Machine Learning Without Batteries - The Case for Light-Powered TinyML

Explore a comprehensive talk on light-powered tinyML and batteryless machine learning systems. Delve into the growth of energy harvesting technologies and their potential applications in embedded machine learning and IoT. Learn about a design methodology for smart batteryless sensors capable of data collection, processing, and wireless transmission of inference results. Discover the newly developed gesture detection feature for the open-source MiroCard, and examine the cost-benefits and privacy implications of batteryless sensing systems. Gain insights into topics such as device life cycles, predictability resilience, energy considerations during design, measuring energy production, and the compatibility and reliability of batteryless systems. Understand the importance of data streaming, public data, and training results in this emerging field. Witness a demo and explore collaboration opportunities while considering execution time energy and strategic partnerships in the realm of light-powered tinyML.

Introduction
Internet of Things
Current sensors
About me
Device life cycle
Predictability resilience
Recharge cycles
Energy during design
Questions
Measuring energy production
Results
Batteryless systems
Compatibility
Leakage
Reliability
Privacy
Data Streaming
Public Data
Data Acquisition
Training Results
Cold Start
Demo
Collaboration
Execution time energy
Strategic partners