In this presentation we will cover microprocessor (MPU) role in machine learning applications. MPU cover two important roles at the edge of the network: acting as a gateway and broker for smaller nodes and also providing increased processing capacity both for local and remote machine learning applications. We will discuss how MPUs offer intelligent connectivity and how they provide secure integration of smart devices and applications with the cloud. We will detail how Deep Neural Networks should be defined and optimized for MPUs: architecture selection, inference acceleration and heterogeneous processing. We will show how knowledge transfer between nodes and cascaded learning creates a synergistic intelligent network.