Explore the state of TinyML in this 53-minute panel discussion featuring experts Frédéric Pétrot, Etienne Balit, and Loic Lietar. Delve into the landscape and potential of ultra-low-power applications, recent advances, and challenges in TinyML. Learn about AI needs, current neural network architectures, hardware acceleration, and power efficiency. Discover commercial applications, disruptive potential, and use case evaluation for TinyML. Examine tradeoffs, IoT applications, voice commands, face ID, and object detection. Compare embedded vs. cloud solutions, and understand the challenges of AI in embedded systems. Investigate RISC-V architecture for AI modeling, and contrast TinyML with microcontrollers. Gain insights into distributed AI and participate in an interactive Q&A session with the audience.
Overview
Syllabus
Introduction
What kind of AI do we need
Current neural network architectures
Hardware acceleration for neural networks
Power efficiency
Summary
Intel
Etienne
Commercial applications
Is TinyML a disruptive technology
How do we know if a use case is doable
Negative use cases
Positive use cases
Tradeoffs
Use cases
Question
Point of view
IoT
Voice Command
Face ID
System Dimensions
The Problem
Conclusion
Object detection
Embedded vs cloud
AI and Embedded Systems
Challenges
Risk 5 and AI modeling
TinyML vs microcontrollers
Distributed AI
Taught by
tinyML
Related Courses
-
TinyML Talks South Africa - An Introduction to TinyML for All Backgrounds with Hands-On Introduction
-
TinyML Talks Morocco - Enabling Ultra-Low Power Always-On Computer Vision at Qualcomm
-
TinyML Talks France - How to Design a Power Frugal Hardware for AI - The Bio-Inspiration Path
-
TinyML Talks - The New Neuromorphic Analog Signal Processor Concept and Technology Platform
-
Lightweight Face Detection for Ultra-Low Power AI Processors - tinyML Asia 2021
