Explore a tinyML Talks webcast featuring Vicki Moran and Will McDonald from Harvey Mudd College as they discuss training neural networks for sensors. Delve into the 2019-2020 Harvey Mudd College Clinic program, where a team of students collected data from a 6-axis accelerometer and trained neural networks to recognize gestures on a Syntiant NDP101 device. Learn about the project investigation, data collection process, and neural network performance achieved for three unique gestures. Discover insights on wrist-based gestures for smartwatch applications, data augmentation techniques, network architecture, and the effects of time shifting. Witness a watch-checking demonstration and understand the benefits of custom printed circuit boards in implementing low-power components for additional features.
Overview
Syllabus
tiny ML. Talks
Harvey Mudd College Clinic Team
Harvey Mudd College Clinic Program
Problem Statement
More than Speech Recognition
Wrist-Based Gestures for Smartwatch Applications
Prototype Breakdown
Data Collection Hardware
Data Collection Involves Multiple Gestures
Typical Data Instances
Data Augmentation
Data Visualization
Network Architecture
Dataset Composition
Network Performance
Effect of Time Shifting
Framing for Input Window
Watch-Checking Demonstration
Benefits of a Custom Printed Circuit Board
Implementation of the Board
Components Consume Minimal Power
Other Components Enable Additional Features
Taught by
tinyML
Related Courses
-
Deep Learning: Convolutional Neural Networks in Python
-
TinyML - Using Machine Learning on Microcontrollers to Recognize Speech
-
Build a Deep Face Detection Model With Python and Tensorflow - Full Course
5.0 -
Introduction to Neural Networks in Python - Tensorflow-Keras
-
Leaf Tracking, Counting and Occlusion in Plants Using Artificial Neural Networks
