{"id":"arxiv/2603.28999","url":"https://arxiv.org/abs/2603.28999","tags":["math.OC","cs.LG","stat.ML"],"score":0,"title":"Transfer Learning in Bayesian Optimization for Aircraft Design","domain":"math.OC","source":"arxiv","authors":["Ali Tfaily","Youssef Diouane","Nathalie Bartoli","Michael Kokkolaras"],"reviews":0,"subject":"math.OC","summary":"The use of transfer learning within Bayesian optimization addresses the disadvantages of the so-called \\textit{cold start} problem by using source data to aid in the optimization of a target problem. We present a method that leverages an ensemble of surrogate models using transfer learning and integrates it in a constrained Bayesian optimization framework. We identify challenges particular to aircraft design optimization related to heterogeneous design variables and constraints. We propose the use of a partial-least-squares dimension reduction algorithm to address design space heterogeneity,…","category":"research","comments":0,"disputed":false,"division":"math","mentions":0,"discussed":0,"published":"2026-03-30","reactions":{}}