Widening Access to Applied ML with TinyML

I’m delighted to announce that the paper, I’ve worked on with folks from Harvard and Google, has been published online to arxiv! We discuss the strategies that we used in designing a curriculum to widen access to Applied ML, focusing on the rapid growth of TinyML and how a new teaching methodology can open doorways previously shut to the traditionally underrepresented.

Course Design

It’s critically important to do this type of teaching at this time because with the rapid growth of ML, and TinyML in particular, I anticipate that we’ll soon end up with several prominent players in this space and that these players will likely be in the traditional high tech areas. By opening up easy access to teaching materials for everyone, we hope to level the playing field.

With that in mind, we designed the course to have the academic rigor and depth of a top-tier university like Harvard, with the breadth of understanding required to work in a major company like Google. Our goal was linking the two to bring students holistically through a journey of understanding ML, using it, and then deploying it to embedded systems and microcontrollers.

One observation we made while working on this is an inverse relationship between the number of available ML educational resources and the number of systems in the field! We wanted to fix that!

Picture of numbers

Please check out the paper, or indeed the entire course at edX where you can audit it at no cost.