Machine Learning for mobile
summary
ML Kit offers 13 production-ready on-device APIs, and TF Lite runs on more than 4 billion devices while adding on-device training and a Play Services runtime to avoid APK bloat. XNNPack yields about 2.3x faster CPU ops. Microcontroller support fits models in a few kilobytes. MLOps tooling includes TFX with fairness and Model Card integration plus LIT v0.4 (TCAV support). Managed Vertex AI reports training +70%, predictions and GPU hours ×2, API requests ×3, and Waze sees up to 19% lift.
AI-written summary · about
chapters (10)
- 0:00Introduction to on-device ML
- 0:30ML Kit for common use cases
- 1:04TensorFlow Lite and on-device
- 2:29Model deployment and examples
- 3:44TinyML for microcontrollers
- 4:22Introduction to MLOps and TFX
- 5:33OpenX production case study
- 6:20Responsible AI and tools
- 7:12Vertex AI managed platform
- 8:21Waze case study and conclusion
related talks
39mFrom TensorFlow to ML Kit: power your mobile application with machine learning
21mGetting the best of on-device ML with ML Kit
38mML Kit: Machine Learning for Mobile with Firebase
29mFrom TensorFlow to ML Kit: power your Android application with ML
13mWhat's new in Android machine learning
24mML Kit: Machine learning for mobile developers
12mML Kit: Turnkey APIs to use on-device ML in mobile apps
39mMachine Learning on Android Demystified