on 9 March 2005
Uniting information theory and inference in an interactive and entertaining way, this book has been a constant source of inspiration, intuition and insight for me. It is packed full of stuff - its contents appear to grow the more I look - but the layering of the material means the abundance of topics does not confuse.
This is _not_ just a book for the experts. However, you will need to think and interact when reading it. That is, after all, how you learn, and the book helps and guides you in this with many puzzles and problems.
on 27 January 2011
This is unique among the books I have encountered on information theory at this level, indeed one of the most reader-friendly accounts of any mathematically complex topic that I have ever read. The style makes the (difficult) subject matter very accessible. There are plenty of illustrations, which really do help with understanding, as well as examples with (mostly) answers provided, which are also valuable. The provision of answers to examples is frowned upon by purists, who say readers should just work them out for themselves, but we can't always succeed with every one, and I personally hate to be hung up on an example that I can't do.
To appreciate the benefits of Mackay's approach, compare this book with the classic 'Elements of Information Theory' by Cover and Thomas. That book was first published in 1990, and the approach is far more 'classical' than Mackay. It is certainly less suitable for self-study than Mackay's book. That said, I find Cover and Thomas very useful for providing the formal mathematical proofs of the theorems. After reading Mackay and understanding a topic, I then read Cover and Thomas on the same area and find the formal exposition of it, which complements Mackay nicely. I would not be without either book.
PS: I have subsequently discovered an excellent series of lectures by the author available online, essentially covering the main topics of the book. The lectures clarify the rather dense presentation in the book, and I have found them invaluable. They can be found by Googling "Mackay information theory lectures".
I have been able to use this book as extra background material for several courses of my final undergraduate year.
First I have been able to find a lot of usefull information on coding theory. Although this book isn't meanth to be a treatise on several coding, decoding techniques it gives the reader a lot of insight in the connection between coding and information theory. You won't find how matrix decoding algorithms, cyclic codes etc work but you will find out how the limits of information theory restrict coding theory.
I cannot compare the information theoretic approach to any other book as this was my first introduction but I can say the information theoretic treatise was a good read and I make myself strong I now have a solid information theory background.
Another course for which I have been able to use this book was a course on uncertainty reasoning. Mckay's book covers inference in great depth and introduces the reader to several different area's such as belief networks, decision theory, bayesian networks and several other inference methods. As before I cannot compare the ising, monte carlo like methods but it did give me a good introduction. Concerning the bayesian probability/inference, decision theory I can only say this is THE best introduction I have read!
I have read several introductions on Neural Networks (Kevin Geurny). This book keeps up with the standard set by several other good introductions.
Inference/Learning is a vast research area and this books gives a good introduction in all areas. Even as the part on neural networks may be as good as some other books on the topic I would definitely advise this book as for the same price you get so much more introductions to other learning techniques. The last thing which I like very much is the fact that several excercies are solved or come with hints which makes it for a student a very good book accompanying other courses. The author has a very clear writing style and knows when to add a good joke to make the reading more enjoyable.
My conclusion: if you are an undergraduate student interested in learning and inference -> "Go get this book asap!!!"
on 26 September 2008
This is an unqualified classic, to shelve with the likes of 'Structure and Interpretation of Computer Programs', 'Concrete Mathematics' and 'Mathematical Methods of Classical Mechanics'. If you are involved with, or interested in, high-end data analytics, then you _need_ this.
However 'high-end data analytics' does not even begin to do the book justice, so let me try again.
This is a magnificent compendium of fascinating stuff presented in a coherent information-theoretic framework. It covers everything from how digital television data compression and CD error correction work to a detailed commentary on neural networks, and discussion of principled AI methods such as clustering, Gaussian processes and probabilistic graphical models, together with Monte-Carlo techniques and a bunch of statistical physics. It even throws in a complete course in Bayesian statistics. It reads like a really good 'popular' 'science' book (I often wonder where the scare quotes should be) that doesn't bother to try to be popular.
In fact I bought this originally as bedside reading, for pleasure. It was only later that I actually used it for anything.