Customer Reviews

Artificial Intelligence: A Modern Approach

5 star:		(19)
4 star:		(12)
3 star:		(4)
2 star:		(4)
1 star:		(2)

 

 

This book is useful as an introductory textbook. After reading this book, I produced an game playing system with randomized behavior that was able to beat very good (but not expert) players at several different games with quite quick play. If you want to build a game opponent, or do some basic expert-system logic, or take a stab at a simple neural net classifier, this book can get you started.

There's a great deal more to AI than this book presents, and if you want to do anything really sophisticated this book won't get you there.

Everyone I have ever met with this book had problems with the binding coming apart. Complaints to the publisher have not produced books with any better bindings. It makes it difficult to use as a reference, because the pages keep falling out. If this was just one copy, I wouldn't mention it, but I have seen at least 30 copies with the same problem.

i got this book from my local bookstore. This book covers most topics in AI. It is more comprehensive than most AI book i have read. However i dislike the writing style employed in this book. The author needs to refine his writing style and review the contents to include more interesting examples and some humor. It is not wise to have long paragraphs in thick books. The author should also include pratical examples of AI applications and some AI codes instead of general theories. Still a great book !

Artificial Intelligence: A modern approach is definitely the most comprehensive book on AI I have come across. The latest edition covers everything from KR & Machine Learning to Robotics and Statistical learning, and has new chapters on Constraint Satisfaction Problems and Planning.
Artificial Intelligence has diverse branches and it's hard to see the correlations between them. This book manages to take a very innovative "agent" approach and tries to show some parallels between very disparate methods.
In my opinion, each branch of AI has made great progress by itself but more work needs to be done in trying to combine the different approaches together and create more comprehensive systems. The book does a very good job by making all these different approaches accessible in one volume.
This book also covers each topic in depth as well. Interested readers can always look at the exhaustive list of references for further reading. I have seen a variety of people use this book: from professors who have been in the fields for many years, to my friends in humanities who are just curious to know what AI really is. I highly recommend this book to anyone interested in AI.

in hindsight this was a good book...i used it a lot because i couldn't understand my prof very well...i got an A

This is my first book in the field of AI and I really enjoyed reading it in a three-week vacation (four weeks would have been a little more relaxed and appropriate). I am not a computer scientist but could follow most of the book easily (Well I do have a strong mathematical background.). The book is not very formal, but still based on sound arguments. It has a tremendously wide range of subjects incorporated: both within AI and between disciplines as differently as Philosophy and Operations Research. Nearly every one of them is dealt with at an excellent level. Only the physics parts in the perception and robot chapters are a little mediocre in presentation quality. Anyhow they are pretty much at the end of the book.

Also I liked very much the extensive historical overviews. The book contains lots of reading recommendations to explore further fields. I would have liked solutions to selected problems.

The books print is laid out fine and with a lot of care. A two-column layout might be for the width of the book more appropriate. I read the international edition (2nd), which is paperback. This book is very heavy and it took quite a bit of ability to hold it reasonably comfortly over the many reading hours required.

I had finished reading this book 3 months ago. My comments are :-
- It covers most of modern AI topics.
- It is quite easy to understand in some topics. Also quite hard in some topics.
- It is well-explained, but will be great if they used a more understandable way.

This book is suitable for a CS-student, or anyone who interesting in AI (But require some CS background).

The first edition of this book was exceptional, and this one respects its fine quality, and in some parts exceeds it greatly. The field of artificial intelligence has advanced considerably since the first edition, and many of these very exciting developments are reflected in the book. Anyone, regardless of their level of background or expertise in artificial intelligence will benefit greatly from its perusal.

Some of the better parts in the book:
1. The Website that is linked to the book, containing source code in LISP, PYTHON, and JAVA for the algorithms in the book.

2. The opening discussion on the history of artificial intelligence. The authors emphasize that the enhanced use of the scientific method is responsible for the rapid advances in AI in the last decade.

3. The historical summaries at the end of each chapter, which are fascinating reading by themselves and discuss ancillary developments not included in the main text.

4. The use of the "pointer hand" to emphasize important concepts. This should be helpful in the use of the book in a classroom setting.

