What’s TinyML good for
summary
Evidence that embedded ML runs at scale: TensorFlow Lite powers over 3 billion phones, and TensorFlow Lite Micro targets 1 mW budgets. Demonstrated artifacts include a 13‑kilobyte wake‑word model for always‑on DSPs, an 80 MB full speech recognizer running locally on Pixel‑class ARM, and peel‑and‑stick vibration sensors on Ambiq Cortex‑M4 that last five years on batteries for predictive maintenance. Expect low‑memory vision, music‑detection always‑on use cases, toy gesture demos, and industrial sensor deployments.
AI-written summary · about
chapters (17)
- 0:00Intro
- 0:48What is TinyML
- 1:27Machine Learning
- 3:49Machine Learning on Phones
- 5:13Unmet Needs
- 7:11TinyML Basics
- 9:53Challenges
- 11:09Wake Word
- 12:59Speech Recognition
- 14:54Voice Interfaces
- 17:20Pixel Music Detection
- 18:13Predictive Maintenance
- 20:07Toys
- 21:50Motive
- 22:26Environmental
- 23:25Building Automation
- 24:29Medical
related talks
01h 53mTinyML: Machine Learning’s Future is Tiny and Bright
31mKeynote: The next big thing in machine learning is tiny
25mMachine Learning on Microcontrollers; Tiny, Useful and Fun
01h 01mML and the IoT: Living on the Edge
32mAI for Mobile and IoT Devices: TensorFlow Lite
25mTinyML as-a-Service
29mFrom TensorFlow to ML Kit: power your Android application with ML
54mSmarter Everything with ML & the IoT