TinyML: The future of embedded Machine Learning is to change lives for the better

Some insights and real-world applications that can improve the lives of people with physical limitations or disabilities

Photo by Vishnu Mohanan on Unsplash

“A small device with an enormous impact on our lives”

This sentence could resume Tiny Machine Learning (TinyML), an emerging field in artificial intelligence. TinyML is a cutting-edge field that brings the transformative power of machine learning (ML) to the performance and power-constrained domain of embedded systems. [01]

This concept has been around for a while, 2 to 3 years ago but only recently with the popularization of more efficient algorithms such as TensorFlow Lite, for example, we were able to see more applications that directly impact everyone’s life.

In this article, I do a brief review on the subject and go into real applications and recent scientific articles of TinyML that have a great impact on the lives of many people with physical limitations or disabilities.

And finally, I show you where you can learn more about TinyML and start creating your own devices, joining many other professionals and companies who are working harder to bring more devices that can have a positive impact on everyone’s life.

Introduction

An ABI research said that Global Shipments of TinyML Devices will reach 2.5 billion by 2030 and could reach more than US$ 70 billion in economic value. At that moment, there are several companies developing chips and frameworks to be used to build more efficient TinyML devices:

1. Arduino with a new TinyML kit
2. Google with Tensorflow Lite for Microcontrollers
3. Microsoft Azure Sphere, an integrated security platform to develop faster and secure IoT devices.

And the list goes on.

We all know that AI is transforming the world through technology, from how we interact with each other to medical advances. There is an excellent article on this subject on the USC (University of Southern California) blog:

5 Ways Artificial Intelligence Will Change the World by 2050 • Trojan Family Magazine (usc.edu)

But how can TinyML play a role in this transformation? How can this transformation change our lives for the better?

By bringing AI to small microcontrollers, we can use the power of billions of gadgets that we already use in our lives without depending on extra costly equipment. We can build cheaper devices that adapt to our daily lives and have a high impact on how we deal with the environment around us.

I selected some of the best scientific publications from the last two years (in my opinion), with some insights and real-world TinyML applications that can improve the lives of people with an illness or disability:

1. An Intelligent Bed Sensor System for Non-Contact Respiratory Rate Monitoring
2. TinyLSTMs: Efficient Neural Speech Enhancement for Hearing Aids
3. GesturePod: Gesture-based Interaction Cane for People with Visual Impairments
4. A Tiny CNN Architecture for Medical Face Mask Detection for Resource-Constrained Endpoints
5. Rethinking Generalization in American Sign Language Prediction for Edge Devices with Extremely Low Memory Footprint

There’s also a great video who explain the idea and algorithms of the last article of the list: Rethinking Generalization in American Sign Language Prediction for Edge Devices

The options are limitless, only limited by our imagination.

And where to learn more about TinyML?

If you are interested in the topic and would like to learn more and develop your TinyML devices there are several free courses and tutorials on the internet, such as this hands-on Google Codelabs.

Photo by Chenyu Guan on Unsplash

If you want a full specialization, Harvard University and Edx have partnered to deliver a series of free courses that range from basic to advanced:

1. Fundamentals of TinyML

• Fundamentals of Machine Learning (ML)
• Fundamentals of Deep Learning
• How to gather data for ML
• How to train and deploy ML models
• Understanding embedded ML

2. Applications of TinyML

• The code behind some of the most widely used applications of TinyML
• Real-word industry applications of TinyML
• Principles of Keyword Spotting
• Principles of Visual Wake Words
• Concept of Anomaly Detection
• Principles of Dataset Engineering

3. Deploying TinyML

• An understanding of the hardware of a microcontroller-based device
• A review of the software behind a microcontroller-based device
• How to program your TinyML device
• How to write your code for a microcontroller-based device
• How to deploy your code to a microcontroller-based device
• How to train a microcontroller-based device

Conclusions

For the first time in history with the internet revolution, the power to change the world is in everyone’s hands. We can choose to use technology to keep changing people’s lives for the better.

I hope this tiny article (bazinga!) will act as a foundation and inspiration for the development of additional devices that can change people’s lives.

The future is tiny and bright.

Thanks for reading this article, please feel free to ask or provide feedback in the comments section.