5. The chapter on probabilistic language processing. The authors discuss an interesting example dealing with segmentation, namely that of finding the word boundaries in a sample text with no spaces. The Viterbi algorithm is used to do the word segmentation. The discussion on information retrieval in this chapter is also very interesting, and very important since the rise of the Internet and the consequent need for efficient search engines.

6. The discussion on the future developments in AI, given in the last chapter of the book. The authors discuss briefly a few of the more recent research topics, such as hierarchical reinforcment learning, anytime algorithms, and decision-theoretic metareasoning. Although short, the chapter motivates further reading on these topics, and references are given.

7. The discussion on "satisficing-making" decisions and its use in reducing the complexity of rational decisions. Although short, the discussion is helpful for those readers (like myself) who are not familiar with the concept of "satisficing".

8. The discussion of genetic algorithms, the authors pointing out their use widespread use in optimization problems, but observing, correctly, that more research needs to be done on genetic algorithms in order to determine when their use is optimal. The authors also discuss briefly the connection of quadratic dynamical systems with the performance of genetic algorithms. They also mention and interesting attempt by the researchers in genetic algorithms to justify population-based search in terms of Bayesian learning. References are given for both of these developments.

9. The discussion on constraint satisfaction problems is especially well-written, and this is good considering their enormous importance in enterprise and industrial applications in the last decade. Constraint logic programming is also discussed briefly.

10. The discussion on planning and acting in the real world is fascinating, especially the role of unexpected events or serendipity, which all successful planning must take cognizance of.

11. Applications of Bayesian networks have skyrocketed in recent years, and this justifies the authors thorough presentation of probabilistic reasoning and statistical learning in the book.

Some of the disappointments in the book:
1. The discussion on support vector machines is too short, given the importance of this new learning paradigm. Support vector machines can be pretty deep from a mathematical standpoint so this might be the reason the authors chose not to give the details.

2. For those interested in automatic theorem proving, the discussion on this topic might be too short, considering what has been accomplished since the first edition, especially in the proof of the Ramsey conjecture.

3. Only a brief overview of inductive logic programming is given. Through the use of the language Progol, inductive logic programming has made some headway recently, especially in the field of bioinformatics.

4. No discussion at all of abductive reasoning, which has become very important in applications recently, such as event correlation in telecommunication networks.

5. The chapter on "Philosophical Foundations" should have been omitted entirely, given the pragmatic but still rigorous approach the authors take in the book. The title assumes that the field of artificial intelligence is in need of some sort of philosophical underpinning or justification, which it does not. Further, the attempted refutation of philosophical arguments against AI is a misappropriation of time. Instead, efforts should be concentrated on building better thinking machines and developing more sophisticated algorithms in AI. The presence of thinking machines actually working in the field performing useful tasks will serve as a concrete counterexample against any "impossibility proofs" of the philosophers.

In the last paragraph of the book, the authors ask the reader to consider whether AI will succeed in its goals. I disagree with them somewhat when they say that AI has not yet had a major impact. It has. The problem is that many researchers are too modest to acknowledge their achievements, and many times are too easily convinced that what they have done does not in any way approach what could be called intelligent. As the authors remark, much work remains to be done. But much work has been done. One can say with confidence the future of AI will be very exciting and full of surprises, and witnessing this future will be a deep privelege.

Comments from my students

Comments on AI: A Modern Approach
Having used this book, I have found it to be an excellent overview of modern AI techniques. In fact, I knew nothing about many of the subjects in this book before this course, and this book has helped intrigue my interest sufficiently into the field.
This book uses an intelligent agent as the unifying theme throughout. Namely, the problem of AI is to describe and build agents that receive percepts from the environment and perform actions, and each such agent is implemented by a function that maps percepts to actions. It covers different ways to represent these functions, such as production systems, reactive agents, logical planners and decision-theoretic systems. It explains the role of learning as extending the reach of the designer into unknown environments, and shows how it constrains agent design. All in all, the book is quite readable and well organized.
In addition, the references to important works in the field were extremely useful, especially since many of them are now available online. What¡¯s more, these exercises are complete and thorough. They cover the entire breadth of the chapter, and each are detailed enough to challenge the students as more than a trivial problem.
While each topic is treated carefully and explanations are clear, unfortunately the book would run to many volumes if each topic was covered completely. Since this is not the aim of the book, this is not a drawback. AI:AMA is, after all, an introductory text to AI used in many undergraduate AI courses. Every one who is learning AI should have this book on their shelf.

