The Internet of Things Might Have Less Internet Than We Thought?

While machine learning is traditionally associated with heavy duty power-hungry processors, it is beginning to look like the future of machine learning is on the edge. The ability to run trained networks “at the edge” nearer the data without access to the cloud — or in some cases even without even a network connection  at all—means that we can interpret sensor data in the field, turning raw accelerometer data into information about whether a machine is working well or malfunctioning, or recognising voice commands in real-time without storing potentially privacy infringing data. 

Over the last year custom silicon, intended to speed up machine learning inferencing on the edge, has started to appear and the arrival of this new hardware designed to run machine learning models at vastly increased speeds, and inside a relatively low power envelopes, has made the emerging ecosystem around edge computing feel far more mature. Recent work has even allowed us to run the same machine learning networks on micro-controllers, rather than micro-processors, pushing the boundaries of our power envelope down to allow us to embedded devices that can run for a year or more on a single coin cell battery. 

The future of ’smart’ devices, and the Internet of Things, might have a lot less Internet than we previously thought. We look at the implications of machine learning on the edge, and the possible implications around privacy and security.