#Machine Learning Projects for Mobile Applications
My book on building Android and iOS applications with TensorFlow Lite and Core ML. Hands-on machine learning that runs on the device, not in the cloud.
I wrote Machine Learning Projects for Mobile Applications for Packt to answer a specific question: how do you take a trained model and actually run it inside a mobile app, on the phone, at a size and speed a real device can handle? It works through projects, image classification, text and video processing, and deep-learning tasks, each built end to end for Android and iOS using TensorFlow Lite and Core ML.
It is the earlier of my two mobile-AI books. I followed it with Mobile Artificial Intelligence Projects, which widens the same idea into seven projects spanning NLP, computer vision, and robotics. If you are choosing between them: this one is the tighter, model-to-mobile deployment focus; the later book is the broader project tour.
## What it covers
The core is the part most tutorials skip: getting a model off your laptop and onto a phone. Converting and quantizing models for TensorFlow Lite and Core ML, running inference on-device, and handling the constraints, size, memory, latency, that a mobile target imposes and a cloud API lets you ignore. It is written for engineers who can already build a model and want it to ship on a device.
## Why on-device, and where it went next
Both books are built on one conviction: the intelligence should run on the phone. That was a harder argument in 2018 than it is now, and I have kept building on it. DailyVox is the current version, a privacy-first on-device iOS app, and the modern architecture behind it is written up in the on-device AI app architecture and the economics of on-device AI. If you want my honest take on how to use books like this one, see why you should read fewer ML books and build more.