Intro to On Device Machine Learning

Timecodes: 00:00 Topic Introduction 00:24 Edge ML Explosion 03:05 Apps on Device Machine Learning 04:51 What is TensorFlow Lite? 08:03 4B Mobile Devices 08:33 Text Classification 12:10 High Level APIs 18:19 What is a microcontroller or DSP? 20:15 TensorFlow: Easy for Developers 21:00 Hardware for Optimised Performance 21:29 How to get started? 21:58 Optimisation is important 22:49 Methods 23:15 Model Architecture 24:43 Quantisation 26:02 Pruning 27:33 Hardware Accelerator 28:42 TensorFlow Models 29:30 Image Segmentation 31:35 Preparing a Training Data 35:32 Q1 Interested in deploying a coral on edge inspection machine. How common is it? 36:15 Q2 Can you use a knowledge distillation to close the accuracy gap between big/small? This session of #GoogleIO Extended is organised by GDG Kuala Lumpur and WomenTechmakers Kuala Lumpur. * Name: Poo Kuan Hoong (Lead Data Scientist) Topic: An Introduction to On Device Machine Learning There are many different type of devices with different hardware specific