Parallel Population and Parallel Human A Cyber-Physical Social Approach

Proposes a new paradigm to investigate an individual’s cognitive deliberation in dynamic human-machine interactions

Today, intelligent machines enable people to interact remotely with friends, family, romantic partners, colleagues, competitors, organizations, and others. Virtual reality (VR), augmented reality (AR), artificial intelligence (AI), mobile social media, and other technologies have been driving these interactions to an unprecedented level. As the complexity in system control and management with human participants increases, engineers are facing challenges that arise from the uncertainty of operators or users.

Parallel Population and Parallel Human: A Cyber-Physical Social Approach presents systemic solutions for modeling, analysis, computation, and management of individuals’ cognition and decision-making in human-participated systems, such as the MetaVerse. With a virtual-real behavioral approach that seeks to actively prescribe user behavior through cognitive and dynamic learning, the authors present a parallel population/human model for optimal prescriptive control and management of complex systems that leverages recent advances in artificial intelligence. Throughout the book, the authors address basic theory and methodology for modeling, describe various implementation techniques, highlight potential acceleration technologies, discuss application cases from different fields, and more. In addition, the text:

• Considers how an individual’s behavior is formed and how to prescribe their behavioral modes
• Describes agent-based computation for complex social systems based on a synthetic population from realistic individual groups
• Proposes a universal algorithm applicable to a wide range of social organization types
• Extends traditional cognitive modeling by utilizing a dynamic approach to investigate cognitive deliberation in highly time-variant tasks
• Presents a new method that can be used for both large-scale social systems and real-time human-machine interactions without extensive experiments for modeling

Parallel Population and Parallel Human: A Cyber-Physical Social Approach is a must-read for researchers, engineers, scientists, professionals, and graduate students who work on systems engineering, human-machine interaction, cognitive computing, and artificial intelligence.

Peijun Ye is an Associate Professor at the State Key Laboratory for Management and Control of Complex Systems with the Institute of Automation at the Chinese Academy of Sciences. His research mainly focuses on cognitive computing, artificial intelligence, computational social science and intelligent transportation systems. He is an Associate Editor for several IEEE Transactions and journals.

Fei-Yue Wang is Director of the State Key Laboratory for Management and Control of Complex Systems with the Institute of Automation at the Chinese Academy of Sciences (CAS), where he founded the Intelligent Control and Systems Engineering Center. He is an AAAS Fellow, ASME Fellow, IFAC Fellow, INCOSE Fellow, IEEE Fellow and an Outstanding Scientist of ACM. He has been serving as the Editor-in-Chief of several IEEE Transactions and journals.

Preface xi

Acknowledgments xv

1 From Behavioral Analysis to Prescription 1

1.1 Social Intelligence 1

1.2 Human–Machine Interaction 4

1.3 From Behavior Analysis to Prescription 6

1.4 Parallel Population and Parallel Human 9

1.5 Central Themes and Structure of this Book 11

References 13

2 Basic Population Synthesis 17

2.1 Problem Statement and Data Sources 18

2.1.1 Cross-Classification Table 18

2.1.2 Sample 21

2.1.3 Long Table 22

2.2 Sample-Based Method 23

2.2.1 Iterative Proportional Fitting Synthetic Reconstruction 23

2.2.2 Combinatorial Optimization 24

2.2.3 Copula-Based Synthesis 26

2.3 Sample-Free Method 30

2.4 Experiment Results 37

2.4.1 Copula-Based Population Synthesis 37

2.4.2 Joint Distribution Inference 43

2.5 Conclusions and Discussions 52

References 53

3 Synthetic Population with Social Relationships 55

3.1 Household Integration in Synthetic Population 56

3.2 Individual Assignment 60

3.2.1 Heuristic Allocation 62

3.2.2 Iterative Allocation 65

3.3 Heuristic Search 67

3.4 Joint Distribution Fitting 70

3.5 Deep Generative Models 75

3.6 Population Synthesis with Multi-social Relationships 78

3.6.1 Limitations of IPU Algorithm 78

3.6.2 Population with Multi-social Relationships 85

3.7 Conclusions and Discussions 93

References 95

4 Architecture for Agent Decision Cycle 97

4.1 Parallel Humans in Human–Machine Interactive Systems 98

4.2 Why and What Is the Cognitive Architecture? 100

4.3 Architecture for Artificial General Intelligence 103

4.4 Architecture for Control 110

4.5 Architecture for Knowledge Discovery 113

4.6 Architecture for Computational Neuroscience 116

4.7 Architecture for Pattern Recognition 121

4.8 Other Representative Architecture 123

4.9 TiDEC: A Two-Layered Integrated Cycle for Agent Decision 129

4.10 Conclusions and Discussions 134

References 135

5 Evolutionary Reasoning 141

5.1 Knowledge Representation 142

5.2 Evolutionary Reasoning Using Causal Inference 147

5.3 Learning Fitness Function from Expert Decision Chains 156

5.4 Conclusions and Discussions 158

References 158

6 Knowledge Acquisition by Learning 163

6.1 Foundation of Knowledge Repository Learning 164

6.2 Knowledge Acquisition Based on Self-Supervised Learning 167

6.3 Adaptive Knowledge Extraction for Data Stream 170

6.3.1 Neural-Symbolic Learning 171

6.3.2 Explanation of Deep Learning 176

6.4 Experiment on Travel Behavior Learning 183

6.5 Conclusions and Discussions 194

References 195

7 Agent Calibration and Validation 199

7.1 Model Calibration for Agent 199

7.2 Calibration Based on Optimization 202

7.3 Calibration Based on Machine Learning 209

7.4 Calibration Based on Cybernetics 214

7.5 Calibration Using Variational Auto-Encoder 227

7.6 Conclusions and Discussions 233

References 233

8 High-Performance Computing for Computational Deliberation Experiments 237

8.1 Computational Acceleration Using High-Performance Computing 237

8.1.1 Spark with Hadoop 238

8.1.2 MPI/OpenMP on Supercomputing 244

8.2 Computational Deliberation Experiments in Cloud Computing 249

8.3 Computational Deliberation Experiments in Supercomputing 258

8.4 Conclusions and Discussions 262

References 263

9 Interactive Strategy Prescription 265

9.1 Hierarchical Behavior Prescription System 266

9.2 Dynamic Community Discovery for Group Prescription 270

9.3 Strategy Prescription Based on Content Match 273

9.4 Active Learning in Strategy Prescription 278

9.5 Conclusions and Discussions 284

References 285

10 Applications for Parallel Population/Human 287

10.1 Population Evolution 287

10.2 Computational Experiments for Travel Behavior 292

10.3 Parallel Travel Behavioral Prescription 295

10.4 Travel Behavioral Prescription for Sports Event 301

10.5 Conclusions and Discussions 305

References 306

11 Ethical and Legal Issues of Parallel Population/Human 307

11.1 Relationships Between the Parallel Population/Human and Its Individual Users 308

11.2 Authority of the Parallel Population/Human System 309

11.3 Risk Management and Responsibility Identification 310

11.4 Conclusions and Discussions 311

References 311

Appendix A Convergence for Multivariate IPF 313

Index 321

The New copy of this book will include any supplemental materials advertised. Please check the title of the book to determine if it should include any access cards, study guides, lab manuals, CDs, etc.

The Used, Rental and eBook copies of this book are not guaranteed to include any supplemental materials. Typically, only the book itself is included. This is true even if the title states it includes any access cards, study guides, lab manuals, CDs, etc.