Next generation machine learning for mobile and embedded platforms
summary
Company claims a production ML platform that builds and compiles sensor models into C libraries, proven on ARM Cortex‑M0–M4 targets and used to ship 200 million licenses. Pipeline automates data cleaning, sensor selection, feature extraction, model search (logistic, RF, GBM, neural nets), quantization and model compression, reports cross‑validation, latency and library size, and generates flashable binaries. Handles many phone sensor hardware variants (up to 24 per model) and demonstrates live building of gesture classifiers in minutes with real‑time latency tests.
AI-written summary · about
chapters (9)
- 0:00Start
- 0:47Company and Core Competencies
- 2:12Qeexo AutoML: Genesis
- 6:01Qeexo ML Automation Platform for Sensor Data
- 7:46Fully Automated, Behind the Scenes Machine Learning
- 19:30Models and Training Interaction
- 20:14Performance Reporting
- 21:10Qeexo AutoML Advantage
- 24:20Supported Sensors
related talks
10mMachine Learning for mobile
01h 01mHow to Optimize, Validate and Deploy ML Models on Device
25mMachine Learning on Microcontrollers; Tiny, Useful and Fun
15mSimplified Machine Learning on mobile, browsers, and edge devices with on-device ML
09mMachine Learning on the Edge
43mWhat’s new in Android machine learning
32mAdvanced on-device ML made easy
24mML Kit: Machine learning for mobile developers