Customer Reviews

Multiagent Systems: Algorithmic, Game-Theoretic, and Logical Foundations

5 star:		(2)
4 star:		(2)
3 star:		(1)
2 star:		(0)
1 star:		(0)

 

 

On the positive side, the book gives an indication of the wide variety of work that has been done, and is being done, in the area of multiagent systems. On the negative side, I would have liked to see more work done in producing a taxonomy of the field before digging into the details of particular approaches. The authors also seem to be uncertain exactly how to integrate logic (especially modal logic) with existing and current work. As a result, rather than playing a foundational role, the authors have added logic mostly as an afterthought. This may be an honest reflection of how research in multiagent systems is done these days, with computer scientists mostly ignoring the often impenetrable modal logic literature.

This book is a fantastic introduction to game theory where the authors are cleverly worried about the algorithms used to solve the problems. Therefore, it provides a great link among computer science, economic theory and operational research.

The structure of the book is based on very clear definitions, elucidating examples and theorems (not all proved, but several ones are proved). A great point of the book is the updated list of references in the last section of each chapter that can help the interested the reader find additional information about each topic is being taught in the book.

Chapter 1 introduces the problem of Distributed constraint satisfaction that will be used later in the book to compute the Nash equilibrium.

Chapter 2 is a basic introduction to dynamic programming and auction theory.

Chapters 3, 4 and 5 provide an introduction to decision theory and basic game theory. The bonus of these chapters is the list of algorithms that are presented in order to compute the equilibriums. It is worth mentioning that these chapters are as good as the best available introductions to game theory. All the basic topics are clearly covered.

Chapter 6 presents more specific representations of games. Some of them are very common in other books (such as Repeated games and Bayesian games), but others (such as congestion games) are not very common.

Chapter 7 deals with learning and teaching. The main topics of this chapter are two very uncommon topics in game theory books, namely reinforcement learning (you may find for instance in the fantastic book of Suton and Barto or Bertsekas) and evolutionary game theory (you may find in Game theory evolving - Gints, Evolutionary game theory - Weibull, Evolutionary Games and Population Dynamics - Hofbauer and Sigmund, but not the most common books of game theory).

Chapter 8 is about communication. This is a very interesting presentation of the topic Doing by talking (cheap talk) X Talking by doing (signaling games). Furthermore, an unusual topic of this chapter is presentation of the speech act theory dealing with rules of conversation and game theoretical view of speech acts.

Chapter 9 is about the difficult to aggregate preferences.

Chapter 10 is a clever introduction to mechanism design.

Chapter 11 is a very good introduction to auctions.

Chapter 12 presents coalitional game theory. This topic may be found for instance in A Course in Game Theory - Martin J. Osborne and Ariel Rubinstein, but the presentation in Osborne and Rubinstein is much more demanding.

Chapter 13 and 14 is about the logics of knowledge and belief.

Summarizing... This book is not another introduction to game theory, it is much more. It presents a lot of topics not common in other introductions and also presents several useful algorithms. Furthermore, it presents a good balance between mathematics and intuition.

While I generally appreciate Shoham's contributions, this book is disappointingly narrow. While building on the author's previous work in game theory, it ignores many of the major contributors to multiagent systems including Robert Axtell, Joshua Epstein, Nigel Gilbert, Kathleen Carley, Lars-Erik Cederman, Michael North, William Griffin, Scott Page and many more. If you are interested in a few particular approaches and techniques, the book may be useful, but it cannot be viewed as a full or adequate overview of this vital field.

have been interested in constraint optimization problems. MAS casted a fresh perspective on solving such problems, which led to more system-wise algorithms.

This book was very useful to me. It teach me everything we need to know about Multiagent systems, as game theory.