Learning Compact Representation with Less Labeled Data from Sensors
Offered By: tinyML via YouTube
Course Description
Overview
Explore data-efficient learning techniques for compact representations in sensor-based human behavior modeling. Delve into the challenges of limited labeled data in IoT environments and discover innovative approaches like domain adaptation and pretraining. Learn about contrastive learning, change point detection, and their applications in behavior recognition and COVID-19 cough sound detection. Examine evaluation datasets and experiment results to gain insights into scalable contrastive networks and their performance in real-world scenarios.
Syllabus
Intro
Human activity and/or behaviour recognition
Representation learning requires LOTS of labelled d
Typical pipeline for behaviour recognition
Challenges with Sensor Data
What makes good representation from sensor data?
Various types of changes
The Main Idea of Contrastive Learning
Change Point Detection
Evaluation Datasets
Detecting COVID-19 cough sounds with
Scalogram Contrastive Network
Experiment Results
Taught by
tinyML
Related Courses
Web Science: How the Web Is Changing the WorldUniversity of Southampton via FutureLearn Develop Java Embedded Applications Using a Raspberry Pi
Oracle via Independent Introducción a Raspberry Pi (Ver-2)
Galileo University via Independent Fog Networks and the Internet of Things
Princeton University via Coursera Digital Media and Marketing Principles
University of Illinois at Urbana-Champaign via Coursera