Enabling On-Device Learning at Scale
Offered By: tinyML via YouTube
Course Description
Overview
Explore the future of AI-powered personalized experiences in this 37-minute webinar by Joseph Soriaga, Sr. Director of Technology at Qualcomm. Dive into the world of on-device learning and its potential to revolutionize data processing on edge devices while maintaining user privacy. Discover the latest research in few-shot learning, continuous learning, and federated learning. Gain insights into the challenges and solutions for moving from research to commercialization in on-device learning. Learn about self-supervised learning, keyword spotting, semantic segmentation, and user verification techniques. Understand the implementation of back propagation, quantized training, and memory reduction strategies for efficient on-device learning at scale.
Syllabus
Introduction
Qualcomm AI research
Selfsupervised learning
Proof of concept
Challenges
Agenda
FewShot Learning
Keyword Spotting
Personalization
Continuous learning
Semantic segmentation
Solution
State of progress
Aggregation
User verification
User verification without embeddings
User verification with embeddings
Federated learning
Back propagation
Back propagation implementation
Quantized training
Reducing memory
Questions
Conclusion
Taught by
tinyML
Related Courses
Foundations of Teaching for Learning: IntroductionCommonwealth Education Trust via Coursera PLM14101: Global Information Technology Outlook
World Economic Forum via Independent Leadership and Influence
University of Illinois at Urbana-Champaign via Coursera Creating a Team Culture of Continuous Learning
University of Pennsylvania via Coursera قيادة الفرق
University of Michigan via Coursera