Hands‐on tinyML for IoT, bringing intelligence to the edge

ID: Tut-04

 

Description

tinyML is a quickly advancing field in deep learning, focusing on deploying neural networks at the edge. However, it presents many challenges, as it is at the intersection of two research areas, artificial intelligence and embedded systems, which did not use to overlap in the past. tinyML is gaining momentum in the scientific community and forcing AI research to deal with typical properties of IoT applications, such as limited resources, real-time constraints on tiny platforms, data scarcity, and sensor noise. tinyML enables big companies and SMEs to offer novel, enhanced, low-power, low-cost smart products for the IoT market while guaranteeing privacy-preserving technologies. This tutorial will cover the main techniques developed throughout the years to create, adapt and deploy neural networks at the edge. These techniques include knowledge distillation, neural architecture search, quantization, and hardware-aware scaling. The tutorial will comprise a hands-on experience in developing an AI-enhanced IoT application. At the end of the tutorial, participants will have completed the development, adaptation and deployment cycle of a neural network for image classification on a microcontroller.

 

 

If you have any questions, please contact Dr. Francesco Paissan: fpaissan@fbk.eu 

 

Chairs

Francesco Paissan, Fondazione Bruno Kessler, Italy

 

 

 

 

 

 

 

 

Alberto Ancilotto, Fondazione Bruno Kessler, Italy

 

 

 

 

 

 

 

 

Elisabetta Farella, Fondazione Bruno Kessler, Italy