Benchmarking the performance of Bayesian optimization across multiple experimental materials science domains

Abstract Bayesian optimization (BO) has been leveraged for guiding autonomous and high-throughput experiments in materials science. However, few have evaluated the efficiency of BO across a broad range experimental domains. In this work, we quantify performance with collection surrogate model acquisition function pairs five diverse systems. By defining acceleration enhancement metrics objectives, find that models such as Gaussian Process (GP) anisotropic kernels Random Forest (RF) comparable BO, both outperform commonly used GP isotropic kernels. demonstrated most robustness, yet RF is close alternative warrants more consideration because it free from distribution assumptions, smaller time complexity, requires less effort initial hyperparameter selection. We also raise awareness about benefits using future campaigns.