It seems there is a growing number of developments with machine learning on the edge…
Machine learning at the edge? What is it?
In a nutshell, a tiny neural network on a microcontroller with really low power requirements. Sometimes as low as <1mW.
Why is this important?
Well the future will have many devices on the edge that can do really cool things we thought not possible before.
- Think of several smart sensors deployed across environments like buildings or farms that give real-time data of the environment and act without human intervention for extended periods.
- Smart sensors in a factory that monitor the manufacturing process in the assembly line and replace humans so they can do more useful and more meaningful work.
- Many sensors on heavy equipment that are used for real-time preventative maintenance and predict when they breakdown based on the data over time.
This is just a snippet of the many potential applications on the edge…
Lastly, there is a growing group in the Bay Area that is pushing for this technology.
They call themselves TinyML.
It’s the merging of Embedded Systems and Machine Learning!
Check out this great book from the creators on TinyML:
References:
Link: TinyML – community of embedded and machine learning developers
Link: Medium – Blog post
Link: Amazon – TinyML book