Student ID: 0020309059

Comments on ¡°Artificial Intelligence-A modern Approach¡±

I think this book is the leading textbook in Artificial Intelligence I have ever read. The Artificial Intelligence is a big area, this book give me a whole view of it.
Besides all those good idea in this textbook, I think there is flaw in AIMI 2nd.
First of all the whole book doesn¡¯t have a detailed example that can lead reader the way to build AI application from scratch although there are many little examples and figures in the book. I think AI is a practical subject.
Second, Agent is hot and useful in AI now. I think there should be more pages on Agent and the application of Agent.

Comments on ¡°Artificial Intelligence-A modern Approach¡±
I think this book is fairly valuable both for the beginners and for the specialists. This book has the following three characteristics compared with the books on AI that I read before:
A new perspective of building intelligent agents: The book is organized around intelligent agents: what the ideas of agent are; how they make decisions and get information; how they learn knowledge. From this perspective a wide array of materials on AI is well organized and presented to the readers clearly and readably. Finished this book, I found myself have attained an overview of Artificial Intelligence.
This text book is organized using the latest principles of good textbook design. It includes historical notes at the end of every chapter, exercises, margin notes, a bibliography, and a competent index. The historical notes are much helpful to me, from which I could get much information on the corresponding domains and I could read more that I am interested.
The text book contains many valuable and comprehensive examples to illustrate the ideas it presents. The exercise is also good. They are both helpful to me to grasp the knowledge of AI.
Besides those small examples which are responsible for illustrate the basic ideas and concepts, one or two ¡®big¡¯ examples which illustrate the process of building practical artificial intelligent systems must be helpful to us.

I am a graduate student who major in Computer Science. I use the book ¡°Artificial Intelligence¡± as my textbook in the course of AI. After one semester¡¯s study, I find this book is a good one. As we all know, AI is a big field, and it is very difficult to write a good book of AI because that book must be a big one. But the style, the concept and the practicability of this book are excellent.
Firstly, when I read the book, I was deeply absorbing by its style. It describes the key concepts of a problem and its solution first, then give examples, and finally give a theoretical explanation, with rigorous theoretical analysis where appropriate. The point of the theory is always to get the concepts across and explain how they are used in actual, fielded systems. During the course of my reading, I had a clear mind. I could find the right reason why this problem could be promoted, what¡¯s the mathematic model of it, and how it could be solved in several methods.
Secondly, I enjoy the concept of the book. I think the writer must have a good sense of humor. I clearly remembered I fell into a big laughed when I read the example that ¡°if you left your car outside all the night, you might find your spare tyro missing next morning.¡± What a funny story it is!
Finally, the practicability of the book is good. Because of its comprehensive coverage and large number of detailed algorithms, it is useful as a primary reference volume for me.
Just as Prof. Tom Dietterich has said: ¡°I recommend it to anyone who wants to have an introductory overview of the state of AI. And I recommend it to experts in the field, who will enjoy its unified description of the field.¡±

Good coverage of rationality, the structure of agents, search algorithms, and decision trees. However, the book is highly theoretical and mathematical, it would not be appropriate for any business majors. I teach Information Systems in a college, I still use Kurzweil's book because I cannot find anything better suited to our needs.

This was my introductory text to AI and I admit that I've read only the first edition of this book. However, while browsing through the 2nd edition in a bookstore I found that the things that annoy me are still there - the pervasive taxonomy of tomatoes and the simplistic philosophy are just not appropriate for a book that tends to be considered the Bible of AI.

It is otherwise a great and well-targeted book and I enjoyed reading it. The second edition seems to have grown a bit bigger, including such topics as the EM algorithm, which I find very good.

I heartily recommend it to anyone new to AI, except if you're really afraid of finding out how tomatoes are classified :)