Data Centric Artificial Intelligence: A Beginner’s Guide (Data-Intensive Research)

This book discusses the best research roadmaps, strategies, and challenges in data-centric approach of artificial intelligence (AI) in various domains. It presents comparative studies of model-centric and data-centric AI. It also highlights different phases in data-centric approach and data-centric principles. The book presents prominent use cases of data-centric AI. It serves as a reference guide for researchers and practitioners in academia and industry. Preface 6 Contents 8 About the Authors 11 1 Introduction 13 1.1 Building Blocks of AI 13 1.2 AI Current State 15 1.3 Motivation 17 1.4 Need for Paradigm Shift from Model-Centric AI to Data-Centric AI 18 1.5 Summary 20 References 21 2 Model-Centric AI 23 2.1 Working Principle 23 2.1.1 Supervised Learning 24 2.1.2 Unsupervised Learning 25 2.1.3 Reinforcement Learning 25 2.2 Learning Methods 26 2.2.1 Supervised Machine Learning Algorithms 26 2.2.2 Unsupervised Machine Learning Algorithms 30 2.2.3 Deep Learning Algorithms 31 2.3 Model Building 37 2.4 Model Training 38 2.5 Model Testing 38 2.6 Model Tuning 39 2.7 Use Cases: Model-Centric AI 40 2.8 Summary 42 References 43 3 Data-Centric Principles for AI Engineering 45 3.1 Overview 45 3.2 AI Engineering 46 3.3 Challenges 51 3.4 Data-Centric Principles 53 3.5 Summary 56 References 56 4 Mathematical Foundation for Data-Centric AI 58 4.1 Overview 58 4.1.1 Statistics 58 4.1.2 Linear Algebra 58 4.1.3 Calculus 59 4.1.4 Probability Theory 59 4.1.5 Multivariate Calculus 61 4.1.6 Graph Theory 61 4.2 Statistical Data Analysis 62 4.3 Data Tendency and Distribution 64 4.3.1 Data Tendency/Measure of Central Tendency 64 4.3.2 Measure of Dispersion 66 4.3.3 Data Distribution 69 4.4 Data Models 72 4.5 Optimization Techniques 73 4.6 Summary 77 References 77 5 Data-Centric AI 78 5.1 Data Acquisition 78 5.1.1 The Data Acquisition Process 79 5.1.2 Key Insights for Big Data Acquisition 79 5.1.3 Case Study: Data Acquisition for Retail Company 80 5.2 Data Labeling 81 5.2.1 How Does Data Labeling Work? 82 5.2.2 Data Labeling Approaches 83 5.2.3 Importance of Data Labeling 85 5.2.4 Case Study: Data Labeling for Autonomous Vehicle Training 86 5.3 Data Annotation 87 5.3.1 Types of Data Annotation 87 5.3.2 Case Study on Data Annotation 88 5.4 Data Augmentation 89 5.4.1 How Does Data Augmentation Work? 90 5.4.2 Case Study on Data Augmentation 91 5.5 Data Deployment 92 5.5.1 Case Study on Data Deployment 92 5.6 Data-Centric AI Tools 93 5.6.1 Case Study: Predicting Customer Churn for a Telecommunications Company 95 5.7 Summary 95 References 96 6 Data-Centric AI in Healthcare 97 6.1 Overview 97 6.2 Need and Challenges of Data-Centric Approach 99 6.3 Application Implementation in Data-Centric Approach 101 6.4 Application Implementation in Model-Centric Approach 101 6.5 Comparison of Model-Centric AI and Data-Centric AI 103 6.6 Summary 105 References 106 7 Data-Centric AI in Mechanical Engineering 107 7.1 Overview 107 7.2 Need and Challenges of Data-Centric Approach 108 7.3 Application Implementation in Data-Centric Approach 110 7.4 Application Implementation in Model-Centric Approach 112 7.5 Comparison of Model-Centric AI and Data-Centric AI 114 7.6 Case Study: Mechanical Tools Classification 116 7.7 Summary 117 References 117 8 Data-Centric AI in Information, Communication and Technology 119 8.1 Overview 119 8.2 Need and Challenges of Data-Centric Approach 120 8.3 Application Implementation in Data-Centric Approach 123 8.4 Application Implementation in Model-Centric Approach 125 8.5 Comparison of Model-Centric AI and Data-Centric AI 128 8.6 Summary 130 References 134 9 Conclusion 135 9.1 Summary 135 9.2 Research Areas 136 References 137