Microsoft Research today announced a new project that embeds artificial intelligence onto bread-crumb size computer processors. This project is called Embedded Learning Library (ELL) and it will help developers build and deploy machine-learned pipelines onto embedded platforms including Raspberry Pis, Arduinos, Micro:bits, and other microcontrollers.
Once deployed, the machine learning model can run without connecting to the internet. Removing the need of internet reduces bandwidth constraints and eliminates concerns about network latency. Also, the on-device machine learning limits battery drain from communication with the cloud. Privacy also is protected by keeping personal and sensitive information within the device.
The researchers imagine all sorts of intelligent devices that could be created with this method, from smart soil-moisture sensors deployed for precision irrigation on remote farms to brain implants that warn users of impending seizures so that they can get to a safe place and call a caregiver.
Microsoft compared a state-of-the-art neural network with and without quantization trained and deployed for computer vision on Raspberry Pi 3s. The models were equally accurate, but the compressed version ran about 20 times faster.