AutoML + TinyML with Edge Impulse's EON Tuner

Explore the intersection of AutoML and TinyML in this hour-long talk featuring Edge Impulse's EON Tuner. Learn how AutoML techniques are making neural network training and optimization more accessible, enabling use-case experts to discover novel applications for machine learning in embedded systems. Dive into the workings of the EON Tuner, which helps select optimal embedded machine learning models within specific device constraints. Gain insights into the unique benefits of AutoML for embedded systems, and discover how to implement these tools in your own projects. The presentation covers Edge Impulse's background, TinyML overview, traditional vs. data-driven engineering approaches, and includes a live demo showcasing the EON Tuner's capabilities. Engage with topics such as neural network design, hardware selection, manual tuning, DSP optimization, and performance considerations. Conclude with an informative Q&A session addressing audience queries on various aspects of the EON Tuner and Edge Impulse platform.

Introduction
Edge Impulse Background
Developer Focus
Overview of TinyML
Traditional Engineering
DataDriven Engineering
Design Phase
Traditional Pipeline
EON Tuner
Demo
Project Overview
Neural Network
AutoML
Constraints
Performance Comparison
Tuner Features
Audience Questions
Hardware Selection
Manual Tuning
Compare Tuning
Vendors
Shape tuning
DSP optimization
Performance considerations
Video support
Image of arbitrary size
Uploading data to Edge Impulse
Is Edge Impulse free
Does the EON Tuner scale
Is the EON Tuner available for download
Do you support NN architectures
Do we have to pick and specify a particular device
Wrapping up
Questions
Thanks