Learning to Defer for Causal Discovery with Imperfect Experts

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Clivio, Oscar, Mahajan, Divyat, Taslakian, Perouz, Magliacane, Sara, Mitliagkas, Ioannis, Zantedeschi, Valentina, Drouin, Alexandre
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866929719380803584
author Clivio, Oscar
Mahajan, Divyat
Taslakian, Perouz
Magliacane, Sara
Mitliagkas, Ioannis
Zantedeschi, Valentina
Drouin, Alexandre
author_facet Clivio, Oscar
Mahajan, Divyat
Taslakian, Perouz
Magliacane, Sara
Mitliagkas, Ioannis
Zantedeschi, Valentina
Drouin, Alexandre
contents Integrating expert knowledge, e.g. from large language models, into causal discovery algorithms can be challenging when the knowledge is not guaranteed to be correct. Expert recommendations may contradict data-driven results, and their reliability can vary significantly depending on the domain or specific query. Existing methods based on soft constraints or inconsistencies in predicted causal relationships fail to account for these variations in expertise. To remedy this, we propose L2D-CD, a method for gauging the correctness of expert recommendations and optimally combining them with data-driven causal discovery results. By adapting learning-to-defer (L2D) algorithms for pairwise causal discovery (CD), we learn a deferral function that selects whether to rely on classical causal discovery methods using numerical data or expert recommendations based on textual meta-data. We evaluate L2D-CD on the canonical Tübingen pairs dataset and demonstrate its superior performance compared to both the causal discovery method and the expert used in isolation. Moreover, our approach identifies domains where the expert's performance is strong or weak. Finally, we outline a strategy for generalizing this approach to causal discovery on graphs with more than two variables, paving the way for further research in this area.
format Preprint
id arxiv_https___arxiv_org_abs_2502_13132
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Learning to Defer for Causal Discovery with Imperfect Experts
Clivio, Oscar
Mahajan, Divyat
Taslakian, Perouz
Magliacane, Sara
Mitliagkas, Ioannis
Zantedeschi, Valentina
Drouin, Alexandre
Machine Learning
Artificial Intelligence
Integrating expert knowledge, e.g. from large language models, into causal discovery algorithms can be challenging when the knowledge is not guaranteed to be correct. Expert recommendations may contradict data-driven results, and their reliability can vary significantly depending on the domain or specific query. Existing methods based on soft constraints or inconsistencies in predicted causal relationships fail to account for these variations in expertise. To remedy this, we propose L2D-CD, a method for gauging the correctness of expert recommendations and optimally combining them with data-driven causal discovery results. By adapting learning-to-defer (L2D) algorithms for pairwise causal discovery (CD), we learn a deferral function that selects whether to rely on classical causal discovery methods using numerical data or expert recommendations based on textual meta-data. We evaluate L2D-CD on the canonical Tübingen pairs dataset and demonstrate its superior performance compared to both the causal discovery method and the expert used in isolation. Moreover, our approach identifies domains where the expert's performance is strong or weak. Finally, we outline a strategy for generalizing this approach to causal discovery on graphs with more than two variables, paving the way for further research in this area.
title Learning to Defer for Causal Discovery with Imperfect Experts
topic Machine Learning
Artificial Intelligence
url https://arxiv.org/abs/2502.13132