tinyML on embedded microprocessors and the broad avenue of opportunities
TimeThursday, December 9th2:00pm - 2:30pm PST
Special Session (Research Track)
Hosted in Virtual Platform
DescriptionTinyML represents a collaborative effort between the embedded ultra-low power systems and machine learning communities, which traditionally have operated largely independently. This union has opened the floodgates for new and exciting applications of on-device machine learning. However, the knowledge that deep learning and embedded microprocessors are a perfect match has been exclusive, hidden behind the walls of the industry. This becomes more obvious when you identify that this paradigm of running “tiny” deep learning models on microcontrollers is responsible for the always-on functionality that has been around for the last few years such as the Google Assistant, or Apple’s Siri. In this talk therefore, I will give a technical overview of state-of-the-art HW and SW solutions for TinyML on ARM-based systems.