Employing formal methods for software development usually involves using a multitude of tools such as model checkers and provers. Most of them again feature different backends and configuration options. Selecting an appropriate configuration for a successful employment becomes increasingly hard. In this article, we use machine learning methods to automatise the backend selection for the ProB model checker. In particular, we explore different approaches to deep learning and outline how we apply them to find a suitable backend relying on given input constraints.