This chapter focuses on the application of Machine Learning (ML) for cyber-physical systems. These systems consist of both software and hardware components that interact with their physical surroundings. Using these systems can present a number of security, reliability, and performance concerns. In this context, the application of ML could be a viable approach for ensuring these critical features. This chapter focuses specifically on the application of ML to ensure the appropriate operation of CPSs in unexpected contexts and scenarios. We begin by offering an overview of both CPSs and ML. Next, we outline the benefits of employing ML for CPSs and provide two particular examples. The first example is related to a road state scanning application while the second one is about the detection of electricity theft in a smart grid. We then outline the key issues and constraints in this scenario. Finally, we provide some recommendations and guidance for ML users in CPSs, as well as future directions in this subject.