Inference on mobile and IoT devices

Smartphone use has grown exponentially over the last few years and continues to grow in an unabated fashion. Other IoT devices are also becoming increasingly commonplace in our day-to-day lives. These upward trends in usage adoption have interesting consequences for machine learning systems. These platforms are typically resource-constrained in comparison to normal host machines. As a result, additional optimizations are required to run inference on such devices. The TensorFlow platform supports building machine learning and deep learning-based applications that can run on different kinds of edge devices such as mobile phones and other IoT devices. The primary tool made available to this effect is the TensorFlow Lite platform. TensorFlow Lite can be used to build machine learning applications for a variety of platforms including Android, iOS...