Explore the optimization of Automatic Machine Learning (AutoML) for tinyML applications in this conference talk from the tinyML Summit 2022. Delve into the challenges of balancing accuracy, latency, and power efficiency in resource-constrained environments. Learn how Qeexo AutoML leverages hardware-aware techniques to optimize the entire signal flow from raw sensor data to inference results. Discover the process of automatic selection of signal processing filters for STMicroelectronics' Machine Learning Core (MLC) and witness a demonstration of an activity detection wakeup model. Gain insights into the importance of AutoML tools and full signal flow optimization in shaping the future of tinyML.
Overview
Syllabus
Intro
What is kixo
What is AutoML
Basic AutoML capabilities
kixo AutoML
Multiobjective optimization
Hardwareaware optimization
Demonstration
New Project
Results
Signal Processing Chains
AutoML Filters
AutoML Performance
Conclusion
Sponsors
Taught by
tinyML
Related Courses
-
EON Tuner - AutoML for Constrained Devices in TinyML
-
Getting Started with Machine Learning at the Edge on Arm
-
AutoML + TinyML with Edge Impulse's EON Tuner
-
Sensors and Machine Learning: Waking Smarter for Less
-
Adaptive Neural Networks for Agile TinyML - Keynote
-
1 kB and Not a Bit More - The Ideal Weight for a TinyML Model
