What's new in Android Machine Learning
summary
Android claims an updatable, integrated on‑device ML stack: TensorFlow Lite for Android ships preinstalled via Google Play Services so apps no longer bundle TF Lite, and automatic acceleration uses model metadata to pick CPU/GPU/other backends. Numbers: 730 million devices expose NNAPI and ML Kit hits >500 million monthly users. Android moves NNAPI out of the OS into Google Play Services for updateable NNAPI, Qualcomm is a launch partner, and an Early Access Program is at g.co/androidmleap.
AI-written summary · about
chapters (5)
- 0:00Start
- 1:54Common challenges
- 2:34Announcing Android's updateable, fully integrated ML inference stack
- 5:07Optimising performance
- 6:57Neural Networks API ... like 2 extra CPU cores
related talks
13mWhat's new in Android machine learning
43mWhat’s new in Android machine learning
47mMyths and Reality of Android Machine Learning
39mMachine Learning on Android Demystified
10mMachine Learning for mobile
43mMyths and Reality of Android Machine Learning
38mNEURAL NETWORKS ON YOUR MOBILE PHONE. DIVING INTO THE NN API
36mMaking intelligent Android apps