Michael Faller %A Patrick Blöbaum %A Elke Kirschbaum %A Shiva Kasiviswanathan %A Kyra Gan %B Proceedings of The 29th International Conference on Artificial Intelligence and Statistics %C Proceedings of Machine Learning Research %D 2026 %E Emtiyaz Khan %E Yingzhen Li %E Arno Solin %E Aaditya Ramdas %F pmlr-v300-hiremath26a %I PMLR %P 1459--1467 %U https://proceedings.mlr.press/v300/hiremath26a.html %V 300 %X Causal discovery algorithms often perform poorly with limited samples. While integrating expert knowledge (including from LLMs) as constraints promises to improve performance, guarantees for existing methods require perfect predictions or uncertainty estimates, making them unreliable for practical use. We propose the Guess2Graph (G2G) framework, which uses expert guesses to guide the sequence of statistical tests rather than replacing them. This maintains statistical consistency while enabling performance improvements. We develop two instantiations of G2G: PC-Guess, which augments the PC algorithm, and gPC-Guess, a learning-augmented variant designed to better leverage high-quality expert input. Theoretically, both preserve correctness regardless of expert error, with gPC-Guess provably outperforming its non-augmented counterpart in finite samples when experts are "better than random". Empirically, both show monotonic improvement with expert accuracy, with gPC-Guess achieving significantly stronger gains.
APA
Hiremath, S., Janzing, D., Faller, P.M., Blöbaum, P., Kirschbaum, E., Kasiviswanathan, S. & Gan, K.. (2026). From Guess2Graph: When and How Can Unreliable Experts Safely Boost Causal Discovery in Finite Samples? . Proceedings of The 29th International Conference on Artificial Intelligence and Statistics, in Proceedings of Machine Learning Research 300:1459-1467 Available from https://proceedings.mlr.press/v300/hiremath26a.html.

Related Material