Microsoft is taking a huge leap in artificial intelligence and machine learning by incorporating it into a tiny device such as the Raspberry Pi 3.
Microsoft research lab senior researcher in India Manik Varma has developed an innovation in machine learning software that allows melding artificial intelligence onto a credit card sized single board computer processor like a Raspberry Pi.
Varma along with his team was able to compress neural networks from 32 bits down to a single bit and use it on Raspberry Pi 3. It is now available for download via Github so developers can lay hands on this new machine learning software.
Ofer Dekel, Machine Learning and Optimization group lead at Microsoft's research lab in Redmond, Washington, said today's machine learning uses a massive amount of memory since it operates on cloud computing, collecting and sending data through sensors. This newly-developed software can do just that, fitting machine learning in just a speck of a device.
In Microsoft's blog post, Varma, who is visually impaired, expressed how this technology can help the humanity, including people like him. "If you're driving on a highway and there isn't connectivity there, you don't want the [AI] implant to stop working. In fact, that's where you really need it the most", says Varma.
This move is a highlight of Microsoft chief executive Satya Nadella's vision to keep up with "a new world that is going to be made up of an intelligent could and intelligent edge", as he described it in his keynote speech at the firm's Build 2017 conference in Seattle.
In line with that purpose, the software giant has created this "intelligent edge" as a leap to fulfill the vow to dominate the world with smart and tiny machines whichever way humans go.
While this innovation in machine learning is considered a part of the so-called Internet of Things, it needs not an Internet connection to function.