Machine Learning for iOS Developers

Harness the power of Apple iOS machine learning (ML) capabilities and learn the concepts and techniques necessary to be a successful Apple iOS machine learning practitioner! Machine earning (ML) is the science of getting computers to act without being explicitly programmed. A branch of Artificial Intelligence (AI), machine learning techniques offer ways to identify trends, forecast behavior, and make recommendations. The Apple iOS Software Development Kit (SDK) allows developers to integrate ML services, such as speech recognition and language translation, into mobile devices, most of which can be used in multi-cloud settings. Focusing on Apple’s ML services, Machine Learning for iOS Developers is an up-to-date introduction to the field, instructing readers to implement machine learning in iOS applications. Assuming no prior experience with machine learning, this reader-friendly guide offers expert instruction and practical examples of ML integration in iOS. Organized into two sections, the book’s clearly-written chapters first cover fundamental ML concepts, the different types of ML systems, their practical uses, and the potential challenges of ML solutions. The second section teaches readers to use models—both pre-trained and user-built—with Apple’s CoreML framework. Source code examples are provided for readers to download and use in their own projects. This book helps readers: Understand the theoretical concepts and practical applications of machine learning used in predictive data analytics Build, deploy, and maintain ML systems for tasks such as model validation, optimization, scalability, and real-time streaming Develop skills in data acquisition and modeling, classification, and regression. Compare traditional vs. ML approaches, and machine learning on handsets vs. machine learning as a service (MLaaS) Implement decision tree based models, an instance-based machine learning system, and integrate Scikit-learn & Keras models with CoreML Machine Learning for iOS Developers is a must-have resource software engineers and mobile solutions architects wishing to learn ML concepts and implement machine learning on iOS Apps.

Introduction xix Part 1 Fundamentals of Machine Learning 1 Chapter 1 Introduction to Machine Learning 3 What is Machine Learning? 4 Tools Commonly Used by Data Scientists 4 Common Terminology 5 Real-World Applications of Machine Learning 7 Types of Machine Learning Systems 8 Supervised Learning 9 Unsupervised Learning 10 Semisupervised Learning 11 Reinforcement Learning 11 Batch Learning 12 Incremental Learning 12 Instance-Based Learning 13 Model-Based Learning 13 Common Machine Learning Algorithms 13 Linear Regression 14 Support Vector Machines 15 Logistic Regression 19 Decision Trees 21 Artificial Neural Networks 23 Sources of Machine Learning Datasets 24 Scikit-learn Datasets 24 AWS Public Datasets 27 Kaggle.com Datasets 27 UCI Machine Learning Repository 27 Summary 28 Chapter 2 The Machine-Learning Approach 29 The Traditional Rule-Based Approach 29 A Machine-Learning System 33 Picking Input Features 34 Preparing the Training and Test Set 39 Picking a Machine-Learning Algorithm 40 Evaluating Model Performance 41 The Machine-Learning Process 44 Data Collection and Preprocessing 44 Preparation of Training, Test, and Validation Datasets 44 Model Building 45 Model Evaluation 45 Model Tuning 45 Model Deployment 46 Summary 46 Chapter 3 Data Exploration and Preprocessing 47 Data Preprocessing Techniques 47 Obtaining an Overview of the Data 47 Handling Missing Values 57 Creating New Features 60 Transforming Numeric Features 62 One-Hot Encoding Categorical Features 64 Selecting Training Features 65 Correlation 65 Principal Component Analysis 68 Recursive Feature Elimination 70 Summary 71 Chapter 4 Implementing Machine Learning on Mobile Apps 73 Device-Based vs Server-Based Approaches 73 Apple’s Machine Learning Frameworks and Tools 75 Task-Level Frameworks 75 Model-Level Frameworks 76 Format Converters 76 Transfer Learning Tools 77 Third-Party Machine-Learning Frameworks and Tools 78 Summary 79 Part 2 Machine Learning with CoreML, CreateML, and TuriCreate 81 Chapter 5 Object Detection Using Pre- trained Models 83 What is Object Detection? 83 A Brief Introduction to Artificial Neural Networks 86 Downloading the ResNet50 Model 92 Creating the iOS Project 92 Creating the User Interface 95 Updating Privacy Settings 100 Using the Resnet50 Model in the iOS Project 100 Summary 109 Chapter 6 Creating an Image Classifier with the Create ML App 111 Introduction to the Create ML App 112 Creating the Image Classification Model with the Create ML App 113 Creating the iOS Project 117 Creating the User Interface 118 Updating Privacy Settings 122 Using the Core ML Model in the iOS Project 123 Summary 132 Chapter 7 Creating a Tabular Classifier with Create ML 135 Preparing the Dataset for the Create ML App 135 Creating the Tabular Classification Model with the Create ML App 143 Creating the iOS Project 147 Creating the User Interface 148 Using the Classification Model in the iOS Project 156 Testing the App 172 Summary 173 Chapter 8 Creating a Decision Tree Classifier r 175 Decision Tree Recap 175 Examining the Dataset 176 Creating Training and Test Datasets 180 Creating the Decision Tree Classification Model with Scikit-learn 181 Using Core ML Tools to Convert the Scikit-learn Model to the Core ML Format 186 Creating the iOS Project 187 Creating the User Interface 188 Using the Scikit-learn Decision Tree Classifier Model in the iOS Project 193 Testing the App 201 Summary 202 Chapter 9 Creating a Logistic Regression Model Using Scikit-learn and Core ML 203 Examining the Dataset 203 Creating a Training and Test Dataset 208 Creating the Logistic Regression Model with Scikit-learn 210 Using Core ML Tools to Convert the Scikit-learn Model to the Core ML Format 216 Creating the iOS Project 218 Creating the User Interface 219 Using the Scikit-learn Model in the iOS Project 225 Testing the App 232 Summary 233 Chapter 10 Building a Deep Convolutional Neural Network with Keras 235 Introduction to the Inception Family of Deep Convolutional Neural Networks 236 GoogLeNet (aka Inception-v1) 236 Inception-v2 and Inception-v3 238 Inception-v4 and Inception-ResNet 239 A Brief Introduction to Keras 244 Implementing Inception-v4 with the Keras Functional API 246 Training the Inception-v4 Model 259 Exporting the Keras Inception-v4 Model to the Core ML Format 269 Creating the iOS Project 270 Creating the User Interface 271 Updating Privacy Settings 276 Using the Inception-v4 Model in the iOS Project 277 Summary 286 Appendix A Anaconda and Jupyter Notebook Setup 287 Installing the Anaconda Distribution 287 Creating a Conda Python Environment 288 Installing Python Packages 291 Installing Jupyter Notebook 293 Summary 296 Appendix B Introduction to NumPy and Pandas 297 NumPy 297 Creating NumPy Arrays 297 Modifying Arrays 301 Indexing and Slicing 304 Pandas 305 Creating Series and Dataframes 305 Getting Dataframe Information 307 Selecting Data 311 Summary 313 Index 315