Explore recent advancements in tinyML technologies and always-on machine capabilities in this 47-minute keynote presentation from the IEEE Computer Society Annual Symposium on VLSI 2020. Delivered by Evgeni Gousev of Qualcomm AI Research, the talk covers the definition and constraints of tinyML, its exciting potential across various industries, and implementation strategies. Delve into the future of tinyML, focusing on always-on voice and vision applications, MEMS sensors, environmental sensing, and edge measurement. Gain insights into specific examples such as face detection and image quality improvements, and understand the system approach required for successful tinyML implementation. Discover how tinyML is revolutionizing machine learning on resource-constrained devices and opening up new possibilities in the field of artificial intelligence.
Recent Progress on TinyML Technologies and Always-On Machine Opportunities
Overview
Syllabus
Introduction
Title
Outline
What is tinyML
Constraints
Defining tinyML
Why we are excited about tinyML
Opportunities in all verticals
Animal Foundation
Community Events
How tinyML is implemented
Is tinyML good enough
The future of tinyML
Alwayson voice
MEMS sensors
Environmental sensing
Time use and edge measurement
Alwayson Vision
Applications
Examples
Face Detection
Image Quality
System Approach
Summary
Taught by
tinyML
