I. Introduction
As technology advances and software applications become widespread, the requirements for multiple functionalities increase at a fast rate. In the automotive industry, high-end products nowadays embed more than 100 million lines of code that realize a variety of functions. From robust safety features to increased comfort in the cabin, the role of software has taken as much importance as the mechanical integrity of the vehicles. The industry now has a clear focus on artificial intelligence (AI) applications that handle very large amounts of data. Mainly due to such large amounts of data, most time and development efforts in this domain are spent on understanding, preparing, monitoring, and logging of data, rather than actually implementing the machine learning algorithms and models [1]. Such high demands on software can only be realized through mechanisms that allow for increased hardware performance and energy efficiency at a reasonable cost.