Interactions in Multiagent Systems: Fairness, Social Optimality and Individual Rationality

verfasst von: Jianye Hao, Ho-fung Leung

Verlag: Springer Berlin Heidelberg

Enthalten in: Springer Professional "Wirtschaft+Technik" , Springer Professional "Technik" , Springer Professional "Wirtschaft"

Einloggen, um Zugang zu erhalten

Über dieses Buch

This book mainly aims at solving the problems in both cooperative and competitive multi-agent systems (MASs), exploring aspects such as how agents can effectively learn to achieve the shared optimal solution based on their local information and how they can learn to increase their individual utility by exploiting the weakness of their opponents. The book describes fundamental and advanced techniques of how multi-agent systems can be engineered towards the goal of ensuring fairness, social optimality, and individual rationality; a wide range of further relevant topics are also covered both theoretically and experimentally. The book will be beneficial to researchers in the fields of multi-agent systems, game theory and artificial intelligence in general, as well as practitioners developing practical multi-agent systems.

Inhaltsverzeichnis

Frontmatter

Chapter 1. Introduction

Abstract

Multiagent systems (MASs) have become a commonly adopted paradigm to model and solve real-world problems. Many competing definitions exist for a MAS. In this book, we consider a typical MAS as a system involving multiple autonomous software agents (or humans) interacting with each other with (possibly) conflicting interests and limited information, and the payoff of each agent is determined by the joint actions of all (or some) agents involved. Therefore, different from single-agent environments, in multiagent interaction environments, each agent needs to take other agents’ behaviors into consideration when it makes its own decisions, since others’ behaviors can directly influence what it expects from the system. The major question we seek to answer in this book can be summarized as follows: how can a desirable goal be achieved in different multiagent interaction environments where each agent may have its own limitations and (possibly) conflicting interests?

Jianye Hao, Ho-fung Leung

Chapter 2. Background and Previous Work

Abstract

In this chapter, we first review some concepts and terminologies from game theory and multiagent systems areas that will be used throughout the book in Sect. 2.1. After that, we explore previous work in multiagent learning literature by dividing them into two major categories: cooperative multiagent systems and competitive multiagent systems, distinguished by the underlying intentional stance of the agents within the system. Within each category, we review previous work in three different parts distinguished by the different goal that the work targets at. In Sect. 2.2, we focus on reviewing previous work whose goal falls into one of the following major solution concepts: Nash equilibrium, fairness, or social optimality. In Sect. 2.3, we focus on investigating previous work targeting at either of the following major solution concepts: Nash equilibrium, maximizing individual benefits, and Pareto-optimality.

Jianye Hao, Ho-fung Leung

Chapter 3. Fairness in Cooperative Multiagent Systems

Abstract

In cooperative MASs, the interests of individual agents are usually consistent with that of the overall system. In the cases of conflicting interest, each agent is assumed to have the willingness to cooperate toward a common goal of the system even at the cost of sacrificing its own benefits. Therefore, the behaviors of each agent in cooperative environments can be determined by the designer(s) of the system, which thus allows for intricate coordination strategies to be implemented beforehand. To achieve effective coordinations among agents, one traditional approach is to employ a superagent to determine the behaviors for all other agents in the system. However, there exist a number of disadvantages by adopting this approach. First, the scalability problem will become serious when the number of agents is significantly increased, since the computational space of the superagent increases exponentially to the number of agents. Second, it explicitly requires the superagent to be able to communicate with all agents in the system and has the global information, which may not be possible in distributed environments, where the communication cost can be very high. Lastly, it makes the system very vulnerable since the malfunction of the superagent would lead to the failure of the whole system.

Jianye Hao, Ho-fung Leung

Chapter 4. Social Optimality in Cooperative Multiagent Systems

Abstract

In this chapter, we turn to look at another important solution concept called social optimality which targets at maximizing the sum of all agents’ payoffs involved. A socially optimal outcome is desirable in that it is not only optimal from the system-level’s perspective but also Pareto optimal. To achieve socially optimal outcomes in cooperative environments, the major challenge is how each agent can coordinate effectively with others given limited information, since the behaviors of other agents coexisting in the system may significantly impede the coordination process among them. The coordination problem becomes more difficult when the environment is uncertain (or stochastic) and each agent can only interact with its local partners if we consider a topology-based interaction environment [1, 6].

Jianye Hao, Ho-fung Leung

Chapter 5. Individual Rationality in Competitive Multiagent Systems

Abstract

In competitive MASs, each individual agent is usually interested in maximizing its personal benefits only, which may have conflicts with the utility of others and the overall system as well. Thus, one natural research direction in competitive MASs is to consider how an agent can learn to obtain as much utility as possible against different opponents based on its local information. Another important question is raised from the system designer’s perspective, i.e., how can the selfish agents be incentivized to coordinate their behaviors to maximize the system-level performance (i.e., maximizing social optimality)? In this chapter, we focus on the first research direction by considering an important competitive multiagent interaction scenario: bilateral negotiation [1]. The second research direction will be the focus of Chap. 6

Jianye Hao, Ho-fung Leung

Chapter 6. Social Optimality in Competitive Multiagent Systems

Abstract

In this chapter, we turn to investigate the question of how socially optimal solutions can be achieved in competitive MASs which consist of individually rational agents. In competitive multiagent environments, the individually rational agents are assumed to be only interested in maximizing their individual benefits and may not be willing to follow the (socially oriented) strategy specified by the system designer. If each agent behaves in a purely individually rational manner, it usually leads to the nonsocially optimal outcomes, thus decreasing the overall system’s utilities. Therefore we usually need to resort to designing effective mechanisms to motivate those selfish agents to change their behaviors toward coordinating on socially optimal outcomes. Specifically in this chapter we look at how to handle this problem within two major learning frameworks in MASs. The first one is the infinitely repeated game learning framework [1, 2], which will be introduced in Sect. 6.1. The second one is the social learning framework which will be introduced in Sect. 6.2 [3].

Jianye Hao, Ho-fung Leung

Chapter 7. Conclusion

Abstract

This book mainly investigates the general question of how a desirable goal can be achieved given that each agent may have some limitations (in terms of communication and environment information) within two major multiagent interaction environments: cooperative multiagent environment and competitive multiagent environment. In cooperative multiagent environment, we focused on the solution concepts of fairness and social optimality and described a number of efficient learning strategies for agents to coordinate on fair or socially optimal outcomes for different multiagent interaction problems, which are summarized as follows:

Jianye Hao, Ho-fung Leung

Backmatter

Titel: Interactions in Multiagent Systems: Fairness, Social Optimality and Individual Rationality
verfasst von: Jianye Hao
Ho-fung Leung
Verlag: Springer Berlin Heidelberg
Electronic ISBN: 978-3-662-49470-7
Print ISBN: 978-3-662-49468-4
DOI: https://doi.org/10.1007/978-3-662-49470-7