Build an iOS App With Deep Learning
Use Core ML to add intelligence to your applications
Apple’s CoreML library allows developers to use state-of-the-art deep learning in their iOS and macOS applications. With the advent of Apple silicon, the use of Core ML is about to enter enterprise data centers. Coupled with Apple’s strategy for privacy-preserving AI and the need to analyze sensitive data in fast-growing domains such as AI analysis in medical applications, there’s no better time to learn Apple’s ML ecosystem and CoreML than now.
Join experts Till Lohfink and Stefan Zapf for an exciting tour of the CoreML library. You’ll learn the foundations of deep learning with CoreML by applying it to two real-life scenarios: an app that allows lawyers to use privacy-preserving question-and-answering BERT AIs to search for relevant facts in large amounts of documents right on an iPad in the courtroom, and an iOS brewing diary that allows coffee enthusiasts to take photos of a bag of beans and use a transfer learning and semantic search ResNet AI to search for previous brews with the same coffee.
What you'll learn-and how you can apply it
By the end of this live online course, you’ll understand:
- The CoreML library
- The foundations of state-of-the-art deep learning in NLP question answering (BERT) and image classification (ResNet)
- How to implement privacy-preserving AI
And you’ll be able to:
- Integrate and use CoreML in your iOS and macOS apps
- Use transfer learning to expand CoreML’s capabilities and adapt CoreML to new use cases
- Use on-device machine learning
This training course is for you because...
- You’re an iOS or macOS developer who wants to break into the world of machine learning with CoreML and deep learning models.
- You’re a data scientist or machine learning engineer who wants to apply your skills to the Apple platforms.
- A computer with macOS Big Sur (v.11—required to use Xcode) with the latest stable version of Xcode 12 installed
- Familiarity with the Swift programming language
- Read “Getting Started” and “The Swift Programming Language” (chapters 1 and 2 in Learning Swift, third edition)
About your instructors
Stefan has worked in the field of machine learning for more than a decade. As VP of Engineering, he led the development of the Deep Learning systems for Rainforest Connection, a silicon valley startup that protects the rainforest with AI now in cooperation with Google. As a consultant, Stefan researched Deep Learning, training, and developed AI strategies for clients in the social media and security sectors. He is now working with Till on NLP and privacy-preserving AI in DATEV eG, a German financial services cooperative.
Till helped the German automotive industry make data driven decisions in sales and supply chain management during his consultancy career. He came from a traditional high dimensional data analysis background and embraced Deep Learning solutions. Currently, he is working with Stefan on a privacy-preserving multi-tenant AI solution serving hundreds of thousands of NLP Machine Learning models at the same time in a near real-time scenario for DATEV eG.
The timeframes are only estimates and may vary according to how the class is progressing
Introduction to CoreML (55 minutes)
- Presentation: The CoreML library; foundations of deep learning
- Group discussion: How would you go about building a detector for bad posture while working at the desk? Where would you expect to find the models in CoreML?
Break (5 minutes)
Scenario 1: A legal app for learning to apply Core ML (75 minutes)
- Scenario overview: An app that allows lawyers to use privacy-preserving question-and-answering BERT AIs to search for relevant facts in large amounts of documents right on an iPad in the courtroom (A lawyer shares the requirements for legal—how they could use for briefs and in court.)
- Presentation: NLP (BERT) models
- Hands-on exercise: Using a skeleton app, add CoreML models
- Group discussion: What are other applications of NLP Q&A?
Break (5 minutes)
Scenario 2: A coffee app for learning to apply transfer learning with Core ML (75 minutes)
- Scenario overview: An iOS brewing diary that allows coffee enthusiasts to take a photo of the bag of beans and uses transfer learning and semantic search ResNet AI to search for previous brews with the same coffee (A coffee brewing company shows how to brew speciality coffee, why a brewing diary app is needed, and what their requirements are.)
- Presentation: Image models (ResNet)
- Hands-on exercise: Using a skeleton app, use transfer learning to adapt CoreML models
- Group discussion: What are other applications of semantic search?
Break (5 minutes)
Final thoughts about privacy and scalability (20 minutes)
- Presentation: The future of sensitive AI analysis in fields such as medicine; how privacy and deep learning work together; the scalability of deep learning models
- Group discussion: Can we use adapt-and-deploy strategy outside of iOS apps?