Chapter 1 Introduction 1
1.1 Mechanism Design 2
1.1.1 Social Choice Function 2
1.1.2 Mechanism 2
1.1.3 Implementation 3
1.1.4 Revelation Principle 4
1.1.5 Efficient Mechanisms 5
1.2 Auctions 7
1.3 Why AI-Driven 11
1.3.1 Challenges in Auction Design 11
1.3.2 The AI-Driven Framework 12
1.4 Organization of the Book 13
References 14
Chapter 2 Multi-Dimensional Mechanism Design via AI-Driven Approaches 16
2.1 Recovering Optimal Mechanisms with Simple Neural Networks 16
2.1.1 Background 17
2.1.2 Setting 19
2.1.3 Revisiting the Na\\\i ve Mechanism 21
2.1.4 Network Structure of MenuNet 24
2.1.5 Recovering Known Results 27
2.2 Discovering Unknown Optimal Mechanisms 30
2.2.1 Experiment Results 31
2.2.2 Theoretic Analysis and Formal Proofs 34
2.3 Performance 52
References 56
Chapter 3 Dynamic Mechanism Design via AI-Driven Approaches 59
3.1 Dynamic Cost-Per-Action Auctions with Ex-Post IR Guarantees 60
3.1.1 Background 60
3.1.2 Our Contributions 62
3.1.3 Related Works 63
3.1.4 Setting and Preliminaries 64
3.1.5 Mechanisms 70
3.1.6 Truthfulness and Implementation 74
3.1.7 Impossibility Result 80
3.2 Dynamic Reserve Pricing via Reinforcement Mechanism Design 80
3.2.1 Background 81
3.2.2 Settings and Preliminaries 86
3.2.3 Bidder Behavior Model 88
3.2.4 Dynamic Mechanism Design as Markov Decision Process 93
References 103
Chapter 4 Multi-Objective Mechanism Design via AI-Driven Approaches 109
4.1 Balancing Objectives through Approximation Analysis 110
4.1.1 Background 110
4.1.2 Settings and Preliminaries 113
4.1.3 Generalized Virtual-Efficient Mechanisms 114
4.1.4 Experiments 126
4.2 Balancing Objectives through Machine Learning 128
4.2.1 Background 129
4.2.2 Market Clearing Loss 132
4.2.3 Theoretical Guarantees 138
4.2.4 Empirical Evaluation 140
References 146
Chapter 5 Summary and Future Directions 151
References 153
內容試閱：
In recent decades， the area of mechanism design has undergone remarkable advancements. Among all its applications， online ad auctions stand out as one of the most important industries deeply rooted in mechanism design theory. These auctions have become a major revenue source for Internet giants like Google， Amazon， Alibaba， and Facebook.
Despite the huge success of mechanism design， a large gap persists between theory and practice. Take， for instance， the design of auction rules. While we have a clear understanding of revenue-maximizing mechanisms for simple scenarios， such as selling a single item， the problem becomes very challenging when multiple items are involved due to the vast design space. Furthermore， mechanism design theory often operates under the assumption that all buyers are fully rational actors and have access to enough information and computational power to figure out the optimal strategy.
This draws a sharp contrast to the diverse goals and irrational behaviors of real-world buyers. Besides， online ad auction platforms possess a large amount of bidding data that the conventional mechanism design theory overlooks， data that could revolutionize the way of designing auctions optimized for real-world performance.
This book emerges as a bridge over these gaps， uniting mechanism design theory with the powerful tools of artificial intelligence. This fusion harnesses the flexibility of AI techniques to manage vast datasets while preserving the economic properties from theoretical analyses. We aim to demonstrate the multifaceted applications of AI techniques in the domain of mechanism design. We hope this perspective will offer both researchers and practitioners a fresh point of view for studying these intricate problems.
Moreover， we explore how computer science and economics can mutually enrich each other， promoting interdisciplinary collaboration. The union of these disciplines not only addresses the deficiencies in current theory but also opens up new possibilities for research and application.
購物車/收銀台(
0
) | 在線留言板
| 付款方式
| 運費計算
| 聯絡我們
| 幫助中心 |
加入書簽
HOME
新書上架
