Survey Paper Accepted by TMLR

I am pleased to announce that our paper, “Causal Bayesian Optimization: Foundations, Methods, and Applications,” has been accepted by Transactions on Machine Learning Research (TMLR). The decision was accepted with minor revision.

This paper grew out of my research assistantship at North Carolina State University with Dr. Hien Tran. For me, the TMLR acceptance is the perfect full stop on that chapter: the last piece of work from the assistantship, and a journal paper that closes the collaboration we started together. I am deeply grateful for his mentorship, which shaped how I think about causal decision-making and how I learned to carry a project from an idea to a finished manuscript. I also thank my collaborators, Thuy T. Le and Zixuan Ma, for the work we did along the way.

The paper is a survey of Causal Bayesian Optimization (CBO). We organize the literature through a unified BO-loop perspective, covering intervention search spaces, surrogate construction, acquisition design, and decision policies. To make comparisons more reproducible, we also introduce a benchmark with a trajectory-aware Path-Aware GAP (PA-GAP) metric and evaluate seven CBO methods alongside a non-causal baseline.