Machine Learning in Mobile Applications

Explore scenarios for using machine learning within mobile development.

Introduction

• Introduction to machine learning in mobile applications
• What you should know to take this class
• Setting up your machine
• Using the exercise files

1. Mobile Developers Primer on Machine Learning

• What is machine learning?
• Required concepts
• Why does this matter for my app?
• Training a model
• Machine learning vs. deep learning vs. generative AI
• What can I do with machine learning?
• Server-side vs. client-side ML
• ML frameworks

2. Server Models: IBM Watson

• Overview of Watson
• Natural Language Understanding: Setup
• watsonx.ai™ AI studio: Setup
• watsonx.ai™ AI studio: Training
• Deploying the model
• Authenticating against a deployed model
• Installing the Watson SDK into your mobile app
• Calling Watson Natural Language Understanding
• Returning a watsonx access token
• Calling a watsonx custom model
• Running the app
• Challenge: Use Natural Language Understanding features
• Solution: Use Natural Language Understanding features

3. Server Models: Azure

• Azure Machine Learning overview
• Language Understanding: Setup
• Language Understanding: Using Language Studio
• Language Understanding: Train, deploy, and test
• Custom Vision: Setup
• Azure Machine Learning Studio: Setup
• Azure Machine Learning Studio: Create a model
• Azure Machine Learning Studio: Deploy and test a model
• Install the SDK in a mobile app
• Tie to Language Understanding
• Tie to Custom Vision
• Prepare Android and iOS apps to consume non-SSL endpoints
• Tie to the Azure Machine Learning Studio model
• Running the app
• Challenge: Create a custom Language Understanding model
• Solution: Create a custom Language Understanding model

4. Client Models: Core ML

• Core ML overview
• Core ML: Create a natural language model
• Core ML: Create a visual recognition model
• Core ML: Create a regression model
• Client tied to a natural language model
• Client tied to a visual recognition model
• Client tied to a regression model
• Running the app
• Challenge: Create a custom model
• Solution: Create a custom model

5. Client Models: ML Kit

• Introduction to ML Kit
• Selecting a model
• Adding the SDK to a mobile app
• Calling the model
• Running the app
• Challenge: Implement the image labeling model
• Solution: Implement the image labeling model

6. Understanding the Offerings

• Different philosophies of the vendors
• Why use client-side vs. server-side models?
• When to use one or another of these solutions

Conclusion

• Where to go from here