Implementierung einer Tensor Processing Unit mit dem Fokus auf Embedded Systems und das Internet of Things


Machine learning is more and more applied in our everyday life, but also safety critical systems are increasingly equipped with ML procedures. This paper gives an insight into the implementation of a machine learning co-processor for embedded systems and IoT devices. A scalable architecture based on Google’s Tensor Processing Units was imple- mented. This allows smaller systems to be equipped with this accelerator and to perform other real-time tasks in addition to the parallel execution of ML models.