Causal Discovery via Bayesian Optimization
Bao Duong, Sunil Gupta, and Thin Nguyen, Applied Artificial Intelligence Institute (AI)Deakin University, Geelong, Australia · 2025
Read Causal Discovery via Bayesian Optimization by Bao Duong, Sunil Gupta, and Thin Nguyen, Applied Artificial Intelligence Institute (AI)Deakin University, Geelong, Australia (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2501-14997.txt.
Description
Bao Duong, Sunil Gupta, and Thin Nguyen propose a new way to learn causal graphs from observational data. Score-based methods search over directed acyclic graphs for the one that best fits the data, but the search is slow and often inaccurate. Their framework, called DrBO, hands the search to Bayesian optimization, a method that picks the most promising candidates to try next.
Standard Bayesian optimization does not scale to this problem, so the team replaced its usual Gaussian process model with dropout neural networks trained continuously. The swap lets the model track uncertainty in graph scores without overfitting, and it grows gracefully with the number of trials. In wide tests on synthetic and real data, DrBO found the correct graph in fewer trials and less time than leading methods. The authors have released their code publicly. The work aims to make causal discovery practical on harder problems.
Catalog record
| CREATOR | Bao Duong, Sunil Gupta, and Thin Nguyen, Applied Artificial Intelligence Institute (AI)Deakin University, Geelong, Australia |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 94 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2501-14997