1: \begin{abstract}

2: Bayesian Optimization (BO) is a class of surrogate-based, sample-efficient algorithms for optimizing black-box problems with small evaluation budgets.

3: The BO pipeline itself is highly configurable with many different design choices regarding the initial design, surrogate model, and acquisition function (AF).

4: Unfortunately, our understanding of how to select suitable components for a problem at hand is very limited.

5: In this work, we focus on the definition of the AF, whose main purpose is to balance the trade-off between exploring regions with high uncertainty and those with high promise for good solutions.

6: We propose Self-Adjusting Weighted Expected Improvement (SAWEI), where we let the exploration-exploitation trade-off self-adjust in a data-driven manner, based on a convergence criterion for BO.

7: On the noise-free black-box BBOB functions of the COCO benchmarking platform, our method exhibits a favorable anytime performance compared to handcrafted baselines and serves as a robust default choice for any problem structure.

8: The suitability of our method also transfers to HPOBench.

9: With SAWEI, we are a step closer to on-the-fly, data-driven, and robust BO designs that automatically adjust their sampling behavior to the problem at hand. 

10: \end{abstract}

11: