This is an introductory book for professionals and undergraduate students without any background in artificial intelligence. The book describes how the new artificial intelligence technology, neural network, totally changes our everyday life. It first describes the neural network development history and introduces the classical neural network – convolutional neural network architecture. After that, it covers the artificial development in six different areas; healthcare, finance, retail, manufacturing, agriculture, and smart city. This book targets students with different backgrounds, such as, business, humanity, arts, science and engineering, allowing them to understand the artificial intelligent general applications and how this impacts on their future.

Albert Liu, PhD, is the CEO of Kneron and an Adjunct Associate Professor at National Tsing Hua University, National Chiao Tung University, and National Cheng Kung University, Taiwan.

Oscar Ming Kin Law, PhD, is currently a senior staff member at Qualcomm Inc. He has over 20 years' experience in the semiconductor industry and has published more than 60 patents in various areas.

Iain Law is a rising senior at Canyon Crest Academy, San Diego. He has worked on several artificial intelligence projects including the LEGO smart robot and DJI Tello smart drone.

1 Introduction 1

1.1 Overview 1

1.2 Development History 3

1.3 Neural Network Model 6

1.4 Popular Neural Network 7

1.4.1 Convolutional Neural Network 7

1.4.2 Recurrent Neural Network 8

1.4.3 Reinforcement Learning 9

1.5 Neural Network Classification 9

1.5.1 Supervised learning 10

1.5.2 Semi-supervised learning 10

1.5.3 Unsupervised learning 11

1.6 Neural Network Operation 11

1.6.1 Training 11

1.6.2 Inference 12

1.7 Application Development 12

1.7.1 Business Planning 14

1.7.2 Network Design 14

1.7.3 Data Engineering 14

1.7.4 System Integration 15

Exercise 16

2 Neural Network 17

2.1 Convolutional Layer 19

2.2 Activation Layer 20

2.3 Pooling Layer 21

2.4 Batch Normalization 22

2.5 Dropout Layer 22

2.6 Fully Connected Layer 23

Exercise 24

3 Machine Vision 25

3.1 Object Recognition 25

3.2 Feature Matching 27

3.3 Facial Recognition 28

3.4 Gesture Recognition 30

3.5 Machine Vision Applications 31

3.5.1 Medical Diagnosis 31

3.5.2 Retail Applications 32

3.5.3 Airport Security 33

Exercise 34

4 Natural Language Processing 35

4.1 Neural Network Model 36

4.1.1 Convolutional Neural Network 36

4.1.2 Recurrent Neural Network 37

4.1.2.1 Long Short-Term Memory Network 38

4.1.3 Recursive Neural Network 39

4.1.4 Reinforcement Learning 40

4.2 Natural Language Processing Applications 41

4.2.1 Virtual Assistant 41

4.2.2 Language Translation 42

4.2.3 Machine Transcription 43

Exercise 45

5 Autonomous Vehicle 46

5.1 Levels of Driving Automation 46

5.2 Autonomous Technology 48

5.2.1 Computer Vision 48

5.2.2 Sensor Fusion 49

5.2.3 Localization 51

5.2.4 Path Planning 52

5.2.5 Drive Control 52

5.3 Communication Strategies 53

5.3.1 Vehicle-to-Vehicle Communication 54

5.3.2 Vehicle-to-Infrastructure Communication 54

5.3.3 Vehicle-to-Pedestrian Communication 55

5.4 Law Legislation 56

5.4.1 Human Behavior 57

5.4.2 Lability 57

5.4.3 Regulation 58

5.5 Future Challenges 58

5.5.1 Road Rules Variation 58

5.5.2 Unified Communication Protocol 58

5.5.3 Safety Standard and Guideline 59

5.5.4 Weather/Disaster 59

Exercise 60

6 Drone 61

6.1 Drone Design 61

6.2 Drone Structure 62

6.2.1 Camera 63

6.2.2 Gyro Stabilization 63

6.2.3 Collision Avoidance 64

6.2.4 Global Positioning System 64

6.2.5 Sensors 64

6.3 Drone Regulation 65

6.3.1 Recreational Rules 65

6.3.2 Commercial Rules 66

6.4 Applications 66

6.4.1 Infrastructure Inspection 66

6.4.2 Civil Construction 67

6.4.3 Agriculture 68

6.4.4 Emergency Rescue 69

Exercise 70

7 Healthcare 71

7.1 Telemedicine 71

7.2 Medical Diagnosis 72

7.3 Medical Imaging 73

7.4 Smart Medical Device 74

7.5 Electronic Health Record 76

7.6 Medical Billing 77

7.7 Drug Development 78

7.8 Clinical Trial 79

7.9 Medical Robotics 80

7.10 Elderly Care 81

7.11 Future Challenges 82

Exercise 84

8 Finance 85

8.1 Fraud Prevention 85

8.2 Financial Forecast 88

8.3 Stock Trading 89

8.4 Banking 91

8.5 Accounting 94

8.6 Insurance 95

Exercise 96

9 Retail 97

9.1 E-Commerce 98

9.2 Virtual Shopping 100

9.3 Product Promotion 102

9.4 Store Management 103

9.5 Warehouse Management 104

9.6 Inventory Management 106

9.7 Supply Chain 108

Exercise 110

10 Manufacturing 111

10.1 Defect Detection 112

10.2 Quality Assurance 113

10.3 Production Integration 114

10.4 Generative Design 115

10.5 Predictive Maintenance 117

10.6 Environment Sustainability 118

10.7 Manufacturing Optimization 119

Exercise 121

11 Agriculture 122

11.1 Crop and Soil Monitoring 123

11.2 Agricultural Robot 125

11.3 Pest Control 126

11.4 Precision Farming 127

Exercise 129

12 Smart City 130

12.1 Smart Transportation 131

12.2 Smart Parking 132

12.3 Waste Management 133

12.4 Smart Grid 134

12.5 Environmental Conservation 135

Exercise 137

13 Government 138

13.1 Information Technology 140

13.2 Human Service 141

13.3 Law Enforcement 144

13.3.4 Augmenting Human Movement 147

13.4 Homeland Security 147

13.5 Legislation 149

13.6 Ethics 152

13.7 Public Perspective 155

Exercise 159

14 Computing Platform 160

14.1 Central Processing Unit 160

14.1.1 System Architecture 161

14.1.2 Advanced Vector Extension 164

14.1.3 Math Kernel Library for Deep Neural Network 165

14.2 Graphics Processing Unit 165

14.2.1 Tensor Core Architecture 167

14.2.2 NVLink2 Configuration 167

14.2.3 High Bandwidth Memory 169

14.3 Tensor Processing Unit 170

14.3.1 System Architecture 170

14.3.2 Brain Floating Point Format 171

14.3.3 Cloud Configuration 172

14.4 Neural Processing Unit 173

14.4.1 System Architecture 173

14.4.2 Deep Compression 174

14.4.3 Dynamic Memory Allocation 174

14.4.4 Edge AI Server 175

Exercise 176

Appendix A Kneron Neural Processing Unit 178

Appendix B Object Detection (Overview) 179

B.1 Kneron Environment Setup 179

B.2 Python Installation 180

B.3 Library Installation 184

B.4 Driver Installation 185

B.5 Model Installation 186

B.6 Image/Camera Detection 186

B.7 Yolo Class List 190

Appendix C Object Detection - Hardware 192

C.1 Library Setup 192

C.2 System Parameters 193

C.3 NPU Initialization 194

C.4 Image Detection 195

C.5 Camera Detection 197

Appendix D Hardware Transfer Mode 199

D.1 Serial Transfer Mode 199

D.2 Pipeline Transfer Mode 201

D.3 Parallel Transfer Mode 203

Appendix E Object Detection – Software (Optional) 205

E.1 Library Setup 205

E.2 Image Detection 207

E.3 Video Detection 208

Reference 211