As a new era of gravitational wave detections rapidly unfolds, the importance of having accurate models for their signals becomes increasingly important. In this context, we will discuss numerical simulations of black hole binaries. While most of the simulations so far have focused on the quasi-circular case, some events detected by the LIGO-Virgo-Kagra collaboration challenge this interpretation. With this motivation, we will consider dynamical captures which are expected to be relevant in dense media such as globular clusters. We will argue that Active Learning, a data-driven strategy to explore parameter space with costly experiments, could be particularly relevant in this scenario.